Social Policy Review 33: Analysis and Debate in Social Policy, 2021 9781447359739

Citation preview

SOCIAL POLICY REVIEW 33 Analysis and Debate in Social Policy, 2021 In association with the Social Policy Association Edited by Marco Pomati, Andy Jolly and James Rees

SOCIAL POLICY REVIEW 33 Analysis and Debate in Social Policy, 2021 Edited by Marco Pomati, Andy Jolly and James Rees

First published in Great Britain in 2021 by Policy Press, an imprint of Bristol University Press University of Bristol 1-9 Old Park Hill Bristol BS2 8BB UK t: +44 (0)117 954 5940 e: [email protected] Details of international sales and distribution partners are available at policy.bristoluniversitypress.co.uk © Bristol University Press/Social Policy Association 2021 British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 978-1-4473-5972-2 hardcover ISBN 978-1-4473-5974-6 paperback SPA members’ edition (not on general release) ISBN 978-1-4473-5973-9 ePdf The right of Marco Pomati, Andy Jolly and James Rees to be identified as editors of this work has been asserted by them in accordance with the Copyright, Designs and Patents Act 1988. 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, electronic, mechanical, photocopying, recording, or otherwise without the prior permission of Bristol University Press. Every reasonable effort has been made to obtain permission to reproduce copyrighted material. If, however, anyone knows of an oversight, please contact the publisher. The statements and opinions contained within this publication are solely those of the editors and contributors and not of the University of Bristol or Bristol University Press. The University of Bristol and Bristol University Press disclaim responsibility for any injury to persons or property resulting from any material published in this publication. Bristol University Press and Policy Press work to counter discrimination on grounds of gender, race, disability, age and sexuality. Cover design: Bristol University Press Front cover image: iStock / Imagesines Bristol University Press and Policy Press use environmentally responsible print partners. Printed in Great Britain by CPI Group (UK) Ltd, Croydon, CR0 4YY

Contents List of figures and tables Notes on contributors

v vii

Part I

COVID-19: responses and implications for social policy Marco Pomati and James Rees

1

1

Locked down or locked in? Institutionalised public preferences and pandemic policy feedback in 32 countries Hung H.V. Nguyen, Nate Breznau and Lisa Heukamp New directions for European Union social policy in challenging times Carla Valadas Lesson-drawing for the UK government during the COVID-19 pandemic: a comparison of political, scientific and media lenses Sophie King-Hill, Ian Greener and Martin Powell On the periphery of the global spotlight: Sweden’s social policy responses during the COVID-19 pandemic Jayeon Lindellee Social policies put to the test by the pandemic: food banks as an indicator of the inadequacies of contemporary labour markets and social policies Jean-Michel Bonvin, Max Lovey, Emilie Rosenstein and Pierre Kempeneers ‘We have been left to go it alone’: the wellbeing of family carers of older people during the 2020 COVID-19 pandemic in Wales Maria Cheshire-Allen and Gideon Calder Gender crisis, or not? A comparative analysis of the impact on gender equality in Sweden and Germany due to the COVID-19 pandemic Marlene Haupt and Viola Lind Older adults’ access to information and referral services using technology in British Columbia, Canada: past learnings and learnings since COVID-19 Karen Lok Yi Wong, Andrew Sixsmith and Leslie Remund

5

2

3

4

5

6

7

8

29

49

75

95

115

135

161

iii

Social Policy Review 33

9

The pandemic as a litmus test for social security systems in transition economies – the case of Georgia Ana Diakonidze

Part II Migration, welfare and public health in Europe Andy Jolly 10

11

12

201

All of the same type? The use of ‘welfare tourism’ to limit 203 the access of EU migrants to social benefits in the UK and Germany Angie Gago Where is the vulnerability assessment tool? Disabled 223 asylum seekers in Direct Provision in Ireland and the EU (recast) Reception Conditions Directive (2013/33/EU). Keelin Barry Rethinking exclusionary policies: the case of irregular 243 migrants during the COVID-19 pandemic in Europe Marie L. Mallet-Garcia and Nicola Delvino

Index

iv

181

265

List of figures and tables

Figures 1.1 1.2 1.3

1.A1 4.1 7.1 7.2

7.3

7.4 8.1 8.2

Theoretical framework Scatterplots with correlation coefficients of relevant variables Aggregate public social behaviours and the relative discrepancy between public preferences and government pandemic response, after adjusting for current government intervention Density plots of government intervention over two stages Number of people who were notified of the termination of their employment January 1990 to May 2020 Public spending on family benefits in Germany and Sweden, 2005–15, as a percentage of gross domestic product Enrolment rates in early childhood education and care services, 0- to 2-year-olds and 3- to 5-year-olds in Germany and Sweden, 2005–17 Recipients/users of publicly administered parental leave benefits or publicly administered paid parental leave in Germany and Sweden, 2005–16 Employment rate of women, age 15–64, by age of youngest child in Germany and Sweden, 2005–19 Types of participants Municipalities of participants

8 13 18

27 80 141 142

143

144 167 168

Tables 1.1 1.2

1.3

1.A1

Descriptive statistics of key variables OLS regressions predicting government intervention using public preferences, outbreak stage, 1–15 March 2020, 32 countries OLS regressions predicting public behaviours in the public reaction stage (20 March–7 April) with discrepancy score from the outbreak stage (1–15 March 2020), 32 countries Metrics from measurement invariance models, public preferences

13 16

17

25

v

Social Policy Review 33

1.A2 1.A3

1.A4

1.A5

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.A1 3.A2 3.A3 6.1 7.1 9.1

vi

Metrics from measurement invariance models, public 25 behaviours OLS regressions predicting public behaviours in the public 26 reaction stage (20 March–7 April) with discrepancy score, dichotomised, from the outbreak stage (1–15 March), 32 countries OLS regressions predicting public behaviours in the public 26 reaction stage (20 March–7 April) with discrepancy score from the outbreak stage (1–19 March), 32 countries OLS regressions predicting public behaviours in the public 27 reaction stage (20 March–7 April) with discrepancy score from the outbreak stage (1–15 March), 31 countries, Croatia excluded Learning from scientific data by topic, frequency and 54 direction Summary of media perspective inter-crisis data by topic, 55 frequency and direction Summary of media perspective intra-crisis data by topic, 55 frequency and direction Political perspective by topic, frequency and direction 56 Media perspective summarised by country, frequency and 58 direction Comparing perspectives from the past: positive and 58 negative lessons Comparing perspectives from the past 58 Comparing perspectives from abroad: positive and negative 59 lessons Comparing perspectives from abroad 60 Coding for scientific data 66 Coding for media perspective 68 Coding for political perspective 71 Participant demographics 119 Selected characteristics of the three worlds of welfare 139 capitalism Unemployment insurance schemes across transition 184 economies

Notes on contributors Keelin Barry is an Irish Research Council PhD candidate at the Irish Centre for Human Rights, at the National University of Ireland, Galway. Her work focuses on human rights, refugee and disability law. Keelin has a particular interest in the rights of asylum seekers and refugees with disabilities. Jean-Michel Bonvin is Full Professor of Sociology and Socioeconomics at the University of Geneva. His fields of expertise include social and labour market policies, especially in favour of vulnerable people, as well as social justice theories and the capability approach. Nate Breznau is a social scientist and principal investigator of projects exploring the reciprocal relationship of public opinion and social policy, and comparative social policy across all countries of the globe, at the University of Bremen. He is an open science advocate. Gideon Calder is Associate Professor of Sociology and Social Policy at Swansea University. His work focuses mainly on social justice, currently in relation to (variously) childhood, care and co-production. He co-edits the journal Ethics and Social Welfare. Maria Cheshire-Allen is a PhD candidate and Researcher at the Centre for Innovative Ageing at Swansea University. She has worked on several large research projects including the ESRC-funded programme ‘Sustainable Care’ (2017–21). Her research interests lie in ageing, care, care ethics and social care policy. Nicola Delvino is a Senior Researcher at the Global Exchange on Migration and Diversity at COMPAS, University of Oxford, where his research focuses on EU and national laws and policies on irregular migration, and on local practices responding to the presence of migrants with irregular status. Ana Diakonidze is an Associate Professor of Sociology at the Georgian Institute of Public Affairs, Tbilisi, Georgia. Her research focuses on labour and employment policies in transition economies, social protection and new forms of labour. Ana also acts as a policy

vii

Social Policy Review 33

consultant to European Union Technical Assistance project to the Government of Georgia on Employment and Vocational Education and Training (VET) reform. Angie Gago is a Senior Researcher at the Centre of Comparative, European and International Law (CDCEI) of the University of Lausanne and a post-doctoral Fellow of the National Center of Competence in Research (NCCR-on the move) for migration and mobility studies. Her research interests are multilevel politics, institutions and public policy. Her current work studies the relationship between EU free movement and national welfare states. Ian Greener is Professor of Social Policy at the University of Strathclyde. He worries about a lot of things, but mostly how well different health systems work, the relationship between social justice and social policy, and the absence of Marmite-coated cashew nuts in his local supermarket. Marlene Haupt is Professor of Economics and Social Policy at the RWU, Ravensburg-Weingarten University. Her research interests include comparative social policy with a focus on the Scandinavian welfare states, pension systems and financial literacy. She is co-editor of the German Review of Social Policy (Sozialer Fortschritt). Lisa Heukamp is a Research Fellow at the SOCIUM Research Center on Inequality and Social Policy and a PhD fellow at the Bremen International Graduate School of Social Sciences. Her research areas are comparative social policy with a focus on the German welfare state, party politics, and public opinion and attitudes. Andy Jolly is a Research Fellow at the Institute for Community Research and Development (ICRD) at the University of Wolverhampton. His research interests are food poverty and bordering practices in social care. Pierre Kempeneers is a Research Fellow at the Geneva School of Social Sciences. His recent work focuses on the evaluation and implementation of public policies in the fields of employment and social policies. He holds a PhD in economics from the University of Montreal.

viii

Notes on contributors

Sophie King-Hill is a Senior Fellow within the Health Services Management Centre at the University of Birmingham. Her research areas include policy transfer, policy success, social policy, controversial issues and sexual behaviours in children and young people. She teaches on the NHS Leadership Academy award in Senior Healthcare Leadership and the MSc in Healthcare Leadership. Viola Lind is Gender Equality Officer at the Ludwig-MaximiliansUniversity Munich (LMU). Her research interests include feminist research and comparative social policy. Jayeon Lindellee is a postdoctoral researcher at the School of Social Work at Lund University in Sweden. Currently, she works with comparative studies of leaders of civil society organisations in several European countries. Jayeon is also working to design new social policy proposals for sustainable welfare through citizen involvement in Sweden. Karen Lok Yi Wong is a registered social worker in British Columbia, Canada, and has been practising in diverse settings related to older adults, including community senior service centre and longterm care. She is currently practising in acute care at Mount St Joseph Hospital in Vancouver. She is also affiliated as a researcher with Simon Fraser University Science and Technology for Aging Research (STAR) Institute. Max Lovey is a PhD candidate at the Institute of Demography and Socioeconomics of the University of Geneva and the Swiss National Centre for Competence in Research ‘LIVES’. His research areas include social policies, the access to social rights and the non-take-up of social benefits, vulnerability and the capability approach. Marie L. Mallet-Garcia is a Marie Skłodowska-Curie Researcher at COMPAS, University of Oxford. Marie holds a PhD from the Sorbonne University (Paris IV, 2013). Her research interests include immigrants’ integration into their host society through access to and use of social services. She is particularly interested in Latin American immigrants in the United States and Europe, as well as irregular and other vulnerable migrant groups. Hung H.V. Nguyen is a Research Fellow at the Research Center on Inequality and Social Policy (SOCIUM) at the University of Bremen

ix

Social Policy Review 33

and a PhD candidate at the Bremen International Graduate School of Social Sciences (BIGSSS). He studies public opinion, comparative political economy, party politics and quantitative text analysis. His current works focus on Germany and the bargaining spaces of parties in the Bundestag. Marco Pomati is a Senior Lecturer at Cardiff University. His recent work focuses on the creation and validation of policy-relevant living standards measures in Europe, the UK and Africa. Martin Powell is Professor of Health and Social Policy at the Health Services Management Centre, School of Social Policy, University of Birmingham. He has research interests in health policy, especially the British National Health Service. James Rees is Reader and Deputy Director of the Institute for Community Research and Development (ICRD) at the University of Wolverhampton. His research focuses on the third sector, public service delivery and reform, as well as leadership, governance and citizen involvement. He is co-editor of Voluntary Sector Review (Policy Press). Leslie Remund is the Executive Director of the 411 Seniors Centre Society and the 411 Foundation, where she works alongside an active board of directors and membership, who are all seniors, to advance the socio-economic status of older adults in Vancouver, British Columbia, Canada. Emilie Rosenstein is Lecturer in Sociology at the University of Geneva. Her main research areas include social policies, especially in the field of disability and youth welfare, the capability approach and the life course perspective. Andrew Sixsmith PhD is the joint Scientific Director of AGEWELL NCE, the Director of the Simon Fraser University (SFU) Science and Technology for Aging Research (STAR) Institute and a professor in the SFU Gerontology department. He is past President of the International Society of Gerontechnology and was previously Director of the Gerontology Research Centre and Deputy Director of the IRMACS Centre at SFU. His research interests include ageing, social isolation, technologies to connect people, technology for independent living, technology policy and the digital divide, theories

x

Notes on contributors

and methods in ageing and understanding the innovation process. His work has involved him in a leadership and advisory role in numerous major international research projects and initiatives with academic, government and industry partners. He received his doctorate from the University of London and was previously a lecturer at the University of Liverpool in the Institute of Human Ageing and Department of Primary Care. Carla Valadas is Researcher in Sociology at the University of Coimbra, Centre for Social Studies, Portugal. Her research interests include the social dimension of the EU, Portuguese employment policies and labour market transformations, and the implications of active employment policies in Southern European countries. Her current works focus on precarious employment relationships affecting different groups of people in different settings.

xi

Part I

COVID-19: responses and implications for social policy Marco Pomati and James Rees

Governments around the world reacted in different ways and showed varying levels of preparedness to the COVID-19 pandemic. Yet they all had to reconsider assumptions about personal freedoms, social security, health and social care, childcare and education. This volume showcases examples of responses to the pandemic and their implications for social policy from a wide range of countries. Compared with previous volumes, this year’s contributions are much more international and were written in the summer of 2020 in an ever-changing and uncertain landscape. They share a commitment to grasp some the most important issues highlighted by the pandemic, yet they can only do so with limited information. With time, the implications of COVID-19 will emerge in full, and the contributions in this volume are important for setting the research agenda of social policy for the years to come. Chapter  1 by Nguyen, Breznau and Heukamp explores the relationship between public preferences around restrictions of personal freedom and governments’ ability to curtail social behaviours. They measure public preferences using a measure of public attitudes that predates COVID-19 and correlate this with an online poll data on risky behaviours such as attending social gatherings during the pandemic. Despite the very clear limitations of the data, which they explain, they provide clear and thought-provoking theories on the trade-off between government restrictions and public behaviour during the pandemic, and their interesting if tentative results will shape future contributions in this field. Chapter 2 by Valadas is broad in scope, and sets out to examine the extent to which the COVID-19 pandemic, and earlier crises such as the global financial crisis of 2007–8, have accelerated the move towards a more coherent ‘Social Europe’. Providing detailed analysis of the social policies that have been put in place, particularly in relation to efforts to tackle job insecurity and extend welfare payment, and implement social rights, the chapter explores the nuanced policy and

1

Social Policy Review 33

politics of the recent trajectory that, haltingly and sometimes painfully, seems to be moving the EU towards a more unified approach to social policy. King-Hill, Greener and Powell bring together in Chapter 3 the literatures on policy learning and lesson-drawing with intra-crisis learning literature in order to examine three different sources of ‘learning lessons’ from overseas and the past in the COVID-19 pandemic; namely, the political, scientific and media lenses. Their responsive and timely chapter sheds light on how the UK polity – putting government decisions and responses in a broader context – looked for policy learning, and the extent to which it absorbed lessons from elsewhere. They conclude, perhaps unsurprisingly, that the UK government appear to have taken a rather piecemeal and unstrategic response to COVID-19, but also that the scientific ‘gaze’ was surprisingly insular too. Lindellee, in Chapter  4, also focuses primarily on policy, and government actions in response to COVID-19, this time in the much remarked-upon case of Sweden, famous (infamous?) for its singular approach to the pandemic based on avoiding lockdown and school closures, and a high degree of trust and social consensus. The chapter focuses on the broad range of economic and labour market measures swiftly implemented by the state, which were very much ‘tweaks’ to existing policies rather than panic measures. Nevertheless, Lindellee shows that even these relatively advantageous factors haven’t helped Sweden avoid some of the deleterious economic consequences, and uncomfortable questions remain over the country’s handling of the pandemic. Of course, the impact of the pandemic on older citizens has been a huge consideration for social policy. In Chapter 6, Cheshire Allen and Calder draw on 30 semi-structured interviews to explore the negative impact of the first wave of COVID-19 in Wales on the subjective, material and relational wellbeing of family carers caring for an older adult. Wong, Sixsmith and Remund examine an interesting case study from British Columbia, Canada, in Chapter 8, focusing on senior citizens’ access to information and referral services using technology. Clearly, the ‘digital divide’ and issues of inclusion/exclusion have been uppermost across the globe, and in many cases the pandemic has forced services and basic human interaction online – to be mediated by digital devices. The authors add a nuanced understanding of the challenges facing older people in using technology to access information, and the ways in which barriers have been overcome.

2

COVID-19: responses and implications for social policy

Chapter  5 by Bonvin, Lovey, Rosenstein and Kempeneers and Chapter 9 by Ana Diakonidze focus on the lack of adequate financial assistance for individuals on the edge of the formal labour market. They may represent a small minority in Switzerland or a much larger group in Georgia, yet the COVID-19 pandemic has provided clear evidence that too many workers are being knowingly treated as outsiders and as undeserving by governments around the world. The lockdown enforced in 2020, combined with the lack of a safety net, has simply driven many of these groups from poverty to the brink of destitution. Haupt and Lind’s Chapter 7 also shines a light on hidden work, this time the unpaid childcare work still carried out primarily by women, and how COVID-19 has the potential to reinforce gender inequalities. Overall, these three chapters outline a clear connection between the weaknesses of pre-COVID-19 national policies and the increase in inequalities during the pandemic.

3

1

Locked down or locked in? Institutionalised public preferences and pandemic policy feedback in 32 countries Hung H.V. Nguyen, Nate Breznau and Lisa Heukamp

Introduction The COVID-19 pandemic is the most severe worldwide public health crisis since the ‘Hong Kong Flu’ of the late 1970s.1 According to the Johns Hopkins Coronavirus Resource Center, the number of confirmed deaths worldwide from COVID-19 by December 2020 surpassed 1.5 million. This rapidly spreading virus tested the limits of governance soon after its outbreak. On the one hand, the strength and speed of government responses to a pandemic prevent infection and save lives (Wilder-Smith et al, 2020); but on the other hand, the public will only tolerate so much government ‘lockdown’ and will engage in behaviours that cause political turmoil and increase infections in response. Most theories of governance suggest that government actors respond to public opinion when making public policy. Therefore, we explore whether the variation in response to this pandemic met public preferences, and if not, whether it led to risky public behaviours. Governments around the world varied in the timing and severity of measures such as national lockdowns, imposing curfews and contact tracing. Surprisingly, strict lockdowns early in this pandemic often led the public to become more supportive of incumbent governments and democratic institutions (Baekgaard et al, 2020; Bol et al, 2020; Reeskens et al, 2020), known as the ‘rally around the flag’ effect. However, after some time, more and more members of the public began acting against government regulations by refusing to wear masks, holding social gatherings or even engaging in mass protests. We question how many of these anti-lockdown behaviours can be explained as a result of policy feedback from governments’ initial responses to the pandemic.

5

Social Policy Review 33

In a basic policy responsiveness model, governments act as representatives of the public and thus should formulate policies that follow public expectations (Page and Shapiro, 1983; Stimson et al, 1995; Wlezien, 1995; Manza and Cook, 2002). From an institutional theory perspective, path dependency suggests that governments are unlikely to engage in policymaking that is far removed from what the public expect (Pierson, 2004; Rasmussen et  al, 2019), as the cost is simply too high from an electoral maximisation and practical standpoint (Huber and Powell, 1994; Adams et al, 2004, 2006; Adams, 2012). At the same time, policy shapes public opinion, leading some to propose that opinion and policy operate in equilibrium with one another, which creates feedback keeping things on a stable trajectory (Jones and Baumgartner, 2012; Breznau, 2017). The conventional link between public preferences and what policymakers do, however, might be interrupted in a pandemic owing to imminent threats to human lives. This would explain the rallying of the public in support of the government at first, even though government policies severely limited their freedoms and rights. Given the preceding discussion of how opinion and policy are theoretically related, we expect that after the initial outbreak the public will begin to express their dissatisfaction as a function of what they normally see as acceptable government intervention policies. Therefore, if public preferences are deeply institutionalised in opposition to government intervention into people’s freedoms, but the government takes severe and long-lasting lockdown measures, the public will react negatively – either by simply refusing to follow the measures or engaging in political actions such as protests. This is a prediction based on institutional theories of public preferences when governments encroach by taking away long-standing rights such as social security (Pierson, 1994). If true, this could explain why many democratic countries still struggle to convince some members of the public to follow guidelines such as wearing masks, even when such guidelines are supported by scientific evidence (Prather et al, 2020). To test the link between embedded public preferences and government response, we investigate two stages of the pandemic. In the outbreak stage in the first half of March 2020,2 we expect governments to have reacted fiercely, implementing freedom-restricting measures that would go against most public preferences. This might have led to a temporary disconnection between long-standing public preferences and public policy. After the initial shock of the outbreak, a public reaction stage followed, in which people experienced fatigue from what they perceived as harsh restrictions or gradually lost trust in the

6

Locked down or locked in?

effectiveness of government restrictions, as observed in both Germany and the United Kingdom (Colfer, 2020; Naumann et al, 2020). We expect such negative affect among the public to be a function of the discrepancy between government responses and what the public normally deem appropriate. Therefore, we discuss this discrepancy as leading to two potentially opposing outcomes in public reaction that should follow in this phase. On the one hand, the public may react in opposition to the discrepancy, creating negative feedback and potentially undermining the government’s response. When governments restrict freedom less than the public normally prefer, people would fill in the missing role of the government and protect themselves by taking additional precautions. In contrast, in societies where government measures are stricter than the public normally prefer, people become dismayed and disobedient, resulting in less cautious activities. This is a form of self-undermining feedback (Jacobs and Weaver, 2015). On the other hand, the public may react in accord with the restrictions, creating positive feedback. Following a policy feedback perspective (Pierson, 1993; Soroka and Wlezien, 2010), if governments take weaker measures than the public are institutionally conditioned to prefer, people might develop a false sense of security and take weaker actions themselves. This may explain why death tolls in some northern European countries were so high early on, while public rebellion and concerns were low (Breznau et al, 2020). In a similar fashion, people seeing their governments taking actions that were more critical than expected might also become more prudent and virus-conscious, and less likely to engage in risky behaviours. From this perspective, stronger government intervention would be self-reinforcing. Our policy feedback logic suggest that public responses could support or undermine pandemic policies. Therefore, we leave these competing hypotheses open for empirical investigation. What is crucial here is that these public reactions should occur net of the actual severity of the outbreak and whether early interventions worked well, which varied widely across countries (Colfer, 2020). Figure 1.1 visualises our theoretical model.

Data and methods Data selection The beginning of March 2020 is when the pandemic emerged as a global emergency and governments everywhere took measures in

7

Social Policy Review 33 Figure 1.1: Theoretical framework Outbreak stage

Public preferences

Government intervention

Discrepancy Public reaction stage Negative feedback

Undermining behaviours

Positive feedback

Reinforcing behaviours

Note: In the outbreak stage, government intervention is a function of public preferences. In the public reaction stage, the gap between government intervention and public preferences, measured as the residual error, should explain one of two outcomes: self-reinforcing (positive) feedback or self-undermining (negative) feedback on public behaviours. Source: Drawn by the authors

response (Breznau, 2020). We accordingly conceive of the outbreak stage as 1  March to 15  March and the public reaction stage that followed as 20 March to 7 April. We assume by the latter stage the public had adjusted to the initial shock and began to evaluate the adequacy of their government’s initial response. Our selection of time periods is also somewhat determined by the data. First, our measure of public behaviours is only globally available after mid-March; secondly, it comes from survey questions asking about respondents’ behaviours in ‘the past week’, necessitating a lag between the two stages. Public preferences We are aware of only one measure of public preferences for government intervention into personal liberties with a broad sample of countries. This is the International Social Survey Programme (ISSP), which asked

8

Locked down or locked in?

three new questions in 2016 about governments’ right to encroach on personal freedoms and privacy ‘in the name of national security’ or ‘for the sake of public security’. They are (translated verbatim): • Do you think that the [COUNTRY] government should or should not have the right to do the following: keep people under video surveillance in public areas? • Do you think that the [COUNTRY] government should or should not have the right to do the following: monitor e-mails and any other information exchanged on the internet? • Some people think that governments should have the right to take certain measures in the name of national security. Others disagree. Do you think that the [COUNTRY] government should or should not have the right to do the following: …collect information about anyone living in [COUNTRY] without their knowledge? Although there were no questions on a pandemic specifically, these offer insights into the abstract and normative principles analogous to policies infringing on personal liberties for a greater public good that governments implemented at the early stages of the pandemic, namely regional and national lockdowns, contact tracing and prohibition of public gatherings. With these data, 32 mostly middle to high income countries can be compared (the countries are shown in Figure 1.2 and Figure 1.3). Apart from the above-mentioned, there are other questions on public/national security, but these either involve collecting information about people living outside the country or suspected terrorist acts. These questions fall outside the national context or are too specific for our research. The same can be said for batteries of questions regarding government spending or government responsibilities: although social insurance and basic welfare provisions are interventions into individual lives, they do not inhibit personal freedoms as pandemic lockdowns do. We do not take a strong stance on measurement here, in terms of whether individuals do or do not have a single latent opinion regarding the rights of the government to intervene in their freedoms and private lives. We are certain that these are highly salient issues to the public, given some of the violent acts and protests we have witnessed worldwide in response to lockdown measures, or even reactions to simple face mask regulations in stores or on public transportation (Pavlik, 2020). An extensive search using Google Scholar revealed that although the Design Group of the 2016 ISSP had

9

Social Policy Review 33

good reasons for including these new questions (Edlund and Lindh, 2019), no scholars have yet taken the time to investigate their scaling properties. We thus conduct our own confirmatory factor analysis (CFA), finding reasonable metric invariance across the 32 countries (see Table 1.A1, Appendix). Crucially, they have ostensible validity as reflective indicators of government intervening in personal freedoms. The resulting fit measures points towards invariance that we deem appropriate for our purposes (comparative fix index [CFI] = 0.968, Tucker-Lewis index [TLI] = 0.950, root mean square error of approximation [RMSEA] = 0.086); again, the substantiation of a single latent reflective psychometric indicator was not our goal.3 We used this model to create a new variable called public preferences, which contains the predicted factor scores from the three indicators. This provides us with the independent variable measuring embedded public preferences in the Outbreak Stage (see Figure 1.1). For this index, higher values indicate more support of government intervention. Deaths per million We assume one of the core predictors in both stages is the local severity of the pandemic. In the outbreak stage, governments should have responded directly to the seriousness of the situation, which should also have shaped public behaviours in the public reaction stage. Accordingly, we measure this local severity using the countrylevel number of confirmed deaths per million owing to COVID-19 in each period. We opt to use it here because it causes subjective severity perceptions among publics and governments: it was a key piece of information available to both that was separate from mixed predictions from scientists. Moreover, it is a measure that is more directly comparable across countries, unlike COVID-19 confirmed cases or infection rate measures. Government intervention We measure government response to the pandemic, which we label government intervention, using the COVID Stringency Index (also known as the Oxford COVID-19 Government Response Tracker), created and maintained by a team of researchers from the Blavatnik School of Government at the University of Oxford (Hale et al, 2020). Data are publicly available on the project’s GitHub page as well as on the Our World in Data website. This project collects 17 different indicators of government responses to the pandemic, ranging from

10

Locked down or locked in?

economic to containment and health measures. As the predictor in this case revolves solely around containment issues (lockdowns, contract tracing, etc), we use the Containment and Health Index in lieu of the general Stringency Index because its subset of variables links directly to personal freedoms. We create two versions of the index: the respective averages from each period standardised. The former is our dependent variable in the outbreak stage; the latter is a crucial control variable in the public reaction stage, because public behaviours are limited by current laws and because we observe a large increase in stringency between the stages (see Figure 1.A1 in the Appendix). To assess the validity of our hypothesised lag, we also take a mean index for the period from 1 March to 19 March as a sensitivity analysis. Discrepancy score To capture the gap between public preferences and government intervention, we create a discrepancy score as the standardised difference between the two. This is our main test variable. As in most public preferences research, we cannot know exactly how much intervention the public want from their government. Therefore, we can only identify relative public preferences based on the average across countries, and then compare deviations from this average to deviations from the average government intervention. As a result, a larger value in either direction indicates that government intervention was further away from the cross-country average of public preferences. We also explore the idea of a categorical difference between negative and positive relative discrepancies by dichotomising this score. In our very first theoretical test model, we calculated this score while adjusting for deaths per million, but subsequent tests revealed that this introduced an endogeneity problem, so we abandoned this model in favour of the simple subtraction method. Public behaviours For public behaviours in the public reaction stage, we employ Global Behaviours and Perceptions from the COVID-19 Pandemic data set, published by a team of researchers from various institutions (Fetzer et al, 2020). This large-scale internet survey received answers from 175 countries between 20 March and 7 April 2020. The original study attempted to offer weights, but it seems these are not available for all cases or in all countries, so we opt not to use them. The sample overrepresents younger, more educated, higher income and more frequent

11

Social Policy Review 33

internet users, and we discuss the limitations of this in the conclusion. In this survey, we identify three questions that ask about preventative social behaviours of respondents in the past week, including staying at home, not attending social gatherings and keeping a distance of 2 metres from others. We leave out other available questions because they ask about private activities such as washing hands and reporting symptoms. Respondents use a slider to reflect how frequently they practised each of the said actions on a scale 0–100. The exact wordings of the questions (translated verbatim) are: To what extent do the following statements describe your behaviour for the past week? • I stayed at home. • I did not attend social gatherings. • I kept a distance of at least 2 metres from other people. Following a similar procedure as with the ISSP data, we find reasonable standardised factor loadings for the baseline CFA model (Table 1.A2, Appendix) and goodness-of-fit indices for the metric invariance model with the three questions (CFI = 0.972, TLI = 0.957, RMSEA = 0.078).4 We created a new variable called public behaviours, using predicted factor scores. This variable was subsequently aggregated to country level. Control variables Besides the main variables, we also controlled for other country-level factors. These are economic development, measured as gross domestic product (GDP) per capita in thousands of US dollars, disposable income inequality (Gini) and social spending as a percentage of GDP. Table  1.1 gives basic descriptive statistics of all key variables discussed above, as well as other control variables in our subsequent regression models. Scatterplots with correlation coefficients for the main variables are shown on the right of Figure 1.2. The upper plot shows the bivariate relationship of public preferences and government intervention in the outbreak stage. The lower scatterplot visualises the relationship of this discrepancy score and public behaviours in the public reaction stage; to remind the reader, this is where we did not expect to find a relationship, yet the residuals are relevant to our discrepancy score.

12

Locked down or locked in? Table 1.1: Descriptive statistics of key variables Variable

N

Mean

SD

Min

Max

Public preferences

32

3.07

0.44

2.32

4.08

Deaths per million – outbreak stage

32

0.14

0.26

0

0.93

Deaths per million – public reaction stage

32

13.34

28.83

0.03

131.74

Government intervention – outbreak stage

32

0.06

0.95

–2.22

3.05

Government Intervention – public reaction stage

32

0.02

1.01

–2.34

1.99

Discrepancy score

32

0

0.95

–2.29

2.91

Discrepancy score (dichotomised)

32

0.44

0.5

Public behaviours

32

–0.03

1.02

GDP (k$) per capita

32

33.45

Disposable income inequality (Gini)

32

Social spending (% of GDP)

32

0

1

–2.57

1.1

22.39

2.1

81.99

33.17

7.75

23.3

57.5

18.07

7.86

2.2

31.68

Figure 1.2: Scatterplots with correlation coefficients of relevant variables KOR

Government intervention in the outbreak stage

3

2 FRA

1

CZE

DEU

SVK HRV

0 SVN

–1

PHL

JPN

ESP TUR

ISR

CHE RUS USA

ISL LVA LTU ZAF

VEN

TWN DNK NZL

BEL AUS

IND

FIN NOR THA

GBR SWE

R = –0.055, p = 0.76

–2 CHL

2.5

3

3.5

4

Institutionalised public preferences (continued)

13

Social Policy Review 33 Figure 1.2: Scatterplots with correlation coefficients of relevant variables (continued) SVN

Public behaviours in the reaction stage

1

ESP

CZE BEL DNK PHL CHE SVK NOR ISR VEN DEU LVA FIN ISL IND TUR NZL GBR ZAF LTU

USA

0

CHL

THA

–1

FRA

AUS HRV KOR

SWE

–2

JPN

RUS

R = –0.11, p = 0.55 TWN

–2

–1

0

1

2

3

Discrepancy score in the outbreak stage Source: Plotted by the authors

Research design Outbreak stage: alignment between institutionalised public preferences and government intervention? We start our analysis by predicting government intervention with public preferences using ordinary least squares (OLS) regression in the outbreak stage (M11). In the next models, M12 to M15, we gradually include our control variables as potential confounders, starting with deaths per million. Public reaction stage: does the discrepancy between expectation and reality predict public behaviours? In the second phase of the analysis, our discrepancy score is the main predictor of public behaviours (M21). Again, this discrepancy score is the difference between predicted government intervention and observed values of government intervention in the outbreak stage.

14

Locked down or locked in?

We then add in government intervention and deaths per million in the Public Reaction Stage as the other main test variables (M22 and M23). The point of including both government intervention and discrepancy score is to see how the gap between public preferences and actual stringency of government measures influences public behaviours net of current level of intervention and the severity of the pandemic. M24–M26 then contain the other control variables in the same order as in the first stage. Next, we check whether positive and negative discrepancies are qualitatively different phenomena by recoding the discrepancy score to be binary (one for positive scores and zero otherwise) for another round of regression models (Table 1.A3, Appendix). Although we believe it takes time for the public to react, we also test a scenario without a time lag between stages. For this, we recode the outbreak stage from 1 March to 19 March and keep the public reaction stage from 20  March to 7  April for sensitivity models M41–M46 in Table 1.A4 (Appendix). Finally, we remove an influential data point detected using the Cook’s distance on M24, namely Croatia, from the data set,5 and rerun the main models (M21–M26) to learn whether results are potentially distorted by it. These models are found in Table 1.A5 (Appendix).

Results Outbreak stage Regression results suggest the government did not react to institutionalised public preferences in the outbreak stage, as we expected. In other words, the coefficient for public preferences is near zero (see Table 1.2) and has a wide confidence interval (p-value = 0.75). Deaths per million is the major predictor of government intervention, with a statistical effect significantly different from zero with at least a 95 per cent confidence interval. The effect size is moderate to large. For every additional death per million, government measures are more stringent by 1.49 standard deviations (standardised beta = 0.41). We could explain around 25 per cent (R2 equals 0.250 in M15) of the variance in government intervention. The implicit assumption in this approach is that the remaining 75 per cent unexplained variance is ‘random’ in the sense that it is not causally associated with the variables on the right-hand side of our regression equation. Our main focus and finding is therefore the non-existent statistical association between public preferences and government intervention.

15

Social Policy Review 33 Table 1.2: OLS regressions predicting government intervention using public preferences, outbreak stage, 1–15 March 2020, 32 countries

Predictors

M11 Estimates

M12 Estimates

M13 Estimates

M14 Estimates

M15 Estimates

Intercept

0.43

0.24

0.85

2.15

2.20

–0.12

–0.12

–0.06

0.02

0.07

Public preferences Deaths per million

1.38**

Disposable income inequality (Gini)

1.25* –0.02

Social spending

1.44** –0.05

*

–0.04

GDP (k$, per capita)

1.49** –0.05* –0.03 –0.01

Observations

32

32

32

32

32

R /R adjusted

0.003/ –0.030

0.151/ 0.093

0.186/ 0.099

0.242/ 0.130

0.250/ 0.106

2

2

* p