Japanese Democracy and Lessons for the United States: Eight Counterintuitive Lessons 2019049370, 2019049371, 9780367440053, 9781003008620

Citation preview

Japanese Democracy and Lessons for the United States

This book presents a collection of lessons on how best to run elections and politics, using examples from the Japanese experience and showing how elections operate in a non-Western democracy. Featuring extensive data and evidence from both Japan and the United States, the themes covered include one-party rule, ballot security and voting procedures, election regulations, malapportionment and gerrymandering, court interventions, voter attachments, and distortions of the public will by election rules. In so doing, the analysis challenges conventional wisdom in both Japan and the United States, highlighting surprising and counterintuitive findings from decades of observation. This book also explicitly compares Japan to other, similarly situated democracies. Japan is therefore not treated as a standalone case but, rather, the lessons from Japan are contextualized for greater understanding and can be used to inform discussions about comparative elections and democracy. Offering practical advice in relation to elections and the functions of democracy, Japanese Democracy and Lessons for the United States will be useful to students and scholars of Japanese, US, and comparative politics. Ray Christensen is an Associate Professor of Political Science at Brigham Young University, USA. His research focuses on aspects of Japanese elections including gender issues, gerrymandering, malapportionment, corruption, and electoral alliances.

Politics in Asia series

Risk Management Strategies of Japanese Companies in China Political Crisis and Multinational Firms Kristin Vekasi The Political Economy of Press Freedom The Paradox of Taiwan Versus China Jaw-Nian Huang Inequality and Democratic Politics in East Asia Edited by Chong-Min Park and Eric M. Uslaner India’s Maritime Strategy Balancing Regional Ambitions and China Shishir Upadhyaya Post-Politics and Civil Society in Asian Cities Spaces of Depoliticization Edited by Sonia Lam-Knott, Creighton Connolly, and Kong Chong Ho Chinese Constructions of Sovereignty and the East China Sea Conflict Czeslaw Tubilewicz Japan’s Cold War Policy and China Two Perceptions of Order, 1960–1972 Yutaka Kanda Japanese Democracy and Lessons for the United States Eight Counterintuitive Lessons Ray Christensen Governance and Democracy in the Asia-Pacific Political and Civil Society Edited by Stephen McCarthy and Mark R. Thompson For the full list of titles in the series, visit: www.routledge.com/Politics-in-Asia/ book-series/PIA

Japanese Democracy and Lessons for the United States Eight Counterintuitive Lessons Ray Christensen

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Ray Christensen The right of Ray Christensen to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Names: Christensen, Ray, 1960- author. Title: Japanese democracy and lessons for the United States : eight counterintuitive lessons / Ray Christensen. Description: Abingdon, Oxon ; New York, NY : Routledge, 2020. | Includes bibliographical references and index. Identifiers: LCCN 2019049370 (print) | LCCN 2019049371 (ebook) Subjects: LCSH: Elections–Japan. | Elections–United States. | Democracy–Japan. | Democracy–United States. | Japan–Politics and government–1945- | United States–Politics and government–1989Classification: LCC JQ1692 .C57 2020 (print) | LCC JQ1692 (ebook) | DDC 324.60952–dc23 LC record available at https://lccn.loc.gov/2019049370 LC ebook record available at https://lccn.loc.gov/2019049371 ISBN: 978-0-367-44005-3 (hbk) ISBN: 978-1-003-00862-0 (ebk) Typeset in Times New Roman by Swales & Willis, Exeter, Devon, UK

Contents

List of figures List of tables

vi vii

1 Introduction 

1

2  One-party rule

6

3  Campaign regulations

25

4  Drawing the boundaries of election districts

44

5  Supreme Courts and election procedures

72

6 Corruption

100

7  Democracy and political instability

123

8  Acceptance of election results

171

9 Conclusion

200

Index202

Figures

4.1  Population deviation of election districts from a national average 52 4.2  Deviation in votes cast from a national average 53 5.1 Ratio of the most malapportioned Japanese election districts at election day 89 5.2 Conservative votes in prefectures by levels of malapportionment, Japan 1958–2017 91 6.1 Distribution of margins of victory in Japan, Canada, and the United States 108 6.2  Margins of victory in Japan, by political party 109 6.3  Margins of victory in Canada, by political party 109 6.4  Margins of victory in the United States, by political party 110 7.1 Comparative vote percentages for the LDP and the Socialist/ Democratic opposition 127 7.2  Electoral volatility in Japan 129 7.3  Party switching and volatility compared, in Japan 131 7.4 Prime minister/presidential approval ratings in Japan and the United States133 7.5 Variation in approval ratings for one-year periods in Japan and the United States with the same prime minister or president 134 7.6  Highs and lows in approval rating for Japanese prime ministers 134 7.7 Unaffiliated voters (yearly averages) in Japan and the United States, 1947–2019137 7.8  Average turnout in rural and urban Japan, 1958–2017 139 7.9 The percentage increase in turnout explained by making an election district 10 percent more rural, 1958–2017 140 7.10  Comparing changes in turnout for rural and urban Japan 141 8.1  All democratic reversals 187 8.2  Governments elected by a minority of voters 188 8.3  Largest party not in government 189 8.4  Change in government inconsistent with party popularity 190

Tables

4.1  Redistricting rules in five democracies 51 4.2  Average deviations from the ideal district, population, and votes 52 4.3  A hypothetical redistricting in Ohio, using Japanese rules 63 4.4  Possible division of three Ohio counties with large populations 65 6.1 Predicted and actual numbers of ultra-close races, by party in Japan, Canada, and the United States 111 7.1 Troughs in Japanese prime minister approval ratings and party responses to those troughs 135 7.2 National swing and muting of that swing for conservative ­candidates in Japan 143 7.3 Dependent variable: district level turnout, Japanese House of ­Representatives elections, 1958–2009 168 7.4 Dependent variable: vote share won by LDP candidates in district ­elections in Japan’s 2003 House of Representatives election 169 7.5 Dependent variable: vote share won by Democratic candidates in ­district elections in Japan’s 2003 House of Representatives election170 7.6 Dependent variable: vote share won by Communist candidates in ­district elections in Japan’s 2003 House of Representatives election170 8.1  Consequential democratic reversals for selected countries 180

1 Introduction

Lesson 1: Countries should look to how other countries solve problems. In the realm of campaigns and elections, Japan is the country with the system most similar to the US system and provides many lessons and warnings for would-be political reformers in the United States. Ezra Vogel (1979) based his seminal work on the Japanese economic miracle on the premise that the United States was remiss in not studying what other countries do and learning from those examples. He contrasted American insularity with the well-documented Japanese penchant to study the best from abroad and adopt and modify those practices for a Japanese environment. Forty years later, Vogel’s observations remain true. It is rare for US elites, in any policy discussion, to examine or discuss how other nations deal with similar problems. The only times other nations are brought into the discussion are either as epithets to discredit an opponent’s position—socialized medicine like they have in the UK, wait lists like they have in Canada—or as stylized symbolic representations of a nirvana-like state—equality like they have in Sweden, renewable energy like they have in Germany. This book goes against the grain of expected discourse in two fundamental ways. First, I suggest that Americans should study how democracy functions in other countries as a way of informing and improving the functioning of democracy in the United States. Second, and perhaps even more startling, is my recommendation that the starting place for comparative study is Japan. This second recommendation would be even more startling to a Japanese person than to an American because the Japanese have long known that their political system is corrupt and inferior to the better political systems in other comparable democracies. In fact, the Japanese have a common saying that their economics is first tier, but their politics is third tier. How then can I possibly suggest that there is anything that the United States can learn from the Japanese example in the realm of politics? A first response is to note that many lessons can be learned from another example, both good and bad. Most of the lessons in this book are positive examples, many aspects of democracy work well in Japan, and reformers in the United States should consider emulating some of these practices. Some of the lessons, though, are negative lessons, practices in Japan that don’t work well.

2 Introduction Advocates of reform in the United States would be smart to pay attention to how some reforms have failed in Japan. This book is also part of a larger debate about democracy and its failings. Several of the lessons outlined in this book reflect a healthy appreciation of the potential excesses of democracy and the need for some anti-democratic checks on the unbridled power of the people. Other lessons, though, are a warning about the potential excesses of anti-democratic checks on democracy. The guiding principle of the observations in this book is practicality rather than theoretical coherence. My goal was to explain and learn from what works and what doesn’t work in Japanese electoral and democratic practice. I had little interest in shoehorning all of my findings into one, neat, theoretically coherent analysis of either the failings of democracy or the failings of anti-democratic checks on the powers of a democracy. This approach makes it difficult to label the perspective of this book as either pro- or anti-democratic. For example, Chapter 5 suggests that the US Supreme Court should learn from its Japanese counterpart and intervene less in politicized issues of election procedures. This lesson is a pro-democracy lesson, one that suggests reining in the power of one of the most prominent anti-democratic checks on unfettered democratic powers. Similarly, Chapter 7 teaches that as voters become detached from loyalty to parties or candidates, election outcomes can become more volatile unless political entrepreneurs modify their appeals to better reflect popular, short-term policies. Though this chapter highlights a prominent failing of democracies, that voters often support parties and candidates for reasons other than a careful consideration of party policies and proposals, ultimately the best solution for this problem is a democratic one. Greater potential volatility is addressed by the rise of a new type of politician that can effectively put forward and enact policy proposals, even in an electoral environment that is hostile towards any policies that take a long-term perspective and might require sacrifice from the voters. In contrast, Chapter 3 contains a strong anti-democratic message: many campaign regulations created by the people’s representatives to supposedly improve the quality of elections do not work well and are often counterproductive. The solution for this problem might be for the courts to step in and restrain excessive or misguided campaign regulation efforts (as the US Supreme Court has done) or to simply educate legislators about the consequences of some of their initiatives. Thus, the solution to this problem might involve more democracy or it might involve an anti-democratic check on democracy, but regardless the problem itself stems from a failing of democracy. Similarly, Chapter 8 uncovers a curious information paradox in elections. It seems that the more an election system gathers specific information from voters, the more likely that system will be viewed as illegitimate. This is an obvious failing of democracy, but one possible solution is perhaps even more troubling than the problem, designing electoral systems so as to minimize the people’s input into the process. Both the problem and the potential solution to the problem are clearly anti-democratic in their approach.

Introduction  3 Finally, the lessons of several other chapters are mixed or neutral in their approaches to democracy and its possible failings. Chapter 2 argues that oneparty regimes in a democratic system are just as healthy and functional as regimes in which the parties regularly alternate in power. Though expectations are that democracy will naturally lead to alternation in power, there is little in democratic theory that would suggest that the people can’t be satisfied with one political party and keep returning that same party to power in election after election. Chapter 4 describes the tendency of politicians to rig the boundaries of their election districts to advantage themselves or their political parties. The Japan-inspired solution to this common problem is for an anti-democratic entity such as a court to set some outer limits to the undesirable tendencies to draw self-serving district boundaries while also allowing the democratic process to function within those prescribed boundaries. This lesson clearly anticipates a positive role for both democratic and anti-democratic institutions in solving this problem. Chapter 6 also presents another Japan-inspired solution to one aspect of the problem of election corruption (stealing elections on election night). The solution involves gathering ballots into one public location before beginning the public count of the ballots. This solution is neither democratic nor anti-democratic but is simply a procedure that helps democracies to function better. A second reason why the United States can and should learn from the Japanese example of elections and democracy is to avoid the tendency to mistakenly ascribe success in one area to other unrelated areas. It a mistake to judge all of the facets of a country’s policies by one overall measure just as it is a mistake (and a very common one among political pundits) to judge all of the facets of a politician’s campaign just by one overall measure (victory or defeat). Because Donald Trump defeated Hillary Clinton for the US presidency in 2016, is it appropriate to conclude that Trump also had the best fundraising organization? Did he also have the best media campaign? Did he have the best advisors and the best organized campaign structure? Were all of his strategic decisions on where to allocate resources better decisions than those of Hillary Clinton? A fair and clear-headed analysis would have no problem identifying what Trump did best and what Clinton did best. However, the natural human tendency is to see success in whatever the winner did and failure in whatever the loser did. There are many deficiencies in Japan’s democracy, just as there are many deficiencies in US democracy. It is not my intent to argue that Japan’s democracy is overall superior to US democracy. Rather, I argue that there are many particular aspects of Japanese democracy that the United States can learn from. There are undoubtedly areas of democratic practice in the United States that the Japanese can learn from too, but chances are they have already looked at and studied those practices. The United States should do the same. Third, and perhaps most important, is the appropriateness of the Japanese example for informing democratic practices in the United States. At first blush it might seem more appropriate for the United States to learn from Canada,

4 Introduction or Britain, or France, or Sweden. These countries have longer democratic traditions, and the cultural and historical differences between the United States and these countries are much less than between the United States and Japan. However, Japan has one aspect that makes it an appropriate point of comparison for the United States, especially in the area of elections and campaigns. Politics in both the United States and Japan is candidate oriented rather than party oriented. Political parties play important roles in both countries, but their efforts are far overshadowed by the efforts of individual candidates. Candidates raise money, run campaigns, and take positions, all of which, in the aggregate, determine the success or failure of their respective parties. In contrast, elections in most other advanced industrial democracies are determined largely by the actions of the party and the party leader. Candidates in these countries play important electoral roles, but their roles are much less consequential than the actions of the political parties. There are, of course, other countries that share the US and Japanese penchant for candidate-oriented elections, Ireland and Brazil are two prominent examples that come to mind. These other countries would also be appropriate examples to study, but they are perhaps not as good a fit for comparison to the United States as Japan is. In addition to candidate-oriented elections, Japan and the United States also share a predilection for extensive fundraising and expensive campaigns, a detached and relatively apathetic electorate, generally two-party systems, conflict over election district boundaries, and extensive concerns about electoral corruption. Many of these similarities derive from the candidate orientation of both systems, but they also further buttress the appropriateness of the comparison to Japan. Nevertheless, in this book I adjust the countries of comparison for each chapter according to what the best comparison would be for the issues raised in that chapter. Every chapter is about Japanese practices, but Japan is compared exclusively to the United States on issues where this comparison is the most apt because of similarities in candidate-centered elections. On issues that involve all industrialized democracies, the comparison includes the United States, but it also includes Canada, Australia, New Zealand, and the countries of Europe. On issues of democracy generally or of election management, the comparison includes an even broader selection of countries. For example, discussions of the information paradox (Chapter 8) and ­ballot-handling rules (Chapter 6) all include examples from the advanced industrialized democracies and other countries. One-party rule (Chapter 2) and volatility (Chapter 7) are issues that affect all democracies, so the discussion in these chapters is focused on the advanced industrial democracies though some information from other democracies is also included. In contrast, the discussion in Chapter 3 about campaign regulations is exclusively about a comparison to the United States because the incentives to comply with or evade these regulations are strong in countries that have candidate-­ centered elections. The lessons of this chapter also apply to all countries that hold elections, but the specific findings do not transfer as well to countries with

Introduction  5 party-centered elections. Similarly, the discussions of court interventions in political issues (Chapter 5) and the drawing of election boundaries (Chapter 4) in theory apply to a broad range of countries, but Chapter 4’s analysis is ­limited to the advanced industrial democracies that use single-seat districts (Australia, Canada, United States, United Kingdom) in which boundaries are an extremely important political issue. Chapter 5’s analysis of court interventions also conceivably applies to all of the many nations that have high courts that review legislation for legality, but in practice the discussion is limited primarily to the two nations on opposite ends of this continuum, the United States with high levels of court intervention and Japan with extremely low levels of court intervention.

Reference Vogel, Ezra. 1979. Japan as Number One: Lessons for America. Cambridge, MA: Harvard University Press.

2 One-party rule

Lesson 2: One-party rule is a common outcome in democratic systems, and it produces effective and capable governments.

Most people believe that one-party rule is bad It is quite common to believe that the longer that politicians are in office the more out of touch they become with the people that they are supposed to be serving. A prominent example of this belief is the eight-year presidency of Ronald Reagan. The number of newspaper articles saying that President Reagan was ‘out of touch’ increased with each passing year of his presidency, rising from nine in his first year of the presidency (1981) to 65 in 1987, the second to last year of his presidency.1 This simple example from the Reagan presidency illustrates a common belief of pundits, politicians, political scientists, and voters: politicians, the longer that they are in office, become increasingly out of touch and non-responsive to the concerns of the voters. The Democrats may have controlled the US House of Representatives for 40 years from 1954 to 1994, but, it is argued, they finally lost control of that chamber in 1994 because of Democratic hubris. Similarly, the Republicans lost control of the same chamber in 2006 because of a similar tin ear to the mood of the public. Opposition politicians know how to attack a long-time incumbent: simply claim that she is out of touch with the concerns of regular people. The claim is much more likely to be believed if a politician or party has been in office for one or more decades. So pervasive is the belief that occasionally the politicians themselves buy into this belief after they have been kicked out of office. ­Margaret Thatcher observed in her first interview after she was removed as Britain’s prime minister ‘politics should not be a full-time job. I’ll tell you why—if it is, you get totally out of touch with the real world.’2 If even the victims of this political malady accept it as an explanation for their defeat, it would seem to be true that politicians become progressively out of touch the longer they are in office. Two conclusions follow from a belief in the natural decay of the responsiveness of parties and politicians the longer that they are in office: (1) there will be

One-party rule  7 regular alternation in power, and by implication, something is wrong with the political system if one party stays in power for too long; and (2) government policies and political outcomes are superior when parties alternate in power. In addition, characteristics of one-party regimes also seem to support these conclusions. The majority of one-party states are not functioning democracies. Countries ruled by only one political party also typically lack political or economic freedom. Human rights abuses and corruption seem to be more common in such countries. Economic growth is typically stagnant, even if the trappings of democracy such as elections are well established in the country. Even within the United States, the same pattern holds. The one-party states of the solid Democratic South were not known for their protection of civil rights nor their economic vitality.3 Rather, these one-party states were known for political corruption, stagnant economies, and poor performance on a host of socioeconomic indices ranging from infant mortality rates to adult literacy. Thus, alternation in power seems to be strongly correlated with both a functioning democracy and good government policies.

The Japanese example The experience of Japan, however, suggests a sharply contrasting view to this conventional wisdom. Japan, though solidly democratic, is a one-party nation. The Liberal Democrats have dominated not only national politics but also local politics. They have dominated Japanese politics not just for a long time but for multiple generations. If alternation in power is the natural result of a free and functioning democracy (or if one-party rule occurs only when democracy has been tampered with), why then in Japan, with all of its highly developed protections of democratic values, has one-party rule thrived? Is it possible that one-party rule could also result from a free and fair democratic system? Similarly, Japan has been one of the world’s most successful countries of the post-World War II period. If one-party rule is so detrimental to the political and economic fortunes of a country, why has Japan fared so well despite its domination by one political party? The Japanese case suggests several modifications to assumptions about oneparty rule and alternation in power. Reed (1993, 14) correctly noted that oneparty rule in Japan was not anomalous, citing one-party rule in Sweden or the US House of Representatives. One-party rule is actually quite common, even in democracies that hold free and fair elections. In fact, in the first half of the 1900s one-party rule was the expected outcome even at the national level of the US government. Frequent alternation in power at the national level has become more common only in recent decades. In addition, one-party rule is quite common, then and now, at the local or state government level. Many states or cities have been run by one political party for decades. Similarly, for every politician who is derided by his opponents as having been in power too long, there is a well-regarded politician whom people are sad to see leave at the end of her term of office. The enthusiasm for a presidential run by

8  One-party rule Michelle Obama, wife of former President Barack Obama is evidence of this ­phenomenon. Opinions about the appropriateness and incidence of alternation in power should be modified to reflect these realities. One-party rule also has distinct policy advantages that are easily overlooked in a rush to embrace alternation in power as the natural and appropriate outcome of a free democratic process. It is true that one-party rule is more prone to corruption and potentially less responsive to the concerns of the voters than governments that regularly alternate in power and face greater scrutiny and competition in elections. However, corruption and responsiveness are not the only facets of government that should be evaluated. Established one-party rule has the benefit of being much more efficient and less politicized than rule by governments that result from close races in highly competitive political systems. One-party rule also makes clearer the lines of responsibility in the political system, making it easier for voters to reward or punish those in power. Perhaps these advantages are not worth the increased chance of corruption under long-term one-party rule, but contrary to conventional wisdom, there are some important advantages to having one party consistently in power. In addition, good governments do not necessarily result from the actual alternation in power of governments. Rather, good governments likely spring only from the potential to alternate in power, for example, a government that operates in the presence of a competitive opposition party. Unfortunately, because there is such a strong correlation between non-democratic regimes and one-party dominant regimes, some assume that the benefits of competition cannot exist in a system unless there is alternation in power. This assumption is wrong; it is possible to obtain the benefits of political competition and still have one party that consistently wins elections (Caughey, 2018; Webb, 1997). Thus, criticism of one-party systems should be redirected to that subset of one-party governments that lack competition or the potential for alternation in power. It is fine to criticize Vladimir Putin for dismantling the democratic infrastructure that aided his political opponents in Russia, but the stable rule of the LDP in Japan or the Democrats in the state of Massachusetts is not the same political phenomenon. Competition, or the potential for alternation in power, is important for the proper functioning of a democratic system, however, actual alternation in power need not occur for the system to enjoy the benefits of a full democracy and political competition.

The frequency of one-party domination in functioning democracies Though there are many one-party states around the world, they seem to be rare in true democracies. Many dictatorships are one-party dominant regimes, whether or not they hold elections, but these one-party states are not the focus of this study. For example, it is not surprising that the PRI ruled Mexico for 60 years; the PRI rigged the elections to make sure that it would win, and party leaders were quite efficient in this task, winning every senatorial, gubernatorial, and presidential election held in Mexico from the late 1920s until 1988. Thus,

One-party rule  9 Mexico, during this period, is an example of a dictatorship that augmented its legitimacy by holding sham elections. In contrast, it is not an example of a one-party dominant regime functioning in a democratic environment.4 Setting aside the obvious and many cases of nondemocratic one-party states leaves as one-party regimes those true democracies in which one party is the consistent winner for an extended period of time. Pempel (1990) identified four democratic nations with such a one-party dominant system: Israel, Sweden, Italy, and Japan. Nevertheless, in each of these countries, the dominant party subsequently lost power (Israel 1984, Sweden 1976, Japan 1993, and Italy 1994). It would seem that one-party dominant regimes are just accidents. No party, in a true democracy, can consistently hold on to power for a long period of time. This conclusion, though, is factually unsupported. Japan is an important and obvious contrary example. The LDP had a brief hiatus from power in 1993 to 1994 (11 months), but by 1994 it was back in power as the main party in a coalition government. It lost its first national election in 2009 and was in opposition until regaining power in the 2012 election. Thus, the LDP has been in power for 60 of the past 64 years (94 percent, 1955–2019). Even more striking is the fact that three of the six men who served as prime ministers during the four years the LDP was out of power were former members of the LDP. Prime Minister Hosokawa, who led the non-LDP coalition government for ten of its eleven months in 1993, was not only a former LDP representative and governor, he was also the grandson of Japan’s prime minister during the late 1930s. Prime Minister Hatoyama who served from 2009 to 2010 was also the grandson of a former LDP prime minister who was one of the founders of the LDP. Though it is true that the LDP was out of power for two periods in the past 64 years, it is hard to distinguish the LDP’s replacement from the actual LDP when the replacement government was led, at least some of the time, by politicians who had served for decades as leading political figures in the LDP. Though Japan’s record of one-party rule is unparalleled, other countries have also had extensive periods of rule by one political party. Sweden’s Social Democrats had longer periods away from power than Japan’s LDP (1976–82, 1991–4, and 2006–14), but they also returned to power and have been part of the government for 70 of 86 (81 percent) years since 1932. The conservatives in France, though divided between two parties, have controlled the presidency for 39 of 60 years (65 percent) and the national assembly for 38 of 60 years (63 percent) since the founding of the Fifth Republic. The Liberals in Canada, though their rule has been regularly punctuated by the opposition Conservatives, have also ruled for a total of 83 of the past 124 years (67 percent).5 Even the United States has seen long periods of one-party rule despite the past 60 years of regular alternation of power in the presidency. In fact, the first 50 years of the 1900s were so clearly dominated by a political party (the Republicans from 1896 and the Democrats from 1932) that one-party dominance was seen by some as the natural order of politics, resulting from stable electoral

10  One-party rule coalitions (Lubell, 1965, 190–4), the absence of crises (Bowles, 1956, 9–11), or sectionalism (Schattschneider, 1956, 210). This one-party dominance characteristic of national politics in the United States for the first half of the 1900s extended also to state races. For example, a governor from a party different than her predecessor won election in only 24 percent of state elections from 1900 to 1945, but parties changed in 32 percent of the elections from 1946 to 2018. Though party control of state legislatures switched only in about 11 percent of elections in both periods,6 the percentage of states with unified state governments dropped from 78.8 percent of states in the earlier period to only 58.0 percent of states from 1946 to 2018. Clarke reports a similar pattern, finding 41 of 48 states with unified governments in 1945 but only 21 of 50 states with unified governments in 1986 (1998, 5). The case for alternation in power as a natural product of healthy political competition falls apart even further when party dominance is compared across different elections in the same country. In contrast to the presidency, the US House of Representatives was controlled for 40 years by the Democrats and then switched control, in 1994, only to be controlled for 12 more years by the Republicans. In fact, Democrats won the House for 58 of 62 years from 1932 to 1994, and Republicans have won the House for 20 of 25 years from 1994. If alternation in power were the natural outcome of a healthy democracy, why is it that voters want to alternate the political power of the presidency but not the political power of the House of Representatives? Similarly, is it more important to have control of the Senate alternate, which it has done eleven times (in elections) since 1932 in contrast to the House which has changed hands only eight times since 1932? Are representatives therefore less responsive to voters than senators as shown by the fact that control of the House is more stable than control of the Senate? The comparison to races at the state level is also striking. For example, from 1946 to 2019 the control of the presidency has switched political parties nine of eighteen times (50 percent), but for governors during the same period, party control has switched in only 32.3 percent of the elections. Another marker of one-party rule, unified governments, also shows this same pattern. At the national level, the United States has had only 28 (38.4 percent) years of unified government (House of Representatives, Senate, and presidency controlled by the same party) from 1946 to 2019. At the state level, however, states have had unified governments 58.0 percent of the time during the same period. This pattern of greater stability at the local level also extends to the separate voting record of each US state. For the same period, a state switched its winning party in a presidential election in 28.6 percent of the elections, but in races for the Senate, a state switched (from the party that previously won the Senate seat six years earlier) in only 22.0 percent of races. Even more telling are the few changes in the majority party in a state’s delegation to the House of Representatives, something that changed in only 11.8 percent of elections. The majority party in state legislatures has also been stable with control of a chamber switching parties in only 10.2 percent and 12.8 percent of upper house and

One-party rule  11 lower house elections respectively (1946–2018). Though many Southern states had decades of uninterrupted Democratic rule in their state legislative houses, many states outside of the South have also had long periods of one-party domination of their legislative chambers. For example, the Hawaii lower house has been controlled by the Democrats from 1954 to 2019. The North Dakota upper house was run by the Republicans from 1889 to 1984. Maryland’s upper house has been a Democratic bastion from 1899 to 2019. New Mexico’s upper house has had only one partisan switch in its history, having been run by the Republicans from 1888 to 1930 and then by the Democrats from 1932 to 2019. A couple of explanations present themselves for these patterns of data in the United States. First, it is easier to have alternation in power in executive positions (president or governor) than it is to have alternation in power in the control of a legislative body. A single candidate may lose an election for idiosyncratic reasons, but in an election for the many seats in a legislature, the effect of idiosyncratic factors can cancel out across all of the legislative races. The Republicans might have won the governorship of Hawaii in 2002 for the first time in forty years because they had an unusually strong candidate against a weak opponent, but in the many races for a legislature, one loss by a weak candidate can often be balanced by a win in another district where the weak candidate was in the other party. Another important difference is the amount of attention paid and the sense of accountability assessed against national leaders in contrast to local leaders. The race for the president receives much more attention than a gubernatorial race, making it more likely voters will know and have an opinion about the successes or failures of the incumbent and the incumbent’s party. The president, is also, rightly or wrongly, seen as being responsible for the economy, the crime rate, foreign policy, quality of schools, and many other issues and policies. In contrast, governors can often sidestep blame for some problems that are clearly national in nature. In addition, many voters don’t have a clear sense, beyond party identification, of the issues that separate gubernatorial candidates. The greater prevalence of one-party rule in local races can also be explained, in part, by the positions that the parties take at the national level. Competitive forces at the national level push parties to develop an appeal that will work with a majority of voters, leading national parties towards policy positions that are near the middle of voter preferences (Downs, 1957). Obviously, any national party that takes policy positions at the extremes, away from the typical voter, will have a difficult time cobbling together a coalition of a majority of voters to win the election. These positions taken at the national level, however, can have serious consequences for elections in a state or province that is disproportionately conservative or liberal. Thus, two parties that are intensely competitive at the national level may not be competitive in a local jurisdiction that is skewed in its political ideology away from the national average. Local politicians in such skewed jurisdictions may try to reposition themselves to the middle in their specific state or election district, but the dissonance between

12  One-party rule those local positions and the positions taken by the national party may doom such efforts of the local party or candidate to disassociate themselves from the positions of the national party.7 Japan, with its record of one-party dominance at the national level, illustrates some of these same trends, but, more importantly, Japan’s record shows how thoroughly the LDP has dominated Japanese politics, both at the national and prefectural levels.8 Power has alternated in only 18.2 percent of Japan’s elections for control of the Diet and the prime minister’s office.9 Prefectural delegations to the Diet’s lower house have also mirrored the LDP’s national dominance with the LDP (or its conservative predecessors) having a majority in 81.3 percent of all prefectural delegations from 1955 to 2019. Control of a prefectural delegation has switched in only 17.4 percent of elections over that same period. Japan’s governors have also been remarkably stable with a switch in the ruling party occurring in only 10.7 percent of gubernatorial elections held from 1955 to 2017. Opposition candidates won only 10.4 percent of Japan’s gubernatorial races, and in seven of Japan’s 47 prefectures, the opposition has never won the governorship since 1955.10 The record for control of Japan’s prefectural assemblies (1955 to 2017) tells an even more extreme story with the LDP winning the largest share of seats in 738 of 748 elections (98.7 percent). In only 2.3 percent of prefectural assembly elections has the largest party changed. This data shows that in both Japan and the United States, one-party dominance, rather than being rare, is quite common in both legislative bodies and local elective offices. Even in races for the national executive, Japan boasts an unparalleled record of one-party rule, and even the United States has had, in its history, long periods of one-party dominance. Though the levels of oneparty dominance in Japan are unprecedented, there are similar levels of oneparty dominance in some elections in the United States. Both countries have many examples of one-party dominance at both the local and national levels.

Competing theories to explain one-party dominance or alternation in power Despite the frequency of one-party rule, the theories that suggest that alternation in power is the normal pattern of politics remain well regarded. These theories, however, are flawed, and they must be modified to explain the relatively frequent occurrence of one-party rule. For example, the argument that politicians become out of touch and less responsive to the needs of the voters the longer that they have been in office doesn’t account for why politicians would become so lazy or uninterested. After all any incumbent government is led by politicians who were the most successful, in a previous election, at selling their agenda to the voters. Why would they stop caring about selling that same agenda to the voters after they had obtained office? Indeed, once they are in office, wouldn’t they have better resources to not only track the mood of the electorate but also respond to that mood?

One-party rule  13 Perhaps it could be argued that politicians no longer care once they have served for a long period and are nearing the end of their careers, or perhaps they become arrogant and corrupted with power. However, politicians are surrounded by advisors and allies who have supported them. Even if politicians are near retirement, they will work hard to make sure that their successors are ideologically similar. The arrogance argument is equally puzzling. No national leader comes to power without having already served in prominent positions in government or the private sector. If power were to corrupt such leaders so that they could no longer make sound judgments, then those people would likely have never been elected to an important political position. Yes, people can become arrogant and unresponsive when they are placed in positions of power, but typically people with those tendencies are discovered long before they rise to the top position in a nation, in a state, or even in a legislative district. A second problem with the out-of-touch explanation is that it flies in the face of the reality of very successful politicians who win elections throughout their careers and retire after a stunning career of electoral successes. For every politician who loses an election to the charge that he or she was out of touch with the voters there are five other long-time incumbents who are reelected by those same voters. Most US senators retire from office, voluntarily, after decades of service. Presidents Obama, Clinton, Reagan, and Eisenhower could have likely served third terms if they had been allowed. Yes, Democratic hubris might explain the Republican victory in the House of Representatives in 1994, but why did hubris suddenly become a problem after forty uninterrupted years of Democratic rule? Why wasn’t it a problem at year twenty or year thirty? It does not seem to be the case that the natural progression of all long-time incumbents is into arrogance or unresponsiveness. Many politicians retire after long careers with the continuing strong support of their constituents. Another explanation for why alternation in power might be expected comes from the inherent dynamics of competition in a democracy that holds free and fair elections. Downs (1957) points out that because ruling parties are forced to stake out positions on contentious issues first, because they are the government, the opposition can nearly always fine-tune its position, in response to the government position, to appeal to a slightly larger portion of the electorate. This can occur because either the voters have inconsistent positions (they want lower taxes but they also want subsidized mass transit) which makes it possible for the opposition to exploit the government position in a way that earns a majority of support in the next election, or it can occur because voters care differently about a variety of different issues. The opposition can exploit the preferences of various minorities that each care strongly about a specific issue, choosing positions on enough issues to cobble together a new majority composed of these groups that care passionately about their specific issues (58–62). Downs recognizes that seeing the proper path to victory may not be so quick or so easy with only the partial information that parties and candidates have about voters and their preferences in a specific election. He argues that a party may have to lose several elections before it identifies the potential coalition

14  One-party rule that would be large enough to take power (60). The fact, however, that voters’ preferences can be inconsistent or varied across several issues and the fact that the opposition has the luxury of choosing its policy positions in response to the government’s positions give the opposition an advantage and makes alternation in power the expected outcome of political competition. Downs’ argument is persuasive, but it ignores an important feature identified by Pempel (1990) in his work on one-party dominant regimes. Pempel points out that once a party is in power, it not only has better resources than the opposition with which to woo potential supporters, it also controls the agenda and can actually help shape the preferences of the voters to be more in line with the policy positions of the ruling party. He labels this phenomenon the virtuous cycle and points out how a well-established party with competent leaders should be able to transform its turn at power into long-term rule (16). If the electorate changes its position on an issue, the ruling party can respond to that change and use its more prominent position as the government to explain and enact its new policies, in response to changing voter preferences. If a new issue comes to the forefront, the government is best positioned to respond in a way to that new issue that the people will support. Indeed, it seems odd that once a party has attained power it should ever lose it because in a position of power it has better information and resources than the opposition to formulate and sell a winning policy position to a majority of the voters. Pempel’s argument applies to ruling parties, but it is similar in logic to the well-documented studies of the advantage of individual incumbents in securing their own reelection. Downs’ response is that parties find it difficult to change their positions, so a party or politician would find it difficult to accurately follow the changing positions of voters (105–13). Pempel recognizes this difficulty and agrees that there is a cost for parties or politicians to change their policies (1990, 6). Indeed, a party obviously incurs a cost by changing a policy position, but the argument about the ideological flexibility of parties turns not on whether a cost exists but whether the potential benefits of changing a position outweigh the costs of that same change. Political history is full of examples of parties and politicians that have successfully changed their positions on important issues. Two additional insights help reconcile these two competing explanations of whether one-party rule or alternation in power should be more likely. First is the importance of good information. Most politicians are willing to do what is necessary to assure their own reelection, including moderating their stance on issues in response to changes in the mood of the voters. It is difficult, however, to know five months in advance of an election what the most salient issues will be to the voters on election day. In retrospect, it becomes easy to identify the mistakes of a political campaign and argue that a party or candidate should have taken a different policy position, but candidates do not have the benefit of hindsight when they are forced to stake out their policy positions in advance of an election campaign. Incumbents, therefore, may lose elections and parties will alternate in power, not because incumbents are inherently more likely to be arrogant or more likely to be unresponsive to voters but simply because

One-party rule  15 incumbents, despite their informational and resource advantages, occasionally make mistakes and miscalculate what the most important issues and what the winning positions on those issues might be. In addition, some incumbents might be unwilling to change their position to the winning position because they would rather lose than give up certain core principles. It might be the case, then, that most incumbents can avoid losing in a specific election because of the numerous advantages of incumbency, but even a low failure rate will eventually lead to considerable alternation in power. Imagine that in any given year only 2 of 100 incumbents will take a losing political position or commit a serious political gaffe that would cause them to lose the next election. Even this low rate of error would result in 40 percent of the incumbents losing because of their errors after 20 years. Thus, even a very low error rate occurring over a long period of time will cause a significant attrition and naturally lead to some alternation in power. In addition to these theories is the importance of non-policy factors that also affect voting decisions. Both Downs and Pempel emphasize the importance of a politician matching her policy promises to the policy concerns of voters, but other factors certainly play a role in the political choices of voters. For example, Kuroda (2005) identifies cultural factors, Scheiner (2006) points to clientelist politics, and Curtis (1999, 141) fingers internal competition and change within the LDP as reasons the LDP has held on to power so long in Japan.11 Each of these other factors could be shoehorned into the label of policy preferences of the voters, but they seem to be operating according to a logic that is different than simply matching the policies that the voters want with the policies that the politicians espouse. For example, Scheiner’s argument seems to imply that the LDP satisfies supporters by providing tangible benefits to those same voters, thus earning their support, even if the LDP may not be entirely in tune with those voters’ policy preferences. Occasionally a politician might be more liberal or more conservative than her constituents but might continue to be reelected with broad support across the ideological spectrum because of what she has done in constituency service or securing federal funding for important local projects. Her appeal may be built on constituency service, pork barrel, or personal contact with voters, but each of these elements is not a policy position in the typical sense of the word. These other factors may also help explain some of the patterns of one-party dominance that exist in the United States and Japan. It could be that issues have become more important in US elections with increasing numbers of unaffiliated voters and the decline of strong ties to political parties. Voters of a previous era might have voted in greater numbers based on party identification. Pork barrel politics and constituency service might also have been stronger factors in deciding how to vote in previous decades. Similarly, it might be the case that, in local and legislative elections, not only is less attention paid to the specific policy issues (and hence greater attention is paid to party affiliation) but also constituency service and pork barrel politics might be greater factors in the political calculations of voters. It is perhaps easier, in local or legislative

16  One-party rule races, for a candidate to run based solely on what he or she has done tangibly for the district and avoid or downplay policy positions that might be a less reliable means of winning support.

The advantages and disadvantages of one-party governments Having established that one-party rule is much more common than many realize and having explained why one-party rule might occur in some democratic systems, I now turn to the consequences of one-party rule. Is it true that oneparty dominant regimes are worse (less responsive to the voters, less productive in improving political and economic life in country) than more competitive systems in which alternation in power is a regular occurrence? This analysis is complicated by two factors: (1) a predisposition, especially in the United States, to view the concentration of power as inherently evil, and (2) a tendency to conflate non-democratic and democratic one-party dominant regimes into the same category. When both of these caveats are remembered, the record in Japan and other jurisdictions does not support the conclusion that alternation in power produces superior policy outcomes. The political ideology of most Americans is built on a strong distrust of political power. This ideology taints any discussion of one-party rule. Not only does the United States have separation of powers with its checks and balances at the national level, but power is further devolved between federal, state, and local governments. Americans distrust government and celebrate the frequent checks on the power of political leaders. If some checks on that power are good, more checks would seem to be better. These attitudes are not shared in many other advanced industrial democracies, including Japan. Competition among political leaders and checks on the power of the ruling party exist in each of these countries, but rarely do the checks reach the level of development that exists in the United States. It is important to remember that the United States is the extreme case in the level to which power is fragmented.12 Despite substantial evidence that the US president has gotten stronger in recent decades, the executives of most comparable advanced democracies have more relative power (compared to other domestic actors) than the US president (Howell and Pevehouse, 2007). The US president is powerful only because he or she leads a very powerful country, but that power is shared more fully with other domestic political actors than is common in most other democracies. In addition, one-party rule in democracies is often inappropriately conflated with authoritarian rule. The phrase one-party rule conjures up images of the PRI rule in Mexico, Communist rule in North Korea, or the rule of the Democratic Party in the American South for three generations. Indeed, each of these is an example of a one-party state, but these are not one-party states created by the operation of a democracy holding free and fair elections. An evaluation of one-party rule in contrast to alternation in power should not be based on a comparison of democratic competitive systems with undemocratic one-party systems. Such a comparison is misleading because it compares democracy with

One-party rule  17 autocracy rather than comparing one-party rule with alternation in power. The governments of South Carolina or Mexico fifty or one hundred years ago were non-responsive governments that did little to enhance the health and prosperity of their peoples, but neither of these governments were democratically elected. Rather, the better examples of one-party democratic regimes are Sweden or Japan, or in the United States the domination of Kansas or North Dakota by the Republican Party. For example, in recent decades Utah has trended as one of the most Republican states in the United States, and many Democrats decry this trend as somehow inappropriate, unseemly, or simply a bad idea.13 Prominent Democrats in Utah toyed with the idea of not running candidates in many districts in the 2000 state legislative elections in order to show voters how bad one-party rule was in Utah.14 However, an objective observer would be hard pressed to claim that politics is worse in the one-party state of Utah than it is in neighboring Nevada which is evenly divided between the Republicans and the Democrats. Alternation in power, which occurs regularly in Nevada, does not necessarily produce better political outcomes for the voters. Similarly, Massachusetts is clearly a one-party state with the same complaints about the ruling Democrats that are made about the ruling Republicans in Utah.15 Is it obvious, however, that politics is worse in Massachusetts than it is in other more competitive states such as Florida or Ohio? Thus, a key feature of one-party rule in true democracies is not an actual alternation in power but the possibility of alternation in power. As V. O. Key (1950) noted early on in his study of the one-party Democratic state governments of the South, in the states with a functioning opposition, the Democratic party organization was better than in those states without any organized opposition. Southern politics, bad as it was, was better in states in which alternation in power was a possibility, even if it never occurred. Thus, the crucial feature seems to be only the possibility of competitive elections and alternation in power. As long as this check on the power of the ruling party exists, the benefits of political competition follow, even if alternation in power is merely a possibility and never realized. Furthermore, the US fixation on the desirability of alternation in power ignores the several benefits of one-party rule. A first benefit is efficiency. The American political system is famous for its gridlock, resulting from the many, many checks on power that exist in the system. Most Americans accept the gridlock as a cost of having a more responsive and deliberative system, but certainly an important facet of responsiveness is the ability of a government to enact timely and coherent legislative proposals. In fact, Steinmo even argues that this lack of efficiency is what causes Americans to dislike or distrust government (1994). A second, and even less recognized, benefit is the advantage of less frequent politicization of issues in a one-party dominant system. In a closely competitive system, it will be a common occurrence that issues that both parties might have agreed on will quickly become intractable as parties stake out positions in

18  One-party rule an attempt to embarrass their opponents or deny them a legislative outcome that might help the party in an upcoming election.16 Thus, the political debate on these issues quickly transforms from what would be the best for the country to who will win politically on this issue, and issues that should have been dealt with last year are delayed until after elections have occurred. In contrast, under one-party rule, much less can be gained by such political grandstanding. Parties and politicians still, of course, look to ways to embarrass their opponents or deny them political victories, but an opposition party with only one third of the seats in a legislature will not be able to delay or deny its political opponents the legislative victory that they seek. Thus, the emphasis turns from this politically dubious task to the more constructive negotiations over what modifications to the legislation should be made. The minority can later claim responsibility for the modifications, and the majority can tout its legislative accomplishments. Because the legislation is going to pass, both parties work to make a better bill rather than just score political points off their opponents at the cost of sound or comprehensive legislative reforms. For example, despite the enormous unpopularity of tax increases, Japan has been able to implement a new sales tax and increase the taxation level to 5 percent, then 8 percent, and now 10 percent. Each of these increases came at a high cost to the ruling political party, but the LDP was able to implement the initial tax as well as the increases to 5 and 10 percent. Similarly, Utah’s Republican lawmakers have undertaken to restructure the state sales tax to cover the gradual and inexorable change from most purchases being goods to many more services being purchased. Tax increases are some of the most difficult and unpopular policy decisions for politicians to make, yet they have been successfully done in low tax jurisdictions by conservative parties in part because the ruling party controlled all of the levers of power and was relatively secure in its position as a ruling party. Of course, these benefits of one-party rule must be balanced against the obvious costs of greater possibility of corruption when a party doesn’t face a realistic chance of losing power. Here, again, lies the importance of competition and the possibility of alternation in power. Even in Massachusetts, an occasional Republican will win office, and even in Utah, occasionally a Democrat will win her race. When competition exists, the threat of widespread corruption declines sharply, even if actual alternation in power is a rare occurrence. In fact, it is probable that the incentives for certain types of corrupt behavior, specifically electoral corruption, are considerably weaker in one-party dominant states because the elections are less competitive. The incentives to resort to electoral corruption are certainly much higher in highly competitive regimes where alternation in power is a regular occurrence and a small amount of electoral corruption could be the deciding factor in a closely contested race. The actual empirical record suggests that one-party regimes (if non-­ democratic regimes are excluded) produce as good if not better political outcomes than their highly touted competitors—regimes in which power

One-party rule  19 alternates. Japan is the first and best case in support of one-party dominant regimes. It is hard to think of a nation that has been more successful, not only in terms of economic growth, but also in quality of life indicators. Japan successfully transformed after World War II from an authoritarian to a highly developed democratic system. Japan rates at the top or near the top of most nations on a host of indicators. The Japanese economy is still the third largest in the world, even after two decades of on and off again recession and deflation. Japan’s literacy rate is nearly universal despite an extremely difficult written language, and Japan boasts the highest life expectancy rates of any country in the world. Japan is also a world leader in the diffusion of technology among its citizens.17 Comparing Japan to the most similar nations of Western Europe, North America, Australia, and New Zealand shows the strong performance of Japan. Japan is in the top group of even these advanced industrial democracies on many important indicators, including lower inflation, lower unemployment, lower infant mortality, lower income tax levels, higher life expectancy, fewer homicides, and higher total GDP. Japan is at the middle of this group of nations on indicators such as: the Human Development Index, the Human Poverty Index, total external debt, GDP growth, industrial growth, per capita GDP, death rate, total fiscal burden, and levels of corporate taxation.18 Japan is in the bottom third of advanced industrial democracies on only a few indicators: suicide, perceptions of corruption, and the empowerment of women. Japan does lag in several areas. It has fewer women in its parliament than any other advanced industrial democracy, a major reason for its low rating for the empowerment of women. It also rates near the bottom of the advanced industrial countries in terms of its corruption, or at least the perception of corruption. Japan has higher relative suicide rates and greater levels of pessimism than other comparable countries. Indeed, there is much that can be improved in Japan, but most countries would gladly change their position on most of these indicators with Japan. Japan is not just an economic powerhouse. It is also stable, peaceful, relatively crime free, and healthy; all indicators that reflect well on the government and its responsiveness to citizen concerns. Of course, each of these indicators can be argued to be the result of factors that have little to do with the actions of governments, and measures of happiness, suicide rates, and longevity may have more to do with cultural practices and attitudes than with actual government policies or programs. Indeed, there is no clear way to measure the actual outcome of politics that disentangles government actions from the myriad of other factors that also affect any of these indicators. Nevertheless, it is hard to imagine that Japan could have done so well on so many indicators with an unresponsive or irresponsible government. It is also telling that Japan’s worst performance is on one of the indicators (corruption) that is directly attributable as a negative feature of one-party dominant regimes, suggesting that these indicators are picking up, in part, one of the negative impacts of one-party rule on the economic and political welfare of the Japanese.

20  One-party rule The case for one-party governments is also buttressed by other one-party nations. Among the advanced industrial democracies, Sweden, France, Canada, and Italy stand out as having significant periods of one-party rule. This list of countries is diverse, suggesting that neither one-party rule nor alternation in power tends disproportionately towards better or worse policy outcomes. Though Italy’s performance on some of these indicators has been comparatively low, Sweden, France, and Canada are similar to Japan in their strong records on many of these same indicators. Similarly inconclusive results are obtained by a statistical analysis of the 50 US states by their periods of unified government, a form of one-party rule. Some states have had relatively long periods of one political party controlling all of the levers of power in that state, and other states have had much less time under the control of only one political party. For example from 1980 to 2019, Georgia and South Dakota have had one-party rule 95 percent of the time. At the opposite end of the spectrum are New York (14 percent) and Nevada (16 percent).19 I used a statistical procedure called a regression analysis to predict three different measures of state-level corruption and state-level GDP growth for roughly the same periods. In four separate regressions the measure of unified state government failed to be a statistically significant predictor of either corruption or economic growth.20 Other studies have also evaluated one-party rule as a causal factor of policy outcomes, and some have failed to find any disadvantages for one-party rule in comparison to more competitive states. For example, Alt and Lowry (1994) find that one-party state governments respond better to budgetary pressures than states with divided governments. Obikili (2019) finds better economic growth in South African municipalities that have less political competition. Dye (1984, 1099) identifies five other studies that failed to find a correlation between the degree of competitiveness in a state and spending or taxation policies of the states. Dye also finds that within competitive states, party positions are correlated with actual welfare spending levels, a finding that suggests greater responsiveness in competitive systems. Besley, Persson, and Sturm (2010) also find that one-party states have higher taxes, lower capital spending, fewer right to work laws, and lower income growth. Their study is largely historical, and they obtain weaker results outside of the one-party states of the US South. In addition their findings about higher taxes and lower capital spending are less persuasive as indicators of good government, though they certainly are related to pro-growth economic policies. Dash and Mukherjee (2015) analyze the Human Development Index, a more comprehensive measure, and find that political competition in Indian states is correlated with higher growth in the index. They also survey the extensive literature on the relationship between political competition and economic growth and conclude that the results are mixed. Some of these studies, though, measure economic growth contrasting democracies with authoritarian regimes rather than comparing the level of political competition within democracies.

One-party rule  21

Conclusion The Japanese record, coupled with important examples from the 50 US states, shows several counterintuitive outcomes. First, one-party governments are much more common than is commonly thought. Second, the performance of one-party governments does not suggest that these governments are consistently less responsive to their people or produce worse policies than those produced through a more intensely competitive process. Both of these outcomes run counter to theory and conventional wisdom. Most observers expect democracy to lead to competition and expect that the more competitive a ­system is, the more natural and better the outcome of the system will be. In practice, however, it does not seem to be the case that if some competition is good, more is necessarily better. The Japanese example seems to make the more subtle point that competition is necessary for the proper functioning of democracy, but that competition need not result in the alternation of power. It is only necessary for competition to exist; it is only necessary for alternation in power to be a theoretical possibility. Beyond this threshold, the struggle between competing political ideas can occur within a dominant political party (as occurs in Japan or states with one-party governments) or between political parties (as occurs in US presidential races). This competition of ideas at either level, coupled with the existence of formal electoral competition (the possibility that a ruling party could be replaced in an election), makes governments adequately responsive and responsible to the voters. Critics of one-party governments should tone down their rhetoric. They are, of course, welcome to criticize the ruling party using whatever arguments they feel are correct or are likely to sway voters. The facts, however, do not seem to justify a claim that one-party governments either do not represent the voters or produce inferior policies. Even the evidence that one-party governments are more susceptible to corruption than governments that alternate in power more regularly is inconclusive. The Japanese case suggests that corruption is more common, but in the United States, there is no significant correlation between one-party government states and levels of corruption.

Notes  1 The numbers dropped to only 15 when attention turned to the race for his replacement.   2 Nicholas Timmins, Independent December 15, 1990, 5 ‘Parliament and politics, Thatcher says that she will go on fighting for her beliefs.’   3 Besley, Persson, and Sturm (2010, 1344) estimate a long-term increase in per capita income of 15 percent in the South as a result of the political competition created by the Voting Rights Act.  4 Another example of one-party dominant regimes is countries such as South Africa where a party that led the country in an independence battle or revolution maintains control because of its historical legacy (Giliomee and Simkins, 2005, 2;

22  One-party rule Pempel, 1990, 5). These cases are not relevant to an analysis of one-party rule in advanced industrial democracies.   5 Smiley (1958) argues that one-party domination accurately characterizes Canadian politics at both the national and provincial levels.   6 10.8 percent of elections from 1900 to 1944 compared to 11.5 percent of elections from 1946 to 2018.  7 Robbins and Norpoth (2010) do not make this causal argument, but they do find regional anomalies in US political history in which one party dominates congressional elections in one part of the country while the national balance between the two parties is much closer to parity.   8 The Japanese equivalent of a state or province is called a prefecture. Japan has 47 prefectures.   9 Several caveats are important to note. Japan is not a two-party system, so it has been quite common for the LDP to fail to win a majority of seats. I measured whether the election resulted in the LDP being ousted as the ruling party or part of the ruling coalition of parties. The LDP lost power in 1993 as the result of an election, but it regained power in 1994 without a new election by simply enticing some members of the ruling opposition coalition to break ranks and join an LDP-led coalition government. There have been 22 elections from 1955 to 2019. The LDP loss in 1993 and then the coalition change in 1994 count as two of the four alternations in power, leading to the 18.2 percent number that is reported in the text. 10 Most Japanese governors run as independents, but for many of them their party allegiances can be ascertained by party endorsements. I counted opposition candidates as those endorsed by a major opposition party that were not also endorsed by the LDP. When the LDP made no endorsement (suggesting a split in the local LDP organization) I did not count the winner as an opposition candidate unless the winner had been endorsed by a major opposition party and a loser had been endorsed by the LDP. 11 Kuroda’s argument is that Japanese politics has two layers and that Western notions of democracy and competition become merely a façade behind which Japanese norms of consensual resolution of conflict operate. Curtis points to factional competition within the LDP and Japan’s multimember districts which facilitated the replacement of discredited LDP leaders with new LDP leaders without having to actually remove the LDP from power. 12 Steinmo (1994) disagrees that this distrust of power stems from cultural reasons, but he reaches the same conclusion, that power is diffused, producing less efficient and efficacious politics. For Steinmo, inefficient and ineffective political outcomes are what produces the distrust of government, not the distrust of government producing divided power and ineffective policies. 13 ‘Shame on Utah’s GOP’ by Matt Asay, Deseret News, September 28, 2001, A16; ‘Overkill by the GOP?’ by Ray Diehl, Deseret News, November 28, 2001, A18; ‘Hear both sides,’ Salt Lake Tribune, January 25, 2002, A14; ‘Mathesons looking good to Dems’ by Ted Wilson and LaVarr Webb, Deseret News, July 6, 2003, A1. 14 ‘Will Utah Demos run no legislative candidates at all in 2000?’ by Bob Bernick Jr., Deseret News, March 27, 1998, A23. 15 ‘“I ask you to stand with us on this election,” Kerry Healey Republican nominee for governor’ by Frank Phillips and Scott Helman, Boston Globe, April 30, 2006, B1; ‘Healey’s dire warnings and long odds’ by Scot Lehigh, Boston Globe, April 30, 2006, E9; ‘Dems will pay price at the ballot box,’ Boston Globe, June 14, 2004, 22.

One-party rule  23 16 Hinchliffe and Lee (2016) find greater polarization in state legislatures where parties are more competitive. 17 The World Bank’s Digital Adoption Index. 18 The Human Development Index, Human Poverty Index, and the Gini coefficient of inequality are reported in the Human Development Report 2005 at http://hdr. undp.org/reports/global/2005/under Human Development Reports, United Nations Development Programme, accessed September 19, 2006. 19 It might seem puzzling that heavily Democratic New York is the US state with the longest period of divided government, but this result occurs because of the long domination of the state senate by Republicans, with Democrats only winning control of the body in the 2018 elections. 20 The three corruption measures were FBI numbers of local politicians convicted for corruption crimes from 1976 to 2010 using (1) raw numbers, (2) ratio of state population, and (3) ratio of state and local government employees. The measure of state-level GDP growth was included in a regression by itself and in a regression with state population growth (from 2000 to 2010) as a control variable.

References Alt, James E., and Robert C. Lowry. 1994. “Divided Government, Fiscal Institutions, and Budget Deficits: Evidence from the States.” American Political Science Review 88, no. 4 (December): 811–28. Besley, Timothy, Torsten Persson, and Daniel M. Sturm. 2010. “Political Competition, Policy, and Growth: Theory and Evidence from the US.” Review of Economic Studies 77, no. 4 (October): 1329–52. Bowles, Chester. 1956. American Politics in a Revolutionary World. Cambridge, MA: Harvard University Press. Caughey, Devin. 2018. The Unsolid South: Mass Politics and National Representation in a One-Party Enclave. Princeton, NJ: Princeton University Press. Clarke, Wes. 1998. “Divided Government and Budget Conflict in the US States.” Legislative Studies Quarterly 23, no. 1 (February): 5–22. Curtis, Gerald L. 1999. The Logic of Japanese Politics: Leaders Institutions, and the Limits of Change. New York: Columbia University Press. Dash, Bharatee Bhusana, and Sacchidananda Mukherjee. 2015. “Political Competition and Human Development: Evidence from the Indian States.” Journal of Development Studies 51, no. 1: 1–14. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Addison Wesley. Dye, Thomas R. 1984. “Party and Policy in the States.” Journal of Politics 46, no. 4 (November): 1097–116. Giliomee, Hermann, and Charles Simkins. 2005. The Awkward Embrace: One-Party Domination and Democracy in Industrializing Countries. London: Routledge. Hinchliffe, Kelsey L., and Frances E. Lee. 2016. “Party Competition and Conflict in State Legislatures.” State Politics and Policy Quarterly 16, no. 2 (June): 172–97. Howell, William G., and Jon C. Pevehouse. 2007. While Dangers Gather: Congressional Checks on Presidential War Powers. Princeton, NJ: Princeton University Press. Key, V. O. 1950. Southern Politics in State and Nation. New York: Knopf.

24  One-party rule Kuroda, Yasumasa. 2005. The Core of Japanese Democracy: Latent Interparty Politics. New York: Palgrave Macmillan. Lubell, Samuel. 1965. The Future of American Politics, 3rd ed. Revised. New York: Harper & Row. Obikili, Nonso. 2019. “The Impact of Political Competition on Economic Growth: Evidence from Municipalities in South Africa.” South African Journal of Economics 87, no. 1 (March): 3–21. Pempel, T. J., ed. 1990. Uncommon Democracies: The One-Party Dominant Regimes. Ithaca, NY: Cornell University Press. Reed, Steven R. 1993. Making Common Sense of Japan. Pittsburgh, PA: University of Pittsburgh Press. Robbins, Suzanne M., and Helmut Norpoth. 2010. “Balance or Dominance? Party Competition in Congressional Politics.” Political Research Quarterly 63, no. 2 (June): 316–27. Schattschneider, E. E. 1956. “United States: The Functional Approach to Party Government.” In Modern Political Parties: Approaches to Comparative Politics, edited by Sigmund Neumann, 194–215. Chicago, IL: University of Chicago Press. Scheiner, Ethan. 2006. Democracy without Competition in Japan: Opposition Failure in a One-Party Dominant State. New York: Cambridge University Press. Steinmo, Sven H. 1994. “American Exceptionalism Reconsidered: Culture or Institutions?” In The Dynamics of American Politics: Approaches and Interprentations, edited by Lawrence C. Dodd and Calvin Jillson, 106–31. Boulder, CO: Westview Press. Webb, Samuel L. 1997. Two-Party Politics in the One-Party South: Alabama’s Hill Country, 1874–1920. Tuscaloosa, AL: University of Alabama Press.

3 Campaign regulations

Lesson 3: Strict campaign regulations often become counterproductive, forcing campaign activities underground and becoming tools to punish political opponents.

Introduction All Americans decry the state of US elections. The presidential campaigns are far too long. Millions of dollars are raised and spent in seemingly mindless campaign activities. Though some television advertisements are clever and entertaining, most insult the intelligence of many voters. Presidential debates are an elaborately choreographed exercise in spinning the media and soundbite politics. Yard signs, phone calls, and door-to-door canvassing, though effective, contribute to the sense that the election is not about choosing options but simply manipulating apathetic voters. In contrast to this stunning display of election dysfunctionality that is the United States, the hyper-regulated election system of Japan must seem to be the most enlightened election system in existence. Campaigns for the House of Representatives are limited in length to only a mercifully short 12 days. That is not a typographical error. There are only 12 legally allowed days of election campaigning for these elections. The only advertisements are carefully regulated introductions to specific candidates, aired minimally, and party advertisements. Money is prevented from playing a role in elections by state subsidies for the publication of candidate introductions and postcards mailed to voters. In fact the only significant monetary barrier to running for office is the requirement of a campaign deposit of 3,000,000 yen (approximately US$30,000) that is forfeited by candidates who win fewer than 10 percent of the votes in a district. Government subsidies are also given to political parties to help defray the additional costs of election campaigns, and contributions to parties and candidates are strictly regulated. In practice, though, the strict regulations of nearly all forms of campaigning in Japan produce even worse results than the much-criticized US campaign system. Because nearly all effective means of campaigning are prohibited or so narrowly circumscribed as to be meaningless in determining the outcome of

26  Campaign regulations an election, all successful candidates put nearly all of their money, time, and efforts into underground campaign activities, what Japanese laws call ­‘political activities,’ and what pundits call pre-campaign activities. Thus, Japanese politicians are almost always campaigning, even if they are required by law to never mention words such as candidate, office, election, support, or endorsement. They raise large sums of money, as much as many US candidates raise, to support their pre-campaign activities. The actual activities of the campaign period become a meaningless exercise in doing what one is expected to do in a campaign, with almost no discussion of issues. Finally, and perhaps most disturbing, the heavy hand of government regulation in almost every facet of campaign activity means that many regulations are retained or expanded only because they harm political opponents, typically those out of political power. Thus, the lesson to be learned from Japanese campaign regulations is a negative lesson. Good intentions to clean up and improve elections, in many situations, can lead to harmful, unintended consequences. These regulations, born of good intentions, can even become tools of partisan punishment. Though there are many regulatory efforts that work well in a variety of settings, the incentives that surround highly competitive electoral environments increase the probability that well-intentioned campaign regulations will backfire. The Japanese lesson shows that though some campaign regulations are well-crafted and work well in practice, many suffer from fundamental flaws.

A cost–benefit analysis of government regulations Looking at the costs and benefits of government regulations can help explain (1) why effective campaign regulations are particularly difficult to craft in highly competitive electoral systems and (2) why some campaign regulations in those settings, nevertheless, are effective and work as intended while others become counterproductive. This analysis seems obvious: if the benefits of c­ omplying with a government regulation are greater than the costs of non-compliance, then candidates, voters, parties, citizens, whoever is the target of the regulation, will comply with the regulation. This simplistic analysis leads to the only partially correct conclusion that if there is significant non-compliance, then the costs of non-compliance (greater fines, prison sentences) need only be increased to obtain the desired outcome. Such a superficial analysis, however, ignores the complexity of the components that are built into the general and poorly defined concepts of costs and benefits. There are the costs and benefits of compliance, but these costs and benefits are always compared to the comparable costs and benefits of both non-compliance and alternative behaviors. For example, the government can ban the sale of high-flow showerheads to encourage water conservation. The obvious cost of compliance is purchasing and installing a new, conforming showerhead and enduring low-flow showers. The only benefits of compliance (since there are no enforcement actions taken against consumers) are the psychological benefits of being law abiding and taking personal action to

Campaign regulations  27 conserve water. An evaluation of only these costs and benefits would lead to the conclusion that very few people would comply with laws requiring low-flow showerheads. These calculations, however, must also consider the costs and benefits of alternatives. People could stop taking showers and take baths instead, and that choice would have its own list of costs and benefits. Alternatively, people could try to purchase non-conforming showerheads and install them. In addition, if such showerheads were difficult to obtain, people could modify the conforming showerheads, build their own showerheads, or install multiple showerheads in the same shower. Each of these actions also has different costs and benefits. In the end, most Americans comply with showerhead regulations because the added benefit of having greater water flow in the shower is small, and though the costs of most of the alternatives are low, they are difficult enough to deter most people. If the only cost of one of these alternatives was to buy and install a high-flow showerhead, then many more people would likely pursue this alternative, but because the cost includes such actions as altering the flow regulator within a showerhead, most people are deterred enough by that action that they simply put up with the small inconvenience of low-flow showering. The contrast with government regulation of light bulbs is informative in how the differences in the costs of alternatives affects compliance. Government bans on energy inefficient incandescent light bulbs imposed a cost similar  to low-flow showerheads on the public. There was also no punishment cost to non-compliance (similar to the showerhead example) because enforcement was never extended to consumers. A crucial difference in the two (and an explanation why light bulb compliance is likely higher than showerhead compliance) is that the costs of the alternatives to light bulb compliance were dramatically higher. If someone wanted a high-flow shower, she need only remove the flow regulator in a standard showerhead and reinstall the showerhead, a one-time action, that might seem daunting to the less-handy, but hardly a difficult action. In contrast, non-conforming light bulbs would have to be bought from a supplier, likely outside of the United States. And once that purchase was made, the consumer would have to arrange to buy the contraband light bulbs again, in a few months. In other words, the costs of the non-compliant alternative were much higher than for the shower example because the costs are repeated. In addition, the compliant alternatives for light bulbs quickly became comparable (in terms of light output and quality) to what was replaced at only a somewhat higher price, a price that is at least comparable to the costs of importing non-compliant light bulbs. These two examples show how well government regulations can work when the costs and benefits are relatively low. However, the potential effectiveness of regulations becomes much more difficult to predict when the costs and benefits increase. The US government’s use of social security numbers to encourage compliance with other laws shows this lesson well. When the IRS, in 1987, started requiring a social security number for each person claimed as a tax deduction on a tax return, suddenly about 7 million dependents disappeared

28  Campaign regulations from the tax system (Levitt and Dubner, 2006, 21). The analysis of this ­phenomenon is ­simple. Tax filers could gain $1,900 for each fraudulent claimed child on their tax returns. Though the cost of non-compliance was high (a potential audit and tax penalty), the likelihood of actually paying this cost was probably low for the low-income people for whom a $1,900 tax benefit was significant enough to encourage violation of tax law. What changed in 1987? The use of social security numbers automated compliance checks for the IRS, making entanglement with IRS enforcement a much more likely outcome for people using bad social security numbers for dependents. There were (as there always are) alternatives for non-compliance. A taxpayer could seek out fake social security cards and numbers. Such actions would likely require working with organized criminal elements who knew which numbers could be used to fool the IRS systems. The cost of these alternatives just shot up dramatically, especially in contrast to the relatively small benefit of a tax reduction of $1,900. In contrast, there still is a healthy market for fake social security cards and numbers for undocumented workers in the United States. Why are those people willing to pay the much higher cost of social security card falsification that so many taxpayers were unwilling to pay once the IRS computerized up its identification procedures for fraudulent dependent claims? The answer lies in the extreme cost of compliance to undocumented workers. Compliance means giving up their life in the United States and returning to their country of citizenship. That is a much higher cost than losing a $1,900 tax benefit. Similarly, government officials that make assignments of public schools to children in order to create socioeconomic, racial, or ethnic balances, surely laudable goals, suffer from the same compliance problem. To many parents, the quality of the school that their children attend is a matter of great importance. Despite the high cost of alternatives (private school tuition), a good number of parents choose to pay this high cost if they have concerns about the quality of their assigned public school. Canada prohibits private medical care to ensure that all Canadians have the same quality of health care, care that is publicly provided. However, even banning the alternatives does not produce perfect compliance as shown when the premier of Newfoundland and Labrador traveled to Miami, Florida for a heart valve replacement surgery, a certainly ironic action for the nominal leader of his province’s health care system.1 The premier was willing to pay both the higher costs of US surgery and the potential political embarrassment of his choice in order to receive what he perceived as better care, joining an estimated 63,000 Canadians who received health care outside of Canada in 2016 (Ren and Labrie, 2017). These last three examples are the most appropriate for an analysis of campaign regulations. Strict campaign regulations can sometimes become crippling, so much so that any candidate who actually follows the regulations would likely lose the election. For a candidate, much like the undocumented worker, the parents considering their children’s education, or the Canadian seeking private medical care, the costs of compliance are potentially ­astronomical. There is no

Campaign regulations  29 calculation of costs and benefits that will justify knowingly losing an election. The only way around this calculation is to force compliance, with no exceptions, on all candidates, so that compliance will not result in a lost election. However, this is much easier said than done. Candidates will come up with creative alternatives, some legal, some illegal, and some very close to the line between the two, and the candidate that can get away with the most effective of the alternatives (and not be caught and punished for non-­compliance) will be the victor. Just as undocumented workers risk arrest and deportation daily as they live their lives in the United States, candidates who labor under unrealistically restraining campaign regulations develop alternatives that will allow them to win elections, even if those alternatives carry some risk of punishment. This analysis should not imply that campaign regulations can’t be effective. Regulations that don’t seem to prohibit activities seen as necessary for a successful campaign will likely be followed. Regulations that are easily monitored will also likely be followed. Similarly, regulations or bans that are coupled with efficacious alternatives will also likely be followed. Finally, regulations in less competitive systems are more likely to be effective. These four characteristics are a guidebook to the types of regulations, especially campaign regulations, that will or will not succeed. Campaign regulations will be more likely followed in party-oriented electoral systems rather than candidate-oriented electoral systems. It isn’t that a party-oriented system lacks electoral competition; however, the competition in such systems is between parties rather than between candidates. It is often more difficult for political parties to evade campaign regulations, making their compliance more likely. For example, party fundraising and spending are typically centralized and uniform. It is easier to monitor one set of fundraising and spending records than it is to monitor 300 or 600 sets of records, all with potential differences in their accounting methods. In addition, party campaign activities are typically widely advertised or accessible public events. Parties buy advertisements; they host debates or massive rallies. The very public nature of these events makes it easy for regulators or opposition parties to monitor behavior for compliance with campaign rules. In contrast, some of the campaign activities of individual candidates are private—a small meeting with potential donors, a meet and greet activity with voters, or a campaign swing by a workplace. Monitoring of campaign activities in such settings is harder to do and is less likely to occur. In the decentralized environment of candidate-oriented elections (such as occur in Japan and the United States) every campaign regulation has to be evaluated and complied with by each candidate running in the election. If one candidate finds a way to evade compliance or develops an alternative, that candidate has an advantage, and these actions will discourage compliance by other candidates. Japan and the United States should be two of the countries with the most likely occurrence of ineffective campaign regulations, and for this reason, Japan and the United States are the main points of comparison for this chapter.

30  Campaign regulations Nevertheless, campaign regulations, even in highly competitive, candidate-­ oriented countries, will be effective when the regulations themselves are inconsequential, at least in terms of deciding who wins the election. For example, the United States has required the reporting of minimal information for all campaign donations, and this requirement has nearly perfect levels of compliance. Japan’s laws only allow candidates to have two versions of a campaign flier or pamphlet. Candidates in both countries follow both of these rules, and they have not tried to develop new, non-compliant alternatives. Compliance occurs because the reporting of campaign donors or the limiting of the versions of campaign pamphlets to only two alternatives are the type of restrictions that will not materially affect the success of a campaign. As a result, candidates follow these rules, without complaint or evasion. A related factor is the existence of comparable alternatives. Direct corporate donations to campaigns could be effectively banned in the United States (as they were before the Citizens United decision) because there were alternative, legal means for campaign funds to flow from corporations to campaigns. Prior to the Citizens United decision of the US Supreme Court, corporations (and labor unions) could give money to candidates as long as the money was funneled through a political action committee of the corporation or the union. In addition, corporate executives could, as any individual can, make their own direct donations to a campaign as individuals, and they could make unlimited contributions to electoral efforts that were not coordinated with official campaigns. The existence of all of these alternatives made compliance more likely with the pre-Citizens United ban on direct corporate contributions to a campaign. Similarly, rules that regulate donations while still allowing some methods for candidates and parties to raise sufficient funds are also largely successful because other alternatives exist. Finally, easily monitored regulations are also usually successful. Ease of monitoring includes two aspects: is the regulated activity (1) public and quickly identifiable and (2) unique, with no analogs in the normal activities of people? Thus, Japan’s bans on candidate-specific advertisements have been successful as campaign regulations, in part because it is impossible to conceal an illegal candidate-specific advertisement and in part because there are no easy legal analogs to this activity. Non-politician citizens do not run media advertisements about themselves. The absence of such advertisements makes it difficult for candidates to develop alternatives to the banned behavior that have a similar effectiveness.

Japan’s campaign regulations Accepting that the candidate orientation of Japanese elections complicates election law compliance, I now examine the three remaining characteristics— non-necessity for victory, ease of monitoring, and existence of efficacious alternatives—to understand and evaluate Japan’s many campaign regulations. Most of Japan’s regulations are ineffective, primarily because the behaviors

Campaign regulations  31 are crucial to campaign success and are difficult to monitor. The prohibited campaign activity is also often easily transformed into seemingly innocent, everyday actions of people.2 Short campaigns The most consequential of all election regulations in Japan is the limit on the time of the campaign to only 12 days. The House of Councilors has a slightly longer election period of 17 days. Some municipalities have only five-day election periods. These limits have a long pedigree in Japanese politics, with the first limits put in place in 1934 (Yamamoto, 1998, 144). These campaign period regulations suffer from an enormous monitoring problem. Campaigning can occur simply by talking to another person. Thus, campaigning must be defined in a way that distinguishes it from other, normal conversations. Japanese law, therefore, defines campaigning as a communication that includes specific references to elections, candidates, voting, support, or endorsements. These rules can be followed quite easily while evading the intent of the rules by campaigning without ever saying the forbidden words. It is impossible to forbid prospective candidates from talking to people who might, in the future, become the constituents of the candidates. A prospective candidate can get to know prospective voters, become friends with those voters, discuss political issues generally, or even help out the voters with a particular problem long before the official campaign period has begun. In fact, the literature on Japanese politics has documented well the pre-­ campaign campaign that most successful politicians in Japan run (Curtis, 1971). Most Japanese politicians assume that the election is decided by the end of the pre-campaign period, and the formal activities of the campaign period are inconsequential. With the rise of party media campaigns after the 1994 electoral reforms, this conclusion is perhaps less warranted now, but as far as the actual election activities of the candidates go, it is still the case that the formal activities of the campaign period matter much less to the election outcome than the informal, pre-campaign campaign. By most measures, the limit on campaign periods is an ineffective campaign regulation because, even though there are few violations of the regulation, the regulation hasn’t actually shortened the length of campaigns. In fact, by forcing candidates to run surreptitious campaigns, the actual length of the campaign season has been extended (because candidates have to use less direct and inefficient methods to contact voters) so as to require that all politicians in competitive districts run a permanent campaign. Door-to-door campaigning A second campaign regulation is the longstanding ban on door-to-door campaigning. This regulation was the first consequential campaign regulation and was put into place in 1925 after first being proposed in 1909 (Soma, 1992, 46).

32  Campaign regulations The regulation remains in place despite a strong effort to remove the ban and an actual lifting of the ban for a two-year period in the 1950s. The motivations behind the ban were well-intentioned. Japan has long had a problem with politicians giving voters money for their votes, and it was known that the most likely location of such illegal exchanges was in the voters’ homes. However, from the earliest discussions of the ban, observers noted the likely irrelevance of the ban on such bribes because the politician was not the person who paid the money (Soma, 1992, 49–50, 56). Rather a local notable would visit supporters, delivering the small bribes. No ban on a candidate campaigning door to door would prohibit the city official who lived in the neighborhood from visiting some of her friends and neighbors. Candidates also worked around the ban by developing candidate support groups called koenkai (personal support organizations). Koenkai essentially became a vehicle by which a candidate could visit with constituents either in groups or individually as long as the candidate met the requirements of not actually saying any words or phrases that could be construed as campaigning. A candidate might hold a meeting in which he invited existing koenkai members to bring a friend. The candidate would make the meeting worth the while of the attendees by serving good food, inviting a popular figure, or holding the meeting at a local resort. The candidate would get to visit with voters at the meeting in the same way that a candidate might visit with voters in a doorto-door canvassing. Similarly, in the pre-campaign period, a potential candidate could call on koenkai members at their homes for short visits. These visits would not be construed as door-to-door campaigning because the potential candidate was not talking about elections and was going to visit the homes of friends (koenkai members). The extensiveness of koenkai networks and their use to effectively contact a large portion of the electorate are shown by surveys of koenkai memberships in which many voters belong to multiple koenkais. The door-to-door campaigning ban is another example of a failed regulation because of the difficulty of monitoring the behavior. Candidates have equivalent alternatives to the behavior which are even harder to monitor. Their use of these alternatives means that the ban is complied with superficially, but the actual behavior targeted by the ban continues unimpeded. The regulation has the perverse effect of pushing the undesirable behavior underground where it is even less easily monitored. In addition, this example shows how easily regulations can be turned into tools to punish political opponents. In the 1994 reforms, there was a groundswell of support to once again allow door-to-door campaigning. The election reform commission recommended this change, and the major political parties initially adopted removal of the ban as part of their reform proposals (Asahi Shimbun, May 30, 1991, 1; March 12, 1993, 4; March 25, 1993, 2; August 27, 1993, 1). However, some members of the ruling LDP mounted a rear-guard effort to stop the lifting of the ban. They succeeded first in changing their party’s reform proposal and subsequently got their position reflected in the compromise reform legislation which eventually passed (Asahi Shimbun, March 31, 1993, 1; November 4, 1993, 2). Though the politicians

Campaign regulations  33 who fought to keep the ban in place cited other reasons, the acknowledged ­justification for keeping the ban was to disadvantage parties such as Koumeitou that had more willing volunteers who would use the tool of door-to-door campaigning more effectively than the ruling LDP (Asahi Shimbun, September 23, 1993, 7; October 5, 1993, 1). Thus, a regulation that was born out of a sincere (but perhaps misguided) desire to clean up campaigning has become a tool to advantage one political party over another. Campaign posters Political posters are one of the most unusual aspects of Japanese political campaigns. It is illegal to post campaign posters during the campaign period (and the ban has been extended to begin when the election is called or 6 months before the expiration of the House of Representatives’ term of office). The one exception to this ban is the placement of large pieces of plywood by the government at various locations across election districts. The plywood is marked off into many poster-sized sections. Each candidate is allocated one section on each plywood board, and the candidate can place her poster only on that designated spot. These poster regulations developed over time in the wake of the ban on door-to-door campaigning. Prior to 1925, door-to-door campaigning was the primary method that candidates used to campaign. With the new ban on door-to-door campaigning in 1925, candidates were encouraged to use printed materials to contact voters, and posters, along with other printed materials, filled the gap caused by the ban on door-to-door campaigning (Soma, 1992, 48; Tamai, 2017, 4). Thus, the use of posters dramatically increased after 1925. For example, in the 1928 election there were nearly 35 million posters for an average of 35,000 posters per candidate (Tamai, 2017, 16). Immediately concerns arose about the regulation of this new form of campaigning because there were so many posters. In addition, government officials were treating the posters differently depending on the location and the candidate in question (Tamai, 2017, 102–3). New regulations were subsequently enacted for campaign posters in subsequent elections (Tamai, 2017, 21). It is somewhat ironic that the new and supposedly better form of campaigning that was to replace door-to-door campaigning was, as soon as it became popular, the next target for government regulations. In the postwar era the extensive regulation of posters produced, until the 1994 reforms, an odd phenomenon, which illustrates how ineffective certain types of campaign regulations can become. Campaign posters were only allowed during the campaign period and only on municipally provided plywood boards. However, like the ban on campaigning generally, it is impossible to ban all posters, just like it is impossible to ban all conversations between potential candidates and constituents. Thus, potential candidates would regularly advertise a speech that they would give at some point in the future. The posters advertising the speech would use most of their space on a picture of

34  Campaign regulations the person and the person’s name. In small print at the bottom was the ­location and time of the speech. Government rules limited such speech posters to only times and places reasonably relevant to the intended speech (Yamamoto, 1998, 148), but these limits, in practice mattered little. The one absurdity of the entire exercise was that these posters had to all come down when the formal campaign started, to be replaced by the limited number of posters placed on the plywood boards. In practice, neighborhoods would have many, many posters that to all intents and purposes looked like campaign posters (with only a small notation giving the time and place of a speech), and when the campaign formally began, those posters disappeared and only a few officially sanctioned posters remained. Responding to this phenomenon, the 1994 reforms explicitly banned posters advertising speeches, leading to a doubling in the number of citations for illegal posters in the subsequent 1996 elections (Yomiuri Shimbun, April 1, 1993, C; September 28, 1996, 2). These bans and regulations of posters have been largely ineffective because they seem important to the success of campaigns and there are no other allowed alternatives that create a lasting, visual presence for campaigns. Candidate advertising of all forms is banned. In addition, though the presence of an illegal campaign poster is easily monitored, the line between legal posters and illegal campaign posters is a difficult line to maintain and defend. There are certainly fewer campaign-type posters now that meeting posters have been banned, but they have been replaced by party rather than candidate posters or posters that identify a neighborhood branch office of a party or a candidate’s koenkai. Funeral and wedding gifts In most countries, government regulation of funeral and wedding gifts would not seem to be an appropriate area of government regulation of campaigns, but in Japan, two features have combined to bring these practices to the forefront of campaign regulations. First, Japanese culture highly values the exchange of gifts. Gifts are given in the middle of the summer, on attaining the age of majority, to a person departing on a trip, to friends of a person returning from a trip, at weddings, and at funerals. Second, as the usual means of campaigning have been banned or severely regulated, other ways to ingratiate oneself with voters have become more important. It is likely that there would be little incentive for Japanese politicians to give large sums of money to constituents using Japanese gift-giving practices if politicians could use the money that they raised to effectively run their campaigns in other venues. In other words, if Japanese politicians could spend hundreds of thousands of dollars on media campaigns or direct voter mobilization efforts, then they would have no need to use their campaign funds on funeral gifts to the bereaved. However, Japanese politicians have no legitimate outlets to use the money that they raise for campaigns, so they use that money in other methods to make themselves known to voters and to create a sense of obligation among voters to the potential candidate.

Campaign regulations  35 The extent of this surreptitious form of campaigning is revealed in ­ iscussions about the new regulations of gift giving passed in 1989. A group of d LDP legislators explained that in an average year they spent about $1.3 ­million maintaining their personal support organizations (largely pre-­ campaign activities). Of this total, they spent from $100,000 to $200,000 each year on wedding and funeral gifts.3 These politicians pushed for new regulations that would make it illegal for a politician to give money unless the politician actually attended the funeral or wedding. These new regulations seem to have been a cease-fire in the intense competition for votes among conservative candidates in which they would all send gifts to the bereaved, and, of course, no potential candidate wanted to be perceived as the cheap candidate who didn’t give a gift or gave a smaller amount than her competitors. Thus, the new rules served to block the spiraling cost of wedding and funeral gifts. It is interesting that these new rules were created by the politicians themselves to self-regulate their behavior and prohibit cheating against the rules by other candidates. The new rules are, in and of themselves, humorous in their effort to stop the most egregious forms of payments to voters while retaining obvious allowances for normal social interactions between politicians and voters. For example, it is common in Japan to send New Year’s and summer greeting cards to friends and acquaintances. Japan’s rules allow politicians and candidates to send these cards as long as they are handwritten. Candidates and potential candidates can also still make wedding and funeral gifts if they actually attend the wedding or funeral. The differentiation between what is allowed and what is not is a huge area of concern for candidates and especially politicians, so much so that government agencies produce extensive guidelines to help candidates and politicians know what is allowed and what is not. For example, politicians can make gifts at funerals if they attend the funeral and the gift is made on the day of the funeral. A funeral gift made the day after the funeral is illegal, even if the politician attended the funeral on the previous day. Similarly, a politician was not allowed to donate a prize to be given to the winning team in a local baseball competition. Japanese people often write messages on commemorative placards (shikishi), and politicians are allowed to write such messages only if the voter provides the placard. Politicians may not write m ­ essages on placards that they provide themselves (Yomiuri ­Shimbun, ­December 21, 1989, 2). In a counter-intuitive way the new regulations on funeral gifts have been effective. Their goal was to reduce the amount of money spent on funeral gifts, and they have succeeded in that goal. Of course, if the rules are judged by whether or not they eliminated the efforts by potential candidates to ingratiate themselves with voters through all the methods of pre-campaign campaigning that are done, then no, the rules have failed. But of course, that was not the goal of the new regulations. They were simply an attempt to stop one form of pre-campaign campaigning that some politicians had judged to have become too expensive (and likely less than cost effective because funeral and wedding gifts in Japan are expensive).4 These new rules have succeeded, in part,

36  Campaign regulations because they are easily monitored and there are other, less costly, methods for ­politicians to ingratiate themselves with voters. Campaign pamphlets and postcards The culmination of the early efforts to rein in the initially encouraged use of printed materials for campaigns has come into fruition as extremely detailed regulation of the distribution of any printed materials to constituents, ­limiting their distribution to 70,000 pamphlets of only two types and 35,000 ­government-subsidized postcards. These regulations, like the ban on door-to-door campaigning have been effective at a superficial level. Candidates don’t send out postcards or distribute campaign fliers in addition to what they are allowed to do by law. Compliance is easily monitored. In addition, there is very little incentive to evade these regulations because of the presence of equivalent, legal alternatives. Candidates and potential candidates use other forms of print to enhance their standing and notoriety among constituents. For example, koenkai members will receive newsletters several times a year full of information about issues and the potential candidate’s activities and background. It would be difficult, far in advance of an election to prohibit private citizens who belong to a political organization from distributing a newsletter to the members of the organization, and these newsletters, though carefully written, become the functional equivalent of printed campaign materials, though they make no mention of campaigns and are distributed well in advance of the formal campaign period. Fundraising and campaign spending It would not be surprising that Japan also has extremely strict regulations on the raising of campaign funds and their spending. Politicians find loopholes, and new reforms efforts plug those loopholes. In the 1980s, it was common to sell tickets to koenkai events as a form of fundraising, and contributors would buy up large blocks of tickets to the events in order to surreptitiously contribute to a campaign. As the various fundraising organizations came under stricter government regulation, savvy politicians managed to set up multiple organizations to evade contribution limits. New reforms plugged these loopholes, and an analysis of reported fundraising and spending indicates that candidates are spending less and have fewer affiliated organizations than in the period before the reforms (Carlson, 2007). In addition, perceptions of corruption in Japan have also declined. Both pieces of evidence point to a moderately effective set of campaign reforms regarding money, reforms that didn’t eliminate but channeled and reduced the influence of money on Japanese politics. The moderate success of these reforms stems first from their tolerance of large amounts of spending. Japanese rules didn’t unrealistically limit fundraising or spending to such small amounts as to make them meaningless. Candidates and parties can still effectively compete in both fundraising and spending

Campaign regulations  37 within the new limits. Reforms that still allow a sizeable amount of competitive behavior are more likely to be adhered to because a cost of compliance doesn’t include a likely election loss. Despite limits on certain types of contributions, Japan’s election spending rules allow largely unlimited spending and also have no limits on fundraising in the aggregate (just on specific types of donations). A second reason for the success of these reforms is the opening up of additional alternatives to channel the competitive aspect of election campaigns. As greater scrutiny was given to some forms of fundraising and fundraising organizations, the government also established government subsidies of political parties, allowed greater ease of fundraising by political parties, and opened the door for political parties to run party-based campaign advertisements. These new rules changed the focus of fundraising and spending from primarily an individual candidate activity to one that is shared by the candidates and their parties. Thus, it is easier to gain compliance for individual campaign spending rules if other alternatives for fundraising and spending are created or augmented, even if some of these other paths are party-centered. Of course, this conclusion that these reforms have been moderately successful comes with some caveats. All of our data on campaign spending in Japan comes from publicly reported spending documents. It is possible that much campaign spending occurs out of the sight of government regulators and never shows up on reports submitted to the government. For example, when ­Kanemaru Shin was arrested (under the looser rules of the past), he had $85 million in gold, bearer bond certificates, and cash at his home. Even under the less strict reporting requirements of the past, Kanemaru had this much money in untraceable assets, presumably to prevent creating a paper trail that could be used in a prosecution. If that much money was on hand in an LDP leader’s home under the less strict rules, it might be the case that similar amounts of funding are changing hands surreptitiously under the stricter, current rules that likely increase the incentives to hide contributions that are illegal. It is entirely possible that politicians scrupulously keep their books and file their reports while also taking and spending some amounts of money off the books. For example, if payments are still being made to voters, it is unlikely that politicians report the money given to intermediaries to make the payments. Soundcar and loudspeaker regulations In what must seem absurd to an outside observer, Japan’s campaign regulations also limit, along with the number of offices and office signs, the number of soundcars and loudspeakers that a candidate can use. These limits come from the peculiar form of campaigning that occurs during the official campaign period. Candidates will stand in public places (for example, the entrance to a train station) and talk to the voters as they stream in and out of the station. Each candidate may only have one such voice amplification device. Similarly, politicians typically drive around their districts in a car, festooned with posters and a loudspeaker that repeats over and over the name of the politician

38  Campaign regulations and a call for support. In fact, campaign regulations ban the repeated saying of a candidate’s name, but that ban has an explicit exception for the candidate’s campaign vehicle. Each candidate is only allowed one such car, and both activities are also limited to the hours of 8 am to 8 pm. These regulations are largely effective, in part because they are inconsequential. Why would a candidate need two soundcars to drive around the district (and annoy the voters)? Why would a candidate need two loudspeakers if he can only be in one place at a time. Indeed, it is possible that these two very public activities of the actual campaign period are of very limited effectiveness in winning votes. Candidates do these activities during the official campaign period in order to be seen as trying hard. If they don’t make these very public appearances during the official campaign period, some voters would conclude that they weren’t working hard for election. Telephone calls Though telephone calls were initially included in the ban on door-to-door campaigning enacted in 1925 (Soma, 1992, 66) post-World War II campaign laws removed the ban on campaigning by telephone. Thus, it is an oddity that, in the midst of all of the extensive campaign regulations and banned activities in Japan, telephone calls are largely unregulated. Indeed, perhaps the most consequential activity of the official campaign period is having supporters call voters, especially voters that they personally know, and ask for their support in the election. The only limit on this form of campaigning is that it can’t occur prior to the official election period (though of course people can have discussions about politics, just not formal mentions of words like candidate, elections, support, or endorsement). Internet campaigning With the rise of the importance of the internet, websites, and social media in the gathering of political information, it was only a matter of time before Japan’s campaign regulators would attempt to regulate this area as well. In fact, the new regulations for internet campaigning were billed as a relaxation of regulations (Soumusho, 2017) because until the new regulations took effect, most internet publications were treated as banned activities because they were unauthorized written communications of campaign messages. Of course, this didn’t prevent political material from being on the internet. Rather, internet postings just had to avoid endorsing candidates or specifically talking about an election. The new rules allow campaign messages to be freely posted on the internet. Of course, these activities can only occur during the formal election period, and those not allowed to campaign (non-citizens, youth) are barred from these as well as other campaign activities. Curious, though, is the strict ban on using email as a campaign tool (Soumusho, 2017). This ban extends not just

Campaign regulations  39 to candidates but to everyone in Japan. Thus, it is legal for a voter in Japan to post on a website ‘I am voting for Katsumi Tanaka. You should vote for her too.’ as long as it is during the official campaign period. But if the voter were to send the same message to a friend in an email message, the action would be illegal.5 Similarly, if the voter sent an email saying ‘check out my endorsement at this website,’ that message would also be illegal. The logic behind this ban is elusive. In some ways an email message is very much like a telephone call, and that is the one area of campaigning that the regulators have not touched. An email message is also like a letter or postcard and those forms of written campaigning have long been heavily regulated and banned. Nevertheless, it is hard to square a ban on email communications with the allowance of the exact same message posted on a website. This ban is nearly impossible to enforce, at least for individual emails. It requires that everyone in Japan be versed in the nuances of which words and phrases are campaign language (which are banned) and which are mere political discussion (which is allowed). It is likely that this ban is meant only as a prohibition on mass emails that would be used by campaigns in the absence of such a ban. If this is the real purpose of the ban, then it is likely effective. Pre-printed ballots Though not technically a regulation or ban of a campaign activity, Japan’s outdated adherence to handwritten ballots shares some important characteristics with the preceding analysis of campaign regulations. In Japan’s elections to the House of Representatives, voters receive a blank piece of paper with a large box outlined in the middle of the paper. The voter must write the name of his preferred party or candidate within that box. The names of candidates and parties are posted in the voting booth, but because the Japanese language uses complex ideographs for writing, it is easy to make a mistake when writing the name of a candidate or party. Thus, approximately 5 percent of ballots are spoiled because of mistakes in writing names, writing that goes outside of the outlined box on the paper, or other deficiencies in the ballot. For example, in the 1995 House of Councilors election there were 3.88 million invalid ballots, including 5.6 percent of the ballots cast in the national proportional representation race and 3.5 percent of the ballots cast in the district races (Asahi Shimbun, October 31, 1995, 4). Despite allowing pre-printed ballots (where a voter simply checks the box for the preferred candidate or party) for local elections in Japan, national elections are still conducted using handwritten ballots because members of the ruling LDP see such ballots as a barrier to casting votes for non-incumbents or new political parties (Asahi Shimbun, October 20, 1995, 2; October 21, 1995, 5; December 17, 1997, 5). In earlier periods it was also noted that illiterate voters were disproportionately supporters of some of the opposition parties, making the rule advantageous to the ruling LDP during that period. In fact, the 1994 reforms made the transition to pre-printed ballots at the national level, but after the victory of the newly created New Frontier

40  Campaign regulations Party in the 1995 Upper House elections, LDP politicians quickly rescinded the reform (before it had ever been used in a House of Representatives election), and went back to the handwritten ballots. The manipulation of this ballot rule is strikingly similar to the discussion surrounding door-to-door campaigning. A rule or practice that starts out with noble goals such as reducing corruption or encouraging the discussion of issues, quickly becomes perceived as advantaging or disadvantaging a particular political party. The regulation then takes on a life of its own as a tool to punish political opponents.

Application of the Japanese lesson to the United States The political lesson of Japanese campaign regulations is a negative lesson— what the United States should not do. This lesson, though, is important. Though blocked by a Supreme Court that insists that First Amendment rights exist and are protected in campaign donations and spending, many Americans would like to have a highly regulated campaign environment (Teachout, 2014). They yearn for an environment in which money doesn’t matter, elites have no more influence than the average voter, and candidates eruditely argue the merits of competing policy positions in front of attentive and well-informed ­voters. The Japanese example shows that the path to this political nirvana does not lie through the garden of strict campaign regulations.6 The Japanese example does help explain some noteworthy developments and the rather efficient functioning of the few campaign regulations that exist in the United States. For example, voluntary limits on presidential campaign expenditures worked (were followed) for 30 years because the limits were set high enough and were compensated by generous subsidies to create an effective alternative for competitive efforts. Only as the limits became more constraining were they abandoned by candidates, first by Barack Obama in 2008 and then by both major-party candidates in 2012. Similarly, the ban on some forms of corporate donations was efficacious prior to the Supreme Court’s decision in Citizens United in part because there were easy alternatives (political action committees and individual donations) to the ban. Despite the vitriolic criticism of the Citizens United decision, its loosening of the rules for corporate donations has not dramatically skewed campaign finance (Hansen et al., 2015; Mutch, 2016, 131). Corporations have always been able to make political contributions though political action committees and have done so (Corrado, 2000, 43; Gilpatrick, 2011, 411–12). Ignored in the furor about the decision are the dramatic increases in other sources of campaign finance including individuals, bundlers, non-corporate organizations, and the ultra-rich. Similarly, it has long been the case, for example, that labor union political contributions are similar to or even greater than corporate contributions (Hansen et al., 2015, 533), but only corporate contributions are discussed in the post-Citizens United clamor for additional regulation. Though the decision itself and the ensuing discussion and clamor

Campaign regulations  41 for repeal have been intensely politicized, the ban that preceded the decision was an effective regulation because sufficient alternatives existed to channel competition, increasing the likely compliance with the ban. The politicization of the decision and the ensuing discussion is also another example of how seemingly well-meaning campaign regulations quickly become tools to punish political opponents. Japan’s examples also throw cold water on the dream of taking money out of US elections. Many observers yearn for a new Supreme Court that will interpret the First Amendment as not applying to campaign contributions and campaign spending (Cain, 2015). Under such a scenario, governments would be free to limit both contributions and spending, perhaps subsidizing all campaigns so that it would become unimportant how much money a campaign had. Rather, all competitive efforts would be channeled to debating the issues in government-sanctioned public presentations. Though such hypothetical campaigns have enormous appeal, the Japanese example suggests that the reality would look very different from the purported idealistic scenario. First, money would still be spent in elections; it would just be spent in a way that couldn’t be tracked as actual campaigning. Organizations would be set up to encourage turnout or educate voters about issues. The campaign between candidates would become one step removed as the campaign shifted to supposedly non-partisan mobilization efforts (that just so happened to target groups most likely to vote a certain way) or voter education and discussion of issues, which again could easily develop partisan impacts. Just as the Japanese pre-campaign became more important than the formal campaign, these non-campaign efforts would increase dramatically in importance and would become the new receptacle for political donations. Second, the rules would become easily susceptible to partisan manipulation. Incumbents would love rules that strictly regulated fundraising and spending because those same rules would cripple any effort to replace the incumbents (Samples, 2014). Rules that tightly circumscribe campaign efforts give a natural advantage to the already well-known incumbent, who has an enormous advantage if it becomes extremely difficult to broadcast a new message to the voters. In addition, the various aspects of the rules would be manipulated in ways that created partisan benefits. Increasing the number and severity of regulations only increases the ways in which the campaign regulations can be manipulated for partisan benefits. Rather, the Japanese example shows that campaign regulations in fiercely competitive electoral systems such as Japan or the United States are hard to design so that they will be effective. They work only when the regulations are inconsequential to the eventual outcome of the race, or they are coupled with significant alternatives for competition. In addition, campaign regulations are most likely to be successful when they regulate a behavior that is not shared by the people, a behavior that is unique to politicians and is easily monitored. This isn’t to say that most forms of campaign behavior can’t be strictly regulated or even banned. They can, and they are in Japan. The problem arises that

42  Campaign regulations though these bans are enforced, they give rise to much less salutary alternative forms of campaigning, forms that are more hidden and less efficient. Thus, the behavior hasn’t been effectively banned but rather has been driven underground. In addition, these bans and strict regulations persist, even when their usefulness no longer is apparent simply because they start to serve partisan interests.

Notes  1 Globe and Mail, Toronto, February 5, 2010, ‘Williams’s heart surgery choice was based on ignorance.’   2 Curtis (1971) exhaustively describes these campaign regulations.  3 Yomiuri Shimbun, January 3, 1989, 2 ‘Dou torikumu 89 seiji kadai (1) seiji kaikaku.’   4 The most common amounts for such gifts are the equivalents of $100, $300, and $500. $100 gifts are seen as being stingy, leaving the much more expensive options of $300 and $500 gifts. Gifts for $200 and $400 are inappropriate because the numbers 2 and 4 in Japanese share the same sounds as the words for suffering and death.   5 Though it may seem odd to ban any type of campaigning between voters as illegal, Japan, in fact, had a ban in place on all campaign activities by anyone outside of the formal campaigns, with only two exceptions for speeches and endorsements (Yamamoto 1998, 153). This ban was lifted in the post-World War II election reforms.   6 Though Cain (2015) supports these goals, his analysis also recognizes the unintended consequences and potential politicization of many reform attempts.

References Cain, Bruce E. 2015. Democracy More or Less: America’s Political Reform Quandary. New York: Cambridge University Press. Carlson, Matthew. 2007. Money Politics in Japan: New Rules, Old Practices. Boulder, CO: Lynne Rienner. Corrado, Anthony. 2000. Campaign Finance Reform. New York: Century Foundation Press. Curtis, Gerald L. 1971. Election Campaigning Japanese Style. New York: Columbia University Press. Gilpatrick, Breanne. 2011. “Removing Corporate Campaign Finance Restrictions in Citizens United v. Federal Election Commission, 130 S. Ct. 876 (2010).” Harvard Journal of Law and Public Policy 34, no. 1 (Winter): 405–20. Hansen, Wendy L., Michael S. Rocca, and Brittany Leigh Ortiz. 2015. “The Effects of Citizens United on Corporate Spending in the 2012 Presidential Election.” Journal of Politics 77, no. 2 (April): 535–45. Levitt, Steven D., and Stephen J. Dubner. 2006. Freakonomics: A Rogue Economist Explores the Hidden Side of Everything, revised edition. New York: HarperCollins. Mutch, Robert E. 2016. Campaign Finance: What Everyone Needs to Know. New York: Oxford University Press. Ren, Feixue, and Yanick Labrie. 2017. Fraser Institute, Research, Health Care, Archive, “Leaving Canada for Medical Care, 2017.” June 29. https://fraserinstitute.org/ studies/leaving-canada-for-medical-care-2017. Accessed May 10, 2019.

Campaign regulations  43 Samples, John. 2014. The Fallacy of Campaign Finance Reform. Chicago, IL: University of Chicago Press. Soma, Masao. 1992. Nihon senkyo seido shi, futsu senkyohou kara koushoku senkyohou made. [A History of Japan’s Election System from the Regular Elections Law to the Public Officers Election Law]. 2nd edition. Fukuoka, Japan: Kyushu daigaku shuppankai. Soumusho. 2017. [Ministry of Internal Affairs and Communications]. Senkyo [Elections]. Senko seido kaikaku no torikumi. Inta-netto senkyo undou no kaikin ni tsuite. Chirashi. [Grappling with Election System Reform, about Lifting the Ban on Internet Campaign Activities, flier]. www.soumu.go.jp/main_content/000225177. pdf. Accessed May 26, 2017. Tamai, Kiyoshi. 2017. Daiikkai fusen to senkyo posutaa: Shouwa shotou no senkyo undo ni kansuru kenkyuu [The First Regular Election and Election Posters: An Examination of Campaign Activities at the Beginning of the Shouwa Period]. Tokyo: Keio gijuku daigaku shuppankai. Teachout, Zephyr. 2014. Corruption in America: From Benjamin Franklin’s Snuff Box to Citizens United. Cambridge, MA: Harvard University Press. Yamamoto, Shinichirou. 1998. Jitsumu to kenshu to tame no wakariyasui koushoku senkyohou. [An Easy to Understand Explanation of the Public Officers Election Law, for Training and Practical Application]. 10th edition. Tokyo: Gyosei.

4 Drawing the boundaries of election districts

Lesson 4: Election district boundaries should follow local government boundaries. This method, used in Japan, prevents the creation of politically biased districts while only slightly increasing population inequality among districts.

Introduction It is not surprising that some of the worst examples of political behavior occur when politicians write the laws that govern their own actions. Problems can arise when politicians determine their own pay raises, pensions, or benefits. Politicians are poor investigators when looking at their own scandals or ethical lapses. Politicians also have strong incentives to bias elections in their favor when they write election rules, decide election district boundaries, or use the tools of government to create a favorable electoral environment. The solutions to this age-old problem of who watches the guardians generally take one of three forms. First is to delegate these decisions to non-­politicians. For example, monetary policy in many advanced industrial democracies is delegated to a central bank that is independent of politicians. This is done, in part, to prevent politicians from using a country’s monetary policy to create short-term booms that coincide with elections, actions that harm long-term fiscal stability (Bernhard, 2002, 2). In the electoral context, the tendency has been to entrust the task of drawing election district boundaries to neutral and non-partisan commissions (Karch et al., 2007). Cain (2015, 183) argues that courts rather than partisan elected state officials or political parties should oversee all election procedures. Similarly, civil service reforms in the United States were enacted to remove the possibility of patronage in the awarding of government jobs, one of the important tools politicians used to augment their political standing (Grindle, 2012). A second strategy has been to use the inherent competition among political parties to have one party keep in check the excesses of the other party. Thus, members from the opposition party might demand and hold investigations of the ruling party (Kriner and Schwartz, 2008; Parker David and Dull, 2009). This system tends to break down when one party controls all of the levers of government, but even in that situation, the press and voters can often

Drawing the boundaries of election districts  45 function as a check on the excesses of the ruling party (Canes-Wrone et al., 2002; ­Carson et al., 2010). The third method is the tool of democratic accountability to counterbalance the excesses of a ruling party. For example, in the United States politicians who vote to raise their own salaries run a high risk of electoral rebuke at the hands of their constituents (Theriault, 2004). This tool, however, loses its effectiveness when self-serving political actions are complex or lost in the details of other more significant issues. Thus, it is hard to marshal voter anger, for example, against election district boundaries that give a partisan political advantage to one political party (Fougere et al., 2010). Politicians can thus evade democratic accountability and ‘choose their voters,’ as the well-known quip puts it, rather than the voters choosing their representatives. One of these problems, the drawing of election district boundaries, is more severe in Japan and the United States than it is in most other advanced industrial democracies. Both Japan and the United States have strong candidate rather than party orientations in their elections. Voters in both countries, of course, are influenced by political parties—their leaders and their platforms. However, because voters cast ballots for specific candidates rather than political parties, voters are primed to consider attributes of specific candidates in addition to party information.1 Quite a few countries have single-seat districts in which voters cast ballots for candidates rather than parties. There are, in addition, several other factors which encourage voters to give greater weight to candidate attributes rather than party membership (Carey and Shugart, 1995). Japan and the United states are at the far end of a continuum that ranges from candidate-centered elections to party-centered elections. As such, individual politicians in both countries pay close attention to the boundaries of their election districts. In contrast, in Canada, Australia, or the United Kingdom, parties rather than candidates show the greater interest in the effects of election district boundaries. Thus, individual politicians in Japan and the United States strongly oppose any loss of their control over the boundary-drawing process. Japan’s innovations to help solve this problem, then, are especially useful for the United States, the country that is most similar to Japan among the advanced industrial democracies in terms of the candidate-centered orientation of its elections. In contrast to Japan’s recent innovations, the United States has historically been a world leader in solving the problems that surround the drawing of election district boundaries. The long history in the United States of election districts with unequal populations (malapportioned) ended with Supreme Court decisions beginning in 1962 that required that all election districts be of equal population. Since then, US districts have stopped being malapportioned, but now district boundaries are drawn to benefit a particular political party ­(gerrymandered). US election districts might have exactly the same number of people, but the boundaries of the districts are contorted, either to create districts in which a majority of the voters are members of a specific minority group or the districts are contorted to give one political party an advantage.

46  Drawing the boundaries of election districts Thus, in the 2018 election, Republicans won all four congressional races in Arkansas with only 62.6 percent of the votes in the state. Similarly, Democrats won all five of Connecticut’s races with only 61.6 percent of the statewide vote. Even in California where the boundary lines were drawn by a non-partisan commission, in the 44 congressional districts that had both Republican and Democratic candidates, Democrats won 86.4 percent of the seats with only 73.3 percent of the vote. In the 2014 elections the California disparity was even greater: Democrats won 72.1 percent of the seats both parties contested with only 55.9 percent of the vote. Japan is the opposite of the United States. It doesn’t have a problem with gerrymandering. Rather, Japan’s election districts are malapportioned. Though the 2017 reapportionment was put into place in time to be used for the 2017 elections, the previous reapportionment in 2012 came almost two years after a Japanese Supreme Court decision that required action be taken. The legislators also delayed the effect of the reapportionment until after the 2012 election. This and similar controversies in Japan make it seem as though Japan is unique among the advanced, industrial democracies for not having yet solved its malapportionment problem. This characterization, though, ignores the substantial reduction in malapportionment that has occurred in Japan. At the peak of malapportionment, Japan’s 1972 election, the Osaka 3rd district had 397,696 voters per representative in contrast to the rural Hyogo 5th district which had only 79,403 voters per representative. The voters of Hyogo 5 enjoyed five times the representation as the voters of Osaka 3. The insistent prodding by the Japanese Supreme Court and a new electoral system in 1994 that required the drawing of new election boundaries both helped reduce malapportionment levels dramatically. In the 2017 election, the worst disparity was only 1:1.98, meaning that the 473,879 voters of the Tokyo 13th district only had 1.98 times the representation as the 238,961 voters of the Tottori 1st district. Nevertheless, it might seem odd to suggest that the United States can learn anything from Japan’s less than stellar record of redistricting. After all, the United States, despite its other political shortcomings, is the nation that spearheaded the malapportionment revolution and still leads the rest of the world in its commitment to the ‘one person one vote’ principle. The United States, however, has erred in being rigidly committed to vote equality, so much so that other, equally important concerns, have been completely ignored in the drawing of election district boundaries. The US commitment to absolute vote equality has opened the door to rampant gerrymandering, creating a worse problem than small deviations in the population of election districts. In contrast, despite an extensive literature that criticizes the disparate sizes of Japanese districts (Hata, 1990; Hickman and Kim, 1992; McElwain, 2013; Kobayashi, 2012; Wada, 2012), the Japanese have actually succeeded in creating a system that balances vote equality with district cohesiveness.2 In fact, the interplay between these two factors creates boundary-drawing rules that are a model for other countries. Japan’s rules reduce the discretion of line drawers

Drawing the boundaries of election districts  47 to a minimal level by using local government boundaries to determine the ­composition of election districts. With discretion so reduced, gerrymandering is hardly a possibility. In addition, malapportionment is constrained by requiring that districts created be within maximum ranges of tolerated population disparities. The Japanese system sensibly allows for small amounts of population inequality while preventing gerrymandering by removing much of the discretion in the boundary-drawing process. In addition, the Japanese system is better suited for countries that have candidate-centered elections such as Japan or the United States (Nemoto and Tsai, 2016, 165–6). When voters are influenced more by candidates than by political parties, the incentives to manipulate district boundaries increase significantly. Solutions such as non-partisan commissions, which work well in countries with weaker incentives to manipulate, don’t work as well in countries such as Japan or the United States for two reasons. First, non-partisan solutions are less likely to be adopted. Second, even when they are adopted, they are more likely to be coopted by partisan interests. The heavy reliance of the Japanese on local government boundaries may be the best solution for the problems of political redistricting, especially in countries with candidate-centered elections.

Perfect equality among election districts is an impossible goal The drawing of election district boundaries in the United States, whether done by politicians, non-partisan commissions, or the courts, is driven largely by an illusory commitment to the absolute equality of election districts, a commitment that in practice is impossible to attain. For example, California used a newly created commission to draw its 53 districts after the 2010 census so that 20 districts had the exact target population of 702,905 people. The remaining 33 districts had either one more person (12 districts with 702,906 people) or one less person (21 districts with 702,904 people). Similarly, Utah’s politicized redistricting process also created four new districts, all within one person of the target district population in Utah of 690,971 people. The absurdity of these artificially precise efforts is perhaps apparent. Why should it matter whether districts in California have populations that differ by 1, 10, 100, or even 1,000 people if California’s districts already each have 12,000 more people than Utah’s districts? This absurd outcome, however, is required by the judicial overseers of the redistricting process. The US Supreme Court set the standard by rejecting plans with average variations from the population ideal within a state of only 1.6 percent (Kirkpatrick v. Preisler (1969), 394 US 526, 530–31) and 0.14 percent (Karcher v. Daggett (1983) 462 US 725, 739–43). Pennsylvania’s state legislature proposed a plan for the state’s congressional districts in 2002 that had a difference in population between the state’s smallest and largest districts of only 19 people. Citing this 19-person difference, a federal court rejected the plan for having unequal districts (Vieth v. Pennsylvania, 195 F. Supp. 2d 672 (M.D. Pa. 2002)). Though this decision was actually a use of the principle of equality of district size as a way to strike down a gerrymandered district plan,

48  Drawing the boundaries of election districts the principle remains that even the tiniest of population ­discrepancies can be used to overturn a districting plan in the United States. The federal court system’s fixation on absolute population equality is meaningless because other features of the US election system make it impossible to create districts that are absolutely equal in size. These other features include: 1 2 3 4

state boundaries that make it impossible for districts in different states to be of equal size, the allocation of seats to states based on population numbers rather than eligible voter numbers, a constantly changing population that makes the supposedly equal districts already out of date when the first election occurs, and differences in voter turnout that give voters in one district more representation than voters in another district, even if the districts are of equal population.

For example, if all 435 congressional districts in the United States were perfectly equal in size (according to the 2010 census), each district would have 710,766.5 people.3 California’s average-sized district, however, has only 702,905 people. Thus, California’s perfectly equal districts are already about 8,000 people deviant from the national average. The closest state above this national average is Wisconsin with 712,278 voters per district and the closest state below the average is Tennessee with 708,381. Even the two states closest to the national average still have a 4,000 vote disparity in district sizes, even with districts that are of perfectly equal size within each state. This 4,000 vote disparity (0.5 percent) is the smallest possible difference between the 17 ­election districts of these two states. Of course, Wisconsin and Tennessee represent the best-case scenario for equal apportionment in the United States. The worst case is a comparison of Rhode Island, the smallest state that qualifies for two seats, with Montana, the largest state that has only one seat. The average for Rhode Island is 527,623 voters per district, and Montana has 994,416 people in its state-wide district. Thus, even if election district boundaries were perfectly apportioned within every state, Montana would still have close to twice as many voters for each congressional representative (88.5 percent more) than Rhode Island would have. These calculations are not some peculiar function of US state boundaries. Japan is similarly divided into 47 provinces that are labeled prefectures in Japan.4 Japan’s 2015 census dictates that the average election district should have 430,882 people. The two prefectures closest to this average are Nagano and Kyoto. In contrast, the two prefectures with the greatest deviations are Tokyo and Tottori. A perfectly equal apportionment in Tottori would result in 286,824 voters per representative, but in Tokyo, there are 540,549 voters per representative. Thus, the smallest possible disparity between states in the United States is

Drawing the boundaries of election districts  49 0.5 percent (Wisconsin and Tennessee), and in Japan it is 3.5 percent (Nagano and Kyoto). In contrast, the largest disparity is 88 percent (1:1.88) in both the United States and in Japan. The goal of absolute population equality in the United States is also vexed by the fact that the US Constitution requires apportionment on the basis of population, but not every person in the population can actually vote in elections. Children, obviously, do not vote. In addition, noncitizens do not vote. Some states restrict the voting rights of felons or those who are mentally incompetent. Differences in the populations of these groups between states and districts make it impossible for districts to have both equal numbers of residents and equal numbers of eligible voters. For example, Utah in 2010 has 31.2 percent of its population under the age of 18. In contrast, Vermont has only 20.1 percent of its population under 18. If Utahns had as few children as Vermonters, the total population of Utah would decline by 300,000 people, the population of nearly half of a congressional seat. These differences in the eligible voting population, combined with differences in turnout among eligible voters can produce stark differences in the number of votes won by congressional candidates elected from districts that are supposedly of equal size (Campbell, 1996). For example, in 2018, T. J. Cox won the 21st district of California with only 57,239 votes while Pramila Jayapal won Washington’s 7th district with 329,800 votes. Part of this disparity is due to differences in the closeness of the races: Cox only won 50.4 percent of the vote in contrast to Jayapal’s 83.6 percent of the vote. However, even counting all of the votes cast in each district, California 21 had only 113,616 votes in contrast to 394,681 votes in Washington 7. Both districts had about the same population in 2018 with 2017 census estimates of 723,549 for ­California 21 and 768,408 for Washington 7. A contributing factor to this difference is the number of noncitizen residents in each district. California 21 had 156,278 noncitizens (21.6 percent); Washington 7 had only 70,633 noncitizens (9.2 percent).5 Citizenship differences, along with different levels of political engagement meant that there were only 232,379 registered voters in California 21 in 2019, meaning that if every registered voter in California 21 had actually voted, this perfect, hypothetical turnout would still have been only 58.9 percent of the actual turnout in Washington 7. A third deficiency in the absolute equality requirement is that district populations constantly change, making absolute equality an impossible goal to attain in an actual election. From the 2010 census until July 2018, the US ­Census Bureau estimates that California added 2.3 million people to its population. Divided among California’s 53 election districts, this means these districts, on average, have 43,443 more people than the numbers that were used to draw district lines. Of course, if the population of all congressional districts grew at the same rate, this rate of population growth would not matter. However, as some districts grow in population, others shrink or grow slower, producing large disparities among district populations. In the eight years from 2010 to  2018, Texas added 3.5 million people, which even using a higher national average

50  Drawing the boundaries of election districts district size still entitles Texas to an additional 2.2 members of Congress. Illinois, in contrast, lost about 90,000 people. Illinois now enjoys one more congressional seat than it is entitled to with its current population. Despite these large population shifts, no changes will be made in electoral votes, congressional seat allocations, or district boundaries until the 2022 election. Creating absolutely equal election districts is also a misguided enterprise because turnout rates vary significantly, giving the actual voters in different states or election districts disparate political influence. Even if the numbers of registered voters are absolutely equal between two election districts, if twice as many people turn out to vote in one district as in the other, then those voters have only half the political effect as the voters in the low turnout district. The example of California 21 and Campbell’s (1996) work illustrate this problem. These four reasons make it irrational for US redistricting to be so dominated by concerns of absolute population equality among districts. This enforced equality is maintained only for congressional districts within the same state, while larger differences that naturally occur between states are completely ignored. Population inequalities are further exacerbated by differences between the population and potential voters and also differences between potential voters and actual voters. Finally, any precise calculation is rendered out of date before even the first election can be held in the new districts. Population equality is a laudable and appropriate goal; in contrast, inflexible requirements of extremely precise equality do not work as promised.

Malapportionment in a comparative perspective The irrationality of the US commitment to absolute population equality of election districts within states (but not among states) is also shown by the fact that no other nation has followed the US standard. Though other countries have followed the general example of the United States in moving towards greater equality of district size, no other country has followed the absolute equality standard that the United States uses for districts within the same state. In fact, redistricting in other countries is complicated by the fact that other countries purposely increase malapportionment by creating special rules that give extra representation to certain areas, a redistricting problem that the United  States does not have. For example, Canada gives additional seats to smaller provinces, Quebec, and its three northern territories. In fact, under the 2012 redistricting, only Ontario, British Columbia, and Alberta did not get additional representation because of Canada’s several special provisions. Britain gives additional seats to Scotland, Wales, and Northern Ireland, though these provisions will be mostly eliminated in the next redistricting.6 However, Britain still specially allocates four seats to three sets of islands. Australia’s constitution guarantees a minimum of five seats to the state of Tasmania. Japan’s 1994 redistricting also gave a one-time, special advantage to smaller prefectures by allocating one seat to each prefecture before allocating the remaining

Drawing the boundaries of election districts  51 253 seats by population. This rule, though, has lapsed, and subsequent redistrictings in 2002, 2013, and 2017 reversed the effects of this rule by reallocating seats among prefectures based only on the population of prefectures. The effects of special rules can be small, as in the case of Australia where the additional Tasmanian seats are only two of 150 total seats, or significant, as in the case of Canada, where special rules account for 24 seats of the Canadian parliament’s 338 seats (Baldini and Pappalardo, 2009, 3). Japan’s special rule, in force from 1994 to 2002, also had a significant effect, exempting 47 of 300 seats from population equality considerations. Table 4.1 summarizes the most important rules that affect malapportionment levels for these five countries.7 The level of malapportionment in a given country is the product of the combined effect of these several factors. Thus, though the United States is better than the other countries in having no special rules and a tolerance of 0 percent in disparities in district sizes within states, the United States also has state boundaries that Britain lacks (though even future redistributions in Britain will initially allocate seats to four regions, and districts cannot straddle two different regions). The United States also has a relatively long time lag between apportionments. The US apportionment is already two years out of date by the time the first election is held under that apportionment. In contrast, Australia requires that the deviation between districts be no more than 3.5 percent of the estimated population at the point halfway in time between the current apportionment and the next apportionment. US apportionments are also less exact than those in Australia

Table 4.1  Redistricting rules in five democracies Country

Special rules Equality rule

State/ province

Timing

Allocation basis

United States Great Britain

None

Yes

10 years

Population

Yes, small

Australia Yes, small

Canada

Yes, major

Japan

Past major, now none

States: 0 percent National: none Past was 20 percent in regions, future is 5 percent national 10 percent in states, but 3.5 percent in between apportionments Provinces: 25 percent National: none Provinces: 33 percent National: 33 percent

Past yes, 5 years (but future lag from minor census time) Yes

Registered voters

Yes

About every States by 7 years population, districts by registered 10 years Population

Yes

10 years

Population

52  Drawing the boundaries of election districts 0.6

Percent of election districts

0.5 0.4

Japan

US

0.3

Canada Britain

0.2

Australia

0.1 0

2%

6% 10% 14% 18% 22% 26% 30% 34% 38% 42% 46% 50% Percent deviation of district population from the national average

Figure 4.1  Population deviation of election districts from a national average

and Britain because the United States uses population numbers rather than registered voters for its calculations of district size. The net effect of all of these different rules produces some surprising patterns in comparative malapportionment. Figure 4.1 presents recent, actual distribution of seats according to population or registered voters as reported by the national electoral commissions of each country.8 Because Figure 4.1 reports the percentage deviation from the ideal district for each election district in each county, the least malapportioned countries will have more districts at lower percentages and fewer districts at higher percentages. The country with the most equal apportionment of districts is the United States, followed by Australia. Britain has the next best distribution, and Canada’s distribution is the one closest to Japan’s. Another way to report Table 4.2  Average deviations from the ideal district, population, and votes Country Australia (2016) Britain (2017) United States (2015, 2018) Canada (2015) Japan (2017)

Average deviation, population or registered voters (percent)

Average deviation, actual votes cast (percent)

6.0

6.7

8.9 3.1

11.9 16.4

12.0 15.9

15.2 15.2

Drawing the boundaries of election districts  53 these same findings is to give the percentage deviation for the average district in each country (see Table 4.2). Because these measurements take several factors into account at the same time, these results do not mean that each and every rule that the United States uses in its redistricting is superior to Australian or British methods or that Japanese rules are all inferior to the redistricting rules used in other countries. Each country has a different time lag from its last reapportionment, different special rules, different rules for the maximum disparity tolerated, different baselines of population or registered voters, and different state or provincial boundaries to consider. Each of these differences contributes to a country’s overall level of malapportionment, some rules making the districts less malapportioned and others making the districts more malapportioned. The comparative waters are further muddied if the actual weight that voters have when they vote for their representatives is calculated (the averages are reported in Table 4.2). Because the United States allocates seats based on population and many people living in the United States are not eligible to vote or choose not to vote, the disparities between how many votes elect a representative in different election districts are significant in the United States. I reproduce Figure 4.1 but instead of using the population or registered voters for a district, I calculate instead the actual number of votes cast in a district, compared to the number of votes cast in the average district in a recent national election in each country (see Figure 4.2).9 Australia’s record stands out as the best by this measure, in part because district boundaries are drawn in states based on registered voters and in part because voting is compulsory in Australia which reduces the possible differences between registered voters and actual turnout. Japan’s record is similar to the

0.3 0.25 0.2 0.15 0.1 0.05 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26% 28% 30% 32% 34% 36% 38% 40% 42% 44% 46% 48% 50% 50%+

0

Japan

United States

Canada

Figure 4.2  Deviation in votes cast from a national average

Britain

Australia

54  Drawing the boundaries of election districts Canadian or US records. Japan’s average deviation (reported in Table  4.2) is better than the US average and the same as the Canadian average. In addition, Japan is the only one of the five reported countries which has no districts with more than a 50 percent deviation from the national average. These high deviation districts are explained in Britain, Canada, and Australia by special rules that advantage certain districts. US deviations are explained by a small number of minority districts that have both low turnout and low numbers of eligible voters. These results show that malapportionment is a much more complex issue than simply saying ‘the United States has equal-sized districts’ and ‘Japan’s electoral districts are hopelessly malapportioned.’ Though there is less malapportionment in the United States than in Japan, both countries are actually more similar than they are different. In addition, there are a variety of factors that affect malapportionment. Both countries have higher disparities caused by (1) using population rather than registered voter numbers, (2) respecting state/provincial boundaries, and (3) having long periods between apportionments. These other factors are more significant factors than Japan’s relative tolerance of disparities (as much as a 33 percent disparity at the national level) and the US policy of no tolerance of any disparity among districts within the same state.

Less malapportionment often leads to more gerrymandering If malapportionment and gerrymandering were completely independent of each other, then it would be appropriate to conclude that Japan should follow the US (or the Australian) example of different rules that would reduce the amount of inequality among Japan’s election districts. Indeed, at first consideration, there seems to be no reason for malapportionment and gerrymandering to be linked. It is possible to have districts of different sizes that also have contorted boundaries. It is also possible for unequal districts to have compact boundaries. Similarly, equal-population districts could also have either contorted or compact boundaries. Furthermore, even if the two concepts are linked, shouldn’t malapportionment and gerrymandering occur at the same time? Isn’t it logical for politicians to either reap all possible political benefits by creating districts that are both malapportioned and gerrymandered? Alternatively, if neutral redistricting criteria are used, wouldn’t that prevent both unequal and contorted election districts? The reality, at least in Japan and the United States, is that in a politicized redistricting process, election districts will tend to be either malapportioned or gerrymandered but not both. In addition, the stricter the rules against malapportionment, the greater the likelihood that gerrymandered districts will crop up. The link between these two concepts stems from the most widely used tool to combat malapportionment: the requirement that districts meet a minimum threshold of deviation from the average or ideal district. As line drawers strive to minimize population deviations, they will also ignore more local

Drawing the boundaries of election districts  55 government boundaries that might have constrained their discretion in drawing district boundaries. Whenever a city or a county is split between election districts, line drawers gain discretion in drawing the line that splits that city or county. The more discretion that line drawers have to split local units, the more freely they can also gerrymander those boundaries to benefit a political party or a group of voters. In contrast, when higher levels of malapportionment are tolerated, it is common for local government units to become the building blocks of election districts. Because cities and counties are treated as units and not split into pieces, the disparities in size among the districts become larger, increasing malapportionment. As malapportionment increases, however, it becomes harder for line drawers to gerrymander boundaries because there are fewer possible combinations of cities and counties that can constitute a district. As the discretion of the line drawers drops, the ability to gerrymander also declines. Thus in both Japan and the United States, higher malapportionment is associated with intact local government units and less gerrymandering. Similarly, lower malapportionment is associated with split local government units and more gerrymandering. This relationship between malapportionment and gerrymandering occurs most prominently in politicized reapportionment processes (Matsui, 1986, 44). If redistricting is depoliticized by placing it in the hands of the judiciary or a non-partisan commission, then the likely outcome will avoid both malapportionment and gerrymandering. Indeed, the record of commission countries such as Britain, Canada, and Australia all support the conclusion that the easiest way to eliminate both malapportionment and gerrymandering is to create non-partisan commissions, tasking them to create equal and compact districts that respect community ties. There are two problems, however, with concluding that non-partisan commissions will also solve the gerrymandering problem in the United States and the malapportionment problem in Japan. First, the candidate-centered nature of US and Japanese elections makes it less likely that non-partisan commissions will actually be non-partisan. Second, even if non-partisan commissions remain politically neutral, the use of strict equality standards still opens the door for gerrymandering for other non-partisan goals. Even with commissions, the tradeoff between malapportionment and gerrymandering can still appear. The fact still holds that the stricter the standard of equality, the more local government units that will be split. The more local government units that are split, the greater the potential for the commission to manipulate district boundaries in pursuit of other goals that commission members see as appropriate, even if the goals are not partisan goals. A first example comes from the work of California’s newly empowered Citizens Redistricting Commission. Conservative critics of the Commission’s work claim that a member of the Commission was associated with special interest groups that had partisan affiliations, that evidence favoring one political party was presented to the Commission as if it were politically neutral, and that

56  Drawing the boundaries of election districts witness testimony to the Commission was manufactured to create political advantage (Greenhut, 2011; Pierce and Larson, 2011). Malanga (2012) adds claims of political bias in the workings of non-partisan commissions in both Colorado and Arizona. It is telling that defenders of the California Commission’s work don’t argue that the process was depoliticized, rather they point out that Republicans also tried as hard as Democrats to subvert the process in their favor (Calbuzz, 2011, Daily Kos, 2011, Cruickshank, 2011). Thus, the defenders of the process accept the politicized nature of the process even while defending the ultimate outcome as fair (Cain, 2015, 38, 83–5, 126). Such an outcome is not unique to California. British parties have long used public consultations to organize and present data so that boundaries are drawn in a politically advantageous manner. This bias was the main reason that the rules for the current redistribution have severely truncated the public input process (Johnston et al., 2013). Boundary commissions can also gerrymander boundaries, even when their goals are politically neutral. Their gerrymanders are for different reasons, such as better representation of minority groups. For example, after the 2000 census, Arizona’s boundary commission created an extreme gerrymander, the Arizona 2nd district. The absurd shape of Arizona 2 was not the result of partisan calculations. Rather, line drawers wanted to put the Hopi Indian Reservation in a different congressional district than the Navajo Indian Reservation because of a longstanding conflict between the two tribes. The Hopi voters were connected to the bulk of the voters in the district who live at the opposite end of the district (in the suburbs of Phoenix) by a chain of small towns and rural areas. In places this connection was only as wide as the Colorado and Little Colorado Rivers, meaning that the district ran down the middle of the river with both sides of the river being in a different congressional district. The district managed to take in portions of half of the counties in the state of Arizona while only including one county in its entirety in the district. It also took in territory from the northeast and northwest corners of the state, combined with voters in the west central part of the state. A final example is the recent work of the Pennsylvania Supreme Court, striking down the state’s congressional district map and replacing it with a map that achieves partisan balance in the state by favoring the Democratic Party. Democrats face a problem in Pennsylvania: their voters are concentrated in the urban areas of Philadelphia and Pittsburgh. Any map that is drawn without paying attention to who is likely to win a district will likely result in Republican advantage because of the concentration of Democratic voters in urban areas. In contrast, the state Supreme Court adopted a plan drawn up by an academic that made it possible for the Democrats to win close to the number of congressional seats in the state as is proportional to their share of the statewide vote.10 In order to achieve this outcome, though, some district boundaries were drawn in order to produce a few more Democratic-leaning districts rather than what would have likely resulted if district partisanship had been ignored in the boundary-drawing process. One group of analysts concludes, ‘Democrats

Drawing the boundaries of election districts  57 couldn’t have asked for much more from the map. It’s arguably even better for them than the maps that they proposed themselves’ (Cohn et al., 2018). Political controversy can and does swirl around the work of the few boundary commissions that operate in the United States. This is consistent with the record in Japan, where the operations of its boundary commissions have been controversial and the rules governing their operations have retained some elements of political control. Specifically, the Japanese legislature only empowers its commission to draw lines after the legislature has manipulated variables such as the number of seats to be created or how those seats are to be allocated to prefectures. After this political input, the actual drawing of district boundaries within prefectures follows clear sets of rules that are politically neutral. The United States is the opposite, with the allocation of seats to states being done by a politically neutral set of rules that operate in a largely uncontroversial manner. The actual drawing of district boundaries within states, however, is controversial and politicized. Japan’s model of politicizing the allocation of seats while reining in politicization in the actual drawing of district lines is similar to the operation of boundary commissions in Britain, Canada, or Australia. In these three nations, political calculations lead to deviations from norms of equality such as favoring certain islands or regions in the UK or special considerations for smaller states such as Tasmania or territories in both Australia and Canada. Japan also politicized the allocation of seats in 1994 by first giving each prefecture a seat before allocating the remaining seats by population. In the subsequent redistrictings, the allocation of seats was also politically influenced by the legislature specifying which prefectures would lose or gain seats (though these specifications were driven by a need to reduce the maximum disparity among districts). All four nations, though, draw the actual boundaries of districts largely without partisan motivations or influence. In Japan, the 2002, 2012, and 2017 redistrictings eliminated the effect of the special seat-allocation rule used in 1994 by specifically targeting the prefectures with the most egregious disparities for a reallocation of seats.11 An unrecognized advantage of the Japanese method is its superiority to other alternatives given a highly politicized redistricting environment in countries such as Japan and the United States which have candidate-centered rather than party-centered elections. Though non-partisan boundary commissions are an obvious solution to partisan redistricting practices, such commissions are unlikely to develop in intensely partisan political environments, and when they do develop (as they have in Japan and some US states), there is intense political pressure to influence their work or limit the applicability of their work in a way that preserves significant political influence in the process. In addition, these commissions may still have their own agendas and can still engage in gerrymandering, especially if they are following a strict population equality standard that makes it easier to contort district boundaries to fit any number of goals that the commission members have.

58  Drawing the boundaries of election districts The Japanese system helps solve this problem by aligning redistricting rules with the strong incentives that politicians face to intervene in the redistricting process. Politicians, in most situations, prefer redistricting rules that reduce the discretion of line drawers. Certainly politicians can use gerrymandering to advantage themselves, but gerrymandering is also dangerous; it can be used just as easily to hurt a politician as it can be used to help a politician. Requirements that boundaries follow, as much as possible, a variety of local government and administrative boundaries guarantee to politicians that gerrymandering cannot be used against them. This guarantee of safety makes it more likely that politicians will embrace Japanese-style redistricting reforms even when they would be unlikely to embrace a non-partisan boundary commission. The potential danger of gerrymandering to a politician, even when done by a non-partisan boundary commission, is illustrated by another outcome of the California boundary commission’s work. In order to create a Hispanic-­ majority district in the San Fernando Valley area of Los Angeles, the remaining non-Hispanic areas of the Valley were combined into the newly reconfigured 30th congressional district. Unfortunately, this new district combined the core areas of two powerful Democratic incumbents, Howard Berman and Brad Sherman. The boundary commission gerrymandered district boundaries to create a Hispanic-majority district, but this gerrymander forced two powerful incumbents to run against each other in the neighboring district.12 This example shows how politicians would likely prefer a boundary regime that minimizes gerrymandering by using local boundaries rather than the rules of an ostensibly neutral commission that might be subverted politically or used to gerrymander for other goals.

Japan’s redistricting method Japan’s redistricting method, despite its several prominent flaws, has one feature that recommends the method to other countries. Japan’s line drawers are required to respect several types of local government and administrative boundaries and are allowed a reasonable amount of deviation from population averages to create districts with no or a minimal number of divided local units. The requirement that local boundaries be followed, along with the use of several sets of local boundaries, removes most of the discretion in drawing the boundaries, making it nearly impossible to gerrymander boundaries. In a highly politicized redistricting setting, reducing the discretion of line drawers is perhaps the most effective method to create a completely impartial redistricting process. Key to the success of this method is the reduction in discretion of line drawers by using multiple sets of local boundaries. For example, in the 1994 redistricting, the Commission was instructed to use the following sets of local boundaries: cities, counties, wards of cities, and the previous boundaries of the larger multi-seat election districts. In addition, the 1994 boundary commission

Drawing the boundaries of election districts  59 largely followed the work of the 1991 boundary commission that also followed administrative boundaries in Japan that grouped cities and counties together.13 The important difference in this method as compared to the typical requirement that line drawers respect local boundaries is that this method uses additional sets of boundaries that are both larger than cities and counties and smaller than cities. These additional sets of boundaries dramatically reduce the amount of discretion that line drawers have when creating new election districts. For example, the entire nation of Japan is divided up into municipalities. The municipalities are grouped into three types, depending on their populations. The largest municipalities are called cities. Smaller municipalities are called towns and villages. Japan further groups only towns and villages into larger administrative units called counties. Thus, urban areas are in cities but not in counties. Rural areas are in towns or villages and those units are grouped into counties. The largest cities (those with at least one million people) are further subdivided into wards. These divisions alone are not much different than the local government divisions used in most other advanced industrial democracies. The Japanese process (used in 1994) then adds administrative regions that are used by the central government. Regions are smaller than prefectures, but typically group together all of the cities and counties of an identifiable area within a prefecture. In addition, the election districts used prior to 1994 were also used as guides in creating the new election districts. Prior to 1994 Japan used large multi-seat districts that had been put in place in 1947 and changed very little since then. These election districts were not gerrymandered to begin with and served as appropriate surrogates, along with the administrative regions, for logical and reasonable groupings of municipalities into election districts. In order to minimize the number of times that these various local units would have to be split in order to create new election districts, the 1994 reforms tolerated a maximum disparity from the national average of 33 percent. Of course, this disparity seems ridiculously high considering the 0 percent tolerance in the United States, the 10 percent tolerance in Australia (and 3.5 percent tolerance at the projected midpoint between apportionments), and the 20 and 25 percent tolerances used in Britain and Canada. However these numbers can be misleading because the Japanese tolerance was from the national average whereas all of the other tolerances reported above are for averages within states, provinces, or regions. For example, the 0 percent tolerance within states in the United States typically becomes a maximum 40 percent deviation if calculated at the national level. Similarly, if Japan’s districts in 1994 had been reapportioned with absolute equality within prefectures and all 300 seats were apportioned according to population, the maximum national disparity would still have been 40 percent, which by coincidence is the same as the maximum disparity in the United States. Thus, the British requirement that the maximum disparity from the national average among districts be less than 5 percent

60  Drawing the boundaries of election districts for its forthcoming redistricting process is achievable only because Britain has four regions and no state or provincial boundaries that would have to be followed in drawing new district boundaries. The cost of tolerating higher levels of malapportionment is mixed. For example, allowing a hypothetical 10 percent deviation within a state or province usually does not affect the maximum deviation of districts within a country. Often the most deviant district is the largest state or province that has only one seat (Montana in the United States). Thus, allowing a ten percent deviation does not increase that district’s size because there are no lines to be drawn for district boundaries in that one-seat state or province. Allowing a 10 percent deviation, however, in the drawing of district lines does increase the average deviation of districts. For example, if Japan had redistricted after the 2010 census requiring all districts within a prefecture to have the same population, the average deviation would have declined to only 3.8 percent whereas the actual average deviation after the 2010 census was 17.4 percent. Under either allocation, however, the maximum deviation would still have been 38 percent. This difference between the average deviation and the maximum deviation has negatively affected the recent redistrictings in Japan, increasing the overall inequality of Japan’s election districts from what would have been possible. Court and public oversight of Japan’s redistricting focuses on the maximum deviation between the smallest and largest districts and the total number of districts that exceed a certain level of inequality (rather than the average deviation). Thus, when district lines are redrawn, as they were in 2002, 2012, or 2017, boundaries are not shifted to equalize the size of districts unless the inequality in a district exceeds the maximum tolerable deviation of 1:2. For example, the 2002 redistricting reduced the maximum deviation in the country from 1:2.57 to 1:2.06 and reduced the number of districts with more than twice the population of the smallest district from 95 to only 9. These changes affected 55 of 300 districts. In 13 additional districts, boundaries were adjusted to match boundary changes in local government units, including the elimination of one split municipality (caused by a municipal merger), which reduced the total number of split municipalities in the country from 17 to 16 (Soumusho, 2001).14 In 2012, operating under a more stringent Supreme Court decision, the maximum disparity declined from 1:2.52 to 1:1.998 and the number of districts with more than twice the population of the smallest district dropped from 97 to zero. In addition, twelve cities that had previously been divided between districts were reunited while three cities were newly divided (many of the formerly split municipalities were caused by mergers among municipalities that occurred after the district boundaries were drawn) (Soumusho, 2013). The 2017 reapportionment was similar. It only affected the boundaries of 97 of the 295 existing districts. All 32 districts that had more than double the population of the smallest district had their boundaries redrawn so that no districts remained with more than double the population (the new maximum

Drawing the boundaries of election districts  61 disparity ratio was 1:1.956). For the first time the redistricting used future (2020) population estimates rather than past census data. The 71 districts that were estimated to have more than double the population of the smallest district were all revised so that no district was projected to have more than double the population of the smallest district in the 2020 census. 94 of the 97 districts that had boundary changes were altered in order to reduce the maximum population disparity. In addition, minor boundary changes were made to three additional districts in order to unify two rural regions that had previously been split among the three different districts (Soumusho, 2017). The work of these boundary commissions was enacted into law by the Diet. Similarly the 1994 boundary commission (and the Diet) made only 19 alterations to the boundaries first proposed by the boundary commission in its 1991 recommendations. Of these 19 changes, only three were to improve the national maximum disparity by changing boundaries in the three proposed districts with the largest and smallest of populations. Japan’s redistricting method decreases discretion and makes gerrymandering extremely difficult to do. However, these barriers to gerrymandering come at the cost of increasing the average deviation of districts from the national average. Japan’s rules, however, do not increase the maximum deviation among districts. Japan’s rules are also attractive to self-interested politicians because they reduce the risk that any reapportionment may change boundaries in a way that might harm a politician’s reelection chances.

Using the Japanese apportionment method The advantages of using the Japanese method to draw district boundaries can be shown by a practical application of this method to a hypothetical drawing of congressional district boundaries in a US state. The first step of the Japanese method is to identify local government divisions that group counties together and also help divide cities and counties that are so large in population that they have to be split between two or more congressional districts. Japan’s bureaucratic guidelines crafted for the 1994 line-drawing process did this well and reduced discretion so much that in some prefectures there was only one possible way to draw the district boundaries (Christensen, 2004). In US states, a common administrative boundary that groups counties together is state judicial districts. Another common problem in the United States is the division of large, urban counties. When large counties have to be divided, city boundaries help guide that process, but many large counties are a crazy patchwork of cities and unincorporated areas that give little guidance to line drawers as how to best divide a large county that must be divided into multiple election districts. Fortunately, most counties combine municipalities and unincorporated areas into school districts or other administrative divisions. Similarly, when a large city must be divided, most large cities are also subdivided into neighborhoods or administrative districts. All of these types of units are useful, in addition

62  Drawing the boundaries of election districts to city and county boundaries, in drawing the lines for new congressional districts. Japan uses the equivalent of these types of boundaries when drawing the boundaries of new election districts. Boundary commissions (or courts) in the United States typically don’t look at administrative subdivisions beyond cities and counties, but they should in order to reduce the discretion of line drawers. In a hypothetical redistricting in the United States, the Japanese method would result in the following redistricting rules: 1 2 3 4

5

All congressional districts must fall within a maximum population deviation of only 10 percent from the state average.15 District lines must minimize the number of local units (see requirement 4) that are split while still remaining within the maximum allowed 10 percent deviation. Districts also must be contiguous, and district compactness is also a consideration. Panhandles or peninsulas of districts should be avoided. District boundaries first must respect groupings of counties, typically state judicial districts; then counties; then groupings of cities within a county, typically school districts; then cities; then administrative divisions within cities. Where multiple options exist (that are contiguous, within 10 percent of the average, and minimize the number of local units divided), district compactness and closeness to the average district population guide the decision of which option to choose.

I tested this Japanese method in a hypothetical redistricting of the 16 congressional districts in the state of Ohio, using the 2010 census data. Ohio is a difficult test for this method, not only because the state’s actual redistricting was a highly effective Republican gerrymander but also because the many congressional districts, counties, and cities in the state make it a difficult task to see if following these rules can create only a few, largely identical redistricting maps that would reduce the discretion of line drawers, whoever they might be, to almost zero. Even in this difficult test case, the outcome is almost entirely determined by the line-drawing rules, and what little discretion remains would not significantly affect any particular district’s population or geographic makeup. Ohio’s population of 11,536,504 divided among 16 congressional districts yields a target population of 721,031 for each district. Allowing a 10 percent deviation both above and below this target gives a range of acceptable district populations from 648,928 to 793,134. Ohio also combines its counties into 12 judicial districts. Two judicial districts have the right populations to include congressional districts entirely within their boundaries. For the remaining ten judicial districts, there is only one possible judicial district that each district can be combined with that (1) minimizes the number of congressional districts that straddle judicial district

Table 4.3  A hypothetical redistricting in Ohio, using Japanese rules Judicial district number

Population (1,000)

Number Location of seats

Combined?

Divided?

1

802

1.11

Cincinnati

See Table 4.4

2

1,031

1.43

Dayton and West Central rural

Only option is 12 Can only combine with 12 (3 not available)

3

787

1.09

No need

4

634

0.88

Northwest rural South rural

5

1,485

2.06

Central rural

No need

6

887

1.23

Toledo and Northwest rural

Can only combine with 9 (5 and 3 not available)

7

561

0.78

8

1,280

1.78

9

1,130

1.57

Young-stown Can only and East rural combine with 4 (5 and 11 not available) Cleveland Can only combine with 11 (if 9 would require additional splits) Can only Akron and combine with Cleveland suburbs 6 (8, 11, and 5 not available)

Must combine with 7 and 10 (12 and 5 not available)

3 West counties must be together (population and contiguity) 3 East counties must combine with 12 No need Several options possible for East counties to 7 Remainder must combine with 10 Several options possible East–West split is compact and equal populations 5 West counties only possible split (population and contiguity) East counties to 9 No need

See Table 4.4

Akron and neighbor must be together (population) Remaining counties to 6 (continued )

64  Drawing the boundaries of election districts Table 4.3  (continued ) Judicial district number

Population (1,000)

Number Location of seats

Combined?

Divided?

10

1,163

1.61

Columbus

See Table 4.4

11

797

1.11

Northeast rural

Can only combine with 4 (5, 3, and 12 not available) No additional combinations needed if with 8

12

980

1.36

Southwest suburbs and rural

Must take excess from 11 Must combine with 2 because 2 has no other options

West 3 counties must be together (population and contiguity) 5 South and East counties are left to combine with 2

Only 1 county can go to 8 (population and contiguity)

boundaries, (2) keeps congressional districts within required population ranges, and (3) keeps districts contiguous and compact. Table 4.3 gives the details of this process. The result of this process is 16 congressional districts with populations that range from 656,301 to 787,269. Ten districts lie entirely within one judicial district. Six congressional districts (the minimum possible number) straddle the boundaries of two judicial districts. Only three counties are split between congressional districts, and all three must be split because their populations are too large to be in only one congressional district. Even more fascinating is the fact that of the ten required divisions of either counties or judicial districts, only four of the ten had more than one possible, acceptable (according to the rules) division. In addition, all four of the divisions that had multiple options did not create significant opportunities for line-drawing mischief as in each case the several options available were largely identical in terms of outcomes. Two divisions had options between similarly situated suburbs located at the county line. The other two divisions had options between similar rural counties with small populations. These results perhaps seem rather unremarkable. It is obvious that anyone can draw a map of congressional districts in a state that are both compact and have relatively equal populations. What is different about this outcome is the process that removes almost all discretion by following very simple rules. What discretion remains is politically irrelevant. These rules reduce the discretion of

Drawing the boundaries of election districts  65 Table 4.4  Possible division of three Ohio counties with large populations County

Division

Cuyahoga (Cleveland)

Cleveland City and east suburbs would have too many people to combine with the county from JD 11. Cleveland City must be combined with its west and then south suburbs. Grouping by school districts and contiguity, must draw line between Parma and Strongsville. No other options exist. A minimum of 25,000 voters must be shifted north to JD 12. Hamilton County has about 23 school districts. There are multiple school districts that are contiguous to JD 12. A combination of smaller school districts in the northeast and northwest corners of the county are the best options for combination with JD 12, though multiple options exist. Columbus City has the necessary population to be a congressional district, and the rest of Franklin County could then be combined with JD 4. Unfortunately the city creates non-contiguous areas of the county. Another option would be to take the southernmost areas of the city and the county and combine them with JD 4.

Hamilton (Cincinnati)

Franklin (Columbus)

line drawers, whether they be politicians, a neutral boundary commission, or election department bureaucrats. I have also used this method in a variety of other states, including California, Georgia, and Utah, and have produced similar results (Christensen, 2004). California had 13 counties and four cities that had to be split because they had too many people to be a congressional district in the first place. Only one of the state’s six judicial districts had a population that required that it share a congressional district with its neighbor. Only two additional counties had to be split beyond the 13 splits that were unavoidable. In Utah, only one judicial district and one county were so large that they had to be split. With this split the rest of the state could be divided following administrative boundaries, but the result had higher disparities and one less compact district than would result if a second split of a judicial district were allowed. With that additional split, three possibilities emerged and all of them reduced the population disparity among Utah’s four congressional districts to between 3 and 5 percent and improved the compactness of the resulting districts. Georgia’s results were similar with a few judicial districts having no obvious division of counties within the district, but no significant consequential differences in the resulting districts even if different options were used. The conclusion of this exercise in the drawing of congressional districts in Ohio, California, Utah, and Georgia is that tolerating a reasonable difference in district size while also using a variety of local government and administrative boundaries produces an outcome that minimizes the discretion of line

66  Drawing the boundaries of election districts drawers. This loss of discretion is desirable as it makes it nearly impossible to gerrymander district boundaries. As long as line drawers are constrained to minimize the number of divisions of local units to only that which is required to meet the population targets of a maximum deviation of 10 percent, even if there are several possible alternative maps, the political differences between the maps are likely to be small. Even in those situations, the line drawers should be constrained to choose the option with the least disparity among districts.

Conclusion Japan’s record on redistricting is not, overall, an admirable record. Its districts are among the most malapportioned districts of the advanced industrial democracies. This disparity is caused by several factors. Some of them are unavoidable: Japan allocates seats to prefectures, and prefectural boundaries are respected, Japan typically redistricts at ten-year intervals, Japan uses population numbers rather than registered voter numbers. Some of the factors are easily changed: Japan has begun reallocating seats according to a more frequent calendar and anticipating population changes in its reallocations. In addition, this redistricting process could be made mandatory and automatic. Japan could also work to minimize disparities in all districts, not just the districts that have the greatest disparities. These changes should be made. However, these flaws should not obscure the excellent method that Japan has developed to actually draw district lines when it does redistricting. In its line drawing, Japan tolerates levels of disparity that are only slightly greater than those that exist in other countries, and given the division of the country into prefectures, these levels of disparity are not much higher than what would necessarily exist under even the most stringent of population equality rules. More importantly, the aggressive use of a variety of administrative boundaries in Japan constrains the discretion of line drawers, making it impossible for line drawers to contort districts in ways that either benefit political parties or serve other non-partisan goals that a redistricting committee might have. The combination of population targets (maximum 10 percent deviation) and several levels of local boundaries reduces a complex redistricting process with a seemingly endless number of possible boundary permutations into a much simpler exercise in finding the only possible outcome or choosing among a few, similar outcomes. There are, of course, several other factors that would affect the use of the Japanese line-drawing rules in the United States. Japanese-style rules would make it impossible to contort district boundaries to pull Hispanic or African American voters together from different areas to create districts that had a majority of voters from a particular ethnic or racial group. Such minority districts would still exist, where minority voters are geographically concentrated, but the number of such minority districts would decline as the

Drawing the boundaries of election districts  67 gerrymanders used to create such districts would not be allowed. Perhaps for this reason alone, the Japanese model is not a desirable alternative for the United States. On the other hand, it is possible that minority districts might actually be counterproductive to the interests of the minorities represented (Thernstrom, 1987, 187–8, 205, 221–46, 243; Swain, 1995, 210–16, 229–39).16 There is also some political and judicial discomfort with the open-ended use of race or ethnicity as a requirement in the drawing of district boundaries. Finally, voting by race or ethnicity that was one of the reasons used by the US Supreme Court to justify the creation of such minority districts seems to be declining (Bullock and Dunn, 1999; Pildes, 2002, 1529–35; Harvard Law Review, 2003, 2219). For  example, in the 2016 elections, 15 Black, Hispanic, or Asian candidates were elected in white-majority districts and eight white candidates were elected in ­majority-Black, -Hispanic, or -Asian districts. In addition, a 2017 US Supreme Court decision on racial gerrymandering in North Carolina turned largely on the fact that there was no evidence of bloc voting by white voters against Black candidates in the challenged districts (Cooper v. Harris 581 U.S. ____ 2017). In fact, of the 134 US House districts with more minority residents than white residents, minority candidates won 77 seats and white candidates won 57 seats in the 2016 elections.17 Perhaps voting patterns in the United States are evolving to the point where gerrymandered districts of minority voters are no longer needed to ensure reasonable levels of minority representatives in Congress. A second problem with implementing Japan’s rules in the US redistricting process is the Supreme Court decisions that specifically require absolute equality among the districts within a state. It is the point of this chapter to show that this requirement is misguided in two ways: (1) the standard that is supposedly upheld is unrealistic and (2) adherence to the absolute equality standard has opened the door to egregious gerrymandering. In addition, in the next chapter I argue that the court’s role in redistricting disputes is harmful because the very types of arguments that are used in court disputes are inimical to the type of political compromise that is necessary in many political contexts, including redistricting.

Notes   1 Japanese voters have, since 1994, also cast an additional ballot for their preferred party, in addition to their ballot for their preferred candidate.  2 Most commentators recognize that malapportionment levels have declined, making malapportionment less of a factor in recent Japanese elections. They fail to give proper emphasis, though, to the fact that even at its worst, malapportionment rarely affected the outcome of Japanese elections (Christensen and Johnson, 1995; Baker and Scheiner, 2007; Jou, 2009). In addition, they ignore the comparative context, that levels of malapportionment in Japan are qualitatively similar to levels in other countries.   3 This number is based on the ‘apportioned’ population calculated by the US Census Bureau for reapportionment purposes.

68  Drawing the boundaries of election districts  4 Prefecture is the English word chosen by the Japanese government to describe their provinces. The Japanese actually use four names for their provinces: ken, fu, to, and dou.  5 American Community Survey, 2017. US Census Bureau. Cain (2015) notes this same problem in the apportionment of state legislative districts in California.   6 A 2018 redistricting plan (not yet enacted) in the United Kingdom assigned seats to four regions by population, but the assignment could be made even more equal. For example, there would be slightly less disparity among regions if one of Northern Ireland’s seats were given to England.   7 Australia staggers its reapportionments by state with every state reapportioning every seven years. At the time of reapportionment, the deviation of districts within a state may be no more than 10 percent, but in addition, the predicted deviation 3.5 years in the future (the midpoint between the current and the next reapportionment) must be no more than 3.5 percent.  8 The data is taken from the following websites: Japan (Ministry of Internal Affairs and Communications [Soumushou], Elections division [Senkyobu]), www.soumu.go. jp/senkyo/48sansokuhou/index.html; United States (US Census Bureau, American FactFinder) factfinder2.census.gov/faces/naf/jsf/pages/index.xhtml; Britain (Electoral Commission, 2017 General Election) www.electoralcommission.org.uk/our-work/ourresearch/electoral-data; Canada(Elections Canada, 2015 election results) www.elections. ca/content.aspx?section=res&dir=rep/off/42gedata&document=summary&lang=e; Australia (Australian Electoral Commission, 2016 Federal Election) results.aec.gov. au/20,499/Website/HouseDownloadsMenu-20,499-Csv.htm.   9 The elections used are reported in Table 4.2. The data sources for Figure 4.2 are the same as Figure 4.1 except the US election data was taken from the clerk of the US House and Japanese data came from www.soumu.go.jp/main_content/000513918.pdf. 10 In the first election under the new plan (2018), the parties split the Congressional delegation evenly (9 seats each). However, because the Republicans did not contest one district, in the districts in which both parties ran candidates, the Democrats won 42.1 percent of the seats with 53.7 percent of the vote. 11 The 47 seats were distributed, one to each prefecture, meaning that only a few of the 47 seats contributed significantly to malapportionment. Of the 15 prefectures that had only two or three seats in the 1994 redistricting, six of them had lost a seat by the 2012 redistricting and two of them had actually gained a seat because of population increases. The 2002, 2012, and 2017 reapportionments reallocated seats among prefectures only on the basis of population. 12 The shape of the Hispanic-majority district is not unusual, though its creation forced the other San Fernando Valley district to include both the southern and western portions of the Valley in an unusual L-shaped district. More important than the shape of the district is the fact that non-Hispanic voters in the Valley were grouped together into one district that forced two long-time incumbents to run against each other. 13 The 1994 districts only had minor changes in 19 of Japan’s 47 prefectures from the 1991 districts that were proposed but never enacted. Thirteen of the 19 changes were in response to public testimony requesting that a city be switched to an adjoining district. Three changes were made to reduce the maximum population disparity, and two changes eliminated non-contiguous districts. One change reflected changes to the internal boundaries of a city’s administrative districts. Yomiuri Shimbun, August 12, 1994, 12. 14 Mainichi Shimbun, December 12, 2001, 3.

Drawing the boundaries of election districts  69 15 This level seems stricter than the 33 percent deviation allowed in Japan, but that deviation was from the national average. This deviation is from the state average, an average that is ignored in Japanese reapportionments because of their focus on the maximum, national disparity. 16 Canon (1999) and Swain (1995) argue that many minority politicians from majorityminority districts moderate their positions so that they represent non-minority voters, in addition to representing the minority voters in their districts. 17 The classification of districts is taken from the 2015 estimates of congressional district populations by the US Census Bureau. In addition to districts with a majority of their people being African American, Asian American, or Hispanic, I also counted districts in which those reporting minority status constituted a majority of a district’s population. I included those reporting being Hispanic as well as all of the Census’s racial categories, including other, but excluding white and two or more races.

References Baker, Andy, and Ethan Scheiner. 2007. “Electoral System Effects and Ruling Party Dominance in Japan: A Counterfactual Simulation Based on Adaptive Parties.” Electoral Studies 26, no. 2 (June): 477–91. Baldini, Gianfranco, and Adriano Pappalardo. 2009. Elections, Electoral Systems and Volatile Voters. New York: Palgrave Macmillan. Bernhard, William T. 2002. Banking on Reform: Political Parties and Central Bank Independence in Industrial Democracies. Ann Arbor, MI: University of Michigan Press. Bullock, Charles S. III, and Richard E. Dunn. 1999. “The Demise of Racial Districting and the Future of Black Representation.” Emory Law Journal 48, no. 4 (Fall): 1209–53. Cain, Bruce E. 2015. Democracy More or Less: America’s Political Reform Quandary. New York: Cambridge University Press. Calbuzz. 2011. “Why the ProPublica Remap Yarn is Nonsense.” Calbuzz, December 23. www.calbuzz.com/2011/12/why-the-propublica-remap-yarn-is-nonsense. Accessed July 20, 2012. Campbell, James E. 1996. Cheap Seats: The Democratic Party’s Advantage in US House Elections. Columbus, OH: Ohio State University Press. Canes-Wrone, Brandice, David W. Brady, and John F. Cogan. 2002. “Out of Step, Out of Office: Electoral Accountability and House Members’ Voting.” American Political Science Review 96, no. 1 (March): 127–40. Canon, David T. 1999. Race, Redistricting, and Representation: The Unintended Consequences of Black Majority Districts. Chicago, IL: University of Chicago Press. Carey, John M., and Matthew Soberg Shugart. 1995. “Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas.” Electoral Studies 14, no. 4 (December): 417–39. Carson, Jamie L., Gregory Koger, Matthew J. Lebo, and Everett Young. 2010. “The Electoral Costs of Party Loyalty in Congress.” American Journal of Political Science 54, no. 3 (July): 598–616. Christensen, Ray. 2004. “Redistricting in Japan: Lessons for the United States.” Japanese Journal of Political Science 5, no. 2 (November): 1–27.

70  Drawing the boundaries of election districts Christensen, Raymond V., and Paul E. Johnson. 1995. “Toward a Context-Rich Analysis of Electoral Systems: The Japanese Example.” American Journal of Political Science 39, no. 3 (August): 575–98. Cohn, Nate, Matthew Bloch, and Kevin Quealy. 2018. “The New Pennsylvania Congressional Map, District by District.” The Upshot, New York Times, February 19. https://nytimes.com/interactive/2018/02/19/upshot/pennsylvania-new-housedistricts-gerrymandering.html. Accessed May 16, 2018. Cruickshank, Robert. 2011. “ProPublica’s Absurd Attack on California’s Redistricting.” Calitics, December 22. www.calitics.com/diary/14076/propublicas-absurd-attack-oncalifornia-redistricting. Accessed July 20, 2012. Daily Kos. 2011. “ProPublica CA Redistricting Report: GOP Manipulation They Missed in OC.” Daily Kos, December 27. www.dailykos.com/story/2011/12/27/1049203/ProPublica-CA-Redistricting-Report-GOP_Manipulation-They-Missed-in-OC. Accessed July 20, 2012. Fougere, Joshua, Stephan Ansolabehere, and Nathaniel Persily. 2010. “Partisanship, Public Opinion, and Redistricting.” Election Law Journal 9, no. 4 (December): 325–47. Greenhut, Steven. 2011. “Removing Politics from Politics: California’s Redistricting Commission Draws Slanted Lines on the Political Map.” City Journal, August 12. www.city-journal.org/html/removing-politics-politics-10899.html. Accessed November 23, 2019. Grindle, Merilee Serrill. 2012. Jobs for the Boys: Patronage and the State in Comparative Perspective. Cambridge: Harvard University Press. Hata, Hiroyuki. 1990. “Malapportionment of Representation in the National Diet.” Law and Contemporary Problems 53, no. 2 (Spring): 157–70. Harvard Law Review. 2003. “The Future of Majority-Minority Districts in Light of Declining Racially Polarized Voting.” Harvard Law Review 116, no. 7 (May): 2208–29. Hickman, John C., and Chong Lim Kim. 1992. “Electoral Advantage, Malapportionment, and One Party Dominance in Japan.” Asian Perspective 16, no. 1 (Spring– Summer): 5–25. Johnston, Ron, Charles Pattie, and David Rossiter. 2013. “Local Inquiries or Public Hearings: Changes in Public Consultation over the Redistribution of UK Parliamentary Constituency Boundaries.” Public Administration 91, no. 3 (September): 663–79. Jou, Willy. 2009. “Partisan Bias in Japan’s Single Member Districts.” Japanese Journal of Political Science 10, no. 1 (April): 43–58. Karch, Andrew, Corrine M. McConnaughy, and Sean M. Theriault. 2007. “The Legislative Politics of Congressional Redistricting Proposals.” American Politics Research 35, no. 6 (November): 808–25. Kobayashi, Yoshiaki. 2012. “Giin Teisuu Fukinkou Ni Yoru Minshu Shugi No Kinou Fuzen—Mini Futaku, Kokkai Shingi, Seisaku Keisei No Yugami” [How Malapportionment Produces Dysfunctional Democracy: Distorting Citizen Mandates, Diet Debate, and the Policy-Making Process]. Senkyo Kenkyuu [Japanese Journal of Electoral Studies] 28, no. 2: 15–25. Kriner, Douglas, and Liam Schwartz. 2008. “Divided Government and Congressional Investigations.” Legislative Studies Quarterly 33, no. 2 (May): 295–321. Malanga, Steven. 2012. “Redistricting Wars: The Hidden Story of the 2012 Elections.” City Journal 22, no. 2 (Spring): www.city-journal.org/2012/issue_22_2.html.

Drawing the boundaries of election districts  71 Matsui, Shigenori. 1986. “The Reapportionment Cases in Japan: Constitutional Law, Politics, and the Japanese Supreme Court.” Osaka University Law Review 33 (March): 17–46. McElwain, Kenneth Mori. 2013. “The Nationalization of Japanese Elections.” In Japan under the DPJ: The Politics of Transition and Governance, edited by Kenji E. Kushida and Phillip Y. Lipscy, 45–72. Stanford, CA: Walter H. Shorenstein AsiaPacific Research Center. Nemoto, Kuniaki, and Chia-hung Tsai. 2016. “Post Allocation, List Nominations, and Preelectoral Coalitions under MMM.” In Mixed-Member Electoral Systems in Constitutional Context, Taiwan, Japan, and Beyond, edited by Nathan F. Batto, Chi Huang, Alexander C. Tan, and Gary W. Cox, 165–93. Ann Arbor, MI: University of Michigan Press. Parker, David C. W., and Matthew Dull. 2009. “Divided We Quarrel: The Politics of Congressional Investigations, 1947–2004.” Legislative Studies Quarterly 34, no. 3 (August): 319–45. Pierce, Olga, and Jeff Larson. 2011. “How Democrats Fooled California’s Redistricting Commission.” Pro Publica, December 21. www.propublica.org/article/how-democratsfooled-californias-redistricting-commission. Accessed July 19, 2012 . Pildes, Richard H. 2002. “Is Voting-Rights Law Now at War with Itself ? Social Science and Voting Rights in the 2000s.” North Carolina Law Review 80, no. 5 (June): 1517–73. Soumusho. 2001. [Ministry of Internal Affairs and Communications]. “Shuugiin Shousenkyoku Senshutsu Giin No Senkyoku No Kaiteian Ni Tsuite No Kankoku, Heisei 13 nen 12 gatsu 19 nichi” [Recommendations regarding the Redistricting Plan for House of Representatives Members Elected from Single-Member Districts, December 19, 2001]. www.soumu.go.jp/main_sosiki_singi/senkyoku/pdf/senkyoku_ shingi.06_03.pdf. Accessed May 19, 2017. Soumusho. 2013. [Ministry of Internal Affairs and Communications]. “Shuugiin Shousenkyoku Senshutsu Giin No Senkyoku No Kaiteian No Gaiyo, Heisei 25 nen 3 gatsu 28 nichi” [Overview of the Redistricting Plan for House of Representatives Members Elected from Single-Member Districts, March 28, 2013]. www.soumu. go.jp/main_content/000215094.pdf. Accessed May 19, 2017. Soumusho. 2017. [Ministry of Internal Affairs and Communications]. “Shuugiin Giin Senkyoku Katei Shingikai Setchihou, Oyobi Koushoku Senkyohou No Ichibu O Kaisei Suru Houritsu No Ichibu O Kaisei Suru Houritsu No Shikou Ni Tsuite, Heisei 29 nen 6 gatsu 16 nichi” [Concerning the Implementation of Partial Legal Reforms, Partial Reforms of the Public Officers Elections Law and Law Establishing the House of Representatives District Boundary Advisory Committee, June 16, 2017]. www.soumu.go.jp/main_content/000496842.pdf. Accessed May 16, 2018. Swain, Carol M. 1995. Black Faces, Black Interests: The Representation of African Americans in Congress, enlarged ed. Cambridge, MA: Harvard University Press. Theriault, Sean M. 2004. “Public Pressure and Punishment in the Politics of Congressional Pay Raises.” American Politics Research 32, no. 4 (July): 444–64. Thernstrom, Abigail M. 1987. Whose Votes Count? Affirmative Action and Minority Voting Rights. Cambridge, MA: Harvard University Press. Wada Junichiro. 2012. “Teisuu Haibun to Kuwari—Keizaigaku No Shiten Kara” [Apportionment and Districting: From the Viewpoint of Economics]. Senkyo Kenkyuu [Japanese Journal of Electoral Studies] 28, no. 2: 26–39.

5 Supreme Courts and election procedures

Lesson 5: Courts should intervene in the election process minimally, allowing political processes to freely operate within specified boundaries.

Introduction Political decision making and judicial decision making are fundamentally different. Political processes are responsive to democratic pressures; they reflect the views of disparate interests and voter groups. Political decisions are also typically based on compromises. As such, their results may not even be consistent with an internal logic. As the well-known quip puts it, you don’t want to see legislation or sausage being made. The contrast with judicial decision making is stark. Judicial decisions must follow an internal logic, and that logic must be reconciled across all prior judicial decisions on that same topic (Cain, 2015, 32–33, 130–31). In addition, only legal arguments, especially constitutional arguments, can be the basis of judicial decisions. For example, a court must have a constitutionally based reason before it can strike down the actions of a democratically elected legislature. Though it is true that courts and legislatures can consider any type of argument—political, legal, emotional, economic, constitutional, moral, etc.—courts, unlike legislatures, relegate non-legal arguments to the margins of their analyses. Arguments that are the bread and ­butter of political decision making become, when a court sits, inferior to both legal and constitutional arguments. These differences make courts a good forum for certain kinds of decision making and legislatures a better forum for other types of decisions. For example, there is a long history of debate over whether the courts are the appropriate forum for especially contentious social issues (Bickel, 1975; Cox, 1976). Regardless of the outcome of this debate, all sides agree that legislative decisions typically have greater legitimacy and are more likely to resolve the issue conclusively than a similar decision made by a court. On the other hand, courts seem to provide the best forum for protecting individual rights, especially individual rights of people with unpopular views or members of minority groups.

Supreme Courts and election procedures  73 Similarly, courts seem to be better suited to regulate the actual procedures of democracy, especially elections. Politicians, it is argued, should not write or administer the very rules from which they will personally benefit (Cain, 2015; Ely, 1980; Issacharoff, 2002; Pildes, 2004). Thus, politicians should not draw their own district boundaries, investigate their own ethical lapses, or count the ballots in their own election campaigns. Contrary to Justice Frankfurter’s warning in the US Supreme Court’s malapportionment cases (Colegrove v. Green, 328 US 549, 1946), courts should wade into the political thicket, especially when there is no hope that legislatures will actually respond to and solve a particular problem in the electoral process. The historical record also supports this conclusion. In both Japan and the United States, courts have eliminated serious malapportionment problems that the legislatures were unwilling to solve. These interventions were appropriate and strengthened the democratic nature of elections in both countries. There is, however, an equally strong argument that points to the conclusion that the legislature is better suited for solving political problems, including problems in elections and the procedures of democracy. This argument has a long history in the political question doctrine that has a prominent place in jurisprudence in both Japan and the United States. At its simplest, this doctrine states that courts are ill suited to have the information, sensitivities, legitimacy, or legal basis to intervene and substitute their judgment for the judgment of the people’s elected representatives on political matters. Japan’s Supreme Court has balanced these two competing concerns most effectively in its handling of malapportionment cases. On the one hand, it has limited the self-interest of politicians by setting a maximum level of tolerable malapportionment. On the other hand, the political process of the legislature has operated with a high level of freedom in crafting compromise solutions. Thus, the political process still functions, making it possible to consider a wide range of factors and information. The outcome also retains a greater sense of legitimacy, being the result of the debate and compromise of a democratically elected legislature. The Japanese Supreme Court has created a rare and underappreciated example of judicial restraint that combines the real need for judicial intervention to set limits on manipulations of the political process while also refraining from completely supplanting the legislative process with a judicial process that is both inflexible and less legitimate.

The overly legalistic judicial oversight of US elections In contrast to the Japanese Court’s restraint, the US Supreme Court has taken a highly interventionist position on many issues of election procedures. This stance began with Baker v. Carr in 1962 when the Court necessarily pared back the political question doctrine to make it possible to push into the political thicket that Justice Frankfurter had warned against. The Court limited the areas in which it would exercise restraint under the political question doctrine to only six: clear constitutional authority given to another branch of

74  Supreme Courts and election procedures government, the lack of a clear standard by which a court could intervene, policy decisions that are clearly not appropriate for the courts, court actions that would violate respect owed to the other branches of government, unusual situations that require consistency with previous political decisions, and the potential for confusion if different parts of the same government reached different conclusions on the same issue (369 US 186, 217). On its face, Baker v. Carr’s long list of possible uses of the political question doctrine might have seemed to be a strengthening of the doctrine. However, in the subsequent fifty years of cases, observers have concluded that the political question doctrine is largely dead (Barkow, 2002), or that it is operating in only a few limited areas (Choper, 2005). Curiously, the Court has also manifested the entire continuum of responses from restraint to intervention in its decisions in the areas of malapportionment and gerrymandering. On one extreme, the Court has intervened and regularly intervenes in redistricting to ensure the absolute equality of population of election districts within states. On these issues, the Court is standing in the middle of the political thicket, with court challenges to every redistricting plan expected, even if the legislatures and commissions have learned to anticipate and respond to many of the possible legal challenges that will inevitably be raised. On the other extreme, the Court has stayed steadfastly outside of the political thicket in the area of political gerrymandering. Though a majority of the Court has never accepted that the political question doctrine puts partisan gerrymandering questions outside of the Court’s purview, a majority has consistently failed to find any clear judicial standard that would make it possible for the Court to rule in issues of partisan gerrymandering. This lack of a clear standard in the political gerrymandering cases contrasts with the clear standard of one-person, one-vote, the basis of court interventions in redistricting for malapportionment reasons. In between these two extremes are the Court’s decisions about gerrymanders done for racial or ethnic balancing rather than for political reasons. The Court has recognized these gerrymanders as required under the prevailing interpretation of the Voting Rights Act, but the Court has also constrained its own intervention by setting outer limits for political action (no retrogression in the number of minority districts and districts cannot be so convoluted as to appear that race was the predominant consideration in the drawing of district boundaries). Justice Stevens rightly notes the oddity that the Court finds no clear standard that would allow court intervention in cases of political gerrymandering while using a standard that allows intervention in cases of racial gerrymandering (Vieth v. Jubelirer 541 US 267, 2004). Not surprisingly, Justice Stevens suggests that the Court intervene in political gerrymandering cases using the same standards that it uses in cases of racial gerrymandering. Indeed, it is odd that the Court has created three different standards for restraint or intervention in the politicized process of drawing the lines of election districts in the United States: absolute equality of population (malapportionment); maximum possible number of majority minority districts, without

Supreme Courts and election procedures  75 resorting to extreme measures (racial gerrymandering); and no constraints on political gerrymandering. Zelden (2010, 80) summarizes this diversity of outcomes of the Court’s case law on redistricting issues as follows: Shifting between engaged to disengaged, the Court has struggled to find a coherent response to these challenging questions. In the process, its answers have run the gamut from practical to ideologically extreme and everything in between. A majority of the US Supreme Court supports and justifies this diversity of outcomes because of concerns about the appropriateness of judicial intervention into the politicized world of gerrymandered election districts. The Court’s efforts, however, would have better internal logic and be practically more effective if the Court were to simply adopt maximum standards within which the political process could operate, similar to the standard described in the previous chapter. A first concern, though, of such an approach is identifying the actual harm that occurs in a redistricting dispute. In a malapportioned district, the harm is easy to identify: voters in one district have two or three times the electoral clout of voters in a different district. Both the Japanese and US Supreme Courts have identified this harm and used it to justify setting population equality requirements in the redistricting process. In a political gerrymander, however, the same clarity as to the harm is missing. It is true that a minority of voters can actually elect more than half the members of a legislative body. Indeed, these were the facts brought to the Court in its first direct case on political gerrymandering, Davis v. Bandemer (178 US 109, 1986). Indiana Republicans gerrymandered boundaries for state legislative districts so that in 1982 Democrats won only 43 percent of the seats in the House or Representatives despite having won 53.1 percent of the vote. But in the exact same election, Democrats also won 52 percent of the seats up for election in the state Senate with 53.1 percent of the total vote, casting doubt on the conclusion that a gerrymander will consistently produce biased results. Another prominent gerrymandering case, Vieth v. Jubelirer (541 US 267, 2004) shows this same problem. At issue in Vieth was a Republican gerrymander of Pennsylvania’s congressional districts that in 2002 allowed the Republicans to win 63.2 percent of Pennsylvania’s congressional delegation with only 58.0 percent of the statewide, two-party vote. However in the 2006 and 2008 elections, the districts actually worked against the Republicans. In both elections the Republicans won only 37 percent of the seats, despite winning about 44 percent of the statewide, two-party vote.1 Issacharoff (2002, 604) also documents an example of gerrymandering in North Carolina in which a court found election districts biased against one political party only to have the party that was the victim of the gerrymandered districts win the subsequent election. In addition to identifying the harm of a partisan gerrymander, courts must also link that harm to a constitutional right that is violated by the gerrymander.

76  Supreme Courts and election procedures A Supreme Court cannot supplant a law passed by a legislature just because the law had a partisan motivation or runs contrary to conventional notions of good governance. Rather, laws must have a fatal constitutional flaw before they can be overturned. In the United States, the obvious constitutional argument is that partisan gerrymanders violate the guarantee of equality of all citizens in the 14th Amendment because privileging the supporters of one party over other voters creates unequal treatment of the voters. A related argument is made using the 1st Amendment, that voters cannot be disadvantaged because they are associated with a particular political party. In racial gerrymandering cases the US Supreme Court has used equality concerns to strike down electoral arrangements that diluted the influence of minority voters. Thus, election districts cannot be drawn in such a way as to spread a hypothetical 40 percent of the electorate that is African American among districts equally so that ­African American voters are a minority in all of the election districts (and unlikely to elect any representatives if voting is racially polarized). The analogy to political gerrymandering is obvious. If Republicans in Indiana give their supporters a majority in most districts, creating only a few heavily Democratic districts, then this strategy seems the same as an unconstitutional dilution of the votes of African Americans. However, the crucial difference is that unlike racial discrimination, political discrimination is at the heart of all aspects of the political process in the United States. Should the equality clause of the 14th Amendment or the associational rights of the 1st Amendment also prevent one party from passing legislation that disproportionately benefits its likely supporters? Should legislative rules that give parliamentary advantages to the majority party over the minority be party be struck down in the name of equality? Should presidents be required to appoint office holders from political parties in equal proportion to the votes cast in the most recent presidential election? In addition, unlike victims of racial discrimination, political parties are typically the victims of political gerrymandering in one state but are the perpetrators of a gerrymander in a different state. Political discrimination is at the heart of the political process and the operations of both the legislative and executive branches. This fact makes it difficult for a court to constitutionally justify finding a harm to a political party if it is disadvantaged in an election when similar advantages and disadvantages exist throughout the political process. Proponents of a stronger judicial response to gerrymandering in the United States have identified an additional harm of gerrymandered districts, though this harm has an even weaker link to constitutionally protected rights. They argue that partisan gerrymanders create uncompetitive districts, making elections uncompetitive (Carson and Crespin, 2004; Issacharoff, 2002; Issacharoff and Karlan, 2004; Jacobson, 2004, 8–11; Lublin and McDonald, 2006; ­Pildes, 2004). In a common form of partisan gerrymandering, ruling parties pack their opponents into only a few districts giving themselves majorities in the rest of the districts. This process creates an intended winner in all or nearly all of the districts in the state. In bipartisan gerrymanders, the two parties collude to

Supreme Courts and election procedures  77 make a maximum number of districts safe for the incumbents of both parties. Under both results, the voters are no longer choosing their representatives. The representatives are choosing the voters in such a way as to predetermine the outcomes of elections. Similar to discrepancies between the seats and votes that a party wins, declining competitiveness of elections is also an undesirable outcome and something that seems appropriate for a court to prevent. The analysis, however, stops at this point because of the difficulty in answering the following question: what constitutionally protected interest is harmed if election districts are uncompetitive? Additionally, what constitutional guarantee is thwarted by having the average margin of victory in congressional districts increase from 53 percent to 58 percent? If such a constitutional guarantee to have competitive election districts is found, then are legislatures required to gerrymander district boundaries in order to create competitive districts? Do the largely Democratic voters of Philadelphia have to be divided among each of Pennsylvania’s 18 congressional districts in order to make each and every one of Pennsylvania’s congressional districts equally competitive? What should be done in heavily Democratic Massachusetts or heavily Republican Utah where no drawing of district lines can make all or even a majority of election districts competitive? Additionally, a possible future plunge of the US Supreme Court into the political gerrymandering thicket is hampered by the legal obligation of finding a practical and easy to administer standard by which to guide the Court’s intervention, a requirement that does not exist in potential political solutions to the same problem (Berman, 2005; Ely, 1998, 609; Kasper, 2007, 409). A comparison of seats and votes won by a political party would seem to be an acceptable standard, similar to the one-person, one vote standard used in malapportionment cases. A seats/votes disparity standard, though intuitive, has two substantial flaws: (1) it isn’t known until after an election has occurred and (2) a wide seats/votes differential is also caused by single-member districts and regional concentrations of supporters, two causes that can be completely unrelated to gerrymandering. In addition, a strictly enforced seats/votes measure would in essence require the use of some form of proportional representation in US elections. It would indeed be odd to argue that guarantees of equality in the US Constitution require that a version of proportional representation be used in US elections, especially given the other political effects of proportional representation: coalition governments and unclear accountability of politicians, political parties, and governments. Perhaps, when all factors are considered, a proportional representation electoral system is best for the United States, but there is little justification for the Supreme Court to make this decision rather than the elected representatives of the people. Commentators have tried to solve this problem by proposing standards that  could be used to evaluate partisan gerrymandering, but these other standards also fail because they are either too complex, they rely on data gathered after a redistricting is put into effect, or they are too far removed

78  Supreme Courts and election procedures from a constitutional basis for the standard. The most prominent of these standards is what its proponents have called the efficiency gap, which is simply a measure of wasted votes in an election.2 Because every election has wasted votes (all votes cast for losing candidates are wasted, and every vote cast for a winning candidate over and above the minimum number of votes needed for victory are also wasted), proponents have to pick a cutoff point of what ratio of wasted votes between competing parties is too much. A 7 ­percent difference has been suggested as an appropriate cutoff point. This measure, popular though it might be, suffers from several of the problems that afflict all proposed measures of partisan gerrymandering. First, it relies on data that changes from election to election and is not known before an election occurs. Second, it is hard to argue that an efficiency gap difference of 8 percent is unconstitutional but a gap of 6 percent is constitutional. Thus, political parties could gerrymander a certain amount, but not too much: a conclusion that is hard to square with relevant constitutional language about equality, due process, or speech and association rights. The measure, does, however, benefit from simplicity. Other possible measures fare no better. For example, Niemi (1990) suggests that the swing ratio is an appropriate measure, and indeed this measure helps sort out the usual advantage that large parties earn in single members districts from a gerrymandered outcome that gives an even larger bonus to one political party. The formula, however, is not intuitive and it requires implementation of a redistricting plan before accurate calculations can be made. McDonald and Engstrom propose a measure that is even more complex (1990). Backstrom, Robins, and Eller (1990) solve the timing problem of evaluations by suggesting that the results from a recent, low-profile, partisan race be used as a baseline from which predicted effects of a redistricting plan can be calculated. They then argue that an impermissible gerrymander occurs when a party with less than a majority of votes wins a majority of seats. Their standard is intuitive and simple, but it suffers from the potential flaw of actual election results defying the predictions made based on the baseline election used to measure partisanship. It is one thing to argue that a redistricting plan will likely result in a minority party winning a majority of seats, but if the predicted outcome never actually occurs in an election, the standard will quickly be condemned and discarded. Others try to solve the standard problem by including multiple factors in what it typically called a ‘totality of the circumstances’ test (Baker, 1990). Justices Souter and Ginsburg advocate a version of this standard in their Vieth dissent. A multi-faceted standard, however, doesn’t solve the various problems of finding a constitutionally supported standard, but by including multiple factors, it becomes easier for a court to supplement or work around a particularly problematic factor. For example, it is easier to ignore some of the problems of using a seats/votes ratio as a standard if the ratio is only one of several factors that are considered before concluding that an impermissible partisan gerrymander has occurred.

Supreme Courts and election procedures  79 Despite these advantages, though, a totality of the circumstances type of standard suffers from a different problem, the inability of courts to balance competing interests well, especially when one set of interests is clothed in the magic robes of constitutional language. The development of the one-person, one-vote standard in the Court’s malapportionment cases illustrates well this tendency of legal analysis. In the first malapportionment decision, the majority opinion did not give a standard by which to judge the equality of districts. Justice Clark, writing in a concurrence went so far as to say ‘No one—except the dissenters advocating the Harlan “adjusted ‘total representation’” ­formula—contends that mathematical equality among voters is required by the Equal Protection Clause’ (369 US 186, 258 (1962)). Justice Stewart agreed with this sentiment in his own concurrence: ‘Contrary to the suggestion of my Brother Harlan, the Court does not say or imply that “state legislatures must be so structured as to reflect with approximate equality the voice of every voter”’ (265). From this starting point, Justice Warren enshrined the one-person, one-vote standard in a subsequent case, Reynolds v. Simms (377 US 533, 1964)), but even Warren was careful to still accept the possibility of other factors having influence on redistricting: A State may legitimately desire to maintain the integrity of various political subdivisions, insofar as possible, and provide for compact districts of contiguous territory in designing a legislative apportionment scheme. Valid considerations may underlie such aims. …  So long as the d ­ ivergences from a strict population standard are based on legitimate considerations incident to the effectuation of a rational state policy, some deviations from the equal population principle are constitutionally permissible. (578–79) This promised flexibility of the standard has been maintained in cases challenging the redistricting for state legislatures, but in congressional redistricting, the Court has instead elevated equality as the only allowed consideration (Fuentes-Rohwer, 2005). In Karcher v. Daggett (462 US 725, 1983) the Court struck down a plan for congressional districts in New Jersey that had a maximum deviation from perfect equality of population of less than 1 percent. The Court nominally held out the possibility that minor deviations could be justified by an unspecified state purpose, but because an alternative plan with smaller deviations is always possible, the practical effect of the decision was to require all congressional districts within a state to be of absolutely equal population. Thus, the Court moved from first denying the use of an absolute equality standard in its first cases to the elevation of that standard above all other possible considerations in later cases because the constitutional basis of the equality standard necessarily trumps all other considerations that have no basis in the constitution. The contrast of the legislative process to the judicial process is quite striking. Unlike a legislature, a court can neither give equal weight nor even consider other factors if those factors are not constitutionally based and they are

80  Supreme Courts and election procedures competing against a constitutionally based factor. Legislatures, however, may produce redistricting plans that are politically motivated and perhaps defy common sense, but part of the reason that they defy common sense is that they are based on a variety of concerns, concerns that vary from district to district and from plan to plan. This obvious weakness of the political process is also its strength. A variety of interests and stakeholders all have input into the process, and though some voices are excluded, it is at least possible to argue for other factors to be included in the redistricting process, factors that would not be considered in a judicially mandated review (Persily, 2002, 677–9). This inability to compromise in balancing a variety of constitutional and non-constitutional factors is precisely what has gone wrong in the US Supreme Court’s malapportionment jurisprudence. At least for congressional districts, boundaries based on communities of interest, contiguity, compactness, or local administrative boundaries are not allowed if boundaries based on these considerations deviate even minutely from the standard of absolute population equality. The rigid requirement of population equality also forces politicians to ignore these other considerations, at least as long as these other factors conflict with population equality. Furthermore, legislators, being politically motivated, are more than happy to jump on the bandwagon of ignoring all of these other factors. They know that their work will not be evaluated nor will it be criticized for ignoring these other factors. Rather, the courts will examine only population equality, creating the perverse outcome of the Court falling victim to its own restrictive logic. In some situations courts have used tiny population differentials to justify striking down an apportionment plan that the Court objects to for other reasons (Issacharoff and Karlan, 2004, 569). The judges, however, can’t voice their other concerns because of a long line of court precedent decreeing that the only relevant standard for judicial review is the absolute population equality of districts. It is easy to imagine the Supreme Court ending up in a similar logical dead end if it were to intervene in the realm of political gerrymandering. Perhaps an initial decision would have a multi-faceted standard that included a seats/votes differential as just one factor of several factors used in deciding whether an unconstitutional gerrymander occurred. Over time, however, it is likely that the seats/votes differential would become the only factor that a court considered as it evaluated redistricting plans for their constitutionality. Perhaps this fear is what restrains the bare five-vote majority of justices from plunging ahead into the issue of partisan gerrymandering as its predecessors did in the 1960s for the issue of malapportionment. The net result, though, is a tortured jurisprudence on redistricting that satisfies no one. The paring back of the political doctrine question to allow the interventions in Baker v. Carr has led some to question the wisdom of the Court’s intervening in other election cases, most prominently the 2000 presidential election (Barkow, 2002). In contrast the logic that a majority of the Court’s justices have used to not intervene in political gerrymandering cases has been greeted by a chorus of criticism, with many suggesting the missing constitutional basis for such an intervention

Supreme Courts and election procedures  81 or the standards by which the Court could do such an ­intervention (Berman, 2005; Calidas, 2008; Fuentes-Rohwer, 2005; Issacharoff, 2002; Pildes, 2004). What is missing from this discussion is a healthy understanding of how a similarly situated country, Japan, has effectively solved both the malapportionment and gerrymandering problems. Japan combined a highly restrained judicial intervention with a political process that aligned both the self-­interest of legislators with constitutionally mandated redistricting standards. This combination allowed for a variety of factors to be considered in redistricting rather than limiting the process to only the factors that are based on constitutional language.

Redistricting problems in a comparative perspective On redistricting issues, the United States strikingly stands out from other similarly situated countries. While Japan, Canada, Britain, and Australia have all moved in the direction of non-partisan or constrained redistricting practices, the United States seems alone among advanced industrial democracies in its outdated adherence to partisan redistricting methods. The outlier status of the United States is also surprising given the early and frequent intervention into the redistricting process by US courts. It is ironic that the nation with the most vigorous court interventions in redistricting is also the nation with the most politicized and problematic redistricting process. Pildes (2004, 79–80) cites three unique characteristics of US democracy in explaining the longevity of partisan redistricting in the United States. (1) The Founders wrote the US Constitution at a time when there was less awareness of potential problems of the oversight of the democratic process. Thus, important guarantees of the integrity of this process were not included in the Constitution. (2) Many potential tools of oversight of the democratic process have a history of being politicized in the United States, for example the election of state judges. (3) Americans generally distrust claims of impartiality by ostensibly non-partisan or independent organizations. It is difficult to argue with any of Pildes’ explanations. They are factually correct. The characteristics of the United States that Pildes identifies help explain why innovations in redistricting have taken root in recent decades in some countries while falling on barren soil in the United States (with some state reforms being exceptions.) There has been a clamor for reapportionment reform in the United States as well as in Japan, Canada, Australia, or Great Britain. In most of these countries the politicians themselves have been the ones who have enacted reforms or reform efforts. Thus, Pildes’ explanations are plausible: state legislators in the United States inherently distrust non-­ partisan institutions and are not constrained by constitutional language or legitimately non-partisan oversight organizations. The Japanese experience with reapportionment reform, however, suggests an important factor that is missing from Pildes’ analysis, specifically that a non-partisan redistricting practice is much more likely to take hold in

82  Supreme Courts and election procedures party-centered rather than candidate-centered election systems. Indeed the two advanced industrial democracies most prominent for their resistance to redistricting reforms have been Japan and the United States, with both countries only undertaking significant reforms when forced to do so by their respective Supreme Courts. It is likely not a coincidence that both countries have the most candidate-centered electoral systems of any of the advanced industrial democracies. This resistance is explained by the great importance of district boundaries in a candidate-centered election system. In such electoral systems, candidates win elections, in part, because of their personal connections to the voters of their districts. Though party identification and support matter and determine the choices of many voters, in a candidate-centered system, a relatively large group of voters base their vote on attributes of the candidates rather than the political affiliation of the candidate. Because of these election imperatives, individual MPs in candidate-centered electoral systems care intensely about their own district boundaries. They are likely to resist any efforts to change those boundaries and in many cases would even oppose a redistricting plan that improved the partisan makeup of their district if that advantage was created by adding significant numbers of new voters to the district (Persily, 2002, 662–3). The development of redistricting processes in both Japan and the United States illustrates well the priorities of politicians in candidate-centered election systems. Prior to Baker v. Carr, most states had state legislative districts that were dramatically biased to rural areas. A reapportionment of seats to more urban areas would have advantaged at least one of the major political parties, yet regardless of which party might gain by redistricting, redistricting did not occur. The sitting legislators were more concerned about protecting their own election districts than creating a partisan advantage for their political party. In fact, the few instances of redistricting that occurred prior to the Supreme Court’s intervention were because of the intervention of state courts in Massachusetts or a political response to a failed court intervention in Illinois (Ansolabehere and Snyder, 2008, 97, 109). Similarly, in Japan, the ruling LDP tried on multiple occasions to change the electoral system to a system of single-member districts that would have had strong partisan advantages for the LDP. Despite these partisan advantages, individual LDP legislators opposed the plans because they would change district boundaries. The members of the LDP whose votes were necessary to pass the several electoral reform proposals preferred to sacrifice an obvious advantage to their own party in exchange for knowing that their own election districts would remain untouched. This priority of the electoral fortunes of individual members of the LDP shows up also in analyzing the four limited reapportionments done under the previous multi-seat district electoral system. Each of these reapportionments was done in a way that minimized the disruption of election seat boundaries. For example, only 13 of Japan’s original 119 election districts (11 percent) had a significant boundary change from 1947 to 1993. Eleven of the districts

Supreme Courts and election procedures  83 were split, creating a new district, and in only one instance were two districts merged.3 Of the total 130 multi-seat districts used from the 1947 to the 1993 elections, 78 districts (60 percent) had no changes, and an additional 28 districts (22 percent) had changes in the number of representatives elected without altering district boundaries.4 LDP incumbents have also fared well under most of Japan’s reapportionments. In the 1964 and 1975 reapportionments, which only added seats to the Diet, the LDP actually increased the percentage of seats that it won in the reapportioned districts, going from winning 37 percent of the seats in the districts before reapportionment to winning 49 percent of the seats in the newly reapportioned districts. More significantly, 28 of the 31 conservative incumbents who ran for reelection in the reapportioned districts won reelection. The numbers for the later two reapportionments (1986 and 1992) when some districts lost seats were worse for party interests, but not for individual politicians. The Party won 54 percent of seats in districts before reapportionment but won only 38 percent of seats after reapportionment. However, the 64 conservative incumbents who ran for reelection in the reapportioned districts fared much better, with 58 of them winning reelection.

The Japanese solution to malapportionment and gerrymandering The Japanese experience suggests a two-step solution to solving both the problems of potential malapportionment and potential gerrymandering in ­candidate-centered election systems with single-member districts. The first step is for the courts to show more restraint in dictating the outcome of an inherently political process, redistricting. In order to preserve some of the benefits of allowing the political process to operate, the Court should set only the maximum boundaries within which the political process may operate, a realistic standard of equality between districts that understands that a certain level of inequality is unavoidable, but also allows the political process to operate as long as its results don’t exceed this reasonable boundary. The Japanese Supreme Court has recently moved to a 1:2 maximum standard, and a similar maximum standard is appropriate for the United States. This maximum standard would apply to all districts in the United States, and it would not be much different than the unavoidable difference in district sizes that exists now because of different state populations. US courts could continue to require a stricter standard for district equality within a state, perhaps a maximum 10 percent deviation from the average district size for that state. If the judiciary tolerated slightly higher levels of inequality, it would become possible for legislators to consider other redistricting factors, in addition to population equality. The current legal regime in the United States makes it impossible for legislators to consider other factors if those factors reduce population equality. Typically these other factors (contiguity, communities of interest, local boundaries, compactness) all reduce population equality. In contrast, political factors such as partisanship or incumbent preservation can all be served without in any way affecting population equality. In fact,

84  Supreme Courts and election procedures the requirement of population equality (and its legal dominance over any other redistricting factors) makes it easier to justify contorted boundaries that serve both the requirement of absolute population equality and also partisan interests. Of course, simply pulling back to a maximum standard of tolerated population deviation will not transform line drawers in the United States into paragons of redistricting virtue. Legislators are used to manipulating boundaries for partisan benefit, so allowing greater leeway for small population inequalities would likely just give partisans opportunities to gerrymander unequal-sized districts rather than gerrymandering equal-sized districts. If only the first half of the Japanese solution is applied by itself, it would likely worsen a bad redistricting situation in the United States without creating any benefit. The second part of the Japanese solution is an additional requirement of a second standard in redistricting that removes the possibility of g­ errymandering. This standard is the prioritizing of local boundaries as a limiting mechanism in the redistricting process. Legislators would be free to come up with any boundaries that worked as long as those boundaries (1) were within the maximum range of allowed district populations, and (2) the number of local units divided were the minimum possible, unless an additional division was justified by a non-political concern, for example creating a majority African American district or maximizing the number of competitive districts in the state. In practice, as shown in Chapter 4, such requirements would reduce the available options to only one, in many states, or to only a small number of relatively similar options in states with more complicated boundaries. As discussed in Chapter 4, the requirement, to truly constrain the impulse to gerrymander, would need to include a variety of local administrative, not just city and county boundaries. Groupings of counties, groupings of cities within populous counties, and groupings of neighborhoods within populous cities would all need to be included in the list of local administrative boundaries that would be respected in redistricting plans. This priority could be created by court action, but the problem remains of finding a constitutional basis for requiring that local boundaries be followed. Perhaps this requirement could be transformed into a constitutional argument treating local boundaries as surrogate measures for non-political redistricting concerns. As such, a fairness or equality argument could be based on making sure that redistricting is insulated from politics, treating all partisans equally. The Japanese example, however, shows how this same outcome can be reached without judicial intervention. Three factors came together to make local boundaries so important in Japanese redistricting. First, the Japanese bureaucracy has its own independent power and reputation that give its preference for minimizing the divisions of boundaries of administrative units significant weight in the redistricting process. Second, any deviations from this bureaucratic standard were noted immediately by the media and pundits; each new redistricting proposal was greeted by media commentary on how many additional local boundaries were split and what were the justifications for those splits. Third, the politicians

Supreme Courts and election procedures  85 themselves had little incentive to oppose the use of local boundaries for redistricting. Though it might have been in a party’s interests to carve up a city or a county four different ways in order to maximize partisan benefit, Japanese MPs preferred, in contrast, that cities, neighborhoods, and counties remained intact. The partisan benefits of a gerrymandered boundary were outweighed by the desire of MPs to keep their election districts the same, with the same voters that the MPs knew well. Acquiescing to the bureaucratic desire to keep local administrative units whole meant that MPs gave up a possible partisan gerrymander that might have benefited their party in exchange for the assurance that they would never be the victim of a partisan gerrymander. A similar outcome in the United States is hampered by the weakness of bureaucratic actors in the United States. Nevertheless, politicians and pundits could start talking about redistricting plans in terms of how many local units are split and the justifications for those splits. US legislators could be enticed to embrace reforms that enshrined the minimization of split units in redistricting proposals by appeals to their self-interest. Legislators could be guaranteed that such a reform would protect them from a potential adverse redistricting. In this way, selfish incentives of politicians can be turned in a way that produces a better redistricting outcome. This is one of the advantages of allowing the political process to operate within reasonable boundaries in solving political questions. No longer is the Court legislating to try and force politicians to stop behaving like politicians, rather the political process allows politicians to act in a self-interested way that will produce the equivalent of a non-partisan outcome by using boundary rules that align the self-interest of politicians with an outcome that is best for the entire nation.

Problems with and alternatives to the Japanese solution One obvious difference between Japan and the United States is that in Japan redistricting rules and proposals are passed by the very MPs whose districts are affected by those changes, but in the United States, congressional boundaries are often decided by a different set of legislators (the members of a state legislature). Thus, these state legislators are less likely to be influenced by the selfish desire of representatives to ensure that their districts aren’t changed; after all, the districts under consideration are somebody else’s district. The political effect of this difference, though, can be mitigated in several ways. First, in a redistricting law passed in a state legislature, the rules can and should be written to govern both districts for the state legislature (where self-interest prevails) and seats for Congress. Second, the link between state legislators and members of Congress is strong, and many state legislators aspire to positions in Congress and could be similarly motivated to draw districts that reflect the concerns of individual candidates rather than partisan advantage. A second potential problem arises from the US practice of creating ­majority-minority districts to ensure representation of African Americans and other minority groups in the US Congress. These concerns can easily

86  Supreme Courts and election procedures be accommodated by explicitly allowing a few additional divisions of local units in order to facilitate the creation of a majority-minority district. The US Supreme Court has also signaled its unease with the use of extreme measures to create such majority-minority districts. The counting and justification of local unit divisions would create a transparent and easy to use standard for judging when efforts to create a majority-minority district have become extreme. Thus, for example, if a redistricting plan is challenged for having too few minority districts, it would be easy for a court to rule that an additional minority district that only required two additional divisions of local units should have been created while an additional minority district that required the division of eight local units would have been unacceptable. There are, in addition, many other ways that the courts, legislators, or voters could eliminate the problem of partisan gerrymandering. The most prominent of these alternative solutions is the non-partisan redistricting commission. Indeed this has been the solution of choice in most comparable countries. Even Japan uses a boundary commission to draw the actual boundaries after the politicians have determined the allocation of seats to provinces. Japanese politicians, however, still retain the authority to approve, augment, or ignore the proposals produced by the boundary commission. Some US states have begun the use of boundary commissions, and even more states end up with the equivalent of a boundary commission when court-mandated redistricting plans are used. Boundary commissions are an excellent alternative solution to the problem of partisan gerrymandering. The major limitation on their usefulness is the reluctance of legislators to embrace them as a solution to problems of gerrymandering. In this way, the Japanese solution is perhaps superior, as its appeal to the self-interest of legislators increases its prospects for adoption. As politicians see their districts will be protected (even from a gerrymander by a boundary commission) they will be more likely to enact the Japanese solution, even without the requirement of action by judicial intervention.

Other explanations of Japanese court interventions The Japanese Supreme Court’s measured response to redistricting problems is a model for US courts to follow. This conclusion, though, is hardly conventional wisdom, and it is to be expected that some observers of the Japanese Supreme Court have reached different conclusions. One group complains that the malapportionment problem persists in Japan because the Court has not intervened enough. A second group argues that the Court was doing the bidding of LDP leaders in its gradual movement towards enforcing increasingly stringent malapportionment guidelines. The timing of the court cases and the political response of MPs suggest that both interpretations are wrong. The Court has significantly constrained levels of Japanese malapportionment, though the process has been gradual, and there is no evidence that LDP leaders actually favored the Court’s interventions that constrained malapportionment.

Supreme Courts and election procedures  87 The first criticism of the Japanese Supreme Court’s restrained approach to redistricting cases ignores two important facts. First, that the US Court is by far and away the most active Supreme Court in the world (Dahl, 1989, 188–9; Lijphart, 1984, 194). Second, the US Court has come to this position of prominence only gradually, developing its own robust doctrines of judicial review over centuries. For example, though judicial review was a power exercised by colonial courts and by the Supreme Court from its inception, the Court waited 54 years after Marbury v. Madison (1803) until its decision in Scott v. Sanders (1857) to actually strike down as unconstitutional a law passed by Congress5 (Bolz, 1980, 133–4; Mavcic, 2001, 19). Indeed, the Supreme Court’s record on matters of grave national importance has more often been one of restraint rather than political leadership, enshrining in constitutionality the fugitive slave act, segregation, the imprisonment of Communists, and the internment of the Japanese (Dahl, 1989, 190–1). The Court’s greatest activism, prior to the Warren Court, was in the economic realm, striking down some progressive and New Deal-era economic reforms. Despite this slow start, the US Supreme Court has become the most likely tool for advancing political goals in a variety of cutting-edge political debates from a woman’s right to choose, to policies dealing with suspected terrorists, campaign spending, school admissions, immigration policy, and gay rights. In these and other areas, passage of legislation is but the first step in a protracted political battle that will not end until a court decision has been handed down, and even then the political battle may continue in an effort to reshape the Court in order to get a different decision. In contrast, Japan stands on the opposite extreme: countries with weak systems of judicial review. There are many countries in which courts still lack any power of judicial review; however, both the number of countries with judicial review and the extent to which courts exercise their judicial review powers have been increasing. Japan was part of the large number of countries that enacted judicial review after World War II (Henderson, 1968; Mavcic, 2001, 21). The innovation in Japan came at the insistence of the US-led, postwar occupation (Danelski, 1992; Higuchi, 2001; Otake, 2001; Tomatsu, 2001). Some countries such as France and Canada, which gained powers of judicial review after Japan, have plunged ahead with active courts that regularly rule on the constitutionality of laws (Law, 2011). The courts in Germany and the United States have each struck down approximately 600 and 900 laws respectively in recent decades (Law, 2011). In contrast, Scandinavian countries have had few instances of judicial review (Lijphart, 1984, 192–4). Similarly, Japan’s Supreme Court has struck down only eight laws on constitutional grounds in 60 years (Law, 2009, 1547, 2011; Matsui, 2011; Satoh, 2008).6 Two of these eight cases come from the Court’s malapportionment decisions. The Japanese Supreme Court has rejected many claims of an unconstitutional law by adhering to a robust definition of political questions that the Court is prohibited from ruling on. For example, the Court has consistently refused to rule on whether the US–Japan Security Treaty violates Article 9 of

88  Supreme Courts and election procedures its Constitution which states that Japanese ‘land, sea, and air forces, as well as other war potential, will never be maintained’ (Bolz, 1980; Tomatsu, 2001; Yokota, 1968). The Court used this same doctrine in its first malapportionment decision in which it refused to act, but the Court left open the door for later action by allowing for the possibility of court review if the malapportionment was egregious (Yokota, 1968, 155). The Court walked through this door in a subsequent 1976 decision that found that egregious malapportionment did violate a constitutionally protected right to the equality of votes. Thus, in context, Japan’s Supreme Court has been timid, not only in comparison to trends in US Supreme Court jurisprudence, but also to trends in most European countries where, similar to Japan, judicial review does not have a long history. Unlike Japan, however, judicial review in these countries has developed quickly into an important part of the political process. The German Supreme Court is taking an increasingly political role in evaluating legislation, a stance seen most prominently in the Court’s rulings on Euro area bailouts and whether the proposed plans meet constitutional requirements. Even more telling is the prominent role that the European Court of Human Rights has played in forcing European countries to rewrite legislation to meet standards of fundamental human rights as enshrined in the European Charter of Human Rights. Even European countries that still lack formal judicial review now have a form of judicial review, at least in the area of fundamental human rights, as interpreted by the European Court of Human Rights. Against this backdrop, skepticism and criticism of the Japanese Supreme Court’s malapportionment decisions is perhaps understandable and to be expected. The Court has rarely exercised its powers of judicial review, and its efforts to force the legislature to reapportion districts have been met with delay and only the most minimal possible action to satisfy the Court’s demands.7 Law (2011, 1452) argues, ‘The Court has insisted for decades that electoral malapportionment has reached unconstitutional levels, yet the LDP has yet to introduce an apportionment scheme that satisfies the standards set forth by the Court.’ Koshiji (2001) and Bailey (1997) both recognize that the legislature has responded in recent apportionments to prodding by the Supreme Court, but they criticize the Court for failing to adopt the 1:2 standard, which the Court eventually did in its 2011 decision. Matsui (1986, 43) also explains the limitations of the approach used by the Japanese courts to evaluate the constitutionality of an apportionment: requiring that an election be held under the apportionment before it can be evaluated as constitutional. If an election has already been held under the apportionment plan, then the Court faces the dilemma of invalidating the election that created the legislature that must pass the required reapportionment legislation. This difficulty has resulted in the oddity of Supreme Court decisions that hold that the redistricting is in an ‘unconstitutional state’ rather than hold that the redistricting is unconstitutional. This method also delays any response of the Court, requiring first a census and an election before the Court can find the apportionment to be unconstitutional (Bailey, 1997). In addition, even

Supreme Courts and election procedures  89

6 5 4 3

Ratio

2 1

17

14

20

10

20

06

20

02

20

98

20

94

19

90

19

86

19

82

19

78

19

74

19

70

19

66

19

19

19

19

62

0 58

Ratio of the most over- and under-represented districts

when the legislature acts, the new reapportionment is out of date even before the first election is held. For example, the legislative response to the 2011 decision was a law passed in 2013 that reduced the maximum disparity (based on the 2010 census) to 1:1.99, just a whisker under the constitutionally required standard of 1:2. This standard was already exceeded (1:2.13) by the first election, held only one year later, under the new apportionment. All of these criticisms are correct, but perhaps these critics are guilty of refusing to acknowledge the Court’s reasonably strong record on malapportionment because of its almost non-existent record of judicial review in other cases. In fact, a very good case can be made that the Japanese Supreme Court, whatever its possible timidity in other areas, has been the primary cause of improvement in malapportionment in Japan. Figure 5.1 shows the ratio of the two worst election districts in Japan on the day of each lower house election since 1958. In addition, the Diet has reapportioned seats fairly regularly, in 1964, 1975, 1986, 1992, 1994, 2002, 2013, and 2017. Though the first two apportionments preceded any court decisions that questioned the constitutionality of malapportionment (Koshiji, 2001), it does seem that the political branch has been responsive to court directions regarding malapportionment, and despite Japan’s poor record on the issue, reapportionments, at least limited reapportionments, have happened regularly as they do in other advanced democracies. My purpose is not to argue that Japan’s Supreme Court should become more activist or that other Supreme Courts should become less activist. It is hard to argue against the defense of constitutional rights, and courts are uniquely prepared to defend those rights against the whims and passions of more ephemeral legislative majorities. It was appropriate for the Japanese Supreme Court to have intervened in the malapportionment dispute because political incentives were such that legislative action to redress the imbalance would have been rare and insufficient. My point, however, is more narrow. I simply praise the manner of the intervention done by Japan’s Supreme Court as being better suited to effectively solving the problem than the more robust intervention

Election year

Figure 5.1 Ratio of the most malapportioned Japanese election districts at election day

90  Supreme Courts and election procedures method done by the US Supreme Court. Perhaps the message of this comparison is that greater judicial intervention is not always the best solution to every problem, especially intensely political problems. The second set of criticisms of the Court comes from the opposite perspective, that the Court is controlled by the ruling LDP which appoints the justices of the Supreme Court (Ramseyer and Rasmusen, 2001).8 These critics of the Court interpret the change in the Supreme Court’s position as being the result of LDP political pressure on the Court. The LDP changed its position from not wanting reapportionment to desiring reapportionment, and this change was communicated to the Court which then went about implementing the change by changing its jurisprudence from deference to the legislative branch on apportionment issues to setting standards within which reapportionment activities must occur. This criticism, though, suffers from several enormous logical and evidentiary flaws. The first is the incongruity of the LDP using the Court to force reapportionment on itself when it had the votes to simply pass legislation reapportioning districts. The only potential response to this obvious argument is that LDP leaders wanted to reorient their party’s appeal away from overrepresented rural districts to underrepresented urban districts. Lacking support from individual LDP MPs to pass such redistricting legislation, party leaders used the Court to force this change on recalcitrant party members. Two parts of this argument are obviously correct: (1) the LDP would be better served by broadening its appeal to include more urban voters and (2) LDP MPs are generally opposed to redistricting because of the negative effect boundary changes have on their reelection prospects.9 These two truths, however, do not, by themselves, support the logic of Ramseyer and Rasmusen’s argument. Improving electoral performance among a certain group of voters is not the same thing as shifting additional political power to that same group of voters. Republicans in the United States might be well served if they enhanced their appeal to urban voters, but that is a clearly different argument than wanting to abolish the Electoral College which would shift political influence from states with fewer urban voters to states with more urban voters. What party would shift political power to the stronghold of their political opponents prior to developing a heightened voter appeal in those strongholds of their opponents? There is no logical reason to shift seats away from LDP-dominated rural areas to opposition-dominated urban areas before the LDP improves its electoral performance in those areas. This timing issue shows up in two additional factual problems with Ramseyer and Rasmusen’s claim. First, the LDP has consistently done limited reapportionments, about every ten years, since the 1960s. It is odd to claim that LDP leaders needed the help of the Supreme Court to pass reapportionment legislation when its first two reapportionments preceded the 1976 decision finding malapportionment unconstitutional. In addition, the LDP took no additional reapportionment action until a decade after the 1976 decision. The LDP pattern of enacting minimal reapportionments at the latest possible date,

Supreme Courts and election procedures  91 both before and after the change in Supreme Court precedent, casts doubt on the idea that a Supreme Court decision was necessary to force reluctant LDP members to vote for reapportionment legislation. Clearly, court decisions added pressure on the LDP to act, along with other public pressure. In addition, as the Supreme Court has reduced the range of tolerable malapportionment, the LDP has been forced to act within a narrower range of tolerated disparity.10 These two truths, however, do not support the claim that LDP leaders had to have a court decision in order to pass reapportionment legislation when the party was quite capable of passing and did pass such legislation without any pressure from a Supreme Court decision. A second timing problem is related to the argument that reapportioning seats to underrepresented areas would somehow help the LDP increase its popularity in those areas. Ramseyer and Rasmusen explicitly make this argument when they say that by 1985 ‘LDP leaders were solidifying the party’s position as an urban party and abandoning the farmers to the socialists and communists’ (2001, 336). Figure 5.2 seems to contradict this claim.11 It reports the vote percentages for all LDP-affiliated candidates in House of Representatives elections from 1958 to 2017. These percents are broken down into three categories of prefectures, those that gained consequential numbers of seats in Japan’s eight reapportionments that occurred from 1964 to 2017 (underrepresented districts), those that lost consequential numbers of seats in the same reapportionments (overrepresented districts), and those that maintained about the same level of representation (middle districts).12 Figure 5.2 does show that by 1976, when the LDP supposedly instructed the Supreme Court to attack malapportionment, the LDP had suffered ten years of considerably lower levels of support in Japan’s underrepresented prefectures (largely urban areas). However, the Supreme Court decisions and the various reapportionments seem to have no effect on this urban weakness of 60

Percent of vote

50 40 30 20 10

17

14

20

12

20

09

20

05

20

03

20

00

20

96

20

93

19

90

19

86

19

83

19

80

19

79

19

76

19

72

19

69

19

67

19

63

19

60

19

19

19

58

0

Election year Under

Middle

Over

Figure 5.2 Conservative votes in prefectures by levels of malapportionment, Japan 1958–2017

92  Supreme Courts and election procedures the LDP until after 1993 when Japan creates single-member districts across the nation. It is possible that by 1985, the LDP had a more urban and less agricultural policy perspective, but this change in policy perspective did not bear fruit in LDP efforts to win a greater share of the urban vote. This increase in the LDP’s urban appeal occurred only after parties consolidated in response to Japan’s new single-member districts, and the smaller parties that were popular in urban areas stopped running candidates or won fewer and fewer votes in those district races. The LDP’s urban performance peaks in the anomalous 2005 election in which Koizumi expels 37 LDP incumbents who voted against his postal reform legislation. The issues surrounding this election were perhaps the best example of the LDP attracting urban voters and sacrificing the interests of traditional LDP support groups in rural areas. Thus, Figure 5.2 shows that electoral reform did help the LDP win more votes in urban areas, but the important reform was the use of single-seat districts, not the gradual shift of seats from overrepresented areas to underrepresented urban areas. A third problem with Ramseyer and Rasmusen’s argument is the actual statements by LDP leaders at the time of crucial Supreme Court malapportionment decisions. If LDP leaders wanted to use the decisions to force reapportionment actions, we might expect them to praise the decisions for showing a necessary path forward for the party. However, their statements stand out as the only statements made against the decisions by any politicians or commentators. For example, the LDP’s Secretary General, Nakasone Yasuhiro, cited the reapportionment done the previous year and said that there was no problem now, despite some malapportionment remaining (Mainichi Shimbun, April 15, 1976, 3; Asahi Shimbun, April 14, 1976, 3). The head of the Ministry of Home Affairs said that he would study the decision and noted that the 1975 reapportionment would reduce inequality between districts in future elections (Mainichi Shimbun, April 14, 1976, 2). Head of the LDP’s elections committee said that there was no need for another reapportionment for four or five years (Mainichi Shimbun, April 15, 1976, 3). Only one LDP MP representing an underrepresented Tokyo district called for immediate reform and criticized the current levels of malapportionment (Asahi Shimbun, April 14, 1976, 8). In contrast, the leaders of all the opposition parties, academic commentators, and the editorial boards of Japan’s main newspapers all called for immediate reapportionment (Asahi Shimbun, April 15, 1976, 5; Yomiuri Shimbun, April 15, 1976, 5; Mainichi Shimbun, April 14, 1976, 2). In addition, various sources pointed out the harm that additional reapportionment would do to LDP political prospects. Both Asahi Shimbun (April 14, 3) and Mainichi Shimbun (April 15, 3) pointed out that the LDP would lose its parliamentary majority if it fairly reapportioned seats. Mainichi Shimbun pointed out that under a fair apportionment, Tokyo would go from 43 to 102 seats, and Osaka would increase its representation from 26 to 70 seats. Mainichi added that in 16 election districts spread across seven prefectures the LDP wins less than 30 percent of the vote. Asahi piles on quoting an LDP publication which cites how lucky the party is that urban votes have less weight

Supreme Courts and election procedures  93 than rural votes. Shinohara Hajime, a professor at Tokyo University, adds that the recently completed 1975 reapportionment perhaps made it easier for the Court to issue its unprecedented decision. Justices knew that there would be no expectation of government action in response to the Court’s decision (Mainichi Shimbun, April 15, 1976, 3). Itoh similarly indirectly quotes Justice Ito of the Supreme Court as saying the Supreme Court was lucky not to have irritated and upset incumbent LDP politicians when it declared the 1972 general election for the House of Representatives unlawful although it had not disqualified elected candidates in the Chiba malapportionment case (1976), in which the inequality of voting value had reached as much as 5 to 1. (2010, 213) It is truly odd (for Ramseyer and Rasmusen’s argument) that a member of the Supreme Court that was supposedly doing the bidding of LDP leaders in finding malapportionment unconstitutional in 1976 was worried about LDP criticism and the LDP reaction to the decision. Of course, it is always possible that LDP leaders were playing a double game, saying to LDP MPs that they were unhappy to be forced to do reapportionments while pressuring the Supreme Court to force those same reapportionments. The problem with this argument, beyond the lack of any evidence to support the argument, is that LDP leaders didn’t even respond to the decision by saying that they would even consider doing another reapportionment. The immediate response of LDP leadership was that no further reapportionments were needed, and that is indeed the path that the LDP followed, waiting more than a decade before enacting the next reapportionment. Finally, it can be argued that Ramseyer and Rasmusen (2001) have produced data that supports their explanation: judges in Japan’s judicial system who ruled contrary to the supposed LDP preferences were disadvantaged in their subsequent careers in contrast to judges who ruled in line with these supposed LDP preferences. This argument, however, ignores a much more likely and obvious explanation. Prior to 1976, Supreme Court precedent held that malapportionment was constitutional. After 1976, new precedent of the Court was that high levels of malapportionment were unconstitutional. It is not surprising that judges who followed the Supreme Court’s precedents benefited and those who made decisions contradictory to the Supreme Court’s precedents were harmed in their subsequent careers. Ramseyer and Rasmusen (2001, 339) understand this logic as they also include in their analysis whether or not a court decision about malapportionment was overturned. By including both decisions overturned and the supposed LDP preferences, their analysis implies that these two different explanations of court decisions are actually similar. It is a mistake, though, to assume that lower court judges following the precedent of their Supreme Court is somehow proof that political pressure on the Court from the LDP caused the precedent to change.

94  Supreme Courts and election procedures

Conclusion Japan’s Supreme Court has laid out a path to constrain the worst excesses of partisan election rules while retaining an appropriate ability to work out political compromises in crafting a solution that fits within a broad, judicial framework of constitutionality. Though most disputes about fundamental rights should probably be handled by courts, there is room to consider a less rights-based and less constraining approach by courts, especially in areas where political compromise is necessary. Unfortunately a court’s rights-based analysis often makes it impossible to craft a political compromise. Conservatives have long made this point in contentious areas in which cultural attitudes are changing. The United States has one of the most liberal legal frameworks for abortion because the rights-based analysis of the US Supreme Court has made it possible for states, if they want to, to allow abortion up until the moment of birth, something that is allowed in only six nations other than the United States. This outcome results because a woman’s right to choose is enshrined as a constitutional right, and the right to life of a fetus is not recognized by the Court. Instead, the Court has allowed states, after fetal viability, to protect a vaguely defined right of the state to preserve life. Thus, states that choose not to protect life may allow abortions up until the moment of birth. A compromise position that would protect the right to choose but also protect the life of a viable fetus, however, is not possible under this analysis because the US Supreme Court has refused to recognize any rights of a fetus. Thus, the relatively extreme form of abortion law in some US states is dictated by the fact that this issue operates almost entirely within the rights-based analysis put forth by the US Supreme Court. Cain (2015, 167) remarks extensively on this same type of problem in the area of campaign finance law. The Supreme Court early on put this issue within the framework of free speech rights, a framework which severely constrains state regulation of campaign funding because such funding is a form of political speech. This analytical framework means that governments can regulate campaign contributions and spending only in ways that directly prevent quid-­proquo corruption. Regulating against money for access or influence or regulations that equalize the voice of different citizens is not allowed by the Constitution. Just as with abortion law, a political compromise that respected free speech rights while also allowing greater regulation of campaign financing would likely be possible if politicians were allowed to craft such a compromise, but such compromises are largely disallowed by the strict free speech rights that the Court has directed must be respected. It is difficult, though, to identify which issues are fundamental rights that deserve clear court protection and which issues should be part of the messy compromises that are part of the political process. Certainly a woman’s right to control her own body and free speech rights in political campaign are both fundamental rights. However, both of these issues have been identified as areas in which a political compromise might be more suitable than a restrictive rights-based analysis. The easier example, though, of this phenomenon is the

Supreme Courts and election procedures  95 much more attenuated claim of a fundamental right in malapportionment litigation. Voting is clearly a fundamental right, but this right is much more attenuated when discussing whether small differences in district sizes are actionable under a constitutional rights framework. In addition, because these election rules are core features of the political process, it is perhaps appropriate to let the political process produce compromises as long as the most egregious and self-serving manifestations of the process are constrained by the courts. The Japanese example in its malapportionment cases is an excellent example of an appropriate level of restraint by a court.

Notes   1 Much of this effect results from the amplification of vote majorities into seat majorities when single member districts are used. The best way to disentangle these two effects is to compare parties at similar vote percentages to see if the seat percentages differ depending on the party. The Republican’s 58 percent of the vote (2002) and the Democrat’s 56.3 percent of the vote (2006) each produced identical seat shares of 63.2 percent. These results suggest that the partisan bias of the gerrymander was small or easily overcome by changes in voting patterns.  2 Gill v. Whitford (585 US, 2018) was the vehicle for the court to consider the efficiency gap as a standard to evaluate partisan gerrymandering, but this case was remanded on the issue of standing, setting aside any evaluation of this standard. The Supreme Court, though, agreed in 2019 to hear another challenge to partisan gerrymandering.   3 An anomalous one-seat district had been created by the reversion of some islands to Japanese sovereignty well after the conclusion of World War II. The islands remained their own election district for nearly forty years before they were merged with an adjoining mainland district.   4 Occasionally district boundaries are changed to match changes in local administrative boundaries. These changes are not included in this analysis. The remaining 18 percent of districts is larger than the 11 percent reported because some of Japan’s 130 districts were newly created districts caused by a split of an existing district. When a district is split, both the new and the old district can be counted as an altered district even though the new district was not one of Japan’s original election districts.   5 Of course Marbury v. Madison also found a law passed by Congress to be unconstitutional, but there was no effect of the Marbury decision because the Court found other reasons to deny Marbury his requested relief. Similarly, prior to Scott v. Sandford the Court struck down some laws passed by Congress that regulated the judiciary.   6 Matsui (2011, 1388–95) describes the eight laws that were held unconstitutional. They were laws about patricide, overseas voting, liability for lost mail, locations of pharmacies, two for malapportionment, forest divisions, and illegitimacy. He also describes the six other government actions that have also been held unconstitutional.   7 Bolz (1980, 102) argues, in contrast, that the Japanese Court may be giving the legislature time to make amendments before striking down a law as unconstitutional, and he cites a constitutional challenge to a law that allowed the confiscation of property owned by third parties without notice that the Court first found to be constitutional but later found to be unconstitutional after the Diet failed to amend the law.

96  Supreme Courts and election procedures  8 Other authors address the conservative nature of the Japanese Supreme Court, but they credit different institutional features as explanations of the conservatism. For example, Law (2009) argues that short tenure of judges, promotion and vetting practices in the Japanese court system, the large number of judges, and the shared educational and social backgrounds of Japanese elites produce conservative justices who are aligned with the ruling LDP leaders who formally nominate the justices to the Supreme Court. Satoh (2008) also notes the high turnover of court members and adds the role of the Cabinet Legislative Bureau and its review of the constitutionality of legislation as explanations for the relative passivity of the Court. Itoh (2010, 213) finds no evidence of the political appointment process affecting the conservatism or liberalism of individual justices.  9 The only evidence that Ramseyer and Rasmusen give to support their claim is Ramseyer and Rosenbluth, 1997, chapter 3, but this chapter only makes the uncontroversial claim that the declining numbers of rural voters encouraged the LDP to switch its policy appeals to more urban-oriented issues (see also Ramseyer and Rosenbluth, 1993). 10 Itoh (2010, 149–54) also describes the Court’s malapportionment jurisprudence as switching back and forth between more conservative and more liberal decisions rather than the Court making a one-time switch in 1976 from a conservative position to a more liberal position. 11 The claim is also contradicted by Ramseyer and Rosenbluth (1993, 40), who show the percentage of farmers supporting the LDP rising from 52 percent in 1955 to 70 percent in 1980. Though their share of the electorate had declined dramatically, at least in 1980, the LDP had not relegated the agricultural vote to the socialists and the communists. 12 I calculated these numbers based on prefectures rather than individual election districts because reapportionments enacted by the Diet first assign seats to prefectures, with any changes in boundaries happening after the initial assignment of seats to prefectures. I also divided the prefectures according to increases or decreases in seats rather than a straight urban–rural measure. I did this because some largely rural prefectures located near large metropolitan areas have not lost seats in reapportionments. For this analysis it seems more accurate to assess LDP vote shares in districts that gained or lost seats rather than just looking at the urban or rural nature of a prefecture. I also calculated the votes of LDP candidates, independents affiliated with the LDP, and members of the New Liberal Club to reduce the effect of the splintering of the LDP vote among multiple LDP-affiliated candidates. Finally, I calculated these percentages based only on districts in which there were conservative candidates. From 1996 on, the LDP often has some districts in which it endorses a candidate of a coalition partner, and these districts are excluded from the prefectural-level calculations.

References Ansolabehere, Stephen, and James M. Snyder, Jr. 2008. The End of Inequality: One Person, One Vote and the Transformation of American Politics. New York: W. W. Norton. Backstrom, Charles, Leonard Robins, and Scott Eller. 1990. “Establishing a Statewide Electoral Effects Baseline.” In Political Gerrymandering and the Courts, edited by Bernard Grofman, 145–70. New York: Agathon Press.

Supreme Courts and election procedures  97 Bailey, William Somers. 1997. “Reducing Malapportionment in Japan’s Electoral Districts: The Supreme Court Must Act.” Pacific Rim Law & Policy Journal 6, no. 1 (January): 169–98. Baker, Gordon E. 1990. “The ‘Totality of Circumstances’ Approach.” In Political Gerrymandering and the Courts, edited by Bernard Grofman, 203–11. New York: Agathon Press. Barkow, Rachel E. 2002. “More Supreme than Court? The Fall of the Political Question Doctrine and the Rise of Judicial Supremacy.” Columbia Law Review 102, no. 2 (March): 237–336. Berman, Mitchell N. 2005. “Managing Gerrymandering.” Texas Law Review 83, no. 3 (February): 781–854. Bickel, Alexander M. 1975. The Morality of Consent. New Haven, CT: Yale University Press. Bolz, Herbert F. 1980. “Judicial Review in Japan.” Hasting International and Comparative Law Review 4, no. 1 (Fall): 87–142. Cain, Bruce E. 2015. Democracy More or Less: America’s Political Reform Quandary. New York: Cambridge University Press. Calidas, Douglass. 2008. “Hindsight Is 20/20: Revisiting the Reapportionment Cases to Gain Perspective on Partisan Gerrymanders.” Duke Law Journal 57, no. 5 (March): 1413–47. Carson, Jami L., and Michael H. Crespin. 2004. “The Effect of State Redistricting Methods on Electoral Competition in United States House of Representatives Races.” State Politics and Policy Quarterly 4, no. 4 (December): 455–69. Choper, Jesse H. 2005. “The Political Question Doctrine: Suggested Criteria.” Duke Law Journal 54, no. 6 (April): 1457–523. Cox, Archibald. 1976. The Role of the Supreme Court in American Government. New York: Oxford University Press. Dahl, Robert A. 1989. Democracy and Its Critics. New Haven, CT: Yale University Press. Danelski, David J. 1992. “Documenting the Establishment of Judicial Review: Japan and the United States.” In Comparative Judicial Review and Public Policy, edited by Donald W. Jackson and C. Neal Tate, 17–28. Westport, CN: Greenwood Press. Ely, John Hart 1980. Democracy and Distrust: A Theory of Judicial Review. Cambridge, MA: Harvard University Press. Ely, John Hart. 1998. “Law and the Political Process, Gerrymanders: The Good, the Bad, and the Ugly.” Stanford Law Review 50, no. 3 (February): 607–41. Fuentes-Rohwer, Luis. 2005. “Domesticating the Gerrymander: An Essay on Standards, Fair Representation, and the Necessary Question of Judicial Will.” Cornell Journal of Law and Public Policy 14, no. 3 (Summer): 423–42. Henderson, Dan Fenno. 1968. “Japanese Judicial Review of Legislation: The First Twenty Years.” In The Constitution of Japan: Its First Twenty Years, 1947–67, edited by Dan Fenno Henderson, 115–40. Seattle, WA: University of Washington Press. Higuchi, Yoichi. 2001. “The 1946 Constitution: Its Meaning in the Worldwide Development of Constitutionalism.” In Five Decades of Constitutionalism in Japanese Society, edited by Yoichi Higuchi, 1–8. Tokyo: University of Tokyo Press. Issacharoff, Samuel. 2002. “Gerrymandering and Political Cartels.” Harvard Law Review 116, no. 2 (December): 593–648. Issacharoff, Samuel, and Pamela S. Karlan. 2004. “Where to Draw the Line?: Judicial Review of Political Gerrymanders.” University of Pennsylvania Law Review 153, no. 1 (November): 541–78.

98  Supreme Courts and election procedures Itoh, Hiroshi. 2010. The Supreme Court and Benign Elite Democracy in Japan. Burlington, VT: Ashgate. Jacobson, Gary C. 2004. The Politics of Congressional Elections, 6th ed. New York: Pearson Longman. Kasper, Michael J. 2007. “The Almost Rise and Not Quite Fall of the Political Gerrymander.” Northern Illinois University Law Review 27, no. 3 (Summer): 409–25. Koshiji, Masami. 2001. “Constitutional Issues concerning the Franchise.” In Five Decades of Constitutionalism in Japanese Society, edited by Yoichi Higuchi, 133–54. Tokyo: University of Tokyo Press. Law, David S. 2009. “The Anatomy of a Conservative Court: Judicial Review in Japan.” Texas Law Review 87, no. 7 (June): 1545–94. Law, David S. 2011. “Why Has Judicial Review Failed in Japan?” Washington University Law Review 88, no. 6: 1425–66. Lijphart, Arend. 1984. Democracies: Patterns of Majoritarian and Consensus Government in Twenty-One Countries. New Haven, CT: Yale University Press. Lublin, David, and Michael P. McDonald. 2006. “Is it Time to Draw the Line? The Impact of Redistricting on Competition in State House Elections.” Election Law Journal 5, no. 2 (June): 144–57. Matsui, Shigenori. 1986. “The Reapportionment Cases in Japan: Constitutional Law, Politics, and the Japanese Supreme Court.” Osaka University Law Review 33 (March): 17–46. Matsui, Shigenori. 2011. “Why Is the Japanese Supreme Court so Conservative?” Washington University Law Review 88, no. 6: 1374–424. Mavcic, Arne. 2001. The Constitutional Review. Den Bosch: Book World. McDonald, Michael D., and Richard L. Engstrom. 1990. “Detecting Gerrymandering.” In Political Gerrymandering and the Courts, edited by Bernard Grofman, 178–202. New York: Agathon Press. Niemi, Richard G. 1990. “The Swing Ratio as a Measure of Partisan Gerrymandering.” In Political Gerrymandering and the Courts, edited by Bernard Grofman, 171–177. New York: Agathon Press. Otake, Hideo. 2001. “The Making of the Japanese and West German Constitutions.” In Five Decades of Constitutionalism in Japanese Society, edited by Yoichi Higuchi, 43–71. Tokyo: University of Tokyo Press. Persily, Nathan. 2002. “In Defense of Foxes Guarding Henhouses: The Case for Judicial Acquiescence to Incumbent-Protecting Gerrymanders.” Harvard Law Review 116, no. 2 (December): 649–83. Pildes, Richard H. 2004. “The Constitutionalization of Democratic Politics.” Harvard Law Review 118, no. 1 (November): 28–42. Ramseyer, J. Mark, and Eric B. Rasmusen. 2001. “Why Are Japanese Judges so Conservative in Politically Charged Cases?” American Political Science Review 95, no. 2 (June): 331–44. Ramseyer, J. Mark, and Frances McCall Rosenbluth. 1993. Japan’s Political Marketplace. Cambridge, MA: Harvard University Press. Ramseyer, J. Mark, and Frances McCall Rosenbluth. 1997. Japan’s Political Marketplace, with a New Preface. Cambridge, MA: Harvard University Press. Satoh, Jun-ichi. 2008. “Judicial Review in Japan: An Overview of the Case Law and an Examination of Trends in the Japanese Supreme Court’s Constitutional Oversight.” Loyola of Los Angeles Law Review 41, no. 2 (Winter): 603–28.

Supreme Courts and election procedures  99 Tomatsu, Hidenori. 2001. “Judicial Review in Japan: An Overview of Efforts to Introduce US Theories.” In Five Decades of Constitutionalism in Japanese Society, edited by Yoichi Higuchi, 251–77. Tokyo: University of Tokyo Press. Yokota, Kisaburo. 1968. “Political Questions and Judicial Review: A Comparison.” In The Constitution of Japan: Its First Twenty Years, 1947–67, edited by Dan Fenno Henderson, 141–166. Seattle, WA: University of Washington Press. Zelden, Charles L. 2010. The Supreme Court and Elections: Into the Political Thicket. Washington, DC: CQ Press.

6 Corruption

Lesson 6: The United States should adopt Japan’s uniform rules for the collection, handling, and counting of ballots which makes stealing an election on election night practically impossible. Japan has a well-deserved reputation for political corruption. Politicians are ­regularly embroiled in scandals, and these scandals typically involve election activities (Carlson and Reed, 2018). Many politicians illegally raise and spend campaign funds. Often the raising of these funds, at least in past decades, involved kickbacks paid to politicians by companies that got government contracts (Johnson, 1995, 183–211; Woodall, 1993). Candidates spread money around in their districts prior to election day (Curtis, 1971, 232–41). The ubiquity and notoriety of these actions help contribute to Japan’s low ranking, at least among industrialized democracies, on measures of the perception of corruption (Transparency International, 2019).1 It also has led to reform movements, anti-corruption crusades, and new laws to reduce levels of corruption (Carlson, 2007). Though these anti-corruption efforts have made progress, it still seems odd to hold up anything about Japanese elections, at least in the realm of electoral integrity, as an example to other countries. This potential confusion stems from a failure to distinguish between the actions of many politicians, which have been notoriously corrupt, and the actions of the electoral bureaucracy and the functioning of election rules, which have compiled an impressive record of election integrity in the collection, handling, and counting of ballots. It is these practices that provide an example to other countries, especially the United States. The United States could dramatically improve the integrity of its elections by following some of the rules that Japan uses in its elections. For this analysis of Japanese rules on the collection, handling, and counting of ballots, I first define election night corruption and how it differs from corruption generally and more specific forms of electoral corruption. I then measure the frequency of occurrence of election night corruption in Japan, the United States, and Canada. This measurement is made possible by the special characteristics of election night corruption and the unique evidentiary trail that this form of corruption creates. I focus on Japan because its ballot-­ handling practices make election night corruption practically impossible to do.

Corruption  101 I also analyze the United States because tales of stealing elections on election night are part of US electoral lore, and US ballot-handling practices have all the deficiencies that Japanese practices do not have. I include Canadian data as an additional reference point because Canada also has a long history of single-seat elections in which the incentive to alter ballots on election night in order to secure a victory for a specific candidate are as high as they are both in Japan and the United States. After reporting my results on the relative frequency of election night corruption in all three countries, I last examine Japan’s ballot-handling rules in order to create a model of best practices that could be used by any country hoping to reduce the potential for election night corruption or improve the integrity of the entire election process.

Election night corruption Corruption is difficult to define. At its core, corruption is defined by the law. What is corrupt or illegal in one country might be perfectly legal and common behavior in another country. Japanese politicians break the law if they ask for support outside of the short, official campaign period. In contrast, all US politicians regularly ask for such support at any and all times. In addition, there are many unethical practices that are not illegal. For some, these legal yet shady practices are corrupt. For others, these practices are not corrupt. Carlson and Reed (2018, 1–16) recognize these deficiencies of the legalistic definition of corruption as well as the deficiencies of other, competing definitions that are based on private gains or corrupt exchanges. Rather, they suggest that the best definition of corruption, at least in the Japanese context, focuses on subverting democratic procedures. This definition, therefore, includes perfectly legal actions that change democratic procedures, typically to give an advantage to a particular party or candidate. Fortunately, this analysis of Japanese election procedures does not require wading into the debates about the most appropriate definition of corruption. Instead, this analysis focuses on only electoral corruption and even more specifically on stealing an election on election night (or shortly thereafter). For this analysis a legalistic definition of corruption is the most appropriate because all of the actions discussed are illegal. Both generic electoral corruption and the more specific form of election night corruption share, with other corrupt activities, a difficulty in discovering their occurrence. Both types of corruption are difficult to identify with certainty because participants hide their actions from authorities (Duggan and Levitt, 2002, 1594; Kallina, 1988, 182–93). In addition, there is no victim’s body or cache of stolen goods to prompt a police investigation. Hence both generic electoral corruption and the more specific act of stealing an election on election night typically go undetected and unpunished. For example, if a corrupt election worker stuffs a ballot box with bogus ballots for her favored candidate, who will discover this action? Certainly the corrupt worker has no incentive to confess. Voters might notice if records show them as having voted

102 Corruption in an election in which they didn’t vote, but such a check of the records is unlikely. If the bogus ballots replaced legitimately cast ballots or were cast in the names of dead or incapacitated voters, then only the determined research work of the opposing camp could ferret out the ballot stuffing. Similarly, if on election night a partisan operative adjusts the vote totals reported for a specific precinct and pads the electoral records to support the altered vote totals, who but the operative would be able to show that such alterations were made? Despite these evidentiary difficulties, election night corruption (the adjustment of the last returns to ensure a candidate has a narrow margin of victory) does leave a distinct evidentiary trail. This evidentiary trail makes it easier to detect the likely occurrence of election night corruption, even without the aid of eyewitnesses. When an election is stolen on election night—after nearly all of the ballots are counted—the corrupt officials know exactly how many additional votes are needed to secure a victory. Candidates hold back precincts that they control (colluding with local electoral officials) until they know how many votes are needed to ensure victory (Campbell, 2005, 207, 247). If the candidate is facing sure defeat or sure victory, then no adjustments are made, and the results from those precincts are simply released. If however, the candidate is facing a narrow loss, local electoral officials make last-minute adjustments to ballot totals to create a slim margin of victory for their preferred candidate. Thus, typically only a bare margin of victory is created in election night corruption, just enough new votes to ensure victory, but not so many as to raise suspicions or trigger investigations. Thus, an anomalous number of ultra-close races suggests that election night corruption is occurring. Political history in Nevada illustrates this point. Senate races in 1964, 1974, and 1998 were won with 48, 624, and 401 vote margins respectively. Is it just a coincidence that Nevada had three nail-biter US senate elections in 34 years? Or, were vote totals adjusted in one or two of these races to provide a bare margin of victory for a candidate who was otherwise headed for defeat? The disputed 1964 Senate election in Nevada is a good candidate for possible election night corruption. Lieutenant Governor Paul Laxalt challenged the incumbent Howard Cannon in a race that went down to the wire. With 90 percent of the vote counted and a 6,000-vote lead, Cannon declared ­victory late in the evening. Laxalt, however, refused to concede because most of the uncounted precincts were located in rural counties that strongly favored ­Laxalt. By the next morning, the nearly complete returns gave Laxalt an 18-vote lead. Later that morning two county clerks revised their vote tallies, correcting what they claimed were errors in their previously announced tallies for their counties.2 In a strange coincidence, the only counties to revise their tallies were the home counties of Cannon and Laxalt, and both revisions favored the hometown politician. Cannon won the race, in part, because the number of new Cannon votes found the morning after the election in populous Clark County (Las Vegas) swamped the additional Laxalt votes found in his much smaller base of Ormsby County (Carson City). I am from Nevada, and I still

Corruption  103 remember my father joking as he pantomimed how he thought Cannon had won the election: he would act out the counting of hundred-dollar bills in payment to a local election official. A similar story is told about Lyndon Johnson’s 1941 loss and 1948 ­victory in Democratic primary races for Texas Senate seats. Johnson lost in 1941 because he urged his supporters to release the returns from heavily pro-­Johnson precincts early on election night to build up a large margin and discourage his opponent. In contrast, his opponent held back the precincts he controlled, saw how many votes were needed for victory, and allegedly manufactured enough votes on election night to defeat Johnson by a slim ­margin. President Roosevelt later teased Johnson for his mistake in 1941 saying, ­‘Lyndon, up in New York the first thing they taught us was to sit on the  ballot boxes’ (Dallek, 1991, 224). In 1948, in contrast, Johnson had learned his lesson and was better prepared to ensure his own victory over fellow Democrat Coke Stevenson. After Johnson’s supporters knew how many votes would be needed to turn a narrow Johnson loss into a narrow Johnson victory, Johnson allies created enough bogus ballots to provide this margin of victory. Unfortunately for Johnson, the bogus ballots in Precinct 13 in Alice, Texas were recorded in alphabetical order. These 200 people managed to vote in alphabetical order, and they were just the amount needed to win the Senate seat for Johnson. The original and all copies of this suspect voting list were lost or stolen before they could be examined by a court, so the legend of the alphabetical voting list in Alice lives on in the statements by those who saw the list in the first few days after the election (Caro, 1991, 375–6; Dallek, 1991, 340). More recently, John Fund (2004, 78–9) has raised questions about Tim Johnson’s narrow victory over John Thune in the 2002 South Dakota senate race. Thune held a narrow lead over Johnson until the last returns came in from Shannon County. Those returns gave Johnson just enough votes to win the election. Fund points out that turnout and the Johnson vote in Shannon County were both disproportionately higher than other similarly situated pro-Johnson counties. This anomaly opens the door to questions that perhaps the very last votes counted in South Dakota, the Shannon County votes, were altered to give Johnson just enough votes to squeak out a victory over Thune. In contrast to these possible examples of election night corruption, all other campaign activities—corrupt or legitimate—are not able to affect final vote totals with the precision that is only available to those who manipulate vote totals after almost all of the votes have been counted. Ballot boxes may be stuffed or votes may be bought, but these actions occur before anyone knows how many votes will be needed to assure victory. In the absence of such clear information, more bogus votes are better than fewer bogus votes. Operatives try to create as many votes as they can for their candidate without being caught. These additional votes are obtained even though, ultimately, the votes might simply add to a candidate’s already ample margin of victory or be wasted in a losing effort that no amount of corruption is able to reverse.

104 Corruption For example, a candidate might redouble her efforts in the final days of a campaign because all reports indicate that the race will be very close. She might try to buy more votes, run more advertisements, or enhance her get out the vote effort. Any of these activities should raise her share of the vote over what she would have won if she had not made that last-minute, additional effort. If, for example, her efforts increased her vote share by 2 percent, how would her efforts affect the outcome of the election? The candidate and her advisors only knew that the race was going to be very close. Perhaps her additional efforts turned what would have been a 49 percent loss into a 51 percent victory. Perhaps the efforts turned what would have been a 51 percent victory into a 53 percent victory or a 47 percent loss into a 49 percent loss. Thus, all electoral efforts (other than election night corruption) reverse the outcomes of some races (a 49 percent loss becomes a 51 percent victory), make some races closer (a 47 percent loss becomes a 49 percent loss), and make some races less close (a 51 percent victory becomes a 53 percent victory). The candidates only know that the race is likely to be close and that additional efforts (legal or illegal) may be needed to secure a victory (Eggers et al., 2015). These activities do not create a distinct evidentiary trail. Only election night corruption, in contrast to other forms of election corruption, stops producing votes once a narrow margin of victory has been achieved. This production of bogus ballots stops because going beyond a slim margin of victory increases the risk of being caught. It might seem odd that those who steal elections would stop manufacturing votes for their candidate when only a small margin of victory is created. However, it is difficult to create bogus votes because the number of votes that can be created in any given precinct is limited by the number of registered voters, expectations of typical turnout in a given precinct, and expectations of typical ratios of partisanship in a specific precinct. A corrupt election official may be able to add 50 fraudulent votes to the totals for a precinct in which there were 400 registered voters and 300 actual votes were cast, but adding 100 votes to that precinct would certainly call attention to that precinct. Similarly, if a candidate won 200 of the 300 votes cast in that precinct, those totals could be adjusted to 250 of the 300 votes without perhaps attracting attention, but any greater margin is likely to trigger an investigation of the voting returns. Each precinct can produce only a certain number of fraudulent votes over and above the legitimate votes cast for a candidate without attracting unwanted attention, and the manipulation of each precinct’s data requires the corruption of a local election official in that precinct. Given these constraints, it is not surprising that election night corruption rarely results in large margins of victory for the candidate stealing the election. Another factor also puts a ceiling on the number of votes that can be created, destroyed, or altered in an effort to steal an election after almost all of the votes have been counted. These election night efforts come on top of an effective campaign that will already have boosted a candidate’s margins to near maximum levels before the counting begins on election night. In addition, the places where a candidate is most likely to have corrupt election officials are

Corruption  105 where the candidate’s support is strongest, and those are precisely the areas where there is the least room to increase a candidate’s vote totals. If a candidate is already winning 85 percent of the vote in a precinct that already has 90 percent turnout levels, there is little room to add votes to the candidate’s already impressive turnout and vote totals. Several examples show the limits on the creation of new ballots and the perils of pushing too far in creating, altering, or destroying ballots. In the 2018 midterm election in the 9th Congressional District of North Carolina, state election officials refused to certify the narrow victory of the Republican candidate for a variety of reasons, all based on election irregularities. Attention, however, was first drawn to this race because of the absentee voter numbers in one county of the district, Bladen County. That county had an unusually high number of requests for absentee ballots, had an unusually high number of such ballots that were not returned, and though registered Republicans were only 19 percent of the absentee voters in that county, the Republican candidate won 61 percent of the absentee votes. These anomalous numbers attracted scrutiny and ultimately led to the calling of a new election in this congressional district. Though this is not an example of election night corruption, this example shows the problem that corrupt operatives face. They can create, destroy, or change ballots, but if they affect too many ballots, anomalies show up. Thus, there is a cost to every additional vote obtained and the cost increases as the number of bogus votes increases. The Nevada example also shows this limitation. If Laxalt were trying to create additional votes, he was hobbled by the fact that there were only 5,000 voters in his base of Ormsby County. Laxalt had already won 60 percent of those votes. In contrast, Cannon’s base of Clark County had more than 64,000 votes cast with Cannon winning a similar 61 percent of the vote. Cannon could have increased his vote share in Clark County by a hardly noticeable 1 percent and created 640 additional Cannon votes, but Laxalt would have had to increase his margin of victory in Ormsby County by a much more noticeable 12 percent in order to create the same number of votes. With both counties adjusting up their vote totals in the morning after the election, Ormsby County simply couldn’t match the vote increase posted by much larger Clark County. Similarly, Lyndon Johnson’s blatant manipulation of the votes in Alice, Texas was a mark of his supporters’ desperation. Johnson allies had created every additional vote that they could in the areas that they directly controlled, but Coke Stevenson, Johnson’s opponent, was still managing to best Johnson’s vote totals. In a brazen and ultimately successful attempt to win, Johnson allies reached out of their base to nearby Alice because there still was a large gap there between the number of registered voters and ballots cast, making it possible to create a significant number of additional Johnson votes. Those opportunities had all been used up in more solid Johnson precincts, but by reaching outside of their area of direct control, Johnson’s allies made themselves vulnerable to claims of cheating by a local official in Alice who was not part of the Johnson machine.

106 Corruption Distinct evidentiary trails created by certain forms of corrupt behavior also show up in settings outside of elections. For example, Duggan and Levitt (2002) describe how incongruities in the patterns of victories by some sumo wrestlers suggest the existence of deals to throw matches. Similar methods have been used to detect favoritism by soccer referees (Garicano, Palacios-Huerta, and Prendergast, 2005), Medicare abuse (Becker, Kessler, and McClellan, 2005), and teachers giving illegal help to students taking standards exams (Jacob and Levitt, 2003). Duggan and Levitt (2002) also describe the historical example of how there were suspiciously fewer French military conscripts whose height was just above the minimum height and disproportionately large numbers of those who avoided conscription because their height was just below the minimum. In each of these cases, corruption cannot be confirmed in the case of a specific conscript, a specific exam, or a specific sumo match, but the existence of multiple incidents of corruption can be confirmed. This confirmation comes by comparing the actual distribution of results to a hypothetical distribution of results that should have existed if there had been no corruption. Some recent political science work has used related methods to assess the voting preferences of a subset of absentee voters in Florida (Imai and King, 2004). Others have similarly applied statistical methods in the study of electoral corruption (Mebane, 2006; Simpser, 2013), but other than these studies, the comparison of actual results against what the distribution of outcomes should have been without corrupt activities has rarely been used in the study of political corruption. This analysis fills this gap, looking at the difference between actual election results and a hypothetical distribution of election results if no election night corruption occurred. Of course, many factors could be used to predict differences in both corruption generally and election night corruption specifically among countries. Federal systems are seen as being more prone to corruption (Gerring and Thacker, 2004; Treisman, 2000). Parliamentary regimes are claimed to be less prone to corruption (Gerring and Thacker, 2004). Strong candidate-centered regimes also create incentives for greater election corruption. Countries have cultures that are more prone or less prone to corruption, at least as measured in perception studies of corruption (Alesina and Weder, 2002; Anderson and Tverdova, 2003; Davis, Camp, and Coleman, 2004; Gerring and Thacker, 2005; Xin and Rudel, 2004) or in studies of actual corruption prosecutions (Golden and Chang, 2001; Meier and Holbrook, 1992). This study, however, examines election bureaucracy and rules about the handling and counting of ballots in order to understand the frequency of a specific form of corruption, election night corruption.

Using aggregate measures to identify election night corruption in Japan, Canada, and the United States Though the Nevada and Texas stories presented above are part of the political lore of the United States, it is noteworthy that there are no examples of similar stories of corrupt campaign officials in Japan facilitating the illegal

Corruption  107 manufacture or destruction of ballots, sufficient to change an outcome of an election. This lack of such examples in Japan stands out given the notorious corruption of Japanese politics generally. Fortunately, it is possible to go beyond stories and analyze the occurrence of election night corruption by looking for distinctive evidentiary trails such as: 1 2 3

a disproportionate number of ultra-close races compared to very close races; one political party winning a disproportionate number of ultra-close races compared to very close races; a correlation between the winners of ultra-close races and their party’s control of the local electoral machinery.

A first and obvious test, then, for election night corruption is to compare the number of ultra-close races (defined as victory margins of less than 0.5 percent) against the number of very close races (victory margins less than 2.5 percent). I group races into half-percent categories, creating, for example, 20 categories of victors of those that won by 10 percent or less. These categories can then be compared to see if ultra-close races significantly differ from very close races. Comparing similarly situated races, for example, all races won by 2.5 percent or less, controls for the many factors, legal or illegal, that can affect the closeness of a race generally. Because factors other than election night corruption cannot distinguish between a very close race (victory margin of less than 2.5 percent) and an ultra-close race (victory margin of less than 0.5 percent), the differences that do show up are likely caused, in part, by election night corruption.3 Thus, in jurisdictions in which there is no election night corruption, the distribution of the margins of victory, plotted according to half-percent increments should produce a distribution of only incremental changes between half-percent intervals. These incremental changes would reflect the impact of all of the factors that generally affect the closeness of races. There might be twice as many 1-percent victories as there are 10-percent victories because races in that jurisdiction are quite competitive. On the other hand, there might be twice as many 10-percent victories as there are 1-percent victories because one party dominates races in that jurisdiction or because bipartisan gerrymandering has eliminated most of the competitive districts. Under either scenario, however, the changes in half-percent categories moving from 1 percent to 10 percent should be incremental. None of the factors (other than election night corruption) that affect the closeness of races generally are able to distinguish between victories of 0.5 percent and victories of 1 or 2 percent. Thus, if election night corruption is not occurring, the difference between the number of races won by 0.5 percent and races won by 1 percent should be consistent with the trend line of incremental differences between, for example, races with a 1.5-percent and 2-percent margin of victory. In contrast, if election night corruption is common there will be a sharp break between the number of ultra-close races and the number of very close

108 Corruption

Number of election districts

300 250

200 150 100 50

Japan

Canada

9% 9. 50 % 10 %

8% 8. 50 %

7% 7. 50 %

6% 6. 50 %

5% 5. 50 %

4% 4. 50 %

3% 3. 50 %

2% 2. 50 %

1% 1. 50 %

0. 50

%

0

United States

Margin of victory

Figure 6.1 Distribution of margins of victory in Japan, Canada, and the United States

races, rather than an incremental change between these two categories. Figure 6.1 reports aggregate totals for Japan, Canada, and the United States, ­scaling the results so that all three countries can be placed in comparable positions on the graph.4 Figure 6.1 also truncates the distribution, showing only the results for the first 20 categories of half-percent increments (all races with a 0–10-­percent margin of victory). There is no sharp or disproportionate increase for the ultra-close races (category 1). Japan, in contrast to the United States and Canada, shows a skewing in favor of close races which is explained by the multi-seat electoral system that Japan used until the 1993 elections. However, this trend occurs across all ultra-close and very close races, creating no suggestion that election night corruption occurred. This first cut of the data suggests that election night corruption is not occurring in Japan, the United States, or Canada. It is possible, however, for significant election night corruption to occur without altering the overall distribution of margins of victory. This can occur if one party disproportionately benefits from election night corruption. For example, election night corruption might be disproportionately used by one party to convert what would have been ultra-close victories for their opponents into ultra-close victories for their candidates. It stands to reason that election night corruption is more likely to be attempted when an opponent’s margin of victory would have been small and easy to reverse. If this scenario is correct, then the total number of ultra-close races would not be disproportionately larger, but the party distribution of ultra-close races would stand out in contrast to the party distribution of very close races. Figures 6.2, 6.3, and 6.4 break down the data for each country reported in Figure 6.1 into partisan groups. The Japanese data shows a consistent pattern across all percentage groupings of victories. The LDP and affiliated independents maintain a consistent and small advantage over all non-LDP candidates which is consistent with the

Corruption  109 180 Number of election districts

160 140 120 100 80 60 40 20

9% 9. 50 % 10 %

8% 8. 50 %

%

7. 50

%

7%

6. 50

%

LDP

6%

5. 50

%

5%

4. 50

%

4%

3. 50

%

3%

2. 50

%

2%

1. 50

% 0. 50

1%

0

Others

Margin of victory

Figure 6.2 Margins of victory in Japan, by political party 90 Number of election districts

80 70 60 50 40 30 20 10

Liberals

Conservatives

NDP

%

% 10

50

9.

%

9%

50

8.

8%

% 50

7.

7%

% 50

6.

6%

% 50

5.

5%

% 50

4.

4%

% 50

3.

3%

% 50

2.

%

2%

50

1%

1.

0.

50

%

0

Other

Margin of victory

Figure 6.3  Margins of victory in Canada, by political party

fact that the LDP has won a small majority of the seats over time also. This data shows no evidence of election night corruption. Similar results are obtained looking at Canadian data, broken down into four party groupings. No strong pattern emerges of one party doing better in ultra-close elections in contrast to very close elections. The data for the United States, however, is intriguing. There are always more Republican victors than Democratic victors for every category of close races from 0- to 10-percent margins of victory except for the closest category, those candidates who won by less than 0.5 percent of the vote. There, for the only time, do the Democrats win more races than the Republicans. Though it is hard to see in Figure 6.4, there is also a sizeable jump in the number of winners

110 Corruption

Number of election districts

500 450 400 350 300 250 200

150 100 50

Democrats

Republicans

9% 9. 50 % 10 %

8% 8. 50 %

7% 7. 50 %

6% 6. 50 %

5% 5. 50 %

4% 4. 50 %

3% 3. 50 %

2% 2. 50 %

1% 1. 50 %

0. 50

%

0

Other

Margin of victory

Figure 6.4  Margins of victory in the United States, by political party

in the ultra-close category for the candidates of other parties, a category that includes all independents and candidates who run on party lines other than the Republican and Democratic parties. Thus, of the three countries, only the United States shows a footprint of possible election night corruption, but are these results just random variation or are they statistically significant? After all, the number of opposition victors in Japan and Conservative Party victors in Canada also drop for the ultra-close category. At what level is it possible to be confident that these patterns aren’t just random noise in the data? It is possible to use the data to calculate a trend line that will predict the number of ultra-close winners for each party grouping. This predicted number can be compared to the actual number of ultra-close winners, and the difference between these numbers can be assessed to see if the difference is statistically significant, meaning that the results are less likely to be just random noise in the data. Thus, for example, the Democrats in the United States won 365 seats with margins of victory of 2.5 to 3.0 percent. They won 393 seats with margins of 2.0 to 2.5 percent. They also won 383, 385, and 383 seats in the next categories of 1.5–2, 1.0–1.5, and 0.5–1.0 percent. I used a statistical procedure that uses all five observations, weighting later observations more, to predict the number of victors in the 0–0.5-percent margin of victory category. This predicted value was 386.8 victors in the ultra-close category, with a 95-percent confidence interval of the actual number being within the range of 363.5 to 392.1 victors. In reality, the Democrats won 425 seats in this category, suggesting that the number of Democratic victors who won by less than 0.5 percent of the vote is both anomalous and statistically significant. Similar calculations for all nine party groupings are reported in Table 6.1. The results reported in Table 6.1 seem like stunning evidence of election night corruption occurring only in the United States and being centered in the Democratic Party and in third parties and independents. Indeed, some authors have also speculated that election corruption might be more common in the

Corruption  111 Table 6.1  Predicted and actual numbers of ultra-close races, by party in Japan, Canada, and the United States Party/country Democrats US Republicans US Other US LDP Japan Other Japan Liberals Canada Conservatives Canada NDP Canada Other Canada

Predicted number

Actual number

Difference

Confidence interval

Statistically significant?

386.8 432.5 45.5 169.2 171.2 59.4 59.2

425 424 62 163 146 59 52

+38.2 −8.5 +16.5 −6.2 −25.2 −0.4 −7.2

363.5–392.1 398.3–466.7 31.82–59.3 147.2–191.3 144.2–198.2 42.5–76.3 41.0–77.5

Yes No Yes No No No No

11.3 23.8

14 28

+2.7 +4.2

4.7–18.0 17.5–30.0

No No

Democratic Party in the United States (Sabato and Simpson, 1996, 299; Fund, 2004, 6–7). Similar drops in the number of Canadian Conservative victors or opposition victors in Japan, in contrast to the US data, are not statistically significant. However, as a robustness check for these results, I calculated every half-percent category of victory between 0 percent and 10 percent victories (a total of 20 categories). For each calculation, I used the trend line for the five prior categories. Of a total of 180 calculations (20 calculations for each of 9 party groupings), 64 of these calculations produce statistically significant results. In addition, these results are also affected by the number of categories included in the trend line calculations. I used the five previous categories to create trend lines, but slightly different results obtain if only the three previous categories, the ten previous categories, or all previous data are used. These varied results show that the number of victors in a certain margin of victory category are fairly volatile, and calculations of statistical significance should not be the end of the analysis but rather a suggestion of a possible relationship that needs further investigation.

A comparison of election night corruption using individual measures One problem with the analysis of aggregate data is the possible cancelling out of events of election night corruption by different political actors. For example, the US data shows a statistically significant advantage for the Democratic Party of 38 additional seats (over the predicted number of seats) in the ultra-close margin of victory category. However, it could be the case that Democratic candidates stole elections in 70 races and Republican candidates stole elections in 32 races, producing a net advantage to the Democrats of 38 additional seats. Thus, the true magnitude of election night corruption might be underestimated if only aggregate data is analyzed. In order to ascertain if

112 Corruption mutual corruption efforts are masking some of the incidents of election night corruption, I used a statistical procedure called a logistic regression. This procedure predicts the likelihood of a very close race ending up in the category of an ultra-close margin of victory, while controlling for the effect of other factors. I controlled for the following factors: the province or region of the race, the type of race (for US elections), the party of the candidate, the ruling party in the local jurisdiction of the race, decade groupings of races, and incumbency status. I am most interested in the following as predictors of election night corruption: 1 2 3

whether or not incumbents are more likely to win ultra-close races; whether or not candidates from the party that controls the local jurisdiction are more likely to win ultra-close races; whether or not time periods or certain jurisdictions have more ultra-close races.

I tested these three possible correlates of election night corruption in separate data sets for each country with the dependent variable being whether or not a specific victory was by an ultra-close margin in contrast to being by only a very close margin. Because this variable only takes two values (either in the ultra-close category or not), a logistic regression is the appropriate statistical procedure. For my analysis of Japan, as predictors (independent variables) I  calculated how urban a specific district was (given that much analysis of Japanese politics turns on differences in urban versus rural voting, a difference that could be relevant to measures of electoral corruption), whether or not the winner of the race was an incumbent, the time period of the election (in decade groupings from 1947 to 2003), and geographical groupings of districts into eight regions of Japan. I also calculated a measure of political influence that matched the party of the candidate with the party that controlled the governorship in a particular prefecture, thinking that the party in charge at the local level might be more able to affect ballot counting in its areas of jurisdiction. I also ran this test with the local party control variable separate from the variable for the candidate’s party. Finally, because there is nothing magical about saying that election night corruption will always end with a margin of victory of 0.5 percent or less, I calculated this effect or dependent variable both for margins of 0 to 0.5 percent and for 0 to 1 percent victories. Similarly, I limited the entire data set of very close victories to one set of cases of all victories from 0 to 2.5 percent and another set of cases of all victories from 0 to 5 percent. As a check, I also ran regressions on the entire data set of victories, though I don’t believe that a comparison of ultra-close victories against all victories will show an effect that is necessarily linked to election night corruption. There should be quite a few factors that distinguish close contests from all other contests, and these factors would likely have nothing to do with election night corruption. Thus, I ran many independent regression tests, and they all showed no evidentiary footprint of election night corruption in Japan. LDP candidates

Corruption  113 (including affiliated conservative independents) are not more likely to win ultra-close races (in contrast to winning very close races) when a copartisan controls the local government. The answer remains no regardless of whether the categories used for very close races are 0 to 0.5 percent or 0 to 1 percent and regardless of whether the entire data set includes only races with margins of victory of 0 to 2.5 percent or 0 to 5 percent (or even all the races). Reversing the test to see if opposition candidates are more likely to win ultra-close races when a copartisan controls the governorship also produced negative results across all three data sets.5 Similarly, running tests with separate variables for candidate parties and party control of the local administration also produced no results that were statistically significant, except when opposition control of local administration is considered in the data set of all election victories, an outcome that is not interesting for this research question.6 In addition, with only a few exceptions, decades, regions, urban versus rural, and incumbency did not matter in explaining the occurrence of ultra-close races, regardless of the size of the categories used for either the dependent variable or the entire data set.7 In contrast, a similar series of regressions for the Canadian and US data produce results suggesting the possibility of election night corruption in these countries. In Canada, some decade variables are consistently statistically significant regardless of the size of the data set used in the regressions, and incumbency is statistically significant in regressions using the 0 to 5 percent category of close races. However, incumbency is of an unanticipated sign, with a negative coefficient, meaning that there are fewer incumbents winning races by ultra-close margins, not a surprising outcome generally (incumbents, should, all things being equal, be more likely than non-incumbents to win races by large margins), but contradicting the idea that incumbents might be better situated to manipulate local election officials in order to steal an election on election night. The most intriguing Canadian outcome is the near statistical significance in both regressions (ultra-close victory is defined as less than a 0.5-percent or less than a 1-percent victory) for the 0 to 5 percent category of close races for Liberal candidates running in provinces where the premier is also a liberal. The p values for this variable in both regressions in this category of close races are .055 and .071, not quite statistically significant, but suggestive that election night corruption might have occurred in Liberal-controlled provinces in favor of some Liberal candidates. The results for the United States are similarly suggestive. When comparing ultra-close races (0–1-percent margins of victory) to close races (0–5-percent margins), Democratic candidates, in comparison to Republican candidates are more numerous with a nearly statistically significant p value of .053. Other party candidates and independents have a p value of .002. When the ultra-close category is only 0–0.5-percent margins, the Democrat’s p value is .018 and the other party’s p value is .000. When the close races are considered to be only those with margins of 0–2.5 percent, the Democrat’s p value rises to .088 and

114 Corruption the other party’s p value is .012.8 Incumbency was also statistically significant in some of the regressions, but this result was similar to the Canadian result where there were fewer incumbents winning ultra-close races in contrast to the numbers of incumbents winning close or very close races. These US results strongly support the aggregate finding for the United States in which Democrats and third-party candidates seemed to have a distinct advantage in winning ultra-close races in contrast to the virtually indistinguishable very close races. One potential explanation of this repeated outcome is not that the Democrats are more corrupt than the Republicans, but that the Democrats, with their support bases in more densely populated cities, are more successful when they engage in election night corruption in being able to produce more votes for their victories. For example, Howard Cannon had an advantage in his late-night contest to find more votes because his base was in the more populated Clark County (Las Vegas). Similarly, Democrats in ­Chicago or Philadelphia have many potential votes to work with when padding ballot totals. The curious result of third-party and independent candidates also being overrepresented as winners in ultra-close races can also perhaps be explained by the fact that third-party candidates and independents who win races are likely to be candidates with very strong local organizations and popularity, to allow them to be able to win a race without an affiliation to a national party. Thus, such candidates may have strong local connections that make them well placed to find a few more votes to push them over the top in a close race on election night. These findings also address a curious problem that has emerged in related scholarship on US elections. Studies that use a regression discontinuity analysis which assumes that two outcomes that are very close to each other are essentially randomized on either side of an arbitrary dividing line (for example candidates who win with 50.2 percent of the two-party vote in comparison to candidates who lose with 49.8 percent of the two-party vote) have speculated that these two similar categories of candidates may not be randomly distributed (Caughey and Sekhon, 2011; Eggers et al., 2015). These scholars have found a curious difference in the ultra-close races for the US House of Representatives, after 1945. Caughey and Sekhon (2011) explain the difference as being the result of the advantage of incumbency. Eggers et al. (2015) is skeptical of the incumbency advantage explanation. Both sets of scholars dismiss claims of election fraud because (1) recounts rarely reverse results (Caughey and Sekhon, 2011), (2) this pattern doesn’t appear in other elections and US House races before 1945 (Eggers et al. (2015), and (3) qualitative studies of close US House elections don’t show significant evidence of corruption being reported (Caughey and Sekhon, 2011). Perhaps these analyses are correct, but election night corruption occurs on election night and is less likely to be picked up in a formal recount. In fact, if corrupt election officials have replaced legitimate ballots for candidate A with bogus ballots for candidate B and have altered election records to include serial numbers for the bogus ballots, there is no way that a recount could identify this form of election night corruption.

Corruption  115 Similarly, if an election official has cast bogus votes on behalf of citizens who did not vote, an investigation in which people were contacted about their voting might reveal that election night corruption occurred, but a simple recount of the votes cast would not uncover such corruption. In addition, the absence of election night corruption in some jurisdictions is one of the points of this chapter. Thus, the lack of evidence of such occurrences in different time periods or different jurisdictions should not be taken as proof that it never occurs. My data is consistent with Eggers et al.’s data in finding that incumbency is not correlated with election night corruption, but partisan differences in outcomes rather than incumbency seem to be correlated with election night corruption, a possible explanation that is not considered in these other analyses. At a minimum, the differences described in this chapter have also been noticed by other researchers though there is no agreement as to what the cause for the differences might be.

Differences in election regulations Having shown that results in Japan are free from election night corruption while similar results in Canada and the United States show at a minimum a suggestion of some election night irregularities, I now turn to the rules governing the casting, handling, and counting of ballots in all three countries to find an explanation for Japan’s superior record. A first and obvious explanation is the centralization of Japanese election procedures in contrast to the decentralized nature of election procedures in federal countries such as Canada and the United States. The mere fact that different procedures are used in different jurisdictions means that some of those procedures will be more conducive to corruption than others. Thus, even if the average Canadian election rule is ­better than the one, national regulation that operates in Japan, it is possible and perhaps likely corruption will develop in the few Canadian jurisdictions that have relatively weaker rules. Beyond this systemic factor though, there are also specific rules in place in Japan that appear to be ideally designed to thwart any attempt at election night corruption. The most important of these rules is the requirement that all of the ballots cast in a municipality be collected before the counting of any ballots begins (Public Officers Election Law, Article 65; Ministry of Home Affairs, 1996, 41). This rule is in place to make it difficult to speculate as to how a specific person or group of people voted by not reporting vote totals by small election district units. Rather, all votes are aggregated first and then results are only counted and reported at the municipality level. Thus, there is no such thing as precinct-level electoral data in Japan. In order to accomplish the law’s purpose of obscuring the origins of specific groups of ballots, vote counting does not begin until all the votes are collected from the various polling places in a municipality. In addition, the votes from each of these places are mixed together before counting begins, producing a result that cannot be differentiated by polling place.9 Typically, the municipality will decide on the

116 Corruption time when counting will begin, and the votes are delivered to the counting place from the various polling locations prior to the start time for the vote mixing and counting. This procedure, designed to maximize the secrecy of votes, incidentally makes it much more difficult to engage in election night corruption. The counting of votes does not begin until all of the votes have been cast and delivered to a central location. It becomes impossible for a corrupt election official to create new votes or change existing cast votes in order to generate a few more votes for the favored candidate. The possibility remains that different municipalities in the same district might announce their returns at different times, making it theoretically possible for a corrupt municipality to alter its vote totals after it sees how many votes will be needed to ensure a favored candidate’s victory. Though this potential differential in the announcing of counts among different municipalities exists, in practice, municipalities begin their counts at the same time, which means that all of the ballots have been delivered and secured in a public place prior to any one municipality beginning its count. In addition, different candidates have representatives at each municipality to observe the count, and their presence, along with the public and a large number of people used for the count would make it impossible for one corrupt official in a municipality to ‘find’ additional ballots beyond those previously delivered for the count. In contrast, the counting systems in both the United States and Canada are decentralized with wide variety in counting and ballot-handling rules in the United States among different jurisdictions and smaller variations within the Canadian system. For example, it is fairly common in the United States to have a few precincts still outstanding (the votes not delivered or counted) after all of the other precincts have been counted. Ballot boxes are under the control of poll workers until they are delivered, making it easy for one corrupt worker to tamper with ballots in a ballot box or stored in a machine. Counting is sometimes centralized and sometimes not, with some precincts calling in vote totals and only later delivering the hard ballots, whether they be paper or stored in machines. This process of decentralization with many different people having control over ballots at some point in the process is made even less secure by the growth in early voting and voting by mail. Under these procedures, the time period for which ballots must be kept secure is extended, more people have access to ballots (including anyone who can access the US postal system), and more information is given about vote totals during the extended vote-counting process. This information change has two components. First, votes are counted on election night and publicized, but additional votes will be counted that have timely postal designations but are still not delivered to the election department. It is then possible for a corrupt person to create additional ballots, sufficient to gain victory and insert them into the postal flow of materials as long as the postal designations are created or altered to make the ballots appear timely. A second information component is that many US jurisdictions release daily updates on which voters have requested ballots and which voters have returned their

Corruption  117 ballots. Thus, an unscrupulous campaign operative knows which voters could be targeted for the creation of bogus, last-minute absentee ballots. Canada’s election procedures are standardized at the national level with some provincial discretion in how they are implemented. However, election rules are established at the provincial level with very little discretion at the riding (election district) or municipality level. For example, Canadian elections are administered by a variety of chief electoral officers (at the p ­ rovincial level) and returning officers (at the riding level), and all of these officers are appointed by members of the provincial government, with slight differences between provinces as to who controls the appointment power (Elections ­Canada, 2017). Up until about 2000, though, the district returning officers were appointed by the party that had won the previous election in that riding, and though parties could recommend other officers to help with balloting, the district returning officer had sole discretion over the selection of all of the election workers in that riding (Elections Canada, 2007). In addition, vote counting in Canada is decentralized like the United States, though it is largely uniform (unlike the United States). Vote counting is done initially at the polling location where the results are phoned in to a central location (Elections Canada, 2007). The actual physical evidence of voting is later delivered to the election bureaucracy. Thus, though Canada’s rules are more centralized than the rules in the United States, various opportunities existed to engage in election night corruption, especially given the partisan nature of many of the election officials in the Canadian system.

Suggestions for improving election administration in the United States The lessons learned from Japanese practices are relatively simple. First, information should not be released about partial ballot counts or about partial voting actions until all of the ballots have been collected and they are under the control of the central election bureaucracy where public observation is possible. Second, ballots should not be handled by private parties nor should they be under the control of an election worker who has partisan leanings. Of course all citizens have partisan leanings, but ballots should not be placed under the control of election bureaucrats who are elected or who are appointed by politicians. If an election bureaucracy does not have any non-partisan workers, then ballots should always be under the control of multiple people and should be able to be observed by the public. In practice these lessons suggest a dramatic change in how ballots are handled and counted in US elections. The election night count should be postponed until all of the ballots are collected. No counting should begin until all of the ballots are delivered to the designated counting location. Similarly, election bureaucracies should not release names or totals of people who have either requested ballots or returned ballots before all of the ballots have been collected and the count is completed. It is fine to release this data after the election results are known, but campaigns should not be able to access personal voting

118 Corruption data before the election results are known and announced. US jurisdictions should also radically rethink their use of voting by mail, a method of voting susceptible to corruption (National Commission on Federal Election Reform, 2002, 43–4, 174; Fund and von Spakovsky, 2012, 101–21). Because votes in transit in the mail system are inherently insecure, deadlines for the arrival of votes cast by mail should be prior to the beginning of counting. In other words, votes should not be counted if they are postmarked by a certain day or time; rather votes should only be counted if they are delivered to the election bureaucracy by a certain day or time. Similarly, even if the votes are delivered in a timely manner, third parties should not be allowed to play a role in the distribution, casting, or delivery of ballots, including vote by mail ballots. Japan’s practices also include an additional rule for provisional ballots. In Japan, the validity of all provisional and absentee ballots are determined prior to the counting of ballots, and these ballots are mixed in with the regular ballots prior to the count (Ministry of Home Affairs, 1996, 43). Similarly, in the United States, the determination of whether a provisional ballot will be counted should be made prior to the counting of ballots. Deciding which ballots will be counted after the rest of the votes are counted opens the door to corruption of the process of deciding the status of provisional ballots. These decisions should all be made in advance of the start of any vote counting. Some of the problems of vote by mail could be reduced by adopting a secure electronic system for such votes, similar to what is used by companies, credit cards, and some government agencies such as the Social Security Administration or the Internal Revenue Service (National Research Council, 2006, 58). Even if such practices are adopted, the lesson from Japan remains that any efforts to vote electronically should also be protected by rules that no votes can be delivered after the counting of ballots has begun, and no information should be made public about who has or has not voted (or registered to vote electronically) until after the voting process has been completed, the ballots have been counted, and the results of the election are publicly known. Finally, the controversial practice of ballot harvesting in some US jurisdictions can also be improved by learning from the Japanese example. This example suggests stringent rules on ballot harvesting, making it illegal for partisan operatives to assist, collect, or deliver ballots. Japanese practices suggest that the chain of custody of ballots should be minimized with ballots always in public view and under the control of election officials and members of the public who are assisting in the balloting process or are working as observers.

Notes   1 In 1998, Japan rated a 5.8 on Transparency International’s 10-point scale, making it 25th best, a relatively low rating for an advanced industrial democracy. The rating rose to 17th best in 2006 and 14th best in 2011. In 2017, the rating dropped to a 73 (on a 100-point scale) which is only 20th best.   2 All of these announced tallies were unofficial. The official tally did not come until several days after the election.

Corruption  119   3 A percent measure is superior to an absolute measure because the number of votes that can be created or manipulated using election night corruption increases with the size of the electorate.   4 The Canadian data is for House of Commons races from the 1945 to 2000 elections. The data is from the official parliament website (Parliament of Canada, 2005). The Japanese data covers House of Representatives elections from 1947 to 2003 and is based on a data set originally created by Steven R. Reed and supplemented with data from recent elections. The US data set is an ICPSR data set (1991) that includes all presidential, congressional, gubernatorial, and some state office races, and in some analyses I have included Southern primaries from the ICPSR (1994) data set.  5 p value measurements of statistical significance were .83 (LDP) and .99 (Opposition) in the smaller data set and .86 (LDP) and .25 (Opposition) in the larger data set. Even when all of the data is included, close and not close races, the p values are .50 and .68 respectively.   6 The p value for opposition control for LDP candidates is .055, but this finding only shows that in Japanese prefectures that are controlled by the opposition parties, there are a larger number of ultra-close races (because the comparison group is all Japanese races). It is to be expected that in the few cases in which the opposition controlled the local government that parliamentary elections in those same areas might be closer than in other parts of Japan. In addition, it is hard to otherwise theoretically explain why LDP candidates win more ultra-close races when the opposition controls the local government.   7 Several regional variables showed up as statistically significant in one or the other regressions based on the size of the category of close races (0–2.5-percent margins of victory or 0–5 percent victories).  8 Occasionally a decade variable was also statistically significant, and I grouped states by Mayhew’s (1986, 196) categories of state party system, from traditional to modern, with the expectation that the more traditional party systems (categories 4 and 5) would be more prone to election night corruption than the more modern party systems (categories 1 and 2). The dummy variables for Mayhew’s categories were consistently not statistically significant. I did not calculate a variable for party membership and state control because of the extreme decentralized nature of US elections. I had no reason to expect that the party of the governor would be determinative of election night shenanigans being done in a specific county or city in that same state. There are, also, other possible measures of the potential corruption of a state electoral bureaucracy. Elazar (1972) presents cultural explanations that Meier and Holbrook (1992) reject. Meier and Holbrook, among others, use numbers of prosecutions of public officials. Of these and other measures, Mayhew’s measure had two distinct advantages. First, his measure seemed most relevant to election night corruption rather than general patterns of corruption and dishonesty. Second, his measure was specifically created to cover state politics in the 1960s and even earlier periods, an important fact for this data set which extends back to the 1870s.   9 Public Officers Election Law, Article 66, part 2 says that all of the ballots are mixed together at the counting location before counting begins. ‘kaihyou kanrisha wa, kaihyou tachiainin to tomo ni tougai senkyo ni okeru kaku tohyousho oyobi kinichi mae tohyousho no tohyou o kaihyouku goto ni kondou shite, tohyou o tankan shinakereba naranai.’

120 Corruption

References Alesina, Alberto, and Beatrice Weder. 2002. “Do Corrupt Governments Receive Less Foreign Aid?” American Economic Review 92, no. 4 (September): 1126–37. Anderson, Christopher J., and Yuliya V. Tverdova. 2003. “Corruption, Political Allegiances, and Attitudes toward Government in Contemporary Democracies.” American Journal of Political Science 47, no. 1 (January): 91–109. Becker, David, Daniel Kessler, and Mark McClellan. 2005. “Detecting Medicare Abuse.” Journal of Health Economics 24, no. 1 (January): 189–210. Campbell, Tracy. 2005. Deliver the Vote: A History of Election Fraud, an American Political Tradition—1742–2004. New York: Carroll & Graf. Carlson, Matthew. 2007. Money Politics in Japan: New Rules, Old Practices. Boulder, CO: Lynne Rienner. Carlson, Matthew M., and Steven R. Reed. 2018. Political Corruption and Scandals in Japan. Ithaca NY: Cornell University Press. Caro, Robert A. 1991. The Years of Lyndon Johnson, vol. 2: Means of Ascent. New York: Vintage Books. Caughey, Devin, and Jasjeet S. Sekhon. 2011. “Elections and the Regression Discontinuity Design: Lessons from Close US House Races, 1942–2008.” Political Analysis 19, no. 4 (Autumn): 385–408. Curtis, Gerald L. 1971. Election Campaigning Japanese Style. New York: Columbia University Press. Dallek, Robert. 1991. Lone Star Rising. New York: Oxford University Press. Davis, Charles L., Roderic Ai Camp, and Kenneth M. Coleman. 2004. “The Influence of Party Systems on Citizens’ Perceptions of Corruption and Electoral Response in Latin America.” Comparative Political Studies 37, no. 6 (August): 677–703. Duggan, Mark, and Steven D. Levitt. 2002. “Winning Isn’t Everything: Corruption in Sumo Wrestling.” American Economic Review 92, no. 5 (December): 1594–605. Eggers, Andrew C., Anthony Fowler, Jens Hainmueller, Andrew V. Hall, and James M. Snyder, Jr. 2015. “On the Validity of the Regression Discontinuity Design for Estimating Electoral Effects: New Evidence from over 40,000 Close Races.” American Journal of Political Science 59, no. 1 (January): 259–74. Elazar, Daniel J. 1972. American Federalism: A View from the States. New York: Thomas Y. Crowell. Elections Canada. 2007. Interview with Elections Canada official. By phone. August 17. Elections Canada. 2017. Compendium of Election Administration in Canada: A Comparative Overview. August 30. www.elections.ca/content/aspx?section=res&document= index&dir=loi/com/arc/com2017&lange=e. Accessed March 1, 2018. Fund, John. 2004. Stealing Elections: How Voter Fraud Threatens Our Democracy. San Francisco, CA: Encounter Books. Fund, John, and Hans von Spakovsky.. 2012. Who’s Counting? How Fraudsters and Bureaucrats Put Your Vote at Risk. New York: Encounter Books. Garicano, Luis, Ignacio Palacios-Huerta, and Canice J. Prendergast. 2005. “Favoritism under Social Pressure.” Review of Economics and Statistics 87, no. 2 (May): 208–16. Gerring, John, and Strom C. Thacker. 2004. “Political Institutions and Corruption: The Role of Unitarism and Parliamentarism.” British Journal of Political Science 34, (April): 295–330.

Corruption  121 Gerring, John, and Strom C. Thacker. 2005. “Do Neoliberal Policies Deter Political Corruption.” International Organization 59, no. 1 (January): 233–54. Golden, Miriam A., and Eric C. C. Chang. 2001. “Factional Conflict and Political Malfeasance in Postwar Italian Christian Democracy.” World Politics 53, no. 4 (July): 588–622. Imai, Kosuke, and Gary King. 2004. “Did Illegal Overseas Absentee Ballots Decide the 2000 US Presidential Election?” Perspectives on Politics 2, no. 3 (September): 535–50. Inter-university Consortium for Political and Social Research (ICPSR). 1991. Southern Primary Candidate Name and Constituency Totals, 1920–1972 [computer file]. ICPSR. ed. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor]. Inter-university Consortium for Political and Social Research (ICPSR). 1994. Candidate and Constituency Statistics of Elections in the United States, 1788–1990 [computer file]. 5th ICPSR. ed. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor]. Jacob, Brian A., and Steven D. Levitt. 2003. “Rotten Apples: An Investigation of the Prevalence and Predictors of Teacher Cheating.” Quarterly Journal of Economics 118, no. 3 (August): 843–77. Johnson, Chalmers. 1995. Japan: Who Governs? The Rise of the Developmental State. New York: W. W. Norton. Kallina, Edmund F. 1988. Courthouse over White House: Chicago and the Presidential Election of 1960. Orlando, FL: University of Central Florida Press. Mayhew, David R. 1986. Placing Parties in American Politics: Organization, Electoral Settings, and Government Activity in the Twentieth Century. Princeton, NJ: Princeton University Press. Mebane, Walter R., Jr. 2006. “Detecting Attempted Election Theft: Vote Counts, Voting Machines, and Benford’s Law.” Paper presented at the Midwest American Political Science Association annual convention, Chicago, IL. Meier, Kenneth J., and Thomas M. Holbrook. 1992. “‘I Seen My Opportunities and I  Took ’em:’ Political Corruption in the American States.” Journal of Politics 94, no. 1 (February): 135–55. Ministry of Home Affairs. 1996. Election System in Japan. Tokyo: Jichi Sogo Center. National Commission on Federal Election Reform. 2002. To Assure Pride and Confidence in the Electoral Process: Report of the National Commission on Federal Election Reform. Washington, DC: Brookings. National Research Council. 2006. Asking the Right Questions about Electronic Voting, Computer Science and Telecommunications Board. Washington, DC: National Academies Press. Parliament of Canada. 2005. “Library of Parliament: History of the Federal Electoral Ridings since 1867.” www.parl.gc.ca/information/about/process/house/infer/infer. asp?Language=E. Accessed August 2, 2005. Sabato, Larry J., and Glenn R. Simpson. 1996. Dirty Little Secrets: The Persistence of Corruption in American Politics. New York: Times Books. Simpser, Alberto. 2013. Why Government and Parties Manipulate Elections: Theory, Practice, and Implications. New York: Cambridge University Press. Transparency International. 2019. “Corruption Perceptions Index 2017.” https:// transparency.org/news/feature/corruption_perceptions_index_2017. Accessed 16 January 2019.

122 Corruption Treisman, Daniel. 2000. “The Causes of Corruption: A Cross-national Study.” Journal of Public Economics 76, no. 3 (June): 399–457. Woodall, Brian. 1993. “The Logic of Collusive Action: The Political Roots of Japan’s Dango System.” Comparative Politics 25, no. 3 (April): 297–312. Xin, Xiaohui, and Thomas K. Rudel. 2004. “The Context for Political Corruption: A Cross-national Analysis.” Social Science Quarterly 85, no. 2 (June): 294–309.

7 Democracy and political instability

Lesson 7: When voters are less loyal to political parties or candidates, politics can become more unstable and governments less effective.

Introduction In the five years from 2006 to 2012, Japan had six different prime ministers, with each one serving in office for about one year. This period of volatile leadership straddled both the rule of the LDP and Democrats. It was preceded and followed by two popular LDP prime ministers, Koizumi Junichiro and Abe Shinzo, who both enjoyed higher popularity, won successful reelection campaigns, and stayed in office for five years or longer. Ironically, Prime Minister Abe initially was one of the six failed prime ministers. He followed Koizumi Junichiro into the office in 2006, but he resigned unexpectedly after only twelve months in office. Abe was followed by Fukuda Yasuo, less flamboyant than Abe and son of a 1970s-era prime minister. Fukuda surprised commentators when he also resigned after only twelve months in office. Aso Taro, grandson of an LDP prime minister of the 1950s, was next, and some saw him as a charismatic politician who could lead the LDP to victory in the elections that it faced in 2009. Aso, however, failed to significantly revive LDP popularity, and after the landslide victory of the Democrats, Hatoyama Yukio, leader of the Democratic Party, became the new prime minister. Hatoyama was the only one of these six prime ministers chosen by voters in an election rather than by party leaders. However, even the landslide victory of the Democrats didn’t protect Hatoyama when he resigned nine months later after a series of scandals and missteps. His replacement was Kan Naoto who lost popularity after proposing a tax increase and then led what was perceived to be an inadequate response to the 2011 East Japan Earthquake. Kan resigned after fifteen months in office, and his replacement was Noda Yoshihiko. Noda pushed the tax increase into law and suffered terribly at the hands of the voters, leading the Democrats to a spectacular loss in 2012 and paving the way for Abe Shinzo’s return to the prime minister’s office in 2012. Abe cemented his hold on the office by leading the LDP to another landslide

124  Democracy and political instability victory, after calling an early election in 2014. He did a repeat performance, leading the LDP to a third consecutive landslide victory in the 2017 elections. It is fascinating that Koizumi’s five years in office (2001 to 2006) and Abe’s seven years and counting (2012 to 2019) should bracket an equally long sixyear period during which Japan had not one but six prime ministers. Japan’s political instability from 2006 to 2012 is a rare, but occasional, phenomenon in other countries. For example, France had 21 prime ministers in the 12 years of its Fourth Republic (1947–58). Though constitutionally it is difficult for political leaders in the United States to change so frequently, midterm elections in the United States suggest a related type of volatility. Three of the past four US presidents saw their parties lose control of the House of Representatives in their first midterm election. For example, President Obama posted a win in 2008 that was the best performance of a Democrat in 44 years, yet two years later Republicans won the greatest share of House seats for their party in 64 years of campaigns. President Obama returned to the ballot in 2012 and again earned the highest share of the popular vote of any Democratic candidate in 48 years, with the exception of his previous 2008 victory. The Republicans, responded in kind, winning in 2014 the greatest number of House seats for their party in 86 years.1 Japan’s recent instability, however, stands out because it is in such stark contrast to the extreme political stability that characterized Japanese politics for a half century in the second half of the 1900s. The LDP or its predecessors ruled Japan continuously from 1945 to 2015 with only two brief interruptions: ten months in 1993–4, and the Democrats’ three-year rule from 2009–12.2 Unchanging LDP political domination over five decades created a record unmatched in any true democracy, a record that makes the period of ­revolving-door prime ministers even more puzzling. This instability also stands in contrast to the relatively stable politics of both Koizumi’s and Abe’s tenures. What has changed in Japan to make political instability more common? What is it about certain prime ministers that make it possible for them to craft stable politics when otherwise political volatility is the norm? The contrast with past decades is especially telling when politics was stable, regardless of who was the prime minister. Additionally, not only does stability now seem to depend on who is the prime minister, but even the same person in that office can be ineffective as Abe was in 2006–7 and effective as Abe has been from 2012 to 2019. Japan’s recent political volatility is shown not only by the rotation of prime ministers in and out of office with stunning rapidity but also by dramatic shifts in voter preferences. Koizumi engineered a landslide election victory for the LDP in 2005 only to have the voters give an unprecedented landslide victory to the opposition Democrats in 2009. The pendulum then swung back with the LDP winning landslide victories in 2012, 2014, and 2017. Japan’s contrasting records of political stability and instability point to an often-ignored weakness of democratic politics in any country: only some voters are informed and carefully weigh political issues before voting (Achen and Bartels, 2016). The political impact of the small number of issue voters can easily be outweighed by the voting decisions of low information/low interest

Democracy and political instability  125 voters. These other voters take any of a number of alternate routes to their voting decisions: the recommendations of friends, family, or important interest groups; partisan cues; candidates who share the same gender, religion, race, or ethnicity; the genuineness or likeability of a candidate or a party leader; television ads; debate skills; significant political gaffes or scandals; and even the height or physical attractiveness of a candidate. In past decades, Japanese politics was stable, in part, because many Japanese voted based on their strong personal ties to individual candidates. These personal ties to candidates were never praised as a positive feature of Japanese democracy. Rather they were decried as the remnants of a feudal system or the corrupt effects of political system based on pork barrel politics. For example, the report of the 8th Commission on Electoral Reform made this perspective clear in its listing of party-centered and issue-centered elections as its goal for election reforms. The report blamed candidate-centered elections on the ­multi-member election system used in Japan until 1994 because that system forced candidates from the same party to run against each other in the same election district (Munetaka, 1997, 389–90). Reformers used these criticisms as justifications to change Japan’s electoral system. They wanted to make issues more central to the decisions of voters, reducing the importance of both pork barrel politics and candidate loyalty in Japanese elections (Christensen, 1994, 602; Reed and Thies, 2001, 155). Unfortunately, however, the weakening of voter loyalties to candidates has not improved Japanese politics in the way that some reformers intended. It is true that parties have become more relevant to the political process for several reasons: (1) voters now cast a ballot for a party as well as a candidate, (2) candidates from the same political party no longer run against each other in the same election district, (3) government subsidies flow to parties not to candidates, and (4) restrictions on campaign activities are looser on parties than they are on candidates. These changes appear to have weakened, though not eliminated, candidate-centered campaigns and personal support groups of candidates (Christensen, 1998; Krauss and Pekkanen, 2011). Despite these changes, though, there is no evidence that political reforms in Japan have caused the Japanese to study political issues more earnestly now than they did under the previous election system (Kobayashi, 2012, 97–110). Rather, for many Japanese voters, one set of factors (loyalty to a particular candidate) has been replaced by another set of factors (candidate or party leader appearance, speaking skills, portrayal in television advertising, notable gaffes or scandals, etc.). Thus, the net effect of the reforms has not been to create a more careful or thoughtful Japanese electorate, but rather to create a more volatile electorate, one that punishes politicians for making difficult decisions and rewards politicians who pander to the voters. Unfortunately for the nation, the most popular position in the short term may not be the best policy for the nation in the long term. Often popular political positions serve the short-term interests of voters while ignoring long-term problems that a nation faces (Prisching, 1995).

126  Democracy and political instability The Japanese experience also seems to echo similar changes in European and American voting patterns. The lodestone of voting in such long-­established democracies has been party loyalty rather than candidate loyalty, but as party loyalty has weakened, other factors have taken a more prominent role in voting (Renwick and Pilet, 2016). Thus, politics has also become more volatile (Converse and Depeux, 1962; Dalton, McAllister, and Wattenberg, 2000), and it is also becoming harder for politicians to take actions that are necessary but that would hurt their popularity with voters. Several factors, however, make the US case less severe than the current Japanese situation. First, though weaker than they used to be, party attachments are still quite strong in the United States. Second, American politicians serve set terms of office, so an unpopular American president will serve out her term rather than be shuffled out of office in favor of a new face who might lead the party to electoral victory. Third, Japan has more frequent elections than the United States, increasing the political need to always be appealing to increasingly fickle voters. Fourth, US voters also have stronger loyalties to candidates and parties based on race, religion, and ethnicity. For example, voter surveys done in Japan by Akarui Senkyo (2018, 50) don’t even list religious, gender, race, or ethnicity similarities between candidates and voters on their surveys while they do list age and occupational similarities between candidates and voters. These factors in the United States mute rising volatility there, a rise that nevertheless seems to be occurring in both Japan and the United States. The Japanese experience tells an important, cautionary tale: sometimes the cure for a political problem can be worse than the disease. In this chapter, I examine several complex political issues: voter volatility, candidate popularity, and voter motivations for supporting candidates. In addressing these issues, I compare Japan to the United States and other similarly situated democracies. These complex issues play out in any functioning democracy, so unlike some of the issues addressed in other chapters, there is no need to limit this analysis to only countries that have, for example, single-­ member districts or strong candidate-centered voting. In addition, the arguments identified in this chapter are certainly some of the factors that produce these complex outcomes, but I do not claim that they are the exclusive reasons for the observed phenomenon. For example, the bulk of this chapter shows how volatility has increased in Japan, and evidence is presented that candidate loyalty has declined. I then discuss the consequences of these changes without ever explaining why candidate loyalty has declined or why volatility has increased.

Volatile voters I begin with a comparative discussion of whether or not politics in Japan and elsewhere is becoming more volatile. The starting point for this inquiry is the behavior of voters in elections. Three of Japan’s recent elections have shown unprecedented swings in the preferences of the voters. In 2005 the LDP won its highest share of the popular vote (49.5 percent) since 1963 and its second

Democracy and political instability  127 highest number of seats (61.7) in the Party’s 50-year history.3 Four short years later, the LDP won its second worst share of the popular vote in its history (40.1 percent), and the opposition Democrats won the greatest percent of seats (64.2) ever won by any political party in Japan from 1945 to 2019. The Democrats also won the highest share of the vote (52.0 percent) won by any political party in 54 years. The Democrats then suffered the most dramatic drop in popular support of any party in Japan’s history (26.6 percent), while the LDP rebounded almost to the record level of support enjoyed by the Democrats in 2009: the LDP won nearly 50 percent of the vote in both 2014 and 2017 and 61 percent of the seats in 2012, 2014, and 2017.4 Figure 7.1 puts recent volatility, a 26.6 percent drop in the Democrats’ vote (2012) and a 9.4 percent drop in the LDP’s vote (2009), in historical perspective.5 Even so, the smaller 9.4 percent swing in the LDP’s vote is a rare occurrence in any established democracy. For example, Barack Obama won the US presidency in 2008 with 52.9 percent of the popular vote, the highest percentage won by a Democratic candidate for the presidency since 1964. In fact, only Lyndon Johnson, Franklin Roosevelt, and Andrew Jackson have won a higher percentage of the popular vote as Democrats in 180 years of US presidential elections. However, Obama’s impressive performance was less than a 5 percent increase over John Kerry’s 48.3 percent of the vote.6 Similarly, only one British and one German election since 1945 have shown increases in support of the winning political party of greater than 9.4 percent, and both of them were in the chaotic aftermath of World War II. In the 1945 British election, the Labour Party increased its share of the vote by 11.7 percent, and in 1953 the Christian 70

60

Percent of vote

50 40 30

LDP Vote Percent

20

Soc./Dem Vote

10

1958 1960 1963 1967 1969 1972 1976 1979 1980 1983 1986 1990 1993 1996 2000 2003 2005 2009 2012 2014 2017

0

Election year

Figure 7.1  Comparative vote percentages for the LDP and the Socialist/Democratic opposition

128  Democracy and political instability Democrats in Germany improved on their performance in the first postwar election by 11.2 percent. In contrast, Margaret Thatcher’s victory in 1979 and Tony Blair’s victory in 1997 were both accomplished by an increase in the popular vote of less than 9 percent. The French UPM suffered a 12 percent drop in its share of the first-round vote from 2007 to 2012, but that decline reduced to less than 9 percent in second-round voting. Though Canada has had a few elections in which the winning party increased its share of the vote by more than 9.4 percent, even in 1993 when the Conservatives went from being the ruling party to winning only two seats in the election, the winning Liberals increased their share of the vote by only a similar 9.3 percent.7 There are, of course, more systematic ways of measuring political volatility than vote swings between just two elections or the numbers of prime ministers who have served in a given time period. The most widely recognized measure examines the extent to which voters change their support from one party to another across a series of elections. Pedersen (1979) developed this measure, adding together all the changes in party support from one election to the next and then dividing that total by 2. In a simple two-party system in which a party’s support shifted from a 49–51 loss in one election to a 53–47 victory in a subsequent election, Pedersen’s measure would add that party’s 4 percent gain to the other party’s 4 percent loss and divide the sum by 2 for a total volatility number of 4 percent. Of course, party mergers, splits, and name changes must be accounted for in calculating the measure so that these changes do not artificially inflate Pedersen’s measure. Bartolini and Mair (1990) make these adjustments and calculate political volatility for many Western European nations for most of the 1900s. I have made similar calculations for Japan for all elections since 1945 and have reported the Japanese results against Bartolini and Mair’s results for Western Europe in Figure 7.2.8 Figure 7.2 shows Japan’s two periods of political instability that bracket more than thirty years of political stability. My point of comparison is Bartolini and Mair’s (1990), Mair’s (2002, 131), and several other authors’ averages for groupings of European countries holding elections from 1945 to 2017 (designated as ‘Europe average’ in Figure 7.2). In the elections held from 1945 to 1985, European countries had only 4 percent of their elections with a volatility measure greater than 20 (Bartolini and Mair, 1990); in contrast Japan had 20 percent of its elections (1945 to 2014) with that same high level of volatility.9 Similarly, Japan had 24 percent of its elections (six elections) with a volatility level below 4 while in Europe there were only 18.8 percent of elections (57 elections) with that low level of volatility. Thus, Japan has the odd distinction of having both more volatile and less volatile elections than comparable elections in European countries from 1945 to 1985. Moving to include more recent elections, volatility has increased in both Japan and Western Europe, with the stable European average of around 8 percent in earlier decades rising to 12.6 in the 1990s (Mair, 2002). Baldini and Pappalardo (2009, 139) find a similarly higher volatility average of 11 percent from 1990 to 2007. Dalton, McAllister, and Wattenberg’s (2000, 41)

Democracy and political instability  129 35 30

Volatility

25 20

Japan Districts Japan PR List

15

Europe Average 10

2014

2009

2003

1996

1990

1983

1979

1972

1967

1960

1955

1947

0

1952

5

Year

Figure 7.2  Electoral volatility in Japan

study of volatility in OECD countries in the 1990s also shows a similar average of 12.9. Renwick and Pilet (2016, 3) corroborate these numbers in Europe for earlier decades and extend Europe’s average to 16 percent in the first half of the 2010 decade. These numbers mirror Japan’s increasing volatility over the same time period, though Japan’s increase is larger. Other countries also have higher levels of volatility. Roberts and Wibbels (1999) report an average volatility in Latin American countries of 23.2. Two separate studies of Eastern Europe find average volatility ratings of 32.3 and 25.9 (Tavits, 2008, 551; Sikk, 2005, 396). It is, however, not surprising that new democracies such as those in Eastern Europe would have higher levels of volatility. Japan’s first decade of democracy (1945–55) also had high levels of volatility, similar to levels in the new democracies of Eastern Europe or the postwar elections in Britain and Germany. Voters in new democracies such as Japan in the 1940s and 1950s and Eastern European voters in the 1990s are not set in patterns; politicians are not members of political parties with long histories and traditions. Not only are the politicians more willing to create new political parties but also the voters are more willing to support new political parties. For these reasons, the high levels of volatility in the new democracies of Eastern Europe or in Japan’s first postwar decade (average volatility of 19.9) are not that surprising. The data reported in Figure 7.2, though, seem to contradict Bartolini and Mair’s (1990) claim that recently high levels of volatility in Europe are comparable to similarly high historical levels of volatility. Of course, their comparison of rising volatility ended in 1990 before much of the increase in European volatility occurred. Dalton, McAllister, and Wattenberg (2000) argue, in contrast,

130  Democracy and political instability that levels of volatility are rising in most of the established democracies. There is, however, some truth in both claims, and the Japanese data supports both claims. Volatility is increasing in Japan as Dalton, McAllister, and Wattenberg and others claim, but there are also, at least in some countries, historical periods of extremely high political volatility, consistent with Bartolini and Mair’s claim. Though there are some differences in how party mergers and divisions are counted which calls for some caution in making comparisons across these different studies,10 the Japanese trend towards greater instability seems corroborated in other nations, consistent with Dalton, McAllister, and Wattenberg’s and Renwick and Pilet’s (2016) findings. Thus, Japan’s current and past experiences with high volatility are not unique to Japan. Japan’s experiences, therefore, can shed light on political instability generally, especially as it occurs in advanced industrial democracies.

Volatile politicians One lesson from Japan’s record of volatility is how politicians can contribute extensively to voter volatility. When political parties are weak and candidates are strong, as is the case in Japan, individual politicians face few barriers to shifting and remaking political groupings. Lin (2016) additionally argues that the post-reform frequency of coalition governments in Japan also gives smaller parties more bargaining power which encourages party splinters. When politicians change parties, political volatility increases, even if the voters remain constant in their political preferences. Though not the only explanation of high voter volatility levels in Latin America, some Latin American countries have also had periods of extreme elite-generated volatility. For example, Mainwaring (1991, 28; 1999, 37, 142–7) points out that members of the Brazilian Congress on average had been members of 2.9 parties. Of those elected in 1986, one third switched parties before the next election. Mainwaring also finds similar levels of volatility in Ecuador and Russia (1999, 37). Of course weak political parties do not necessarily produce high levels of elite volatility. Japan’s political parties have been consistently weak, but there have been only three periods of politicians switching parties, 1945–55, 1993–2000, and 2012–17. In contrast, from 1955 to 1993, Japanese politicians rarely switched parties, and party labels were stable. This insight explains part of the puzzle of contrasting periods of political stability and instability in Japan. The most significant jumps in volatility in Japan were largely because of politicians switching parties. For example, the LDP’s biggest drop in its share of the vote came in 1993, but this drop is largely explained by the defection of 46 LDP MPs to two new political parties immediately prior to the 1993 election. When these candidates ran in districts and split the conservative vote, the LDP share of the vote necessarily dropped. After these defections, however, the LDP contested the election with only 222 incumbents remaining on the party label, and the Party won 223 seats, for a net gain of one seat. Thus, the LDP didn’t lose the 1993 election because

Democracy and political instability  131 % candidates switched parties and volatility

80 70

60 50 40

Switched Parties

30

Volatility

20 10

17

12

20

05

20

00

20

93

20

86

19

80

19

76

19

69

19

63

19

58

19

53

19

19

19

49

0

Election year

Figure 7.3  Party switching and volatility compared, in Japan

voters abandoned the LDP, rather it lost the election because so many of its ­incumbents switched parties away from the LDP. Figure 7.3 makes a strong case that much of Japan’s political instability can be explained by politicians switching party labels. The three periods of high volatility in Japan correspond to large numbers of politicians changing their party affiliation from the previous election. This pattern shows up even though party mergers, divisions, and name changes are not included when calculating the volatility measure. Even excluding such changes in volatility calculations, there still is a strong correlation between the remaining volatility that is counted and party switching by politicians. It is noteworthy, however, that, in the most recent period of volatility (2012 to 2017), politicians switching parties lags behind voter volatility, in contrast to the earlier periods in which elite volatility was greater than voter volatility. Despite its prominence as a potential explanation of political instability in places as diverse as Brazil, Japan, Ecuador, or Russia, elite volatility is rare in most established democracies. Examples of elite volatility in Western Europe or North America stand out precisely because they are so rare. Jim Jeffords’ party switch gave the Democrats control of the US Senate in 2001, and Joseph Lieberman won reelection to the Senate as an independent after previously running on the Democratic presidential ticket, but these cases are the exception rather than the rule. Japan provides a convenient window into understanding elite volatility because in some decades it has tiny numbers of party switchers similar to most advanced industrial democracies and in other decades it has fluid party politics similar to Brazil, Ecuador, or Russia.

132  Democracy and political instability

Volatility in public opinion Having explained much of Japan’s occasional periods of volatility as being the result of candidates switching parties, it remains to be seen if voters have also become more volatile in their party preferences, even in elections when party labels have been consistent. One way to answer this question is to examine voter attitudes as recorded in public opinion polls and see if these polls also record significant swings in popular support for parties or politicians, even when party labels have been stable. A first measure to check is the approval rating of the prime minister, and this measure does show increasingly volatile political attitudes in recent decades, at least when a popular prime minister hasn’t been in power for a long period of time. The amount of change in month-to-month ratings has increased significantly since the mid-1990s, with the exception of Koizumi’s and Abe’s long terms in office. In addition, Japanese parties have become less tolerant of party leaders with low approval ratings than they were in previous decades. Thus, though politicians may get low approval ratings as often now as they did in past decades, the descent to those low ratings is often quicker, and parties now often respond more quickly to these low approval ratings. A party leader with low approval ratings is much more likely to be jettisoned by others in the party, and this change has increased Japanese political instability. After Koizumi Junichiro left office with a 43 percent approval rating, the next five prime ministers fared much worse, leaving office with abysmal approval ratings: Abe 26, Fukuda 16, Aso 13, Hatoyama 19, and Kan 13 percent. In fact, each also saw a dramatic drop in approval: Abe 36 percent in ten months, Fukuda 28 percent in eleven months, Aso 28 percent in ten months, Hatoyama 42 percent in seven months, and Kan 23 percent in nine months (Jiji tsushinsha, n.d., 1981, 1992, 2011; Japor, 2011; Roper Center, 2011). Comparing these figures to the similarly recorded approval ratings for US presidents shows how unpopular Japan’s recent prime ministers have been and how volatile their support has also been. In nearly sixty years of Gallup (2019) opinion polling, the lowest recorded number for a US president is 24 percent (Nixon), yet four of these five prime ministers of Japan have left office with lower approval ratings than even Richard Nixon. Similarly, only the two most recent US presidents have seen dramatic changes in their approval ratings during their first year in office: Obama’s approval dropped 28 percent and Bush’s rose 27 percent. In contrast, most other presidents have had relatively stable approval numbers during their first year: Trump −4, Clinton −4, Bush +10, Reagan −4, Carter −14, Ford −21, Nixon +3, Johnson −9, Kennedy +7, and Eisenhower +3 percent. Figure 7.4 presents this comparison visually, showing the gradual decline in popularity during the last four years of the Bush administration and the first three years of the Obama administration along with the gradual increase in Obama’s rating in the last years of his presidency. These gradual trends ­contrast sharply with the high peaks and low valleys of

Democracy and political instability  133 80

Percent approving

70 60 50 40 30 20

10 2006 2006 2006 2007 2007 2008 2008 2008 2009 2009 2010 2010 2011 2011 2011 2012 2012 2013 2013 2013 2014 2014 2015 2015 2016 2016 2016 2017 2017 2018 2018 2018

0

Japan

United States Year

Figure 7.4  Prime minister/presidential approval ratings in Japan and the United States

approval ratings for each of Japan’s five prime ministers who served between Koizumi and Abe from 2006 to 3013. The comparison in Figure 7.4, however, is misleading because approval ratings change dramatically with each new president or prime minister. Thus, with five new prime ministers, approval ratings will be much more volatile in Japan than over the same period in the United States during which there is only one change in the president. A better comparison would examine periods of the same length in both countries during which the president or prime minister did not change. In Figure 7.5 I have selected every one-year period in both countries during which the political leader did not change. I have calculated the standard deviation of approval ratings taken at one-month intervals during each one-year period. The standard deviation essentially measures how much the approval figures vary from month to month during each one-year period. Low standard deviations suggest that the approval numbers are consistent across the one-year period; high standard deviations show much greater volatility in the month-to-month approval numbers. Figure 7.5 shows jumps in the volatility at expected times. Presidential approval ratings in the United States were more volatile from the Watergate era to the beginning of the Reagan administration, during the first Iraq war, and after 9/11. In Japan the most prominent spike is around the 1972 scandals of Tanaka Kakuei. Then begins a sustained increase in volatility, beginning in the late 1990s and only reduced by the long periods of high approval ratings of both Koizumi and Abe. For much of the period, US and Japanese approval ratings seem equally volatile, but Japan’s volatility has risen to higher levels for the past two decades, with some exceptions. Perhaps Japan is going through a simply greater period of volatility as the United States did in the 1970s, but it is also possible that Japan is entering a period of higher volatility, and the

134  Democracy and political instability

Standard deviation of approval ratings

18 16 14 12 10 Japan

8

US

6 4 2

17

14

20

11

20

08

20

05

20

02

20

99

20

96

19

93

19

90

19

87

19

84

19

81

19

78

19

75

19

72

19

69

19

66

19

63

19

19

19

60

0

Year

Figure 7.5 Variation in approval ratings for one-year periods in Japan and the United States with the same prime minister or president 90 High and low approval ratings

80 70 60 50 40

High

30

Low

20 10 Ikeda Sato Tanaka Miki Fukuda Ohira Suzuki Nakasone Takeshita Kaifu Miyazawa Hosokawa Murayama Hashimoto Obuchi Mori Koizumi Abe Fukuda Aso Hatoyama Kan Noda Abe

0

Prime minister

Figure 7.6  Highs and lows in approval rating for Japanese prime ministers

record of approval ratings suggests that this period is not just limited to a few elections but started in the 1990s. The Japanese data suggests a second point about approval ratings that is relevant to the changing nature of Japan’s politics. Though approval ratings were very low (by United States standards) for the five prime ministers from 2006 to 2011, in a historical comparison, many of Japan’s past prime ministers have also had similarly low levels of public approval. Figure 7.6 shows

Democracy and political instability  135 that the highs and lows of recent prime ministers in Japan do not differ dramatically from the highs and lows of past prime ministers. Though two of the four most popularly rated prime ministers have been during the past two decades, four past prime ministers left office with even lower approval ratings than the five prime ministers of the 2006–11 period. Figure 7.6 most strongly makes the point that poor approval ratings for prime ministers are nothing new in Japan. Table 7.1, however, tells a more nuanced story of the history of low approval ratings. Past prime ministers often suffered low approval ratings in the 20 percent range without being removed from office. The four that eventually dropped below 20 percent were removed from office quickly (Tanaka, Takeshita, Miyazawa, and Mori), but even three of these four had won reprieves from earlier bouts of unpopularity and continued to serve in office until their unpopularity became overwhelming. Recent prime ministers have dropped a little bit further and more quickly than their predecessors. There have also been fewer reprieves than occurred in the past, with only two of the recent prime ministers managing short extensions after hitting a low in approval ratings. Aso was the fourth LDP prime minister to hold office since the 2005 elections, and elections had to be held within seven months of when Aso’s popularity plummeted. LDP leaders decided it better to wait and see if Aso could improve his ratings rather than switch to a fifth LDP prime minister right before the election. Similarly, Kan benefited from the sense of crisis surrounding the aftermath of the East Japan Earthquake, directly using the crisis to justify his staying in office until crucial legislation had been passed. In contrast, Koizumi’s popularity dropped only once to only 36 percent, and Abe’s drop to 30 percent was anomalous—his popularity rebounded to an average of 40 percent in the following four months. Table 7.1  Troughs in Japanese prime minister approval ratings and party responses to those troughs Prime minister

First low approval rating (percentage)

Consequence

Decline from high to low (points)

Ikeda

33 (June 1961)

11

Sato

26 (December 1966)

Tanaka

21 (May 1973)

Miki

23 (March 1976)

Fukuda

24 (March 1977)

Ohira

21 (October 1979)

In office 3.3 more years In office 7.5 more years In office 1.5 more years In office 9 more months In office 1.6 more years In office 9 more months

21 35 19 6 14 (continued )

136  Democracy and political instability Table 7.1  (continued ) Prime minister

First low approval rating (percentage)

Consequence

Decline from high to low (points)

Suzuki

28 (July 1981)

14

Nakasone

27 (March 1983)

Takeshita

24 (December 1988)

Uno

10 (July 1989)

Miyazawa

21 (April 1992)

Mori

10 (March 2011)

Abe

23 (August 2007)

Fukuda

19 (June 2008)

Aso

16 (February 2009)

Hatoyama

19 (May 2010)

Kan

18 (February 2011)

Noda

22 (April 2012)

Abe

30 (July 2017)

In office 1.3 more years In office 4.5 more years In office 5 more months Replaced immediately In office 1.3 more years Replaced next month Replaced next month In office 3 more months In office 7 more months Replaced immediately In office 6 more months In office 8 more months In office 1.6 more years

8 17 10 27 23 28 25 23 42 23 21 31

Thus, the slumps in approval ratings for recent prime ministers have been only a little bit steeper and deeper than comparable slumps of previous prime ministers. The crucial difference is that recent prime ministers have been consistently replaced when their approval has plummeted in contrast to past prime ministers who often stayed in office despite dips in their approval ratings. Perhaps influential MPs are much more sensitive to the electoral effect of having an unpopular party leader in office when facing an election, and they replace party leaders immediately when they first become unpopular (Krauss and Nyblade, 2005). A second measure of the volatility of public opinion, one that is especially linked to the idea of weakening ties to parties and candidates, is the number of Japanese who fail to identify with a particular political party. If Japanese voters are becoming less partisan and less tied to a particular candidate, these numbers should be increasing over time. Figure 7.7 shows the yearly averages for both Japan and the United States for those who report no political affiliation in monthly opinion polls for the past 70 years, excluding independents who lean towards a political party (Keith et al., 1992).

Democracy and political instability  137

Average percent of unaffiliated voters

80 70 60

50 40

Japan

30

United States

20 10

1947 1950 1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019

0

Year

Figure 7.7   Unaffiliated voters (yearly averages) in Japan and the United States, 1947–2019

Japan had more independents than the United States during the turbulent first postwar decade, but then was similar to the United States in the latter 1950s and 1960s. While the number of independents increased gradually in the United States, Japan’s numbers jumped dramatically in the late 1960s and early 1970s and then again in the 1990s. The United States passed a milestone in 2011 with the highest annual average for independents surpassing the 40 percent mark (Jones, 2011), but Japan passed that milestone nearly forty years earlier in 1974 and posted its own new record for a monthly high for unaffiliated voters in July, 2012: 71.4 percent. Though some specific periods of volatility in countries such as Japan can be explained by the newness of the democratic system or elites changing parties, no obvious explanation presents itself for the gradual rise in volatility seen in Japan and other countries. Nevertheless, this change is occurring, and it is affecting politics in these countries.

Volatility in turnout Public opinion polling has shown both that prime minister approval ratings are more volatile in Japan (in recent decades) and the number of unaffiliated voters in Japan is higher (for the past forty years) than comparable numbers for the United States. This well-documented evidence, however, does not measure whether Japanese voters are also becoming less loyal to individual candidates. This question, however, is asked in public opinion polling in Japan, showing mixed results. Nemoto and Tsai (2016, 165–6) show Japanese support for candidates dropping from about 60 percent in the 1960s to about 50 percent on the eve of electoral reforms in the 1990s. Polling by Akarui Senkyo

138  Democracy and political instability (2018) shows that in 1969, 49 percent of Japanese cast votes based on party affiliations and 33 percent voted for candidates, but in 2018, the respective numbers were almost unchanged: 48 percent voting according to party and 32 percent voting according to the candidate. Their election surveys show that candidate-oriented voting declined in elections from 1996 to 2009, but from 2012 to 2018, candidate-oriented voting has posted small increases. Polling data, however, on this specific question is perhaps not entirely reliable because many voters are unlikely to admit that their vote is determined by their relationship to a particular candidate. For example, when Japanese voters are asked to give reasons for their support of a particular candidate, the answers, in order of popularity, are as follows: candidate’s party (54 percent), candidate’s policy positions (47 percent), candidate’s characteristics (24  percent), local benefits provided by the candidate (23 percent), appeal of the party leader (12 percent), candidate’s age (11 percent), balance among parties (8 percent), recommendations (6 percent), candidate’s occupation (5 percent), and advertisements (5 percent) (Akarui Senkyo, 2018, 50). It is quite possible that fewer voters are persuaded by a candidate’s policy positions than is reported here, and more voters are actually influenced by recommendations than is reported here. To help avoid the problem of some people giving the expected answer to pollsters rather than the true answer about voting motivations, I have examined the effectiveness of candidate support organizations on voter turnout for candidates to see if this effectiveness has changed over time. Renwick and Pilet (2016, 254–5) also analyze similar relationships, hypothesizing that reforms that personalize elections will boost turnout, but they find no evidence of such an effect in European elections. However, in Japan loyalty to specific candidates has been a distinguishing characteristic of Japanese politics rather than the strong party loyalty that is common in other similar democracies.11 Personal support groups in Japan are more robust the more dense interpersonal connections are, and those connections are strongest in rural areas of Japan where family, friendship, and employment ties can span generations of voters (Maeda, 2009; Scheiner, 2006; Curtis, 1971). In contrast, these traditional campaign vehicles are at their weakest in new urban or suburban neighborhoods where family, friendship, employment, and neighborhood ties are weaker. Japanese political strategists have long called these urban and suburban voters floating voters precisely because of the difficulty in reaching them through the tried and true method of interpersonal networks and memberships in candidate support organizations. Thus, a primary goal of personal support organizations in Japan is to increase turnout among likely supporters, and they still seem to serve that purpose. Even after 25 years under a new electoral system, polling during the 2018 national elections showed that 90 percent of members of personal support groups voted in contrast to only 68 percent turnout among labor union members or 65 percent turnout among voters with no group memberships (Akarui Senkyo, 2018, 64). A few other groups also boast similarly high turnout rates: senior citizen clubs, 92 percent; religions, 87 percent, and chambers of commerce, 97 percent (Akarui Senkyo, 2018, 64). Because I expect personal support organizations to be more

Democracy and political instability  139 90 80

Average turnout

70 60 50 Rural

40

Urban

30 20 10 1958 1960 1963 1967 1969 1972 1976 1979 1980 1983 1986 1990 1993 1996 2000 2003 2005 2009 2012 2014 2017

0

Election year

Figure 7.8  Average turnout in rural and urban Japan, 1958–2017

effective and dense in rural areas, if turnout rates in rural Japan are becoming more like rates in urban Japan, then this finding would be strong indirect evidence that candidate loyalty is also declining, suggesting that the Japanese electorate is becoming dominated by floating voters, even though a weakened effect of personal support groups remains intact. Figure 7.8 shows that the turnout gap between the most urban and the most rural election districts has narrowed considerably over time, suggesting that differences in urban and rural voting patterns are narrowing.12 In addition, the pattern suggests that rural voters are becoming more like urban voters rather than urban voters becoming more like rural voters. However, one of the declines in rural turnout corresponds with the introduction of single-seat districts in 1994 that made many rural districts much less competitive (and therefore less likely to have high turnout levels) than they had been under the previous electoral system that used multi-seat districts. Thus, part of this change in turnout may have been caused by Japan’s new election system rather than changes in rural loyalty to specific candidates. A more sophisticated test of urban and rural turnout is needed, and such a test can be done using a regression, a common statistical technique which makes it possible to simultaneously estimate the effects of multiple factors, such as how urban a district is and how competitive a district is. When the effect of one factor is estimated in a regression, it is the effect of that factor holding constant the effects of all the other possible causes included in the regression analysis. Thus, a regression makes it possible to measure the change in rural turnout after having controlled for the fact that rural elections are, on average, less competitive under the new electoral system than they were under the old election system.

140  Democracy and political instability I calculated a turnout rate for each election district in Japan for each of the 21 House of Representatives elections held from 1958 to 2017. I then added to the regression analysis five factors that might affect turnout rates: how urban a district was, how competitive a district was, and how small (in terms of the number of voters) a district was, how many serious candidates were running in each district, and what percentage of the voters in a district were voting for candidates who ran in the previous election.13 All of these explanations affected turnout levels: if a losing candidate was 5 percent closer to the winning candidate, turnout increased by 0.64 percent; for every additional major candidate, turnout increased by 1.1 percent; for every 10,000 additional votes that a candidate needed to win a district, turnout increased by 0.36 percent; and if 10 percent more of the voters in the district voted for someone who ran in the district in the previous election, turnout increased by 0.09 percent. I also included separate variables for each election year and interaction variables for the urban/rural nature of a district combined with the election year to tease out the changing influence of the urban/rural dimension over time. Thus, in 1958, if a district was 10 percent more rural, its turnout rate increased by 2.2 percent, but in 2017, the same change in how rural a district was caused turnout to increase by only 1.1 percent. (In 2012, the rural advantage for a 10 percent increase in ruralness dropped to only a 0.4 percent increase in turnout.) These results are as expected: even controlling for other factors, the urban or rural nature of a district still exerts the greatest effect on turnout in Japan, though that effect has declined from 1958 to 2017.14 In Figure 7.9 I have plotted the values for the interaction term for the election year and urban/rural. These values show how the effect of rural areas on turnout has changed over 60 years of elections. The values shown are the 3.5

Percent increase in turnout

3 2.5 2 Turnout increase

1.5 1

0.5

1958 1960 1963 1967 1969 1972 1976 1979 1980 1983 1986 1990 1993 1996 2000 2003 2005 2009 2012 2014 2017

0

Election year

Figure 7.9  The percentage increase in turnout explained by making an election district 10 percent more rural, 1958–2017

Democracy and political instability  141 expected increase in turnout in a district if the rural nature of the district increases by 10 percent. From 1958 to 1986, the rural advantage in turnout did not decline appreciably, but beginning in 1990, the turnout advantage of rural districts begins to dramatically decrease, giving rural districts in 2012 or 2014 only one fourth the advantage over urban districts that rural districts had in 1958 or even 1986. Most interesting, however, in Figure 7.9 is the fact that the decline in the turnout advantage of rural voters began before Japan changed its electoral system. In fact the beginning of the decline corresponds most closely with the rise of higher voter volatility as the 1989 House of Councilors election was the first election in which conservative voters abandoned the LDP, giving the opposition coalition its first victory in a national election. In subsequent elections, the voters oscillated back and forth between the LDP and new alternatives to the LDP. These swings appear to have affected both urban and rural voters, reducing the urban/rural difference in turnout, a difference that was more pronounced when voter preferences were relatively stable. Rural and urban voters have also become more similar in that the rural tendency to be more consistent in voting patterns has completely disappeared, making rural voters equally volatile as urban voters, at least in terms of turnout. Figure 7.10 shows the change in turnout, from one election to the next, averaging this change for Japan’s most urban and most rural districts.15 Prior to the increase in voter volatility in the 1990s, urban voters oscillated much more dramatically in their turnout levels while rural voters followed the same general pattern but with much smaller peaks and shallower valleys. Arguably the stronger candidate loyalty in rural districts muted the dramatic swings in 15

Average change in turnout

10 5

0

Rural Urban

–5 –10

14–17

12–14

09–12

03–05

05–09

00–03

96–00

90–93

86–90

83–86

80–83

79–80

76–79

72–76

69–72

67–69

63–67

60–63

58–60

–15

Election year pairings

Figure 7.10  Comparing changes in turnout for rural and urban Japan

142  Democracy and political instability turnout that were occurring in the more volatile urban districts. In contrast, beginning in the 1990s, the peaks and valleys of the rural and urban districts are nearly identical. If turnout goes up, it increases by the same amount in both urban and rural districts. If turnout declines, it declines by the same amount in both urban and rural districts. Thus, candidate loyalty no longer provides a buffer in rural districts to candidates trying to buck national trends. Rural voters are responding to national trends in turnout with the same level of volatility as urban voters. All three sets of turnout data show that rural voters are behaving more like urban voters in Japan. Not only has the gap between rural and urban turnout narrowed in recent elections, the rural swings in turnout, which used to be more muted, have become as extreme (and even more extreme in recent elections) as urban swings in turnout. Even controlling for other factors that also affect turnout using a regression procedure, the impact of turnout has more than halved, and the decline in a rural area’s impact on turnout began with the swing to the opposition in the 1989 House of Councilors election rather than the subsequent change in the electoral system.16 This data suggests that a stable cornerstone of Japanese voting behavior, the personal support networks of Japanese politicians, is exercising much less of a buffering role than it used to. At least with turnout, Japanese rural voters today look a lot more like their urban voters than they used to in past decades. A final test of candidate loyalty among Japanese voters can be done by comparing (1) the vote for all conservative candidates against the vote for conservatives who ran in both the current and previous elections and (2) the vote for candidates who ran in rural districts in consecutive elections against candidates from urban districts who ran in consecutive elections.17 Both comparisons are suggested from the observation that candidates who run in consecutive elections are more likely to have effective personal support groups. The support of these groups should mute the effects of national swings in the vote for conservative candidates. Similarly, if personal support groups are more effective in rural areas, then there should be more muting of the national swing in rural districts than occurs in urban districts. The calculations presented in Table 7.2 show dramatically the weakening effect of personal support groups in muting the swing, either for or against the LDP. For election pairings from 1958 to 1993, candidates who run in both elections nearly always have a much smaller change in their vote support levels than the average for the entire LDP.18 Though this muting may occur for many reasons, including having effective personal support organizations, all of the reasons tie into the phenomenon of some voters having greater loyalty to a candidate rather than a party. It is also interesting that a difference in this muting between rural and urban districts only appears in the 1980s, suggesting that in the 1960s and 1970s there were perhaps fewer differences in candidate loyalty levels between urban and rural areas. The overall muting effect largely disappears, however, under the new electoral system, perhaps explained by the fact that in most of these districts there

Democracy and political instability  143 Table 7.2  National swing and muting of that swing for conservative candidates in Japan Election years

National swing (all LDP candidates)

Swing for consecutive candidates

Rural swing (consecutive candidates)

Urban swing (consecutive candidates)

1958–60 1960–63 1963–67 1967–69 1969–72 1972–76 1976–79 1979–80 1980–83 1983–86 1986–90 1990–93 1993–96 1996–00 2000–03 2003–05 2005–09 2009–12 2012–14 2014–17

−2.87 −1.25 −5.63 −1.64 −0.97 −4.82 +2.31 +2.3 −0.83 +4.66 −2.74 −12.6 +2.32 +4.65 +2.1 +1.85 −9.32 +4.49 +5.35 −0.29

+0.80 +0.38 +0.18 +0.16 −0.38 +0.02 +0.63 +2.11 −2.39 +1.21 −1.97 +0.44 +25.40 +5.84 +0.59 +1.22 −8.24 +6.56 +5.14 −0.46

+0.54 +0.63 +0.29 +0.26 −0.31 −0.60 +0.30 +1.73 −0.75 +0.85 −1.93 +1.55 +32.33 +6.20 −2.15 −3.95 −4.28 +8.39 +3.06 −0.74

+0.52 +0.58 −0.55 +0.38 −0.69 +0.88 +0.08 +4.42 −4.45 +1.38 −1.77 −1.59 +18.21 +6.22 +4.40 +5.76 −12.09 +3.24 +6.40 −0.30

are only two, major-party candidates, making it more likely that conservative candidates will benefit or be harmed by the movement of swing voters in their districts, similar to how candidates in any other district are affected. There still appears to be a slight muting of this swing in rural districts in contrast to urban districts with a few exceptions: Koizumi’s 2005 expelling of some LDP MPs (concentrated in mostly rural districts) and the rebound of the LDP in 2012 after the Democrats’ 2009 victory.

Alternatives to candidate and party loyalty Electoral data and polling data about party identification in Japan mirror an extensive literature in US politics that discusses changes in party identification there (Bartels, 2000; Lewis-Beck, Jacoby, Norpoth, and Weisberg, 2008). Of course, the question arises of what exactly explains voting decisions if fewer voters are motivated by party or candidate loyalty. The most commonly cited answer is issues. Both in Japan and the United States, voter decisions are based on the issues of the campaign and the positions that candidates take on those

144  Democracy and political instability issues (Bafumi and Shapiro, 2009). Thus, volatility is explained by the campaign issues and the positions of candidates. For example, dramatic shifts in votes in 1964 or 1972 in America are explained by the extreme positions that Goldwater and McGovern took in their campaigns, leading to their electoral drubbings. Similarly, swings in Japanese preferences from 2005 to 2009 are explained by issues such as the voters’ evaluation of the Aso administration as well as evaluations of the Democrats’ ability to govern (Masahiro, 2010). Though it is undoubtedly true that most voters are either voting based on party/candidate loyalty or alternatively weighing the issues before deciding their preferred candidates, there are several problems with these explanations of voting decisions. First, issue preferences of voters can be quite short-term and malleable (Bertrand and Mullainathan, 2001; Carsey and Layman, 2006; Miller and Shanks, 1996). For example, if a voter says she is supporting Prime Minister Abe because he didn’t raise the sales tax, how sure can we be that the sales tax is actually the reason why she supports Abe? Perhaps she supports Abe for other reasons and cites the sales tax as a justification because she has heard Abe talk about the sales tax issue. This calculation is complicated by the fact that the parties regularly try to muddy the waters on seemingly clear-cut issues to enhance their issue advantages in the minds of voters. Thus, the abortion issue is transformed into a discussion of the appropriateness of late-term abortions, or the illegal immigration issue is transformed into arguments about whether high-performing children of illegal immigrants raised in the United States should be exempt from deportation orders. Just because voters answer pollster questions saying that they agree with certain policies of a political party or candidate that they support doesn’t mean that those policy positions actually caused them to support that party. These theoretical concerns are supported by evidence that many voters either (1) don’t know the issue positions of candidates or (2) state issue preferences that seem to lack coherence (Achen and Bartels, 2016; Baldassarri and Gelman, 2008; Bernstein, 1989; Carpini, Michael, and Keeter, 1996; Carsey and Layman, 2006; Converse, 1964). In contrast, other researchers have found high (Ansolabehere, Rodden, and Snyder, 2008) or rising levels of issue consistency (Nie, Verba, and Petrocik, 1979) among voters. Despite these differences, all the researchers have found lower levels of issue consistency among low information voters. An example of this type of voter is undecided Nebraska voter Kevin Sheen who was interviewed one day before the 2008 election and had not yet decided whether he was going to vote for John McCain or Barack Obama. Sheen explained that his vote came down to the issue of abortion and that he was ‘definitely pro-life.’ He added that he was trying to decide which candidate’s position on abortion was closer to his.19 It is perfectly fine for voters to decide at the last minute, but Sheen was an example of a low information voter who did not know the obvious issue differences between the candidates on the abortion issue. There is also a vigorous debate about whether or not party voting is actually declining (Bartels, 2000; Renwick and Pilet, 2016, 3) and whether voters can be

Democracy and political instability  145 just as partisan when they weigh issues as they were when they used to follow partisan cues (Bafumi and Shapiro, 2009; Ansolabehere, Rodden, and Snyder, 2008). Nevertheless, in contrast to well-informed issue voters, there are also many voters who are low interest/low information voters (Bartels, 1996; Converse, 1964; Dahl, 1961; Campbell et al., 1960; Norris, 2011, 160–3; Popkin, 1994; Zaller, 1992). The inattention of such voters in a previous generation perhaps mattered less because, if these people voted, they tended to be loyal party voters, or if they were in Japan, they had loyalties to a particular candidate. Now with weakened party or candidate attachments, these voters are supposed to be making decisions based on issue positions, but it is likely that such low information voters are not making candidate support decisions based on issues, and they are much less influenced by partisan or candidate loyalty than they used to be. What then are the alternatives that these voters have turned to in making their decisions? Direct evidence of voting for other reasons is difficult to develop because voters are cued, as was Mr Sheen, to give issues as their reasons for voting, even if they are not the actual reason for that person’s vote. Despite this bias, however, a first alternative for many low information/low interest people is simply to not vote (Bernstein, 1989; Carpini, Michael, and Keeter, 1996, 224–7; Lewis-Beck Michael et al., 2008, 97–9). Indeed, lower turnout is a predictable consequence of declining party and candidate loyalty. A second option is for voters to use other shortcuts or cues to help make their voting decisions (Cutler, 2002; Dahl, 1961, 323–4; Downs, 1957; Lupia and McCubbins, 1998; Lupia, 1994; Popkin, 1994; Wittman, 1995). A voter watching a debate or a television commercial might be either impressed or repulsed by a specific image or action. A voter might vote against a certain candidate because something about that candidate reminds them of his or her ex spouse. Some voters follow the recommendations of a close friend, family member, labor union, religion, or special interest group. Most voters would be less than forthcoming about many of these motivations, making it difficult to measure their actual presence in the thinking of voters. Despite these evidentiary difficulties, some analysts have argued that elections are being decided by irrelevant or even harmful preferences of voters. For example, Caplan (2007) argues that voting in a way that indulges the erroneous beliefs of voters is costless, making it rational for voters to indulge their erroneous beliefs at the voting booth (whereas market incentives restrain such behavior in other settings). Kuklinski et al. (2000) describe human tendencies to infer missing information, to make that information consistent with other beliefs, and to be confident in those beliefs, and these three tendencies combine to make voters not just uninformed, but misinformed. Similarly, Brennan and Lomasky (1993) explain how expressive voting can dominate the calculations of voters and that such expressive voting may result in outcomes that actually run counter to a voter’s narrowly defined self-interest. Though Popkin (1994) argues that information deficiencies in voters can be cured, he also explains how signaling to low information voters, framing of

146  Democracy and political instability issues, and perceptions of whether or not a politician is responsible for a particular outcome can all change the predicted outcome of an election or cause politicians to invest greater time and effort into less relevant issues rather than more consequential issues. Politicians begin to worry more about how their actions and policies will be interpreted by the voters than the actual effectiveness of those same actions and policies. Bartels (1996) finds the differences between informed and uninformed voters are not random and do not cancel each other out, giving incumbents and some political parties advantages in elections. Prisching (1995) cites Schumpeter for claims that voters are manipulable, ill-informed, emotional, and too focused on the short term. Quattrone and Tversky (1988) show that voters make different risk calculations depending on whether they are facing a gain or a loss and condition their responses to equivalent situations depending on the status quo. These irrational preferences, they argue, help explain the advantages that incumbents have in elections. Redlawsk (2002) finds that people search more for information that corroborates their prior preferences and in fact become stronger in those preferences after reading negative information about their preferred position. Somin (1998) adds to this list a concern that the scope of government today makes it nearly impossible for voters to be informed on all of the relevant issues of governing. In line with these studies, some intriguing evidence of the importance of just one possible form of irrelevant voting motivations, the physical attractiveness of candidates, shows up in an analysis of recent Japanese voting. I have analyzed the results of the 2003 Japanese elections to see to what extent the physical beauty of candidates affects the number of votes that candidates win. Japan presents an excellent case for testing the effect of physical attractiveness on candidate voting for two reasons. First, candidate campaigns are carefully circumscribed in Japan, making it likely that the only image of the candidate that the voter sees are the posters put up by candidates in the district before the election. Media advertisements by the party are prohibited from showcasing specific candidates, and candidate advertising is largely prohibited. The typical voter in Japan has only seen a photo of the candidate, perhaps in a handbill passed out in public setting, but more likely on a poster, prominently showing the candidate’s face and name. Thus, the candidate’s campaign photo likely represents the entire visual image that Japanese voters have of that candidate. Second, under Japan’s new electoral system, voters simultaneously cast two ballots, one for their preferred candidate and one for their preferred party. Having a simultaneous measure of a party’s support makes it easier to disentangle candidate support that is independent of the party from candidate support that comes from his or her membership in a party. In fact, in order to best measure the support for candidates that comes from purely personal characteristics of the candidate rather than party support, I have calculated a predicted measure of candidate support based on party support and the number of candidates who are running from each of the major parties in each election district. When, for example, the Socialist Party doesn’t run a candidate in a

Democracy and political instability  147 district, I have allocated Socialist voters to the remaining party candidates in likely percentages that take into consideration both the endorsements of candidates made by parties and the ideological proximity of the various parties. I have used this predictor of a candidate’s vote in a regression analysis with a variety of other factors that are related to the personal characteristics of a candidate. This list includes: (1) whether or not the candidate was an incumbent, (2) whether or not the main opponent was an incumbent, (3) the age of the candidate, (4) the gender of the candidate, (5) the relative amount of money the candidate spent on her campaign, (6) the educational background of the candidate, and (6) the relative physical attractiveness of the candidate, compared to her main opponent.20 I ran separate regressions for all of the LDP, Democrat, and Communist candidates who ran in the 2003 Japanese House of Representatives election. The results showed that physical attractiveness exerts a significant influence on the votes that a candidate wins, even controlling for many other personal characteristics of candidates and the party strength in the district. For example, if an LDP candidate is one point more physically attractive than her main opponent on a five-point scale of physical attractiveness, then that candidate earned 1.7 percent more votes. Thus, a reasonably attractive four on the physical attractiveness scale won 3.4 percent more votes if he ran against a relatively unattractive two on the same five-point scale than an LDP candidate who was only as attractive as her opponent. This advantage is similar in size to the 3.8 percent advantage that a district incumbent had and significantly larger than the 2.4 percent disadvantage that older candidates had. Only the candidate’s gender, money spent on a campaign and the relative partisan breakdown of the district had a larger impact on a candidate’s share of the vote than the relative physical attractiveness of the candidate. Other personal attributes of candidates (youth and schooling) were not statistically significant in their effect on a candidate’s share of the vote. The regression for Democratic candidates in 2003 produced similar results. The same variables were statistically significant and statistically insignificant. The partisan breakdown of the district and the amount of money spent on campaigning also had the greatest effect on a candidate’s share of the vote. Being an incumbent increased a candidate’s vote share by 4.6 percent, but being two points more attractive than an opponent also increased a candidate’s vote share by a smaller 2.4 percent. The only results that did not show the importance of physical attractiveness for candidates were the results for the Communist Party candidates, but these results actually support the observation that when candidate or party loyalty declines, more ephemeral and less convincing rationales for voting will take their place. In 2003, the Communist Party ran candidates in all 300 election districts even though the Party’s candidates were unlikely to win any district seats. The Party went into the election with no district incumbents and the closest it came to winning a district seat was in Kyoto 1 where the Communist candidate came in third with only 25.6 percent of the vote. Thus, the voters

148  Democracy and political instability who cast ballots for Communist candidates were likely those who were the most committed to the Communist Party. They were voting for hopeless candidates because they were committed to the ideology of the Communist Party. When the voter percentages of these candidates are analyzed, only the partisan breakdown of the district, candidate gender, and candidates holding previous elective office in the Diet mattered.21 Women candidates of the Communist Party, on average, win 1 percent more votes than men candidates. If a candidate had previously served a term in the Diet, she would on average win 1.8 percent more votes. In contrast, candidate schooling, age, relative physical attractiveness, and the amount of money spent on campaigns did not affect the candidate’s share of the vote. Thus, voters who supported Communist candidates who had no chance of victory were not influenced by the physical attractiveness of a candidate. They were voting for ideological reasons. In contrast, LDP and Democratic candidates were competing for votes from a wide swath of the electorate, not just the ideologically motivated voters. Among this broad cross-section of voters, the physical attractiveness of a candidate mattered as much as or more than other important factors such as the incumbency status of a candidate. Of course, it is perhaps not surprising that physical attractiveness would affect some voters’ evaluations of candidates. A host of other research has found that physical attractiveness affects not only elections in other countries but also a variety of decisions including employment, income, and trust. (­Biddle and Hamermesh, 1998; Megumi, Stone-Romero, and Coats, 2003; Mobius and Rosenblat, 2006; Tews, Stafford, and Zhu, 2009; Wilson and Eckel, 2006). My research suggests that with a decline in party or candidate loyalty, the number of voters who will be swayed by other factors such as the physical attractiveness of a candidate will likely increase and become a more prominent factor in voting decisions (even if the voters themselves are unlikely to admit the motivation). Physical attractiveness is just one of several less defensible factors that can influence voting, but it is these factors, along with the more defensible position of issue voting, that are likely filling the void in voter determinations that has been created as candidate and party loyalty has weakened. Physical attractiveness is not just a minor footnote in campaigns, at least in Japan, it has as much of an effect on voting as the much-studied incumbency advantage of candidates. Candidate attractiveness can also explain some of the popularity of Koizumi Junichiro in contrast to his five successors. I used a mixture of Japanese and American raters to evaluate the physical beauty of candidates. Each rater had only a few seconds to view a candidate’s picture and assign a rating of physical attractiveness on a 1 to 5 scale. Japanese raters were used because they would be more sensitive to cultural differences in physical attractiveness between the United States and Japan (Englis, Solomon, and Ashmore, 1994). American raters were used because they would not know any of the politicians, so their ratings of physical attractiveness would be unbiased by knowledge of famous politicians. I used two panels of raters, one with Japanese raters and

Democracy and political instability  149 one with US raters. In either case, the ratings between US and Japanese raters were not significantly different. In fact, there was greater variation in ratings between the gender of the raters than there was between the country of origin of the raters (see Cunningham et al., 1995; Johnston and Oliver-Rodriguez, 1997).22 It was interesting, though, that American raters gave Koizumi the 3rd highest rating of all LDP politicians and the 14th highest rating of all the candidates in the 2003 election without knowing Koizumi’s position or fame. Perhaps a portion of Koizumi’s national appeal was derived from his physical appearance. It is also likely that physical attractiveness plays a role in American elections. Despite the extensive discussion of the role that race played in the 2008 presidential election, age (39 percent) appeared to be a much greater factor in voting decisions than race (19 percent), at least as reported in exit polls (CNN, 2008). Though obviously race would be underreported in polling data, it is surprising that nearly 40 percent of voters cited age as a voting decision factor. Perhaps all of these voters were only concerned that John McCain would be 72 years old in 2009. It is also possible that at least some of these so-called age voters were motivated by differences in the physical attractiveness of John McCain and Barack Obama. Looking old is a thinly veiled way of describing physical attractiveness. Certainly the Obama campaign was well aware of the power that Obama’s image had on voters: young, thin, and stylish. His campaign made sure that candidate Obama was never seen smoking a cigarette, as that image would have run counter to the attractive image that candidate Obama was successfully portraying. Unfortunately for candidate McCain, thin white hair with a scarred and round face that clearly showed his age was not something that could be manipulated as easily as making sure that no photographs were taken while he was smoking. Pundits and scholars have worked overtime trying to measure the effect of race on the 2008 presidential election, but perhaps an even bigger effect was caused by the age difference in candidates, and I suggest that the age difference was for some voters a polite way of describing how differences in physical attractiveness affected their voting choices.

Greater volatility makes it difficult to solve political problems Even granting that it is true that a subgroup of Japanese and US voters have moved from candidate or party loyalty to more superficial reasons for casting votes, still no case has been made why such a change, in contrast to the more desirable option of voting on issues, would be significant for politics. For example, Renwick and Pilet (2016) study the effects of electoral system changes that personalize elections more, and they find no consistent effect on (1) numbers of women MPs, (2) voter satisfaction, or (3) turnout. There is, however, a long and surprising tradition in political science that says that the more demands that the voters place on a government, the harder it is for that government to perform well. The argument follows that governments need a breathing space in which to make decisions without the unrelenting scrutiny

150  Democracy and political instability and demands of the people. Because people’s demands of government tend to be short-term and selfish, governments that are more closely tied to those demands create policies that are harmful in the long run, internally contradictory, or biased towards the particular good of the most vocal groups against the collective good. Corollaries of this core distrust of democracy manifest themselves in a variety of political settings. Campbell et al. (1960, 541–5) argue that voter lack of information and attention give politicians greater freedom to act. In fact, they specifically link volatility to low information/low interest voters (547). Dahl (1961, 305–10) argues that the gap between voters’ potential influence and actual influence gives political entrepreneurs the leeway to push specific policies, as long as their efforts cause voters to become more involved in politics. A debate exists between those who argue that politicians are largely unconstrained by voters (Bernstein, 1989) and those who argue that voters do constrain the actions of their representatives (Ansolabehere and Jones, 2010; Canes-Wrone, Brady, and Cogan, 2002). Clientelism, pork barrel politics, and corruption, though not the exclusive domain of democracies, all seem to root well and flourish in democratic settings in which politicians have incentives to woo the support of voters through these methods. In the international setting, Valentino, Huth, and Croco (2010) argue that democracies are more likely to minimize the costs of war in order to placate public opinion. In analyzing the responses of Latin American governments to economic crises, O’Donnell (1973) argues that authoritarian regimes are better able to implement unpopular but necessary economic reforms than are democratic regimes. Haggard and Kaufman (1992) analyze government responses to inflation and find a more complex, but similar, response pattern. Authoritarian regimes respond more effectively than democratic regimes respond when politics in a democracy is polarized. In contrast democracies effectively respond to inflation when popular opposition has been coopted into the ruling system. These findings show that the demands of the people must either be partially coopted by the ruling regime or ignored by an authoritarian government in order to make difficult economic reforms. A more widely known manifestation of this idea occurs in Thomas Friedman’s New York Times column in which he praised China’s authoritarian government for being able to ‘just impose the politically difficult but critically important policies needed to move a society forward in the twenty-first century.’23 Labor leader Andy Stern agreed, favorably citing Nobel laureate Robert Engle that ‘while China is making five-year plans for the next generation, Americans are planning only for the next election.’24 Though experts disagree on the empirical question of how free representatives actually are to go against their constituents’ preferences, they all agree on the potential for a politician’s actions to be constrained by voters in such a way as to prevent or make difficult the taking of unpopular political actions. Thus, in wartime, unity governments are sometimes formed to depoliticize the necessary decisions made in conducting the war effort. Similarly, in forcing the

Democracy and political instability  151 implementation of initial reform measures in the Euro crisis, France and Germany insisted on the replacement of political governments with technocratic caretaker governments in Italy and Greece. These actions all show the belief that partisanship and the demands of constituents must be subordinated when difficult decisions need to be made. Two prominent actions of Kan Naoto, Japan’s prime minister in 2010 and 2011 illustrate these incentives well. When Kan came into office in the spring of 2010, Japan faced (as it does now) a severe fiscal crisis. Japan’s debt to GDP ratio was at the level of 1.8 placing it well above any comparable advanced industrial democracy. The consensus among elites was that the solution to this problem was to raise the sales tax in Japan from its level of 5 percent.25 Previous decisions to create the national sales tax and to raise its level from 3 to 5 percent were difficult decisions, but the ruling LDP had managed to take these actions with only some damage at the polls, the Party retaining its hold on power. Furthermore in 2010, the opposition LDP had also called for an increase in the sales tax to 10 percent, making it seem easier for the ruling Democrats to take the same position. Kan did, and he was praised for his action. An editorial in Asahi Shimbun called it a step in the right direction and praised Japan’s political parties for overcoming their history of kneejerk opposition by the parties out of power to all proposals for tax increases.26 Despite the praise, Kan and his party were punished at the polls in Japan’s 2010 House of Councilors election. In the wake of his party’s defeat, his actions were quickly seen as a massive political blunder. In contrast, the model of how to win a Japanese election is shown by the same Democratic Party only one year earlier in the 2009 election. The same Asahi Shimbun panned the Democrats in 2009 for failing to address Japan’s fiscal problems. An editorial raised the question of whether the voters would be persuaded that the Democrats could be trusted with power if the new party leader Hatoyama continued the same fiscal promises as those made by his predecessor, Ozawa ­Ichirou. Specifically the newspaper criticized plans to increase government payments to citizens without any corresponding increase in government revenue other than vague assurances to cut bureaucratic waste.27 Fiscal irresponsibility was unpopular with the paper, but the Democrats used this irresponsibility to win their only electoral victory, a massive landslide in the 2009 elections. The true motivations for why Kan failed to copy the successful electoral strategy of 2009 and instead embark on a risky electoral strategy are unknown, but it is likely that he saw the need to take action and noted the consensus among not only the experts but also the opposition LDP that a tax increase was needed. Unfortunately for Kan, the rise of several small parties that explicitly attacked Kan for his proposal for a tax increase and his reneging on his party’s electoral promise not to raise taxes led to the Democrats’ loss in 2010. The LDP vote share did not rise compared to the previous 2007 election. Rather the Democrats’ vote share dropped, and the difference was picked up by three new parties, depriving the Democrats of a working majority in the upper house and contributing to a precipitous drop in Kan’s popularity.

152  Democracy and political instability In contrast, at the end of Kan’s year in office, he seemed to have learned the lesson of volatile voters and political popularity. In response to the disaster at the Fukushima nuclear power plant, Kan made several statements and took actions that suggested his willingness to commit Japan to a future without nuclear power. His efforts in this regard were done despite expert opinion that Japan’s reliance on nuclear power was so great as to make a quick transition away from nuclear power nearly impossible. In this situation, Kan took a politically popular but unfeasible position, and his reputation rose as a result. One year earlier he took the politically unpopular but necessary position and saw his popularity plummet. Perhaps Kan learned the lesson in his one year in office that first and foremost a Japanese politician must remain popular, even if that means delaying a response to a looming fiscal crisis or creating an energy crisis by a too quick embrace of a highly popular anti-nuclear power stance. The tax story continues in the actions of subsequent prime ministers. Kan’s successor, Noda, actually passed legislation that raised the sales tax rate from 5 percent to 8 percent with an additional increase to 10 percent to take effect 1.5 years later. His actions led to the Democrats’ electoral drubbing in the 2012 elections and paved the way for Abe’s return to power. Abe, in contrast, twice delayed the implementation of the second tax increase (to 10 percent), putting it off until October 2019, at the end of his political career. Clearly, Abe was aware that tax increases could have destroyed his political career. The importance of keeping high levels of popularity also seem to be thwarting the making of difficult but necessary decisions in the United States. Though analysts argue about whether American voters are becoming more partisan (Abramowitz and Saunders, 2008; Fiorina, Abrams, and Pope, 2006; Layman et al., 2010), they all agree that elites are more partisan than they were in the past and that voters have sorted more cleanly into ideologically coherent political parties, at least in comparison to previous decades. Thus, partisanship is blamed for the failure in the US government to effectively resolve fiscal crises in recent decades (deficits and social security funding). In contrast, I argue that partisanship was just a byproduct of a common desire in both the Republican and Democratic Parties to avoid taking politically unpopular action or at least delay those actions until they would have a lesser effect on elections. Thus, rather than seeing a partisan inability to agree on issues as the main cause of legislative gridlock in the United States, I see a bipartisan desire to avoid making tough political decisions as the real cause of the failure to act. For example, comprehensive health care reform passed in 2010, but the authors of the legislation specifically delayed the implementation of many of the legislation’s requirements until after the 2012 elections. The bipartisan compromise reached in 2010 to extend the Bush tax cuts was for two additional years, punting the need to make a final decision on the future of the tax cuts until immediately after the 2012 election. When cuts to the popular Medicare Advantage program were scheduled to appear to voters in information mailed out to senior citizens during October 2012, the Obama administration devised a demonstration project that would put sufficient funds into

Democracy and political instability  153 the program to make it appear to voters that no changes had been made in the program until after the 2012 election. Republicans finally agreed to raise the debt limit ceiling after paying a political cost, but oddly both sides were interested in raising the limit only as much as would be needed to get past the 2012 election. A final piece of evidence that election timing and presidential popularity rather than partisanship are what cause fiscal paralysis in Washington is the fact that no action on these difficult fiscal issues was taken during 2009–10 when the Democrats had control of both the Congress and the presidency, nor in 2016–18 when the Republicans had similar control. Perhaps the volatility of US voters combined with the likely unpopularity of any measures that would put the US fiscal house in order (either less spending or higher taxes) make it unlikely that either party would take the strong action needed to address this crisis for fear of being blamed for the action in the next election. These tendencies continue in the present political climate. In 2019 Democrats have proposed an overhaul of social security with tax increases (though only on the wealthy) coupled with a much more palatable expansion of benefits. Democrats understand that a social security proposal that only includes tax increases will be unpopular, so they have coupled that proposal with an enticing increase in benefits, even if the benefit increase is fiscally irresponsible. The repeated failures of Republicans to repeal the Affordable Care Act also show this same political calculation. The Republicans never found a way to signal to voters that repeal would not eliminate popular features of the Act, health care coverage for dependents until age 26 and the ban on insurers considering pre-existing conditions when signing up people for health insurance. Surely politicians have always considered the popularity of proposals when making or supporting such proposals. My argument is only that with more volatile voters, popularity, especially short-term popularity has become perhaps the most important feature of policy discussions.

Differences between Japan and the United States Despite the many similarities between Japan and the United States in the weakening of party and candidate loyalty, the paths of both nations have also significantly differed. Party and candidate loyalty seem to have weakened much more in Japan than in the United States, and political volatility seems to have increased more in Japan than in the United States. These differences can be explained by several factors. First, Japanese volatility is exacerbated by a parliamentary system in which the prime minister can be removed at any time by her colleagues rather than serving a defined term of office as is the case in presidential systems. Thus, it is more difficult in Japan for a prime minister to ride out a period of unpopularity with the assurance that the political landscape will have changed by the time of reelection. Second, Japan has a staggered election system with a significant election almost always on the horizon. This fact, combined with the ability of the ruling party to remove its leader, creates a strong incentive to get rid of unpopular leaders sooner rather than later. Japan, for example has no timetable for lower

154  Democracy and political instability house elections (except that they be held at a minimum of every four years), but its upper house elections are held at three-year intervals and most of its local elections are held at four-year intervals. Thus, in 35 of the 49 years from 1971 to 2019 there has been a major election, and only twice in that 49-year period have Japanese voters gone two consecutive years without a major election. Even more than the United States in which members of the House of Representatives seem to be always running for reelection, Japan’s political leaders face a referendum by the voters nearly every year. Third, Japan should have an advantage in dealing with difficult political issues because it is a parliamentary system where the party that controls the executive branch of government also controls the legislature. Such systems make it easier to pass and implement difficult or unpopular legislation. For example, Canada reduced its government deficits dramatically under the leadership of the Liberal Party in the late 1990s and avoided any penalty at the polls for cutting government spending (Barnes, 2011). However, Japan does not have a pure parliamentary system. After World War II, Allied occupiers grafted on a bicameral system in Japan that makes it possible for Japan to have divided government, just as occurs in the United States. Thus, the ruling party or coalition of parties in Japan controls the lower house, but a different coalition of parties might control the upper house, whose approval is needed for most legislation. When government is divided, it is ­easier for the opposition to politicize policy disagreements and block or delay legislative proposals. In a divided government setting, it is much more likely for unpopular political proposals to be blocked and politicized. In contrast, in a pure parliamentary system, a ruling party can ignore opposition and implement new policies without having to worry about voter response for several years. It is precisely this dynamic that has also stymied efforts to address fiscal problems in the United States. Though Japan has a parliamentary system, the frequency of its elections and the fact that the upper house is more often than not controlled by a different coalition of parties makes addressing difficult fiscal issues in Japan more like the United States than like the Canadian example. Fourth, Japan’s voters were never strong in their partisan loyalty, preferring instead, at least among conservative voters, to have a strong candidate loyalty. However, as that loyalty has declined, the voters have become much more volatile, making possible broad and sweeping changes in voter preferences. This increased volatility, combined with the lesser importance of candidate loyalty and its concomitant rise in other factors such as the popularity of the party leader (Krauss and Nyblade, 2005), has led to a polity in which sometimes party leaders change with rapidity, but also party leaders, to be successful, must avoid unpopular decisions and work, even harder, to please a volatile electorate. In contrast, though party identification has declined in the United States, there is ample evidence that partisanship remains as strong as ever. Thus, though American voters may claim that they are politically unaffiliated, there is strong evidence to suggest that many of these independents are actually as loyal to political parties as the party loyalists of previous generations.

Democracy and political instability  155 Fifth, voter volatility is also ameliorated in the United States by a surrogate which largely doesn’t exist in Japan: voter loyalty to candidates of the same race, ethnicity, religion, or gender (Achen and Bartels, 2016). This loyalty appears strongest in the absence of partisan cues. Thus, in party primaries women are more likely to support women candidates, evangelical Christians to support evangelical candidates, and Hispanic voters to support Hispanic candidates. However, these tendencies do carry over to general elections where they can create political stability that is as good as or better than the stability created by strong party identification among voters. For example, African American voters have the strongest partisan loyalty of any group of voters. African American voters reliably support Democratic candidates. Of course, African American voters make some demands on the Democratic Party, but if those demands are met on symbolic issues of racial equality and sensitivity to racial concerns, the Democratic Party has few other demands that it must answer in order to receive the strong support of African American voters. As a result, Democratic candidates such as President Obama go into any election with about a 10 percent advantage over their Republican opponents. Democrats need only to win 45 percent of the remaining voters to have an absolute majority. In contrast, Republican candidates, unable to win the support of most African American voters, must persuade a much larger percentage of the remaining voters, 55 percent, to be able to win an election. Exit polling in the 2008 US presidential election shows this advantage dramatically. In exit polls, President Obama won 53.8 percent of the votes of those polled. However, if the solid support of African Americans is excluded from this total, President Obama’s support drops to 47.8 percent of the remaining white, Latino, Asian, and Other voters (CNN, Election Center, 2008, 1). The loyal support of African American voters turned a 47.8 percent deficit among other voters into a 53.8 percent advantage for President Obama among the entire electorate. There are, of course, other groups that are loyal to the Republican Party or the Democratic Party, and certainly such groups also exist in Japan. This greater loyalty of some groups in the United States does give political parties and candidates greater freedom to address difficult political issues with less fear of a voter backlash at the polls. Ethnic, racial, religious, or gender loyalty can all work to make it easier for a party or a candidate to take an unpopular but necessary political position. These other forms of partisan loyalty are either missing or much weaker in Japan than they are in the United States.

Conclusion In both Japan and the United States, the supposed rise of issue voting is hailed as a positive development for democracy. Right-thinking people would never say that they preferred an election in which local notables signaled how a town should vote and the voters meekly followed those directives like so many sheep. Floating voters may be the bane of candidates in Japan because they are fickle

156  Democracy and political instability and hard to contact, but they represent mature democracy, a modern Japan where voters carefully weigh the issues and vote according to their preferences. Similarly, in the United States there are, for example, few defenders of straight ticket voting by which voters can choose one political party and automatically cast a ballot for all of the candidates from that political party. Few believe that voters should be simply deciding their votes based on partisan allegiance; rather they should be voting after carefully weighing the issues and the qualifications of the candidates. This progression of democratic ideals is a worthy goal, and the decline of partisan voting, to be replaced by a more nuanced and informed vote choice should be a welcome change and improvement in the quality of democracy in both Japan and the United States. Unfortunately, the reality is far from this depiction of a model democracy. Though many voters have improved the quality of their deliberations in how they decide to vote, others have not transferred their prior allegiance to parties or candidates to a new, careful consideration of the issues. Rather, they have transferred their allegiance from parties or candidates to a hodgepodge of influences, some laudable (such as issue positions) and other not so praiseworthy (such as who is more physically attractive). Scholars have long recognized that the democratic ideal of a careful consideration of the issues is a model that few voters actually follow. Most American voters don’t even know the names of many of the candidates running for office. Downs (1957) argues that it is not rational for voters to put forth the effort to gather even small amounts of information about candidates to inform their vote. Rather, voters resort to cues or shortcuts in voting. These cues take many forms: party membership; endorsements by unions, religions, or special interest groups; shared race, ethnicity, religion, or gender; physical attractiveness; intangible evaluations of character, trust, or likeability; campaign gaffes; and advertising. Though some voters have moved away from using party cues to an informed weighing of the issues, perhaps even greater numbers of voters have moved away from party cues to cues from other organizations and cues based on beauty, being telegenic, media campaigns, or shared characteristics (gender, race, religion, age) with a candidate. Of course, there is nothing in Japan’s cautionary tale that says that volatility and policy immobilism must result from declining party and candidate loyalty.28 The long tenures of both Koizumi and Abe show how popularity can be used to cultivate a stable electorate through which some difficult reforms can be made. Though Koizumi’s and Abe’s popularity can be explained by many factors, both used short-term, symbolic gestures to boost their own and their party’s support. Koizumi also successfully used symbolic gestures to vanquish opponents of his reform efforts within his own political party. I have no explanation for why Koizumi and Abe were successful in these appeals to short-term popularity while other politicians such as Aso, Kan, or Noda were unsuccessful. There is something to be said for raw political talent and the ability to effectively read and respond to public moods. Koizumi and Abe seem to have this talent.

Democracy and political instability  157 Canada’s Liberal Party made a similar successful attempt to change the discourse in Canada to support dramatic declines in government spending in order to address a fiscal crisis there. Mitt Romney’s selection of Paul Ryan as his running mate and the elevation of Medicare reform as an election issue in the 2012 campaign was a failed attempt to turn, as Koizumi successfully did, a potentially unpopular government reform into a winning electoral issue. The record, however, suggests that successes such as Koizumi’s will be few and far between. Only an extremely talented political entrepreneur will be able to ­create a consensus in favor of taking tough political action and then convert that consensus into stable political support. These findings might also explain the rise and success of unconventional politicians in many other democracies. Though Japan is farther ahead on the curve of increasing volatility, the detachment of voters from stable partisan alignments in many countries has opened the door for talented political entrepreneurs to rise to power. Donald Trump in the United States, Silvio Berlusconi in Italy, Sebastian Kurz in Austria, Park Geun-Hye in South Korea are just a few of the examples of unconventional politicians who have won elections by their popular appeals to voters. Though perhaps the only similarity between these politicians is their unconventional nature, perhaps their appeals would have failed in an earlier era in which party or candidate loyalties of voters were stronger. Finally, this cautionary tale of the negative consequences of the decline of candidate loyalty in Japan is also limited by several conditions. Though similar outcomes seem to be occurring in a wide variety of nations, Japan also is a unique case because of its frequent elections and weak partisan identification. In addition, my analysis is limited by the fact that many voters do vote according to party or candidate loyalty or a careful consideration of the issues. In any election the low interest/low information voters might be of no consequence because they choose not to vote or are influenced to vote in the same proportions as the high interest voters. Again, there is nothing about low interest/low information voters that makes them necessarily more volatile than other voters, they just have a greater potential to be more volatile in some circumstances. This potential has been seen in Japan in the rapid changes in prime ministers and the back-to-back landslides for different parties in 2005 and 2009. The stability, however, in strong LDP victories in the three subsequent elections also shows how an astute party leader can ride their popularity to political success.

Notes  1 Of course, part of this volatility is due to differences in the electorate between presidential elections and midterm elections. However, these differences can’t explain the popularity of President Obama when he is on the ballot or shifts among groups of voters between these elections that are independent of changes in the electorate.   2 In 1947–8 the government was a coalition with the Socialists and the Democrats (a precursor to the LDP). This coalition government is similar to the Socialist-led government of Murayama Tomiichi in 1994 and 1995. In these and other situations,

158  Democracy and political instability the LDP or its predecessors were in power, even if the prime minister came from a coalition partner of the LDP.   3 The calculations of party shares of the popular vote are based on the district races for all elections even after 1993 when there was a simultaneous party vote conducted along with the district races. In calculating party support using district races, I have counted only the districts in which a party ran candidates. Parties occasionally back the candidates of other parties in some districts, so using an unadjusted national total would artificially decrease national party strength by including districts in which the party ran no candidates. I have also not included independent candidates affiliated with the LDP in these calculations. The true conservative share of the vote would be several percentage points higher in each of these elections if conservative independents were included in the vote totals.   4 The LDP’s vote share in 2012 was not as remarkable as in the other landslide elections (44.6 percent) because third parties won 30 percent of the vote. Strong third parties made it possible for the LDP to win more seats with a smaller percentage of the vote compared to its win in 2014 when third parties won a smaller 14 percent of the vote. The comparison party for the 2017 election is difficult because the Democratic Party split. I used the pacifist alliance for this election which includes the Constitutional Democratic Party, the Socialists, and the Communists.   5 The only comparable swing in the LDP vote was the similar 9.5 percent drop in the LDP vote in 1993, but this drop occurred because many incumbents left the LDP to run under new party labels. The politicians left the LDP in 1993 in contrast to 2009 when the voters left the LDP.   6 There are US presidential elections in which candidates increased their shares of the vote by more than 9 percent. For example President Carter in 1976 improved the Democrats’ share of the presidential vote by much more than 9 percent in contrast to George McGovern’s defeat in 1972. Presidential elections, however, should be more volatile as the number of candidates and the quality of candidates change with each election. It is much more significant when the average for all the legislative elections in a country changes by as much as 9 percent.  7 The Conservative Party’s share of the vote dropped a much more dramatic 27 percent, but part of this drop occurred because of the emergence of the Reform Party which split the conservative vote. Similarly, in 2017 the Liberals enjoyed a phenomenal 20.6 percent increase in their share of the vote.   8 In the Appendix to this chapter I list all of the name changes, mergers, and splits of political parties in Japan that I used in calculating the Pederson measure for Japanese elections.   9 Japan had 5 of 25 such elections. In Europe, there were only 12 of 303 extremely volatile elections. 10 Dalton and Wattenberg report a volatility measure of 41 comparing two early postwar elections. I calculate only 18.7 for these two elections. I can obtain a similar average of 41 for Japan in the 1952 and 1953 elections by (1) not comparing the new Left and Right Socialists to the Socialist Party of the 1949 election, (2) not comparing the Kokumin kyoudoutou’s vote in 1949 to the Kaishintou’s vote in 1952 despite the fact that the party merged into the Kaishintou for the 1952 election, and (3) comparing the Liberal vote in 1952 to the Yoshida Liberal vote in 1953 while treating the Hatoyama Liberal vote in 1953 as a new party that increased its vote from 0 percent in 1952. In contrast my calculations compare each of these parties to its predecessor party vote.

Democracy and political instability  159 11 This generalization applies mostly to conservative or moderate voters. Leftist voters in Japan have long been committed more to their political parties than to the candidates of their parties. 12 For this measurement, rural districts were those election districts in which less than 40 percent of the district’s population lived in what the Japanese government classifies as densely inhabited districts. Urban districts were those districts in which more than 80 percent of the district’s population lived in densely inhabited districts. Because Japan has urbanized generally over time, keeping these thresholds constant meant that the number of districts in each category changed. For example 68 percent of all election districts were rural in 1958, but only 27 percent were rural in 2017. Data for densely inhabited districts for recent elections is from Maeda (2019). 13 I measured competitiveness as the vote percentage difference between the lowest winner in a district and the highest loser in a district. The number of candidates was calculated using the formula for the effective number of candidates, subtracting the district magnitude of the district. The size of the district was calculated by the number of votes won by the average victor in the district. The percent of the voters voting for returning candidates was calculated as the percentage of voters who voted for a candidate who had also run in that same district in the previous election. 14 A better, though less intuitive, way to compare the effect of these five variables on turnout is to compare their standardized coefficients, which takes into account differences in the range of variables and compares the increase in turnout across similar ranges of each independent variable, a one standard deviation change in each independent variable. The standardized coefficients for the entire period 1958 to 2017 are reported in the Appendix, but for only the 1958 election, the urban/rural nature of a district has 28 times the effect of district competition, five times the effect of the number of candidates, three times the effect of district size and 42 times the effect of candidate repetition in explaining turnout. By 2017, urban/rural becomes much like the other factors with only 0.3 times the effect of competitiveness, and six, .4, and 10 times the effect of candidates, district size, and candidate repetition respectively. 15 Because volatility is affected by the number of districts in each group (a smaller number of districts in either category would be more likely to show higher volatility), I defined approximately the most rural third and the most urban third of districts for each election. Thus, moving from 1958 to 2017, the definition of the most rural third of districts changes from districts with a ratio of the population living in densely inhabited districts increasing from only 20 percent in 1958 to 47 percent in 2017. Similarly, the most urban districts were those with at least 60 percent of the population living in densely inhabited districts in 1958 changing to those with 86 percent of the population in such areas in 2017. 16 This data doesn’t show voting in the House of Councilors election, but that swing is mirrored in the subsequent 1990 House of Representatives elections. 17 This analysis is done of conservative candidates because they typically used personal support groups much more than opposition candidates did. From 1958 to 1993 all LDP candidates and LDP-affiliated conservative independents are considered. Under the new electoral system, the phenomenon of LDP-affiliated conservative independents has largely disappeared. For the national swing, only districts in which there was a conservative candidate were used for the calculations, making the swings reported here slightly different than what is typically reported (where the national LDP vote is compared to the total national vote, ignoring that in some districts the LDP does not run candidates because of electoral alliances with another party). All numbers are calculated by adding the votes together rather than averaging

160  Democracy and political instability the percentages from each election district. Urban and rural districts are divided by grouping the third of districts with the lowest or highest percentage of their populations living in the census-defined densely inhabited district. 18 The exceptions to this trend are the 1979–80 and 1980–3 pairings of elections which are possibly explained by the anomalous 1980 election which was one of only two double elections (simultaneous election of both the upper and lower house of the Diet) ever held in Japan. The other double election was the 1986 election. A double election complicates interpreting election results because candidates in the upper house elections are also mobilizing their supporters for the same voting. 19 CNN. 2008. CNN Politics. ‘1 Day Left and Still Undecided.’ November 3rd. edition. cnn.com/2008/POLITICS/11/03/undecided.voters/index.html. Accessed July 1, 2012. 20 The age of candidates was broken down into two dummy variables: old for candidates over the age of 65 and young for candidates younger than 40. Education was also a dummy variable, distinguishing candidates who attended one of Japan’s top universities in contrast to candidates with other educational backgrounds. Relative campaign spending is the percentage of the total campaign spending in the district that was spent by a particular candidate. Relative beauty is the average score of the candidate’s main opponent subtracted from that candidate’s average beauty score. 21 Because there are so few Communist incumbents in district races, I used for a surrogate measure of incumbency, a dummy variable for all Communist candidates who had ever before been elected to the House of Representatives, even if they were not currently the incumbent in that district. I also calculated relative beauty for Communist candidates as their average difference in beauty scores compared to both the LDP and Democratic candidates in their districts. Full regression results are reported in the Appendix. 22 I used an average of all ratings (Japanese and US) in my calculations, but the results do not change if I use only Japanese or only US ratings. 23 Thomas Friedman. 2009. ‘Our One-Party Democracy.’ New York Times. September 8. www.nytimes.com/2009/09/09/Opinion/o9friedman.html. Accessed July 1, 2012. 24 Andy Stern. 2011. ‘China’s Superior Economic Model.’ December 1. Wall Street Journal. online.wsj.com/article/SB10001424052970204630904577056490023451980. html. Accessed July 1, 2012. 25 A survey of Japanese businesses had 61 of the 62 companies that responded supporting the sales tax increase (Asahi Shimbun, June 21, 2010, 6. ‘Shouhizei younin 6 wari zeiritsu 10% ga saita zenkoku juuyou 100 sha chousa’). 26 Asahi Shimbun, June 18, 2010, 3: ‘saninsen manifesto shouhizei tabu o koete.’ 27 ‘Seiken o makaserareu to iu yukensha no shinrai o eru koto ga dekiru no darou ka,’ May 17, 2009, 3: ‘seiken koutai e nattokuryoku o migake.’ 28 Renwick and Pilet (2016, 64) argue that populist and right-wing parties and leaders will be the most supportive of electoral reforms that further weaken parties, creating a spiral effect where the decline of party voting leads to political leaders who enact changes that further weaken party voting.

References Abramowitz, Alan I., and Kyle L. Saunders. 2008. “Is Polarization a Myth?” Journal of Politics 70, no. 2 (April): 542–55. Achen, Christopher H., and Larry M. Bartels. 2016. Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton, NJ: Princeton University Press.

Democracy and political instability  161 Akarui Senkyo, Suishin Kyokai. 2018. “Dai 48 Kai Shuugiin Giin Sousenkyo Zenkoku Ishiki Chousa, Chousa Kekka No Gaiyo” [The National Political Opinion Poll for the 48th General Election for the House of Representatives, Poll Results and Summary]. www.akaruisenkyo.or.jp/wp/wp-content/uploads/2018/07/48syuishikicyosa-1.pdf. Accessed January 14, 2019. Ansolabehere, Stephen, and Philip Edward Jones. 2010. “Constituents’ Responses to Congressional Roll Call Voting.” American Journal of Political Science 54, no. 3 (July): 583–97. Ansolabehere, Stephen, Jonathan Rodden, and James M. Snyder, Jr. 2008. “The Strength of Issues: Using Multiple Measures to Gauge Preference Stability, Ideological Constraint, and Issue Voting.” American Political Science Review 102, no. 2 (May): 215–32. Asahi Shimbun. 1997. Asahi Senkyo Taikan [Asahi’s General Survey of Election]. Tokyo: Asahi Shimbunsha. Bafumi, Joseph, and Robert Y. Shapiro. 2009. “A New Partisan Voter.” Journal of Politics 71, no. 1 (January): 1–24. Baldassarri, Delia, and Andrew Gelman. 2008. “Partisans without Constraint: Political Polarization and Trends in American Public Opinion.” American Journal of Sociology 114, no. 2 (September): 408–46. Baldini, Gianfranco, and Adriano Pappalardo. 2009. Elections, Electoral Systems and Volatile Voters. New York: Palgrave Macmillan. Barnes, Fred. 2011. “Lessons from Canada.” National Affairs, no. 8 (Summer): 77– 90. www.nationalaffairs.com/publications/detail/lessons-from-canada. Accessed August 17, 2012. Bartels, Larry M. 1996. “Uninformed Votes: Information Effects in Presidential Elections.” American Journal of Political Science 40, no. 1 (February): 194–230. Bartels, Larry M. 2000. “Partisanship and Voting Behavior, 1952–1996.” American Journal of Political Science 44, no. 1 (January): 35–50. Bartolini, Stefano, and Peter Mair. 1990. Identity, Competition, and Electoral Availability. New York: Cambridge University Press. Bernstein, Robert A. 1989. Elections, Representation, and Congressional Voting Behavior: The Myth of Constituency Control. Englewood Cliffs, NJ: Prentice Hall. Bertrand, Marianne, and Sendhil Mullainathan. 2001. “Do People Mean What They Say? Implications for Subjective Survey Data.” American Economic Review 91, no. 2 (May): 67–72. Biddle, Jeff E., and Daniel S. Hamermesh. 1998. “Beauty, Productivity, and Discrimination: Lawyers’ Looks and Lucre.” Journal of Labor Economics 16, no. 1 (January): 172–201. Brennan, Geoffrey, and Loren Lomasky. 1993. Democracy and Decision: The Pure Theory of Electoral Preference. New York: Cambridge University Press. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: John Wiley & Sons. Canes-Wrone, Brandice, David W. Brady, and John F. Cogan. 2002. “Out of Step, Out of Office: Electoral Accountability and House Members’ Voting.” American Political Science Review 96, no. 1 (March): 127–40. Caplan, Bryan. 2007. The Myth of the Rational Voter: Why Democracies Choose Bad Policies. Princeton, NJ: Princeton University Press. Carpini, Delli, X. Michael, and Scott Keeter. 1996. What Americans Know about Politics and Why It Matters. New Haven, CT: Yale University Press.

162  Democracy and political instability Carsey, Thomas M., and Geoffrey C. Layman. 2006. “Changing Sides or Changing Minds? Party Identification and Policy Preferences in the American Electorate.” American Journal of Political Science 50, no. 2 (April): 464–77. Christensen, Ray. 1998. “The Effect of Electoral Reforms on Campaign Practices in Japan: Putting New Wine into Old Bottles.” Asian Survey 38, no. 10 (October): 986–1004. Christensen, Raymond V. 1994. “Electoral Reform in Japan: How It Was Enacted and Changes It May Bring.” Asian Survey 34, no. 7 (July): 589–605. CNN, Election Center 2008. “Results, Exit Polls.” National Exit Poll. www.cnn.com/ Election/2008/results/polls/#val=usp00p6. Accessed February 28, 2012. Converse, Philip E. 1964. “The Nature of Belief Systems in Mass Publics.” In Ideology and Discontent, edited by David E. Apter, 206–61. New York: Free Press of Glencoe. Converse, Philip E., and Georges Depeux. 1962. “Politicization of the Electorate in France and the United States.” Public Opinion Quarterly 26, no. 1 (Spring): 1–23. Cunningham, Michael R., Alan R. Roberts, Anita P. Barbee, Perri B. Druen, and Cheng-Huan Wu. 1995. “Their Ideas of Beauty Are, on the Whole, the Same as Ours: Consistency and Variability in the Cross-Cultural Perception of Female Physical Attractiveness.” Journal of Personality and Social Psychology 68, no. 2 (February): 261–79. Curtis, Gerald L. 1971. Election Campaigning Japanese Style. New York: Columbia University Press. Cutler, Fred. 2002. “The Simplest Shortcut of All: Sociodemographic Characteristics and Electoral Choice.” Journal of Politics 64, no. 2 (May): 466–90. Dahl, Robert A. 1961. Who Governs? Democracy and Power in an American City. New Haven, CT: Yale University Press. Dalton, Russell J., Ian McAllister, and Martin P. Wattenberg. 2000. “The Consequences of Partisan Dealignment.” In Parties without Partisans: Political Change in Advanced Industrial Democracies, edited by Russell J. Dalton and Martin P. Wattenberg, 37– 63. Oxford: Oxford University Press. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Addison Wesley. Englis, Basil G., Michael R. Solomon, and Richard D. Ashmore. 1994. “Beauty before the Eyes of Beholders: The Cultural Encoding of Beauty Types in Magazine Advertising and Music Television.” Journal of Advertising 23, no. 2 (June): 49–64. Fiorina, Morris P., Samuel J. Abrams, and Jeremy C. Pope. 2006. Culture War? The Myth of a Polarized America, 2nd ed. New York: Pearson Longman. Gallup. 2019. Presidential Job Approval Center. www.gallup.com/poll/124922/ Presidential-Approval-Center.aspx. Accessed May 29. Haggard, Stephan, and Robert R. Kaufman. 1992. “The Political Economy of Inflation and Stabilization in Middle-Income Countries.” In The Politics of Economic Adjustment: International Constraints, Distributive Conflicts, and the State, edited by Stephan Haggard and Robert R. Kaufman, 271–315. Princeton, NJ: Princeton University Press. Japor. 2011. Yoron Chousa Indekksu [Index of polls]. August 24. www.socnii.ac.jp/ Japor/search/kensaku.html Jiji tsushinsha. 1981. Sengo Nihon No Seitou to Naikaku, Jiji Yoron Chousa Ni Yoru Bunseki [Political Parties and Cabinets in Postwar Japan, an Analysis Using Jiji Press’s Polling]. Tokyo: Jiji Press.

Democracy and political instability  163 Jiji tsushinsha. 1992. Nihon No Seitou to Naikaku 1981–1991, Jiji Yoron Chousa Ni Yoru Bunseki [Japan’s Parties and Cabinets 1981–1991, an Analysis Using Jiji Press’s Polling]. Tokyo: Jiji Press. Jiji tsushinsha. 2011. Jiji Dotto Comu [Jiji Dot com]. Naikaku Shiji Ritsu No Suii [Changes in Cabinet Approval Ratings]. August 24. http://jiji.com/jc/v?p=ve_ pol_politics-support-pgraph Jiji tsushinsha. n.d. Jiji Yoron Chousa Tokuhou [Special Report of Jiji Polling]. Various volumes and issues. Johnston, Victor S., and Juan C. Oliver-Rodriguez. 1997. “Facial Beauty and the Late Positive Component of Event-Related Potentials.” Journal of Sex Research 34, no. 2, 188–98. Jones, Jeffrey M. 2011. “Democratic Party ID Drops in 2010, Tying 22-Year Low.” January 5. gallup.com/poll/145463/democratic-party-drops-2010-tying-year-low. aspx. Accessed October 24, 2011. Keith, Bruce E., David B. Magleby, Candice J. Nelson, Elizabeth A. Orr, Mark C. Westlye, and Raymond E. Wolfinger. 1992. The Myth of the Independent Voter. Berkeley, CA: University of California Press. Kobayashi, Yoshiaki. 2012. Malfunctioning Democracy in Japan: Quantitative Analysis in a Civil Society. New York: Lexington Books. Krauss, Ellis S., and Benjamin Nyblade. 2005. “‘Presidentialization’ in Japan? The Prime Minister, Media and Elections in Japan.” British Journal of Political Science 35, no. 2 (April): 357–68. Krauss, Ellis S., and Robert J. Pekkanen. 2011. The Rise and Fall of Japan’s LDP: Political Party Organizations as Historical Institutions. Ithaca, NY: Cornell University Press. Kuklinski, James H., Paul J. Quirk, Jennifer Jerit, David Schwieder, and Robert F. Rich. 2000. “Misinformation and the Currency of Democratic Citizenship.” Journal of Politics 62, no. 3 (August): 790–816. Layman, Geoffrey C., Thomas M. Carsey, John C. Green, Richard Herrera, and Rosalyn Cooperman. 2010. “Activists and Conflict Extension in American Party Politics.” American Political Science Review 104, no. 2 (May): 324–46. Lewis-Beck Michael, S., William G. Jacoby, Helmut Norpoth, and Herbert F. Weisberg. 2008. The American Voter Revisited. Ann Arbor, MI: University of Michigan Press. Lin, Jih-wen. 2016. “The Consequences of Constitutional Systems on Party Systems.” In Mixed-Member Electoral Systems in Constitutional Context: Taiwan, Japan, and Beyond, edited by Nathan F. Batto, Chi Huang, Alexander C. Tan, and Gary W. Cox, 52–72. Ann Arbor, MI: University of Michigan Press. Lupia, Arthur. 1994. “Shortcuts versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections.” American Political Science Review 88, no. 1 (March): 63–76. Lupia, Arthur, and Mathew D. McCubbins. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? New York: Cambridge University Press. Maeda, Ko. 2009. “Has the Electoral-System Reform Made Japanese Elections PartyCentered?” In Political Change in Japan: Electoral Behavior, Party Realignment and the Koizumi Reforms, edited by Steven R. Reed, Kenneth Mori McElwain, and Kay Shimizu, 47–66. Stanford, CA: Walter H. Shorenstein Asia-Pacific Research Center. Maeda, Ko. 2019. “DID% of Japan’s Single-Member Districts, after the 2017 Redistricting.” February 18. politicalscience.unt.edu/~maeda/. Accessed April 10, 2019.

164  Democracy and political instability Mainwaring, Scott. 1991. “Politicians, Parties, and Electoral Systems: Brazil in Comparative Perspective.” Comparative Politics 24, no. 1 (October), 21–43. Mainwaring, Scott. 1999. Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press. Mair, Peter. 2002. “In the Aggregate: Mass Electoral Behaviour in Western Europe 1950–2000.” In Comparative Democratic Politics: A Guide to Contemporary Theory and Research, edited by Hans Keman, 122–40. Thousand Oaks, CA: Sage. Masahiro, Yamada. 2010. “2009 Nen Sousenkyo Ni Okeru Seiken Koutai to Suingu Uoteingu” [The Power Shift on the 2009 General Election in Japan and Swing Voting]. Senkyo Kenkyuu [Japanese Journal of Electoral Studies] 26, no. 2: 5–14. Megumi, Hosoda, Eugene F. Stone-Romero, and Gwen Coats. 2003. “The Effects of Physical Attractiveness on Job-Related Outcomes: A Meta-Analysis of Experimental Studies.” Personnel Psychology 56, no. 2 (Summer): 431–62. Miller, Warren E., and J. Merrill Shanks. 1996. The New American Voter. Cambridge, MA: Harvard University Press. Mobius, Markus M., and Tanya S. Rosenblat. 2006. “Why Beauty Matters.” American Economic Review 96, no. 1 (March): 222–35. Munetaka, Tanaka. 1997. Seiji Kaikaku Roku Nen No Doutei [The Six-Year Journey to Political Reform]. Tokyo: Gyousei. Nemoto, Kuniaki, and Chia-hung Tsai. 2016. “Post Allocation, List Nominations, and Preelectoral Coalitions under MMM.” In Mixed-Member Electoral Systems in Constitutional Context, Taiwan, Japan, and Beyond, edited by Nathan F. Batto, Chi Huang, Alexander C. Tan, and Gary W. Cox, 165–93. Ann Arbor, MI: University of Michigan Press. Nie, Norman H., Sidney Verba, and John R. Petrocik. 1979. The Changing American Voter. Cambridge, MA: Harvard University Press. Norris, Pippa. 2011. Democratic Deficit: Critical Citizens Revisited. New York: Cambridge University Press. O’Donnell, Guillermo. 1973. Modernization and Bureaucratic Authoritarianism: Studies in South American Politics. Berkeley, CA: Institute of International Studies. Pedersen, Morgens N. 1979. “The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility.” European Journal of Political Research 7, no. 1 (March): 1–26. Popkin, Samuel I. 1994. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns, 2nd ed. Chicago, IL: University of Chicago Press. Prisching, Manfred. 1995. “The Limited Rationality of Democracy: Schumpeter as the Founder of Irrational Choice Theory.” Critical Review 9, no. 3 (Summer): 301–23. Quattrone, George A., and Amos Tversky. 1988. “Contrasting Rational and Psychological Analyses of Political Choice.” American Political Science Review 82, no. 3 (September): 719–36. Redlawsk, David P. 2002. “Hot Cognition or Cool Consideration? Testing the Effects of Motivated Reasoning on Political Decision Making.” Journal of Politics 64, no. 4 (November): 1021–44. Reed, Steven R., and Michael F. Thies. 2001. “The Consequences of Electoral Reform in Japan.” In Mixed-Member Electoral Systems: The Best of Both Worlds?, edited by Matthew Soberg Shugart and Martin P. Wattenberg, 152–72. Oxford: Oxford University Press. Renwick, Alan, and Jean-Benoit Pilet. 2016. Faces on the Ballot: The Personalization of Electoral Systems in Europe. New York: Oxford University Press.

Democracy and political instability  165 Roberts, Kenneth M., and Erik Wibbels. 1999. “Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations.” American Political Science Review 93, no. 3 (September): 575–90. Roper Center. 2011. “JPOLL: The Japanese Public Opinion Database.” August 24. http://roperweb.ropercenter.uconn.edu/jpoll/Jpoll.html. Scheiner, Ethan. 2006. Democracy without Competition in Japan: Opposition Failure in a One-Party Dominant State. New York: Cambridge University Press. Sikk, Allan. 2005. “How Unstable? Volatility and the Genuinely New Parties in Eastern Europe.” European Journal of Political Research 44, no. 3 (May): 391–412. Somin, Ilya. 1998. “Voter Ignorance and the Democratic Ideal.” Critical Review 12, no. 4 (Fall): 413–58. Tavits, Margit. 2008. “On the Linkage between Electoral Volatility and Party System Instability in Central and Eastern Europe.” European Journal of Political Research 47, no. 5 (August): 537–55. Tews, Michael J., Kathryn Stafford, and Jinfei Zhu. 2009. “Beauty Revisited: The Impact of Attractiveness, Ability, and Personality in the Assessment of Employment Stability.” International Journal of Selection and Assessment 17, no. 1 (March): 92–100. Valentino, Benjamin A., Paul K. Huth, and Sarah E. Croco. 2010. “Bear Any Burden? How Democracies Minimize the Costs of War.” Journal of Politics 72, no. 2 (April): 528–44. Wilson, Rick K., and Catherine C. Eckel. 2006. “Judging a Book by Its Cover: Beauty and Expectations in the Trust Game.” Political Research Quarterly 59, no. 2 (June): 189–202. Wittman, Donald A. 1995. The Myth of Democratic Failure: Why Political Institutions Are Efficient. Chicago, IL: University of Chicago Press. Zaller, John R. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press.

Appendix

A. Classifications of party mergers and divisions in Japan The source for this information is Asahi Shimbun (1997, 126–7, 440) and various internet sources.   1 The Kyoudou minshutou merged in 1947 with the Kokumintou to form the Kokumin kyoudoutou.   2 The Nihon shimpotou became the Minshutou in 1947.   3 The Nihon jiyuuyou became the Minshu jiyuutou in 1948.   4 The Minshu jiyuutou became the Jiyuutou in 1950.   5 The Kokumin kyoudoutou merged with part of the Minshutou to form the Kokumin minshutou in 1950. The remainder of the party merged with the Jiyuutou. I only compare the new Kokumin minshutou with the Kokumin kyoudoutou. I cannot compare it with the Minshutou because this party split between two different parties. The Kokumin minshutou became the Kaishintou in 1952.   6 The Kaishintou became the Nihon minshutou in 1954.  7 The Jiyuutou split between the Nihon jiyuutou and Jiyuutou in 1953. These two parties are distinguished by their leaders’ names, Hatoyama and Yoshida. These two parties subsequently merged together under the name Jiyuutou.   8 In 1948 the Roudousha noumintou split off from the Nihon shakaitou.   9 The Nihon shakaitou split into two parties, the Saha shakaitou and the Uha shakaitou in 1951. These two parties merged together recreating the Nihon shakaitou in 1955. The Roudousha noumintou also joined the Nihon shakaitou at this time. 10 In 1960 the Minshu shakaitou split off from the Nihon shakaitou. 11 In 1970 the Minshu shakaitou changed its name to the Minshatou. 12 In 1977 the Shakai shimin rengou split off from the Nihon shakaitou, later changing its name to the Shakai minshu rengou. 13 In 1976 the Shin jiyuu kurabu split off from the Jiyuu minshutou; in 1986 it rejoined the party. 14 The Nihon shintou formed in 1992. 15 The Koumeitou formed in 1964.

Democracy and political instability  167 16 The Shintou sakigake and Shinseitou were formed from the LDP in 1993. 17 In 1994, the Shakai minshu rengou joined the Nihon shintou. 18 In 1994 the Nihon shintou, Koumeitou, Minshatou, and Shinseitou merged forming the Shinshintou. 19 In 1996 the Minshutou formed drawing its members primarily from the Nihon shakaitou and the Shintou sakigake, but both of these other parties remained in existence. 20 In 1996 the Nihon shakaitou changed its name to the Shakai minshutou. At this time the Shin shakaitou split off from the party. 21 In 1997 the Shinshintou disbanded. Some of its members joined the LDP or the Minshutou. Others formed the following new parties: Jiyuutou, Shin koumeitou, Hoshutou. In computing volatility for the 2000 election I compared the 1996 results for the Shinshintou with the results in 2000 for the Jiyuutou, Shin Koumeitou, and Hoshutou. 22 Shintou sakigake partially becomes Midori no kaigi in 2002, but becomes irrelevant before that. 23 In 2003, the Jiyuutou officially joined the Minshutou. 24 After the 2003 election Hoshu Shintou joined with the LDP. 25 In 2012 Nihon ishin no kai was formed from members of JDP, LDP, Minna no tou. In computing volatility for the 2012 election I treated Nihon ishin no kai as a new party. 26 In 2012, Nihon mirai no tou becomes Seikatsu no tou. I treat them as the same political party. 27 In 2014 Nihon Ishin no kai breaks up into Nihon ishin no tou and Jisedai no tou. Both new parties are small, so they are included in the Other category for calculations. The Ishin no tou merges with the Minshutou in 2016 to form the Minshintou. 28 For the 2017 election, Kibou no tou was treated as a new party because it pulled members from several existing parties. 29 The Rikken minshutou was treated as a successor to the Minshintou because nearly all the members of the Rikken minshutou came from the Minshintou. 30 The Kibou no tou merges with the Rikken minshutou to form the Kokumin minshutou in 2018. Party names Jiyuu minshutou, Liberal Democratic Party (LDP) Minshutou, Democratic Party Nihon shakaitou, Japan Socialist Party, also known as the Shakai minshutou, Social Democratic Party (JSP) Roudousha noumintou, Farmer Labor Party (Rounoutou) Shintou sakigake, New Party Harbinger (Sakigake) Minshatou, Democratic Socialist Party (DSP) Shinseitou, Renewal Party Shinshintou, New Frontier Party (NFP)

168  Democracy and political instability Jiyuutou, Liberal Party Hoshutou, Conservative Party Koumeitou (Shin koumeitou), Clean Government Party (CGP) Nihon shintou, Japan New Party (JNP) Shakai minshu rengou, Social Democratic League (SDL) Shin jiyuu kurabu, New Liberal Club (NLC) Saha shakaitou, Left-Wing Socialist Party (L. shakaitou) Uha shakaitou, Right-Wing Socialist Party (R. shakaitou) Shintou Nippon, New Party Nippon (NPN) Nishon ishin no kai, Japan Restoration Party, also Ishin no tou Nihon mirai no tou, Tomorrow Party of Japan Jisedai no tou, Party for Future Generations Seikatsu no tou, The People’s Life Party Kibou no tou, Party of Hope Rikken minshutou, Constitutional Democratic Party Minshintou, Democratic Party Kokumin minshutou, Democratic Party for the People

B. Regression results for turnout in Japan Table 7.3  Dependent variable: district level turnout, Japanese House of Representatives elections, 1958–2009 Independent variables

Coefficient (standard error) Standardized coefficient

Intercept Urban percent Competition Number of candidates District size Repeating candidates Interact year60/urban Interact year63/urban Interact year67/urban Interact year69/urban Interact year72/urban Interact year76/urban Interact year79/urban Interact year80/urban Interact year83/urban Interact year86/urban Interact year90/urban Interact year93/urban

81.735 (.912)** −.222 (.017)** −.127 (.008)** 1.068 (.128)** .000036 (.000)** .009 (.004)* −.051 (.024)* −.090 (.024)** −.020 (.023) −.064 (.023)** −.058 (.023)* −.017 (.022) −.105 (.022)** .013 (.022) −.049 (.022)* −.057 (.022)* .038 (.022) .050 (.023)*

−.630 −.162 .075 .095 .022 −.039 −.069 −.016 −.053 −.054 −.016 −.102 .014 −.051 −.058 .040 .054

Interact year96/urban Interact year00/urban Interact year03/urban Interact year05/urban Interact year09/urban Interact year12/urban Interact year14/urban Interact year17/urban

.053 (.019)* .056 (.019)* .080 (.019)** .126 (.019)** .132 (.019)** .159 (.019)** .155 (.019)** .115 (.019)**

.093 .097 .141 .223 .233 .283 .275 .205

N = 4016 **p value < .01 *p value < .05 adjusted R2 .812 Independent variables for each election year (1960–2017) were also included in the regression, but are not reported in this table. Both the election year variable for 1958 and the interaction term for 1958 and Urban were omitted from the regression, making 1958 the comparison year for all other year-specific variables.

C. Regression results for the physical attractiveness of candidates Table 7.4  Dependent variable: vote share won by LDP candidates in district elections in Japan’s 2003 House of Representatives election Independent variables

Coefficient (standard error)

Intercept LDP predicted percent Elite schooling Young Old Female District Incumbent Opponent is Incumbent Campaign spending Beauty

−.016 (.035) .843 (.073) ** −.006 (.009) .002 (.013) −.024 (.011)* −.053 (.024)* .038 (.011)** −.012 (.012) .167 (.030)** .016 (.005)**

N = 245 **p value < .01 *p value < .05 adjusted R2 .650

Table 7.5  Dependent variable: vote share won by Democratic candidates in district elections in Japan’s 2003 House of Representatives election Independent variables

Coefficient (standard error)

Intercept Dem. predicted percent Elite schooling Young Old Female District Incumbent Opponent is Incumbent Campaign spending Beauty

−.047 (.033) .908 (.073) ** .012 (.010) −.001 (.009) −.045 (.017)** −.025 (.013)* .046 (.011)** −.019 (.010) .192 (.031)** .012 (.004)**

N = 244 **p value < .01 *p value < .05 adjusted R2 .653

Table 7.6  Dependent variable: vote share won by Communist candidates in district elections in Japan’s 2003 House of Representatives election Independent variables

Coefficient (standard error)

Intercept Com. predicted percent Elite schooling Young Old Female Previous Incumbent Opponent is Incumbent Campaign spending Beauty

−.023 (.004)** 1.094 (.038) ** −.002 (.008) −.001 (.003) .001 (.006) .010 (.002)** .018 (.005)** .001 (.002) .015 (.019) .002 (.001)

N = 244 **p value < .01 *p value < .05 adjusted R2 .850

8 Acceptance of election results

Lesson 8: Japan’s election system illustrates an unusual information paradox. The more input that voters have in an election, the less legitimate the election system will seem.

Zombie politicians in Japan When Japan ran its first election under its new, mixed electoral system, one of the most surprising outcomes of the new system was the creation of what were called zombie politicians. The label zombie seems to have been a creation of English-language commentators on Japanese politics, based on the Japanese word fukkatsu which means resurrected or restored (rather than zombie). The imagery behind all of the labels, however, is consistent. Some candidates lose their races in a specific district, but because they are also candidates on regional party lists, some of these district losers are brought back from electoral death and nevertheless win seats in the national Diet. They serve, however, from a regional party list rather than as the representative of a particular election district. Thus, whether these candidates are said to be resurrected, brought back from the dead, restored, or made into zombies, the concept underlying each name is the same. The appearance of these MPs under the new electoral rules attracted an unusual amount of attention in Japan. For example, in a compendium of information about this first election under the new system, the issue of zombie politicians appeared repeatedly. In a one-page overview of the new electoral system, the journalist authors of the compendium noted that the new electoral system had not produced all of the results that its proponents intended. They added that ‘there are many problems that arose after actually using the new system, such as dual-listed candidates who lose in their districts but have a “restored victory” on party lists’ (Asahi Shimbun, 1997, 13). The authors call this phenomenon ‘a problem of the new electoral system that has specially attracted attention after the election’ (Asahi Shimbun, 1997, 13). The authors then elaborate on this problem in a prefecture by prefecture summary of the election results. In the section on Tokyo elections, one of their three points was, ‘8 Zombies Elected’ (Asahi Shimbun, 1997, 28). In addition to

172  Acceptance of election results eight losers from district elections being elected on the party lists of the Tokyo region, two Tokyo districts ended up with three representatives. One was the winner from that district, but two losers in both districts also won seats on party lists, making each of the districts have, in effect, three representatives. The authors then tie this result to the malapportionment of Japan’s districts noting that one of those districts easily becomes Japan’s most overrepresented district in terms of representatives per voter. They further point out that one of the zombies could manage only a fifth place showing in his district. He won so few votes in his district that he was required under Japanese law to forfeit his election deposit of 3 million yen or approximately $30,000. This candidate, despite winning fewer than 14,000 votes, served in the Diet in contrast to the losing candidate in a different Tokyo district who won more than 67,000 votes in his district, yet was placed too low on his party list to win a seat in the Diet. They conclude by calling these resurrected candidates a ‘contrary result’ that clearly illustrates problems of the new electoral system. On a separate page they list the 84 members of the Diet elected on party lists after losing their district races (Asahi Shimbun, 1997, 107). This intense interest in Japan in resurrected beings or zombies in the legislature continues to the present. A search in Japan’s largest newspaper, Yomiuri Shimbun, reveals 3,153 articles that use the phrase ‘zombie victory’ from the years 1996 to 2019. The phrase does not appear in the newspaper prior to the 1996 election. Eleven articles still used the phrase in 2018. This massive media attention to what Asahi Shimbun calls a contrary result bore fruit with a 2000 electoral reform package that prohibited any loser in a district race that won fewer than 10 percent of the total votes cast in the district from being eligible to win a seat on a party list. This new requirement seemed to calm some of the upset about zombie politicians, but in practice the new rule only affected the candidates of the Social Democratic Party of Japan (Socialists). Only small parties have a difficult time winning a minimum 10 percent of the district vote, and Japan’s two other small parties typically don’t dual list candidates in both districts and party lists. Even at the time of passage, it was noted that there were 94 zombies elected in the previous 2000 election, and the new rules would have only affected three Socialists who were next in line to take a seat if one of the elected Socialists resigned or died in office. In fact, facing this new requirement, the Party election chair said that the Party would simply put non-duallisted candidates at the bottom of its party lists so that seats would be filled if all the dual-listed candidates were disqualified for not winning at least 10 percent of the district vote. (Yomiuri Shimbun, May 18, 2000, 4). The perverse and illogical effect of this new rule is shown by comparing results before and after the new rule went into effect. In the first election under the new rules (2003) all of the Socialists elected on party lists who also ran in a district race managed to win more than 10 percent of the district vote. However in the next election in 2005, two of the six party-list winners were insulated from potential disqualification by being put only on the party list and not also running in a district race. Because these placeholder candidates were

Acceptance of election results  173 not dual listed, the 10 percent requirement did not apply to them. One of these winners was Hosaka Nobuto, the same man cited in Asahi Shimbun’s summary of the 1996 election who came in fifth place in his district and also lost his election deposit. So what had changed between 1996 and 2005? The same candidate from the same party was elected on the same Tokyo party list. The only difference was that in 2005 his name was only on the party list and not listed as candidate in a district race. The voters were not allowed to express a preference for or against Mr Hosaka in 2005, in contrast to 1996, so there could be no outrage about how unpopular he was despite his winning a seat in the Diet. This example shows the oddity of the concern about zombie politicians. Many parties put candidates on their party lists who would not be able to win many votes in a district race. No concern is ever raised about these candidates because no one knows for sure exactly how popular or unpopular the candidates might be. In contrast, the Japanese system makes it possible to see, at least for dual-listed candidates, exactly the popularity or unpopularity of specific candidates on a party list. This information, gained from the voters, undercuts the legitimacy and approval levels of the candidates on the party lists and also the election system itself because such a system allows such unpopular candidates to be elected. These events highlight an interesting puzzle about the support for and legitimacy of election systems. Why does it matter that a candidate who is known to be unpopular is elected on a party list but it is fine to elect many other candidates on the same party list who are potentially just as unpopular, but the extent of their unpopularity is merely a conjecture, unsupported by any specific electoral results? There is an information paradox of electoral systems which explains this puzzle, and this information paradox also appears in many other electoral settings. Understanding this paradox helps explain the popularity of a wide variety of electoral practices, from party quotas to promote women candidates, to Iran’s religious authorities not authorizing candidates, to the popularity of proportional representation electoral systems over majoritarian electoral systems. In all of these settings, from zombies in Japan to mullahs in Iran, the information paradox helps explain commonly accepted electoral practices and preferences. Because all electoral systems (indeed all political institutions) distort the public will in some ways, electoral systems that gather less information about the public will are seen as more legitimate and will be more popular. In contrast, electoral systems that gather more information from voters will seem to have more glaring inconsistencies or distortions, and as a result, these electoral systems will be seen as less legitimate and will be less popular than other alternatives. Thus the paradox: the most democratic electoral systems that give voters the greatest choice and ability to express multiple and detailed preferences are precisely the electoral systems that will be seen as the least legitimate because the extensive information gained from the voters will highlight any distortions of the electoral system. In contrast, electoral systems that ask for less information from the voters will not have their corresponding distortions highlighted and will appear to be better and more

174  Acceptance of election results legitimate than other electoral systems. Because the information paradox of electoral systems affects all electoral systems, the examples and comparisons used in this chapter cover a wide variety of regime types and electoral systems.

Zombies and their Caribbean and American cousins It isn’t quite true that zombie politicians are noticed only in Japan (Ferrara, Herron, and Nishikawa, 2005, 103). ‘That these legislators are referred to by names like “zombies” or ‘back door MPs’ calls attention to their perceived illegitimacy’ (Shugart and Tan, 2016, 271). Zombie politicians exist in every electoral system which is mixed and allows the dual listing of candidates, and at least twenty countries use or have recently used such rules for their elections.1 In fact, the whole point of the dual listing of candidates is to create zombies. Dual listing makes it possible for a candidate to win under either of two different parts of the electoral system. Obviously, some of these candidates will lose under one part of the electoral system while winning under a different part of the system. Thus, when the Japanese tried to solve the problem of zombie politicians, they prohibited only the election of candidates who do very poorly under one of the systems. The great bulk of the candidates who won party-list seats after losing district races were allowed to continue winning seats, for after all, that is precisely what the system was designed to allow. It seems, though, that zombie politicians don’t attract the same notoriety in other countries that they attract in Japan. Though some other countries dual list candidates, the popularity or unpopularity of a dual-listed candidate does not typically become an issue. For example, in Germany’s most recent 2017 election, 1,674 candidates were dual listed of the 4,828 candidates. Many of the party-list winners in Germany were also losers in district races, but the fact that they won a seat in the Bundestag despite losing their district race does not seem to be an issue of importance in discussions of the German electoral system. For example, the Green Party elected 11 candidates off of its party list in the province of Bavaria. All 11 candidates also ran in district races. The 11 candidates won between 3.6 and 16.3 percent of the vote in their district races. Seven of these 11 candidates would not have met the Japanese requirement that a district candidate must win at least 10 percent of the vote in a district race to be elected from a party list. Thus, a relatively common occurrence in Germany is an extreme occurrence in Japan that is also prohibited by law. This difference in reactions can be explained by one significant difference between Japan and Germany in the rules of how dual-listed candidates are elected. On all German party lists, candidates are clearly and unambiguously ranked, and these rankings are unaffected by how many or how few votes a dual-listed candidate actually wins in a district race. In Japan, in contrast, parties are allowed to list multiple candidates at the same ranking position on a party list, and the candidates that come closest to winning their district races are the ones that win seats on the party lists. This Japanese rule focuses attention on how many votes a party-list candidate won in a district race, and this attention then

Acceptance of election results  175 begins to undercut the legitimacy of candidates like Mr Hosaka who didn’t win many votes in his district race. At least Mr Hosaka, however, won 13,904 (5.9 percent) in his much derided 1996 victory. In contrast, Erhard Grundl won a seat on the Green Party list in Bavaria in 2017 despite winning only 4,602 votes (3.6 percent) in the Straubing district. The difference in reaction to these two examples lies with the fact that district numbers are deemphasized in Germany. They play no role in deciding who wins party-list seats. Mr Grundl won election because he was 8th on the Green Party list in Bavaria, and the Greens won enough votes to elect 11 candidates off of their list. The number of votes that he won in his district race was irrelevant to his favorable position on the Green Party list. A similar version of the information paradox occurs in some Caribbean democracies. Governments there have been known to appoint to their upper houses people who had previously lost an election for a seat in the lower house. The discomfort with this practice is heightened when those same appointees are then named to cabinet positions (Donahoe, 2002, 49). This practice is not controversial for other appointees to the upper house (who have never run for a lower house seat), even when those appointees also serve in the cabinet. It is possible to speculate about whether these appointees, who were never candidates in any election, could have won an election if they had run, but such speculations would not be supported by information about actual voter preferences. In contrast, appointees become controversial only when voters know exactly how unpopular the appointee was because he or she had lost a previous election. Survey results also weakly support this difference between knowing a candidate’s unpopularity in contrast to only assuming that a certain candidate is unpopular. In a survey experiment of 475 Americans, a randomly selected half of the survey respondents were asked to rate the fairness on a scale from 0 (not fair at all) to 100 (perfectly fair) of several electoral outcomes. Unknown people winning seats on party lists with no evidence of their popularity earned a fairness rating of 18.6. In contrast losers in a previous election later winning seats on a party list because of favorable placement on the list earned a slightly lower fairness rating of 16.2.2 The 2008 presidential primary election in the United States provides another example of this information paradox. Party nomination procedures used by both the Democratic and Republican Parties in the United States are actually more democratic and more inclusive than those used by most political parties in other well-respected democracies. Very few parties in other countries use primaries that are open to all voters willing to declare an affiliation with a party. Rather, limited elections among dues-paying party members, selection by application and committee review, votes by party MPs, or selection by a small group of party leaders are much more typical methods of candidate selection. These methods can produce nominees that the voters, if given a choice, would not have supported. Typically, however, such procedures are not controversial because there is no information about voter preferences for

176  Acceptance of election results the nominees. In addition, most parties typically select popular candidates, the candidates that the voters would most likely have selected if they had voted in a primary-type nomination process.3 The more democratic and open procedures used in the United States, however, are controversial because the extensive information about voter preferences collected in the process reveals small distortions of the process which stem from: (1) the use of caucuses rather than primaries in some states, (2) the Republican use of winner-take-all primaries in some states, and (3) the Democratic use of superdelegates. Though any distortions which result from these rules are part of an extremely democratic selection system and similar or worse distortions occur under any selection method, the huge amount of information about voter preferences collected by the extensive system of primaries and caucuses in the United States makes even small distortions in the selection process appear glaringly obvious. Though such distortions pass unnoticed when the nominee easily wins the nomination, the 2008 Democratic race was the exception because the outcome ran a significant chance of being decided by small distortions. In addition, the ample information available about actual voter preferences fueled extensive criticisms of the nomination process and rules. Specifically, though Barack Obama won more delegates from the state selection processes than Hillary Clinton won, Clinton actually won slightly more votes than Obama. Obama had more delegates because he won more of the caucus states where turnout for caucuses is much lower than turnout for a primary vote. Clinton initially also had an advantage in the superdelegates, delegates not selected by the primary and caucus system. For a time there was also a concern that she might win the nomination based on this advantage. In contrast, if the Democratic Party had selected its nominee in 2008 by a simple conference of party leaders or party MPs as is done in many other democracies, the selection process, though less democratic, would also have been less controversial. Less information about voter preferences would have been available, making it harder to criticize either the outcome or the process as antidemocratic.

Expectations of equality These several examples of the information paradox illustrate the importance of expectations of equality and the form that those expectations take in established democracies. Though theoreticians of democracy disagree on some of the components of democracy, equality plays an important role in any defensible definition of democracy (Katz, 1997, 100–1). Similarly, McGann (2006) gives equality primacy in his definition of democracy. Even Thompson’s (2002, 9) position that some concerns may take precedence over equality is still qualified by a test of ‘equal respect.’ These authors explicate an idea that is generally accepted in all democracies: voters should be equal, and the expressed preferences of voters should be reflected equally in election outcomes. Election

Acceptance of election results  177 outcomes that contradict voter preferences suggest that greater weight is being given to the preferences of some voters over the preferences of other voters, violating deeply held expectations of equality. Election distortions that do not result in an outcome contrary to the public will are of less concern because the violation of norms of equality produces no significant consequences. Thus, few voters care when a party or candidate wins the most votes and wins an election even if the party or candidate failed to win a majority of the votes cast. The outcome seems fair because the winner won the most votes. In contrast, a losing party or candidate that actually won more votes than the eventual winner is a shocking outcome, even if the actual difference in votes between the two is tiny. The fact that the National Party won four more seats with 4,000 fewer votes than the Labour Party in New Zealand’s 1981 election is a shocking outcome. In contrast, the National Party won 50 percent of the seats in parliament in 2014 with only 47 percent of the vote, but this distortion in voter preferences passed unnoticed. The inequality of the first example is obvious: the more popular party should be the victor. The inequality of the second example is muddled: who cares that a majority of voters seemed to be voting against the National Party? The Party won the most votes, so it seems natural that it also won the election. The public reaction to an electoral distortion depends on (1) whether there is clear information about voter preferences, and (2) whether the distortion produces a consequential outcome. Despite the importance of public attitudes about election outcomes and election systems, the public, however, easily misunderstands the operation of all election systems. For example, few voters understand that all election procedures, to some degree, distort electoral outcomes (Farrell, 1997, 7; Rush, 1998). In addition, few voters understand that it is impossible to design, for example, a single set of election rules that maximizes or protects minority rights, communities of interest, governability, and equality (Rush and Engstrom, 2001, 78; Reeve and Ware, 1992, 156–7). The desirable features of election systems are actually contradictory, and the public rarely recognizes this contradiction. For example, single-member districts give representation to local areas but make it harder to provide fair representation to dispersed minorities. Similarly, some argue that greater proportionality leads to less government accountability (Ferrara, Herron, and Nishikawa, 2005, 145; Mann, Przeworski, and Stokes, 1999, 46–8). No set of electoral rules is perfect. All represent tradeoffs. Americans complain about the confusing paradoxes of decentralized parties and divided government, Italians and Hollanders complain about government coalitions whose formation seems divorced from electoral outcomes. The English complain about underrepresentation of some voters and government dictatorship between elections by the representatives of others. … Students of elections and democracies will always have something to complain about because no set of election arrangements can satisfy conditions for all the desirable electoral rules. (Powell, 2000, 18–19)

178  Acceptance of election results In contrast to these realities, the average voter erroneously believes that election rules operate like a calculator, the data is entered in and the appropriate result comes out. Voters are uninterested in the details of a specific election system, assuming that nearly all election systems report election results fairly. They become concerned only when it is obvious that an election rule has operated unfairly, and the information collected from the voters along with the amount of the distortion both play a large role in deciding whether or not public attitudes are affected by the distortion.

Democratic reversals The most consequential of electoral distortions is an election outcome that contradicts the directly expressed will of the voters. As these are a reversal of the democratic will, as expressed by the voters, they are, for lack of a better name, democratic reversals. Democratic reversals range from the most egregious example of a candidate with fewer votes than an opponent actually winning the election to parties that win more votes nationwide than their opponents but win fewer seats than those opponents. Even the most politically obtuse Americans know how the Electoral College can and has created democratic reversals. Hillary Clinton won almost 3 million more votes than Donald Trump in 2016, but she lost to Trump. Al Gore won 500,000 more votes than George Bush in 2000, but he lost to Bush. In 1876, Samuel Tilden won 250,000 more votes than Rutherford Hayes, but he lost to Hayes. It is not my purpose to argue about the merits or demerits of the Electoral College in the United States. However, comparing the democratic reversals in 1876, 2000, and 2016 to the 1960 presidential election shows how the voters’ reaction to the most egregious type of electoral distortion, a democratic reversal, also depends on the availability of clear data about voter preferences. A strong case can be made that the 1960 presidential election was also a democratic reversal with 50,000 more Americans voting for Richard Nixon than voted for the winner, John Kennedy. Critics of the Electoral College and others regularly note the other democratic reversals (Sterling, 1978, 68; Adkinson and Elliot, 1997, 78; Dahl, 2003, 31; Sabato, 2007, 134), but the 1960 election is almost never mentioned, except in a few academic papers (Longley and Pierce, 1996, 48–50; Kallina, 1988, 106–7; Gaines, 2001).4 After the 2000 election, critics of that outcome jumped back to 1888 for evidence of democratic reversals in the United States rather than mentioning the questionable outcome of the 1960 race (Babington, 2000, A1; Will, 2000, A29; Berke, 2000, A1; Stout, 2000, A26; Leonard, 2000, D1). The 1960 election is typically considered to not have been a democratic reversal because clear information does not exist about popular vote totals in this election. Disgruntled Democrats in some Southern states ran slates of unpledged electors that supported Senator Harry Byrd and not the party nominee, John Kennedy. In Mississippi, voters had the choice between Byrd, Kennedy, or the Republican candidate, Richard Nixon. The Byrd slate of electors won a plurality of the vote in Mississippi

Acceptance of election results  179 and won all of Mississippi’s electoral votes. Votes cast for these Byrd electors in Mississippi are not counted in Kennedy’s popular vote totals; he is only credited with the votes cast for the Kennedy slate of electors in Mississippi. In contrast, in neighboring Alabama, voters cast 11 separate votes for each of Alabama’s 11 electoral votes. In a Democratic primary held prior to the general election, Alabama’s Democrats split in electing their 11 electors to the Alabama Democratic slate. Five of the winners supported Kennedy; the remaining six were Byrd supporters, representing a split in the Democratic vote in Alabama, similar to the split that also emerged in Mississippi voting. On election day, Alabama’s 11 democratic electors defeated all 11 Republican electors, however Kennedy only garnered the electoral votes of his five pledged electors. The remaining six unpledged electors voted for Senator Byrd. The ultimate outcome of the popular vote total in Alabama is ambiguous. Credible arguments can be made to justify different outcomes. Most credit Kennedy with the average of all of the votes cast for all 11 Democratic electors even though a majority of those electors were anti-Kennedy electors. Alternatively, some credit him with only the pro-rated average of the votes cast for his five electors. Alabama voters had to cast 11 separate votes, and they had to choose in every instance between only one Democratic elector and only one Republican elector, unlike Mississippi where voters could choose between three complete slates of electors. Ambiguity exists because we don’t know what an Alabama voter’s intent was if she cast five votes for Kennedy electors and six votes for anti-Kennedy electors. Nevertheless, if the popular vote total for Kennedy is reduced proportionally by the votes cast for Byrd electors (more than 180,000 votes), then Nixon actually won the popular vote in 1960 by more than 50,000 votes, while Kennedy won the electoral vote. The 1960 election is not recognized as a democratic reversal because of the complexity of the information that surrounds the popular vote total in Alabama. The peculiar circumstances in Alabama make it easy to reach contradictory conclusions about Kennedy’s popular vote. In contrast, if Alabama had used the same voting system that Mississippi had used, then clear information would exist about the preferences of Alabama voters between Byrd and Kennedy, and the 1960 election would be widely recognized as a democratic reversal. Information also plays a decisive role in responses to the much more common type of democratic reversal, a reversed outcome for the composition of an entire legislature. For example, in 1978 and 1981, 11,000 and 4,000 more New Zealanders cast ballots for the Labour Party than the National Party, but in both elections the National Party won enough parliamentary seats to select New Zealand’s government. Similarly, in the 1996 House of Representatives elections, 190,000 more Americans voted for the Democratic Party than voted for the Republican Party, but it was the Republican Party which won control of the House. The anomalous election outcomes in New Zealand are well documented and widely known, and they eventually helped lead to a change in New Zealand’s electoral system. The anomalous outcome of the 1996 House

180  Acceptance of election results race is unknown and never mentioned. Information differences explain these broadly different reactions to the wrong party winning a legislative election. Democratic reversals in legislatures occur for a variety of reasons. In the single-seat districts of the United States, they occur because of variations in support across districts, malapportionment of districts, differentials in turnout across districts, variations in the numbers of qualified voters across districts, and the gerrymandering of districts (Campbell, 1996). Some of these factors can affect any single-seat district system, and they also occasionally can affect the outcomes of proportional representation electoral systems. Surprisingly, democratic reversals in legislative elections occur quite often. Powell (2000, 130) goes so far as to propose a 15-percent failure rate, representing the approximate percentage of elections that are democratic reversals in elections for legislatures in countries that use majoritarian electoral systems. Table 8.1 shows the incidence of democratic reversals in a variety of electoral settings, focusing on legislative elections in countries that use single-seat districts.5 Of the outcomes reported in Table 8.1, one of the most interesting is the 1996 election for the US House of Representatives. Both Congressional Quarterly (2008) and the Clerk of the US House of Representatives (Office of the Clerk, 2008) report national vote totals showing the Democrats winning several hundred thousand more votes than the Republicans despite the fact that the Republicans won more seats than the Democrats. Two other commentaries report nationwide vote totals with the Republicans in first place in the popular vote (Scammon, McGillivray, and Cook, 1998, vii–viii; Abramson, Aldrich, and Rohde, 1998, 86). Either outcome, however, is typically ignored in the press and in the many volumes written on this election. For example, Cohen (1997), Cesar and Busch (1997), Jacobson (1997) and Hershey (1997) Table 8.1  Consequential democratic reversals for selected countries Country and time period

Percentage of elections Reversal elections as democratic reversals

US House of Representatives, 1870–2018 US Presidency 1872–2016 Canada 1867–2015 United Kingdom 1900–2005 New Zealand 1938–93 Japan 1945–2005 Australia 1901–2007

10.7

Various countries using proportional electoral systems

2.5

13.5 11.9 17.9 10.5 4 20.5

1880, 1888, 1914, 1916, 1930, 1942, 1996, 2012 1876, 1888, 1960, 2000, 2016 1896, 1926, 1957, 1962, 1979 1910, 1910, 1929, 1951, 1974 1978, 1981 1947 1903, 1925, 1949, 1954, 1961, 1969, 1987, 1990, 1998 Austria (1953, 1959); Sweden (1914); Spain (1993); Finland (1929, 1948)

Acceptance of election results  181 do not mention national vote totals for house races in their analyses; others report national percentages without additional comment (Center for Voting and Democracy, 1998). In contrast to the lack of American interest in popular vote total for the House of Representatives, reversal elections in Britain, Canada, or New Zealand regularly attract press attention and occasionally lead to electoral reform efforts (Farrell, 1997, 13; Hart, 1992, 279; Amy, 1993, 210–12; Dahl, 1989, 149). Democratic reversals, when they occur in other countries are regularly noted in the press, even if no action is taken in response to the reversal. For example the 1979 democratic reversal in Canada was noted immediately in press reports of the election. ‘The Conservatives rode to their minority win on only 36 per cent of the popular vote while the losing Liberals received 40 per cent’ (Toronto Globe and Mail, May 24, 1971, 1; May 23, 1971, 1). After the 1978 reversal election in New Zealand, the press also noted immediately the occurrence of a reversal: ‘The Labour Party gained 11,000 more votes than National in the general election on Saturday—633,328 against 622,041. But it still did not become the Government’ (New Zealand Herald (Auckland) November 25, 1978, 4). The lack of attention paid to democratic reversals in US elections stems not from an unwillingness to look at the popular vote. For example, observers, including House Democratic leader Richard Gephardt, pointed out that control of the House of Representatives would have switched in 1996 and in 1994 if only a few thousand voters in several key districts had switched their votes (Dilulio, 1997, 168; Cesar and Busch, 1997, 134). Others noted that President Clinton won his Electoral College landslide in 1996 with less than 50 percent of the popular vote (Scammon, McGillivray, and Cook, 1998, vii). US politicians and commentators do comment on popular vote totals, but surprisingly, they never mention that a democratic reversal has occurred. Information availability explains this anomaly. Most democracies, including federal countries such as Canada and Australia, have a national election bureaucracy which administers national elections, tallies the votes cast, and publicizes the official vote count. In contrast, the United States lacks a nationwide election bureaucracy and has few national election rules. In fact, in 2000 only eight US states had uniform statewide election systems (Pound, 2002, 127). This lack of national consistency makes it difficult to calculate a definitive popular vote total for House of Representative elections. Similar to the problems of interpretation of the Alabama popular vote in the 1960 presidential election, US elections are vexed by voting idiosyncrasies that vary from state to state. For example, calculating the national popular vote totals for the 1996 US House elections is complicated by the following factors: (1) Three Louisiana districts and one Florida district had no votes cast because the race was uncontested, but in eighteen other uncontested districts, votes were cast, accruing only to the winner’s party and minor party candidates. (2) The outcomes in two other Louisiana districts were decided in an open primary held in September. No votes were recorded in these two districts in November

182  Acceptance of election results balloting, making a total of six uncontested districts in which no votes were cast in November. (3) Thirteen congressional districts in Texas voted in an open primary on election day with runoff elections occurring in December in three districts. In one of the runoff districts, both candidates were Republicans. (4) Only two of Louisiana’s seven congressional districts voted in November (in runoff elections), but in one of these districts both candidates were Democrats. (5) In Connecticut and New York, major party candidates can also run on minor party labels. These complicating factors make it possible to calculate several different totals for the parties. Indeed, Congressional Quarterly (2008) Voting and Elections Collection credits the Democrats with 360,000 more votes nationwide than the Republican Party which won a majority of seats. The Office of the Clerk (House of Representatives) (2008) agrees, but finds a smaller Democratic popular vote advantage of 272,708.6 Scammon, McGillivray, and Cook, in contrast, find that the Republicans won more votes (275,833) by counting open primary rather than runoff voting in Texas and Louisiana and counting minor party votes for major party candidates in New York and Connecticut (1998, iv). None of the calculations try to impute what national vote totals might have been if votes were included for the 22 other House districts in which one or both parties did not receive any votes. The appropriate outcome, however, is less important for this analysis than the simple fact that there is no single, universally accepted popular vote total for US House of Representatives elections. The competing calculations and justifications for different totals make it much less likely that observers will even note or comment on a democratic reversal when it occurs in the United States. Problems with calculating party vote totals also mean that other democratic reversals, when they occur in US legislative elections, are typically ignored. Occasionally more voters in one state vote for US representatives from one party but a different party wins more seats in the state delegation to the US House of Representatives. From 1870 to 2004, such reversals occurred in 5.8 percent of House delegations from states.7 Amy (1993, 29–30, 37) also notes their occurrence in the 1980 congressional elections. Similarly, democratic reversals in races for state legislatures also occur with some frequency, but again, their occurrence usually passes unnoticed (Amy, 1993, 37). Massicote (2005, 104) also notes the frequency of reversals in Canadian provincial elections.8 The informational problems of US House elections are not unique to the United States; they are, however, much more severe than potential ambiguities in popular vote totals that occur in other countries or even voting for the US president. For example, the presidential popular vote total is typically not affected by districts in which a major party ran no candidate or the state was uncontested. US presidential elections also have very few instances in which the voting in November is not the decisive voting for president. Voting in other nations also lacks this variation in voting procedures because usually a national election administration exists that requires uniform procedures for

Acceptance of election results  183 national elections. In addition, though there might be some uncontested races, the number of uncontested races is lower in most other countries because these countries have stronger political parties that run candidates in every district. For example in the Canadian democratic reversal of 1979 the Conservatives and the Liberals ran candidates in every one of Canada’s ridings. Even the New Democrats ran a candidate in all but one of Canada’s ridings. In contrast, Hart (1992, 74, 140–1) notes that early discussions of electoral reform in Britain were hampered by the problem of how to count nationwide vote totals when parties did not run candidates in many districts. Though confounding factors exist in other countries, they are much less common than they are in the United States. In addition to a greater number of uncontested races, US legislative elections suffer from additional forms of informational uncertainty that make it nearly impossible to arrive at an undisputed, national vote total for a US political party. It is likely that differentials in information explain disparate reactions to democratic reversals, but, of course, other factors are at play in explaining these responses. Some electoral rules are more legitimate or traditional, making them less likely to be challenged. Others serve popular goals, making any distortions that occur seem a necessary cost to achieve a laudable goal. Political parties, provincial or state boundaries, election district boundaries, ballot structures, qualifications for candidacy, and quotas for women and minority candidates are all examples of common and legitimate election rules practices that potentially distort election outcomes. The legitimacy or purpose of a distorting institution affects the public acceptance of any distorted election outcomes produced by that institution. For example, voters are more likely to accept quotas for women candidates to redress past discrimination than voters are likely to accept malapportioned district boundaries that advantage rural voters. Despite these various confounding factors, in a clean, hypothetical situation that sets aside problems of information difficulties, US survey respondents react as expected to different types of democratic reversals. A democratic reversal in the race for the presidency earned a fairness rating of 32.9 in contrast to a democratic reversal in a House election which was perceived as more fair, with a fairness rating of 52.3. US survey respondents, at least, understand the unavoidable reasons that might lead to a reversal in a legislature, and they see such a reversal as more fair. In contrast, the same respondents recognize the fundamental unfairness in a reversal for a specific race. Similarly, political entrepreneurs can also play an important role in framing the debate about a particular election rule. Equally distorted election outcomes can occur in two different countries, but if a political party in one country makes an issue of the distortion, the distortion is much more likely to attract public attention and outrage, potentially leading to electoral reform. Though these other factors are important, the availability of information is a crucial factor in determining the public reaction to anomalous election outcomes. Indeed, in the next example I compare two equally legitimate systems to increase the number of women in a legislature. The two systems produced

184  Acceptance of election results nearly identical outcomes, but one system creates information that will weaken support for the system while the other creates no information that could be used to attack that country’s quota system as anti-democratic. Though other explanatory factors are important, the availability of information plays a crucial and independent role in determining public attitudes about democratic reversals.

Quotas for women Quotas for women candidates are a much more popular tool for increasing women’s representation in legislatures than reserving seats in the legislature for women. Only 26 of the 149 countries that mandate some form of gendered representation reserve seats for women (IDEA, 2019). The information paradox explains the popularity of quotas as a tool to increase the legislative representation of women over reserved seats for women. For example, consider the radically different operation of rules put in place to ensure that both Iraq and Afghanistan would have significant numbers of women in their newly elected national legislatures. Both rule systems were successfully implemented, with women winning 25.5 percent of seats in 2005 in Iraq and a comparable 27.7 percent of seats in 2005 in Afghanistan. Though the outcome of the two quota systems was remarkably similar, the different methods used produce radically different amounts of information. In Iraq, parties are legally required to list at least one woman in every three seats on the party lists used for proportional voting. Thus, the quota is implemented in advance of the election by an institution (a political party) that is separate from the election itself. The preference for women candidates is made by a party and not by the voters or ad hoc adjustments to lists of winners, locating the potentially distorting effects of the quota away from the voters’ preferences as expressed at the ballot box. One might argue that Iraqi voters might have preferred different candidates than the ones that were on the party lists. Perhaps they would have preferred that half of the candidates be women or perhaps they might have preferred that only 10 percent of the candidates be women. These are simply unfounded speculations because the election system collects no information as to the actual preferences of Iraqi voters for male or female candidates. The election system used in Iraq recorded the preferences of voters for political parties but not preferences for specific candidates or specific genders of candidates. In contrast, Afghanistan used multimember districts in which voters could cast one ballot for one of many candidates. Each of Afghanistan’s 34 provinces was an election district with multiple winners being elected from each province (the single nontransferable vote—SNTV—system as it is known in the electoral systems literature). The number of representatives elected from each province ranged from a low of two in Nuristan to a high of 33 in Kabul. The quota for women representatives was put in place by rules that required that approximately one fourth of the candidates elected from a province be

Acceptance of election results  185 women. Thus, in Baghlan province, two of the eight representatives were to be women. However, in the 2005 elections, the two highest placed women running in Baghlan came in 10th and 15th. They were each elected, however, jumping ahead of men who had won more votes and were initially in positions 7, 8, 9, 11, 12, 13, and 14. Nationwide, 68 women were elected to Afghanistan’s Wolesi Jirga in 2005, and only 18 of these 68 women won their positions without displacing any men. The remaining 50 women each displaced an average of 12.5 men who each won more votes than the woman candidate. The most extreme example was Parween Momand Talwasa, the third woman elected in Kuchi province. She won 1,241 votes, coming in 46th in a ten-seat district, but because she was the third highest woman in terms of votes, she was elected to the tenth and last position in Kuchi province, leaping ahead of 34 other men who collectively won 99,818 votes. Each of the 34 men that she displaced won, on average, more than double (2,935) the number of votes that she won. The quotas used in Iraq and Afghanistan are equivalent in many ways. Both elected approximately the same percentages of women in 2005. Both seem to have been well received by the voters and outside observers (Ballington and Dahlerup, 2006; Reynolds, 2006, 114–15; Coleman, 2006, 24). The information paradox, however, makes the Afghan electoral system more fragile than the Iraqi system because Afghan’s system will remind voters at every election that the reserved seats are distorting the actual preferences of the voters. In contrast, potentially similar distortions of the voters’ actual preferences may be occurring in Iraq, but the electoral system collects no information about those preferences, so the distortions, even if they are occurring, pass without notice or comment. Again, there is a paradox that collecting more detailed information from voters in Afghanistan (preference for candidates rather than just for parties) makes the inevitable distortions more glaring, the greater amount of information collected from the voters leads to greater criticism of the results of the election and of the electoral system itself. This difference also shows up again in survey results. Half of survey respondents, randomly selected, were asked to rate gender quotas done by parties, and the other half of respondents rated gender quotas accomplished by reserved seats. The reserved seats earned a fairness score of only 15.6, in contrast to the party quotas earning a fairness ratio of 27.1.9

Majoritarian and proportional electoral systems A final type of election distortion in which the information paradox plays a crucial role is the contrast between distortions which regularly occur in majoritarian electoral systems and the much less noticed distortions that regularly occur in proportional election systems. Part of the popularity of proportional election systems can be explained by the fact that its distortions are accompanied by little information regarding voter preferences. In contrast, majoritarian election systems create clear information about distortions, amplifying the criticisms of majoritarian electoral systems.

186  Acceptance of election results The strength of majoritarian systems is that they select governments well (Baldini and Pappalardo, 2009, 43). In contrast, proportional systems are best at selecting legislatures that faithfully reflect the party preferences of voters. Each system’s strength is the other system’s weakness (Powell, 2000; Baldini and Pappalardo, 2009, 39–40; Cain, 2015, 92). Legislatures in majoritarian countries are skewed towards the largest parties with smaller parties sometimes excluded from any representation. Governments in proportional countries are often made and unmade by party negotiations rather than the direct choices of voters. The distortions typical in majoritarian systems are most commonly distortions in legislative representation; thus these distortions are usually accompanied by electoral data that shows that the distortion occurred. In contrast, any distortions created when parties negotiate the participants in a ruling coalition are not accompanied by recorded voter preferences about the composition of government coalitions. Though it is possible to speculate that the voters might have objected to an alliance that excluded the most popular party from the government, no one knows for sure that such a coalition government would be contrary to voter preferences. Similarly, coalition negotiations might add a party to the government that the voters had just repudiated at the polls. Though this outcome might be objectionable to many voters, there is no record of voter preferences about that particular party’s (or any party’s) participation in the government. There are, therefore, a variety of distortions that can occur under both majoritarian and proportional electoral systems. A party can win more votes than another party but actually win fewer legislative seats than the other party. The largest party could be excluded from the government. A government can be formed that is supported by only a minority of voters. Parties could be added to or removed from government in a way that is inconsistent with changes in the popularity of that party as expressed by the voters. Of these four potential distortions, only the first directly contradicts the will of the voters. It seems clearly unfair that one party could win 2 million votes and win 15 seats in a legislature and a different party win only 1.5 million votes but win 20 seats in the same legislature. In contrast, who is to say that the voters actually wanted the largest party to be a part of the government? Does it matter that a majority of voters cast ballots for parties that are not included in the government? If a party’s share of the vote dropped by 5 percent in the most recent election, but the party joined the coalition government after the election, is that a violation of the voters’ will? These three last examples are all questionable because voters don’t directly express a preference for governments when they vote. It isn’t odd to assume that voters prefer that their supported party also be a part of the government, but this information is not directly collected from the voters. Thus, proportional electoral systems, which have more coalition governments and more political parties will have more distortions in the government formation process but will have fewer distortions in legislative representation. Because of the clear information available about distortions in legislative representation,

Acceptance of election results  187 proportional electoral systems will seem to be more fair and will be more popular than majoritarian electoral systems. Majoritarian electoral systems have fewer distortions in government formation, but these distortions are much less apparent. The information paradox helps explain why proportional electoral systems are more popular than majoritarian electoral systems. Figure 8.1 begins to show this difference. All democratic reversals, even if they are small, are included in Figure 8.1 (Table 8.1 only counted democratic reversals that affected which party controlled the government). Surprisingly, there are a fair number of democratic reversals, even in some countries that use proportional electoral systems. If a country with a proportional election system has regional districts or thresholds that exclude some parties, reversals can occur. These distortions, though, often pass unnoticed because they are not consequential. No one really cares that in the 2007 Finnish elections the Swedish People’s Party won 9 seats with only 126,520 votes while the Christian Democrats won only 7 seats, despite winning 134,790 votes.10 Grouping the countries reported in Figure 8.1, countries with proportional electoral systems had 19.4 percent of elections with a democratic reversal.11 Majoritarian countries had a similar ratio of 23.4 elections with reversals.12 Semi-proportional countries saw 48.9 percent of elections with reversals.13 Thus, though democratic reversals that change the composition of the government are concentrated in countries that have majoritarian or semi-proportional electoral systems, when all reversals are counted, even inconsequential reversals, the record of countries with proportional electoral systems is indistinguishable from the record of countries with majoritarian electoral systems. Even more interesting is the analysis of distortions in the types of governments created. One might expect, for example, that the disproportionality of majoritarian systems would make it so that a greater number of governments in majoritarian systems were elected by a minority of the voters than would occur

80

Percent of elections

70 60 50 40 30 20 10

Country

Figure 8.1 All democratic reversals

ly Ita

A N ustr et he ia rla nd s Sw ed D en en m ar Po k rtu ga G er l m U an ni y te U ni d S ta te te d s K in gd om Ire la nd N o N ew rwa Ze y al an d Ca na d a A us tra lia Ja pa Be n lg iu m Fr an ce Fi nl an d

0

188  Acceptance of election results 120

Percent of elections

100 80 60 40 20

ly Ita

rtu

ga Ire l la nd D U e ni te nm d K ark in gd om Sw ed en Ca na da Ja pa n A us tra lia N or N ew way Ze al an d

G

Be l

Po

gi um er m an y Fr an ce

s

es at

te

d

St

d

nd ni

U

N

et

he

rla

an nl

Fi

A

us tri

a

0

Country

Figure 8.2 Governments elected by a minority of voters

under proportional election systems. However, proportional electoral systems are also much more likely to result in coalition governments, and minority governments are common in coalition governments. For example, the frequency of governments being elected without the support of a majority of voters can be just as high under proportional electoral systems as it is under majoritarian electoral systems. Figure 8.2 shows the percentage of recent governments that have been elected by a minority of voters. Proportional countries are well represented among those countries with most or all of their governments being elected by a minority of the voters. Of the governments analyzed, 59.6 percent of the governments elected in majoritarian countries and a similar 57.8 percent of the governments elected in proportional countries were elected by a minority of the voters.14 Katz (1997, 164) presents data that seems to contradict this conclusion (only 21 percent of governments in proportional countries are elected by a minority of voters), but his calculations specifically exclude governments that do not hold a majority of seats in a legislature. When those governments are included, only 46.1 percent of governments in proportional countries are elected by a majority of the voters. Another distortion in voter preferences can occur when the political party with the greatest support of the voters is excluded from participation in the government. Figure 8.3 shows the relative occurrence of this distortion. This distortion occurs with greater frequency in countries that use proportional election systems (21.4 percent) than in countries that use majoritarian electoral systems (6.4 percent). Semi-proportional systems see this distortion in 17 percent of cases. Powell (2000, 147) mentions this distortion, citing Dutch governments formed in 1977 and 1981 that excluded the plurality-winning Labor Party (Dahl, 1989, 149). Survey data also marginally supports the conclusion that information can play a role in reaction to a distortion about coalition government

Acceptance of election results  189 60

Percent of elections

50 40 30 20 10

Ja pa n er m an A y us tri a Ca U n ni ad te a d St at e Be s lg i N et um he rla nd s Ire la nd Fr N ew anc Ze e al an d Fr an ce D en m a A rk us tra lia N or w ay G

l ga

gd in

K

d te U

ni

om

ly

rtu

Ita

Po

Fi

nl

an

d

0

Country

Figure 8.3 Largest party not in government

participants. Half of a survey sample gave a fairness rating of 30.6 to a coalition government that excluded the largest party (a party that won 40 percent of the seats in a legislature). When the other half of this survey sample were additionally told that a majority of voters approved the largest party being in the government while one of the three coalition parties was not approved by a majority of voters as a governing party, the fairness rating drops to 28.1 percent. Though the difference is not statistically significant, the difference shows that the perception of fairness of an election procedure is affected by whether or not the result of the procedure directly contradicts expressed voter preferences. A final distortion occurs when a party that is increasing in popularity is dropped from the government or a party that is declining in popularity is added to a government. Both actions seem to contradict voter preferences. Of 397 government changes that coincided with elections that were analyzed, 23.7 percent of them occurred in a direction that was contrary to voter preferences. 27.4 percent of the changes in proportional countries, 22.1 percent in semi-proportional countries, and 0 percent in majoritarian countries were in directions that were contrary to expressed voter preferences (see Figure 8.4). Powell (2000, 73–4) concurs: Election results were followed by bargaining between most of the parties. The party gains and losses had only indirect impact, at best, on the formation of governments. Sometimes the connections between party gains/ losses and their exclusion/inclusion in government seemed quite perverse. Powell and others, nevertheless, make a compelling argument that proportional systems produce governments that are actually more responsive than governments produced by majoritarian systems, despite the distortions described above. Powell cites the permeability of most proportional systems on

190  Acceptance of election results 80 Percent of elections

70 60 50 40 30 20 10 ar k us tri a

y

A

m

an

en D

en

um

er m

G

gi

Be l

ed

an

Sw

d

Ja p

ly

an nl

Ita

Fi

li Ze a al an N d et he rla nd s Ire la nd N or w ay

us tra

A

ew

N

a

om

ad te

d

K

in

gd

l ga

Ca n ni

rtu

Po

U

U

ni

te

d

St

at

es

0

Country

Figure 8.4  Change in government inconsistent with party popularity

other dimensions such as committee structures. Other commentators disagree with Powell’s conclusions (Stokes, 1999; Maravall, 1999; Golder and Stramski, 2007). Still others argue that distortions in the formation of governments are not a serious problem (Farrell, 1997, 154–6; Gallagher, Laver, and Mair, 1992, 303–4). At a minimum though, understanding the information paradox better informs this debate: the distortions that occur under proportional systems are perhaps less serious as Powell and others argue, but they are surely less likely to be noticed or criticized than the distortions that occur under majoritarian systems. Perhaps proportional election systems should be favored because they are more likely to be accepted and supported by voters, but it is the information paradox which creates this seeming advantage of proportional systems over majoritarian systems. Proportional systems create legislatures that are much closer to voter preferences than those legislatures created by majoritarian systems. Majoritarian systems, in contrast, create governments that are much closer to the assumed preferences of the voters for governments, but this advantage is not supported by data because voters aren’t asked to choose governments in elections, only candidates or parties. In majoritarian systems vote choice is closely connected to the government choice, because the winning party typically forms the government. In proportional systems, the connection between vote choice and the eventual coalition government is weaker. Voters are less sure that a vote for a party will result in that party being a part of the government. Thus, it is hard to even identify, much less criticize, this distortion of the democratic will by proportional electoral systems.

Practical implications of the information paradox A first implication is the often unrecognized importance of accessible and unambiguous information about voter choices, independent of the information paradox. Perceptions matter, and information is crucial in evaluating

Acceptance of election results  191 electoral systems. Thus, proponents of electoral reform should do all that they can to simplify and make accessible information about election rules and other democratic procedures. Attitudes about democracy generally are affected by public perceptions of the fairness of election rules. The best designed set of election rules may still lead to apathy and cynicism if the public perceive the rules as unfair. Alternatively, poorly designed election rules may nevertheless contribute to government legitimacy if people believe that the rules work fairly. Accessible information is therefore crucial in building and maintaining popular support for both election rules and democracy. Some activists have learned this lesson. For example, proponents of proportionality in US elections realize that the title ‘instant runoff’ system makes more sense to voters than the label ‘single transferable vote’ which is the name that political scientists use (FairVote, 2008). ‘Instant runoff’ makes a complex electoral system easily understood and therefore more palatable to the average voter. Hart’s (1992, 268) partial explanation for why Britain failed to adopt a more proportional electoral system during the 1920s is similar. Proponents of electoral reform did not use the most easily understood examples. They relied heavily on the obscure example of Tasmanian state elections rather than the more familiar national elections in neighboring Ireland, opening the door for one parliamentary opponent to quip ‘Is there no lesson to be learnt from Lithuania? How do they elect their borough councils in Yugoslavia and aldermen in Lapland?’ The specialized and esoteric nature of election rules only increases the burden on those who would discuss these rules to make information as accessible as possible to the general public. In addition to the importance of information generally, the information paradox clearly shows how the information that is created by an electoral system can reap long-term consequences for the ultimate success or failure of those election rules. For example, if a politician wants to sabotage a proposed election rule, he or she should make sure that any potential distortions of voter choices caused by the election rules should be glaringly obvious if and when they occur. In contrast, if a politician wants to make it more likely that an election rule will remain in place, he or she should design the system to collect and record less information about the direct preferences of the voters. Instead, voter preferences should be mediated by other political institutions that neither collect nor record voter preferences about specific election outcomes. This second lesson on the creation of viable electoral rules is, in its most crude form, a lesson that is not lost on some dictators. They understand that it is easier to constrict the voters’ choice of candidates before an election than it is to overturn voters’ choices once they have recorded their preferences for candidates. For example, Iran’s Guardian Council disqualified 40–50 percent of candidates prior to voting in the 1996 and 2004 elections (Saghafi and Ehsani, 2004, 16–17). Disqualifying candidates is an egregious distortion of voter preferences, but, similar to party quotas for women candidates, this distortion occurs before the election when no information about voter preferences exists. Thus, the distortions are less controversial than they would otherwise be.

192  Acceptance of election results The information paradox also suggests two areas of study of electoral systems that should be examined more closely: (1) the extent to which distortions are actually consequential and (2) public opinion and the public reaction to elections and electoral rules. Most studies of electoral systems develop sophisticated measures that make it possible to evaluate different electoral systems on important factors such as the proportionality or fairness of electoral systems (Rae, 1967, 183; Loosemore and Hanby, 1971; Taagepera and Shugart, 1989, 104–11; Gallagher, 1991; Lijphart, 1994, 58–67; Taagepera and Grofman, 2003), the number and viability of parties (Rae, 1967, 47–64; Laakso and Taagepera, 1979; Taagepera and Shugart, 1989, 77–91), and the congruence between voter policy preferences and government actions (Golder and Stramski, 2007; Huber and Powell, 1994; McDonald and Budge, 2005; Powell, 2000). In contrast, the information paradox suggests that the result of a distortion is more important, at least to electoral system legitimacy and public perceptions, than simply measuring the amount or size of the distortion. Conventional measures of election systems assess distortions or other features on a continuum. It is likely, though, that most voters don’t care about distorted electoral outcomes unless the distortion produces outcomes that directly contradict voter preferences. For example, few people care that Bill Clinton won two presidential elections with large seat bonuses of 25 and 21 percent.15 These large seat bonuses are unremarkable because the outcome of the election was consistent with voter preferences (Clinton won the presidency, and he also won a plurality of votes). George Bush’s seat bonus in the 2000 election, in contrast, was a tiny 2.5 percent, one tenth the disproportionality of the Clinton election of 1992, but his election was extremely controversial because he won the election despite having won fewer votes than Al Gore. The extensive literature on proportionality and seat bonuses is important and valuable, but this literature misses an important qualitative difference between election outcomes that directly contradict public preferences and simple distortions that are consistent with voter preferences. Public opinion and public reaction to election rules should also take a higher priority in analyses of electoral systems. With few exceptions (Johnsson and Maddock, 2002, 81; Ferrara, Herron, and Nishikawa, 2005, 145), the extensive literature on election systems typically neglects the opinions of average citizens when evaluating different election systems. In many situations, this omission is appropriate. Why should uninformed and sometimes misguided public opinion be considered in a sophisticated discussion about the merits of a particular election system? Yet, this omission can become dangerous when public distrust of a particular electoral system leads to institutional reform. For example, Hart (1992, 26) notes that Thomas Hare wrote his first work on electoral systems because he was upset that so many talented politicians were defeated in the 1857 British elections. Similarly, discussion of proportional representation was reignited in the early 1900s by John H. Humphreys who was ‘disturbed by the results of the elections to London Borough Councils.’ ‘Sometimes the party with least votes obtained the majority of seats’ (145). In a modern example, Matthew Soberg Shugart (2005, 26) notes

Acceptance of election results  193 although I cannot speak for other scholars in this field, I certainly suspect that many can tell a personal story similar to my own, which was a self-conscious entry into the field through observation of, and often misgivings about, the way electoral systems have worked in my own country and elsewhere. Indeed, public opinion about the need for reform is often a crucial factor in explaining why election reform occurs. For example, though many politicians seemed unwilling to entertain meaningful proposals for electoral reform in Italy, New Zealand, Japan, or Thailand, the public demand for reform provided the crucial impetus for shepherding along radical electoral reform proposals in all four countries (Katz, 2005, 69; Christensen, 2000, 25–6; Englehart, 2003, 270–2). In fact, public demand for reform was necessary in each of these countries to overcome the self-interested reluctance of many politicians to seriously consider election reform. The lessons of the information paradox in the analysis of electoral systems are: (1) public perceptions matter, (2) voters will only notice distortions that seem to directly contradict the expressed public will, and (3) support and legitimacy of electoral systems can be enhanced by restricting the amount of information gathered from the voting public. The information paradox should not dictate that it is better to gather less information from voters, though such an outcome might be a practical concern to be kept in mind by those arguing for electoral reform. Rather, it is important to remember that evaluations of electoral systems are subtly influenced by the amount of information gathered from the voters and the extent to which that information produces outcomes that violate expectations of equal treatment. Being aware of this influence will make discussions and analyses of electoral system more informed. The lessons of zombie politicians in Japan illustrates these points well. Zombie politicians, in contrast to placeholder politicians, became an issue in Japan because the election system gathered data about their popularity and the method of winning party-list seats called attention to this data because it decided winners based on that same data from district races. In contrast, placeholder candidates (people at the bottom of party lists who only win seats if the party has an unexpectedly large victory) are likely even less popular than zombie politicians. The extent of their unpopularity, though, is unknown, so they aren’t noticed; their numbers aren’t counted or reported, and they don’t even have a name—I had to coin the term placeholder candidates to describe this type of MP in Japan. This information paradox created a public outcry for reform, and the Japanese government obliged by creating a new rule that required that the most egregious zombie politicians be replaced by placeholder MPs. Will Japan further change its rules to completely eliminate the possibility of zombie politicians? The probable answer is no. The whole system of dual listing was created precisely to give losers in one constituency a chance to win election in a different constituency. As long as the rules prevent the most egregious examples of

194  Acceptance of election results losers becoming winners, there is little chance of a public outcry forcing another change in the rules. Just as voters don’t mind disproportionality if the results align with expectations, Japanese voters don’t mind that some district losers win seats on party lists. However, when expectations of fairness are upset by a district loser that forfeits an election deposit (and by implication is a non-credible candidate) and still wins a seat in the Diet, then the election rule is discredited in the eyes of the voters. The rule change that made such outcomes impossible has gone a long way to simply defusing the zombie politician issue in Japan.

Notes  1 The countries using both a mixed system with dual listing are Germany, Japan, Hungary, Philippines, Pakistan, Romania, Albania, Lithuania, Armenia, Sudan, Lesotho, Tajikstan, Monaco, and Seychelles. In addition, Russia, Ukraine, Italy, Venezuela, Bulgaria, and Kyrgyzstan have used a mixed system with dual listing in some of their recent elections.   2 The actual wording was ‘In a country in which voters vote for parties, winning parties choose legislators from ranked lists of people published by each of the parties prior to the election. Occasionally, parties put unknown people high on their lists, and these people are guaranteed to win a seat because of their high placement on the party list.’ The alternative version had a different second sentence ‘Occasionally, parties put people who previously lost elections high on their lists, and these people are guaranteed to win a seat because of their high placement on the party list.’ The difference in the two averages was not statistically significant.   3 Renwick and Pilet (2016) note that when electoral changes are made in Europe, they tend more towards the direction of personalizing elections, consistent with this trend of party elites trying to anticipate which candidates would be popular with voters.   4 The typical list of reversal elections includes 1876 and 1888. Some also include 1832. Those published after the 2000 election also include the 2000 election.   5 For US elections, Congressional Quarterly also lists the 1950 and 1952 elections as reversals, but the Office of the Clerk of the House does not list those elections as reversals. In addition, the 1930 election became a reversal, in part, because of special elections, and the 1916 reversal was caused, in part, by a coalition with independents. For Japan’s elections, from 1947 to 1993 the electoral system was a multi-seat, SNTV (single nontransferable vote) system. From 1996 on, Japan’s system was a mixed system, meaning that for the entire period Japan’s electoral system was semi-proportional. For Australian elections, Australia used preference voting beginning with the 1919 election. Preference voting could easily increase the possible discrepancy between the first-place preferences of voters and the eventual election outcome when votes are reallocated from losing candidates. For proportional representation countries, the data comes from: Austria 1919–30 and 1945–2006, Belgium 1946–2007, Denmark 1920–2007, Finland 1917–2007, Germany 1919–33 and 1949–2005, Italy 1948–2008, Netherlands 1933–2006, New Zealand 1996–2005, Norway 1930–36 and 1945–2005, Portugal 1975–2005, Spain 1977–2008, and Sweden 1908–2006. This data comes from Mackie and Rose (1991), McHale (1983), Muller and Strom (2000), and various internet sources.

Acceptance of election results  195   6 The discrepancy between these two totals is accounted for by a possible error in the Congressional Quarterly database: ten Texas districts that did not do a December runoff have greatly reduced vote totals for both parties that are inconsistent with the vote totals recorded for these districts in the open primary held on election day in November.   7 The data includes congressional races from 1870 to 2004, calculating only races in which a reversal was possible, excluding one-seat delegations and delegations that were split evenly. This excludes 701 instances. Of the remaining 2,323 cases, 144 of them were democratic reversals with 75 advantaging the Democratic Party and 69 advantaging the Republican Party. Interestingly, if the data set is broken down into two periods, before and after the reapportionment revolution in 1964, the incidence of reversals in state delegations to the House of Representatives increases from 3.8 percent before 1964 to 10.6 percent after 1964.   8 His list includes Quebec (1949, 1966, and 1998), Ontario (1985), British Columbia (1996), New Brunswick (1974), and Saskatchewan (1986 and 1999).  9 The prompt for party quotas read ‘In a country which has a law requiring that 1/3 of elected representatives be women, parties nominate women as one third of their candidates, making sure that one third of the people elected are women.’ The prompt for reserved seats read ‘In a country which has a law requiring that 1/3 of elected representatives be women, if not enough women win their races, men who won more votes are displaced by women candidates who won fewer votes to meet the legal requirement.’ The difference in means is statistically significant, with a p value of less than .001. 10 Because Finland elects MPs from party lists in districts and the Swedish People’s Party is concentrated in only a few districts, its votes are more effective at electing MPs than other parties that have supporters in every district. Finland’s parliament created slightly larger electoral districts for the 2015 election to help reduce the potential for such disparities. 11 The data includes Portugal (1976–2015), Belgium (1974–2014), Netherlands (1977– 2017), Denmark (1975–2015), Norway (1977–2017), Sweden (1976–2018), Finland (1976–2015), Austria (1975–2017). The data is taken from various internet sources. 12 The data includes the United Kingdom (October 1974–2017), US House of Representatives (1974–2018), and Canada (1974–2015). Data is from various internet sources. 13 The data includes Japan (1976–2017), Australia (1974–2016), New Zealand (1975– 2017), Italy (1976–2018), France (1978–2017), Germany (1972–2017), and Ireland (1977–2016). Data is from various internet sources. It is also less than precise to lump together the French two-ballot system, Ireland’s SNTV system, Australia’s alternative vote, mixed systems, and countries that used different systems over the time period (New Zealand, Italy, and Japan) into one category, but all of these variations are more proportional than a pure majoritarian system and are less proportional than a wholly proportional system. For data on mixed systems a crude average of the vote totals of the two systems was calculated. 14 70.9 percent of the governments in semi-proportional systems were supported by a minority of the voters. 15 In 1992 Clinton won 68.8 percent of the electoral vote with only 43 percent of the popular vote, and in 1996 he won a similar 70.4 percent of the electoral vote with 49.2 percent of the popular vote.

196  Acceptance of election results

References Abramson, Paul R., John H. Aldrich, and David W. Rohde. 1998. Change and Continuity in the 1996 Elections. Washington, DC: CQ Press. Adkinson, Danny M., and Christopher Elliot. 1997. “The Electoral College: A Misunderstood Institution.” PS: Political Science and Politics 30, no. 1 (March): 77–80. Amy, Douglas J. 1993. Real Choices, New Voices, the Case for Proportional Representation Elections in the United States. New York: Columbia University Press. Asahi Shimbun. 1997. Asahi Senkyo Taikan [Asahi’s General Survey of Election]. Tokyo: Asahi Shimbunsha. Babington, Charles. 2000. “Electors Reassert Their Role; Bush Wins Vote; Protest Costs Gore.” Washington Post. December 19, p. A1. Baldini, Gianfranco, and Adriano Pappalardo. 2009. Elections, Electoral Systems and Volatile Voters. New York: Palgrave Macmillan. Ballington, Julie, and Drude Dahlerup. 2006. “Gender Quotas in Post-Conflict States: East Timor, Afghanistan, and Iraq.” In Women, Quotas, and Politics, edited by Drude Dahlerup, 249–58. New York: Routledge. Berke, Richard L. 2000. “Bush and Gore Vie for an Edge with Narrow Electoral Split.” New York Times. November 8, p. A1. Cain, Bruce E. 2015. Democracy More or Less: America’s Political Reform Quandary. New York: Cambridge University Press. Campbell, James E. 1996. Cheap Seats: The Democratic Party’s Advantage in US House Elections. Columbus, OH: Ohio State University Press. Center for Voting and Democracy. 1998. Electing the People’s House: 1998. Washington, DC: Center for Voting and Democracy. Cesar, James W., and Andrew E. Busch. 1997. Losing to Win: The 1996 Elections and American Politics. New York: Rowman & Littlefield. Christensen, Ray. 2000. Ending the LDP Hegemony: Party Cooperation in Japan. Honolulu: University of Hawai’i Press. Cohen, Richard E. 1997. “Campaigning for Congress: The Echo of ’94.” In Toward the  Millenium: The Elections of 1996, edited by Larry J. Sabato, 163–88. Boston: Allyn & Bacon. Coleman, Isobel. 2006. “Women, Islam, and the New Iraq.” Foreign Affairs 85, no. 1 (January–February): 24–38. Congressional Quarterly. 2008. “House General Elections, All States, 1996 Summary.” CQ Voting and Elections Collection, CQ Electronic Library. http://library.cqpress. com/elections/document.php?id=avg1996-us1&type=hitlist&num=12& Accessed April 15, 2019. Dahl, Robert A. 1989. Democracy and Its Critics. New Haven, CT: Yale University Press. Dahl, Robert H. 2003. How Democratic Is the American Constitution? New Haven, CT: Yale University Press. Dilulio, John J., Jr. 1997. “Conclusion: Valence Voters, Valence Victors.” In The Elections of 1996, edited by Michael Nelson, 167–74. Washington: CQ Press. Donahoe, Arthur. 2002. “Do Unicameral or Bicameral Parliaments Function More Effectively?” In Democracy, Parliament and Electoral Systems, edited by Mary Anne Griffith-Traversy, 47–57. London: Pluto Press.

Acceptance of election results  197 Englehart, Neil A. 2003. “Democracy and the Thai Middle Class: Globalization, Modernization, and Constitutional Change.” Asian Survey 43, no. 2 (March–April): 253–79. FairVote. 2008. Solutions. Ranked Choice Voting/Instant Runoff. fairvote.org/ rcv#where_is_ranked_choice_voting_used Accessed October 2, 2019. Farrell, David M. 1997. Comparing Electoral Systems. New York: Prentice Hall. Ferrara, Federico, Erik S. Herron, and Misa Nishikawa. 2005. Mixed Electoral Systems: Contamination and Its Consequences. New York: Palgrave Macmillan. Gaines, Brian J. 2001. “Popular Myths about Popular Vote–Electoral College Splits.” PS: Political Science and Politics 34, no. 1 (March): 70–5. Gallagher, Michael. 1991. “Proportionality, Disproportionality, and Electoral Systems.” Electoral Studies 10, no. 1 (March): 33–51. Gallagher, Michael, Michael Laver, and Peter Mair. 1992. Representative Government in Modern Europe, 2nd ed. New York: McGraw-Hill. Golder, Matt, and Jacek Stramski. 2007. “Ideological Congruence and Two Visions of Democracy.” Presented at the Annual Convention of the American Political Science Association, Chicago, IL, September 2007. Hart, Jennifer. 1992. Proportional Representation: Critics of the British Electoral System 1820–1945. New York: Oxford University Press. Hershey, Marjorie Randon. 1997. “The Congressional Elections.” In The Election of 1996: Reports and Interpretations, edited by Gerald M. Pomper, Walter Dean Burnham, Anthony Corrado, et al., 205–39. Chatham, NJ: Chatham House. Huber, John, and Bingham G. Powell. 1994. “Congruence between Citizens and Policymakers in Two Visions of Liberal Democracy.” World Politics 46, no. 3 (April): 291–326. IDEA. 2019. “Gender Quotas Database.” https://idea.ing/data-tools/data/genderQuotas/country-overview. Accessed March 21, 2019. Jacobson, Gary C. 1997. “The 105th Congress: Unprecedented and Unsurprising.” In The Elections of 1996, edited by Michael Nelson, 143–66. Washington: CQ Press. Johnsson, Anders, and Diana Maddock. 2002. “Alternative Electoral Systems.” In Democracy, Parliament, and Electoral Systems, edited by Mary Anne GriffithTraversy, 73–87. London: Pluto Press. Kallina, Edmund F. 1988. Courthouse over White House: Chicago and the Presidential Election of 1960. Orlando, FL: University of Central Florida Press. Katz, Richard S. 1997. Democracy and Elections. New York: Oxford University Press. Katz, Richard S. 2005. “Why Are there So Many (or So Few) Electoral Reforms?” In The Politics of Electoral Systems, edited by Michael Gallagher and Paul Mitchell, 57–76. Oxford: Oxford University Press. Laakso, Marku, and Rein Taagepera. 1979. “Effective Number of Parties: A Measure with Application to West Europe.” Comparative Political Studies 12, no. 1 (April): 3–27. Leonard, Mary. 2000. “Election 2000/Electoral College; How Popular Mandate Can Mean Defeat.” Boston Globe. November 9, p. D1. Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945–1990. Oxford: Oxford University Press. Longley, Lawrence D., and Neal R. Pierce. 1996. The Electoral College Primer. New Haven, CT: Yale University Press.

198  Acceptance of election results Loosemore, John, and Victor J. Hanby. 1971. “The Theoretical Limits of Maximum Distortion: Some Analytic Expressions for Electoral Systems.” British Journal of Political Science 1, no. 4 (October): 467–77. Mackie, Thomas T., and Richard Rose. 1991. The International Almanac of Electoral History, 3rd ed. Washington, DC: Congressional Quarterly. Mann, Bernard, Adam Przeworski, and Susan C. Stokes. 1999. “Elections and Representation.” In Democracy, Accountability, and Representation, edited by Adam Przeworski, Susan C. Stokes, and Bernard Mann, 29–54. New York: Cambridge University Press. Maravall, Jose Maria. 1999. “Accountability and Manipulation.” In Democracy, Accountability, and Representation, edited by Adam Przeworski, Susan C. Stokes, and Bernard Mann, 154–96. New York: Cambridge University Press. Massicote, Louis. 2005. “Canada: Sticking to First-Past-the-Post for the Time Being.” In The Politics of Electoral Systems, edited by Michael Gallagher and Paul Mitchell, 99–118. Oxford: Oxford University Press. McDonald, Michael D., and Ian Budge. 2005. Election, Parties, Democracy: Conferring the Median Mandate. New York: Oxford University Press. McGann, Anthony. 2006. The Logic of Democracy: Reconciling Equality, Deliberation, and Minority Protection. Ann Arbor, MI: University of Michigan Press. McHale Vincent, E. ed. 1983. Political Parties of Europe. Westport, CT: Greenwood Press. Muller, Wolfgang C., and Kaare Strom eds. 2000. Coalition Government in Western Europe. New York: Oxford University Press. Office of the Clerk. 2008. US House of Representatives. Statistics of the Presidential and Congressional Election of Nov. 5, 1996. April 15. http://clerk.house.gov/member_ info/electioninfo/1996/96stat.htm. Pound, William. 2002. “Democracy, Parliament, and Electoral Systems.” In The United States Presidential Election: Year 2000, edited by Mary Anne Griffith-Traversy, 127–30. London: Pluto Press. Powell, Bingham G., Jr. 2000. Elections as Instruments of Democracy: Majoritarian and Proportional Visions. New Haven, CT: Yale University Press. Rae, Douglas. 1967. The Political Consequences of Electoral Laws. New Haven, CT: Yale University Press. Reeve, Andrew, and Alan Ware. 1992. Electoral System: A Comparative and Theoretical Introduction. New York: Routledge. Renwick, Alan, and Jean-Benoit Pilet. 2016. Faces on the Ballot: The Personalization of Electoral Systems in Europe. New York: Oxford University Press. Reynolds, Andrew. 2006. “The Curious Case of Afghanistan.” Journal of Democracy 17, no. 2 (April): 104–17. Rush, Mark E. 1998. “Postscript: The Promise of Electoral Systems and the Perils of Electoral Reform.” In Voting Rights and Redistricting in the United States, edited by Mark E. Rush, 261–70. Westport, CN: Greenwood Press. Rush, Mark E., and Richard L. Engstrom. 2001. Fair and Effective Representation? Debating Electoral Reform and Minority Rights. Lanham, MD: Rowman & Littlefield. Sabato, Larry J. 2007. A More Perfect Constitution. New York: Walker Publishing. Saghafi, Morad, and Kaveh Ehsani. 2004. “The New Landscape of Iranian Politics.” Middle East Report, no. 233 (Winter): 16–23.

Acceptance of election results  199 Scammon, Richard M., Alice V. McGillivray, and Rhodes Cook. 1998. America Votes 22: A Handbook of Contemporary American Election Statistics 1996. Washington, DC: Congressional Quarterly. Shugart, Matthew Soberg. 2005. “Comparative Electoral Systems Research: The Maturation of a Field and New Challenges Ahead.” In The Politics of Electoral Systems, edited by Michael Gallagher and Paul Mitchell, 25–55. Oxford: Oxford University Press. Shugart, Matthew S., and Alexander C. Tan. 2016. “Political Consequences of New Zealand’s MMP System in Comparative Perspective.” In Mixed-Member Electoral Systems in Constitutional Context: Taiwan, Japan, and Beyond, edited by Nathan F. Batto, Chi Huang, Alexander C. Tan, and Gary Cox, 247–77. Ann Arbor, MI: University of Michigan Press. Sterling, Carleton W. 1978. “The Electoral College Biases Revealed: The Conventional Wisdom and Game Theory Models Notwithstanding.” Western Political Quarterly 31, no. 2 (June): 159–77. Stokes, Susan C. 1999. “What do Policy Switches Tell Us about Democracy?” In Democracy, Accountability, and Representation, edited by Adam Przeworski, Susan C. Stokes, and Bernard Mann, 98–130. New York: Cambridge University Press. Stout, David. 2000. “The 2000 Campaign: The Electoral College; How Winner of the Popular Vote Could Lose After All.” New York Times. November 3, p. A26. Taagepera, Rein, and Bernard Grofman. 2003. “Mapping the Indices of Seats–Votes Disproportionality and Inter-Election Volatility.” Party Politics 9, no. 6 (November): 659–57. Taagepera, Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven, CT: Yale University Press. Thompson, Dennis F. 2002. Just Elections: Creating a Fair Electoral Process in the United States. Chicago, IL: University of Chicago Press. Will, George F. 2000. “The Framers’ Electoral Wisdom.” Washington Post. November 2, p. A29.

9 Conclusion

The various lessons described in this book are counterintuitive and perhaps highly controversial. Many readers will not like conclusions that one-party rule has its advantages, campaign regulations are often ineffective, voter loyalty to parties or candidates is often better than the alternatives, election systems seem better the less input that they have from the voters, some malapportionment can be better than no malapportionment, and stricter rules on the handling of ballots are needed. These unexpected outcomes, though, are what should be expected when dealing with the contradictions inherent in democracy and the necessary vagaries and distortions that occur when individual preferences are aggregated into government policies. A realistic approach to the internal contradictions and foibles of democratic governance, as it is practiced, is preferred over an idealized and somewhat myopic view of pure democratic practices. The Founders of the United States understood this well, and it is good to follow their healthy skepticism of some democratic practices as the best method of building the best possible democratic governance. It must also seem jarring to take lessons from the Japanese experience, given the shorter time frame of true democratic governance in Japan and Japan’s notoriety for corruption, one-party rule, malapportionment, and a weak Supreme Court. Of course, the same criticisms could be lobbed at any other country. Canada has many flaws and weaknesses; the United States is not and never has been a paragon of democratic virtues. Sweden, Germany, France, and the United Kingdom all have their strengths and their weaknesses. Thus, the overarching lesson of this book is not that the Japanese model of democracy is superior to the US model. Rather, the lesson is that much can be learned from democratic practice in other countries. Additionally, because of some similarities between the electoral systems in Japan and the United States, the Japanese experience serves as a particularly good model or perhaps experiment for testing and analyzing the virtues and vices of specific electoral and democratic practices. It is thus not surprising that there are many Japanese practices that the United States would be well served to emulate, as well as a few practices that reformers in the United States should avoid. It will be a disappointment to some readers that I do not add my voice to either the chorus of pro-democracy reformers or the chorus of anti-democracy

Conclusion  201 reformers. People are free to draw their own conclusions about the appropriate findings from a comparison to Japanese electoral and democratic practices, but I see no evidence here to support efforts to abolish the Electoral College, elect Supreme Court justices, or reduce voter apathy by adopting a proportional representation electoral system in the United States. These and other reforms may be desirable and well suited for the United States, but the Japanese experience sheds little light on these pro-democratic initiatives. Similarly, I see no justification to join those who call for the United States to stop its single-minded encouragement of procedural democracy in authoritarian countries, want rigid gender quotas for all elected offices, or endorse the EU’s or the British government’s penchant to ignore the democratically expressed will of the voters regarding either enhancements of EU unity or the departure of a member from the Union. Again, these policy positions are perhaps laudable and correct, but I see no evidence from the analysis of this book to support these anti-democratic solutions to various problems. Rather, the message of this book is harder and more difficult to accept and use. What is the best electoral or democratic practice often depends on the country in which the practice is used. Strong candidate-centered electoral systems such as Japan and the United States have such strong incentives for candidate behavior that perfectly reasonable redistricting practices that work well in Canada or Australia fail or are distorted if they are used in the United States or Japan. Similarly, the limited intervention of the courts that I point out is necessary to solve some political problems in Japan and the United  States would  likely not be necessary in other, party-oriented democracies. The ­Japanese electoral system’s reliance on district vote totals to assign positions on party lists creates a legitimacy problem for dual-listed candidates that does not exist in other countries where positions on party lists are decided in advance of the election. These and other examples that come out of a detailed analysis of Japanese electoral and democratic practices strongly show that a reform that works well in one setting may fail in a different setting. Many of the practices described in this book will transfer well to other democratic settings, but some will require refinement to adjust to important differences between countries.

Index

Abe Shinzo 123–4, 132–3, 135, 144, 152, 156 abortion 87, 94, 144 accountability 6, 8, 10–14, 16–17, 19–21, 44, 72, 77, 177, 189 advertisements, political 25, 29–30, 37, 138, 146 Afghanistan 184–5 age 147–9, 156 agenda control 14 alternation in power 3, 7–12, 14–21 approval rating 132–7, 151 arrogance 6, 13–14 Aso Taro 123, 132, 135, 144, 156 associational rights 76, 78 Australia 4–5, 19, 44, 50–5, 57, 59, 81, 181, 191, 201 authoritarian 150, 191, 201 bailouts 88 ballots: counting 3, 73, 100–4, 106, 112, 115–18; handling 4, 100–2, 106, 114–18, 200; harvesting 118; preprinted 39–40; provisional 118; spoiled 39 boundary commissions 44, 46–7, 55–9, 61–2, 65–6, 74, 81, 86 bureaucracy 84–5, 100, 106–7, 117–18, 181 Brazil 4 Britain 1, 4–6, 44, 50–7, 59–60, 81, 127, 129, 181, 183, 191, 200–1 Cabinet Legislative Bureau 96n8 Cain, Bruce 44, 94 campaign: deposit 25; finance 4, 25–6, 29, 35–7, 40–1, 87, 94, 147–8; length 25, 31, 33–4, 37, 101; regulations 2, 4, 25–40, 94, 125, 146, 200; subsidies 25, 37, 40–1, 125

Canada 1, 3–5, 9, 20, 22n5, 28, 44, 50–5, 57, 59, 81, 87, 100–1, 108–11, 113–17, 128, 154, 157, 181–3, 200–1 candidate: attributes 44, 82, 125, 138, 146–7; beauty 125, 145–9, 156; centered elections 4, 29–30, 37, 44, 47, 55, 57, 82–3, 106, 125–6, 138, 201 Cannon, Howard 102–3, 105 Carlson, Matthew 101 central bank 44 centralization 115–17, 181–3 children 49 China 150 Citizens United 30, 40–1 civil service 44 clientelism 15, 44, 125, 138, 150 Clinton, Bill 13, 181, 192 Clinton, Hillary 3 coalition: electoral 11, 13–14; government 9, 22n9, 77, 130, 154, 177, 186, 188–90 competition, electoral 8, 10–11, 13–14, 16–18, 20–1, 26, 29–31, 41, 76–7, 84, 107, 139–40 compromise 67, 72–3, 80, 94–5 corruption 7–8, 13, 18–21, 36, 94, 100–1, 106–7, 150, 200; electoral 3–4, 18, 32, 100–17 cost-benefit analysis 26–29 Curtis, Gerald 15, 31 decision making 72 deficits 151–2 democracy, checks on 2–3; failings of 2–3, 200 Democratic Party: Japan 123–4, 127, 144, 147–8, 151–2; United States 7–9, 11, 13, 16–18, 46, 56–7, 75–7, 109–11, 113–14, 124, 127, 131, 143, 153, 155, 175–6, 178–82, 201

Index  203 democratic reversal 178–84, 187 district: candidates 171–5, 193; cohesion 46, 54–5, 62–5, 79–80, 83, 177; multiseat 59, 82–3, 107, 125, 139, 184–5; single seat 5, 44, 77–8, 82–3, 92, 101, 126, 139, 177, 180 distortion 173, 176–78, 183, 185–93, 200–1 door-to-door campaigning 25, 31–33, 36, 38, 40 Downs, Anthony 11, 13–15, 156 dual listing 171–5, 194n1, 201 earthquake 123, 135 economic growth 1, 7, 11, 16, 19–20 efficiency: energy 27; gap 77, 95n2; government 8, 17 election: deposit 172–3, 194; district boundaries 3–5, 44–67, 73–5, 79–80, 82–3, 183, 201; frequency 126, 153–4, 157; primary 155, 175–6, 179 Electoral: College 90, 178, 201; reform 82–3, 125, 137, 141–2, 171–2, 179, 181, 183, 191–4, 201 email 38–39 equality 76–80, 83, 88, 94 176–7, 193 Europe 4, 88, 128–9, 131, 138, 201 European Court of Human Rights 88 federalism 16, 106, 115, 181 feudal 125 Finland 187, 195n10 France 4, 9, 20, 87, 106, 124, 128, 151, 200 free speech 40–1, 78, 94 freedom 7 Fukuda Yasuo 123, 132 Fund, John 103 gender 147–9, 155–6, 184–5, 201 Germany 1, 88, 127–9, 151, 174–5, 200 gerrymandering 45–8, 54–9, 61, 66–7, 74–8, 80–1, 83–6, 107, 180 gifts 34–36 government: distrust of 16–17; efficiency see efficiency; formation 186–90; minority 186–8; performance 123, 145, 149–50; unified 10, 20 governors: Japan 12, 22n10; United States 10–11 gridlock 152, 154 Hatoyama Yukio 9, 123, 132, 151 Hopi 56

Hosokawa Morihiro 9 House of Councilors 31, 39–40, 141–2, 151, 154 House of Representatives: Japan 12, 25, 39–40, 91, 93, 114, 126–7, 140, 147, 154; United States 6–7, 10, 33, 124, 154, 179–82 human rights 7, 88 ideology 16, 147–8 incremental 107 incumbency advantage 14–15, 39, 41, 83, 112–15, 146–8 independents 15, 22n10, 110, 113–14, 131, 136–7, 154 India 20 information 13–15, 103, 145–6, 150, 156, 171, 173–6, 178–81, 183–8, 190–3 internet 38–9 investigations 44, 104–5 Iran 173, 191 Iraq 184 Ireland 4, 191 Israel 9 issue-based campaigns 40–1, 124–5, 138, 143–5, 148–9, 155–7 Italy 9, 20, 193 Johnson, Lyndon 103, 105, 127, 132 Johnson, Tim 103 judicial: districts 62–5; restraint 73–4, 81, 83, 86–7, 89–90, 95; review 5, 80, 87–9; standards 74, 77, 80–1, 83–4, 86 Kan Naoto 123, 132, 135, 151–2, 156 Kanemaru Shin 37 Kennedy, John 178–9 Key, V. O. 17 koenkai 32, 34–6, 125, 138–9, 142 Koizumi Junichiro 92, 123–4, 132–3, 135, 143, 148, 156–7 Koumeitou 33 Latin America 129–30, 150 Laxalt, Paul 102–3, 105 legitimacy 2, 9, 72–3, 171, 173–4, 183, 185, 191–3, 201 Liberal Democratic Party 7–9, 12, 15, 18, 22n9–11, 32–3, 35, 37, 39–40, 82–3, 86, 88, 90–3, 108–9, 112–13, 123–4, 126–7, 130–1, 135, 141–3, 147–9, 151, 157 Liberals, Canada 113

204 Index local government 7, 10–12, 15–16, 46–7, 51, 53–5, 58–9, 61–2, 65–6, 79–80, 83–6, 112–13, 154, 191 logic, internal 72 long term perspective 125 loyalty 123, 125–6, 137–8, 141–5, 148, 153–7, 200 low information voters 124, 144–6, 150, 157 McCain, John 149 majoritarian 173, 180, 185–90 malapportionment 45–57, 61, 66, 67n2, 73–5, 77, 79–81, 83, 86–9, 91–2, 95, 172, 180, 183, 200 margin of victory 102–14 media 44–5, 84, 92, 151, 156, 171–2, 180–1 median voter 11 Medicare 106, 152, 157 Mexico 8–9, 16–17 minorities 45, 56, 66–7, 72, 74–6, 84–6, 126, 149, 155–6, 177, 183 monetary policy 44 monitoring compliance 29–32, 34, 36, 39, 41 Nakasone Yasuhiro 92 Navajo 56 new democracies 129, 137 New Frontier Party 39–40 New Zealand 4, 19, 177, 179, 181, 193 Nixon, Richard 132, 178 Noda Yoshihiko 123, 152, 156 noncitizens 49 North Korea 16 nuclear power 152 Obama, Barack 8, 13, 40, 124, 127, 132, 149, 155, 176 occupation, US 87 one-party rule 3–4, 6–21, 44, 200 opposition, Japan 12, 22n10, 92, 111, 113, 119n6, 127, 141–2, 154 out of touch, politicians 6, 12–13 Ozawa Ichirou 151 pandering 125, 152–3 paradox 173–7, 185, 187, 190–3 parliamentary systems 106, 154 party: centered elections 4, 125, 201; control 10, 187; identification 82, 136–7, 143, 154, 157; leader 4, 123–4, 126, 132–3, 136, 138, 153–4; list 171–5, 184, 193–4, 201; switching 130–2, 137 Pedersen, Morgens 128

Pempel, T. J. 9, 14–15 personal support groups see koenkai Pildes, Richard 81 placeholder candidates 172–3, 193 policy: popular 125–6, 146, 152–5; positions 11–12, 145, 200–1 political: question doctrine 2, 5, 73–4, 80, 87–8; thicket 73–4, 77 politicization 2, 8, 17–18, 55–8, 75, 81, 154 postal reform 92 posters, political 33–4, 146 Powell, Bingham 180, 188–90 pre campaign 26, 31, 35–6, 38–9, 41 prefectures 12, 22n8, 48, 51, 57, 59, 66, 68n4, 112, 171 presidency, US 6–8, 10–11, 16, 21, 25, 124, 132–3, 153, 176, 178–9, 181–2 prime minister, Japan 9, 12, 123–4, 132–6, 151–3, 157 proportional representation 77, 173, 177, 180, 184–92, 201 public observation 116–18 Putin, Vladimir 8 quality of life 19 quotas see women, political representation race see minorities Ramseyer, J. Mark 90–93 Rasmusen, Eric 90–93 Reagan, Ronald 6, 13, 132–3 recount 114–15 Reed, Steven 7, 101 registered voters see voters, eligible regression discontinuity analysis 114 Republican Party 9, 11, 13, 17–18, 46, 56–7, 62, 75–7, 90, 109–11, 113–14, 124, 153, 155, 175–6, 178–80, 182 responsiveness see accountability rights 72, 89, 94–5 rural 90–3, 102, 112–13, 138–43 Russia 8 Scheiner, Ethan 15 school: districts 61, 65; private 28 second mover advantage 13–14 Senate 10, 131 separation of powers 16 self interest 85–6, 145, 150 simplicity 78, 191 soccer 106 social security 153 sound cars 37–8 South Africa 20, 21n4

Index  205 South, US 7, 11, 16, 17, 20, 21n3 state legislatures 10–11, 17, 75, 79, 81–2, 85, 182 subsidies see campaign suicide 19 sumo 106 supreme court: Japan 2, 46, 60, 73, 75, 82, 86–95, 200; United States 2, 30, 40–1, 45, 47, 67, 73–5, 77, 79–80, 82, 86–8, 90, 94, 201 Sweden 1, 4, 7, 9, 17, 200 swing 78, 126–8, 141–4 symbolic gesture 156 talent, political 156–7 Tanaka Kakuei 133, 135 tax: compliance 27–8; increase 18, 123, 144, 151–3 Thatcher, Margaret 6, 128 totality of circumstances 78–9 turnout 41, 48–50, 138–42, 145, 149, 157, 176, 180 Trump, Donald 3, 132, 157, 178

undocumented workers 28–9, 144 United Kingdom see Britain US–Japan Security Treaty 87–8 virtuous cycle 14 Vogel, Ezra 1 volatility 4, 123–57 voter: apathy 4, 25, 201; eligible 48–50, 52–4, 66, 180; preferences 13–15, 176–8 184–6, 188–92 Voting: absentee 105–6, 117–18; cues 125, 145, 155–6; early 116; by felons 49; by mail 116, 118; recommendations 125, 138, 145; Rights Act 74; straight-ticket 156 wasted votes see efficiency gap water conservation 26–7 women, political representation 19, 149, 173, 183–5, 191, 201 World War II 87, 127, 154 zombie politicians 171–4, 193–4