Faculty of law blogs / UNIVERSITY OF OXFORD

Policy Failure: The Role of “Economics” in AT&T-Time Warner and American Express


Marshall Steinbaum
James Biese


Time to read

11 Minutes

The recent AT&T and Amex decisions showcase the pitfalls of considering antitrust cases solely on the basis of economic analysis and may have the effect of immunizing tech giants from serious antitrust scrutiny, argue Marshall Steinbaum of the Roosevelt Institute and James Biese in this op-ed.

Antitrust cases are increasingly driven solely by “economic” analysis. Yet in many cases, economic evidence provides false certainty that leads courts astray. The very idea of consumer harm under the antitrust laws has largely become synonymous with economic inefficiencies and net costs to consumers, typically in the form of higher prices. This restrictive view of antitrust requires courts and enforcers to consider false, biased, and irrelevant evidence in satisfaction of misguided legal requirements supposedly grounded in economics. As a result, the real-world economic problems that should be the focus of antitrust economists are too often ignored.

Two recent blockbuster antitrust cases illustrate the pitfalls of relying on the wrong economics when enforcing the law. First, Judge Richard Leon’s ruling in U.S. v. AT&T allowing the AT&T-Time Warner merger to proceed was a spectacle of flawed economic analysis on both sides of the courtroom. Second, the Supreme Court’s ruling in Ohio v. American Express accepted as gospel a skewed understanding of economic theory concerning “two-sided platforms” that may have the unfortunate effect of immunizing tech giants like Google, Amazon, Uber, and Facebook, as well as powerful labor market monopsonists, from serious antitrust scrutiny.


In what some called the merger “trial of the century,” AT&T faced off against government antitrust lawyers, fighting over whether AT&T could purchase Time Warner. But the real fight was the battle of economists, with star government expert Carl Shapiro—a professor and former Department of Justice official—against star AT&T expert Dennis Carlton.  Both economists fell into serious analytical and credibility traps.

Shapiro, for his part, came out of the gate conceding that AT&T had an excellent case. The government and Shapiro accepted an old defendant-friendly theory in vertical integration (the ownership of multiple stages of the supply chain by a single firm): that consolidation results in the “elimination of double-marginalization,” which means that instead of two firms charging a profit, the vertically integrated entity only charges once, and that profit is less than the sum of its parts. In this case, AT&T would apparently charge its Dish and U-Verse subscribers less because of eliminated margins on Time Warner content, to the tune of $352 million per year.

The reason that this $352 million-per-year concession was so problematic was because the government’s theory of the case boiled down to an allegation that the merger would cause prices to go up. Rather than starting from zero, however, the government started out in a $352 million per year hole. (And, in the real world, AT&T’s prices did go up almost immediately after the merger was approved.)

The elimination of double marginalization also counts in favor of a merger only if it creates truly offsetting benefits to the potential harms caused by a merger. In other words, benefits to one set of stakeholders (in this case, AT&T’s existing customers) cannot offset harm to others (customers of its TV distribution rivals), especially when there is no mechanism in the merger to compensate those who are harmed.

Additionally, and importantly for AT&T, one must assume that the cost savings from the merger cannot be had through alternative business strategies short of a merger. However, for as much as AT&T touted the possibility of new business models with the merger, no one, and certainly not Shapiro, seriously examined whether these new business models could either partially or wholly be employed without having to merge.

Shapiro’s empirical prediction that AT&T’s customers would benefit so immensely from the merger was therefore a highly questionable strategic move. Unfortunately, it was not the only one.

Shapiro’s prediction that prices would go up for customers of AT&T’s competitors relied upon a theoretical economic model that attempted to show that Time Warner would have increased bargaining leverage due to its ability to “black out” unaffiliated distributors and still reach AT&T’s customer base. The basic idea is that, facing a loss of their own subscribers to AT&T should Time Warner content become unavailable, those rivals would be forced to deal on Time Warner’s terms and pass the ensuing price increases along to their customers. Exactly how bad the resulting price increases would be significantly depends on one of the assumptions baked into Shapiro’s model: how likely a rival’s subscribers are to leave. If that “loss rate” is high, Time Warner has all of the bargaining power and will charge a higher price. Crucially, no extended blackout is likely ever to occur—it is the threat that one might occur that influences prices.

"The AT&T-Time Warner merger is a significant step in turning the telecoms sector into a series of 'walled gardens' where consumer choice of content is determined by their Internet service provider or by powerful tech sector platforms. We can and should question why these issues were not aired at the trial in any significant way."

In estimating the “loss rate,” Shapiro apparently did not look too deeply into the numbers on which he relied, which had methodological problems of their own. Shapiro was forced to concede that under an alternative set of numbers, the net cost increase that he projected in his model would be largely eliminated. Because the government’s case rested heavily on showing price increases to consumers for Time Warner content, this was tantamount to an admission that no harm would result from the merger—yet another signal from the government’s economic expert that AT&T had an excellent case.

The trial judge sided with AT&T on almost every point relating to Shapiro’s model. He noted that as it became clearer over the course of the trial that Shapiro’s model had crumbled under cross-examination, the government increasingly “minimized” it. The judge even compared the model to a “Rube Goldberg” machine, commenting wryly that “at least [Mr. Goldberg’s] contraptions would normally move a pea from one side of a room to another.”

Such a contraption is inevitably vulnerable on multiple points related to the assumptions necessary to sustain it. For that reason, academic economic research has moved away from heavy reliance on theoretical models, instead opting to estimate real-world effects directly. To be sure, there are still faults in this approach, but this development within the economics profession does call into question the government’s choice to place so many eggs in the basket of Shapiro’s model. Given the courts’ apparent reliance on this type of speculative economic theory, however, the government may have had no choice—either go along with whatever testimony Shapiro was willing to offer (which itself may have been constrained by his own history of expert testimony on behalf of defendants), or don’t challenge the merger at all. That’s not a tenable place for antitrust and specifically merger policy to be, given that the law prohibits mergers in order to prevent the accumulation of market power “in its incipiency.”

On the other side of the courtroom, AT&T’s economist only added to the economic trainwreck. AT&T’s lead expert, Dennis Carlton, offered a transparently Pollyannaish view of the merger that assumed no costs to consumers. He also performed econometric analysis of past vertical mergers, suggesting that they too have not caused any competitive harm. These types of arguments could be expected from AT&T’s economist—but they were particularly suspect here because of Carlton’s history of advancing incorrect arguments in other merger cases.

For example, Carlton also worked for AT&T in its blocked merger with T-Mobile. In that case, he elaborated what’s known as a “failing firm defense.” Essentially, he predicted that if AT&T and T-Mobile were not permitted to merge, T-Mobile would go out of business. They were not allowed to merge, and T-Mobile did not go out of business. That and similar failed predictions in past merger cases cast serious doubt on Carlton’s sunny predictions in the AT&T case; yet, perhaps because of his counterpart’s numerous economic missteps, Carlton’s analysis was left looking more credible than the government’s alternative.

This battle of economists, out of touch as it may seem, is important because the probable consequences of the merger are so significant. AT&T has its tentacles in many markets critical to the economy of the future, and contrary to the economic analysis in the AT&T case, it does more than simply distribute television content to paid subscribers. It is also one of the largest wireless carriers in the country, as well as an Internet service provider.

The problems with the AT&T-Time Warner merger therefore extend well beyond the price of Time Warner’s content. It now seems clear that Time Warner assets will be leveraged to entrench AT&T’s wireless business. And it is now set to either withhold HBO and other streaming content from its wireless rivals, or at the very least prioritize subscribers to AT&T’s wireless service. Without Net Neutrality to protect the public anymore, AT&T has both carrots and sticks available to destroy competition: the ability to promote its own streaming content (through “zero rating”), and the ability to hold the threat of throttling competitors’ streams over their heads to chill competition.

These actions do not represent the type of innovation or technical change that could be thought to benefit the public, and yet the economists testifying in the case didn’t really speak to any of these issues. Rather, the economic evidence shows that proliferating vertical integration and new ways of using technology to discriminate in favor of one’s own products are causing substantial harm to the economy. The AT&T-Time Warner merger is a significant step in turning the telecoms sector into a series of “walled gardens” where consumer choice of content is determined by their Internet service provider or by powerful tech sector platforms. We can and should question why these issues were not aired at the trial in any significant way.

Ohio v. American Express

The Supreme Court’s decision in Ohio v. American Express is a master class in how to use industry-backed theoretical economic speculation to overturn findings of fact. In that case, several states and the federal government accused American Express of violating the antitrust laws by using “anti-steering” provisions in its contracts with merchants. This so-called non-price vertical restraint prevents merchants from steering or incentivizing consumers to use other credit cards that may have lower fees for the merchants—for example, the merchants may pay less to Discover to use its network and could suggest that their customers use a Discover card rather than an American Express card at the checkout counter. On a broad scale, that “steering” practice would ultimately lower consumer prices as merchants are forced to pay less to credit card companies, and credit card companies would actually have to compete with each other.  However, with a contractual “anti-steering” provision in place, a company like American Express does not need to worry about being undercut on price by another credit card company because merchants are prohibited from steering customers away from more expensive cards.

If this sounds like a straightforward attempt to artificially inflate prices and destroy competition, that’s because it is. However, American Express argued that there is no antitrust problem unless the government could show net harm to both merchants and cardholders. The economic theory behind this argument relates to so-called two-sided markets, or markets where two distinct sets of counter-parties interact through a platform or network that connects the two. For credit cards, that means merchants are matched with cardholders through the credit card network for a transaction. However, there are numerous other examples of two-sided markets in nearly every industry, from things like newspapers to tech platforms like Google, Amazon, and Uber. The designation is therefore so general as to be near-useless as a guide to policy.

"There is at least an appearance that the credit card industry actually created its own legal loophole under the auspices of supposedly neutral economic research."

In American Express, the trial court found that the anti-steering provisions had the effect of increasing merchant fees, without any offsetting pro-competitive effect on prices for consumers. This makes sense: if businesses have to pay more to credit card companies every time they swipe a card, some or all of the added cost will likely be passed on to consumers. But that finding and logic was ignored by the Supreme Court, which held that the government had not actually proven harm on both sides of the market: higher fees for merchants, and reduced output for customers. Because the number of credit card transactions has increased over the relevant time period, the Supreme Court determined that “output” actually increased, and thus, there was no consumer harm from the anti-steering provisions.

This new way of defining a market is at odds with empirical economic research that holds that if anything, markets are too narrowly defined and customers (or workers, or suppliers) usually substitute between a small number of alternatives. And it is not at all clear that there is an obvious link between what the Supreme Court referred to as “output” and the prices charged to credit card merchants or consumers. Again, this is for a common-sense reason: most consumers don’t know how much the merchant has to pay to the credit card company when they hand their cards over at the checkout counter, and thanks to restraints like American Express’s non-steering provisions, merchants are actively blocked from giving consumers the facts. Both consumers and merchants can be harmed by paying higher prices—as high profits in the credit card industry strongly suggests—even as the number of transactions rise.

The economic reasoning for such a wrongheaded result in American Express was supplied by a body of economic research with disconcerting links to the credit card industry. Thus, there is at least an appearance that the credit card industry actually created its own legal loophole under the auspices of supposedly neutral economic research. 

While the lingering questions over the reliability of credit card industry–funded research are bad enough, it is clear that the defendant-friendly burdens of proof endorsed by the court in American Express will be used as a blueprint to chill antitrust enforcement in other industries. Despite an ill-defined attempt by the majority opinion to limit its reach to markets similar to the credit card market, past experience shows that lower courts will read the theoretical speculation in the Supreme Court opinion broadly. Indeed, defendants in two labor market monopsony cases, one pertaining to the NCAA and the other to UFC Mixed Martial Arts, are already arguing that, just like the Supreme Court’s decision in American Express, they cannot be held liable for exercising their monopsony power over their workers unless plaintiffs show harm to consumers. This is yet another example of the fact that existing antitrust law is not strong enough to target the issue of labor market monopsony.

Furthermore, we know that some of the most flagrant exercises of market power apparent in the modern economy are to be found in tech-sector platform firms who will attempt to claim the type of “two-sided market” mantle that was effectively immunized by American Express.

Questions of Expertise

The decisions in both AT&T and American Express are significant setbacks for anyone who thinks competition policy has been insufficiently enforced and the economy is suffering as a result. And both outcomes depended to a significant degree on flawed antitrust economics.

Unfortunately for the American public, the economic issues debated in both cases will now become a real-world experiment with their pocketbooks, their smartphones, their Internet connections, and their TV sets. Restructurings and consolidations in both telecoms and tech are already underway. All of us will soon know exactly what the government and the Supreme Court missed in their economic analyses, and what those things will really cost us.

Reprinted with permission from promarket.org – the blog of the Stigler Center at the University of Chicago Booth School of Business.


With the support of