A Bayesian Change Point Model for Historical Time Series Analysis

Political relationships often vary over time, but standard modelsignore temporal variation in regression relationships. We describea Bayesian model that treats the change point in a time seriesas a parameter to be estimated. In this model, inference forthe regression coefficients reflects prior uncertainty aboutthe location of the change point. Inferences about regressioncoefficients, unconditional on the change-point location, canbe obtained by simulation methods. The model is illustratedin an analysis of real wage growth in 18 OECD countries from1965–1992.