Political trust and job insecurity in 18 European polities

Abstract

Several decades of trust research has confirmed that difficult national economic conditions help explain citizens' low levels of political trust. But research points to a much less important role for personal economic factors. The latter finding, it is argued here, is a result of flawed survey questions and model misspecification. We actually know very little about the precise economic concerns that may generate low levels of trust and about the mechanisms via which they do so, resulting in a rather thin causal story. This paper seeks to address this lacuna, focusing on an issue of increasing importance in advanced economies: job insecurity. Using individual-level data from 18 European polities at two different time points, the paper finds that job insecurity generates lower levels of trust in politicians, political parties and political institutions and lower levels of satisfaction with democratic performance. Importantly, job insecurity's effect does not diminish as one moves from specific to more diffuse objects of political trust, as previous research suggests it should. The paper also finds that the effect of job insecurity is exacerbated if citizens have negative perceptions of the performance of the wider economy. Finally, and drawing on the occupational psychology literature, the paper proposes a novel causal mechanism to link job insecurity to political trust. The intuition is that job insecurity violates a ‘psychological-democratic’ trust contract between workers and the state. The mechanism is consistent with the observed results. The paper thus contributes to both the empirical and theoretical debates on the linkages between political trust and economic performance.

Keywords:

• political trust
• job insecurity
• economy
• European Social Survey data

Acknowledgements

Many people read early drafts of this paper and their critical commentary and advice have improved it immeasurably. These include the JTR reviewers, Amanda Gosling, Matt Loveless, Ben Seyd, Paul Stoneman, Robert Worcester and participants in the Work in Progress Seminar in the School of Politics and IR at the University of Kent.

Notes on contributor

Andrew Wroe is a lecturer in US politics in the School of Politics and IR at the University of Kent, UK. His current research interests include the causes and consequences of political trust, the political effects of economic insecurity and political polarisation in the USA.

Notes

1. Trust in the ‘employing organisation’ was measured using a two-item scale, based on responses to the following two questions: ‘I trust this organization to look out for my best interests’ and ‘I believe in the top management of this organization’ (Ashford et al., 1989, p. 813).

2. They examined job security's effect on job satisfaction, job involvement, organisational commitment, physical health, mental health, performance and turnover intention, in addition to trust.

3. Estonia and Portugal featured in both rounds but were excluded from the analysis because of missing data.

4. ‘Please tell me on a score of 0–10 how much you personally trust each of the institutions I read out. 0 means you do not trust an institution at all, and 10 means that you have complete trust: the [country's] parliament; the legal system; politicians; political parties’. ‘And on the whole, how satisfied are you with the way democracy works in [your country]?’ where 0 means extremely dissatisfied and 10 extremely satisfied.

5. Formally, as the dependent variable is ordinal rather than cardinal, an ordinal regression procedure such as ordered probit is more appropriate than OLS. We thus compared the base model (with no interactions) from an OLS and from an ordered probit and found, first, that the relative size and significance of all coefficients were similar in both models, and, second and more importantly, the estimated thresholds were spaced equally, suggesting any inconsistencies with OLS would be very small. We thus decided to retain the OLS model as this facilitates the easy use and interpretation of interaction terms, which are problematic to operationalise in nonlinear models.

6. The unweighted sample sizes for each country in 2004 are: Belgium 1778; Czech Republic 3026; Demark 1487; Finland 2022; France 1806; Germany 2870; Greece 2406; Hungary 1498; Ireland 2286; Netherlands 1881; Norway 1760; Poland 1716; Slovakia 1512; Slovenia 1442; Spain 1663; Sweden 1948; Switzerland 2141; and the UK 1897. And for 2010: Belgium 1704; Czech Republic 2386; Demark 1576; Finland 1878; France 1728; Germany 3032; Greece 2715; Hungary 1561; Ireland 2576; Netherlands 1829; Norway 1549; Poland 1751; Slovakia 1856; Slovenia 1403; Spain 1885; Sweden 1497; Switzerland 1506; and the UK 2422.

7. Multilevel modelling is neither necessary nor appropriate in this case. It is not necessary because the fixed effects model generates robust estimates for reasons stated in the main text and accompanying footnotes. It is not appropriate because econometric analysis suggests strongly that at least 50 level 2 (country) observations are required (Maas and Hox, 2005; Clarke et al., 2010 provide an excellent discussion of the relative merits and demerits of fixed effects and multilevel approaches), and also because the analysis seeks to control cross-national heterogeneity, not explain it.

8. The fixed effects model is also robust to the issue that unobserved factors may be correlated strongly within countries. Failure to control for this potential correlation could bias the estimates of the standard errors, even if these unobserved factors are uncorrelated with any of the regressors.

9. The three interaction terms were constructed using mean-centred variables. The coefficients represent simple conditional effects, not main or average effects as per the base model. See Brambor, Clark and Golder (2005) and Jaccard and Turrisi (2003) for discussions about variable coding and coefficient interpretation in interaction models. The base model above was run separately to the interaction term model in order to first facilitate a clear interpretation of job insecurity's average effects, before proceeding to explore the product terms' effects.

10. According to ESS data, job security declined between 2004 and 2010 by 23 percentage points in Ireland and, from a lower base, 18 points in Greece and Portugal. But it is likely that job insecurity has grown and political trust fallen since the ESS collected its data in 2010. The Eurozone crisis peaked in mid-2012. The huge IMF and ECB loans taken on by Greece, Spain, Portugal, Ireland, Italy and others, and the accompanying austerity programmes, have exacerbated already difficult economic conditions (Peston, 2013).