The Effects of Electing Working Class Candidates

• In Event: The Selection and Retention of Politicians of and by Underrepresented Groups

Abstract

Political candidates often campaign on their working-class roots, and working-class representation is broadly understood to matter for political outcomes. Prior work finds that working-class legislators take distinct positions on economic policy. However, beyond this relatively narrow set of policy domains, the effects of working-class representation remain unclear. Moreover, existing studies tend to rely on simple or statistically adjusted comparisons of elected working class vs. non-working class, a strategy that is ill-equipped to credibly estimate the effects of politician characteristics.

We advance the literature on working-class representation on both these fronts, considering a broader range of outcomes and providing a stronger foundation for causal inference. Focusing on Denmark, a case with abundant historical variation in working-class representation, we first use computer vision methods to digitize every candidate-level national election result ever recorded. The resulting data set consists of thousands of candidate-level results with known candidate occupations since the first democratic elections---spanning more than 150 years.

To capture the class background of candidates, we develop a novel measurement strategy, using word embeddings pre-trained on millions of newspaper pages to accurately scale the class alignment of each candidate occupation. We then use these class alignment measures to identify general election races pitting working-class against non-working-class candidates. Lastly, we implement regression discontinuity designs based on close elections at the candidate level to credibly estimate the effects of electing a working-class candidate on legislative rhetoric and behavior.

Our study provides novel insights into the effects of working-class descriptive representation. Specifically, our results show that electing a working-class candidate affects representation along more dimensions than documented in the extant literature. This in turn implies that the substantive costs of descriptive underrepresentation of the working class are greater than previously anticipated.

Authors

• Frederik Kjøller, University of Copenhagen

• Olivia Viktoria Jensen Levinsen, University of Copenhagen

• Anna Lund Sørensen, University of Copenhagen

• Frederik Hjorth, University of Copenhagen