You’ve read one of your four complimentary articles for this month.
You can read four articles free per month. To have complete access to the thousands of philosophy articles on this site, please SUBSCRIBE!
Economics and Rationality
Brendan Larvor reviews The State We’re In by Will Hutton and explains why consumers aren’t crazy after all.
Will Hutton’s The State We’re In offers a many-layered argument. At one level, it is an essay on the recent political and economic history of Britain. It is also a call for a certain sort of political programme. I shall not concern myself with Hutton’s case at either of these levels. I shall focus exclusively on Hutton’s argument against classical economic theory, where that theory intersects with more general questions about the nature and scope of human rationality.
Hutton makes two kinds of argument about rationality and classical economics. The first type goes as follows: classical economics predicts that rational people will do X. In fact, they do not do X, and they have good reason not to do X. Therefore, classical economic theory fails empirically (since it does not describe what people actually do) and it fails normatively (since it does not describe what they rationally ought to do). The second type of argument goes like this: classical economics predicts that rational people will do Y. In fact, they do not do Y, and we have scientific proof that people are irrational in the relevant respects. Therefore, classical economic theory fails empirically (since it does not describe what people actually do), but its normative claim (that it describes what people ought to do if they wish to be rational) is bolstered.
Examples of these two sorts of argument follow shortly. 1 shall then claim that Hutton accepts the deliverances of empirical psychology too uncritically. The allegedly scientific proofs of human irrationality depend on crude notions of rationality which are shared by classical economics. If Hutton were to adopt a more critical stance towards formal rationality theories in general (including game theory and classical economics), he could turn type 2 arguments into type 1 arguments. This would refute the normative claims of classical economic theory by knocking away the conception of rationality on which it stands.
Arguments of the first type
The clearest examples of the first sort of argument occur in Hutton’s discussion of the labour market (pp.95-105). Classical economic theory predicts that workers will push up their wage rates as high as possible, and employers will pay at the lowest rates they can get away with. Labour, in this model, is a commodity for sale like any other.
Of course, all things being equal workers will always prefer a higher rate of pay and employers a lower one, but there is a whole world packed into that “all things being equal” clause. To begin with, work is not simply a burden. A workplace is a finely-graded social system which offers the worker a detailed identity (as a member of precisely this team or that unit) and a set of similarly closely-specified goals (to go on this training course, to take over that job). This in turn locates a person in society as a whole. It supplies one with an answer to the question strangers always ask, “What do you do?” Hutton cites empirical evidence (p. 100) to suggest that some workers are prepared to accept lower wages than they might otherwise obtain in order to enjoy the benefit of a position towards the top of a company hierarchy. It has always been obvious that classical economic theory has a hard time explaining why anyone would go into nursing. The examples Hutton cites are drawn from companies such as estate agencies and car dealerships, where one might have expected the market model to work. These people (like nurses) do not behave as the model says they should, and are therefore irrational according to the model. Hutton cites recent research to argue, plausibly, that the flaw lies in the model.
On the other hand, employers often pay over the odds. They do not always lay off their staff or impose wage cuts as readily as the classical model predicts, because it is bad for morale. They do not always hire new staff at a lower rate (as they might if unemployment is high) because a two-tier pay system would be resented. They try not to hire overqualified applicants because such people upset the delicate connections between skills and rates of pay already established within the firm. Of course, these results vary from sector to sector, and firms sometimes do behave as the classical model predicts. When they don’t, they are irrational by the lights of the theory. But these firms are not irrational and their failure to screw wages down as far as possible seems to be grounded in clear-eyed commercial judgment. To succeed financially, a company must be an effective social organisation. This means, for example, that the desire for fairness must be respected. If two people are doing the same work for different levels of pay, there had better be a good reason – and “labour was cheaper when you were hired” is not good enough.
The various studies which Hutton cites do not present a simple account of why it is that firms and individuals do not behave as the classical model predicts. Nevertheless, there is a connecting idea which is that the workplace is a complex network of human relationships. Firms are more successful, and work more satisfying, if these relationships are characterised by some level of trust and decency. It is for this reason that it is entirely rational for players in the labour market to give away more than classical economics predicts that they need to. Hence, the theory fails empirically and normatively.
Arguments of the second type
Hutton’s book contains three arguments for his claim that classical economics fails empirically because it fails to take human irrationality into account. The first of these concerns the ‘matching law’ (pp.158-160, 251). This ‘law’ states that “the attraction of a reward is inversely proportional to the delay in receiving it, so that most people will opt to be paid a lower amount today than a greater amount in the future – even though by waiting they can enrich themselves” (p.251). In other words, people value present rewards very much more highly than rewards in the future. The effect remains even if the subjects of the experiment are assured that the future benefits will definitely materialise. This law, says Hutton, is “as true of rats and pigeons as it is of human beings.” (p.158). I shall not pursue the question of exactly how one assures a rat or a pigeon that a promised future benefit really will appear if an immediate benefit is forgone. The claim is that people (and rats and pigeons) discount for uncertainty far more heavily than they should.
The relevance of this for Hutton is that this allegedly irrational time-discounting is exhibited on the stock markets. Classical economics fails to model these markets because the theory assumes that the dealers are rational, when experimental psychology has shown that they are not (in this respect at least). The trouble with Hutton’s argument here is that it requires there to be a rationally calculable discount rate for uncertainty. The alleged irrationality lies in the fact that an event with a probability of (say) 0.8 is valued by the markets as if its probability were only (say) 0.5.
But where could such figures come from? The claim that people are irrational in this respect depends on the mistaken idea that we can measure the likelihood of future events. It is only in the light of an objective measure of probabilities that our subjective judgments can be found wanting. But such objective probabilities are not available to us (except in artificial cases involving bags full of coloured balls). What is the objective probability that a long-established bank will collapse overnight? Is it the same now as it was before the Barings fiasco? Real people spending real money worry about events like the Barings or B.C.C.I. collapses, and they are rightly suspicious of expert assessments of risk. In other words, (whisper who dares) financiers do not believe in actuarial science! Nor should they. Financial traders live the problem of induction every day of their working lives and they know that the premises underlying statistical analyses of the future do not always hold.
What about the experimental case in which the future benefit is guaranteed by the psychologist? Surely it is irrational to prefer a smaller present benefit under these circumstances? Not necessarily. We all know that psychologists sometimes mislead their experimental subjects in the interests of science. It is quite common to conceal the real point of an experiment from its subjects (for example, in one experiment – to test the effect of confidence on teachers – an undistinguished sample of teachers were told that they had been selected for their excellence). For all the subjects of the ‘matching law’ experiments know, the point could be to test their reactions to broken promises. In short, the matching law shows that humans are programmed to discount very heavily for uncertainty. This makes sense, given that our fundamental psychology was presumably fixed at a time in the distant past when life was extremely uncertain. The fact that we continue to discount heavily for uncertainty even when we are assured that the large deferred benefit will materialise shows not that we are irrational, but that we are innately suspicious of such assurances. Life is incalculably uncertain for rats, pigeons and City dealers, and their propensity to live in the now is entirely reasonable.
Indeed, it is not even clear that the experimental psychology Hutton cites on this point will do the work he wants of it. Robert Frank (Hutton’s chief source on this topic) remarks that “the inconsistency implied by the matching law is not the result simply of the fact that we discount future rewards. There is, after all, nothing irrational about that. In an uncertain world, a bird in the hand often really is worth two in the bush.” (Passions within Reason p.80). His point is more subtle. Given a choice between $100 now and $120 in three days time, many people choose $100 now; on the other hand, given the choice between $100 in twenty-eight days and $120 in thirty-one days, most people choose to wait an extra three days for the $120. Then, if we assume that the function mapping the proximity of a reward to its attractiveness is continuous there must be a time when subjects would be indifferent between $100 and $120. Such indifference would be irrational, but it is also irrational to assume without argument that the mathematics of experimental psychology will only yield continuous functions. There are plenty of discontinuities in physical nature (such as phase transitions, for example). The difference between ‘now’ and ‘then’ is just the sort of place where one might expect to find a discontinuity in psychology.
Frank goes on to warn that, “This is not to say that material reward and delay feed mechanically into the matching law to produce a deterministic theory of human behaviour. On the contrary, people obviously have considerable control over the way they perceive the rewards from competing activities.” (Passions within Reason p.81). This remark leads us away from experimental psychology and back to historical and political questions about how the City of London views British industry – precisely the area where Hutton’s argument is strongest.
This analysis permits us to turn the argument from a type 2 into a type 1. If heavy discounting for uncertainty is rational after all, then the empirical failure of classical economics to describe the behaviour of stock markets is now compounded by a normative failure. The theory cannot be said to describe what stockbrokers ought to do if they wish to be rational.
At another point (p.230), Hutton cites the well-known phenomenon of intransitive preferences as further proof of human irrationality. One would expect someone who preferred Coca-Cola to lemonade and lemonade to fizzy water to prefer Coca-Cola to fizzy water – but people do not always behave as expected. Here too, Hutton accepts the psychologists’ conclusion uncritically. Consider an example (borrowed from Paul Anand). A fruit-bowl contains two out of the following: a large apple, a mediumsized orange, and a small apple. You are a hungry guest. If the bowl contains the large apple and the orange, you choose the apple (since it is bigger and you are hungry). If the bowl contains the orange and the small apple, you choose the orange (since it is bigger and you are hungry). If the bowl contains the two apples, transitivity requires that you choose the larger. However, etiquette requires that you choose the smaller apple (since it is greedy to pick the largest out of a selection of similar objects). Since most humans value politeness higher than transitivity, they make an apparently irrational choice. In fact the fault lies in the crudity of the psychology. More generally, transitivepreference experiments ignore the fact that each option constitutes part of the context of the alternatives: a large apple in the presence of an orange is not the same as a large apple in the presence of a small apple. So people who make intransitive choices are not necessarily irrational after all.
On the same page, Hutton mentions the ‘peak end rule’, that: “our memory of an event is conditioned by the peak of emotion we experienced during its course and how we felt when it ended.” (p.230). These two factors determine our recollection of an experience, whereas its duration has little effect on the memory. In Hutton’s example, a long, cumulatively painful operation may be remembered relatively favourably if a nice moment occurs during it and it ended pleasurably. A shorter operation with a painful peak and painful ending will be remembered less favourably. Thus, a more painful event may be remembered better than a less painful one. Therefore, humans are irrational in this respect.
This argument assumes that pain can be measured and plotted against time. The psychologist can then calculate the area under the graph and check to see if the subjective comparison of remembered pains ranks episodes in the same order as this ‘objective’ calculation. If the orderings differ, then the psychologist records that the subject is irrational in his memories. This result only follows, however, if we assume that the total amount of pain over time is all that matters. Anyone who has ever been in pain knows that nothing is further from the truth. Pain doesn’t just hurt, it debilitates. A long-lasting dull ache may register more pain-units on the psychologists’ meter, but it does not interfere with one’s day in the way that a short period of excruciating pain does. The subjective quality of pain also depends on whether there is an end to it in sight, and whether one is able to do things to take one’s mind off it. It may make a difference if a painful experience is the latest in a sequence of similar episodes. Long-term invalids tend to develop complex relationships with their diseases. Part of the distress of pain is to see the effect on close family and friends, who feel both sympathy for the sufferer and frustration at their inability to help. In this way pain overflows into the sufferer’s social circle. Some of these effects are lessened if one is a hospital patient (or laboratory subject). One is not required to perform many daily tasks, and medical staff are not normally helpless in the face of suffering. Nevertheless, it remains the case that pain in humans affects us not only as physical organisms but also as active and social beings. Simple measures of sheer intensity miss all these subtleties.
Moreover, the connection between subjectivity and time is a rich source of philosophical puzzlement. Why are we more concerned about pain in the future than identical pain in the past? According to game theory, this result is irrational: a pain represents the same disutility regardless of whether it has already happened. Game theory notwithstanding, everyone leaves the dentist’s chair saying (or thinking), “I’m glad that’s over.” Should we concur that absolutely everyone is irrational, or should we re-examine the underlying theory?
Hutton argues that classical economics fails partly because it assumes, falsely, that people are rational in certain respects. I think that this analysis grants too much normative force to classical economics. It leaves the market theorist free to argue that his theory shows how things would be if people behaved as rationally as they ought – he is not responsible for human stupidity. In other words, Hutton’s type 2 arguments refute classical economics as an empirical theory, only to reinstate it as an ideal against which we can all be measured and found wanting. Rather than leaving this escape-route, we should say instead that classical economics fails because practical reason is too subtle and sensitive an instrument to be captured in the clumsy formalisms of classical economists and experimental psychologists. The correct conclusion is not that “Rationality… is too difficult for human beings to cope with.” (p.230). It is that the kind of consistency demanded by formal rationality theories is but one goal among many. It is often rational to sacrifice this kind of consistency in favour of some other worthwhile goal such as smooth social life or effective action. The fruit example above works because good manners often yield greater dividends than good logic. On those occasions it would be irrational to behave in the way that classical economists and experimental psychologists think we should. In general, then, Hutton should try to turn type 2 arguments into type 1 arguments.
Of course, I am not suggesting that people are never irrational. I do think that we should be suspicious of theories which convict humanity as a whole of irrationality. And we should be aware of the false assumptions shared by classical economics and certain strands of experimental psychology. Philosophers have encountered (and on the whole rejected) this rationality-as-calculus approach in the form of utilitarian ethics. It is rare (though not unheard of) to find a philosopher who thinks that practical reason is a matter of cost-benefit analysis. Instead of playing the psychologists against the economists, we Huttonistas should work to expose their common errors.
© Dr B. Larvor 1996
Frank, Robert Passions within Reason Norton, 1988
Hutton, Will The State We’re In London: Jonathan Cape, 1995
Brendan Larvor is a philosophy lecturer at Liverpool University