Milton Friedman and the Rational-Agent Model

Paul Krugman's analysis of Milton Friedman's intellectual legacy is one of the best articles I've read recently. Krugman not only paints a balanced portrait of Friedman's accomplishments - great economist, bad popularizer - but ably summarizes the rival tensions in 20th century economics. It's all fascinating stuff, but I was particularly interested in this section on the rational agent model:

For most of the past two centuries, economic thinking has been dominated by the concept of Homo economicus. The hypothetical Economic Man knows what he wants; his preferences can be expressed mathematically in terms of a "utility function." And his choices are driven by rational calculations about how to maximize that function: whether consumers are deciding between corn flakes or shredded wheat, or investors are deciding between stocks and bonds, those decisions are assumed to be based on comparisons of the "marginal utility," or the added benefit the buyer would get from acquiring a small amount of the alternatives available.

It's easy to make fun of this story. Nobody, not even Nobel-winning economists, really makes decisions that way. But most economists--myself included--nonetheless find Economic Man useful, with the understanding that he's an idealized representation of what we really think is going on. People do have preferences, even if those preferences can't really be expressed by a precise utility function; they usually make sensible decisions, even if they don't literally maximize utility. You might ask, why not represent people the way they really are? The answer is that abstraction, strategic simplification, is the only way we can impose some intellectual order on the complexity of economic life. And the assumption of rational behavior has been a particularly fruitful simplification.

The question, however, is how far to push it. Keynes didn't make an all-out assault on Economic Man, but he often resorted to plausible psychological theorizing rather than careful analysis of what a rational decision-maker would do. Business decisions were driven by "animal spirits," consumer decisions by a psychological tendency to spend some but not all of any increase in income, wage settlements by a sense of fairness, and so on.

But was it really a good idea to diminish the role of Economic Man that much? No, said Friedman, who argued in his 1953 essay "The Methodology of Positive Economics" that economic theories should be judged not by their psychological realism but by their ability to predict behavior. And Friedman's two greatest triumphs as an economic theorist came from applying the hypothesis of rational behavior to questions other economists had thought beyond its reach.

It doesn't take a degree in neuroscience to realize that the rational agent model is deeply flawed. From the perspective of the brain, Keynes' arm-chair psychologism was far more accurate than Friedman's idealized assumptions. But even more damaging to economic science was Friedman's attitude that the validity of the economic assumptions was largely irrelevant. As long as economists could make reasonably accurate predictions about economic phenomena - like the rate of unemployment, or the value of the dollar - then our bewildering behavior in the real world was largely irrelevant. (Friedman often used Quine to justify his position.)

But now this naivete is beginning to crumble. There are two reasons. The first reason is neuroeconomics, which has its roots in behavioral economics, which is similarly rooted in prospect theory. David Laibson, an economist at Harvard, ably summarized the philosophy underlying neuroeconomics: "Natural science has moved ahead by studying progressively smaller units. Physicists started out studying the stars, then they looked at objects, molecules, atoms, subatomic particles, and so on. My sense is that economics is going to follow the same path. Forty years ago, it was mainly about large-scale phenomena, like inflation and unemployment. I think the time has now come to go beyond the individual and look at the inputs to individual decision-making. That is what we do in neuroeconomics."

The second reason economics is beginning to probe the limitations of the rational-agent model come from attempts to explain empirical phenomena that neo-classical macroeconomics can't explain. I won't go into the details here, but for those interested George Akerlof gave a terrific Presidential Address to the American Economic Association this year. His basic theme was that Keynes' armchair psychology was often more accurate in explaining consumer behavior than the intricate, mathematical models favored by neo-classical economists. It's time for economists to rediscover the mind:

This lecture has shown that the early Keynesians got a great deal of the working of the economic system right in ways that are denied by the five neutralities. As quoted from Keynes earlier, they based their models on "our knowledge of human nature and from the detailed facts of experience." They used their intuitions regarding the norms of how consumers, investors, and wage and price setters thought they should behave. There is systematic reason why such knowledge and experience is likely to be accurate: by their nature norms are generated and known by a whole community. They are known to those who abide by them, and those who observe them as well.


More like this

I feel the same way and get irritated when economists tout these so-called rational actors.

Economics should figure out better ways of predicting irrationality and their consequences, as long as consumers are influenced by advertisement, and not reality, it will be more accurate.

A third contribution comes from game theory. In particular, I'm thinking of Gintis book Game Theory Evolving. He goes "beyond Homo economicus" to model "Homo egualis", "Homo reciprocans", and "Homo parochius". It's a clear discussion of experimental results regarding altruism, social norms, and us vs them.

By Rick Thomas (not verified) on 29 Jan 2007 #permalink

Per Krugman, "You might ask, why not represent people the way they really are? The answer is that abstraction, strategic simplification, is the only way we can impose some intellectual order on the complexity of economic life." The big challenge for behavioral and neuroeconomics is not to simply elucidate the specific mechanisms of action, but to use those findings as a basis for developing models that can do a better job making predictions about aggregate behavior. Extrapolations from behavioral and neurological models need to be empirically verified in order to demonstrate accuracy and generalizability. "Armchair psychology" is just as dubious as unrealistic assumptions of rationality in the absence of empirical verification.

"Armchair psychology" is just as dubious as unrealistic assumptions of rationality in the absence of empirical verification.

I disagree. I think the assumption of rationality is conclusively been proven to be inconsistent with consumer behavior. Just look at anything from alternative medicines, fashion, and fads to the power of advertisement over consumer choice. People buy brand name drugs when generics are available. People listen to Britney Spears and buy ugly purses for 4000 dollars (or buy knock-offs that are still overpriced) because some silly hotel heiress whore was carrying one a week ago. In fact, I'd be willing to bet that except for some pretty basic staples, the overwhelming majority of economic decisions people make are irrational from an unemotional, abstract standpoint. The rational actor represents the minority of such decisions and is about as far from the truth as you could possibly me.


Armchair psychology, while it may only weakly correlate is doing better. Even stronger would be models predicting irrational human responses to economic challenges. Based on what Jonah has blogged in the past, I'm sure this is coming (in fact I heard a great lecture on this sometime last year).


Actually, empirical studies of how well the rational actor assumptions held up under controlled condition have been conducted (this is a good chunk of what Vernon Smith is famous for), yielding results that both verified neoclassical equilibrium models in some contexts and indicated their flaws in others. You also seem to underestimate how big of a role goods where people make simple comparisons play in the economy. People spend more on car insurance and phone carriers where decisions are largely rational (hence the prominence of appeals to cost and service in advertising rather than more abstract branding) than they do on designer purses.

As for the effects of advertising, they're an excellent example of where the phenomena are not as simple as the "armchair psychology" suggests - see the work by Gary Becker and Kevin Murphy on advertising.

Or better yet, I'll explain why I spent $80 on a pair of Lucky Brand faux-vintage jeans around 2 years ago. I can buy totally servicable pair of jeans for $20 or less which fit me as well, hold up even better, are just as comfortable, and look just as good. So was my decision irrational? Nope. The reason I bought the jeans was to signal two things 1) I have some rudementary understanding of what constitutes stylish clothing 2) I can blow $80 on a pair of jeans - both of which are very useful things for me to signal while I was on the dating scene. Now, I'm in a committed relationship - my gf is already familiar with my financial situation and sense of style, so I no longer need to signal, so the last couple pairs of jeans I bought cost me $15 each.

The "armchair psychology" approach isn't any more reliable that rational actor approach because they both require the assumption of a highly simplified model of decision making that probably accurately describes a subset of the population, is a reasonable approximation for another fraction, and completely off-base for another subset. The rational actor assumption's advantage lies not in that it is inherently more accurate, but that in because it is easy to predict how rational actors will behave, its relative applicability can easily be assessed. It's easy to cite behavioral effects, but hard to make and test concrete predictions about them unless you're very systematic in how you study them.

Thank you everybody for your comments. Per the Matt/Quitter discusion: I think both abstract models of human decision making - loose psychologizing ala Keynes and mathematical modeling of rational agents ala Friedman- are flawed. So what's the alternative? My own hope is that neuroeconomics provides economics and pscyhology with a more rigorous model of human decision-making in the real world. While neuroeconomics remains a very young science, it does show some real promise. By studying the messy reality of the mind, it might one day ground our economic models in something more reliable than untested theory. I think Herbert Simon said it best:
"It has sometimes been implied that the assumptions of rational behavior underlying the classical theory of economics are not merely irrelevant, but are not even empirically testable in any direct way, the only valid test being whether these assumptions lead to tolerably correct predictions at the macroscopic level. That would be true, of course, if we had no microscopes, so that the micro-level behavior was not directly observable. But we do have microscopes."

I don't personally see that there's anything particularly wrong with the mathematical model of human decision making. The way microeconomics defines "rationality" is not the way people normally think of "rational". All it means is that a person, when presented with the same set of choices and the same set of circumstances, will choose the same thing each time (consistency), and that there is an ordering of preferences with respect to budgets. These are both perfectly acceptable assumptions, I think, and it's the basis for classical microeconomics.

In a sense, whatever "new" knowledge that comes from neuroeconomics and psychology can fit perfectly well in a classical economic framework; all it means is that we don't know enough about individual utility functions. Any supposedly "irrational" things can be encoded in terms of economic utility, from the emotional satistfaction of revenge to the sense of fairness to compassion for the poor. After all, what is "utility" but simply whatever we tend to maximize, as "rational" agents? It's all kind of tautological, but that's the way any fundamental theory is.

It's true that traditional economics doesn't take into account psychological phenomena other than our frontal cortex rational brain.

Just reading these blog comments show's that we have a rational mind (that can reason)

Yet we are debating which ideas are worth pursuing further (we are debating the intrinsic value and utility of the idea's of each model)

Idea's have their own market place.

Yet their is also an emotional appeal. I feel a certain emotional connection to milton friedman, I find his love for economics, freedom, and capitalism inspiring.

As a person who has studied marketing for a long time, it comes to no suprise to me that the human mind is not purely rational.

By simply changing the way you phrase a request, you can increase your response rate by 50%.

In other words, by asking for something, but by simply changing the way you ask it, you can change another persons behavior.

If you want to understand how people are influenced, I highly recommend you read influence: the psychology of persuasion.

By Robben Salter (not verified) on 22 Jan 2011 #permalink