4 Conclusion
In this study we create a new measure of disagreement based on institutional trades in order to examine how the divergence of opinion among investors affects stock returns. We use the imbalance of stock-level buying and selling as a percentage of institutional trading to measure agreement (low values imply greater disagreement). Three hypotheses regarding the effect of disagreement on stock returns currently exist: the sidelined investor hypothesis, which predicts high initial and low subsequent abnormal returns; the information asymmetry hypothesis, which predicts low initial and high subsequent abnormal returns; and the unbiased prices hypothesis, which predicts that abnormal returns concurrent with and subsequent to periods of investor disagreement will on average be insignificant.
Our initial results, taken from the full sample, show little if any support for the information asymmetry hypothesis or the sidelined investor hypothesis. Thus at first blush the results appear consistent with the unbiased prices hypothesis. However, after sorting our sample by firm size, we find that for the smallest firms, disagreement among institutions does correspond to negative returns initially and an almost exact reversal in the following month as predicted by the information asymmetry hypothesis. When we sort our sample by institutional ownership as a proxy for shorting constraints and examine cases of disagreement, we find no evidence of positive initial returns followed by a reversal as predicted by the sidelined investor hypothesis. Additionally, when we employ alternative proxies for information asymmetry and shorting constraints, the results enumerated above remain largely unchanged.
