-5. advantage of subtle nonlinear patterns-see Goodman(1982),Engel and Hamilton(1990), and Bilson(1990).Finally,the problem of joint hypothesis testing has led to some tests that have very low power to reject plausible deviations from market efficiency.A priori faith in the power of no-arbitrage has probably clouded the common sense recognition that human errors might persist or even compound rather than cancel each other. Nonetheless the efficient markets hypothesis provides a natural benchmark,and thus financial markets are a promising arena in which to investigate deviations from rational behavior.The research is facilitated by the facts that financial markets are liquid,and individual motivations are relatively straightforward,in accord with wealth-maximizing hypotheses (possibly modified by risk aversion).Moreover,the operation of these markets is thoroughly recorded,with rich price and volume data readily available. 1.2 Typology of markets The nonrational behaviors of individuals,although well documented,may not be apparent at the market level because many financial markets offer poaching potential(that is,players can take advantage of poor decisions made by others).More generally,aggregate social outcomes reflect both prevailing levels of individual rationality and opportunities for poaching,as table 1 shows schematically.Poachability will play an important role in determining which macro outcomes we study to draw inferences about levels of individual rationality. Table I.Models for Outcomes of Social Interactions: Poachability and Rationality Poachability/Arbitrage Potential Individual High Low Rationality Substantial/Full I.Efficient markets. II.Deviations due to sluggish information flow (e.g., mispriced open-end mutual funds). Bounded/Low III.Continual movement IV.Gross behavioral (behavioral to restore efficiency (see outcomes (e.g.,marriages- anomalies observed) Table 2). poor fits frequent;bargains -beneficial deals not struck; finance-inadequate personal savings)