Bayesian Causal Relation Effect in Quantiles Regression Models

Abstract

Analysis of causal relationships held an important part of the theoretical and empirical contribution in quantitative economic theory. This research explored the performance of Bayesian quantile regression with Granger causality showing that Bayesian inference can be undertaken in the context of quantiles regression. Causality Bayesian inferences in the context of quantile regression were achieved by applying the framework of the generalized linear model using asymmetric Laplace distribution for the error term. The developed scheme allows assessing the impact of the explanatory variables on all quantiles range of the conditional distribution of GDP growth. In Practical usage of macroeconomics variables, the scheme can be used to estimate parameters with causality effect which is synonyms to time series data. This research contributed to the versatile application of quantile regression in the contest of statistical research, the study estimated the regression quantiles parameter estimate applying Bayesian procedures. Furthermore, compared to frequentist estimates, Bayesian estimates established the superiority of the Bayesian regression method to the frequentist approach.