Mergers Often Improve Efficiency
Evidence from power plants
As a society, we have a certain hostility to “horizontal” mergers. When a firm buys its suppliers or what distributes its product to the public, we understand this vertical integration. Perhaps it wants to avoid a recalcitrant monopolist playing chicken with badly needed materials, as numerous papers have suggested (and on which, I wrote a blog post). Horizontal mergers, or those between similar enterprises, raise alarm bells. There are now fewer firms in the market. The new firms may have more market power, and a greater ability to set prices above marginal cost. On the other hand, horizontal mergers may allow for greater productive efficiency. Which effect predominates?
We cannot have a universal answer, as it will vary from industry to industry. What I am convinced of is that the rise in productive efficiency is greatly underrated. Firms vary greatly in their efficiency, with firms in the same industry being far worse than others. Efficient firms are often unable to drive out the inefficient firms, with them hanging around year after year. These differences in productivity are largely driven by management, good or bad, which leads to a host of differences in production which cannot be easily imitated by the outside observer. Firms have internal spillovers, from department to department. American firms in Europe, for example, are more likely to adopt the use of IT which has led to so much of America’s gains in productivity relative to Europe.
What we should expect to find, if this is true, is that acquisition by a better managed firm will lead to a change in management, better production choices, and a rise in efficiency. The recent paper, “Do Mergers and Acquisitions Improve Efficiency? Evidence from Power Plants”, by Demirer and Karaduman, gives us some suggestive evidence in favor. Power plants are an ideal setting to study this. Power generation is quite similar from plant to plant. The product is homogeneous, thus eliminating the possibility that quality might be varying over time. They are able to observe inputs and outputs directly, rather than having to infer productivity from revenue. They use an extraordinary dataset of all power plants in the US from 2000 to 2023. During the period, there were 505 transactions, with 3,515 generators changing ownership. They can then use a difference-in-differences estimator. Difference-in-differences is when you show that two things, in this case power plants, were in parallel before an event, and afterward diverged. You can then argue that the divergence is due to the event.
A specific plant being acquired led to a substantial increase in productivity. Changes at the parent level did not, but an efficient network acquiring a less efficient power plant from a competitor led to a 5% increase in productivity. This is not due to them trading off against more outages, as outages also declined.This did not occur due to simply increasing capital intensity either. They found no evidence for any greater expenditures after. What they did find is that the new management was more educated than prior management. Gains were especially high when the acquiring firm had many other acquisitions, or was larger, indicating that some firms specialize in buying underperforming assets and bringing them in line with their normal.
Of course, other markets will have different characteristics than power plants. We can’t assume that all acquisitions will increase efficiency — of course not. We should simply be less instinctively hostile to them. Firms have culture, and common ownership is how the better cultures spread.