Book Description

This thesis studies two strategies that households may use to keep their consumption smooth in the face of fluctuations in income and expenses: credit (borrowing and savings) and insurance (state contingent transfers between households). The first chapter asks why insurance among households in rural Thai villages is incomplete. The second chapter analyzes the impacts of micro-credit. The third chapter examines the interaction between interpersonal insurance and access to savings. The first chapter is motivated by the observation that interpersonal insurance within villages is an important source of insurance, yet consumption, while much smoother than income, is not completely smooth. That is, insurance is incomplete. This chapter attempts to identify the cause of this incompleteness. Existing research has suggested three possibilities: limited commitment-the inability of households to commit to remain within an insurance agreement; moral hazard-the need to give households incentives to work hard; and hidden income-the inability of households to verify one another's incomes. I show that the way in which "history" matters can be used to distinguish insurance constrained by hidden income from insurance constrained by limited commitment or moral hazard. This history dependence can be tested with a simple empirical procedure: predicting current marginal utility of consumption with the first lag of marginal utility and the first lag of income, and testing the significance of the lagged income term. This test is implemented using panel data from households in rural Thailand. The results are consistent with insurance constrained by hidden income, rather than limited commitment or moral hazard. I test the robustness of this result to measurement error using instrumental variables and by testing over-identifying restrictions on the reduced form equation for consumption. I test robustness to the specification of the utility function by nonparametric ally estimating marginal utility. The results suggest that constraints arising from private information about household income should be taken into account when designing safety net and other policies. My second chapter (co-authored with Abhijit Banerjee, Esther Duflo and Rachel Glennerster) uses a randomized trial to analyze the impacts of micro credit in urban South India. We find that more new businesses are created in areas where a micro credit branch opens. Existing business owners increase their spending on durable goods but not non-durable consumption. Among households that did not have a business before the program began, those with high estimated propensity to start a business reduce non-durable consumption and increase spending on durables in treated areas. Those with low estimated propensity to start a business increase non-durable consumption and spend no more on durables. This suggests that some households use micro credit to pay part of the fixed cost of starting a business, some expand an existing business, and others pay off more expensive debt or borrow against future income. We find no effects on health, education, or women's empowerment. My third dissertation chapter (co-authored with Arun Chandrasekhar and Horacio Larreguy) is motivated by the observation that the ability of community members to insure one another may be significantly reduced when community members also have the ability to privately save some of their income. We conducted a laboratory experiment in rural South India to examine the impact of savings access on informal insurance. We find that transfers between players are reduced when savings is available, but that, on average, players smooth their consumption more with savings than without. We use social network data to compute social distance between pairs, and show that limited commitment constraints significantly limit insurance when risk-sharing partners are socially distant, but not when pairs are closely connected. For distant pairs, access to savings helps to smooth income risk that is not insured interpersonally.