This grant renews support for an ambitious and productive project on economies with uncertainty and with private information. This phase of the project studies the presence or absence of insurance, credit, and risk reduction arrangements in poor high risk economies. It develops and makes tractable economic models with private information, limited communication, and spatial separation. It measures which of these impediments are important in practice and determines what information-communication structures prevail in poor high risk villages. More specifically, the research contains the following: (1) Empirical tests of full insurance for three villages surveyed by the International Crops Research for the Semi-Arid Tropics (ICRISAT) in India from 1975-1984; (2) The development of general equilibrium prototypes with particular emphasis on private information to explain dispersed land holdings, modeled from historical material for medieval England, but applicable with modifications to India and Thailand; and (3) The evaluation of survey questionnaires administered to villagers in northern Thailand to quantify which, if any, risk reduction arrangements are used and to measure the nature and type of information impediments. A variety of important policy issues turn on whether or not there are private information problems or impediments to trade in poor high risk villages and whether or not risk reduction devices are available. For example, rice banks in Thailand are run like standard text-book banks, not respecting the erratic nature of crop yields. This project should show whether or not rice banks are needed and, if needed, how they might be improved. But the major contribution of this project is methodological. Standard economic theory relies on a paradigm of full insurance and complete markets. The poor agrarian villages studied in this project are used as natural laboratories in which to evaluate full insurance models. Alternatives to full insurance models exist but they are unrealistic or so complex that they can not be estimated empirically. This project for the first time merges three different strands of theoretical research with actual applied research.
