The section also discusses the possible motives for, and effects of, vertical integration in this industry. Section 3.4 presents the econometric specification, followed by a description of the estimation strategy in Section 3.5. After describing the data in Section 3.6, I present the estimation results, including the findings from policy simulation, in Section 3.7. A concluding section follows.Vertical integration has three main types of efficiency effects. The first one is the elimination of double marginalization . Double marginalization occurs when oligopolistic markups are charged in both the upstream and downstream segments. By eliminating the markup in the upstream segment, a vertically integrated firm enjoys a cost advantage over its unintegrated downstream rivals. Chipty finds evidence from the cable TV industry that is consistent with the elimination of double marginalization by vertically integrated firms. The second type of efficiency effect arises from the ability of vertically integrated firms to carry out higher levels of non-contractible relation-specific investments . There is an abundance of empirical research – recent examples of which include Woodruff and Ciliberto – indicating the existence of such investment-facilitation effects. The third type of efficiency effect relates to the ability of vertically integrated firms to secure the supply of an intermediate good or more generally, vertical growing racks to improve coordination in logistics. Theoretical models that explore this aspect of vertical integration – namely, Carlton and Bolton and Whinston – find that the overall effect on market outcomes is indeterminate. Meanwhile, Hortac¸su and Syverson’s empirical analysis of the cement and ready-made concrete industries finds that vertical integration motivated by logistical concerns has a price-lowering effect.

One issue that has not been addressed in the literature is the possibility that the positive efficiency effects of vertical integration may spill over to other firms that are not vertically integrated. Such efficiency spillovers would reinforce the price-lowering or quality-enhancing effects of vertical integration. Foreclosure typically occurs when a vertically integrated firm restricts supply of the intermediate good with an aim to raise the final good price. The significance of such practices has been the subject of continuing debate; Riordan summarizes the notable theoretical models that have shaped the discussion. The models are roughly divided into two groups: models where vertical integration raises downstream rivals’ costs by dampening competition in the upstream market , and models where vertical integration allows upstream units to restrict the supply of the intermediate good and restore monopoly power . Most of the empirical analysis on vertical foreclosure looks directly at the effect of vertical integration on market outcomes. A general conclusion from this literature is that the effect of vertical integration varies across industries; higher final good prices due to vertical integration is found in some industries but not in others. This may be because foreclosure effects exist only in certain industries. It is also possible that in many industries, any foreclosure effects that do exist are offset by the efficiency effects of vertical integration. Useful experimental evidence on vertical foreclosure exists. Normann finds that vertically integrated players often employ strategies that raise their rivals’ costs, as in Ordover et al. . Similarly, Martin et al. demonstrate that the monopoly restoration model of Rey and Tirole is partially supported by experimental data. Thus, the experimental literature provides support for vertical foreclosure theory, not least because efficiency effects are absent by design. Rosengren and Meehan and Snyder look at the effect of vertical integration on rival profits to make inferences about vertical foreclosure.

Both papers focus on the effect of a vertical merger announcement on the stock prices of unintegrated rivals. Rosengren and Meehan do not find that vertical mergers have a significant effect on independent downstream rivals. Thus, they find no support for foreclosure theory. Christopher Snyder’s study of the British beer industry, described in Snyder , finds that an independent upstream brewery was harmed by vertical integration between rival breweries and downstream pubs. He interprets this as support for foreclosure theory. A common feature of the existing empirical work on foreclosure is that they assume exogenous changes in market structure. A defining feature of recent studies such as Hastings and Gilbert and Suzuki has been to design dataset construction and estimation methods so that the exogeneity assumption can be made plausible. Hart and Tirole’s theoretical paper contains some analysis on the effect of vertical integration on market structure formation. Essentially, the changes in profits brought about by vertical integration may induce unintegrated firms to become integrated themselves or to exit the market. Ordover et al. also investigates the possibility that vertical integration by one firm may lead another to become vertically integrated. The possibility that vertical integration can affect the market structure formation process – in other words, that vertical integration exhibits “market structure effects” – is an area that has only recently begun to receive attention from empirical economists. The leading example is Hortac¸su and Syverson . They find that in the cement and ready-made concrete industries, unintegrated upstream firms had higher exit probabilities in markets where a higher proportion of entrants were vertically integrated. This was apparently caused by higher productivity levels among vertically integrated firms. In other words, the efficiency effects of vertical integration may have led unintegrated upstream firms to exit.

The vertical entry model presented in this chapter is designed to estimate the effect of rival actions, including vertical integration, on firm payoffs. For instance, the estimated parameters can be used to calculate how an unintegrated firm’s payoff changes when a rival pair consisting of one upstream firm and one downstream firm is replaced by a vertically integrated one. In this sense, the model is closest in spirit to the event studies of Rosengren and Meehan and Snyder that look at the effect of rival vertical integration on firm value. The weakness of the Rosengren and Meehan and Snyder studies – that the impact of vertical integration on market outcomes is not directly observed – is thus shared by the current model. A major concern is that foreclosure effects and efficiency effects often affect unintegrated firms’ profits in the same manner, and thus tend to be indistinguishable . For example, if an unintegrated downstream firm’s profit decreases as a result of rival vertical integration, it could be due to a foreclosure effect, an efficiency effect, or both. Therefore, even if a significant payoff impact is found, one may not be able to conclude anything about the existence of either of these effects. The advantage of my model is that different types of payoffs can be observed. For a few of the payoff functions, the direction of foreclosure effects is different from that of efficiency effects, so that one is distinguishable from the other. For example, if we find that unintegrated upstream profits increase in response to rival vertical integration, the existence of foreclosure effects is implied. This is because efficiency effects can only have a negative impact on an unintegrated upstream firm’s payoff. Similarly, if unintegrated downstream profits increases in response to vertical integration, it must be due to the positive spillover of efficiency effects, because any foreclosure effect would affect unintegrated downstream profits negatively. Another characteristic of the vertical entry model is that, unlike in existing studies such as Hastings and Gilbert and Suzuki , one need not assume that firms’ vertical integration decisions are exogenous. In fact, entire market structures, including the vertical integration status of individual firms, are modeled as endogenous outcomes. This implies that the data requirements for the current model may, in some sense, not be as demanding as that of existing methods. There is, however, a rather stringent requirement that the dataset contain observations from multiple markets where complete vertical market structures are observed. An additional strength of the vertical entry model lies in its ability to examine how vertical integration influences market structure formation. In addition to asking what happens to an unintegrated firm’s payoff when a rival pair becomes integrated, vertical grow room design one can ask whether the payoff impact is so large that the firm’s entry decision changes. In this connection, a useful application of the model is to evaluate the effect of a policy that bans vertically integrated entry. How does such a ban affect the number of entrants in the upstream and downstream segments? While the answer is not clear a priori, the model and parameter estimates can be used to obtain one by simulation. This field has been motivated by the technical challenge of how to handle the number of rival entrants – a variable that is clearly endogenous – as a key argument of the firm’s payoff function. The earliest studies are Bresnahan and Reiss and Berry .

Building on the pioneering work of Bjorn and Vuong , their econometric models explicitly allow market structure outcomes to be equilibria of entry games. For example, Berry’s model contains an equilibrium finding algorithm that is run at each iteration of the parameter search. These early papers focus exclusively on horizontal competition among firms that produce a homogeneous good. Coefficients on the number-of-rivals variables represent rival effects; from them, information on the degree of competitiveness in the market can be inferred. More recent papers such as Mazzeo , Seim , and Orhun expand the entry model framework to allow for product and spatial differentiation. For instance, in Mazzeo’s study of motel markets, potential entrants choose between entering the low-quality segment or the high-quality one. His results provide insight not only into the degree of competition within and across different market segments, but also into the process of market structure formation. For instance, the estimated parameters are used to predict how the product-differentiated market structure changes in response to increases in population and traffic. Another group of papers uses the entry model framework to investigate the existence of complementarities in firm actions . For example, Vitorino finds that the existence of agglomeration effects allows stores to profit from co-locating inside shopping centers. The present model examines the formation of vertical market structures in which suppliers and buyers trade and compete. As in the papers on horizontal entry, the estimates provide information on the degree of competition within each vertical segment. In addition, the complementarity between upstream entry and downstream entry can be examined. Finally, and most interestingly, the model should provide evidence on the competitive role of vertically integrated firms. Do vertically integrated firms hurt upstream rivals more than they harm downstream ones? Can some firms benefit from facing a vertically integrated competitor instead of an unintegrated pair of firms? Such questions are empirical in nature, and answering them is the subject of this chapter.This section describes the process of vertical market structure formation in the generic pharmaceutical industry to motivate the econometric model. As described in Chapter 2, drug markets open up to generic competition when patents and data exclusivities that cover the drug expire. In each market, generic drug manufacturers make entry decisions a few years before the market opening date. If an upstream unit decides to enter, it develops the active pharmaceutical ingredient and submits a dossier, called the Drug Master File , to the Food and Drug Administration . A downstream entrant, on the other hand, procures the API – either from an outside supplier or from its own production – and develops the finished formulation. It then conducts bio-equivalence tests using the finished product and files an Abbreviated New Drug Application to the FDA. Two peculiar aspects of the generic entry process need to be addressed before providing a stylized description. The first is the possibility of patent challenge by generic entrants. As described at length in Chapter 2, the regulatory rules governing generic entry incentivize generic entrants to challenge the ability of originator patents to block entry, by way of a 180-day generic exclusivity awarded to the first-to-file paragraph IV ANDA applicant. The existence of such incentives pushes firms into a race to be first whenever a paragraph IV patent challenge is involved. The economics of such a race is very different from that of a conventional entry game where firms move simultaneously. For this reason, this chapter focuses only on markets that are not subject to a paragraph IV patent challenge.