What do these hypotheses imply about abnormal returns in the future? None suggests any obvious pattern for the relation ship between G and returns. Under Hypothesis I, if we interpret our test as a long-run event study, then there is no reason to expect any relationship once the market has fully priced the underlying "event" of corporate governance. The fact that this price adjustment is taking such a long time does not seem so surprising in light of the lengthy intervals necessary for much more tangible information to be incorporated into prices.24 Thus,to the extent that end-of-sample price adjustment is incomplete,complete, or has overreacted, the future relationship between G and returns could be negative, zero, or positive. Under Hypothesis II there is a similar dependence on whether past insider information has been fully incorporated into prices. Under Hypothesis III future return differences would be driven by the
relevant omitted characteristic; clearly, this hypothesis yields no clear prediction.