Abstract
This article investigates the existence of contagion between countries on the basis of an analysis of returns for stock indices over the period 1994 to 2003. The econometrics methodology used is that of multivariate Generalized Autoregressive Conditional Heteroscedasticity (GARCH) family volatility models, particularly the Dynamic Conditional Correlation (DCC) models in the form proposed by Engle and Sheppard (2001). The returns were duly corrected for a series of country-specific fundamentals. The relevance of this procedure is highlighted in the literature by the work of Pesaran and Pick (2003). The results obtained in this article provide evidence favourable for the hypothesis of regional contagion in both Latin America and Asia. As a rule, contagion spread from the Asian crisis to Latin America, but not in the opposite direction.
Acknowledgement
P.L. Valls Pereira author would like to acknowledge the partial support from CNPq Grant No. 480831/2007-6.
Notes
1 This work does not aim to compare the gains from financial globalization with the losses arising from contagion. Contagion brings economic losses for countries and their populations. In this sense, autarchic countries are less susceptible to become victims of contagion. At the same time, while countries may be relatively closed in terms of world trade, they may be exposed or vulnerable to crises by the virtue of external debt, as occurred in Latin America during the 1980s.
2 The investigation could also concentrate on positive shocks. In this case, it would be a question of ‘positive’ contagion, but this is not the object of this study.
3 We also inserted dummy variables for each weekday and in order to distinguish returns calculated using data with a 1-day interval from data with a greater interval.
4 We also calculated t-statistics on the basis of variance estimators robust to heteroscedasticity, and there is good evidence that the fundamentals listed contain information to explain the analysed returns.
5 We tested whether there is residual autocorrelation in the squares of the residues of the regressions and it was not possible to reject the hypothesis of autocorrelation for the series. In this way, these may be used as the starting point for modelling the volatility structure and correlations on the basis of multivariate GARCH models.
6 The GJR model was formulated in the article by Glosten et al. (1993) and permits the introduction of asymmetric effects into the volatility.
7 All of the procedures carried out for the GARCH-DCC(2,2) model were done for a GARCH-DCC(1,1) model with essentially similar results.
8 The same procedures were also carried out using the value of two SDs without a substantial alteration in the results.