Are Markups Really SO Bad?

A version of this post originally appeared at Brian’s Substack newsletter Economic Forces.

Cite this Article
Brian Albrecht, Are Markups Really SO Bad?, Truth on the Market (September 04, 2023),

Concentration is a terrible measure of [insert basically anything people actually care about]. Have I said that before? Concentration tells us nothing about market power, efficiency, or whether policy changes can do anything to increase welfare. Economists know that, especially industrial organization (IO) economists.

If we want to measure market power for a seller, a better measure is the markup, defined as the ratio of price over marginal cost. If we want to measure market power for a buyer, we can look at the markdown. Either one is a better measure of market power (possibly bad) and often the very definition of market power.

Let’s focus on markups. The logic of measuring markups as a proxy for something we care about is straightforward Econ101.

With perfectly competitive markets, the outcome is efficient. Great! And look, in the simple model, each firm has zero markup! In a basic monopoly model, the outcome is inefficient. Oh no! And look, in the simple model, the monopolist has a positive markup! If we just crank up that markup (How? Don’t ask so many questions!), we get a larger deadweight loss.

This comparative static is the basic mechanism behind why many economists are quite concerned when they see rising markups, exemplified by the response to the findings in a famous paper by Jan De Loecker, Jan Eeckhout, and Gabriel Unger.

We must be careful, however, about what it means to be cranking up the markup. We need to think seriously about the cause. For most economists, there is an implicit comparison between the competitive and monopoly models in Econ101. The supply curve remains the same. How do we compare the monopoly world to the competitive world? We imagine that we somehow get more sellers, and they compete to drive down the price. And voila, output and welfare increase.

To be explicit, this comparative static tells us that the causal effect of adding more sellers (that fall from the sky) would be to decrease price, increase output, and increase welfare. Seems plausible to me. To continue my analogy from a previous post, if we had another company like Apple fall from the sky, the price of Apple products would fall, and welfare would improve. It’d be great to have more companies producing good stuff, all else equal. As always in economics, we need to be careful about that damned all-else-equal clause.

But there are more than these two Econ 101 models! (Somehow, I’m accused of being overly committed to simplistic Econ101 models.) That’s not the only comparative static we can do that changes markups.

Consider an alternative cause for markup increases. There is a monopoly seller of a product. Everyone likes their product somewhat, but everyone likes it the same amount. This is captured by a horizontal demand curve: everyone is willing to pay the same amount. In that case, price equals marginal cost and all of the surplus goes to the seller. (It is also perfectly competitive, even with a monopoly, but let’s save that for another day.)

What changes? Suppose the firm said, “Yes, it’s great we are making a profit, but what if we made even more? What if we made a better product?” The change is they discover a better product. Some people really love the improvement, while others are less impressed, but still like it more than the old product. Overall, now the demand curve slopes downward. It also creates a markup! This seller improved their product, raising prices, increasing profits, and charging a markup. Are these rising markups a bad thing? Obviously not. Yes, markups increased, but by assumption, everyone is better off. The gap between price and cost reflects that marginal value has risen. We could also imagine a technological improvement that lowers costs but increases markups. Both cases generate a markup, and that’s great! It may also be worth pointing out that I didn’t label the intersection of MC and D and draw out the deadweight loss. That’s partly because making these graphs is a pain. But also, I’m not sure that point is feasible. It depends. What policy is available to get us there in this case? Should we subsidize output? Are we okay with firms making more profit? It may be possible to get there, but we must be explicit about what could bring us there. In Econ 101, we often act like it is trivial to get to that efficient point. Maybe. Maybe not. You need to tell me what’s feasible and why we aren’t there already. While a trivial example, I think the comparison of the two figures above highlights a big disconnect between how economists talk about profits, margins, and markups and how scholars in strategy or entrepreneurship talk about them. The strategy people see this creation of a better product as a source of value creation, value capture, and profit. Are profits a bad thing? Surely, it is good to take lower-valued items and turn them into higher-valued items. As a wise snowman once said: all good things, all good things.

We can then ask, will product imitation and competition drive prices back down so that welfare further increases? That will depend. What is the property right that generates that markup?

Maybe the improvement was an idea that other firms could quickly replicate. In that case, markets will be driven back to the original graph. We need to realize, however, in that case, the graph won’t show consumers’ welfare gain, but it is there.

Or it could be that the product’s improvement required a big upfront investment by the firm. Maybe they trained all of their employees better. Maybe they invested in higher-tech equipment. In that case, maybe other firms won’t so easily replicate the improvement. The extra measured profits are not really economic profits, but simply returns to the investment.

For example, people discover some technological improvements. By assumption, it’s an improvement and a welfare improvement, allowing producers to lower marginal costs. The catch is that taking advantage of those improvements requires an upfront payment. For example, after 1997, a firm can pay for a computer, which allows them to send zero-marginal-cost (electronic) mail.

We can expect to see prices diverge from marginal costs whenever there are fixed costs. Some revenue needs to cover those fixed costs. I discussed this in my post on electricity markets (which have high fixed costs). With fixed costs, markups don’t necessarily reflect any inefficiency. As I said before, markups will exist, even in a perfectly competitive market with fixed costs, “even if there is a perfect substitute waiting in the wings for the firm to raise prices by a penny.”

That doesn’t mean any markup we measure is efficient. Markups may be too high or too low in some sense. See William Baumol and David Bradford on optimal deviations from marginal-cost pricing. Some policies may improve efficiency by changing markups somehow. It just means we cannot just look at the markup or changes in the markup to tell what has happened to efficiency or welfare. We can’t just use those Econ 101 models!

If fixed costs rise and marginal costs fall over time, the efficient outcome would be exactly what we see in the abovementioned De Loecker, Eeckhout, and Unger paper! In that case, rising markups reflect underlying technological improvements. Two cheers for markups.

I’m not saying the increased markups over the past more than 40 years (assuming that this has, in fact, happened) has been a net beneficial outcome. I’m not saying that policy can’t make things better in the future. I’m just saying that the measurement of a markup itself tells us nothing about efficiency without an explicit model of what caused the rise in markups.

The data doesn’t speak for itself! Luckily, price theory has models to help us.