| Summary: | This dissertation examines implications of models of differential information that formalize the following intuition: securities for which there is relatively little information are perceived as relatively more risky because of the higher uncertainty surrounding the exact parameters of their return distributions.
The testable implications of the theoretical model depend on the nature of the assumed economy. In a small market setting we find that the presence of differential information will lead investors to make an upward adjustment to the measure of systematic risk that they employ for the pricing of low information securities relative to its measurement without regard for differential information. Further, the model predicts that the adjustment will decrease at a decreasing rate as information increases. Lastly, in the face of such an adjustment, the typical estimate of beta employed by researchers is found to be upwardly biased. On the other hand, the model predicts that the size of the adjustment to beta at each information level falls with the degree of positive cross-correlation between the low and high information securities, and the lack of cross-correlation in the information patterns for the low information securities. Thus, in unrestricted economies, where these sources of diversification are available, the presence of differential information should have little or no impact on market equilibrium.
The implication that systematic risk (ft) for low information firms should decline as information increases is confirmed with three independent samples. First, a sample of 142 newly-listed firms is used to examine the relationship between systematic risk and time following the listing date. Next, a sample of 348 NYSE firms is employed to further explore this relationship. Finally, a sample of 140 newly-listed firms is used to examine the change in systematic risk at the time of the first annual earnings announcement. We also consider the suggestion that an association between information availability and both firm size and period of listing might provide a partial explanation for the firm size and period of listing anomalies. Our evidence suggests that this is unlikely where historical returns are used to estimate beta. Prior historical return associations tend to overstate beta at a point in time rather than understate it. === Business, Sauder School of === Graduate
|
|---|
