A0981
Title: An economic evaluation of exchange rates higher order moments timing
Authors: Francesco Violante - IESEG School of Management (France) [presenting]
Stefano Grassi - University of Rome 'Tor Vergata' (Italy)
Abstract: The short-term predictive ability of empirical exchange rate models is evaluated within a framework that allows for volatility, correlation, (co-)skewness and (co-)kurtosis timing. The Mixture of (Student's) t by Importance Sampling Weighted Expectation-Maximization (MitISEM) is adapted to the estimation of multivariate Skew-t models featuring observation-driven time-varying covariance matrix, as well as asymmetry parameter and degrees of freedom. Compared to model-specific Markov chain Monte Carlo (MCMC) procedures, the MitISEM is more general, parallelizable, and it is considerably faster. Additionally, it provides the model's marginal likelihood, which is useful for model selection. Using GBP, EUR and JPY exchange rates relative to the USD, the economic value of these models' forecasting power is assessed by evaluating the impact of predictable changes in the conditional moments of foreign exchange returns on the performance of dynamic allocation strategies.
