Harrison Hong and Jeremy Stein: Disagreement and the Stock Market
They make the case for "disagreement models" as the way forward for behavioral finance:
http://www.economics.harvard.edu/faculty/stein/papers/Disagreement-Dec-2006.pdf The enduring appeal of classical asset-pricing theory over the last several decades owes much to its success in forging a consensus around a foundational modeling platform. This platform consists of a core set of assumptions that have been widely-accepted by researchers working in the field as reasonable first-order descriptions of investor behavior, and that—-just as importantly—-lend themselves to elegant, powerful and tractable theorizing.
If behavioral finance is ever to approach the stature of classical asset pricing, it will have to move beyond being a large collection of empirical facts and competing one- off models, and ultimately reach a similar sort of consensus. While this goal seems well within sight in the part of the field that explores limits to arbitrage, it is much further away in the part that seeks to understand the origins of market mispricings. Many horses are still running in this latter race, and it is not at all clear whether a decisive winner will emerge in the foreseeable future.
Nevertheless, our view is that, taken collectively, the disagreement models described above represent the best horse on which to bet. Disagreement models uniquely hold the promise of being able to deliver a comprehensive joint account of stock prices and trading volume, which we consider to be one of the highest priorities for theoretical work in asset pricing. Moreover, the modeling ingredients are highly tractable, for two reasons. First, preferences are taken to be completely standard—-so the modeler can work with constant absolute risk aversion utility or whatever other functional form makes life easiest. Second, in disagreement models, investors’ beliefs are often a simple function of just their own priors and the signals that they each observe directly; this is in contrast to rational-expectations models, where each investor must also update based on inferences about others’ priors and signals. As a result, disagreement models can be usefully deployed in everything from simple illustrative examples (like the two-period model of momentum discussed earlier in this paper) to elegant continuous-time formulations (like Scheinkman and Xiong, 2003). We hope that future research with disagreement models continues to develop their potential.
The classical finance assumptions were never reasonable first-order descriptions of investor behavior; the most that was sever claimed was that they were reasonable first-order descriptions of those components of investor behavior that did not cancel themselves out.
And the fact that people in disagreement models aren't making inferences from price patterns strikes me as a defect, not a feature...
