Institutions Have Knowledge, Not People
Charlie Plott gave us the Fundamental Equation for an economic explanation. The simplified version is
Preferences + Beliefs + Institutions = Outcomes.
To explain differences on the right-hand side, you need differences on the left-hand side.
Given our understanding of decision theory since Savage, decision-makers are simply endowed with beliefs and preferences. Those are starting assumptions. We don’t think of those changing directly. After all, de gustibus non est disputandum.
Instead of a preference-based explanation, the general approach is that we think of the institutions as generating different outcomes. The institutions are outside the decision-maker (although ultimately explainable only through individual choices). North Korea and South Korea have different economic outcomes not because of preferences or beliefs but because of institutions. The same is true on a smaller scale.
Now we may have a tension. In my last newsletter, I stressed the importance of knowledge and ideas within economics, so it may be tempting to point to beliefs/knowledge/ideas for why outcomes emerge.
While knowledge is important, it can’t be the final explanation. To have a theory, we need it to be causal. We want to say something like “If X changes, then Y changes.” It’s not much of a theory if we need to say “When people believe different things, they do different things.” Big, if true. But that doesn’t teach us anything.
Instead, we still need an institutional explanation of knowledge (which I’ll lump in with beliefs), even if it means abandoning the nice Plott equation separation of institutions and beliefs. In fact, I will argue that we cannot separate knowledge from institutions in the same way that we separate preferences from institutions.
The knowledge that people have does not come fully formed like Athena from the brow of Zeus. Instead, knowledge is institutionally contingent.
A wise man (me) once wrote:
As Armen Alchian pointed out in his great article on evolution, the important part is not that any person is super-rational and calculates optimal business decisions, but that the profit and loss system weeds out those who choose decisions that lose money. We are left with a systemic process that may lead to beneficial outcomes, even though no one can explain why.
With hyper-rational people, the institutions do not matter much. When we drop rationality assumptions we see the importance of institutions even more.
Take a particular example from a paper of mine with Rafael Guthmann on informational efficiency in markets. In that paper, we consider three different market institutions. In the first, trade occurs through a centralized exchange where everyone is buying and selling in one place. In the second, each buyer searches for a seller and then bargains over a price. In the third, intermediaries help connect buyers and sellers.
In all three markets, assume we have the same group of substitutable buyers and sellers. We have the same preferences/beliefs. If buyers and sellers have enough information and are sufficiently rational, the outcome is the same; it is the competitive outcome where supply equals demand. The particular institution does not matter for the outcome!
The institutions matter more when we drop our focus on rational people, which, as Josh has explained is just a benchmark and simplifying assumption.
As we argue in our paper, bargaining between buyers and sellers on a bilateral basis requires a lot of mental work while the other two markets require much less. “Good” institutions substitute for the knowledge of the individual. Ours is a theoretical argument but the classics in experimental economics suggest the same point.
First, Ed Chamberlain’s experiments in the 1940s showed when buyers and sellers traded and negotiated in private, markets failed to reach the competitive prediction. Prices were lower and volume was higher. Then came along Vernon Smith who changed the institutional environment and the information. He added central bids and offers (versus decentralized negotiation) and let people trade over multiple rounds (which allowed learning). Smith’s experiments matched the competitive prediction quite well.
If the students in the experiment were hyper-rational, the institution would not matter! The outcomes should be the same. It is only because the people aren’t such lightning-fast calculators that the institutions generate different outcomes. The open cry double auction substitutes for the individual’s rationality. That’s the sense in which I mean that the institution has the knowledge, not each particular individual. (I understand only individuals can have knowledge.)
The starkest example of institutional knowledge or rationality comes from Gode and Sunder (another JPE paper by the way). Instead of working with “low rationality” students, they worked with computer-generated “zero intelligence” agents. The simulated traders were non-strategic and made bids or asks at random. This does not lead to the competitive allocation. Obvious enough.
Gode and Sunder then add a budget constraint for each zero intelligence agent: Buyers could not offer more than their income and sellers could not sell at a loss. Beyond that, they are still random. With that simple change, the market moved close to the competitive prediction. For efficient outcomes, we don’t need hyper-rational agents; we need the right institutions. The institutions substitute for the rationality/knowledge of the people.
Tying back to Plott’s equation, we now have an institutional causal theory and prediction. Completely idiotic behavior by firms and households that occurs within markets subject to the discipline of profit and loss can generate near-maximal gains from trade. In environments without budget constraints (such as with a soft budget constraint), there is no such group rationality. As Gary Becker pointed out back in 1962, a binding budget constraint does a lot of work. Sticking with our favorite UCLA themes, it all boils down to property rights and information!