The Econometric SocietyEdited by: Guido W. Imbens • Print ISSN: 0012-9682 • Online ISSN: 1468-0262
Econometrica: Mar, 2025, Volume 93, Issue 2
https://doi.org/10.3982/ECTA20496
p. 463-501
Kei Kawai|Takeaki Sunada
We estimate valence measures of candidates running in U.S. House elections from data on vote shares. Our identification and estimation strategy builds on ideas developed for estimating production functions, allowing us to control for possible endogeneity of campaign spending and sample selection of candidates due to endogenous entry. We find that incumbents have substantially higher valence measures than challengers running against them, resulting in about 3.5 percentage‐point differences in the vote share, on average. Eliminating differences in the valence of challengers and incumbents results in an increase in the winning probability of a challenger from 6.5% to 12.1%. Our measure of candidate valence can be used to study various substantive questions of political economy. We illustrate its usefulness by studying the source of incumbency advantage in U.S. House elections.
Kei Kawai and Takeaki Sunada
In the Online Appendix, we provide proofs that we omit from the main text as well as details regarding the model, estimation, data construction and applications to other environments. In Section 10.1, we describe the model of open-seat elections. In Section 10.2, we show that at the estimated parameter values, the challenger’s value function, vC, is not too decreasing in qC: the condition that ensures that the challenger’s entry decision follows a cutoff strategy. In Section 10.3, we provide a derivation of GqC,pC (·|s) and p(s, qC, pC) as a function of ¯qC(s, pC) for general N. In Section 10.4, we prove Proposition 1 (Injectivity) and Proposition 2 (Sufficient statistic). In Section 10.5, we discuss how we forwardsimulate the continuation value. We provide details of the estimation procedure in Section 10.6 and data construction in Section 10.7. In Section 10.8, we show how our approach can be extended to environments in which qI is time-varying, and one in which there are few uncontested elections. In Section 10.9, we present histograms of our valence measures after accounting for sampling error by applying Bayes shrinkage. In Section 10.10, weKei Kawai and Takeaki Sunada
The replication package for this paper is available at https://doi.org/10.5281/zenodo.14172367. The authors were granted an exemption to publish parts of their data because either access to these data is restricted or the authors do not have the right to republish them. Therefore, the replication package only includes the codes and the parts of the data that are not subject to the exemption. However, the authors provided the Journal with (or assisted the Journal to obtain) temporary access to the restricted data. The Journal checked the provided and restricted data and the codes for their ability to reproduce the results in the paper and approved online appendices. Given the highly demanding nature of the algorithms, the reproducibility checks were run on a simplified version of the code, which is also available in the replication package.
