Do political factors affect stock returns during presidential elections?

This study investigates whether or not political factors such as government policy and political connections affected stock returns during the 2008 Taiwanese presidential election. We find that firms that benefitted from (were threatened by) the proposed Three-Links policy of the winning party experienced positive (negative) stock returns during the election. We use the sensitivities of firms’ returns to bilateral trade flows between Taiwan and China to measure the government-policy effect. Our results show that the effects of political connections weakly exist, but they become more significant when the support ratio of the winning party increased in polling data. We also find that only the government-policy effect holds for different crash-risk and corporate-governance levels. Finally, investment strategies based on both political factors can generate positive abnormal returns with respect to the Fama-French Three-factor Model.     

Private polling in elections and voter welfare

We study elections in which two candidates poll voters about their preferred policies before taking policy positions. In the essentially unique equilibrium, candidates who receive moderate signals adopt more extreme platforms than their information suggests, but candidates with more extreme signals may moderate their platforms. Policy convergence does not maximize voters' welfare. Although candidates' platforms diverge in equilibrium, they do not do so as much as voters would like. We find that the electorate always prefers less correlation in candidate signals, and thus private over public polling. Some noise in the polling technology raises voters' welfare.     

Forecasting multiparty by-elections using Dirichlet regression

By-elections, or special elections, play an important role in many democracies – but whilst there are multiple forecasting models for national elections, there are no such models for multiparty by-elections. Multiparty by-elections present particular analytic problems related to the compositional character of the data and structural zeros where parties fail to stand. I model party vote shares using Dirichlet regression, a technique suited for compositional data analysis. After identifying predictor variables from a broader set of candidate variables, I estimate a Dirichlet regression model using data from all post-war by-elections in the UK (n=468). The cross-validated error of the model is comparable to the error of costly and infrequent by-election polls (MAE: 4.0 compared to 3.6 for polls). The steps taken in the analysis are in principle applicable to any system that uses by-elections to fill legislative vacancies.     

Election forecasting: Too far out?

We consider two criteria for evaluating election forecasts: accuracy (precision) and lead (distance from the event), specifically the trade-off between the two in poll-based forecasts. We evaluate how much “lead” still allows prediction of the election outcome. How much further back can we go, supposing we tolerate a little more error? Our analysis offers estimates of the “optimal” lead time for election forecasts, based on a dataset of over 26,000 vote intention polls from 338 elections in 44 countries between 1942 and 2014. We find that optimization of a forecast is possible, and typically occurs two to three months before the election, but can be influenced by the arrangement of political institutions. To demonstrate how our optimization guidelines perform in practice, we consider recent elections in the UK, the US, and France.     

Polling games and information revelation in the Downsian framework

We investigate the incentives faced by poll respondents when candidates use polling data to inform their selection of policy platforms. Focusing on models with a unidimensional policy space, single peaked preferences and two office-seeking candidates observing a summary statistic from polls that ask respondents their preferences, we find that for most environments honest poll response cannot occur in a perfect Bayesian equilibrium. However, simple partially-revealing equilibria exist when the poll only asks respondents which party or candidate they prefer. When the candidates learn the sample average or see all the data, there are partially revealing equilibria that mimic those of the binary message game. Interpretation of polling data requires knowledge of the equilibrium played as the meanings of poll responses are endogenously determined. The analysis suggests that naive use of polling data may be misleading.     

Model-based pre-election polling for national and sub-national outcomes in the US and UK

We describe a strategy for applying multilevel regression and post-stratification (MRP) methods to pre-election polling. Using a combination of contemporaneous polling, census data, past election polling, past election results, and other sources of information, we are able to construct probabilistic, internally consistent estimates of national votes and the sub-national electoral districts that determine seats or electoral votes in many electoral systems. We report on the performance of the general framework in three applications that were conducted and released publicly in advance of the 2016 UK Referendum on EU Membership, the 2016 US Presidential Election, and the 2017 UK General Election.