Willing to Pay?: A Reasonable Choice Approach 019879682X, 9780198796824

Citation preview

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay? A Reasonable Choice Approach S V E N  S T E I N M O A N D J O H N D’ AT T OM A

1

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

1 Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Sven Steinmo and John D’Attoma 2022 The moral rights of the authors have been asserted First Edition published in 2022 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2021942735 ISBN 978–0–19–879682–4 DOI: 10.1093/oso/9780198796824.001.0001 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

For Rita and Sonia

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Preface and Acknowledgments This book has been a long time coming. We began the research project almost ten years ago when Sven Steinmo was the lucky recipient of a European Research Council (ERC) “Frontier Grant” for his proposal “Willing to Pay? Testing Institutionalist Theories with Experiments.”1 ERC Frontier grants are designed to support senior scholars who are willing to go outside their traditional research arena or methodologies. This grant and the research which flowed from it did just that. Steinmo is best known for his work in Historical Institutionalism (HI). Willing to Pay? proposed to test some of the arguments and hypotheses made by Historical Institutionalists with a totally different methodology—laboratory experiments. Like all HI scholars, he was well aware of the criticism that ­historical analyses cannot be falsified. The analyses presented by HI scholars can be quite persuasive (hopefully), but they are difficult to prove precisely because they focus on a small number of cases over time. After all, history is in the past and cannot be recreated. Taking this issue seriously, the central idea behind the ERC proposal and project was to try to test the central proposition of Historical Institutional analysis: History and ideas matter in politics. This may seem like an absurdly obvious point—to a non-­social scientist—but the truth is that the dominant theories in political economy and economics today ignore both history and ideas. No one can peruse the major journals in political science and economics today without noticing the predominance of statistical analysis and theoretical modeling techniques in the vast majority of successfully published articles. While there is a wide variety of subjects and topics addressed in these publications, they also tend to share two basic assumptions about human behavior: (1) human beings are primarily motivated by their individual short-­ term self-­interest, and (2) people’s behavior is shaped by institutions (rules) because institutions present them with different incentives and disincentives. In other words, people are basically the same everywhere, but they behave differently in different contexts because of the different institutional rules they face. Many economists and political scientists may agree that history and people’s ideas are interesting. But they sometimes argue that for serious social

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

viii  Preface and Acknowledgments science analysis, these factors do not really matter . . . or at least they are too difficult to analyze to be of use in any formal scientific analysis. If we want to understand political and economic outcomes, these scholars suggest, we need to focus on institutions and interests. Of course, not all of the scholars publishing in this vein believe that humans are simply self-­interested ­ra­tional actors, but the truth is that it is extremely difficult to build models or test hypotheses using these more nuanced, dynamic, and sometimes vague variables like ideas, values, and/or culture. But for the HI scholar, it is these variables that are the most interesting. In short, the HI scholar is interested in understanding why people sometimes have different preferences (or “utilities”) at different times or places. As a scholar who has built much of his career as an institutionalist, Steinmo has long argued that institutions are important. But he also believes that ideas, norms, and values also matter. The HI dilemma, however, is that since we ­generally study single cases, or at best a small set of cases over time, it is essentially impossible to “prove” that variable as ephemeral as “ideas” or “history” have a causal effect on people decisions, no less entire political economies. Thus, the ERC Advanced Grant offered an opportunity to use some of the techniques developed by behavioral scientists to try to test the basic question: If people in different societies face identical rules, incentives, and disincentives (institutions), will they behave similarly, or differently? In short, do their ideas, norms, beliefs, values, or national histories matter? This book is an attempt to briefly summarize some of the major empirical findings from this multi-­country, five-­year, 2.1-­million-­euro research project. We specifically focus on tax compliance. It is well known that compliance rates vary enormously across societies. In some countries, tax evasion (which is illegal tax avoidance) can cost the public treasury over 30 percent of total tax revenue. In some other countries, people are much less likely to cheat on their taxes and the state can lose as little as 6–8 percent of tax revenue to evasion. The obvious question is WHY? If you ask the average person from a country that suffers high levels of tax evasion why so many people evade their taxes, they are likely to tell you something like “it’s in our culture.” Similarly, if you ask a person from a country where most people pay their taxes fully and honestly why there is so little evasion, they are likely to respond something like “that’s just who we are,” or, “we are an honest people.” Traditional economists and political economists have a very different set of explanations. They would likely argue that differences in compliance rates across countries are simply a matter of different enforcement regimes and penalties. No one wants to pay taxes, they argue,

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Preface and Acknowledgments  ix and the only reason they do is that they are afraid of the consequences they might suffer if they get caught. In short, “culture” has nothing to do with it. It is all about institutions. Our Willing to Pay? project had two basic prongs. First, we brought together country experts who could teach us and each other about the fiscal histories of each of the countries we were to study. Second, we brought in several behavioral scientists to help us build experiments to test the arguments made by the country experts. The result has been two books: The Leap of Faith: The fiscal foundations of successful government in Europe and America (Oxford University Press, 2018), and the current volume, Willing to Pay? A Reasonable Choices Approach. There has been a lot of research using behavioral experiments to analyze compliance. Most of these studies (and indeed most behavioral experiments) have demonstrated that real people do not behave according to the assumptions made in traditional political economy and economic analyses. The evidence is clear that people have additional concerns, motives, and considerations when making decisions. It is reasonable to assume that this is because their ideas, beliefs, and norms matter, and that these variables may differ across societies. In this book we specifically attempt to provide a simple yet sophisticated theory for why people pay the taxes that they pay. Of course, there is forty years of outstanding scholarship on tax compliance that has come before us without which we could have not embarked on this massive five-­year study. Many of these people sat on our advisory committee and helped us along the way and others were colleagues and friends who read our work and gave us feedback. Our first debt, therefore, goes to the scholars who paved the way for this study. Next, we would like to acknowledge an enormous debt of gratitude to the European University Institute (EUI) and the Schumann Center for Advanced Study in Fiesole, Italy. Our experiences there had incalculable influences on our academic growth. From the intellectually stimulating lunches at the Badia to the Coppa Pavone, the EUI was an incredibly enriching, and flat-­out fun, experience. The EUI embodies the best traits in academia by bringing together many of the finest scholars and intellectuals in the world and then encouraging them to mingle, share ideas, and argue. They invite creativity and deep scholarship at the same time. In so doing they represent exactly what collegiality is (or should be) all about. It was our enormous privilege to spend so many years there.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

x  Preface and Acknowledgments Several scholars both at the EUI and beyond have been integral to this project at different stages. Our first debt goes to Giulia Andrighetto. Giulia is one of the preeminent young scientists in Italy today and it has been our pleas­ure to work with her and learn from her since she first came to the EUI as a Max Weber Fellow in 2011. It is no exaggeration to say that this project would not have been possible without her. A special debt of gratitude also goes to Jim Alm, who was enormously helpful in helping us get our feet on the ground regarding laboratory experiments and tax compliance. Jim is perhaps the most well-­known expert in tax compliance and his generous support and suggestions were critical for us both in terms of putting together a successful ERC proposal and building experiments that would make sense. The husband-­and-­wife team of Stefania Ottone and Ferruccio Ponzano also deserve special thanks. Very early in the project, Stefania and Ferruccio helped us design our first set of experiments and ran several of them in their home universities in Milan and Alessandria, Italy. Ryan Murphy and Nan Zhang also contributed enormously to the design, development, and interpretation of the experimental portion of this project. We would specifically like to acknowledge the contributions of the scholars who have contributed to our understanding of the tax histories and political economies of countries we have studied in this project. The very idea of this project was unusual, for it brought together both historically oriented scholars and behavioral scientists. None of this could have been possible without the country specific knowledge and understanding offered up by Michelle D’Arcy, Marcelo Bergman, Martin Daunton, Joseph Hien, Romain Huret, Jenny Jansson, Carolyn Jones, Marina Nistotskaya, Liam Stanley, Arpad Todor, and Clara Volintiru. This volume was also significantly advanced by the cogent and thoughtful comments from a few scholars and policy experts who participated in one or several of the “Willing to Pay?” workshops and conferences we held over the duration of the project: Cornel Ban, Pablo Beramendi, Mark Blyth, Christiano Castelfranchi, John Cullis, Ray Duch, Jonas Edlund, Philipp Genschel, Benedikt Herrmann, Edgar Kiser, Isaac Martin, Matthias Matthijs, Monica Noll, Fred Pempel, Stefano Pisani, Alan Plumley, Bo Rothstein, Mario Scala, and Lennart Wittberg. Their criticisms and suggestions have significantly improved this volume in specific and the research project as a whole. It has been an honor and inspiration to work with these scholars, policymakers, and friends. John would also like to express his specific thanks to some of his fellow tax scholars and friends at the University of Exeter and within the Tax

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Preface and Acknowledgments  xi Administration Research Centre (TARC). Lynne Oats would have to be first on this list. Lynne’s support and mentorship were far beyond the norm one often finds in academia. She was always there to bounce ideas off, to provide feedback, or to just chat. Similarly, Kim-­Lee Tuxhorn, Duccio Gamannossi degl’Innocenti, Antoine Malézieux, Jingnan (Cecilia) Chen, Miguel Fonseca, Oliver James, and Stephen Brooker were wonderful colleagues and friends to whom he will always be grateful. Finally, we would like to thank our wives, Sonia Lalla and Rita Kungel, to whom this book is dedicated. Both these wonderful women have been willing to move across continents—multiple times—put up with frustrated academic spouses, and at the same time offer their love and support for a project that has taken too long. They have always encouraged us and supported some of these life-­decisions that might have just seemed crazy to anyone outside of this academic bubble. We dedicate this book to them.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Table of Contents List of Figures List of Tables

xv xvii

Willing to Pay? A Reasonable Choice Approach

1

1. Why Should I Pay? A Cognitive Theory of Tax Morale

8

Introduction1 Willing to Pay? Testing Institutional Theory with Experiments 3 A Reasonable Choice Approach Institutions, Norms, and Values in Successful Societies

9 10

2. The Experimental Design

18

3. Reasonable Choices and Tax Compliance

37

Conclusion: Everyone Hates a Cheat

77

Experiments and Tax Behavior Running Experiments in Multiple Countries The Willing to Pay Experiments

19 20 21

Culture vs. Institutions? 37 Reasonable Choices 38 Self-­Interest and Institutions 39 Perfidious Albion 46 The Social Impulse 48 Do Tax Rates Matter? 49 Does Progressivity Matter? 52 Norms and Social Behavior 57 The Tragedy of Economics 62 Norms and the “Theories in our Heads” 66 Values68 Interests, Norms, Values Successful Societies and Institutional Performance Explaining National Variation

78 79 81

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

xiv Table of Contents Appendix85 Appendix 1: Full Experiment Instructions85 Appendix 2: Questionnaire to the Baseline Experiment96 Appendix 3: Analysis and Tables114 Endnotes 121 References 127 Index 135

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

List of Figures 1.1. Distribution of Individualists and Cooperators in Society

13

2.1. Screen Capture of the Clerical Task

23

2.2. Screenshot of Clerical Task

24

2.3. Screenshot of Reporting Screen

25

2.4. Social Value Orientation Example Allocation Decision

32

3.1. Average Compliance Rate in Rounds 1 through 3 (all countries, all sessions)

43

3.2. Average Compliance Rate in Rounds 1–3 by Country

45

3.3. Average Compliance Rate in Rounds 4–6

50

3.4. Average Compliance Rate for Rounds 4–6 by Country

51

3.5. Average Compliance Rate in Rounds 7 and 8

53

3.6. Average Compliance Rate in Round 7 and 8 by Country

54

3.7. Histogram of Reported Income in Round 8

55

3.8. Average Compliance Rate by Income Group

56

3.9. Average Compliance Rate by Perception of Other’s Reporting Behavior

60

3.10. Perception of Others’ Reporting Behavior by Country

61

3.11. Average Compliance Rate by Major in Each Round

62

3.12. Average Compliance Rate by Major and Country

63

3.13. Percent who Report 0 by Major

64

3.14. Perceived Earnings by Major

65

3.15. SVO Choice Set

70

3.16. Distribution of SVO-­angle

71

3.17. Average Compliance Rate for Individualists and Prosocials

72

3.18. Distribution of the SVO-­angle by Country

73

3.19. Distribution of the SVO-­angle by Economists

74

A.1. Predictive Margins for Country Interactions for Compliance Rate Rounds 1–3

116

A.2. Predictive Margins for Country Interactions for Compliance Rate Rounds 4–6

117

A.3. Predictive Margins for Country Interactions for Compliance Rate Rounds 7 and 8

118

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

List of Tables 1.1. Tax Gap by Country

15

2.1. Example Survey Questions

34

2.2. Summary Statistics

35

2.3. Summary Statistics for the Compliance Rate

36

3.1. Employment by Enterprise Size

47

A.1. OLS Regressions; DV = Average Compliance Rate

115

A.2. OLS Regressions; DV = Average Compliance Rate

119

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay? A Reasonable Choice Approach Introduction “The art of taxation consists in so plucking the goose as to obtain the largest possible amount of feathers with the smallest possible amount of hissing.” —Jean-­Baptiste Colbert, French Minister of Finance – a view attributed to Jean Baptiste Colbert who served as Louis XIV’s Finance Minister for the period 1665 to 1683.

No one loves taxes. But taxes pay for things that we need, want, and sometimes even love. Taxes fund the government and everyday services we rely on and without them, governments cannot function. Indeed, taxation is the key lynchpin between government accountability and citizen responsibility. In this book, we consider this lynchpin. Specifically, we seek to explain people’s willingness to pay taxes. Typically, when scholars examine taxes and tax compliance, they are usually concerned with the question of why people evade taxes. Rarely do scholars (or policymakers) consider the flipside of this coin. Why are so many people willing to pay taxes? The fact is that in most countries, most people pay their taxes honestly and completely. Of course, it is not true that everyone pays all the taxes they should, but it is true for large majorities in all modern democracies. But why? Most people do not enjoy paying taxes, and most people would not mind if politicians reduced their tax bills. If asked on the street why they pay their taxes, intuitively people would probably say that they have no choice. Yet, if people are given time to think about their taxes, they might say that they pay taxes because taxes pave their roads, pay for their kids’ schools, or pay for the military that keeps them safe. Paying taxes, in other words, may be painful or unpleasant, but it is a necessary pain. Tax money funds the public services that we rely upon, and yet, tax compliance varies greatly both across societies and between countries. In some countries, the revenue losses from tax evasion are enormous and can nearly Willing to Pay? A Reasonable Choice Approach. Sven Steinmo and John D’Attoma, Oxford University Press. © Sven Steinmo and John D’Attoma 2022. DOI:10.1093/oso/9780198796824.003.0001

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

2  Willing to Pay? equal tax revenue. In other countries tax evasion is less than 10 percent of revenues collected. The question is why. What explains this remarkable variation? There are three classic explanations for why people appear to be more willing to pay in some countries than others. First, when citizens believe they receive poor services from their government, they will evade more than when they are satisfied with their public services. Second, in some countries tax collection is inefficient, corrupt, or both, and in these countries, taxpayers may be able to get away with evasion and/or have less fear of punishment even if they are caught. Lastly, tax evasion may be more socially acceptable in some countries than others.1 Most likely, each of these explanations tells part of the story. But how can we know? Are they all equally valid? Is tax evasion, or tax compliance, mostly driven by institutional or cultural differences? Given that tax evasion (what is often referred to as the “Tax Gap”) can account for several hundred million dollars in revenue loss each year, these are not simply “academic” questions. Every government (and most citizens) would like increased tax compliance— not only because it could provide increase revenues, but also because doing so could make the entire system fairer and more equitable. Certainly, where public services are efficient and plentiful, tax collection is effective, and citizens hold high levels of social obligation (e.g. Sweden), we can expect fewer people to evade than in places where the opposite is true (e.g. Italy). But what can a country like Italy do to increase compliance?2 Is it true, as some political scientists have suggested, that a lack of “civic culture” explains the behavioral differences between countries (Banfield  1958; Putnam et al.  1993)? Or, are the differences in behavior between countries better explained by differences in the political economy, the institutions, or the enforcement mechanisms used in different countries? In other words, do Italians, for example, simply avoid their taxes more than Swedes because they are more likely to get away with it? Or are Swedes more compliant because they are more honest, compliant, or “prosocial”? A central problem for traditional comparative social science is that the variables shaping citizens’ decisions in different countries are so numerous that it is nearly impossible to tell what affects what. In social science terms, the problem is over-­ determined. While it is obvious that Scandinavians behave differently from Italians, we do not know whether this is because they live in different systems (i.e. institutions) or because there is something deeper in their “culture” that leads them to behave differently. We know that tax compliance is higher in Sweden than Italy. But we do not know if this is

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay? A Reasonable Choice Approach  3 due to Swedish institutions being more efficient and producing “better” public policies than Italian public institutions, or is it because Swedes are better ­citizens, or simply more compliant than Italians. To make matters worse, these variables are dynamic and often endogenous. In other words, it could be that cultural differences explain why some countries were able to build more efficient institutions and therefore citizens today are more willing to hand over so much of their income to those same institutions than others. But it could also be that there is something about a shared set of ideas, values, and beliefs that have allowed some countries to build political and economic institutions that are more efficient and reliable.

Willing to Pay? Testing Institutional Theory with Experiments The European Research Council (ERC) funded us to conduct a large multi-­ country, multi-­year project to investigate these questions. Our approach was twofold. On one hand, we assembled a substantial network of national experts from economics, history, sociology, and political science to unravel the substantive and historical differences which could help us to understand the roots of fiscal differences between nations. At the same time, we brought together a group of scholars versed in experimental methods and behavioral science. The overarching goal was to see if and how we could successfully and meaningfully bring together the insights from national experts and comparative historical analysis with the tools and techniques of laboratory experiments and behavioral science. As far as we are aware, a study of this type on this issue has never been attempted before. This book represents the second half, or the “summary report,” of this research project. Our first book, The Leap of Faith: The fiscal foundations of successful government in Europe and America (Oxford University Press, 2018), examined the formal institutions and their historical fiscal foundations in five countries (United Kingdom, Sweden, the United States, Italy, and Romania). In The Leap of Faith, we sought to examine and explain how and why these countries developed such disparate fiscal systems and institutions. We encourage the reader of this book to explore the analyses presented in The Leap of Faith. But, as interesting and persuasive as we believe this book is, comparative historical analysis cannot answer the question: Would Italians be more tax compliant if tax-­funded political institutions were as efficient as Swedish counterpart institutions? Similarly, would Swedes cheat on their taxes if their government were as inefficient as the Italian government?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

4  Willing to Pay? To test these questions, we turned to behavioral science. The study of why humans behave the way they do is not new. Cognitive psychologists, after all, are specifically interested in how the mind works and why people behave the way they behave. The works of many great psychologists such as Amos Tversky, Daniel Kahneman, and Dan Ariely are particularly fascinating and relevant to the study of politics and political outcomes. But surprisingly, they pay very little attention to the role that institutions play in structuring or shaping behavior. In the chapters that follow, we elaborate on the details of this study. The central idea driving the study reported in this book was to build a set of behavioral experiments that we could test in multiple countries to see if citizens of one country would in fact behave differently from those in other countries when faced with exactly the same institutions, incentives, and constraints. To preview our findings: we found that there do appear to be some systematic differences in how citizens in different countries behave when faced with identical public goods dilemmas or problems. BUT we also found that: (1) differences across countries are far smaller than we expected; (2) these differences do not fall into the expected cultural stereotypes; (3) there are greater differences across types of individuals within countries than differences between countries. Ultimately, we did find that Italians were somewhat less compliant than Swedes on average. However, much to our surprise, we found that on average Italians were more compliant and more willing to contribute to the public good through taxation than British subjects. We also found that certain personal characteristics have systematic effects across countries. For example, males cheat more than females under every experimental condition that we tested. In short, women are more honest than men—everywhere. For this study, we went to extreme lengths to ensure that our experiments were as consistent as possible in different national, linguistic, and economic contexts. Overall, we conducted 116 laboratory sessions in 14 different locations in the 4 countries and recruited over 1,700 subjects for our experiments.3 As far as we know, we believe this to be the largest cross-­national laboratory tax compliance experiment conducted anywhere to date. Most of the experimental research today on the topic relies on subjects from one or two locations with recruits between 80 and a few hundred subjects and is able to report sometimes fascinating findings regarding specific questions within a particular cultural or national context. Our study, in contrast, required a broader net so that we could be sure we were not biasing our sample with particular locations (or regions of a country) and that we had sufficient data to truly explore cross-­national differences—to the extent that we found them.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay? A Reasonable Choice Approach  5 By now, a significant amount of work in behavioral economics focusing on tax compliance behavior has been conducted. Tax compliance experiments date back forty years, including fascinating studies by James Alm, Benno Torgler, Ernst Fehr, Simon Gächter, Luigi Mittone, Benedikt Herrmann, and many others. Based on the works of these scholars and others, it is already well understood that taxpayers do not behave as classical economic theory would expect them to behave. Instead, there seem to be other motivations for paying taxes driven by what scholars have termed “Tax Morale.” In addition, it appears that tax morale varies between countries. Several of the aforementioned scholars explicitly argued that more studies should be done using laboratory experiments to test for differences in tax compliance behavior across countries. In this volume, we explore how people behave when asked to pay taxes that would be redistributed to people when they are literally sitting in the same room. Some skeptics have argued, therefore, that the results or findings drawn from such laboratory experiments are irrelevant and can teach us nothing about “real” human behavior. As the reader will see, we can also be skeptical of laboratory experiments, but we do not agree that they are either useless or irrelevant. We believe instead that they can be remarkably informative and indeed teach us quite a lot about how people react to different stimuli or incentives and disincentives as well as how people in varied contexts may act differently.

A Reasonable Choice Approach This book highlights three variables that help explain variance in tax compliance both between individuals and across countries: interests, norms, and values. We discuss these variables and how they interact in the following chapters. Our intention is to push the implications of this specific analysis beyond the study of tax evasion in different countries. We suggest that tax compliance behavior is but one example of cooperative or compliance behavior more generally. Thus, our broadest ambition is to better understand why compliance and cooperation in general varies across countries and individuals. We believe that a careful study of something as narrow, and specific, as tax compliance can provide a window through which we can better understand the relationships between citizens, their society, and their public institutions. If it were the case that Swedes had high tax compliance rates but otherwise exhibited few other prosocial behaviors, or if Italians tended to evade taxes

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

6  Willing to Pay? but otherwise were model civic citizens—then this study might be interesting and important simply as an examination of tax compliance. But we believe it offers more than that. Tax compliance behavior is only one form of social or civic behavior. Ultimately, we want to know why some societies exhibit a preponderance of prosocial behaviors, while others much less so. To answer this question, we offer what we have titled a “Reasonable Choice Approach.” We elaborate on this model in the following chapter. Here we can say that our approach builds on some of the micro-­foundations of classical rational choice theory but integrates more complex understandings of human preferences and motivations than classical Rational Choice. Certainly, people can be quite self-­interested. But, as an overwhelming body of cognitive, psychological, and behavioral economic studies demonstrate, our preferences are not exclusively—or even primarily—individually motivated. Instead, we are social animals who have strong preferences to associate with and model ourselves on others. We argue that in addition to self-­interest and normative preferences, humans have values that matter. In short, what we do in any given context is in part motivated by what we can get out of a situation, in part by what we think others will do, but also by what we believe we ought to do.

Plan for the Book The book is structured as follows: In the first chapter we introduce the reader to some of the basic questions regarding compliance behavior. We focus on three key variables that we will follow throughout this book: interests, norms, and values. We argue that these three variables are key to explaining why people cooperate, share, and pay their taxes. In Chapter 2 we offer a brief introduction to experimental methodology and our experiments as a tool for comparing societies. In Chapter 3 we present our evidence. We demonstrate how self-­interest, individual norms, and social values each contribute to a more complete understanding of human behavior in choice situations. A key point here and throughout this book is that societies are made up of a wide diversity of individual types. In every society, there are some people who are extremely self-­ interested and will try to get the most for themselves in almost every situation. But there also exist highly altruistic people in every society who often are quite willing to sacrifice their own self-­interest for the good of the whole.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Willing to Pay? A Reasonable Choice Approach  7 As scholars who have come to behavioral experiments relatively recently— but who have used these methods intensively and extensively for the last seven years—we believe we can offer a perspective that is both honest and straightforward. We are both fans of experimental methods and skeptics of the way in which many laboratory experiments have been conducted. Like any other method, when poorly designed, you can easily obtain junk results. Well done and carefully constructed, they can offer insights that are sometimes quite remarkable. In our view, experimental methods hold great promise and can teach us a great deal about politics, but their results cannot be interpreted in isolation from the context in which they were conducted. We hope this short book will be helpful to anyone interested in the mechanisms that shape individuals’ tax compliance decisions. Our goal is to provide a simple and generalizable theory for why most people willingly pay their taxes. It will certainly be helpful to scholars and tax professionals interested in taxation. This book will also be useful to those interested in embarking on a large multi-­country study that involves experiments, or any social scientist embarking on an experiment conducted within a single laboratory.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

1 Why Should I Pay? A Cognitive Theory of Tax Morale “It must not be forgotten that although a high standard of morality gives but a slight or no advantage to each individual man and his children over the other men of the same tribe, yet that an increase in the number of well-­endowed men and an advancement in the standard of morality will certainly give an immense advantage to one tribe over another.” —Charles Darwin in The Descent of Man 1871, p. 159 “The dynamic nature of history implies that the centrality of beliefs— how humans form their beliefs and how they learn—is fundamental to a new social science. This in turn leads us to two inquiries: first, how the mind and brain work to understand their environment; second, how humans learn from one another, for example through culture.” —Douglass North in The Oxford Handbook of Political Economy 2006, p. 1005

All large complex societies have free-­rider problems. Whereas small-­scale societies can monitor the behavior of group members and observe and punish transgressions against a group’s rules and norms, large-­scale societies are far more anonymous. It is virtually impossible for a large complex society to consistently monitor all of its members to ensure that they behave according to the rules and norms they are assigned. Hobbes’s solution to this dilemma was the dictator. He feared that the true “nature of man” would lead individuals to defect whenever they had the opportunity and thus make complex and efficient societies impossible absent strong authority that could impose order (Hobbes 1651; reprinted 1909). Of course, many large complex societies in the world today do not depend on the imposition of strong authoritarian rule in order to function. Instead, social life and organization is built, at least to some large extent, on the fact that people willfully follow the rules of society.

Willing to Pay? A Reasonable Choice Approach. Sven Steinmo and John D’Attoma, Oxford University Press. © Sven Steinmo and John D’Attoma 2022. DOI:10.1093/oso/9780198796824.003.0002

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Why Should I Pay? A Cognitive Theory of Tax Morale  9 A massive body of literature exists today that demonstrates people will quite ­normally cooperate with others, sacrifice their individual self-­interest at times, and follow social rules even when there are no direct incentives for them to do so.1 Indeed, for a complex society to succeed and prosper, norms of trust and cooperation must be internalized by the members of that society (Elster 1989). One of the great puzzles in modern social science is the question: How and why does this happen? Why do individuals choose to follow the rules—even when they are not being monitored and even when so doing gives them no personal advantage? Tax compliance is a perfect example of this; though paying taxes is compulsory, most people will never be audited. Though most people today have their taxes automatically deducted from their paychecks by their employer, many have opportunities to under-­report a part of their income or over-­claim expenses or deductions from that income in ways that would reduce what they pay to the state. From a rationalist perspective, reporting all your income honestly to the tax authorities when the odds of being caught are minimal makes no sense. If people were truly “rational,” they should only cooperate with others, follow rules, or obey laws when it is in their self-­interest.2 But, a very large body of experimental evidence produced in recent years demonstrates that most people follow legal rules and social norms even when they have nothing to gain (Milgram 1974; Fischbacher et al. 2001; Alm and Torgler 2011). In fact, real people do not behave in the ways that many economists and political scientists predict.

A Reasonable Choice Approach In the following analysis, we offer an explanation for rule-­following behavior that pushes beyond classical instrumentalist theories and builds on the empirical evidence suggesting that humans are motivated by multiple needs or preferences. We call this a “Reasonable Choice Approach.” We use this term intentionally as we borrow from some of the key insights from Rational Choice Institutionalism (RCI), and at the same time introduce what we believe are some of the most important findings from the burgeoning research in the fields of evolutionary and social psychology and human cognition. In this chapter, we elaborate on this approach and set the stage for our empirical investigation of tax compliance behavior. As we suggested in the introductory chapter, we

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

10  Willing to Pay? believe tax compliance is a particularly useful arena to study cooperative behavior. Paying taxes, after all, is a social act, but typically it is done in private. This fact also lends itself to studying this behavior in a laboratory. Our approach is distinguished from a classical rationalist perspective in two major ways. First, instead of assuming people are autonomous individuals who rationally calculate their individual self-­interest, we begin from the basic fact that human beings are “social animals.” It has been analytically useful for scholars to attempt to understand human action when reduced to the individual. But one need not take the intellectual leap to assume that the individual is singularly motivated by their individual self-­interest. Humans, like all social animals, are normative creatures. We prefer to adopt the norms because we want to “fit in.” This is not just because acting outside the norm can invoke social disapproval, but also because most humans prefer to behave in accordance with local norms. This is why we tend to follow social norms— even when no one can view our behavior. Indeed, the very identity of an individual—that is to say the definition of self—cannot be separated from the social context in which this self is defined. Indeed, our “self-­interest” is a social construct. Second, our approach recognizes that humans are able to reason, but this does not mean that they are necessarily “rational.” As a number of psychological studies have shown, the human mind is a “sense-­making machine” that tries to build explanations, or theories, for the world around us (Kahneman 2011). It is a fundamental human desire to explain both the world around us and our own behaviors. Thus, we attempt to “rationalize” both what we see and how we behave. Our desire to rationalize does not mean that we use our brains to rationally calculate all our decisions, however. Real-­world decisions are in fact highly biased by our cognitive frames and past experiences. Indeed, as any student of human behavior knows, sometimes the theories we build can be quite irrational, and we can even become quite committed to these theories. We argue below that these “theories in our heads” help us to explain (rationalize) past behavior and to provide us a roadmap for future behaviors. Over time, they are internalized to the point that we do not actively consider them. Eventually, they become our values.

Institutions, Norms, and Values in Successful Societies The central finding in our earlier book, The Leap of Faith, was that some societies (most notably the Scandinavians) have built political institutions that:

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Why Should I Pay? A Cognitive Theory of Tax Morale  11 (1) effectively monitor their societies (are able to enforce norms), (2) practice procedural justice (are fair), and (3) deliver value for the money they collect in taxes. In these successful societies, taxpayers can see that it is in their self-­ interest to contribute (they get value for their money) and they believe others will also pay their fair share. Less successful societies, where institutions are less efficient, do not effectively monitor behavior in their society and thus allow some individuals to defect. In these cases, taxpayers see significant defection and are therefore more likely to defect themselves. In a society where there is much defection from socially positive norms, many people will decide that the system is not “fair” and consequently may themselves begin to behave in socially irresponsible or selfish ways. Repeating this behavior may be rationalized into the belief that cheating is justified. In short, when the rules of society are clear, and (near) universally enforced, most people will comply. When most people comply, more people comply. When more people comply, they create more social goods. More social goods lead people to believe that society is fair and just. If society is fair and just, reasonable people will desire to behave appropriately and correspondingly behave in a fair manner. You will be more socially responsible. If you are more socially responsible, the people around you are more likely to behave similarly. Also, if you and most of those around you behave in pro-­social ways, you will be more likely to shun those who are selfish.

Political Institutions and Self-­Interest Society’s rules (institutions) shape a self-­ interested individual’s behavior because they can reward some behaviors and penalize others. If these rules/ institutions are effective, they can shape a group’s, or a whole society’s, behavior in similar ways. Since we are social creatures, when we participate in a group in which most or all members behave in a particular way, we tend to adopt those behaviors (norms) as well. As we follow patterns of behavior, we build theories that help us explain our actions and predict what we should do in novel situations. These theories can eventually become internalized as values. For most of us, most of the time, each of these factors comes into consideration when we make a decision. Do I do what is in my self-­interest? Do I do what others expect of me? Do I do what I believe to be “right”? It is precisely because of our advanced reasoning capacities and our deep-­ seated social preferences that political institutions play such a powerful role

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

12  Willing to Pay? in shaping both our individual identities and behaviors. Once again, we ­certainly are responsive to the direct incentives and constraints of institutions (if there is a death penalty for not paying taxes, you’re more likely to pay them than if there is no penalty at all), but this captures only part of the motivations we have for paying or not paying our taxes. Rules set by our governments tell us what to do. They tell us both what is expected and what is likely to happen if we do not follow the rules. The most successful societies are ones in which there is a significant degree of coherence or consistency between society’s rules, public norms, and personal values. In these societies, acting in your self-­interest, acting like others around you, and doing what is “right” are brought into greater alignment. In these societies, behaving in socially responsible ways can become a kind of virtuous cycle. Less successful societies suffer from a vicious cycle. In these cases, significant defection from pro-­social behaviors (like tax evasion) creates a scenario where individuals come to believe that anti-­social behavior (like cheating on their taxes) is normal and even morally acceptable. Effective institutions encourage consistent behavior in their societies and reinforce positive social norms. Ineffective and/or inefficient institutions, in contrast, apply rules haphazardly and or allow some members of the community to be exempted from, or even to outright disobey, the rules. In other words, they do not establish or maintain consistent norms. When people observe (or believe) that others are not behaving according to the social norm, they are inclined to defect as well. In some ways you can say that the norm itself becomes “it’s okay to cheat or free ride.”

Comparing Countries and Individuals Even though there are differences between societies, it is important to remember that there are also significant differences in preferences, behaviors, and values within societies. All societies comprise a mix of different individuals. Some of us are taller, some are shorter. Some are more generous while some are meaner. Some are risk-­takers and some are risk averse. The key difference between societies is not whether everyone in that society differs from everyone else in another society, but rather the distribution of preferences and/or behaviors in each society. We may casually observe that many Americans are creative and/or risk-­takers, and we often naturally conclude that “Americans” are creative risk-­takers. But, obviously, not all Americans are risk-­taking individualists, no more than are all Swedes socially oriented altruists. This is not

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Why Should I Pay? A Cognitive Theory of Tax Morale  13 to say that there are no differences between societies, but rather the key issue is the distribution of behaviors, norms, and preferences within societies. As we shall see, differences within societies can sometimes be greater than differences between societies.3 In short, we see differences between societies as matters of distribution. While we can see patterns of difference, we also find huge overlaps. Even in the most cooperative/compliant societies, there are plenty of people who will cheat if they believe they can get away with it. Even in the most “individualistic” societies, there are plenty of people who fully cooperate and fully report absolutely all of their incomes all of the time. Figure 1.1 illustrates this point. Here we see two societies along a single dimension, “Individualists vs. Cooperators.” In society A, there are more individualists while in B there are more cooperators. It might make sense to say that Sweden is “more cooperative” than Italy. But there are nearly as many cooperators in Italy as in Sweden. Clearly, one cannot (or should not) think that any particular person in Italy is an individualist, or any particular person in Sweden is a cooperator. This sort of thinking is intellectually sloppy and can lead to gross stereotyping and misperception. With that said, however, it remains true that there are differences in the distribution of motives between these societies, and these differences can be important. The inherent problem with an overly simple culturalist analysis is that it can treat culture as monolithic, coherent, and/or stable. Cultures are none of these things. We argue that there are important cultural and normative variables that help explain variation in tax behavior (and by implication other behaviors as well), but we must understand these variables are neither fixed, constant over time, nor internally coherent and consistent. While there are tendencies towards certain types of attitudinal or normative “packages” in different societies, we must remember that not everyone in any society is like everyone else. While it might make sense to talk about Swedes, or Italians, in Society A

Individualists

Society B

Cooperators

Figure 1.1  Distribution of Individualists and Cooperators in Society

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

14  Willing to Pay? some contexts, it makes no sense if one thinks that this means that all Swedes are like X or all Italians are like Y. Swedes may well be slightly more honest and trustworthy then many other people today, but not radically so. Moreover, many Swedes are at least as willing to cheat their neighbors and/or their colleagues as any other people in the world. Averages can be misleading. But at the same time, small differences may be amplified by temptations toward confirmation bias. If you believe the people around you are honest, you will view many behaviors as demonstrations of that honesty. If you believe the people around you to be dishonest, that very same behavior may confirm the skepticism of others. The question is: If you’re cynical about the behavior of others, would you fudge a little yourself? Clearly, people can be self-­interested, some people will only think of themselves. But most people are driven by a mix of conflicting preferences and traits. Simply put, most people have a preference for cooperation and are self-­interested. This is why institutions matter so much. If you are motivated both by a desire to cooperate and by a desire to get the most for yourself, what you actually do in any particular situation depends a great deal on the normative and institutional context in which you make the decision. If the incentives are extremely clear in favor of cooperation, even those who are more self-­oriented are likely to cooperate. If the incentives for maximizing your individual self-­ interest by defecting or cheating on others are very strong, even those who have a preference for cooperation may well “defect.” If, on the other hand, it is not clear whether you are personally advantaged by cooperation—or behaving in a more self-­interested way—then some people are likely to cooperate and others to defect. Institutions structure incentives and disincentives. This is why tax compliance has increased in Sweden in recent years even while Swedish society has become more individualistic and diverse. This is because Swedish ­institutions—especially concerning taxes—have become more efficient over time. As a result the norm of tax compliance has been reinforced in this country over time. In the United States, by contrast, tax compliance has decreased in recent years. This is largely because successive administrations have cut back on IRS enforcement mechanisms. Americans do not like “cheaters” any more than Swedes. But if you believe many of those around you are cheating on their taxes (including the President of the United States) you may be a fool not to do so as well. In the American case, the norm of compliance is eroding. Consequentially, not only has tax compliance decreased, but equally importantly, both social and institutional trust are collapsing as well.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Why Should I Pay? A Cognitive Theory of Tax Morale  15 Figure  1.1, shows a normal distribution of individualist type people and cooperative type people in two idealized societies. We would predict that in Society A, slightly more people are likely to cooperate while in Society B, more people are likely to pursue their individual self-­interest. But, while there is significant overlap, there are also clear and identifiable patterns of behavior that can have huge consequences for the state and society at large. Tax evasion is just one good example. As evident in Table 1.1 below, tax evasion (meas­ ured as the “Tax Gap”) varies enormously across countries. Authorities in Sweden estimate a loss of approximately 5 percent of tax revenue to tax evasion. Even though this 5 percent loss accounts for billions, it is far less than the 32 percent Italian authorities lose to tax evasion annually. If the Italian state were as efficient in collecting taxes as the Swedish state, then it would be entirely possible for them to balance their budget and begin to pay off their huge public debt. The huge variation in tax evasion in these countries is not idiosyncratic. If one spends any time in these countries it is easy to observe a whole set of social (or anti-­social) behaviors that mimic the patterns indicated in this table. Whether it is obeying the posted speed limit, or not littering the streets, it can be easy to find examples where Swedes are more likely to follow social rules than Italians (or Americans, for that matter). Even while we must recall our warnings against cultural stereotyping, it is obvious that there exist patterns of behavior that cry out for explanation. Differences in tax compliance rates are highly correlated with levels of social trust (Rothstein and Uslaner 2005) as well as with perceptions of the benefit side of the budget (Coughlin 1980; Svallfors 1997; Taylor-­Gooby 1985). Svallfors and Edlund, for example, have argued that the willingness of Swedes to pay high taxes grows out of the fact that they believe they receive more out of the taxes they pay (Svallfors 1997; Edlund 1999; Taylor-­Gooby 1985; Scholz and Lubell 1998). It is also reasonable to expect, as some scholars have argued, that citizens are willing to pay taxes if they believe that public spending is allocated fairly and efficiently (Rothstein 1998; D’Attoma 2020). Table 1.1  Tax Gap by Country4

Italy UK Sweden US

National Tax Gap

Revenue Loss

Annual Deficit

32% 5.6% 5% 18–19%

$190 Billion $45 Billion $13.6 Billion $381 Billion

$38 Billion $59 Billion $3.2 Billion (surplus) $1 Trillion

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

16  Willing to Pay? We believe that this book, coupled with the first book we published from this research project, The Leap of Faith, provides an appropriate window through which we can better understand pro- and anti-­social behaviors both across and within societies. By using laboratory experiments, we are able to control the institutional context, while letting culture vary, and observe real behaviors. We can also vary different aspects of the tax system, such as tax rates, to see how institutional variables affect tax compliance. Further, the historical institutionalist analysis in The Leap of Faith provides the more fine-­ grained analysis that can help us explain the differences that we find in the lab. This deep historical perspective is not available in typical behavioral studies. Finally, we also provide a systematic study in four countries, which we believe to be one of the largest tax compliance experiments so far conducted anywhere in the world.

Willing to Pay? To test the Reasonable Choice theory, we conducted a set of experiments in 4 rich democratic countries in 14 different laboratories with over 1,700 subjects.5 The experiments were carefully designed to test how people in different societies would behave if they were given exactly the same choices. The experimental sessions were divided into three different sections. The first section, what we call the “baseline” experiment, was similar to a classical tax compliance experiment. In this section, subjects completed a typing task for which they were paid for each line they typed correctly. They were then given a series of different scenarios in which they would pay taxes on their earning. The tax collected would then be redistributed to the participants in the room. Tax rates as well as the different types of redistributive regimes were varied in the different scenarios. The key issue we were trying to ask was quite simply: if Swedes, Brits, Americans, and Italians faced exactly the same rules and choices—would they behave differently? In short, is the huge variation in tax compliance we can see in different countries a product of different institutions (a.k.a. “rules”), or do people in different societies cooperate and comply with rules differently due to some deeper differences in their societies and cultures? In the second section of the experiment, we attempted to test for variation in values directly. In this section, we presented the participants with a Social Value Orientation (SVO) exercise which measures an individual’s level of altruism. In this section subjects were faced with set of choices in which they

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Why Should I Pay? A Cognitive Theory of Tax Morale  17 were paired with another anonymous person in the room and given the opportunity to share a financial windfall with the anonymous “other.” They could keep all the money for themselves or share part of it. Once again, we were interested to learn whether people in some countries would be more “altruistic” than other countries and whether there was a connection between their levels of altruism and their tax compliance behavior. Finally, in section three, we gave our subjects a questionnaire in which we asked them about themselves and their attitudes. We asked questions which were specific to our experiment, as well as standard public opinion questions. Here we wanted to explore the relationship (if any) between individuals’ stated beliefs and attitudes, and their behavior in the two previous sections of the study. These experiments generated a huge amount of fascinating data, to say the least, which we will elaborate in the following chapters. Across countries we found many highly consistent behavioral patterns, as well as large differences. We describe the experiment in detail in Chapter 2 and examine the results in Chapter 3. Briefly, here we can offer that our evidence clearly demonstrates that institutions, social norms, and values are all important for tax compliance. We will show that institutional incentives and disincentives have predictable effects. For instance, most people in all countries increase their compliance as we increase the payoffs for contributions—or penalties for evasion. We will also show that social norms are important. When people believe that those around them contribute, they are more likely to contribute themselves. Finally, we demonstrate that individuals’ social values contribute to explaining why some people contribute to the public good (pay their taxes) or not. In the end, we show that people are not perfectly rational individualists, but instead, they are imperfectly reasonable social animals who react in ways that are cognitively consistent with their social environment. We conclude in the final chapter by tying all of this together and discussing its implications for successful gov­ern­ ance. As scholars who came into this with very little experience in experimental methods, we believe we have something valuable to offer those looking to embark on experiments in social sciences, and more specifically, laboratory experiments.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

2 The Experimental Design Virtually all students of tax compliance have realized the conclusion that tax compliance is driven by more than pure self-­interest (Halman 2009; Alm and Torgler 2006; Cummings et al. 2009; Frey and Torgler 2007). Several scholars have begun to identify this package of attitudes toward tax evasion as “Tax Morale” (Torgler 2007; Cummings et al. 2004; Frey and Torgler 2007).1 These studies by and large suggest that tax compliance goes far beyond the basic rational-­choice model; that is, variables such as gender, religion, income, ideology, and political institutions have important implications for tax compliance across the globe. For example, Benno Torgler and co-­ authors’ (Torgler 2002, 2003; Torgler and Schaltegger 2005; Torgler 2007; Torgler and Schneider 2007) work on tax morale examines a variety of different institutional settings across the world, demonstrating that political institutions, such as direct democracy, are strongly correlated with tax morale. Torgler and Schneider (2007) use the World Values Survey, the Latinobarómetro, and a controlled field experiment in Switzerland in which a test group of citizens receive a letter from the Swiss government outlining the importance of paying taxes. They found that while trust in the legal system is an important predictor of tax morale, the same can’t be said for tax rates. Similarly, direct democracy and referendum voting increase citizens’ information, which, in turn, affects tax morale (Torgler 2002). Surveys also suggest a strong relationship between many demographic variables and tax morale. Research has shown that women have significantly higher levels of tax morale than men (Torgler and Valev 2010). McGee, Ho, and Li (2008) examine cross-­cultural differences in tax morale in six countries (the United States, Brazil, Russia, India, China, and Germany), arguing that enforcement and education are key factors for tax compliance. The Chinese were the most adverse to tax evasion, possibly due to strict enforcement by the state, followed by the United States and Germany. Opposition to tax evasion and correlations between tax morale and education also varied from country to country. In the United States and India, the most educated showed the greatest opposition to tax evasion. Conversely, in Brazil, Russia, and China, the least educated were the most opposed to tax evasion. Willing to Pay? A Reasonable Choice Approach. Sven Steinmo and John D’Attoma, Oxford University Press. © Sven Steinmo and John D’Attoma 2022. DOI:10.1093/oso/9780198796824.003.0003

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  19 The literature also suggests that older people have more positive attitudes toward taxation than younger people (Ross and McGee 2012; Torgler 2007). Braithwaite and Ahmed (2005), utilizing cross-­sectional data from a survey of 447 recent college graduates in Australia, also made the claim that greater tax morale is a function of age, and Torgler and Valev (2010) confirmed the connection between age and tax morale. Existing literature also shows a link between religion and tax morale (Torgler 2007). In their six-­country study, Ross and McGee (2012) found that Catholics were more opposed to tax evasion than other religious groups, while Buddhists were least opposed. Finally, Gentry and Kahn (2009) used Internal Revenue Service (IRS) and zip-­code level data to examine the relationship of ideology and tax morale. They found that locales with voters who are more conservative actually shelter less income. Although we can learn much about tax morale through these surveys, surveys alone do not inform us about real behavior. They do not tell us about how human beings will behave given a set of parameters, incentives, and disincentives. Moreover, tax morale is framed within a particular tax compliance environment, and thus it is incredibly difficult, if not impossible, to control that environment. The tax morale literature might argue, for example, that people in some countries are more supportive of taxation than others, but it is very difficult to capture the reasons for those differences.

Experiments and Tax Behavior Early work using experimental methodologies was developed by psychologists, but economists were quick to realize that these techniques could be used to test many of their assumptions about human decision making (Smith 2008). This work has instead led to a far more nuanced and complex understanding of how and why individuals come to their decisions (Gintis et al.  2006; Kahneman et al. 1982; Tversky and Kahneman 1992). James Alm, one of the world’s leading scholars on tax compliance using experimental economics, summarizes the major findings in the tax compliance literature as follows: “Compliance is driven by far more than the purely financial considerations of detection and punishment, but in ways that are not yet fully understood.” “The challenge,” Alm tells us, “is to design precise experiments that parallel the essential elements of the naturally occurring world that are of interest, so that the experiments can demonstrate the external validity necessary to inform policy” (Alm 2010, 651). Our research team

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

20  Willing to Pay? did just this. Over five years, we designed, tested, and implemented a number of experiments in multiple countries that were specifically tailored to test the relationship between different institutions and citizens’ preferences in different contexts. We will describe the precise questions to be asked in each experiment below. But our chief aim was to build a series of scenarios that would allow us to test how different institutional contexts frame or shape citizens’ decisions. Through examining these decisions across countries, we hoped to gain a better understanding of the different choices and trade-­offs people would make in different societies. Several cross-­country studies done by behavioral economists to date have shown that differences in tax compliance are related to differences in both attitudes and beliefs (sometimes called “tax morale”) as well as differences in formal institutions. However, as far as we are aware, no one has systematically examined how people would respond to exactly the same institutions, rules, and incentives (under controlled laboratory conditions) in multiple countries and multiple laboratories under a tax compliance scenario as we have done here (but see, for example, Gërxhani and Schram  2006; Bergman 2009).

Running Experiments in Multiple Countries The key argument for using experiments in multiple countries is that this methodology could allow us to control the rules, or institutions, facing citizens in different societies. In an experiment, you can present exactly the same choices to people in different locations. Rather than asking people if they think cheating on taxes is morally wrong, in an experiment you can test their individual willingness to trust the authorities while providing them the opportunity to cheat on their fellow “citizens” and then measure how they actually behave. In short, through experiments, it might be possible to discern whether people behave so differently in different countries due to some deep-­seated (perhaps even cultural) differences, or simply because they face different choices in different societies. In other words, do Italians cheat more on their taxes because they live in a world of “amoral familialism,” whereas Swedes are culturally predisposed to be more honest and/or trusting? Do Italians cheat on their taxes because it is easy to get away with cheating, or because they feel that their government doesn’t offer them value for their money, or because they believe that everyone else is cheating?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  21

The Willing to Pay Experiments The Baseline Experiment In this section, we will describe the basic outlines of the experiments we conducted in this project and attempt to explain what we were hoping to achieve. We began the study by constructing what we call the “baseline” experiment. The baseline experiment is divided into three basic sections: (1) a tax compliance experiment; (2) a values experiment which attempts to measure individuals’ altruism; and (3) an attitudinal survey. The reason we chose three different sections or parts is that we wanted to see if there was a correlation between the way people behaved in a tax compliance experiment and their values and political social attitudes. We will describe each of these instruments or parts in turn.

Section I—Tax Compliance The first part of our experiment focused explicitly on measuring our subject’s willingness to pay taxes under different rules or conditions. In many ways, this was the heart of our study. This first section was divided into three “stages.” We describe each of these stages in detail below, but generally, we can say that Stage 1 tested the effect of different redistributive regimes on people’s willingness to contribute to a public good (or common pot of money); Stage 2 examined how different tax rates affected people’s willingness to contribute to the common pot; and Stage 3 looked at how different types of tax progressivity affected people’s contributions. It is important to understand that when our subjects (participants) came to the experiment, they did not know in advance what kinds of questions they would be asked, or even what the experiment was about. We recruited our subjects through a database of subjects called ORSEE (Online Recruitment Software for Economic Experiments) (Greiner  2015). Once they entered the room, they were randomly assigned to different computer terminals which were separated so that no one could view how others were answering. We then gave them the following instructions (note that while all instructions were provided to them on their computer screens, we also read them aloud):2

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

22  Willing to Pay? Welcome. Thank you for agreeing to participate in this experiment. This experiment will not be particularly difficult or involve trick questions. You will simply need to follow the instructions as they gradually appear on your screen. The answers you will provide will be confidential. Those who p­ rocess the data from the experiment will not be able to match your name to your choices. During the experiment, you will be asked to make choices. It is therefore ­important for the success of the experiment that you do not talk to each other and that you read the instructions very carefully. If you have questions during the experiment, please raise your hand. At the end of the experiment, you will receive your payment in cash. The actual amount will depend partly on your choices, partly on the choices of the other participants and partly on chance. Please press “Continue.” We informed the participants, first, that all of their answers and performance in the experiment would remain confidential, and that no one would ever know what choices they had made, how they had performed, or how much money they eventually made during the experiment. They were then told that they would be paid in an envelope at the conclusion of the experiment. Then they were provided with some basic information about the experiment to follow: The first three stages of the experiment are similar to each other in many ways. In each of the first three stages, you will be asked to perform a clerical task for 5 minutes. Based on your performance, you will earn currency units; these will be converted into dollars at the end of the experiment. You will participate in several rounds, during which you will be asked to report your income for tax purposes. You are free to report any amount, from 0% to 100% of your earnings. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, the computer will choose a number between 1 and 100 for each of you in each round. If the result is a number between 1 and 5, you will be audited for that round. If you are audited and you have under-­reported your earnings, you will have to pay the tax you should have paid, plus a fine equal to that tax. In other words, you will have to pay twice the tax you should have paid. Note here that we told the subjects that they would earn money by typing and that the money they earned would be “taxed.” We also told them that they

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  23 will only be taxed on the income they reported, but that they might be “audited” and would be fined if they were caught for under-­reporting. It is important to note that we intentionally framed the experiment at the very outset as a “tax” frame. We wanted people to understand that this experiment was about taxes and not a simple gambling choice. If we want to understand why “tax morale” differs in different countries, then it was obviously ­important to invoke the morality of taxes as we did here. We then handed participants sheets of paper titled “Clerical Task.” Figure 2.1 is a screen capture of the clerical task, and Figure 2.2 is a screen capture of how they entered the rows on the computer. Each sheet had lines of text which were essentially gibberish. These lines of text included capital letters, lowercase letters, and numbers. It was intentionally written in this manner to be rather difficult to type, and so that no language speakers would be advantaged over others. We strived to make it hard for everyone and equally hard for people in multiple different countries and languages. Participants were then told that they would be paid “experimental currency units” (ECU) for each line they typed correctly and that these currency units would be converted to real dollars/euros/kronor/pounds at the end of the experiment. The exact payment amount depended on the country, but essentially, we can say they earned about $0.10 per correct line and the average payment at the end of the experiment was roughly between $15 and $25 or at least 1.5 times the hourly minimum wage (or equivalent in local currency). The participants were told that their final earnings would be based partly on how much they typed, partly on the choices about how much they would report, partly on the choices of others in the room, and partly on luck.3 Row 1 2 3 4 5 6 7 8 9 10 11 12 13

ID number 015678 124578 259076 013457 134567 987231 018973 967453 689043 234567 012348 980531 123567

Last name Gnpuwy Wekrzb Dfioqvw Bdefiruy Fmhyzni Chkopvwx Bgjkly Fhnptwx Fmorvw Abehlpuv Dfkmqrs Jkpqtvwy Cprsxv

Figure 2.1  Screen Capture of the Clerical Task

First name Hgjkwz Dfhlnx Cikvwx Guzyhmi Cfikmw Unpzmyio Bhiqsvx Bcgipst Cdjipgr Bgipquvw Bemorstw Dfhijx Efikmrv

Vote 7 0 4 9 5 0 5 1 8 8 4 1 9

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

24  Willing to Pay?

Figure 2.2  Screenshot of Clerical Task

Before they began the experimental treatments, participants were provided with a two-­minute opportunity to practice the clerical task, after which they were given the first “real” clerical task for five minutes. Stage 1, Round 1 Once they had completed the task, they were told how many lines they had typed correctly and how many ECUs they had earned. They were then told the following: In the first round, we ask you to report your income for tax purposes. The tax rate is 30%. Please remember that you are free to report any amount of your earnings from 0% to 100% for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and are caught under-­ reporting your income, you will pay the tax you should have paid, plus a fine equal to that tax. In other words, you will pay twice the tax you should have paid.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  25 Next, they were given several examples to show how different kinds of choices might affect their final earnings, including what would happen if they did not report their full income and what might happen to their final pay if they under-­reported their full income and were audited. After it was clear that the participants understood the instructions and the examples, they were shown their earnings from the typing task and asked to “report their income for tax purposes.” Once again, the tax was 30% and they were told nothing about what would happen with the taxes they paid. Figure 2.3 is a screenshot of the reporting screen. Before we move forward, let us look at an example for how participants’ decisions affected their payoff. EXAMPLE 1 Let’s assume that we have three participants: X, Y and Z. All three participants copy 10 names correctly during the clerical task and each participant therefore earns 100 currency units. If Participant X accurately reports his total real earnings of 100 currency units and the tax rate is 30%, he will pay 30 in taxes. His final income will therefore be 70 currency units.

Figure 2.3  Screenshot of Reporting Screen

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

26  Willing to Pay? If Participant Y reports only 50 currency units, she will pay 15 currency units (since you are only taxed on the income you report). In this case, her final income will be 85 currency units, if she is not audited and fined at the end of the experiment. If Participant Z reports 0 currency units, he will pay 0 currency units in taxes and his final income will be 100 currency units, unless he is audited and fined at the end of the experiment. EXAMPLE 2 What happens if all three participants (X, Y, and Z) are audited and fined at the end of the experiment. If Participant X is audited, nothing happens: since he reported his total real  earnings, he will not incur any penalty and his final income will be 70 currency units. If Participant Y is audited, then since she underreported her earnings, she will have to pay twice the tax she did not pay. Since she reported 50 currency units, she owes 15 currency units in taxes. She will therefore have to pay double that (30 currency units) and her final income will be 55 (since 85 − 30 = 55). If Participant Z is audited, then since he reported 0 currency units, he owes 30 currency units in taxes. He will therefore have to pay 60 currency units and his final income will be 40 (since 100 − 60 = 40). Stage 1, Round 2 Once each participant reported their incomes, they were provided their second set of choices. In this case, they were told that the choice was identical except that in this “round” the money collected in tax revenues would be placed into a “general fund” which would be redistributed to all participants of the experiment in equal amounts regardless of how much each individual had contributed to the fund. They were told that their choices in this round would be completely independent of Round 1. In other words, they began this stage with the same amount of money they had earned in their typing task. They were given multiple examples of how this would work so they clearly understood that they had a choice about whether they would contribute to the general fund, or not, while with the full knowledge that they would receive money from the fund whether they had put money in or not (this is similar to a classic public goods experiment with a tax frame). The major difference between this round and the first is that the tax money was put into a group fund and distributed equally between participants. For example:

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  27 EXAMPLE Let’s assume that 3 people participate in the experiment, and that each one of them earns 100 currency units during the clerical task. One participant reports 100 currency units, one reports 50 currency units and the other reports 0 currency units. The amount collected in taxes and put into the general fund is therefore 45 currency units (since 30 + 15 + 0 = 45). Each participant receives 15 currency units from the “general fund.”

Participants, once again, were shown their earnings from their original typing task and reminded that the amount they had paid in the last case would not affect their current earnings or their current choices. Stage 1, Round 3 Once everyone had registered their choices in Round 2, they were given a final round in this stage of the experiment which was identical to Round 2, except that in this case the amount of money contributed to the general fund would be doubled and then redistributed in equal parts to all participants. Once again, they were provided with examples to show how this would work. The point was to help them to understand that they had a collective interest in contributing to the fund, while at the same time they had an individual “self-­ interest” in cheating the fund. In other words, there would exist a conflict between the public good and self-­interest. To be sure that they understood what they were doing, we gave them the following example: EXAMPLE If the total amount of tax collected is 500 currency units, the amount in the general fund will be doubled and the amount redistributed in equal parts among all participants will be 1000 currency units. Let’s look at another example. Let’s assume that 20 people participate in the experiment and that each of them pays 100 currency units in taxes. In this case, the amount in the general fund will be doubled to 4000 (since 20 × 100 = 2000 × 2 = 4000) and each participant will receive 200 currency units from it.

The participants were once again shown the screen displaying how much they earned in the original typing task and then asked to “report their income for tax purposes.” We examine our results from each part of our study in greater detail in Chapter 3, but to preview these results, there is a bit we can say. First, most

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

28  Willing to Pay? people comply even when there is no direct payoff or return benefit for the taxes they pay. Second, increasing the common good payback or redistributed amounts significantly increases people’s compliance rates. And third, we did not find the kind of national variation that we expected to find. Stage 2, Rounds 4–6 Once our participants completed reporting their incomes in Round 3, we moved on to the “Second Stage” of the experiment. Whereas in the first three rounds of the experiment we were interested in how people in different countries responded to different allocations and payoffs from the public good, in rounds four through six we were interested in how raising taxes affects compliance. The classic Allingham and Sandmo model (1972) suggests that increasing rates should decrease compliance, and of course, it is a hotly debated political topic. However, given this, behavioralists have spent surprisingly little time analyzing how tax rates affect compliance. Moreover, real-­world observation would suggest that tax rates might not matter so much, since evasion in countries such as Sweden is very low even though they have very high taxes. It is here, we thought, we could contribute something with our study. Our participants were once again told that this part of the experiment would be very similar to the last one, but that the choices they would face would differ slightly. We then asked them to complete a second typing task, “Clerical Task #2,” in which they would once again earn currency units that would later be converted into their local currency. They were informed that their earnings and reported incomes in this stage were completely in­de­pend­ ent of their performance and earnings from Stage 1; they then began typing for five minutes. In this stage, the participants were asked to report their incomes in three different “Rounds,” each of which would present them with a different tax rate (10%, 30%, and 50%). It is important to understand that the subjects were unaware of how many rounds there would be, or what kind of questions would be asked in each subsequent round. After they completed the second typing task, they were told that they would again be asked to contribute to a general fund and the money in the fund would be redistributed to all participants in equal shares, no matter how much they had contributed. In all three rounds (4–6), the amount of money collected in the “general fund” was doubled and redistributed equally to all participants—regardless of how much they had paid into the fund, just like in Round 3 of Stage 1. We will not detail the language used, but we did provide examples to the

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  29 participants in the experiment as it was quite straightforward and similar to the other rounds. Importantly, the participants did not know what would follow in each round. Stage 3, Rounds 7–8 Finally, we moved on to the third and last stage of our compliance experiment. Stage 3 of the experiment was quite similar to the first two stages. Subjects were provided with another clerical typing task and were asked to type it in five minutes. They were then asked to make a set of choices about how much tax they would pay under different conditions. In this third stage, however, they were presented with two different types of progressive tax systems and finally a scenario in which the tax money collected would be donated to a public charity.4 In this round (Round 7), as in the rounds before, they were reminded that there was a 5% probability that they would be audited and if they under-­reported their real earnings they would be assessed a fine equal to twice the amount of tax that they had avoided. They were also assured in the last round that the money truly would be donated to the specified charity. Below we provide examples of the decision and how the participants’ decision affected their total payout. EXAMPLE 1 Let’s look at an example. Let’s assume that ten subjects participate in the experiment. Of these ten participants, one copies 20 names correctly, one ­copies 5 names correctly, and the other eight copy 10 names correctly. Every correct name is awarded with 10 currency units. Let’s assume too, that all participants then report their total real earnings for tax purposes. The participant who copies 20 names correctly has an income equal to 200 currency units and pays 50% of it (=100 currency units) in taxes. These 100 currency units are collected in the “general fund.” The participant who copies 5 names correctly has an income of 50 currency units and pays 10% of that income (=5 currency units) in taxes. These 5 currency units are collected into the “general fund.” Each of the participants who copies 10 names correctly has an income equal to 100 currency units and pays 30% of that income (=30 currency units each) in taxes. A total of 240 currency units (8 × 30) is collected in the “general fund.” The general fund is then doubled and distributed equally among participants.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

30  Willing to Pay? EXAMPLE 2 There are now 345 currency units in the general fund (100 + 240 + 5 = 345). The amount collected in the general fund is then doubled (345 × 2 = 690) and redistributed in equal parts among all participants (690/10 = 69). Every ­participant therefore receives 69 currency units from the “general fund.” In this example, the participant reporting the highest income (200 currency units) pays 100 in taxes, receives 69 from the “general fund,” and therefore has a final income of 169 currency units. The participant reporting the lowest income (50 currency units) pays 5 in taxes, receives 69 from the “general fund,” and therefore has a final income of 114 currency units. Each of the other participants (reporting 100 currency units each) pays 30 in taxes, receives 69 from the “general fund,” and therefore has a final income of 139 currency units.

In reality, progressive tax systems do not work this way, but this is the way many people think they work. For Round 8, we implemented a progressive system that was more like the type of tax system that most people face in the real world. Therefore, everyone paid 10% for the first 50 currency units of reported income, 30% for any reported income between 51 and 100 currency units, and 50% for any reported income above 100 currency units. In other words, all participants paid the same tax on the same amount of income but paid a different tax rate on different levels of reported income. The instructions were as follows: – INSTRUCTIONS The second round is very similar to the first one. As in the first round, we ask you to report your income for tax purposes. Again, you are free to report any amount of your earnings (from 0% to 100%) for tax purposes. The probability of being audited at the end of the experiment is, again, 5%. Once again, the money collected in taxes from all participants will be put into a “general fund,” doubled and redistributed in equal parts among all participants, regardless of how much each participant paid in taxes. The new thing about this round is that you pay 10% for the first 50 currency units of reported income, 30% for any reported income between 51 and 100 currency units, and 50% for any reported income above 100 currency units. In other words, all participants pay the same tax on the same amount of income, but you pay a different tax rate on different levels of reported income.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  31 EXAMPLE Let’s assume that Participant X reports 80 currency units for tax purposes. The amount paid in taxes will be equal to 10% on the first 50 currency units, plus 30% on the next 30 currency units, so 14 (= 5 + 9) currency units. With these two versions of progressive taxation we hoped to discover whether individuals are particularly sensitive to tax rates at higher levels of income. As you will see in the following chapter, we found that they most certainly are.

Section II—Measuring Social Values Once subjects finished the first three stages of our experiment we moved onto the next major phase of our study. Stages 1 through 3 measured tax compliance under different institutional conditions. But we also wanted to know if their basic value differences would correlate with individuals’ willingness to contribute and if there were significant differences in value orientation across societies. To measure this, we turned to what is called a Social Value Orientation (SVO) exercise designed by Murphy, Ackermann, and Handgraaf (2011). This instrument was designed to measure levels of prosociality. In this case, subjects were tasked with fifteen allocation decisions (one decision per screen) to elicit their “social value orientation.” We told subjects the following: We now begin the fourth and final stage of the experiment. This stage is different from the first three. In this stage, you have been randomly paired with another person, whom we will refer to as “the other.” This other person is someone you do not know and will remain mutually anonymous. All of your choices are completely confidential. You will be making a series of decisions about allocating resources between yourself and this other person. For each of the questions that will follow, you will be asked to indicate the distribution of resources you prefer by marking the respective position along the midline, as in the example below. You can only make one mark for each question. There are no right or wrong answers; this is all about personal preferences. We then provided them with an example of the SVO choice. Figure 2.4 is a screenshot of the SVO choice. These decisions required participants to choose an allocation of tokens (money) between themselves and an anonymous partner. Here it is also important to understand that the SVO choices that subjects were provided

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

32  Willing to Pay?

Figure 2.4  Social Value Orientation Example Allocation Decision Notes: In each allocation, the first value is the number of tokens the decision-­maker keeps for themselves and the second value is the number of tokens the other person receives.

with were sequentially offered and were independent of each other. At no point were the subjects aware of how many of these choices they were going to make, or more generally, what would follow. The first six decisions were constructed to assess different individuals’ willingness to share an endowment with a random stranger. As seen in Figure 2.4, subjects can choose to maximize their own share of their endowment—or to share part of it with a random “other person” in the room. The key to this instrument is to measure the degree to which a participant is willing to sacrifice a share of their own money with another person—even when that person would never know who they were, and they would not know who they were “donating” to.5 For each choice they were asked to make, they were instructed to indicate the distribution of resources they would prefer by marking the respective position along the midline, as demonstrated in Figure  2.4. They could only make one mark for each question. Finally, they were told that there were no right or wrong answers; this is all about personal preferences. In the example above, a person has chosen to distribute money so that she receives 50 tokens, while the anonymous other person receives 40 tokens. Every set of 10 tokens

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  33 had the value of 0.03 dollars. Just to reiterate, participants played this out over a series of fifteen decisions, but we only use the first six to measure levels of altruism. The participant’s SVO angle is then constructed based on their responses to the six decisions and can be used to classify people according to their social preferences as either caring about others, that is being prosocial (22.45°–57.15°), or caring primarily about oneself, that is being an individualist (−12.04°–22.45°). The modal SVO angle is associated with perfectly individualistic preferences, simply choosing the option that maximizes one’s own payoff. The mean SVO Angle was 22.50°, meaning that the average individual in our total pool had prosocial leanings, although there were a decent ­number of individualist types as well. We will dig much deeper into this in the following chapters.

Section III—Measuring Attitudes Finally, at the end of the experiment, our participants were asked to complete a survey asking them questions about themselves and regarding their views on various policy issues. Here we wanted to see if there was a correlation between how people behaved in the tax compliance experiment and/or the social value orientation experiment and their political attitudes and/or policy preferences. It is reasonable to assume, for example, that people who believe in economic redistribution at the policy level might be more likely to comply in our tax experiment than those who are more hostile to taxes, for example. We provide this entire survey in Appendix 2. Table 2.1 contains examples of just a few of the types of questions we asked beyond the basic demographic questions. It is also important to note that our survey questions were drawn from the World Values Survey and the European Values Survey and that the national responses to our survey generally aligned with these much larger, nationally representative samples. To sum up: our baseline experiment was divided into three sections. The first section was similar to a classical tax compliance experiment. For each stage of this experiment, we were interested in a different institutional change; in Stage 1 we varied the amount of redistribution; in Stage 2, we varied the tax rates; and, in Stage 3 we varied the tax regime (i.e. different forms of progressivity). The second section presented the participants with a SVO exercise which measured their level of individual altruism. Finally, in section three, we asked them about themselves and their attitudes.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

34  Willing to Pay? Table 2.1  Example Survey Questions Question

Possible Responses

People can only get rich at the expense of others – Wealth can grow so that there’s enough for everyone For those with high incomes, are taxes

1 to 10

For those with low incomes, are taxes

Cheating on taxes if you have a chance

1 = Much too high 2 = Too high 3 = About right 4 = Too low 5 = Much too low 1 = Much too high 2 = Too high 3 = About right 4 = Too low 5 = Much too low 1 (Never justifiable) 23456789 10 (Always justifiable)

For the most part, we found consistent results within countries, providing some validation to our experimental findings. Across countries we found many highly consistent behavioral patterns, as well as large differences. Tables 2.2 and 2.3 describe the basic characteristics of our data in all countries and the summary statistics for our compliance rate—our main de­pend­ ent variable. As the reader can see, the average compliance rate was high (65 percent) considering the “optimal” decision according to rational choice would always be to report 0. Exactly half our subjects were women, and the vast majority of our subjects were students. In the next chapter we analyze this data. We look at the effects of increasing redistribution, increasing taxes, and changing the tax regime on overall compliance in the experiment. We link these three institutional variables to self-­interest. Then we investigate how norms fit into the tax compliance decision. In other words, how do my beliefs about others’ behaviors affect my willingness to pay taxes? Finally, we follow this up with an analysis and discussion of values. How does my belief in what one ought to do affect my willingness to pay taxes? For each of these variables, we test how they affect compliance in all countries overall, and how they vary by country. Needless to say, our results are interesting and not always as we predicted.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

The Experimental Design  35 Table 2.2  Summary Statistics Personal Characteristics

Mean

Std. Dev.

Past-­Participation (dummy) Male (dummy) Economics Student (dummy) Age (years) SVO-­angle

0.714 0.506 0.152 24.757 22.503

Lab Country

Median

Min

Max

Obs.

0.452 0.500 0.359 10.395 14.809

1.000 1.000 0.000 22.000 23.857

0 0 0 18 −16

1 1 1 255 61

1,611 1,706 1,730 1,709 1,730

Freq.

%

Cum. %

Italy (IT) United Kingdom (UK) United States (US) Sweden (SE)

311 526 566 327

17.977 30.405 32.717 18.902

17.977 48.382 81.098 100.000

Lab Region

Freq.

%

Cum. %

Milan Bologna Rome Oxford Exeter RHUL Essex Oxford adults Boulder Hawaii Santa Cruz Appalachian Stony Brook Stockholm Gothenburg Total

116 106 89 126 120 114 70 96 130 118 106 108 104 227 100 1,730

6.705 6.127 5.145 7.283 6.936 6.590 4.046 5.549 7.514 6.821 6.127 6.243 6.012 13.121 5.780

6.705 12.832 17.977 25.260 32.197 38.786 42.832 48.382 55.896 62.717 68.844 75.087 81.098 94.220 100.000

Notes: Abbreviations: Std. Dev. for Standard Deviation; Obs. for Observations; Freq. for Frequency; Cum. For Cumulative.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

36  Willing to Pay? Table 2.3  Summary Statistics for the Compliance Rate Average Compliance Rate

Lab Country Italy (IT) United Kingdom (UK) United States (US) Sweden (SE) Total

Mean

Std. Dev.

Median

Min

Max

Obs.

0.647 0.577 0.694 0.676 0.647

0.422 0.442 0.404 0.430 0.427

0.944 0.750 1.000 1.000 1.000

0.000 0.000 0.000 0.000 0.000

1.000 1.000 1.000 1.000 1.000

311 526 566 327 1,730

Notes: The table reports summary statistics of the compliance rate, operationalized as a subject’s reported income divided by their total earned income. This is a continuous variable ranging from 0 to 1. Abbreviations: Std. Dev. for Standard Deviation; Obs. for Observations.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

3 Reasonable Choices and Tax Compliance Culture vs. Institutions? There are essentially two competing explanations for why tax compliance is so much lower in some countries than in others. Perhaps the most common explanation for the fact that Italy, for example, loses nearly 40 percent of potential tax revenue to tax evasion is that “it’s the culture.” This implies that Italians are less socially responsible than citizens in other (northern European) countries like Sweden. The alternative explanation is that people are basically the same in all countries and the key explanation for variance in compliance rates (tax or otherwise) is institutional. According to this view, Italians are more likely to cheat on their taxes than Swedes because (1) their opportunities for evasion are greater, (2) the odds of being caught are lower, or (3) the penalties they will likely suffer if they are caught are significantly lower in Italy than in Sweden. In our view, both explanations make some sense. Having lived in several countries, it is undeniable that different countries have different cultures, and even while this variable is difficult to pin down, it is intuitive that cultural variables should affect something as specific as tax compliance. At the same time, it is equally clear that institutional rules, and the effectiveness and efficiency of public institutions (like tax collection), vary significantly across countries. Quite simply, even causal observation while living in Italy teaches you that being caught and fined for tax evasion is far less common than the behavior. (The same is true for speeding on the highways.) Sweden seems to be quite different. In order to test this rather basic issue we structured our compliance experiment to test for these alternative hypotheses. Quite simply, we expected that if humans simply calculate their narrow self-­interest when making compliance decisions, then there should be no variation in responses across countries since the incentives were identical. If it were true that “culture” is the best explanation for the high compliance rates in Sweden and low compliance

Willing to Pay? A Reasonable Choice Approach. Sven Steinmo and John D’Attoma, Oxford University Press. © Sven Steinmo and John D’Attoma 2022. DOI:10.1093/oso/9780198796824.003.0004

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

38  Willing to Pay? rates in Italy, for example, then we would expect our experiment to show this as well. As the reader already knows, this is not what we found. Instead, what we found was that there is signficant variation both between and within each of these countries. Surprisingly, though, we did not find that the patterns of behavior in our experiment matched the behaviors one would predict given a simple culturalist explanation.

Reasonable Choices In our view, the problem with both institutionalist and culturalist explanations is not that they are wrong, but rather they are too simple. Our approach, which we call Reasonable Choice theory, attempts to integrate the insights from institutionalist and culturalist approaches. We argue that almost everyone is influenced by their self-­interest, their norms, and their values. All three of these variables matter, and together these variables help us explain why some people are more “Willing to Pay” than others. If classical economists were right, then we would expect that there is very little variation when citizens in different nations are faced with the same incentives and disincentives. If sociologists are right, it should be that differences in value orientations would be the main driver of variation in compliance. If psychologists are right, we would expect that social norms to be the most important explanation for variance and tax compliance. If we are right, each of these variables should provide partial explanations for why some individuals comply more so than others and also help us explain why there is so much variation between countries in their compliance rates overall. The chapter proceeds as follows. First, we demonstrate the impact of self-­ interest; this is the classical perspective. The standard literature on tax compliance focuses on punishment—suggesting that people only pay taxes due to fear of being caught and punished for evasion. As will be seen, self-­interest does powerfully affect citizens’ willingness to pay (contribute to the public purse), but instead of simply focusing on the negative side of taxes, we are also interested in understanding the conditions under which citizens are willing to increase their contributions to the public good. After all, you normally should expect to get something back from the taxes you pay. Taxes pay for real things, such as roads, schools, security, a cleaner environment, etc. As we will see, what individuals get back from their taxes can have as great, or even greater, effect on their compliance rates than how much they are required to pay.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  39 Second, norms matter. Only a purely classical economist (or someone on the spectrum of a classical economist) could believe that people do not care about the behaviors of others when making their own decisions. Our data support the existing evidence offered in psychological and cognitive studies that have repeatedly demonstrated the effects of norms on behavior. We show that the perception of others’ behavior has a significant impact on an individual’s behavior and that the expectations of other’s also vary across societies. Finally, we show that social values are central to tax compliance decisions. The belief that someone ought to pay their taxes plays an important role in one’s decision to pay taxes or not. Few would argue against this proposition, but (as far as we are aware) no one has tried to empirically test this proposition. We employ the unique Social Values Orientation (SVO) study to explore the relationship between compliance and social values. We find that values, like norms and self-­interest, are all important factors contributing to individual compliance decisions. We also find that the distribution of values (prosocial vs. individualist) varies both within and across societies.

Self-­Interest and Institutions “It is not from the benevolence of the butcher, the brewer or the baker, that we can expect our dinner but from their regard to their own interest”1 —Adam Smith in The Wealth of Nations

At the foundation of modern economics and tax policy is the theory that individual decisions are largely cost–benefit calculations. Since the beginning of the modern state, this is the basic logic that has driven most revenue officials and tax policymakers. The belief has been that through increased penalties and audit rates, compliance would increase. Remarkably, however, several empirical studies have shown that the “punishment model” which underlies most actual tax policy seems to be significantly less effective than classical rational choice theory predicts. Still, there is no doubt that self-­interest is an essential feature of human behavior. Interestingly though, in the real world, audit rates and punishments are surprisingly low—in fact, they are far below what a pure homo economicus model would predict is necessary to ensure tax compliance. Yet, compliance rates are quite high in many countries. This is not to say that individuals do not make a cost–benefit analysis when it comes to paying taxes;

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

40  Willing to Pay? instead, the cost–benefit analysis or utility maximization tenets are defined within a particular compliance environment (Bergman 2009). Audit rates and penalties vary significantly between countries,2 and con­ sist­ent and effective tax auditing does appear to deter evaders.3 Increasing fines for evasion does tend to increase compliance—but only up to a point.4 This is clearly related to an individual’s threshold for risk and their utility derived from taking risks.5 The bulk of both the experimental6 and theoretical literatures7 suggest that audits and fines do work up to a point, but they can also have unintended effects, especially when the deterrence mechanism is too forceful. For example, Mittone et al. (2016) show that audits can actually increase the likelihood of evasion in subsequent years as taxpayers believe they will not be audited more than once. Supporting this conclusion are studies using Internal Revenue Service (IRS) data in the US. One study reports that operational audits had a large counterdeterrent effect on taxpayers who received a tax assessment in one year, but not in the subsequent year. They argue this is due to what is called the “bomb-­crater” effect, where audits can reduce the tax morale of compliant taxpayers (Beer et al. 2019). Similarly, high fines appear to also have negative consequences for overall compliance. Several research studies have shown that high fines ultimately undermine citizens’ willingness to pay, because they may engender resistance and hostility toward the state. Kastlunger et al. (2011) summarizes this nicely by saying, Overall, tax compliance was not affected by the rewards. However, a change in compliance strategies was observed. It seems that rewards provoked an all-­or-­nothing behavior. Whereas in the reward conditions, participants were either completely honest or evaded all their income, in the control condition, the amount of evasion varied more strongly. Furthermore, audited compliant taxpayers who are rewarded evaded less in the following period compared with audited compliant taxpayers who experienced no rewards (p. 150).

The above does not suggest, though, that self-­interest is of no importance. The point is instead that self-­interest is mitigated by several other variables. Clearly, if the death penalty were imposed on tax evaders, and the state was good at capturing tax evaders, few people would risk tax evasion. Yet, in the real world there appears to be no direct correlation between tax penalties and rates of tax evasion. In fact, the country in our study with the mildest penalties for evasion (Sweden) has the highest level of compliance. The US, which

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  41 is known internationally for imposing significantly heavy fines and even lengthy prison terms for tax evaders (think Al Capone), still loses USD $381 billion dollars a year to tax evasion (IRS 2019). This is one reason that the efficiency and effectiveness of formal institutions are so important. Institutions structure behavior because they provide the rules (incentives and disincentives) in which people make decisions and thereby can calculate their self-­interests. Efficient tax institutions provide the incentives for individuals to behave in predictable ways. Inefficient institutions (e.g. where the odds of being caught are low, or where authorities ­themselves are easily corrupted) offer greater incentives for self-­interested individuals to act in their own self-­interest and shirk their public financial responsibilities (e.g. pay their taxes).8 This last point suggests an additional variable in explaining how self-­ interest contributes to tax compliance, which is too often ignored by public authorities. As argued in Chapter 1, self-­interest does not have to be defined as simply our personal and autonomous interest. Since we are social creatures who live in social environments, “self-­interest” can also include collective goods. For example, I may see it as my “self-­interest” to live in a society which has good public services, less crime, etc.—from which the whole society ­benefits—even if I have to pay out of my own pocket for these collective goods. In other words, it is in our collective self-­interest if everyone pays their fair share of taxes as taxes pay for things, we all benefit from, regardless of country. Paying taxes is part of an exchange relationship between the state and its citizens (Webley et al. 1991). The state provides a service (i.e. public goods) and citizens reciprocate by paying their taxes. This is what Levi called “contingent consent” (Levi 1997). Of course, from a purely homo economicus perspective, this makes no sense. If I evade my taxes and everyone else pays, then I can benefit from their payments and save my money. But this is not how (most) people behave. Indeed, as we will show below, even when the odds of  being caught for evasion are very low, most people—in all countries— contribute to the public good through taxes. In our study, we tested for this social aspect of self-­interest by offering collective as well as individual incentives. The results show very clearly that as the collective payoff increases, all but the most individually selfish will contribute more to the public purse.9 We vary the types and conditions of “collective self-­interest.” In brief, we wanted to know: how do individuals in different countries respond to differing incentives (such as an increase in the provision of the public good or an increase in the tax rate); and second, how does the intensity in response vary by country?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

42  Willing to Pay? This setup makes our study unique. Systematic studies across countries regarding how individuals respond to redistribution and the provision of the public good are rare in themselves. As far as we are aware, no one has tested whether there might be variation across countries in individuals’ willingness to pay under different redistributive regimes. Recall that our study had subjects first earn money through a random typing task. We then told them that we wanted them to pay a 30 percent tax on their earnings, and there was a minimal (5 percent) probability that they would be audited at the end of the session. We also gave them examples showing that if caught for evasion, they would pay the tax that they had evaded plus a fine equal to the tax they had evaded. In the first “round,” this is all they were told. In the second “round,” they were provided the same audit/fine rules but told that the money collected would be redistributed to everyone who participated in the experiment—regardless of how much they put in. In other words, even if an individual reported earning zero income, they would get the same amount back from the collective purse as everyone else. In the third round, we gave the same basic rules, but in this case, we told them that the amount collected in the purse would be doubled and redistributed equally to all participants. To be clear, the direct individual disincentives (fines, audit rates, etc.) for cheating remained constant throughout the tax compliance experiment. We wanted to see if people would pay more in if they felt that everyone would get more out of it. We aggregate the responses from all countries in Figure  3.1, which represents the average compliance rate, pooling our data for rounds one through three of our experiment. We see from the first bar in Figure 3.1 that across all of our countries the average compliance rate in Round 1 is 51 percent—even when there was no redistribution. To some, this finding is surprising. Why, after all, would anyone pay taxes when the odds of being caught for evasion are so low and the penalties, even if caught, are also so low? We believe that the answer is found in our use of the term “tax.” In fact, Lawerence et al. (2016) tested the differences in compliance between different frames. Their framed experiment was a typical framed tax compliance experiment for which students earned money and paid taxes on that money with some probability of being caught if they failed to report their full income.10 They demonstrated that neutral framing elicits significantly less compliance than the tax frame. Most people believe they ought to pay taxes. In the case of this experiment, our subjects earned money in the experiment and many thought they

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  43

70% 0.74 60% 0.61

Tax Compliance Rate

50%

40%

0.51

30%

20%

10%

0% R1: No Redistribution

R2: Redistribution

R3: Redistribution x 2

Figure 3.1  Average Compliance Rate in Rounds 1 through 3 (all countries, all sessions)

were obligated to pay taxes for participating in the experiment.11 Even if participants paid a tax of 30 percent, they still kept 70 percent.12 In Figure 3.1, the first bar demonstrates that the average compliance rate in our experiment, across all countries in Round 1 (no-­redistribution) was 51 percent. When we introduce a redistributive payback from the general fund (Bar 2), then compliance increases to 61 percent. When the general fund is doubled (Bar 3), we see a 23-­percentage point increase from the baseline round in the average amount contributed (73 percent). This result is con­sist­ ent across countries. Out of all the institutional variables that we looked at in our experiment, this double-­pot variable had the strongest effect on compliance. This is consistent with other studies. For example, in Alm et al. (1992a), compliance rose by approximately 30 percent when the money was placed in a group fund, which is quite similar to our data. Gërxhani and Schram (2006) reported that when no public good was presented, Albanian students reported zero income in 75 percent of their decisions, but when their money was

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

44  Willing to Pay? placed into a public fund, as in a public goods game scenario, the subjects reported at least some income in 75 percent of their decisions. We interpret these outcomes to support the “self-­interest” hypothesis, but not in the narrow “individual” way. Again, the penalties for evasion are the same across the three rounds of our experiment; in other words, the incentives for cheating the group are constant across rounds. So, why do our subjects contribute so much more when we double the money redistributed from the public purse? The answer is clearly that the collective benefit is larger. If I  pay taxes when there are no direct benefits to either myself or my group, I  only pay because I am extremely risk-­averse, or because I feel it is moral obligation to pay my taxes (see section on Values below). But when there is a collective benefit derived from my contribution, there is now a social benefit. As a social animal, the social benefit has value to me. Recall that as all decisions are private, no one can know how anyone else in the experiment has acted. In short, no one will know if I contributed 100 percent of the “tax due” or if I cheated 100 percent (technically, there is a marginal increase in the size of the purse when I contribute, but when divided by the group, this amounts to significantly less than the amount I “sacrifice” by being honest and reporting all my earnings). In sum, our data strongly support the idea that humans can be driven by their self-­interest. But that self-­interest must be understood in broader terms than our individual or autonomous, self-­interest. We disaggregate these data in a number of ways below, but the next obvious question is whether there are significant differences between countries. Are people in some countries more honest than others?13 Quite frankly, we were very surprised by our findings here. We fully expected Italy to be the least compliant of the four countries examined—both because the actual Tax Gap is significantly larger in Italy than in other countries, but also because of the common academic narrative suggesting that Italy suffers from “amoral familialism.” In other words, according to Banfield (1958, 9–10), Southern Italian backwardness can be explained “largely but not entirely” by “the inability of the villagers to act together for their common good or, indeed, for any end transcending the immediate, material interest of the nuclear family.” Figure 3.2 displays the average compliance rate in each round for each country. As can be seen in Figure 3.2, Italians did contribute to the common purse less than Swedes or Americans, but it was our British subjects that contributed the least. This was true for every redistributive round.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  45 R1: No Redistribution

R2: Redistribution

R3: Redistribution x 2

80%

0.68

Tax Compiance Rate

60% 50%

0.8

0.77

70%

0.64

0.64

0.76

0.65

0.6 0.53

0.51

0.51

40%

0.42 30% 20% 10% 0% Italy U.K. U.S. Sweden

Italy U.K. U.S. Sweden

Italy U.K. U.S. Sweden

Figure 3.2  Average Compliance Rate in Rounds 1–3 by Country

Figure  3.2 is fascinating at multiple levels. Once again, contrary to our expectations, it was our British subjects, and not the Italians, who contributed the least on average. We can see, for example, that in Round 1 (no redistribution), the average compliance rate for British subjects was 42 percent whereas for Americans it was 60 percent. In the “double pot,” Round 3, the average compliance rate for British subjects was around 65 percent, whereas subjects from every other country reported over 75 percent of their incomes for tax purposes. Secondly, while contributions increased for all countries, as redistribution increased, subjects in each country contributed more, but the effects of increasing the redistributed amounts differed across countries. In short, redistribution appeared to affect responses somewhat differently in each country. We address each of these puzzles in turn (see Appendix 3, Table A.1, Model 1 for our regressions, and Figure A.1 for predictive margins).

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

46  Willing to Pay?

Perfidious Albion It is surprising that British subjects are so much less willing to contribute to the public good. Therefore, we begin with the British puzzle. Recall, that the Tax Gap in Britain is quite low when compared to the American and Italian. Therefore, it would have been reasonable to expect that tax compliance in our experiment would reveal that Brits are more compliant than others. Clearly, as Figure 3.2 demonstrates, this is not the case. Why? To say the least, this is a complicated and controversial question. Indeed, the first time we ran these experiments in the UK, we were skeptical of our own results, so much so that we insisted that we run them over multiple times in multiple locations. We found the results consistent across locations and across the time span of our study. We cannot hope to fully explain these results. However, we do believe we are able to offer some insights. First, we should recognize that there are two sides to our puzzle: Why was compliance so low in England, but also why is it relatively higher in Italy? We were surprised by the low compliance rate in Britain in our experiments because the results appeared to be so different from the compliance rate of British citizens to other national tax authorities. To be fully honest, we were also surprised as we expected Italians to be less cooperative or compliant than they turned out to be in our experiment. This is because we knew that the Italian compliance rate in the “real world” is so much lower than it is in the other countries. So, the question for us became, what is going on here? In our dozens of interviews with tax authorities and British citizens, along with our examination of British physical history and institutions,14 we concluded that Britain’s relatively small tax gap has more to do with the efficiency of the national tax authorities and administration on the one hand, and the structure of the British economy on the other. In comparison with the Italian case, British tax authorities are remarkably efficient and effective. Moreover, and this is equally important, British employers deduct taxes directly from the paychecks of the vast majority of their people. Compare this to the Italian case in which a large proportion of workers are self-­employed or work for a small business. Table 3.1 shows the percentage of employment by the size of the enterprise. As the reader can easily see, nearly half of Italian workers work for companies of fewer than ten employees. In these cases, it is quite common for employers to under-­report their employees’ salaries and taxes. In fact, it is remarkably common for self-­ employed people (like doctors and dentists, for example) to offer their customers two prices—one if you pay in cash, another higher price, if you want a receipt. In fact, in Italy, cash usage as a percent of total purchases is much

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  47 Table 3.1  Employment by Enterprise Size Percentage of all persons employed, 2014, or latest available year COUNTRY

1–9

10–19

20–49

TOTAL LESS THAN 50

Italy UK US Sweden

45.9 17.9 10.2 25.6

10.9 8.4 6.7 9.6

9.6 10.9 9.8 12.5

66.4 37.2 26.7 47.7

Note i: Data were provided by OECD (2017).

higher than any other country in our study. Successive Italian governments have attempted to reduce cash transactions in the economy, but at the time of our study, a remarkable, 86 percent of all transactions in Italy are made in cash compared to only 20 percent of all in Sweden. The US (32 percent) and UK (46 percent) are also comparatively low (G4S  2018). In other words, there are few opportunities for tax evasion in Britain . . . or for that matter in Sweden—but many more opportunities for evasion in Italy. When presenting these data to British fiscal experts, we found that they were not surprised. In several interviews and presentations made by Steinmo to members of Her Majesty’s Revenue & Customs (HMRC), its members repeatedly told him that these findings were completely in-­line with their expectations and understandings. British taxpayers, apparently, are often inclined to try and cheat the tax authorities when provided the opportunity. The point, though, is that British taxpayers rarely have the opportunity. Quite the opposite is true in Italy. Out of the overall Italian workforce, 45.9 percent of Italians work for companies with fewer than ten employees and many of these companies are primarily cash businesses. It is near universally known in Italy that this is because small businesses generally under-­report their incomes for tax purposes. Indeed, in Italy, it is commonly understood that if many small businesses actually paid their full tax burden, they would go bankrupt. Moreover, because so many Italians evade their taxes it is extremely difficult for tax authorities to catch evaders. In a system like the British, or the Swedish, where tax compliance is very high, it is much easier to discover anomalies in the payments and deductions in the system. Simply put, where taxpayers have greater opportunity to evade due to the nature of their work, they evade more, and similarly, they pay less where the government’s “institutional capacity” to effectively collect revenues is limited (see Kleven et al. 2011).15 These large structural factors can outweigh—and indeed as we argue below—even help explain some of the cultural differences we so easily observe on the surface when comparing nations and peoples.16

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

48  Willing to Pay?

The Social Impulse The data in Figure 3.2 demonstrate that virtually everyone is more likely to contribute to a public good/service when there is more public good/service to be had. We set up this part of our experiment to test people’s responses to different kinds of “welfare states.” To the most cynical, our “no redistribution” (Round 1) represents a system in which the taxpayer is asked to contribute but expects nothing in return. Round 2 represents a case where the taxpayer gets back something equal to what their community has contributed. Round 3 represents what might be thought of as a positive welfare state (we sometimes referred to Round 3 as the Swedish state). In this case, the group gets back two times more than they collectively contributed. As we expected, people’s willingness to pay increased quite substantially as the amount they would receive back collectively was increased. This was true everywhere. We see this as an example of the “fiscal contract.” Still, it is interesting to note that citizens in different countries responded differently to these incentives. Americans, for example, were the most compliant in Round 1, reporting 60 percent of their income, on average, even when they received nothing in return. When the pot was doubled, Americans increased their contribution to 76 percent, which is a modest increase of 16 percentage points. Swedes, in contrast, are less compliant than Americans when they saw no redistribution but increased their contributions to 80 percent under the 2× system. In other words, this treatment produced nearly twice the effect on Swedes as it did on Americans.17 Why? One possible explanation for the ways that subjects in different countries responded differently to these treatments is that individuals in some countries are somehow less “rational” than others. One can consider Round 1’s “no redistribution” as a kind of “baseline,” indicating citizens’ basic willingness to comply with rules or directives. Following this logic, one could say that Americans are simply less individually rational, and more rule compliant than other people irrespective of the incentives or disincentives they face. There is anecdotal evidence to support this proposition. But does it make sense? Why, then, does compliance increase more for Swedes than it does Americans? Are Swedes more, or less, “rational” than Americans? The reader will not be surprised to learn that this is not our explanation. Instead, we infer that Swedes are more likely to understand the positive and iterative effects of high taxes and redistribution. Swedes, after all, live in a society that has a strong reputation for its high taxes, redistributive welfare state, and positive egalitarian outcomes. Our survey evidence supports this thesis. Swedes, in effect, “get it” when it comes to the relationship between taxes and social benefits.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  49 In our survey, we asked subjects if they preferred that incomes be made more equal (1), or if we need larger income differences as incentives for individual effort (10). Our American subjects had significantly higher responses, on average than subjects from the other countries, indicating that they were more comfortable with greater inequality. On the other hand, Swedes had the lowest, meaning they supported higher levels of equality. When confronted with the question of whether they were willing to pay higher taxes if everyone received more benefits, Swedes more easily saw the collective good in this idea. Americans have far less experience with, or confidence in, the positive effects of high taxes and/or the redistributive welfare state. Following this logic, we would expect British and Italian citizens to fall somewhere between the American and Swedish extremes; and this is exactly what we observed. This treatment had nearly the same positive effect on British (an increase of 23 percentage points) and Italian subjects (an increase of 24 percentage points).18 In short, the sensitivity to redistribution, and how much people are willing to pay for that redistribution, is connected to people’s experience with redistribution and the state in society.19 As Pampel et al. (2019, 1359) argue: institutional rules and individual attitudes may reinforce one another. Isolating the separate influence of institutions and attitudes in some ways may be artificial, and tests for their combined influence may be more realistic. The tests suggest that combining redistributive institutional conditions and positive attitudes makes for particularly high compliance.

Do Tax Rates Matter? The reader will recall from Chapter 2 that the tax compliance section of our experiment was divided into three main stages. In the first stage, we attempted to test how subjects would respond to different redistributive regimes. In the next stage of our experiment, we wanted to understand how different tax rates would affect people’s willingness to contribute to the public purse. Surprisingly, there are few studies that have investigated this question. Of course, there exists much economic theory that suggests citizens should be highly responsive to tax rates. But there is remarkably little empirical data that supports this theory. Moreover, most economic theory which discusses the disincentive effects of tax rates fundamentally ignores the fact that citizens also get something back from their taxes. Allingham and Sandmo’s (1972) famous and well-­cited

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

50  Willing to Pay? early work on tax compliance, for example, predicts that increasing the tax rate will simultaneously decrease tax compliance—but totally ignores the question of what happens to that tax money.20 There are some studies that find positive, mixed, or null results (Alm et al. 1995a; Beck et al. 1991; Becker et al.  1987), but overall, very few economists and tax policymakers have brought this basic point into their models or understanding of tax policy. At best, as Andreoni et al. (1998, 839) observe, “the effect of tax rates on evasion remains unclear” and “given the importance of this topic, it surely deserves further investigation.” In an abstracted world, in which you only pay taxes but receive nothing in return (as in Round 1), it would be quite reasonable to assume that if the tax rate is increased, taxpayers will reduce their contributions. But of course, this is not how a real tax system works. When people pay taxes, they receive something in return. And as far as we know, the connection between the disincentive effects of increased tax rates has never been tested across different countries when citizens get something in return for their taxes. This is what we attempted to do in Stage 2 of our experiment. In Figure 3.3, we look at the second stage of our experiment in which we vary the tax rates. The tax rates in these rounds were 10 percent, 30 percent, and 50 percent.

70% 60%

0.7 0.64

Tax Compliance Rate

50%

0.58

40% 30% 20% 10% 0% R4: 10% Tax Rate

R5: 30% Tax Rate

Figure 3.3  Average Compliance Rate in Rounds 4–6

R6: 50% Tax Rate

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  51 As expected, as we increase tax rates, overall compliance declines. However, the decline in compliance advancing from a 10 percent to a 50 percent tax rate was surprisingly small.21 This suggests, to us at least, that the “supply-­side” economic theory, which suggests that the state will collect more money if it drops tax rates significantly, is wrong. Indeed, in our experiments, raising the tax rate from 10 percent to 50 percent lowered average tax compliance by only 12 percentage points. Interestingly, this is a much smaller change than what we observed in Stage 1 of our experiment when moving from no redistribution to 2× redistribution.22 Figure 3.4 reports the average compliance rate in Stage 2 with results broken down by country. Once again, we find that the effect of increasing the tax rate differs for each country. While reported incomes decline in all cases, they decline substantially more in Italy—a 24-­percentage point decrease in compliance—compared to the other countries. On the other hand, tax rates seem to have a much smaller effect on our American and British compliance. We found these results surprising as well, especially since both the UK and US R4: 10% Tax Rate

R5: 30% Tax Rate

R6: 50% Tax Rate

80% 70%

0.79

0.76 0.71

Tax Compliance Rate

60%

0.63

0.69 0.69 0.63

0.62 0.57

0.55

0.61

0.53

40% 30% 20% 10% 0% Italy U.K. U.S. Sweden

Italy U.K. U.S. Sweden

Italy U.K. U.S. Sweden

Figure 3.4  Average Compliance Rate for Rounds 4–6 by Country

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

52  Willing to Pay? have relatively low tax burdens for OECD countries and a higher quality of government, according to the Quality of Government Index (see Charron et al. 2019 for survey measures). There does seem to be a relationship between the tax burden and quality of government in one’s home country and how sensitive our subjects are to tax increases in the lab (see Appendix 3, Table A.1, Model 2 for our regressions and Figure A.2 for predictive margins).

Does Progressivity Matter? Most advanced democracies have progressive tax systems. Therefore, we also wanted to test how or whether progressivity would affect taxpayers’ willingness to pay. There is significant literature about fairness (Fehr 2006; Kastlunger et al. 2011; Hayek and Shenfield 1983) suggesting that people have a general preference for fairness. People are also averse to inequality (Bolton and Ockenfels 2000; Fehr and Schmidt 1999), especially when inequality is visible. In many cases, the concern for fairness is greater than the concern for efficiency (Fehr and Schmidt 1999). Indeed, fairness lies at the heart of most tax debates. It is important to note that there were significant differences in typing abilities across subjects in our experiments. Some participants were significantly more adept than others at the clerical task, and therefore earned much more than others. We subsequently discovered that our subjects were actually fairly good at predicting whether they were high or low earners (at least by the third stage of our experiment)—even though no one’s income was ever reported to anyone else. Whether a tax regime is flat or progressive, obviously affects how much income a person earns after taxes and how much inequality there is in a society. There is but one study (of which we know) that examines how taxpayers respond to different tax regimes. Heinemann et al. (2013) exposed participants to a progressive tax system and a flat (proportionate) rate system; uncovering that evasion was higher under the proportional tax system. That said, they found that moving from a progressive tax system to a proportional tax system had a more positive effect on compliance than the reverse. For us, it turned out that it was considerably more difficult to implement progressivity in the lab. This is because individuals generally do not understand how progressive income tax systems work.23 In this stage of our experiment our subjects once again typed random letters and received their

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  53 earnings. They were then presented with two different progressive tax systems (Round 7 and 8).24 In Round 7, those who reported the highest incomes (the top 10 percent in session) paid a 50 percent tax rate on that income. Those who had the lowest reported income paid only 10 percent on that income. Everyone else paid a 30 percent rate. (No real tax system works like this, but this is actually how many people think a “progressive” tax system works.)25 In Round 8 we created a tax structure somewhat more akin to what we find in most countries. In this case, the tax rate increases as incomes rise, but only the incomes over specific thresholds bear the increased tax rate. The first 50 ECU reported was taxed at 10 percent. Reported incomes between 51 and 100 faced a 30 percent tax rate. Reported incomes over 100 ECU were taxed at 50 percent. Our expectation was that subjects would care about fairness and would be more willing to pay taxes when taxes were more progressive.26 Figure 3.5 illustrates the average compliance rate in Rounds 7 and 8. As you can see, the type of progressive regime imposed had very little overall impact on citizens’ willingness to pay. On average, around 62 percent of income was reported in both rounds seven and eight.

60% 0.62

0.62

R7: Progressive 1

R8: Progressive 2

Tax Compliance Rate

50%

40%

30%

20%

10%

0%

Figure 3.5  Average Compliance Rate in Rounds 7 and 8

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

54  Willing to Pay? R7: Progressive 1

R8: Progressive 2

70%

0.67

60%

0.65

0.65

0.68

0.6

Tax Compiance Rate

50%

0.57

0.57

Italy

U.K.

0.55

40%

30%

20%

10%

0% U.S.

Sweden

Italy

U.K.

U.S.

Sweden

Figure 3.6  Average Compliance Rate in Round 7 and 8 by Country

If we look at the effects of progressivity by country in Figure 3.6, we see that our results are consistent with the findings from other parts of the study. For example, British subjects report a lower percentage of their incomes than anyone else, and Americans and Swedes report the most on average. Interestingly, it appears that our subjects reported approximately the same incomes in the two progressive tax rounds as they had in the 30 percent round (#5). It is interesting to note that while the average amounts contributed to the public purse are similar between the 30 percent flat tax and the progressive rounds, our subjects’ behavior was clearly shaped by the structure of the tax system. Figure  3.7 is a histogram of the distribution of reported income in Round 8 with the number of subjects displayed on the Y-axis. As Figure 3.7 shows, 21 percent of our subjects reported zero income, 13 percent reported incomes just below 50 ECU, and 13 percent of our subjects reported just under 100 ECU. There is clearly bunching just below each threshold. A significant percentage of our subjects chose to report their income in such a way

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  55

25%

Percentage of participants

20%

15%

10%

5%

0% 0

30

60

90

120 150 180 Reported income

10% Rate

30% Rate

210

240

270

50% Rate

Figure 3.7  Histogram of Reported Income in Round 8

that they did not enter the next tax bracket. In other words, even people who were mostly honest, chose to “fudge” a little bit and these threshold tax rates offered them the markers where they chose to report. Evidence from real world tax returns indicates that this kind of behavior occurs in real world tax systems as well (Le Maire and Schjerning 2013; Saez 2010) (see Appendix 3, Table A.2, model 3 for our regressions and Figure A.3 for predictive margins). Finally, if we stratify our sample by three income groups based on how much they earned from the typing task, we find that when high income earners faced a 50 percent tax rate they were much more likely to lie about their earnings (Figure  3.8). The average compliance rate for high earners was 57 percent whereas low earners (slow typists) reported an average of 71 percent of their incomes. In other words, in a progressive tax system, high income earners tended to cheat on their taxes more so than low earners for the simple self-­interested reason that their payoff for doing so would be greater (see Table A.3, Model 1 in Appendix 3 for the regressions).

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

56  Willing to Pay? 70% 0.71 60%

0.65 0.57

Tax Compliance Rate

50% 40% 30% 20% 10% 0% 1

2 Poor

Middle

3 Rich

Figure 3.8  Average Compliance Rate by Income Group

To sum up this section: it is clear that self-­interest is an important driver for individual behavior. Our experiments demonstrate that when the incentives for contributing to a public good increase (e.g. when you get more back for your tax contribution), taxpayers increase the amount they put into the public purse in every country. It is also clear, that when people have an opportunity to get away without paying into a public good, some people do not contribute as much as others. In other words, it appears that some people are more socially responsible than others. Some individuals in every country are more self-­centered and more willing to cheat others when it is in their individual self-­interest. The data also indicate that each society has some mix of each of these types of individuals. In each country we studied, some individuals cheated 100 percent of the time and some individuals reported 100 percent of their income, 100 percent of the time. But it appears that the distribution of individually oriented or socially oriented individuals differs across societies (we will return to this issue below). Finally, the distribution of different types of individuals is correlated with the different incentives or disincentives they face in different countries.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  57

Norms and Social Behavior People pay their taxes in part because they are afraid of punishment from the state. But, as a large body of research has shown, other non-­economic factors, such as social norms shape tax compliance as well.27 In this section, we explain why social norms are central to tax compliance decisions. In other words, in addition to the incentives that rational people face, tax compliance— like other forms of cooperative social behavior—is strongly influenced by norms of our community. Social and evolutionary psychologists have long known that humans are social creatures who want to be part of groups, and want to fit into that/those groups. This is why norms matter. As we show below, most people are more willing to contribute to a public good (by paying taxes) if they believe that others are paying as well. Conversely, more people will cheat (or refuse to contribute) when they believe that others are cheating. In short, most humans frame their own actions in the context of others’ behavior. Why? Because as social creatures, we desire to do what is expected of us and act according to the norms of the society/group that we are in. “The motivation to form and sustain at least a minimum amount of social connections is one of the most powerful, universal and influential human drives” (Baumeister  2011, 121; Baumeister and Leary 1995). We don’t join groups (just) to get more for ourselves. Instead, we join in groups because it is basic to our human nature. We want to be part of something. This is why rejection and social exclusion have such profound effects on us and can even trigger aggressive behavior, emotional numbness, and strong hormonal responses (Baumeister 2011).28 Even our individual identity is built in the context of social groups; we are not autonomous actors abstracted outside of our social context. We desire to fit into a group, and we prefer to associate with others whom we share some common attributes, behaviors, or attitudes and values. This implies that along with our desire to be part of a group, we also have a basic desire to conform to that group’s rules and norms. The psychological research on this point is overwhelming. We adapt to and abide by “the rules” governing our society in part due to the incentives and constraints that confront us, but also because of a deep-­seated preference for conformity itself (Janes and Olson 2000; Mead et al. 2011). Indeed, violations of conformity can cause physical pain (Kross et al. 2011). This is what psychologists call, Social Identity Theory (Tajfel 1982; Turner 1982). This group-­related self-­definition in turn prompts us to behave according to the norms of the respective groups.29

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

58  Willing to Pay? Though more complex than a simple theory which claims that everyone pursues their individual self-­interest, a theory that acknowledges that we are also motivated by a desire to be part of a group is more in line with the empirical facts. It is also common sense. Ask yourself—are you really a purely autonomous actor who only does what is in your individual self-­interest, or do you also like being a part of something? Do you really calculate your actions according to maximizing what is good for you alone, or do you sometimes also follow along with the group norms? Most of us enjoy being part of a group, doing things with the group, and even behaving and dressing like other members of the group. Whether we are Americans, French, or Swedish; Chelsea fans or Manchester United supporters, Democrats or Republicans, we often adapt our behaviors and even attitudes in the direction of others in that group. We want to “fit in.” In short, our decisions are often led by the desire to conform to the tastes and beliefs of others (Cialdini and Goldstein 2004; Elster 1993). This fundamental need to be part of society has significant implications for our understanding of rule compliance, especially with regard to tax compliance. In some countries, paying your taxes is considered a social responsibility, while in others it is normal and expected to avoid as much as you can. The differences in behavior in these cases, is not due to differences in the citizens’ desire to follow rules or be part of a community; they are instead due to differences in the particular social rules and norms that govern the society at hand. It may be easy to believe that norms or social rules are stronger in some societies than they are in others. But there is no evidence for this. We should instead see that varied rules and norms are enforced in different societies. When a tourist or a visitor arrives in a foreign land, they often immediately notice that the norms of behavior they expect are not enforced. This can then lead them to interpret this to mean that there “are no rules.” In fact, we suggest that what can appear to be anti-­social behavior (like ignoring stop signs, throwing trash out the window, or not paying your taxes) should be understood as examples of rule-­following behavior when those are the common “rules” followed in that society. Ironically, then, anti-­ social behavior can be a product of the desire to be like everyone else. Social norms consist of “prescribed behaviors shared within a community” (Andrighetto et al. 2013). They provide information to individuals about what ought to be done and are enforced through informal sanctions, like social exclusion, negative gossip, shame, or disapproval. Norms are so indispensable to social life that they are defined as the “cement” (Elster 1989) or the “grammar” of society (Bicchieri 2006). Formal institutions (legal rules, or laws) are important for shaping and structuring behavior because they structure incentives. But

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  59 external authorities, like the state, do not create nor impose norms. Social norms emerge from reiterated interactions and in particular from peer pressure. This is why social norms are so important to tax compliance; they provide people with cues and expectations about what they should and shouldn’t do. Just as formal institutions provide formal sanctions, informal social sanctions following social norm violations can provide just enough pressure to elicit compliance. A simple eye-­roll might suffice to provide enough weight to provoke compliance to the norm. However, the evidence clearly shows that people comply with social norms even in situations where there is complete anonymity and where social punishment is not an option (Fehr and Fischbacher 2004). In general, we choose to follow formal rules if we believe that others are doing the same. This is sometimes called conditional cooperation (Fischbacher et al. 2001). That is, I pay my taxes if I know others are paying their taxes. Frey and Torgler (2007) used the European Values Survey from the years 1999 and 2000 to explore conditional cooperation and taxation. The authors reported a high correlation between tax morale (justifiability of tax evasion) and a perceived level of tax evasion. Bergman (2009) tested the effect of social contagion on tax compliance in a laboratory experiment. The experiment suggested that when participants received information that tax evasion is widespread in the environment, they significantly decreased their tax compliance; when the message reported a high level of compliance, the effect in increasing compliance was much weaker. Similarly, Bazart and Bonein (2014) found evidence for reciprocal tax behavior in a laboratory setting. They showed that participants tended to adjust their reported income according to the information about their peers’ behavior, which they obtained in previous rounds.30 As Bergman (2009, 42) puts it, “the institutionalization of norms depends on the predictability of sanctions: the higher the expectation that sanctions will be imposed, the more institutionalized the norm. Informal sanctions are effective to the extent that prestige and close-­knit relations are important within a community.” Bergman goes on to say “[When there is] voluntary compliance, [the] norms efficiently resolve collective problems where the perceived benefits of compliance exceed the perceived costs” (Bergman 2009, 44). There is even evidence demonstrating that people are willing to sacrifice their own self-­interest (incur a cost) to enforce norms and punish free riders (Fehr and Gächter 2000; Andrighetto et al. 2013; Herrmann et al. 2008). Voluntary compliance can then relieve tax administrations of much of the burden related to tax enforcement. The most obvious distinction amongst our cases is between Italy and Sweden. Undoubtedly, the job of tax collector is much easier in Sweden than it is in Italy, as evinced by the popularity of the

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

60  Willing to Pay? 80% 70%

0.82

Tax Compliance Rate

60%

0.67

50% 40% 0.45

30% 20% 10% 0% All

Some Think Others Report

Little

Figure 3.9  Average Compliance Rate by Perception of Other’s Reporting Behavior

Skatteverket (Swedish Revenue Agency). In fact, the Swedish tax authority (Skatteverket) is the most popular public bureaucracy in Sweden . . . and Swedes tend to love their state generally.31 In our experiments we tested the impact of norms on behavior by asking the simple question: “Do you think most of the participants in the experiment reported their total earnings (All), less than their total earnings (Some), or much less than their total earnings (Little) for tax purposes?” Figure 3.9 represents the average compliance rate for each level of our “norms” variable. As Figure 3.9 suggests, those individuals who believed that most other people in the experimental session reported all of their earnings, reported almost twice as much of their individual earnings (82 percent) compared to those who believed that most people in the session reported little of their earnings (45 percent). Beliefs-or expectations of how others would behave-were far from con­sist­ ent across countries, however. Figure 3.10 shows the means of each category for those who think others report all, some, or little of their income by country. As Figure 3.10 shows, there were very significant differences in the expectations of how others would behave in our different laboratory sessions in different countries.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  61

0.7 0.6

0.7

0.6

Percentage of respondents

0.6

0.4

0.3 0.3 0.2

0.2 0.2

0.2

0.2

0.1

0.1 0.0 Italy

UK

US

Sweden

Perceived Amount of Total Earnings Reported by Fellow Participants All

Some

Little

Figure 3.10  Perception of Others’ Reporting Behavior by Country Note: Due to rounding, not all numbers will add to 100%

In the graph, the first bar shows that fewer than 10 percent of Italian subjects believe that others reported their total income. The second bar demonstrates that 60 percent of Italian subjects believe others reported some of their income and the third bar shows over 30 percent of Italian subjects believe that most others reported little income. This is fascinating because norms have a similar effect on tax compliance in each country; the more you think others will report the more you are likely to report. However, the big difference—and what explains the big variation between countries—is that the distribution of “perceived amount of total earnings reported by fellow participants” varies greatly by country. In Sweden, for example, where tax compliance is both the highest in the real world and in our experiments, roughly 10 percent of participants believe that most other participants will report little income. This is a full three times lower than in Italy. On the other hand, less than 10 percent of Italians believe that most others will report all of their income, while in Sweden that number is over 20 percent (see Table A.2, Model 2 in Appendix 3 for regressions).

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

62  Willing to Pay?

The Tragedy of Economics Every society is made up of many different types of individuals, groups, and sub-­cultures. It would be reasonable to expect, then, that some groups would tend to be more “self-­interested” than others. When breaking down our data to explore how different educational backgrounds might affect behavior we found what many might have expected: economists are more selfish than any other group we studied (Figure 3.11). There is a limited body of extant literature on why economists behave ­differently, but several studies have found that they do. Marwell and Ames (1981), for example, demonstrated that economists were more likely to contribute less in public goods games than non-­economists. Another study by Carter and Irons (1991), in which they used an ultimatum game, uncovered that economists were less willing to share than non-­economists. An ultimatum game is a simple experiment in which a subject is given an endowment and paired with

80%

0.72

Tax Compliance Rate

60%

50%

40%

0.79

0.76

70%

0.67

0.64

0.62

0.65

0.61

0.53

0.6 0.57

0.47

0.66

0.64

0.5 0.45

0.46

0.42

30%

20%

10%

0% R1

R2

R3

R4 R5 R6 Experimental rounds Other Major

R7

R8

Economics Major

Figure 3.11  Average Compliance Rate by Major in Each Round

R9

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  63 another subject. He/she is then told to allocate a certain amount of their endowment to the random partner who can either accept or decline the amount. The optimum decision is, of course, to keep something like 99% of the endowment and allocate something like one percent to the partner. The partner should always accept anything over zero since it is a one-­time situation, and 1% is better than 0. Although the optimally rational decision is to keep almost all of the money for yourself and give the partner just enough to give them an incentive to accept the offer, most allocations which are agreed to are closer to a 50–50 split. Carter and Irons however, demonstrated that economists allocate significantly less than non-­economists. In another example of the economist effect, Rubinstein (2006) asked students to imagine themselves as Vice Presidents of a company who had to either maximize profits by laying off workers or decrease profits by maintaining current levels of employment; in that study, economists chose to fire 20 percent more workers than non-­economists. Under each of our experimental scenarios, and in every country, we found that economists cheat more than any other identifiable group (Figure 3.12). Whether they select into economic majors because they are more selfish or 70%

60%

0.7

0.69 0.6

0.69

0.61

0.6

Tax Compliance Rate

50% 0.52 40% 0.37

30%

20%

10%

0% Italy

UK

US

Sweden

Experimental rounds Other Major

Economics Major

Figure 3.12  Average Compliance Rate by Major and Country

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

64  Willing to Pay? driven by money or if they socialized once they become economics majors is still up for debate. However, our data demonstrate that economics majors free ride more than non-­economics majors. Another way to look at this would be to examine the extent to which economists evade. As we have mentioned before, a rational decision would always report 0. We would expect individuals who have been trained that everyone is a “utility maximizer” would be more likely to cheat the group than those who have not been taught that everyone is selfish. We can therefore examine the percent of economists and non-­economists who report zero in each round. Indeed, looking at Figure 3.13 we can observe that the number of economists who report zero (cheated 100 percent) is considerably higher in every round. Of course, we cannot fully know what kind of economics training every individual has had, but we can see that there is a strong correlation between 0.8 0.76

Percent Reporting 0 Income

0.6

0.68

0.65

0.62

0.53

0.5

0.5 0.5

0.49

0.72

0.7

0.69

0.55 0.51 0.47

0.4 0.39 0.35 0.3

0.2

0.0 R1

R2

R3

R5 R6 R4 Experimental rounds Other Major

Figure 3.13  Percent who Report 0 by Major

R7

R8

Economics Major

R9

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  65 economics training and our subjects’ expectation of how much others will contribute to the public good. As Figure 3.14 shows, economists are significantly more likely to believe that “most” or “many” will under-­report their income. Figure 3.14 illustrates that about 17 percent of non-­economics majors thought that others would report all their income, while only 5 percent of economists thought that others would report all of their income. On the other hand, 31 percent of economists thought that others would report little income, whereas 21 percent of non-­economics majors thought the same. In sum, norms provide people with cues about how they should and/or should not behave. They help people maintain cognitive consistency while at the same time engendering a sense of belonging to the social group. On the one hand, social norms render tax compliance a context-­dependent decision, meaning that the environment in which we live has a large effect on our taxpaying decision. On the other hand, how we internalize these norms can also have lasting consequences on a wide range of socially relevant decisions.

Percentage of respondents

0.6

0.63

0.63

0.4

0.31 0.2

0.21 0.17

0.05

0.0 Other Major

Economics Major

Perceived Amount of Total Earnings Reported by Fellow Participants All Some Little

Figure 3.14  Perceived Earnings by Major

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

66  Willing to Pay? We think of these internalized norms as social values, and we consider their implications for tax compliance below.

Norms and the “Theories in our Heads” It is of course possible that people respond to local norms because they fear social disapproval, or perhaps they wish to enhance their social standing as “reliable” participants/partners. But many socially responsible behaviors, like tax compliance, are done in private. Why would people abide by a “norm” if no one can see them doing so? We argue that in order to understand why we behave according to social norms and/or follow rules even when no one is looking, we need to appreciate two specific features of the human mind: (1) our need and ability to build theories to help us explain the world around us and, (2) our need for cognitive consistency. The act of following rules—or even being able to know how general rules (a.k.a. institutions) apply to a particular situation—is an act of theory building. The human brain has a remarkable ability to build such theories, or what Walter Lippmann called “pictures in our heads” (Lippmann  1922). Indeed, perhaps the most unique feature of human cognition is this remarkable ability to apply the general to the specific and to be able to see how a specific case fits into a pattern that can be generalized. The point here is that not only can human beings build theories that help them make sense of the world around them but that they also want to make sense of the world around them. Our mental models, or theories, are not simply efficiency devices that help make decisions easier. They become preferences. We become committed to these theories—even when they are wrong. A very large body of psychological and cognitive research demonstrates that humans have strong preferences for cognitive consistency. Indeed, acting in ways that violate our mental norms create a phenomenon called “cognitive dissonance,” and can create a host of emotional, hormonal, and even physical maladies (Abelson and Bernstein 1963; Festinger  1957). Why? Because that conflicts with our models about how the world works and how “I” should act undermines our “strong need to enhance their self-­concepts by behaving consistently with their actions, statements, commitments, beliefs, and self-­ ascribed traits” (Cialdini and Goldstein 2004; Cialdini and Trost 1998). It is clear that compliance behavior is partially explained by self-­interest and our preference to follow and adapt to group norms—even when the likelihood of being caught and/or costs of evasion are low. But absent an

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  67 appreciation of our need for cognitive consistency, these variables alone are insufficient. After all, for the most part, we pay our taxes privately. It would be perfectly possible to be part of a community that diligently pays its taxes, but still cheat when you yourself fill out your tax form all by yourself. So, why do we (or most of us) do it? Why do we pay our taxes if we feel we can get away with part of our income? Some scholars have argued that we behave according to group norms only because we get more for ourselves from being part of the group. Surely, we are able to achieve more when working together, and therefore, maybe our group behavior is simply a way of maximizing our self-­ interest in the long run (Ellemers et al. 2004; Hornsey et al. 2006; Kaplan and Miller  1987; Thibaut and Kelley  1959). But we believe this explanation is insufficient. The marginal benefit we each receive for paying our individual taxes in a society with millions of taxpayers is so small that it would indeed be completely “irrational” to pay our individual share in the hope that this would increase the general social good. We argue, instead, that to not contribute while most others are paying would be inconsistent with our desire to follow group norms. This basic drive for consistency has several implications for why we follow rules generally, and ultimately, how and why complex societies are even possible. First, due to our need for accuracy and consistency, we adapt our behavior as well as our attitudes to the institutions surrounding us. As we saw above, most human beings want to be part of a group. Indeed, their very identity, is built in the context of their group (or groups). Quite logically, then, if I am part of a group, I should want to behave, look, and even act like members of that group. When we consider choices such as whether we should drive above the speed limit or declare all our income to the taxman, we naturally look around us for references to what we are “supposed to do.” Even when we are outside that social context, perhaps we are alone (or on vacation outside away from home), we will still behave according to the rules/norms of the group with which we identify. This is because acting differently would be in­con­sist­ ent with our understandings of who we are, and this behavior is con­sist­ent with this identity. As John Rawls and Erin Kelly (2001, 30) argued, “Many of our most serious conflicts are conflicts within ourselves. Those who suppose their judgments are always consistent are unreflective or dogmatic.” Consider the curious act of tipping in a restaurant. Most people tip according to the norms and customs of their local community—even when they are away from home. This is completely “irrational” of course. Surely it may be in your self-­interest to tip at local restaurants since you may return there and hope for good service and/or you want to demonstrate your generosity to

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

68  Willing to Pay? others who may observe your behavior. But why should we tip if we are at a café to which you will never return and where no one who is there will ever see you again? The answer is simply that to not tip at this random café would be inconsistent with what you normally do and what you normally expected of yourself. To act one way while traveling across the country and another way while at home would create cognitive dissonance (Festinger 1957). James March and Johan Olsen’s term “Logic of Appropriateness” (1989) captures many of these insights. March and Olsen argue that in addition to the “Logic of Consequences,” humans operate according to what they feel appropriate. But how does one know what “appropriate” behavior is? Certainly, we can and do observe others’ behavior. But if all we did was follow the crowd or masses, we would not know how to behave in novel situations. Due to our developed cognitive capacities and our desire for accuracy and consistency, even when we are totally alone and unobserved, most of us will abide by certain norms and rules. Why? Because to behave differently would violate our need for accuracy and consistency. While there are exceptions to every rule, almost no one wants to think of themselves as acting randomly or behaving inconsistently. Even when we do so, we invent narratives through which we can “make sense” of our behaviors.

Values “Justice is the first virtue of social institutions.” —John Rawls in a Theory of Justice

One of the most difficult and controversial things to measure (and perhaps even discuss) in polite social science is “values.” To begin with, “values” can have very different meanings in different disciplines. In economics, the term “value” is generally meant to mean “economic value,” which is generally understood to be a type of “marginal utility”—“which means, roughly, usefulness relative to the costs of other objects” (Turner 1990, 747). The advantage to this definition is that values in these terms can be measured or referred to as the price. Sociologists and political scientists have used the term “values” to mean a great many different things. For Moore (2001, 5), “Values are ideas about the correct form of behavior, which stress how we ought to behave in general terms.” For example, one general value might be that it is wrong to hurt

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  69 other people. Values differ from “norms” in that norms are defined by the behavior of others around you, whereas values are prescriptive in the sense of what one “ought” to do. It may be “normal” to take a bribe in some contexts, but almost no one thinks that bribe-­taking is a good thing or socially desirable because bribe taking runs against most people’s basic values (Rothstein and Uslaner 2005). Similarly, tax evasion is prevalent in some societies even when most people in those societies think paying taxes is a social value and an obligation of citizenship. This simple fact is intriguing. Why is tax evasion prevalent in many countries even though paying taxes is thought to be a basic duty of citizenship? In the following section we set out to test whether social values, measured as a willingness to share a subject’s endowment with others (Social Value Orientation), affect the willingness to pay taxes.32 Values are not just hard to define, they are also hard to measure. We wanted to have a measure that could be used alongside our tax compliance experiment and that was relatively straightforward. Fortunately, a group of game theorists developed an instrument called the Social Value Orientation (SVO) index (Murphy et al. 2011). The SVO is a unique and rather clever instrument designed to measure an individual’s willingness to give up something for the benefit of someone they do not know. The SVO experiment works like this: All participants are told that they will be matched with someone else in the room (whom we call “the other”) and then presented with a scale that will divide a sum of money between themselves and the other person. Neither they, nor the other can know who the other is, either during or after the experiment. In other words, all their choices are completely anonymous. Subjects are not told how many times they will be asked to make these choices. They are also specifically told that “there are no right or wrong answers,” and that each person should follow their personal preferences. Finally, they are informed that at the end of the experiment one of the pair’s choices will be randomly selected and that person’s choices will decide the ultimate distributions of money to each participant. The idea here was to give people several different ways of distributing money between themselves and someone they do not know and who will never know how they behaved or what they did. An individual could be profoundly selfish and keep 100 percent of the money for themselves, or they could divide the money evenly, or even give more to the “other” and no one, but themselves will ever know what choices they made.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

70  Willing to Pay?

Figure 3.15  SVO Choice Set

Each participant was given an example first and then told they could move their computer cursor to anywhere on the scale and choose their preferred allocation.33 Figure 3.15 is one example of one of the SVO sets. Classical economic theory predicts that everyone should choose whatever point on the scale that maximizes the chooser’s income, and that they should be relatively neutral about what the other receives—since the other person will never know who had made the allocations. This is not what we found. Instead, the majority of the 1,700+ people who took this test in our experiments chose to give some money to the random other—even though this cost them personally and the “other” would never know who they were. In Rational Choice terms, this is “irrational.” There were indeed a large number of different types of responses, but in fact only 46 percent chose to only maximize their own incomes. The data from the SVO experiment can be used in a variety of different ways (see Murphy et al. 2011), but in order to simplify the analysis, we aggregated the responses into a set of categories that make them easier to visualize and compare.34 We present the distribution of our SVO-­angle below in

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  71

Propotion of participants

15%

10%

5%

0% –20

–10

0

Competitive

10

20 SVO angle

Individualist

30

40

Prosocial

50

60

Altruist

Figure 3.16  Distribution of SVO-­angle

Figure 3.16. Starting from the far-­left side of Figure 3.16 are what we termed “competitive” individuals (SVO < 12.04), next were what we termed “individualist” (SVO between 12.04 and 22.45), followed by “pro-­ socials” (SVO between 22.45 and 57.5), and finally on the far right are “altruists” (SVO > 57.5). At one end of the scale the competitive types actually sacrificed some of their own payoffs in order to minimize the payoff to the random other. At the other end of the scale, the altruistic types chose to maximize the payoffs of the random other even when it reduces their own final income. Of course, the vast majority of our subjects were somewhere in the middle. “Individualists” ignored the payoffs to others and simply maximized their incomes and “Prosocials” were willing to sacrifice something to increase the income of the other. Once again, we see that there are many different types of individuals in any society. Not everyone acts like an economist thinks they should, nor is everyone prosocial.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

72  Willing to Pay? 90% 80%

0.86

0.85 0.8

70%

0.78

0.75

0.77

0.76

0.7

Tax Compliance Rate

60%

0.67

0.64 0.6

0.59

50% 40% 30%

0.48

0.45

0.44

0.44

0.45

0.36

20% 10% 0% R1

R2

R3

R4 R5 R6 Experimental rounds

R7

Individualist

Prosocial

R8

R9

Figure 3.17  Average Compliance Rate for Individualists and Prosocials

Our next two questions were whether there is a correlation between an individual’s SVO score and their “willingness to pay” in our tax compliance experiment, and whether the distribution of these “types” differs between countries. Figures 3.17 and 3.18 (below) show that the answer is “yes” to both questions. Figure 3.17 presents the average compliance rate in all rounds of the experiment for prosocials and individualists. From Figure 3.17, we can clearly see that social values have an enormous effect on tax compliance. Prosocials report significantly more income than individualists’ in every round. Indeed, in our tax compliance experiments, the average compliance rate for prosocials is 77 percent, compared to just 51 percent for individualists. The first bar in Figure 3.17 illustrates that the compliance rate for individualists is 36 percent in Round 1, whereas prosocials report 64 percent of their income on average.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  73 10%

Italy

UK

US

Sweden

7.5%

Propotion of participants

5% 2.5% 0% 10% 7.5% 5% 2.5% 0% –20 –10 0

10 20 30 40 50 60

–20 –10 0

10 20 30 40 50 60

SVO-angle Competitive

Individualist

Prosocial

Altruist

Figure 3.18  Distribution of the SVO-­angle by Country

Figure  3.18 below adds an interesting piece to our puzzle. Figure  3.18 exhibits the distribution of the SVO across countries. On the X-axis is the proportion of subjects and on the Y-axis is SVO angle for each country. Looking at Figure  3.18 by row, the top row reveals that the modal Italian subjects and the modal Brit are individualists; the bottom row reveals that the modal American and Swede are prosocial. In other words, prosocials are more compliant in every country, but as Figure 3.18 shows, social values are distributed differently between countries. Whereas in Sweden and the United States, prosocials outweigh individualists, Italy and the United Kingdom have more individualists than prosocials (see Table A.2, Model 3 in Appendix 3 for regressions). Finally, if we look at the behavior of economists in the SVO we once again find them to be significantly more selfish than our subjects from any other educational background (see Figure 3.19). In sum, values are the third part of our compliance puzzle. If a person is motivated both by a desire to get the most for him or herself, to cooperate,

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

74  Willing to Pay? Other Major

Economics Major

Propotion of participants

15%

10%

5%

0% –20 –10

0

10

20

30

40

50

60

–20 –10

0

10

20

30

40

50

60

SVO-angle Competitive

Individualist

Prosocial

Altruist

Figure 3.19  Distribution of the SVO-­angle by Economists

and by what they think is right, what they actually do in any particular situation depends a great deal on the institutional incentives they face, their perception of local norms, and their basic social value orientation. When the incentives are extremely clear in favor of cooperation, even those who are more self-­oriented are likely to cooperate. When the incentives for maximizing your individual self-­interest by defecting or cheating on others are very strong, even those who have a preference for cooperation may well “defect.” Real people have multiple and sometimes conflicting motivations and preferences. Most of us will act in our individual self-­interest when the incentives for doing so are very high and the price of defection is very low. At the same time, most of us will contribute to a public good (pay our taxes) when we feel that it is in society’s interest, even if it costs us individually (e.g. paying taxes). Finally, most of us are neither purely self-­interested “individualists” or pure “altruists.” In sum, self-­interest, norms, and social values are all pieces of the story that are necessary to help explain why, or predict when, an individual or a group, will contribute to or cheat on their social obligations . . . And pay their taxes.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Reasonable Choices and Tax Compliance  75

-----------------------------This study shows that there are remarkable similarities across countries, some of which we expected and others that we did not. It also demonstrates that institutions matter because they can both provide incentives for preferred behaviors and disincentives for anti-­social behaviors. This is why we have laws that penalize those who cheat on their taxes and why we have high taxes on items considered to be detrimental to our health. Institutions structure behavior because we are self-­interested. Efficient and effective institutions are able to structure behaviors to the point where that behavior is repeated enough that it becomes the norm. At the same time, norms motivate individuals to behave in accordance with others in their group . . . even when no one is looking. We want to behave like those around us behave because we are social creatures. Finally, as we repeat certain behaviors, we try to explain these behaviors to ourselves by building theories in our heads. These theories can become values. Values, furthermore, are often passed down through generations. In the end, our behavioral experiments do not support simple-­minded cultural stereotypes, such as the idea that southern Europeans are basically more dishonest and/or less willing to cooperate than their northern European counterparts (Banfield 1958; Putnam et al. 1993). Our experiments demonstrate some cross-­national variation, but not always in the ways that we expected. For example, whereas we expected that Americans would be quite hostile to the idea of paying taxes and would be culturally predisposed to evading when possible, we find that our American subjects are actually quite rule-­abiding, and even if they do not like paying taxes, they are generally quite compliant. To our great surprise, it was the British participants in our study who stood out for their “unwillingness” to pay taxes and/or contribute to the public good. Instead, what we found is that in all countries the same three variables seem to affect tax compliance in similar ways. Still, we do find significant differences between countries in our experiments. They are just not what we originally expected. Morever, there are substantial and robust differences in the behavior of subjects in different countries that cannot be explained by institutional incentives. To understand these differences, we need to look more closely at the historical and political context. In our volume, The Leap of Faith, we engage much more comprehensive examinations of the particular countries’ tax histories and behaviors. In sum, our laboratory experiments combined with our historical analysis have helped us to generate what we believe to be a simple, but sophisticated,

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

76  Willing to Pay? theory for why (most) people willingly pay their taxes and others do not. We began from the starting point that institutions and specifically political institutions matter. Institutions structure behavior and provide cues and signals that shape people’s cost–benefit analyses, their beliefs and expectations about other citizens, and ultimately, we believe, their values.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Conclusion Everyone Hates a Cheat

One of the dominant and recurring complaints against x-­ President Donald J. Trump was the suspicion that he did not pay his taxes. Trump was the first and only presidential candidate since Richard Nixon to have not shared a copy of his tax returns with the American public while running for office.1 Despite Trump eventually being elected—even though he failed to reveal his tax returns—the issue continued to plague the President throughout his presidency and beyond. Indeed, even in the 2020 election year, the release of Trump’s tax returns formed the basis of two Supreme Court cases. For many Americans, it proved to be just one more example of the fact that Mr. Trump was untrustworthy and more concerned with protecting himself than the well-­being of the American society to which he had sworn an oath. Still, Trump’s tax returns present us with a puzzle: Why do so many Americans care about a candidate’s taxes? Taxes, after all, are considered private in America, and the US has long been a peculiarly anti-­tax country. Indeed, the very foundations of the American republic are built on anti-­tax revolts, and anti-­tax sentiment still runs deep in the American psyche. Why, then, should any politician—especially one who promised to cut taxes—be expected to pay any more than he can legally, if underhandedly, avoid? We suggest that the answer to these questions is that the vast majority of Americans, and indeed the vast majority of citizens in every democratic society, believe that it is the civic duty of everyone to pay “their fair share.”2 Trump’s unwillingness to open his tax accounts led many to suspect that he may have legally avoided or even illegally evaded paying taxes by manipulating America’s complex tax laws in his personal self-­interest. In so doing, Trump did not pay his “fair share.” Many also suspected, with good reason, that he had also violated a social norm that we should all pay into society in proportion to how much we earned in that same society. Indeed, his protestations that all he did was use the laws to lower his burden demonstrated to many that this was yet one more example of his flagrant promotion of self-­interest over social well-­being. In other words, his behavior suggested something nefarious about his values as well. Willing to Pay? A Reasonable Choice Approach. Sven Steinmo and John D’Attoma, Oxford University Press. © Sven Steinmo and John D’Attoma 2022. DOI:10.1093/oso/9780198796824.003.0005

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

78  Willing to Pay?

Interests, Norms, Values Tax compliance and tax evasion are usually studied in the negative. This book started from the opposite question; instead of asking how we explain tax evasion, we began with the question, why do so many people willingly pay their taxes in the first place? Taxation can be understood as a collective action problem (Olson 2009); the collective is better off if everyone pays their taxes, but everyone has a financial incentive to pay as little as they can get away with even while benefiting from the contributions of others. The classical economic explanation for tax compliance is that people will pay only when they feel that the costs and risks of being caught for evasion outweigh the benefits from cheating. But note also, this classical view only considers the cost–benefit calculation regarding taxes themselves. How much one benefits from the social goods financed by taxation is outside this model. As we have seen in this study and in many of the other works cited herein, this narrow view of self-­interest as the primary motivator for social cooperation is plainly inadequate. People—both in the real world and in our experimental world—clearly have multiple motivations for contributing to a public good like taxation. We have suggested that people reason in broader ways when making choices. In addition to pursuing narrow individual self-­interest (a.k.a. homo economicus), most people can and also do ­consider the broader interest of the society or group. They can sometimes see their self-­interest as tied to the well-­being of the group and are sometimes willing to sacrifice their individual interests for the good of the group or society. Additionally, because we are social animals, we desire to be part of a group and act according to the group’s norms. To be sure, we may do this in part to enhance our reputation, etc. (e.g. advance our individual self-­interest), but this explanation is also inadequate. As our experiments demonstrate quite clearly, most people are profoundly motivated to behave according to their perception of group norms even when no one in the group will ever be able to monitor their private behavior. Finally, people’s values matter. Most people believe that they ought to contribute to a public good, because it is the right thing to do. Some will behave this way when it is clearly not in their self-­ interest, and even when they believe that most other people in their group or society do not. Of course, in any society there will also be someone like Donald J. Trump. In this book we made the case that these three distinct, but interrelated, set of motivations determine whether an individual is more or less likely to

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Conclusion  79 contribute to a public good. In other words, if you want to know why person X is more likely to pay all the taxes he or she should pay than person Y, you should look to the institutional environment in which he or she lives (the calculation of the financial costs and benefits of compliance vs. evasion), the local norms in the society or group within which they are embedded (the expectation of how others around them will behave), and their values (their understanding of what is right and wrong). When we examine the “micro-­foundations” of human behaviors, as we’ve done here, we do not encounter homo economicus. Instead, we find human beings that display a complex mix of motivations, preferences, and desires. The mix or weighting of these motivations may differ according to both the individual concerned and the society in which they live. In other words, every individual (or almost every individual) is motivated by all three factors. They want to maximize their interest in a cost–benefit way, they want to go along with and get along with the norms of the social group in which they find themselves, and they want to do what they think is “right.” Clearly, different individuals weigh these motivations differently. Some people are much more likely to make a choice according to a cost–benefit analysis than others. Similarly, some people are more likely to try and meas­ ure what is expected of them within the group and according to what the group’s norms are. Finally, some people will act according to what they believe to be right, correct, or fair, virtually irrespective of how they believe others will behave, and, even up to a point, irrespective of their own individual self-­interest.

Successful Societies and Institutional Performance Some societies are more successful than others. Successful societies tend to be relatively wealthy, have less poverty, less corruption, less violence, and more efficient institutions. In successful societies people tend to obey laws, pay their taxes, trust each other and are happier. You do not have to be a rocket scientist, or even a social scientist, to know that all these factors are interrelated. But how? Are some societies more successful than others because they have better institutions, or are the institutions better because they have better people? Is the key difference between Sweden and Italy the fact that Swedes are more honest and trustworthy, whereas Italians are more dishonest and untrustworthy? In other words, are the differences between the more successful and the less successful societies matters of culture or institutions?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

80  Willing to Pay? In this book and project, we examined a small set of democratic countries that range quite broadly both in terms of institutions and cultures. We explicitly tried to examine whether there are basic “cultural” differences between the societies that would explain something as obvious as to why people are more likely to pay their taxes in one country than in the other. By now the reader knows that the answer to this basic puzzle is that differences in behavior cannot be simplified down to differences in culture. Italians are not more dishonest than Swedes, Brits, or Americans. People behave differently in different societies, to be sure. Certainly, we can say that there are different “cultures,” if what we mean by culture are the observed behaviors that vary from society to society. This project was conceived when Steinmo witnessed a man toss a bag of garbage out of his car right next to a large garbage bin near Fiesole, Italy. He couldn’t help but think that this kind of behavior could scarcely exist in a country like Sweden. Not only was this man throwing garbage onto the street, but in doing so he was also teaching his children that it was okay to behave in what is an obviously antisocial way. Sven couldn’t help but wonder “why?” The result was an ERC advanced grant. Over the next several years we studied something that seems quite different from littering, but in our view is actually quite similar. The act of throwing garbage on the street and the act of cheating on your taxes have much in common in our view. Both are expedient individual behaviors that undermine the society as a whole. Studying tax compliance seemed like a perfect way to understand why people in different countries behave in such different ways. Every society has taxes and every society has free-­riders, but the extent of taxes and free-­riders varies greatly between countries, and so do the formal and informal institutions that structure society. The basic idea of this project was to see if we could combine comparative historical analysis, micro-­analysis of individual behaviors, and cognitive science. We believe that The Leap of Faith volume coupled with our experiments provides a novel approach to studying tax compliance. Our ambitious five-­ year study of the fiscal foundations of several democratic states and our tax compliance experiments make this one of the largest studies of tax compliance (of which we know) to date. Over the years, we gained an enormous amount of insight into the reasons for which people pay their taxes. We have built upon forty years of existing tax compliance research and in doing so, confirmed some findings while contradicting others. Overall, what we found, however, was both predictable and surprising.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Conclusion  81

Explaining National Variation The significant differences in tax compliance behavior across societies matters. When a government loses revenue through tax evasion, everyone in that society suffers. Not only does this result in less public money to spend on programs and projects that citizens want, but it also leads to increases in tax rates and levels to pay for the revenues lost. Perhaps even more importantly, tax evasion builds distrust in society. If many people “get away” with not paying their taxes, other taxpayers quite reasonably conclude that the system is unfair. If this kind of inequity is pervasive, it is also reasonable for citizens to ignore or disobey other rules when they feel they can get away with it. The result is ever more distrust of public institutions and ultimately of each other. This is precisely why Donald Trump’s tax evasion story is so important. If even the President of the United States can get away with not following basic civic rules like tax compliance or mask-­wearing, why should anyone pay taxes or wear a face mask? We believe that these variations are firstly a product of the different incentives and disincentives that individual citizens face in those societies when making decisions. In other words, the most important explanation for why more people pay their taxes in Sweden than in Italy, is that the efficiency of the state institutions is far greater in Sweden than in Italy. The cost–benefit calculation that citizens make drives them more toward compliance in Sweden than Italy partly because the perceived opportunity to get away with evasion is significantly smaller in Sweden than in Italy. Public institutions are the formal rules and laws set out by public authorities that are meant to guide and structure behavior in society. Informal institutions are the social rules and norms that govern our behavior. As we have shown in the Leap of Faith, and in chapters 2 and 3 of this book, in some places there is a fairly strong relationship between the formal and informal institutions. In “successful societies” both the formal and informal institutions are aligned. In some places there is a wide divergence between the formal rules and actual social behavior; these are less successful societies.3 Successful societies have more efficient institutions that signal and monitor desired behaviors or choices. Less successful societies have less efficient or efficacious institutions that neither give clear signals nor effectively monitor the behavior of their members. Since we are all partially motivated by norms and partially motivated by our self-­interest, when we live in a society in which violating social rules or laws is rarely punished, more people will defect if they find it to be in their self-­interest. The more people that defect, the more others

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

82  Willing to Pay? will begin to do the same because they see that as the norm. Conversely, in successful societies few people defect, and consequently, the cost of defection grows (even for those who are more motivated by a cost–benefit self-­interest calculation). Andrighetto et al. (2013) provided empirical evidence for this complex dynamic between social norms and institutions. Using a public goods game, which represents a situation in which individuals face a conflict between the maximization of personal gain and collective interest, the authors show that just providing clear information about the socially appropriate behavior in the game, that is, the social norm, is sufficient to promote cooperation among participants. However, if this information about the norm is not enforced by sanction, cooperation soon breaks down. When no sanction is available, an increasing number of participants defect and consequently the cooperative norm erodes, and participants eventually turn back to a cost–benefit calculation as to how to maximize the money they can get from the game. We have shown in several chapters in The Leap of Faith that the most successful societies built effective monitoring and measuring mechanisms before they began to extract heavy revenues. Sweden, for example, very early on established local community institutions (based on the Cadastral system) which had the capacity to monitor both who paid taxes and who didn’t.4 As a consequence, the state developed the ability to enforce rules broadly and relatively effectively. Thus, even those who might otherwise have defected tended to cooperate and contribute. In the Italian case, in contrast, the state failed to develop efficient institutions that could monitor its citizens and/or taxpayers effectively. Consequently, many taxpayers defected. The result was that the Italian state has had difficulty to this very day in generating the high levels of compliance observed in Scandinavia. This is not because Italians are more dishonest than Swedes. It is instead because the inefficiencies of Italian institutions never incentivized individuals to always comply. The result has been a norm of non-­compliance. Margaret Levi’s often quoted phrase “no one wants to be a sucker” (Levi 1988, 53, our emphasis) only makes sense in this context. You don’t want to be a sucker because you don’t want to contribute when no one else is contributing. Being a sucker or not has little to do with the odds of being caught for defecting and instead has much to do with norms. Even in countries like Sweden, some people still defect, cheat, and sometimes even steal from their neighbors. Also, even in a low compliance country like Italy, most people contribute fully, even when they know there are many others who do not. In other words, some people will violate the social norms, and perhaps even their self-­interest, because their weighting of the different motivations for compliance varies from others.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Conclusion  83 Aggregating this up we see that the weighting of self-­interested norm-following versus prosocial behaviors may differ across societies, but this does not imply that in any one society everyone is either X or Y (see Chapter 1). Our study demonstrates that there are very significant differences in the way people behave within and across societies—even when offered exactly the same incentives and institutions. In short, not all people are the same. Some people choose to comply (with the experimenters in a lab field study and/or with real world authorities), others shirk. Still, the differences across societies are not random. In some countries we find more people willing to cooperate and comply, while in other societies we find a relatively larger percentage of ­people who are willing to cheat. Thus, political institutions are important because they shape people’s cost–benefit analyses, their beliefs, and expectations about other citizens, and ultimately, we believe, their values. An institution which has a very low payback (or redistribution in our experiments) provides substantially lower incentives for the individual to contribute. There are very significant differences between societies in how much taxpayers feel they “get back” from their tax money. Moreover, in the real world, some tax administrations have limited ability to monitor and effectively control the level of tax compliance in their countries. Therefore, all other things being equal, an individual is more likely to pay their taxes in Sweden, or for that matter Britain, than they would in Italy, for the obvious reason that they are far more likely to be caught for tax evasion in Britain or Sweden than in Italy. This is not because Italians are more dishonest or less moral than Swedes or Britons, but because of the structure of their political economy, along with the inefficiencies in their tax administrative systems; there are quite simply more opportunities to evade tax in Italy than Sweden, Britain, or the United States. In sum, our interpretation of the fact that the Tax Gap is so much higher in Italy than in the other countries is driven by: (1) the fact that Italians have greater opportunities to evade without being caught or fined than do taxpayers in other countries; (2) Italian citizens believe (either correctly or not, it does not matter) that they receive less back for their tax money; and 3) they tend to believe that others are evading and getting away with it. At the individual level, Italians have greater opportunities to cheat and at the social level they have fewer incentives to contribute. On the flip side, we argue that tax compliance is so much higher in Sweden, for example, than Italy or America, because the Swedish tax system is radically more efficient and difficult to evade, and Swedes tend to believe that they get good value for their tax money.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

84  Willing to Pay? We came into this as institutionalists, believing that institutions fundamentally structure and shape individual choices. We leave this project as institutionalists as well. But no one who has actually lived in multiple diverse societies could have failed to notice that people behave in quite different ways in different societies. This research allowed us to delve further into the dynamic relationship between institutions, norms, and values, and how central it is to explaining variation across countries. In the end, what we put forth is a simple theory for social cooperation that we believe will be useful to both scholars beyond the study of tax compliance. But we also hope that these findings can be useful to governments and policy makers who want to devise strategies to improve tax compliance.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

APPENDIX

This appendix is separated into three parts: First, we provide our full instructions of the experiments. Second, we provide our experimental survey. Third, we provide a more technical analysis of our experiments.

Appendix 1:  Full Experiment Instructions We outlined our experimental design in Chapter 2 with detailed examples, but we did not provide the full instructions. To reiterate, we recruited all of our students using Online Recruitment Software for Economic Experiments (ORSEE), a software commonly used throughout the economic discipline to recruit subjects. Every laboratory has its own preferred way of recruiting subjects, but usually it entails asking subjects, who are most often students or related to the university, to proceed to a website and enter their information to be contacted for experiments. In our case, typically a week before the experiments, we contacted potential subjects, asking them to sign up to participate in our experimental sessions. On the day of the experiment, the participants would arrive at the lab where we asked them to wait outside until everyone showed up. Once we reached the number of subjects necessary to conduct the experiment, we provided them with a random seat number and asked them to take a seat at an apportioned computer that corresponded with their ticket number. Once everyone was seated, we welcomed them and began to read the instructions. Appendix 1 is an exact copy of the experimental instructions to our baseline experiment, which we read to our participants, and they followed along on their computer screens. As the participants read through the instructions, they saw the word SCREEN # in orange. This indicates the screen number on which the text appeared. So, SCREEN 1, for example, means that the following text appeared on the first screen that the participants saw. Once the participants either finished reading or made their decision and were ready to move on to the next screen, they were then asked to press “continue” to move on to the next screen. This comes at the bottom of each screen. Moreover, as we mentioned in Chapter 2, the study was separated into three parts: a tax compliance experiment, an SVO task, and an attitudinal survey. The tax compliance experiment was broken up into three stages: one that changes the redistribution, another that changes the tax rates, and the last one that changes the tax regime. Before each stage there was a CLERICAL TASK lasting five minutes. Within each stage there are three rounds. Each Stage and Round is denoted by the text INSTRUCTIONS 1.1 (for example)—this means Stage 1, Round 1. The reader will also notice the text EXAMPLE. Here we provide an example for each round so that the participant has a clear idea of how their decisions affect their final payoff. This is usually followed by a REPORTING SCREEN which is the screen where participants report their income. Between decisions the reader will see a TRANSITION SCREEN to indicate that they are moving on to the next round, task, decision, etc. Once everyone completed the tax compliance experiment, they proceeded to the SVO task. We detailed this experiment in Chapter 2, so here we simply provide instructions on how to read the text below. Each set of the SVO INSTRUCTIONS provides a detailed

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

86 Appendix explanation of the task for participants. This is followed by the RESULTS screen which provides the subjects a detailed breakdown of the decisions they made and how we calculated their payoff (results). Finally, we ask subjects to fill out a brief questionnaire (QUESTIONNAIRE INSTRUCTIONS below) of which the exact question wording can be found in Appendix 2.

Willing to Pay? Baseline experiment instructions Below are the instructions given to the subjects both orally and written on their computer screens. The coordinator reads the following aloud as the subjects read along on their computer screens. ------ - --------------------------------------------------------------------SCREEN 1 Before we begin, please type your experiment ID code into the field on your computer screen. You can find your experiment ID code by opening the envelope we handed to you as you entered the room. If you have any questions about this, please raise your hand now. Please take a moment to type in your experiment ID code now. Please press “Continue.” -------------------------------------------------------------------------SCREEN 2 – WELCOME Welcome. Thank you for agreeing to participate in this experiment. This experiment will not be particularly difficult or involve trick questions. You will simply need to follow the instructions as they gradually appear on your screen. The answers you will provide will be confidential. Those who process the data from the experiment will not be able to match your name to your choices. During the experiment, you will be asked to make choices. It is therefore important for the success of the experiment that you do not talk to each other and that you read the instructions very carefully. If you have questions during the experiment, please raise your hand. At the end of the experiment, you will receive your payment in cash. The actual amount will depend partly on your choices, partly on the choices of the other participants and partly on chance. Please press “Continue.” -------------------------------------------------------------------------SCREEN 3 – OVERVIEW Today’s experiment is divided into four stages. Every stage is, in all respects, separate from the others. This means that the choices you make in each stage will have no effect on your earnings in the other stages. Your final earnings will be the sum of the earnings obtained in all four stages. At the beginning of each stage, we will read the relevant instructions and give examples to ensure that the experiment is clear. At the end of the experiment, you will turn in your envelope with your experiment ID number inside. Your earnings will be issued based on this number, not based on your name, so it is important that you retain your ID number. The payments will be issued in

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  87 cash in the next room, where only one participant at a time may enter. This way none of the other participants will know how much you earned. Please press “Continue.” -------------------------------------------------------------------------SCREEN 4 – THE FIRST THREE STAGES The first three stages of the experiment are similar to each other in many ways. In each of the first three stages, you will be asked to perform a clerical task for 5 minutes. Based on your performance, you will earn currency units; these will be converted into dollars at the end of the experiment. You will participate in several rounds, during which you will be asked to report your income for tax purposes. You are free to report any amount, from 0% to 100% of your earnings. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, the computer will choose a number between 1 and 100 for each of you in each round. If the result is a number between 1 and 5, you will be audited for that round. If you are audited and you have under-­reported your earnings, you will have to pay the tax you should have paid, plus a fine equal to that tax. In other words, you will have to pay twice the tax you should have paid. Please press “Continue.” -------------------------------------------------------------------------SCREEN 5 – CLERICAL TASK The work that you will complete involves copying data about fictional students from a sheet of paper onto your computer screen, starting from the first line on the list and following the correct order. This piece of paper is in front of you, entitled “Practice Clerical Task.” Once you have completed the first line, click the NEXT button. If the computer detects a mistake, you will see an error message, and you will have to correct the mistake before proceeding. If there is no mistake, clicking NEXT will take you to a new screen on which you can start the next line. For each row copied correctly, you will be paid 10 currency units. Each currency unit in this task is worth 0.01 dollars. We now ask you to try a practice round of this clerical task for 2 minutes. Your performance in this practice round will not count towards your earnings. Afterward, we will begin the real clerical task, for which you will be paid, and which will last 5 minutes. Please press “Continue.” ................................................................................ Your screen should now show the number of rows you copied correctly. When you have finished viewing your results, please press “Done.” Now that everyone has completed the practice clerical task, we will begin the real clerical task, for which you will earn real currency units that you can later exchange for dollars. Please turn to the next page in your packet, entitled “Clerical Task #1.” You will start with the first line on this page, which is row number 101. ................................................................................ Your screen should now show the number of rows you copied correctly. When you have finished viewing your results, please press “Done.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

88 Appendix -------------------------------------------------------------------------SCREEN 6 – TRANSITION SCREEN Now that everyone has completed the clerical task, you will have to report your income for tax purposes. You will be asked to do this multiple times during the next three rounds. Please press “Continue.” -------------------------------------------------------------------------SCREEN 7 – INSTRUCTIONS 1.1 In the first round, we ask you to report your income for tax purposes. The tax rate is 30%. Please remember that you are free to report any amount of your earnings from 0% to 100% for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and are caught under-­reporting your income, you will pay the tax you should have paid, plus a fine equal to that tax. In other words, you will pay twice the tax you should have paid. Please press “Continue.” -------------------------------------------------------------------------SCREEN 8 – EXAMPLE 1 Before we begin, let’s look at some examples. Let’s assume that we have three participants: X, Y, and Z. All three participants copy 10 names correctly during the clerical task, and each participant therefore earns 100 currency units. If Participant X accurately reports his total real earnings of 100 currency units and the tax rate is 30% he will pay 30 in taxes. His final income will therefore be 70 currency units. If Participant Y reports only 50 currency units, she will pay 15 currency units (since you are only taxed on the income you report). In this case, her final income will be 85 currency units, if she is not audited and fined at the end of the experiment. If Participant Z reports 0 currency units, he will pay 0 currency units in taxes and his final income will be 100 currency units, unless he is audited and fined at the end of the experiment. Please press “Continue.” -------------------------------------------------------------------------SCREEN 9 – EXAMPLE 2 What happens if all three participants (X, Y, and Z) are audited and fined at the end of the experiment? If Participant X is audited, nothing happens: since he reported his total real earnings, he will not incur any penalty, and his final income will be 70 currency units. If Participant Y is audited, then since she underreported her earnings, she will have to pay twice the tax she did not pay. Since she reported 50 currency units, she owes 15 currency units in taxes. She will therefore have to pay double that (30 currency units) and her final income will be 55 (since 85 − ­30 = 55). If Participant Z is audited, then since he reported 0 currency units, he owes 30 currency units in taxes. He will therefore have to pay 60 currency units and his final income will be 40 (since 100 − ­60 = 40). During the experiment, do not worry about making these calculations. The computer will do them for you. Please press “Continue.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  89 -------------------------------------------------------------------------SCREEN 10 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 30%. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 11 – INSTRUCTIONS 1.2 The second round is very similar to the first one. As in the first round, we ask you to report your income for tax purposes, at the tax rate of 30%. Again, you are free to report any amount of your earnings (from 0% to 100%) for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and it is discovered that you have under-­reported your earnings, you will have to pay a penalty equal to twice the tax you should have paid on the income you did not report. The only difference between the first and the second round is that in the second round, the money that is collected in taxes from all participants will be put into a general fund and then redistributed in equal parts among all participants, regardless of how much each participant paid in taxes. Please press “Continue.” -------------------------------------------------------------------------SCREEN 12 – EXAMPLE Let’s look at an example. Let’s assume that 3 people participate in the experiment, and that each one of them earns 100 currency units during the clerical task. One participant reports 100 currency units, one reports 50 currency units, and the other reports 0 currency units. The amount collected in taxes and put into the general fund is therefore 45 currency units (since 30 + 15 + 0 = 45). Each participant receives 15 currency units from the “general fund.” Please press “Continue.” -------------------------------------------------------------------------SCREEN 13 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 30%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous round. In other words, your income is exactly the same amount as in the beginning of the first round. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 14 – INSTRUCTIONS 1.3 The third round is similar to the second one. The only difference is that the amount of money that is collected and put into the general fund is now multiplied by 2 before being redistributed in equal parts among all participants. Again, everyone receives an equal amount from the “general fund,” regardless of the amount that each participant paid in taxes. Please press “Continue.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

90 Appendix -------------------------------------------------------------------------SCREEN 15 – EXAMPLE Let’s look at an example. If the total amount of tax collected is 500 currency units, the amount in the general fund will be doubled and the amount redistributed in equal parts among all participants will be 1000 currency units. Let’s look at another example. Let’s assume that 20 people participate in the experiment and that each of them pays 100 currency units in taxes. In this case, the amount in the general fund will be doubled to 4000 (since 20 × 100 = 2000 × 2 = 4000) and each participant will receive 200 currency units from it. Please press “Continue.” -------------------------------------------------------------------------SCREEN 16 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 30%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous round. In other words, your income is exactly the same amount as in the beginning of the first round. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 17 – INSTRUCTIONS 2.0 Now let’s begin the second stage of the experiment. The second stage is very much like the first stage. We will again ask you to perform a clerical task for 5 minutes. Based on your performance, you will earn currency units, which will constitute your income in this stage, and which you will be able to turn in for real money at the end of the experiment. As in the last stage, you will participate in three different rounds, during which we will ask you to report your income for tax purposes. You are free to report any amount (from 0% to 100%) of your earnings. You will only be taxed on the income you report. There is, once again, a 5% probability of being audited. At the end of the experiment, if you are audited and it is discovered that you have under-­reported your earnings, you will have to pay a penalty equal to twice the tax you should have paid on the income you did not report. In each of the following rounds, the money that is collected in taxes from all participants will be put in a “general fund.” The general fund will then be multiplied by 2 and distributed in different ways throughout the next several rounds. Please press “Continue.” -------------------------------------------------------------------------SCREEN 18 – CLERICAL TASK We now ask you to perform the clerical task for 5 minutes. Afterwards, you will be told how many currency units you earned. The currency units you earn will constitute your income for the next three rounds. Please turn to the next page in your packet, entitled “Clerical Task #2.” You will start with the first line on this page, which is row number 201. Once again, we ask you to focus on the clerical task, so that you can earn as much as possible by correctly completing it. Please press “Continue.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  91 ................................................................................ Your screen should now show the number of rows you copied correctly. When you have finished viewing your results, please press “Done.” -------------------------------------------------------------------------SCREEN 19 – TRANSITION SCREEN Now that everyone has completed the clerical task, we will present you with the different scenarios in which you will have to report your income for tax purposes. Please press “Continue.” -------------------------------------------------------------------------SCREEN 20 – INSTRUCTIONS 2.1 In the first round, we ask you to report your income for tax purposes. The tax rate is 10%. The money collected in taxes from all participants will be put into a “general fund,” multiplied by 2, and redistributed in equal parts among all participants, regardless of how much each participant paid in taxes Please remember that you are free to report any amount of your earnings (from 0% to 100%) for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and are caught under-­reporting your income, you will pay the tax you should have paid, plus a fine equal to that tax. In other words, you will pay twice the tax you should have paid. Please press “Continue.” -------------------------------------------------------------------------SCREEN 21 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 10%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous rounds. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 22 – INSTRUCTIONS 2.2 The second round is similar to the first one. The only difference is that the tax rate is now 30%. Please press “Continue.” -------------------------------------------------------------------------SCREEN 23 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 30%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous rounds. In other words, your income is exactly the same amount as in the beginning of the first round. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

92 Appendix -------------------------------------------------------------------------SCREEN 24 – INSTRUCTIONS 2.3 The third round is very similar to the first and second rounds. The only difference is that the tax rate is now 50%. Please press “Continue.” -------------------------------------------------------------------------SCREEN 25 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 50%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous rounds. In other words, your income is exactly the same amount as in the beginning of the first round. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 26 – INSTRUCTIONS 3.0 Now let’s move on to the third stage of the experiment. The third stage is very similar to the previous ones. We will ask you once again to perform a clerical task for 5 minutes. Based on your performance, you will earn currency units, which will constitute your income in this stage, and which you will be able to turn in for real money at the end of the experiment. As in the last stage, you will participate in different rounds, during which we will ask you to report your income for tax purposes. You are free to report any amount (from 0% to 100%) of your income. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and it is discovered that you have under-­reported your earnings, you will have to pay a penalty equal to twice the tax you should have paid on the income you did not report. Please press “Continue.” -------------------------------------------------------------------------SCREEN 27 – CLERICAL TASK We now ask you to perform the clerical task for 5 minutes. Afterwards, you will be told how many currency units you earned. The currency units you earned will constitute your earnings in the three rounds of this stage, in which we will present you with different scenarios and ask you to report your income for tax purposes. Please turn to the next page in your packet, entitled “Clerical Task #3.” You will start with the first line on this page, which is row number 301. Once again, we ask you to focus on the clerical task, so that you can earn as much as possible by correctly completing it. Please press “Continue.” ................................................................................ Your screen should now show the number of rows you copied correctly. When you have finished viewing your results, please press “Done.” -------------------------------------------------------------------------SCREEN 28 – TRANSITION SCREEN Now that everyone has completed the clerical task, we will present you with the different scenarios in which you will have to report your income for tax purposes. Please press “Continue.”

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  93 -------------------------------------------------------------------------SCREEN 29 – INSTRUCTIONS 3.1 In this round, participants will be divided into three groups based on reported income. Those with the highest reported income (the top 10%) will pay a 50% tax rate; those with the lowest reported income (the bottom 10%) will pay a 10% tax rate; and everyone else will pay a 30% tax rate. In other words, those with the highest incomes pay more in taxes, while those with lower incomes pay less. Note, however, that you will not know for sure whether you are a high income, low income, or middle income earner. As in the previous stage, the money collected in taxes from all participants will be put into a “general fund,” multiplied by 2, and redistributed in equal parts among all participants, regardless of how much each participant paid in taxes. As in the last stage, you are free to report any amount (from 0% to 100%) of your income for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and are caught under-­reporting your income, you will pay the tax you should have paid, plus a fine equal to that tax. In other words, you will pay twice the tax you should have paid. Please press “Continue.” -------------------------------------------------------------------------SCREEN 30 – EXAMPLE 1 Let’s look at an example. Let’s assume that ten subjects participate in the experiment. Of these ten participants, one copies 20 names correctly, one copies 5 names correctly, and the other eight copy 10 names correctly. Every correct name is awarded with 10 currency units. Let’s assume, too, that all participants then report their total real earnings for tax purposes. The participant who copies 20 names correctly has an income equal to 200 currency units and pays 50% of it (=100 currency units) in taxes. These 100 currency units are collected in the “general fund.” The participant who copies 5 names correctly has an income of 50 currency units and pays 10% of that income (=5 currency units) in taxes. These 5 currency units are collected into the “general fund.” Each of the participants who copies 10 names correctly has an income equal to 100 currency units and pays 30% of that income (=30 currency units each) in taxes. A total of 240 currency units (8×30) is collected in the “general fund.” Please press “Continue.” -------------------------------------------------------------------------SCREEN 31 – EXAMPLE 2 There are now 345 currency units in the general fund (100 + 240 + 5 = 345). The amount collected in the general fund is then doubled (345 × 2 = 690) and redistributed in equal parts among all participants (690/10 = 69). Every participant therefore receives 69 currency units from the “general fund.” In this example, the participant reporting the highest income (200 currency units) pays 100 in taxes, receives 69 from the “general fund,” and therefore has a final income of 169 currency units. The participant reporting the lowest income (50 currency units) pays 5 in taxes, receives 69 from the “general fund,” and therefore has a final income of 114 currency units.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

94 Appendix Each of the other participants (reporting 100 currency units each) pays 30 in taxes, receives 69 from the “general fund,” and therefore has a final income of 139 currency units. Please press “Continue.” -------------------------------------------------------------------------SCREEN 32 – REPORTING STAGE We now ask you to report your income for tax purposes. You will pay a tax rate of 50% if your reported income puts you in the top 10% of income earners; a tax rate of 10% if your reported income puts you in the bottom 10% of income earners; and a tax rate of 30% if you are in the middle 80% of reported income earners. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous round. In other words, your income is exactly the same amount as in the beginning of the first round of this stage. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 33 – INSTRUCTIONS 3.2 The second round is very similar to the first one. As in the first round, we ask you to report your income for tax purposes. Again, you are free to report any amount of your earnings (from 0% to 100%) for tax purposes. The probability of being audited at the end of the experiment is, again, 5%. Once again, the money collected in taxes from all participants will be put into a “general fund,” doubled and redistributed in equal parts among all participants, regardless of how much each participant paid in taxes. The new thing about this round is that you pay 10% for the first 50 currency units of reported income, 30% for any reported income between 51 and 100 currency units, and 50% for any reported income above 100 currency units. In other words, all participants pay the same tax on the same amount of income, but you pay a different tax rate on different levels of reported income. Please press “Continue.” -------------------------------------------------------------------------SCREEN 34 – EXAMPLE Let’s look at an example. Let’s assume that Participant X reports 80 currency units for tax purposes. The amount paid in taxes will be equal to 10% on the first 50 currency units, plus 30% on the next 30 currency units, so 14 (= 5 + 9) currency units. Please press “Continue.” -------------------------------------------------------------------------SCREEN 35 – REPORTING STAGE We now ask you to report your income for tax purposes. You will pay a tax equal to 10% for the first 50 currency units of reported income, 30% on reported income between 51 and 100 currency units, and 50% on reported income above 100 currency units. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous round. In other words, your income is exactly the same amount as in the first round.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  95 Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 36 – INSTRUCTIONS 3.3 We now begin the final round. In this round, we ask you to report your income for tax purposes. The tax rate is 30%. As in all the other rounds, you are free to report any amount of your earnings (from 0% to 100%) for tax purposes. You will only be taxed on the income you report. There is, however, a 5% probability of being audited. At the end of the experiment, if you are audited and are caught under-­reporting your income, you will pay the tax you should have paid, plus a fine equal to that tax. In other words, you will pay twice the tax you should have paid. In this round, however, the money collected in the general fund will be multiplied by 2 and the proceeds will be donated to Red Cross (note: we really will do this). Please press “Continue.” -------------------------------------------------------------------------SCREEN 37 – REPORTING STAGE We now ask you to report your income for tax purposes. The tax rate is 30%. Please remember that the amount you report in this round is in no way connected to the amount you reported and the tax you paid in the previous rounds. In other words, your income is exactly the same amount as in the beginning of the first round. Your earnings from the clerical task are on your screen. Please report your earnings for tax purposes now. When you have finished reporting your earnings, please press “Continue.” -------------------------------------------------------------------------SCREEN 38 – TRANSITION SCREEN You have finished the first three stages of the experiment. Now let’s begin the fourth stage. Please press “Continue.” -------------------------------------------------------------------------SCREEN 39 – SVO INSTRUCTIONS 1 We now begin the fourth and final stage of the experiment. This stage is different from the first three. In this stage, you have been randomly paired with another person, whom we will refer to as “the other.” This other person is someone you do not know and will remain mutually anonymous. All of your choices are completely confidential. You will be making a series of decisions about allocating resources between yourself and this other person. Please press “Continue.” -------------------------------------------------------------------------SCREEN 40 – SVO INSTRUCTIONS 2 For each of the questions that will follow, you will be asked to indicate the distribution of resources you prefer by marking the respective position along the midline, as in the example below. You can only make one mark for each question. There are no right or wrong answers; this is all about personal preferences.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

96 Appendix In the example below, a person has chosen to distribute money so that she receives 50 tokens, while the anonymous other person receives 40 tokens. Note that 10 tokens are worth 0.03 dollars. Please press “Continue.” -------------------------------------------------------------------------SCREEN 41 – SVO INSTRUCTIONS 3 Note that at the end of this stage, only one person in each pair will have their decisions applied. In other words, if you are selected to have your decisions applied, then your choices will determine the ultimate payout that both you and the other person receives at the end of this stage. If you are not selected, then the choices made by “the other” will be implemented, and your earnings will depend on the choices they have made about the distribution of resources. The selection will be at random, and neither of you will know beforehand whose decisions will be applied. The earnings from this stage will be added to your earnings from the previous three stages of the experiment to determine your final income. You may move through the following screens at your own pace; you do not need to wait for permission to move to the next screen. Please press “Continue” to start the final stage. -------------------------------------------------------------------------SCREEN 42 – SVO RESULTS We have now completed the four stages of the experiment. Next, we will show you your results. Please press “Continue.” -------------------------------------------------------------------------SCREEN 43 – RESULTS INSTRUCTIONS The next few screens will reveal your earnings for each stage and each round of the experiment, including whether you were audited and fined for a particular round. You may move through these screens at your own pace; you do not need to wait for permission to move to the next screen. The final screen will convert your earnings from currency units into dollars, so that you can see how much you will be paid for your participation in this experiment. Please press “Continue” to begin viewing your earnings. -------------------------------------------------------------------------SCREEN 44 – QUESTIONNAIRE INSTRUCTIONS Finally, we will ask you to complete a short questionnaire, which we will hand out shortly. Please remain seated and refrain from talking to other participants. Once everyone has completed the questionnaire, we will give you instructions about how to receive your payment.

Appendix 2:  Questionnaire to the Baseline Experiment Over the course of the experiment, the questions included in the survey and the specific form of some of the questions changed based on incremental results from the experimental sessions. It should be noted that the experimental design, however, was not subject to change. It also should be noted that not all questions were asked in all countries; those that fall into this category are listed below under the heading Country-­specific variables.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  97 Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options

age Age: Discrete [Respondent selects from dropdown menu] – male Gender: Discrete [Multiple choice radio buttons]: 0 = Female 1 = Male – ethnic Ethnicity String [Respondent types answer] – employed Are you currently employed? Discrete [Multiple choice radio buttons:] 0 = No 1 = Yes – student Are you currently enrolled as a student? Discrete [Multiple choice radio buttons:] 0 = No 1 = Yes – major Major field of study String [Respondent types answer] – minor Minor field of study String [Respondent types answer] – progyear Year in your program [displayed only if student = 1] Discrete [Multiple choice radio buttons:] 1 = First 2 = Second 3 = Third 4 = Fourth or above 5 = Grad Student

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

98 Appendix Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options Notes

– birthplace Country of birth String [Respondent types answer] – m_origin Where is your mother from? String [Respondent types answer] – f_origin Where is your father from? String [Respondent types answer] – past_part Have you participated in other experiments before? Binary [Multiple choice radio buttons:] 0 = No 1 = Yes – times_part If yes, how many times? Discrete [Multiple choice radio buttons:] 1 = Once 2 = Twice 3 = 3–5 times 4 = More than 5 times 5 = More than 10 times (code as 0 if past_part = 0) The scale of this question has evolved over time. In earlier versions of the survey, respondents would type their number of past participations, while later versions included the scale reported here. num_friends How many other participants in today’s session do you know by name? Discrete [Respondent selects number from pull down list of numbers from 0 to N] –

Listed below are various areas of government spending. Please indicate whether you would like to see more or less government spending in each area. Remember that if you check “much more,” it might require a tax increase to pay for it.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  99 Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

spend_envir The environment Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_health Health Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_police Police and Law enforcement Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_educ Education Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_military Military Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

100 Appendix Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes

[Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_pension Old age pension Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_unemployment Unemployment insurance Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above spend_culture Arts and culture Discrete [Multiple choice radio buttons:] 1 = Spend much more 2 = Spend more 3 = Spend the same as now 4 = Spend less 5 = Spend much less 9 = Can’t choose [code as missing] See question introduction above

Now we’d like you to tell us your views on various issues. How would you place your views on this scale? “1” means you agree completely with the statement on the left; “10” means you agree completely with the statement on the right; and if your views fall somewhere in between, you can choose any number in between.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  101 Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text

Variable type Answer options Notes Variable name Question text

Variable type Answer options Notes Variable name Question text

Variable type Answer options Notes Variable name Question text

equality_vs_incentive Incomes should be made more equal – We need larger income differences as incentives for individual effort Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above private_vs_public_ownership Private ownership of business and industry should be increased – Government ownership of business and industry should be increased Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above govt_vs_idv_resp Government should take more responsibility to ensure that everyone is provided for – People should take more responsibility to provide for themselves Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above competition_good_bad Competition is good. It stimulates people to work hard and develop new ideas – Competition is harmful. It brings out the worst in people Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above hard_work_vs_luck In the long run, hard work usually brings a better life – Hard work doesn’t generally bring success—it’s more a matter of luck and connections Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above zero_vs_positive_sum People can only get rich at the expense of others – Wealth can grow so that there’s enough for everyone

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

102 Appendix Variable type Answer options Notes

Discrete [Row of radio buttons:] 1 2 3 4 5 6 7 8 9 10 See question introduction above

Please tell me whether you think the following actions can always be justified, never be justified, or something in between. “1” means you think the action can never be justified, and “10” means you think the action can always be justified. Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes

cheat_tax Cheating on taxes if you have a chance Discrete 1 (Never justifiable) 23456789 10 (Always justifiable) See question introduction above cheat_benefits Claiming government benefits to which you are not entitled Discrete 1 (Never justifiable) 23456789 10 (Always justifiable) See question introduction above

Generally, how would you describe taxes in your country today? We mean all taxes together, including wage deductions, income taxes, taxes on goods and services, and all the rest. Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

tax_high For those with high incomes, are taxes Discrete [Multiple choice radio buttons:] 1 = Much too high 2 = Too high 3 = About right 4 = Too low 5 = Much too low See question introduction above tax_mid For those with middle incomes, are taxes Discrete [Multiple choice radio buttons:] 1 = Much too high 2 = Too high 3 = About right 4 = Too low 5 = Much too low

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  103 Notes Variable name Question text Variable type Answer options

Notes

See question introduction above tax_low For those with low incomes, are taxes Discrete [Multiple choice radio buttons:] 1 = Much too high 2 = Too high 3 = About right 4 = Too low 5 = Much too low See question introduction above

The following are different statements made about the [national] tax system. Please indicate how much you agree with each of these statements, where 1 means you completely disagree and 4 means you completely agree. Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

opinion1 The more a person earns, the higher their taxes should be Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion2 Paying taxes is a fundamental duty of citizenship Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion3 Not paying taxes is one of the worst crimes a person can commit because it damages the entire community Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion4 It is right not to pay taxes if you think they are unfair Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

104 Appendix Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

[Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion5 Even if a person considers a tax unfair, they should pay the tax and then protest later Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion6 High taxes force individuals to evade taxes in order to keep their businesses afloat Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion7 It is justifiable not to pay taxes because the government uses tax money to support people who are lazy Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion8 Most people who cheat on their taxes do so because the tax system is too complicated Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion9 Most people who cheat on their taxes do so because they feel that politicians are corrupt Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  105 Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

[Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion10 Most people who cheat on their taxes do so because the tax rates are too high Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion11 Most people who cheat on their taxes do so because they feel they will not be caught Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion12 It is right to pay taxes because through taxes, we can help the most vulnerable in our society Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion13 If the government worked more efficiently, people would be more willing to pay taxes Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion14 Tax money should be spent where taxes are collected Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

106 Appendix Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

[Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion15 People would be more willing to pay taxes if they had some control over how tax dollars are spent Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion16 Paying taxes is stressful because even people who try to pay their taxes correctly are afraid of making a mistake and being audited Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion17 A person is more willing to pay taxes if he or she believes that other people pay their taxes Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion18 Wealthy people and businesses can usually find legal ways to avoid paying their taxes Discrete [Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above opinion19 Paying taxes is the rational thing to do because our tax dollars fund useful and important services Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  107 Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text

Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text

[Multiple choice radio buttons:] 1 = completely disagree 2 = disagree 3 = agree 4 = completely agree See question introduction above leftright In political matters, people talk of “the left” and “the right.” How would you place your views on this scale, generally speaking? Discrete [Row of radio buttons:] 1 (Left) 23456789 10 (Right) – risk Are you generally a person who is completely willing to take risks, or do you normally try to avoid taking them? Indicate your feelings about this issue below, in which “1” means completely unwilling to take risks, and “10” means completely willing to take risks Discrete [Row of radio buttons:] 1 (Completely unwilling to take risks) 23456789 10 (Completely willing to take risks) – think_self_earn Compared to the other participants in this experiment, do you think your earnings are above average, average, or below average? Discrete [Row of radio buttons:] 1 = Above Average 2 = Average 3 = Below Average – think_self_perform Compared to the other participants in this experiment, do you think your performance on the clerical task was above average, average, or below average? Discrete [Row of radio buttons:] 1 = Above Average 2 = Average 3 = Below Average – think_others_report Do you think most of the participants in the experiment reported their total earnings, less than their total earnings, or much less than their total earnings for tax purposes?

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

108 Appendix Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text

Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

Discrete [Row of radio buttons:] 1 = Reported their total earnings 2 = Reported less than their total earnings 3 = Reported much less than their total earnings – self_report In the experiment, did you report your total earnings, less than your total earnings, or much less than your total earnings for tax purposes? Discrete [Row of radio buttons:] 1 = Reported my total earnings 2 = Reported less than my total earnings 3 = Reported much less than my total earnings – motivations Our experiment was created to help us better understand individuals’ motivations for tax compliance. Could you please explain your motivations for the decisions you made in today’s experiment? String [Respondent types answer] – q70 Are you religious? Discrete [Multiple choice radio buttons:] 0 = No 1 = Yes – rel_affil What is your religious affiliation? [Show only if religious = 1) String [Respondent types answer] – q72 Please rate the importance of religion in your life: [show only if religious = 1] Discrete [Multiple choice radio buttons:] 1 = Very important 2 = Fairly important 3 = Not very important – rel_freq How frequently do you attend religious services? [show only if religious = 1] Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  109 Answer options

Notes

[Multiple choice radio buttons:] 1 = daily 2 = multiple times per week 3 = about once a week 4 = about once a month 5 = only once every few months –

Country-­specific variables Italy Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options

ita_residence Region of residence String [Respondent types answer] – ita_birthplace Region of birth String [Respondent types answer] – ita_m_origin Mother’s region of birth String [Respondent types answer] – ita_f_origin Father’s region of birth String [Respondent types answer] – ita34 Tenendo in considerazione le esigenze dello Stato nell’assicurare il funzionamento dei servizi pubblici, indica con una “x” quale tra le seguenti 4 affermazioni è la più vicina al tuo pensiero. Discrete [Multiple choice radio buttons:] 1 = La funzione dello Stato è quella di fornire a tutti i cittadini il maggior numero di servizi pubblici (es.: scuola, sanità, pensioni, trasporti, ecc.) anche se il livello di tassazione che ne deriva è molto elevato. 2 = Vi sono spese necessarie per il benessere sociale che lo Stato deve sostenere, quindi bisogna fare in modo che le tasse ed imposte coprano queste spese aumentandole ogni qual volta sia necessario. 3 = Il prelievo fiscale è troppo alto dunque se non ci sono soldi a sufficienza, bisogna diminuire le spese riducendo i servizi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

110 Appendix

Notes

4 = Lo Stato dovrebbe prelevare il minimo indispensabile di tasse e imposte per i servizi pubblici veramente essenziali (es.: difesa, giustizia, polizia, ecc.) e lasciare il resto all’iniziativa privata –

Secondo te, tra quelli elencati di seguito, quali potrebbero essere gli strumenti più efficaci per combattere il fenomeno dell’evasione fiscale? (indicare con una “x”; massimo tre risposte). Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options

ita35_1 Inasprire le sanzioni in denaro Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_2 Inasprire le pene detentive Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_3 Aumentare il numero dei controlli Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_4 Adottare in maniera diffusa una tassazione forfetaria per i lavoratori autonomi Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_5 Controllare dettagliatamente tutti i cittadini con un tenore di vita elevato Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  111 Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes Variable name Question text Variable type Answer options Notes

See question introduction above ita35_6 Fare maggiori accertamenti bancari Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_7 Diminuire il numero degli adempimenti previsti Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_8 Aumentare le informazioni che lo Stato dà ai cittadini Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above ita35_9 Limitare l’uso del contante Discrete [Multiple choice radio buttons:] 0 = Not mentioned 1 = Mentioned See question introduction above

UK How often do you read the following newspapers? Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type

uk66_1 The Guardian Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_2 The Times Discrete

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

112 Appendix Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text

[Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_3 The Mirror Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_4 The Daily Mail Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_5 The Independent Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_6 The Observer Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_7 The Telegraph

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  113 Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes Variable name Question text Variable type Answer options

Notes

Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_8 The Sun Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_9 The Morning Star Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_10 The Daily Star Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above uk66_11 The Daily Express Discrete [Multiple choice radio buttons:] 1 = Regularly 2 = Occasionally 3 = Rarely 4 = Never See question introduction above

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

114 Appendix

Appendix 3:  Analysis and Tables Finally, in this appendix, we provide a more technical analysis of our data. In the main text we only examined the difference in means between groups. Although this provides us with a basic understanding of the effects of different experimental parameters and attitudes on tax compliance, it doesn’t allow us to control for variables that could also be driving this tax compliance behavior. In other words, looking at averages only provides us with prima facie evidence that our main variables—institutions, norms, and values—are significant. For the analysis below we use multivariate regressions, to look at the effect of our main independent variable while controlling for other possible confounds. We do this by utilizing simple linear regressions (OLS), since our dependent variable is the average compliance rate which is continuous. We felt comfortable providing only the simplified means comparisons in the main text as our analyses in the main text are robust to all our specifications below.

Redistribution: Rounds 1–3 Below we carry out a series of multivariate Ordinary Least Squares (OLS) to provide our individual level analyses, while controlling for a number of variables. In the first model, which corresponds to section 1 our analysis of our first three rounds of the experiment, we examine the effects of redistribution on tax compliance. Our dependent variable is the average tax compliance rate in rounds 1–3, which we operationalize as reported reported income income divided by earned income (tax compliance rate = earned income ). It ranges from 0 percent to 100 percent and has a standard deviation of .427. Our main independent variables are experimental rounds 1—(dummy variables for each round) and interactions between country and experimental round. We control for gender (dummy: Female = 0; Male = 1), economics majors (dummy: non-­economics = 0; economics = 1), past participation in laboratory experiments (dummy: No = 0; Yes = 1), and we cluster by each individual subject (see Table 2.2 in Chapter 2 for more summary statistics). Table A.1, model 1, presents the results of our full OLS with interactions and controls. From our OLS, we can see that the effect of redistribution on tax compliance is robust to a number of controls. Specifically, going from no redistribution to redistribution increases tax compliance by approximately 9 percentage points, while moving to double redistribution increases compliance by approximately 17 percentage points. Moreover, to examine the interaction effects we keep the US and Round 1 as the reference point. Here the interaction effects are not significant for Round 2, but each of our interactions in Round 3 are significant, meaning that there is a significant increase in compliance moving from Round 1 to Round 3 in Italy, UK, and Sweden, compared to the United States, which confirms our explanation from Figure 3.2 in the main text. Moreover, Figure A.1 below presents the predictive margins for our interaction terms. The predicted margins demonstrate, for example, that our model predicts that while Italians increase their compliance by approximately 23 percentage points moving from Round 1 to Round 3, compliance among Americans only increase by approximately 14 percentage points. For Brits, the result is nearly the same as for Italians, and for Swedes, there is almost a 30-­percentage point increase.

Redistribution: Rounds 4–6 Table A.1, model 2, presents the results of our OLS for the tax rate rounds of our experiment (rounds 4–6). The model is identical to model 1 with the exception that our

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  115 Table A.1  OLS Regressions; DV = Average Compliance Rate Model 1

Model 2

Model 3

R2: Redistribution

0.0889*** (0.0164)

R3: Redistribution × 2

0.168*** (0.0174)

Italy

0.0249 (0.0309)

0.129*** (0.0293)

−0.00789

UK

−0.143***

−0.0610**

−0.0750***

Sweden

−0.0362 (0.0301)

0.0848*** (0.0282)

(0.0277)

R2: Redistribution # Italy

−0.00231 (0.0279)

R2: Redistribution # UK

−0.00716 (0.0235)

R2: Redistribution # Sweden

0.0368 (0.0268)

R3: Redistribution × 2 # Italy

0.0596** (0.0300)

R3: Redistribution × 2 # UK

0.0596** (0.0275)

R3: Redistribution × 2 # Sweden

0.116*** (0.0296)

(0.0280)

R5: 30% Tax Rate

−0.0167 (0.0163)

R6: 50% Tax Rate

−0.0785*** (0.0183)

R5: 30% Tax Rate # Italy

−0.110*** (0.0271)

R5: 30% Tax Rate # UK

−0.0425* (0.0247)

R5: 30% Tax Rate # Sweden

−0.0532** (0.0251)

R6: 50% Tax Rate # Italy

−0.130*** (0.0310)

R6: 50% Tax Rate # UK

−0.00991 (0.0288)

R6: 50% Tax Rate # Sweden

−0.0697** (0.0298)

R8: Progressive 2

(0.0307) (0.0274) 0.0270 (0.0289)

−0.0162 (0.0124) Continued

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

116 Appendix Table A.1  Continued Model 1

Model 2

Model 3

R8: Progressive 2 # Italy

0.0570** (0.0229)

R8: Progressive 2 # UK

0.000786 (0.0193)

R8: Progressive 2 # Sweden

0.0409* (0.0217) 0.744***

Constant

0.827***

(0.0216) Controls Observations

0.817***

(0.0203)

Yes 4821

(0.0210)

Yes 4830

Yes 3222

Standard errors in parentheses Note: Standard errors clustered at the individual level * p < 0.10, ** p < 0.05, *** p < 0.01

No Redistribution

Redistribution

Redistribution x 2

.8

76.4%

76.2%

Average Compliance Rate

.6

68.4%

62%

64%

79.9%

61.7%

59.5%

53.4%

.4

47% 51.4% 38.9%

.2

0 Italy

UK

US Sweden Italy

UK

US Sweden Italy

UK

US Sweden

Figure A.1  Predictive Margins for Country Interactions for Compliance Rate Rounds 1–3

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  117 10% Tax Rate

30% Tax Rate

50% Tax Rate

.8 76.2% 70.8%

Average Compliance Rate

.6

75.6 69.1%

63.6%

62.9%

59.4% 53.5%

.4

68.6% 60.8%

55.4% 50.6%

.2

0 Italy

UK

US Sweden Italy

UK

US Sweden Italy

UK

US Sweden

Figure A.2  Predictive Margins for Country Interactions for Compliance Rate Rounds 4–6 de­pend­ent variable is now the average compliance rate for rounds 4–6, and our main independent variables are dummy variables for rounds 4–6. Again, we also include interactions for country and round, with the US as our reference country. From our OLS, we can see that the effect of increasing the tax rates from 10 percent to 50 percent on tax compliance is significant, negative and robust to a number of controls. Compared to the 10 percent tax rate round, increasing compliance to 50 percent decreases the average compliance rate by approximately 8 percentage points. However, there is not a significant effect on compliance going from 10 percent to 30 percent. Here the interactions tell us that the Italians and Swedes reduce their compliance significantly more as the tax rate increases from 10 percent to 30 percent compared to our American subjects. The predictive margins in Figure  A.2 demonstrate that Americans barely decrease their compliance with a 20-­percentage point increase in the tax rate, while Italians decrease their compliance by about 13 percentage points and Swedes approximately 7 percentage points. There is not a significant difference between our American and British subjects between rounds 4 and 5. As expected, the effect is larger when comparing the 10 percent round to the 50 percent round. The coefficient is twice as large (though post hoc testing shows this is not statistically different) for Italy than Sweden, which translates into about a 21 percentage point reduction in compliance for Italians and only a 15 percentage point reduction for Swedes (see Figure A.2). For both Americans and Brits, it is approximately a 9 percentage point reduction in compliance. Again, the differences between the UK and the US are not statistically significant.

Redistribution: Rounds 7 and 8 For our final analysis of the institutional rounds, we turn to Table  A.1, model 3. The de­pend­ent variable is now the average compliance rate for rounds 7 and 8, and our main

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

118 Appendix Progressive 1

Progressive 2

.8

Average Compliance Rate

.6

67%

65.1%

65.3%

61.2%

67.7%

57.1% 51.4%

52.9%

.4

.2

en Sw ed

S U

K U

ly Ita

en Sw ed

S U

K U

Ita

ly

0

Figure A.3  Predictive Margins for Country Interactions for Compliance Rate Rounds 7 and 8 independent variables are dummy variables for round 7 and 8. We also include interactions for country and round, with the US as our reference country. From our OLS, we can see that the effect of changing the tax regime on tax compliance is not significant. Our only significant effect is on the coefficient corresponding to the interaction between Round 8 and Italy. This tells us that Italians increase their compliance significantly more with the change in tax regimes compared to Americans. Post hoc tests tell us that Italians also increase their compliance significantly more than Brits, who like Americans, decrease their compliance with the change in regime. Specifically, looking at Figure A.3, we can see that the predicted compliance rate for Italians decreases by about 4 percentage points when changing tax regimes, while for Americans it actually decreases by 2 percentage points and for Brits it decreases by about 1 percentage point. The interactions aren’t significant for the other countries.

Income Table A.2, model 1, presents the results for the effects of income on tax compliance. The dependent variable is the average compliance rate for all rounds and individuals. Again, we control for gender, economics majors, past-­participation in experiments, round and country fixed-­effects, and we cluster our standard-­errors by subject. The model confirms Figure 3.8 in the main text; that is, as income increases, compliance significantly decreases. Specifically, being “rich” decreases compliance by 16 percentage points compared to being poor with all else being equal.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Appendix  119 Table A.2  OLS Regressions; DV = Average Compliance Rate Model 1 Middle

−0.0710*** (0.0303)

Rich

−0.152*** (0.0396)

Model 2

Some

−0.143*** (0.0192)

Little

−0.336*** (0.0233)

Model 3

Model 4

0.00732*** (0.000509)

SVO angle

−0.111*** (0.022)

Economics Major Constant

0.894*** (0.0322)

0.911*** (0.0287)

0.489*** (0.0220)

Round FE Country FE Controls Observations

Yes Yes Yes 1611

Yes Yes Yes 14448

Yes Yes Yes 14484

0.690*** (0.0165)  Yes Yes Yes 14484

Standard errors in parentheses Note: Standard errors clustered at the individual level * p < 0.10, ** p < 0.05, *** p < 0.01

Norms We now move on to discuss norms. Table  A.2, model 2 presents the results of our OLS model for norms. The model is the same as in model 1, except our main independent variable is “think others report,” ranging from 1 (All), 2 (Some), 3 (Little). Again, we control for gender, economics majors, past-­participation in experiments, round and country fixed-­ effects, and we cluster our standard-­errors by subject. The model shows that those who think people report much less than their total income are significantly less compliant than those who think others report all of their income. The effect is not just statistically significant; it represents an enormous effect. Specifically, thinking others reported much less (little) than their total income reduces compliance by 33 percentage points compared to those who think others report their total income (all). Interestingly, the effect is approximately half of that for those who think others report some of their income.

Values Table A.2, model 3 is the same as in model 1, except our main independent variable is our SVO angle, which is a continuous variable ranging from -16 (competitive) to 61 (altruistic) (see Chapter  2 for an explanation of the SVO-­angle). Again, we control for gender, economics majors, past-­participation in experiments, round and country fixed-­effects, and we cluster our standard-­errors by subject. The model shows that as the SVO-­angle increases,

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

120 Appendix compliance also increases. The coefficient on the SVO-­angle variable tells us that a 1 degree increase in the SVO-­angle increases compliance 1 percentage point. Substantively, this means that our model predicts that someone with an SVO-­angle of 10 would have a compliance rate of approximately 57 percent, whereas someone with an SVO-­angle of 30 would have a compliance rate around 65 percent.

Economics Majors Finally, Table A.2, model 4, examines the effect of being an economics major on tax compliance. Our main independent variable is whether the student is an economics major or another type of major. Again, we control for gender and past-­participation in experiments, round and country fixed-­effects, and we cluster our standard-­errors by subject. The model shows that being an economics major has both a significant and negative effect on tax compliance. The coefficient on the economics major variable tells us that being an economics major, decreases tax compliance by approximately 11 percentage points compared to non-­economics majors.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Endnotes Preface and Acknowledgments 1. Funding from the European Research Council under the European Union’s Seventh Framework Programme/ERC Grant Agreement n. 295675. The content of this publication reflects the authors’ views and the EU is not liable for any use that may be made of the information contained herein.

Introduction 1. In fact, the former Prime Minister, Silvio Berlusconi, was being tried for tax fraud at the time and no one doubted that he was guilty. As the New York Times reported, “Italians have been accused by some of making tax evasion a national sport” (Elisabetta Povoledo, NYT, Aug. 8, 2011). Similarly, The Economist (In Flagrant Sep. 4, 2012) writes about the Greeks that “eradicating a culture of tax evasion is no small task.” 2. In Sweden, for example the government estimates that it loses roughly 5% of potential tax revenue through what’s called the “Tax Gap.” The Bank of Italy estimates that nearly 32% of public revenue is lost in its “Tax Gap.” See Pisani and Vega Pansini 2017 and Skatteverket 2014. 3. The findings discussed in this book explore four countries (Britain, Sweden, Italy, and the USA). The reader may note that we included Romania in the Willing to Pay? Project and the Leap of Faith book. Romania was added rather later in the original design and though we did run several of our experiments in Bucharest, we came to the unfortunate conclusion that the results from these sessions were not directly comparable with those from the other countries—at least for the purposes of this book. Our collaborators in Romania, Arpad Tudor and Clara Volintiru, were fabulous collaborators and did heroic work setting up and conducting experiments in Bucharest. Unlike the other countries studied here, neither John nor Sven speak Romanian. As a result, we could not effectively monitor the experimental session in Romania as we have personally done for every other session conducted. The final result is that while the findings are largely similar to the findings from our other experiments, there are significant enough issues or questions that we decided to not include them in the present analysis.

Chapter 1 1. See for example, (Monroe  2001; Ostrom  1998; Tversky and Kahneman  1974; Weingast 2005; Yee 1997; Boyd and Richerson 2005; Cook and Levi 1990; Levi 1997).

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

122 Endnotes 2. Of course, what is in a person’s self-­interest can be quite complicated and is obviously not simply a cost/benefit calculation about whether they “earn” more or less by following the rules. People have long-­term and short-­term interests, are embedded in many different social/political/economic networks, have reputations for integrity and honesty to protect, etc. 3. There are numerous studies and scholars who look at differences in social norms between countries. See for example: Andrighetto et al.  2013; Hirschfeld and Gelman 1994; Jahoda 1993; Logan and Ganster 2007; Shore 1996. 4. Latest estimates for the Italian tax gap use data from audits conducted between 2010 and 2014 (Pisani and Vega Pansini 2017); UK tax gap is estimated for 2017–18 and published in 2019 (HMRC 2019); The most recent report from Skatteverket uses data from 2007 to 2012 published in 2014 (Skatteverket 2014); the US released its latest tax gap estimations in 2017 using tax data from 2011 to 2013 (IRS  2019). It should be noted that these measures are estimates and are not without political bias. For example, the UK does not consider revenue losses from multinational’s transfer pricing arrangements, which likely underestimates their tax gap; see Eurostat (2021) for annual deficits. 5. The experimental design is described more fully in Chapter 2. The entire experimental design is presented in the Appendix.

Chapter 2 1. It should be noted that tax morale is generally drawn from attitudinal surveys, whereas tax compliance experiments elicit revealed preferences regarding tax compliance. 2. It is also important to note that at least one researcher from the project was at every experimental session, and our instructions were read by a native speaker, usually from the region in which the experiments were conducted. 3. We aimed the rate of compensation to be roughly twice the average hourly wage in the location in which we were conducting the experiment. 4. The specific charity that received these funds varied according to locations. In the US and Sweden, the money was donated to the Red Cross, in Britain it went to Oxfam, and in Italy these “tax” monies were donated to UNICEF. The participants were assured that the money really would be donated to these charities. 5. The last nine decisions determined whether prosocial behavior was driven by inequality aversion or joint gain maximization. Since the motivation for prosocial behavior is not relevant for this research question, these decisions are omitted from Table 2 and the subsequent analysis.

Chapter 3 1. Smith, A., 1976, The Glasgow edition, Vol. 2a, pp. 26–7. 2. Gërxhani and Schram 2006.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Endnotes  123 3. Alm et al. 1992a; Alm et al. 1999; Alm and Malézieux 2020; Friedland 1982; Kirchler et al. 2003; Spicer and Hero 1985. 4. Alm et al.  1992a; Alm et al.  1999; Alm et al.  1995b; Beck et al.  1991; Collins and Plumlee  1991; Cummings et al.  2009; Kirchler et al.  2003; Park and Hyun  2003; Webley 1987. 5. As Kastluner et al. have shown, penalties can cut both ways, however. “Both coercive power and legitimate power were correlated with enforced compliance. However, the effect of enforced compliance leads to increased evasion. The results evidence the multifaceted nature of power and trust and their relation with tax compliance, and the importance of power and trust in political regulatory strategies” (Kastlunger et al. 2013, 36). See also Bergman 2009; Bensel 1990; Bergström 1987; Berns et al. 2005. 6. Alm et al. 2006; Alm et al. 1995a; Alm and Torgler 2011; Andrighetto et al. 2016. 7. Andreoni et al. 1998; Erard and Feinstein 1994. 8. In our experiments we chose to hold the audit rates and penalties for evasion constant so that we could more effectively compare behaviors across societies. 9. Several studies which have linked tax compliance directly to real public goods. Doerrenberg (2015) and Mittone (2006) demonstrate, for example, that peoples’ compliance increases when they get something back from the public good or their tax money pays for a real public good. See also Alm et al. 1992a; Bosco and Mittone 1997; Gërxhani and Schram 2006. 10. For the neutral experiment, the pay-­off and decision was numerically equivalent to the framed experiment, but there was no reference at all to tax. For example, subjects had two accounts, and they had to decide how much money to allocate to each account, but the accounts differ in their returns. In account A there was a 100% percent chance that subjects would lose .25 currency units, similar to if subjects put that money in a public fund. However, in Account B, there was an 80% chance the subject would keep all his/her money (free rider), but a 20% chance that he/she would lose .5 ECU, like if they were caught cheating. 11. We ran a test of this thesis by replacing the word “tax” with “fee” in one set of experiments run in the UK. We found that compliance was significantly lower in all rounds when the subjects were asked to pay a fee instead of a tax. See https://themonkeycage. org/2015/04/the-­word-­tax-­is-­unpopular-­but-­the-­word-­fee-­is-­worse-­2/ for a closer look at this, but on average people who received the “Tax” treatment were about 10 percentage points more compliant that in the “Fee” treatment. 12. In our post-­experiment surveys, we found that subjects who tended to be very honest throughout the experiment responded that they felt that it was a moral obligation to pay taxes. Those who evaded their taxes in the experiment (especially in the first round) often commented that they were simply trying to make as much money as they could and did not feel it was an obligation to pay taxes. 13. For a more technical elaboration of these findings, see Andrighetto et al. 2016. 14. See especially Daunton 2018; Stanley 2018. 15. Kleven (2014) demonstrates precisely this in his study of third-­party reporting vs. self-­reporting in Denmark. They find that even in Denmark, individuals who have some self-­reported income and some third-­party reported income are significantly more likely to evade on their self-­reported income.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

124 Endnotes 16. There is also reason to believe that compliance amongst our subjects in the UK was depressed by the fact that the British government had recently increased student fees by 300 percent in the year before we began our study. We do not have definitive proof of this, but there was clearly a general sense in the UK at the time (2013–16) that students were being “screwed.” This variable speaks to one of the central difficulties of laboratory experiments. Even when the researcher does her best to set up comparable experiments, there are inevitably external variables that simply cannot be controlled for in the laboratory. We return to this issue in our epilogue. 17. The interactions with the US as our base country are statistically significant at .05. 18. See the online appendix for predictive margins, specifically Table 3.1 and 3.2. 19. See Pampel et al. 2019 and D’Attoma 2020 for this. 20. Several studies also confirm this (Alm et al.  1992a; Baldry  1987; Collins and Plumlee 1991; Friedland et al. 1978; Bernasconi et al. 2014; Alm et al. 1992b). 21. This is similar to Alm et al.’s findings (1992b), who find a 17 percentage point decrease in compliance, but not Alm et al. (2009), who report a 12 percentage point increase when the tax rate increased from 35 percent to 50 percent. It’s important to note, however, that most tax systems are not designed like this, and most people don’t pay average tax rates of 50%. 22. To take but one example: In a group of twenty-­five subjects, if the average income was 20 ECU, at a 10 percent tax rate approximately 70 percent of earned income was reported. Therefore, the total pot would be 38. At the 50 percent tax rate, if 57.8 percent of income was reported the total amount in the pot would be 142 ECU. 23. See, for example, Lewis 1978, 1982. 24. All other rules regarding redistribution, audit, and penalties remained the same as they had in earlier stages of the experiment. 25. Block 1986; Ellis 1940. 26. As part of this project, we also conducted a public opinion survey using an interactive budget simulation. See Tuxhorn et al. 2021 for a working paper. Our research finds that most Americans support progressive taxation, which confirms Ballard-­Rosa et al. 2017. 27. Alm et al.  1999; Blumenthal et al.  2001; Frey and Torgler  2007; Traxler  2010; Wenzel 2004, 2005; Weber et al. 2014; Onu and Oats 2015. 28. It is widely recognized that the most effective punishment one can apply in the modern prison is to isolate the individual prisoner and prevent him or her from having contact with other individuals. Some consider this punishment “inhumane.” 29. The starting point for social identity research was set by Tajfel et al.’s (1971) “minimal group paradigm,” a series of experiments showing that individuals tend to favor their own “in-­group” at the detriment of an “out-­group”—even if the groups consist of unknown others and membership to groups is randomly assigned. 30. Wenzel (2004) has argued that tax compliance can either increase or decrease tax compliance depending on the norms with which the individual identifies. Since paying taxes is a private act, typically citizens do not have exact information about the behavior of other taxpayers nor clear indirect measures from which to infer it, and this opacity may negatively affect people’s willingness to pay taxes. For example, it might be more efficient to announce increased auditing efforts as a means “to further

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

Endnotes  125

31.

32.

33.

34.

improve the high level of tax compliance” rather than as a response to “a recent increase in tax evasion” (Traxler 2010). Wittberg (2010) explains that this is because the Swedish tax authority has embraced the behavioral evidence, like that presented here, which suggests that people want to do what other’s do and want to behave in socially appropriate ways. Therefore, the Swedish tax authority has eschewed the punishment/penalty model in favor of presenting themselves as a “service” organization. Their approach is that they want to help people pay their correct taxes, rather than punish people that cheat. Of course, they still try to capture evaders, but their evidence shows that they actually collect much more revenue when they try to aid taxpayers rather than frighten them. Tax “morale” has been a major topic for students of tax compliance for the past several years. Alm and Torgler (2004, 269) define it as: “the intrinsic motivation to pay taxes.” A number of scholars quite early on suggested that there is more to paying taxes than the basic rational choice models considered such as taxpayers’ own welfare, the welfare of others (Bosco and Mittone 1997), shame, and guilt (Erard and Feinstein 1994). Above we surveyed the literature on tax morale, which suggests that institutions, culture, and basic demographic characteristics are all determinants of this intrinsic motivation to pay taxes. Note that the SVO allows subjects to give small amounts to increase the payout to the other in different ways. This avoids the problem identified by Fehr and Gächter who argue that even an altruistic person will not normally incur heavy costs (Fehr and Gächter 2000). In the original model of the Murphy et al. versions of this experiment, the individual’s responses are calculated into an absolute figure that then is placed in a Venn Diagram to show the degree of altruism or individualism of each individual.

Conclusion 1. A copy of Trump’s 2005 tax return was leaked to the press, and in that year he paid 25 percent of his income in taxes. This was nearly $38,000,000 (Siddiqui et al. 2017). It is widely suspected that these returns were real precisely because it was one of the few years in which he did actually pay taxes. But then again, of course, no one knows. 2. Eighty-­six percent of Americans believe that it is a moral duty to share in the costs of government—even when many of these same voters believe the costs of government are too high and or that much of that money is wasted (Diaz 2016). 3. For an excellent discussion of the differences between formal and informal institutions, see Hodgson 2006. 4. Nistotskaya and D’Arcy 2018.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

References Abelson, Robert  P., and Alex Bernstein. 1963. “A Computer Simulation Model of Community Referendum Controversies.” Public Opinion Quarterly 27 (1): 93–122. Allingham, Michael, and Agnar Sandmo. 1972. “Income Tax Evasion: A Theoretical Analysis.” Journal of Public Economics 1 (3–4): 323–38. Alm, James. 2010. “Testing Behavioral Economics Theories in the Laboratory.” National Tax Journal 63 (4): 535–658. Alm, James, John Deskins, and Michael McKee. 2009. “Do Individuals Comply on Income Not Reported by Their Employer?” Public Finance Review 37 (2): 120–41. Alm, James, Betty  R.  Jackson, and Michael McKee. 1992a. “Deterrence and Beyond: Toward a Kinder, Gentler IRS.” In Why People Pay Taxes: Tax Compliance and Enforcement, ed. Joel Slemrod: 311–29. Ann Arbor: University of Michigan Press. Alm, James, Betty R. Jackson, and Michael McKee. 1992b. “Estimating the Determinants of Taxpayer Compliance with Experimental Data.” National Tax Journal 45 (1): 107–14. Alm, J. and Malézieux, A., 2020. 40 Years of Tax Evasion Games: A Meta Analysis. Experimental Economics, 1–52. Alm, James, Gary McClelland, and William Schulze. 1999. “Changing the Social Norm of Tax Compliance by Voting.” Kyklos 52 (2): 141–71. Alm, James, Jorge Martinez-Vazquez, and Mark Rider. 2006. The Challenges of Tax Reform in a Global Economy. New York, NY: Springer Science+Business Media. Alm, James, Isabel Sanchez, and Ana De Juan. 1995a. “Economic And Noneconomic Factors In Tax Compliance.” Kyklos 48 (1): 3–18. Alm, James, and Benno Torgler. 2004. “Estimating the Determinants of Tax Morale.” Proceedings. Annual conference on taxation and Minutes of the annual meeting of the National Tax Association, vol. 97: 269–74. Minneapolis: National Tax Association. Alm, James, and Benno Torgler. 2006. “Culture Difference and Tax Morale in United States and Europe.” Journal of Economic Psychology 27 (2): 224–46. Alm, James, and Benno Torgler. 2011. “Do Ethics Matter? Tax Compliance and Morality.” Journal of Business Ethics 101 (4): 635–51. Andreoni, James, Brian Erard, and Jonathan Feinstein. 1998. “Tax Compliance.” Journal of Economic Literature 36 (2): 818–60. Andrighetto, Giulia, Jordi Brandts, Rosaria Conte, Jordi Sabater-Mir, Hector Solaz, and Daniel Villatoro. 2013. “Punish and Voice; Punishment enhances cooperation when combined with norm-signalling.” PloS one 8 (6): e64941. Andrighetto, Giulia, Nan Zhang, Stefania Ottone, Ferruccio Ponzano, John D’Attoma, and Sven Steinmo. 2016. “Are Some Countries More Honest Than Others? Evidence from a Tax Compliance Experiment in Sweden and Italy.” Frontiers in Psychology 7. DOI: 10.3389/fpsyg.2016.00472. Baldry, Jonathan. 1987. “Income Tax Evasion and the Tax Schedule: Some Experimental Results.” Public Finance = Finances Publiques 42 (3): 357–83. Ballard-Rosa, Cameron, Lucy Martin, and Kenneth Scheve. 2017. “The Structure of American Income Tax Policy Preferences.” The Journal of Politics 79 (1): 1–16. Banfield, Edward. 1958. The Moral Basis of a Backward Society.: Glencoe: Free Press.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

128 References Baumeister, Roy  F. 2011. “Need-to-belong theory.” In Handbook of Theories of Social Psychology: Volume Two, ed. P. A. M. Van Lange, A. W. Kruglanski, and E. T. Higgins. London: Sage. Baumeister, Roy F., and Mark R. Leary. 1995. “The Need to Belong: Desire for Interpersonal Attachments as a Fundamental Human Motivation.” Psychological Bulletin 117 (3): 497. Bazart, Cécile, and Aurélie Bonein. 2014. “Reciprocal Relationships in Tax Compliance Decisions.” Journal of Economic Psychology 40: 83–102. Beck, Paul J., Jon S. Davis, and Woon-Oh Jung. 1991. “Experimental Evidence on Taxpayer Reporting Under Uncertainty.” Accounting Review 66 (3): 535–58. Becker, Winfried, Heinz-Jürgen Büchner, and Simon Sleeking. 1987. “The impact of public transfer expenditures on tax evasion: An experimental approach.” Journal of Public Economics Sleeking 34 (2): 243–52. Beer, Sebastian, Matthias Kasper, Erich Kirchler, and Brian Erard. 2019. Do Audits Deter or Provoke Future Tax Noncompliance? Evidence on Self-employed Taxpayers. Washington, DC: International Monetary Fund. Bensel, Richard Franklin. 1990. Yankee Leviathan: The Origins of Central State Authority in America, 1859–1877. New York: Cambridge University Press. Bergman, Marcelo. 2009. Tax Evasion and the Rule of Law in Latin America. University Park, PA: University of Pennsylvania Press. Bergström, Hans. 1987. Rivstart?: om övergången från opposition till regering. Stockholm: Tiden. Bernasconi, Michele, Luca Corazzini, and Raffaello Seri. 2014. “Reference Dependent Preferences, Hedonic Adaptation and Tax Evasion: Does the Tax Burden Matter?” Journal of Economic Psychology 40: 103–18. Berns, Gregory  S., Jonathan Chappelow, Caroline  F.  Zink, Giuseppe Pagnoni, Megan E. Martin-Skurski, and Jim Richards. 2005. “Neurobiological Correlates of Social Conformity and Independence During Mental Rotation.” Biological psychiatry 58 (3): 245–53. Bicchieri, Christina. 2006. The Grammar of Society: The Nature and Dynamics of Social Norms. New York: Cambridge University Press. Block, H+R. 1986. The American Public and the Federal Income Tax System. Kansas City: H+R Block. Blumenthal, Marsha, Charles Christian, Joel Slemrod, and Matthew Smith. 2001. “Do Normative Appeals Affect Tax Compliance? Evidence from a Controlled Experiment in Minnesota.” National Tax Journal 54 (1): 125–38. Bolton, Gary  E., and Axel Ockenfels. 2000. “ERC: A Theory of Equity, Reciprocity, and Competition.” American economic review 90 (1): 166–93. Bosco, Luigi, and Luigi Mittone. 1997. “Tax Evasion and Moral Constraints: Some Experimental Evidence.” Kyklos 50 (3): 297–324. Boyd, Robert and Richerson, Peter J. 2005. The Origin and Evolution of Cultures. Oxford: Oxford University Press. Braithwaite, Valerie, and Eliza Ahmed. 2005. “A threat to tax morale: The case of Australian higher education policy.” Journal of Economic Psychology 26 (4): 523–40. Carter, John R., and Michael D. Irons. 1991. “Are Economists Different, and If So, Why?” Journal of Economic Perspectives 5 (2): 171–7. Charron, N., V. Lapuente, and P. Annoni. 2019. “Measuring Quality of Government in EU Regions Across Space and Time.” Papers in Regional Science 98 (5). DOI: 10.1111/ pirs.12437. Cialdini, Robert  B., and Noah  J.  Goldstein. 2004. “Social influence: Compliance and Conformity.” Annual review of psychology 55: 591–621.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

References  129 Cialdini, Robert  B., and Melanie  R.  Trost. 1998. “Social Influence: Social Norms, Conformity and Compliance.” In The Handbook of Social Psychology, ed. D. T. Gilbert, S. T. Fiske, and G. Lindzey. Boston: McGraw-Hill. Collins, Julie  H., and R.  David Plumlee. 1991. “The Taxpayer’s Labor and Reporting Decision: The Effect of Audit Schemes.” Accounting Review 66 (3): 559–76. Cook, Karen S., and Margaret Levi. 1990. The Limits of Rationality. Cambridge: Cambridge University Press. Coughlin, Richard. 1980. Ideology, Public Opinion & Welfare Policy: Attitudes Toward Taxes and Spending in Industrialized Societies. Berkeley: Institute of International Studies, University of California. Cummings, R. G., J. Martinez-Vazquez, M. McKee, and B. Torgler. 2004. “Effects of Culture on Tax Compliance: A cross check of experimental and survey evidence.” In CREMA Working Paper Series 2004–13. Basel: Center for Research in Economics, Management and the Arts. Cummings, Ronald, Jorge Martinez-Vazquez, Michael McKee, and Benno Torgler. 2009. “Tax Morale Affects Tax Compliance: Evidence from Surveys and an Artefactual Field Experiment.” Journal of Economic Behavior & Organization 70 (3): 447–57. Darwin, Charles. 1871. The Descent of Man: And Selection in Relation to Sex. London: J. Murray. D’Attoma, John. 2020. “More bang for your buck: tax compliance in the United States and Italy.” Journal of public policy 40 (3): 1–24. Daunton, Martin. 2018. “Creating Consent: Taxation, War, and Good Government in Britain, 1688–1914.” In The Leap of Faith: The Fiscal Foundations of Successful Government in Europe and America, ed. S. Steinmo. Oxford: Oxford University Press. Diaz, Daniella. 2016. “8 in 10 Trump backers say paying taxes is a ‘civic duty’.” CNN Politics. Available from: https://edition.cnn.com/2016/10/03/politics/cnn-orc-poll-taxes-donald-trump-2016-election/index.html (accessed June 1, 2021). Doerrenberg, Philipp. 2015. “Does The Use of Tax Revenue Matter for Tax Compliance Behavior?” Economics Letters 128: 30–4. Edlund, Jonas. 1999. “Attitudes towards Tax Reform and Progressive Taxation: Sweden 1991–96.” Acta Sociologica 42 (4): 337–55. Ellemers, Naomi, Bertjan Doosje, and Russell Spears. 2004. “Sources Of Respect: The Effects of Being Liked by Ingroups and Outgroups.” European Journal of Social Psychology 34 (2): 155–72. Ellis, Elmer. 1940. “Public Opinion and the Income Tax, 1860–1900.” Mississippi Valley Historical Review 27: 225–42. Elster, Jon. 1989. The Cement of Society: A Study of Social Order. Cambridge and New York: Cambridge University Press. Elster, Jon. 1993. Political Psychology. Cambridge and New York: Cambridge University Press. Erard, Brian, and Jonathan Feinstein. 1994. “The Role of Moral Sentiments and Audit Perceptions in Tax Compliance.” Public Finance 49: 70–89. Eurostat. 2021. “Government finance statistics”. https://ec.europa.eu/eurostat/statisticsexplained/index.php/Government_finance_statistics_-_quarterly_data (accessed June 1, 2021). Fehr, Ernst. 2006. “Inequality Aversion, Efficiency and Maximum Preferences in Simple Distribution Experiments: Comment.” American Economic Review 96 (5):1912–17. Fehr, Ernst, and Urs Fischbacher. 2004. “Third-Party Punishment and Social Norms.” Evolution and Human Behavior 25: 63–87.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

130 References Fehr, Ernst, and Simon Gächter. 2000. “Cooperation and Punishment in Public Goods Games.” American Economic Review 90 (4): 980–94. Fehr, Ernst, and Klaus  M.  Schmidt. 1999. “A Theory of Fairness, Competition, and Cooperation.” The Quarterly Journal of Economics 114 (3): 817–68. Festinger, Leon. 1957. A Theory of Cognitive Dissonance. Stanford: Stanford University Press. Fischbacher, Urs, Simon Gächter, and Ernst Fehr. 2001. “Are People Conditionally Cooperative? Evidence From a Public Goods Experiment.” Economics Letters 71 (3): 397–404. Frey, Bruno  S., and Benno Torgler. 2007. “Tax Morale and Conditional Cooperation.” Journal of Comparative Economics 35 (1): 136–59. Friedland, Nehemiah. 1982. “A Note on Tax Evasion as a Function of the Quality of Information about the Magnitude and Credibility of Threatened Fines: Some Preliminary Research.” Journal of Applied Social Psychology 12 (1): 54–9. Friedland, Nehemiah, Shlomo Maital, and Aryeh Rutenberg. 1978. “A Simulation Study of Income Tax Evasion.” Journal of Public Economics 10 (1): 107–16. G4S. 2018. “The World Cash Report.” https://www.g4scashreport.com/ (accessed June 1, 2021). Gentry, William M., and Matthew E. Kahn. 2009. “Understanding Spatial Variation in Tax Sheltering: The Role of Demographics, Ideology, and Taxes.” International Regional Science Review 32 (3): 400–23. Gërxhani, Klarita, and Arthur Schram. 2006. “Tax Evasion And Income Source: A Comparative Experimental Study.” Journal of Economic Psychology 27 (3): 402–22. Gintis, Herbert, Samuel Bowles, Robert Boyd, and Ernst Fehr, eds. 2006. Moral Sentiments and Material Interests: The Foundations of Cooperation in Economic Life (Economic Learning and Social Behavior). Cambridge, MA: MIT Press. Greiner, Ben. 2015. “Subject Pool Recruitment Procedures: Organizing Experiments with Orsee.” Journal of the Economic Science Association 1 (1): 114–25. Halman, Loek. 2009. “Value Change in Western European Societies: Results from the European Values Study.” Kwansei Gakuin University School of Sociology Journal 107: 35–48. Hayek, Friedrich A. von, and Arthur A. Shenfield. 1983. Knowledge, Evolution and Society. London: Adam Smith Institute. Heinemann, Friedrich, and Martin Kocher. 2013. “Tax Compliance Under Tax Regime Changes.” International Tax 20 (2): 225–46. Herrmann, Benedikt, Christian Thöni, and Simon Gächter. 2008. “Antisocial Punishment Across Societies.” Science 319 (5868): 1362–7. Hirschfeld, Lawrence  A., and Susan  A.  Gelman. 1994. Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press. HMRC. 2019. HMRC exchange rates for 2019: monthly. Published December 27, 2018. Last updated November 25, 2019. https://www.gov.uk/government/publications/hmrcexchange-rates-for-2019-monthly (accessed June 1, 2021). Hobbes, Thomas. 1909. Hobbes’s Leviathan: reprinted from the edition of 1651. Oxford: Clarendon Press. Hodgson, Geoffrey. 2006. “What Are Institutions?” Journal of Economic Issues 40 (1): 1–25. Hornsey, Matthew  J., Jolanda Jetten, Brendan J McAuliffe, and Michael  A.  Hogg. 2006. “The Impact of Individualist and Collectivist Group Norms on Evaluations of Dissenting Group Members.” Journal of Experimental Social Psychology 42 (1): 57–68. IRS. 2019. “The Tax Gap.” Available from: https://www.irs.gov/newsroom/the-tax-gap (accessed June 1, 2021).

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

References  131 Jahoda, Gustav. 1993. “The Colour of a Chameleon: Perspectives on Concepts of Culture.” Cultural Dynamics 6 (3): 277–87. Janes, Leslie  M., and James  M.  Olson. 2000. “Peer Pressure: The Behavioral Effects of Observing Ridicule of Others.” Personality and Social Psychology bulletin 26 (4): 474–85. Kahneman, Daniel, Paul Slovic, and Amos Tversky, eds. 1982. Judgement Under Uncertainty: Heuristics and Biases. New York: Cambridge University Press. Kahneman, Daniel. 2011. Thinking, Fast and Slow. New York: Farrar, Straus, and Giroux. Kaplan, Martin  F., and Charles  E.  Miller. 1987. “Group Decision Making and Normative Versus Informational Influence: Effects of Type of Issue and Assigned Decision Rule.” Journal of Personality and Social Psychology 53 (2): 306. Kastlunger, Barbara, Edoardo Lozza, Erich Kirchler, and Alfred Schabmann. 2013. “Powerful Authorities and Trusting Citizens: The Slippery Slope Framework and Tax Compliance in Italy.” Journal of Economic Psychology 34: 36–45. Kastlunger, Barbara, Stephan Muehlbacher, Erich Kirchler, and Luigi Mittone. 2011. “What Goes Around Comes Around? Experimental Evidence of the Effect of Rewards on Tax Compliance.” Public Finance Review 39 (1): 150–67. Kirchler, Erich, Boris Maciejovsky, and Friedrich Schneider. 2003. “Everyday Representations of Tax Avoidance, Tax Evasion, and Tax Flight: Do Legal Differences Matter?” Journal of Economic Psychology 24 (4): 535–53. Kleven, Henrik Jacobsen. 2014. “How Can Scandinavians Tax So Much?” Journal of Economic Perspectives 28 (4): 77–98. Kleven, Henrik Jacobsen, Martin  B.  Knudsen, Claus Thustrup Kreiner, Søren Pedersen, and Emmanuel Saez. 2011.“Unwilling or Unable to Cheat? Evidence From a Tax Audit Experiment in Denmark.” Econometrica 79 (3): 651–92. Kross, Ethan, Marc G. Berman, Walter Mischel, Edward E. Smith, and Tor D. Wager. 2011. “Social Rejection Shares Somatosensory Representations with Physical Pain.” Proceedings of the National Academy of Sciences 108 (15): 6270–5. Lawerence, Choo  C.  Y., Miguel Fonseca, and Gareth Myles. 2016. “Do Students Behave Like Real Taxpayers In The Lab? Evidence from a Real Effort Tax Compliance Experiment.” Journal of Economic Behavior & Organization (124): 102–14. Le Maire, Daniel, and Bertel Schjerning. 2013. “Tax Bunching, Income Shifting and SelfEmployment.” Journal of Public Economics 107: 1–18. Levi, Margaret. 1997. Consent, Dissent and Patriotism. New York: Cambridge University Press. Levi, Margaret. 1988. Of Rule and Revenue. Berkeley: University of California Press. Lewis, Alan. 1978. “Perceptions of Tax Rates.” British Tax Review (6): 358–65. Lewis, Alan. 1982. The Psychology of Taxation. Oxford: Martin Robertson. Lippmann, Walter. 1922. Public Opinion. New York: Macmillan. Logan, Mary S., Daniel C. Ganster. 2007. “The Effects of Empowerment on Attitudes and Performance: The Role of Social Support and Empowerment Beliefs.” Journal of Management Studies 44 (8): 1523–50. McGee, Robert, Simon Ho, and Annie Li. 2008. “A Comparative Study on Perceived Ethics of Tax Evasion: Hong Kong vs the United States.” Journal of Business Ethics 77 (2): 147–58. March, James, and Johan Olsen. 1989. Rediscovering Institutions. New York: Free Press. Marwell, Gerald, and Ruth  E.  Ames. 1981. “Economists Free Ride, Does Anyone Else?: Experiments on the provision of public goods, IV.” Journal of Public Economics 15 (3): 295–310. Mead, Nicole  L., Roy  F.  Baumeister, Tyler  F.  Stillman, Catherine  D.  Rawn, and Kathleen  D.  Vohs. 2011. “Social Exclusion Causes People to Spend and Consume Strategically in the Service Of Affiliation.” Journal of Consumer Research 37 (5): 902–19.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

132 References Milgram, Stanley. 1974. Obedience to Authority: An Experimental View. New York: HarperCollins. Mittone, Luigi. 2006. “Dynamic Behaviour in Tax Evasion: An Experimental Approach.” The Journal of Socio-Economics 35 (5): 813–35. Mittone, Luigi, Fabrizio Panebianco, and Alessandro Santoro. 2016. “The Bomb-Crater Effect of Tax Audits: Beyond Misperception of Chance.” No. 583, Working Papers, IGIER, Bocconi University, Milan. Available from: http://www.igier.unibocconi.it/ files/583.pdf (accessed June 2, 2021). Monroe, Kristen  R. 2001. “Paradigm Shift: From Rational Choice to Perspective.” International Political Science Review 22 (2): 151–72. Moore, Steven. 2001. Sociology Alive. Cheltenham: Nelson Thornes. Murphy, Ryan  O., Kurt  A.  Ackermann, and Michel Handgraaf. 2011. “Measuring Social Value Orientation.” Judgment and Decision Making 6 (8): 771–81. Nistotskaya, Marina, and Michelle D’Arcy. 2018. “Getting to Sweden: The Origins of High Tax Compliance in the Swedish Tax State.” In The Leap of Faith: The Fiscal Foundations of Successful Government in Europe and America, ed. S.  Steinmo. Oxford: Oxford University Press. North, Douglass  C. 2008. "What is Missing from Political Economy." In The Oxford Handbook of Political Economy, ed. D.  Whittman and B.  Weingast. Oxford: Oxford University Press. OECD. 2017. Entrepreneurship at a Glance 2017: OECD Publishing. Olson, Mancur. 2009. The Logic of Collective Action: Public Goods and the Theory of Groups, Second Printing with a New Preface and Appendix. Vol. 124. Cambridge, MA; London: Harvard University Press. Onu, Diana, and Lynne Oats. 2015. “The Role of Social Norms in Tax Compliance: Theoretical Overview and Practical Implications.” Journal of Tax Administration 1 (1): 113–37. Ostrom, Elinor. 1998. “A Behavioral Approach to the Rational Choice Theory of Collective Action: Presidential Address, American Political Science Association, 1997.” The American Political Science Review 92 (1): 1–22. Pampel, Fred, Guilia Andrighetto, and Sven Steinmo. 2019. “How Institutions and Attitudes Shape Compliance: a Cross-Nation Experiment and Survey.” Social Forces 97 (3): 1337–64. Park, Chang-Gyun, and Jin Kwon Hyun. 2003. “Examining the Determinants of Tax Compliance by Experimental Data: A Case of Korea.” Journal of Policy Modeling 25 (8): 673–84. Pisani, Stefano, and Rosaria Vega Pansini. 2017. Bottom-up estimates of Tax Gap by the Italian Revenue Agency. Agenzia delle Entrate. Available from: https://ec.europa.eu/ transparency/regexpert/index.cfm?do=groupDetail.groupDetailDoc&id=38178&no=3. Putnam, Robert  D., Robert Leonardi, and Raffaella Nanetti. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Rawls, John. 2009. A Theory Of Justice. Cambridge, MA: Harvard University Press. Rawls, John, and Erin Kelly. 2001. Justice as Fairness: A Restatement. Cambridge, MA: Harvard University Press. Ross, Adriana  M., and Robert  W.  McGee. 2012. “Education Level and Ethical Attitude toward Tax Evasion: A Six-Country Study.” Journal of Legal, Ethical and Regulatory Issues 15 (2): 93. Rothstein, Bo. 1998. Just Institutions Matter: The Moral and Political Logic of the Universal Welfare State. Cambridge: Cambridge University Press.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

References  133 Rothstein, Bo, and Eric  M.  Uslaner. 2005. “All for All: Equality, Corruption, and Social Trust.” World Politics 58 (1): 41–72. Rubinstein, Ariel. 2006. “Dilemmas of an Economic Theorist.” Econometrica 74 (4): 865–83. Saez, Emmanuel. 2010. “Do Taxpayers Bunch at Kink Points?” American Economic Journal: Economic Policy 2 (3): 180–212. Scholz, John  T., and Mark Lubell. 1998. “Trust and Taxpaying: Testing the Heuristic Approach to Collective Action.” American Journal of Political Science 42 (2): 398–417. Shore, Bradd. 1996. Culture in Mind: Cognition, Culture, and the Problem of Meaning. Oxford: Oxford University Press. Siddiqui, Sabrina, Jon Swaine, and Julia Carrie Wong. 2017. Trump tax returns: president had to pay millions due to tax law he aims to scrap 2017 [cited 26 June 2017]. Available from: https://www.theguardian.com/us-news/2017/mar/14/donald-trump-tax-return-leaked-­ alternative-minimum (accessed June 2, 2021). Skatteverket. 2014. “The development of the tax gap in Sweden 2007–12.” [cited March 2, 2020]. Available from: https://www.skatteverket.se/download/18.15532c7b1442f25 6baeae28/1395223863657/The+development+of+the+tax+gap+in+Sweden+2007-12. pdf (accessed June 2, 2021). Smith, Vernon. 2008. Rationality in Economics: Constructivist and Ecological Forms. New York: Cambridge University Press. Smith, A., 1976, The Glasgow edition, Vol. 2a, pp. 26–7. Spicer, Michael  W., and Rodney  E.  Hero. 1985. “Tax Evasion and Heuristics.” Journal of Public Economics 26 (2): 263–7. Stanley, Liam. 2018. “‘When We Were Just Giving Stuff Away Willy Nilly’: Historicizing Contemporary Tax Morale.” In The Leap of Faith: The Fiscal Foundations of Successful Government in Europe and America, ed. S. Steinmo. Oxford: Oxford University Press. Svallfors, Stefan. 1997. “Worlds of Welfare and Attitudes to Redistribution: A Comparison of Eight Western Nations.” European Sociological Review 13 (3): 283–304. Tajfel, Henri. 1982. Social Identity And Intergroup Relations. Vol. 7. Cambridge: Cambridge University Press. Tajfel, Henri, Michael  G.  Billig, Robert  P.  Bundy, and Claude Flament. 1971. “Social Categorization and Intergroup Behaviour.” European Journal of Social Psychology 1 (2): 149–78. Taylor-Gooby, Peter. 1985. “Attitudes to Welfare.” Journal of Social Policy 14 (1): 73–81. Thibaut, John  W., and Harold  H.  Kelley. 1959. The Social Psychology of Groups. New York: Wiley. Torgler, Benno. 2003. “Does Culture Matter? Tax Morale in an East-West-German Comparison.” FinanzArchiv: Public Finance Analysis 59 (4): 504–28. Torgler, Benno. 2002. “Speaking to Theorists and Searching for Facts: Tax Morale and Tax Compliance in Experiments.” Journal of Economic Surveys 16 (5): 657–83. Torgler, Benno. 2007. Tax Compliance and Tax Morale: A Theoretical and Empirical Analysis. Cheltenham: Edward Elgar Publishing. Torgler, Benno, and Christoph Schaltegger. 2005. “Tax Amnesties and Political Participation.” Public Finance Review 33 (3): 403–31. Torgler, Benno, and Friedrich Schneider. 2007. “What Shapes Attitudes Toward Paying Taxes? Evidence from Multicultural European Countries.” Social Science Quarterly 88 (2): 443–70. Torgler, Benno, and Neven Valev. 2010. “Gender and Public Attitudes toward Corruption and Tax Evasion.” Contemporary Economic Policy 28 (4): 554–68. Traxler, Christian. 2010. “Social Norms and Conditional Cooperative Taxpayers.” European Journal of Political Economy 26 (1): 89–103.

OUP CORRECTED PROOF – FINAL, 02/11/21, SPi

134 References Turner, Frederick. 1990. “The Meaning of Value: A New Economics for the Future.” The New Literary History 21 (3): 747–62. Turner, John  C. 1982. “Towards A Cognitive Redefinition of the Social Group.” In Social Identity and Intergroup Relations, ed. H. Tajfel. Cambridge: Cambridge University Press. Tversky, Amos, and Daniel Kahneman. 1992. “Advances in Prospect Theory: Cumulative Representation of Uncertainty.” Journal of Risk and Uncertainty 5 (4): 297–323. Tversky, Amos, and Daniel Kahneman. 1974. “Judgment under Uncertainty: Heuristics and Biases.” Science (185): 1123–31. Tuxhorn, Kim-Lee, John D’Attoma. and Sven Steinmo. 2021. Assessing the Stability of Fiscal Attitudes: Evidence from a Survey Experiment. Public Administration. Weber, Till Olaf, Jonas Fooken, and Benedikt Herrmann. 2014. “Behavioral Economics and Taxation.” Taxation Papers 41. Brussels: Directorate General Taxation and Customs Union, European Commission. Webley, Paul. 1987. “Audit probabilities and tax evasion in a business simulation.” Economics Letters 25 (3): 267–70. Webley, Paul, Robben Henry, Henk Elffers, and Dick Hessing. 1991. Tax Evasion: An Experimental Approach. Cambridge: Cambridge University Press. Weingast, Barry R. 2005. “The Constitutional Dilemma of Economic Liberty.” The Journal of Economic Perspectives 19 (3): 89–108. Wenzel, Michael. 2004. “An Analysis of Norm Processes in Tax Compliance.” Journal of Economic Psychology 25 (2): 213–28. Wenzel, Michael. 2005. “Motivation or Rationalisation? Causal Relations Between Ethics, Norms and Tax Compliance.” Journal of Economic Psychology 26: 491–508. Wittberg, Lennart. 2010. “Bemötande och förtroende (Encouragement and Trust).” In Skatteverket. Stockholm: Swedish Tax Agency. Yee, Albert  S. 1997. “Thick Rationality and the Missing ‘Brute Fact’: The Limits of Rationalist Incorporations of Norms and Ideas.” The Journal of Politics 59 (4): 1001–39.

OUP CORRECTED PROOF – FINAL, 13/11/21, SPi

Index Note: Tables and figures are indicated by an italic t and f following the page number. Ackermann, Kurt A.  31 age factors, and tax morale  19 Ahmed, Eliza  19 Albania 43 Allingham, Michael  28, 50 Alm, James  5, 19, 43, 124n21, 125n32 altruists 71f, 71, 73f, 74f see also Social Value Orientation (SVO) experiment Ames, Ruth E.  62 Andreoni, James  50 Andrighetto, Giulia  82 anti-social behavior  58, 75, 80 Ariely, Dan  4 attitudinal survey  17, 33, 96–114 example question  34t instructions  86, 96 questionnaire 96–113 audit rates  39–40 authoritarianism 8 Banfield, Edward  44 baseline experiment  16, 21 instructions 85–96 summary statistics  35t, 36t see also attitudinal survey; Social Value Orientation (SVO) experiment; tax compliance experiment Baumeister, Roy F.  57 Bazart, Cécile  59 behavioral economics  4–6 Bergman, Marcelo  59 Berlusconi, Silvio  121n1 Bonein, Aurélie  59 Braithwaite, Valerie  19 Brazil, and tax morale  18 Buddhism, and tax morale  19 Carter, John R.  62, 63 cash transactions  46–7 Catholicism, and tax morale  19

charitable donations  29, 95, 122n4 China, and tax morale  18 citizenship, paying taxes as duty of  69 cognitive consistency  65–8 cognitive dissonance  66, 68 Colbert, Jean-Baptiste  1 comparing countries and individuals  12–16 competitive individuals  71f, 71, 73f, 74f conditional cooperation  59 confidentiality 22 confirmation bias  14 conservatism, and tax morale  19 cooperators, distribution in society  13–15, 13f culture vs. institutions  37–8, 79–80 Darwin, Charles  8 Denmark, third-party reported vs. self-reported income  123n15 dictators 8 direct democracy, and tax morale  18 Doerrenberg, Philipp  123n9 economics majors  62–6, 62f, 63f, 64f, 65f analysis 120 Social Value Orientation  73, 74f Edlund, Jonas  15 education levels, and tax morale  18 enforcement, and tax morale  18 equality and inequality, attitudes towards 49 progressive tax systems  52 European Research Council (ERC)  3, 80 European Values Survey  33, 59 experiments running in multiple countries  20 and tax behavior  19–20 see also attitudinal survey; baseline experiment; Social Value Orientation (SVO) experiment; tax compliance experiment

OUP CORRECTED PROOF – FINAL, 13/11/21, SPi

136 Index fairness lack of  81 political institutions  11 progressive tax systems  52 fees, and compliance rates  123n11 Fehr, Ernst  5, 125n33 fines for tax evasion  39–40, 75 fiscal contract  48 free-riders  8, 80 economics majors  63 norms and social behavior  59 political institutions  12 Frey, Bruno S.  59 Gächter, Simon  5, 125n33 gender factors honesty 4 tax morale  18 Gentry, William M.  19 Germany, and tax morale  18 Gërxhani, Klarita  43 Greece, and social acceptability of tax evasion 121n1 Handgraaf, Michel  31 Heinemann, Friedrich  52 Herrmann, Benedikt  5 Ho, Simon  18 Hobbes, Thomas  8 ideology, and tax morale  19 income effects on tax compliance  55, 56f, 118, 119t India, and tax morale  18 individualists 71f, 71, 72f, 72, 73f, 73, 74f distribution in society  13–15, 13f inequality and equality, attitudes towards 49 progressive tax systems  52 institutions  17, 75 comparing countries and individuals  14 vs. culture  37–8, 79–80 norms and values  10–12 performance  79–80, 81–3 and self-interest  11–12, 39–45 tax morale  18 see also tax collection systems interests see self-interest

Irons, Michael D.  62, 63 Italy 3–4 analysis and tables  114–20 average compliance rate  44, 45f, 46 cash transactions  46–7 comparing countries and individuals  13–14 culture vs. institutions  37–8, 79, 80 economics vs. non-economics majors 63f employment by enterprise size  46, 47t, 47 norms and social behavior  59–60, 61f, 61 opportunities for tax evasion  46–7 progressive taxation  53, 54f social acceptability of tax evasion  2, 121n1 social impulse  49 Social Value Orientation angle  73f, 73 tax collection system  46, 81, 82, 83 tax gap  2–3, 15, 15t, 44, 46, 83, 121n2, 122n4 tax rate effects  50, 51f justice, and institutions  11 Kahn, Matthew E.  19 Kahneman, Daniel  4 Kastlunger, Barbara  40, 123n5 Kelly, Erin  67 Kleven, Henrik Jacobsen  123n15 Lawerence, Choo C. Y.  42 Leap of Faith, The (ed. Steinmo)  3, 10–11, 16, 75, 80, 81, 82 legal system, and tax morale  18 Levi, Margaret  41, 82 Li, Annie  18 Lippmann, Walter  66 littering behavior  80 McGee, Robert W.  18, 19 March, James  68 Marwell, Gerald  62 minimal group paradigm  124n29 Mittone, Luigi  5, 40, 123n9 Moore, Steven  68

OUP CORRECTED PROOF – FINAL, 13/11/21, SPi

Index  137 moral obligation to pay taxes  44, 123n12 US  77, 125n2 Murphy, Ryan O.  31 national variation, explaining  81–4 Nixon, Richard  77 norms  5, 10, 17, 78–9 analysis 119 comparing countries and individuals  12–14 national variation in tax compliance  81–3 political institutions  11–12 reasonable choice theory  38, 39 and social behavior  57–61, 60f, 61f, 65, 75 “theories in our heads”  66–8 Trump’s tax returns  77 and values, difference between  69 North, Douglass  8 Olsen, Johan  68 ORSEE (Online Recruitment Software for Economic Experiments)  21, 85 Pampel, Fred  49 peer pressure  59 penalties  39–40, 75 political institutions see institutions progressive tax systems  52–6, 53f, 54f, 55f, 56f analysis and tables  117–18 tax compliance experiment  29, 93–5 prosocial individuals  71f, 71, 72f, 72, 73f, 73, 74f punishment for tax evasion  39–40, 75 Quality of Government Index  51 quality of public services  2, 11, 15 Rational Choice Institutionalism (RCI)  9 rationality 10 Rawls, John  67, 68 reasonable choice approach  5–6, 9–10, 16, 38–9 redistribution of taxes  42–5 analysis and tables  114–18, 115t, 116f, 117f, 118f

social impulse  48–9 referendum voting, and tax morale  18 religion, and tax morale  19 risk attitudes  40, 44 Romania  3, 121n3 Ross, Adriana M.  19 Rubinstein, Ariel  63 Russia, and tax morale  18 Sandmo, Agnar  28, 50 Schneider, Friedrich  18 Schram, Arthur  43 self-employment, and opportunities for tax evasion 46 self-interest  5, 55–6, 78–9 cognitive consistency  66, 67 comparing countries and individuals  14 culture 37 economics majors  62–6, 62f, 63f, 64f, 65f and institutions  11–12, 37, 39–45 national variation in tax compliance 81–3 norms and social behavior  58, 59 reasonable choice theory  38 social aspect  41–4, 78 tax compliance experiment  27 Trump 77 Smith, Adam  39 social animals, humans as  10, 17, 56 cognitive consistency  67 collective goods and self-interest  41–4 norms  57–61, 60f, 61f, 65, 78 political institutions and self-interest  11 Social Identity Theory  57 social trust  15 Social Value Orientation (SVO) angle  70–1, 71f, 73f, 74f Social Value Orientation (SVO) experiment  16–17, 31–3, 69–73 Choice Set  70f example allocation decision  32f instructions  85–6, 95–6 Social Value Orientation (SVO) index 69 social values see values student fees  124n16

OUP CORRECTED PROOF – FINAL, 13/11/21, SPi

138 Index successful societies  79–80, 81 Svallfors, Stefan  15 Sweden 3–4 analysis and tables  114–20 average compliance rate  44, 45f cash transactions  47 comparing countries and individuals 13–14 culture vs. institutions  37–8, 79, 80 economics vs. non-economics majors 63f employment by enterprise size  47t norms and social behavior  59–60, 61f, 61 opportunities for tax evasion  47 progressive taxation  53, 54f punishment for tax evasion  40 Skatteverket, popularity of  60 social impulse  48–9 Social Value Orientation angle  73f, 73 tax collection system  81, 82, 83 tax gap  2–3, 15, 15t, 28, 121n2, 122n4 tax rate effects  51f trends in tax compliance  14 Switzerland, and tax morale  18 Tajfel, Henri  124n29 tax collection systems  2, 15, 37, 81–2 Italy  46, 81, 82, 83 Sweden  81, 82, 83 UK  46, 83 US 83 tax compliance experiment  16, 21–31 analysis and tables  114–19 clerical tasks  23–4, 23f, 24f, 28, 29 instructions  85, 87–95 reporting screen  25f tax gap  2–3, 15, 122n4 Italy  2–3, 15, 15t, 44, 46, 83, 121n2, 122n4 Sweden  2–3, 15, 15t, 28, 121n2, 122n4 UK  15, 15t, 46, 122n4 US  15, 15t, 41, 46, 122n4 tax morale  5, 18–19, 125n32 audits 40 and perceived level of tax evasion  59 tax compliance experiment  23 tax rates  49–51, 50f, 51f analysis and tables  114–17, 115t, 117f

on harmful items  75 progressive tax systems  54, 55, 55f tax compliance experiment  28–9 tax morale  18 “theories in our heads”  10 tipping behavior  67 Torgler, Benno  5, 18, 19, 59, 125n32 Trump, Donald J.  77, 78, 81, 125n1 trust in legal system  18 loss of  81 successful societies  79 social 15 Tversky, Amos  4 United Kingdom  3–4 analysis and tables  114–20 average compliance rate  44–7, 45f, 75 cash transactions  47 culture vs. institutions  80 economics vs. non-economics majors 63f employment by enterprise size  47t norms and social behavior  61f opportunities for tax evasion  46, 47 progressive taxation  54, 54f social impulse  49 Social Value Orientation angle  73f, 73 student fees  124n16 tax collection system  46, 83 tax gap  15, 15t, 46, 122n4 tax rate effects  50–1, 51f United States  3–4 analysis and tables  114–20 anti-tax sentiment  77 audits 40 average compliance rate  44–5, 45f, 75 cash transactions  47 culture vs. institutions  80 economics vs. non-economics majors 63f employment by enterprise size  47t moral obligation to pay tax  77, 125n2 norms and social behavior  61f progressive taxation  54, 54f, 124n26 punishment for tax evasion  40–1 social impulse  48, 49 Social Value Orientation angle  73f, 73

OUP CORRECTED PROOF – FINAL, 13/11/21, SPi

Index  139 tax collection system  83 tax gap  15, 15t, 41, 46, 122n4 tax morale  18 tax rate effects  50–1, 51f trends in tax compliance  14 Trump’s tax returns  77, 81 Valev, Neven  19 values  5, 10, 17, 68–76, 78–9 analysis 119–20

comparing countries and individuals  12–13 national variation in tax compliance  83 and norms, difference between  69 political institutions  11–12 reasonable choice theory  38, 39 Wenzel, Michael  124n30 Wittberg, Lennart  125n31 World Values Survey  33