A market with 1,000 tiny sellers is not some ideal market. Concentration can be extremely beneficial, leading to economies of scale and stiffer competition to win a big share of the market.
Yet the Federal Trade Commission (FTC) and U.S. Justice Department’s (DOJ) draft merger guidelines double down on the idea that concentration is inherently a problem. They add new structural presumptions against concentration, for both horizontal and vertical mergers. Even worse, the agencies won’t recognize any efficiencies “if they will accelerate a trend toward concentration.”
The problem? Concentration is not a good proxy for competition. This post will go through some of the empirical work that generally finds that competition increases concentration. The FTC/DOJ are completely at odds with the economic literature on this point.
What Long-Time Readers Know
Despite what you may have taken away from a misleading Econ 101 class, there is not a simple sliding scale between one seller (high concentration, low competition) and perfect competition (low concentration, high competition).
From a theoretical point of view, even perfect competition does not require infinitely many buyers and sellers. Two sellers can generate perfect competition. That’s what the standard Bertrand model is. Price equals marginal cost. It’s efficient. All the good stuff, with just two sellers.
This is more than just a theoretical matter. In the lab—even in settings with few buyers and sellers and where the experimenters are trying to put the finger on the scale against competitive forces—we see the fruits of competition.
In a disgustingly-under-appreciated experiment, Dan Friedman and Joseph Ostroy even construct the “worst-case” scenario for competition that gave players lots of market power to try to exploit. Still, when people choose their prices and quantities to offer for trade, the market becomes competitive.
So we can have competition without many buyers and sellers. We can also have market power (which signifies a lack of competition) with many buyers and sellers. As Chad Syverson argued in a famous Journal of Economic Perspectives paper on markups that I’ve quoted too many times: “concentration is worse than just a noisy barometer of market power. Instead, we cannot even generally know which way the barometer is oriented.”
What’s the Relationship Empirically?
Ok, you may respond, from pure theory, we do not know the direction of the relationship between competition and concentration. But what do we know empirically? Do some relationships tend to emerge for most markets? Or, in some cases, can we guess which way the relationship goes based on prior evidence?
Yes, we can, but we have to be careful.
Suppose we wanted to explore the relationship between competition and concentration. Generally, in economics, we want to discover a causal relationship: If X changes, then Y changes. Here, we want to ask, if competition increases, what happens to concentration? Or maybe, if concentration increases, what happens to competition?
But what does increasing competition even mean? What would we observe and conclude that competition increased? We can’t measure competition directly. We can measure prices and quantities. Competition is an organizing concept that economists have found useful.
In lots of economics papers, there is a type of Cournot model, where competition simply means the number of sellers or buyers. You start with two sellers, then add another. Surely, another seller satisfies any reasonable definition of more competition. In that world, increases in competition (a new seller falling from the sky) decreases concentration. This is the bad Econ 101 reasoning.
To be clear, I agree that adding more sellers would decrease concentration and increase competition. If we had another company like Apple fall from the sky, the concentration in the market for Apple products would fall, and so would the price of Apple products.
Lots of papers amount to proving that. Or they show the opposite: if the two Apple companies merge, then prices rise.
But that’s not the kind of causal reasoning we could bring to data. Where did that new seller come from? Another Apple won’t fall from the sky. I’d love for that to happen (or any innovative company), but it’s not happening. It’s not a reasonable change to search for in the data to back out the effect on concentration. In modern empirical work, we can’t simply look at changes in the number of competitors as a measure of competition. The number of sellers is endogenous to the market. Firms may be entering and exiting for a variety of reasons. We cannot regress prices or markups on the number of sellers. Definitely don’t regress concentration on the number of sellers.
Instead, we want to look for an exogenous change that we think captures changes in the competitiveness of a market. A common thing to look for is a reduction in search/trade costs or an increase in how substitutable products are.
The idea is simple. If a consumer’s search costs drop, the goods are more substitutable to that consumer. Think about the difference between flipping between Walmart and Aldi on Instacart, versus going out to the physical stores. From the firms’ perspectives, the consumers’ demand curves in the Instacart example are more elastic, and each firm needs to compete more intensely for each consumer. In equilibrium, markups will fall, and prices will be driven toward marginal cost.
These search costs aren’t competition directly, but in a large class of models, there is a simple mapping from these costs to other measures of competition. Basically, if you have heterogeneous firms selling differentiated products, you get some parameter (search costs or substitutability) that closely matches a measure of competition. I’ve gone through a few of these models before here.
Beyond increasing competition, reducing search costs can increase concentration, since more consumers are able to purchase from the highest-productivity/lowest-cost producers. Compare the concentration measures between a world with a bunch of isolated towns, each with one retailer, and a world where everyone can buy from Amazon. For the nerds around, think of a Melitz model. In those models, lower trade costs increase competition and concentration.
Theory—always start with theory—gives us a reason to believe there is a causal relationship between something like trade costs (as a proxy for competition) and concentration. Moreover, there is reason to expect a positive relationship. More competition causes more concentration.
What do we observe in the data? I’ll once again turn to Chad Syverson:
Many empirical studies in varied settings have found that greater substitutability/competition—resulting from, say, reductions in trade, transport, or search costs—shifts activity away from smaller, higher-cost producers and toward larger, lower-cost producers. As an example, in Syverson (2004a, b), I show that increases in the ease with which consumers can substitute among producers’ spatial differentiation is limited, or products are more physically similar—force out the least efficient producers and increase skewness in the size distribution. In Goldmanis et al. (2010), we demonstrate that search cost reductions reallocate market share toward lower-cost and larger sellers, increasing market concentration even as margins fall. It is not an exaggeration to say that there are scores, perhaps hundreds, of such studies. (links and emphasis added)
The key takeaway from a huge literature in trade and industrial organization (IO) is that you cannot assume trends toward concentration are bad. Empirically, we tend to see more competition increasing concentration. That doesn’t mean it is always the case. A new upstart that is super efficient will initially decrease concentration, before winning over a large portion of the market and increasing concentration. But we should be aware of the trends in the data.
Returning to the merger guidelines (sorry), we cannot directly apply mergers to this literature. A merger is a different causal change than those that are studied in the papers cited. A merger may increase concentration while decreasing competition. That’s not my point.
The argument is simply that some pro-competitive actions increase concentration, and we see that a lot in the data. This is why I think it is unscientific to have a blanket skepticism toward concentration. It would be a shame for the FTC and the DOJ to ignore such a large body of evidence accumulated over the past 20 years.
[A version of this post originally appeared at Brian’s Substack newsletter Economic Forces.]