Research

Learning-by-Losing: Do Political Parties Widen Representation to Win Elections? (Job Market Paper)

Although the right to vote has long been near-universal in many countries, there are still significant inequalities in political representation. One contributing factor may be political parties selecting candidates to try and win elections. I argue that parties may learn to be more inclusive in their selection in the aftermath of a defeat, through a `learning-by-losing' process. The case study of the Conservative Party after the shock landslide defeat in the UK 1906 election provide a unique setting to identify the response of parties to electoral defeat through their selection of candidates. Using comprehensive hand-collected biographical data and machine-learning algorithms to classify over 2000 candidates, I find evidence of the Conservatives choosing candidates from a wider range of backgrounds in response to electoral defeat, and evidence of the targeted use of candidates to win votes. These results may have implications for labour market inequalities in other occupations, as well as for policy decisions.

Local Information, Contextual Effects, and Geographical Polarisation: Evidence from the Brexit Referendum with Alexandru Savu

Despite evidence of increasing geographical political polarization, there is little work investigating why such polarization occurs. We propose a local social learning mechanism whereby policy preferences become more homogeneous within spatial units, yet increasingly heterogeneous between units as a result of voters becoming better informed about the preferences of those in their geographic vicinity. To empirically assess how such contextual effects may shape policy preferences, we exploit the delayed implementation of Brexit and its salience in the elections following the 2016 withdrawal referendum. Using a difference-in-differences design, we observe a two-percentage point relative decrease in the Liberal Democrat vote share in constituencies where Leave won a majority, mirrored by a similar increase for the Conservatives. Our findings suggest that voters updated their preferences concerning Brexit after observing the referendum's local results, and acted upon their new beliefs in subsequent elections - in a manner consistent with increased geographical polarization on the matter.

Female Representation in US Politics

Female political representation has grown across the world in the past century, but has somewhat stagnated in the U.S. since the 1990s. Using a data from all fifty states' legislatures from 1968 to 2010, I exploit the number of women elected by a close margin as a quasi-random treatment to find that the election of women does increase future female representation. However, this effect only lasts for two election cycles and driven largely by those initial women running for re-election, rather than new candidates. This result is robust across parties and time.

Impact of Competitive Elections on Strategic Spending in US State Legislatures

This paper investigates whether governments in nine US states strategically allocate public spending geographically to win elections. I use term limits as a proxy for an increase in the electoral competitiveness of a district, which is confirmed by finding a significant incumbency advantage of fielding the same candidate again of around 10-15 per cent of the vote. I find that using politician-district fixed effects there is a `term limit effect’ of 4 per cent increase in spending per capita, and a long-term effect of 10 per cent, primarily by Democrat governments. Using close elections as exogenous variation in majority size, I also find Democrat and Republican governments react differently to having more influence in their state legislature. With a larger majority, Democrats increase spending across the state but focus less on targeting competitive districts while Republicans do the opposite