Information asymmetry. The principal’s perceived performance, comes from the linear interpolation of the level of performance she perceives from inspections. Therefore, the information asymmetry is the difference between the perceived performance and the real performance. In Figure
Information asymmetry. The information asymmetry/signaling hypothesis predicts low stock returns prior to a repurchase, thus we control for the firm’s pre-repurchase stock returns, as well as the market returns (defined as the value-weighted market return from CRSP) to deal with the possibility that information asymmetry may also be related to the market condition. Yet our agreement hypothesis also predicts pre-repurchase low stock returns due to low investor-management agreement. Controlling for stock returns alone, thus, is not sufficient to draw a distinction between these two hypotheses. Furthermore, one may argue that some of our agreement proxies may be related to information asymmetry. As discussed later, we rely on firm fixed-effect estimation to explore within-firm variation in examining a firm’s repurchase decision. Therefore, the effect of any unobserved time-invariant firm characteristics that are related to information asymmetry will be eliminated. Additionally, large firms are more likely to be included in our sample as we use institutional ownership and analyst forecast data. The advantage of the estimation method and our sampling of relatively large firms both work to help alleviate the information asymmetry problem to some extent, which is more of a severe concern in cross-sectional studies and in small firms. Nevertheless, to further disentangle the effect of agreement from that of asymmetric information, we include several measures of information asymmetry that are not related to agreement.19 One proxy is the firm-specific stock return variation, Psi, developed in Durney, Xxxxx, Xxxxx and Xxxxxxx (2003). 20 Based on Roll’s (1988) observation that low R2 statistics for common asset pricing models is due to firm-specific return variation that is not associated with public information, they show that firms and industries with lower market-model R2 statistics exhibit higher association between current returns and future earnings. The idea is that greater firm-specific variation in stock prices implies that more information about future earnings is incorporated into the stock price and hence there is less information asymmetry. Specifically, the firm-specific stock return variation of a given firm-year is defined as Psi = log(1–R2), where R2 is obtained from regressing a firm’s weekly stock returns on market return and industry return (defined at the two-digit SIC level) for the given year. Intuitively, the higher the level of Psi, the lower is the information asym...
