If you're so smart, why aren't you someone in an economics model?
Does economics require rationality?
You are reading Economic Forces, a free weekly newsletter on economics, especially price theory, without the politics. Economic Forces arrives weekly in the inboxes of 5,500 subscribers. You can support our newsletter by sharing this free post or becoming a paid subscriber:
Every few months, a major news outlet—often in the UK—will run an opinion piece about how economics is an utter failure and needs to be burned to the ground so that out of the ashes can arise [insert the author’s preferred approach that conveniently, completely aligns with their political views].
One of the most common critiques of economics in these screeds is that economics obviously ridiculous because it relies on the assumption that “people are rational.” The pieces that really want to trigger me will then have some vague citation to Kahneman and Tversky.
But does economics require the rationality assumption? Josh and I have both previously written why we think the answer is no. Yet, in every paper I have written, I assume the people in the model are rational. Everyone starts with someone maximizing utility (or profit), subject to constraints that determine what is feasible.
In this week’s newsletter, I will try to convince you that rationality is not necessary, and we definitely don’t need everyone to be a lighting-fast calculator of pleasure and pain for the models to work. Instead, a little rationality can go a long way.
Why we can drop rationality: Becker, Gode, and Sunder
Building on Alchian’s famous evolution paper, Gary Becker gave the most compelling early argument against needing rationality within economics. Becker provides a defense of the theorems of economics that does not rely on assuming rationality (or even survivorship like Alchian does). As he points out, “of course, the only ultimate defense is an empirical one,” but I’m going to focus on theory.
Instead of rationality or evolutionary arguments, the critical force in Becker’s model is scarcity: the budget constraint. Budgets generate downward-sloping demand curves, at least at the aggregate level.
The easiest way to illustrate Becker’s point is through his example of random behavior. There are two goods, X and Y. The prices start such that each household’s budget constraint is the line from A to B. If households randomly pick a bundle that spends their whole budget, the average household will pick on the A-B line in the middle, around point P=(X_0,Y_0).
Now suppose the relative price of the X good increases. The household’s budget constraint rotates and becomes the line from C to D. Students of economics will recognize this as a “compensated” price change, so the households can still consume their original bundle at P.
But since people behave randomly, the average household again chooses a bundle in the middle of the budget constraint. This time that is a new point P', which involves less consumption of good X and more consumption of good Y, compared to the original budget.
There’s the result. If the price of good X increases, the (compensated) demand for good X decreases. No rationality required.
For Becker, the key theorem of rational behavior is that demand curves slope down. And since everything is demand and there is no such thing as supply, if you prove demand slopes down without rationality, you’ve captured most of economics: supply and demand.
Israel Kirzner wasn’t convinced and had a back-and-forth in the JPE with Becker.
Kirzner’s agreed we could generate a downward-sloping curve. Still, according to Kirzner, a key theorem of traditional theory is that the market will end up and the intersection of the supply and demand curves. Supply and demand is about more than the slopes.
Suppose we start with a market price that isn’t the equilibrium price but is too low. Kirzner lays out the standard logic that every intro econ teacher gives for how markets move toward equilibrium:
Any such (below equilibrium) price results, therefore, in the disappointment of plans of buyers. Buyers will have discovered that their buying plans were laid in error; the prices which they offered were too low to secure what they had expected to secure. These discoveries will result in the systematic revision of market decisions. Buyers will bid higher prices; market price will rise…
The essence of this market process, it will be observed, is the systematic way in which plan revisions are made as a consequence of the disappointment of earlier plans. (emphasis in original)
This revision based on earlier disappointments requires some amount of rationality.
Economists well understand that people have the incentive to buy low and sell high to earn a profit, and that force helps markets reach equilibrium. In his textbook, Becker uses the standard, Kirznerian logic to explain the stability of equilibrium. Becker goes further than Kirzner even and argues, “The usual incentives to invest in knowledge would work to stabilize each market.” People will actively invest to clear markets.
But remember, the question isn’t “does the story make more sense when people are rational?” The question is, “does the result require rationality?”
We now know the answer is no. There exist types of markets that reach approximately equilibrium prices without any rationality. It is not valid for all markets, as I will elaborate on below.
Gode and Sunder (1993) put Becker’s argument to the test with computer-generated “zero intelligence” agents. They keep the double-auction format from Vernon Smith’s experiments. This is like a stock market where people call out prices with a board of prices for all to see.
What Gode and Sunder do differently from Smith is that they simulated traders that were non-strategic and made bids or asks at random, just like in the Becker example above. Without a budget constraint, random actions do not generate the equilibrium allocation. That’s expected.
But, with a budget constraint, things change. Gode and Sunder induce a budget constraint by making it so buyers cannot offer more than their income, and sellers cannot sell at a loss. Beyond that, the simulated agents were still doing their non-rational, random thing. With the simple addition of a budget, the market moved close to the equilibrium prediction.
Budget constraints with a public price double auction can generate equilibrium outcomes. We don’t need the Kirznerian story of learning from mistakes. As I’ve argued before, the institutions can replace the rationality of the agents.
Supply and Demand Applies outside of Stock Markets
The double auction is a fundamental but special institution. Everyone can buy and sell with everyone without much thinking.
The supply and demand results would be minimal if they only applied in that setting. The learning/rationality/arbitrage point that Kirzner is making becomes clear and has more bite in situations where the institutions are not quite so friendly.
To see what I mean, I’ll walk through the simple arbitrage model in my paper with Rafael Guthmann. This is also the implicit model at the beginning of Kirzner’s Competition and Entrepreneurship.
Take Gode and Sunder’s random behavior agents with budget constraints, but split agents into two markets. There are no “physical” barriers to trading (these are computer simulations, after all), but the agents only see offers from half of the participants. They have the same amount of rationality, but they have slightly less information than in the full, double auction. Gode and Sunder’s results show that each market will have an approximate equilibrium price.
However, nothing forces those two prices to be similar if the supply and demand curves differ across the two markets. The agents only trade within a market.
The central proposition in economics is the law of one price; that does not hold here.
Yes, these are technically different goods because no one knows about the other market. Still, the economist observing these markets in the real world would likely expect these prices to converge since everyone in the simulation would consider them perfect substitutes.
We need some person—a Kirznerian entrepreneur—to learn about the difference and try to make an arbitrage profit, which will drive the prices in the two markets toward each other. We need two for complete arbitrage, zero profits, and the law of one price to hold.
Alternatively, we could imagine enough zero-intelligence traders that can trade in both markets that their behavior swamps the people who can only trade in one. Then we are back in the original Gode and Sunder environment. A law of large numbers almost replaces rationality.
The combination of insights from Becker, Gode and Sunder, and the modified Kirzner example leads me to a middle ground in the debate about rationality. We don’t need everyone to be that quintessential economic agent, a lighting-fast calculator of pleasure and pain. But we get a lot of traction from one person or a few people being rational.
Rationality plays a more minor role in economic theories than you’d probably think.
Theories Don’t Fall from the Sky
There is one more benefit of assuming rationality. The above discussion acts as if these theories and results exist in the ether. But people (often called economists) have to come up with these theories to apply them and test them.
As an empirical observation, it’s just a fact that economists have had more luck coming up with useful theories by assuming rationality. It’s not a perfect theoretical assumption, but it has been powerful.
For example, while his textbook, Economic Theory, starts with scarcity (instead of choice), Becker himself assumes rationality all over the place. I found no other examples from Becker’s work that take the random behavior point seriously.
His next big paper to be published after “Irrational Behavior and Economic Theory” was “A Theory of the Allocation of Time” (1965). That paper explicitly builds a model of maximizing utility subject to constraints. Becker is telling us to take the time component of maximization more seriously.
More than that, Becker was the original economist with results that were basically “actually this behavior is rational.” Crime? Rational. Discrimination? His with Kevin Murphy is “A Theory of Rational Addiction,” not “A Theory of Random Behavior Addiction, which is as-if it was Rational.”
Here is my takeaway from Becker on rationality, summarized in meme form (since this is Economic Forces):
As a friend of mine once put it, the Becker irrationality result is too clever by half.
That’s what makes it great. It’s provocative enough to get you thinking, even if you don’t ultimately put full weight behind it. Becker didn’t.
"Rationality plays a more minor role in economic theories than you’d probably think."
Becker's intention with that paper was that many of the conclusions of economics are robust to alternative assumptions, including the two forms of "irrational" behavior modeled there.
Becker was my dissertation advisor in the late 1980s and I was debating these issues with him then, coming to pretty much the same place you come to here. Becker was extremely pragmatic and empirical. He was open to the possibility of irrational behavior being influential in human behavior, and followed behavioral economics closely. But his take was that the vast majority of critiques of rational choice were lazy. If one could obtain testable predictions from an assumption of stable preferences and rational choice, then given Becker's extraordinary track record of obtaining new insights using such a framework, one should do so. If that framework led to empirically false conclusions (as, indeed, behavioral economists could occasionally demonstrate), then that was interesting.
But despite Becker's analytical rigor and commitment to empiricism, there has remained a vast cottage industry critiquing economics for "assuming rationality." I've never seen any of these critiques recognize Becker's sophisticated analysis as summarized above. For instance, David Sloan Wilson is often criticizing the assumptions of economics, without (to my knowledge) ever recognizing Becker's nuance on this issue. Becker would have Wilson show exactly how a specific assumption of irrationality would change predictions and then ask whether those predictions were, in fact, more accurate. Again, in some circumstances they are more accurate. But rational choice is a remarkably good empirical approximation in most circumstances. Moreover, there is also a literature via experimental economics showing that in some circumstances individual irrationality is irrelevant in a market in which some rational agents move markets towards rational outcomes, as you note. I'd love to have you educate Wilson if you have the time. He means well but he really does not understand these issues clearly.
As an aside, if academic research were an "efficient market" in information, there would not be a need to recall arguments from 70 years ago which have been widely forgotten. They would have been incorporated into all subsequent research. I estimate there are many thousands of peer reviewed papers criticizing economic rationality which show no awareness of this line of thought at all. David Sloan Wilson is only one of many.
So why do economists still need utility theory?