There is little doubt that economic conditions affect voting behavior (e.g., Lewis-Beck
and Stegmaier 2000). This paper attempts to determine how this happens, a subject on which
there is considerably less agreement. The lack of consensus on how changes in individual or
national economic circumstances are transformed into votes is particularly troubling. For one
thing, the political economy of economic growth is significant in its own right. In the long
run, it can affect political stability and, certainly, the well-being of the population. For
another, the failure to understand this basic link between citizens and government challenges
any claim to understand democratic politics: if we cannot explain the substantively important
and relatively clear case of economic voting, it is doubtful we can adequately explain voting
behavior in general.
According to familiar political lore, individuals vote their pocketbooks, rewarding or
punishing each administration based on its economic record. Early time series studies (e.g.,
Kramer 1971) provided reassuring support for this conventional wisdom. In the U.S., an
increase in per capita income during an administration is indeed associated with support for its
reelection. However, confirmation of a significant pocketbook motivation at the individual
level has been more elusive (see, e.g., Nannestad and Paldam 1994).
Attempting to move beyond the tribulations of ecological inference, Kinder and Kiewiet
(1979) suggested that the discrepancy between aggregate and individual-level results is not
methodological at all. Rather, individuals directly base their vote on perceived national
economic conditions. Such sociotropic voting, they noted, would account for both Kramer’s
finding and the literature’s weaker results on pocketbook voting at the individual level. In
response, Kramer (1983) argued that, on the one hand, cross-sectional analysis as undertaken
by Kinder and Kiewiet cannot by itself uncover the relevant causal relations. On the other
Convention