The Metropolis-Hastings Algorithm: A Bayesian Inference Application for the Argentine Pension System

Abstract

The Metropolis-Hastings algorithm is a Markov Chain Monte Carlo (MCMC) method that allows sampling from complex distributions, facilitating Bayesian inference. It is an essential tool when the conditional posteriors lack an analytical form. This work presents an application of the Metropolis-Hastings algorithm to specify the posterior distributions of the parameters in a Bayesian logistic regression model, which captures the determinants of contributory density in the Argentine pension system. The method is implemented using the R statistical software, and the data source used is the Longitudinal Sample of Registered Employment (MLER) from the Integrated Argentine Pension System (SIPA). Among the results, it is noteworthy that all chains converge, and the significant coefficients exhibit the expected signs.