How does unemployment and labour market downturns affect electoral participation? This question, which dates back to the classical studies in political sociology of the 1930s, has received renewed attention in the aftermath of the Great Recession. Yet, scholars differ in their views on how voter turnout is affected by the state of the economy. Some have argued that unemployment and economic hardship lead to political distrust, apathy and support for radical populist parties. Males are often said to be particularly susceptive to these effects when they are unable to uphold the traditional role of the family-breadwinner. Others, however, claim that high unemployment instead has the potential to mobilize voters for policy change and can thus result in higher turnout. Existing empirical evidence, mainly based on surveys, is mixed and suffers from two major shortcomings. The causal effect of unemployment on political behaviour is difficult to separate from other confounding factors and survey data imply issues connected to selection, data quality and limited samples.
The present study aims to improve on earlier scholarship on the effect of unemployment on turnout by making use of Swedish state-of-the-art individual-level register panel data for the period 1991-2014. These data allow for the use of empirical methods that go beyond existing correlational evidence, giving stronger support for causal inference. They also eliminate problems of non-response, social desirability bias and other data quality issues. In addition, the large number of observations opens up for the study of how the effect of unemployment differs across subgroups with great precision, in particular across the sexes. We may further study how differences between males and females are conditioned by the family structure, such as who is the main bread-winner. The empirical strategy primarily builds on using the variation in unemployment risk across occupational sectors and over time. Data on several elections imply opportunities for studying the effect of unemployment under different economic circumstances. By studying the effect within occupational sectors and by determining a person’s occupational sector at an earlier time point, we can mitigate problems of selection and reversed causation. Different methods for fixing a person’s occupational sector give further robustness to the results.
