تغییرات ناگهانی در تداوم واریانس و نوسانات در بازار ارز خارجی

Participants in the foreign exchange markets should be concerned with how a major event may lead to a sudden change in volatility as well as the role that shocks play in determining the persistence of volatility over time. This paper examines these issues by first identifying the time periods of sudden changes in volatility and then examining economic events surrounding those shifts. This research detects time periods of sudden changes in variance (i.e. regime shifts) by using the iterated cumulated sums of squares (ICSS) algorithm. Examining five major exchange rates from January 1990 to September 2000, it is found that accounting for volatility shifts in the GARCH model considerably reduces the persistence in volatility. The results suggest that many previous studies may have significantly overestimated the degree of volatility persistence that exists in financial time series.

The volatility of exchange rates has been an important issue in financial economics since the collapse of the Bretton Woods system. Volatility in exchange rates may make the exchange rate-adjusted value of foreign sales less predictable, thus discouraging domestic firms from engaging in international trade. Engel and Hakkio (1993) show that high volatility in exchange rates can also hinder the flow of capital across countries. The disrupting effects of exchange rate volatility on international trade and the economy have been widely documented in the literature (see Eichengreen and Irwin, 1993, Kroner and Lastrapes, 1993 and Caporale and Doroodian, 1994). Investors and policymakers may benefit from a better understanding of how major economic events can correspond to sudden changes in volatility in exchange rates and how shocks will affect volatility over time. The persistence in volatility is a key ingredient for accurately predicting how events can affect future exchange rate volatility and consequently its effects on international trade, capital flows and the economy. The present paper studies these issues and examines weekly nominal exchange rates for five major countries from January 1, 1990 to September 6, 2000. An iterated cumulated sums of squares (ICSS) algorithm is used to identify the time periods of volatility shifts. Economic events surrounding the time points of increased volatility are then analyzed. These volatility shifts are then introduced in a GARCH model to compute the effect of a given shock on persistence of volatility. Porterba and Summers (1986) have developed an asset pricing model which explicitly shows that the amount of persistence in volatility directly affects the price of an asset. Specifically, they show that an increase in expected volatility persistence will reduce the current price of an asset. Since exchange rates are asset prices, finding the ‘true’ persistence in volatility in exchange rates is important for building accurate asset pricing models, forecasting future exchange rate volatility and will further our understanding of exchange rate markets.1

The present paper detects time periods of sudden changes in volatility by using the ICSS algorithm for five major currencies. Major economic and political events correspond to volatility breaks in foreign exchange markets. These volatility shifts are incorporated within a standard GARCH framework to provide a more accurately measured effect of a shock on volatility persistence. The results are consistent with earlier research that argued that incorporating regime shifts in the GARCH model may significantly lower volatility persistence. However, unlike most of these earlier studies, the results presented here are based on endogenously determined regime shifts. Since the strength of the relationship between asset prices and volatility is directly linked to the extent of volatility persistence (see Porterba and Summers, 1986), the above information is a key element in building more accurate asset pricing models and improving forecasts of exchange rate volatility. These results suggest that it may be appropriate to re-evaluate the conclusions of studies on volatility persistence that do not take into account regime shifts as they may overstate the degree to which shocks affect volatility.