How Do Economic Incentives Affect Social Preferences and Behavior Essay

For decades economic theories have relied heavily on the effectiveness of material incentives (Fehr & Gachter, 2001). According to the traditional exchange theory all people are exclusively motivated by their own material self-interest. It predicts that the introduction of a penalty will reduce the occurrence of the behavior that is subject to the fine. On the other hand it states that introducing a material incentive will lead to an increase of the behavior related to the bonus. Based on economic theory, incentives have become increasingly popular and are used to increase certain behaviors in various fields including environmental policy (Andersen & Sprenger, 2000; Barde & Smith, 1997; Baumol & Oates, 1988; Kahn, 1995; all cited in ThOgersen, 2003), household surveys (Singer, 2002) and education policy (Fryer, 2011). On the other side, penalties have been used to reduce free-riding (Feldman, Papadimitriou, Chuang, & Stoica, 2006), and crimes (Akerlof & Dickens, 1982). There is much evidence that supports the basic premise of economics that incentives are effective (Gibbons, 1997; Prendergast, 1999; Lazear, 2000; all cited in Benabou & Tirole, 2004). However, a large body of literature in psychology has shown that explicit incentives lead to decreased motivation and reduced performance in the long run (Deci & Ryan, 1985; as cited in Benabou & Tirole, 2004). Titmuss (1970, as cited in Benabou & Tirole, 2004) was the first who claimed that people might adopt a ‘market mentality’ when they are exposed to explicit economic incentives. He found that paying blood donors for donating blood could actually reduce supply. In the beginning there was little hard evidence that social preferences affected individual behavior, but empirical and theoretical advances over the past decades provide the basis for more support. For example, Gneezy and Rustichini (2000a) found that introducing a monetary fine for late-coming parents in day-care centers led to a significant increase in late-coming. There was no reduction in late-coming after the fine was removed. Also Fryer (2011) didn’t find evidence that providing financial incentives to teachers to increase student performance had any effect. Partly because of these findings, terms as trust, reciprocity, gift exchange and fairness have appeared in the empirical study and modeling of principal-agent relationships (Bowles & Polania-Reyes, 2012). This highlights the importance of the influence that social preferences have on incentives. Based on the contradictions mentioned above I conclude that a more thorough analysis is needed in order to understand the influence of incentives on behavior. I’ll focus on the interplay between incentives and social preferences and how this affects behavior. In this paper I will review several mechanisms that can explain how incentives can be less effective than economic theories predict and how they can even have counterproductive effects. Furthermore I will indicate the implications of the (non-)effectiveness of incentives for economic policy. Overview of past research According to the definition of Bowles and Polania-Reyes (2012), social preferences refer to â€Å"motives such as altruism, reciprocity, intrinsic pleasure in helping others, inequity aversion, ethical commitments and other motives that induce people to help others more than would an own-material-payoff maximizing individual† (p. 4). Fehr and Fischbacher (2002) have indicated the most important types of preferences that have been uncovered by the literature. I will shortly review them below. The first important type of social preference is the preference for reciprocal fairness or reciprocity. An individual is reciprocal when he responds kindly to actions that are perceived as kind, and when he responds hostile to actions that are perceived as hostile. Whether some action is perceived as hostile of kind depends on the unfairness or fairness of the intention and on the consequences that are associated with the action. A second social preference type is inequity aversion. According to Fehr and Schmidt (1999; as cited in Fehr & Fischbacher, 2002) â€Å"inequity averse persons want to achieve an equitable distribution of material resources† (p. C3). Inequity averse persons show altruistic behavior if the other persons’ payoffs are below an equitable level. However, if the other persons’ payoffs are exceeding the equitable level an inequity averse person want to decrease the other persons’ payoffs. There are a lot of similarities in the behavior of reciprocal and inequity averse individuals, since both concepts depend in some way on the perception of fairness. Pure altruism is the third type of social preference, which is very different from the former two. Altruism can be seen as an unconditional form of kindness (Fehr & Fischbacher, 2002), as an altruistic person would never take an action that decreases another person’s payoff. The problem with pure altruism is that it cannot explain conditional cooperation, that is, people want to increase their voluntary cooperation in response to cooperation of others. The last social preference type that Fehr and Fischbacher (2002) mentioned is envious or spiteful preferences. An envious or spiteful person always values the payoff of other agents negatively. Therefore the envious person is willing to decrease the other agent’s payoff even if it brings along a personal cost to himself. This happens irrespective of fair or unfair behavior of the other agent and irrespective of the pay-off distribution (Fehr & Fischbacher, 2002). However, spitefulness can’t explain why it is that the same individuals sometimes are willing to help others at a personal cost, while sometimes they harm other people. Over the past decades, many studies have confirmed that a significant fraction of individuals engage in reciprocal or altruistic behaviors (Buraschi & Cornelli, 2002; as cited in Benabou & Tirole, 2004; Fehr & Gachter, 2000). Thus, many individuals do not only care about the material resources allocated to them, but also care about material resources allocated to other relevant agents. To give an overview of the incentive effects on preferences, two distinctions are made: the nature and the causes of incentives (Bowles & Polania-Reyes, 2012). Concerning the nature of incentives, people often respond to the mere presence of incentives, rather than to their extent (Gneezy, 2003; as cited in Bowles & Polania-Reyes, 2012). However, the extent of an incentive may also play a role. Therefore the effects of incentives on social preferences can be either categorical or marginal or a combination of the two. Bowles and Polania-Reyes (2012) also make a distinction between 2 causes of incentive effects on preferences. First, incentives can affect the environment in which preferences are learned. When this happens, the preferences are referred to as endogenous preferences. Second, the extent or presence of incentives affect the behavioral salience of an individual’s social preferences. When incentives constitute different states, we refer to social preferences as state-dependent preferences. There are three mechanisms that make social preferences state-dependent. First, by implementing an incentive, the principal discloses information about his intentions, about his beliefs about the target of the incentives and about the targeted behavior. This information might affect the agent’s social preferences which in turn affect the agent’s behavior. Second, incentives provide situational cues for appropriate behavior. Finally, incentives may lead to a crowding out of intrinsic motivations. The crowding-out effect is based on the intuition that the presence of punishments or rewards spoils the reputational value of good deeds. This creates doubt within the individual about the extent to which he performed because of the incentives rather than for himself. This phenomenon is also referred to as the ‘overjustification effect’ (Lepper, Greene, & Nisbett, 1973; as cited in Benabou & Tirole, 2004). In the next part of this paper I’ll give experimental evidence for both endogenous preferences and for all 3 mechanisms that make preferences incentive-state-dependent. Furthermore, I’ll give examples of experiments where crowding in has been found and explain the underlying mechanisms. 1. Endogenous preferences: incentives alter how new preferences are learned Preferences are endogenous if someone’s experiences lead to durable changes in motivations and eventually result in a change in behavior in certain situations (Bowles, 2008). In most cases, experiments have a few hours duration and therefore it’s unlikely to uncover the mechanisms that are involved in the process of durable change of preferences. Although it’s hard to explore the causal mechanisms at work, there exist some experiments that do show a durable learning effect (Irlenbausch & Sliwka, 2005; Falkinger, Fehr, Gachter, & Winter-Ebmer, 2000; all cited in Bowles, 2008). Gneezy and Rustichini (2000a), for example, examined if the introduction of a monetary fine for late-coming parents in day-care centers would lead to reduction of late-coming. However, the amount of late-coming parents didn’t decrease, but increased significantly. Thus incentives led to more self-interested behavior. More importantly, after the fine was removed no reduction in late-coming parents was shown, meaning that there was some durable learning effect going on. 2. State-dependent preferences: incentives provide information about the principal When an incentive is imposed on an agent, he may infer information about the principal who designed the incentive. He may, for example, infer information about the principal’s beliefs regarding the agent, and about the nature of the task that has to be done (Fehr & Rockenbach, 2003). This information can lead to a negative response to fines that are imposed by principals. Fehr and Rockenbach (2003) designed a sequentially played social dilemma experiment and examined how sanctions intended to prevent cheating affect human altruism. Participants in the role of ‘investor’ could transfer a certain amount of money to another player, the ‘trustee’. The experimenter tripled this amount. After tripling the money, the trustee was given the opportunity to back-transfer some of this money to the investor. The investor could indicate a desired level of the back-transfer before he transferred the money to the trustee. In the incentive-condition the investor even had the option to impose a fine if the trustee would send a back-transfer that was less than the desired amount. Instead of imposing a fine the investor could also choose to decline the use of the fine. The decision of imposing or declining the fine was known to the trustee. In the trust-condition the investor could not make use of incentives. Fehr and Rockenbach (2003) found that generous initial transfers by investors were reciprocated with greater back-transfers by trustees. However, the use of the fine reduced the return transfers, while renouncing the fine in the incentive-condition increased back-transfers. This means that sanctions revealing selfish or greedy intentions destroy altruistic cooperation almost completely (Fehr & Rockenbach, 2003). In another experiment by Fehr and Schmidt (2007), principals could choose between offering a bonus contract or a combination contract (which was a combination of the bonus contract with a fine) to the employee. What they found was that agents perceive that principals who are less fair are more likely to choose a combined contract and are less likely to pay the announced bonus. Furthermore the effect of effort on the bonus paid is twice as large in the pure bonus condition compared to the combined contract condition. The positive response to the principal’s renunciation of the fine option can be seen as a categorical effect. The threat of a fine led to diminishment of the trustee’s reciprocity. 3. State-dependent preferences: incentives may suggest permissible behavior The experiments that will be described here, differ from the experiments mentioned above in the way that here incentives are implemented exogenously by the experimenter. This means that incentives do not provide any information about the beliefs or intentions of other experimental subjects. In a lot of situations people look for clues of appropriate behavior. These are often provided by incentives. These framing effects have been investigated in many studies. Hoffman, McCabe, Shachat and Smith (1994; as cited in Bowles & Polania-Reyes, 2012) found that by making a game sound more competitive after relabeling it, generosity and fair-minded behavior in the participants were diminished. In some other studies (Ellingsen, Johannesson, Munkhammar, & Mollerstrom, 2008; as cited in Bowles & Polania-Reyes, 2012) the framing effect even appeared to have changed subjects’ beliefs about the actions of others. Framing effects can also be induced in other ways than simply renaming the experiment. Providing an incentive may already provide a powerful frame for the decision maker. In an experiment of Schotter, Weiss and Zapater (1996) subjects played an Ultimatum Game experiment in which player 1 is given an endowment and asked to propose a part of this endowment to player 2. Player 2 can either accept or reject this division. If he accepts, the proposed division is implemented. However, if he rejects both players receive nothing. Schotter et al (1996) found that if a market-like competition was included in the game, that is, subjects with lower earnings would be excluded from the second round in the game, player 1 proposed less generous divisions to player 2. Furthermore, lower offers were accepted by player 2. The authors interpreted these results as that implementing market-like competition â€Å"offers justifications for actions that in isolation would be unjustifiable† (p. 38). Thus, providing incentives in the form of a competition can lead to moral disengagement. The framing effects of incentives can occur in cases of government-imposed incentives as well. An example comes from an experiment from Cardenas, Stranlund and Willis (2000) where they studied the effects of external regulatory control of environmental quality. Participants were asked to choose how much time they would spend collecting firewood from a forest, while being aware that this activity has a negative effect on local water quality. Two treatments were considered to examine whether external control may crowd out group-oriented behavior. All subjects played eight initial rounds of the game without any treatment, that is, without being able to communicate with each other and without external regulation. After the initial rounds, one subset of groups played additional rounds in which they were able to communicate. The other subset of groups was confronted with a government-imposed regulation. The regulation also involved the possibility of imposing a fine to subjects that would withdraw too much of the firewood. Although standard economic theory predicted that the regulation would increase group-oriented behavior, this wasn’t the case. When subjects were able to communicate they made way more efficient decisions. However, regulatory external control caused subjects to make decisions that were closer to their self-interest. This means that the fine, although it was insufficient to enforce the social optimum, extinguished the subjects’ ethical aptitudes. 4. State-dependent preferences: incentives may compromise intrinsic motives and self-determination A third reason why social preferences may be state dependent is because providing incentives may lead to motivational crowding out. As Bowles (2008) put it: â€Å"where people derive pleasure from an action per se in the absence of other rewards, the introduction of explicit incentives may ‘overjustify’ the activity and reduce the individual’s sense of autonomy† (p. 607). According to Deci (1975; as cited in Bowles, 2008) the underlying psychological mechanism appears to be a desire for â€Å"feelings of competence and self-determination that are associated with intrinsically motivated behavior† (p. 1607). There is a large body of literature on the psychology of intrinsic motivations going back to the early work of Festinger (1957; as cited in ThOgersen, 2003) and his cognitive dissonance theory. In the past decades a lot of experiments have been done to test the crowding out of intrinsic motivation. One of these studies comes from Gneezy & Rustichini (2000b) who tested the effects of monetary incentives on student performance. 180 students were asked to answer 50 questions of an IQ test. They were all paid 60 NIS (New Israeli Shekel) for their participation in the experiment. The students were divided into 4 different groups, which were all corresponding to 4 different treatments. The students in the first treatment group were only asked to answer as many questions as possible. The students in the second group got an extra payment of 10 cents of a NIS per question that they answered correctly. Subjects in the third group were promised 1 NIS, and subjects in the fourth group 3 NIS per question that they answered correctly. The average number of questions correctly was approximately 28 in the first group and declined to 23 in the second group. Furthermore, the number increased to 34 in both the third and the fourth group. The differences in performance were significant. In a second experiment Gneezy & Rustichini (2000b) tested the effect of incentives on volunteer work performed by high school children. 180 children were divided into three groups. The subjects in the first group constituted the control group and they were only given a speech about the importance of volunteer work. The second group was given a speech as well, but was also promised to receive 1 per cent of the total amount of donations collected. The third group was promised 10 per cent of the amount collected. The average amount collected was highest in the first group and lowest in the second group. The average amount that was collected by the third group was higher than that of the second group but not as high of the amount that was collected in the first group. Also these results were significant. It appears to indicate that the effect of incentives can be detrimental, at least for small amounts. In another experiment, Falk and Kosfeld (2006; as cited in Bowles 2008) tested the idea that control aversion based on the self-determination motive is the reason that incentives reduce performance. They used a principal-agent game where agents could choose a level of production that was beneficial for the principal, but costly for themselves. If the agent chose to produce nothing, he would get a maximal pay-off. Before the agent’s decision the principal could decide to leave the choice f production level completely to the agent or to impose a certain lower bound on the agent’s production level. The experimenter varied the bounds across the treatments and the principal could only choose to impose it or not. Results showed that when the principal imposed the bound, the agents chose a lower production level than when the principal didn’t impose a bound. The ‘untrusting’ principals earned half of the profits of those who did trust the agents and thus didn’t impose a bound. In post-surveys, the agents indicated that imposing the lower bound was perceived as a signal of distrust. The results of this experiment suggest that the desire for self-determination and control aversion are not the only effects of imposing the bound. Imposing this minimum was informative for the agents about what the principals’ beliefs were regarding the agents: the principals who imposed the bounds had lower expectations of the agents. Thus, the results in the experiment of Falk and Kosfeld (2006; as cited in Bowles 2008) seem to be the result of both negative information about the principal (or incentive designer) as well as the result of self-determination. 5. Crowding in Although a lot of experiments show that providing incentives has a negative effect on social preferences, there is also some evidence that crowding in can occur, that is, social preferences and incentives enhance the effect on each other. This might happen when an incentive provides good news about the principal’s type or intentions, for example when he offers the agent a reward rather than a fine. It is also seen in experiments where the incentive designers are peers in a public goods game who pay to punish free riders in order to sustain cooperative behavior (Bowles & Polania-Reyes, 2012). The phenomenon of crowding in is interesting since it indicates how policies could be implemented optimally and how incentives and social preferences could become complements rather than substitutes (Bowles & Polania-Reyes, 2012). Besides that, it appears that crowding in happens often in Public Goods games and Common Pool Resources games, which display the same characteristics as public policy settings. Below I’ll give an example of an experiment in which crowding in was found. Fehr and Gachter (2000) conducted a public good experiment with and without the opportunity to punish. In the no-punishment treatment the dominant strategy is complete free-riding. In the punishment treatment free-riders could be punished by their altruistic peers, since it was costly for them to punish. Therefore, if there were only selfish individuals, as assumed in economic theory, there wouldn’t be a difference between the two treatments. However, in the no-punishment treatment the contributions of the players were substantially lower than in the punishment treatment. This suggests that powerful motives drive the punishments of free-riders. Furthermore there was evidence that the more free-riders deviated from cooperation, the more they were being punished. There are several mechanisms that can explain the effect of crowding in. In the first place when a peer imposes a fine on a free-rider, this may activate a feeling of shame. Barr (2001; as cited in Bowles & Polania-Reyes, 2012) found that just a verbal message of disapproval already can have a positive effect on the free riders’ contributions. A second mechanism that appears to be at work it that nobody wants to be the cooperator while all others are defecting. Shinada and Yamagishi (2007, as cited in Bowles & Polania-Reyes, 2012) found that students cooperated more in a public goods experiment when they were assured that defecting free-riders would be punished. They just didn’t want to be exploited by defectors. A third mechanism underlying crowding in was consistent with the findings of an experiment by Vertova and Galbiati (2010, as cited in Bowles & Polania-Reyes, 2012). They found that when a stated obligation was introduced, this produced a larger effect when it was accompanied with a small monetary incentive, rather than with a big incentive or than when no incentives were offered. The authors interpreted this phenomenon as that the salience of the stated obligation is enhanced by large explicit incentives. The latter phenomenon was also found in Ireland, where a small tax was imposed on plastic grocery bags (Rosenthal, 2008; as cited in Bowles & Polania-Reyes, 2012). After two weeks there was a 94% decline in the use of these bags. This result can be explained by the fact that the introduction of the tax was preceded by a large publicity campaign. Thus, the incentive was implemented jointly with a message of social obligation and it seems that it served as a reminder of the importance of one’s civic duty. Implications for policy Many policies are based on the self-interest hypothesis that predicts that all individuals are self-regarding. However, as we have seen social preferences play an important role as well when it comes down to behavior. This would mean that a lot of current policies are non-optimal. Therefore a big challenge is facing the mechanism designer: how to design optimal fines, taxes or subsidies when the individual’s responses depend on his preferences which in turn are determined by the incentive imposed? In most experiments the effects of incentives were studies and afterwards the mechanisms were identified that could explain the results. However, one of the problems that the designer is facing is that he must determine beforehand how incentives will affect behavior. Based on the experiments that have been done, several guidelines can be drawn. The first is that when crowding out is found, social preferences and incentives are substitutes. This means that a negative effect of incentives is less likely to be found when the social preferences are minimal. In contrast, when social preferences are prevalent among a society, it may be more convenient to reduce the use of incentives. Also, policies that are implemented in order to enhance social preferences will be more effective when incentives are little used. The second stems from Titmuss’s claim that if the crowding out effect is so strong that the incentive has an opposite effect than intended, incentives should be used less. However, in many cases the effectiveness of incentives is not reversed, but blunted and then the implications for the optimal use of incentive isn’t that obvious (Bowles & Hwang, 2008). How Bowles & Hwang (2008) state it: â€Å"the reduced effectiveness of the incentive associated with crowding out would entail a larger incentive for a planner designing a subsidy to ensure compliance with a quantitative target† (p. 4). Present evidence is insufficient in providing enough guidelines to the policy maker who wants to know ex ante what the effects are of the incentives that he considers to implement (Bowles & Polania-Reyes, 2012). What we do know is that the same incentives imposed by individuals who have no personal benefit but only want to promote pro-social behavior (as in the experiment of Fehr & Gachter, 2000) are more likely to increase contributions than when imposed by an untrusting principal (Fehr & Rockenbach, 2003). Furthermore it seems to be important to let the agent understand that the desired change in behavior would be socially beneficial rather than that the incentive is perceived as a threat to her autonomy or reflecting badly on the designer’s intentions (Bowles & Polania-Reyes, 2012). Conclusion The self-interest hypothesis assumes that individuals are only motivated by their own material self-interest. This assumption is used in the design of many policies. However, in the past decades a lot of experiments have shown that other-regarding social preferences rather than self-regarding preferences play a role in behavior. We have seen that some mechanisms can induce pro-socially oriented individuals to behave as they are selfish. On the other hand, there are also examples of experiments in which mechanisms induce self-interested individuals to behave at a more pro-social level. Thus, incentives can lead to both crowding out and crowding in phenomena. Whereas negative information about the principal and the over-justification effect may lead to crowding out of intrinsic motivation to contribute to a good, altruistic punishment by peers who do not benefit personally is more likely to increase contributions. Furthermore it seems important to make individuals aware of their civic duty, as was shown in Ireland where a small tax was imposed on plastic bags. Regarding to public policy, we have seen that small differences in institutional design can lead to many different outcomes. This imposes a big challenge on the policy designer who has to know ex ante what the effects of the incentive that he is considering to implement will be. When social preferences are not present, incentives may have a positive effect, predicted by economic theory. However, in areas where social preferences do play a role, the use of monetary incentives needs to be reconsidered.