Contagion in Latin America: Definitions, Measurement, and Policy Implications

This paper analyzes bond and stock markets in Latin America and uses these patterns to investigate whether contagion occurred in the 1990's. It defines shift-contagion' as a significant increase in cross-market linkages after a shock to one country or region. Several coin-toss examples and a simple model show that the standard tests for contagion are biased due to the presence of heteroscedasticity, endogeneity, and omitted-variable bias. Recent empirical work which addresses these problems finds little evidence of shift-contagion during a range of crisis periods. Instead, this work argues that many countries are highly interdependent' in all states of the world and the strong cross-country linkages which exist after a crisis are not significantly different than those during more stable periods. These findings have a number of implications for Latin America.