Research

Working Papers

Community Matters: Psychological Well-Being and Spillovers in an Asset-Building Anti-Poverty Program in Paraguay

Abstract
Asset-building anti-poverty programs that follow BRAC's graduation approach generally yield positive average treatment effects on economic outcomes. However, these averages often mask substantial heterogeneity, and the psychological effects of such programs remain understudied. Leveraging a randomized controlled trial with a staggered rollout and saturation design, I examine how and for whom the Paraguayan government's graduation program works. Midline findings indicate that while the program improves key economic outcomes for most treated households, impacts vary widely across participants. I also find that the program worsens the psychological states of beneficiaries at midline, with measures of depression, locus of control, aspirations, and self-efficacy suggesting that the expectation for program participants to transform their livelihoods may induce stress. A saturation analysis reveals that this psychological decline seems to be attenuated in communities with a higher share of beneficiary households, highlighting the importance of community dynamics in supporting participants. Psychological factors may act as an important source of spillover effects, as beneficiaries in higher-saturation communities experience better economic outcomes than those in communities where fewer neighbors receive the program. The paper discusses what these findings imply for the cost-effective design and implementation of graduation programs.

The Economics of Sovereign Insurance in Low- and Middle-Income Countries (with Michael R. Carter)

Abstract
The increased frequency and severity of natural disasters has spawned the creation of multilateral risk management facilities that offer sovereign parametric insurance contracts that provide governments with budgetary support for infrastructure replacement and excess social protection payments that accumulate in the wake of hurricanes and droughts. While the argument for pre-arranged financing is compelling, there is a paucity of economic analysis concerning how much coverage a government should optimally purchased, especially taking into account the fact that parametric insurance will sometimes over- or under-predict true losses. We answer this question using a formal model of the decision to purchase parametric insurance to cover stochastic social protection payments. Assuming that the government has a fixed social protection budget and that its goal is to maximize the expected well-being of the poor population, we show that optimal insurance coverage is highly sensitive to both the predictive accuracy of the parametric disaster index and the pricing of the insurance relative to its actuarially fair level. Using realistic parameters drawn from a scheme designed to insure Kenya's social protection program for its drought-prone regions, we further show that while the optimal amount of insurance is positive, it is lower than typically imagined. Our analysis underscores the need to improve the predictive accuracy of parametric insurance and to stress-test it with rigorous public finance models.
Working paper available upon request.
Abstract
We study how individuals' risk preferences, subjective beliefs about future shocks, and related behavior change following a natural disaster. We focus on the impact of Typhoon Ketsana in 2009—one of the most devastating storms to hit Southeast Asia in recent times. Our analysis reveals that individuals who were affected by the typhoon become more risk averse a year after landfall. This effect persists up to four years later. We base our findings on household-level panel data from Vietnam and a difference-in-differences strategy with a continuous treatment variable that exploits variation in the intensity of the typhoon. We conclude that a standard deviation (SD) increase in excess rainfall during the typhoon leads to a 0.16–0.24 SD increase in risk averseness one year after landfall, and a 0.21–0.26 SD increase four years after landfall. Moreover, individuals exposed to higher excess rainfall are more likely to believe that storms will not transpire in the following five years or will occur with reduced frequency. This result supports the view that the main observed effects on risk preferences indeed reflect updated risk attitudes, rather than changes in the subjective probability structure assigned to the occurrence of storms. Finally, we show that individuals exposed to the typhoon increase their insurance purchasing in the long term. Our paper contributes to the literature that empirically documents how negative shocks may alter risk preferences and helps illuminate the way climate-related hazards can induce changes in the attitudes and economic behavior of individuals.
Working paper available upon request.

Asset Protection and Disaster Risk Financing to Improve Shock-Responsive Social Protection in Latin America and the Caribbean (with Michael R. Carter)

Abstract
In Latin America and the Caribbean (LAC), the poorest and most vulnerable households predominantly engage in informal work or self-employment and are often beyond the reach of conventional income maintenance programs that provide shock-responsive social protection for the formally employed. While the poorest households sometimes benefit from conditional cash transfers and economic inclusion asset-building programs, making these programs shock-responsive requires recognizing these households as active economic actors who need to build and protect the assets crucial to their livelihood and resilience. Although there has been some progress toward making cash transfer schemes shock-responsive, we argue that effective social protection for this population must extend beyond scalable cash payments and include customizable asset protection schemes tailored to individual asset exposure. The payoffs from such schemes could be substantial in terms of improved asset accumulation incentives, long-term poverty reduction, and resilience. These schemes could leverage parametric disaster risk financing instruments of the sort already present in the region to provide predictable, reliable, and customizable support to the target population. In addition to our core argument concerning the protection of the poorest in LAC, we also examine the role of parametric or index insurance products in underwriting scalable social protection obligations of governments. Our analysis suggests a cautionary perspective on the conventional logic regarding the important role that parametric insurance products can play as a form of stochastic government support. We show that the degree of reliance on these products depends critically on the reliability of the underlying parametric index. This point is particularly important given our emphasis on the need for reliable social protection programs that enhance and sustain asset building and protection for the poorest.
Working paper available upon request.

Other Reports

“Evaluating the Case for National Disaster Risk Insurance” (with Michael R. Carter, Francesco Fava, and Nathaniel Jensen). MRR Innovation Lab Evidence Insight.

Blog Coverage

“How Earth Observations Can Help Governments Protect Vulnerable Families from Disaster Risk.” BASIS Feed the Future Innovation Lab for Markets, Risk and Resilience at UC Davis.