A long-standing theoretical claim is that children serve as their parents' retirement assets in the absence of formal social security systems. But how can this hypothesis be tested empirically? The paper studies the context by examining the introduction of a comprehensive and financially lucrative pension system in Brazil. While employees in metropolitan areas have been paying into a state pension system for several decades, employees and self-employed people in rural areas of Brazil only gained access to a state pension scheme with the constitutional reform of 1991. The research shows that women who were able to qualify for the state pension system at a young age actually had fewer children. Compared to the control group (women in urban settlement areas for whom no change was made in pension provision), the women affected by the introduction of the pension system had on average 1.3 fewer children at the age of 45, and thus fewer than 4 children per woman. This accelerated the decline in birth rates in rural Brazil that had already been occurring for many decades. The pension reform thus exacerbated the demographic undermining of the pay-as-you-go pension system in Brazil.

Empirical evidence on the temporal stability of risk aversion and the role of exogenous shocks is inconclusive. In this paper we revisit this problem and analyze whether and to what extent risk aversion changes in response to rainfall shocks in an agrarian setting, and the role of changes in yields and output prices as two potential channels. We study this both in a simple theoretical model and in an extensive empirical study. The theoretical model predicts that under prudence, households that are either a net seller, a net buyer, or autarkic, increase risk aversion. To test the model predictions, we use a panel data set of the same 1,005 respondents from Northeastern Thailand and Central Vietnam (TVSEP) interviewed across five survey waves between 2008 and 2017 and combine it with historical rainfall data at the village level to capture negative rainfall shocks. Our empirical strategy exploits exogenous variation in the timing, location, and magnitude of rainfall shortages to identify their effect on risk aversion. We find that rainfall shortages increase respondents’ risk aversion, a finding remarkably stable across alternative specifications. The magnitude of the main effect of rainfall shortages on risk aversion is largest among net buyers, and approximately half the size among net sellers. Autarkic households do not exhibit a significant effect. Although rainfall shortages lead to significant reductions in agricultural yields and significant increases in commodity prices, the mediation analysis suggests that these market mechanisms do not play a significant role, except for a small statistically significant mediating effect of prices among the net buying household group. The finding that rainfall shortages lead to significant increases in risk aversion, especially among households that depend on local markets to buy food, has potentially important implications for food security and poverty dynamics. An increased level of risk aversion can undermine investments, e.g., in a beneficial technology, induce forgone returns, and increase the likelihood to fall or being pushed further below the poverty line.

Further Information.