Deindustrialization and the electoral strength of the far right in Brazil: the role of the decline of unions

The rise of far right parties and politicians in democracies has drawn the attention of the public and scholars from various fields of study. In Brazil, the election of far-right leader Jair Bolsonaro in 2018 marked the end of the electoral hegemony of the Workers' Party (PT), which had won the previous four presidential elections and governed the country from 2003 to 2016, when Dilma Rousseff underwent an impeachment process. In this article, we argue that deindustrialization, as an exogenous political shock, contributed to the increase in votes for right-wing candidates. Specifically, we are interested, as a causal mechanism, in the decrease in union strength in municipalities that underwent deindustrialization. Our hypothesis is that the loss of union density caused by deindustrialization negatively affects the supply of candidates rooted in unions, which favors the electoral growth of the right. Systematic declines in union density contribute to a relatively weaker public opposition to regressive measures in terms of labor legislation, wage policy, trade liberalization, and income inequality. As a result, parties historically rooted in union movements, such as the PT, lose electoral strength vis-à-vis parties and candidates that defend more right-wing agendas. In order to test the suggested hypothesis, we use mediation analysis with causally dependent multiple mechanisms, following the methodology developed by Imai and Yamamoto. This method allows estimating indirect effects of the loss of union density and the decreased supply of union-related candidates as causally dependent mechanisms that potentially help explain how, in a causal chain, deindustrialization (treatment) affects the right-wing vote (outcome), controlling for a set of demographic, economic, and social factors. The analysis uses data on Brazilian municipalities that experienced deindustrialization between 2002 and 2018. To measure deindustrialization, we use the industrial-electoral balance, computed as the difference in the share of industry in the GDP of municipalities from one election cycle to the next, with data from the Brazilian Institute of Geography and Statistics (IBGE). The main source of electoral data is the Superior Electoral Court (TSE).