Government responsiveness to voters’ economic vulnerabilities: evidence from 17 European democracies

Abstract

Do governments represent the material needs and preferences of their voters? This is one of the central questions of political science with great consequences for how we should evaluate the state of democracy. A thorough test of the link between voters’ economic vulnerabilities and public policies is presented in the article. Using data covering 17 European democracies, it is shown empirically that economically vulnerable voters support generous unemployment protection; that parties adopt policy positions that correspond with these voter signals; that entire party systems can meaningfully be described by the economic vulnerabilities of parties’ voters; and that parties, when entering office, pursue policies aligned with their voters’ needs and preferences.

Keywords:

• Policy representation
• social protection
• welfare state

Disclosure statement

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

Notes

1 We have benefitted from the comments of the reviewers and colleagues at presentations at University of Heidelberg, Aarhus University, Danish Political Science Association’s Annual Conference 2018, and the Swedish Institute for Social Research.

2 It is important to stress that the party family approach has other analytical merits. As pointed out by Kwon and Pontusson (2010), it is entirely legitimate to investigate if parties traditionally associated with left- or right-wing policies still pursue such policies. However, it is a different question than the one we are exploring currently.

3 Respondents who refused to answer, did not answer, or answered with a ‘don’t know’ are excluded.

4 Respondents who refused to answer, did not answer, or answered with a ‘don’t know’ are excluded.

5 Respondents are excluded if they refused to answer, did not answer, answered with a ‘don’t know’, or if they were not applicable to answering the question.

6 In Online appendix C, we include a subjective unemployment risk-measure (in ESS-round 4 and 8, respondents have been asked to estimate how likely it is for them to be unemployed within the next 12 months).

7 This conceptualization reduces the informational complexity, because information about each voter’s current income and risk of future income loss is compressed to the four boxes of vulnerability. In Online appendices C, D, and E, we therefore supplement each analysis below with an analysis, where we test the effect of the two dimensions separated.

8 Refusal, ‘don’t know’, and no answers are coded as missing.

9 Our two primary independent variables are dichotomous. The first one is coded such that the respondent has the value of 1 if s/he is advantaged, and 0 if s/he is not (i.e. disadvantaged or cross-pressured), while in the second one, the respondent has the value of 1 if s/he is disadvantaged and 0 if s/he is not (i.e., advantaged or cross-pressured).

10 In Online appendix C, we also include a control for the respondents’ self-reported ideology.

11 Respondents who did not answer the vote choice question, answered with a ‘don’t know’, refused to answer, was ‘Not applicable’ to answer, or answered ‘other’ are excluded. When there is more than one ESS-round between two elections, we use all ESS-rounds to calculate the economic vulnerabilities of the party voters.

12 The finding that AfD voters are characterized by relatively high risk of unemployment might at first seem to fly in the face of research showing that voters of the AfD are not significantly more likely to be unemployed than other parties’ voters (Schmitt-Beck 2017). Yet our data captures the risk of unemployment as measured by the occupational unemployment rate, not whether the voters themselves are unemployed.

13 The weight is calculated such that the value, for instance, is multiplied by 1, if the government was in office the whole year, while it will by multiplied by 0.5, if it was only in office half of the year (we exclude days between an election and the formation of a new government and calculate the weight relatively to the total number of days, one or more governments were in office during a year).

14 In Online appendices E4 and E5, we provide robustness tests with two other measures of the dependent variable: The first one is public unemployment spending relative to unemployment rate, while the second one is public unemployment spending measured in euro per inhabitant and relative to unemployment rate.

15 In Online appendix E, we also show that the effects are not changing across time (we include a dummy variable with two time periods – 1999–2008 and 2009–2018 – as a moderator).

Additional information

Notes on contributors

Carsten Jensen

Carsten Jensen is Professor in Political Science, Aarhus University. His books include The Right and the Welfare State (2014, OUP) and The Politics of Inequality (2018, Palgrave). He has published articles in the American Journal of Political Science, British Journal of Political Science, Comparative Political Studies¸ European Journal of Political Research, among others. [[email protected]]

Mathias Bukh Vestergaard

Mathias Bukh Vestergaard is a PhD student in the Department of Political Science at Aarhus University. He is doing research on political representation, welfare state politics, election pledges, party behaviour, and the media coverage of politics. [[email protected]]