I study whether households make Pareto‐efficient intrahousehold resource allocation decisions. Combining randomized variation in women’s income generated by the evaluation of the Mexican PROGRESA program with variation attributable to localized rainfall shocks as distribution factors in the collective model, I find evidence favoring Pareto optimality. More specifically, female‐specific income changes have positive effects on children’s goods expenditures, whereas changes due to rainfall shocks have a smaller influence on household public goods expenditures. The evidence is consistent with female partners having greater sensitivity to own‐income changes and norms that oblige women to devote their earnings to meet collective consumption needs.

This paper identifies neighborhood peer effects on children's school enrollment decisions using experimental evidence from the Mexican PROGRESA program. We use exogenous variation in the school participation of program-eligible children to identify peer effects on the schooling decisions of ineligible children residing in treatment communities. We find that peers have considerable influence on the enrollment decisions of program-ineligible children, and these effects are concentrated among children from poorer households. These findings imply that policies aimed at encouraging enrollment can produce large social multiplier effects.

Using U.S. state-level data we show that agglomeration externalities influence the level of foreign-invested capital in a location. Our empirical model allows the separation of agglomeration effects from the rate of capital stock adjustment, two forces that previous research has conflated. We estimate an agglomeration elasticity of investment of 0.11 to 0.15 with respect to same-source-country investment, lower than previous estimates. We also investigate the influence of state policies and find that although general investment incentives do not affect the location of FDI, targeted policies such as unitary taxation and state foreign offices influence investment.

Does monitoring corrupt activities induce a sustained reduction in corruption? Using longitudinal data on audits of municipal governments in Puerto Rico, we show corruption is considerably lower in municipalities with timely audits—before elections. However, these municipalities do not exhibit decreased levels of corruption in subsequent audits, even while mayors in these benefit from higher reelection rates. Our results suggest that audits enable voters to select responsive but corruptible politicians to office. Audit programs must disseminate results when they are most relevant for voters—shortly before an election—and ensure that these programs are sustained, long-term commitments.

This study argues that economic vulnerability causes citizens to participate in clientelism, a phenomenon with various pernicious consequences. To examine how reduced vulnerability affects citizens’ participation in clientelism, we employ two exogenous shocks to vulnerability. First, we designed a randomized control trial to reduce household vulnerability: our development intervention constructed residential water cisterns in drought-prone areas of Brazil. Second, we exploit rainfall shocks. We find that reducing vulnerability significantly decreases requests for private goods from politicians, especially among citizens likely to be in clientelist relationships. Moreover, reducing vulnerability decreases votes for incumbent mayors, who typically have more resources for clientelism.