Social Assistance Policies and the Presidential Vote in Latin America

Politics and Economy

A recent Insights report of the Americas Barometer shows that there is a correlation between participation in social assistance programs and support for the incumbent presidential candidate or party.  Analyzing relevant data from nine Latin American countries surveyed in the 2010 round of the Americas Barometer, we found that in almost every country examined, social assistance recipients are more likely to vote for the incumbent than non-recipients, even after accounting for social class, economic perceptions, and national context.

This first figure shows the percentage of citizens receiving support from existing government assistance programs in Latin America. Data was taken from a 2010 Americas Barometer survey question asking:  “Do you or does someone in your household receive monthly assistance in the form of money or goods from the government, for example, from [name of program(s)]?” 

Percentage of population on Social Assistance Programs

The graphic shows that large proportions of the populations in these countries receive government assistance; in fact, most countries analyzed here assist somewhere between 20 and 30 percent of their populations through social programs. But the question to pose now is: Why might social assistance programs affect presidential voting? Despite the differences across nations and programs, it is important to assess what the political effects of such social programs may be.

On the one hand, incumbent presidents seek to increase their vote share by claiming credit for social assistance programs. On the other hand social assistance recipients have a strong self-interest in maintaining benefits. Because there is always the uncertainty of whether the opposition candidate or party would continue to fund the program, those who currently receive social assistance will vote for the incumbent at higher levels. They will do so because they are familiar with that candidate or party’s first term behavior (Stokes 2001, 61) including implementing or maintaining the social policy that directly improves their economic condition.

The report also contains preliminary evidence that there are statistically significant differences between social assistance program recipients and non-recipients in a hypothetical vote for the incumbent – at least in some of the countries of Latin America. To assess who respondents would vote for, interviewers asked the following question:” If the next presidential elections were being held this week, what would you do?”

Percent voting for the incumbent

Without exception, a higher percentage of social assistance recipients would cast a hypothetical vote for the incumbent than would non recipients. Pooling all the countries together, there is a statistically significant mean difference between recipients and non-recipients of 15.45 percentage points (p<0.000). Still, there is considerable cross-national variation in the percentage of difference.

But although these results are suggestive of a relationship between social assistance and a hypothetical vote for the incumbent, they are not definitive nor do they distinguish between the theoretical causes of such a relationship. The key problem is that, given its nature, social assistance is not randomly distributed in any of these countries. Therefore, these aggregated differences of means may disguise what are the true correlates of incumbent vote based on social class and individual level attitudes.

In order to isolate the relationship between social assistance receipt and the vote, independent of social class, evaluations of the incumbent president, and national context, we develop a logistic regression model (see “Social Assistance Policies and the Presidential Vote in Latin America” by Matthew L. Layton and Amy Erica Smith).

The model includes controls for level of wealth, level of education, gender, and size of area of residence. It also includes an index of support for interventionist policies, to make sure that we are not simply picking up on the possibility that those who are predisposed to support public intervention may be more likely to make the effort to enroll in social assistance programs.

It also includes a series of variables accounting for a respondent’s perception of the current national economic situation. It captures the overall performance of the current government using an indicator for crime victimization, an index of corruption victimization and an index of government effectiveness. Finally, country-level fixed effects (not shown here for ease of presentation) account for important differences across countries in level of social assistance receipt and in support for the incumbent.

Confidence interval

The key finding from Figure 3 of this report is that there is a significant relationship between participating in a social assistance program and presidential voting, holding all else constant. It also shows that evaluations of government performance are by far the strongest predictor of vote for the incumbent. Those who see the current government as more effective are more likely to vote for the incumbent candidate or party.

Second only to these evaluations, perceptions of the national economy are also an important predictor of the vote. Individuals who perceive a better current national economic situation are more likely to vote for the incumbent. Neither corruption nor crime victimization, however, are significantly associated with support for the incumbent.

All of the other variables show some relation to the vote for the incumbent but in lower significance. To read the complete analysis of the results, visit Americas Barometer Insights series I0866

The results presented in this report indicate that in almost every country examined, social assistance recipients are more likely to vote for the incumbent. In multivariate analysis we find that the association is attenuated once we take into account other variables related to social status and evaluations of the incumbents’ performance. Nonetheless, even accounting for these confounding and mediating factors, we still find that social assistance receipt has a significant independent association with the vote.

*This blog is a short version of the article “Social Policies and the Presidential Vote in Latin America”, by Matthew L. Layton and Amy Erica Smith, published in the 2011 Insights series of the Latin American Public Opinion Project. To access the complete article and references, click here.



Share this