Bayesian Forecasting (the Levels) of Vector Autoregressive Log-transformed Time Series Bayesian Forecasting (the Levels) of Vector Autoregressive Log-transformed Time Series Bayesian Forecasting (the Levels) of Vector Autoregressive Log-transformed Time Series

Bayesian dynamic models, stochastic simulation and Bayesian econometrics. of Rio de Janeiro in 1993 and is presently a lecturer of Statistics at Federal University of Parann a (Brazil). Research interests include Bayesian inference, stochastic simulatio n and Bayesian dynamic models. Abstract Forecasting the levels of vector autoregressive (VAR) log-transformed time series has shown to be awkward by Ari~ no and Franses (1996) who realised that just exponentiating the forecasts was a naive procedure due to the ocurrence of bias. They pr oposed a new manner to forecast untransformed VAR through correcting the log-transformed forecasts, and they also showed that their procedure outperforms the naive scheme. Our objective is to show that from a Bayesian viewpoint does not exist any theoretical problem and that it is feasible to obtain forecasts, and interval forecasts, of the untransformed time series. To do this, we use Monte Carlo simulation of the posteri or distribution of the parameters of the VAR adjusted to the log-transformed data.