Betas y correlaciones dinámicas del sector Oil & Gas en Colombia 2014-2021 : una aproximación GARCH multivariada

Abstract

The objective of this document was to test the constancy of the coefficients under the framework of Fama and French, through the methodology of multivariate auto-regression models of conditioned heteroscedasticity that consider dynamic correlations (MGARCH – DCC), and considering in the estimation of parameters, variances and fluctuating covariances in time, distributions other than Normal (Student's t and general exponential, GED). Evidence was supported in favor of the hypothesis of variable coefficients in time for market factor, market value/book value ratio and including as a fourth factor changes in the price of crude oil for six stocks and two portfolios of the Oil & Gas sector in Colombia, between February 2014 and April 2021, with daily frequency that are listed in United States or Canada stock exchange markets. Weak evidence of variable coefficients was found for

the size factor. In the variable coefficients (betas) a coverage effect or mitigation Hedge (negative) was found in the response of the factor to shocks in the covariances (correlations), between the return on the share and the respective factor.