Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Many empirical studies of financial volatility within the GARCH framework tend to exclude terms in the mean equation that are often proved to be statistically significant. We analyze analytically how omission of various mean terms affects the value of the ARCH parameter in the simple ARCH(1) model. We focus on the following terms missing from the mean equation when they are in fact present: a constant, a seasonal dummy, an autoregressive term, and a time-varying risk premium. We track how the relative distortion in the value of the ARCH parameter depends on amount of misspecification, and calibrate it to actual daily and monthly returns. It turns out that the effect on the variance equation of missing elements in the mean equation tends to be quite benign.