Spare the Rod, Spoil Society

Does punishing free riders increase long-run cooperation?

Ronald Bailey | December 9, 2008

Want to punish the fat cats on Wall Street who have allegedly wrecked the economy? Of course you do! And it's only natural, according to research done by two economists whose work focuses on the puzzle of human cooperation. As Swiss researchers Ernst Fehr and Simon Gächter noted in 2002, "Unlike other creatures, people frequently cooperate with genetically unrelated strangers, often in large groups." They argued that one significant key to cooperation is the existence of "altruistic punishment."

During the course of human evolution, people frequently engaged in cooperative activities such as big game hunting and the preservation of common property resources like fisheries. But it's all too easy for individuals to free ride on such projects. So how does true cooperation occur? Fehr and Gächter point to altruistic punishers: people who respond with strong emotion or even violence when someone else benefits from the labor of others without contributing something themselves. It may cost something to act as altruistic punisher, but going postal on non-cooperators does encourage everybody to contribute to the public good. The New York Times even speculated that this drive to punish free riders was behind the American public's disinclination to support the Congressional bailout proposals back in September.

To test their hypothesis, Fehr and Gächter set up a series of public goods experiments in which each player chose how much money to contribute to a joint investment—without knowing beforehand how much the other players will contribute. If everyone puts in a lot, they maximize their profits. However, the games were rigged such that non-cooperators could gain even more by taking a share of the profits while retaining their own initial endowments. The researchers found that when punishing non-cooperators was possible (say, spend $1 to reduce the free riders' endowments by $3), it substantially increased the amount that nearly all subjects invested in the public good. For example, in experiments done in 2000, where free riders could be punished, experimental subjects contributed two to four times more than when there was no punishment option.

However, Dutch experimenters noted that public goods games where punishment was allowed actually produced lower overall returns than did games in which no punishment occurred. Why? Because the destruction of the non-cooperators' resources was greater than the subsequent gains from cooperation. Punishment increases cooperation, but it also makes the group poorer. This is not a particularly inspiring outcome.

In a new study published last week in Science, however, Gächter and his colleagues show that the lower overall returns from games in which punishment is possible may be an experimental artifact resulting from the number of rounds in which the games are played. In most experimental conditions, public goods games are played for ten rounds or less. The new research compares the outcomes of games lasting for 10 rounds versus 50 rounds, both when punishment is possible and when it is not.

What happens when players have only ten rounds in which to invest? As satisfying as it is to punish free riders, the average payoffs after ten rounds are indeed lower than when no punishment is allowed. The results are quite different when the games last 50 rounds. Interestingly, the payoffs in the initial rounds when punishment is possible are lower than the payoffs when punishment can't occur. However, as rounds of play accumulate, the payoffs in the games where free riders can be punished rise rapidly, while the payoffs in games in which free riders are not punished drop throughout the duration of play.

Another happy result is that once players understand that they can be punished for free riding, they start investing enthusiastically in the common pool, causing the costs of punishment to drop to near zero. "Overall, our experiments show that punishment not only increases cooperation, it also makes groups and individuals better off in the long run because the costs of punishment become negligible and are outweighed by the increased gains from cooperation," conclude the researchers. "These results support group selection models of cooperation and punishment, which require that punishment increases not only cooperation but also group average payoffs."

So punishing free riders increases cooperation and boosts incomes over the long-run. But that is not always the case (at least in shorter-term games). Earlier this year, Gächter and his colleagues reported the results from a series of public goods games using players from 16 different societies. Their research turned up profound cross-cultural differences in response to punishment. All groups punished free riders, but the free riders did not all respond with increased cooperation. Instead, some sought revenge by punishing their punishers—if you whack me, I'll whack you, in other words. So a cycle of vendettas broke out.

For example, players from Muscat, Greece, and Saudi Arabia were the most vengeful. On the other hand, players from the United States, Australia, and Britain were the least vengeful and most likely to respond to punishment with increased cooperation. The researchers concluded that revenge is stronger among participants from "societies with weak norms of civic cooperation and a weak rule of law." Not surprisingly, the overall payoffs were significantly lower in the games in which participants indulged in cycles of vengeance.

We've come a long way from the bands of Pleistocene hunter-gatherers in which these psychological tendencies evolved. In today's complex economy, which encompasses globe-spanning webs of cooperation, how do people correctly identify free riders who merit punishment? Are the investment bankers with big bonuses free riders? What about hedge fund managers? Government agencies? Politicians? Perhaps the good news from experimental economics is that while Americans want to punish free riders as much as the next guys do, we are unlikely to engage in a self-defeating cycle of financial vengeance that will make us all poorer.

Ronald Bailey is reason's science correspondent. His book Liberation Biology: The Scientific and Moral Case for the Biotech Revolution is now available from Prometheus Books.

Comments

|12.9.08 @ 4:38PM|#

Punisher vs. Polar Bear

|12.9.08 @ 5:22PM|#

"What's to prevent free riders while encouraging real cooperation?"

Sorry, I'm not understanding the problem. If free riders did not prevent real cooperation in our past, where's the problem? Sure some academic economists get their panties in a twist that the real world doesn't fit their nice neat model, but so what? As the story shows, our ancient ancestors could deal with it, so why can't we?

Pac|12.9.08 @ 5:28PM|#

Punisher vs. Polar Bear

Damn you beat me to it.

C'mon Reason where's the Ennis/Dillon love

Fluffy|12.9.08 @ 5:33PM|#

Fehr and Gächter point to altruistic punishers: people who respond with strong emotion or even violence when someone else benefits from the labor of others without contributing something themselves.

By this definition, libertarians seem to be altruistic punishers. And objectivist libertarians would seem to be the biggest altruistic punishers of all, although I'm sure they would bristle at such a description.

However, Dutch experimenters noted that public goods games where punishment was allowed actually produced lower overall returns than did games in which no punishment occurred. Why? Because the destruction of the non-cooperators' resources was greater than the subsequent gains from cooperation. Punishment increases cooperation, but it also makes the group poorer. This is not a particularly inspiring outcome.

I have a problem with this part of the article, though. The methodology seems off. If they are measuring the total group income at the end of each round, including the income of people who attempted to free ride, that doesn't seem like the right methodology to me. The "destruction of the non-cooperators' resources" is a feature, not a bug. We should be measuring the total income of the set of cooperating persons, and excluding from consideration the set of noncooperating persons, when performing relative calculations. In addition, the total group income returns are skewed by the fact that resources spent on punishing non-cooperators are treated as a reduction in income, instead of as a luxury expense. If as noted at the beginning of the article, there are persons who are powerfully driven emotionally to undertake these punishments, then the people who spend $1 to punish haven't experienced an income drop - they spent some income on an activity they find pleasurable. So the real group income we should consider is "[total income of set of cooperating persons] plus [add back in the funds expended by cooperating persons on the pleasurable activity of smiting the non-cooperative]".

|12.9.08 @ 5:58PM|#

The net effects of this behavior on societal welfare are still ambiguous.

People can cooperate to do some pretty evil/stupid stuff. The "free-riders" in those cases just happen to be called "draft dodgers" during popular wars, "kulaks" during coop-farming schemes, "dissidents" under one-party states, "hoarders" under the New Deal, etc.

They were all punished or shunned for being uncooperative.

|12.9.08 @ 5:58PM|#

Fehr and Gächter point to altruistic punishers: people who respond with strong emotion or even violence when someone else benefits from the labor of others without contributing something themselves.

By this definition, libertarians seem to be altruistic punishers. And objectivist libertarians would seem to be the biggest altruistic punishers of all, although I'm sure they would bristle at such a description.

I thought about this as well. However, I wonder how much of this is dependent on perception. What happens if the someone thinks the other person is free riding even if isn't so?

A common perception among many is that the upper management-types in a company don't do "real" work. So they are perceived to be getting a free ride. So when tax time comes around, for certain people, they want not just to have those types taxed fairly but punished for their perceived lack of contribution.

|12.9.08 @ 6:06PM|#

Fluffy-

Are folks like Irwin Schiff, Joe Bannister, Bob Schultz, Wesley Snipes et al free riders? In other words, are incone tax protestors free riders? What would you call the IRS agents, assistant united states attorneys and federal district court judges who investigae, prosecute and jail the income tax protesters? ALtruistic punishers? Ditto for the Treasury dept. employees, state revenue employees and all others who dutifully parrot the Leviathan's line that "we all most contribute our fair share"?

|12.9.08 @ 6:13PM|#

Fluffy-

Naturally, I don't have to tell you what I think.

Some altruistic punihsers just don't get the reality that the commons are wasted on pursuing compliance with the cooperative objective. Of course, others know this and don't care because they are paid quite well to be altruistic punishers. Yes, some of them do get a boner out of being an altruistic punisher-you are right about that.

perilisk|12.9.08 @ 6:41PM|#

"By this definition, libertarians seem to be altruistic punishers. And objectivist libertarians would seem to be the biggest altruistic punishers of all, although I'm sure they would bristle at such a description."

On the other hand, unless it results in increased cooperation, altruistic punishment is negative-sum, irrational, and useless. Withholding handouts from bums might cause them to re-evaluate their priorities, withholding handouts from the severely disabled is likely to result in unnecessary tragedy. Of course, intuition tends to work well in most cases -- it's only in complicated systems where people may be mistakenly identified as free riders that problems emerge. Even there though, the existence (or persistence, rather) of such a tendency suggests that erring on the side of "I don't understand the details, but this guy is screwing me" tended to pay off more often than not.

|12.9.08 @ 7:16PM|#

Are folks like Irwin Schiff, Joe Bannister, Bob Schultz, Wesley Snipes et al free riders?

No they're not free riders, because tax collection is not economic production. BUt they're they're still insane kooks. Snipes got duped and rolled, but the others are the ones who duped and rolled him. They make their money telling people how to win in court then act all surprised when they themselves lose in court. Tax deniers are almost as pathetic as troofers.

|12.9.08 @ 7:16PM|#

And objectivist libertarians would seem to be the biggest altruistic punishers of all, although I'm sure they would bristle at such a description."

No, someone who really understands Objectivism would never behave altruistically. They would seek to maximize their long-term results, with whatever strategy of punishing versus cooperation resulted in maximizing their gains.

|12.9.08 @ 7:36PM|#

Brandybuck-

Irwin Schiff has had the stones to stand up to the Leviathan. He has stuck his neck out and has paid-with his freedom. Tax denier, tax protester or kook, he has taken action. He has entered the arena in pursuit of a cause that is truly noble. If his cause prevails, you win. But, I guess it feels good to sit there and call guys like him kooks.

|12.9.08 @ 7:42PM|#

Bradybuck-

Nothing is more insanely kooky than buying a government conspiracy fable that 19 jihaddis with box cutters...

|12.9.08 @ 9:06PM|#

libertymike,

He has indeed taken action, but it was silly action. I have every respect in the world for people who fight against taxes. But people like Schiff shit all over them. He doesn't fight taxes, he sells books to people trying to convince them that they can legally stop paying taxes and win in court. That is a lie. But is he any sillier than his clients? After all, they continue to beleive his nonsense even after he LOST in tax court. It's like buying a Get Rich Quick book from homeless bum.

Fact time: The 16th Amendment was properly ratified. The IRS was properly instituted. "Income" has been properly defined. You are legally obligated to by taxes on your wages, salaries and tips. I've got a gazillion links to proof of this, but I know you won't read any of them so I won't bother.

|12.9.08 @ 9:28PM|#

if i'm working hard to contribute to a societal benefit and you're getting a free ride, at some point i'm going to feel like a chump. the hazard in this sequence is that i might withdraw my cooperation. i'll still be ok on my own, but some of you might starve, so, take care not to make me feel like a chump.

|12.9.08 @ 10:44PM|#

There's some evidence that altruistic punishment is not strictly altruistic, in that is may be a form a of social signaling and reputation building.

In other words, punishing free riders could be a means of displaying what a zealous cooperator you are.

Also it's troublesome that the same altruistic punishment mechanism is used to enforce all kinds of social norms, including those that have nothing to do with free-riding.

It's basically akin to gay-bashing. Going out of one's way to "altruistically" punish those darn gays for violating sexual norms. All for the good of society, of course. Can't have those people degrading our morals.

|12.10.08 @ 6:45AM|#

Aside from the genuine issue of the need to avoid the adverse impact of free-riders on the willingness to cooperate (see the classic David Axelrod book, 'The Evolution of Cooperation'), it seems that the net pay-off definition in this study is mis-specified. The penalties paid by the cheaters are not necessarily a deadloss: they have to be offset against the reduced volume of activity. Those dollars/pounds/euros do not just evaporate.

Is the study's adverse conclusion on the need for social coordination robust against this recalibration? Maybe yes, maybe no. At some point, free-loading must overwhelm the system.

Neu Mejican|12.10.08 @ 10:49AM|#

At some point, free-loading must overwhelm the system.

Indeed.

Look to New Zealand as an example in recent history.

|12.10.08 @ 1:26PM|#

I have a problem with this part of the article, though. The methodology seems off. If they are measuring the total group income at the end of each round, including the income of people who attempted to free ride, that doesn't seem like the right methodology to me. The "destruction of the non-cooperators' resources" is a feature, not a bug.

You're mistaken. The methodology is correct, since you're talking about wealth maximization. Any destruction of resources WILL make everybody worse off, not better off. What the test is trying to find is how to make everybody better off through cooperation, even though they are looking at a bogus problem: the "free-rider" problem. Under a strict property rights economic system, there would not be free riders.

|12.10.08 @ 1:45PM|#

Neu,
"At some point, free-loading must overwhelm the system".

Indeed.

Look to New Zealand as an example in recent history.

New Zealand has purely public goods??

|12.12.08 @ 8:42AM|#

Crime and punishment...Sounds a lot like the "deterent effect" regarding crime. Crime exists without consequences and is diminished when punishment is swift, certain and severe.
incentive, reward...human nature?