Nils Köbis, a post-doctoral researcher at the Center for Experimental Economics and Political Decision-Making (CREED), University of Amsterdam, contributes today’s guest post:

Whenever I am traveling and take a taxi, I try to strike up a conversation with the driver. The beauty of this situation is that both sides can be really candid. The length is typically short, and chances are you will never meet again. The chat usually kicks off with some small talk about sports, weather, and food. Once warmed up, I have often asked: “What do you think is the biggest problem in your society?” So far, the most common answer has been corruption. And my taxi-driver-based anecdotal evidence is consistent with large  international surveys. Notwithstanding the old canard that people who live in corrupt societies generally tolerate corruption as normal and natural, ample empirical evidence (and my taxi drivers) suggests that this is not true: People widely despise corruption, especially in countries riddled with it. Yet on several occasions the very same taxi driver who has been ranting to me about corruption has stopped by a traffic police officer—and willingly paid a bribe to avoid a ticket.

What explains this apparent paradox? The most frequent explanation for why a person outraged by corruption would nevertheless pay a bribe is that “everybody does it”—as the Nigerian novelist Chimamanda Adichie nicely puts it, “If we do something over and over again it becomes normal. If we see the same thing over and over again it becomes normal.” This notion of normality plays an important role in explaining why corruption is sometimes the exception and sometimes the rule. Scholars who research social norms differentiate between injunctive norms, which concern whether a given behavior is acceptable, and descriptive norms, which indicate whether the same behavior is common. This distinction might help to explain the taxi-driver-paradox: People might often bribe because everybody else is doing it, even though they think it’s wrong.

Although widely endorsed by corruption specialists, rigorous empirical evidence is still relatively rare. My collaborators and I recently put this hypothesis to an empirical test by conducting three experiments in our lab in Amsterdam. Of course, studying corruption in the lab is challenging. For one thing, corruption can take so many different forms. We ultimately decided to design an experiment that simulates a bribery situation, because this seems to be the most commonly experienced form of corruption. In our game (which you can play here, if you are curious), participants get the opportunity to bribe a minister to obtain advantages in the allocation of a construction job; the decision to pay a bribe confers a benefit on the bribe-paying player, but imposes a cost on other players (and these benefits and costs affect the actual material rewards that experiment participants receive at the end of the game). We decided not to include any threat of punishment, in order to capture a situation where corrupt actors can act with impunity—a situation that, alas, exists in many places around the world.

The result? Most participants (around 75%) were willing to pay a bribe that directly hurt another player. The results also confirm the taxi-driver-paradox: The decision to bribe was driven by the perceived frequency and not the moral valence of bribery. Descriptive norms trumped injunctive norms. Although participants perceived bribing the Minister as corrupt and unfair, they did it since they perceived it to be the common thing to do. It confirms the commonly voiced logic: “If everybody does it, I might just do it as well.”

This is both bad and good news. The bad news first: Perceptions about what other people do are subject to all kinds of biases. The very idea of “everybody bribing” is most certainly wrong. Even in countries with endemic corruption, many abstain from it. Telling each other and yourself that “everybody does it” keeps this narrative alive. It sustains an often false belief about the behavior of others and leads to a social trap.

Now the good news: These perceptions can change. Studies show that people quickly adjust when the rules of the game are changing. This can be after a political regime change or after migrating to other societies. Inspired by that idea, we wanted to see whether this could also happen in the lab. So in the last of our three experiments, we gave participants short descriptive norm statements before the game: they either received a statement that bribing in the experiment is common (“almost everybody does it”) or uncommon (“almost nobody does it”). The norm statements giving the impression that hardly anybody engages in this type of corruption drastically reduced the level of bribery in the subsequent game. This suggests that if people believe that corruption is uncommon, they are less likely to do it themselves.

This has a number of implications for anticorruption strategy.

• First, we need to rethink the way we speak about corruption: Although the narrative of “everybody does it” is quite popular, it almost certainly not true—even in highly corrupt societies many abstain from corruption—and saying that “everybody does it” sustains a (false) perceived high descriptive norm that in fact might reinforce corrupt behavior. In fact it might even lead to “pluralistic ignorance”: false beliefs about the behavior of others sustaining this very behavior.
• Second, the news media plays a crucial role in shaping perceptions, and the manner in which a corruption scandal is covered can shape the readers’ social norms. For example, media coverage about a corruption scandal conveying the message that “yet another politician is corrupt” might increase the perceived descriptive norm that corruption is rampant; however, media coverage that focuses on the “heroic” acts of those detecting and persecuting the corrupt activities, might lead to the opposite effect. This alternative focus on the efforts of those fighting against corruption might shape the readers’ perception that corruption is not the social norm, and that punishment is in effect.