Government Partisanship, Elections, and the Stock Market: Examining American and British Stock Returns, 1930–2000


  • For helpful comments and suggestions, we would like to thank three anonymous referees and participants at the Political Economy Seminar, University of California, San Diego; American Politics Research Group, University of North Carolina, Chapel Hill; Workshop on Economic History, Department of Economics, Rutgers University; and the Institute For Behavioral Science, University of Colorado at Boulder. We would also like to thank Barry Eichengreen and Hui Tong for giving us their money supply data, Heather Bell from Dow Jones company for daily data on stock prices from the Dow Jones Industrial Index, Chris Wlezien and Robert Erikson for providing us with their daily U.S. polling data, and to George Krause for his paper.

David Leblang is associate professor of political science, University of Colorado, 106 Ketchum Hall, Boulder CO 80309 ( Bumba Mukherjee is assistant professor of political science, Florida State University, Tallahassee, FL 32306 (


We construct a model of speculative trading to examine how the mean and volatility of stock prices is affected both by government partisanship and by traders' expectations of electoral victory by the right-wing or left-wing party. Our model predicts that rational expectations of higher inflation under left-wing administrations lowers the volume of stocks traded in the stock market. The decline in trading volume leads to a decrease in the mean and volatility of stock prices not only during the incumbency of left-wing governments, but also when traders expect the left-wing party to win elections. Conversely, expectation of lower inflation under right-wing administrations leads to higher trading volume. This leads to an increase in the mean and volatility of stock prices during the tenure of right-wing governments and when traders anticipate the right-wing party to win elections. Daily and monthly data from U.S. and British equity markets between 1930 and 2000 statistically corroborate the predictions from our formal model.