Financial analysts' forecast dispersion has been used in a variety of contexts in accounting and finance studies. In this study, we provide large sample evidence on the cross-sectional determinants of forecast dispersion and examine to what extent analysts' strategic behavior biases the observed dispersion from the dispersion of unmanaged forecasts. We propose a method to estimate the dispersion bias for each sample observation. We find that observed dispersion, on average, understates dispersion of unmanaged forecasts by 53.4 percent and this downward bias varies considerably across firms. We further discuss the implications of the significant cross-sectional variation in the bias in observed dispersion for studies that rely on dispersion to estimate constructs such as consensus and information quality as well as those that use dispersion directly in research design. Data Availability: The data are publicly available for the sources indicated.

We examine the performance and compensation implications of firms' decisions to combine the roles of CEO and board chairman (duality). We document that firms that split the CEO and chairman positions due to investor pressure have significantly lower announcement returns and subsequent performance, and lower contributions of investments to shareholder wealth. Further, these performance outcomes are more negative for firms with higher predicted probabilities of duality based on a model of economic determinants of board leadership structure. We also find that pay-performance sensitivity in CEO compensation contracts are significantly lower following a split in the CEO and chairman positions, and significantly higher following a combination in these positions. Our evidence suggests that on average, board leadership choices by firms and market responses are consistent with efficiency arguments, and recent proposals for all firms to separate the CEO and chairman roles warrant more careful consideration.