Streetwise Professor

There is a boomlet in economics and legal scholarship suggesting that increased market power has contributed to income inequality, and that this can be addressed through more aggressive antitrust enforcement. I find the diagnosis less than compelling, and the proposed treatment even less so.

A recent report by the President’s Council of Economic Advisors lays out a case that there is more concentration in the US economy, and insinuates that this has led to greater market power. The broad statistic cited in the report is the increase in the share of revenue earned by the top 50 firms in broad industry segments. This is almost comical. Fifty firms? Really? Also, a Herfindahl-Hirschman Index would be more appropriate. Furthermore, the industry sectors are broad  and correspond not at all to relevant markets–which is the appropriate standard (and the one embedded in antitrust law) for evaluating concentration and competition.

The report then mentions a few specific industries, namely hospitals and wireless carriers, in which HHIs have increased. Looking at a few industries is hardly a systematic approach.

Airlines is another industry that is widely cited as experiencing greater concentration, and which prices have increased with concentration. Given the facts that a major driver of concentration has been the bankruptcy or financial distress of major carriers, and that the industry’s distinctive cost characteristics (namely huge operational leverage and network structure) create substantial scale and network economies, it’s not at all clear whether the previous lower prices were long run equilibrium prices. So some of the price increases may result in super competitive prices, but some may just reflect that prices before were unsustainably low.

Looking over the discussion of these issues gives me flashbacks. There is a paleo industrial organization (“PalIO”?) feel to it. It harkens back to the ancient Structure-Conduct-Performance paradigm that was a thing in the 50s-70s. Implicit in the current discussion is the old SCP (LOL–that’s the closest I come to being associated with this view) idea that there is a causal connection between industry structure and market power. More concentrated markets are less competitive, and firms in such more concentrated, less competitive markets are more profitable. Those arguing that greater concentration increases income inequality go from this belief to their conclusion by claiming that the increased market power rents flow disproportionately to higher income/wealth individuals.

The PalIO view was challenged, and largely demolished, in the 70s and 80s, primarily by the Chicago School, which demonstrated alternative non-market power mechanisms that could give rise to correlations (in the cross-section and time series) between concentration and profitability. For instance, firms experiencing favorable “technology” shocks (which could encompass product or process innovations, organizational innovations, or superior management) will expand at the expense of firms not experiencing such shocks, and will be infra marginal and more profitable.

This alternative view forces one to ask why concentration has changed. Implicit in the position of those advocating more aggressive antitrust enforcement is the belief that firms have merged to exploit market power, and that lax antitrust enforcement has facilitated this.

But there are plausibly very different drivers of increased concentration. One is network and information effects, which tend to create economies of scale and result in larger firms and more concentrated markets. Yes, these effects may also give the dominant firms that benefit from the network/information economies market power, and they may charge super competitive prices, but these kinds of industries and firms pose thorny challenges to antitrust. First, since monopolization per se is not an antitrust violation, a Google can become dominant without merger or without collusion, leaving antitrust authorities to nip at the margins (e.g., attacking alleged favoritism in searches). Second, conventional antitrust remedies, such as breaking up dominant firms, may reduce market power, but sacrifice scale efficiencies: this is especially likely to be true in network/information industries.

The CEA report provides some indirect evidence of this. It notes that the distribution of firm profits has become notably more skewed in recent years. If you look at the chart, you will notice that the return on invested capital excluding goodwill for the 90th percentile of firms shot up starting in the late-90s. This is exactly the time the Internet economy took off. This resulted in the rise of some dominant firms with relatively low investments in physical capital. More concentration, more profitability, but driven by a technological shock rather than merger for monopoly.

Another plausible driver of increased concentration in some markets is regulation. Hospitals are often cited as examples of how lax merger policy has led to increased concentration and increased prices. But given the dominant role of the government as a purchaser of hospital services and a regulator of medical markets, whether merger is in part an economizing response to dealing with a dominant customer deserves some attention.

Another industry that has become more concentrated is banking. The implicit and explicit government support for too big to fail enterprises has obviously played a role in this. Furthermore, extensive government regulation of banking, especially post-Crisis, imposes substantial fixed costs on banks. These fixed costs create scale economies that lead to greater scale and concentration. Further, regulation can also serve as an entry barrier.

The fixed-cost-of-regulation (interpreted broadly as the cost of responding to government intervention) is a ubiquitous phenomenon. No discussion of the rise of concentration should be complete without it. But it largely is, despite the fact that it has long been known that rent seeking firms secure regulations for their private benefit, and to the detriment of competition.

The CEA study mentions increased concentration in the railroad industry since the mid-80s. But this is another industry that is subject to substantial network economies, and the rise in concentration from that date in particular reflects an artifact of regulation: before the Staggers Act deregulated rail in 1980, that industry was inefficiently fragmented due to regulation. It was also a financial basket case. Much of the increased concentration reflects an efficiency-enhancing rationalization of an industry that was almost wrecked by regulation. Some segments of the rail market have likely seen increased market power, but most segments are subject to competition from non-rail transport (e.g., trucking, ocean shipping, or even pipelines that permit natural gas to compete with coal).

Another example of how regulation can increase concentration and reduce concentration in relevant markets: EPA regulations of gasoline. The intricate regional and seasonal variations in gasoline blend standards means that there is not a single market for gasoline in the United States: fuel that meets EPA standards for one market at one time of year can’t be supplied to another market at another time because it doesn’t meet the requirements there and then. This creates balkanized refinery markets, which given the large scale economies of refining, tend to be highly concentrated.

Reviewing this makes plain that as in so many things, what we are seeing in the advocacy of more aggressive antitrust is the prescription of treatments based on a woefully incomplete understanding of causes.

There is also an element of political trendiness here. Inequality is a major subject of debate at present, and everyone has their favorite diagnosis and preferred treatment. This has an element of using the focus on inequality to advance other agendas.

Even if one grants the underlying PalIO concentration-monopoly profit premise, however, antitrust is likely to be an extremely ineffectual means of reducing income inequality.

For one thing, there is no good evidence on how market power rents are distributed. The presumption is that they go to CEOs and shareholders. The evidence behind the first presumption is weak, at best, and some evidence cuts the other way. Moreover, it is also the case that some market power rents are not distributed to shareholders, but accrue to other stakeholders within firms, including labor.

Moreover, the numbers just don’t work out. In 2015, after-tax corporate income represented only about 10 percent of US national income. Market power rents represented only a fraction of those corporate profits. Market power rents that could be affected by more rigorous antitrust enforcement represented only a fraction–and likely a small fraction–of total corporate profits. If we are talking about 1 percent of US income the distribution of which could be affected by antitrust enforcement, I would be amazed. I wouldn’t be surprised if its an order of magnitude less than that.

With respect to how much of corporate income could be affected by antitrust policy, it’s worthwhile to consider a point mentioned earlier, and which the CEA raised: the distribution of corporate profits is very skewed. Further, if you look at the data more closely, very little of the big corporate profits could be affected by more rigorous antitrust–in particular, more aggressive approaches to mergers.

In 2015, 28 firms earned 50 percent of the earnings of all S&P500 firms. Apple alone earned 6.7 percent of the collective earnings of the S&P500. Many of the other firms represented in this list (Google, Microsoft, Oracle, Intel) are firms that have grown from network effects or intellectual capital rather than through merger for market power. They became big in sectors where the competitive process favors winner-take-most. It’s also hard to see how antitrust matters for other firms, Walt Disney for instance.

Only three industries have multiple firms on the list. Banking is one, and I’ve already discussed that: yes, it has grown through merger, but regulation and government are major drivers of that. There have also efficiency gains from consolidating an industry that regulation historically made horrifically inefficiently fragmented, though where current scale is relative to efficient scale is a matter of intense debate.

Another is airlines. Again, given the route network-driven scale economies, and the previous financial travails of the industry, it’s not clear how much market power rents the industry is generating, and whether antitrust could reduce those rents without imposing substantial inefficiencies.

Automobiles is on the list. But the automobile industry is now far less concentrated than it used to be in the days of the Big Three, and highly competitive.  Oil is represented on the list by one company: ExxonMobil. Crude and gas production is not highly concentrated, when one looks at the relevant market–which is the world. This is another industry which has seen a decline in dominance by major firms over the years.

Looking over this list, it is difficult to find large dollars that could even potentially be redistributed via antitrust. And given that this list represents a very large fraction of corporate profits, the potential impact of antitrust on income distribution is likely to be trivial.

(As an exercise for interested readers: calculate industry profits by a fairly granular level of disaggregation by NAICS code, and see which ones have become more concentrated as a result of merger in recent years.)

In sum, if you want to ameliorate inequality, I would put antitrust on the bottom of your list. It’s not where the money is because the kind of market power that antitrust could even conceivably address accounts for a  small portion of profits, which in turn account for a modest percentage of national income. Market power changes in many profitable industries have almost certainly been driven by major technological changes, and antitrust could reduce them only by gutting the efficiency gains produced by these changes.