The Candidate Factory: Technological Change and Political Supply

The transition from industrial to post-industrial societies is reshaping political landscapes, with studies documenting a decline in working-class representation. Many parties have become dominated by middle-class professional politicians with little experience outside of politics itself, while working-class people find it increasingly difficult to enter politics.
Despite the existing evidence on the compositional change in the population of politicians and the importance of political elites for the democratic process, we still know relatively little about the structural economic determinants of this phenomenon. This paper provides a structural explanation for the changing composition of political elites, focusing on the role of technological advancements such as automation and robotization. In particular, I focus on the occupational background of politicians, given how individual’s work experiences on the job play an important part in shaping political attitudes.
I argue that the distributional effects of automation play a key role in shaping political representation. Specifically, I theorize that automation alters the distribution of economic resources and opportunities across occupations, influencing who runs for office. I propose that those most negatively affected by automation, particularly individuals from working-class or automatable backgrounds, are increasingly underrepresented among political candidates.
I begin by documenting broad changes in elite composition, focusing on state legislative representatives across the U.S. from 1997 to 2012. To explore structural shifts in the broader pool of political candidates, the empirical analysis then centers on the full population of candidates for lower-level offices in California—such as city councils, school boards, and city supervisors—between 1998 and 2022. Candidates are classified by their occupational backgrounds using a text-as-data approach, and this data is linked to district-level indicators, including exposure to automation and robotization. The findings reveal a systematic decline in the share of candidates from working-class occupations and those most adversely affected by automation in counties with greater exposure to robotization.
To further support my findings, I use data from the nationally representative Cooperative Election Studies (CES) to examine the impact of exposure to robotization on political candidacy across social groups. The analysis reveals that automation exacerbates inequality in political participation along skill lines, with evidence pointing to the distributional effects of automation on income as a potential driver.
By exploring the impact of technological and occupational shifts on participation and candidate selection, this research sheds new light to key political developments, including trends in political representation, inequality, and the reconfiguration of political elites.