Policymakers and antitrust authorities have long been concerned that corporate consolidations increase market power, leading to higher prices, fewer choices, and reduced quality for consumers (e.g., Bonaime and Wang (2024); Cunningham, Ederer, and Ma (2021); Cooper et al. (2019); Eliason et al. (2020); Fathollahi, Harford, and Klasa (2022); Garmaise and Moskowitz (2006); Kamepalli, Rajan, and Zingales (2022); Sapienza (2002)). Influenced by this evidence and the increasing body of research indicating a rise in corporate market power across the economy (e.g., De Loecker, Eeckhout, and Unger (2020); Autor et al. (2020)), a wave of policy actions has intensified scrutiny of mergers and acquisitions (M&As) that consolidate markets (e.g., Biden (2021)). However, for antitrust authorities to enforce these policies, they must first be aware of anticompetitive M&A deals. We show that thousands of large deals effectively escape ex ante merger review solely because the statutory reporting criteria that determine antitrust review ignores an increasingly important asset in the economy—intangible capital.
Intangible capital is systematically overlooked because, for M&A transactions within a specified deal-size range (e.g., $90 million to $359.9 million in 2023), antitrust authorities such as the Federal Trade Commission (FTC) and the Department of Justice (DOJ) apply a statutory reporting threshold based on the target firm’s assets to determine which deals are subject to review (i.e., the size-of-person [SoP] test). Originally set at $10 million in 2001, this threshold has since been adjusted annually for Gross National Product. However, this asset size threshold only considers the value of assets as reported under U.S. Generally Accepted Accounting Principles (GAAP), which exclude nearly all self-generated intangible assets. This exclusion means that the FTC and DOJ are not accounting for this increasingly important class of assets in the economy (e.g., Crouzet et al. (2022)). Indeed, acquired intangibles now represent eight times the amount of acquired tangible assets (see Figure 1). In line with the rise of intangible assets, the FTC and DOJ have recently focused explicitly on enforcing competition—including implementing appropriate screening mechanisms—in markets where firms’ intangible capital plays a central role, such as in the pharmaceutical and technology sectors (FTC (2022)). However, little is known about the extent to which accounting rules regarding intangible capital impact M&A enforcement and how such an effect impacts product market competition in turn.
Ratio of acquired intangible assets to tangible assets. This figure displays the ratio of acquired intangible assets to tangible assets from 2002 through 2019. The red line depicts the ratio of identifiable intangible assets plus goodwill, all scaled by tangible assets. The dashed lines depict goodwill scaled by tangible assets (dark gray) and identifiable intangible assets scaled by tangible assets (light gray). We use a sample of 11,436 unique observations that comprise M&As conducted by U.S. publicly traded acquirers and represent $8.8 trillion in total acquired assets. For the purpose of our study, we narrow our focus to 1,918 deals that are subject to the SoP test, as depicted in Figure 3. Our measure of intangible assets is identifiable intangible assets plus goodwill, and our measure of tangible assets is the sum of all tangible assets. We obtain values for each type of asset from the purchase price allocation (PPA) disclosed in acquirers’ 10-K SEC filings, found on EDGAR at www.SEC.gov. An example of the PPA disclosure is found in Internet Appendix Section IV.
We collect novel data on both intangible assets—for example, customer lists, patents, brands, and in-process R&D—and tangible assets of target firms from post-merger purchase price allocations (PPA) in acquirers’ financial statements. Consistent with reporting hinging on the target’s book assets, we find that having assets just over the threshold for premerger review strongly predicts reporting to the FTC and DOJ (see Figure 2). Since these targets’ book assets largely exclude intangibles, this suggests that many acquisitions effectively escape ex ante merger review solely because the intangible capital of target firms is ignored. Indeed, we find that if legislator were to require that firms add intangible capital to the targets’ assets, the number of deals reported to the FTC and DOJ would increase by approximately 263 per year, more than half of which represent horizontal consolidations among competitors. These nonreportable deals are very similar in size to reportable deals but comprise 50% more intangible assets. Thus, nonreportable deals, despite being exempt from premerger notification, are sizable enough to warrant antitrust review. If they were reported, we estimate that total Second Requests—the most stringent form of antitrust enforcement by the FTC and DOJ short of litigation—would increase by approximately 10% per year.1
Percent of transactions that were notified. This figure displays the percent of transactions that triggered premerger notifications under the size-of-person test, covering filing years 2001 through 2019. Transactions are grouped by the target firm’s tangible asset value (in millions of 2005 dollars). The vertical dashed line at $10 million marks the asset threshold (for the filing year ending on 2005) below which transactions are not required to file a premerger notification. Data on premerger notifications are manually collected from public corporate filings available on the U.S. SEC’s EDGAR database (www.sec.gov) and supplemented with internet searches. Internet Appendix Section III provides our method for identifying whether the FTC and DOJ were notified.
We conduct four sets of tests to examine whether deals that escape ex ante merger review create benefits for acquirers’ shareholders and impose costs on other stakeholders such as consumers, through increased acquirer market power. First, we compare the deal premia in nonreportable M&As with those in reportable M&As. If nonreportable deals reduce competition, we expect acquirers in those acquisitions to pay more than those in reportable deals. Consistent with this prediction, we find that deal premia are roughly 10 percentage points (or 20%) higher for nonreportable deals. We also find that deals that consolidate product markets drive our results.
Second, we compare changes in acquirers’ equity values around the announcement of nonreportable and reportable deals. If nonreportable deals provide anticompetitive benefits, then the equity values of acquirers should impound this information (e.g., Kepler, Naiker, and Stewart (2023); Fathollahi, Harford, and Klasa (2022)). We find that nonreportable mergers lead to increased equity values (3.4 percentage points higher) for acquirers around their announcement date. We further expect that the acquirer’s industry peers should also benefit from these consolidations, since they too will benefit from lower competition. Consistent with this view, we find that the equity values of industry rivals also rise (0.8 percentage points higher) following the announcement of a nonreportable deal by a competitor. Also consistent with greater market power following nonreportable deals, we find that these equity value responses are driven largely by deals that consolidate product markets.
Third, we examine markups following nonreportable acquisitions. The intuition behind these tests is that a firm’s ability to charge prices above marginal costs demonstrates market power (De Loecker, Eeckhout, and Unger (2020)). We find that the markups of acquirers whose deals escape ex ante merger review increase by 26 percentage points (or 11.3%) following these acquisitions, with the increase beginning in the subsequent year and persisting for at least three years. These results concentrate among deals that consolidate product markets, and they are driven by acquisitions of intangibles related to developed technologies and brands, consistent with the consolidation of intangibles for these markets having an immediate price impact for consumers (e.g., consolidating competing brands can immediately influence an acquirer’s ability to increase prices).
Finally, we examine the implications of accounting rules for intangible capital in facilitating the avoidance of premerger review in undeveloped product markets. To do so, we focus on post-acquisition development of overlapping pharmaceutical projects. We focus on this sector on account of its reliance on intangible capital (e.g., in-process R&D). Moreover, extant research suggests that acquirers have incentives to preempt competition by consolidating undeveloped pharmaceutical projects (e.g., Cunningham, Ederer, and Ma (2021))—a practice recognized by accounting standard setters as “defensive in-process R&D.”2 We find that nonreportable deals are about three times more likely to involve overlapping projects between potentially competing drugs, and acquirers in nonreportable deals are roughly 40% more likely to discontinue acquired projects than acquirers in reportable deals, despite observing no differences in the ability to develop projects. Moreover, we find that this increased likelihood to discontinue projects in nonreportable deals is most pronounced for deals that are economically important to the acquirers, in particular drug projects in the final phase of clinical trials and drugs for highly concentrated markets. These results continue to hold when we compare acquired projects in nonreportable deals to all of an acquirer’s internally developed projects, consistent with acquirers of overlapping projects in nonreportable deals striving to reduce product market competition.
Our findings have implications for public and private enforcement. We find that reportable deals similar in size to nonreportable deals receive 25% of all Second Requests. The deals are rarely blocked, but deals of this size are clearly of interest to antitrust authorities, although they tend to be overlooked simply because of the accounting treatment of intangible assets. We do find evidence, however, that private enforcement substitutes for a lack of public oversight, albeit imperfectly, given the frictions associated with private litigation. Given that the United States relies on both public and private enforcement (e.g., Baer (2014)), the presence of these frictions suggests that many anticompetitive deals are happening.
We compute a back-of-the-envelope effect of an alternative SoP threshold that also includes the value of acquired intangible assets. Such a policy would be consistent with accounting rules that recognize the value of intangibles after validation in an M&A deal. Our estimates suggest that the Premerger Notification Office (PNO) would review an additional 90 deals involving horizontal rivals each year at an expected cost of 2.6% to 3.5% of the annual antitrust enforcement budget. Factoring in the effect on the reluctance of managers to initiate deals they believe would not pass antitrust review, we estimate that this policy change would deter 23 of these 90 newly reportable deals. However, given the capacity constraints of antitrust authorities, they have limited ability to allocate enforcement effort to challenge large M&A deals that meet statutory reporting thresholds. Thus, absent a proportional increase in resources, the increase in Second Requests resulting from an increase in the reportability of deals after including intangibles is difficult to determine precisely.
Such a policy shift would also likely affect managers’ incentives to manipulate deals to avoid premerger review. Consistent with this view, we find a 50% increase in the proportion of nonreportable deals shortly after the announcement of an accounting standard that moved leases onto firms’ balance sheets but before the accounting treatment was adopted, which implies that relevant deals that would have otherwise occurred after the policy shift increased the size of firms’ balance sheets, making their deals reportable to the FTC and DOJ. These findings suggest that firms do indeed exercise discretion regarding the reportability of deals, and thus our back-of-the-envelope estimates on the enforcement implications of intangible capital likely represent a lower bound.
Our paper relates to several literatures. First, our paper adds to the growing literature on the effects of corporate consolidations on product markets. These studies highlight the potential harm of consolidations for consumers in important sectors of the economy—namely, healthcare (e.g., Wollmann (2026); Eliason et al. (2020); Cooper et al. (2019)), pharmaceuticals (e.g., Cunningham, Ederer, and Ma (2021); Bonaime and Wang (2024)), and technology (e.g., Kamepalli, Rajan, and Zingales (2022)). Building on this evidence, recent studies examine how these deals are allowed to occur and find that many have deal values that are too small to trigger antitrust review (e.g., Wollmann (2019, 2026); Cunningham, Ederer, and Ma (2021); Asil et al. (2024)) or are structured to evade review (Kepler, Naiker, and Stewart (2023)). Our study demonstrates that even larger deals well above statutory reporting thresholds can escape ex ante merger review, and thus firms can consolidate markets and reduce competition simply because antitrust authorities rely on accounting data that largely exclude the value of self-developed intangible assets. Second, our paper contributes to the literature on the enforcement of product market competition (e.g., Asker and Nocke (2021); Nocke and Whinston (2010, 2022)) and to the growing debate on shareholder- versus stakeholder-based corporate governance (e.g., Edmans (2021)). Our evidence suggests that recent policies to intensify merger scrutiny in industries such as healthcare, pharmaceuticals, and technology are likely ineffective, as accounting rules help hide these deals from the view of antitrust authorities. In this regard, our study shows that “lax screening” driven by an overreliance on arbitrary thresholds found in other market contexts (e.g., Keys et al. (2010)) also extends to the enforcement of antitrust laws.
Third, our findings contribute to research on the sources of anticompetitive harm. Recent corporate finance studies highlight the importance of ownership structure (Gupta et al. (2024); Eaton, Howell, and Yannelis (2020); Azar, Schmalz, and Tecu (2018); Aghamolla, Jain, and Thakor (2023); Asil et al. (2024)), product similarity (Fathollahi, Harford, and Klasa (2022)), managerial incentives (Antón et al. (2023)), and political connections (Mehta, Srinivasan, and Zhao (2020)) for anticompetitive practices. Our study shows that the interaction between accounting rules for intangibles and antitrust enforcement can also contribute to anticompetitive behavior in product markets.
Fourth, our paper extends the literature on the link between accounting and regulatory enforcement. Historically, enforcement agencies have used accounting information for monitoring (e.g., Covaleski, Dirsmith, and Samuel (1995); Holthausen and Leftwich (1983); Solomon (1970); Taggart (1981)). Considerable research explores how firms adjust financial reporting and investments in response to regulatory actions (see Leuz and Wysocki (2016) for a review). We add to this work by showing that financial reporting standards influence product market structure when antitrust authorities rely on accounting data to select mergers for review.
Finally, our paper adds to the literature on intangible assets (e.g., Crouzet et al. (2022); Lev (2019)). A growing body of research documents the rising importance of intangibles as the economy shifts from relying on physical assets to services and technology as key production inputs (e.g., Haskel and Westlake (2018); Peters and Taylor (2017); He, Mostrom, and Sufi (2024)). Because the value of most intangibles is difficult to measure (Glaeser and Lang (2024)), this literature focuses on distortions unique to intangible assets, such as the difficulty of contracting on intangibles and the potential for pricing inefficiency (e.g., Eisfeldt and Papanikolaou (2014); Rampini and Viswanathan (2010); Giglio and Severo (2012)). We add to this literature by identifying another potential inefficiency, namely, enforcement agencies’ reliance on asset values that exclude most intangibles, which allows anticompetitive mergers to escape review.
The remainder of this paper proceeds as follows. Section I discusses our setting. Section II describes our data. Section III presents results on the role of intangibles in nonreportable M&As. Section IV separately analyzes developed and undeveloped product markets, and Section V discusses the implications of our results and additional analyses. Section VI concludes.
I. Institutional Setting
A. Antitrust Screening of Proposed Mergers
To promote competition in the United States, the antitrust divisions of the FTC and DOJ rely on the HSR Antitrust Improvements Act of 1976 to review proposed M&As. The Act requires parties in deals above a specific size to file a premerger notification, which allows the FTC and DOJ to review whether the merger might decrease competition. After review, the FTC and DOJ can allow the merger to proceed or can issue a Second Request, seeking additional details before issuing a decision on the transaction. Approximately 3% (6%) of all notified (horizontal) deals receive a Second Request (Billman and Salop (2023)).
For most deals, the FTC and DOJ do not require a premerger notification because the size of the deal or transacting parties falls below certain size thresholds (see Figure 3). Deals with transaction prices below the lower size-of-transaction threshold are exempt from premerger notification and thus effectively escape ex ante merger review, while deals with transaction prices above the upper size-of-transaction threshold must submit a premerger filing and be reviewed by the agencies. In 2001, the lower and upper size-of-transaction thresholds were $50 million and $200 million, respectively. These thresholds have been adjusted since 2004 to track U.S. gross national income. Consequently, as of 2019, the lower threshold was $90 million and the upper one was $359.9 million. Figure 3 displays the annual threshold amounts.
Notification thresholds. The Hart-Scott-Rodino Act established the federal premerger notification program, which provides the FTC and the DOJ with information about M&A deals before they become effective. This figure depicts the premerger notification rules and threshold values used to determine whether a merger is subject to or exempt from notification. Panel A depicts the premerger notification rules, as they relate to the size-of-transaction and size-of-person (SoP) tests. Panel B depicts the values (in $ millions) by year for the lower and upper size-of-transaction thresholds and the SoP threshold.
For deals that fall between the thresholds, the SoP test applies. These transactions require review only if (i) the target has total assets or net sales at or above a specified level (e.g., $18 million in 2019) and (ii) the acquirer has total assets or net sales at or above a specified level (e.g., $180 million in 2019).3 If either the target or acquirer does not meet these SoP conditions, a premerger filing is not required.4 Nearly 50% of all deals reviewed by the FTC and DOJ from 2001 through 2019 were between the lower and upper size-of-transaction thresholds where the SoP test applied.5
When determining the target’s and acquirer’s assets and net sales for the SoP test, firms must use financial information from their “last regularly prepared balance sheet” and “annual statement of income” (HSR Act Rules §801.11(c)(1) and (2)). These values are based on GAAP book assets, which exclude the vast majority of intangible assets. Consistent with these rules, Section I of the Internet Appendix provides an example of an FTC representative agreeing in 2007 that, in accordance with US GAAP, intangibles should be excluded from a target’s total assets for premerger notification purposes, even though including intangibles would trigger antitrust review under the SoP test.6 Cases like this are likely common, as a growing range of firms—for example, innovation-intensive companies—often have few tangible assets, and most of their intangibles are not recognized on their balance sheets. Consequently, these targets can fall below the assets and sales thresholds, allowing them to escape ex ante merger review simply because of accounting standards.
Avoiding premerger review benefits the merging parties. In addition to avoiding filing fees (which range from $45,000 to $125,000), firms avoid the possibility of a substantially more costly Second Request. Second Requests typically last six months and cost $2 million to $9 million, while consuming 1,000 hours of internal management and legal time.78 Perhaps more costly to the merging firms, however, is the fact that the vast majority (roughly 75%) of Second Requests result in orders to terminate the transaction or to divest key assets to mitigate anticompetitive effects. Aware of these risks, firms can take real actions to reduce the value of the target’s assets, in an effort to avoid premerger review. Consistent with such review avoidance, Section I of the Internet Appendix provides an example of correspondence with the FTC in 2004 regarding a target’s plans to pay a special dividend, reducing the size of its tangible assets enough to avoid premerger review.
B. Accounting for Intangible Capital
Measuring total assets according to US GAAP for the SoP test immediately expenses internally generated intangibles rather than recording them as assets on the balance sheet.9 A consequence is that book assets primarily comprise physical assets, leading to underreporting of the true value of the firm’s economic assets. Only after a firm acquires another is the target’s internally generated intangible assets recognized at their fair value (ASC 805-20-30). These intangible assets include those that can be separately identified, including customer relationships, in-process research and development (R&D), trade names, and patents. After determining the fair value of the target’s assets and liabilities, the purchase price is allocated to the identifiable assets less liabilities (collectively called “net assets”). Any remaining amount of the purchase price is then recorded as goodwill on the acquirer’s balance sheet.
These accounting rules lead to considerable differences in a target’s book assets before versus after a merger. This difference has grown over time as intangible capital has become a prominent input in firms’ production. Indeed, Figure 1 shows that the ratio of acquired intangible-to-tangible assets has doubled over the past two decades.10 Acquired intangibles now represent eight times the amount of acquired tangible assets, with this increase driven to equal extents by identifiable intangible assets and goodwill.11
An important result of overlooking the consolidation of intangible capital is that these accounting rules might enable intangible-intensive sectors to escape premerger review. While a vast accounting literature examines how financial accounting standards can shape economic activity (e.g., Kanodia and Sapra (2016); Graham, Hanlon, and Shevlin (2011); Bens and Monahan (2008); Dou, Ryan, and Xie (2018)), little is known about whether and how the use of GAAP rules by noncapital-market enforcement agencies shapes market structure and competition.12
II. Data and Descriptive Statistics
A. Primary Data Sources
Our data come from multiple sources. Data on all completed U.S. M&A transactions announced by publicly traded firms between February 2001 and February 2020 come from the Refinitiv Mergers and Acquisitions database. This data set includes the announcement and completion dates, deal value, names, and industries of the target and acquirer, the acquirer’s pre-deal ownership stake in the target, and the form of payment. To be included in our sample, deal values must fall within the annual HSR premerger notification thresholds for transaction size (see Figure 3).13
The FTC’s Bureau of Competition and the DOJ’s Antitrust Division copublish the HSR Annual Report. We obtain reports for fiscal years 2001 through 2019 from the FTC.gov website. These reports provide data on the number of reported HSR mergers, categorized by transaction size and by three-digit North American Industry Classification System (NAICS) industry. They also include information on the number of Second Requests issued, again broken down by transaction size and industry. We use the frequency of Second Requests to rank industries, focusing on those that averaged at least one Second Request per year. The final list of all industries that meet this requirement appears in Internet Appendix Table IA.II.14
We collect data on acquired assets from PPA disclosed in acquirers’ 10-K and 10-Q filings, obtained from the U.S. Securities and Exchange Commission’s (SEC’s) website (SEC.gov), as well as from annual reports downloaded from acquirers’ corporate websites (see Section IV of the Internet Appendix for additional details). The PPA separately discloses the values of acquired tangible assets, intangible assets, and goodwill. We use the reported tangible asset values as our measure of the target’s size to determine whether a deal is reportable under the SoP test. To be included in our main sample, acquired assets must be reported in gross terms—the relevant basis for the SoP test—rather than net values.15 We exclude deals in which the acquirer consolidates multiple acquisitions into a single PPA disclosure, which occurs in approximately 5% of cases.
Our final sample consists of 1,918 unique M&A transactions, after excluding 5,483 deals due to incomplete acquisition data or because they fall outside our focal industries, as well as 1,608 additional deals that either do not report gross asset values or aggregate multiple acquisitions within a single PPA disclosure (see Internet Appendix Table IA.III).
B. Other Data Sources
We supplement our PPA data with information that we collect from corporate disclosures—such as merger agreements, proxy statements, 8-Ks, 10-Qs, and 10-Ks—as well as news reports and law firm websites, to determine whether the FTC and DOJ were notified of a deal. FTC-issued notices of “early terminations,” which we obtain from FTC.gov, are also used to identify notified transactions.16 Finally, following the approach of Asil, Barrios, and Wollmann (2023), we also use M&A announcement and completion dates from Refinitiv, to assess whether a deal was likely notified based on the number of days between these dates. The full methodology for identifying premerger notifications is described in Section III of the Internet Appendix.
Data used to analyze deal premia, announcement-date returns, and markups come from three main sources: Compustat, the Center for Research in Security Prices (CRSP), and, to support our markup calculations, the publicly available industry-level elasticity data sets of De Loecker, Eeckhout, and Unger (2020). Following the methodology of De Loecker, Eeckhout, and Unger (2020), we calculate firm-year markups by dividing net sales by the cost of goods sold—both from Compustat—multiplied by the firm’s corresponding industry-level elasticity. Descriptions of all outcome and control variables are available in Section VI of the Internet Appendix.
We classify industries using three-digit NAICS codes as the FTC and DOJ adopted NAICS for antitrust analysis starting in 2001—the same year our sample period begins. Because Refinitiv reports Standard Industrial Classification (SIC) codes rather than NAICS, we convert SIC to NAICS using the crosswalk process shown in Internet Appendix Table IA.II.
Finally, we obtain data on pharmaceutical drug projects from Cortellis Competitive Intelligence, which provides start and end dates for all development phases of each project. For every drug, the data set includes information on its intended therapeutic market (e.g., “Parkinson’s disease”), mechanism of action (MOA; e.g., “Growth Hormone Receptor”), and development phase. Following Cunningham, Ederer, and Ma (2021), we use this information to identify overlapping projects. Our sample includes drug projects that began in January 2000 and extends through the end of our study period. We link these projects to acquirers and targets in our main data set using fuzzy matching techniques.
C. Descriptive Summary
Of the 1,918 deals used in our analysis, 1,682 (88%) involve private targets. Horizontal deals, that is those for which the target and acquirer share the same three-digit NAICS, constitute roughly 50% of the sample in terms of both number of deals and transaction value (see Panel C of Internet Appendix Table IA.III). Table I, Panel A, presents the distribution of deals by whether the deal was reportable to the FTC and DOJ. We classify a deal as reportable (nonreportable) if the total assets for the target exceed (fall below) the SoP asset threshold in the given reporting year. To validate our classification method, we collect information—as described in Section III of the Internet Appendix—to assess whether the size of the target’s tangible assets predicts whether a transaction is reportable. The results, shown in Figure 2, indicate that the share of transactions that are notified is near zero just below the asset size threshold. However, this share jumps sharply to around 70% immediately above the threshold and continues to increase beyond that point.17 Table I, Panel A, shows that these nonreportable horizontal deals represent roughly the same percentage as reportable horizontal deals (both about 55%), but are on average smaller (i.e., $121.3 million versus $143.5 million).18
Descriptive Statistics
| This table presents descriptive statistics for our sample of reportable and nonreportable deals from February 2001 through February 2020. A deal is classified as reportable if its total assets are greater than the SoP threshold in that reporting year. A deal is classified as nonreportable if its total assets are less than or equal to the SoP asset threshold in that reporting year but has not been reviewed by the FTC or DOJ. In Panel A, we present descriptive statistics separately for reportable and nonreportable deals. In Panel B, we present descriptive statistics by industry (sorted by total deal value) for only nonreportable horizontal deals. In Panel C, we present the mean percent of tangible assets, intangible assets, and goodwill for reportable and nonreportable horizontal deals. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. | |||
|---|---|---|---|
| Panel A. Reportable versus Nonreportable M&As | |||
| Reportable | Nonreportable | Difference | |
| Type of M&A | |||
| Horizontal (3-digit NAICS) | 766 (55.2%) | 219 (56.6%) | −1.0% |
| Nonhorizontal | 621 (44.8%) | 168 (43.4%) | 1.0% |
| Average deal value (in $ millions) | |||
| Horizontal (3-digit NAICS) | $143.5 | $121.3 | $22.2*** |
| Nonhorizontal | $148.1 | $122.1 | $26.0*** |
| Panel B. Nonreportable Horizontal M&As (by total deal value) | ||
|---|---|---|
| Industry | Horizontal M&As (Nonreportable) | Value (in $ billions) |
| Computer and Electronic Product Manufacturing | 107 (48.8%) | $11.83 |
| Chemical Manufacturing | 62 (28.3%) | $8.72 |
| Professional, Scientific, and Technical Services | 17 (7.80%) | $2.11 |
| Machinery Manufacturing | 10 (4.60%) | $1.62 |
| Telecommunications | 8 (3.70%) | $0.71 |
| Food and Kindred Products | 5 (2.30%) | $0.57 |
| Publishing Industries (except Internet) | 4 (1.80%) | $0.45 |
| Communications | 5 (2.30%) | $0.44 |
| Hospitals | 1 (0.50%) | $0.12 |
| Utilities | 0 (0.00%) | $0.00 |
| Transportation Equipment | 0 (0.00%) | $0.00 |
| Health Services | 0 (0.00%) | $0.00 |
| Merchant Wholesales, Nondurable Goods | 0 (0.00%) | $0.00 |
| Total | 219 (100%) | $26.56 |
| Panel C: Tangible Assets, Intangibles, and Goodwill of Horizontal M&As | |||
|---|---|---|---|
| Reportable | Nonreportable | Difference | |
| Horizontal M&As | |||
| Tangible assets | 35.5% | 6.7% | 28.8%*** |
| Intangibles | 27.7% | 46.8% | −19.1%*** |
| Goodwill | 36.8% | 46.4% | −9.6%*** |
| Total | 100% | 100% | |
Panel B of Table I presents the distribution, by value, of nonreportable horizontal deals. Of the 219 nonreportable horizontal deals, 169 (77%, with total deal value of $20 billion) are in the computer and electronic product manufacturing and chemical manufacturing industries, which several of our subsequent analyses focus on given the prominence of consolidation in these product markets. In aggregate, over $26.5 billion in horizontal deals were nonreportable. The total value of the 1,918 deals in our full sample is $268 billion, and thus 10% of all market consolidation involved nonreportable horizontal mergers.
In terms of the composition of assets in these deals, Panel C of Table I presents PPA for reportable and nonreportable horizontal M&As. Reportable deals comprise roughly similar degrees of tangible and intangible assets (35.5% versus 27.7%, respectively). However, we find that identifiable intangible assets represent 46.8% of nonreportable deal values, roughly seven times the value of tangible assets in nonreportable deals (6.7%).
To illustrate how this omission of intangible capital impacts antitrust enforcement over time, Panel A of Figure 4 plots the number of deals in our sample that were subject to review (blue line) and the number that would have been subject to review if intangible assets were included in the SoP test (red line), where we determine the hypothetical number by adding in the fair value of intangibles from the acquirer’s PPA. Doing so increases the number of reportable deals by 25% to 60% each year. Panel B applies these increases to all reportable deals reviewed by the FTC and DOJ annually. The panel shows that another 5,003 deals ($630 billion of total deal value) would be reportable if intangibles were included in the SoP test (or 263 deals at $33 billion annually).19 This translates to an additional 466 nonreportable horizontal deals (about 25 deals per year) in the United States over our sample period and in turn to approximately $60 billion in deal value, or 10% of all consolidation in the economy over this period.20,21
Trends in reported deals. This figure displays the number of deals reported to antitrust authorities when only tangible assets are included in the size-of-person test (in blue) and the number of deals that would be reportable if both tangible and identifiable intangible assets were included in the SoP test (in red). In Panel A, we present the current HSR regime (blue) and the counterfactual regime (red) for only our sample of deals. In Panel B, we present the current HSR regime (blue) using data from HSR annual reports and then estimate the counterfactual HSR regime (red) using red-to-blue proportions obtained from Panel A.
D. Defining Product Markets
Identifying mergers with the potential to cause anticompetitive harm is crucial to our study. Doing so requires a measure of overlapping product markets, which is challenging to obtain since nearly 90% of our targets are private firms. The FTC and DOJ analyze proprietary, product-level data at the six-digit NAICS level submitted by both the target and the acquirer in premerger review filings to identify and evaluate the potential anticompetitive effects resulting from the consolidation product markets. Absent detailed data, we employ two approaches to define overlapping markets. The first begins by classifying a merger as horizontal if the target and acquirer share the same three-digit NAICS code. To improve the accuracy of our classification, we then review corporate filings and press releases, industry reports, and news articles to determine whether the acquirer and target actually operate in at least one overlapping product market, and we adjust our classification if they do not. The second approach, applied specifically to pharmaceutical mergers, follows prior studies by constructing a measure of scientific overlap in drug development projects (e.g., Cunningham, Ederer, and Ma (2021)).
III. Nonreportable M&A and Intangible Capital
We next study the types of intangible assets in nonreportable deals and examine how deal characteristics differ across nonreportable versus reportable transactions. Our empirical strategy compares deals that undergo the SoP test and are nonreportable to antitrust authorities with deals that undergo the test and are reportable. However, whether a nonreportable or reportable M&A transaction occurs may vary with factors such as industry norms and deal timing. For instance, firms in industries more subject to antitrust enforcement might be more likely to consolidate markets when they are nonreportable. Furthermore, firms might strategically time deals to occur during years of heightened M&A activity when the PNO is unusually busy. We employ industry and year fixed effects to address the first and second concerns, respectively. We also leverage the fact that many acquirers in our sample have both nonreportable and reportable deals, which allows us to conduct within-acquirer tests to help rule out differences in acquiring-firm preferences for deals of a certain kind to go unreported. Furthermore, the sharp discontinuity that we find in reporting likelihood precisely at the SoP asset threshold (as shown in Figure 2), which is difficult to attribute to other deal characteristics, provides further validation for our research design’s treatment of deal nonreportability as exogenous.22
A. Levels and Types of Intangibles in Nonreportable Deals
We first characterize how reportable and nonreportable deals differ in their intensity of intangible capital. Figure 5 plots deal-size density for each type of deal. Panels A and B show that the total deal value and intangible asset value distributions for reportable and nonreportable deals are remarkably similar. Nonreportable deals have slightly more intangible assets than reportable deals, consistent with our earlier finding that intangibles represent a higher proportion of deal value in nonreportable deals. In Internet Appendix Table IA.V, we find no statistically significant difference in the level of intangibles in nonreportable deals relative to reportable deals. However, the proportion of the deal related to intangibles is more than 50% greater for nonreportable deals. Thus, despite the FTC and DOJ perceiving reportable deals as larger, these results suggest that nonreportable deals are quite similar in size but differ in the accounting treatment of intangible assets when determining whether they avoid ex ante merger review.
Distribution of deals: Nonreportable versus reportable. This figure graphically displays the distribution of nonreportable versus reportable deals. In Panel A, we present the distribution of deal values for nonreportable and reportable M&As. Deal values presented exclude the value of equity in the target held by the acquirer before the deal is announced (i.e., the “toehold”). We explain this calculation in Internet Appendix Section II. This adjustment shifts 22 deals at the lower end of the deal-value distribution from below to above the lower size-of-transaction threshold. In Panel B, we present the distribution of identifiable intangible asset values for nonreportable and reportable M&As.
We next examine the types of intangibles acquired in these deals by collecting data on the categories of intangibles disclosed in the PPA of acquirers’ 10-Ks for 1,810 of 1,918 deals with identifiable intangible assets, 75% of which allocate the purchase price into separate intangible categories (e.g., customer relationships, patents, in-process R&D). See Sections IV and V of the Internet Appendix for details of this collection process. Panel B of Table II shows that identifiable intangibles total nearly $79 billion across 22 categories. In Panel C, we find that nonreportable deals have on average approximately four times the level of in-process R&D relative to reportable deals, while reportable deals have twice the level of customer-related intangibles relative to nonreportable deals. Figure 6 shows these patterns visually. One reason for these findings is that customer relationships develop gradually and are more likely to be associated with mature firms with more tangible assets whose deals will therefore be reportable (e.g., Foster, Haltiwanger, and Syverson (2016)). By contrast, early-stage, innovative firms—which have few tangible assets—are more likely to rely on intangible capital.
Categories of Intangibles
| This table presents results of the analysis of categories of intangibles. In Panel A, we present the frequency of intangibles in our sample. In Panel B, we present the amounts (in $ millions) and percents for all categories of identifiable intangible assets in our sample. In Panel C, we present results for difference-in-means tests by category for reportable and nonreportable deals. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. | |
|---|---|
| Panel A. Frequency of Intangibles in M&A | |
| Description | Observations |
| No intangibles | 108 |
| Intangibles (not disaggregated) | 410 |
| Intangibles (disaggregated by category) | 1,400 |
| Total | 1,918 |
| Panel B. Economic Importance by Category of Intangible | ||
|---|---|---|
| Category | Amount($ millions) | Percent |
| Customer Relationships & Lists | $30,491.91 | 38.7% |
| Patents, Technology, & Software | $19,808.12 | 25.1% |
| Trademarks & Brands | $8,906.38 | 11.3% |
| In-Process R&D | $7,663.93 | 9.7% |
| Licenses | $3,212.06 | 4.1% |
| Product Rights | $3,036.69 | 3.9% |
| Distribution Agreements | $1,242.37 | 1.6% |
| Power Purchase Agreements | $628.67 | 0.8% |
| Other Intangibles | $627.16 | 0.8% |
| Non-Compete Agreements | $513.91 | 0.7% |
| Mineral Interests | $475.20 | 0.6% |
| Usage Rights | $391.00 | 0.5% |
| Franchise Rights | $325.60 | 0.4% |
| Databases | $272.60 | 0.3% |
| Lease Intangibles | $247.96 | 0.3% |
| Supplier Agreements | $163.03 | 0.2% |
| Maintenance Contracts | $122.20 | 0.2% |
| Management Agreements | $103.10 | 0.1% |
| Pipeline Capacity Rights | $87.60 | 0.1% |
| Other Contract Rights | $66.90 | 0.1% |
| Assembled Workforce | $50.80 | 0.1% |
| Royalty Agreements | $4.90 | 0.0% |
| Total | $78,760.16 | 100.0% |
| Panel C. Difference-in-Means Tests (by Category) for Reportable versus Nonreportable M&A | |||
|---|---|---|---|
| Category | Mean($ millions) Reportable | Mean($ millions) Nonreportable | Difference |
| Customer Relationships & Lists | $25.19 | $12.04 | $13.15*** |
| Patents, Technology, & Software | $13.78 | $15.05 | -$1.27 |
| Trademarks & Brands | $ 6.54 | $ 4.88 | $1.66* |
| In-Process R&D | $ 2.94 | $12.14 | −$9.20*** |
Categories of intangibles. This figure displays, by reportable versus nonreportable, the percent of total identifiable intangibles that each category represents. We display the top four categories separately and aggregate the remaining 18 categories into “All Others.” See Panel B of Table II for the complete list of categories.
B. Deal Premia for Nonreportable M&A
(1)
where is the proportion of target ’s equity recognized as goodwill in year . We include fixed effects for reporting year () and acquirer industry (). Table III presents the results. Consistent with nonreportable deals providing anticompetitive benefits that acquirers pay more for, column (1) shows that deal premia for nonreportable deals are approximately 10 percentage points (or 20%) higher than those in reportable deals.
Deal Premiums and Nonreportable M&As
| This table presents results from ordinary least squares (OLS) regressions of deal premiums on an indicator for whether the deal was reportable or nonreportable to the antitrust regulators. The main variable of interest in columns (1) and (3), Nonreportable, is an indicator that assumes the value of one if the target firm’s assets are below the size-of-person asset threshold, and zero otherwise. The main variable of interest in columns (2) and (4), Nonreportable ProductMarketOverlap, is an interaction term that assumes the value of one when the acquirer and the target firm share overlapping product markets in a nonreportable deal, and zero otherwise. Across all columns, the dependent variable, DealPremium, is a continuous variable that captures the proportion of the acquired equity that is allocated to goodwill. All variables are described in Internet Appendix Section VI. We vary the inclusion of fixed effects as follows. In columns (1) and (2), we include filing-year and acquirer-industry fixed effects, respectively. In columns (3) and (4), we include filing-year and firm (i.e., acquirer) fixed effects. DealPremium is winsorized at the 1% and 99% levels. Per prior research (e.g., Dong et al., 2006), we exclude deals when the deal premium is larger than 200%. Robust t-statistics are reported in parentheses and calculated using standard errors clustered at the filing-year and the acquirer’s industry level, respectively. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. | ||||
|---|---|---|---|---|
| (1) | (2) | (3) | (4) | |
| Dependent Variable: | DealPremium | DealPremium | DealPremium | DealPremium |
| Nonreportable | 0.099*** | 0.084** | 0.060* | 0.046 |
| (3.16) | (2.45) | (1.86) | (1.34) | |
| ProductMarketOverlap | −0.047* | −0.039 | ||
| (−2.15) | (−1.19) | |||
| Nonreportable ProductMarketOverlap | 0.049* | 0.051*** | ||
| (1.91) | (4.52) | |||
| Observations | 1,663 | 1,663 | 707 | 707 |
| Adjusted | 0.151 | 0.155 | 0.481 | 0.482 |
| Filing-Year F/E | Y | Y | Y | Y |
| Industry F/E | Y | Y | N | N |
| Firm F/E | N | N | Y | Y |
To better attribute the higher deal premia for nonreportable deals to anticompetitive benefits that accrue to acquirers—rather than, say, lower transaction costs associated with deals that avoid ex ante merger review—we consider whether the higher deal premia vary with an indicator for whether the deal consolidated the acquirer’s and target’s product markets (). We interact ProductMarketOverlap with and present the results in column (2). We find that higher deal premia for nonreportable deals are more pronounced in deals that consolidate product markets—acquirers of nonreportable deals involving the consolidation of product markets are willing to pay a 13.3 percentage-point (or 26.6%) higher premium than acquirers of reportable deals. Similar results obtain in columns (3) and (4), which include acquirer (rather than industry) fixed effects and compare deal premia within the same acquirer.23
C. Acquirer Equity Values and Nonreportable M&A
(2)
where is acquirer ’s market-adjusted five-day cumulative abnormal return (centered on the announcement date) and is as previously defined.24 To account for other factors previously shown to be associated with announcement returns in M&A (e.g., Masulis, Wang, and Xie (2007); Bao and Edmans (2011); Officer (2003); Moeller, Schlingemann, and Stulz (2004)), we include the following deal- and acquirer-level characteristics: AllCash, AllStock, DealPremium, DealSize, FreeCashFlow, Leverage, PublicTarget, RelativeSize, Q, and Size. All variables are described in Section VI of the Internet Appendix.
Panel A of Table IV reports the results.25 In column (1), we find no significant difference in announcement returns for nonreportable relative to reportable deals. However, when we interact Nonreportable with ProductMarketOverlap in column (2), we find a 3.4 percentage-point increase in abnormal returns of nonreportable deals that consolidate product markets relative to reportable deals that do the same. This represents a 43% increase over the mean abnormal return for reportable deals that consolidate product markets. We find similar results in columns (3) and (4), which include acquirer (rather than industry) fixed effects and therefore compare announcement returns within the same acquirer.
Announcement Returns and Nonreportable M&As
| This table presents results from OLS regressions of cumulative abnormal returns on an indicator for whether the deal was reportable or nonreportable to the antitrust regulators. In Panel A, the main variable of interest in columns (1) and (3), Nonreportable, is an indicator that assumes the value of one if the target firm’s assets are below the size-of-person (SoP) asset threshold, and zero otherwise. The main variable of interest in columns (2) and (4), Nonreportable ProductMarketOverlap, is an interaction term that assumes the value of one when the acquirer and the target firm share overlapping product markets in a nonreportable deal, and zero otherwise. Across all columns, the dependent variable, AnnReturn, is a continuous variable that captures the five-day market-adjusted cumulative abnormal returns of the acquirer centered on the announcement date. In Panel B, the main variable of interest in columns (1) and (3), Nonreportable, is an indicator that assumes the value of one if the target firm’s assets are below the SoP asset threshold, and zero otherwise. The main variable of interest in columns (2) and (4), Nonreportable ProductMarketOverlap, is an interaction term that assumes the value of one when the acquirer and the target firm share overlapping product markets in a nonreportable deal, and zero otherwise. Across all columns, the dependent variable, RivalReturns, is a continuous variable that captures the five-day market-adjusted cumulative abnormal return, centered on the announcement date, of the industry rivals of the acquirer. Acquirer- and deal-level control variables included, but reported in Internet Appendix Table IA.VI, in the estimations in Panel A are AllCash, AllStock, DealPremium, DealSize, FreeCashFlow, Leverage, PublicTarget, RelativeSize, Q, and Size. In Panel B, we control for DealPremium. All variables are described in Internet Appendix Section VI. In Panels A and B, we vary the inclusion of fixed effects as follows. In columns (1) and (2), we include filing-year and acquirer-industry fixed effects. In columns (3) and (4), we include filing-year and firm (i.e., acquirer) fixed effects. AnnReturn and RivalReturns are winsorized at the 1% and 99% levels. Robust t-statistics are reported in parentheses and calculated using standard errors clustered at the filing-year and acquirer-industry level, respectively. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. | ||||
|---|---|---|---|---|
| Panel A. Acquirer’s Announcement Returns | ||||
| (1) | (2) | (3) | (4) | |
| Dependent Variable: | AnnReturn | AnnReturn | AnnReturn | AnnReturn |
| Nonreportable | −0.002 | −0.009 | 0.023** | 0.008 |
| (−0.31) | (−1.20) | (2.95) | (0.57) | |
| ProductMarketOverlap | 0.011* | −0.015 | ||
| (2.02) | (−0.89) | |||
| Nonreportable ProductMarketOverlap | 0.034* | 0.059** | ||
| (2.00) | (2.32) | |||
| Observations | 1,047 | 1,047 | 502 | 502 |
| Adjusted | 0.037 | 0.046 | 0.186 | 0.199 |
| Controls | Y | Y | Y | Y |
| Filing-Year F/E | Y | Y | Y | Y |
| Industry F/E | Y | Y | N | N |
| Firm F/E | N | N | Y | Y |
| Panel B. Rivals’ Announcement Returns | ||||
|---|---|---|---|---|
| (1) | (2) | (3) | (4) | |
| Dependent Variable: | RivalReturns | RivalReturns | RivalReturns | RivalReturns |
| Nonreportable | 0.005** | 0.003* | −0.001 | −0.002 |
| (2.70) | (2.00) | (−0.15) | (−0.73) | |
| ProductMarketOverlap | 0.001 | −0.003 | ||
| (0.65) | (−0.89) | |||
| Nonreportable ProductMarketOverlap | 0.008* | 0.007* | ||
| (2.13) | (2.16) | |||
| DealPremium | −0.002 | −0.002 | 0.004 | 0.003 |
| (−0.60) | (−0.62) | (0.27) | (0.25) | |
| Observations | 998 | 998 | 458 | 458 |
| Adjusted | 0.010 | 0.010 | 0.032 | 0.026 |
| Filing-Year F/E | Y | Y | Y | Y |
| Industry F/E | Y | Y | N | N |
| Firm F/E | N | N | Y | Y |
Finally, following Eckbo (1983), Chevalier (1995), and Fathollahi, Harford, and Klasa (2022), we examine the abnormal returns of industry rivals around the deal’s announcement.26 The intuition for these tests is that if nonreportable deals soften competition, rents should also accrue to industry rivals, as they can free ride on the benefits. Consistent with this view, in Table IV, Panel B, we find 0.8 percentage points higher abnormal returns, as shown in column (2), for industry rivals following nonreportable deals that consolidate a product market.27 Taken together, the results in Table IV are consistent with equity markets impounding into stock prices the anticompetitive benefits of nonreportable deals, particularly when an acquirer’s market power increases.
IV. Developed and Undeveloped Market Consolidation
We next examine mechanisms, that is, whether acquiring intangible-intensive targets is beneficial simply because it avoids antitrust review, or whether these intangibles are especially advantageous in providing anticompetitive benefits. To do so, we first leverage data on the most prominently consolidated intangibles in developed product markets, namely (i) trademarks and brands and (ii) patents, technology, and software.28 We then examine the role of intangibles in undeveloped product markets (e.g., in-process pharmaceutical drug projects), as a growing literature suggests that the acquisition of undeveloped products can have anticompetitive consequences (e.g., Cunningham, Ederer, and Ma (2021); Kamepalli, Rajan, and Zingales (2022)).
A. Empirical Approach
Our empirical strategy leverages our intangible capital data to compare nonreportable and reportable deals in event-study designs. We first focus on transactions that only involve the acquisition of technology, brands, or in-process R&D. These tests incorporate acquirer-firm fixed effects. This design enables us to assess differences in economic outcomes between nonreportable and reportable deals involving similar intangible assets acquired by the same firm. We next examine pharmaceutical mergers. These tests incorporate a fixed-effect framework that enables comparisons of drug project R&D within therapeutic classes (TCs) and mechanisms of action across nonreportable and reportable deals. They also include a set of variables that allow us to control for other factors that may help explain drug project outcomes. We elaborate on our research designs below.
B. Markups and Developed Product Market Consolidation
(3)
where is acquirer ’s markup in the years through , and is an indicator for whether the markup is in year , , , or .29 All other variables are as previously defined. Table V, Panel A, presents the results.30 Overall, we find that nonreportable deals between firms in overlapping product markets are associated with higher post-acquisition markups. This association begins shortly after the acquisition (see column (2)) and persists for at least three years. The increase in markups continues to hold, as shown in columns (3) and (4), when our control group only includes acquisitions by firms that are not yet-, last-, or never-treated, that is, observations that are “clean” controls, to address problems related to staggered treatment timing and treatment effect heterogeneity (e.g., Baker, Larcker, and Wang (2022); Sun and Abraham (2021); Callaway and Sant’Anna (2021); Cengiz et al. (2019); Borusyak, Jaravel, and Spiess (2024); Athey and Imbens (2022); De Chaisemartin and d’Haultfoeuille (2020); Goodman-Bacon (2021); Kosuke and Song (2021)).31 Panel A of Figure 7 depicts the results from our dynamic specification reported in column (4).32
Markups and Intangible Capital in Nonreportable M&As
| This table presents results from OLS regressions of markups on an triple-interaction term for whether the deal was reportable or nonreportable to the antitrust regulators, whether the acquirer and the target have product markets that overlap, and a time indicator. In Panel A, the main variable of interest in columns (1) and (3), Nonreportable ProductMarketOverlap Post, is an indicator that assumes the value of one if the target firm’s assets are below the size-of-person (SoP) asset threshold, the acquirer and the target have product markets that overlap, and the year the markup is measured is the acquisition year () or after, and zero otherwise. The main variables of interest in columns (2) and (4), Nonreportable ProductMarketOverlap Post (+3), Nonreportable ProductMarketOverlap Post (+2), Nonreportable ProductMarketOverlap Post (+1), Nonreportable ProductMarketOverlap AcqYear, Nonreportable ProductMarketOverlap Before (−2), and Nonreportable ProductMarketOverlap Before (−3), are triple-interaction terms that include a time indicator that takes the value of one if the markup is measured three years after, two years after, the year of, one year after, two years before, or three years before the acquisition, respectively, and zero otherwise. The exclusion year is the year immediately before the acquisition year. In Panel B, the main variable of interest in columns (1), (3), and (5), Nonreportable ProductMarketOverlap Post, is an indicator that assumes the value of one if the target firm’s assets are below the SoP asset threshold, the acquirer and the target have product markets that overlap, and the year the markup is measured is after the acqusiition year, and zero otherwise. The main variables of interest in columns (2), (4), and (6), Nonreportable ProductMarketOverlap Post (+3), Nonreportable ProductMarketOverlap Post (+2), Nonreportable ProductMarketOverlap Post (+1), Nonreportable ProductMarketOverlap AcqYear, Nonreportable ProductMarketOverlap Before (−2), and Nonreportable ProductMarketOverlap Before (−3), are triple-interaction terms that include a time indicator that takes the value of one if the markup is measured three years after, two years after, one year after, the year of, two years before, or three years before the acquisition, respectively, and zero otherwise. Across all columns of Panels A and B, the dependent variable, Markup, is a continuous variable that captures the acquirer’s markup. All variables are described in Internet Appendix Section VI. In Panels A and B, across all columns, we include acquisition-year and firm (i.e., acquirer) fixed effects. Markup is winsorized at the 1% and 99% levels. Robust t-statistics are reported in parentheses and calculated using standard errors clustered at the acquisition-year and the acquirer’s industry level, respectively. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. Coefficients not displayed in columns (2) and (4) of Panel A, and columns (2), (4), and (6) of Panel B, are reported in Internet Appendix Table IA.VIII. | ||||
|---|---|---|---|---|
| Panel A. Markups Following Nonreportable M&As | ||||
| (1) | (2) | (3) | (4) | |
| Dependent Variable: | Markup | Markup | Markup | Markup |
| Sample: | Full | Full | Never- or | Never- or |
| Sample | Sample | Last-Treated | Last-Treated | |
| Nonreportable ProductMarketOverlap Post | 0.244*** | 0.257*** | ||
| (5.48) | (5.59) | |||
| Nonreportable ProductMarketOverlap Post (+3) | 0.352** | 0.371*** | ||
| (2.75) | (2.90) | |||
| Nonreportable ProductMarketOverlap Post (+2) | 0.321*** | 0.331*** | ||
| (3.96) | (3.90) | |||
| Nonreportable ProductMarketOverlap Post (+1) | 0.264*** | 0.269*** | ||
| (3.34) | (3.32) | |||
| Nonreportable ProductMarketOverlap AcqYear | 0.041 | 0.045 | ||
| (0.67) | (0.68) | |||
| Nonreportable ProductMarketOverlap Before (−2) | 0.060 | 0.057 | ||
| (0.64) | (0.61) | |||
| Nonreportable ProductMarketOverlap Before (−3) | −0.058 | −0.066 | ||
| (−0.64) | (−0.71) | |||
| Nonreportable ProductMarketOverlap | −0.162 | −0.163 | −0.038 | −0.035 |
| (−1.24) | (−1.08) | (−0.31) | (−0.26) | |
| Nonreportable Post | 0.025 | 0.021 | ||
| (0.43) | (0.37) | |||
| ProductMarketOverlap Post | −0.033 | −0.046 | ||
| (−0.58) | (−0.82) | |||
| Nonreportable | 0.003 | 0.023 | −0.015 | 0.005 |
| (0.02) | (0.13) | (−0.11) | (0.03) | |
| ProductMarketOverlap | −0.017 | 0.010 | −0.054** | −0.031 |
| (−0.94) | (0.45) | (−2.49) | (−1.36) | |
| Post | −0.012 | −0.008 | ||
| (−0.26) | (−0.17) | |||
| Sample Markup (Mean) | 2.284 | 2.284 | 2.278 | 2.278 |
| Observations | 6,748 | 6,748 | 6,643 | 6,643 |
| Adjusted | 0.886 | 0.886 | 0.885 | 0.884 |
| Acquisition-Year F/E | Y | Y | Y | Y |
| Firm F/E | Y | Y | Y | Y |
| Panel B. Markups and Intangible Capital | ||||||
|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | |
| Dependent Variable: | Markup | Markup | Markup | Markup | Markup | Markup |
| Subsample: | Brand or Tech=1 | Brand or Tech=1 | Brand & Tech=1 | Brand & Tech=1 | Brand & Tech=0 | Brand & Tech=0 |
| Nonreportable ProductMarketOverlap Post | 0.407** | 0.522*** | −0.093 | |||
| (2.75) | (4.19) | (−0.45) | ||||
| Nonreportable ProductMarketOverlap Post (+3) | 0.522* | 0.548** | 0.025 | |||
| (2.03) | (2.35) | (0.14) | ||||
| Nonreportable ProductMarketOverlap Post (+2) | 0.576** | 0.598** | −0.208 | |||
| (2.58) | (2.38) | (−0.96) | ||||
| Nonreportable ProductMarketOverlap Post (+1) | 0.438* | 0.473*** | −0.108 | |||
| (1.88) | (3.37) | (−0.53) | ||||
| Nonreportable ProductMarketOverlap AcqYear | 0.122 | 0.245* | −0.126 | |||
| (1.18) | (1.78) | (−0.70) | ||||
| Nonreportable ProductMarketOverlap Before (-2) | 0.037 | −0.073 | 0.110 | |||
| (0.35) | (−0.49) | (0.80) | ||||
| Nonreportable ProductMarketOverlap Before (−3) | −0.015 | −0.100 | −0.143 | |||
| (−0.09) | (−0.67) | (−0.61) | ||||
| Nonreportable ProductMarketOverlap | −0.164 | −0.171 | 0.172 | 0.228* | 0.488 | 0.499 |
| (−1.20) | (−1.12) | (1.19) | (1.79) | (0.95) | (0.94) | |
| Nonreportable Post | −0.049 | −0.070 | 0.193 | |||
| (−0.64) | (−0.52) | (1.03) | ||||
| ProductMarketOverlap Post | −0.013 | −0.151** | −0.102 | |||
| (−0.20) | (−2.10) | (−1.02) | ||||
| Nonreportable | −0.032 | −0.016 | −0.115 | −0.165* | −0.383 | −0.360 |
| (−0.23) | (−0.08) | (−1.27) | (−1.94) | (−1.08) | (−1.14) | |
| ProductMarketOverlap | 0.020 | 0.060 | 0.049 | 0.068 | −0.023 | −0.024 |
| (0.55) | (1.04) | (0.79) | (1.38) | (−0.22) | (−0.25) | |
| Post | −0.042 | 0.029 | 0.050 | |||
| (−0.80) | (0.35) | (0.73) | ||||
| Observations | 4,207 | 4,207 | 2,100 | 2,100 | 2,436 | 2,436 |
| Adjusted | 0.900 | 0.900 | 0.900 | 0.898 | 0.871 | 0.870 |
| Never- or Last-Treated Sample | Y | Y | Y | Y | Y | Y |
| Acquisition-Year F/E | Y | Y | Y | Y | Y | Y |
| Firm F/E | Y | Y | Y | Y | Y | Y |
Markups following nonreportable deals. This figure graphically displays the evolution of markups before and after acquisitions that consolidate developed product markets. In Panel A, the figure presents coefficients from column (4) of Panel A in Table V. In Panel B, the figure presents coefficients from column (2) of Panel B in Table V. In Panel C, the figure presents coefficients from column (4) of Panel B in Table V. In Panel D, the figure presents coefficients from column (6) of Panel B in Table V. Coefficients correspond to the following interaction terms in the model: Nonreportable ProductMarketOverlap Post (+3), Nonreportable ProductMarketOverlap Post (+2), Nonreportable ProductMarketOverlap Post (+1), Nonreportable ProductMarketOverlap AcqYear, Nonreportable ProductMarketOverlap Before (−2), and Nonreportable ProductMarketOverlap Before (−3). Our exclusion year is (−1).
As shown in column (3), markups increase by approximately 26 percentage points on average over the three years following a nonreportable acquisition that consolidates product markets. Given the sample mean markup of 2.278, this represents an 11.3% increase over the average markup. Compared to recent studies estimating the impact of horizontal mergers on markups, our findings are consistent with those of Blonigen and Pierce (2016), who report increases ranging from 15% to over 50% in the context of U.S. manufacturing plant acquisitions. In magnitude, the 11.3% increase in markups that we estimate is also broadly comparable to the 5.9% announcement returns reported in column (4) of Table IV.
To assess whether our estimates are economically sensible given the relative size of targets to acquirers, we estimate an OLS regression model of Markup—the percentage change in markups from one year before to one year after an acquisition—on Nonreportable ProductMarketOverlap, along with the main effects. The model includes industry and year fixed effects. Summary statistics and cross-sectional results by quantiles of RelativeSize are reported in Internet Appendix Table IA.IX. Panel A shows that a target is on average one-quarter (or 0.259) the size of an acquirer. Panel B presents regression estimates for subsamples based on quantiles of relative size.33 Overall, we find that changes in markups grow following nonreportable acquisitions involving overlapping product markets when the relative size of the target to the acquirer is larger. In the upper decile of relative size—where targets are at least half the size of acquirers—markups increase by an average of 47.2%, consistent with the magnitudes reported in recent studies (e.g., Blonigen and Pierce (2016)).34
Finally, using our specifications from columns (3) and (4) of Table V, we examine whether results vary based on types of intangible assets that, when consolidated, might lead to increased markups. We focus on acquisitions of brands and technologies, given that we expect them to be economically important in developed product markets (e.g., consolidating two competing brands likely has an immediate impact on the acquiring firm’s market power). Panel B of Table V presents results from separately estimating equation (3) for subsamples in which the deal involves (i) brands or technology (columns (1) and (2)), (ii) brands and technology (columns (3) and (4)), and (iii) neither brands nor technology (columns (5) and (6)). Overall, we find that the increase in markups documented in Panel A concentrates among deals that involve intangibles related to developed product markets.35 By contrast, in columns (5) and (6), where we restrict our sample to deals that do not involve either a brand or technology, we find no significant differences in markups. Panels B through D of Figure 7 display these patterns graphically. Taken together, our results in Table V suggest that nonreportable deals between firms in overlapping product markets are more likely to increase acquirers’ market power, which is exacerbated when these deals relate directly to developed product markets.36
C. Undeveloped Product Market Consolidation
We next explore the implications of accounting rules for the avoidance of premerger review in undeveloped product markets, where intangible assets are as prevalent, if not more so, relative to developed product markets.
Since the anticompetitive effects of deals in undeveloped product markets are unlikely to manifest in markups, we turn our attention to the pharmaceutical industry, defined by antitrust authorities as chemical manufacturing (NAICS 325), which receives significant antitrust enforcement, as reflected in the rate of Second Requests (14.72%) reported in Internet Appendix Table IA.II.37 In particular, the early-stage intangible assets involved in pharmaceutical deals—for example, in-process R&D—are a key concern in public and private antitrust-related litigation (see Section VII of the Internet Appendix). Evidence in Cunningham, Ederer, and Ma (2021) suggests that this concern might be warranted, given acquirers’ incentives to preempt competition by consolidating undeveloped pharmaceutical projects. Indeed, accounting standard setters have provided examples of how fair-value measurements of acquired in-process R&D should be conducted when the acquirer does not intend to complete the project but instead wants to lock up the project to “prevent its competitors from obtaining access to the technology.”38 We also find that intangible assets are prevalent in our sample of pharmaceutical deals, especially for those that are nonreportable to antitrust authorities.39 We examine overlapping drug projects acquired in these deals in the sections that follow.
C.1. Acquisitions to Preempt Future Competition
Following Cunningham, Ederer, and Ma (2021), we identify overlapping drug projects by examining the intended TC and MOA. If both the acquirer and the target have a drug project that shares the same TC and MOA, we categorize that project as overlapping. We then calculate the number of overlapping projects scaled by the target firm’s total number of drug projects. Thus, if a target firm has only one project and that project overlaps with one of the acquirer’s, the project is likely the focus of the deal. By contrast, if a target has many projects and one of the projects overlaps with a project of the acquirer, the overlapping project is less likely to be the focus.
(4)
where is either an indicator variable equal to one if at least one project overlaps or a continuous variable equal to the proportion of projects that overlap in year .
Table VI, Panel A, presents results from estimating equation (4). Results in column (1) indicate that, on average, nonreportable deals are associated with a 8.2 percentage point higher likelihood of involving overlapping drug projects relative to reportable deals, that is, overlapping drug projects in nonreportable deals occur at nearly three times the rate of overlapping drug projects in reportable deals (i.e., 13% versus 4.7%). Columns (3) and (4) report results using our continuous measure of overlap as the dependent variable and show that nonreportable deals have a greater proportion of overlap between acquired projects. For roughly half of the horizontal deals in this sample, more than 15% of acquired projects overlap, all of which were nonreportable, in part because in-process R&D comprises nearly 35% (relative to 15% for reportable deals) of the deal but is not accounted for in the SoP test.40
Overlapping Pharmaceutical Projects and Nonreportable M&As
| This table presents results from OLS regressions of pharmaceutical projects on an indicator for whether the deal was reportable or nonreportable to the antitrust regulators. The main variable of interest in Panel A, Nonreportable, is an indicator that assumes the value of one if the target firm’s assets are below the size-of-person (SoP) asset threshold, and zero otherwise. In Panel B, the main variable interest in columns (1) to (3) is Nonreportable; in columns (4) and (5), it is the interaction term Nonreportable EconImportance. In column (4), assumes the value of one if the target’s drug project is in Phase 3 of trials, and zero otherwise, and in column (5), assumes the value of one if there are three or fewer competitors developing an overlapping drug that matches the target’s project, and zero otherwise. The main variable of interest in Panel C, Nonreportable AcquiredProject, is an interaction term that assumes the value of one when an overlapping project is acquired in a nonreportable deal, and zero otherwise. In columns (1) and (2) of Panel A, the dependent variable, Pr(ProjectOverlap), is an indicator variable that assumes the value of one if the target firm and the acquiring firm have at least one drug development project that directly overlaps, and zero otherwise. In columns (3) and (4) of Panel A, the dependent variable, ProjectOverlap, is a continuous variable that measures the proportion of the target firm’s drug development projects that overlap with the acquirer’s drug development projects. In all columns of Panels B and C, the dependent variable, ProjectDisc’d, is an indicator that assumes the value of one if a drug project is discontinued after the acquisition date. All variables are described in Internet Appendix Section VI. For both Panels A and B, we vary the inclusion of fixed effects as follows. In columns (1) and (3) of Panel A, we exclude filing-year fixed effects, and in columns (2) and (4), we include filing-year fixed effects. In column (1) of Panel B, we exclude fixed effects, in columns (2) we include TC fixed effects, and in column (3), we include TC and filing-year fixed effects, respectively. For Panel C, we vary the fixed effects structure across columns. We also vary the inclusion of our control variables, for example, we include control variables in columns (2), (4), (6), and (8). Control variables included in the estimations in Panel C but not reported are Size, Sales, Leverage, EBITDA/Assets, Cash/Assets, CashFlow/Assets, R&D, and Q. Robust t-statistics are reported in parentheses and calculated using standard errors clustered at the filing-year level. *, **, *** represent significance at the 10%, 5%, and 1% level, respectively. | ||||
|---|---|---|---|---|
| Panel A. Overlapping Projects | ||||
| (1) | (2) | (3) | (4) | |
| Dependent Variable: | Pr(ProjectOverlap) | Pr(ProjectOverlap) | ProjectOverlap | ProjectOverlap |
| Nonreportable | 0.082** | 0.073* | 0.015** | 0.012** |
| (2.26) | (2.05) | (2.75) | (2.68) | |
| Observations | 169 | 169 | 169 | 169 |
| Adjusted | 0.016 | 0.051 | 0.045 | 0.051 |
| Filing-Year F/E | N | Y | N | Y |
| Panel B. Drug Project-Level Development and Competition | |||||
|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | |
| Dependent Variable: | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d |
| : | Phase 3 Trials | High Market Concentration | |||
| Nonreportable | 0.148** | 0.332** | 0.595* | 0.597* | 0.306 |
| (2.92) | (2.39) | (2.29) | (2.25) | (1.07) | |
| EconImportance | 0.042 | −0.088*** | |||
| (0.61) | (-6.10) | ||||
| Noreportable EconImportance | 0.643*** | 0.686*** | |||
| (9.44) | (20.33) | ||||
| Observations | 210 | 210 | 210 | 210 | 210 |
| Adjusted | 0.016 | 0.044 | 0.088 | 0.097 | 0.121 |
| Therapeutic Class F/E | N | Y | Y | Y | Y |
| Filing-Year F/E | N | N | Y | Y | Y |
| Panel C. Drug Project-Level Development | ||||||||
|---|---|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | |
| Dependent Variable: | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d | ProjectDisc’d |
| Nonreportable AcquiredProject | 0.161*** | 0.331** | 0.235*** | 0.414** | 0.216*** | 0.361** | 0.282*** | 0.366* |
| (3.51) | (2.26) | (4.60) | (2.82) | (4.31) | (2.52) | (6.44) | (2.08) | |
| Nonreportable | −0.013 | −0.026 | 0.000 | −0.007 | −0.043 | −0.026 | −0.002 | −0.018 |
| (−0.38) | (−0.79) | (0.00) | (−0.23) | (−0.87) | (−1.25) | (−0.05) | (−0.64) | |
| AcquiredProject | 0.209*** | 0.058 | 0.200*** | 0.028 | 0.190*** | 0.074 | 0.137*** | 0.074 |
| (4.76) | (0.41) | (4.17) | (0.19) | (4.15) | (0.52) | (3.40) | (0.42) | |
| Observations | 3,504 | 2,541 | 3,504 | 2,541 | 3,504 | 2,541 | 2,658 | 2,003 |
| Adjusted | 0.038 | 0.065 | 0.043 | 0.073 | 0.071 | 0.104 | 0.265 | 0.328 |
| Controls | N | Y | N | Y | N | Y | N | Y |
| Therapeutic Class F/E | N | N | Y | Y | Y | Y | N | N |
| Filing-Year F/E | N | N | N | N | Y | Y | Y | Y |
| TC-by-MOA F/E | N | N | N | N | N | N | Y | Y |
C.2. Project Development after Acquisition
We next examine how consolidated pharmaceutical projects develop post-acquisition. Acquirers may choose to continue projects if synergies exist (Beneish et al. (2022)). Alternatively, they may discontinue projects when the acquisition aims to preempt competition. Consistent with the latter, Cunningham, Ederer, and Ma (2021) show that incumbents discontinue acquired drug projects when a project potentially substitutes for the incumbent’s project. In our setting, the acquirer’s ability to discontinue overlapping projects is likely greater when the size of the target firm’s assets is below the asset size threshold, allowing the merger to escape ex ante merger review. Moreover, as we show in Section V.A.3, the threat of private litigation is near zero because drug development occurs before commercialization.
To examine whether drug development differs between nonreportable and reportable deals, we exploit our project-level data, which track project development throughout its life cycle. We identify a project as discontinued if, after the acquisition date, the project’s status is “discontinued” or “no development reported.” For this analysis, we use a sample of 210 overlapping projects across the 13 deals that involve overlapping projects, approximately 50% of which (98 of 210) are discontinued after acquisition.
Panel B of Table VI presents results from regressing an indicator variable for whether a project is discontinued (ProjectDisc’d) on Nonreportable. Overall, we find that acquired overlapping projects in nonreportable deals are about 15 percentage points more likely to be discontinued than overlapping projects in reportable deals. This represents a 40% increase over the 37.5% probability of discontinuing a project in reportable deals. These results are robust to the inclusion of TC fixed effects, which control for the possibility that nonreportable and reportable deals differ in development rates (e.g., due to the types of drug projects acquired) as well as filing-year fixed effects (see columns (2) and (3), respectively).
Next, we examine whether the likelihood of discontinuing a drug project increases with economic importance. Given the absence of data on drug sales for yet-to-be-developed drugs, we construct two measures of economic importance: advanced drug development and market concentration. Our first measure, advanced drug development, is an indicator for whether the target’s drug project is in Phase 3 trials, that is, the final-phase marketing approval from the U.S. Food and Drug Administration.41 Our second measure, high market concentration, is an indicator for whether three or fewer competitors are developing an overlapping drug with the same TC and MOA.42 We interact our measures with and report the results in columns (4) and (5) of Panel B. The interaction term in both columns is positive and statistically significant, indicating that overlapping drug projects in nonreportable deals are more likely to be discontinued when the target’s project is in an advanced stage of development (column (4)) and when the benefits to reducing competition are larger (column (5)).
(5)
where is an indicator variable that assumes the value of one if the project is an overlapping project acquired via M&A and zero otherwise. Given the larger sample for this analysis, we can include a vector of controls that proxy for the size and the financial health of the acquirer (e.g., Size, Sales, Leverage, EBITDA/Assets, Cash/Assets, CashFlow/Assets, R&D, and Q). To account for baseline differences in discontinuation rates inherent to a project’s specific drug category, we also include fixed effects for each MOA pair (). All variables are defined in Section VI of the Internet Appendix.
Panel C of Table VI reports the results from estimating equation (6). Overall, we find that overlapping projects in nonreportable deals are 16 percentage points more likely to be discontinued (column (1))—an increase of approximately 77% relative to the discontinuation rate in reportable deals. Notably, the coefficient on Nonreportable is not statistically significant at conventional levels, suggesting that the discontinuation rate of internally developed projects for acquirers with nonreportable deals does not differ from that of internally developed projects in acquirers with reportable deals. Thus, no ex ante differences exist in the development rates across these firms, consistent with prior work on large firms having incentives to stifle innovation (e.g., Seru (2014)).
We obtain similar inferences when we control for the size and financial health of the acquirer (column (2)), but we also find that the discontinuation rate for acquired projects in reportable deals does not differ from that of internally developed projects. We also find similar results when we include TC (filing-year) fixed effects to control for variation in discontinuation rates due to unobservable drug-therapy characteristics (time trends) in columns (3) and (4) ((5) and (6)). Finally, in columns (7) and (8), we replace TC fixed effects with MOA pair fixed effects and find that, even within the same TC and the same MOA, acquired overlapping projects in nonreportable deals have a higher rate of discontinuation than internally generated ones. Taken together, our results in Panel C are consistent with acquirers of overlapping projects in nonreportable deals striving to reduce product market competition.
V. Implications and Additional Analysis
We now discuss enforcement challenges associated with deals that escape ex ante merger review. We also estimate the effects of a change in enforcement policy, consider various threats to our inferences, investigate the impact of a change in intangible capital accounting standards on nonreportable deals, and address how current reporting practices might impact the likelihood of deal completion.
A. Implications for Public and Private Enforcement
A.1. Public Enforcement
Given antitrust authorities are resource-constrained, one possibility is that enforcement of deals subject to the SoP test, that is, deals that fall between the lower and upper size-of-transaction test, receive lax enforcement. To examine this possibility, we obtain data from HSR reports on Second Requests (the highest degree of antitrust enforcement prior to litigation). We find that roughly 25% of all Second Requests correspond to such deals (see Panel A in Internet Appendix Table IA.XII) and are similar in investigative length to the largest U.S. mergers (e.g., 146 days versus 160 days; Tucker (2013)). Thus, public enforcement of deals scrutinized under the SoP appears to be costly from a compliance perspective but does not lead to denial of the merger (these deals are 29 times less likely than those above the upper deal-size threshold to be subject to actual enforcement actions; see Section VIII of the Internet Appendix).
However, firms can reduce the risk of a Second Request by withdrawing and refiling their premerger notifications, which in effect restarts the 30-day waiting period, giving the FTC and DOJ additional time to evaluate the deal before they must initiate a Second Request. Firms typically use “pull-and-refile” when they believe doing so will give them a reasonable chance of a clearance without a Second Request. We compile a data set of pull-and-refiles by searching the public disclosures of public targets and acquirers and report our findings in Panel B of Internet Appendix Table IA.XII. Roughly 11% of pull-and-refiles occur in deals subject to the SoP test, suggesting that relying on Second Requests incidences alone likely underestimates the actual threat of enforcement for SoP deals.43
A.2. Private Enforcement
Given that accounting rules lead to many deals escaping ex ante merger review, one might wonder whether private litigation substitutes for public enforcement against these deals (e.g., Lancieri, Posner, and Zingales (2023)), which is permitted under the Clayton Act. The fixed costs of private antitrust litigation for both plaintiff and defendant are high (e.g., Davis and Kohles (2022)), potentially limiting its prevalence. Nonetheless, we examine the prevalence of private litigation in nonreportable versus reportable deals in Section VIII of the Internet Appendix. We find that 1.2% (i.e., 23/1,918) of the deals in our sample have private antitrust-related litigation, which is comparable to the number of deals litigated by the FTC (1.4%; Billman and Salop (2023)). Among the 23 deals with private litigation, eight relate to nonreportable deals. Thus, 2.1% of nonreportable deals in our sample faced a private antitrust lawsuit, which is 50% higher than the rate of public litigation for reportable deals. Furthermore, most (60%) private antitrust litigation is in the same industries as public litigation (i.e., technology and pharmaceutical sectors). Given that private antitrust litigation is financially costly for defendants (roughly $200 million on average for plaintiff-favorable rulings in one-third of cases), private antitrust litigation seems to at least partially substitute for public enforcement against anticompetitive deals that escape ex ante merger review due to the accounting rules for intangible assets.44
A.3. Frictions to Litigation
While private litigation partially offsets a lack of public enforcement, plaintiffs—usually customers or competitors—face different thresholds for court dismissal than the FTC or DOJ. For instance, competitors must prove both that the merger violates antitrust law and that its alleged harm is anticompetitive (e.g., predatory pricing practices; see the U.S. Supreme Court’s decisions in Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc. and Cargill, Inc. v. Monfort of Colorado, Inc.). Private litigation is further limited by the nature of the markets impacted by intangible-intensive consolidations, like those in undeveloped product markets. Indeed, all of the court complaints in our sample include allegations of anticompetitive harm in either existing product markets or those with sophisticated customers. Given the United States relies on a combination of public and private enforcement (Baer (2014)), when private enforcement faces legal constraints or when no private enforcers intervene, anticompetitive acquisitions will likely go unchecked.45
B. Estimated Antitrust Enforcement Effects
We next provide back-of-envelope calculations of the effect of an alternative treatment that requires firms to include intangible assets in their calculations for the SoP test. Such a rule change would increase the number of reportable mergers as well as the compliance costs to firms and enforcement costs to the DOJ and FTC. Furthermore, it could also deter M&As with increased antitrust costs and enforcement risk. We estimate the magnitudes of these effects in our setting and consider how such a change may impact firms’ incentives to manipulate deals to avoid premerger review. We also consider a recent change to accounting standards to understand the prevalence of manipulation around the threshold.
B.1. Enforcement Costs
To compute our back-of-envelope calculation, we begin with the 263 deals we estimate would be reportable if intangible assets were included in the SoP test (Figure 4). However, 44% of those new filings would involve nonhorizontal deals (Table I, Panel B), which are unlikely to receive a Second Request. The costs to enforcement agencies for reviews that do not require a Second Request are minimal (i.e., less than the filing fees; Wollmann (2026)). Thus, most of the premerger-review enforcement costs would come from the increase in reported horizontal deals, which constitute 55% of reportable deals. Based on these amounts, recognizing intangible assets in the SoP test would increase the number of horizontal deals by 145 each year. We expect that 40% of these deals would be granted an early termination of the premerger review (see Internet Appendix Table IA.IV). Therefore, an additional 90 horizontal deals (i.e., 60% of the 145 deals) would be reviewed each year if intangible assets were included in the SoP test.
Of these 90 additional deals, roughly 6% (five or six) would likely result in a Second Request, assuming a proportional increase in Second Requests (see Section I.A). Thus, at an estimated cost per Second Request investigation of $163,000 to $215,000 (Wollmann (2026)), recognizing intangibles in the SoP test would cost the agencies an estimated additional $815,000 to $1,075,000 each year (a 2.6% to 3.5% increase in total enforcement costs of $31 million to $41 million). When we include the expected effects of deterrence (i.e., we estimate that 23 of the 90 additional horizontal deals would not occur if managers knew the deal would not pass antitrust review; see Section IX of the Internet Appendix), our estimates decrease to $652,000 to $860,000 (2.1% to 2.8% of annual enforcement costs).
An important caveat to these estimates is that the capacity constraints of antitrust authorities may limit the agencies from allocating enforcement effort to challenging large M&A deals that meet statutory reporting thresholds. An increase in reportable mergers from including intangibles in the SoP test could further strain these resources. Such an increase in reportable deals could even relax the scrutiny of other enforcement efforts to substitute resources in scrutinizing additional reportable deals. Thus, an increase in reportable deals may not necessarily lead to a proportional increase in Second Requests or subsequent litigation.
B.2. Incentives to Manipulate Deals to Avoid Enforcement
While including intangibles in the SoP test would likely increase the number of deals subject to premerger review, doing so could also alter firms’ incentives to manipulate deal terms to sidestep review. In line with this kind of manipulation, Kepler, Naiker, and Stewart (2023) find a 45% higher-than-expected number of deals just below the lower deal-size threshold. Applying this magnitude to our estimates of the 90 annual horizontal mergers that our analysis suggests would become reportable, 41 would continue to avoid reporting via manipulation.
In terms of managers’ incentives to avoid reviews, they may prefer that deals be nonreportable if they believe that antitrust authorities block deals that are not anticompetitive. However, our collective results on higher markups, project discontinuation, and private litigation are more indicative of nonreportable deals increasing firms’ market power rather than avoiding an imperfect antitrust authority. Another explanation could be that managers want to avoid reviews because reviews increase deal termination or renegotiation risk, which is costly to acquirers. However, cancellations and renegotiations are rare (3.9% and 3.1% of all deals, respectively), and only 0.2% of these deals cite “antitrust authority concern” as the reason for cancellation or termination (see Internet Appendix Table IA.XIII).
B.3. Changes to Accounting Standards
To better understand managers’ incentives to manipulate around the SoP thresholds, we examine how firms respond to changes in accounting standards that shift some deals from nonreportable to reportable. Given that the SoP test uses the book value of assets to determine premerger review requirements, any change that moves assets to the balance sheet could shift deals to being reportable. For deals with anticompetitive implications, such a shift would heighten the risk of antitrust enforcement solely because the deal would become reportable. If firms internalize these costs, we expect such a change to an accounting standard would influence the decision to acquire or the timing of deals. Consistent with such patterns, we find a 50% increase in the proportion of nonreportable deals shortly after an accounting standard that moved leases onto firms’ balance sheets was announced but before its adoption (see Internet Appendix Section X). We also find that the increase is driven by target firms that, if operating leases had been recognized on the balance sheet, would have shifted from nonreportable to reportable. These results further support the view that requiring firms to include intangible capital in the SoP test would result in additional avoidance strategies by managers.
VI. Conclusion
We find that the use of antitrust screening criteria that rely on accounting information leads to thousands of M&A transactions being nonreportable to antitrust authorities, despite these deals being otherwise similar to reportable deals. These nonreportable deals consolidate product markets in intangible capital-intensive industries that antitrust authorities have expressed particular concerns over (e.g., technology and pharmaceutical markets), as these deals often involve the acquisition of brands, patented technology, and in-process R&D that lead to anticompetitive behavior for a significant fraction of deals. We find that acquirers and their rivals benefit from nonreportable deals in terms of higher equity values and product markups. Furthermore, we find that nonreportable deals in the pharmaceutical industry are more likely to involve overlapping projects that are subsequently discontinued.
Our findings have policy implications. Given antitrust authorities’ reliance on screening thresholds, accounting standards can influence the types of deals that avoid ex ante merger review and thus impact market structure. In this regard, our study suggests that enforcement agencies’ concerns about the limitations set by premerger-review thresholds may be warranted. We add to this debate by showing that industries that are more intangible-intensive are more likely to consolidate, increasing firms’ market power but going undetected by antitrust authorities. Overall, our study suggests that the continued growth of intangible assets may exacerbate market consolidation in the sectors that are of most concern for consumers.
Editors: Antoinette Schoar, Urban Jermann, Leonid Kogan, Jonathan Lewellen, and Thomas Philippon
Leave a comment