Populist governments, judicial independence, and public trust in the courts

ABSTRACT

Can governments make courts politically compliant without undermining public confidence in the judicial system? Many studies show a positive relationship between judicial independence and citizens’ trust in courts. However, most of them have shown static cross-sectional correlations rather than actual effects of court curbing on trust. Factors such as citizens’ level of education and political preferences may also play a role in moderating reactions to court curbing. We analyse how assaults on judicial independence by populist governments in Turkey, Hungary, and Poland affected judicial trust, using a difference-in-differences approach to Eurobarometer data. While we find evidence that court curbing has an adverse effect on judicial trust, this effect is much clearer among citizens who are ideologically distant from their governments. These findings coincide with experimental evidence indicating how citizens tolerate democratic backsliding, suggesting that, for many, trust in the judicial system can subsist even when courts are made politically subservient.

KEYWORDS:

• Trust in courts
• judicial independence
• populism
• Hungary
• Poland
• Turkey

Acknowledgements

We are grateful to Pablo Castillo-Ortiz, Christoph Engel, Eric Helland, Ana Rute Cardoso, Sveinung Arnesen, Dag Arne Christensen, Jonas Linde, Håvard Thorsen Rydland, and to the anonymous reviewers for helpful and stimulating comments and suggestions. We also benefited from comments by participants at the Max Planck Institute for Research on Collective Goods- Bonn workshop. Madeline Conn provided excellent research assistance. The usual disclaimers apply.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material information

All data and code used for this manuscript are available at the Open Science Foundation: https://osf.io/qe654/.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

1 Czech Republic and Slovakia are the illustrations provided by the authors.

2 Hungary and Poland are the two examples discussed by the authors.

3 Source: IDEA (2022).

4 In the Eurobarometer datasets, responses to the question about trust in the judicial system are missing for EB 90.3 and EB95.3 in Turkey. The Appendix describes the Eurobarometer surveys employed.

5 Programme Prawa i Sprawiedliwości, 2014. Available at: https://manifesto-project.wzb.eu//down/originals/2020-1/92436_2015.pdf.

6 We use the Stata sdid package (Pailañir & Clarke, 2022).

7 We use the community-contributed boottest command (Roodman et al., 2019).

8 We use Stata 18’s hdidregress command.

9 In the appendix, Table A2, we show results of estimations of a placebo treatment: dropping all observations after any country was treated, we estimate the effects of a phantom ‘treatment’ that would have occurred in the middle of the pre-treatment periods in Turkey, Poland, and Hungary. There is no evidence of the effect of this placebo treatment.

10 In the Eurobarometer datasets, responses to the question about trust in the judicial system are missing for EB 90.3 and EB95.3 in Turkey. Therefore, since the synthetic difference-in-differences estimation requires balanced panels, we have excluded those waves for the analysis conducted both when the three treated units are included and, obviously, when Turkey is the single-treated unit. See Table A3 in the Appendix for details, as well as Figures A4 to A6, where we show the resulting visualisations of the outcome trends in trust in the judicial system in each treated unit and the control units, as well as the time-weights then used to balance the pre- and post-treatment outcomes and the unit-weights used to make the weighted average pre-treatment outcome for the control units approximately parallel to that of the treated unit for the purpose of the DiD analysis.

11 We also checked for the linearity of these interaction effects, using Stata’s package interflex (Hainmueller et al., 2019). Using a binning estimator, with survey wave and territory fixed effects, the conditional marginal effects of Treatment at low, medium, and high levels of Age stopped full time education (corresponding to the terciles of that variable’s distribution) are placed very close to the linear marginal effect line. See same occurs with the effects at low, medium, and high levels of the variable Ideological distance from government. See Figures A7 and A8 in the Appendix.

12 In fact, as Figure A7 shows, the effect of treatment is already estimated to the negative and significant with cluster-corrected standard errors for ‘low’ (first tercile) levels of Age stopped full time education.

13 A Wald test of the difference between coefficients allows rejection of the hypothesis that the coefficients are equal with greater than 90% confidence.

Additional information

Funding

This research was supported by grants UIDB/50013/2020 , LA/P/0051/2020 & UIDP/50013/2020 of the Fundação para a Ciência e a Tecnologia, Portugal.

Notes on contributors

Pedro C. Magalhães

Pedro C. Magalhães is Senior Research Fellow of the Institute of Social Sciences of the University of Lisbon.

Nuno Garoupa

Nuno Garoupa is Professor of Law, Associate Dean for Research and Faculty Development and Faculty Director of Graduate Studies at the Antonin Scalia Law School of the George Mason University.