After five grueling weeks, the Department of Justice's challenge to the proposed merger between AT&T and Time-Warner reached a milestone of sorts on Tuesday with the government resting its case. The centerpiece of the government's argument has been the model and analysis of economist Carl Shapiro, a former Justice Department official now at the University of California at Berkeley predicting the likely behavior of the merged firm.
Shapiro's model suggests that the merged firm will engage in a form of behavior known as strategic foreclosure, where it will engage in strategic bundling of its products and charge higher prices for its stand-alone components after the merger than the separate firms would have done had the merger not proceeded. Rivals buying individual products to include in their own bundles (or consumers building their own bundles from individual component offerings) face higher prices, while those producing competing products and bundles must discount their own offerings to compete with the deeper bundle discounts offered by the merged firm. While some consumers benefit from the lower bundle prices, others will not. Some rival firms will be unable to survive and must leave the market, while others who may have entered will not - that is, they will be foreclosed from competing. In the long run, this reduces the extent of product variety to the overall detriment of consumers and society as a whole.
Whinston and Choi
The thinking driving Shapiro's model derives from a seminal 1990 American Economic Review paper by MIT economist Michael Whinston. Whinston developed a model where a monopolist producing a product A could "tie" the sale of it to another product B1, which faced competition from the provider of a rival B2, and price the bundle strategically to make it unviable for the rival to remain in the market. "Tying" is a special form of bundling, where the products A and B1 are only ever available together - so a customer wanting A must also buy B1 (and vice-versa).
Michigan State's Jay Pil Choi's 2008 paper expanded on Whinston's model to allow for mergers including "mixed bundling." In Choi's model, four firms produce four products in two markets - A1 and A2, and B1 and B2. Two firms merge and sell their products in a bundle - A1B1. However, they still sell the components separately, so that three other "bundles" can be constructed to compete with theirs - A1B2, A2B1, and A2B2. Choi's model provides the basic intuition for Shapiro's assertions that the merged firm may both bundle its own products and discount them steeply but raise the prices of the components when sold separately.
Importantly, however, Choi found that it will not always be profitable for the merged firm to engage in strategic pricing to foreclose its rivals, as it must trade off the gains from extra sales of the components in the bundle against the lost sales from pricing the separately-sold components higher. It now matters whether or not the components in the relevant markets are close substitutes (where consumers are indifferent to whether they buy A1 or A2, or B1 or B2). If the products are not close substitutes, each bundle can be seen as a separate product, and there is little direct competition between them. In this case, there is little harm to total welfare from the merger proceeding. However, where there are high degrees of substitutability between the bundles, then competition between them intensifies, and harm to welfare is more likely.
The nuances in Choi's paper indicate the importance of both the form of bundling engaged in and how consumers view the various products in the bundles when assessing the effects of a merger where bundled products are involved. However, it also raises the question of how much reliance can be placed in an antitrust case on analysis based on sophisticated theoretical models that rely on a number of assumptions and simplifications for their mathematical tractability. The latter has engaged much of the attention of the defense lawyers in the AT&T-Time Warner case. Elegant though the models may be, they are of little help when they are not an accurate depiction of the behaviors actually observed in the real world cases.
Unfortunately for the Department of Justice, in the real world in which AT&T and Time-Warner operate, the ways in which they bundle and sell their products and the way consumers behave in the face of these offers are much more complex than the simple models.
Many products, different valuations
For starters, both Choi's and Whinston's models presume that consumers purchase at most one product in each category, and they are always purchased in strict proportions to each other. This may well have been the case when considering a single internet connection and a single pay television offer. However, the current reality is that the internet and content (application) bundles offered are now no longer simple one-to-one packages. Consumers can purchase many different forms of internet access from the same or different suppliers - mobile, fixed, satellite, or Wi-Fi. Furthermore, they do not purchase just one entertainment bundle. Yes, they may purchase a television package, but they can also purchase (or have "free" access to) a whole range of other internet-based video and other entertainment packages - Netflix, Hulu, Spotify, or YouTube for example - in bundles or stand-alone from a wide variety of channels. It is far from clear that the same theoretical outcomes will eventuate when proportionality assumptions are violated.
Furthermore, Choi's model also assumes that the products in the bundle are strict complements. That is, any customer buying anything will buy both products. It is not possible for a customer to buy just an internet connection but not a content package. Moreover, it presumes that consumer demand for (valuations of) the products in the bundle is both linear and symmetric - which is most unlikely to be the case for (say) an internet connection and a television content package, let alone any of the many different contents that compete for inclusion in consumers' purchase sets in the internet and content space.
Models and decisions
Ultimately, the relevance of a model for understanding the likely effects in an antitrust case hinges on its ability to capture important real-world effects. The bundling observed in the real world of the AT&T/Time Warner case involves levels of complexity far beyond what these simple models have been able to capture. If Shapiro's model does not address these complex realities, it may not be helpful for the decisions that must ultimately be made.
* Department of Justice (DOJ)
© Copyright © 2018 The American. All rights reserved., source Newspapers