Hirshleifer, David, Silva Lopes, Artur, Floris Heukelom, Day, John B. Taylor, More about this item Statistics Access and download statistics Corrections All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:aea:jeclit:vyip See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact:. If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about. If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form.
If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation. For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: Michael P.
Albert email available below. Please note that corrections may take a couple of weeks to filter through the various RePEc services.
Economic literature: papers , articles , software , chapters , books. FRED data. My bibliography Save this article. Handle: RePEc:aea:jeclit:vyip as. Most related items These are the items that most often cite the same works as this one and are cited by the same works as this one.
More about this item Statistics Access and download statistics. Corrections All material on this site has been provided by the respective publishers and authors.
Louis Fed. This paper investigates the bounded rational route choice problem with the framework of quantal response equilibrium in which users are noisy optimizers to make route choice decisions. In the … Expand. View 8 excerpts, cites methods and background. Addressing the information problem in agriculture via agrobiodiversity: Streamlining the issues, challenges and policy questions.
The purpose of this paper is to show how agrobiodiversity addresses the inherent information problem that agriculture faces and present the issues, challenges and policy questions that need to be … Expand. View 6 excerpts, cites background. Agent sensing with stateful resources. Bounded rationality and irreversible network change.
A network change is said to be irreversible if the initial network equilibrium cannot be restored by revoking the change. The phenomenon of irreversible network change has been observed in reality.
This paper looks at the philosophical foundations of rational-choice theory. It is argued that L. Savage's expected-utility axioms cannot be defended as requirements of instrumental rationality, in … Expand.
Highly Influential. View 14 excerpts, references background. This paper presents calculations of the utility cost to consumers of following alternative decision rules in the environments specified by tests of the intertemporal allocation of consumption on … Expand.
View 9 excerpts, references background and methods. View 17 excerpts, references background. Stochastic learning theory, View 16 excerpts. Some Elementary Selection Processes in Economics. The famous M. Friedman and Leonard Savage billiards expert plays as if a master of the laws of physics. But what of a beginner taking the first shot, in poor light, on a badly warped and randomly moving table, with assorted friends and relatives guiding the cue stick?
Is a young person making life cycle decisions more like the expert player or more like the beginner? Argument 3. Survivors and tricksters. Agents who do not optimize will not survive.
The survival argument is associated with the classic papers by Armen Alchian and M. Friedman , and has been critically evaluated in general terms by many authors, notably Winter , and Nelson and Winter Early on, Koopmans , pp. The models show Argument 3 to be highly conditional. Nonoptimizing firms survive under some conditions but not under others. In the presence of deliberation cost, for example, survival logic may favor a cheap rule of thumb over a costly optimization.
The survival argument carries lesser force for individuals than for firms. We commonly read in the financial pages that firms fail for lack of profits, but we seldom read in the obituary pages that people die of suboptimization.
Consumers who display wasteful shopping patterns can survive at a lower standard of living, and workers who use their talents wastefully can survive at a lower wage. There is a more subtle survival argument for individuals. Rules of thumb are typically exploitable by tricksters, who can in principle money pump a person. As an example of his view, Becker , p. However, Grossman himself notes p.
Did utility maximization suggest the patterns Grossman found, or did it merely package them? Whatever the truth about the particular case, economic research often seems to work backwards from empirical findings to whatever utility maximization will work. Where the empirical arrow falls, there we paint the utility bullseye.
Putting the issue a bit differently, Arthur Goldberger , Simon , and Arrow note that utility maximization has little empirical content without strong auxiliary assumptions on the utility functions and other model ingredients. Because a trained economist can see through a utility maximization, stating auxiliary assumptions is often little different from stating empirical predictions outright, as, say, a sociologist might. In this sense, the utility maximization merely packages the prediction.
Argument 5. Sidewalk twenties. A model of unbounded rationality identifies an agents best opportunity for gain. Because it is implausible for an agent to forgo opportunities for gain, unbounded rationality identifies the agents likely action.
In the rational expectations literature, the sidewalk twenties argument appears as the claim that an agent with suboptimal expectations would be consistently fooled into forgoing opportunities for gain.
In the face of such deliberation cost, the walker may walk on by. Similarly, deliberation cost may make rational expectations cost more than they are worth. Unboundedly rational optima, by neglecting deliberation cost, may identify false opportunities for gain. Argument 6. Discipline and ad hocery. Without the discipline of optimizing models, economic theory would degenerate into a hodge podge of ad hoc hypotheses which cover every fact but which lack overall cohesion and scientific refutability.
Discipline comes from good scientific practice, not embrace of a particular approach. Any approach, including the optimization approach, can lead to an undisciplined proliferation of hypotheses to cover all facts.
Conversely, a bounded rationality hypothesis might produce a parsimonious explanation of a variety of empirical patterns. For example, Shefrin and Statman use their behavioral capital asset pricing model to address various financial anomalies as a group. A merit of the deliberation cost idea is that it suggests a discipline for models of bounded rationalitythat departures from unbounded rationality be systematically related to the deliberation cost involved.
Argument 7. Tractability and definite outcomes. The unbounded rationality postulate, because it can be formulated through well understood mathematical optimizations, confers tractable analysis and definite outcomes. Consider tractability.
Because optimizations may be arbitrarily complex, whereas bounded rationality may be represented by simple rules of thumb, optimization-based models are sometimes more and sometimes less.
Whenever theory and evidence suggest a need to settle the sparsely populated areas between clusters, science says welcome. In summary, the standard arguments for unbounded rationality, despite their great influence, are too extreme to be convincing. Put in more flexible form, however, the arguments contain many useful insights about conditions favoring one or another treatment of rationality.
Fortunately, economists are coming to adopt more flexible interpretations. Even Becker, perhaps pushed by his long-standing interest in nonstandard costs, has recently opened the door a crack for deliberation cost. In his Nobel lecture , Becker says: Actions are constrained by income, time, imperfect memory and calculating capacities, and other limited resources p. No Free Lunch, Yes Bounded Rationality It is evident that the rational thing to do is to be irrational, where deliberation and estimation cost more than they are worth.
Frank Knight , p. A spectacular example of the latter is recent macrotheory. By insisting on rational expectations and intertemporal optimization, which are quite intractable in general settings, macrotheory is often reduced to considering only a single representative agent. Arrow , James Tobin , Robert Solow , and Kirman note the strange sacrifices required for the ritual purity of optimization-only models Akerlof and Yellens phrase, , p.
Consider definite outcomes. For an agent with a well behaved objective function, it is argued that an optimization gives one model and one outcome, whereas adaptation, which may take different forms, may give many models and many outcomes. The main response is that, even if we insist on looking at only one model, evidence and plausibility should be the criteria, not prior bias toward optimizations. In any case, the one optimization model may generate multiple equilibria and thus multiple outcomes, whereas the adaptive models may all converge to the same one of the multiple equilibria and thus generate a single outcome.
This equilibrium-selection issue motivates a number of the adaptive models cited in Section II. Argument 8. Economics is by definition the study of optimizing behavior; bounded rationality is the province of other disciplines. By its most common definition, economics concerns scarcity. Because human reasoning ability is scarce, one could as well argue that economists are by definition required to study bounded rationality.
More important, economics as a science must view every theory, including optimization theory, as open to empirical challenge. Regarding the province metaphor, scientific disci-. Human cognition is a scarce resource, implying that deliberation about economic decisions is a costly activity. To avoid a free lunch fallacy, it can be argued, we are forced to incorporate deliberation cost, and thus bounded rationality, in economic models.
There are special problems. Economizing Economizing: The Regress Issue Unbounded rationality is typically formulated as the assumption that a. Because it is a routine exercise to include one more cost in an optimization model, a treatment of deliberation cost seems straightforward at first glance. Simply include that extra cost. However, we quickly collide with a perplexing obstacle. Suppose that we first formulate a decision problem as a conventional optimization based on the assumption of unbounded rationality and thus on the assumption of zero deliberation cost.
Suppose we then recognize that deliberation cost is positive; so we fold this further cost into the original problem. The difficulty is that the augmented optimization problem will itself be costly to analyze; and this new deliberation cost will be neglected. We can then formulate a third problem which includes the cost of solving the second, and then a fourth problem, and so on. We quickly find ourselves in an infinite and seemingly intractable regress.
In rough notation, let P denote the initial problem, and let F. There are two difficult issues here: i what the operator F looks like and ii how to deal with the regress. Start with ii.
Few authors mention the regress issue, and most mentions are little more. Examples: It might. Though I have not explored the latter possibility carefully, I suspect that any attempt to do so leads to fruitless and endless regression. Savage , p. Winter , p. The question of how far to go. Should one try to analyse the question of how to strike an optimal balance ,. At some point a decision must be taken on intuitive grounds.
Leif Johansen , p. Other early mentions of the regress issue are in Howard Raiffa , p. Perhaps the most succinct summary of the issue is Day and Pingles phrase economizing economizing , p.
If we can economize on economizing, then we can economize on economizing on economizing, and so on. Given the vast number of expositions of choice theory, it is remarkable how infrequently the regress issue is mentioned. I have found only three papersPhilippe Mongin and Bernard Walliser , Holly Smith , and Barton Lipman which discuss the regress issue in any detail. The regress problem seems to block any effort to maintain optimization as the ultimate logical basis for all behavioral modeling.
How can we formulate an optimization problem which takes full account of the cost of its own solution? We seemingly must yield to the idea that some behavioral hypothesis other than optimization, such as learning or adaptation, is needed to escape the regress.
In Johansens words, At some point a decision must be taken on intuitive grounds. In practical modeling, then, what. XXXIV June senting the unpredictability of deliberation else the agent would know the answer to begin with. It seems sensible to focus on only the first two problems, P and F P.
Problem P asks what the perfect decision is, and problem F P asks in addition how much costly deliberation the decision maker should expend in approximating the perfect decision. These are sensible behavioral questions. Problem F2 P asks in addition how much deliberation the decision maker should expend deciding how much deliberation to expend approximating a perfect decision.
This problem seems overly convoluted, and F 3 P , F4 P ,. Although the regress as a whole is worthwhile to notice, because it helps to put issues in perspective, practical modeling might, at least initially, neglect all problems beyond P and F P. In any case, that is what economists have done. Let X have unique optimizer X. Suppose that the decision maker has enough information in principle to compute the value of X for any X and thus to find X.
Thus, consider a deliberation technology by which the decision maker produces an approximation X to the perfect decision X. Let T be the costly effort devoted to approximating X, where C is the cost of one unit of T. Let X T be the actual decision resulting from this costly deliberation, and let X0 be a ruleof-thumb decision that the agent could use for free zero deliberation. Finally, let u be a random disturbance repre-. Consider the intuition of 1 under these assumptions.
In between, 1 gives a mix among rule-of-thumb behavior, deliberation, and random noise. The mix dictates the decision makers degree of rationality for the problem at hand. Algebraically specific deliberation technologies of form 1 are used in Conlisk , forthcoming and Evans and Ramey To the possible criticism that 1 doesnt look much like human cognition, we might note that a CES production function doesnt look much like a factory floor.
In representing a deliberation technology as in representing a production technology, the object is not faithfulness to cognitive science or to engineering. Rather the object is a simple relationship for representing economic tradeoffs. In this example, the original problem. P is to choose X to make X large. Summarizing: Original problem P. Choose X to make X large. Augmented problem F P. Four Rationalities The problems P and F P suggest a rough way of categorizing treatments of rationality in the literature.
Most models treat the original decision problem P. Only a few add a deliberation technology and treat the augmented problem F P. Among models treating P, there are two ways to close the model. Either the decision maker optimizes, or the decision maker uses some other behavioral rule, call it an adaptive rule. Among models treating F P , there are the same two ways to close the model.
Either the decision maker optimizes in the sense of finding the optimal deliberation effort to devote to the choice, or the decision maker follows some adaptive rule in choosing deliberation effort.
This gives four categories of models. Treat problem P. Optimal closure. Adaptive closure. Treat problem F P. The categories are in decreasing order of size. Category 1 comprises models of unboundedly rational choice, the vast majority of models in the literature. Category 2 includes models of bounded rationality in which adaptive choice rules are specified outright, with no deliberation technology or explicit treatment of. This category includes the vast majority of models of bounded rationality surveyed in Section II.
Categories 3 and 4, which require specification of a deliberation technology, contain only the very few models surveyed in the final three paragraphs of Section II.
Consider Category 3. It supposes that we have specified a deliberation technology and that the decision maker chooses the optimal amount of deliberation. Thirty years ago, William Baumol and Richard Quandt , p. In their words, One can easily formulate the appropriate. We might quarrel with the words easily formulate, because Baumol and Quandt did not in fact present a model of optimal imperfection, nor have many authors since.
If F P is viewed as a stopping problem when to stop deliberating and take final action , then optimal imperfection means optimal stopping. However, there is a problem. Why would a decision maker who cannot optimize relative to problem P be able to optimize relative to problem F P , which will often be more complicated?
Optimal imperfection returns us to the regress. Nonetheless, in the literature, the few models which treat problem F P often do invoke optimal imperfection.
What is the defense? Taking a dynamic view, we. Among economists, however, I could have claimed that, given the spatial distribution of lamp posts, the expected utility of bird watching exceeded the expected disutility of a collision. Ex ante, the post probably was not there, and it is entirely rational to collide with an ex post post.
This example illustrates the confounding of rationality issues with information issues. Am I dumb to walk into a post or merely a rational victim of imperfect information? Expanding the deliberation technology idea, it is natural to view decisions as produced by a decision technology with two inputs, costly information-gathering and costly deliberation.
The similarity of information-gathering and deliberation, as joint inputs in producing a decision, suggests that models of deliberation, as they evolve in economics, will inevitably have a general resemblance to existing models of information collection. For example, the illustrative deliberation technology above resembles some sampling models, with T the analog of a sample size. It is curious that such similar economic issues, costly deliberation and costly information collection, have been treated so differently in standard economics, one avoided and the other embraced.
In practice, the difference in treatment has required that anything resembling imperfect deliberation be passed off instead as imperfect information. For example, Williamson , is a towering figure in industrial organization for his insights about transactions cost. Although he sees these costs as rooted in bounded rationality, formal. A model in Category 4 might adapt over time into a model in Category 3, just as, by more familiar adaptive logic, a model in Category 2 might adapt over time into Category 1.
However, the conditions for such convergence from Category 4 to Category 3 seem delicate. Finally, we must assume that the decision maker does manage to converge. Because these conditions are delicate, a modeler may have to justify optimal imperfectionCategory 3as nothing more profound than a compromise of expedience. Category 3, by taking direct and explicit account of deliberation cost and the tradeoffs it implies, is already a big improvement over Categories 1 and 2, which comprise most of the existing literature.
Though optimal imperfection closes a model with an optimization, it is not a retreat to some new form of unbounded rationality.
An unboundedly rational decision maker optimizes every setting; whereas an optimally imperfect agent does not. This difference is large. In the example, an optimally imperfect X mixes rule-ofthumb behavior, deliberation, and random noise. Another example is the famous Gang of Four explanation of why we observe cooperation in finitely repeated prisoners dilemma games even though the familiar unraveling argument of game theory predicts failure to cooperate.
Although the observed behavior appears to be boundedly rational, Kreps et al. They assume that, although both players really are unboundedly rational, one player thinks the other might be boundedly rational. This clever and strained? The Gang of Four approach is in sharp contrast to Seltens approach to the chain store game, another game in which the unraveling logic produces a counterintuitive prediction. Selten faces the bounded rationality issue directly and sketches a theory of bounded rationality, including a brief discussion of deliberation cost and, implicitly, of the regress issue.
See also Selten and Rolf Stoecker and Selten , especially p. To gain perspective, it is entertaining to imagine an accidentally different history for economic theory. Imagine that modern decision theory began, not with perfect rationality and imperfect information, but with the opposite. Observed behavior that seemed to be the result of imperfect information was instead passed off by clever economists as the result of bounded rationality.
As the idea caught on, strict conventions for proper treatment of bounded rationality developed. Scholars departing from the conventions, or even worse from the perfect information postulate, were chastised as ad hoc and were firmly guided back to.
No one claimed that information was literally perfect in real life, merely that agents learned their own situations well enough to act as if perfectly informed; after all, those who didnt would be driven out of business by those who did. Elephants in the Living Room Deliberation cost and bounded rationality, like elephants in a living room, are sometimes just too much to ignore. Standard economics is forced to recognize their presence, if not to refer to them by name.
Consider two examples, human capital and technical change. People spend much on human capital, in large part through schooling. The investment is partly information collection names and dates , partly skill acquisition typing , and partly general cognitive investment learning to think. The cognitive investment must be a response to bounded rationality. Consider a deliberation cost interpretation.
Deliberation cost can be specific to a particular decision, as in the F P illustration above; or it can be the general cost of all-purpose cognitive training used in many decisions over many years. The part of schooling cost which goes into general cognitive development is general deliberation cost, and human capital theory is implicitly concerned with bounded rationality.
The assumption that students invest optimally in schooling is an unusually strong example of optimal imperfection. Explicit recognition of the relation of human capital theory to bounded rationality might bring new insights to the theory. As a second example, consider technical change. Many technological innovations result from insights that would have been made years earlier if people really could draw all possible inferences from existing information.
In this sense,. As with other model ingredients, however, we in practice want to work directly with the most convenient special case which does justice to the context.
The evidence and models surveyed suggest that a sensible rationality assumption will vary by context, depending on such conditions as deliberation cost, complexity, incentives, experience, and market discipline.
Beyond the four reasons given, there is one more reason for studying bounded rationality. It is simply a fascinating thing to do. We can mix some Puck with our Hamlet. An economic theorists book of tales. Cambridge: Cambridge U. Press, Procrastination and Obedience, Amer. Rational Models of Irrational Behavior, Amer. Uncertainty, Evolution, and Economic Theory, J. Judgment and decision making: An interdisciplinary reader. B RIAN. Evolutionary Econ. Inductive Reasoning and Bounded Rationality, Amer.
Yet, according to various models of research and development, decision makers engage in optimal amounts of search for the unexploited opportunities, as if unboundedly rational on that dimension.
We can view the search cost as in part deliberation cost, and we can view the optimal search assumption as an example of optimal imperfection. If the relation of technical change to bounded rationality were recognized openly as in the evolutionary models surveyed in Nelson , standard models of technical change might be better. Final Words Why bounded rationality? In four words one for each section above : evidence, success, methodology, and scarcity. In more words: Psychology and economics provide wide-ranging evidence that bounded rationality is important Section I.
Economists who include bounds on rationality in their models have excellent success in describing economic behavior beyond the coverage of standard theory Section II. The traditional appeals to economic methodology cut both ways; the conditions of a particular context may favor either bounded or unbounded rationality Section III. Models of bounded rationality adhere to a fundamental tenet of economics, respect for scarcity.
Human cognition, as a scarce resource, should be treated as such Section IV. The survey stresses throughout that an appropriate rationality assumption is not something to decide once for all contexts. In principle, we might suppose there is an encompassing single theory which takes various forms of bounded and unbounded rationality as special.
The Economics of Rumours, Rev. Stochastic models for social processes. New York: Wiley, Risk Uncertainty, Mar. The economic approach to human behavior. Chicago: U. Microsimulated transactions model of the United States economy. Baltimore: Johns Hopkins U. Learning to Be Rational, J.
Theory, Aug. Evolution and Market Behavior, J. Theory, Oct. Culture and the evolutionary process. Theory, Apr. Rational Learning and Rational Expectations, in Arrow and the ascent of modern economic theory. New York: New York U. Press, , pp. Rational Routes to Randomness. Economics Working Paper Santa Fe Institute, Bubbles and Fads in Asset Prices, J. Surveys, , 3 1 , pp.
Dordrecht: Kluwer Academic, , pp. ET AL. Cambridge: MIT Press, , pp. Letters, Sept. Strategy and structure. Cambridge: MIT Press, The Nature of the Firm, Economica, Nov. The Institutional Structure of Production, Amer. Costly Optimizers versus Cheap Imitators, J. Competitive Approximation of a Cournot Market, Rev.
Optimization Cost, J. The Utility of Gambling, J. Risk Uncertainty, June a, 6 3 , pp. Oxford: Oxford U. Press, b, pp.
Bounded Rationality and Market Fluctuations, J. Expectations and the structure of share prices. Adaptive Dynamics. Parts I and II. Special issues of Games Econ. Adaptive Dynamics in Coordination. Does the Stock Market Overreact? Finance, July , 40 3 , pp. Do Security Analysts Overreact? Working paper NBER, Rules of Thumb for Social Learning, J. Ulysses and the sirens: Studies in rationality and irrationality. Sour grapes: Studies in the subversion of rationality.
Social Norms and Economic Theory, J. Perspectives, Fall , 3 4 , pp. Hierarchical Decomposition in Economic Analysis. Hierarchical Decomposition in Economic Analysis, in Behavioral decision making: Handbook of behavioral economics. Oxford: Basil Blackwell, Expectation Calculation and Macroeconomic Dynamics, Amer. Calculation, Adaptation and Rational Expectations. Expectation Calculation, Hyperinflation and Currency Collapse, in The new macroeconomics: Imperfect markets and policy effectiveness.
Games, Econometrica, Jan. Recent developments in game theory. Aldershot, England: Edward Elgar, Speculative Dynamics, Rev. A theory of adaptive economic behavior. A behavioral theory of the firm. Oxford: Blackwell, , pp. Anomalies: Cooperation, J. Perspectives, Summer , 2 3 , pp. Recursive programming and production response.
0コメント