Forecasting the 2012 U.S. Presidential Election: What Can We Learn from a State Level Political Economy Model?

Abstract

Most common work in the literature is based on aggregated models (see e.g., Campbell and Lewis-Beck, 2008) with some important exceptions that use data at the state level, e.g., Rosenstone (1983), Holbrook (1991) and Campbell (1992). These aggregated models, are based on either survey results or economic data at the national level. Here, in order to improve the accuracy of predictions, we examine the politico-economic determinants of presidential voting using pooled time series on a state-specific basis. After investigating a wide range of variables among which areas of concentrated strength of parties through time and territories in order to determine an appropriate model, we estimate which party’s candidate can be expected to win each state. Then, taking into account Electoral College voting strength, we aggregate state by state predictions in order to estimate the final winner in 2012 and the candidate’s expected share of the Electoral College vote.