Working Paper: NBER ID: w4834
Authors: Charles Engel; Craig S. Hakkio
Abstract: Exchange rates of currencies in the Exchange Rate Mechanism of the EMS are characterized by long periods of stability interrupted by periods of extreme volatility. The periods of volatility appear at times of realignments of the central parities and at times when the exchange rate is within the ERM bands. We begin by considering a procedure for finding outliers based on measuring distance as a quadratic form. The evidence suggests that the exchange rates of the EMS can be described by a mixture of two distributions. We therefore model the exchange rate as switching between two distributions--one that holds in stable times and the other that holds in volatile times. In particular, we use Hamilton's Markov-switching model. In addition, we extend Hamilton's model by allowing the probability of switching from one state to another to depend on the position of the exchange rate within its EMS band. This model has the interesting implication that near the edge of the band, large movements--either realignments or large jumps to the center of the band--are more likely if the move to the edge of the band has been precipitous.
Keywords: exchange rates; European Monetary System; Markov-switching model; volatility; outliers
JEL Codes: F31; F33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| state of the exchange rate (stable vs. volatile) (F31) | likelihood of outlier observations (C52) |
| exchange rate's position within its EMS band (F33) | probability of switching between stable and volatile states (C62) |
| large movements approaching band edges (F20) | likelihood of volatility (G17) |
| previous state of exchange rate (F31) | probability of being in a volatile state (C62) |
| EMS exchange rates (F31) | more outliers than floating exchange rates (F31) |
| EMS exchange rates (F31) | volatility (E32) |