An Introduction to Behavioral Economics

The purpose of any science is to develop theories that can accurately explain and predict phenomena. Perhaps the most prevalent criticism of economics is that it fails to do this very well in many cases. Of course, economists are often put in the spotlight, making public forecasts regarding the state of the economy, growth, inflation and unemployment rates, stock market indices, currency values, and so on. When these forecasts go awry, as they often do, the status of economics as a valid science is called into question. If the forecasts are wrong, then the underlying theory must be wrong too, no matter how complex it may be.

All theories are based on assumptions, or axioms. If a theory is wrong it is often an indication that these assumptions are misplaced. We have also seen that assumptions are made, not because they are necessarily believed to be realistic, but because they simplify the analysis.

Economic theories should not be criticized simply because they make unrealistic assumptions. Theorizing relies on abstracting key explanatory features from the rich complexity of the phenomena under investigation. Some theories can make some surprisingly accurate predictions even when they are based on rather abstract and prima facie unrealistic assumptions. This is the argument economists use when justifying the ‘as-if’ approach. However, when a theory fails to explain and predict well because of misplaced assumptions, then it is time to reject or revise the theory by constructing it on a sounder base of assumptions. In behavioral economics, this base has been drawn from other sciences, such as psychology, sociology and biology, including neuroscience. These sciences can provide economics with ‘process theories’, and are essential in achieving the unifying and integrative approach advocated throughout this text.

Let us recall at this point the main criteria, discussed in the first chapter, that are generally agreed to be important in establishing any scientific theory as being good and useful:

1 Congruence with reality

This relates to explaining existing empirical observations, and thus having good fit. The effect of this should be to produce more accurate predictions (although we have seen that goodness of fit and prediction do not necessarily go together).

2 Generality

Theories are more useful when they have a greater range of application. Sometimes this attribute is referred to as fruitfulness.

3 Tractability

This characteristic refers to how easily a theory can be applied to different specific situations in terms of making predictions. Mathematical complexity is often an issue in this regard.

4 Parsimony

We have now seen that some theories are more parsimonious than others, meaning that they are based on fewer assumptions or parameters. However, excessive parsimony may adversely affect both congruence with reality and generality.

Behavioral economics criticizes the standard economic model for being excessively parsimonious, resulting in a lack of congruence with reality or empirical fit. We saw in the first chapter that Ho, Lim and Camerer (2006) have proposed the addition of two more criteria: precision and psychological plausibility. They, and other behavioral economists, perceive their task as involving the revision of the basic assumptions underlying the standard model, and placing them on a sounder psychological foundation. This should, and does, improve congruence with reality, at the cost of reduced parsimony and sometimes analytical tractability. However, another benefit of the addition of more complex assumptions and parameters is that this often leads to more precise, as well as more accurate, predictions than those of the standard model. It is helpful at this stage if we illustrate the agenda of behavioral economics by using some of the examples discussed throughout the book.

Examples of behavioral revisions to the standard model

The examples given below illustrate that in some cases modeling a situation in behavioral terms, and using the model to make predictions is relatively easy, while in other cases the relevant behavioral factors are much more difficult to incorporate into a model capable of precise prediction. We will start with those examples where the use of additional parameters is more obvious and straightforward.

Game theory
This example is taken first, since it relates to the point in the previous paragraph regarding predictions. The standard model often predicts multiple equilibria, like in the market entry situation discussed in Case 9.2. Learning theory, thinking steps, and other behavioral concepts of equilibrium like QRE, can be introduced into the analysis by adding new parameters into the relevant models. Behavioral game theory can then provide more precise predictions about what will happen.
Loss-aversion
We have seen that this is a widespread phenomenon. It appears to provide the best explanation for the equity premium puzzle, along with choice bracketing. Loss-aversion can be modeled using a loss-aversion coefficient.
Impatience and time-inconsistent preferences
These factors are often modeled using hyperbolic discounting models. These models add the parameter to the standard discounting model, where represents a preference for immediacy.
Inequality-aversion and reciprocity models
By taking into account social preferences, these models can explain many seemingly irrational or unselfish actions, like refusing small offers in ultimatum bargaining games or cooperating in public goods games. For example, the inequality-aversion model of Fehr and Schmidt (1999) uses the coefficients and as envy and guilt coefficients respectively.
Choice bracketing
This aspect of mental accounting means that people select certain accounting periods over which they measure gains and losses. The choice of shorter periods results in a greater chance of making a loss in any given period.

Lack of fungibility

This factor explains why people may simultaneously borrow at high rates of interest, such as on a credit card, when they have current assets earning a much lower interest rate.

The representativeness heuristic

This and other heuristics have a profound effect on people’s judgments of probability, causing severe errors.

Anchoring effects

These affect both judgment and choice. Purely random influences, like social security numbers, have been shown to affect people’s valuation of items.

Endowment effects

Owners, or sellers, often place a higher value on items than buyers, reducing frequency of transactions. Again loss-aversion coefficients may be relevant.

Framing effects

We have seen that different ways of presenting identical information can cause preference reversals, like in the ‘Asian disease’ situation.

Decision-weighting

This can be used to translate objective probabilities into the subjective probabilities that are used to make decisions under risk.

Obviously this list is not exhaustive, but it does review a wide variety of situations where behavioral theories can improve on the standard model in terms of congruence with reality.

However, there are claimed to be two problems with the behavioral approach that we have come across in relation to accounting for the above factors, and these have led to some criticisms of behavioral economics.

11.2 Criticisms of behavioral economics

A profusion of models

One alleged problem, described by Fudenberg (2006), is that there are too many behavioral models, many of which have few applications. Fudenberg gives the example of modeling mistakes in inference, quoting various different models:

1 The ‘confirmatory bias’ models of Rabin and Schrag (1999) and Yariv (2005), which propose that agents miscode evidence that their prior beliefs indicate is unlikely.

2 The model of Rabin (2002a), which proposes that agents update as Bayesians in terms of modifying prior probabilities in the light of new evidence, but that they treat independent draws as draws with replacement. For example, if a lottery number has just come up, people would regard it as unlikely to come up again next week, even though lottery draws are independent from one week to the next.

3 The model of Barberis, Shleifer and Vishny (1998), which proposes that agents mistakenly see trends in independent and identically distributed data, such as spins of a roulette wheel or daily stock market fluctuations.

There are four points that can be made in response to the above criticisms:

1 Different models are appropriate in different decision situations

This point has been made a number of times in previous chapters: in Chapter 5 in relation to decision-making under risk and uncertainty; in Chapter 8 in relation to inter-temporal decisions; and in Chapter 10 in relation to decision-making when social preferences are relevant. Camerer and Loewenstein (2004) admit that the discipline is not a unified theory, but is rather ‘a collection of tools or ideas’. They also claim that the same is true of neoclassical economics. They repeat the claim of Arrow (1986), that economic models do not derive much predictive power from the single general tool of utility maximization. They then claim that behavioral economics is more like a power drill, which uses a wide variety of drill bits to perform different jobs. Thus the concepts of other-regarding preferences and social norms may be the appropriate analytical tools when examining public goods provision and bargaining games; the concept of time-additive separable utility may be relevant in asset pricing; remembered utility (or disutility) may be relevant in the decision to have a colonoscopy; anticipatory utility may be relevant in considering the decision to postpone a pleasurable experience; and loss-aversion may be relevant in explaining asymmetrical price elasticities.

2 Populations are heterogeneous

Modeling heterogeneous populations is a complex task, and we have seen an aspect of this problem in the inequality-aversion model of Fehr–Schmidt (1999). Not only do different people have different values of the parameters in a model, such as envy and guilt coefficients, but also different models may apply to different people or different cultures.

3 Conflicting theories are a feature of many sciences

In some cases, as in the example given by Fudenberg (2006) above, there are conflicting theories relating to the same phenomenon. Thus we cannot use as a response here the ‘power drill’ analogy of Camerer and Loewenstein (2004) discussed in the first point, where different tools are appropriate for different jobs. However, behavioral economics is hardly unique in terms of being a science where there are many conflicting theories regarding the explanation of a given set of phenomena. One example from psychology relates to the phenomenon where emotional reactions to life-changing events are surprisingly short-lived. There is now a substantial body of research in this area, but there are still five main competing theories relating to it. We agree with Fudenberg and grant that this is not an ideal situation from a scientific point of view. However, to some extent this is inevitable given the status of behavioral economics as explained in the next point.

4 Behavioral economics is a relatively new science

It is going through what may be termed the growth phase of its product life cycle (PLC). Using the PLC analogy, the growth phase is characterized by a profusion of product variations and firms, and then, as the maturity phase is reached, these products and firms are winnowed down by the market forces of competition. A similar process can be applied to the development of models in behavioral economics. As further research is carried out, some models are likely to be winnowed out as lacking support, while others have their status confirmed.

Lack of normative status

We discussed this criticism in Chapter 5 in comparing prospect theory with EUT. It is alleged that, by attempting to describe reality better, behavioral economics has lost its normative status. This is important because normative status is necessary in order for individuals, firms and governments to make good policy decisions.

As we have seen, the term normative is ambiguous when it is used in an economic context. The ambiguity arises from the interpretation of the word ‘should’, as for example in the expression ‘people should contribute toward public goods’. For our purposes here, no moral or value judgment is implied by the term ‘should’. The normative aspect implies a prescriptive status: what is necessary in order to achieve optimization. Therefore, the statement involves a conditional, and can be interpreted as meaning ‘people should contribute toward public goods if they want to maximize their utilities’. Obviously a utility function incorporating social preferences is required here. In this example behavioral economics is able to provide a normative statement in the prescriptive sense.

The area of behavioral economics where such prescriptive status is most inclined to break down is where individuals are making judgments and choices, and was discussed in Chapters 3, 4 and 5. We saw, for example, that preferences and choices are not necessarily based simply on attitudes and judgments. This complication arises from the violation of certain assumptions in the standard model relating to description invariance, procedural invariance and extensionality. There are also different concepts of utility: decision utility; experienced utility, which may be either remembered utility or real-time utility; anticipatory utility, which is based on predicted utility; residual utility, arising from reminiscence; and diagnostic utility, which people infer from their actions. The standard model only considers decision utility, which it assumes is based on attitudes and preferences, which in turn are assumed to be consistent.

Once these demonstrated anomalies in the standard model are taken into account, the simple utility maximization model involving indifference curve analysis is no longer applicable. We have also seen that under conditions of risk a similar situation occurs, in that we can no longer use a simple EUT preference function that is to be maximized. However, in spite of the problems raised by the violation of basic assumptions, this does not mean that the behavioral model cannot be prescriptive in principle. Although the analysis is more complicated, and often difficult to achieve in practical terms, it was shown in Chapters 4 and 5 that prescriptive statements can still be possible. If we use the concepts of real-time, or moment-based, utility, and psychological utility, in order to provide an all-inclusive measure of utility, then we can draw some conclusions about maximizing welfare. For example, we can attempt to resolve the paradox that people may value more highly a job with a higher relative salary but lower absolute salary, because they think it will make them happier, but actually choose a job with lower relative salary and higher absolute salary. Only by becoming aware of these paradoxes to which behavioral economics draws attention can better policy decisions be made.

11.3 Methodology

We have seen at the start of this book that some aspects of this issue are highly controversial in science. For example, reductionism is often opposed in the social sciences, since practitioners find that their authority is being usurped by disciplines lower down the hierarchy or causal chain. It is not proposed here to revisit all the methodological issues described in the second chapter, relating for example to experimental design, but to focus instead on the most fundamental issues. In particular, the methodology advocated here involves using some of the ‘best practice’ from the various behavioral sciences that are currently in conflict with each other, in order to unify these sciences. When economists use the approaches and models commonly associated with psychology, sociology and biology, in addition to their usual methods, in an appropriate combination, then the behavioral sciences will make the most progress in terms of explaining and predicting decision-making. At a normative level, this will also help people to make better decisions, in terms of improving welfare. This aspect is discussed in Section 11.5. In order to facilitate this unification the following issues need to be clarified and resolved: (1) relationships between assumptions and conclusions; (2) the role of gene-culture coevolution; (3) the role of game theory; and (4) the relationship between parsimony and universality.

Assumptions and conclusions

Students frequently misinterpret and misuse these two terms. It is a basic tenet of the scientific method that in any science one starts off with certain assumptions, and based on theory and empirical evidence, one proceeds to form certain conclusions. These conclusions may involve a rejection of some original theory or hypothesis, a confirmation of it, or a modification and refinement of it. Where matters become confusing is that in the causal sequence of events, and this involves the relationships between different sciences, conclusions in one area become assumptions in another. We have seen some examples of this in the first two chapters, but more specific examples are needed here to illustrate this phenomenon in terms of how it relates to behavioral economics.

The standard model assumes that people are purely self-interested. As we have seen, this is an example of an assumption that is made for convenience, to simplify analysis, rather than because it is believed to be universally true. It is obviously a useful assumption, enabling a wide variety of predictions to be made, many of which are reasonably accurate. Behavioral economics attempts to find a sounder psychological foundation for economic theorizing, and therefore takes into account social preferences when trying to explain behavior. In attempting to build this better foundation the behavioral approach aims to search for psychological regularity. For example, it may find psychological regularity in the empirical evidence that people tend to reject low offers in ultimatum bargaining games, and that people also tend not to make such offers. This finding, a contradiction to the standard model, then provides a basis for a behavioral theory incorporating social preferences. In this context this aspect of psychological regularity becomes an ‘assumption’ on which the development of different behavioral theories, like inequality-aversion and reciprocity, can be based.

However, this analysis only goes one level down the causal sequence. We can then ask: what causes people to have both positive and negative reciprocity? Behavioral economics leaves many questions with this status, for example:

Why are people influenced by framing effects?
Why do people exhibit loss-aversion?
Why do people show self-serving bias?
Why do people have endowment effects?
Why do people make choices that do not appear to make them happy?

It is not usually regarded as part of the agenda of behavioral economics to answer these questions. These effects, or psychological regularities, are taken as being the building blocks, or ‘assumptions’ on which alternative theories to the standard model must be constructed. The task of answering these questions is ‘delegated’ to psychology, where these regularities are treated as being ‘conclusions’ that must be explained by underlying psychological theories. Psychologists in turn make certain assumptions regarding neuroscience, in terms of how the brain works at a physiological level, and in terms of how the brain has evolved. Thus we move one further level down the causal sequence into the realms of neuroscience and evolutionary biology. As explained in the second chapter, this is the nature of hierarchical reductionism, which has proven to be extremely effective over the last few centuries in terms of developing theories with wide and deep explanatory and predictive powers.

The role of evolutionary psychology

The discussion of reductionism, assumptions and conclusions, leads us once more to the equally controversial role of evolutionary psychology, which has been touched on several times throughout the book. In order to obtain a proper perspective the role of evolutionary psychology has to be viewed as being a component of the larger role of gene-culture coevolution. Not even the most extreme evolutionary psychologist would claim that all human behavior can be explained by examining, and guessing, what our ancestral lives and environments were like and theorizing about how human brains adapted to this. Unfortunately this is a straw man that has become popular in much of the social science literature critiquing the discipline. More realistically, the problem of evolutionary psychology (EP) is summarized by Camerer and Loewenstein (2004) as follows:

… it is easy to figure out whether an evolutionary story identifies causes sufficient to bring about particular behavior, but it is almost impossible to know if those causes were the ones that actually did bring it about (p. 40).

This problem leads to the main criticism of evolutionary psychology as a science: that it fails to make testable, meaning falsifiable, predictions. It is useful to illustrate this problem by taking one particular example, concerning the tendency to cooperate rather than defect in ‘one-shot’ situations like prisoner’s dilemma games. The claim in EP is that subjects in these ‘one-shot’ games are unable to switch off their inbuilt cooperation mechanisms. Thus, even when they are told explicitly that they are playing a one-shot game, and acknowledge this, in psychological terms they are still playing a repeated game. As Camerer and Loewenstein point out, this makes the EP hypothesis unfalsifiable if one is trying to predict comparative behavior in one-shot games with behavior in repeated games.

Certainly there is a need to be aware of this problem. Indeed, evolutionary psychologists will always face the problem that concrete evidence from our ancestral Pleistocene past is flimsy. The archaeological and biological record is sparse and is likely to remain so, although progress in molecular biology and genetics may make some progress in the latter area. Therefore conjecture and speculation, despised by ‘hard’ scientists in particular, is inevitable. Much theory in EP proceeds on the basis of examining modern observations and making the following comparison:

Prob (theory X from EP| observed values) > or < Prob (other theory Y| observed values)

The EP theory is regarded as being vindicated if the EP theory has greater probability of generating the observed data than any other theory. In many situations there is no other recognized theory. For example, we have seen from Kahneman and Tversky (1973) that people are much better at solving problems when given data involving frequencies rather than probabilities. Single event probabilities are often nonsensical in practical terms. For example, it may make little sense to say that there is a 35% chance that a woman is pregnant: either she is or she isn’t pregnant. Similarly, in evolutionary terms it may seem strange to say that there is a probability of 0.375 of finding berries in a particular valley. On the other hand people understand more readily the statement that, in the last eight visits to the valley, one found berries three times. This may be regarded as another ‘just so’ story, but the author is not aware of any other comparable theory in cognitive psychology which can explain this phenomenon.

In addition, it should also be pointed out that EP predictions are falsifiable in a number of areas. If experiments were to observe a widespread tendency to defect in one-shot games, this would falsify the hypothesis. We have also given an example in Case 10.1 related to the Wason Test, where EP can make some sharp and falsifiable predictions. Let us take a final example from a different area, regarded as being the domain of cultural influences. Parasites are known to degrade physical appearance. Evolutionary psychologists Gangestad and Buss (1993) therefore made a prediction that people living in ecologies with a high prevalence of parasites should place a greater value on physical attractiveness in a mate than people living in ecologies with a low prevalence of parasites. This hypothesis was tested by collecting data from 29 cultures relating to the prevalence of parasites and the importance that people in those cultures attached to physical attractiveness in a marriage partner. The results confirmed the hypothesis, finding significant positive correlation between the two variables. This example is taken because it illustrates how cultural differences, often assumed to be non-evolutionary in nature, can be explained by universal evolved psychological mechanisms that are differentially activated across cultures. Of course, other theories may be developed to explain the same observed relationship, but the essential point is that EP, when used carefully and thoughtfully, can provide testable hypotheses.

The role of game theory

We have seen, particularly in Chapters 2, 9 and 10, that game theory plays a fundamental role in understanding decision-making, simply because so many decisions we make involve some kind of social interaction. However, we have also seen that standard game theory is inadequate for modeling many situations, since it is either silent, for example, indicating multiple equilibria, or it produces anomalies, as in bargaining games. In order to gain a better understanding of decision-making we need to take account of other-regarding preferences and bounded rationality in particular, and recognize the failure of backward induction. This means that other approaches to game theory, meaning behavioral game theory and evolutionary game theory, need to be used in conjunction with the standard approach. The inevitable result is more complexity in terms of the analysis, and this is related to the next issue.

Parsimony and universality

Another criticism leveled at EP is that the explanations that it provides lack universality, and appear to have an ad hoc nature about them. We have come across these terms before in connection with the evaluation of theories. The term ad hoc means ‘for a particular purpose’, and scientists tend to use the term disapprovingly, since it often implies that a theory is being twisted or convoluted in an unnatural way to explain away undesirable observations that cannot be explained by any standard form of the theory. The result is a loss of parsimony, as the theory becomes more cumbersome. A good example from the field of astronomy is the invention in medieval times of the concept of ‘epicycles’ (circular orbits centred on other circular orbits). The purpose was to explain aberrations in the observed movements of the planets that could not be explained by the standard Ptolemaic theory of the geocentric universe. Copernicus’s theory of the heliocentric universe, combined with Kepler’s notion of elliptical planetary orbits, turned out to be a much more parsimonious theory, as well as being more accurate.

However desirable parsimony and universality may be to our structure-loving minds, the irony is that the human mind appears to be good at some things, but is surprisingly incapable of performing other functions. Evolutionary psychologists explain this behavior in terms of the modularity of the brain; it is not a universal problem-solving device. In order for the brain to develop such a broad capacity enormous resources would be required, and this would be very wasteful, since many of the capabilities of such a brain would never be required in real life. The EP model, expressed most clearly in various contributions from Cosmides and Tooby, is that the brain did indeed develop in an ad hoc fashion, as it adapted to deal with the problems that it actually had to face in reality. They liken the mind to a Swiss army knife, with many blades designed for specific tasks.

EP, by proposing that the mind is a device with ad hoc functions, is not alone in terms of appearing to involve ad hoc explanations. The same accusation is often leveled at behavioral economics; this is essentially the criticism of Fudenberg (2006) that was discussed earlier.

This discussion may seem to lead to a messy, even contradictory, conclusion regarding the desirability of parsimony and generality. There are two points that may clarify this:

1 If we accept that the mind is modular (and certainly more evidence is needed regarding the nature of this modularity), then it should come as no surprise that a variety of tools is necessary to study its functions and operations, as they are manifested in human attitudes and behavior.

2 This variety of tools may still have some commonality, just as all the various drill bits can be used with the same drill. In behavioral economics there are various commonalities; in particular the concepts discussed in prospect theory and mental accounting have a very general application to a wide variety of human behavior.

This discussion relating to evolutionary psychology will now help us to draw some conclusions relating to rationality, a recurring theme throughout this book.

11.4 Are we really irrational?

As more studies claiming and documenting the violations of rationality described above have been published, there have also been a number of attempts to re-establish the status of rationality, by defending the notion that people do generally behave rationally. These defenses make various objections to the claims of irrationality. Shafir and LeBoeuf (2002) have classified these objections into three main categories: trivializations, misinterpretations and inappropriate tests. We will follow this useful classification in the following discussion.

Trivializations

This category of objection relates to the claim that the alleged violations of rationality are unsystematic and unreliable. There are five main aspects that can be considered here: randomness, incentives, justification, expertise and need for cognition (NC).

1 Randomness

Sometimes it is claimed that deviations from the norms prescribed by the standard model of rationality are purely random errors, commonly observed in statistical distributions. This claim is easily refuted, however, since a closer look at the evidence in a large number of studies shows systematic errors of overwhelming statistical significance. We have seen that systematic errors are in a predictable direction, for example many preference reversals.

2 Incentives

A common claim is the participants in studies often lack the required motivation to provide true or reliable results. In general, however, it has been observed that incentives do not reduce, let alone eliminate, the observed violations of rationality. This is not to say that incentives do not affect behaviour in experiments. For example, we have seen that, by using earned as opposed to unearned rewards in a dictator game, many subjects do behave purely selfishly, as the standard model predicts (Cherry, Frykblom and Shogren, 2002). However, in the extensive review by Camerer and Hogarth (1999) of 74 studies that manipulated incentives, the authors concluded that ‘there is no replicated study in which a theory of rational choice was rejected at low stakes … and accepted at high stakes’. Most violations persist in the face of even large incentives, for example the amount of one month’s salary in a study involving Chinese workers (Kachelmeier and Shehata, 1992). Furthermore, incentives do not prevent a large proportion of small businesses failing. Neither do incentives help to reduce the incidence of optical illusions. It should also be noted that, even if incentives are successful in increasing motivation, people still need to apply the correct insights in order to improve their performance; mere enthusiasm does not suffice for good decision-making.

3 Justification

Another method that can be used to increase the involvement of participants is to ask them to justify their responses. Although this sometimes reduces inconsistencies like framing effects (Takemura, 1994; Sieck and Yates, 1997), such effects often persist even when justification is provided (Fagley and Miller, 1987; Levin and Chapman, 1990; Miller and Fagley, 1991; Takemura, 1993). As with

incentives, greater involvement does not ensure better performance; insight is still required.

4 Expertise

It is sometimes claimed that more expert subjects are more likely to exhibit rational choice, having more relevant knowledge and familiarity with the tasks involved. A variation of this objection is that people learn from their mistakes, and thus become more expert. An example of a study reporting this finding is List (2004), who found that experience in the market eliminated the endowment effect. The problem with this ‘learning effect’ objection is that learning relies on feedback. In real life this feedback is problematical because many situations are unique, outcomes are often delayed, and they are often influenced by multiple factors. People do not frequently sell their houses, for example, and even when they do repeat this process, it is often under quite different circumstances from previously. Even when people do learn from experience there is much evidence that experts make the same sort of violations that non-experts do. For example, physicians and nurses have been found to make preferences and choices violating the first two criteria for rationality, involving the laws of probability and consistency (Casscells, Schoenberger and Grayboys, 1978; Redelmeier and Shafir, 1995; Redelmeier, Shafir and Aujla, 2001). Financial experts have been prone to making judgments and choices involving preference reversal and framing effects (Benartzi and Thaler, 1995; Siegel and Thaler, 1997; Tversky and Kahneman, 1992). Professional gamblers have also been observed to exhibit preference reversal (Lichtenstein and Slovic, 1973). It appears that experts are subject to certain judgmental biases discussed earlier, like overconfidence (Faust, Hart and Guilmette, 1988) and hindsight bias (Arkes et al., 1981). It is, therefore, highly improbable that such violations can be attributed to lack of motivation or understanding.

5 Need for cognition

This factor relates to the notion that people differ in their tendency to engage in and enjoy thinking. It has been proposed that people with a greater need for cognition (NC) are more motivated to make better decisions, and are more likely to give proper consideration to the relevant aspects. There is indeed some evidence that certain violations may be reduced with high-NC participants, for example framing effects, with improved judgments in conditional probability situations, and less consideration of irrelevant factors like sunk costs (Smith and Levin, 1996; Stanovich and West, 1999). However, in spite of these results, there is no evidence that high-NC participants have better insight in judgments involving hypothesis testing and prisoner’s dilemma situations (Stanovich and West, 1999).

There are some studies that combine a number of the above objections. For example, the discovered preference hypothesis (DPH) developed by Plott (1996) proposes that people’s preferences are not necessarily revealed in their decisions, as we saw in discussing criticisms of prospect theory in Chapter 5. According to this theory, preferences have to be discovered through a process of information gathering, deliberation and trial-and-error learning. Subjects must therefore have adequate opportunities and incentives for discovery, and it is claimed that studies lacking these factors are unreliable. It has been argued that the best type of experimental design to ensure that the requirements of the DPH are met is a single-task individual-choice design (Cubitt, Starmer and Sugden, 2001). Such a design can ensure that subjects get an opportunity to practice a single task repeatedly, with the requisite learning effect, and it can also ensure simplicity and transparency, which are difficult to achieve in market-based studies, where tasks are more complex and involve interactions with others. However, when Cubitt, Starmer and Sugden reviewed the results of nine different experiments involving such a design, they found that the results still violated the criteria for rationality, specifically in terms of the independence axiom for consistent choices in Allais-type situations.

In summary it seems fair to say that violations of rationality cannot be dismissed as trivial errors in performance. Such violations persist even with highly motivated experts who give serious consideration to the problems involved.

Misinterpretations

A second objection to the rationality violations is that the irrationality perceived by researchers arises because participants tend to adopt a different understanding of the tasks from that intended, and that in the light of the participants’ understanding their responses are rational (Macdonald, 1986; Hilton, 1995; Levinson, 1995; Schwarz, 1996; Slugoski and Wilson, 1998; Hertwig and Gigerenzer, 1999). We will discuss four main aspects of this objection.

1 The conjunction error

It is sometimes claimed that subjects may infer that the researcher only gives relevant and nonredundant information. This is particularly important in tasks relating to the representativeness heuristic, like the Linda problem discussed earlier, involving conjuncts and conjunctives. When comparing the statements ‘Linda is a bank teller’ and ‘Linda is a bank teller and is active in the feminist movement’, subjects may infer that in the first statement Linda is not active in the feminist movement and, therefore, rate the statement as less probable than the second statement. In more general terms, subjects may, after seeing the statement with the conjunctive A and B, infer that the conjunct statement involving just A refers to ‘A and not B’. Indeed, some studies have found that rewording the construct and giving some logical clues, like saying ‘Linda is a bank teller whether or not she is active in the feminist movement’ does tend to reduce the conjunction error (Dulany and Hilton, 1991; Politzer and Noveck, 1991). Another study has found that the tendency to make the error was correlated with a person’s conversational skill (Slugoski and Wilson, 1998). However, other studies have found that the conjunction error occurs across a wide variety of problems, and persists in a majority of subjects even when reinterpretations of the conjuncts are given to aid subjects in understanding the logic (Morier and Borgida, 1984; Tversky and Kahneman, 1983).

2 Under-reliance on base rates

Another kind of error relating to representativeness concerns the insufficient reliance on base rates in likelihood judgments. A good example is the AIDS diagnosis situation (Tversky and Kahneman, 1982), where a base rate of one in a thousand people have the disease, but people, even medical experts, are over-influenced by the representativeness description that the positive test is 95% accurate. The critique here is that conversational inferences are again important. When the base rate is stated after the representativeness description, rather than before, this increases the reliance on base rates and improves the normativeness of responses, meaning that they conform more closely to the rationality model (Krosnick, Li and Lehman, 1990). When descriptions are said to be randomly sampled and unreliable, reliance on base rates increases significantly (Ginossar and Trope, 1987; Schwarz et al., 1991). Also, when base rates are varied in the experiment, reliance on base rates improves (Fischhoff, Slovic and Lichtenstein, 1979). In spite of these improvements in performance, however, the evidence suggests that an under-reliance on base rates persists (Fischhoff, Slovic and Lichtenstein, 1979; Schwarz et al., 1991).

3 Framing effects

It has also been claimed that conversational factors may be involved in framing effects, like the Asian disease situation discussed earlier. Some researchers have suggested that the frames in options A and C, supposedly identical, may not be seen as such by some subjects (Berkeley and Humphreys, 1982; Macdonald, 1986). For example, option A, ‘saving 200 lives with certainty’, may be construed as meaning ‘saving at least 200 lives’, while option C, ‘400 dying with certainty’, may be construed as ‘at least 400 people dying’. Obviously, under this construal option A is preferable to option C, and the apparent preference reversal disappears. However, evidence also suggests that the majority of subjects do not interpret the options in this manner, and do in fact interpret the options as identical (Stanovich and West, 1998).

4 Interpretations of probability

A further area where different interpretations of terms may be involved concerns the general concept of probability. There are three main interpretations of the term probability:

(i) Classical – probabilities are determined a priori; typical situations involve gambling, like tossing a coin, rolling a die, or drawing a playing card from a pack.

(ii) Empirical – probabilities are determined a posteriori; past similar situations are observed and relative frequencies are used to calculate probabilities. A typical example is the probability of rain in September in London.

(iii) Subjective – probabilities are estimated based on intuition and experience. An example is the probability that firm A will increase its price next month.

However, it should be stressed that in all three cases the general mathematical maxims of probability hold true, for example the probability of any event must be between 0 and 1, 0 referring to impossible and 1 referring to certain.

It has been objected that some subjects do not interpret probability in mathematical terms, and this causes them to make certain errors, like the conjunction fallacy (Hertwig and Gigerenzer, 1999). This study showed that this error could be reduced by adding further information in the tasks involved. However, other studies have shown that subjects do not have different interpretations of probability than researchers, and that the conjunction fallacy persists even when further information is given to aid judgments (Tversky and Kahneman, 1983; Kahneman and Tversky, 1996; Stanovich and West, 1998).

It should also be stated in summary that we have to be careful not to legitimize all misinterpretations by subjects, for this tends to beg the issue. We must ask why subjects tend to misinterpret certain terms in certain ways.

Inappropriate tests

The last, and most fundamental, group of objections to violations of rationality are based on the appropriateness of tests for rationality. Shafir and LeBoeuf (2002) classify these objections into three main categories: computational limitations, inappropriate problem formats and inappropriate norms.

1 Computational limitations

Some researchers have objected that computational limitations are the source of many violations, and that it is not useful to define rationality in such a way that it is out of reach for the majority of people. This is really the easiest objection to counter. Most of the tasks involved in the experiments described earlier are computationally very simple. It appears to be the conceptual aspects that cause problems and errors (Agnoli and Krantz, 1989; Fong and Nisbett, 1991; Frank, Gilovich and Regan, 1993). Furthermore, when the source of the errors is pointed out, subjects quickly learn to avoid them (Fiedler, 1988; Tversky and Kahneman, 1986; Tversky and Shafir, 1992). It seems therefore that the problem lies in the heuristic procedures used.

2 Inappropriate problem formats

It has been suggested, in particular by evolutionary psychologists, that the type of problem used in many experiments has been the cause of many problems, and that if the nature of the problems set is changed to coincide more with the types of problem encountered during our Pleistocene past, then the errors will tend to disappear (Cosmides and Tooby, 1996; Gigerenzer, 1996b). This point has been discussed in various places in the text; at this stage it is sufficient to point out that, indeed, general performance on many tasks, like the Wason Test discussed in Case 10.1, is improved when the terms of the problem are restated to resemble realistic situations in our past, like enforcing social contracts (Cosmides and Tooby, 1992; Gigerenzer and Hug, 1992). However, this objection is really only valid if the problems set in experiments do not resemble current everyday problems, and this is not the case. Most of the tasks and problems are not entirely abstract, and the interesting point is that, while people may be proficient at solving problems related to our evolutionary past, they are not nearly as proficient at solving current problems. In this situation it appears to make sense to define rationality as relating to current problems, not those of our past. However, this does not invalidate the most important point of the evolutionary psychologists, which is that our minds evolved to solve past, not current, problems.

3 Inappropriate norms

The most fundamental of all objections to the rationality violations is that inappropriate normative standards of rationality are imposed (Wetherick, 1971; Smith, 1990; Gigerenzer, Hoffrage and Kleinbolting, 1991; Lopes and Oden, 1991; Gigerenzer, 1996a; Plott, 1996; Binmore, 1999). This objection relates to the definition of rationality and the discussion in the first chapter. In essence this objection proposes a relaxation of some of the normative rules in the standard model to accommodate certain aspects of choice and behavior that violate principles such as independence and transitivity.

The essential problem with this approach is that the suggestions as to which principles to relax are often somewhat arbitrary. Furthermore, if rationality is redefined in order to allow such a relaxation, rational choice will then involve violations of simple laws of probability, with the result that ‘rational’ individuals will engage in gambles that in general they are bound to lose (Resnick, 1987; Osherson, 1995; Stein, 1996).

It can be argued that there are at least two approaches here that are not arbitrary. A first approach essentially limits the application of EUT to situations involving risk, and excludes situations involving uncertainty (Binmore, 1999; Plott, 1996). This therefore excludes all new situations decision-makers might encounter, like buying a house. In principle there is nothing incorrect about this approach, but in practice it imposes severe limits on the application of EUT. Furthermore, it can still be claimed that violations of rationality will occur even in the more limited scenario.

A second approach is that of Vernon Smith (1990, 1991), described briefly earlier. Smith’s views have much in common with the work of a founding father of behavioral economics, Simon (1956, 1957, 1959, 1978). Both reject the norms of the standard model as far as rationality is concerned. Smith claims that the appropriate norms for judging rationality are the end results of decision-making, in terms of long-run market efficiency. He also stresses that short-run errors and biases may be corrected in the long run by experience and learning.

The most obvious problem here is that short-run errors and biases are often not corrected by learning, as mentioned earlier. Furthermore, there are some glaring systematic and long-run inefficiencies in the market; a prominent example is a phenomenon referred to as the ‘equity premium puzzle’, discussed at length in Case 6.1. Equities in the US have over the long run (since 1926) yielded an annual return about 6% higher than bonds, which is a much greater difference than can be explained by any reasonable degree of risk-aversion. Other examples, such as the tendency to bet on long-shots in the last race of the day, were also discussed in Chapter 5.

A different type of problem, both with Smith’s approach and with the general approach regarding the use of inappropriate norms, is that the normative standards of the standard model which are so widely violated in practice are actually generally adhered to in principle by the majority of people. It therefore seems to make little sense to abandon them. One might even say that it appears irrational to abandon them.

By way of summary we can say that in general the above objections relating to trivializations, misinterpretations and inappropriate tests cannot explain the large body of systematic evidence documenting a wide variety of violations of the standard model of rationality; nor does it make sense to redefine rationality in order to allow for such choices and preferences.

11.5 Welfare and happiness

Measuring happiness

Although happiness, and the search for happiness, is the most fundamental part of our lives, economists are not ‘happy’ with examining the concept. The term ‘happiness’ is not frequently found in the economics literature, unlike the concept of welfare, which is found in abundance. This is partly because welfare can be measured in objective terms, usually in terms of GDP per head. In the last two decades both economists and politicians have taken a greater interest in measuring welfare, and there is some controversy over this issue, with an increasing number of variables now being taken into account. These include not just real income, but also factors such as life expectancy, health levels, environmental quality, job satisfaction and stress levels. However, when these other variables are taken into consideration, some of them (like the last two) have to be measured subjectively, and a subjective weighting system must be used. The consequence of this is that economists tend to refer to such measures as relating to subjective well-being (SWB) as opposed to economic welfare. Nevertheless, it is this concept of SWB that corresponds with happiness, and that people are concerned with in terms of maximization; GDP per head, consumption and other economic variables are simply means to that end.

Figure 11.1 The Easterlin paradox (US, 1973–2004)

Another frequently used method of measuring happiness is simply to ask people. Perhaps the most prominent example of this is the World Values Survey. This has been carried out by a network of social scientists on an ongoing basis since 1981, and encompasses nearly 100 countries. Representative samples of 1000 people from each country are asked to report their satisfaction with ‘life as a whole these days’, on a scale from 1 (dissatisfied) to 10 (satisfied). It has been found that there is a correlation of 0.62 between mean life satisfaction and mean purchasing power income across all countries (World Values Survey, 1981–2007).

These surveys have found that that there are important cultural factors affecting reports of life satisfaction. In general cultural factors vary across two dimensions: (1) traditional versus secular-rational values; and (2) survival versus self-expression values. In cultures where traditional values are important (for example Asia, Africa and the US) people judge their life satisfaction by comparing their individual situations with social norms, such as employment and marital status, while in secular-rational cultures (for example, North-Western Europe) people introspect as far as their emotions are concerned in order to judge their happiness. In cultures where survival values, meaning physical and economic security, are important (for example, Eastern Europe), reports of life satisfaction tend to be lower than in countries where self-expression values (like democracy, civil rights and environmental protection) are more important.

Because (1) happiness can only be measured subjectively; and (2) cultural factors have an important influence, it is difficult to make meaningful comparisons between different countries. Neuroscience is now beginning to allow us to estimate objective measures of happiness, through the use of PET and fMRI scans, but there is still much more to be achieved in this respect. Even if objective measures of happiness are attainable, we have already seen both in this chapter and in Chapter 3 that the relationship between welfare and happiness is a complex one. Even though richer people tend to be happier than poorer people within the same country, and people in richer countries tend to be happier than people in poorer countries, the ‘Easterlin paradox’ shows that increasing income is not associated with increasing happiness in many countries. This situation is shown for the US in Figure 11.1.

Furthermore, a number of studies show that people tend to be ‘mean-reverting’ in response to life shocks like winning the lottery or disability when long-term happiness is measured. This means that they revert to close to their original level of happiness within a period of 12–24 months of the original shock.

Psychological models

Psychologists have proposed several theories of happiness to account for these empirical findings:

1 Happiness is a dispositional trait rather than a reaction to external events (Costa and McCrae, 1984; Lykken and Tellegen, 1996).

2 People adapt to repeated experiences of the same event, as that experience becomes a reference point to which new experiences are compared (Brickman and Campbell, 1971; Kahneman and Tversky, 1979; Parducci, 1995).

3 Happiness results more from pursuing a goal rather than attaining a goal (Davidson, 1994; Diener, 2000).

4 People possess a psychological immune system that speeds recovery from negative emotional events (Freud, 1937; Festinger, 1957; Taylor, 1991; Gilbert et al., 1998; Vaillant, 2000).

5 People reduce the emotional power of events by ‘making sense’ of them; this is also referred to as ordinization, meaning making events ordinary, predictable and explainable (Wilson et al., 2003).

Although each theory may explain some aspects of the empirical data observed, the first four all tend to leave certain aspects unexplained. The first theory does not explain why external events do affect happiness or why this effect is short-lived. The second theory does not explain why single occurrences of an event cause adaptation or a return to the original emotional state. The third theory does not explain why people recover rapidly from negative events, and is therefore asymmetrical. The fourth theory is asymmetrical in the other direction: it cannot explain why people recover quickly from positive events.

The fifth theory is in many ways the most satisfactory in terms of explaining the different findings. In order to understand it further the concepts of homeostasis and allostasis need to be explained. Homeostasis is a well-known biological principle, whereby various systems in the body have an optimal set point, and deviations from this point trigger negative feedback processes that attempt to restore it. Examples are body temperature, the level of blood sugar and electrolyte balance. The term allostasis was introduced by Sterling and Eyer (1988) to refer to a different type of feedback system whereby a variable is maintained within a healthy range, but at the same time is allowed to vary in response to environmental demands. Heart rate, blood pressure and hormone levels are variables in this category. Thus, when we exercise, both heart rate and blood pressure are allowed to rise in order to optimize performance. Wilson et al. suggest that happiness is also a variable in this category.

Wilson et al. also give an explanation regarding why an allostatic system represents a functional adaptation as far as happiness is concerned; in fact they give three reasons. First, they claim that it is dysfunctional for people to remain in an extreme emotional state, since they cannot adjust to new emotional events. Second, people in extreme emotional states tend to make less rational decisions, resulting in self-harming results, as already discussed. Third, extreme emotional states are physiologically debilitating when they continue for sustained periods; a good example of such situations occurs in wartime, when soldiers may be under extreme stress for prolonged periods, resulting in neurasthenia, ‘shellshock’ and similar conditions.

The theory also maintains that the ordinization process evolved as a cognitive mechanism in order to maintain affective or emotional stability. This aspect involves the uncertainty aversion principle, a fundamental factor in many psychological theories. The principle is described by Gilovich (1991):

We are predisposed to see order, pattern, and meaning in the world, and we find randomness, chaos, and meaninglessness unsatisfying. Human nature abhors a lack of predictability and the absence of meaning (p. 9).

Again it is easy to see evolutionary advantages in uncertainty aversion. It drives people to take steps to reduce uncertainty, by trying to explain and predict their environment, thus furthering their chances of survival and reproduction. Reducing uncertainty also plays a large part in increasing pleasure and decreasing pain, relating the principle to the standard model. The implications of the uncertainty aversion principle for the standard model were discussed further in Chapter 5.

The implications of the principle are very general. It has been argued that the main function of religion and art is to help people make sense out of a confusing, unpredictable world (e.g. Jobes, 1974; Pfeiffer, 1982; Dennett, 1995; Pinker, 1997). We now need to consider more specifically how ordinization involves uncertainty aversion, how the ordinization process operates, and finally what its implications are for decision-making.

Ordinization is the process of assimiliation, accommodation and ‘sense making’ (Wilson et al., 2003), and it is a process that occurs automatically, without any conscious awareness. As time goes on painful events become part of our life story, making them seem less novel and surprising. The events cause us less sadness, both because we think about them less and because when we do, the emotional reaction is less strong. In this situation ordinization and uncertainty aversion result in positive hedonic consequences, as pain is reduced. A similar process occurs for positive effects, but in this case there is a paradox, since the hedonic consequences are negative: we seek positive experiences that increase pleasure, but by doing so we rob these experiences of their future hedonic power. This paradox will be discussed further in the next section, in connection with EUT, but we can give an example here for illustration. We may have a goal of seeking promotion, which, when achieved, gives intense satisfaction in the immediate aftermath. As time progresses we think about the event less, and it becomes an accepted and normal part of our life history. It may even come to appear inevitable in retrospect, a phenomenon known as hindsight bias. More generally this bias means that events become more predictable in retrospect than in prospect, and this is an important feature of ordinization, again being a result of nonconscious, automatic mental processes (Pohl and Hell, 1996). However, the paradox is that sometime after gaining the promotion we are essentially in the same emotional state as before the promotion, and in order to increase pleasure and happiness we now have to progress to the next rung. Of course, if and when that is achieved, the same phenomenon will occur again. We are in the situation of the Red Queen from Alice in Wonderland, who has to run faster and faster just to stay still. As Matt Ridley has noted in his book The Red Queen (1993), this phenomenon is ubiquitous in biological systems, because of homeostasis and ‘arms races’.

The discussion of ordinization explains why emotional reactions tend to be short-lived. But why do we consistently tend to underestimate the effects? One reason for this durability bias is that the relevant mental processes are automatic and nonconscious. A second reason is that we may fail to generalize from our experiences; we may come to believe that there was a specific factor involved (we bought a bad TV set, or we were not really that angry), rather than a general process. A final reason involves the phenomenon of retrospective durability bias. For example, we may forget that the product did not make us happy for as long as we had originally anticipated. Both prospective and retrospective durability bias may be caused by focalism, meaning that people think too much about the event in question and fail to consider the consequences of the many other events that are either likely to be going on, or were going on, in their lives (Schkade and Kahneman, 1998; Wilson et al., 2000).

Economic models

Only recently have economists started to develop models of happiness. A good example is the model of Graham and Oswald (2010), which illustrates a unification of various sciences. Their model is based on evolutionary biology, and the assumption that happy individuals are more likely to breed successfully. Unhappy people tend to have higher stress levels and therefore compromised immune systems, and also are more likely to take greater risks, resulting in death. The variables and structure of the model take into account psychological factors such as dispositional traits and a decaying memory of past experiences, but it is analogous in a number of ways to a standard economic capital investment model involving a production function. The capital in this case is referred to as ‘hedonic capital’ rather than physical capital, and this naturally decreases over time unless people invest in it.

The production function in (11.1) shows that hedonic energy (y) in any time period is a function of dispositional traits towards happiness (z), hedonic capital (k) and the parameter α, which is positive but less than one, to reflect diminishing returns to hedonic capital. The variable v_t is a α life events in the relevant period:

(11.1)

Hedonic energy can then be used to either produce happiness or to invest in hedonic capital:

y_t = h_t +i_t

(11.2)

The stock of hedonic capital depreciates over time, but can be increased by investment:

k_t+1 = (1 – )k_t + i_t

(11.3)

where δ is the rate at which hedonic capital depreciates. Thus if people have a bad shock they can draw on their hedonic capital, i.e. disinvest, in order to smooth out this life event. Similarly a fortunate life event, like a promotion, could result in an investment in hedonic capital.

Like other economic models based on evolutionary biology, the situation is viewed as a principal-agent problem, with Nature being the principal and the individual being the agent. The virtue of this game-theoretic approach is that it incorporates the principle of natural selection as far as Nature is concerned, but does not allow Nature to program individuals’ brains completely for making every possible lifetime decision, since this would require a huge brain capacity, which would be very wasteful in terms of using precious energy resources. Thus in the principal-agent model Nature wishes to maximize biological fitness, meaning expected happiness, and determines a set of decision rules or policies in terms of hardwiring the brain; the individual, who is more informed about the specifics of a situation, then has some flexibility in making decisions according to these rules.

Graham and Oswald claim that this model both captures essential psychological traits and successfully predicts the tendency toward mean-reversion of happiness observed in many empirical studies, because of the model’s ‘smoothing’ characteristics.

11.6 Problems in pursuing happiness

The assumption that people act in such a way as to maximize their expected utility is the most fundamental neoclassical pillar underlying the standard model. However, because of the various problems already discussed relating to rationality, substantial doubts are cast regarding the validity of this model. Even if we extend the standard economic measure of welfare to include the hedonic aspects of happiness, there still remain problems with EUT as a normative theory. Schooler, Ariely and Loewenstein (2001) classify these problems into three categories:

1 Limits to hedonic introspection.

2 The adverse effects of hedonic introspection on wellbeing.

3 The self-defeating nature of happiness-seeking.

These problems will now be discussed in turn.

Limits to hedonic introspection

Hedonic introspection refers to the subjective measurement of one’s own happiness or utility by the individual. It might initially seem self-evident that the individual is in the best position to report their own state of happiness, and relying on self-reports is certainly the easiest way to measure happiness. Furthermore, studies indicate that such measurements have relatively stable qualities of validity and reliability. In order for a measure to possess validity it has to measure the variable that it is supposed to measure. Lyubomirsky and Lepper (1999) have shown that individuals’ self-reports of their overall happiness are reasonably well correlated with assessments made by friends and spouses. The quality of reliability means that repeated tests yield similar results. Again, correlations were quite high.

However, there are various problems related to measuring happiness by relying on self-reports, some of which have already been discussed in previous sections. The first point to note here is that there is a fundamental distinction between individuals’ continuous hedonic experience and their intermittent reflective appraisal (sometimes referred to as meta-awareness or meta-consciousness). Thus, although every waking moment involves a hedonic experience, registered as visceral feelings, we only consciously assess these experiences intermittently. We cannot spend our whole lives consciously asking ourselves how happy we feel; sometimes we only realize after the event that we were happy, or unhappy, during that period of time.

When we do intermittently reflect on our happiness state there is then the issue of how we infer it. This is not as simple as it may sound, for it is not like taking a pulse; as we have already seen, there is no single scale we can use here, for we can experience both happiness and sadness simultaneously. There appear to be two main influences here which may not be intuitively obvious. One, which again has already been discussed, relates to the self-perception theory attributed to Bem (1972). Bem’s central premise is that people often lack meta-awareness of their own internal states, with the result that they tend to infer these states, attitudes and preferences from their behavior. This process is subject to considerable error, in terms of misattribution. Not only do people make inferences from their behavior that may be in error, but they may also misattribute sources of visceral arousal. For example, a study by Zillman (1978) showed that arousal induced by exercise could be misattributed to anger, and a study by Dutton and Aaron (1974) showed that arousal induced by fear could be misattributed to sexual attraction.

The second influence on our self-report of happiness is the situational context. Research indicates that people’s valuation of experiences is strongly affected by prior questions they are asked. Such effects are known as anchoring effects, as we have seen. For example, a study by Strack, Martin and Stepper (1988) asked college students how many times they had gone out on a date in the last month, and then asked how happy they had been overall. Other students were asked the same questions but in the reverse order. For the first group who were asked about dates first, the correlation between the two responses was 0.66; for the second group who were asked about happiness first, the correlation was close to zero. Thus the happiness response for the first group was anchored to the question about dating, indicating that the students were inferring the dating question was a cue to their happiness.

In summary, it can be seen that the results of the hedonic introspection research in general shed considerable doubt on our abilities to measure our subjective happiness at all accurately.

Adverse effects of hedonic introspection

‘Ask yourself whether you are happy and you cease to be so’ is a famous quotation by John Stuart Mill (1873, p. 100). There are a variety of studies that indicate that hedonic introspection reduces the experience of happiness compared with situations where such introspection is absent. Some of these studies relate to specific hedonic experiences; some relate to happiness in general.

Taking the specific studies first, one body of research has exposed subjects to a series of painful stimuli of varying intensities, profiles and durations, comparing the reports of those subjects who gave frequent on-line appraisals with those who simply evaluated the whole experience in retrospect (Ariely, 1998; Ariely and Zauberman, 2000). The result was that the first group was less sensitive in general to various aspects of the experience.

Another study with similar implications was discussed earlier, in connection with self-deception, and relates to preferences for different strawberry jams (Wilson and Schooler, 1991). It appears that the requirement to reflect on judgment reduced the ability of the participants to evaluate the jams, by ‘muffling’ their hedonic experience. Wilson et al. (2000) found a similar result in a study examining the evaluation of the quality of relationships with a significant other. Rapid judgments, without time for reflection, gave more reliable results, showing the importance of ‘gut feeling’.

A final area of wellbeing worth mentioning in this context concerns humor. One study has shown that, when people were asked to reflect on why they found certain cartoons funny, they actually found them less funny (Cupchik and Leventhal, 1974). Thus there appear to be a variety of hedonic experiences where reflection spoils enjoyment.

Various reasons have been proposed to explain the above findings. First, introspection, by focusing on the self, automatically detracts from attention to the experience. Subtle features of the experience may therefore be overlooked, detracting from the hedonic appraisal. A second reason is that increased reflection and evaluation may cause the subject to consider features of the experience that were lacking in some way, thus also detracting from enjoyment.

Another group of studies has examined the overall happiness levels of different types of people. The general finding is that happy people tend to be less introspective (e.g. Veenhoven, 1988; Lyubomirsky and Lepper, 1999). Furthermore, unhappy people tend to be more self-conscious and ruminative (Musson and Alloy, 1988; Ingram, 1990). However, there is one main problem in interpreting these results; the relationships are correlational, which gives no indication of causation. It may well be that, rather than introspection causing unhappiness, unhappiness may cause introspection. Likewise, happy people may have no motive for introspection.

Self-defeating nature of happiness-seeking

A number of philosophers and writers have claimed that the pursuit of happiness is self-defeating. Again, an example can be found in John Stuart Mill (1873):

Those only are happy who have their minds fixed on some object other than their own happiness; on the happiness of others, on the improvement of mankind, even on some art or pursuit, followed not as a means, but as itself an ideal end. Aiming thus at something else, they find happiness by the way (p. 100).

Schooler, Ariely and Loewenstein (2003) propose three reasons for the self-defeating nature of seeking happiness:

1 People have faulty theories of happiness

In particular this relates to the phenomenon already discussed in connection with homeostasis and allostasis. People who are motivated by increasing their material wealth tend to underestimate the short-term nature of the increased happiness that results. Studies have shown that such people tend to be less happy than people with other goals, such as achieving psychological growth, having satisfying personal relationships and improving the world (Kasser and Ryan, 1993; 1996). Once again it should be realized that these studies are correlational, and do not prove causation. It may well be that unhappiness may motivate people to increase material wealth, and Kasser and Ryan (1996) have found that people experiencing troubled childhoods are more likely to pursue wealth as a primary goal than people with normal childhoods.

2 Loss of intrinsic value of activities

Considerable research indicates that when people perform activities for external reward, for example money, the activities lose their intrinsic appeal. It is therefore plausible that if people perform activities, like going to a concert, with the primary goal of achieving happiness, they may obtain less enjoyment than if they performed the same activities for their intrinsic value, meaning in this example enjoying the music.

3 Increased monitoring of happiness

It is likely that people pursuing happiness will monitor their happiness more frequently, and we have already seen that increased monitoring may well reduce happiness.

One approach to the empirical investigation of the issue of the self-defeating nature of happiness-seeking is to examine the relationship between selfishness and happiness, using selfishness as a proxy variable for intensity of happiness-seeking. A study by Konow (2000) revealed that people who were more selfish in a dictator game showed markedly lower levels of happiness than people who were more generous to team-mates.

Another empirical approach has been used by Schooler, Ariely and Loewenstein (2001). They performed an extensive study with 475 participants to examine their goals, plans and the realization of these goals for New Year’s Eve 2000. This study has particular value since it was not an experiment, and therefore the goals and plans of the participants were self-determined, not determined by an experimenter. One main conclusion of the study was that those who had made the most ambitious plans and devoted the most energy to their celebrations, presumably meaning those most concerned with the pursuit of happiness, were most likely to be disappointed.

The conclusion from all the studies on the pursuit of happiness therefore appears to be a paradox: if we explicitly pursue happiness and monitor the results we are not likely to achieve our goal. Yet, if we never evaluate our experiences, we can have no idea what kind of activities to pursue at all, and are also unlikely to achieve happiness. Thus it seems that we are damned if we do and damned if we don’t. Schooler, Ariely and Loewenstein, however, are reasonably optimistic. They point to evidence that our unconscious automatic psychological mechanisms may be highly effective at both pursuing goals and monitoring our effectiveness in achieving them (Wegner 1994; Bargh and Chartrand, 1999). In practice we may be like pilots who can fly most of the time on automatic pilot and only occasionally have to engage in manual control. As Schooler, Ariely and Loewenstein conclude: ‘the challenge is determining when it is best to man the controls, and when it is better to simply enjoy the ride’.

What is the significance of these three problems discussed above? It is possible to take the view that we can be rational from the point of view of the four criteria discussed in Chapter 3, in terms of attitudes and preferences adhering to logic and probability theory, being coherent, not being based on irrelevant factors, and not being incompatible with known empirical observations, but we may still not follow the norms of EUT. However, there is one final important point to make at the end of this section, which was also made in Chapter 5 in relation to prospect theory. Criteria for normativeness are not the same, or should not be the same, as criteria for descriptiveness. The fact that there are various problems associated with the pursuit of happiness does not mean that EUT is necessarily flawed from the point of view of predicting behavior, at either the individual level or at the aggregate level. It may not be a good model to follow in terms of its policy implications, but it may in principle still be a good predictor of actual behavior, which is what of course economists are largely concerned with. However, as we have seen in practice throughout the previous chapters, EUT often fails as a good descriptive model also.

11.7 Policy implications

This aspect has been discussed in many chapters, in connection with specific behavioral areas. Now that we have discussed rationality and happiness, it is useful to summarize some of the most important implications, as far as individuals, firms and governments are concerned.

Individuals

Four main aspects of importance will be summarized here, relating to (1) happiness; (2) emotions and memory; (3) inter-temporal conflicts in decision-making; and (4) game theory.

1 Happiness

We have seen that happiness is both difficult to define and measure, and difficult, if not self-defeating, to pursue. There are many different factors involved, and the implications of Pinker’s ‘three-act tragedy’ are of fundamental importance. In evolutionary terms we must remember that we did not evolve as happiness-maximizing beings, but rather instead use happiness for a proxy of how good our chances of survival and reproduction are at any point in time. So, in short, food and sex make us happy, and in that order according to some recent research. The fundamental problem that we find is that this proxy mechanism is easily hijacked and sent off the rails. This can be caused by three main factors: (1) our evolved mechanisms may no longer be suited to our current environment (causing us to like fatty and sugary foods); (2) bounded rationality (causing us to use easily-fooled heuristics); and (3) inter-temporal conflicts, discussed shortly. Evidence suggests that introverted and introspecting people tend to be less happy than extraverts. Another general finding is the paradox that the pursuit of happiness appears to be self-defeating, and it seems to be that we are at our happiest in retrospect when we are not thinking about our happiness. The problem with both observations lies in establishing causation: are we happier because we are not thinking about being happy, or are we not thinking about being happy because we are too busy actually being happy?

2 Emotions and memory

Again there are a number of implications here. The first involves durability bias. Currently experienced emotional or visceral states tend to influence decisions relating to the future, even when those states will not persist for long into the future. For example, we buy consumer durables like TV sets and furniture because we anticipate that these goods will bring us lasting pleasure. Durability bias suggests therefore that we tend to pay too much for such goods, since the resulting pleasure will not last as long as we anticipated. Similarly, negative states may influence behavior, so that when we are angry we may make elaborate plans on taking revenge, which turn out to be a waste of time after our anger has cooled off.

A second implication is that people underestimate the effect of future emotional or visceral states on future behavior. Thus if we fail to predict tomorrow’s pain of awakening we may fail to anticipate the need to place the alarm clock on the other side of the room, to prevent our switching it off immediately in the morning and going back to sleep. Even if we can predict the intensity of a visceral state, like the desire to smoke or eat chocolate, we may not be able to anticipate the effect on behavior (driving for an hour to find an open shop). This factor is also of importance in understanding the relapse behavior of addicts, who, after a period of abstinence, believe that they can indulge in low-level consumption (of alcohol, drugs, gambling) without relapsing.

Another implication is that people forget the influence that emotional or visceral states had on their past behavior. This can mean that they find past behavior influenced by such states to be increasingly perplexing as time goes on. For example, pregnant women can get food cravings very different from their normal tastes, and find it very difficult months later to understand such past cravings. People also tend to remember pain and fear badly, probably because it is difficult to re-experience, as opposed to merely recall, these states. Again, pregnant women tend to forget the pain of previous childbirth; they may initially decide to forgo anaesthesia, but when labor starts they often reverse this decision (Christensen-Szalanski, 1984). An important policy implication of this phenomenon is that the kind of scare tactics which are sometimes used in promotional programs to deter people from smoking, taking drugs or committing crimes, often prove ineffective over a sustained period, and sometimes even prove counter-productive (Finkenauer, 1982; Lewis, 1983).

3 Inter-temporal conflicts in decision-making

These conflicts were discussed in Chapter 8, as an important anomaly in the discounted utility model. According to the DUM there will be a time-consistency in preferences, but in reality time-inconsistent preferences are often observed. Continuing with our food example above, we may decide now that we will not have a tasty dessert after dinner at the restaurant tomorrow, but when the dessert trolley comes around we succumb to temptation. The problem is exacerbated by two other related tendencies that we have discussed: we tend to underestimate the strength of future temptations, and over-estimate our ability to overcome these temptations.

There may be many possible explanations for this behavior, and various alternative models to the DUM have been discussed, but where does this leave the individual in terms of optimizing their behavior or happiness? We have seen that the main policy implication here is that we need to make credible commitments. In the above example this may mean eating something healthy before going to the restaurant, designing specific rewards for ‘good’ behavior, only taking enough money for a main course, or even wearing tight clothes that make it uncomfortable to eat too much. The reader may suspect that the author has had plentiful experience of this particular problem!

Another area where inter-temporal conflicts frequently arise is saving habits (money and wealth can be viewed as a means of acquiring food and sex). Many people do not save enough during their working lives to afford a comfortable retirement. Again credible commitments are important in overcoming this problem. We have seen that the use of automatic savings devices (having amounts deducted from our salaries), and the use of illiquid savings accounts that are costly and inconvenient to withdraw from, may serve as effective commitments. This aspect of behavior is also a problem for public policy, as will be discussed later.

The other fundamental area where inter-temporal conflicts occur relates to sexual behavior. A high proportion of people have indulged in sexual relations that they have later regretted. Again, commitment devices are important. Marriage does not appear to be one of them, in terms of effectiveness, at least in many present societies; in the US about 44% of marriages currently end in divorce. One important point here is that sexual appetite is like appetite for food in at least one respect: it tends to build over time, until it is satisfied; it then falls to a low level and gradually builds again in a series of cycles. We have also seen that, as far as food is concerned, commitment often involves satiating the appetite before temptation takes hold; hence it may be best to go to the supermarket just after eating, not when one is hungry. A similar logic can be applied to developing ways of pre-empting sexual temptation. The main problem in this case is that sexual temptation is often less predictable in timing compared with food temptation.

4 Game theory

There have been numerous examples in Chapters 9 and 10 where an understanding of game theory can help people make better decisions. This is particularly true when our understanding is enriched by the incorporation of behavioral and psychological game theory. It is important to remember that many of the decisions we have to make in social situations are where we are interacting with the same people on a regular and repeated basis. The considerations here are different from those situations when we are interacting with strangers whom we may never interact with again.

Firms

Some of the behavioral aspects discussed above also have policy implications for firms.

Three main factors are discussed here: inter-temporal conflicts, loss-aversion and game theory.

1 Inter-temporal conflicts

If firms know that people succumb easily to temptations in terms of inter-temporal preferences, they may take advantage of this in a number of ways. We have already mentioned the wheeling around of dessert trolleys in restaurants, which plays on the urgency of visceral factors. The distinction between investment goods and leisure goods has also been seen to be important in terms of contract design. It is better for firms to charge below marginal cost for investment goods like health club memberships, and above marginal cost for leisure goods like mobile phone usage. As far as saving and spending is concerned, financial organizations can make attractive offers of loans, sometimes framing these in particular ways: ‘no money down’; ‘no payments for 18 months’; ‘pay all your debts with one low monthly payment’; and flexible repayment schemes that allow lower repayments when a borrower’s income falls, balanced by higher repayments when income rises. Some of these practices may be regarded as ethically questionable, of course, since they take advantage of human frailties and may cause unhappiness later. However, they are certainly frequently used.

2 Loss-aversion

This is another important behavioral concept that firms can profitably employ, especially when coupled with framing effects. We have seen, for example, that it may be better for firms to reduce prices by framing the reduction in terms of a discount. This means that if a firm later wants to raise prices, it can then frame this rise as an ending of the discount, causing less loss-aversion and adverse consumer reaction than a standard price rise. Another method of raising prices in a latent way and avoiding loss-aversion is to reduce the quantity offered for a given price. Consumers are less likely to notice that their cereal box is now 350 grams instead of 375 grams than if the price goes up from £1.99 to £2.13, although the effect is exactly the same in terms of price per unit weight.

3 Game theory

There are a number of lessons to be learned here, particularly as far as bargaining and negotiation is concerned. These situations can arise with employees, customers, and competitors. One example we have discussed is the use of a most-favored-customer-clause (MFCC), which has the twin advantages to the firm of reassuring customers regarding prices, and paradoxically serving as a signaling device to other firms that the firm intends to maintain higher prices. The appropriate use and interpretation of signaling devices is particularly important for firms in both competitive and cooperative environments. As another example, a firm may expand its capacity as a signal that it intends to produce more output in the future, discouraging entry into the market. In oligopolistic markets, direct communication between firms that may imply cooperation is illegal in many countries; for example, a number of private schools in the UK have recently been prosecuted for sharing information regarding costs, because it was suspected that this led to higher fees being charged. In this kind of situation firms can effectively use signals regarding price-setting, without needing any direct communications. Many economists suspect that the tobacco and airline industries use such signals effectively.

As far as relationships with employees are concerned, firms need to understand concepts of fairness in determining salaries and salary differentials. In many financial firms, where salaries can be extremely high, it is strictly against company policy for employees to reveal their individually negotiated salaries to other employees, for fear of causing resentment and envy. As we have seen, an employee may be delighted to learn that they have just been given a 10% raise, until they learn that a colleague has been given a 20% raise. Not only is the level of pay important, but also the structure of pay. We have seen in Chapter 10 that bonus contracts tend to be more efficient and result in greater welfare of both parties than either trust or incentive contracts. An understanding of social preferences, reference points and focal points, along with self-serving bias, is necessary for a firm to have satisfactory negotiations with employees.

Governments

The behavioral concepts and principles applied in the situations above are also applicable as far as government policy is concerned. We will start this time with perhaps the most obvious example, loss-aversion.

1 Loss-aversion

If a government wants to stay in power and retain votes, having a vital and growing economy is extremely helpful. In a recession people are losing jobs and income, and even a stagnant or slowing economy can have adverse electoral implications if conditions are worse than people had expected. The importance of economic fundamentals has been particularly important in many countries since 2007, when global financial crisis threatened many countries. Governments have faced recession, collapsing banking systems and record budget deficits, making policy decisions extremely difficult in view of the trade-offs involved. Many of these trade-offs are inter-temporal in nature, for example the issue whether to employ more expansionary fiscal policy now, to boost the economy and reduce employment, at the cost of greater deficits, risk of loss of confidence in the financial markets, and the necessity of harsher cutbacks in spending in the future. These inter-temporal conflicts are discussed further below.

2 Inter-temporal conflicts

Time-inconsistent preferences are important for governments for a number of reasons. First of all many people would charge the government with the duty of helping to increase their state of happiness. How far a government should interfere with the freedom of individuals in order to further their own interests is of course a controversial topic. We are not going to consider the arguments of parochial government versus ‘nanny’ state here. However, there are some areas of policy where governments can help people overcome their tendencies to succumb to temptation resulting in regrets later that are relatively uncontroversial. One broad area concerns pensions and saving for retirement. In view of the well-established tendency for people to save insufficient funds for a comfortable retirement, there are various routes that governments can take to reduce this problem. One method, used in more socialist countries, is simply to collect higher tax revenues in order to pay a moderate government pension to everyone. This creates a problem of moral hazard, in that people lose the incentive to save, relying solely on government-provided pensions. Many countries with developed economies and ageing populations are finding it increasingly difficult to fund pensions in this way. Some of the more market-oriented of these economies, like the UK, are trying to provide incentives for people to take out sufficient private pensions on a voluntary basis. In order to encourage participation it is better to frame such participation as the default option, rather than having people have to actively opt into such a scheme. Another means of encouraging saving for retirement is to use the tax system, by offering tax breaks on retirement accounts like IRAs in the US.

As far as the effects of paying pensions on government budgets are concerned, one possible way of reducing the burden is to offer people the option of receiving a lump sum rather than an ongoing payment. We have seen evidence that, in the case of public employees, this may save a considerable sum for the state, as people tend to opt for lump sum payments in spite of the high implicit discount rate. Of course, if this practice were to be implemented on a large scale for all government-provided pensions, the savings involved would have to be balanced against the likelihood of hardship later for those people with time-inconsistent preferences, who fritter away their lump sums without planning for the future.

There is one other area of government policy where time-inconsistent preferences are of particular importance. This concerns addiction. The term addiction in this context is applied in the broadest sense to include not just recreational drugs, with which it is most commonly associated, but also alcohol, tobacco, gambling and even food. We have seen that visceral influences play a large role here. Addiction not only causes unhappiness and suffering for addicts themselves, but also imposes heavy externalities. These relate to the effects on the families of addicts, the resulting increase in crime, and additional costs imposed on the state in terms of care and rehabilitation. Tackling the causes of addiction is a formidable task. In particular it involves a unified approach involving all the behavioral sciences. First, there is a sociocultural approach: for example, examining the causes of the drug culture among the young, which appears particularly prevalent in the UK. It also involves a psychological approach: researching the factors that cause particular individuals to be vulnerable to different types of addiction is important here. Finally, it involves a neurophysiological approach: this may also help to identify those people most vulnerable, but it should also lead to the development of medical treatments and drugs which counter the effects of addiction.

An important policy implication here, discussed in Chapter 8, is that making the production and distribution of addictive substances illegal may be counterproductive. This is because it increases transactions costs for suppliers, encouraging larger transactions and greater consumption. High excise taxes may be a more effective method of discouraging consumption.

3 Game theory

Once again we see that game theory is a universal factor in decision-making. Again also the incorporation of behavioral game theory elements is important for government policy. There are a number of policy areas that have possible applications; five of these are discussed here, but this is not intended as an exhaustive list:

a) Auctions

Governments and government agencies are sometimes in a position to sell business operations or licences to businesses to operate in certain areas. Large sums of money are usually involved, so it is important for governments to raise a maximum amount. The amount raised is highly sensitive to the method used for the sale. The various privatizations by the UK government in the 1980s and 1990s did not manage this efficiently. The general method was to invite the public to tender for a number of shares at a fixed price. In some cases the share offer was oversubscribed by a factor of four, indicating that the government, and its investment bank advisers, had considerably underestimated the market value of the operations being sold. When shares were later sold on the secondary market at much higher prices than the par value, often within days of the original issue, huge profits were made by speculators at the expense of the government.

An alternative approach is to auction operations to the highest bidder or bidders. An example of a successful auction relates to the sale of radio licenses by the Federal Communications Commission (FCC) in the US, in 1994. The FCC took the advice of game theorists, notably Binmore, in establishing the rules for the auction’s structure. These were complicated, following a simultaneous, multiple-round format. The end result was that, after 46 rounds of bidding over a five-day period, ten nationwide licences were sold for a total of over $600 million. This sum was over ten times the highest estimates that had appeared in the press.

b) International trade

Governments frequently have to negotiate trade deals on a bilateral or multi-lateral basis. They also often unilaterally determine import tariffs and export subsidies, as well as other less obvious trade barriers. They may even appear to take actions that harm their own country, like imposing voluntary export restraints (VERs). The effects of all of these actions are generally counter-intuitive: tariffs and subsidies often reduce overall welfare within the country imposing them, while VERs may improve welfare (at least compared with the likely alternatives). This is not the context to explore the effects of these policy measures in detail, but frequently a major problem is a lack of game-theoretic analysis: how are other governments and other firms likely to react?

This failure is particularly evident in the imposition of trade sanctions. As we have seen, when these are imposed on poor countries ruled by dictators, they often backfire, causing unnecessary hardship and bolstering the regime at which they are aimed. The problem is again a lack of iterative thought.

c) Environmental protection

This is essentially an aspect of the tragedy of the commons problem, which is increasingly affecting the world economy as a whole. Overfishing is a prime example. However, environmental protection, related to reducing greenhouse gases and climate change, has become a major, some would say the major issue, in the world economy. This is not just because of the catastrophic consequences that could follow current trends in greenhouse emissions, but also because of the huge political problems in dealing with the situation. These problems relate largely to the temporal and spatial aspects involved. Temporally, the current behavior of economic agents has no significant effect on their welfare because of the long time lags, extending into decades. This aspect is aggravated by the spatial aspects, that, because of winds and currents, one agent’s pollution often turns out to be their neighbors’ problem. National governments, even if they are concerned with environmental issues, are keen to ensure that their country is not overburdened with the costs of protection while other countries get a free ride. However, unlike other public goods games, it is difficult to ensure cooperation because of problems in monitoring behavior and enforcing punishment.

d) Counter-terrorism

Another major international issue relates to the so-called ‘war on terror’. Over the last decade various western governments have reduced liberties in a number of ways in order to combat terrorism. The costs of this involve not only reduced liberties, but also increased transactions costs for individuals and firms (for example going through airport security), and of course administration costs. The benefits include not only reduced terrorist activity, but also the perception of greater safety. Evidence suggests that there is a nonlinear relationship between counter-terrorism spending and terrorist activity, featuring diminishing returns (Arin et al., 2011), at least in the UK. However, whether counter-terrorism spending reduces the efforts of terrorists, or whether it just makes these efforts easier to detect is still an open issue.

e) Mediation

Bargaining impasses are often caused by self-serving bias and multiple focal points. However, we have seen that one way to resolve these problems is to have some form of mediation where the bargaining parties can have the weaknesses of their positions explained. Governments, or their agencies, can serve as independent arbiters in industrial relations or employment disputes in particular. An example of such an agency is the Advisory, Conciliation and Arbitration Service (ACAS) in the UK. There is some evidence that the existence of this organization has helped to reduce the number and length of industrial disputes over the last two decades.

Thus we can see that in some situations the government is acting as a player that is either trying to cooperate with or to compete with other players. In other cases the government may be acting as a mediator, which resembles the role of the ‘choreographer’, in trying to achieve a correlated equilibrium that is superior to other outcomes in terms of the welfare of the other players.

11.8 Future directions for behavioral economics

There have been various indications throughout the book where either areas of research or types of research are necessary in order to take the discipline forward. Appropriate topics to discuss at this stage are: decision-making heuristics; the formation of social preferences; learning processes; the theory of mental representations; the role of the emotions in decision-making; and the role of neurobiology.

Decision-making heuristics

We have seen, particularly in the early chapters, that people frequently use heuristics or simplifying rules when they make decisions, and we have seen that the root cause of this is the existence of bounded rationality. These heuristic devices are of wide variety, including the representativeness heuristic, the availability heuristic, the 1/n principle, fuzzy trace theory, and the procedural approach. It is not intended here to review all the heuristics discussed, and their shortcomings. However, there is one area of heuristics that has not been discussed so far. This relates to the case-based decision-making approach. Whereas the standard model examines future outcomes and their associated probabilities, the case-based approach is based on assessing the similarity of the current situation with various past decision situations, evaluating average outcomes from particular actions in these past situations, and weighting them by the similarity of previous cases to the current one (Gilboa and Schmeidler, 1995; 2001). The approach is therefore based on the common observation that we all use past experience in some way as a guide to future experience. However, the choice of relevant criteria for judging similarity, and assessment using these criteria, obviously involves a heuristic process. This is best illustrated by an example. Say that we are offered a new job and are deciding whether to take it. We may have been in a similar situation in the past on several occasions. What outcomes were important? These may include salary, or salary relative to one’s reference group; working hours; amount of time traveling; working conditions; type of boss; quality of working environment, and so on. If we took a job based just on salary in the past, and this proved to be a bad experience, we may discount this criterion in our current decision.

The formation of social preferences

We have reviewed various models incorporating social preferences in Chapter 10, and have also seen that these can explain behavior in both real-life situations and experimental studies better than the simplistic ‘pure’ self-interest model of the standard model. However, more research needs to be carried out on the underlying psychology involved and also the development of social norms. For example, it would be useful to know the relative importance of negative and positive reciprocity, and the relative importance of both of these compared with inequality-aversion. The importance of intentions also needs to be studied further, which means taking into account foregone payoffs. These can be studied by various methods of experimental psychology, and the results incorporated into modified social preference models. These can be tested by using further game-theoretical experiments in an ongoing process of refinement and increased understanding. We have also seen that the development of social norms is a complex phenomenon, relying on the process of ‘team reasoning’, but this area needs much further research, in particular to understand why social norms vary between different social groups and over time. This increased understanding of social preferences may have significant benefits for public policy, as will be discussed shortly.

Learning processes

These processes were examined in Chapter 9. As with social preference models (and many other economic models), better models, in terms of goodness-of-fit and prediction, are likely to be more complex, taking into account all of the elements of information that people seem to use in the learning process. Once again, foregone payoffs, both one’s own and those of other players, are likely to be of importance here in improving our understanding. Learning strategies and processes such as randomization is another area where further research is needed. This research also has important implications for equilibrium analysis. Most conventional and nonconventional approaches assume some kind of equilibrium in their analysis, but, if learning takes a long time, a non-equilibrium type of analysis may be more appropriate. A wider range of experimental situations may also help understanding. Most of the game theory models used to test learning have involved cooperative and competitive situations in a highly stylized manner. Different types of experiment may be required if we are to understand better how children learn language, for example, and also how child learning is different from adult learning. This last aspect is considered further in the last area of direction, neurobiology.

The theory of mental representations

This is a relatively new area in psychology, and is concerned with how people form mental models or perceptions of a situation, in particular the elements in a game situation. In the words of Camerer (2003):

The theory of mental representations maps raw descriptions of social situations into the kinds of familiar games theorists study and the kinds of rules people use to decide what to choose (p. 475).

This theory is currently not well-developed, but it has certain elements in common with the equally new area of case-based decision-making, discussed earlier.

The relevance and importance of developing theory in this area can be illustrated by an example from Camerer (2003). One of his student subjects commented after an experimental session playing the stag hunt game: ‘I can’t believe you guys are still studying the prisoners’ dilemma!’ (p. 474). The confusion of these two games is interesting. The common element is that rationality can lead to inefficiency, in terms of not resulting in a Pareto-dominant outcome. However, the ‘uppity, trigger-happy’ student had ignored that in stag hunt, unlike PD, there are two Nash equilibria, and one of them is Pareto-dominant, while in PD, unlike stag hunt, there is a single dominant strategy equilibrium. His confusion led him to ‘defect’ in each round of play, a rational solution in PD, but not in stag hunt. The student had failed to undertake an essential basic task in any game situation: examine payoffs to see if there are any dominated strategies. The performance of this task would have differentiated between the two types of game.

There is insufficient evidence at present to draw any general conclusions regarding the extent of the type of confusion above, or other types of error in mental representation. It would be interesting to see whether subjects make the same errors when such games are presented in more realistic situations rather than abstract ones. A prediction of evolutionary psychology would be that errors would be substantially reduced in situations that resembled real life, in the same manner as the Wason Test variations. This would make a good topic for a PhD dissertation for any enterprising reader!

The role of the emotions in decision-making

This subject was discussed in Chapters 3 and 8 in relation to preferences and in particular time-inconsistent preferences. We also saw in Chapter 4 that emotions may also play a role in determining our beliefs and how we update them in accordance with new information. Thus the role of emotions is a topic that is related to the more general area of decision-making heuristics discussed earlier, and to the role of neurobiology, discussed next. As far as the issue of whether the emotions have a beneficial or detrimental effect on decision-making, the evidence is mixed. Traditionally, since at least the days of Descartes, rationality, in the sense of the absence of emotion, has been regarded as being optimal. However, since the late 1980s, various researchers in economics, notably Frank and Hirshleifer, have suggested that emotions are an evolutionary adaptation and serve an important beneficial role, in that they serve as credible commitments. This theory is also supported by evidence in neurobiology, in particular the work of Damasio (1994). The issue is discussed in more detail in the first case study at the end of the chapter, in the context of patients suffering brain damage.

The role of neurobiology

No discussion of future directions for behavioral economics would be complete without a reference to the role of neurobiology. This aspect has been touched on at various points in the book, with the importance of neuroscientific evidence being discussed in the contexts of mental accounting, intertemporal choice and social preferences. Many social scientists are wary of the encroachment of neurobiology on their disciplines, being suspicious of a reductionist agenda. However, as was made clear in the first chapter, there have been a number of recent technological developments, particularly in terms of brain scanning and imaging techniques like PET (positron emission tomography), fMRI (functional magnetic resonance imaging), EEG (electroencephalography) and rCBF (regional cerebral blood flow). These can shed considerable light on various topics of interest in behavioral economics, and are especially relevant in decision-making heuristics, learning processes and the role of the emotions. We are finding that different types of thinking or mental process are performed in different parts of the brain, indicating the importance of brain structure or anatomy. We are also finding that different chemicals and hormones have dramatic effects on behavior, although of course the effects of alcohol on behavior and decision-making have been known in general terms for thousands of years. The proportion of the populations in developed countries who are now taking regular prescription drugs, for example, Prozac and Ritalin, to improve mental performance has increased enormously over the last two decades. This has in fact become an important public policy issue, in view of recent evidence that such drugs can have long-term harmful effects. Thus the importance of brain physiology is becoming better recognized.

Another controversial area, relating to the nature of free will, concerns the role of genetic factors. Ongoing research in this area is continually finding links between genetic makeup and behavior. Examples include the tendency to commit various criminal acts, tendencies to addictive behavior, tendencies for thrill-seeking, tendencies to certain diseases, various personality traits and disorders like autism, and of course intelligence. The anti-reductionist lobby regards this trend towards genetic explanations as distasteful, or politically incorrect, labeling the findings as ‘genetic determinism’, but there is no doubt that the trend is gathering scientific momentum. As with evolutionary psychology, one has to ward against the ‘straw man’ accusations. The label ‘genetic determinism’ is such a straw man, since it implies an extreme view that behavior is completely controlled by genetic factors, with no environmental influences (and this is ignoring the interactions between genetics and environment, such as developments in the womb). What the scientific findings usually imply is that genetic factors have a predisposing influence, making certain behavioral effects more likely.

11.9 Applications

Case 11.1 The effects of brain damage on decision-making

The following story might be called ‘the curious case of Phineas Gage’. The year was 1848. Gage, 25 years old, was a construction foreman involved in building a railway in Vermont. It was customary in the course of construction to use explosives to blast and clear away rock, in order to build a straighter and more level track. Gage was an expert in such a task, and was described by his bosses as ‘the most efficient and capable’ man in their employ. However, in the course of tamping an explosive charge, there was an accident, and the ensuing explosion caused the iron rod used to tamp the charge to be blown right through Gage’s head. The rod entered through his left cheek, pierced the base of his skull, traversed the front of his brain and exited through the top of his head. The rod measured just over a meter in length, had a diameter just over three centimeters, and weighed about six kilograms. It was tapered, and it was the pointed end that entered first, or the story may well have been different.

We do not need to dwell on the gory details, but some description of the incident is necessary in order to understand its implications. First of all, Gage not only survived the accident but he also made a remarkable recovery, returning in terms of outward appearances to close to his previous self. Indeed, he was talking within a few minutes of the accident, and, having been carried in an ox cart about a kilometer to get medical assistance, he was able to get out of the cart himself, with only a little help from his men. He was then able to explain ‘perfectly rationally’ to the doctor the nature and circumstances of the accident. Gage was pronounced cured in less than two months. According to his doctor, Gage’s physical recovery was complete. He could touch, hear, and see, and was not paralyzed ‘of limb or tongue’. He had lost the sight in his left eye, but his vision was perfect in the right. He walked firmly, used his hands with dexterity, and had no noticeable difficulty with speech or language.

However, Gage’s unfortunate story was only just beginning. According to his doctor ‘the equilibrium or balance, so to speak, between his intellectual faculty and animal propensities’ had been destroyed. He was now ‘fitful, irreverent, indulging at times in the grossest profanity which was not previously his custom, manifesting but little deference for his fellows, impatient of restraint or advice when it conflicts with his desires, at times pertinaciously obstinate, yet capricious and vacillating, devising many plans of future operation, which are no sooner arranged than they are abandoned … A child in his intellectual capacity and manifestations, he has the animal passions of a strong man’ (p. 8).

This description marks a dramatic change in Gage’s personality. Before the accident he was reported to have ‘temperate habits’ and ‘considerable energy of character’ (p. 8). He had possessed ‘a well balanced mind and was looked upon by those who knew him as a shrewd, smart businessman, very energetic and persistent in executing all his plans of action’ (p. 8). In short, as his friends and acquaintances noted, ‘Gage is no longer Gage’. After the accident his employers were not willing to take him back because of the changes in him, and he drifted from job to job, unable to secure and maintain the type of steady, remunerative job that he had once held. He died at the age of 38, after a series of epileptic fits.

The reason for telling this harrowing tale is that it illustrates a vital point. In the words of Damasio (1994), there are:

… systems in the human brain dedicated more to reasoning than to anything else, and in particular to the personal and social dimensions of reasoning (p. 10).

After exhuming Gage’s body, his skull has been examined using the latest neuroimaging techniques, and it is now possible to draw certain conclusions regarding the extent of his brain damage. It was concentrated in the prefrontal cortices, mainly in the ventromedial sector. According to Damasio, this damage ‘compromised his ability to plan for the future, to conduct himself according to the social rules he previously had learned, and to decide on the course of action that ultimately would be most advantageous to his survival’ (p. 33).

One can only draw limited conclusions from a single medical case study. It was obvious that certain parts of the brain had specialized functions, and could function independently of others, but more detailed conclusions are only possible by examining other similar cases. Damasio, as a professor of neurology and a practitioner, has been in an ideal environment in which to encounter such cases. He describes a modern equivalent, whom he refers to as ‘Elliot’, who suffered similar damage after a surgical operation to remove a brain tumor. The damage was again largely in the ventromedial sector of the prefrontal cortices, the main difference from Gage being that the damage was more on the right side. Like Gage, Elliot’s physical capacities were not affected, nor were most of his intellectual faculties, including language and memory. His IQ was tested and showed no impairment, being in the superior range. He appeared charming, knowledgeable and with a laconic sense of humor. However, after the operation, his personality and judgment were severely affected. He became easily distracted, engrossing himself in the detail of trivial or irrelevant tasks, and became incapable of maintaining a normal schedule, time management, or forward planning. He lost his job, and various others. Simple and basic decisions, like whether to take an umbrella when going out, would take ages to make, as he would tediously weigh up all the pros and cons. He made a string of consistently bad business, financial and personal decisions, and proved incapable of learning from his past mistakes. Most important of all, from Damasio’s point of view as a diagnostic, Elliot appeared emotionally flat, describing his fall from grace and disastrous personal experiences with a wry and detached tone.

It was this last characteristic that provided Damasio with the material from which he developed a new theory, which he refers to as the ‘somatic marker hypothesis’. Essentially, this states that the emotions are necessary in order for us to make good decisions, not just in a social environment, but in any real-life setting. The somatic markers act as a decision-making heuristic, enabling us to narrow down possible strategies from an initially huge range of possibilities. Without this heuristic we tend, like Elliot, to waste a lot of time evaluating courses of action that are simply not worthy of consideration. For example, if we are faced with a decision whether to take a new and highly paid job in a different country, we may immediately be overcome by the emotion that we cannot leave our friends or relatives behind. This somatic marker prevents us from considering and evaluating all the other factors related to the decision, and we immediately make the decision to refuse the position.

This hypothesis is in direct contrast to the belief of Descartes, that we need to use reason alone, and not the emotions, if we are to make the best decisions. Hence the title of Damasio’s book: Descartes’ Error.

Questions

1 What do the cases of Gage and Elliot tell us about the nature of rationality?

2 What does this case study reveal regarding the relationship between the brain and the mind?

3 Behavioral economics seeks a sound psychological foundation; what then is the relevance of neuroscience?

Case 11.2 Pursuing happiness

One of the most striking things about the document that Americans celebrate with such gusto on July 4 is that so much of it is dull – hardly worthy of the tons of fireworks and barbecue that are sacrificed in its honour. There are lists of complaints about the administration of the courts and the quartering of British troops. There is an angry passage about King George’s habit of summoning legislators ‘at places unusual, uncomfortable, and distant from the depository of their public records’. But all this tedium is more than made up for by a single sentence – the one about ‘life, liberty and the pursuit of happiness’ (The Economist, Pursuing Happiness, June 29, 2006).

The sentence was remarkable at the time – a perfect summary, in a few pithy words, of exactly what was new about the new republic. Previous countries had been based on common traditions and a collective identity. Previous statesmen had been exercised by things like the common good and public virtue (which usually meant making sure that people played their allotted roles in the divinely established order). The Founding Fathers were the first politicians to produce the explosive combination of individual rights and the pursuit of happiness. It remains equally remarkable today, still the best statement, 230 years after it was written, of what makes America American. The Book of Job gives warning that ‘man is born unto trouble, as the sparks fly upward’. Americans, for all their overt religiosity, have dedicated their civilization to proving Job wrong.

Everywhere you look in contemporary America you see a people engaged in that pursuit. You can see it in work habits. Americans not only work harder than most Europeans (they work an average of 1731 hours a year compared with an average of 1440 for Germans). They also endure lengthy commutes (who cares about a couple of hours a day in a car when you have a McMansion to come home to?). You can see it in geographical mobility. About 40 million of them move house every year. They are remarkably willing to travel huge distances in pursuit of everything from bowling conventions to factory outlets. You can see it in religion: Americans relentlessly shop around for the church that most suits their spiritual needs. And you can see it in the country’s general hopefulness: two-thirds of Americans are optimistic about the future.

Since Americans are energetic even in deconstructing their own founding principles, there is no shortage of people who have taken exception to the happiness pursuit. They range from conservatives such as Robert Bork, who think the phrase encapsulates the ‘emptiness at the heart of American ideology’, to liberals who think that it is a justification for an acquisitive society.

One criticism is that the pursuit is self-defeating. The more you pursue the illusion of happiness the more you sacrifice the real thing. The flip side of relentless mobility is turmoil and angst, broken marriages and unhappy children. Americans have less job security than ever before. They even report having fewer close friends than a couple of decades ago. And international studies of happiness suggest that people in certain poor countries, for instance Nigeria and Mexico, are apparently happier than people in the US.

Another criticism is that Americans have confused happiness with material possessions (it is notable that Thomas Jefferson’s call echoes Adam Smith’s phrase about ‘life, liberty and the pursuit of property’). Do all those pairs of Manolo Blahnik shoes really make you happy? Or are they just a compensation for empty lives à la Sex in the City?

If opinion polls on such matters mean anything – and that is dubious – they suggest that both these criticisms are flawed. A 2006 Pew Research Centre study, ‘Are we happy yet?’ claims that 84% of Americans are either ‘very happy’ (34%) or ‘pretty happy’ (50%). The Harris Poll’s 2004 ‘feel good index’ found that 95% are pleased with their homes and 91% are pleased with their social lives. The Pew polls show that money does indeed go some way towards buying happiness: nearly half (49%) of Americans with annual incomes of more than $100,000 say they are very happy compared with just 24% of people with incomes of $30,000 or less. They also suggest that Americans’ religiosity makes them happier still: 43% of Americans who attend religious services once a week or more report being very happy compared with 31% who attend once a month or less and 26% of people who attend seldom or never.

Weep, and you weep alone

The pursuit of happiness explains all sorts of peculiarities of American life: from the $700m that is spent on self-help books every year to the irritating dinner guests who will not stop looking at their BlackBerries. It also holds a clue to understanding American politics. Perhaps the biggest reason why the Republicans have proved so successful in recent years is that they have established a huge ‘happiness gap’. Some 45% of Republicans report being ‘very happy’ compared with just 30% of Democrats. The Democrats may be right to give warning of global warming and other disasters. But are they right to give the impression that they relish all the misery? The people’s party will never regain its momentum unless it learns to relate to the guy on the super-sized patio, happily grilling his hamburgers and displaying his American flag.

The pursuit of happiness may even help to explain the surge of anti-Americanism. Many people dislike the US because of its failure to live up to its stated ideals. But others dislike it precisely because it is doing exactly what Jefferson intended. For some Europeans, the pursuit of happiness in the form of monster cars and mansions is objectionable on every possible ground, from aesthetic to ecological. You cannot pursue happiness with such conspicuous enthusiasm without making quite a lot of people around the world rather unhappy.

Questions

1 Discuss the proposition that the pursuit of happiness is self-defeating, commenting on relevant empirical evidence.

2 ‘For some Europeans, the pursuit of happiness in the form of monster cars and mansions is objectionable on every possible ground.’ Explain this statement, commenting on differences between the American and European viewpoints.

3 ‘Religiosity appears to make people happier; therefore I should become religious.’ Discuss.

Case 11.3 The bioeconomic causes of war

The title of this case comes from an article by Hirshleifer (1998). His article relates to many of the concepts discussed in the book, although we have rarely mentioned the phenomenon of warfare. The case study is included here to indicate the broad range of phenomena that the tools of behavioral economics are able to analyze. Thus we will be able to see the relevance of bargaining, game theory, self-serving bias and focal points, social preferences, reciprocity and evolutionary psychology.

Hirshleifer states that the premise of bioeconomics is that:

Our preferences have themselves evolved to serve economic functions in a very broad sense: those preferences were selected that promoted survival in a world of scarcity and competition (pp. 26–7).

The most fundamental drives in both non-human animals and humans concern food and sex. Most conflict between animals is on an individual basis, and is directly related to these objectives. When male animals fight over territory, territory represents resources of food, and also females are often only attracted to males who possess a territory. Territory always represents more access to females. Chimpanzees are the only non-human animals known to fight in groups (although dolphins may also do so), and the objective of the kinds of raid that they carry out into other territories is always to gain access to females.

It may be claimed that humans have the dubious distinction of being the only species that conducts organized warfare. However, in human warfare between primitive tribes the objectives in fighting are usually again food and sex. Chagnon (1983) studied the warlike Yanomamo tribe at length and concluded that female capture was the primary motive. He also noted that Yanomamo men who have killed more enemies in battle also produce more offspring. Keeley (1996) noted that wars were often caused by multiple factors, and estimated that, among American Indian societies, material motives (land, booty, poaching and slaves) contributed to 70% of wars, while women contributed to 58%.

In much of human history the same appears to have been true. Genghis Khan was certainly more concerned with access to women than with access to material resources. It is a tribute to his phenomenal success in this regard that it has been estimated that one in twenty Asians can now claim descent from him. In many ancient wars the vanquished males were universally slaughtered, while females became slaves/ concubines/wives.

Food and sex may have been the dominant issues, but why does war break out? War represents the ultimate breakdown of a bargaining process. An individual or group of individuals wants something (food or sex) possessed by another individual or group, and they cannot agree to the terms of a transaction. Of course, in this impasse situation one party can just walk away dissatisfied, and this frequently occurs in the animal kingdom when one animal backs down from a confrontation. So why do animals and humans fight? For physical combat to occur both parties must evaluate the benefits from fighting to be greater than the costs. If only one party does so but the other does not, then the second party will back down, and combat is avoided. One method animals have of avoiding costly combat is to have some display of arms first; the purpose, unconscious though it may be, is to increase both parties’ information regarding the strengths and weaknesses of the opponent. If both parties have perfect information, and encode it correctly, no combat can occur. Combat therefore tends to occur when the two combatants are pretty evenly matched, as well as the stakes being high in order to justify the potential high costs.

A factor that is therefore clearly relevant in the outbreak of human warfare is over-confidence, or self-serving bias, which in this case may not be so self-serving, since it may lead to costly defeat or death. Many wars have been fought and lost through underestimating either the strength of the enemy or the costs involved in fighting in terms of resources. Certainly this accusation can be leveled at both Germany and Japan in World War II.

Historically, most wars were fought between groups of people who were related in terms of kinship. The same is still true of tribal wars today. As civilizations grew in size, often due to returns to scale from industrialized production and increased productivity, kinship was greatly diluted and the related concept of affiliation became important. Many social scientists have noted that, even when assignment to different groups is purely arbitrary, bonds of affiliation are soon formed (Sherif and Sherif, 1964). Experiments dividing subjects into ‘prisoners’ and ‘guards’ have had frightening consequences, as the ‘guards’ come to regard the prisoners as inferior beings to be punished and humiliated. An essential part of our human psychological makeup is an ‘us and them’ mentality, which evolved from kinship ties.

Affiliation can occur on the basis of many kinds of tie: class, interest, age and location are common. Nationality, ethnicity and religion are particularly important in causing strong bonds – bonds strong enough to kill and risk being killed for. Diamond (1997) argues that a key factor in the formation of large warlike groups has been religion. He defines religion as an ideology that manipulates group members to become peaceful and obedient internally, and suicidally brave when it comes to external warfare. It is also notable that in societies where affiliation is particularly strong, the members often use kinship terms, like ‘brother’ and ‘sister’, in referring to other members; the leader of the group may be referred to as ‘father’ or ‘mother’.

As affiliation has become more important and kinship less so, the issues of food and sex appear to have disappeared into the background. Material welfare is still important, although this is often referred to in abstract terms. For example, in 1914 German war aims were linked to finding its ‘place in the sun’. In the 1930s, the war movement was linked to a similar concept, lebensraum, but by this time there was a strong feeling of aggrievement, particularly related to racial hatred. As the importance of food and sex has declined in explicit terms, the concepts of dominance, prestige and honor appear to have taken their place.

In the twentieth century in particular, it seemed that biological motives were no longer relevant in warfare. The historian Ferguson (1998) has claimed that in World War I the main motive for ordinary men fighting was that war was regarded as sport, as fun. Certainly there is a primitive, visceral thrill of the hunt involved, often evoked by soldiers’ writings of the time. However, it might well be that this was a minority view, considering the massive slaughter, appalling conditions, and futility of the fighting. Writers and poets like Graves, Sassoon and Owen were certainly more affected by the latter factors.

Ferguson further contends:

Today’s neo-Darwinian genetic determinism may be more scientifically respectable than Freud’s mixture of psychoanalysis and amateur anthropology, but the latter seems better able to explain the readiness of millions of men to spend four and a quarter years killing and being killed. (It is certainly hard to see how the deaths of so many men who had not married and fathered children could possibly have served the interests of Dawkins’s ‘selfish genes’.) (p. 358)

This passage displays a misunderstanding of ‘genetic determinism’ and the operations of ‘selfish genes’. As explained in the previous chapter, genes do not operate in the teleological manner suggested by Ferguson, but in a purely mechanistic way. We have seen that this is the main reason why they can so easily be fooled or hijacked to serve irrational and self-destructive ends.

Hirshleifer (1998) on the other hand claims that one main factor why sex has become less important as an issue in war is that war leaders have far greater internal opportunities than before; they do not need to invade other countries in order to obtain access to females. Maybe this is why President Clinton never declared war.

Questions

1 Explain the fundamental flaw in the argument by Ferguson: ‘It is certainly hard to see how the deaths of so many men who had not married and fathered children could possibly have served the interests of Dawkins’s ‘‘selfish genes’’.’

2 Explain why self-serving bias is important in causing war.

3 How have the aims of war changed over time?

Case 11.4 Getting children to eat vegetables

Obesity and health-related problems like heart disease and diabetes are increasing in many high-income countries. Furthermore the incidence of these problems appears to be increasing fastest among school-age children. The costs to national economies are substantial in terms of decreased productivity and increased health care costs. Health experts are fairly unanimous in laying the blame on bad diet and lack of exercise. The problem appears to be greatest among low-income households, either because of lack of knowledge or because of the high perceived cost of buying healthy foods such as fruit and vegetables.

A couple of experiments have been carried out in a study by Just and Price (2011) in order to examine the effects of incentives on the consumption of fruit and vegetables.

Objectives of the study

There were three main objectives involved:

1 To examine and compare the effects of different types of incentive.

2 To examine and compare the effects on children with different parental incomes. 3 To examine the cost-effectiveness of an incentive program.

Experimental design

These experiments involved providing a variety of different incentives in 15 different schools (in the first experiment) in order to see how the consumption of fruit and vegetables at school lunches would respond. At some of these schools as many as 77% of the pupils received free or reduced price lunches because of low parental income, while in other schools the proportion was only 17%. Thus it was possible to see how parental income affected increased consumption. Three different variations of types of incentive were used: amount of cash reward (5 cents and 25 cents per day); cash and prize; and immediate reward and delayed reward. This allowed the researchers to use six different treatments: (1) large immediate cash reward, (2) large delayed cash reward, (3) immediate prize, (4) delayed prize, (5) immediate small cash reward, and (6) no incentive (as a control). The delay in each case was that the reward was received at the end of the month.

The prize involved participation in a raffle, where the total value of the prizes was equal to the number of kids who ate their fruits or vegetables for that day multiplied by 25 cents. Many of the prizes were related to some form of active recreation such as rip-sticks, tennis rackets, soccer balls and swim goggles.

The treatment days occurred over five lunch periods spanning 2–3 weeks. On each treatment day there was a message in the morning announcements about the reward students could receive by eating a serving of fruit or vegetables that day. Prizes were displayed near where data were collected, visible to all students.

Results

1 Overall, the proportion of children consuming at least one serving of fruit and vegetables increased from a baseline of 33% before the experiment to 60% (an increase of 80%). The incentive resulting in the largest increase was the larger immediate cash reward, which increased the proportion to 71%, whereas the delayed cash reward of the same amount increased the proportion to 63%. Surprisingly, the delayed prize increased the proportion more (to 60%) than the immediate prize (53%). The researchers propose that this difference is explained by the fact that, when the prize was delayed, all the prizes for the five-day period were displayed as opposed to just the prizes for one day.

2 The experiment revealed a substantial difference in effect between the schools with the highest-income children and those with the lowest-income children. For the richest children the increase in consumption was only 18%, whereas for the poorest the increase was 38%.

3 The program was found to be highly cost-effective compared to other programs designed to increase consumption of fruit and vegetables. The main aspect of this was that wastage, in terms of items thrown in the trash, was reduced by 43% compared with the baseline situation. Administrative costs, involving the cost of the rewards and the labor costs of distributing the rewards, were also low.

Two other findings emerged from the study:

1 There was no evidence of the incentive program reducing the number of servings consumed by children who normally ate more than one serving a day. The researchers originally feared that the program might create a social norm of consuming exactly one serving of fruit/vegetables per day, but this did not appear to be the case.

2 In a second experiment, involving a further eight schools, the after-effects of the incentive program were monitored in order to examine the long-run effects of such a program. Although the proportion of children consuming fruit and vegetables remained higher than the original level for the two weeks immediately following the program, in the next two weeks (two to four weeks after the incentive program ended) consumption returned to the baseline level before the program started.

Questions

1 What aspects of behavioral economics are relevant in this study?

2 Explain the significance of the difference in behavior between low-income and high-income children.

3 Comment on the long-run effects of the incentive program.

4 What other incentives do you think might be effective in increasing children’s consumption of fruit and vegetables, apart from the ones used in the study?