This paper introduces a compound GARCH/markov switching model to add flexibility to the GARCH model in order to model the volatilities of exchange rates in target zones subject to realignments. The compound volatility model endogenizes the weights given to realignments (and all other shocks) in the GARCH process. Previous GARCH applications to EMS exchange rates took polar positions by arbitrarily placing full or zero weight on realignment shocks. Markov switching in the student-t degrees-of-freedom parameter is shown to make the difference between rejection and acceptance of goodness-of-fit tests for four of the six EMS currencies studied.
https://doi.org/10.20955/wp.1994.016
follow @stlouisfed