Abstract
This dissertation explores the estimation of non-linear multivariate systems in their reduced forms. The first essay develops a new method to solve multivariate discrete-continuous problems and applies the model to measure the influence of residential density on households’ vehicle holdings choices and vehicle usage. Traditional discrete-continuous modeling of vehicle holdings choice and vehicle usage becomes unwieldy with large numbers of vehicles and vehicle categories. I propose a more flexible method of modeling vehicle holdings in terms of number of vehicles in each category, with a Bayesian multivariate ordinal response system. I also combine the multivariate ordered equations with Tobit equations to jointly estimate vehicle type/usage demand in a reduced form, offering a simpler alternative to the traditional discrete/continuous analysis. Using the 2001 National Household Travel Survey data, I find that increasing residential density reduces households’ truck holdings and utilization in a statistically significant but economically insignificant way. The method developed above can be applied to other discrete-continuous problems. The second essay (with Ivan Jeliazkov) quantifies the interaction between political governance and macroeconomic performance in the United States by estimating a dynamic system: a vector autoregression (VAR) model involving macroeconomic variables and a presidential partisan dummy, and a regression equation of the presidential election outcome on the economic outcomes. The joint analysis of these components allows us to explore the dynamics of political business cycles and the impact of the economy on electoral uncertainty, and permits us to study their interaction. Our estimates of the short-run economic effects of elections are broadly consistent with the established view that short-run upturns in growth and employment follow the election of Democratic governments, while the opposite is true for Republicans. However, we show that the long-run outcomes are opposite to the short-run effects, which is in contrast to results in the existing literature where the long-run outcomes, although smaller in size, are found to be similar to those in the short-run. Our results from the electoral part of the model show that the incumbency effect in the U.S. is minimal, and that output growth has a noticeable and largely symmetric effect on the election outcomes for both parties. The last chapter (with David Brownstone) explores the influence of residential density on households’ vehicle fuel efficiency and usage choices with a sample of a national scale. A Bayesian approach that corrects for the endogeneity of the residential density is used to mitigate the problem of sample selectivity. The results show that an increase in residential density has a negligible effect on car choice and utilization, but reduces truck choice and utilization with a modest scale marked by statistical significance. The effects are larger than, but qualitatively consistent with, those obtained in Chapter 1, in which a California sample was used and the endogeneity of the density variable left uncorrected. Out-of-sample forecasting accuracy results are also reported to test the robustness of the model.