Econ 101 ignores 50+ years of economic science

Insights from Harold Demsetz

Re-post of a core Economic Forces idea that is worth repeating

---

Econ 101 is behind the times. No, I don’t mean because it’s too free-market or doesn’t include behavioral economics, feminist economics, game theory, or any of the other things every article complaining about Econ 101 mentions.

The major problem with most Econ 101 courses is that they take an outdated and successfully refuted approach to market structure and how prices are determined. As a price theory guy, that makes me sad.

Here’s a standard course or textbook: After focusing on perfect competition for one-third of the semester, the focus moves to the monopoly model. What supposedly differentiates these two situations is the market structure. Competition involves many buyers and sellers; monopoly has only one seller. From that starting difference, all the other results flow: competition is efficient, monopoly is inefficient, etc.

The problem? This approach, which used to be common in economics research, has been thoroughly discredited within the relevant economics subfield: industrial organization (IO). 50 years ago, Harold Demsetz taught economic researchers the errors in a little paper, “Industry Structure, Market Rivalry, and Public Policy.”

It’s time for Econ 101 to learn the lesson.

Pre-Demsetz: The Old Structure-Conduct-Performance Paradigm

Before getting to how we teach 101, it’s worth doing a quick recap on the history of industrial organization. Pre-Demsetz, the structure-conduct-performance (SCP) paradigm had an intellectual monopoly in IO. Monopoly. Get it?

The big name in this area was Joe Bain, but sometimes you also see it associated with Edward Chamberlin or even Joan Robinson. Under SCP, the performance of the market was something like consumer welfare. That’s what we care about. Performance was determined by the conduct of the firms in the market, say their prices. Ultimately, the conduct was determined by the structure of the market: how many firms are there? How high of a market share do the top 4 firms have?

We have the following simple, causal structure.

Structure → Conduct → Performance

There are multiple mechanisms for why this makes sense. In a Cournot model, the number of firms (structure) determines the equilibrium price (conduct) and, therefore, the consumer welfare (performance). It’s not restricted to Cournot, though. George Stigler’s 1964 “A Theory of Oligopoly” was really an SCP model of collusion. Fewer firms (structure) makes it easier to collude (conduct), which hurts consumers (performance). Basically, in all of these models, the core result—really the core assumption— was that fewer firms ultimately led to higher profits. In the applied work, you had a bunch of regressions of profit on the number of firms or price on a concentration measure.

What did they learn? A whole lot of nothing. They found zero, positive, negative, and everything under the sun correlations. A plurality of papers did find that higher concentration was associated with a higher price, but that was in no way a sure thing.

Today, we still find similar regressions being run outside of the field of industrial organization. For example, last year, there was an online debate about the relationship between concentration and inflation. Was concentration to blame for our recent inflation? Hal Singer ran one regression that found a positive relationship: higher concentration correlated with higher price increases. Jason Furman found a negative relationship. More recently, Conlon, Miller, Otgon, and Yao found no correlation between concentration and markups (not exactly prices but part of the same debate).

The whole debate is a mess, as I’ve discussed before.

Is Concentration Driving Inflation?

Problems with Structure-Conduct-Performance

The fundamental problem with this type of regression and the SCP paradigm, more generally, is that both variables are endogenous; they are the outcome of some competitive process. Demsetz was the first to hammer this home.

Demsetz argued that market structure was an outcome and could be shaped by firm behavior and innovation. For example, he starts out his article by discussing the possibility of "concentration through competition." Instead of concentration falling from the sky (as it does in Cournot), concentration is an outcome of some previous competition. For example, high prices and profits can induce entry so that you will find less concentration.

Or maybe one firm is just more efficient. They will capture most of the market, but it doesn’t make sense to say the market was more competitive before they entered. If the threat of entry is real and we have what Demsetz called “competition for the field” or what Baumol and coauthors called “contestable markets.” The market price can be the competitive price, and profits can be zero.

The exact relationship between structure and performance depends on lots of features of the competition that exists in the market. This is especially true once we consider the dynamic nature of any of these markets. Prices and profits today drive dynamism tomorrow.

As I’ve cited before, Chad Syverson nicely summarizes the problem with the SCP approach:

Perhaps the deepest conceptual problem with concentration as a measure of market power is that it is an outcome, not an immutable core determinant of how competitive an industry or market is… ​​As a result, concentration is worse than just a noisy barometer of market power. Instead, we cannot even generally know which way the barometer is oriented.

Another way to see the problem with SCP is to think about the comparative statics approach that economists love to use. With comparative statics, you imagine something outside of the system you are studying (in jargon, an exogenous parameter) changing: the tax rate rises, oil prices spike, or an outsider develops a new technology. You then trace the effects of the parameter on some outcome variable (an endogenous variable).

For a basic comparative static, sometimes we assume buyers are passive price-takers, and then we can imagine the price being exogenous to the buyers and derive how their behavior changes. That is how we get the law of demand; when the price goes down (exogenously), the quantity demanded goes up.

But we should never do comparative statics on price for the whole market since the price is determined within the market. Price is an outcome determined by the interplay of the supply and demand system; it’s not outside.

What does it mean for the market price, not just the buyers’ price, to drop exogenously? What forces the price to go down? By assumption, we are fixing supply and demand, which means the price is determined already. But we are forcing price to change??? It’s an incoherent exercise.

This is the same reason we never reason from a price change. Sometimes we can consider the effects of the price of oil on the price of natural gas. That’s not what we mean. But we cannot reason from a price change for the market we are studying. The reason is that the price change in the market we are studying is not out of that market. Our predictions about the effects of a price change depend on the exact cause. For example, if the price goes down, we need to know whether that was a supply shift or a demand shift. Depending on what changed, the predictions about what happens to quantity are the opposite! That’s why we can’t start the story at “price change.”

In the same way, we can’t start the story with five firms simply existing in the market, and we can't do comparative statics on the number of sellers without imposing a very particular form of competition like Cournot.

Suppose a firm drops out of the market. What happens to the price? It matters why the firm left. If it is a Cournot model, the assumption is lightning strikes, and the firm dies. That begs the whole question; there is no real cause. But in the real world, there is a cause (or many). Did other competitors cut costs? Then we will see firms exiting, but prices drop. Did the government say a firm cannot produce anymore? Then we will see prices rise. The first step in the causal chain to prices and consumer welfare matters, and that first step is never simply “structure.”

In reality, firms are constantly innovating and adapting to changes in the market. For example, they may invest in new technologies to reduce costs, or they may develop new products to differentiate themselves from competitors. None of those key features of markets show up in SCP.

Demset’z insight absolutely destroyed SCP and became core to the IO approach that came to prominence next, the so-called new learning approach. But even when new learning was supplanted by the game theory revolution and what is now called modern industrial organization, this insight from Demsetz remains prominent. The dynamic nature of competition matters, and the structure is not exogenous. In fact, I’d say game theory was such a success in IO exactly because it was able to capture aspects of this dynamic competition. Game theory emphasizes the strategic interactions between firms, which opens up the theory of contestable markets mentioned above. As Berry, Gaynor, and Scott Morton put it, "the structure-conduct-performance approach has been discredited for a long time."

This is not one of those areas where I am just crazy and out of step. Demsetz’s insight is mainstream IO today.

Structure-Conduct-Performance in Econ 101

With that whirlwind history of IO behind us, we can now turn back to Econ 101. (That’s enough spinning metaphors for one newsletter.)

Despite the significance of Demsetz's contributions, introductory economics courses have largely ignored the evolution of IO over the past 50 years. Instead, they continue to present the structure-conduct-performance approach as the primary framework for comparing market structures.

As I said above, the introductory study of economics has traditionally focused on two types of market structures: perfect competition and monopoly. Introductory economics courses often present these structures as polar opposites, with perfect competition representing the ideal market and monopoly representing a market with significant inefficiencies. Under perfect competition, firms are assumed to be price-takers with no ability to influence the market price. In contrast, under a monopoly, firms are assumed to be price-setters with significant market power that allows them to charge high prices and earn large profits.

We should now recognize that this is just the discredited SCP paradigm with all the problems listed above. Despite the advances from Demsetz and others, introductory economics courses have failed to incorporate these new ideas. This is partly due to the inertia of the textbook publishing industry, which often takes years to incorporate new research findings into textbooks. Some of that lag is good. I would never say that Econ 101 should follow the latest fads in the research community, even those rare fads that I like.

Instead, the goal of 101, in my eyes, is to provide lasting knowledge that the students can take with them. Something being central to IO for 50 years crosses the “lasting” threshold. Getting rid of the SCP approach also has the benefit that, unlike adding something like behavioral economics to Econ 101, post-SCP is actually accepted within the relevant economics community. Unlike adding more game theory, it requires no new tools, so it is easy for a 101 course.

The biggest way to improve on the current approach is to remove the monopoly/competitive framing. It’s a language issue. The structure or number of firms is not the important feature. It’s even extremely misleading. The monopoly model doesn’t apply only to monopoly but to any firm with market power. Firms can have market power even when there are infinitely many of them. Similarly, the competitive model doesn’t require many sellers but sometimes goes through with two sellers (as in standard Bertrand) or even one active seller (if the market is contestable). The structure is largely irrelevant!

Common Chicanery about Competition

Moreover, many (though not all) of the insights of supply and demand go through even in a world that isn’t perfectly competitive. This is why I focus so much on supply and demand. For example, when oil prices rise because of an invasion by Russia, the quantity drops, and the price rises across many models. It’s just easier to show that using supply and demand, so that’s where we show it.

Why would we undersell the main topic of our course and the main tool economists have (supply and demand) by filling students’ heads with nonsense like “this holds with many buyers and sellers with identical goods and perfect knowledge,” all of which are wrong claims in theory? They are also wrong empirically. Competition outcomes occur all over the places with few sellers and imperfect knowledge.

If we need a new framing, let me suggest the old UCLA distinction between price-takers and price-searchers. This is the distinction Alchian and Allen make. While not perfect, it has two advantages over the competitive/monopoly dichotomy. It removes the focus on the structure of the market for all the reasons above. It also naturally leads into questions of why Firm X is a price-taker while Firm Y is a price-searcher. Ultimately, it must be that it is more profitable for each firm to take this approach. But why? That’s a result to derive and discover about the market. It is an invitation to study, not an assumption.

Ultimately, I’m not perfectly satisfied with Alchian and Allen’s framing either. I am all ears for better approaches. Let me know in the comments. But I am sure we can’t keep going the way we have been in Econ 101.