Overview
George S. Tavlas is an economist at the International Monetary Fund (IMF). He co-developed with P.A.V.B. Swamy the general random coefficient model with concomitants and the direct/indirect effects decomposition framework, extending the Swamy (1970/71) i.i.d. model to its full general form with persistent coefficient dynamics and observable auxiliary variables.
Key Contributions
- General RC model with concomitants (Swamy-Tavlas 1992, 1995): Extended the Swamy simple model to include observable concomitant variables zt that identify the mean coefficient path and separate direct causal effects from indirect proxy effects. The general model (eqs. 19–21 in the 1995 survey) subsumes 19 conventional fixed-coefficient special cases including OLS, GLS, ARMA, and GARCH.
- GARCH as RC special case (Swamy-Tavlas 1994, Economics Letters 46: 7–10): Showed that GARCH models can be rewritten as a restricted instance of the RC framework, connecting the time-varying parameter and conditional heteroskedasticity literatures.
- Empirical applications (Swamy-Tavlas 1989, IMF Staff Papers 36: 63–101; 1992, Journal of Forecasting 11: 17–33; 1995): Documented superior out-of-sample RMSE performance of RC models in monetary demand, exchange rate, and interest rate forecasting. Table 3 of the 1995 survey reports RC models beating the best CFC alternative by 11–77% across nine studies.
Related