A number of studies investigate whether various stochastic variables explain changes in return volatility by specifying the variables as covariates in a GARCH(1, 1) or EGARCH(1, 1) model. The authors show that these models impose an implicit constraint that can obscure the true role of the covariates in the analysis. They illustrate the problem by reconsidering the role of contemporaneous trading volume in explaining ARCH effects in daily stock returns. Once the constraint imposed in earlier research is relaxed, it is found that specifying volume as a covariate does little to diminish the importance of lagged squared returns in capturing the dynamics of volatility.
INTRODUCTIONThe extensive use of generalized autoregressive conditional heteroscedasticity (GARCH) models in financial economics is testimony to their success in capturing volatility dynamics. As low-order GARCH and exponential GARCH (EGARCH) models typically perform well relative to more complex specifications (see, e.g., Hansen & Lunde, 2005a), researchers often use these models to investigate the relation between changes in return volatility and various stochastic variables. In particular, they assess whether the stochastic variables explain changes in volatility by including the variables as covariates in a GARCH(1, 1) or EGARCH(1, 1) model. The variables considered in the literature include interest rate levels (Engle & Patton, 2001;Glosten, Jagannathan, & Runkle, 1993), interest rate spreads (Dominguez, 1998;Hagiwara & Herce, 1999), forward-spot spreads (Hodrick, 1989), implied volatilities (Blair, Poon, & Taylor, 2001;Day & Lewis, 1992;Lamoureux & Lastrapes, 1993), futures open interest (Girma & Mougoue, 2002), a proxy for the information flow during the overnight market closure (Gallo & Pacini, 2000), and contemporaneous trading volume (Fujihara & Mougoue, 1997;Lamoureux & Lastrapes, 1990;Marsh & Wagner, 2005).Here a closer look is taken at the specification of GARCH models with stochastic covariates, highlighting a specification issue that makes it difficult to draw reliable inferences from many of the models considered in the literature. These models impose an implicit constraint that requires the coefficients on the lagged squared returns and the lagged stochastic variables to decline with the lag length at the same rate. This constraint is problematic for cases in which the stochastic variables provide little information about future return volatility beyond that contained in lagged squared returns. Obtaining precise fitted volatilities in such cases requires giving little weight to the lagged stochastic variables, but owing to the implicit constraint, this also requires giving little weight to lagged squared returns. Hence, if there is a strong contemporaneous relation between the stochastic variables and return volatility, the covariates can drive ARCH effects out of the fitted models regardless of whether they capture volatility persistence.The problem is illustrated by reconsidering the role of contemporaneous trading volume in explai...