Following my post last week on the topic of which economic conditions are most important for understanding variation in US presidential election results — and the difficulty inherent in this kind of enterprise when there are only 16 elections in your data set — University of Connecticut political scientist Matthew Singer kindly offered to provide a follow up guest post more thoroughly reviewing the relevant comparative evidence in this regard:
In his review of Nate Silver’s attempt to tease out the relationship between various economic indicators and election outcomes, Josh makes the correct point that the limiting factor is the small number of American elections that we have to work with. Thus this question is one that is ripe for comparative evaluation. Yet there has been relatively little explicit theorizing about why voters might focus more on one set of outcomes than others and the empirical literature reaches mixed conclusions.
In discussing this literature, it is worth highlighting the relatively limited set of variables that have received attention. Studies look at some form of growth (GDP and GNI are the two most common), inflation, and unemployment. The last indicator has spotty data coverage in the developing world, and thus is not always considered in models focused on those regions. But the economic picture comparative work focuses on is very broad, perhaps because getting comparable data on the nuanced indicators Silver looks at is very difficult.
Considering the empirical evidence, Lewis-Beck and Paldam (2000) argue that inflation and unemployment are the two driving economic outcomes in established democracies. Yet van der brug et al 2007 find that growth and inflation have the most consistent effect on political outcomes generally in the same set of established democracies while unemployment only affects a small subset of parties. Inflation trumped other economic concerns in Latin America during the 1980s (Remmer 1991) but by the 1990s growth had grown in importance (Remmer 2003) and by the 2000s only growth is significantly tied to election outcomes in the region (Singer forthcoming). Unemployment is important in post-communist election contests (Tucker 2006, Roberts 2008). In a pooled sample of 39 post-election studies that combines established and newly democratizing countries, Singer (2011) finds that government approval is more consistently tied to growth than to either inflation or unemployment. Thus the specific indicator which most closely corresponds to election outcomes seems to greatly vary across regions and samples.
The instability in empirical findings highlights the need for a theory about how voters decide which aspect of the economy to focus on. Various studies have focused on the question of whether right parties are punished for different outcomes (e.g. inflation) than left governments (unemployment) are (e.g. Whitten and Palmer 1999) or whether a weak economy prospectively forces voters to vote for the party that “owns” it-e.g. unemployment helps parties of the left (e.g. Swank 1993, Carlsen 2000, Golder 2003). In considering the variation across cases, Singer forthcoming argues we might expect the salience of specific indicators to vary according to the degree to which they are a problem-e.g. inflation has faded as an electoral issue in Latin America over time because hyperinflation has (largely) faded as an economic concern, allowing space for voters to focus on other economic questions with less concern about the impact on prices. Finally, one might expect that groups differ in their exposure to kinds of economic trends, with rich voters responding more heavily to inflation that affects the value of capital while poor voters are less well-equipped to navigate the dislocations associated with unemployment (Hibbs et al 1982; Dorrusen and Taylor 2002, chapter 5), a finding that van der brug et al 2007 dispute. A recent MPSA panel discussed this possibility, although the results were mixed (email the authors for the papers, since none of us have posted them yet…). But across all these studies, my guess is that most of the variance in why some issues become salient while others do not remains unexplained. Thus there is substantial room for more theorizing about how parties/the media frame economic issues and how specific forms of economic vulnerability focus voter attention to issues that affect them the most.