It is unlikely that Mylan and Watson were slower than their rivals at noticing the efficiency effects of vertical integration, given their long histories and large scale of activities. More plausibly, their decisions were made in response to the expectation that generic drug markets were going to become increasingly vertically integrated. The next section discusses how we can test the two leading explanations for the increase in vertical integration within the generic pharmaceutical industry: the existence of bandwagon effects, and the importance of relationship-specific investments to support patent challenges.In the existing theoretical literature on vertical integration, bandwagon behavior is deemed to occur when a firm integrates in response to vertical integration by rivals . In generic drug markets, firms make their entry and vertical integration decisions more or less simultaneously so that we do not observe firms choosing their vertical structures in response to the actions of their rivals. Nevertheless, bandwagon effects can still exist in the sense that firms may become vertically integrated in response to the expected prevalence of vertical integration among rivals. Such a possibility can be examined by seeing if the change in a firm’s payoff from becoming vertically integrated is increasing in the incidence or prevalence of vertical integration among rivals. In other words, we can check whether firms’ payoff functions exhibit strategic complementarity in vertical integration decisions. As Buehler and Schmutzler point out, vertical integration decisions are shown to be strategic substitutes rather than complements in most theoretical models. However, hydroponic rack there are a few important studies such as Ordover et al. , Hart and Tirole , and McLaren that demonstrate the possibility of strategic complementarity.
For instance, one US cement company’s annual report for 1963 mentioned that while it was not inclined to acquire assets in the ready-made concrete industry, the wave of vertical integration among its rivals was forcing the firm to follow suit. I now show, using a simple duopoly model, that when firms’ payoff functions are characterized by strategic complementarity in vertical integration decisions, the following testable prediction arises: a firm’s probability of vertical integration decreases with its rival’s cost of vertical integration. When vertical integration decisions are strategic substitutes, the opposite result holds: the firm’s vertical integration probability increases with the rival’s cost of vertical integration. These results allow us to design a simple econometric test of strategic complementarity. The prediction that vertical integration facilitates early API development during a patent challenge can be tested by seeing if ANDA applicants who make a paragraph IV certification are more likely than other applicants to be vertically integrated. However, my dataset only records whether or not each market is subject to one or more entrants making a paragraph IV certification. I therefore construct a market-level variable that indicates the occurrence of a paragraph IV patent challenge. This indicator variable essentially signifies a switch in the entry process: markets with no paragraph IV patent challenge are characterized by simultaneous entry, while paragraph IV markets are characterized by a race to be first. The empirical strategy is to see whether this switch in the entry process affects firms’ incentives to become vertically integrated. Using the market-level paragraph IV indicator variable as a determinant of firm-level behavior introduces a potential endogeneity problem: markets that are the subject of paragraph IV certification may be attractive to generic entrants in unobservable ways, and those unobserved factors may also influence entry and vertical integration decisions.
This endogeneity can be taken care of by modeling the process of paragraph IV certification, and allowing the error term in the firm-level equations and that in the paragraph IV equation to be correlated. Many authors note that paragraph IV patent challenges have become more common in recent years . Patent challenges may also be more likely in larger markets that offer greater profits to the first-to-file entrant during the exclusivity period. In addition, Grabowski and Hemphill and Sampat note that certain types of secondary patents – particularly those that cover formulations and new uses – tend to be more vulnerable to patent challenge, presumably because it is easier to invalidate or avoid infringing such patents. This suggests the following as possible market-level determinants of paragraph IV certification: market size, the number of originator patents of different types, and year dummy variables. The generic drug markets used for analysis are selected from a database of the US Food and Drug Administration , called the Orange Book, which contains the population of all drug approvals. I begin by selecting a subset of drug markets that opened up to generic competition between January 1, 1993 and December 31, 2005. The set of markets is further narrowed down to those where the relationship between the upstream and downstream segments is relatively straightforward. This is done by first restricting the downstream products to finished formulations containing only one API. When there are multiple single-ingredient formulations containing a given API, I choose only the first of these to open up to generic competition. This is based on the belief that when generic companies make their entry decisions in the first downstream market for a given API, the upstream market structure is not yet formed.
Therefore, it makes sense to view downstream and upstream entry decisions as being made simultaneously. By the time the other downstream markets using the same API open up, the upstream market structure may already be fixed. Because it is not realistic to assume that upstream and downstream actions are decided simultaneously in such markets, they are excluded from the analysis. I also restrict the sample to the following dosage forms which constitute the majority of generic drugs: oral solids, injectables, and topicals. This leaves 177 downstream markets, each defined by a distinct combination of an API and a dosage form. 128 markets remain after removing observations for which market characteristics data could not be obtained. There are 125 corresponding upstream markets, each defined by a distinct API. For three APIs , two different dosage forms went generic on the same day. In these cases, I consider different dosage forms of the same API to constitute independent markets, and combine each of them with data for their respective API markets. Thus, for the three APIs mentioned above, the same upstream market data are used twice. Table A.1 in the Appendix contains a list of the drugs in the sample. A processed version of the FDA data was obtained from a proprietary database called Newport Sourcing, developed and maintained by Thomson Reuters. Table 2.1 and Figure 2.1 presented in Section 2.2 are constructed from the dataset of 128 markets. The econometric model is estimated using observations on 85 of those markets that opened up to generic competition between 1999 and 2005. The reason for restricting the time period in this way is as follows. Between 1992 and 1998, the FDA did not grant 180-day generic exclusivity to the first-to-file paragraph IV applicant. Therefore, during this period generic firms had little incentive to develop their products early in order to engage in patent challenges. Thus, the paragraph IV status of a market is likely to have been irrelevant for the decision to vertically integrate. By limiting the sample to the post-1998 period, we can analyze the role of paragraph IV certification more accurately. To record the two firm-level outcomes – downstream entry and vertical integration – it is first necessary to pinpoint the date when each market opens up to generic competition. Previous authors such as Scott Morton define the market opening date as the approval date of the first ANDA. After comparing ANDA approval dates with the dates when the generic products actually began to be marketed, I find that this definition is not always appropriate. In some cases, vertical growing systems the first generic product is not marketed until several months after its ANDA is approved. During those months,subsequent generic products are not approved by the FDA. I also find a few cases where drugs that appear to be generics are marketed before their ANDAs are approved. The first phenomenon arises when pending patent litigation between the generic entrant and the originator firm, or a settlement between the two, prevent the generic from entering immediately upon ANDA approval. The latter phenomenon is related to a practice called “authorized generics”: the originator gives the generic company a license to sell the product based on the former’s New Drug Application rather than the latter’s ANDA. To accommodate these special cases, I define the market opening date as the first generic approval date or the first generic marketing date, whichever is later. Firm-level entry actions are defined on the basis of market opening dates. Specifically, a potential downstream entrant is considered to have entered the downstream segment if its ANDA is approved by the FDA either before the market opening date or not later than one year after the market opening date.
The relatively narrow window is justified on the grounds that entry timing is an important determinant of profits in generic drug markets; because prices fall rapidly in response to additional entry, most firms enter in the first few months after market opening . As for actions in the upstream segment, a downstream entrant is deemed to have vertically integrated if it submits a Drug Master File to the FDA before the market opening date or no later than one year after the market opening date. I identify a potential downstream entrant in market m as a firm who has entered the downstream segment of any other generic market, including one outside the sample, on a date that is earlier than market m’s opening date but that is no more than five years before that date. Thus, I allow a firm to remain a potential downstream entrant for five years after its last entry. Similarly, a firm is identified as a potential upstream entrant of market m if it has entered the upstream segment of another generic market prior to, but not more than seven years before, market m’s opening date. Therefore, potential entrant status in the upstream segment is allowed to last for seven years after the last entry event. The reason for setting a wider window for potential upstream entrants is that DMF submissions sometimes occur a few years before the market opening date. Firm i is a potential upstream-only entrant in market m if it is a potential upstream entrant but not a potential downstream entrant. To evaluate the potential entrant status of a given firm, it is necessary to accurately identify its previous entries. This requires correct names for the ANDA applicants and DMF holders contained in the FDA data. Similarly, identifying firms’ vertical integration actions, which involves matching the firms found in the downstream ANDA database with those in the upstream DMF database, requires accurate data on firm names. These tasks are complicated by the several mergers and acquisitions that took place in the generics industry during the observation period. As described in Appendix A.2, I use the Newport Sourcing database to attach accurate firm names to the FDA data. Changes in firm ownership are taken into account by assuming that the past entry experience of an acquired firm is fully carried over to the acquiring firm. Table 2.5 presents the distribution of actual entry actions taken by potential downstream entrants in the dataset. The data consist of 92 firms facing 2,539 choice situations spread across 85 markets. 406 of these choice situations result in downstream entry. 76 of the downstream entries lead further to vertical integration. Table 2.6 presents summary statistics for the covariates. The first fourteen variables are market characteristics. “User Population” is a measure of market size, which is expected to have a positive impact on a firm’s probability of downstream entry . However, its impact on a firm’s propensity to vertically integrate is an open question: while Stigler hypothesizes that vertical integration would occur less frequently in larger markets, others note that under certain conditions, the incidence of vertical integration may actually rise with market size. The user population variable is defined as the estimated number of users of each drug in the US during the period immediately before generic entry. It is constructed from results of the National Ambulatory Medical Care Survey and the National Hospital Ambulatory Medical Care Survey .