Growth, Crisis, and the Evaluation of Financial Systems: Micro Underpinnings for Macro Models

National Opinion Research Center, Chicago IL

Investigators

Abstract

This research utilizes data collected by the PI in Thailand to understand financial markets and institutions, their role in providing safety nets during the recent crisis, and their role during the previous period of high growth. Both informal institutions (e.g., the extended family and networks of friends and relatives) and formal institutions (e.g., village level financial funds and national level private and public banks) are considered. More specifically, various models of saving, credit, and insurance widely used in the literature are used here to examine the micro underpinnings of the Thai economy and its actual financial system. This examination is based not only on institutional data but also on data from households and firms, directly. Econometric tests of risk-sharing, efficiency in production, more limited consumption smoothing, and occupation choice are extended to the Thai context, and generalized so as to allow an analysis of the potential role of these financial institutions. The various models are compared to another, via nesting or other means. The estimated underpinnings and associated parameter values are then used in various of the models of growth, inequality, and financial deepening and the predictions tested against the Thai Socio-Economic Survey and other national-level data. The larger goal here is to improve the science of macro-economics, and evaluation and policy recommendations based on macro data. The project began with a review of existing theoretical and empirical literature. This led to the design of a series of questionnaires, to measure the key variables suggested by all these models and related theoretical and empirical literatures. A large cross-sectional survey was fielded in May of 1997, of 2880 households, 606 small businesses, 192 village headmen, 161 local financial institutions, 262 joint liability groups, and soil samples from 1920 agricultural plots. Using a cross-section to mimic aspects of growth, we picked two provinces close to Bangkok, including semi-urban and industrial zones, and two in the relatively poor and semi-arid northeast. With the devaluation of the Thai baht in July 1997, an unanticipated resurvey of 33% of this cross-section was accomplished in May 1998, 10 months into the crisis. We completed in May 1999 a third resurvey, and thus we have an extensive household and business panel covering household composition, occupation, income, consumption, physical assets and consumer durables, saving, lending, borrowing, inputs and output in agriculture and in business, the use of networks, and other variables as well as information on local financial institutions and the history and contemporary situation of the villages. The yet more intensive monthly survey began in July 1998 and covers 16 villages, 45 households per village, or 720 households per month. In five villages every single household is sampled, giving us an excellent opportunity to study networks. The variables are as listed above, but here with more detail on labor supply, weekly consumption, barter versus monetary transactions, and so on. These annual and monthly data are to be examined using the standards suggested by a complete, perfect markets hypothesis or an equivalent set of institutions (e.g., an optimal allocation risk-bearing and efficiency in production), by the permanent income model (and as alternatives the buffer stock model, against fluctuations, and limited credit models), by models with entrepreneurship constrained by credit and human capital, and by models which are explicit about private information and incentives with an emphasis on the moral hazard problem at the level of individual borrowers.. Econometric tests build on the literature, but go beyond what can typically be done, making use of the information on institutional affiliation, actual institutional use, or a non-institutional alternative, all measured in the data on savings, borrowing, risk contingencies, and sharing (rice, money, labor, and equipment). More generally, these models with the various alternative micro underpinnings will be nested or otherwise compared. The parameters of some of the better known models of growth with inequality and financial deepening can be estimated with the Thai project data, and then dynamic simulations used to forecast the movement and changing shape of the income distribution and macro variables. Mean-squared error, Cremer-von Mises, and Kolmogorov-Smirnov test statistics will be used to compare the models' predictions against SES data and national income accounts. The micro, econometric tests of the permanent income model, buffer stock model, human capital occupation choice models, and mechanism design models suggest alternative micro underpinnings for the construction and testing of new micro/macro models of growth. Going from micro underpinnings to macro phenomena and back again, we will come to see more clearly the picture of actual financial markets and institutions in Thailand. This will be important for policy recommendations on structural adjustments and reformation of the financial system, e.g., bank regulation. The method applied here will be useful, in addition, for other countries, including the U.S.