Are behavioral asset-pricing models structural外文电子书.pdfVIP

Journal of Monetary Economics 49 (2002) 229–233 Comment on: Are behavioral asset-pricing models structural? $ Jessica A. Wachter* Stern School of Business, New York University, New York, NY 10012, USA Received 14 May 2001; received in revised form 5 September 2001; accepted 6 September 2001 1. Introduction According to Stanley Zin, financial economists on both sides of the behavioral finance debate have failed to ask the right questions. To Zin, the most important question we should ask about behavioral models is whether they make accurate predictions. Structural models at least have prediction as their goal, whether or not they achieve it. Implicit in this view is the notion that structural models somehow ‘‘explain what is going on’’, that they can offer insights into why economic phenomena occur. In most other fields of science, evaluating a structural model would not be a matter of debate. Because the goal of a structural model is prediction, the model would be evaluated directly using a controlled experiment. If in finance we could somehow change the parameters of our economy, or at least statistically detect such a change, we would also have a direct test of a structural model. The former is clearly impossible; the latter, Zin argues, is unreliable. The question of how to evaluate a structural model without experiments is clearly a difficult one. Instead of attempting an answer, Zin argues that two commonly used criteria are wrong. One of these criteria, that a model appear realistic, looks good at first glance. But, quoting Friedman and Lucas, Zin argues that we are not trying for $ Comments on ‘‘Are Behavioral Asset-Pricing Models Structural?’’ by Stanley E. Zin prepared