Working Paper: CEPR ID: DP1594
Authors: Michael J. Artis; Wenda Zhang
Abstract: This paper considers the evidence for volatility clustering and transmission in six bilateral Deutsche mark ERM exchange rates. Data on daily exchange rate changes are described by a mixture of two normal distributions. One of these contains observations of volatile exchange rate changes while the other pertains to tranquil periods. Using the information given by the two distributions, each observation is classified to one or other category. The phenomenon of volatility clustering in a given bilateral exchange rate series is then studied by means of a non-parametric test, while volatility transmission across exchange rates is studied by means of non-parametric tests for independence.
Keywords: exchange rates; volatility clustering; volatility transmission; ERM
JEL Codes: C14; F31; G10; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| average volatility fell during the EMS period (F36) | decline in the mean value of volatile exchange rate changes (F31) |
| volatility clustering persists despite this decline (E32) | tendency for large movements to cluster remained unchanged (E32) |
| volatility transmission across exchange rates is significant (F31) | French franc acted as an alternative anchor to the Deutsche Mark (F31) |
| shocks in the earlier and later periods were stronger than in the middle period (E65) | time-variation in transmission strength (C22) |