Forecasting US Presidential Elections Using Economic and Noneconomic Fundamentals

Abstract

It commonly is said that presidential elections are determined by the fundamentals, and that the campaign brings these fundamentals to the voters (especially see Gelman and King 1993). But what exactly does this mean? As we describe in The Timeline of Presidential Elections (Erikson and Wlezien 2012a), the fundamentals come in two varieties: internal and external. Internal fundamentals are part of the voter’s political predispositions. The campaign reinforces the voter’s party identifi cation, ideological proclivities, and demographically based partisan loyalties. External fundamentals are environmental forces unique to the campaign, such as, but not limited to, the state of the economy. As voters consider the external fundamentals, which can change over the course of the campaign, the outcome diverges toward the unique outcome predicted by those factors. We can think of these two types of fundamentals as centripetal forces pushing the electorate toward 50-50 (a party line vote) and centrifugal forces pushing voters outward toward some unique outcome specifi c to the issues of the day. In this article we illustrate how this dynamic works. Via interpolation from available poll data, we estimate the presidential preferences of the electorate on a daily basis over the fi nal 200 days of each of the 15 elections from 1956 to 2012. (For details, see Erikson and Wlezien 2012a; 2014.) We also do the same for economic perceptions, specifi cally, about recent economic growth, as measured by the Survey of Consumers at the University of Michigan.1 Using these two resources, we divide daily presidential preferences in diff erent election years into two components—the portion that is predictable from economic perceptions on that date and the residual portion that is unrelated to those perceptions. This division, of course, does not precisely separate national aggregate preferences into external and internal fundamentals; it provides a rough approximation. We ask: how do the two sources of voter preference (predictable versus not predictable from economic perceptions) evolve over the fi nal 200 days of presidential campaigns?