7

Estimation of causal structural models regarding Housing Price and Housing Credit using Bayesian Inference, Random-Walk Metropolis-Hasting, Kalman Filters, State-Space models techniques.

Implements model averaging to improve the performance total credit gap as a predictor of financial crises. The model uses quarterly panel data of 50 years across 40 countries. The methods used are Bayesian model average, partial Area under the ROC curve (pAUC), index synthesizing, n-fold cross-validation, and policy function optimization.

Decomposes Credit series into long-run and short-run components. We use the cyclical property of the short-run component to improve out-of-sample prediction of the total credit series. The methods we used includes the Bate-Granger forecast combination, one-sided adaptive average.