Frontiers of Research in Economic Theory The Nancy L Schwartz Memorial Lectures, 1983-1997 "Leading economists presenting fundamentally important issues in economic theory" is the theme of the Nancy L Schwartz Memorial Lecture series held annually at the J L Kellogg Graduate School of Management of Northwestern University This collection of essays, drawn from the lectures delivered in the years 1983 through 1997, discusses economic behavior at the individual and group levels and the implications to the performance of economic systems Using nontechnical language, the speakers present theoretical, experimental, and empirical analysis of topics such as decision making under uncertainty and under full and bounded rationality, the influence of economic incentives and habits, and the effects of learning and evolution on dynamic choice Perfect competition, economic development, social insurance and social mobility, and negotiation and economic survival are other major economic subjects analyzed, advancing our understanding of economic behavior Econometric Society Monographs No 29 Editors: Peter Hammond, Stanford University Alberto Holly, University of Lausanne The Econometric Society is an international society for the advancement of economic theory in relation to statistics and mathematics The Econometric Society Monograph Series is designed to promote the publication of original research contributions of high quality in mathematical economics and theoretical and applied econometrics Other titles in the series: G S Maddala Limited-dependent and qualitative variables in econometrics, 521 33825 Gerard Debreu Mathematical economics: Twenty papers of Gerard Debreu, 521 33561 Jean-Michel Grandmont Money and value: A reconsideration of classical and neoclassical monetary economics, 521 31364 Franklin M Fisher Disequilibrium foundations of equilibrium economics, 521 37856 Andreu Mas-Colell The theory of general economic equilibrium: A differentiable approach, 521 26514 2, 521 38870 Cheng Hsiao Analysis of panel data, 521 38933 X Truman F Bewley, Editor Advances in econometrics - Fifth World Congress (Volume I), 521 46726 Truman F Bewley, Editor Advances in econometrics - Fifth World Congress (Volume II), 521 46725 X Herve Moulin Axioms of cooperative decision making, 521 36055 2,0 521 42458 L G Godfrey Misspecification tests in econometrics: The Lagrange multiplie principle and other approaches, 521 42459 Tony Lancaster The econometric analysis of transition data, 521 43789 X Alvin E Roth and Marilda A Oliviera Sotomayor, Editors Two-sided matching: A study in game-theoretic modeling and analysis, 521 43788 Wolfgang Hardle, Applied nonparametric regression, 521 42950 Jean-Jacques Laffont, Editor Advances in economic theory - Sixth World Congress (Volume I), 521 48459 Jean-Jacques Laffont, Editor Advances in economic theory - Sixth World Congress (Volume II), 521 48460 X Halbert White Estimation, inference and specification, 521 25280 6, 521 57446 Christopher Sims, Editor Advances in econometrics - Sixth World Congress (Volume I), 521 56610 X Christopher Sims, Editor Advances in econometrics - Sixth World Congress (Volume II), 521 56609 Roger Guesnerie A contribution to the pure theory of taxation, 521 23689 4, 521 62956 X David M Kreps and Kenneth F Wallis, Editors Advances in economics and econometrics - Seventh World Congress (Volume I), 521 58011 0,0 521 58983 David M Kreps and Kenneth F Wallis, Editors Advances in economics and econometrics - Seventh World Congress (Volume II), 521 58012 9, 521 58982 David M Kreps and Kenneth F Wallis, Editors Advances in economics and econometrics - Seventh World Congress (Volume III), 521 58013 7,0 521 58981 Nancy L Schwartz (Photo courtesy Northwestern University Archives) Frontiers of Research in Economic Theory The Nancy L Schwartz Memorial Lectures, 1983-1997 Edited by DONALD P JACOBS EHUD KALAI MORTON I KAMIEN n CAMBRIDGE ^ P UNIVERSITY PRESS CAMBRIDGE UNIVERSITY PRESS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo Cambridge University Press The Edinburgh Building, Cambridge CB2 2RU, UK Published in the United States of America by Cambridge University Press, New York www Cambridge, org Information on this title: www.cambridge.org/0521632226 © Donald P Jacobs, Ehud Kalai, and Morton I Kamien 1998 This publication is in copyright Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press First published 1998 A catalogue recordfor this publication is available from the British Library Library of Congress Cataloguing in Publication data Frontiers of research in economic theory: the Nancy L Schwartz Memorial Lectures, 1983-1997 / edited by Donald P Jacobs, Ehud Kalai, Morton I Kamien p cm - (Econometric Society monographs ; no 29) "Include[s] the lectures previously published by the Kellogg school in pamphlet form between 1983 and 1991, and again between 1993 and 1997, in the Econometric Society Monograph series" Foreword Includes bibliographical references ISBN 0-521-63222-6 - ISBN 0-521-63538-1 (pbk.) Economics I Schwartz, Nancy Lou II Jacobs, Donald P III Kalai, Ehud, 1942 IV Kamien, Morton I V J L Kellogg Graduate School of Management VI Series HB171.F783 1998 330-dc21 98-6159 CIP ISBN-10 0-521-63222-6 hardback ISBN-10 0-521-63538-1 paperback Transferred to digital printing 2005 Contents Editors' Foreword The Schwartz Lecturers Nancy L Schwartz Morton I Kamien Publications of Nancy L Schwartz page ix x xvii xxiii THE LECTURES 1983 Hugo Sonnenschein The Economics of Incentives: An Introductory Account 1984 Andreu Mas-Colell On the Theory of Perfect Competition 1985 Menahem E Yaari On the Role of "Dutch Books" in the Theory of Choice Under Risk 1986 Robert J Aumann Rationality and Bounded Rationality 1987 Robert E Lucas, Jr On the Mechanics of Economic Development 1988 Truman E Bewley Knightian Uncertainty 1989 Reinhard Selten Evolution, Learning, and Economic Behavior 1990 Vernon L Smith Experimental Economics: Behavioral Lessons for Microeconomic Theory and Policy 16 33 47 61 71 82 104 viii Contents 1991 Gary S Becker Habits, Addictions, and Traditions 1993 Peter A Diamond Issues in Social Insurance 1994 Robert B Wilson Negotiation with Private Information: Litigation and Strikes 1995 Roy Radner Economic Survival 1996 Nancy L Stokey Shirtsleeves to Shirtsleeves: The Economics of Social Mobility 1997 David M Kreps Anticipated Utility and Dynamic Choice 123 142 160 183 210 242 260 David M Kreps the level of a smell test at least, although (one hopes) something more systematic can be employed Recall Frank Fisher's (1989) critique of game theory as applied to Industrial Organization Fisher distinguished between exemplifying theory - theoretical models constructed to show logically consistent possibilities - and generalizing theory - theories that cut across a broad range of models, to establish conclusions that hold generally An example establishing the possibility of a Giffen good is exemplifying theory; the Theorems of Welfare Economics are examplars of generalizing theory Fisher asserted that Industrial Organization didn't need game theory to create exemplifying theory; Chamberlain, writing before game theory, already noted that in multi-period competition almost anything was possible, and all that game theory added along this dimension was unnecessary high-tech gloss And since game theory has proven so flexible a language, it has provided little or nothing in the way of generalizing theory I don't want to spend time in this essay debating Fisher's evaluation of game theory's contributions to I.O (The curious should see Kreps [1990, chapter 4].) But his dichotomy is helpful in responding to the Stokey criticism: The sort of models I am advocating will certainly make the task of exemplifying theory easier This, of course, weakens the impact of a piece of exemplifying theory per se However, there are empirical phenomena that are (at least) hard to exemplify with hyperrational individuals but that are easily exemplified with individuals who are limitedly rational in the sense of this essay (See Kreps, 1998.) To come to grips with these phenomena, I assert we need additional flexibility in modeling dynamic choice that matches the reality of individual dynamic choice On the other hand, to the extent that one grants the validity of models with this sort of individual behavior included, generalizing theory that generalizes across these models as well as across the standard hyperrationality-based models becomes more difficult But if one doesn't believe in hyperrationality, then generalizing theories that require it become suspect That is, allowing these sorts of models into the tent provides our (few) generalizing theories with higher hurdles to clear, but these are hurdles that ought to be cleared if the theories are to be accepted What Fisher decries - a decline in the number of generalizing theorems as the universe to be generalized becomes richer - is inevitable and desirable if we believe that the universe is richer than we are accustomed to assuming Assuming that preferences all exhibit weak gross substitutes allows us to prove that general equilibrium is unique - a nice piece of Anticipated Utility and Dynamic Choice 261 generalizing theory in terms of the strength of the conclusion, but not a useful one, given the strong assumption needed to obtain it So what substitutes for fewer generalizing theorems? Presumably the answer is more empirics, directed at the assumptions and the conclusions of specific models I don't want to go too far in this direction Economics has gotten a lot of mileage out of adherence to the twin rubric of rational (purposeful) behavior and equilibrium A wholesale abandonment of those standards is unwarranted But careful, limited, and circumspect abandonment ought to be attempted (at least) before it is condemned as heresy (5) While economists may think these models stray too far from the paradigm, other social scientists may think they have not gone far enough In assuming that current choices (from currently available options) are rational, I've dismissed things like framing effects applying at that level This is not done because I think framing effects are unimportant, but because I want to direct attention to a different array of choice-theoretic considerations; viz., those that have to with dynamics This is somewhat disingenuous: To the extent that dynamics complicate the decision problem, they (probably) make it more likely that the decision maker will be subject to framing effects For example, as the decision maker casts around for a simple model of her environment, the set of options presented to her may influence the model she builds Suppose her choice set consists of a and a!, which differ chiefly along dimension X If these are her only two options, she is likely to concentrate her thinking and model building exercises on models that feature dimension X But if her choice set consists of a, 6, c, d, and a\ she may be led to models that feature a different critical dimension, models that reverse the "apparent" relative values of a and ,; we might want to speak of fi(dc|)|i|^), where op, are conditions observable from the cabin prior to setting out on date t (Having mentioned these possible modeling complications, I will not carry through with them in subsequent discussion.) But what if the decision maker is unaware of the general conditions of the forest The story, after all, is that the decision maker is new to this forest Thus he may be unsure of the distribution of ) vis-a-vis w(x) In particular, there is no interaction whatsoever between x and any other x', or between x and the total weight being carried Thus the decision to carry x or not can be separated from the same decision for other members of X And now, following from the analogy of the multi-armed bandit problem, it is easy to argue that (with the constant discount rate) if x is ever dropped from the daypack, it will never be added back The precise calculation of the value of information from carrying x (given a prior on \i) is not impossibly hard, and if it is still too hard to carry out, reasonable heuristics (involving the distribution of vt(x, (()) above w(x), the length of time until the vacation ends, the precise value of a, and so on) are available But all this depends on the strong linear separability assumption (for 268 David M Kreps each day's utility) Items from x are neither substitutes for one another nor complementary This assumption may be reasonable in some contexts, but certainly it is too strong for most, and so we move on to simplifying assumptions a bit less stark than this one For example, suppose that the items of X are perfect substitutes That is, one and only one x e z will be brought out of the daypack Absent the value of information, the "natural" model is then ut(z,w) = JJmaxxezvt[x9w9^y[d^[d\i)9 where vt(x,w,$t) gives the value of the object x if w is being carried in circumstances