Abstract

CRUZ-TORRES, Cristian. A bayesian estimation of Bivariate Garch-M Models. Rev. Mat [online]. 2024, vol.31, n.1, pp.99-126. ISSN 1409-2433.  http://dx.doi.org/10.15517/rmta.v31i1.53186.

The generalized autoregressive conditional heteroskedasticity (GARCH) model is a statistical model for time series used to describes the variance of the current error as a function of past squared errors terms and previous variances. These GARCH models are commonly used in modeling time varying volatility and volatility clustering. If, in addition, the effect of the variance is included in the observations to predict the mean, we have the GARCH-M (GARCH in mean) models. In this paper, the above issues are analyzed in a bayesian approach to modeling a bivariate time series, where the observations is assumed to behave as a VAR-GARCH-M model. An application of a bivariate model is fitted to measure the effects of inflation variability and uncertainty growth on inflation and output growth mean.

Keywords : bivariate GARCH-M models; bayesian inference; Hamiltonian Monte Carlo; inflation and output growth..

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )