Bryan Caplan  

Behaviorism in Economics: A Funeral

PRINT
Nuclear Power Regulation... International Health Care Comp...

One reaction to my recent piece in Econ Journal Watch is "economics isn't about what people say or believe; it's about what people DO." The easy response is: Not anymore, it isn't! Survey research has exploded in economics. So has the whole field of Psychology and Economics. If economics is what economists do, then what people say and believe is now economics.

It is pretty obvious that the latest generation of economists has lost - or never had - interest in methodological injunctions against studying anything other than "observable behavior." Check out Timur Kuran's cool work on preference falsification. But was there ever any reason to abide by this restriction? It is grossly counter-intuitive. I know by introspection that my beliefs affect my behavior, and I know by experience that asking people what they are doing is often informative. So how did a doctrine so contrary to common sense ever become conventional wisdom?

All the economists I asked traced behaviorism in economics back to Milton Friedman's essay, "The Methodology of Positive Economics." (Here's an excerpt). Upon re-reading it, I was surprised to learn that he barely touches on the issue. Virtually the whole essay is about the alleged impossibility of testing assumptions, as opposed to predictions. His central complaint about survey research is NOT "economics is about what people do, not what they say." Instead, his central complaint is that surveys are yet another futile effort to test assumptions.

Friedman's only specific arguments against asking people are "throw-aways":

1. "The evidence cited [against profit-maximization] is generally taken either from the answers given by businessmen to questions about factors affecting their decisions - a procedure for testing economic theories that is about on par with testing theories about longevity by asking octogenarians how they account for their long life - or from descriptive studies of the decision-making activities of individual firms."

2. "Little if any evidence is ever cited on the conformity of businessmen's actual market behavior - what they do rather than what they say they do - with the implications of the hypothesis being criticized..."

3. "I do not mean to imply that questionaire studies... are useless... They are extremely valuable in... constructing new hypotheses or revising old ones. Whatever their suggestive value in this respect, they seem to me almost entirely useless as a means of testing the validity of economic hypotheses."

Pretty flimsy stuff. Point by point reply:

1. I'd say asking businessmen what they do is a lot more "on par" with asking a satellite installer how he installs satellite t.v. systems. Listen up, he probably knows the answer. Of course, many people digress when you try to pump them for information, but the common sense solution is to keep questioning them to keep them on track. Don't lead the witness, but discourage him from off-roading.

2. I doubt Friedman's charge was ever correct. But suppose it was. The solution to one-sided focus on talk is not one-sided focus on behavior. The solution is doing both, and taking both seriously.

3. How can asking people be so useful for getting new ideas, but so useless for testing existing ideas? It's not impossible, but highly implausible. If people have insightful new things to tell us, they probably have informative old things to tell us too.

Admittedly, Friedman's essay was not primarily a defense of behaviorism. I've heard better arguments. The leading candidates:

1. People lie.

2. People make mistakes.


Both are true, but it is a weird over-reaction to use these inadequacies to excommunicate talk from economics. My rebuttal:

1. Yes, people lie. But there are times to raise your eyebrows, and times to take people at their word. Two big rules:

Be skeptical of self-serving statements. "Are you a hard worker? Yes!"

Doubt "socially desirable" responses.
"Do I maximize profit? No, my workers are my first priority!"

2. Yes, people make mistakes. But some people are a lot more likely to make mistakes than others. That's why, for example, you should:

Give less credence to hasty and emotional reactions. "Do I maximize profits? I get up every morning and bust my hump all day, unlike you and the other eggheads!"

Trust experts more than laymen, and people with first-hand experience more than people without it. An engineer I know frequently accuses businesses of backroom collusion - except in his own business, where he tells story upon story about cut-throat competition. I discount the first class of statements, but largely buy the second.

In the past decade, behaviorism in economics died. It's time to give it a proper burial.

I hate to speak ill of the dead, but duty calls. Behaviorism had a lot of smart adherents, but their arguments on its behalf were lame from the start. Furthermore, I strongly suspect that even in its heydey, a lot of economists didn't believe it, but were too scared to say so.


Comments and Sharing





COMMENTS (11 to date)
simon writes:

Byran,

It is interesting that economists are entering the world of psychology and marketing. In doing so they are joining a field with a large installed base of research. In making the move they are clearly second class citizens given that they are now just renters.

Arianto writes:

Why not combining both: stated preference approach and revealed preference approach as has been done in the areas like environmental economics, health economics, etc.

Arianto

Bill Woolsey writes:

Survey data proves that the law of demand doesn't apply to labor. A higher minimum wage expands employment! Hmmmm.....

Steve Miller writes:

"Survey data proves that the law of demand doesn't apply to labor. A higher minimum wage expands employment! Hmmmm....."

That probably falls under Bryan's caveat of discounting the beliefs of non-experts.

Randy writes:

In my experience, the "experts" nearly always have an agenda.

"Lies, damn lies, and statistics."

Tom writes:

"I know by introspection that my beliefs affect my behavior, and I know by experience that asking people what they are doing is often informative. So how did a doctrine so contrary to common sense ever become conventional wisdom?"

Yes "beliefs affect behavior" and "asking people what they are doing is often informative." But stated beliefs don't reliably affect behavior, and people often don't give informative answers. Most people say, for example, that they oppose government spending, but most of those same people will scream like mad when the programs they favor are threatened.

The reliable prediction of economic choices on the basis of expressed beliefs or attitudes requires a degree of skill in posing questions that is beyond the ability of most surveyors. The rare, skillful survey is so intrusive or annoying as to deter all but the two-sigma cases who enjoy responding to surveys. That is to say, surveys are likely to produce either garbage or unrepresentative views.

Talk is cheap, inconsistent, and often at odds with behavior. The only reliable way to understand behavior is to observe behavior.

As the old saying goes (revised slightly to fit the occasion): Don't believe a word I say, just watch what I do.

"3. How can asking people be so useful for getting new ideas, but so useless for testing existing ideas? It's not impossible, but highly implausible. If people have insightful new things to tell us, they probably have informative old things to tell us too."

That's sloppy reasoning. Here's why: "Asking people" can suggest testable hypotheses, which can be tested only by collecting data about economic behavior. But, as I explain above, "asking people" isn't a valid way of collecting data with which to test hypotheses.

Joshua Allen writes:

"My beliefs affect my behavior"

Any good hypnotist knows that this is only partially true. Beliefs affect behavior, yes; but the human mind is incredibly good at inventing beliefs and rationalizations to justify behavior that is initially spurred by subconscious. In fact, most decisions are made by the subconscious first; and then the conscious mind constructs rationalizations to explain the decisions. As such, a study of human beliefs is not going to yield much more than a bestiary of the excuses that people invent for themselves.

Kevin Brancato writes:

I love surveys -- whether of beliefs or activites.

Your practical rules of skepticism are prudent, but they add a subjective component to the analysis of survey data. This will put one's prior beliefs about 1) the survey topic, and 2) the reliability of surveys, in a very dominant position over the survey evidence, to the point of making the survey almost worthless in some instances.

Skepticism is not an objective rule; skepticism is subjective judgment. Coming to a consensus about "what survey results mean" becomes difficult when one person sees a survey result as an objective reflection of actual beliefs of the survey frame, and another person sees the result as reflecting the unmeasurable lies and misstatements of respondents.

Take an example: a fictional allegedly representative sample of businessmen finds that on average, they will fire 2 out of 10 minimum wage workers if the minimum wage is raised to $10.00.

Economist R finds this reasonable -- perhaps a bit of an underestimation. Economist L finds this an exaggeration of gross proportions. L and R have no means of coming to a consensus, and the survey results do not "correct" either prior.

Ross Emmett writes:

The real methodological statement of "behaviorism" in economics is not the Friedman essay, but the Stigler/Becker piece "De Gustibus Non Est Disputandum" (AER). Their argument is that, confronted with a change in a person's behavior, economists should look at changes in the person's constraint set (usually observable and measurable items like the price of other goods) rather than speculate about possible changes in the person's preferences. The methodological rule that preferences generally stay constant and are universal across all people. A predictive economics requires such a rule.

Naturally, many of us are uncomfortable with an excessive dependence on the notion of preference stability, but it does help us avoid some of the more common errors that other social sciences fall into: such as the trap of thinking that culture is all that matters.

Steve Miller writes:
IIn my experience, the "experts" nearly always have an agenda.

They're often accused of having an agenda, and sometimes they do, but there are ways to test claims of expert bias. In the case of economists, the most common claims of expert bias don't hold up.

Lawrance George Lux writes:

Bryan,
A graduate engineer friend of mine did a Graduate paper, in which he claimed all Businesses surveyed did not set Piece-Rates with an internal Capital return. I knew the Business personnel interviewed, and their Business practice; maybe it is only I who wants to make a living. lgl

Comments for this entry have been closed
Return to top