Financial crises and contagion effects: Stock markets, debt levels, and exchange markets in East Asia and Latin America in the 1990s

This dissertation focuses on an empirical investigation of the causes and spread of financial crises in selected Asian and Latin American countries in the 1990s. The first set of econometric estimates uses the time series methodology of Granger causality to test for the existence of contagion effects between the stock and currency markets of certain crisis originating countries (Mexico, Thailand, and Hong Kong) and the stock and currency markets of other countries in each region (as well as across regions). The econometric results indicate strong evidence for significant contagion effects of the Thai exchange market on currency markets in most other Asian countries, and that these effects were stronger and more significant during specific crisis episodes than during the full sample period. On the other hand, most of the estimated effects of Hong Kong's exchange markets on exchange markets in other Asian countries were insignificant. In addition, Thai and Hong Kong stock markets had significant contagion effects on other stock markets in the region, and these effects were more pronounced during the specific crisis episodes. In Latin America, there was little evidence of any significant contagion effects from the Mexican exchange market or stock market to the three other Latin American countries' markets considered (excluding Argentina). Also, there was little or no evidence of cross-regional contagion effects.; The second set of econometric estimates tests for whether large short-term debts contributed significantly to increasing the index of exchange market pressure in five highly-indebted Asian countries, after controlling for other variables including contagion effects from Thailand or Hong Kong. Four alternate measures of debt are considered and the results show that different measures were significant factors in exacerbating exchange market pressure in each of the countries, using country-level, time-series data. The debt analysis also tests for a parabolic or U-shaped relationship between debt and exchange market pressure. This hypothesis is tested by using a non-linear specification in which both debt and debt-squared are entered separately in the regressions, and the evidence for some of the countries supports this hypothesis.