Academic journal article Revue Canadienne des Sciences de l'Administration

Persistence of Stock Return Volatility in Canada

Academic journal article Revue Canadienne des Sciences de l'Administration

Persistence of Stock Return Volatility in Canada

Article excerpt

It is widely recognized that returns on high frequency financial data exhibit a certain regularity in the variance represented by volatility clustering. This clustering phenomenon, which is indicative of a temporal dependence in the conditional variance of returns, was not successfully modelled until the introduction of the ARCH (autoregressive conditional heteroscedasticity) model by Engle (1982) and its subsequent extension, the GARCH model, by Bollerslev (1986). Since then, both models have been used in studies of stock returns (Akgiray, 1989), exchange rate behaviour (Baillie & Bollerslev, 1989; Hsieh, 1989), and the term structure of interest rates (Engle, Lilien, & Robins, 1987), to name just a few (see Bollerslev, Chou, & Kroner, 1992, for a review).

The results have been unequivocal: Strong ARCH and GARCH effects have been detected in most if not all of the studies undertaken, indicating that there is a lasting persistence in conditional variance. In a large number of studies, the findings point to conditional variances that are integrated, with the implication that many financial series may be wide-sense nonstationary in variance. This is consistent with the considerable literature which finds significant excess volatility in financial markets (see West, 1988).

Nevertheless, the detection of excess price volatility and of integrated conditional variances has not gone unchallenged. Critics such as Fama (1991) have emphasized that those findings could be the result of improperly modelling the conditional mean and/or variance of the financial series being examined. In particular, the use of ARCH models presupposes that the stochastic process is stationary, i.e., the unconditional variance of the series is constant through time-an assumption very hard to sustain when the financial data analyzed often span several decades. Moreover, the assertion has been made that failure to incorporate sources of nonstationarity in the ARCH process could lead to an exaggeration of the degree of persistence in conditional variance. It is in this vein that Diebold (1986) conjectured that models that do not take into account structural changes in the economic environment will bias GARCH estimates towards a unit root.

Monetary regime shifts have been identified by several researchers as potential structural changes susceptible of impacting on conditional volatility. Lastrapes (1989) noted that monetary policy regimes significantly affect the mean and variance of nominal exchange rates. Choi and Kim (1991), using a GARCH model in a study covering the period 1975 through 1989, found evidence of a significant empirical relationship between the conduct

of monetary policy and the behaviour of the foreign exchange risk premium. Their main result is that the time-variation of the risk premium depends on changes in the monetary regime. Simonato (1992) found that if potential structural changes (e.g., monetary regime shifts) are not taken into account, spurious GARCH residuals are obtained and the degree of integration is magnified. Diebold (1986) suggested that recognizing monetary regime changes may lead to stable (i.e., nonintegrated) GARCH movements within regimes but with unconditional jumps between the regimes. On this very issue, Hsieh (1991) claimed that changes in the operating procedures of the U.S. Federal Reserve Board (FRB) can shift the volatility of financial markets.

It is the main hypothesis of this paper that the accommodation of variations in expected returns and of nonstationarities in the variance process of stock return series leads to a significant reduction in the degree of persistence in conditional variance. To that purpose, a GARCH methodology is applied to the Toronto Stock Exchange (TSE) daily value-weighted index over the period 1976 to 1991. Following the example of the authors mentioned above, we first focus on changes in monetary policy regimes, using U.S. dates as proxies for the ensuing shifts in Canadian monetary policy, as potential sources of nonstationarities in the mean and variance of stock returns. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.