Empirical essays on partisan differences in stock market returns

This dissertation consists of three essays which examine the connection between stock market returns and presidential elections in the United States. The first essay, "Stock Market Returns and Political Business Cycles," examines the link between fluctuations in the real economy that depend on the party of the president and differences in stock market returns that depend on the party of the current president. Statistically significant partisan return differences are found to occur only in the first year of presidential administrations. These return differences correspond to partisan differences in capital income growth, also found only in the first year of terms. Unexpected excess returns are shown to have forecasting power for capital income growth, employment growth, and output growth. The conclusion reached is that anticipated partisan effects in the real economy are the likely source of partisan differences in stock market returns.;The last feature of election cycle effects to be investigated in the first essay is the impact of election outcomes. The degree of surprise in election outcomes is found to have no impact on unexpected returns. But election surprises are found to have an effect on employment growth and inventory accumulation. It may be that unanticipated election results create a mismatch between firms' pre-election configuration of labor and capital inputs and that required to take advantage of the opportunity expected to prevail over the term.;In the second essay, "Political Business Cycles and Common Risk Factors in the Returns on Stocks," the cross-section of average stock returns is examined to investigate whether there are differences in risk premia that are related to the political party of the president. In the intertemporal capital asset pricing model (ICAPM) framework of Merton (1973), an asset's expected return depends on its covariance with the market portfolio and with state variables that proxy for changes in the investment opportunity set. As investment opportunities change it is an asset's exposure to these changes and its market beta that are the important determinants of average return. Changes in the investment opportunity set can also bring about changes in the price of risk associated with exposure to these factors.;Changes in the investment opportunity set might be expected to result from the partisan differences in employment growth observed in the early part of presidential terms. Consistent with this notion, significant partisan differences are found to exist in three variables that describe important dimensions of the investment opportunity set---namely the dividend yield, the term spread, and the default spread. These differences are found to translate into partisan differences in the price of risk.;Innovations that are priced risk factors are then examined to see whether they possess any forecasting ability with respect to employment growth. Innovations in the T-Bill yield are found to covary positively with employment growth two months ahead while innovations in the default spread covary negatively with employment growth two and three months ahead. These results suggest that partisan differences in the price of risk are indeed related to partisan differences in the real economy.;The relationship of election uncertainty to the priced risk factors is examined next. There is weak evidence that the degree of surprise in election results contributes to partisan differences in the price of risk.;In the third essay, "Election Outcomes and Stock Market Returns: A Variance Decomposition Approach," the channels through which presidential election cycles might impact stock market excess returns are investigated more rigorously. The first line of inquiry is to see whether accounting for election cycle effects changes the relative importance of expectations regarding returns versus expectations regarding cash flows in explaining the variance of unexpected returns. Intercepts that vary by party and year of term are included in a vector autoregressive (VAR) model to embed election cycle variation within expected returns. This results in less persistence in expected excess returns and a lower portion of the variance of unexpected returns that can be attributed to news about future returns. But, on balance, the variance of excess stock returns is still found to be due primarily to changes in expectations regarding future excess returns.;An interesting effect of taking election effects into account, however, is that the covariance between news about future excess returns and news about future dividend growth becomes negative. Thus, the contribution by news about dividends may be reinforced by the contribution from news about future returns. Since the election surprise impacts are restricted to election years in this study, this negative covariance must be an election year phenomenon.;A decomposition of unexpected returns during election years reveals that the impact of elections can be substantial when the results are surprising. (Abstract shortened by UMI.)...