Reward Now, Punish Later
Skip to content
Economics Strategy Leadership Policy Nov 1, 2010

Reward Now, Punish Later

Punishment alone may not change social norms

Based on the research of

Florian Herold

Over the years, economic theory has generally held that self-interest predominantly governs the behavior of individuals in small groups.

However, recent experiments have suggested a situation less cut and dried: some experimental subjects, they indicate, are willing to reward good behavior by their peers and to punish bad behavior, even at some cost to themselves. Now Florian Herold, an assistant professor of managerial economics and decision sciences at the Kellogg School of Management, has developed a theoretical model that clarifies those findings. The model, he says, explains why rewarding and punishing behaviors can not only coexist with selfish behavior, but also induce cooperative behavior in small groups.

To illustrate his findings, Herold suggests two ways to fulfill the recommendation of many doctors that people sneeze into their elbows rather than into their hands. Here the carrot and the stick—rewarding and punishing behaviors—can work in concert. “Do you punish everybody who does the wrong thing by sneezing into the hand or reward those who use their elbows?” he asks. “Punishment is not viable in the beginning because you’d be punishing almost everybody.

Rewarding works at that time, as few people will be getting rewarded in the beginning. But at some point punishment takes over, as by then few people will be doing the wrong thing, and hence few will be punished.” That shows “that it’s much easier to change an established behavior by offering rewards, rather than threatening with punishments,” he adds. “But once you have established a norm, sustaining it by the threat of punishment is cheap. Only a few people will violate the norm, so you will rarely have to follow through with your threat.”

“It’s much easier to change an established behavior by offering rewards, rather than threatening with punishments.” — Florian Herold

As that example indicates, while both the carrot and the stick play strong roles in influencing changes in a group’s actions, they operate in different ways. Herold makes the point more formally in a paper on his research. “Rewarders,” the paper states, “enhance the evolution of preferences for punishing.” That means, Herold explains, that “reward may help to get you to a situation in which you can establish punishment.” The model also leads to a conclusion that echoes the much-quoted comment of baseball manager Leo Durocher: “Nice guys finish last.” Says Herold, “People who are always nice have a hard time in my model.”

Three Settings of a Simple Game
Herold has developed his theoretical model to explain experimental observations by other researchers. A specialist in game theory, he bases the model on a formal mathematical treatment of a simple game played by “haystack groups”—small collections of individuals familiar with each other. He applies the treatment to three versions of the game.

The first version includes rewards offered for cooperation. Individual “proposers” can choose to cooperate, and hence become eligible to receive a reward, or refuse cooperation; the other group, called “seconders,” can present a reward for cooperation or choose not to do so. “The proposer will cooperate if he’s confident that he’ll be rewarded,” Herold says.

In the second game only punishments are available, but the same conditions apply: Proposers can choose to cooperate and hence avoid punishment, or not do so and thereby face the possibility of punishment, while seconders can choose to punish noncooperators or not do so. “Punishment hurts both parties,” Herold says. “If the proposer believes he’ll be punished, he’ll want to cooperate. And if a lot of people punish, in the end everyone will. So you establish a norm in which everybody cooperates because they fear punishment, and you get very little punishment. But another equilibrium involves defection by everybody, and ultimately nobody is punished.”

The third game includes both punishment and reward options. “The model shows that the rewarding option is important to get you out of the situation in which everybody will defect because nobody punishes,” Herold explains. “Rewarders can go in fairly cheaply; they have to do it only when they get cooperation. Once enough rewarders are in most groups, the general environment will become one of cooperation; then punishment becomes cheaper.”

The Key Insight
Herold’s approach, which used mathematical methods from evolutionary biology, is not entirely new. “A couple of economists have previously applied this idea to the evolution of strategies or the evolution of preferences,” Herold recalls. However, he has expanded the field by applying the concepts to a specific setting. What’s new about the research project, he says, “is that it looks at both reward and punishment behavior.” And that, he continues, has provided “the key insight that we can try to exploit the fact that, if you want to change a norm, it’s cheaper to start doing so by rewarding people.”

The research originated in Herold’s interest in the foundations of behavioral economics, which he describes as “one of the interesting developments of the last decade; along with other people, I find it interesting to find additional constraints on behavior.” The project represented an effort to discover whether nonselfish preferences can survive.

More broadly, the research “is really about what evolutionary forces are in play to understand how reward and punishment at this level interact,” Herold says. He adds that the term “evolutionary forces” has a different connotation for economists than for life scientists. “It can mean simply successful preferences in certain markets—specifically preferences that are not all selfish,” he says.

Can policy makers and corporate leaders apply the principles revealed by the model when they want to make behavioral changes to their organizations? “I’m careful not to rush to conclusions,” Herold says.

“And you want to consider other things when deciding to apply the carrot or the stick. But it’s useful to consider these basic assumptions to change a culture: You start out by rewarding people who move to the culture you want to establish. Then once you have it, you start to punish those who haven’t made the change.” The key to success is timing the transition from reward to punishment. “You have to think in the long run about changing from carrot to stick,” Herold continues. “You have to consider when the carrot is more expensive or the stick is more expensive.”


Related reading on Kellogg Insight:

Healthy Choices: Do people prefer the carrot or the stick?

Featured Faculty

Faculty member in the Department of Managerial Economics & Decision Sciences until 2011

About the Writer
Peter Gwynne is a freelance writer based in Sandwich, Mass.
About the Research

Herold, Florian. 2012. Carrot or Stick: The Evolution of Reciprocal Preferences in a Haystack Model. American Economic Review, April, 102(2): 914-940.

Read the original

Add Insight to your inbox.
More in Economics