The Social and Individual Factors Behind Gift Exchange

Cheating and similar opportunistic behaviour has been discussed, among others, in the context of cheap-talk games (e.g., Sutter (2009); Gneezy (2005); Charness and Dufwenberg (2006)), where experiment participants can send messages and signals that can be strategically used to deceive other experiment participants for personal monetary gains. Further experimental evidence is provided for example by Bucciol and Piovesan (2011), who conducted and analyzed an experiment with children, in which they have to toss a coin in private and report the result afterwards, knowing that only one of the reported outcomes results in a reward. As 86% of them reported the profitable outcome, a substantial amount of them must have cheated. Freeman and Gelber (2010) reported the participants of their experiment cheating by overstating their number of correctly solved mazes, with varying monetary incentives influencing the individual degree of cheating. In a field study, Pruckner and Sausgruber (2013) observed two-thirds of newspaper readers taking advantage of an opportunity to acquire a newspaper without paying for it. Furthermore, around 90% of the individuals who did in fact pay, paid less than the full price. In an experiment conducted by Fischbacher and Heusi (2008), the authors reported a similar kind of ‘incomplete cheating’. In this experiment, participants had to report the result of a private die roll to determine their pay off. The authors observed that subjects stated their outcome of the roll strategically in order to increase their monetary gains. However, they did not try to get the maximum possible amount obtainable by lying. This behavior could be explained by an individual’s desire to keep a favorable and (at least partly) up-standing image of themselves. (Mazar et al. (2008))

Researching the link between perceived (un-)fairness and personal honesty, Greenberg (1990) reported a field experiment in which a company cut the pay of its workers temporarily due to the loss of a large contract. The experimental variation, more specifically the treatment, was the different kinds of communication the company used in order to communicate this action to its employees at different production sites. While the ones at one production site perceived this pay-cut as unfair, the company's measure was explained in detail to the individuals working at a different site, to try and prevent the employees from feeling this way. Greenberg highlighted that workers from the control group responded in a significant increase in employee theft, which was not observed with workers that believed themselves to be treated ‘fairly’. Fehr et al. (1993) arrived at a similar conclusion in their experiment which showed that the amount of co-operation between ‘workers’ significantly decreases when they are confronted with an unfairly perceived wage by their ‘employers’. These results are in line with observations by Schweizer and Gibson (2008), who had experiment participants indicate that a violation of fairness standards by a counterpart incentivizes them to react reciprocally with unethical behavior. Houser and Winter (2012) report a similar behavior of individuals when confronted with perceived unfairness in dictator games.

There has also been an increasing interest in the cognitive and neuronal processes involved in (dis-) honest decision-making. Greene and Paxton (2009) showed that honest individuals did not show an increased amount of activity in brain regions governing response conflict and cognitive control when they chose an honest option in favor of an opportunity for dishonest gain by using functional magnetic resonance imaging. The authors concluded that honest individuals were mostly not purposefully resisting the temptation of a dishonest option, but that they were rather simply not tempted by it. Wang et al. (2010) conducted an experiment by utilizing video-based eye tracking in order to deduce the level of an individual’s degree of multi-layered thinking (called 'level-k reasoning' in the paper) by analyzing their lookup pattern of pay-offs for alternative choices presented on a screen. The authors then continued to show that the knowledge of this information can be used to profitably exploit the examined individuals.

Gintis et al. (2001) showed that an individual's originally unobservable willingness to work together and cooperate can prove to be a very relevant signal if individuals are put into type-separating conditions. Elaborating on this, Fehrler and Przepiorka (2012) imply that in today's social interactions, altruistic behavior and trustworthiness are correlated, which means that individuals acting altruistically will appear trust-worthier in social situations and exchanges. They prove by using a computerized laboratory experiment that these two characteristics are indeed positively correlated and that a charitable donation increases an individual's trust in the donor. However, in a follow-up experiment, Fehrler and Przepiorka (2012) conclude donors to be indeed significantly more frequently selected as interaction partners, yet the authors do not find donors to actually act more trustworthy than non-donors in the same situation. To test the notion of reciprocity, Falk (2007) conducted a field experiment in collaboration with an international charitable organization helping children in need. The organization sent out roughly 10,000 solicitation letters to potential donors in Zurich, Switzerland. These letters were systematically manipulated so that one-third of the letters contained a large gift, one third a small gift, and one third no gift.

The assignment of the letters was random. Consistent with the gift-exchange hypothesis, Falk (2007) reports a 17 percent increase in the frequency of donations with the small gift treatment and a 75 percent increase for a large gift. Both differences prove to be highly significant. This initiation of a gift-exchange program proved to be profitable for the organization. Similar to the conclusion of the field experiment by Gneezy et al. (2006) and Charness et al. (2003), Falk (2007) mentions the gift exchange relation breaking down after a certain period of time. In this field experiment, testing gift exchange at the workplace, ‘gift-condition’ wages were higher than announced and anticipated. While initial output by the workers with the gift condition surpassed the output with the no-gift condition, the relation unraveled in a period of six hours.

Another limiting factor for gift exchange can be competitive pressure, as seen in List (2006), who studied social preferences in actual market transactions and observed this behavior to be comparable to the predictions of the standard self-interest model. Reciprocity and the other factors governing the gift exchange effect played little role. This means that the importance of gift exchange is not a matter of laboratory evidence versus field data, but that the exact circumstances and the environment of the situation determine whether a gift exchange relation profitable for both sides can be initiated, and also contained, successfully. Key factors seem to be attribution, or how generous, fair, or altruistic an offer/a gift is perceived by the recipient and how much the urge to respond with reciprocity can be evoked.

Charness et al. (2003) mention that laboratory gift-exchange results seem robust at first glance, yet they criticize that most of the conducted experiments have taken place in a rather small region, namely Austria and Switzerland. Only a few gift-exchange experiments had been conducted in the U.S. up until that point. Charness (2004) observes similar patterns of gift exchange experiments conducted in Berkeley to the ones in Fehr et al. (2000). However, Charness et al. (2003) mention that Hannan et al. (2002) observe different behavior patterns between MBA students (resembling a standard pattern; consistent with a different experiment conducted with MBA students by Hannan (2001)) and Pittsburgh undergraduates (showing only very limited gift exchanging). By comparing their designs, Charness et al. notice that Hannan et al. (2002) provided participants with a comprehensive pay-off table relating wages and effort levels to worker’s pay-offs and managers incomes. Those tables were not provided in the various experiments conducted by Fehr et al. either.

The experiment conducted by Charness et al. (2003) explores the role of the comprehensive pay-off table on the level of the reported gift-exchanges by utilizing six laboratory sessions with 114 participants at the Ohio State University during May 2001. The authors show that something as arguably innocuous as the inclusion of a payoff table in a game with complete information can have a significant impact on the relationship between wages, effort and pay-offs for principals and agents alike. At least for undergraduate students in the U.S., gift exchange is not reduced to zero, but it is heavily reduced compared to all other experiments conducted at this point, with Hannan et al. (2002) as the exception, who also use U.S. undergraduate students and a similar payoff table.

Charness et al. (2003) come up with various possible explanations for the ‘pay-off table effect’, like framing or presentation format effects: In particular, the existence of the payoff table could serve as a means to disconnect the relationship between the worker's effort and the principal’s wage decision in the worker’s minds by making them focus on their effort level and resulting payoffs, rather than social factors. Apart from the payoff table effect, Charness et al. (2003) also observe different behavioral patterns of the participants over time, with reciprocal responses diminishing towards the end period, which is consistent with various forms of strategic behavior.