This paper, available here, argues that incumbent firms may acquire innovative targets solely to discontinue the target’s innovation projects and preempt future competition.
Using pharmaceutical industry data, the paper shows that acquired drug projects are less likely to be developed when they overlap with the acquirer’s existing product portfolio, especially when the acquirer’s market power is large. Conservative estimates indicate 5.3% to 7.4% of acquisitions in the authors’ sample are killer acquisitions, which occur disproportionately just below thresholds for antitrust scrutiny.
Section 2 outlines the theoretical framework and develops testable hypotheses.
The authors first build a parsimonious model that combines endogenous acquisition decisions, innovation choices and product market competition. The model looks at acquisitions that occur when the innovative target firm’s project is still under development, and therefore further development is necessary and costly, and the ultimate project success is uncertain. An incumbent acquirer has weaker incentives to continue development than an entrepreneur if the new project overlaps with (i.e. substitutes for) a product or project in the incumbent’s portfolio.
The model shows that this disincentive to innovate can be so strong that an incumbent firm may acquire an innovative start-up simply to shut down the start-up’s projects and thereby stem the “gale of creative destruction” flowing from new inventions. Importantly, some degree of acquirer-target overlap is necessary for the ‘killer acquisition’ motive to exist. Finally, the model shows that killer acquisitions continue to exist even when the entrepreneur’s new project is qualitatively superior to the incumbents’ existing projects or products, when incumbents benefit from development synergies relative to the entrepreneur, and when there are multiple (asymmetric) potential acquirers.
Section 3 describes the empirical data.
To obtain empirical support for the model raises important challenges. One needs to observe project-level development activity and track projects as they move across firms. It is also crucial to accurately measure overlap between the acquiring firm’s portfolio and the target’s project, and to quantify competition in the relevant product market. Pharmaceutical drug development has features that address all of these challenges.
The authors collected detailed development information on more than 16,000 drug projects originated by more than 4,000 companies in the past two and a half decades. They also collected relevant acquisition events from comprehensive data sources which allows them to observe development milestones of drug projects independent of project ownership, i.e. the authors can follow the same projects pre- and post-acquisition.
To establish acquirer overlap with the target’s project, and thus identify potentially competing products, the authors use pharmaceutical categories based on disease and mechanism. More specifically, if the target’s drug project is in the same therapeutic class (e.g. antihypertensive) and uses the same mechanism of action (e.g. calcium channel antagonist) as a drug product or project in the acquirer’s portfolio, the authors consider that acquisition to be an overlapping acquisition. Measuring overlap this way helps to ensure one is capturing potential substitutes (i.e. companies developing drugs that, if successful, would directly compete with the acquirer).
Section 4 presents the main empirical results.
The paper compares projects acquired by overlapping incumbents to those acquired by non-overlapping incumbents and to non-acquired projects. The baseline regression uses a project-year panel to estimate the annual probability of development activity (i.e. lack of project termination). Following the logic of killer acquisitions, the authors expect a decreased likelihood in the development of overlapping projects post-acquisition. In line with this, they find that projects acquired by an incumbent with an overlapping drug are 23.4% less likely to have continued to pursue development activity when compared to drugs acquired by non-overlapping incumbents. At a more fine grained level, decreased development for overlapping projects post-acquisition is driven by drugs that have no further development activity post-acquisition (i.e. by immediate and permanent terminations). Such projects acquired by overlapping acquirers are 20.9% more likely to cease development immediately when compared to those acquired by non-overlapping incumbents.
The model developed in the earlier sections also predicts that incumbents have a stronger incentive to acquire and terminate overlapping innovation in ex ante less competitive markets (i.e. when the incumbent has more to lose if the target’s innovation is successfully developed). To examine this, the authors repeat the baseline analysis in subsamples with low and high levels of existing competition (as measured by the number of competing drugs in the same therapeutic class and mechanism of action), and find that the decrease in development probability for acquired overlapping projects is concentrated in markets with low competition. The model also predicts that when the incumbent’s drug is far from patent expiration, and thus generic competition, incumbents have a stronger incentive to acquire and terminate innovation. Accordingly, the authors find that the decrease in development rates is concentrated in overlapping acquisitions for which the patent on the acquirer’s overlapping drug is relatively far from expiry.
In additional empirical tests, the authors examine the progression of projects through the phases of clinical trials by focusing on projects that start Phase I trials and examining their likelihood of starting Phase II. They find that drug projects are 46.6% less likely to enter Phase II if they are acquired during Phase I by an acquirer with an overlapping drug. As in the main analyses, these findings are concentrated in markets with low competition.
Sections 4 and 5 discuss the practical implications of the results.
The empirical results suggest that killer acquisitions are both strategic and intentional. First, and in line with the model, acquisitions are almost four times more likely when the incumbent acquirer’s products overlap with the target project. Second, acquirers conducting killer acquisitions are much more likely to undertake acquisition deals that do not trigger FTC notification requirements for premerger review, and thereby avoid antitrust scrutiny. Acquisitions of overlapping targets bunch just below the FTC acquisition transaction value threshold, while there is no such pattern for non-overlapping acquisitions. In addition, these below-threshold deals exhibit much higher termination rates and much lower product launch rates.
The authors consider, and exclude, a number of alternative explanations.
- One alternative explanation is optimal project selection under which, for multiproject targets, the acquirer could strategically and optimally choose to continue only the most promising projects while discontinuing those that are less promising. To assess this concern, the authors repeat their analysis for acquisitions of single-drug companies. The results are robust in implying that optimal project selection does not explain the results.
- Another alternative explanation is capital redeployment, in which the acquiring firm’s intention is to acquire and redeploy the acquired target’s core assets – i.e. its underlying technology or human capital – to more productive uses. To address this possibility, the authors separately consider technology and human capital redeployment. To explore technology redeployment, they track the chemical similarity of acquired drugs to pre- and post-acquisition projects of the acquirer – and find no evidence supporting the idea that acquired technologies are integrated into acquirers’ new drug development projects, or that acquirers are more likely to cite acquired and terminated projects’ patents. Regarding human capital redeployment, the authors show that only 22% of inventors from target firms end up working for the acquiring firm. Further, those inventors do not become more productive post-acquisition. These results are inconsistent with explanations regarding technology or human capital redeployment.
- A third alternative explanation is “salvage” acquisitions. in which overlapping acquirers buy already-failing targets to (cheaply) acquire the target’s valuable assets. Following this logic, decreases in development would predate the relevant acquisition, which would also have lower post-acquisition development activity. Contrary to the salvage explanation, however, there is no evidence that overlapping acquisitions have pre-acquisition declines in development or lower valuations, on average, compared to non-overlapping acquisitions.
Section 5 also discusses policy implications.
The authors’ conservative estimates indicate that between 5.3% and 7.4% of all acquisitions in their sample (or about 46 to 63 pharmaceutical acquisitions per year) are killer acquisitions. Eliminating the adverse effect on drug project development from killer acquisitions would raise the pharmaceutical industry’s aggregate drug project development rate by more than 4%.
Despite the ex-post inefficiencies of killer acquisitions and their adverse effect on consumer surplus, the overall effect on social welfare is ambiguous because these acquisitions may also increase ex ante incentives for the creation of new drug projects. However, the authors think it unlikely that acquisitions which generate significant ex post inefficiencies resulting from the protection of market power are the most effective way to motivate ex ante innovation. Instead, and because killer acquisitions are less likely to occur when incumbents face significant existing competition, raising the level of existing competition not only leads to immediate benefits for social welfare but it also deters incumbents from engaging in killer acquisitions of future.
In any event, this paper’s results also suggest that antitrust policy should continue to closely scrutinise the impact of acquisitions on corporate innovation, in particular when such acquisitions plausibly prevent the development of future competing products and technologies. The fact that killer acquisitions appear routinely to avoid regulatory scrutiny by acquiring entrepreneurial ventures at transaction values below the HSR review thresholds exacerbates the concern.
Section 6 concludes.
The paper makes three main contributions to the field. First, it sheds new light on a fundamental impediment to corporate innovation. Specifically, it highlights how the protection of existing products, which is well-known to discourage an incumbent’s own innovation efforts, can also incentivise powerful incumbents to stifle the innovation of other firms. Although the analysis here focuses on the pharmaceutical sector, the core insights extend beyond that specific setting. Acquisitions are the primary form of start up exit and have become increasingly popular as an exit strategy over time across various industries, suggesting that the potentially damaging consequences of killer acquisitions reach beyond pharmaceuticals. The results of this paper caution against interpreting acquisitions of nascent technologies solely as incumbents’ efforts to integrate and foster entrepreneurial innovation. Instead, a substantial part of what is fueling this trend may actually be killer acquisitions that potentially harm innovation and competition. In particular, the large number of acquisitions of small entrepreneurial start-ups by large incumbents in the tech sector would suggest a fruitful opportunity for investigating whether killer acquisitions extend beyond the pharmaceutical industry.
Second, the paper documents the importance of killer acquisitions as an obstacle to innovation in the pharmaceutical industry, an innovation-focused industry crucial to consumer and social welfare. Third, the paper provides new evidence relating to trends and consequences of increasing market concentration. Incumbents in already-concentrated markets further reduce competition by acquiring future product market competitors. Such acquisitions often avoid antitrust scrutiny and may therefore pose concerns for consumer welfare.
This is a long paper – but it rewards reading, and is bound to be a reference in discussions on merger control in coming years. Its modelling and discussion of how incumbents may seek to acquire (future) competitors in order to protect their incumbent position, and in particular their evidence that incumbents do so while strategically avoiding merger control, provides an important contribution to the debate and adds to the growing concern about the impact of mergers on innovation – and to proposals to reform merger notification thresholds.
From a practical standpoint, however, I have some concerns in line with those I already expressed in my reviews of other papers. First, the literature seems to be quite clear that the detrimental effects to innovation occur mainly – or even solely – when the merger concerns overlapping products. It is unclear to me how easy it is to prove this in practice, particularly absent internal documentation to that effect. This links to my concern with this literature’s focus on the pharmaceutical industry. Even in the pharma sector, only a small number of mergers can be identified as killer acquisitions – and this is an industry with a very well structured innovation process and where it is reasonably clear whether a product will be a competitor of an incumbent. How to determine whether a merger is horizontal beyond this sector – or, more to the point, how to extend insights from this sector to others – is a truly challenging proposition. This is particularly clear, in my view, from the debate on acquisitions by technological incumbents, which I will review in coming weeks.