5. How to Make ‘Bad’ Choices

There is ample evidence from both common observation (including introspection) and careful professional psychological studies that many – and probably most, if not all – people frequently make choices that are not really in their best interests. (For present purposes I shall define these as ‘bad’ choices although it is arguable that in many cases such choices can be defended.) Behavioural psychologists have identified various classes of reasons for such ‘self-defeating’ behaviour. One is that people simply lack full self-knowledge of their own preferences. Another is that their strategies are based on an imperfect recollection, or interpretation, of past experiences in certain situations. Also, in particular, emotional stress plays a large part in such behaviour. Under emotional distress, people shift towards favouring high-risk, high-payoff options, even if these are objectively poor choices.

Another common cause of self-defeating action is a concern for social considerations. People may pay too much attention to what other people think. Or they may get very upset in response to a blow to their pride, ‘…and the rush to prove something great about themselves overrides their normal and rational way of dealing with life’.¹ Also a variety of social considerations tend to lock people into consumption habits and constrain their freedom of choice, such as class, family, culture, ideology, national character and so on.²

Thanks to the work of many behavioural psychologists such as Daniel Kahneman and colleagues, various other explanations of such behaviour have been proposed. Kahneman distinguishes between two modes of thinking, namely System 1 and System 2.³ The latter comprises a careful reflection and analysis of the available options. The former relies much more on guesswork, intuition and previous experience. It saves time and mental energy. People need some kind of energy to weigh up all the costs and benefits of any choice and many people do not have it, or have used it up, at some crucial point.⁴ But it means that one is prone to make several mistakes which, as Kahneman and his colleagues have shown, in many common types of System 1 thinking, have certain features in common.

One of these is a common tendency to be optimistic about one’s own capabilities. For example, surveys have shown that, in the USA, about 80% of motorists think that they are better drivers than the average. Another common type of mistake is what is known as the ‘anchoring’ effect. This effect can be seen in numerous transactions, such as in influencing how much one is prepared to bid when negotiating to buy a house or a car, where it is rare to put in a bid very far removed from some initial asking price.

The ‘endowment’ effect is a further example of what seem to be inconsistent choices. This is where people value more highly some object that they happen to own – quite apart from sentimental attachment and so on – than would be justified by some objective valuation based on how much they value other objects that they do not own but that have equal monetary value.

An example of what seem to be inconsistent choices among highly educated people is recent analysis of the choices made by a large sample (386) of employees of the highly respected Boston University. They were all asked their preferences as regards the way they would like their consumption levels to vary over their lives, given their existing income prospects. For example, those who did not care much about a likely severe fall in consumption levels in old age would save less than those who did. So their preferred profiles of consumption over their lives had implications for what their savings and insurance behaviour ought to be. But when their actual savings and insurance behaviour were examined it was shown that most of them were far from following the required strategy. There was, in fact, very little correlation between their preferred lifetime consumption profiles and those that would result from their actual savings and insurance behaviour. And the strategies were just as inconsistent for high-income professors who had significant financial knowledge as for low-income staff without such knowledge.⁵

However, it should not be thought that reliance on System 2 rather than System 1 for purposes of making choices always produces more welfare-enhancing outcomes. Indeed, some studies suggest that even if people are reflective and suffer from none of the mentioned impediments to Kahneman’s System 2, they still sometimes make choices that aren’t in their best interests. For example, some students who were given the time to follow System 2 and reflect on and assess all the attributes of university courses from which they could choose subsequently made less optimal course choices than the students who were not asked to reflect much on the choice. Similarly, other students who were allowed to reflect carefully on their choices were more influenced by irrelevant factors – for example, by the way in which the choice scenarios were framed rather than by the rationality of either option – than were the other test participants. So we have to face the fact that even when we have enough mental resources and motivation to engage in thorough reflection on our choices, we can still make choices that are not in our best interests.

This is particularly the case in choices involving the future. For it is generally accepted that people suffer from what Pigou called ‘defective telescopic faculty’ when it comes to taking account of future costs and benefits. This is particularly important since many choices require forecasting future costs and benefits. Sometimes these involve short-term pleasures but long-term pains. In such cases if people discount future costs too heavily they will fail to maximise their welfare over the relevant time period. Also, in most cases the present benefits are certain and the future costs are uncertain, which makes it even more tempting to discount them very heavily. There are many ways in which failure to take due account of longer-term consequences leads to decisions that reduce people’s longer term welfare. Eating too many fattening foods, smoking, drugs, unprotected sex, excessive drinking, are all examples of this failure to make choices that maximise one’s welfare over time.

One real-world implication of an apparent inconsistency in choice is the experience of some companies’ policies concerning their employees’ contribution to pension schemes. Instead of leaving it up to employees to make the positive decision to contribute to their pensions, many companies in the USA have switched to making the contributions automatic while giving the employees the right to opt out of the scheme. Thus the scheme was not paternalistic in the sense of imposing choices on the employees. Anyway, the result in these cases was a big increase in employees’ participation in the schemes, and only a few opted out. In other words, when it was a question of opting in, few took up the option, which seems to indicate a preference for staying out. But when it was a case of opting out, few took up the option, which indicated a preference for staying in. Clearly there is some inconsistency here with respect to which option employees thought was in their best interests (Sunstein, 2005, ch. 8). It is true that the different income effects of the alternative transactions could explain very small disparities but the large disparities observed go well beyond that.

Of course, even if people are reflective and rational and suffer from none of the impediments to Sunstein’s System 2 thinking mentioned earlier, welfare maximising choice may be very difficult simply because the options will often seem incommensurate. At the beginning of this chapter we considered the conflict between values arising in a student’s choice of career. In that example there was a conflict between the values of the student and the values of his parents. But on either side there was no doubt some conflict of values. The student in question probably took account of the value that a more lucrative occupation would permit, such as more appealing cultural activities, including foreign travel, or going to the opera, not to mention the ability to eat high-class food washed down with Chateau Mouton Rothschild claret. If, in the end, he chose a less lucrative career on account of its inherent interest and so on, this could have been the outcome of a very difficult choice between conflicting values. And conflicts of values are ubiquitous at the level both of the individual and society as a whole.

The general problem of incommensurability of values, which is one of the main objections to classical Utilitarianism, is discussed in Chapter 10. Some of the resulting problems in economic policy are discussed in Chapter 6. But there is one particular conflict of values that arises directly from the previous discussion of the ways in which people’s ignorance of the effects of some of their choices may do them harm. For this raises the question of how far one should force people to make choices that are believed to be in their best interests. In other words how far should society trade off the value of ‘consumer sovereignty’, or ‘personal autonomy’ against paternalism and its accompanying restriction on individual liberty.

Even when people are not irrational in the manner described earlier they will still very often make choices that do not best promote their welfare on account of lack of information. Lack of information is obviously an all-pervasive feature of most decisions that people take, ranging from what career path they embark on to which toothpaste best protects their teeth. This is particularly the case when the choices involve – as is often the case – judgements about risk and probabilities. For example, it is well known that people are far more worried about the probability of being killed in a railway or aircraft accident than in a road accident, though the statistical evidence suggests that the latter probability is much greater.

However, information gaps do not necessarily indicate a failure to behave rationally. There is what is known as ‘rational ignorance’. Consumers may sometimes lack information because they have rationally decided that the costs to them – in terms of time or money – of getting the extra information is not worth it in terms of the likely pay-off. Consider the following example that must be replicated in one form or another every day by thousands of tourists all over the world. Outside the Duomo in Florence there are kiosks that sell, among other things, postcards – photographs of the Duomo, or of Michelangelo’s statue of David after he had slain Goliath, or of some famous Raphael or Botticelli painting and so on. About a 10-minute walk north of the Duomo, in the direction of the Piazza Annunziata, there are a couple of bookshops that sell exactly the same postcards as those outside the Duomo, but at a much lower price. How is this possible? Quite simply, coach-loads of tourists are deposited outside the Duomo for a short time; they take their photographs of the façade and the adjacent Giotto tower and perhaps the Ghiberti low-reliefs around the Baptistery outside the Duomo, and then may spend a few minutes buying postcards from the closest kiosks. It would be uneconomical for them to waste time walking all around the area in order to check whether they could save a few cents by buying the same cards elsewhere. It is this allowance by most people for the costs of acquiring information that is known as ‘rational ignorance’. By contrast, most people go to much more trouble to obtain information when making big investments, such as buying a car, or a house, where it is worth doing so.

And in some cases the relevant information is about the future, which is never known to us completely. Lack of information and inability to predict future tastes and preferences is amply demonstrated by the high divorce rates. Given high divorce rates and the assumption that, in most cases, divorce is a last-resort option, marriage is clearly a gamble. Indeed, I am surprised that it is still legal. Instead – unlike gambling for money – it is actively encouraged, unless one is married to more than one person at a time. At least when one gambles on a horse to win a race it is possible to study the horse’s past form, including its record over similar distances and under similar circumstances. Such information is not usually available for marriage.

Also, while it may often be perfectly rational to make choices without obtaining all the relevant information, information is not a commodity like any other commodity. For one cannot know what is the optimum amount of investment one should make to obtain information until one already has the information. So in many cases lack of information probably does cut the link between preferences and welfare, as when people are not aware that they need the information or that it is available – for example, taking a highly paid job unaware that it was likely to be only a short-term job and that it will not prepare them for subsequent employment. Not knowing what one needs to know may be particularly important in certain situations, such as the trade in toxic waste (discussed in more detail in Chapter 8).

Furthermore, much information has the character of a public good – that is, that there is no extra marginal cost in allowing an extra person to have access to a piece of information once that information is available to other people. It is rather like a radio or TV broadcast, where once the programme is on the air, so that some people can enjoy it, there is no further cost in allowing other people to share in it. It is well known that in a free market the supply of information in public goods is unlikely to be optimal. Hence, in the case of many forms of information, there can be no presumption that people enjoy a socially optimal amount of it.

Given the general scepticism about the crude ‘preference-satisfaction theory of good’, how can individuals’ preference satisfaction be a good starting point for judgements about social welfare? Most economists recognise that people’s ‘manifest’ preferences (to use Harsanyi’s term), namely the ones that determine market outcomes, including prices, are often bad guides to their welfare, for the various reasons enumerated earlier. Consequently, many economists and philosophers believe that one should often overrule ‘revealed preferences’ and force or push consumers into expenditure choices that the authorities believe would better promote their welfare. In other words, consumer sovereignty would often give way to paternalism.

There are, of course, innumerable actual examples of not relying entirely on people’s preferences, including what are known as ‘merit goods’ (or their counterpart ‘demerit goods’). The most important example is basic education. Education benefits partly the community at large, of course, and so has an element of externality in it that would justify a subsidy even if it were a purely private good. But some people might still not take advantage of the opportunity to educate their children more cheaply. They might prefer to be able to spend more on drink, or a better car, or a foreign holiday. Thus education is usually made compulsory.

There are many other examples of a paternalist attitude, as when, for example, regulations are introduced that prevent people doing harm to themselves on account of the sort of psychological forces outlined previously. For example, there are lots of instances where we all accept that public restriction on freedom of choice is justified, such as prescription medicines, Health and Safety Regulations, as well as some instances where it is more debatable – drugs, smoking in public places, fatty foods and so on.⁶ And because people are known to underestimate the probability of their being involved in a road accident, many people would not wear seat belts unless they were compulsory (and did not do so for many years before they were made compulsory). So Parliament decided that people would have to be made to wear them.

In the defence of such measures it can be argued that one is not really challenging the validity of people’s ends, but only helping them to achieve them. In other words, the objection to such interference with people’s choices reflects a failure to distinguish between ends and means – that is, society should not try to interfere with people’s ends, but should be allowed to correct their errors about the means to achieve them. For example, one can assume that most people have top-level values to go on living. But suppose we have good reason to believe that if they postpone giving up smoking, or dieting and so on, they will die earlier – that is, we believe that they are failing to promote their ends. Thus, for example, seat belts in cars, or helmets for motorcyclists, would support their objective of long life.

However, there is a long-established tradition in support of the view that people are better judges of their own interest than are some outside authorities. For example, in his famous book, On Liberty, first published in 1859, John Stuart Mill devoted a whole chapter to the question of the extent to which the authorities should overrule the preferences of ordinary people, in the course of which he wrote that

…the only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or mental, is not a sufficient warrant. He cannot rightfully be compelled to do or forbear because it will be better for him to do so, because it will make him happier, because, in the option of others, to do so would be wise or even right.

And this view has been echoed by other philosophers throughout the ages, as, for example, Robert Nozick, who wrote in 1974 that ‘Two noteworthy implications [of his, i.e. Nozick’s, view] are that the state may not use its coercive apparatus for the purpose of getting some citizens to aid others, or in order to prohibit activities to people for their own good or protection’.⁷

Of course, Mill – and even Nozick – could not have been aware of all the modern evidence for believing that people are not best judges of their own interests. So it is arguable that they were quite wrong about the basic assumption that people are the best judges of their interests. On the other hand, as Robert Sugden has cogently argued, there is not much evidence that authorities are any better.⁸ It is highly plausible that in some cases ‘nanny knows best’, but even in such cases does this really justify overriding consumers’ preferences in most other cases at the cost of a probable mistaken belief of what is in their interests? It would be wrong to assume that the alternative to reliance on fallible consumers is intervention by some public authority that is all-seeing, fully informed, efficient, honest, rational and impartial. Nowhere is the ‘agent-principal’ distinction more important than in most public authorities, where the principals are the public and the agents are the officials who have their own agendas.

Furthermore, freedom to choose, even if it may mean making mistakes, may be an independent top-level value. People may attach ‘end value’ to freedom to choose even if they know they may not make the wisest choices. Hence, the objective of maximising people’s welfare by preventing their free choice in certain situations may be subordinated to the objective of respect for their rights.

In any case, one may be reluctant to brush aside people’s revealed preferences on the grounds that this would be the first step down a slippery slope via paternalism and authoritarianism to tyranny. For one has to be aware of the danger of excessive paternalism. Many tyrannies have been based on the view that people do not know what is best for them or that, even if they do, they ought to be motivated by a sense of duty to do what is best for the society in which they live. A cavalier rejection of democratic respect for people’s preferences can lead to very undesirable outcomes, not to mention an unjustified infringement of people’s liberties.

On the other hand, it can be argued that it is morally admirable to respect and promote the ‘projects’ (life aims) of other people because they have value. So a certain amount of paternalism is justified in order to help them pursue their objective, even though we may think that they are mistaken in their beliefs about how to promote their objectives. In other words a certain amount of paternalism could be defended on the grounds that it does not represent a challenge to people’s values, but is merely a way of helping them promote those values.

Again there is no scientific way of trading off the objective of paternalistic welfare maximisation against the objective of respect for consumers’ ‘rights’, including their right to make their own mistakes about the means they adopt to pursue their ends. Faced with a difficult choice between these conflicting values, one compromise policy is the ‘nudge approach’ that has been advocated notably by Thaler and Sunstein.⁹ In the UK there is now an official ‘nudge unit’, known as ‘The Behavioural Insights Team’, which takes account of recent academic research into behaviour to devise ‘nudge’ policies that will be of value to people and the community by making it easier for people to take the decisions that will best promote their welfare. Most obvious are ‘opt-out against opt-in’ policies for, say, pension schemes. But other ‘nudge’ policies would include, for example, health warnings on cigarette packaging, or clear indication of possibly harmful ingredients in some foodstuffs and so on.

However, even the ‘nudge’ approach is based on the assumption that people would be better off making choice A rather than choice B. This is then assumed to justify framing people’s choices in a way that will increase the chances of their making choice A, without falling into the paternalist trap of forcing them to do so. Neither of these assumptions is clearly compelling. For example, consider the aforementioned example of obliging employees to opt-out of a pension scheme rather than having to opt-in, so that the default is membership of the scheme. It is assumed that people who remain in the scheme but who might otherwise have failed to opt-in will be better off as a result. But it is possible that they would have been better off if they had not joined the pension scheme and had made their own arrangements for their old age, or even not at all for that matter.

Thus, again, there do not seem to be any clear-cut answers. In certain cases – such as the availability of possibly dangerous medicines – the case for paternalism may be very strong, but the case for restrictions on the consumption of fattening foods may be less so. The net social welfare gain from one policy or another may be difficult to access. And even if some reasonable rough balance sheet can be drawn up, there is still the problem of balancing the net gain or loss against the violation of personal liberty.

One important reason why the axioms of rationality specified earlier do not guarantee that people’s choices will promote their ‘welfare’ is that some people may deliberately and consciously choose not to maximise their own welfare. The welfare that most people expect to derive from their choices may depend partly on the effect they expect to have on the welfare of other people. Such modes of behaviour may satisfy the two axioms of rational choice without being motivated purely by self-interest. And modern theories of human behaviour in terms of game theory, particularly their behaviour in repeated games, which is the situation that people are faced with in society, explicitly allow for people’s preferences to include considerations such as altruism. For, over the course of human history, some concern for the interests of others has helped establish viable societies which indirectly promote people’s self-interest. In the non-human animal world ‘reciprocal altruism’, as it has been called in evolutionary biology, suggests that evolution has led to animals being programmed to have concern for other animals in proportion to the probability that the other animals will share their genes. In the human world it is the social evolutionary process that has contributed to the development of altruistic considerations in the human psyche.

The concept of ‘sympathy’, defined widely to include altruism and a capacity to enjoy other people’s good fortunes, played a crucial role in the work of Adam Smith and David Hume. Smith’s Theory of Moral Sentiments begins with the assertion that, ‘How selfish however man be supposed, there are evidently some principles in his nature, which interest him in the fortunes of others, and render their happiness necessary to him, though he derives nothing from it, except the pleasure of seeing it’. Smith was not limiting the concept of ‘sympathy’ to the more common current usage, which refers to sympathy with the misfortunes of other people. It also included a more general ability to empathise with other people. Sympathy and benevolence also play an important role in Hume’s moral and political theory. He wrote that ‘No quality of human nature is more remarkable, both in itself and in its consequences, than that propensity we have to sympathise with others, and to receive by communication their inclinations and sentiments, however different from or even contrary to our own’.¹⁰

In his famous article ‘Rational Fools: A Critique of the Behavioural Foundations of Economic Theory’, Amartya Sen distinguishes between his concept of ‘sympathy’ and what he calls ‘commitment’ (Sen, 1982a). Roughly speaking, sympathy is when you can raise your own welfare by raising somebody else’s, and commitment is when you reduce your welfare in order to raise somebody else’s. For Sen, ‘commitment does involve, in a very real sense, counter-preferential choice, destroying the crucial assumption that a chosen alternative must be better than (or at least as good as) the others for the person choosing it.…’ (Sen, 1982, p. 93).¹¹ Whereas sympathy can be regarded as a form of altruism for beginners, or ‘elementary altruism’, ‘commitment’ can be regarded as ‘advanced altruism’ and gives rise to choices that are expected to reduce one’s own welfare. In both cases one person’s utility function is dependent partly on another’s.

However, it is doubtful whether this particular discrepancy between people’s choices and their own welfare justifies any intervention in the pattern of consumers’ expenditures. The motivations in question – many of which may be highly commendable – do not necessarily detract from the welfare of society as a whole. Nor do they call for any revision of basic economic models of consumer behaviour. They can be seen as being part of the innumerable influences on the position of consumers’ demand curves in price–quantity space, without violating the assumption that these curves slope down from left to right in the normal way. It is not obvious that ‘commitment’ violates this assumption.

Lionel Robbins stressed this point in his famous book, The Nature and Significance of Economic Science. In it he argued that that the economic theory of how producers and consumers behave is in no way invalidated by dropping the assumption that homo economicus is some totally egoistical creature concerned only with the pursuit of his self-interest. In his book Robbins wrote that ‘…our economic subjects can be pure egoists, pure altruists, pure ascetics, pure sensualists or – what is much more likely – bundles of all these impulses’. He went on to say that ‘Considerations of this sort enable us to deal also with the oft-repeated accusation that Economics assumes a world of economic men concerned only with money-making and self-interest’ (Robbins, 1945 edn:95).

Robbins gives the example of a community that had been converted from pure hedonism to the objective of devoting their lives to God. The demand for wine would fall off and the demand for materials to build more churches, synagogues or mosques would rise. These changes in demands and supplies and their consequent changes in relative prices will follow the usual basic laws of economics. The price of wine would fall, and the price of the materials needed to build places of worship would rise. Labour would gradually move out of wine production into the construction industry. Demand and supply would still rule.

Or consider the more mundane case of Mrs X who, out of commitment to the survival of small local shops, goes out of her way to buy apples from a shop near where she lives rather than from a supermarket, which is cheaper and where there is no cat sleeping on the meat-slicer. In terms of a simple demand curve in price–quantity space her demand curve for apples from that particular shop would be further out to the right than would be that of some other customer who does not share her ‘commitment’. But it could still slope down from left to right. For example, if the shopkeeper raised the price of his apples she might tend to buy less from him and more from the supermarket, or she may switch from apples to pears. Indeed, if the shop owner raised the price enough Mrs X might decide that she need not carry commitment too far and would stop buying apples from that shop altogether. Thus, commitment does not make her demand curve for apples from that particular supplier slope the wrong way. It merely means that it is further out to the right than it would have been in the absence of this commitment. There are all sorts of reasons why people’s demand curves for particular products differ from each other, which are usually ascribed to differences in incomes and ‘tastes’. This does not mean that they do not slope downwards from left to right, which has been well-established in a vast amount of empirical research.

One of the important limitations on the normative significance of market outcomes is that the choices that people make in the market do not necessarily really reveal their ‘true’ preferences. And even where they do, their ‘true’ preferences may not always promote their ‘true’ welfare. There are various reasons for this, including irrational behaviour, inadequate information and many other psychological influences on choices other than those encompassed in conventional economic analysis. Thus while the assumption that, on the whole, people’s choices will promote their objectives in a rational manner is fairly compelling, there are various reasons why these choices may often fail to do so. This weakens the link between choices and people’s welfare.

Nevertheless, without the assumption that individuals’ preference orderings correspond fairly closely to their welfare orderings, welfare economics would not even get off the ground, let alone fly as far as it has done. For although welfare economics is not very much concerned with changes in the welfare of individuals as such for their own sake, welfare economics does require a criterion of an increase in the welfare of individuals. This is simply because the welfare of society is usually seen as a logical construction from the welfare of individuals. In other words if we use a theoretical model of micro-economics based on the notion of preference satisfaction in order to see under what conditions the market will allocate resources in a way that will best contribute to society’s welfare, we have to assume that a key variable in the model – namely an individual’s preferences – does have a fairly close correspondence to the individual’s welfare.¹² If we just throw up our hands and say ‘Oh well, all we are saying is that people are playing this game of making choices and ranking alternatives but these have absolutely nothing to do with their welfare’, the moral significance of the outcome of the market transactions could no longer be assessed in terms of its contribution to society’s economic welfare.

Furthermore, some of the welfare that people enjoy comes from their sharing in certain activities or duties or conveying sentiments in a manner for which mere monetary transactions are inadequate. And as has been argued in recent important contributions made by Elizabeth Anderson and Michael Sandel to the problem of the moral limits of the market, this part of their wider welfare can be destroyed, or misrepresented, by excessive marketing of certain goods and services. Sandel emphasizes the manner in which marketing certain goods, such as the sale of blood for medical purposes, diminishes the moral significance that voluntary blood donation gives. Elizabeth Anderson puts more emphasis on the manner in which putting a price on certain social facilities, such as access to parks, restricts their value to the personal values of individuals rather than the social significance of shared facilities. Some marketing procedures – such as Sandel’s example of the practice of allowing some prisoners to enjoy special facilities in jail – offend social notions of the equality with which punishment for crimes should be meted out irrespective of a person’s financial circumstances.¹³