BIVARIATE BAYESIAN VECM-SV MODELS FOR POLISH EXCHANGE RATES

• Authors: Pajor A.
• Languages of publication: EN

Abstracts

EN

In the paper three series of daily growth rates (of PLN/USD, PLN/EUR and EUR/USD over a period: 02.01.2002 -31.12.2004) are used to investigate the effect of the Forex market on the Polish exchange rates. In modelling of the official daily PLN/USD and PLN/EUR exchange rates the relationship: (PLN/USD)/(PLN/EUR) ~ EUR/USD is introduced. We assume that this relation (in log terms) is a cointegration equation in the sense of Engle and Granger and that the EUR/USD exchange rate is weakly exogenous in the Bayesian sense (for inferences on the parameters and on the latent variables, which describe two Polish exchange rates). Under this assumption we build five models with the error correction mechanism (ECM) and with the disturbances, which follow bivariate SV processes. The stochastic volatility processes differ in assumptions on the conditional correlation and in the number of latent processes. The best model assumes that the conditional correlation is time-varying, stochastic, and that the conditional covariance matrix is described by three separate stochastic processes. We show that the presence of the EUR/USD exchange rate and of the ECM term has serious effect on the Polish exchange rates. It reduces volatility and leads to increase in the posterior mean of the conditional correlations.

Keywords

EN

BAYESIAN INFERENCE; BIVARIATE SV MODEL; VOLATILITY

Discipline

• ECONOMICS: ECONOMICS

• Publisher: Główny Urząd Statystyczny
• Journal: Przegląd Statystyczny
• Year: 2006
• Volume: 53
• Issue: 3
• Pages: 9-26
• Document type: ARTICLE

Contributors

author

Pajor A.

• A. Pajor, Akademia Ekonomiczna w Krakowie, Katedra Ekonometrii, ul. Rakowicka 27, 31-350 Kraków, Poland

Identifiers

CEJSH db identifier

07PLAAAA01843927