Democratic accountability for economic performance: Informational and institutional dynamics in economic voting across democracies, 1970--2005

Understanding whether voters hold their government accountable for poor economic performance, and how much economic information they take into account when casting their votes, are the main concerns of this dissertation.; In the existing work on economic voting, surveys and case studies have established a relationship between a government's management of the economy and its electoral fortunes, while empirical cross-national analyses have had trouble finding evidence of economic voting. The lack of robust findings in the empirical literature has led scholars to question whether democratic institutions, such as periodic elections, can induce economic accountability.; This research shows that empirical cross-national analyses have failed to find economic voting not because the vote does not function as an instrument of economic accountability, but rather because the empirical models have been incorrectly specified. Conventional wisdom suggests that voters are short-sighted and only consider, at most, the economic information (i.e., the economic performance of the government) observed one year prior to the election. Few studies have questioned the standard economic measures of the comparative economic voting literature, and almost none have asked whether voters are capable of considering longer periods of economic information, or whether voters' length of vision depends on which economic variable they are evaluating.; Building upon voting models that consider more sophisticated and far-sighted voters (Peltzman 1990, Stokes 2001, Magalom 2006) I hypothesize that voters are far-sighted and consider the economic performance of the incumbent since the moment he takes office until new elections take place. If voters deem his performance poor, they punish him by voting for the opposition. Thus, the probability of reelection of the incumbent party depends on the economic performance of the government.; To address these questions, I use an original data set that includes elections from 1970 to 2005 in 40 Latin American and OECD democracies. To test whether voters have a short or long term vision, I create different variables of economic performance that represent different lengths of vision. I then show empirically which of these measures has the highest impact on the probability of reelection of the incumbent party. To complement the cross-national analysis I investigate the impact that economic outcomes have on voters at an individual level. Specifically, I study the effects of economic performance on Argentinean voters in the presidential elections of 1995 and 1999. I use two public opinion polls for this purpose.; Contrary to previous cross-national empirical studies of economic voting, I find that voters do punish poor economic performance and are able to consider long periods of economic information. In particular, I find that the average growth/inflation rates observed during the incumbent's entire tenure best explain the electoral fortunes of the incumbent party. If the incumbent party manages to produce, on average, high growth rates, its probability of reelection rises; otherwise, it decreases.; In line with some studies, and contrary to others, I find that the clarity of responsibility context has no effect on economic voting. I observe economic voting in both low and high clarity of responsibility contexts. Additionally, I find that even though there is a strong relationship between economic and electoral outcomes across democracies, this relationship weakens in situations where voters have experienced high economic volatility. The data suggests that contexts of economic stability foster economic accountability; whereas contexts of economic instability do not.; In contexts of high economic volatility, voters seem to have a hard time distinguishing a good from a bad economic performance. Thus, they tend to discount economic information when judging the performance of the incumbent. Nevertheless, the results show that, in genera...