Trust is one of the key factors affecting the emergence and sustainability of cooperation. As stated, for instance, by Acedo and Gomila, “In evolutionary game theory and experimental economics, the notion of trust is much simpler: it is an expectation about another’s behavior, a kind of wager, in which the sense of mutual commitment and vulnerability is completely absent” (see their paper here). Therefore, the fact that a certain group of people or type of companies are dishonest is a very important finding, and more so if we are speaking of so crucial a sector as the banking industry. This is what Alain Cohn, Ernst Fehr and Michel A. Maréchal set out to find with an experiment, whose results are reported in a 2014 paper entitled “Business culture and dishonesty in the banking industry“.
(Versión en español aquí)
For their research, they recruited 128 bank employees from a large international bank, with an average of more than 11 years of experience in the banking industry. Half of them worked in proper banking business, such as private banking or asset management, and the other half worked in support units, such as human resources management. The key twist here is that they randomly divided the sample in two in order to have two separate treatments: one in which the professional identity was made salient and one in which it was not. The way to do this is to prime the subjects in the first treament by asking them questions prior to the experiment proper such as “At which bank are you presently employed?” or “What is your function at this bank?”. In turn, the second treatment questions were unrelated to the subjects’ job, an example being “How many hours per week do you watch TV on average?”. Is this enough to give relevance to the subjects’s job? To check it, the experimenters asked them to complete word fragments, such as “__oker”. One possible answer is “broker”, and another is “smoker”. They found that primed subjects indeed provided more bank-related words than the control group (36% vs 26%), indicating that their manipulation did have an effect.
The experiment then went as follows, quoting from the original paper:
After the priming questions, all subjects anonymously performed a coin tossing task that has been shown to reliably measure dishonest behaviour in an unobtrusive way and to predict rule violation outside the laboratory. The rules required subjects to take any coin, toss it ten times, and report the outcomes online. For each coin toss they could win an amount equal to approximately US$20 (as opposed to $0) depending on whether they reported ‘heads’ or ‘tails’. Subjects knew in advance whether heads or tails would yield the monetary payoff for a specific coin toss. Moreover, subjects were informed that their earnings would only be paid out if they were higher than or equal to those of a randomly drawn subject from a pilot study. We introduced this element to mimic the competitive nature of the banking profession9. Given that themaximum payoff is approximately $200, subjects faced a considerable incentive to cheat by misreporting the outcomes of their coin tosses.
The results of the experiment show that bank employees behave more dishonestly in the professional identity condition. As the figure below shows, the histogram of earnings reported in the control condition overlaps quite well with the binomial distribution that would result from a fully honest behavior, the average reported earnings being 51.6%. On the contrary, the histogram in the professional identity condition shows a displacement towards higher earnings, with an average of 58.2%, significantly different from the expected 50%.
By analyzing in more detail the results, and within the natural assumption that subjects did not cheat to their disadvantage, the rate of misreporting in the professional identity condition is 16%, while they were also able to find out that the fraction of subjects who cheated is 26% (see the original reference for details). In addition, they also found:
- The effect is robust even when we control for a large set of individual characteristics such as age, gender, education, income, and nationality, as shown by regression analysis.
- The effect does not arise because the priming increase competitiveness, as shown by the lack of correlation between self-reported competitiveness with succesful coin flips and the similarity of the answers in the two treatments.
- Employees in core business units were more dishonest than those in support units.
- Subjects in the professional identity condition endorse more strongly the statement that social status is primarily determined by financial success, and a stronger endorsement of the materialistic statement is positively correlated with the reported number of successful outcomes.
- This behavior is specific to the banking industry: a separate experiment with 133 employees of different sectors showed no effects of the job identity condition.
- The possibility that the effect arises from the concept of “money”, closely tied to banking concepts, is ruled out by running yet another experiment with students, in which they were primed with questions about the typical chores of a bank employee. In this case, both the primed subjects and the control ones behaved similarly and similarly to the control condition.
The comparison of the differences between control and profesionally primed groups for the three experiments (bank employees, employees from other industries, and students, is shown in panel a) in the figure below, confirming the two last points above. Interestingly, the authors conclude that “the prevailing business culture in the banking industry favours dishonest behaviour and thus has contributed to the loss of the industry’s reputation”, even if bankers, as shown by the control condition, behave honestly on average. That the reputation loss is important is shown by a survey in which people from the general population were asked to guess the percentage of successful coin flips that different types would report. As shown in panel b) in the figure below, the general population believes that medical doctors are more honest than the typical citizen, whereas bank employees are perceived as more dishonest (non statistically siginficantly) than prison inmates.
I concur with the authors that their experiment, with all the additional studies it involves (including some I have skipped here and available in the supplementary material) shows clearly that the culture of the banking industry needs a very serious rethinking. In the meantime, I wonder what would happen if a similar priming procedure were applied to a experiment on cooperation, say between bank employees with each other and bank employees with people extracted from the general population. Would there be differences in their cooperative behavior? I’m guessing there will be, but I’ve done enough experiments already to know that life is full of surprises… Anybody out there wanting to fund the experiment?