An important feature of countries that have not experienced sustained economic development is that a large proportion of the population participates in informal institutional arrangements that provide services that in developed countries are furnished by formal institutions. One important such service is the provision of income insurance. However, it is also thought that the performance of informal institutions in underdeveloped areas is constrained by the absence of formal contracts and a legal system that could serve to enforce compliance. In particular, the absence of these features of developed environments places significant barriers to trade across time and states of the world by reducing the ability of parties to commit to certain actions and thus results in inefficient allocations.
The existing empirical evidence suggests that while informal insurance arrangements involving interhousehold transfers do inhibit large fluctuations in consumption in the face of considerable income variability (Rosenzweig, 1988, and Rosenzweig and Stark, 1989), the proposition that insurance markets are complete has been rejected in the context of rural areas of low-income countries (Townsend ,1995; Morduch 1990; Foster 1994). In Foster and Rosenzweig (1995b), evidence is further provided that the presence of commitment problems is one reason why informal transfer arrangements do not, in themselves, provide complete insurance. This evidence would suggest that the formation and growth of formal institutions that is part of the development process might lead to the disappearance of informal institutional arrangements.
Existing empirical evidence on the relationship between the presence or characteristics of formal institutions that improve households' abilities to smooth consumption and informal insurance arrangements is scarce. One recent exception is Rosenzweig and Wolpin (1994) who show that in the United States the public provision of transfers to low-income families reduces only slightly family-based transfers. In this paper we assess the extent to which the introduction of financial intermediation in the form of formal credit institutions affects informal transfer arrangements among households in poor, rural areas using data from two South Asian countries. We modify the model of Thomas and Worral (1995), which characterizes risk-sharing arrangements under imperfect commitment, by incorporating the possibility that households facing risky incomes can both enter into transfer agreements and make use of liquid assets to smooth consumption.
In particular, we assess the welfare gains compared to autarchy and complete asset illiquidity associated with different scenarios characterized by the presence or absence of commitment in transfer arrangements and the degree of asset liquidity. Simulation results based on the model show that a combination of transfer arrangements and liquid assets appears to be a substantial improvement over having liquid assets alone, at least given the limited liquidity being considered, although the additional gain from transfer arrangements, with or without commitment, fall when households have liquid assets compared to autarchy. However, the gains from being able to commit to transfer arrangements, which results in greater insurance, are significantly greater in the presence of liquid assets compared to cases in which households do not have access to financial institutions. Given that informal insurance arrangements are costly to undertake, these results suggest that while the introduction of financial intermediaries in part "crowds out" private transfer arrangements, the arrangements that survive will exhibit greater commitment and thus provide on average greater insurance.
Simulations are also used to assess how the behavior of transfers differs under the various commitment/liquidity scenarios in order to derive testable implications that can be evaluated using actual data. Based on these results in combination with estimates of transfer and savings decision rules from data on Indian and Pakistani households differentiated by whether they resided in villages proximate to formal credit institutions, we find support for the model. In particular, we find that savings are significantly greater and contribute more importantly to consumption smoothing in villages closer to banks, that in such villages there is a reduced incidence of private transfer arrangements, but such arrangements appear to provide significantly more insurance compared to those in villages with lesser savings opportunities, reflecting an increase in commitment in such arrangements. Overall, the insurance gains from financial intermediation thus appear to be enhanced by household transfer arrangements, although not among those households facing high costs of commitment.
