Rational Expectations and Regime Shifts in Macroeconometrics

Sammanfattning: This thesis consists of three papers. The first two papers explore implications of rational expectations when there is stochastic regime-switching; the third is an independent paper where underlying inflation is defined using rational expectations.Rational Expectations in a VAR with Markov Switching examines how a well known class of rational expectations hypotheses using linear vector-autoregressions (VAR:s) can be extended to allow for unobservable Markov switching between discrete states. This statistical model differs from those commonly used in the literature: the model here is easier to estimate and has the appeal that the state dependence is symmetric. The contribution of the paper is to derive simple expressions for the VAR forecasts under Markov switching; these forecasts are then used to find testable restrictions implied by rational expectations, which are linear when the forecast horizon is infinite. As an illustration, I examine a test of the expectations hypothesis (EH) on the short end of the maturity spectrum - three and six month US bills - and find that a non-rejection of the hypothesis in a previous paper, also with regime shifts, may be fragile.Term Premia Under Switching Regimes uses the methods discussed in the first paper to identify a conditional term premium over the long end of the maturity spectrum for US and Swedish data. Traditional tests of the EH using linear VAR:s have treated the premium as a constant, unexplained deviation from expectations. I use a well known present value model, but introduce a more flexible specification: the premium is assumed to depend on the current state only, thus allowing for different premia across states. Despite using a more flexible model, the EH hypothesis is still statistically rejected; the rejection is not sensitive to small changes in the specification, but the premium is highly sensitive to small changes in the discount factor. Nevertheless, the EH performs well in terms of goodness of fit.Underlying Inflation - A Common Trends Approach uses economic theory based restrictions on a VAR with output and prices to compute underlying inflation. For policy purposes, headline inflation has some undesirable properties, such as being affected by changes in taxes, lack of smoothness, but most importantly, is not consistent with nominal shocks being (exactly) output neutral in the long run. I use an identification scheme proposed in the literature to define underlying inflation as the component of changes in nominal prices which have only transient effect on output, but use a new method to implement the restrictions, which may be simpler to use and interpret. I use this method to calculate core inflation for Canada, Germany, Italy, Japan, Sweden, the UK, and the US.