Detection of Structural Changes Without Using P Values
The econometrics of structural change has evolved a lot since the classical Chow  test. Several approaches have been proposed to find the unknown breakdate. But they could be invalid as it is claimed that the P value has been misused for the past one hundred years. This paper reviews other methods of detecting structural changes. Specifically, the Bayes factor can be used for a pairwise comparison of competing models. The Markov-switching model is an effective way of dealing with a number of discrete regimes. But if the regime is a continuous normal variable, the Kalman filter is a better resolution.
KeywordsStructural changes Bayes factor Markov-switching model Kalman filter