Fast and Flexible Bayesian Inference in Time-varying Parameter Regression Models

In this article, we write the time-varying parameter (TVP) regression model involving K explanatory variables and T observations as a constant coefficient with KT variables. contrast much of existing literature which assumes coefficients to evolve according random walk, hierarchical mixture on TVPs is introduced. The resulting closely mimics specification groups into several regimes. These flexible mixtures allow for that feature small, moderate or large number structural breaks. We develop computationally efficient Bayesian econometric methods based singular value decomposition regressors. artificial data, find our be accurate faster than standard approaches in terms computation time. an empirical exercise inflation forecasting using predictors, models forecast better alternative document different patterns change are found assume walk evolution parameters.