Testing Rational Expectations in a Cointegrated VAR with Abrupt Structural Change

The rational expectation hypothesis is widely used in finance and macroeconomics. A natural research question comprises investigating whether models that use this hypothesis can fit the data well. Researchers have been developing econometric procedures to test rational expectation models. Johansen and Swensen showed how to test rational expectation restrictions in the case where the data generating process is a cointegrated vector autoregressive model. This study aims to achieve three objectives. The first objective is to extend Johansen and Swensen's framework to the case where the data generating process is a cointegrated vector autoregressive model with abrupt structural change (CVAR-SC). The second goal is to show that the type of rational expectation restrictions analysed in this paper imply co-breaking, as defined by Hendry. Finally, the restrictions on the CVAR-SC parameters implied by the present value model, which is a particular rational expectation model, are analysed and derived, and a test is developed. Two empirical exercises are reported. The first is Engsted's dataset and the second uses the dividend and share prices of an important Brazilian retail bank.