Competitive Equilibrium in American Presidential Politics

In the wake of the 2012 victory by Democratic incumbent Barack Obama, many within the Republican Party were deeply troubled by the election returns. Republican challenger Mitt Romney faired poorly among ethnic and racial minorities, to the point where Republican Party Chairman Reince Priebus equated the party’s performance among Latinos (Romney won just 27 percent of the Latino vote) to “a clear two-by-four to the head in the 2012 election.” While this level of support among ethnic minorities has been relatively typical for previous Republican presidential candidates (including winning candidates), the fact that has Republican elites worried is that whites—the Republican Party’s core base of electoral support—will constitute a shrinking proportion of the electorate in every subsequent election. Whites made up 72 percent of the electorate in 2012, down from 74 percent in 2008 and 77 percent in 2004. Whites will likely make up less than 70 percent of the electorate in 2016. Republican Senator Lindsay Graham was quick to acknowledge the political ramifications of America’s changing demographic landscape. “The demographics race we’re losing badly. We’re not generating enough angry white guys to stay in business for the long term,” said Graham during the lead up to the 2012 Republican National Convention.

The potential impact of America’s changing demographic profile has been the subject of considerable debate among pundits and academics alike. Many observers have been quick to point out that the ongoing demographic shifts are working in favor of the Democrats. In fact, several prominent political commentators, most notably Judis and Texeria (also see Abramowitz), have gone so far as to claim that the nation’s growing non-white population will lead to the emergence of a long-lasting “Democratic majority” on the national level. Yet, others are skeptical of these claims. Sean Trende has put forth several arguments as to why the Democrats are unlikely to put together a stable, majority coalition as time progresses. As Trende notes, electoral majorities have been notoriously unstable in American political history and even small shifts in voting behavior have been enough to upend seemingly stable electoral coalitions. So what effect is America’s changing demographic landscape likely to have on party competition? Is the emergence of a long-lasting Democratic majority a plausible or even likely outcome?

My answer is likely not. One of the most remarkable features of American presidential elections since the Civil War is the level of stability. In fact, Stokes and Iversen (1962) established that the American party system is characterized by the presence of a competitive equilibrium, which they defined as the expectation of an even partisan distribution of the vote. Figure 1 displays the Democratic percentage of the two-party vote for every presidential election since the end of the Civil War. The pattern of ebbs and flows in Figure 1 illustrates that any advantage one party is able to gain over their rival eventually erodes and an alternation of power occurs. I demonstrate this fact by regressing the change in the Democratic share of the two party vote from time t-1 to time t on the Democratic share of the two party vote at time t-1 for every election year between 1868 and 2012. In this model, the intercept records the equilibrium point (which is an expected vote margin of zero) while the coefficient for the previous share of the two-party vote records the speed of the reversion back to this equilibrium (1-.66).

The empirically derived expectation is that the winning party will perform worse in the next election—the long-term trend is that the vote margin will return towards a 50:50 equilibrium.

The equilibrium is stable even when I examine party competition over shorter periods of time. When I replicate the previous analysis using 10 election subsamples (e.g. 1976-2012) there are only two periods where the equilibrium (intercept) is statistically significantly different from an even 50:50 distribution (zero). The expected two-party vote drifted in favor of the Republican Party by a magnitude of 53.5:46.5 during two periods of time spanning 1892-1932 and 1952-1992, before eventually regressing back towards an even 50:50 distribution. Overall, there is over 160 years of accumulated evidence that suggests sustained periods of Republican or Democratic dominance do happen, but are a relative rarity.

The question here is whether recent demographic changes could possibly push the equilibrium in favor of the Democratic Party for the foreseeable future? I assess this question with a projection analysis. I use these established patterns of group behavior, shown in the table below (originally reported in Zingher 2014), to construct a baseline projection of how demographic changes will continue to alter the partisan balance in the absence of countervailing behavioral trends. To do this, I assume that white, African American and Latino Republican loyalty will remain at their previously established 30+ year average over the course of the next 20 years (year-by-year two-party group splits shown below).

In addition to this baseline projection I also include several alternate projections that serve as high-end and low-end estimates. For the high-end estimates, I assume that each group of voters will support future Republican candidates at a rate commensurate with the highest level of observed Republican loyalty by the group in any election since 1980. While I assume group behavior will remain constant, I assume that the ratio of groups in the electorate will change. I assume that the African America percentage of the electorate will remain stable at 12 percent while I posit a 3-percentage point increase in the Latino and Asian proportion in each subsequent election cycle—which is consistent with Census projections of the future composition of the electorate.

The results of these projections demonstrate that if group preferences remain unchanged, demographic shifts will reduce the expected Republican share of the two-party vote by seven tenths of a percent in each subsequent election. The Democratic advantage created by changing demographics would be sizable but by no means insurmountable, assuming previous outcomes are a reliable indicator of future elections. Even during prior eras when the equilibrium drifted away from the 50:50 mark the disadvantaged party has still managed to win elections. For example, Woodrow Wilson won two consecutive elections, which interrupted a period of sustained Republican dominance spanning 1896-1932. If left unabated, demographic changes will push the equilibrium point in favor of the Democrats, although not enough to prevent the Republican Party from winning when electoral conditions are particularly favorable (as represented by the top line in Figure 2).

However, prior electoral history suggests that the electoral equilibrium will likely remain at or near the 50:50 point. What types of behavioral changes will have to occur in order to maintain the partisan balance? I put forth several plausible scenarios. In the first, the Republican Party can maintain the partisan balance by continuing to increase its share of the white vote—a continuation of the current trend. Assuming that non-white Democratic loyalty remains fixed at its current average, the Republican Party would have to win 67 percent of the white vote by 2032 in order to maintain an expected vote share of 50 percent. Conversely, the Republican Part could also maintain an expected vote share of 50 percent by obtaining a greater proportion of the Latino and Asian vote. Assuming 60 percent of whites continue to support Republican presidential candidates, the Republican Party will have to win 47 percent of the Latino and Asian vote by 2032 in order to maintain the partisan balance. Of course, a combination of smaller increases in support among both groups would also be sufficient.

The conclusion that can be drawn from this analysis is that demographic changes are certainly putting the Republican Party in an increasingly disadvantaged position—although, the magnitude of this disadvantage is often overstated. The Republican Party’s path to victory is still quite plausible even in the absence of equilibrating behavioral changes. Yet, American political history suggests that a sustained period of Democratic advantage is relatively unlikely. There are several ways the Republican Party could offset the political consequences of changing demographics with countervailing behavioral changes. While the magnitude of the required changes is substantial, they are not implausible or even unprecedented.

*For a full version of this analysis please see my working paper here.