Performance measurement under differential information

The basic assumption of this study is that economic agents might be endowed with differential information. ^ In the first part, we investigate whether managers of firms announcing open-market stock repurchases have information superior to outside investors. To address the issue this study proposes a new measure of performance based on the partially revealing rational expectations model developed by Hellwig (1980) and Admati (1985). The new measure of performance conditions on the level of public information conveyed by the security price and therefore may help discern whether managers possess superior information (undervaluation hypothesis). We apply the new measure to a sample of Real Estate Investment Trusts (REITs) that announced open-market stock repurchases and find strong support for the undervaluation hypothesis. We use a control sample of REITs that did not announce open-market stock repurchases and show that our findings are robust to the interim trading bias emphasized by Ferson and Khang (2002) and other sources of misspecifications discussed by Grinblatt and Titman (1993).^ The second part of the study analyzes the market reaction to stock repurchase announcements. If the market underreacts to the initial announcement, then abnormal returns will be spread over long horizons. Lyon, Barber and Tsai (1999), among others, have argued that measuring performance over long horizon can be “treacherous”. We develop a new methodology to model the autocorrelation of monthly returns into long-horizon buy-and-hold abnormal return estimators. Serial correlation can introduce bias (autocorrelation bias) because the bid-ask bounce documented by Blume and Stambaugh (1983) affects monthly returns for sample firms and non-sample firms in a different fashion. Previous long-horizon event studies have overlooked this source of bias. We apply the new methodology to the case of stock repurchases for REITs and find compelling evidence for the underreaction hypothesis . The evidence is robust to different measures of the variance and the effects of cross-correlation of abnormal returns emphasized by Fama (1998) and Mitchell and Stafford (2000).