Bayesian dynamic systems modelling. Bayesian model averaging for dynamic panels with weakly exogenous regressors

This manuscript introduces the bdsm package, which enables Bayesian model averaging for dynamic panels with weakly exogenous regressors — a methodology developed by Moral-Benito (2016). The package allows researchers to simultaneously address model uncertainty and reverse causality. The manuscript includes a hands-on tutorial accessible to users unfamiliar with this approach. In addition to calculating the model space and providing key BMA statistics, the package offers flexible options for specifying model priors, or including a dilution prior that accounts for multicollinearity. It also provides graphical tools for visualizing prior and posterior model probabilities, as well as functions for plotting histograms and kernel densities of the estimated coefficients. Furthermore, the package enables researchers to compute jointness measures and perform Bayesian model selection to examine the most probable models based on posterior model probabilities.