Mind the Economy / Incentives 7

The logic of incentives and the economics of mirroring

Economists and psychologists around the 1970s eventually resumed a dialogue, first in a rather confrontational manner and then gradually with an attitude of greater and more fruitful collaboration

8 December 2024

Aggiungi Il Sole 24 Ore
ai preferiti su Google

10' min read

Economists and psychologists had stopped talking to each other at the beginning of the 20th century, when Vilfredo Pareto, followed some time later by John Hicks, Roy Allen and Paul Samuelson, succeeded in eliminating any reference to concepts of a psychological nature from economic theory, reconstructing the entire discipline on a hypothetico-deductive basis. It was a veritable process of 'depsychologization' that transformed economics as it had hitherto been understood. The reason behind this process was not so much, as is often wrongly believed, the impossibility of constructing a solid economic theory by basing it on vague and unscientific concepts such as those of psychology at the time, but rather the adherence to a scientific project aimed at constructing an extremely specialised discipline totally based on rational choice theory (Bruni, L., Sugden, R., "The road not taken: how psychology was removed from economics, and how it might be brought back". The Economic Journal 117, pp. 146-173, 2007).

But economists and psychologists around the 1970s eventually resumed their dialogue, first in a rather confrontational manner and then gradually with an attitude of greater and more fruitful collaboration. The publication in 1979 in Econometrica of Prospect Theory by Daniel Kanheman and Amos Tvesky is probably the main event with which we can coincide the resumption of this dialogue. The two psychologists, in fact, presented a set of experimental data that seemed to contradict the economic theory of choice behaviour, while proposing an alternative theory of non-rational behaviour based on psychological assumptions. This work turned out to be seminal, so much so that in 2002 Kanheman was awarded the Nobel Prize in Economics 'for integrating results from psychological research into economic science, especially with regard to human judgement and decision theory under conditions of uncertainty' - as the official motivation reads. It would certainly also have been awarded to his lifelong friend and co-author Amos Tversky had he not died prematurely in 1996.

The dialogue between the two disciplines is not always easy because on many important aspects positions continue to diverge. One field in which the confrontation is very active is the study of motivation and incentives. For economists, as we have seen in last weeks Mind the Economy, incentives act as 'positive reinforcers', stimuli, that is, that increase the frequency of expected behaviour. If workers know that they can get a productivity bonus at the end of the year, they will be induced, according to the principle of operant conditioning, to be more productive. In the same way as a trained elephant showers his trunk with the knowledge that he will be rewarded with a certain amount of food once the task is accomplished. Symmetrically, other behaviours can be disincentivised by, for example, sanctions that act, this time, as 'negative reinforcers'. The psychological approach to the issue is different. Many scholars who have dealt with the issue have come to the conclusion that the same incentives can operate simultaneously as both 'positive reinforcers' and 'negative reinforcers'. It has been seen, in fact, that the prospect of a monetary bonus can, in many cases, decrease productivity just as the threat of sanctions, instead of discouraging certain behaviours, makes them more frequent. In an experiment that has now become a classic, psychologist Edward Deci engages university students to work on solving a puzzle. In a first phase, all participants have to solve the puzzle without any reward. In this way, the psychologist can measure the level of intrinsic interest each participant shows in the given task. In the second phase, one part of the students receives a monetary reward for completing the puzzles, while a second group continues without external incentives. In the third phase of the study, the participants are allowed to continue working on the puzzles without, however, receiving any reward. The results of the study show that the subjects in the first group, the one without a reward, continue to show a high level of interest in the activity even in the final phase, while the subjects who were paid in the second phase choose to spend less time solving the puzzles. That is, they show less interest in the final phase than they did in the first phase. It is concluded that the introduction of an extrinsic reward leads to a subsequent decrease in intrinsic motivation (Intrinsic motivation. Plenum Publishing Co, 1975). This experiment was subsequently replicated many times, with numerous variations with respect to the nature of the task and the type of subjects involved. All replications continued to show the same effect of 'displacement' of intrinsic motivation by extrinsic rewards.

In an attempt to bring these two different perspectives into an organic and coherent framework and to continue the dialogue with fellow psychologists, the two French economists Roland Bénabou and Jean Tirole have recently proposed a formal model that attempts to account for both the intrinsic interest in a certain activity and the effect of external incentive-related drives. The two economists describe intrinsic motivation as the pleasure or satisfaction derived from performing a certain activity in its own right. For example, a person who paints because he loves art acts driven by intrinsic motivation, just as a child who plays for no other reason than to derive pleasure from the game itself. The motivations of teachers, nurses, volunteers and, hopefully, politicians, have, at least in part, an intrinsic nature. Extrinsic motivations, on the other hand, are associated with various external forms of reward such as money, social recognition or the desire to avoid a sanction. A worker, for example, who performs a certain task with the prospect of getting a bonus, or a citizen who complies with a certain rule for fear of being sanctioned, act driven by their extrinsic motivations. The novelty that Bénabou and Tirole introduce in their model is the possibility that the two types of motivations, intrinsic and extrinsic, may sometimes conflict with each other generating an effect known as 'motivational crowding-out'. This phenomenon occurs for many reasons. The model, in particular, focuses on the impact of extrinsic incentives on the way agents perceive the nature of the task being incentivised. If we pay a student to read a book we are 'telling' him, indirectly, that reading books is a boring activity in itself that is not worth doing unless there is some form of external reward beyond the pleasure of reading itself. When sociologist Richard Titmuss compared the blood collection systems in the USA and Great Britain in the 1970s, he noticed that in the British system based on voluntary donation, the availability of blood was greater and more regular than in the USA where blood was purchased from commercial blood banks. Titmuss realised that in the commercial system, altruistic donors were discouraged by the presence of monetary reward while people interested in money were attracted, often people in financial difficulty who saw the donation as an opportunity to earn a few dollars. Many of these were homeless or other marginalised people, among whom the incidence of blood-borne diseases was very high. Their 'donations' therefore proved worthless.

Bénabou and Tirole explain the potential conflicts between intrinsic and extrinsic motivations as a consequence of the interaction of two factors, or rather, two 'gazes': on the one hand, the way we see ourselves, which they call 'self-presentation' and, on the other hand, the way we see ourselves through the eyes of others, our 'reflected self' or 'loocking-glass self'. This idea was originally proposed by sociologist Charles Cooley in his 1902 work entitled Human Nature and the Social Order to describe the process by which we use others as mirrors to observe ourselves. In fact, the same idea can already be found, at least in nuce, in Adam Smith's Theory of Moral Sentiments, published in 1759, in which we read, among other things, that man 'does not merely desire praise, he desires to be worthy of it, that is, he desires to be the natural and proper object of praise, though praised by no one. He does not just fear blame, he fears being worthy of it, that is, he fears being a natural and appropriate object of blame, even if not blamed by anyone'. I commented on this on another occasion that 'Making the distinction between the desire to be praised and that of actually being worthy of praise is important because these two aspects do not always go hand in hand. There are situations, in fact, in which we can be praised for doing something that is not actually praiseworthy, and cases in which an action that is actually praiseworthy is not met with any expression of approval (the same, symmetrically, can be said of being blamed and being blameworthy) (The Paradoxes of Trust. Rational Choices and Interpersonal Dynamics. Il Mulino, 2007). Smith continues, 'Although a wise man derives little pleasure from praise when he knows he does not deserve it, he often takes great pleasure in doing something for which he deserves praise, even though he knows just as well that it will never be bestowed upon him. "Obtaining the approval of mankind when it is not due can never be an important thing for him. Obtaining it when it is really due may be of importance on some occasions. But it will always be of the utmost importance to him to deserve approval. There are, therefore, two distinct 'relational' sources of motivation: the first related to the relationship we have with ourselves and the second conveyed by the relationship we have with others.

The introduction of extrinsic incentives can complicate this dynamic by generating two distinct effects. On the one hand, the 'reputation effect' which is associated with the reasons observers attribute to the agent performing a certain action. Does the student who reads a book do so because he is interested in reading or for the reward? Does the restaurateur who does his job conscientiously do it because he cares about the well-being of his customers or because he fears negative reviews? And then there is the 'self-perception effect', which stems from the way we ourselves evaluate the genuineness of our actions. A person who normally volunteers for altruistic reasons might begin to doubt his or her own authenticity if he or she receives monetary recognition for his or her efforts. Thanks to the loocking-glass effect, the agent - the one who receives the incentive - extracts from this 'signal' information that he does not possess about, for instance, the usefulness, appropriateness or more or less interesting nature of what he is asked to do: read a book, donate blood, work hard. But from incentives we also extract 'signals' that convey information about what the 'principal', i.e. the teacher, society or employer thinks of us. A similar dynamic occurs, for example, with respect to the autonomy or control that employees in a company may experience. "When subjected to close external monitoring or surveillance," writes Jeffrey Pfeffer, "psychologically they may draw the conclusion that their superiors do not trust them. And they don't trust them because perhaps they really are not trustworthy, and this may cause them to actually act in ways that reinforce this perception' (Competitive Advantage Through People: Problems and Prospects for Change. Harvard Business School Press, 2014).

Starting from these theoretical assumptions, it is possible to derive some interesting results from the model. We are always assuming on the one hand a principal - employer, teacher, company - who has private information about the nature of the task to be performed by the agent - the worker, the student, the citizen - and the qualities he imagines the latter to possess. On the other hand, we have an agent who does not possess such information and tries to deduce it from the principal's behaviour, for example, from the type of incentives the principal chooses to offer him. The first conclusion of the model is that incentives affect intrinsic motivation in such a way that when the agent is not confident in his abilities, intrinsic motivation decreases as the incentive increases. Similarly, when there is uncertainty about the nature of the task to be performed - interesting or boring, for example - the agent's perception is negatively affected by the level of the incentive. Put another way - the more you pay me, the less you think I am capable and the more unpleasant the task you want to assign me must be. This is why monetary incentives will only act as positive reinforcers in the immediate term, while in the long term they will contribute to reducing motivation. One of the conditions for these results to emerge is that there is a strong information asymmetry between the principal and the agent regarding the nature of the task and the agent's abilities. This explains, according to Bénabou and Tirole, why the phenomenon of crowding-out is very much present in areas such as schools where there is a large information asymmetry between teacher and student and less so in the workplace where, on the contrary, the information differences between what the employer knows about the task and the qualities of the worker is not too different from what the worker himself knows.

Bénabou and Tirole's work addresses many other issues, including that of dominance within power relationships. Relationships in which each agent, through the mechanism of looking-glass self can try to reduce the self-esteem of others in order to place themselves in a situation of dominance and be able to gain control of the situation. The theoretical model produces other important operational implications with regard to the functioning of public policies, management and education policies. Consider, for example, the use of economic incentives and subsidies to promote environmentally sustainable behaviour. If not well-calibrated, these interventions run the risk of displacing the intrinsic motivations of those most sensitive to these issues by reducing their commitment to them. In the managerial sphere, the use of extrinsic incentives can, under the conditions we have described, increase productivity in the short term, but reduce it in the long term. In education, of course, teachers can use rewards of various kinds to encourage desired behaviour on the part of their students, but they must do so in the knowledge that these incentives may at the same time reduce their intrinsic interest.

The right mix of intrinsic and extrinsic motivations depends on the context and time dynamics. For example, in some cases, extrinsic incentives may be crucial to initiate desirable behaviour, such as when new social norms have to be established, however, once these have become widespread and shared, one should reduce the impact of incentives and focus on promoting intrinsic motivation. With its strengths and weaknesses, Bénabou and Tirole's model represents an important attempt to integrate the economic and psychological approaches in addressing the fundamental issue of incentives and their hidden costs. A topic of great relevance in many areas of our political, economic and social life.