The Rationality Hypothesis: Help or Hindrance?

Rothschild, Kurt W. Article The Rationality Hypothesis: Help or Hindrance? Intervention. Zeitschrift fuer Ökonomie / Journal of Economics Provided in Cooperation with: Edward Elgar Publishing Suggested Citation: Rothschild, Kurt W. (2005) : The Rationality Hypothesis: Help or Hindrance?, Intervention. Zeitschrift fuer Ökonomie / Journal of Economics, ISSN 2195-3376, Metropolis-Verlag, Marburg, Vol. 02, Iss. 2, pp. 56-65, https://doi.org/10.4337/ejeep.2005.02.07 This Version is available at: https://hdl.handle.net/10419/277052 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ 56 Forum Th e Rationality Hypothesis: Help or Hindrance? Kurt W. Rothschild* Th e assumption of rational decision-making, the maximising or optimising homo oecono micus, belongs to the fundamental elements of the neoclassical mainstream. It is also an element that has been hotly attacked and defended for a long time without a convincing victory for one side or the other. Th e following considerations – though none of them is completely new – should help to show why such a one-sided victory is neither likely nor desirable. Before going into a more detailed discussion of the pros and cons for using the Ra tio nality Hypothesis (RH), I want to set aside some widely held views which are based on mis - understandings of the role and contents of the RH. One widespread objection to the RH held by some economists, but mainly by psychologists, sociologists and the public in general is the obvious lack of realism of the hypothesis in view of the numerous and diversifi ed factors which infl uence human decisions. Th at a lack of realism exists is beyond doubt. But in itself this is not an acceptable objection. Th e RH is a theoretical (methodological) concept and as such necessarily an abstraction from the fullness of the real world. Only through abstraction can we obtain better insights into a complex reality. Quite another problem, and an important one, concerns the question whether the abstraction is capable to cover those aspects that are decisive for the problems under discussion. Th is ques tion will be in the foreground in later considerations. Another misunderstanding concerns the motivational structure of the homo oeconomicus. Th e usual defi nitions speak of maximisation or optimisation of individual utility or interests. Th is is interpreted by some people as a characterisation of the homo oeconomicus as a perfect egotist. And against this picture of human nature objections are raised for realistic and moral reasons. But this is a misunderstanding, though it must be admitted that unfortunate formulations of some economists have often fostered such misunderstandings. Th e important point is of course that the rationality of the homo oeconomicus refers to his calculation behaviour, i. e. the careful weighing of alternatives so as to reach a maximum of preference satisfaction, but it says nothing about the preferences themselves. Th ese can include altruism and other non-selfi sh motives the satisfaction of which will lead to altruistic actions. Th is proper argument against moral attacks on the homo oeconomicus involves, however, occasionally a dangerous justifi cation for the homo oeconomicus and the RH, which * Österreichisches Institut für Wirtschaftsforschung (WIFO), Wien.  In his methodological proposals Milton Friedman () rejects the criterion of »realism« altogether. For him the test of a »good« assumption consists in its capacity to deliver good predictions. Whether the RH is generally successful in this respect is highly questionable but need not be discussed here. A more general view is that a higher degree of realism (other things being equal) is desirable because it off ers greater reliance on the role of the assumptions and strengthens the explanatory task of a theory. Kurt W. Rothschild: Th e Rationality Hypothesis 57 helps to make it immune against any criticism. Since we cannot specify the preferences of an individual exactly before an action is taken, we cannot decide after an action whether it has led to the maximum of preference satisfaction. All that »revealed preferences« show is that obviously some preferences are present, but we get no information about the extent to which they are satisfi ed. Of course when one starts with the assumption that the RH is correct and the homo oeconomicus maximises, then the actions reveal his maximum satis faction. But when this approach is accepted the RH becomes a tautology. A person acts in a certain way because of his preferences and his preferences are what they are because of his actions. Th is would always be true. Th is pattern of argument can be found in many attempts to preserve the RH in face of doubts and contradictions. Th us if one fi nds, for instance, that people use rules of thumb in a dynamic environment that has room for »better« adjustments, the RH can still be maintained by stipulating that the »costs« of readjustments (information, time, etc.) are so important that the habitual action is the optimal solution. After these general remarks we can now move to some more detailed questions. When dealing with criticisms of the RH we should distinguish two diff erent lines, though these are not always strictly separable: the RH is partly opposed from »within«, i. e. without ques - tioning rational behaviour as such, and partly from »without«, by pointing to other sources of behaviour, such as psychological and sociological infl uences. Remaining within the realm of rational behaviour we must fi rst stress that the opposition to the RH does not amount to a proposal for an Irrationality Hypothesis. Rather it is a criticism of the narrow defi nition of rationality in neoclassical theory. In this »school« rationality is defi ned as choosing a unique optimum action that delivers the maximum of utility. But diverging from this strong and demanding defi nition does not mean that people act irrationally. Rather we can say that they may act »reasonably« which does not necessarily mean optimally, though it is certainly not irrational. A simple example can illus trate the diff erence. A person wants to cross a busy street, which can only be crossed safely at the regulated crossings. If she or he is somewhere near the middle between two crossings she will act reasonably – assuming she is keen to stay alive – if she goes to one of the two crossings. Th ere is nothing irrational about this. But to act rationally in the economist’s sense (neoclassic style) demands that she fi rst calculates which crossing is nearer and then chooses this one (assuming that time and/or walking are regarded as unwanted costs). Th e choice of the other crossing would not be optimal and therefore »not rational«. Th ough it is – from a realistic point of view – fairly obvious that reasonable behaviour (wider rationality notion) is the general rule and optimising behaviour (narrow rationality notion) only a special case in special circumstances it can still be easily seen why neoclassical equilibrium theorists tend to stick to their RH. As the simple street-crossing example shows, the main diff erence between the »reasonable« approach and the optimising RH  In future I will restrict this to one of the two genders.  Irrationality would for instance mean that the person, in spite of a strong desire to survive, crosses the street immediately because of a strong impulsive urge. 58 Forum is that the fi rst leaves open which of the two crossings is fi nally chosen, while the RH approach gives an exact answer. To arrive at the same »exact« result in the fi rst case requires the introduction of additional facts regarding the person (e. g. a bias to turn to the left) or the circumstances (e. g. unequal crowding on the two sides). Th e beautiful analysis delivering unique equilibrium solutions would be disturbed through individual and situational factors. It is obvious that it is not easy for the big cohorts of economic theorists who have been brought up with the pleasant qualities of the homo oeconomicus and the precise results connected with the RH to give up lightly their fi rm basis for the uncertain variety of outcomes (multiple equilibria and disequilibria) connected with alternative routes and theories. Th e old dream which inspired the th century social scientists in general and many economists in particular to develop theories which can achieve the regularities and precision of mechanical physics is still alive in spite of many modifi cations to the original static equilibrium models (Mirowski ). Th is fear of loosing the fi rm basis of the neoclassical paradigm is sometimes very frankly admitted. So for instance by Hicks in his great opus on general equilibrium (»Value and Capital«) when he defended the maintenance of the perfect competition framework (with its RH). »It has to be recognized«, he wrote, »that a general abandonment of the assumption of perfect competition […] must have very destructive consequences for economic theory […] [T]he basis on which economic laws can be constructed is […] shorn away.« (Hicks :  f.) In a similar vein Melvin Reder defends the RH by complaining that »bounded rationality can be made compatible with almost any story of the process of setting prices.« (Reder : ) Th ese fears explain the strong hold of the RH in mainstream theorising and the steady attempts to include »anomalies« and »paradoxes« in such a way that the RH basis is not endangered. Th e question is whether these tactics, which have resulted in highly sophisticated constructions with some interesting insights, are suffi cient in order to deal successfully with the doubts and objections to the RH coming both from the internal rationality discussion and from outside considerations. Th is question will now be discussed. We start with the internal discussion. Th ough there are several earlier sources dealing with this problem the discussion was set off on a big scale by Herbert Simon’s attack on the universal use of the RH and his introduction of the concept of »Bounded Rationality«. Ever since the publication of his important work on »Models of Man« (Simon ), there has been a fl ood of publications touching this subject. Th ough a great part of the contributions came and comes from sociologists, psychologists, and epistemological writings, economists have also taken part in the discussion. But though the considerations concern-  But as I mentioned before and as we shall see later, the internal and external objections cannot always be clearly separated.  An extensive survey of this literature up to  is given by Conlisk (). Since then further contributions have appeared. Kurt W. Rothschild: Th e Rationality Hypothesis 59 ing Bounded Rationality have been generally appreciated by economists as interesting observations (justifying the Nobel Prize for Herbert Simon), there is an astonishingly small echo of this appreciation in the »normal« theoretical literature or in the history of economic science. In order to see whether this neglect is justifi ed we must look a bit more closely into the origins of the Bounded Rationality discussions and their consequences. To deal fully with the numerous sources and forms of Bounded Rationality that have emerged in the ongoing discussion would surpass the limits of this paper and is not essential for the question asked in the title of this article. But some distinctions are necessary. An important distinction from a research point of view can be made between fundamental Bounded Rationality objections to the RH and the study of behavioural factors leading to Bounded Rationality actions instead of RH actions. Th e fundamental objections rest on the fact that the complexity of a constantly changing environment and in particular the unavoidable uncertainty of the future make it impossible to carry out the exact calculations of the homo oeconomicus and to arrive at clear and unique decisions. People will necessarily have to decide on the basis of incomplete information and this leads to the possibility of a range of »rational« decisions. In addition to these fundamental conditions due to the limits of the human brain and the existence of genuine uncertainty of the future (i. e. not reducible to risk) there exists a multitude of psychological and behavioural infl uences that lead to divergences even in cases where a RH decision could be achieved. Under this heading we can mention such obvious human characteristics as fatigue or resistance towards extended eff ort leading to a renouncement of an optimum solution once a »satisfi cing« solution has been reached, or the rules of thumb for similar, but varying cases in order to simplify the decision process. Particularly interesting and intriguing are the numerous cognitive and psychological factors that can infl uence the decision process, which are extensively treated by psychologists and sociologists particularly since the path-breaking work of Kahnemann and Tversky. Th e – as far as the RH is concerned – »disturbing« result of these theoretical and experimental studies is the disclosure of many decision-infl uencing factors that contradict the basic requirements of rational choice, such as consistency, transitivity, absence of regret etc. Th ey are too numerous to be treated here in detail. To mention just a few: there are several factors that bias decisions such as framing eff ects (decisions depending on the form in which a problem is presented), cognitive dissonance leading to biased perceptions of a problem, social infl uences and moral considerations leading to motivational confl icts, passions, emotions, impulsive actions etc. All these are infl uences on decision-making in general, but they are also relevant for the economic sphere. Th e importance of such infl uences explains why psychologists are often more in demand for marketing strategies than economists. In all these cases – the fundamental and behavioural types – we are faced with empirical phenomena that are not conforming to the high standards of the RH. Does this  »Herbert Simon’s suggestions for a ›turn‹ in economic theorizing have been widely discussed but have not really changed the bulk of economic theorizing.« (Lindenberg : ) 60 Forum matter? If we started with and were contented with purely descriptive studies of various situations it would certainly not matter. We would try to picture as exactly as possible how decisions are taken in each case without any bias for one explanation or the other. But when we deal with theoretical abstractions and generalizations that should »explain« and predict a whole class of decisions the choice of basic assumptions becomes important. Th e theory should be as »effi cient« as possible in dealing with the subject, both in its tractability and its explanatory and predictive successes. Full success will – particularly in the social sciences – not be possible and there can be a trade-off between tractability and explanatory/predictive achievement. From this perspective it can easily be seen why neoclassical equilibrium economists, i. e. the mainstream as far as general microand macroeconomic theory is concerned, adhere so strongly to the RH and try to defend it at all costs. After all they have a long-established framework with rich opportunities for deriving exact and unique (equilibrium) results for a wide range of problems that often yield »realistic« answers and predictions which can compete successfully with the results of other theoretical frameworks. It is there - fore no wonder (but nevertheless a problem) that the mainstream has a long tradition to deal with »anomalies« and contradictions between its conclusions and real developments by try ing to fi nd explanations and theoretical adjustments which leave its axiomatic foundations intact. As far as the RH is concerned, the main defensive activities had to be directed against the obvious divergence between fairly easily observable actual decision-processes and the exact calculation needed by the optimising homo oeconomicus. An early and I think to some extent acceptable argument against the attack on the homo oeconomicus on these grounds was the remark that the RH was not meant to apply to decision-making in general where cognitive and emotional factors may very well play a big role, but only to actions in the »market place« or in the economic sphere in general where the »non-rational« factors were less important. While this argument removes some objections to the use of the RH, it still leaves a great deal of RH problems within the economic sphere. Even the central showpiece of traditional theory, price theory, is aff ected by them. In the wake of the imperfect competition revolution of the thirties of last century the theory of administered prices (Means ) and the famous Oxford study of Hall and Hitch about price setting and price diff erences (Hall/Hitch ), which created quite a stir, indicated that the nice results of equilibrium price theory rested on insecure foundations. Similar exam ples of »nonrational« decisions could then be detected in many other fi elds and were advanced by critics of the RH, not least by pointing out the impossibility or extreme diffi culty of making  Th is basic distinction is of course denied by the »economic imperialists« like Gary Becker or Edward Lazear who recommend the RH for analysing decision-making in wide fi elds of human relations (see Lazear () and Rothschild ()).  Th at these empirical studies created such surprised attention only shows how deeply ingrained the theoretical belief in the general applicability of RH-based decisions was with its stress on competitive prices and the theory of one price. Kurt W. Rothschild: Th e Rationality Hypothesis 61 the calculations at the margin (e. g. of revenue and costs) that characterise the desired maximum solution. While these empirical and principal objections to the world of equilibrium and optimum positions were not (and could not be) denied by the mainstream, it has always tried to defend its basic structure. One important answer to the attacks that is certainly acceptable in several cases if and only if this answer is constantly kept in mind when dealing with practical policy problems rests on the »as if«-argument. Th e fact that strictly maximising behaviour is not the normal case in practice is not denied. But one could assume that people aim at optimal solutions and approximate these gradually – supported through learning – in their incomplete strategies. Observed decisions conform therefore to Bounded Rationality but in the end they will not be so very diff erent from the exact maximum of the RH assumption. To maintain this theoretically convenient assumption could therefore be justifi ed. Alternative defence lines come from the analytically and practically more interesting transaction cost and evolutionary perspectives. Transaction cost theory recognises the cognitive and eff ort-caused limitations of acquiring the full information needed for a fully rational decision. Th ese costs will be taken into account and the decision process will be stopped before the maximum of the »pure« RH decision is reached. Th is can be regarded as a Bounded Rationality situation though – when the transaction costs are included in the preference basis – it can also be regarded as an extension of a simple RH analysis. As far as some types of evolutionary theories are concerned with a strong stress on »the survival of the fi ttest« the RH assumption can be defended for the long run on the assumption that only the profi t-maximising fi rms, earning competitive profi ts, will be able to survive. For all these arguments it can be said that they justify the use of the RH with all its advantages in some fi elds generally, in some fi elds under certain circumstances, and in some fi elds not at all. For good insights and results it would be necessary to be able to distin guish these cases in order to see whether these »as if«-assumptions do apply. Th is is not an easy matter but it has to be done if the RH assumption is to be properly used. Th e current problem is not so much the diffi culty of making such diff erentiations as the bias of the mainstream to adhere to the RH whenever there is the slightest chance to defend it. Th is bias is dangerous because it can be shown that even in obvious »as if«-cases small divergences, e. g. in learning or in the evolutionary process, can lead to disequilibrium or multiple equilibria depending on the initial conditions (»path dependence«). A more important and less tractable problem for the RH than in the »as if«-cases pres - ents itself in two special directions: interdependencies and future uncertainty. Let us take interdependence fi rst. Th e RH refers to the calculating (maximising) individual with given preferences who is confronted with a given and transparent environment (e. g. the competitive market). Th is environment defi nes the constraints under which she can optimise. However this structure breaks down when the action of the individual infl uences the environment and the constraints under which she can act. Th ese changes of the environment  See Akerlof/Yellen () and the literature cited there. 62 Forum must also be taken into account and this creates new problems. Th e typical example for this case is oligopoly. To maximise his profi ts the oligopolist must not only take note of market demand and production costs, he must also consider how the other oligopolists will react to his operations. In other words, »strategic« action is required. Th e beauty of the automatics of perfect competition equilibrium is lost. Th e great advance of economic theory in this respect was of course the introduction of game theory by Neumann and Morgenstern. Th ey delivered a theory that makes interdependence a central element of the analysis. But here too a preference for RH equilibria can be detected. Th e comparatively simple basic example of game theory conforms indeed to the RH assumption. Th e »players« will choose actions that deliver a combination of strategies that is the best of all the available ones. But the assumptions required for such an optimal equilibrium are formidable. Th e participants must know all the strategies available to all participants, further the consequences of all possible combinations, and they must also be able to rely on corresponding »rational« decisions of all players. Once these assumptions are dropped, the individual RH cannot give clear optimal answers and does not necessarily lead to optimal solutions. It all depends on the various assumptions about reactions, about the willingness to take risks, about learning opportunities etc. Th e rapid expansion and the variety of dealing with these problems show how irrelevant or »paradoxical« the RH can sometimes become. In any case the »as if«-assumption does not help in most of these situations and the existence of a variety of relevant assumptions and outcomes has to be acknowledged. Similar complications arise when future events are involved. A fundamentally uncer - tain future (i. e. where the risks of alternative developments cannot be objectively ascertained) cannot be subjected to a defi nite and unique rational calculus. Objective circumstances (»situational logic«) and subjective factors of viewing the future and the role of time will combine to infl uence the decision process. Th is does not mean that these processes cannot be dealt with in theoretical analyses, but they require diff erentiated approaches with multiple results.  Th us we get a sub-optimal result in the famous »prisoners’ dilemma« because the involved persons (equipped with risk aversion) do not know what strategy the opposite party will adopt. Th e Cournot solution for the duopoly case leads ultimately to a sub-optimal equilibrium because both duopolists wrongly assume that the other party will always stick to the given output. Bertrand showed that the wrong assumption that the other party sticks to its given price will lead to constant fl uctuations. Of course when we introduce learning processes the picture can change in the course of time.  Th us in some examples of repeated games with a fi xed end-date where cooperative strategies are profi table but breaking cooperation in the fi nal period becomes advantageous, »rational« recognition of this situation leads to the result that cooperation is abandoned by one party at the very beginning no matter how long the cooperative period could have lasted (»backward induction«). Th is »rational« solution looks paradoxical from a realistic point of view.  Th e quick acceptance and appreciation of the »rational expectations« hypothesis of Robert Lucas is a good example of the desire to maintain the rationality axiom by playing down the uncertainty problem. Th is saves not only the theory but has considerable consequences for policy debates by stressing policy ineff ectiveness and the demand for a slim state. Kurt W. Rothschild: Th e Rationality Hypothesis 63 After all these considerations we can now turn to the question raised in the title of this paper: Is the RH help or hindrance? In my opinion the answer cannot be a straightforward one. As we have seen the RH and the equilibrium theories built on it have enormous analytical advantages and the attraction of delivering unique answers to many problems. One can therefore not give it up light-heartedly even when its realism or its results are often disappointing. On the other hand decision processes may so much diverge from the RH that attempts to stick to it at all costs using all sorts of intellectual acrobatics become counterproductive. It prevents proper explanations and fosters doubts in the signifi cance of economic theory. What is needed is a diff erentiated treatment of economic problems and the realisation that there cannot be a single axiomatic theory which can provide access to all economic problems in a dynamic and complex economic and social environment. Rather than aiming at a fi nal establishment of such a universal theory or even a dominant theory one should recognise that diff erent theoretical approaches are required to meet different classes of economic problems. Th is applies to all basic assumptions and also – and not least – to the RH. However the main problem in the theoretical world of today is probably not that that this need for alternative theoretical structures is completely neglected. Th eoretical disputes, the spread of special approaches and of specialised journals, behavioural and experimental economics etc. off er a broad perspective of diff erent decision motives. Th e trouble is rather that the traditional equilibrium models relying on strict rational decision-making play such a dominant role in the profession. Th at can be seen in the contents of the core journals and in most university curricula. »Heterodox« ideas, let alone »heterodox« theories, are severely underrepresented in these infl uential fi elds. Th is domination is not only due to the conservative force of tradition, but is intensifi ed by the form of neoclassical theory with its »exact« structure which cannot be maintained once the volatility of situations and decision-making is allowed to enter. Alternative theories necessarily sacrifi ce some exactness and defi niteness in order to get nearer to the complexities of real situations with their  It should be clear that this question only arises – as is the case in most of the theoretical literature – if the RH is understood as an abstraction of actual human behaviour. If the RH is regarded as a normative recommendation for optimal behaviour the question of hindrance does not arise. Such a view is, for instance, taken by John Harsanyi with regard to game theory when he states that »our theory is a normative (prescriptive) theory rather than a positive (descriptive) theory. At least formally and explicitly it deals with the question of how each player should act in order to promote his interests most eff ectively in the game and not with the question of how he (or a person like him) will actually act in a game of this particular type.« (Harsanyi : ) Even in this case a certain risk-attitude must be assumed (normally risk-aversion in game theory).  But even experimental economics is not free of the RH bias. »Too many experimentalists are in search for the confi rmation of orthodox theory and go to great length in explaining away deviations which cannot be overlooked. […] Th e necessity for a radical reconstruction of microeconomic theory becomes more and more visible.« (Selten : )  »Arguments for optimisations-only modelling have held powerful sway, shaping the research, the teaching, and the everyday conversations of economists.« (Conlisk : )