In an apparent case of libertarianism run amok, two cognitive scientists and one economist have declared (in effect) that proponents of income redistribution are always and everywhere motivated by exactly the same psychopathic impulses that lead people to scratch other people’s cars, break streetlamps, and engage in other forms of senseless destruction.
That’s not a conclusion; it’s their starting assumption. Then they do an experiment in which some subjects choose to engage in a mild form of income redistribution — and conclude that psychopathic behavior is frighteningly widespread.
Now I too am generally skeptical of the impulse to redistribute other people’s income, but even I concede that redistribution is rarely entirely senseless and pretty much never entirely destructive. In particular, the experimenters consider a case where money has just been redistributed and subjects are given the opportunity to (partly or fully) reverse that redistribution. Those who exercise that opportunity are labeled “destructive”. So apparently, in order to avoid the “destructive” label, you must passively accept any arbitrary redistributions imposed by authority figures without ever feeling any impulse to redistribute in the opposite direction.
In the experiment, each subject is initially awarded 1000 tokens. The paper is so poorly written that it’s hard to be sure, but I think these tokens are ultimately exchangeable for money; it’s mentioned that the average subject earns the equivalent of nine British pounds. One subject is then given the opportunity to destroy some of the other subject’s tokens, thereby returning some funds to the experimenters or to the taxpayers who are funding this experiment.
Rather shockingly (to me) only about 15.5% of the subjects exercised this option; given the presumption of public funding, I’d have hoped for a higher number. The experimenters, however, want to conclude that these 15.5% are the sorts who are likely to key your car.
This stuff is infuriatingly stupid. Even more distressingly, Robin Hanson, who is much much much too smart for this, seems to have fallen for it.
Let’s be clear on this: Destroying claims on wealth (thereby transferring or perhaps returning those claims to someone else) is not the same thing as destroying wealth. The right experiment would invite subjects to destroy not tokens but, say, candy bars. Why do the wrong experiment when you can just as easily do the right one? Maybe because you’re afraid the right experiment won’t yield the results you’re looking for.
I agree you are exactly right to distinguish between claims on wealth and actual wealth.
But you also seem to realize that the inability to distinguish them is extremely, distressingly common. (If anyone doubts that, just read some of the comments that have followed the repeated publication of Steve’s classic Slate piece on Scrooge.)
One might guess that the subjects in the experiment would also fail to distinguish between claims on wealth and real wealth. Which makes it quite plausible that they would have destroyed candy bars, just as they destroyed tokens. That is what Robin Hanson predicts, and I’m not sure he’s wrong. (Of course you are right that the experiment should have been done that way in the first place.)
The authors state that their game has eliminated all the possible motives for destruction. SL has provided a motivation they have not removed.
The authors conclude that the destructors indulge in this behaviour because they enjoy it for some reason. You suggest that they are redressing a redistribution of wealth from the public to the individual.
I am inclined to agree with ed that the experimenters probably have it right that most of the subjects did not make the distinction between claims of wealth and wealth. (It is frustrating that the experiment could have been designed to remove this uncertainty, particualrly as it was conducted by the economics dept. rather than the psychology dept.) My view is not just a hunch – there is some evidence for it.
In the introduction, they cite previous work where destructive behaviour increases when the destruction is “hidden” by a chance that destruction would have happened anyway (i.e. by “nature”). This is difficult to explain if your proposed motivation is correct.
Also, the subjects “earned” their tokens by completing a questionaire, so there reduces the arbitrary redistribution element. However, the payment was “three times the student salary in Hungary”, so it is likely that there was at least some awareness that the payment was larger than had actually been earned.
Also, 8.7% destroyed 20% and only 6.8% destroyed 40%. If motivation were to redress the re-distribution, why only destroy 20%?
Crucially, they do not conclude that “psychopathic behavior is frighteningly widespread.” In fact there was no correlation with any personality type, and particularly with psycopathy.
I don’t know if this is true, but Robin Hanson said that in previous experiments when real resources are destroyed, the results tend to be the same. If this is the case, it would be useful for experimenters to include this in the introduction and literature review.
So while it is not absolutely conclusive, it does seem far more likely that the authors attribution of motive is the correct one.
I agree with the previous comments that it’s not unlikely that participants did not realise that by destroying tokens they were returning money to taxpayers.
It’s hard to tell (although I didn’t read very closely) but did the participants all earn the same amount? If so, what’s the point of saying that they earned it by taking the test? I don’t think the participants felt like they earned the money by taking the test but rather by participating in the first place, not sure if this changes anything but it feels like it should.
I think everyone earned the same – about £10. Some then had either 0%,20% or 40% removed by another participant, or by chance. The maximum anyone could lose was 40%. They were given the tokens after the test as “payment”, but it could have been payment for taking part, I don’t think it matters. I assume that the pay motivated the participants to take part, so it could be called remuneration for work done.
This kind of stuff is why TBQ is the first blog I read each day.
I agree that this experiment doesn’t offer much supporting evidence, but I think they hypothesis that redistributive decisions are influenced by similar mental systems to the ones that make us key someone’s car. By this I mean the feeling of “screw your economic arguments, I don’t care what results in the most total welfare, this is wrong!” where moral outrage trumps efficiency or any of the other goals economists generally want to strive for, especially quantitative ones. Living in a town full of hippies I see this attitude on display frequently, and I suspect it’s part of this deep seated feeling pushing people to defend their in-group against things that seem threatening or hard to understand but that might reasonably be called psycopathic.
But of course that’s just speculation based on observation. If I were actually going to call myself confident of that position I’d hope for some real experimental evidence, and this study clearly offers nothing of the kind once you actually read it.
While certainly most people do not recognize the distinction between claims on wealth and real wealth in the context of an intellectual discussion, there is a very real instinctive difference between destroying “tokens” and destroying real food such as candy bars. Even burning real currency would be better (if legal), as people use currency as their reference point for wealth so much, that it is easy to confuse currency for wealth. I agree with Steve that it is surprising that *only* 15% agreed to burn “tokens,” though not to reverse the redistribution, but rather as a cheap thrill with no real cost. Most people, I think, understand intuitively that since you’re only burning a token, nothing of true value is being destroyed.
Steve, have you considered the possibility that the insight that occurred to you, as an economist, occurred to neither to the 15.5% nor to the other 84.5%? I think most people think that claims on wealth actually ARE wealth. They think that burning a dollar bill actually makes the world poorer.
And you’re also making the point that the study is publicly funded so you are depriving taxpayers of wealth. This is another point that I don’t think would occur to most people. I think thy would behave in almost exactly the same fashion if they though the rest of the money (which wasn’t used in the experiment) would be spent on a completely unproductive activity which benefited no one.
Keshav,
I think most people think that claims on wealth actually ARE wealth.
Of course, this is false. If the average person were presented with a choice to burn a suitcase with $350,000 in it or a $350,000 house, all would choose the suitcase.
Keshav,
All implicitly understand, even when not explicitly conscious of it, that burning the suit case is just burning little pieces of green linen and cotton, whereas burning down a house is burning down someone’s home. In other words, all the arguments that people may not be consciously aware of the differences between wealth and claims on wealth at the time of some decision, if you asked them what the difference is, they can tell you, and you can set up an experiment that shows this understanding.
No one earns money in order to have a room full of green cotton and linen. All earn money to buy other things.
TalF: the tokens were exchanged for real money – by destroying tokens they were depriving people of money. Or, more accurately, they were preventing the transfer of real money from the institution to the individual.
Maybe it’s just more fun for people in these make-believe settings involving piddly stakes to burn stuff?
Call me skeptical of economic ‘lab experiments’ in general. More often I’ve heard of studies implying people are surprisingly altruistic and value ‘fairness’. An obvious concern in those cases is whether people tend to behave the way they think an observer would want them to behave. (In this case, as SL suggests, it’s not entirely clear whether the subjects think they’re redistributing or reversing redistribution?) Freakonomics talked about a Chicago prof named John List who’s done actual ‘field research’ like observing people at trade shows — sometimes when they *don’t know* they’re being observed — which contradicts some of these previous results. If I had to bet (tokens or paper) I’d go with the field over the lab.
Assuming the average person is not a homebuilder.
I grudgingly concede Landsburg’s concern about the experimental design. (Robin Hansen also concedes the point in the comments following his post.)
I share the suspicion of other commentors that the experiment would likely produce similar results even if the glitch were fixed. But I’m also impressed with the arguments that people intuitively distinguish between the destruction of valuable objects and the destruction of currency equal to the value of those objects. So I guess I can’t dismiss Landsburg’s concern as quibbling.
What would we conclude from the (speculative) finding that 15% of people might act this way? Perhaps there’s something adaptive about a population having this degree of in-group loyalty? (Just as it may be adaptive to have some percentage of the population that is predisposed to deviate from the demands of in-group loyalty….)
This in-group thing is speculative – the people deprived of the transfer were part of the same in-group, first year students. The motives for the destroyers could well be different from the car-keyers. The motive attributed to them by the authors is also speculative – the experiment did not test this. The destroyers did not correlate to psycopathy – the only weak correlation was to liking new experiences, which could be a valuable trait.
Ken responding to Keshav,
Keshav: “I think most people think that claims on wealth actually ARE wealth.”
Ken: Of course, this is false. If the average person were presented with a choice to burn a suitcase with $350,000 in it or a $350,000 house, all would choose the suitcase.
While Ken’s point may be correct, I’d need stronger evidence than this. It is EASIER to burn a suitcase of cash than a house. It is SAFER to burn a suitcase of cash than a house. It is likely LESS LEGALLY DANGEROUS to burn a suitcase of cash than a house. Any of these reasons, or others, could explain this preference without positing it as evidence people distinguish real value from mere markers of value.
It’s odd that the authors didn’t ask the participants why they destroyed the tokens. We are just guessing as to why they did. Why not just ask?
Maybe they didn’t understand the rules. (This strikes me as a very real possibility because the game as structured doesn’t make much sense. Perhaps they thought there was some subtlety or trick they weren’t told about and choosing destructin would have effects we haven’t thought of. This explains some of Kahneman and Tversky’s findings because subjects read more into their experiments than the bare bones they were given.) Maybe they were bored. Maybe they actively disliked the passive player for some reason. Who knows? Why not ask?
Zazooba, Steve has a great deal of suspicion concerning people’s self-reflections about their own motives.
My instinct is similar to Zazooba’s. Maybe the people burned the tokens because they didn’t know the rules. Maybe they thought it was a multi-turn game and that by burning they were somehow protecting themselves. Even if the rules were clear, maybe they *suspected* there was a hidden trick and they tried to be protective of themselves. I’ve played enough of these lab games to assume there may be a twist and not take the rules at face value.
A perhaps off-topic comment: These comments are so much more civilized, and insightful, than those typically appearing on, say, Brad DeLong’s blog. I would enjoy some responses from SL.
We must have read different papers. The one you linked to had subject pairs with 1000 tokens each. One of the two (randomly chosen) could choose to “destroy” 0%, 20% or 40% of the other’s tokens. Although the authors don’t spell it out, I take destroy to mean that the tokens were taken away from the player and not turned over to any other player. Thus, I could (if assigned the role) choose to reduce your payment (tokens became cash) by 20% or 40% if I wanted to for no monetary benefit for me.
As one who runs laboratory economics experiments, I would rather like those who choose to destroy. They redistribute income from subjects to me (ok, my budget).
David Wallin: Why do you say we must have read different papers? You are saying what I am saying — those who choose to “destroy” are not in fact choosing to destroy; they are choosing to redistribute, which is not at all the same thing.
Zazooba and Economiser. It seems the authors agree with you. Note the last sentence in the paper:
“It could also be enlightening to explictly ask subjects to state, a posteriori, their reasons to senseless (omitting the label) destroy their partners’ earnings.”
Steve, I goofed up. Amongst interruptions one day, I managed to read your introduction and the paper. I responded thinking I had read all your post. I was wrong. Had I done this right (and I should have looked further given my priors n your accuracy), I would have said:
This is a brilliant observation. So often those of us who create laboratory economies think the entire economy contains only the subjects (and that simplifying assumption may often be ok). But, clearly here, the economy is at least the subjects and the experimenters. The experimenters think they are running a nonzero-sum game (just the subjects), while, as you argue, the larger game (subjects and experimenters) is zero sum. The subjects destroy nothing—they transfer back to the experimenters.
I think about how I would fix this. Certainly, we couldn’t give each subject 10, $1 bills and shred $0, $2, or $4 or one subject’s stash (based on the choice of the partner and nature). That just puts us in the bigger economy, where the “destruction” of $4 given to a subject does not affect the partner or experimenter (well, slightly as follows), but one person loses $4 and all holders of dollars gain an aggregate $4.
I guess you’d be stuck with giving 10 items of value, say candy bars, to each subject. Then, you’d destroy 0, 2, or 4 or one subject’s sugary assets. This, of course, raises issues money generally doesn’t (dominance and satiation at least).
This is relevant here as the subject that can destroy does so at no cost to himself. I suspect this argument of wealth transfer back to the experimenters is less important when the subject pays a cost to redistribute. Overall, excellent point, and sorrow for the premature criticism.
Stephen, this isn’t the first time I’ve seen you object to experimental methods/interpretations because they ignore the experimenter as a player. I’ve never seen solid evidence that participants in experiments consider the experimenter as a player. Why are you so sure that results would change if the experiment design took that possibility into account? Is this a hunch, or if not, what evidence do you have?
I am honestly curious. My prior is the opposite of yours (and there are many experiments that I think would have turned out differently if you were right) but I don’t recall much *direct* evidence from a comprehensive, careful study of such social preferences towards the experimenter.
Err, apologies for misspelling your name there!
Vera:
Is this a hunch, or if not, what evidence do you have?
I suppose that if you restrict the context to actual laboratory experiments, it’s a hunch. But it’s an informed hunch, based on a lifetime of casual observations of analogous situations. Generally speaking, I’ve found that people tend to think that destruction is a bad thing when it serves no social purpose, but can be acceptable when there’s a large enough offsetting benefit.
For example: Many people think it’s a good thing that airport security teams destroy stray luggage that’s suspected of housing bombs. I am quite confident that few people would consider it a good thing to destroy stray luggage for no purpose other than the destruction itself.
For another example, politicians frequently campaign against wasteful government spending. I think it’s a fair inference that those politicians, who have some expertise in the matter, believe that people don’t *like* wasteful government spending, even though there might be much government spending of which they *do* approve.
Those sorts of observations convince me that people generally have very different feelings about social waste on the one hand versus transfers on the other hand. So to do an experiment that measures attitudes toward transfers, and interpret the results as having anything at all to do with attitudes toward waste, seems to me to be utterly bizarre.
OK, in the context of this particular experiment, I agree with you that social preferences towards the experimenter need to be taken into account in order to prove their preferred interpretation of the results. (I didn’t carefully read the experimental methods but it’s pretty clear that it *should* be considered as an alternate possibility, because the entire point of the paper crucially rests on an assumption of no such social preferences towards the experimenter, or as you say, equivalence of wealth and claims on wealth.) But, very few experiments hinge so strongly on that assumption, and it seems quite unlikely to be a strong factor generally (even here, still only 15% exhibit such preferences, and a large fraction of those are probably confused or screwing around or who knows what; there’s a lot of noise and crazy behavior in the most straightforward of games).