A new paper finds that mergers allow firms to raise prices, but finds no evidence that they improve productivity or efficiency.
Do large mergers benefit or harm consumers? Over the years, corporations and economists have argued that mergers benefit consumers by increasing efficiency, reducing production costs, and, in turn, lowering prices.
A new paper by economists Justin Pierce of the Federal Reserve Board of Governors and Bruce Blonigen from the University of Oregon, however, shows the opposite is the case. By utilizing new techniques to isolate the effects of mergers, they find no evidence that mergers increase efficiency, but do find evidence that they increase market power, meaning they allow companies to generate higher profits by raising prices.
Blonigen and Pierce focus on the manufacturing sector, which is responsible for roughly 25 percent of M&A deals in the U.S. Relying on data covering the entire sector from 1997 through 2007, they are able to compare data from factories that were acquired during mergers to similar factories that weren’t, and to factories where an acquisition has been announced, but not yet completed.
While they find no statistically significant evidence that mergers have a positive effect on productivity or efficiency, Blonigen and Pierce do find substantial price increases following mergers, with markups ranging from 15 percent to 50 percent.
In the past two years, as issues related to antitrust and concentration came back to the forefront of American political discourse, economists and policymakers have been increasingly concerned that competition is weakening in most U.S. industries. In order to better understand the effect that mergers have on the economy, and the methods used to generate this new data, we recently interviewed Blonigen, the Philip H. Knight Professor of Social Science at the University of Oregon.
Q: Can you explain the tools you used in your data analysis for this study, and how these newly-developed techniques differ from previous studies?
In the past it has been hard for researchers looking at available data to understand why a firm’s profitability goes up after some event, like a merger. It may be the case that their profitability went up because they were able to lower costs. This is called a “productivity” effect. But another way in which profitability could go up after the event is through an increase in prices, and potentially the ability of the firm to increase the price above their costs more than in the past. This is what is called the markup and stems from what we call the firm’s market power.
It’s difficult to get all the detailed data on prices and costs that you need to analyze the real reason a firm is more profitable: profitability or market power. But newly available ideas and statistical approaches now allow us to decompose profitability changes in productivity or market power effects with more-available data on firms’ revenues and input usage.
Q: What prompted you to embark on this project?
It was the new techniques that Jan De Loecker and coauthors have developed, which allow us to decompose profitability changes into productivity or market power effects with more readily-available data. When my coauthor and I saw this paper, we realized, “Wow, this is something that will allow us to better examine the age old question about what happens when firms undertake mergers and acquisitions across a broad swath of firms and industries.”
And when I looked at the literature, it just seemed like there wasn’t great evidence on this age-old question because of data availability. The most convincing pieces of evidence are a number of case studies where they have really detailed data, including prices of the individual products that the firms are producing in order to truly identify productivity versus markup effects. But these techniques allowed us to look at these questions with a relatively less demanding data requirement. And that it was also something you could do across lots of different types of firms and industries.
Q: How do you define markup and market power?
A markup is how much your price is above your marginal cost. Suppose the cost of producing a product is $2 and you can charge $3. That’s a 50 percent markup.
Market power is how much firms can markup their prices above marginal cost. The classic theoretical case is perfect competition, where there’s so much competition that you can’t charge a price above your marginal costs, and you won’t make any profit. You’ll make just enough to stay in business, but no more.
And then the opposite end of the spectrum is a monopolist, who is the only producer of a product, they definitely will be able to charge prices over the cost of producing the product if demand is reasonably sufficient. But in reality there’s a whole range in between that, and most firms fall somewhere in the middle of that spectrum.
Importantly, when firms have higher markups due to greater market power, they’re gaining more profit, but it also creates some losses to society. Consumers are paying higher prices, so consumers are losing what’s called consumer surplus, and it can be shown that the loss of consumer surplus is larger than the extra profit that the firms are making, and that’s a loss to society. So that’s why we care about market power and markups.
Q: In this paper, you focused on manufacturing industries, which face more competition from imports and international trade. How do you measure concentration in manufacturing industries, and the market share of the imports?
The way in which we’re measuring market power in the paper is not coming from us measuring market concentration, so we weren’t trying to measure market concentration, per se. We’re inferring their ability to markup their prices based on how their output was changing relative to their input usage –we’re not using any information on market concentration to get our estimates. So we don’t know the source of the markups from what we’ve done and it’s not a study about market concentration.
It’s really a two-stage estimation. First, we get estimates of productivity and markups by looking at the relationship between inputs and outputs in the plant (our data are at the plant level, not firm level) through the new techniques developed by Jan de Loecker and coauthors. And in the second stage we examine whether these estimates of productivity and markups change significantly when the plant is involved in a merger or acquisition. We estimate these merger effects after controlling for a lot of different things, including industry characteristics and even plant-level effects.
Q: But do you control for the level of concentration in the industry?
We do control for industry concentration in the second stage, and so we are measuring whether these particular plants were able to increase their markup after a merger acquisition, controlling for what was generally going on in the industry because of market concentration changes. We’re also controlling in that sense for changes in imports or import competition, because we are controlling for what’s happening to the industry in general.
Q: Would you expect to see more pronounced results had you researched service industries, which tend to face less competitive pressures from trade?
Actually, there are a lot of service industries that do have competitive pressures from trade. There’s a fair amount of tradable services. There’s air travel for example, but also think about all the professional services–software services, accounting, auditing—those are really big service industries that face a lot of international competition.
The service sector is incredibly diverse. When people say service sector, they oftentimes mean everything that’s not manufacturing, but that spans everything from domestic transportation to accounting services to restaurants.
Our study only looks at the manufacturing sector, but that’s a very broad and diverse set of industries and firms and sectors as well, from timber companies to high-end electronics to toys to printing services. As social scientists, we want to be careful about extrapolating to things that we haven’t necessarily looked at, but if we’re finding these kinds of general effects across that really wide and diverse set of sectors in manufacturing, it wouldn’t be surprising to us that you would find similar things in the service sectors in the U.S. as well.
Q: The conclusion of this paper, that large mergers often hurt consumers by driving up prices, might be controversial. Did you encounter any rebuttals from economists?
We have presented our study in a variety of academic conferences and venues. We are certainly getting excellent critique and feedback that we will use to refine the study. But the comments have not been ideological or concerned about the results being controversial in the academic community. The comments we have received are simply meant to examine further the robustness of our results.
We have had some media interviews where people want to know whether this study can tell us whether a particular merger will be beneficial or harmful. That’s not possible. Our study is estimating an average effect of all mergers in manufacturing over the 1997 to 2007 period, so you can’t use it to suggest that any single particular merger is not going to be harmful or beneficial.
I will note as well, that our estimated effect on market power is really being driven by horizontal mergers in our sample–mergers of firms in the same industry. So for non-horizontal mergers–mergers that are not in the same industry or industries that don’t seem to have relationships to each other–we’re not finding those effects. So that also tells us that there is a more nuanced story there, potentially. The other part we don’t control for, and certainly I think should be for future study, is markups could go up after a merger in a way that would be very beneficial for consumers as well. If there’s quality upgrading of products, or new products that come out of the merger that consumers value much more so than previous products, the increased markup would not necessarily be lost consumer surplus.
Q: But why should we assume that firms that have more market power will also offer more quality products? The opposite can also be a possibility.
Yes, it is. And so that’s why there needs to be more study of the issue. We’ve had to be careful with how many different, smaller dimensions we look at in our data due to the confidential nature of the data. But one possible dimension we could look at in the future would be industries where there are lots of new products being created. If these industries are a main driver of higher markups in our study, this might suggest it is due to this “new-product” effect and not increased market power. So we may examine that in the future, but haven’t done that yet.
