Thesis (Ph. D.)--Massachusetts Institute of Technology, Engineering Systems Division, Technology, Management, and Policy Program, 2003.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Includes bibliographical references (leaves 157-167).We investigate the options for enhancing the welfare of small farmers in Thailand through subsidies of irrigation infrastructure. Enhanced water storage and irrigation can significantly improve yield and the welfare of the farm community by providing a more reliable water supply during growing periods. Generally speaking, such enhancements require subsidies from the government or other organizations since farmers are not able or willing to finance infrastructure development themselves. In order to maximize the effectiveness of such subsidies it is important to understand how farmers will react to alternative policies. We develop a two-tier approach to this problem. First, we use a government-level optimization model to identify the set of subsidies and water pricing policies that maximizes a stated measure of aggregate social welfare. This government-level model relies on a farm-level model that determines how individual small farmers will react to the policy alternatives. The farm-level model combines hydrologic, economic, and agronomic features since it considers how hydrologic variability affects crop yield, which in turn affects the farmer's utility. Policy decision variables considered in the government subsidy/pricing model include 1) water price and the number of farms served by public storage facilities (i. e. a water supply reservoir and enhanced in-stream storage), 2) maximum sizes of on-farm ponds paid for by the government, and 3) amounts of direct cash subsidies paid to the farmer. The objective is to maximize the aggregate welfare of all farmers served subject to a limit on the total subsidy as well as constraints designed to limit inequities and urban migration. The problem is solved with a deterministic nonlinear programming algorithm.(cont.) Decision variables considered in the farm-level model include 1) whether or not to accept a government-subsidized on-farm pond (which reduces land available for cultivation), 2) how much to consume in each year, 3) whether to devote time to agriculture or off-farm employment, 4) type of crop and irrigation technique, and 5) amount of water purchased from communal storage facilities. The problem is solved using a finite-horizon discrete-time stochastic programming algorithm. Our modeling approach is tested on a study site in Saraburi Province, Thailand. This site serves as a suitable prototype because of its existing irrigation infrastructure, relatively developed market institutions, secure land-use rights, and weak endowment of water resources. To achieve an economic optimum in which the farmers' aggregate utility of consumption is maximized, the government must provide some farmers with free reservoir water. The remaining farmers, however, help pay for the subsidy at a relatively high price. Consequently, the latter seek urban employment during the dry seasons. This cross-subsidy solution resulting from the social optimum criteria is economically efficient yet markedly inequitable. In order to assure equity, the government should construct the reservoir and sell the storage water at the same price to all farms. However, this solution cannot prevent urban migration ...by Sutapa Amornvivat.Ph.D.