Accounting for Goodwill*

Stefan J. Hubera Charles G. McClureb

October 27, 2023

ABSTRACT
A significant portion of a merger’s purchase price is allocated to goodwill. Currently,
goodwill is not amortized but tested annually for impairment. When managers care
about earnings, goodwill’s accounting treatment can have large effects on future earnings
and may influence how much a manager will bid for a target company. We quantify the
effects of goodwill accounting by estimating a structural model of corporate takeovers.
Our estimates suggest that accrual accounting increases buyout premia by an average
of nearly 10 percentage points. If firms needed to amortize goodwill over 10 years, we
estimate premia would reduce by 6 percentage points and M&A volume would shrink
by 4.29% or $68.6 billion per year. Furthermore, the fraction of private equity acquirers
increases by 7.74 percentage points, shifting control over productive assets to the private
and financial sector. Our results suggest the accounting treatment for goodwill has a
meaningful effect on the market for corporate control.
JEL classifications: D44, G32, G34, M40, M41
Keywords: Goodwill, mergers and acquisitions, buyout premia, structural modeling

* We thank Matthias Breuer, Robert Göx, Valeri Nikolaev, Katherine Schipper, Doug Skinner, Joe Weber,

Anastasia Zakolyukina, Stephen Zeff, and seminar participants at the University of Zurich, the Junior
Accounting Theory Conference, and the University of Chicago for helpful comments. We thank Boyie
Chen and Sato Maeda for excellent research assistance and Alexander Gorbenko and Andrey Malenko for
generously sharing their data on corporate takeover auctions. Stefan Huber thanks Rice University for
its financial support. Charles McClure thanks the Liew Faculty Fellowship, the Chookaszian Accounting
Research Center, and the University of Chicago Booth School of Business for financial support.
a Jones Graduate School of Business, Rice University, TX 77005.
b Booth School of Business, University of Chicago, IL 60637.
1. Introduction 1

Corporate mergers and acquisitions (“M&A”) are some of the most important corpo- 2

rate activities, yet nearly half of the aggregate deal value is assigned to non-identifiable 3

intangible assets, which are represented as goodwill. As a result, goodwill has become 4

the largest intangible asset on companies’ balance sheets, resulting in 6.92% of total assets 5

or $4.9 trillion in 2021. The size of goodwill suggests that the accounting treatment of 6

goodwill can have significant implications for the combined company’s future earnings 7

and may alter an acquirer’s willingness to pay (e.g., Graham, Harvey and Rajgopal, 2005; 8

Darrough, Guler and Wang, 2014). In this paper, we examine the real effects of goodwill 9

accounting on the market for corporate control. 10

Current accounting standards treat goodwill as an indefinitely-lived intangible asset 11

tested annually for impairment. However, this accounting treatment is not without contro- 12

versy. Standard-setters have debated how to treat goodwill since at least the 1960s because 13

of goodwill’s potential to significantly affect merger activity (Seligman, 1982; Rayburn and 14

Powers, 1991). Most recently, the Financial Accounting Standards Board (“FASB”) con- 15

sidered changing the accounting of goodwill to amortization before deciding to drop the 16

matter because it was unclear whether amortization was a clear improvement.1 17

Measuring the effect of goodwill accounting on the M&A market activity is difficult 18

because how accounting treats goodwill affects acquirers’ private values of a target. But 19

we do not observe these underlying private values, only the realized transaction prices. 20

The transaction price, however, is an equilibrium outcome determined by the interplay 21

1 Seehttps://www.fasb.org/Page/ProjectPage?metadata=fasb-IdentifiableIntangibleAssetsan
dSubsequentAccountingforGoodwill-022820221200.

1
between the valuation of the buyer and the competition from other potential acquirers.2 1

As a result, it is necessary to disentangle the competition effect from the valuation effect 2

to recover the underlying values and quantify the role of goodwill accounting. 3

To address these difficulties, we develop and estimate a model of corporate takeovers 4

where potential acquirers offer bids based on their private values and the competition from 5

others. Following Gorbenko and Malenko (2014) and Haile and Tamer (2003), we model 6

takeovers as auctions and assume that the current market value serves as a reserve price, 7

bidders do not bid above their private values, and bidders do not permit another bidder to 8

win if they could have bid more and maintained a positive surplus. This structure allows 9

us to disentangle the competition effects from the underlying valuations of the bidders. 10

As a result, we can estimate the effect of goodwill accounting on bidders’ private values 11

and the takeover market. 12

Our model distinguishes two types of bidders: strategic and financial (Gorbenko 13

and Malenko, 2014). Financial bidders, often private equity investors, seek to maximize 14

fundamental value, i.e., expected future cash flows. Strategic bidders, such as competitors, 15

suppliers, or customers, also want to maximize fundamental value but are myopic and 16

also care about short-term stock prices (Stein, 1989). Stock price is a function of reported 17

earnings, which is affected by the accounting for goodwill. We allow the relative preference 18

2 The current “impairment-only” accounting regime went into effect in 2001. Prior to this period, the rules
for merger accounting were determined by APB Opinion 16. APB 16 offered two accounting treatments
for mergers: the purchase method and pooling of interests. Under the purchase method treatment, firms
were required to amortize goodwill but could choose the duration for this amortization. Pooling of interests
allowed for the acquirer to assume the target’s assets and liabilities without having to step up their book
values. As such, it did not create an “earnings drag” from the amortization of goodwill and was therefore
perceived as favorable. Pooling of interests, however, was only permitted if the merger was deemed a merger
of equals and satisfied several other criteria. However, inferences that can be drawn from that regime are
limited by the self-selection of firms using either of the two methods. Furthermore, the importance of
goodwill has significantly increased since 2001, increasing from $771.1 bn to $4.9 trillion in 2021.

2
of accounting to cash flows to vary within the group of strategic bidders to account for the 1

presence of private strategic bidders (who may care less about earnings) and differential 2

earnings preferences among managers of public firms. As such, the valuation of a strategic 3

bidder depends not only on the fundamental value of the target, but also on the amount 4

of goodwill created, and its subsequent accounting treatment. 5

An important implication is that the strategic bidder does not fully internalize the 6

cash outlay to acquire the target. This is because she myopically cares about earnings, 7

and earnings at the time of the acquisition do not decrease from this cash outflow. The 8

future impact of an acquisition on earnings occurs from subsequent impairments and 9

amortizations, which are discounted to the present. Consequently, a strategic bidder has 10

a higher effective valuation and can therefore bid more aggressively. 11

We estimate our model using 861 all-cash deals executed as takeover auctions on 12

public targets effective from July 2001 until September 2022.3 With this sample, we 13

estimate our model using simulated maximum likelihood, where the parameters differ 14

between financial and strategic bidders. Similar to Gorbenko and Malenko (2014), we 15

find that strategic bidders’ valuations are higher than financial bidders and are positively 16

related to current market conditions. Strategic bidders have higher values because they 17

care about the future earnings stream. This preference for earnings effectively reduces 18

the acquisition cost because the bidder defers the earnings impact to an uncertain date 19

when an impairment occurs. As a result, strategic bidders only partially internalize the 20

purchase price for the target. Our estimates imply earnings receive about two-fifths the 21

weight of cash flows for the average strategic bidder so that under the current accounting 22

3 We classify a takeover as an auction if the target reported at least two bidders.

3
regime where firms only test for impairment, the average strategic bidder acts as if she 1

only internalizes 85% of the acquisition price. This estimate quantifies the commonly cited 2

trade-off that firms face between maximizing cash flows and financial-reporting concerns 3

(e.g., Matsunaga, Shevlin and Shores, 1992; Bens, Nagar, Skinner and Wong, 2003; Graham 4

et al., 2005). 5

An important reason why we estimate a model of merger activity is to quantify how 6

the merger market would change under hypothetical accounting regimes. To achieve this, 7

we simulate counterfactual takeover auctions to quantify both deal-level changes, such 8

as deal premia, and broader distributive effects, such as the value of assets controlled by 9

private equity investors. To understand how accounting influences takeovers, we compare 10

the current regime to a benchmark where all bidders care only about cash flows. In this 11

counterfactual, no bidder cares about future earnings and therefore must fully internalize 12

the purchase price. Without the ability to delay recognizing some of the cost, strategic 13

bidders’ valuations fall, so the average takeover premium declines by 13 percentage points, 14

and aggregate deal values would decrease by 9.94%. 15

Having demonstrated the sizeable effect of accrual accounting on takeovers, we also 16

compare the current regime to alternative regimes that would amortize goodwill. We 17

focus on the hypothetical accounting regime where firms must amortize goodwill over 18

ten years, and goodwill is subject to annual impairment testing.4 Relative to the current 19

accounting regime, we estimate that with a ten-year amortization schedule, the average 20

bid premium declines by 6 percentage points, and the number of deals failing increases 21

4 This regime corresponds to ASC 350-20-35, which permits private firms to amortize goodwill over ten
years (or less) instead of treating goodwill as an indefinitely-lived intangible asset.

4
by 10%. Together, these two effects reduce aggregate deal value by 4.29%. Overall, this 1

reduction would equate to a reduction of $68.6 billion in deal value for 2021. 2

Under this alternative accounting standard that amortizes goodwill, not only do trans- 3

action prices and aggregate deal values change, but the type of winners do as well. Given 4

the volume of deals, such changes in the makeup of winners can influence the ownership 5

of a substantial portion of the economy. In particular, adopting an accounting standard 6

that amortizes goodwill reduces the relative strength of strategic bidders because it leads 7

to earlier expensing of goodwill compared to an impairment-only standard. The earlier 8

expensing decreases strategic bidders’ target values but does not affect financial bidders’ 9

values. Our counterfactual simulations indicate that this shift in strategic bidders’ values 10

increases the likelihood of a financial bidder winning the takeover from 29.6% to 37.4%. 11

Combined with the changes in deal value, we estimate the increase in assets held by 12

financial bidders to be 20.7%. 13

We perform several additional counterfactual analyses to further explore the relation 14

between goodwill and mergers. We exploit the heterogeneity in the purchase price allo- 15

cated to goodwill and the presence of financial bidders across industries. The effect of 16

goodwill accounting is amplified in industries like Business Equipment, where a greater 17

fraction of the purchase price is allocated to goodwill. We also examine how the compet- 18

itive environment influences deal characteristics. Increasing the proportion of financial 19

bidders magnifies the estimated effect because financial bidders tend to have lower val- 20

uations than strategic bidders. By contrast, adding an additional financial bidder has 21

the reverse effect because it increases the competition for the target. This result supports 22

Bulow and Klemperer (2009) that more competition yields higher payoffs for sellers. 23

5
 Our paper contributes to two strands of literature. First, we add to the sizeable takeover 1

literature.5 Several papers examine how takeovers are shaped by the composition of po- 2

tential acquirers (Gorbenko and Malenko, 2014; Gorbenko, 2019), the information envi- 3

ronment (Gentry and Stroup, 2019), and the threat of entry (Dimopoulos and Sacchetto, 4

2014). However, this literature, primarily in finance and economics, often disregards the 5

accounting for these acquisitions. Several accounting studies focus on how accounting 6

influences the takeover market, such as firms’ accounting quality on the type of acqui- 7

sition (e.g., McNichols and Stubben, 2015; Marquardt and Zur, 2015) and the effect of 8

goodwill on takeover premia (e.g., Robinson and Shane, 1990; Bartov, Cheng and Wu, 9

2021). Research on goodwill accounting documents how economic incentives shape the 10

purchase price allocation (Shalev, Zhang and Zhang, 2013) and subsequent impairments 11

(Beatty and Weber, 2006; Li and Sloan, 2017; Glaum, Landsman and Wyrwa, 2018). We 12

add to this literature in two ways. First, we quantify how goodwill accounting affects 13

acquirers’ valuations. Second, by explicitly modeling competition, we address not only 14

how goodwill accounting affects deal pricing but also how it influences the allocation of 15

assets between private and public owners. 16

Closest to our paper is Bartov et al. (2021), which provides reduced-form evidence 17

of increased overpayment of public acquirers after the passage of SFAS 142. We differ 18

from the work in Bartov et al. (2021) along two critical dimensions. First, we explicitly 19

model the competition among bidders, which allows us to recover bidders’ underlying 20

valuations and how they are shaped by goodwill accounting. Second, estimating a model 21

of competition between different types of bidders allows us to quantify features of the 22

5 See Betton, Eckbo and Thorburn (2008) and Eckbo (2009) for reviews.

6
M&A market beyond just takeover prices, such as the distribution of assets between 1

financial and strategic bidders. 2

Second, our paper is related more broadly to the real effect of financial reporting.6 3

Surveys suggest accounting can influence firms’ investment decisions (Graham et al., 4

2005). Many studies in this area focus on intangible assets, in part, because intangibles 5

are a perennial focus of standard setters as accounting incompletely reflects their value.7 6

Several studies examine how the imprecision of accounting can alter the incentive of 7

the firm to make value-maximizing investment (e.g., Kanodia, Singh and Spero, 2005; 8

Geng, Zhang and Zhou, 2023; McClure and Zakolyukina, 2023). Kanodia, Sapra and 9

Venugopalan (2004) shows whether or how intangible assets are measured can induce 10

changes in managers’ incentives to invest. However, most of this literature focuses on 11

investment into internally-generated intangibles, such as R&D and advertising (e.g., Terry, 12

2023). We complement this literature by focusing on the largest recognized intangible 13

asset—goodwill—and show it has a meaningful effect on the market for corporate control. 14

By doing so, our paper may interest regulators and standard setters. Our results 15

suggest that accounting for goodwill influences whether strategic or financial bidders are 16

more likely to acquire target companies. Our findings indicate that reducing the advantage 17

of strategic bidders by amortizing goodwill can shift more assets toward financial bidders, 18

who are often private equity funds. These results speak directly to the SEC’s concerns 19

over the public’s inability to invest in large portions of the economy because of the rise of 20

private funding.8 As our paper shows, accounting standards can contribute to the balance 21

6 Fora review of this literature, see Kanodia and Sapra (2016).

7 Fora discussion of the considerations by standard setters, see Appleton, Barckow, Botosan, Kawanishi,
Kogasaka, Lennard, Mezon-Hutter, Sy and Villmann (2022).
8 See, for instance, https://www.sec.gov/news/speech/lee-sec-speaks-2021-10-12.

7
between public and private markets. Our results suggest an additional consideration for 1

standard setters as they continue to debate how to account for intangible assets. 2

2. Institutional Background 3

2.1 Accounting for takeovers 4

In an acquisition, the acquirer values the identifiable assets and liabilities at their 5

fair value. The difference between the purchase price and the fair value of the assets, 6

less the liabilities, is classified as goodwill. Effective December 15, 2001, Statement of 7

Financial Accounting Standards (SFAS) 142 specified the accounting for goodwill.9 Under 8

SFAS 142, goodwill is not subject to amortization but is tested annually for impairment. 9

One challenge with impairing goodwill is it cannot be separately identified, so a firm 10

cannot determine its fair value directly. Instead, firms assess whether they need to impair 11

goodwill by comparing the fair value of the reporting unit to which the goodwill is 12

assigned with the reporting unit’s carrying value. If the fair value is less than the carrying 13

value, the firm must determine the fair value of the identifiable assets and liabilities, with 14

the fair value of goodwill set as the difference between the two. If this difference is less 15

than goodwill’s carrying value, the company must recognize an impairment charge to 16

decrease the carrying value to its fair value. 17

How to account for goodwill has been a perennial topic of interest to standard setters 18

9 Before the adoption of SFAS 142, takeovers were accounted for under Accounting Principles Board (APB)

16 and APB 17. If a takeover satisfies 12 criteria, APB 16 permitted firms to use the pooling-of-interests
method, where the target’s assets and liabilities are carried forward at their recorded amounts and the
retained earnings of the two companies are combined. If a takeover does not satisfy these criteria, APB
16 required firms to use purchase accounting, which entailed valuing the target’s assets and liabilities at
their fair value. The difference between the purchase price and the fair value of the target’s net assets was
classified as goodwill. APB 17 required firms to amortize goodwill for a period of less than 40 years.

8
(Seligman, 1982; Rayburn and Powers, 1991), and the current reporting regime is no 1

exception. Ramanna (2008) finds the creation of the impairment-only approach of SFAS 2

142 was the result of political pressure by firms, as managers valued the discretion that 3

periodic impairment provided relative to amortization (Ramanna and Watts, 2012). 4

In 2018, the FASB re-examined the accounting for goodwill because many considered 5

the current regime of annual tests for impairments as costly to perform and subjective 6

in nature (Maurer, 2022). In its place, the FASB contemplated whether to require firms 7

to amortize goodwill over a 10- to 25-year period. Ultimately, the FASB decided to drop 8

the matter in 2022, with the FASB chair, Richard Jones, citing uncertainty about whether 9

amortization would lead to a meaningful improvement given the significance of the change 10

(Lugo, 2022). This paper helps resolve this uncertainty by examining how the takeover 11

market would change under different accounting regimes. 12

2.2 Corporate Takeover Auctions 13

A takeover auction usually starts when the target decides to sell itself to a potential 14

buyer.10 To facilitate the process, the target retains an investment bank to create a list of 15

potential acquirers, which the bank contacts to solicit their interest in acquiring the target. 16

Interested parties sign confidentiality agreements, allowing them access to nonpublic 17

information about the target, which assists them in determining their value of the target. 18

The bidding process typically proceeds in multiple rounds. In the first several rounds, 19

bidders submit nonbinding bids, which can change in each round and may be withdrawn 20

at any point. After each round of bidding, the target may select a subset of bidders to 21

continue to the next round and provide these bidders with additional information for due 22

10 For a more detailed description, see Hansen (2001) and Boone and Mulherin (2007).

9
diligence. At the end of this process, the target invites the remaining bidders to a final 1

round of bidding. Final-round bids are typically binding, but the target may negotiate 2

with some of the bidders to raise the price further. 3

Within a few days of receiving the final-round bids, the target chooses an acquirer, and 4

a takeover agreement is signed. Until the agreement is signed and the target announces 5

an agreement has been reached, the bidding process and bidders’ identities are kept 6

private.11 However, the target must disclose the bidding process when it puts the buyout 7

to a shareholder vote. This background is disclosed as part of the Merger Background in 8

either the DEF14A or SC-TOT documents, which allows us to observe the bids and the 9

type of each bidding participant. 10

Takeover auctions are most similar to an (ascending) English auction, with bidders 11

offering higher prices until only a single winner remains. However, takeover auctions have 12

several differences from English auctions. Unlike an English auction, takeover auctions 13

have several rounds of bidding during which bidders can exit and reenter the bidding 14

process or revise their bids downwards. Also, bidders are typically only informed about 15

the highest bid and are unaware of the number of other bids or their amounts. Finally, 16

targets design their own process, which may have interspersed rounds of negotiations. 17

One consequence of these negotiations is they can induce bidders to jump their bids and 18

bypass intermediate bids that we would expect from a pure English auction. As such, no 19

theoretical auction model describes the process of a takeover auction perfectly. Therefore, 20

we build upon the approach developed by Gorbenko and Malenko (2014) to estimate 21

11 In some instances, a target will pre-empt this takeover announcement and issue a press release that they

are in the process of looking for acquirers.

10
bidders’ valuations from the unstructured bidding process. 1

3. Stylized Facts and Sample Data 2

To motivate the importance of goodwill and subsequent impairments, we report several 3

stylized facts. Table 1 reports summary statistic data of goodwill impairments from 4

Compustat over the sample period of June 2001 through December 2021. The first column 5

shows that the probability a firm recognizes a goodwill impairment in a given year is 6

9.6%.12 This percentage varies by industry, with oil and gas firms having the highest 7

probability (13.7%) and medical and pharmaceutical firms with the lowest (7.3%). The 8

remaining columns in Table 1 show the fraction of the beginning-year goodwill amount 9

that is impaired, conditional on the firm recognizing an impairment. Although the mean 10

amount is 35.8%, there is significant skew as the median is only 22%. For instance, nearly 11

10% of all impairments are for the entire goodwill amount. For certain industries, this is 12

significantly higher, as oil and gas firms have over 25% of all impairments are for the entire 13

amount of goodwill. This table shows that goodwill impairments are not infrequent, and 14

when they do occur, it is often a large fraction of goodwill. 15

Table 2 reports the fraction of the purchase price allocation for public acquirers from 16

June 2001 until December 2021 that is attributed to goodwill. The data to construct this 17

table are from BVWire’s DealStat database, which is based on public acquirers’ subse- 18

quent filings and the disclosed purchase-price allocation. This table shows nearly half 19

(43.6%) of the purchase price is allocated to goodwill. Table 2 also shows this fraction 20

varies substantially across industries. Oil and gas firms have the lowest average goodwill 21

12 For reference, Potepa and Thomas (2023) provide evidence that only 25% of M&A transactions have a
related impairment charge within 10 years of the acquisition.

11
allocation (23.7%), whereas business equipment firms have the highest, with over half of 1

the purchase price allocated to goodwill (50.1 %). Presumably, this variation implies the 2

effect of goodwill accounting will vary by industry. Overall, Tables 1 and 2 show that 3

goodwill is a sizeable fraction of the purchase price and subsequent impairments are a 4

significant reduction in goodwill’s carrying value. 5

We analyze a sample of 861 corporate takeover auctions that were effective from July 6

1st 2001 to September 2022. The sample start date ensures that all takeovers are subject 7

to SFAS 142.13 We identify takeover auctions and collect the data following Gorbenko 8

and Malenko (2014).14 Briefly, we identify all takeovers of publicly-traded non-financial 9

firms in the Refinitiv SDC Platinum data with a non-missing deal value and where the 10

acquirer sought 100% of the target’s shares. We further restrict the sample to deals 11

that were completed with an all-cash consideration.15 We identify whether a deal was 12

a negotiation from the deal background section of the SEC merger filings of the target 13

company (PREM14A, DEFM14A, SC-TOT, and S4). Consistent with prior literature (Boone 14

and Mulherin, 2007; Gorbenko and Malenko, 2014), we classify a deal as an auction if two 15

or more potential bidders execute confidentiality agreements with the target. For the 16

sample of auctions, we hand-collect comprehensive information on the bidding process 17

from the merger background disclosures. This includes the type of bidder, i.e., strategic 18

or financial, the nature of their bid, formal, informal, no bid, or drop out, the value of each 19

13 Although SFAS 142 went into full effect for fiscal years ending after December 15, 2001, it already applied

to all deals completed after June 30, 2001. See https://www.fasb.org/page/PageContent?pageId=/refer

ence-library/superseded-standards/summary-of-statement-no-142.html&bcpath=tff.
14 Data from September 2001 to 2012 was generously provided to us by Alexander Gorbenko and Andrey

Malenko.
15 This restriction is necessary because the identification strategy relies on the value of the winning bid.

The value of a (partial) stock bid is to some extent uncertain when the merger is consummated (Gorbenko
and Malenko, 2014). Note, however, that we keep non-cash losing bids.

12
formal bid, and the date of any press release relating to the takeover auction. 1

Table 3 presents the summary statistics of our sample. The average bid premium is 2

42.8% above the stock price 4 weeks before the takeover announcement or the stock price 3

one day before the first press release about the auction, whichever is earlier. The average 4

number of bidders is 11; however, this amount has significant skew as the median is 5

only 6. On average 29% of bids in an auction are formal bids. Within the set of bidders, 6

approximately 41% are strategic, 29% are financial, and the remainder are of unknown 7

type. This split also manifests itself in the distribution of winners as we find that strategic 8

bidders win 63% of deals. For the auctions with a public acquirer, we collect purchase price 9

allocation from BVWire DealStats and complement it with hand-collected data whenever 10

missing. Among the 304 deals that we can find PPA information, the average allocation 11

to goodwill equals 46.3%. This fraction is significant and suggests that the accounting 12

treatment of goodwill can meaningfully influence the takeover market. The remaining 13

summary statistics largely comport to findings in prior research. 14

4. Model 15

Bidders decide how much to bid for the target company based on their private value 16

from acquiring the target and the competition from other bidders. Bidders can either 17

be strategic acquirers (e.g., competitors) or financial sponsors (e.g., private equity funds), 18

and their values are drawn from a distribution specific to their bidder type. We make 19

this distinction because prior research observes these two types of acquirers often prefer 20

targets with different characteristics (e.g., Gorbenko and Malenko, 2014; Gorbenko, 2019). 21

We assume a bidder will bid such that the bidder receives a positive surplus from 22

13
acquiring the target. Thus, bidder 𝑖 will only acquire target 𝑗 if the expected benefits, 1

𝑣 𝑖,𝑗 , from acquiring the target exceed the cost of the acquisition, 𝑏 𝑖,𝑗 , i.e., 𝑣 𝑖,𝑗 − 𝑏 𝑖,𝑗 ≥ 0. 2

How close the bid, 𝑏 𝑖,𝑗 , is to 𝑖’s value, 𝑣 𝑖,𝑗 , is in part determined by the number of bidders 3

because bidder 𝑖 does not allow another to acquire the target with a bid that is less than 𝑣 𝑖,𝑗 . 4

Consequently, the winning bidder will receive a smaller surplus because as the number 5

of bidders increases, 𝑏 𝑖,𝑗 → 𝑣 𝑖,𝑗 . Thus, when there are more bidders, potential acquirers 6

must offer more competitive bids to win. 7

4.1 Accounting Impact 8

Each bidder’s value is based on a combination of the discounted stream of expected cash 9

flows and earnings. As a result, accounting influences how much bidders value the target 10

and what they are willing to pay (Baiman, Fischer, Rajan and Saouma, 2007; Marinovic, 11

2017). In particular, bidders must follow the accounting standard, 𝑎, for allocating their 12

purchase price across the target’s assets. This standard determines whether the purchase- 13

price allocation results in future expenses, such as amortization of goodwill. Thus, 𝑎 14

directly affects a bidder’s expected surplus. 15

Let 𝑆 be the surplus bidder 𝑖 receives from winning the auction and acquiring the 16

target. We assume that the value derived from acquiring target 𝑗 is comprised of the 17

discounted stream of expected cash flows and earnings. Discounted expected cash flows, 18

𝑑𝑡 , matter to the bidder because they are the fundamental value the acquisition generates. 19

The bidder, who is the manager of the acquiring company, also cares about earnings 20

because, similar to Stein (1989), she is myopic and wants a high stock price, which is a 21

linear function of the current period’s earnings, 𝑒𝑡 .16 22

16 We suppress 𝑖 and 𝑗 subscripts on earnings and cash flows for readability.

14
 We do not specify the compensation contract that induces the manager to myopically 1

incorporate stock price into her surplus, nor do we solve for a rational-expectations equi- 2

librium. Incorporating these features would add significant complexity to the model that 3

moves beyond the objectives of this paper. Instead, we follow prior literature that esti- 4

mates models of managerial decisions and assume price is a fixed multiple of earnings 5

(e.g., Zakolyukina, 2018).17 6

Although the bidder is myopic, she is also forward-looking and knows in each sub-

sequent period, she will also want to have a high stock price and thus, high earnings.

Therefore, the bidder’s surplus is

∞
Õ ∞
Õ
𝑆 𝑖,𝑗 = E[𝑑𝑡 ]𝛿 𝑡 + 𝜋 𝑖 E[𝑒𝑡 ]𝛿 𝑡 , (1)
𝑡=0 𝑡=0

where 𝛿 is the discount factor, the terms E[𝑑𝑡 ] and E[𝑒𝑡 ] are the expected cash flows and 7

earnings for time 𝑡, and 𝜋 𝑖 is the product of the earnings multiple to determine stock price 8

and the importance of stock price relative to cash flows. 9

The initial cash outlay to acquire the target reduces cash flow by 𝑏 𝑖,𝑗 in period 0, so 10

𝑑0 = −𝑏 𝑖,𝑗 . However, this cash outflow does not impact earnings because cash used 11

for investing (such as for acquisitions) does not affect the income statement. Therefore, 12

𝑒0 = 0.18 13

Because 𝜋 𝑖 is unbounded, we re-scale Equation 1 so that the relative weight on earnings

𝜋𝑖
is bounded between 0 and 1, by defining the weight, 𝑤, as 𝑤 ≡ 1+𝜋 𝑖 and suppress the index

on 𝑤 for ease of notation. We assume that 𝑤 is drawn from the distribution ℎ(𝜃), which is
17 The preference for short-term earnings can also arise from empire-building benefits because managers
can increase their compensation by making acquisitions that increase their earnings (Jensen and Meckling,
1976; Jensen, 1986; Morck, Shleifer and Vishny, 1990).
18 Although price is a linear function of earnings, having 𝑒 = 0 does not imply stock price is zero.
0
Presumably, the acquiring firm has other earnings that are distinct from the acquisition.

15
governed by the parameter 𝜃, because different bidders may overweight or underweight

earnings relative to cash flows, and the earnings multiple can also differ across firms.

Separating the cash outflow from the discounted stream of expected cash flows and using

the relative weight, 𝑤, we recharacterize the bidder’s surplus as

∞
! ∞
Õ Õ
𝑆 𝑖,𝑗 = (1 − 𝑤) E[𝑑𝑡 ]𝛿 𝑡 − 𝑏 𝑖,𝑗 + 𝑤 E[𝑒𝑡 ]𝛿 𝑡 . (2)
𝑡=1 𝑡=1

Cash flows differ from earnings by the non-cash cash charges that result from the

acquisition. Therefore, we disaggregate earnings, 𝑒𝑡 , into cash flows, 𝑑𝑡 , and the fraction

of the purchase price recognized as an expense in period 𝑡 under the accounting regime 𝑎,

𝛼 𝑡𝑎 . For example, if the firm must amortize the purchase price over ten years, then 𝛼 𝑡 = 0.1

for 𝑡 ∈ {1, ..., 10} and 0 afterward. Thus,

𝑒𝑡 = 𝑑𝑡 − 𝛼 𝑡𝑎 𝑏 𝑖,𝑗 . (3)

The fraction of the purchase allocated to the target’s net assets with a finite life will

lead to future amortization expenses. For net assets with an indefinite life and goodwill,

as 𝑡 → ∞, there will eventually be a negative shock sufficiently large such that these assets

𝛼 𝑡𝑎 = 1. Using this identity and Equation 3,

Í∞
will require an impairment. Therefore, 𝑡=0

we can rewrite Equation 2

∞
Õ ∞
Õ
𝑡
𝑆 𝑖,𝑗 (𝑏 𝑖,𝑗 ; 𝑠) = E[𝑑𝑡 ]𝛿 −(1 − 𝑤)𝑏 𝑖,𝑗 − 𝑤𝑏 𝑖,𝑗 𝛼 𝑡𝑎 𝛿 𝑡 (4)
𝑡=1 𝑡=1
| {z }
≡𝑣 𝑖,𝑗
h ∞
Õ i
= 𝑣 𝑖,𝑗 − 𝑏 𝑖,𝑗 1 − 𝑤 + 𝑤 𝛼 𝑡𝑎 𝛿 𝑡 .
𝑡=1

Bidder 𝑖 allocates a fraction 𝑔𝑖,𝑗 of the total purchase price to goodwill and the remaining

1 − 𝑔𝑖,𝑗 to net identifiable assets. Therefore, we disaggregate the earnings impact of the

16
bid 𝑏 𝑖,𝑗 into its portion of goodwill, 𝑔𝑖,𝑗 , and net identifiable assets 𝑛 𝑖,𝑗 = 1 − 𝑔𝑖,𝑗 . Given

that the purchase price is divided between identifiable assets and goodwill, we similarly

disaggregate the amortization schedule into the effects specific to identifiable assets, 𝛼 𝑡𝑎,𝑖𝑑 ,
𝑎,𝑔𝑤
and goodwill, 𝛼 𝑡 . Thus, we can recharacterize the impact of the bid on the manager’s

utility as
∞ ∞ 
𝑎,𝑔𝑤
h i h  i
𝛼 𝑡𝑎,𝑖𝑑 𝑛 𝑖,𝑗
Õ Õ
𝑏 𝑖,𝑗 (1 − 𝑤) + 𝑤 𝛼 𝑡𝑎 𝛿 𝑡 = 𝑏 𝑖,𝑗 1 − 𝑤 + 𝑤 + 𝛼 𝑡 𝑔𝑖,𝑗 𝛿 𝑡
(5)
𝑡=1 𝑡=1
∞
𝑎,𝑔𝑤
h   i
𝛼 𝑡𝑎,𝑖𝑑 𝑛 𝑖,𝑗
Õ
𝑡
= 𝑏 𝑖,𝑗 1 − 𝑤 + 𝑤 + 𝛼 𝑡 𝑔𝑖,𝑗 𝛿 ,
𝑡=1

and rewrite the manager’s surplus from Equation 4 as

∞ 
𝑎,𝑔𝑤

𝛼 𝑡𝑎,𝑖𝑑 𝑛 𝑖,𝑗 + 𝛼 𝑡
Õ
𝑆(𝑏 𝑖,𝑗 ; 𝑠) = 𝑣 𝑖,𝑗 − 𝑏 𝑖,𝑗 1 − 𝑤 + 𝑤 𝑔𝑖,𝑗 𝛿 𝑡
 
(6)
𝑡=1

𝑎
= 𝑣 𝑖,𝑗 − 𝑏 𝑖,𝑗 𝐴 𝑖,𝑗 .

𝑎
The term 𝐴 𝑖,𝑗 is the proportion of the bid price that bidders internalize in their utilities. 1

𝑎
For a cash-focused bidder, 𝑤 = 0, which implies 𝐴 𝑖,𝑗 = 1 because this bidder cares about 2

the initial cash outlay to acquire the target. For bidders who care about earnings, 𝑤 > 0, 3

𝑎
which implies 𝐴 𝑖,𝑗 < 1. These bidders, who are myopic, impound a smaller amount of 4

the bid into their utilities because the bidder incurs the acquisition cost during future- 5

period amortization and impairment expenses, which are discounted to the present by 6

the discount factor 𝛿.19 7

19 For example, consider the case where the identifiable assets in an acquisition are depreciated or
amortized over the next 𝑇 periods while goodwill is indefinitely lived and subject to annual im-
𝑎,𝑔𝑤
pairment testing. Hence, 𝛼 𝑡𝑎,𝑖𝑑 = 𝑇1 , 𝑡 ∈ 1, . . . , 𝑇 and zero otherwise, while 𝛼 𝑡 is generated
𝑎
stochastically. Assuming risk-neutrality, bidder 𝑖 with a given 𝑤 would therefore internalize 𝐴 𝑖,𝑗 =
𝑇) 𝑎,𝑔𝑤 𝑡
h Í i
1 − 𝑤 1 − 𝑛 𝑖,𝑗 𝛿(1−𝛿
1−𝛿 − 𝑔𝑖,𝑗 E ∞
𝑡=0 𝛼˜ 𝑡 𝛿

17
4.2 The bidder’s problem 1

With this setup, we can now write the bidder’s problem. The bidder 𝑖 chooses a bid 𝑏 𝑖,𝑗 2

for target 𝑗 to maximize her expected surplus, multiplied by the probability of winning 3

the takeover auction, 4

𝑎
max(𝑣 𝑖,𝑗 − 𝐴 𝑖,𝑗 𝑏 𝑖,𝑗 )𝑃𝑟(𝑏 𝑖,𝑗 ≥ 𝑏 𝑘,𝑗 ∀𝑘 ≠ 𝑖). (7)
𝑏 𝑖,𝑗

To characterize this problem in the usual auction framework, we reformulate the bidder’s 5

problem as 6
!
𝑎
𝑣 𝑖,𝑗
max 𝐴 𝑖,𝑗 𝑎 − 𝑏 𝑖,𝑗 𝑃𝑟(𝑏 𝑖,𝑗 ≥ 𝑏 𝑘,𝑗 ∀𝑘 ≠ 𝑖), (8)
𝑏 𝑖,𝑗 𝐴 𝑖,𝑗

𝑣 𝑖,𝑗
and define the pseudo-value of bidder 𝑖 for target 𝑗 as 𝑣˜ 𝑖,𝑗 = 𝐴(𝑠)
. With this adjustment, 7

the bidder’s problem is equivalent to 8

max 𝑣˜ 𝑖,𝑗 − 𝑏 𝑖,𝑗 𝑃𝑟(𝑏 𝑖,𝑗 ≥ 𝑏 𝑘,𝑗 ∀𝑘 ≠ 𝑖).

(9)
𝑏 𝑖,𝑗

5. Empirical Strategy 9

This section describes how we identify and estimate the primitives of the auction model. 10

To do so, we need to estimate the parameters governing the distribution of underlying 11

valuations 𝑣 𝑖,𝑗 , including bidders’ earnings preferences relative to cash flows. 12

5.1 Identification 13

The unstructured nature of takeovers does not allow us to impose a standard auction 14

format to identify the distribution bidder valuations. Instead, we follow an approach 15

developed by Haile and Tamer (2003) and adjusted by Gorbenko and Malenko (2014) to 16

estimate the distribution of bidder valuations in non-standard auctions. In particular, 17

18
identification comes from three assumptions: 1

1. A bidder does not submit a bid that would leave her with a negative surplus, i.e., 2

bid more than her value for the target 3

2. A bidder does not allow a bid to win that she could beat with a non-negative surplus 4

3. A bidder does not make informal, noncommittal bids if her pseudo-valuation is 5

below the value of the target as a standalone company, i.e., the target’s market value 6

These assumptions provide five non-parametric restrictions that help identify the dis- 7

𝑣 𝑖,𝑗
tribution of bidder pseudo-values, 𝑎 .
𝐴 𝑖,𝑗
To see how these restrictions aid in identifying 8

the distribution, suppose bids for target 𝑗 are sorted in decreasing order so that 𝑏 1,𝑗 is the 9

winning bid. Further, suppose the distribution of pseudo-values is represented by the 10

black line in Panel A, Figure 1, and the dashed vertical line is the winning bid, then: 11

𝑣 1,𝑗
• A winning bid implies 𝑏1,𝑗 < 𝑎 .
𝐴1,𝑗
Thus, this bidder’s pseudo-value must be in the 12

shaded region in Panel B of Figure 1. 13

𝑣 𝑖,𝑗
• A formal losing bid from bidder 𝑖 implies 𝑏 𝑖,𝑗 < 𝑎
𝐴 𝑖,𝑗
< 𝑏 1,𝑗 . Thus, this bidder’s 14

pseudo-value must be in the shaded region in Panel C. 15

𝑣 𝑖,𝑗
• An informal losing bid from bidder 𝑖 implies 1 < 𝑎
𝐴 𝑖,𝑗
< 𝑏1,𝑗 . Thus, this bidder’s 16

pseudo-value must lie within the shaded region in Panel D. 17

• If a bidder makes neither a formal nor an informal bid, i.e., this bidder declines to 18

𝑣 𝑖,𝑗
bid, then 0 < 𝑎
𝐴 𝑖,𝑗
< 𝑏 1,𝑗 . Thus, this bidder’s pseudo-value is in the shaded region of 19

Panel E. 20

𝑣 𝑖,𝑗
• If a bidder states that their value is below the market value, then 0 < 𝑎
𝐴 𝑖,𝑗
< 1. Thus, 21

this bidder’s pseudo-value is in the shaded region in Panel F. 22

19
By knowing the regions in the distribution where pseudo-values are located and incor- 1

porating additional parameterization assumptions (discussed below), we can trace the 2

distribution of pseudo-values. 3

5.2 Parameterization Assumptions 4

On their own, the assumptions of the previous section only provide set-identification

(Haile and Tamer, 2003). Therefore, we need to impose parametric assumptions of bidders’

valuations to achieve point identification. We follow Gorbenko and Malenko (2014) and

assume that the bidders’ values follow a log-normal distribution with a common and an

idiosyncratic component:

𝑣 𝑖,𝑗 = exp{𝑋𝑖,𝑗 𝛽 𝑖 } exp{𝜖 𝑖,𝑗 } (10)

with 𝜖 ∼ 𝑁(0, 𝜎𝑖 ). The vector 𝑋𝑖,𝑗 corresponds to the observable bidder, target, and 5

time characteristics, representing the target’s common value component. The term 𝜖 𝑖,𝑗 6

corresponds to the idiosyncratic component of value and represents bidder 𝑖’s preferences 7

for target 𝑖 that are unobservable to the econometrician. 8

We assume the parameters that determine 𝑣 𝑖,𝑗 , 𝛽 𝑖 , and 𝜎𝑖 , are the same for bidders 9

of type 𝑧 ∈ {𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙, 𝑆𝑡𝑟𝑎𝑡𝑒 𝑔𝑖𝑐}.20 Thus, 𝛽 𝑖 = 𝛽 𝑧 and 𝜎𝑖 = 𝜎𝑧 . We assume that 10

financial bidders only care about cash flow, so if 𝑖 is a financial bidder, 𝑤 𝑖 = 0 and 11

𝐴 𝑖,𝑗 = 1. Strategic bidders have some preference for future earnings, so if 𝑖 is a strategic 12

bidder, 𝑤 𝑖 ≥ 0 and 𝐴 𝑖,𝑗 ≤ 1. However, we do not assume all strategic bidders are alike 13

in their preferences for earnings relative to cash flows. We assume strategic bidders 14

20 SEC filings that describe the merger history typically only distinguish financial and strategic bidders.
We can only infer whether a bidder was a public or private firm for winning bids, which is not enough
variation to estimate a separate valuation distribution.

20
have heterogeneous preferences because strategic bidders contain both private and public 1

bidders who are known to value earnings differently.21 Furthermore, even among public 2

acquirers, preferences for future earnings vary in the cross-section and time-series, such 3

as for management compensation reasons (Healy, 1985). We therefore assume that the 4

accounting preference of strategic bidder 𝑖 is drawn from a beta distribution with shape 5

parameters 𝑎 = 1 and 𝑏 = 𝜃. Thus, 𝑤 ∼ ℎ(𝜃). 6

The beta distribution offers two desirable properties that make it well-suited as a 7

parametric assumption from which 𝑤 𝑖 is drawn.22 First, beta distributions have support 8

over the interval [0, 1], which ensures that a bidder’s weight on earnings relative to 9

cash flows does not exceed 1 or is negative. Second, the beta distribution is flexible, 10

and depending on the shape parameters, the resulting distribution can be unimodal, U- 11

shaped, left or right-skewed. We assume the distribution is right skewed to account for 12

the fact that some strategic bidders, such as private firms, are primarily focused on cash 13

flows. We ensure the distribution of 𝑤 𝑖 is right skewed by requiring 𝜃 > 1. Doing so 14

allows a sizeable fraction of strategic bidders to have 𝑤 𝑖 near zero, which implies they 15

strongly prefer cash flows over earnings. For instance, this constraint accommodates the 16

possibility that private strategic bidders may not have strong preferences for earnings. 17

5.3 Estimation 18

Bids are a function of bidder, target, and time characteristics and the structural param- 19

eters 𝛽 𝑖 and 𝜎𝑖 . We estimate the structural parameters via simulated maximum likelihood 20

21 We assume private and public strategic bidders bid as if they must treat goodwill under an impairment-

only regime. This ignores the alternate accounting treatment for private firms under ASU-2014-02, where
private firms can elect to amortize goodwill on a straight-line basis over ten years.
22 For other examples that use the beta distribution in structural models in accounting, see Huber (2022)

and McClure (2023).

21
estimation (SMLE). Briefly, SMLE is used when estimating parameters by maximum like- 1

lihood estimation is infeasible because there is no closed-form solution for the likelihood 2

function. SMLE simulates a large number (in our case, 500) of simulated observations for 3

a guess of parameter values, computes the likelihood function, and compares it to the ob- 4

served data. SMLE iterates these steps by changing parameter values until the simulated 5

likelihood converges. See Cameron and Trivedi (2005) for details on the method. 6

Using the distributional assumptions from Section 5.2 and the five restrictions from

Section 5.1, every observed bid maps into a well-defined contribution to the likelihood. For

example, assumption 1 implies that the winning bidder needs to have a pseudo-valuation

weakly greater than her winning bid. Let 𝐿 𝑧 (𝑏 𝑖,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) be the likelihood contribution

from bidder 𝑖 of type 𝑧, which is conditional on the characteristics, 𝑥, structural parameters

𝛽 𝑧 and 𝜎𝑧 , and preference for earnings 𝑤, then the likelihood for the winning bidder, 𝑖 = 1

is
𝑎
log(𝑏 1,𝑗 ) − 𝑋1,𝑗 𝛽 𝑧 + log(𝐴1,𝑗
!
(𝑤))
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 1 − Φ ,
𝜎𝑧

where Φ(·) is the cdf of the standard normal distribution. Hence, this likelihood corre- 7

sponds to the area under the upper tail of the pseudo-value distribution.23 8

To arrive at the unconditional likelihood contribution, note that for financial bidders, 9

𝐴 𝑖,𝑗 = 1 because 𝑤 = 0. The likelihood contributions of all other bids can be constructed 10

similarly and are reported in the appendix. Note that for strategic bidders, because 𝑤 > 0 11

is drawn from the distribution ℎ(𝜃) and we do not observe it directly, we need to take the 12

23 All remaining conditional likelihoods are derived in the appendix.

22
integral over the support of ℎ(𝜃). Accordingly, the likelihood contribution becomes: 1

∫ 1
𝑠𝑡𝑟 𝑎𝑡𝑒 𝑔𝑖𝑐
𝐿 (𝑏 𝑖,𝑗 ; ·) = 𝐿 𝑧 (𝑏 𝑖,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤)ℎ(𝜃) d𝑤. (11)
0

To estimate the parameters, we minimize the following log-likelihood: 2

!
ÕÕ Ö
𝑧 𝑧
𝑝 𝑖,𝑗 𝐿 𝑏 𝑖,𝑗 ,


min log (12)
𝛽,𝜎,𝜃
𝑗 𝑖 𝑧

with 𝑝 𝑧 being the probability of bid 𝑏 𝑖,𝑗 being submitted by a bidder with type 𝑧. Including 3

𝑝 𝑧 allows us to include losing bids where the type of bidder is unknown. 4

Three complications arise in our estimation. First, when bidders determine the ex- 5

pected value of a target, they do not know whether goodwill will need to be impaired in 6

the future and, if so, by how much. We assume rational expectations by the firm, so the 7

firm’s belief of the likelihood of a goodwill impairment mirrors the realized distribution 8

of impairments in Table 1. For instance, in our data, the probability of a firm needing to 9

impair goodwill is 9.6%. We assume bidders correctly infer this probability when deter- 10

mining their private value for the target. Rational expectations also apply to the size of 11

the impairment, conditional on an impairment happening. 12

The stochastic nature of goodwill impairments requires us to model the evolution of 13

the target’s value in the years after it is acquired. To do so, we simulate the evolution 14

of goodwill over the next 200 fiscal years for each of our 500 simulated observations. 15

When discounting future cash flows, earnings, and the effect of goodwill’s accounting 16

treatment on the acquiring firm’s utility, we follow prior research and set the discount rate 17

𝛿 = 0.9 (e.g., Taylor, 2013; Zakolyukina, 2018). In each year, an impairment occurs with 18

the empirically observed probability. If an impairment occurs, the size of the impairment, 19

relative to the firm’s overall goodwill amount, is randomly drawn from the empirical 20

23
distribution of impairment charges. 1

The second complication occurs because we only observe purchase-price allocations 2

when the winning bidder is a public company. Hence, we have to impute the fraction of 3

the goodwill allocation, 𝑔𝑖,𝑗 , when the winning bidder is a privately-held strategic bidder 4

and for all losing bids. To approximate the goodwill allocation when the winning bid 5

is a private firm, we assume they allocate goodwill according to the mean allocation to 6

goodwill in the target firm’s Fama-French 12 industry. For losing bids, we impute the 7

minimum of the industry mean or the allocation of the winning bidder (if observable).24 8

The third complication is we do not observe the type of every losing bidder. Although

many SEC filings with background on the merger distinguish between financial and

strategic bidders for losing bids, it is not universal, especially not in the early stages of a

takeover auction. When we do not observe the bidder type, we infer the type of bidder

using a determinants model with the sample of losing bids where the type is known

following Gorbenko and Malenko (2014). Specifically, we estimate the following logistic

regression

′
𝑆𝑡𝑟𝑎𝑡𝑒 𝑔𝑖𝑐 𝑖,𝑑,𝑡 = 𝑋𝑑,𝑡 𝛽 + 𝜀𝑖,𝑑,𝑡 , (13)

where 𝑆𝑡𝑟 𝑎𝑡𝑒 𝑔𝑖𝑐 𝑏,𝑑,𝑡 is an indicator for whether bidder 𝑖 in deal 𝑑 at time 𝑡 was a strategic 9

bidder, and 𝑋𝑑,𝑡 is a vector of deal and target-firm characteristics that follows Gorbenko 10

and Malenko (2014). We use the results from this regression with industry fixed effects 11

24 When we observe a formal bid, we could have also assumed that the fair value of identifiable assets
would be the same as for the winning bid (if the winning bid is public) and imputed the goodwill allocation
as the remainder. Such an approach would still leave the problem of informal bids, which comprise most of
the bids. For consistency, we impute the goodwill allocation for all losing bids similarly. Such an assumption
is consistent with bidders making bids before having allocated the purchase price to individual assets and
therefore bid with an expectation over the percentage of goodwill rather than on how much identifiable
assets are worth. This distinction matters for whether a marginal increase of the bid is allocated 100% to
goodwill or only a fraction of it.

24
to assign probabilities a losing bidder with an unknown type is a strategic or a financial 1

bidder and, thus, estimate how much this bidder cares about the treatment of goodwill. 2

Table 4 reports the results from this regression. Column 1 shows that strategic bidders 3

are more likely to target firms with a higher Tobin’s Q and R&D and lower leverage, cash 4

flows, and credit spreads. Strategic bidders are also more likely to bid when there are 5

fewer bidders and when a financial bidder does not make the winning bid. Column 2 6

includes Fama-French 12 industry fixed effects and shows similar results. Overall, these 7

results are consistent with the findings in Gorbenko and Malenko (2014). 8

6. Results 9

6.1 Distribution of Takeover Valuations 10

We use a simulated likelihood model, as described in Section 5, to estimate the factors 11

influencing the valuation of the target by different types of bidders. Our findings are 12

summarized in Table 5. In Columns 1 and 2, we present estimates from a reduced model 13

that excludes target and market characteristics for strategic and financial bidders. Columns 14

3 and 4 display estimates from the full model for both bidder types. 15

Similar to the study by Gorbenko and Malenko (2014), we observe that strategic bidders 16

have a lower intercept than financial bidders in Columns 3 and 4. However, the average 17

valuation of strategic bidders is higher, as shown in the first two columns. In Columns 18

1 and 2, the estimates suggest that strategic bidders value a target at approximately 1.07 19

times its market valuation, not taking into account the accounting preferences, while 20

financial bidders value it at around 1.01 times.25 21

25 Note that the mean of a lognormal variable is equal to exp(𝜇 + 𝜎2 /2)

25
 The key differences between financial and strategic bidders are related to target size, 1

market-to-book ratio, and cash holdings. Financial bidders prefer smaller targets with 2

lower market-to-book ratios and higher cash amounts. Both strategic and financial bidders’ 3

valuations are influenced by current equity market conditions. Valuations tend to be 4

higher when the market return in the previous month is higher. Contrary to intuition, 5

valuations also increase with credit spreads, which can be attributed to credit spreads 6

showing little variation over the sample period but significantly increasing during the 7

financial crisis when other indicators decrease. Therefore, the coefficient on credit spread 8

becomes important for the model to explain acquisition activity during that period. 9

We find that valuations exhibit considerable dispersion, with standard deviations of 10

0.173 and 0.219 for strategic and financial bidders, respectively. Under the assumption 11

of a lognormal distribution, this implies that the standard deviation of underlying values 12

is approximately 20% and 24% of the current market value of the target for strategic and 13

financial bidders respectively.26 14

The crucial parameter that governs the distribution of acquirers’ preferences between 15

earnings and cash flows is 𝜃. The estimate of 𝜃ˆ = 2.52 suggests a mean value of approxi- 16

mately 𝐸(𝑤) = 0.28, indicating that strategic bidders, on average, assign more than twice 17

the weight to acquired cash flows compared to the acquired earnings stream. We show 18

the distribution of 𝑤 and its mean based on our estimate of 𝜃 in Figure 2. Taking into 19

account the expected value of 𝑤 and the distribution of the impairment data, we find that 20

𝐴 𝑖𝑚𝑝 = 0.85. In other words, strategic bidders’ only internalize 85% of their bids in their 21

surplus. Consequently, they can afford to be about 15% more aggressive in their bidding 22

26 Note that the variance of a log normal variable equals exp(𝜎2 ) − 1 exp(2 ∗ 𝜇 + 𝜎2 ).

26
compared to financial or private bidders, who are primarily concerned with the cash flow 1

implications of acquiring a target. 2

Figure 3 shows the distribution of bidders’ preferences for strategic and financial bid- 3

ders based on the estimates in columns 3 and 4 of Table 5. In addition to strategic bidders 4

having a higher mean value, their distribution is also wider than financial bidders, which 5

comports with Gorbenko and Malenko (2014). We remove strategic bidders’ preferences 6

for accounting and plot the distribution with the dashed line.27 It indicates that once we 7

eliminate accounting preferences, strategic bidders have a smaller dispersion than finan- 8

cial bidders. Rather than strategic bidders’ having more varied synergies, as suggested 9

by Gorbenko and Malenko (2014), the more disperse values of strategic bidders appears 10

to stem from differing accounting preferences 11

7. Counterfactuals 12

7.1 Counterfactual Simulation Procedure 13

We conduct counterfactual simulations to estimate how the M&A market would change 14

under different competitive environments and alternative accounting rules for goodwill. 15

We focus on three policy experiments. First, we consider the first-best scenario where 16

bidders care only about the cash effect from the acquisition and disregard the accounting 17

implications. Second, we consider how mergers would change under various accounting 18

regimes, including different amortization periods and amortization regimes that annually 19

test for impairment if the underlying value of goodwill falls below the amortized value. 20

Finally, we consider changing the composition of bidders would affect mergers. 21

27 Weremove accounting preferences by setting 𝑤 = 0 for strategic bidders and re-drawing values based
on Table 5.

27
 To set up our counterfactual experiments, we simulate 10,000 auctions and set the 1

characteristics equal to the average values of target and market characteristics. For each 2

auction, we randomly draw the number of bidders from its empirical distribution in the 3

data truncated at the 95% quantile (35 bidders). We randomly assign each bidder a type 4

(strategic or financial) based on the actual proportion in the data. Having assigned each 5

simulated bidder a type, we draw an idiosyncratic component of value, 𝜖 𝑖,𝑗 , and compute 6

the fundamental valuation, 𝑣 𝑖,𝑗 , for each bidder based on the estimates from Section 5. We 7

then draw an accounting preference, 𝑤, for each strategic bidder from the beta distribution 8

𝑎
with our estimated parameter 𝜃ˆ and calculate their valuation adjustments, 𝐴 𝑖,𝑗 . When a 9

bidder is financial, or it is the counterfactual where bidders only care about cash, we set 10

𝑎
the valuation adjustment equal to one, 𝐴 𝑖,𝑗 = 1. 11

With this setup, we simulate the bidding process following the approach of Gorbenko 12

and Malenko (2014). This process is an ascending auction with jump bids to reflect the 13

discrete bid increases observed in the data. In each round of bidding, a bidder is randomly 14

selected to submit a bid. If the bidder’s pseudo-value, 𝑣˜ 𝑖,𝑗 , surpasses the current highest 15

bid, the bidder updates her bid by increasing it by a random percentage of the target’s 16

market value uniformly distributed between 1% and 10%. In instances when this jump 17

exceeds the bidder’s pseudo-valuation, we set their bid equal to their pseudo-value. If 18

the prevailing bid exceeds the chosen bidder’s pseudo-value, the bidder drops out of the 19

auction. If none of the bidders’ pseudo-values exceed the current market value of the 20

target, the auction fails. 21

For each counterfactual, we examine both the valuation and aggregate outcomes of 22

mergers to better understand the impact of these hypothetical changes. For valuation 23

28
outcomes, we focus on the deal premium, the valuation of the second-highest bidder, the 1

percent of deals that fail, which are those deals where no winner has a bid greater than 2

the market value, and the probability a financial bidder (“PE”) wins the auction.28 For 3

aggregate changes, we examine the aggregate amount of deal volume and the change in 4

asset value acquired by financial bidders. 5

7.2 Accounting, Valuations, and the Role of Competition 6

Recall that our estimates imply that under an impairment-only regime, strategic ac- 7

quirers act as if they only internalize 85% of the purchase price. In other words, if they 8

only cared about the underlying cash flows, their valuations would be 15% lower. Our first 9

counterfactual, labeled Cash in Table 6, uses the benchmark where all bidders only care 10

about cash. Under this counterfactual, all bidders are optimizing the fundamental value 11

they would receive from the acquisition, so this counterfactual is the first-best. By com- 12

paring this counterfactual to our as-observed outcomes, we can quantify how preferences 13

for earnings interact with competition in determining deal outcomes. 14

In this counterfactual, where every bidder cares only about price, aggregate deal val- 15

uations decline by 9.94%, and deal premiums decrease by 13.7 percentage points. Both 16

statistics are lower than the 15% decline in strategic bidders’ valuation when they only 17

care about cash. This difference illustrates that competition from bidders less sensitive to 18

accrual accounting can partially offset the valuation change and lead to a muted response 19

in transaction prices. This result also highlights that focusing solely on public bidders’ 20

valuations and disregarding the effect of competition, as in Bartov et al. (2021), provides 21

28 The valuation of the runner-up would be the realized deal value if the takeover auction were structured
as a second-price auction instead of one with jump bidding.

29
an incomplete picture of how underlying valuations change as it ignores the impact of 1

competition. 2

We use this insight that merger outcomes result from competition and valuation effects 3

to examine the impact of goodwill accounting standards. As a first step, we compare the 4

difference in pseudo-valuations for strategic bidders under impairment and amortization- 5

only regimes with varying amortization periods. Panels (a) and (b) of Figure 4 show the 6

expected percent changes in strategic bidders’ valuations across different amortization 7

regimes and goodwill allocations relative to the current impairment-only regime. Fo- 8

cusing on the 10-year amortization horizon with impairments, which corresponds to the 9

option currently available to private companies, we find that valuations are approximately 10

6% lower compared to the impairment-only regime for a 40% allocation of goodwill, which 11

closely aligns with the sample average. 12

This figure shows that as the length of the amortization period increases, the val- 13

uations under impairment and amortization converge, resulting in smaller percent dif- 14

ferences. The alternative of amortization with impairment is strictly more conservative 15

than an impairment-only standard, so the pseudo-valuations never fully converge. For 16

amortization-only regimes, we find an inflection point at the 35-year amortization period. 17

Furthermore, considering the underlying distribution of impairment charges, we find that 18

the expected annual impairment charge of 3% of goodwill closely aligns with the annual 19

amortization charge for a 33-year amortization policy. 20

Consistent with intuition, valuation differences increase as the proportion of deal 21

value allocated to goodwill in an acquisition increases. Additionally, we observe that the 22

sensitivity of pseudo-valuation differences to the amortization horizon intensifies with 23

30
higher levels of recognized goodwill in the acquisition. 1

7.3 The Effect of Accounting Standards on Merger Outcomes 2

Having shown how strategic bidders’ expected valuations relate to alternative account- 3

ing treatments of goodwill, we next examine accounting’s effect on merger outcomes. The 4

results from our counterfactual simulations are summarized in Table 6. The first row 5

reports the simulation results that mirror the observed M&A market: goodwill is only 6

tested for impairment and the composition of bidders mimics what we observe in the 7

data. 8

We compare the as-observed results with two amortization standards that each have 9

varying amortization periods. The block of rows labeled Amortization with Impairment 10

considers an amortization standard with annual impairment testing of goodwill, while 11

the last block of rows (labeled Amortization) considers an amortization-only standard. 12

When discussing the results, we refer to the Amortization with Impairment regime over 10 13

years as our main counterfactual because it corresponds to ASC 350-20-35, which permits 14

private firms to amortize goodwill over a period no longer than 10 years and requires 15

the firm to impair assets should they need to. Moving from an impairment regime to a 16

counterfactual regime of amortization with impairments, we estimate a decrease in deal 17

premium of 5.98 percentage points. This decline reflects strategic bidders’ valuations 18

declining because they must, at a minimum, amortize the cost equally over 10 years 19

instead of delaying recognition until impairment. Consequently, we observe runner-up 20

valuations decline by about 5.6 percentage points and a 10% increase in the likelihood of 21

deals failing. 22

31
 Considering both the reduction in valuations and the increase in failed auctions, our 1

counterfactual results suggest a significant 4.29% reduction in total deal value. To put this 2

decline into perspective, this reduction would amount to approximately $68.6 billion in 3

total M&A deal value for 2021 (4.29% times $1.6 tn). 4

Furthermore, transitioning to an amortization regime would not only affect deal val- 5

uations but also change who controls productive assets in the economy. In particular, 6

strategic acquirers bid less aggressively because their valuations are reduced. Conse- 7

quently, there would be a decrease in the fraction of auctions won by strategic bidders 8

as financial bidders are approximately 7 percentage points, or 25%, more likely to win a 9

takeover auction. As a result, we estimate that the value of assets acquired by financial 10

bidders would increase by 20.73%. When we switch to an amortization-only regime, 11

we find similar but slightly attenuated results. Overall, Table 6 shows accounting rules 12

substantially impact the merger market. 13

7.4 Industry-specific Outcomes 14

The counterfactual analyses presented so far have been based on average target char- 15

acteristics and goodwill allocations, overlooking the substantial heterogeneity observed 16

across different industries as shown in Tables 1 and 2. To address this limitation, we 17

conduct separate counterfactual simulations for each of the 12 Fama-French industries, 18

accounting for their specific characteristics. In these industry-specific counterfactuals, 19

we simulate target valuations by applying the average target characteristics and goodwill 20

allocation corresponding to each industry. 21

Table 7 summarizes the results for two industries: Business Equipment and Oil and 22

32
Gas. We report these two industries because Business Equipment has the highest average 1

goodwill allocation, whereas Oil and Gas has the lowest.29 Generally, the impact of 2

transitioning to a 10-year amortization policy is positively correlated with the amount of 3

goodwill allocation in each industry. For instance, in the Business Equipment industry, 4

where 50.1% of the purchase price is allocated to goodwill, the decrease in aggregate deal 5

values under the amortization policy is more pronounced, with a reduction of 5.00%. By 6

contrast, for the Oil and Gas industry, where only 23.7% of the purchase price is allocated 7

to goodwill, the decrease in aggregate deal values is smaller, at 2.14%. 8

The industry-specific counterfactuals also emphasize how variation in goodwill allo- 9

cations can affect the competitive dynamics among potential acquirers, which can either 10

mitigate or amplify the effects of accounting on valuations. This effect becomes apparent 11

when comparing the fraction of winning bidders classified as financial. In our data, Oil 12

and Gas companies have the highest proportion of private equity winners. But when an 13

amortization regime is imposed, the probability of a financial bidder winning an auction 14

increases more for Business Equipment targets, nearly equalizing the probability. This 15

result implies that the accounting treatment for goodwill has a pronounced impact on 16

deals where the goodwill allocation is the highest. 17

7.5 The role of the competitive environment 18

The previous counterfactuals assume the only change occurs with strategic bidder 19

valuations, while the competitive environment remains unaffected. However, accounting 20

rules impact strategic and financial bidders differently, potentially altering the composition 21

of the bidder pool. We therefore explore the potential impact of the type of bidders 22

29 Due to space limitations, results for other industries are omitted but are available upon request.

33
competing for the target on our analysis of alternative accounting rules. We consider 1

counterfactuals where we increase the prevalence of financial bidders because takeovers 2

become less attractive for strategic bidders when they must amortize goodwill.30 3

We consider bidder composition effects by comparing three competitive environments: 4

first, the competitive environment as observed in the data; second, we add an additional 5

financial bidder; third, we keep the same number of bidders but increase the proportion of 6

financial bidders, effectively changing a strategic bidder to a financial one. For each of the 7

three competition scenarios, we report summary statistics for three different accounting 8

standards: an impairment-only regime, a 10-year amortization schedule, and a 10-year 9

amortization schedule with impairments. We compare the two counterfactuals to the 10

as-observed impairment result. 11

In the first set of rows, we report summary statistics for counterfactuals using the 12

observed competition levels. Comparing these results to the second set of rows, which are 13

the counterfactuals where we examine the extensive margin of financial bidders by adding 14

one financial bidder, we observe that the counterfactual effects of an amortization regime 15

are muted for deal valuation outcomes. This muted effect occurs because the increased 16

competition from an additional bidder improves the payoff for the seller (Bulow and 17

Klemperer, 1996). For example, we find the deal premium under 10-year amortization with 18

impairment to be 2 percentage points higher than under the as-observed counterfactual 19

with amortization and impairment. Similarly, because the number of failed deals also 20

decreases, the effects on aggregate deal values are less than half than in the as-observed 21

30 Achanging competitive environment could be micro-founded by having a cost to enter an auction, e.g.,
such as costly due diligence (Gentry and Stroup, 2019). With lower valuations of strategic bidders, more
financial bidders may find it attractive to enter, whereas some strategic bidders might choose not to enter
the auction in the first place.

34
benchmark. 1

The last set of rows reports the counterfactuals where we change the intensive margin 2

of financial bidders. We do so by increasing the proportion of financial bidders by 8 3

percentage points, effectively replacing one strategic bidder with a financial bidder. This 4

change decreases the effective level of competition because a high-value bidder is replaced 5

by a lower-value bidder, in expectation. Hence, we observe that bid premia are lower, 6

declining by an additional 3% for the impairment standard, and financial bidders are more 7

likely to win. Consequently, the effects of requiring goodwill amortization are amplified 8

by changing the competitive environment. Overall, Table 9 shows that competition and 9

accounting standards meaningfully interact in determining merger outcomes. 10

8. Conclusion 11

This paper examines the impact of the accounting treatment of goodwill on the market 12

for corporate control. In order to disentangle the accounting effects from the competitive 13

effects faced by bidders, we develop and estimate a structural model of rational bidding. 14

The model assumes the bidders’ valuations of the target are a function of the target’s 15

characteristics. We assume strategic bidders can be sensitive to goodwill accounting, 16

if they want to maximize a combination of earnings and cash flows, whereas financial 17

bidders only maximize cash flows. 18

Our counterfactual analyses suggest strategic bidders’ preference for earnings substan- 19

tially boosts average deal premia and deal volume. This preference implies that changing 20

the accounting for goodwill would influence merger activity. We estimate that moving 21

from an impairment-only regime to an amortization regime for goodwill—as recently pro- 22

35
posed by standard setters—will decrease the target valuations of public acquirers because 1

amortization expenses reduce future earnings. Our results suggest these hypothetical 2

changes would reduce the bid premia and deal volume while shifting more assets to 3

financial bidders. As a result, these findings provide insights into the real effect of ac- 4

counting in the M&A market. We believe these conclusions may interest standard setters 5

as they continue to debate whether to modify the accounting for intangible assets. 6

36
References 1

Appleton, A., Barckow, A., Botosan, C., Kawanishi, Y., Kogasaka, A., Lennard, A., Mezon- 2

Hutter, L., Sy, J., Villmann, R., 2022. Perspectives on the financial reporting of intangibles. 3

Accounting Horizons 37, 1–13. 4

Baiman, S., Fischer, P., Rajan, M.V., Saouma, R., 2007. Resource allocation auctions within 5

firms. Journal of Accounting Research 45, 915–946. 6

Bartov, E., Cheng, C.A., Wu, H., 2021. Overbidding in mergers and acquisitions: An 7

accounting perspective. The Accounting Review 96, 55–79. 8

Beatty, A., Weber, J., 2006. Accounting discretion in fair value estimates: An examination 9

of sfas 142 goodwill impairments. Journal of Accounting Research 44, 257–288. 10

Bens, D.A., Nagar, V., Skinner, D.J., Wong, M.F., 2003. Employee stock options, EPS 11

dilution, and stock repurchases. Journal of Accounting and Economics 36, 51–90. 12

Betton, S., Eckbo, B.E., Thorburn, K.S., 2008. Corporate takeovers, in: Handbook of 13

Empirical Corporate Finance. Elsevier, San Diego. Handbooks in Finance, pp. 291–429. 14

Boone, A.L., Mulherin, J.H., 2007. How are firms sold? Journal of Finance 62, 847–875. 15

Bulow, J., Klemperer, P., 2009. Why do sellers (usually) prefer auctions? American 16

Econonmic Review 99, 1544–1575. 17

Bulow, J.I., Klemperer, P.D., 1996. Auctions versus negotiations. American Economic 18

Review 86, 180–194. 19

Cameron, A.C., Trivedi, P.K., 2005. Microeconometrics: methods and applications. Cam- 20

bridge University Press. 21

Darrough, M.N., Guler, L., Wang, P., 2014. Goodwill impairment losses and ceo compen- 22

sation. Journal of Accounting, Auditing & Finance 29, 435–463. 23

Dimopoulos, T., Sacchetto, S., 2014. Preemptive bidding, target resistance, and takeover 24

premiums. Journal of Financial Economics 114, 444–470. 25

Eckbo, B.E., 2009. Bidding strategies and takeover premiums: A review. Journal of 26

Corporate Finance 15, 149–178. 27

Geng, H., Zhang, C., Zhou, F., 2023. Financial reporting quality and myopic investments: 28

Theory and evidence. The Accounting Review 98, 223–251. 29

Gentry, M., Stroup, C., 2019. Entry and competition in takeover auctions. Journal of 30

Financial Economics 132, 298–324. 31

Glaum, M., Landsman, W.R., Wyrwa, S., 2018. Goodwill impairment: The effects of 32

public enforcement and monitoring by institutional investors. The Accounting Review 33

93, 149–180. 34

37
Gorbenko, A.S., 2019. How do valuations impact outcomes of asset sales with heteroge- 1

neous bidders? Journal of Financial Economics 131, 88–117. 2

Gorbenko, A.S., Malenko, A., 2014. Strategic and financial bidders in takeover auctions. 3

Journal of Finance 69, 2513–2555. 4

Graham, J.R., Harvey, C.R., Rajgopal, S., 2005. The economic implications of corporate 5

financial reporting. Journal of Accounting and Economics 40, 3–73. 6

Haile, P.A., Tamer, E., 2003. Inference with an incomplete model of English auctions. 7

Journal of Political Economy 111, 1–51. 8

Hansen, R.G., 2001. Auctions of companies. Economic Inquiry 39, 30–43. 9

Healy, P.M., 1985. The effect of bonus schemes on accounting decisions. Journal of 10

Accounting and Economics 7, 85–107. 11

Huber, S., 2022. Loan loss measurement and bank lending. Available at SSRN 4120628 . 12

Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance, and takeovers. The 13

American economic review 76, 323–329. 14

Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agency 15

costs and ownership structure. Journal of Financial Economics 3, 305–360. doi:https: 16

//doi.org/10.1016/0304-405X(76)90026-X. 17

Kanodia, C., Sapra, H., 2016. A real effects perspective to accounting measurement 18

and disclosure: Implications and insights for future research. Journal of Accounting 19

Research 54, 623–676. 20

Kanodia, C., Sapra, H., Venugopalan, R., 2004. Should intangibles be measured: What are 21

the economic trade-offs? Journal of Accounting Research 42, 89–120. 22

Kanodia, C., Singh, R., Spero, A.E., 2005. Imprecision in accounting measurement: Can it 23

be value enhancing? Journal of Accounting Research 43, 487–519. 24

Li, K.K., Sloan, R.G., 2017. Has goodwill accounting gone bad? Review of Accounting 25

Studies 22, 964–1003. 26

Lugo, D., 2022. In a surprising move, FASB drops project on subsequent accounting of 27

goodwill. Associated Press . 28

Marinovic, I., 2017. Delegated bidding and the allocative effect of accounting rules. 29

Management Science. 63, 2181–2196. 30

Marquardt, C., Zur, E., 2015. The role of accounting quality in the M&A market. Manage- 31

ment Science 61, 604–623. 32

Matsunaga, S., Shevlin, T., Shores, D., 1992. Disqualifying dispositions of incentive stock 33

options: Tax benefits versus financial reporting costs. Journal of Accounting Research 34

30, 37–68. 35

38
Maurer, M., 2022. FASB scraps project on goodwill accounting, disclosure. Wall Street 1

Journal . 2

McClure, C., 2023. How costly is tax avoidance? Evidence from structural estimation. The 3

Accounting Review 98, 353–380. 4

McClure, C., Zakolyukina, A.A., 2023. Non-gaap reporting and investment. The Account- 5

ing Review . Forthcoming. 6

McNichols, M.F., Stubben, S.R., 2015. The effect of target-firm accounting quality on 7

valuation in acquisitions. Review of Accounting Studies 20, 110–140. 8

Morck, R., Shleifer, A., Vishny, R.W., 1990. Do managerial objectives drive bad acquisitions? 9

The journal of finance 45, 31–48. 10

Potepa, J., Thomas, J., 2023. Goodwill impairment after m&a: Acquisition-level evidence. 11

Journal of Financial Reporting , 1–25. 12

Ramanna, K., 2008. The implications of unverifiable fair-value accounting: Evidence from 13

the political economy of goodwill accounting. Journal of Accounting and Economics 14

45, 253–281. 15

Ramanna, K., Watts, R.L., 2012. Evidence on the use of unverifiable estimates in required 16

goodwill impairment. Review of Accounting Studies 17, 749–780. 17

Rayburn, F.R., Powers, O.S., 1991. A history of pooling of interests accounting for business 18

combinations in the united states. Accounting Historians Journal 18, 155–192. 19

Robinson, J.R., Shane, P.B., 1990. Acquisition accounting method and bid premia for target 20

firms. The Accounting Review 65, 25–48. 21

Seligman, J., 1982. The transformation of Wall Street: a history of the Securities and 22

Exchange Commission and modern corporate finance. Boston: Houghton Mifflin. 23

Shalev, R., Zhang, I.X., Zhang, Y., 2013. Ceo compensation and fair value accounting: 24

Evidence from purchase price allocation. Journal of Accounting Research 51, 819–854. 25

Stein, J.C., 1989. Efficient capital markets, inefficient firms: A model of myopic corporate 26

behavior. Quarterly Journal of Economics 104, 655–669. 27

Taylor, L., 2013. CEO wage dynamics: Estimates from a learning model. Journal of 28

Financial Economics 108, 79–98. 29

Terry, S.J., 2023. The macro impact of short-termism. Econometrica 91, 1881–1912. 30

Zakolyukina, A., 2018. How common are intentional GAAP violations? Estimates from a 31

dynamic model. Journal of Accounting Research 56, 5–44. 32

39
Appendix A: Likelihood contributions 1

All bids and valuations are scaled by the market value of target 𝑗. Φ(·) is the cdf of a 2

standard normal distribution. 3

Winning bid 4

From assumption (1) it follows that the likelihood of that bid equals

!
𝑣 1,𝑗
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 𝑃 𝑏 1,𝑗 ≤ 𝑎
𝐴1,𝑗

= 𝑃 𝐴1,𝑗 𝑏 1,𝑗 ≤ 𝑣1,𝑗

= 𝑃 𝐴1,𝑗 𝑏 1,𝑗 ≤ exp(𝑋 𝑗 𝛽 𝑧 ) exp(𝜖1,𝑗 )

= 𝑃 log(𝐴1,𝑗 ) + log(𝑏 1,𝑗 ) ≤ 𝑋 𝑗 𝛽 𝑧 + 𝜖 1,𝑗

= 𝑃 log(𝐴1,𝑗 ) + log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 ≤ 𝜖 1,𝑗

= 1 − 𝑃 𝜖 1,𝑗 ≤ log(𝐴1,𝑗 ) + log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧

𝑎
log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴1,𝑗
!
(𝑤))
=1−Φ
𝜎𝑧

Formal losing bid 5

From assumptions (1) and (2) it follows that the likelihood of that bid equals

𝑣 𝑖,𝑗
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 𝑃(𝑏 𝑖,𝑗 ≤ 𝑎 ≤ 𝑏 1,𝑗 )
𝐴 𝑖,𝑗
𝑎 𝑎
log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗 log(𝑏 𝑖,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗
! !
(𝑤)) (𝑤))
=Φ −Φ
𝜎𝑧 𝜎𝑧

40
Informal losing bid 1

From assumptions (1), (2), and (3) it follows that the likelihood of that bid equals

𝑣 𝑖,𝑗
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 𝑃(1 ≤ 𝑎 ≤ 𝑏 1,𝑗 )
𝐴 𝑖,𝑗
𝑎 𝑎
log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗 log(1) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗
! !
(𝑤)) (𝑤))
=Φ −Φ
𝜎𝑧 𝜎𝑧
𝑎 𝑎
log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗 −𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗
! !
(𝑤)) (𝑤))
=Φ −Φ
𝜎𝑧 𝜎𝑧

No observed bid 2

The likelihood of observing a bidder not submitting equals

𝑎
log(𝑏 1,𝑗 ) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗
! !
𝑣 𝑖,𝑗 (𝑤))
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 𝑃 𝑎 ≤ 𝑏 1,𝑗 = Φ
𝐴 𝑖,𝑗 𝜎𝑧

Statement that valuation is below market value 3

From assumption (2) it follows that the likelihood of observing a bidder leaving the auction
with this reason equals

!
𝑣 𝑖,𝑗
𝐿 𝑧 (𝑏 1,𝑗 ; 𝑥, 𝛽 𝑧 , 𝜎𝑧 |𝑤) = 𝑃 𝑎 ≤1
𝐴 𝑖,𝑗
𝑎 𝑎
log(1) − 𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗 −𝑋 𝑗 𝛽 𝑧 + log(𝐴 𝑖,𝑗
! !
(𝑤)) (𝑤))
=Φ =Φ
𝜎𝑧 𝜎𝑧

41
 Figure 1: Identification of Pseudo-Values: An example

(a) Assumed distribution (b) Winning Bid

winning bid b1 winning bid b1

0 1 2 3 0 1 2 3
Target Valuation Target Valuation
(c) Losing Formal Bids (d) Informal Losing Bids
winning bid b1 winning bid b1

losing bid b2

informal losing bid

0 1 2 3 0 1 2 3
Target Valuation Target Valuation
(e) No Bid (f) Withdrawn Bid
winning bid b1 winning bid b1

No bid Withdrawal

0 1 2 3 0 1 2 3
Target Valuation Target Valuation

Notes: This figure shows an example of how the restrictions described in Section 5.1 induce ranges
for the pseudo-values of bidders. In each panel, the solid line is the distribution, and the dashed
vertical line is the winning bid.

42
 Figure 2: Distribution of Earnings Preferences

E(w)
Density

0.00 0.25 0.50 0.75 1.00

Earnings Preference w
Notes: This figure shows the distribution of strategic bidders’ preference for earnings, 𝑤, based
on the estimate of 𝜃 from column 4 of Table 5. The dashed vertical line is the mean value of 𝑤.

43
 Figure 3: Value distributions

Financial
Strategic
imp
Ai = 0.85
Density

0.0 0.5 1.0 1.5 2.0

Target Valuation

Notes: This figure shows the estimated value distributions of financial and strategic bidders. The
dashed black line represents the distribution of strategic bidders’ fundamental valuation not taking
into account the accounting preferences. The solid black line shows the distribution of strategic bid-
ders’ effective valuation taking into account the earnings preference of the average strategic bidder.

44
 Figure 4: Pseudo-valuations under goodwill amortization

(a) Amortization with Impairment

0%

−3%
Valuation Difference

−6%

−9%

20 40 60
Amortization Period

(b) Amortization only

2.5%

0.0%
Valuation Difference

−2.5%

−5.0%

−7.5%

−10.0%

20 40 60
Amortization Period

10% 30% 50%

Goodwill Allocation
20% 40% 60%

Notes: This figure shows the change in a strategic bidder’s pseudo-valuation for a target,
𝑎
i.e., 𝑣 𝑖,𝑗 /𝐴 𝑖,𝑗 , if the bidder was subjected to amortization instead of impairment of good-
will. It shows how this difference in valuation changes with the amortization period and
the fraction of the purchase price allocated to goodwill. Panel A presents the valuation
changes when the accounting standard requires amortization with impairments. Panel B
presents the valuation changes when the accounting standard requires only amortization.

45
 Table 1: Goodwill Impairments

Impairment Magnitude
Industry P(Impairment) N Mean StDev p10% p25% p50% p75% p90%
All 0.096 6,575 0.358 0.352 0.009 0.044 0.220 0.623 0.997
Consumer Nondurables 0.114 536 0.274 0.326 0.004 0.018 0.114 0.456 0.905
Consumer Durables 0.104 253 0.292 0.314 0.010 0.034 0.168 0.464 0.846
Manufacturing 0.094 941 0.307 0.319 0.011 0.039 0.181 0.495 0.889
Oil and Gas 0.137 230 0.547 0.395 0.034 0.134 0.535 1.000 1.000
Chemicals 0.102 220 0.255 0.313 0.008 0.021 0.111 0.358 0.843
Business Equipment 0.087 1,350 0.439 0.362 0.015 0.090 0.368 0.796 1.000
Communication 0.135 332 0.300 0.327 0.004 0.028 0.163 0.535 0.861
Utilities 0.080 93 0.387 0.359 0.018 0.054 0.222 0.687 0.998
Wholesale and Retail 0.091 875 0.334 0.348 0.007 0.035 0.185 0.599 0.981
Medical and Pharma 0.073 548 0.434 0.392 0.007 0.061 0.298 0.881 1.000
Other 0.104 1,197 0.337 0.332 0.011 0.050 0.212 0.572 0.943
Notes: This table summarizes descriptive statistics on the occurrence and magnitude of goodwill impair-
ments in the universe of Compustat firms and broken down by Fama-French 12 industries for fiscal years
ending after June 6, 2001. P(Impairment) is the unconditional probability of observing a goodwill impair-
ment charge for a fiscal year of a firm. Impairment Magnitude is the proportion of beginning-year goodwill
impaired during a fiscal year with an impairment.

46
 Table 2: Purchase Price Allocation to Goodwill

Target Industry Mean StDev p10% p25% p50% p75% p90% N

All 0.436 0.223 0.130 0.266 0.453 0.602 0.711 860
Manufacturing 0.454 0.187 0.204 0.334 0.474 0.585 0.673 61
Wholesale and Retail 0.464 0.201 0.264 0.365 0.462 0.570 0.681 44
Oil and Gas 0.237 0.206 0.000 0.065 0.204 0.370 0.502 43
Communication 0.386 0.187 0.173 0.256 0.383 0.486 0.635 39
Business Equipment 0.501 0.219 0.196 0.356 0.538 0.657 0.743 308
Consumer Nondurables 0.424 0.206 0.153 0.293 0.468 0.593 0.669 30
Utilities 0.319 0.202 0.055 0.188 0.305 0.479 0.540 25
Medical and Pharma 0.375 0.218 0.094 0.197 0.366 0.530 0.656 176
Chemicals 0.365 0.229 0.127 0.204 0.383 0.506 0.565 14
Consumer Durables 0.486 0.184 0.289 0.392 0.458 0.601 0.608 11
Other 0.460 0.223 0.141 0.322 0.494 0.614 0.719 109
Notes: This table summarizes descriptive statistics on the fraction of the purchase price of a takeover is
allocated to goodwill. The data includes all M&A transactions with a public acquirer and a public target
between June 2001 and September 2022 that are available from the DealStat database.

47
 Table 3: Descriptive Statistics

Mean StDev p10% p25% p50% p75% p90% N

Premium (%) 42.785 36.440 12.795 21.212 34.140 52.603 83.938 861
No. of Bidders 10.595 12.425 2.000 3.000 6.000 13.000 23.000 861
Strategic Bidders 0.410 0.356 0.000 0.100 0.333 0.667 1.000 861
Financial Bidders 0.290 0.337 0.000 0.000 0.143 0.524 0.842 861
Formal Bid 0.295 0.223 0.056 0.125 0.250 0.471 0.500 861
Strategic Winner 0.631 0.483 0.000 0.000 1.000 1.000 1.000 861
Financial Winner 0.369 0.483 0.000 0.000 0.000 1.000 1.000 861
Goodwill PPA 0.463 0.224 0.156 0.297 0.478 0.623 0.715 304
Size 5.670 1.549 3.683 4.535 5.551 6.719 7.795 861
Leverage 0.174 0.229 0.000 0.000 0.078 0.304 0.500 861
Q-ratio 1.601 1.426 0.599 0.866 1.253 1.835 2.926 861
Cash Flow 0.007 0.250 -0.175 0.003 0.062 0.107 0.155 861
Cash 0.243 0.229 0.014 0.055 0.173 0.376 0.574 861
R&D 0.018 0.033 0.000 0.000 0.005 0.028 0.050 861
Intangbiles 0.205 0.213 0.000 0.013 0.139 0.350 0.536 861
S&P 500 0.086 0.150 -0.124 0.024 0.109 0.155 0.227 861
Credit Spread 0.026 0.007 0.017 0.019 0.026 0.030 0.033 861
Notes: This table shows summary statistics on the takeover auctions studied in this paper. Deal Premium
(%) is the premium that the acquirer paid for the target relative to the target’s stock price four weeks before
the takeover announcement or on the day before the target issued a press release that they are engaged in a
takeover process whichever is earlier. No. of Bidders is the number of parties that have signed a confidential-
ity agreement to participate in the auction. Public Winner, Financial Winner, and Private Winner are indicator
variables equal to 1 if the winning bidder is a public company, a PE firm, or a private company respectively.
Goodwill PPA is the fraction of the purchase price that acquirer allocated to goodwill. Size is the log of total
assets of the target in the last quarter before the takeover announcement. Leverage is the target company’s
leverage ration in the last quarter before the takeover announcement. q-Ratio is the target company’s Tobin’s
q in the last quarter before the takeover announcement. Cash Flow is the target company’s total cash flow
over the last four fiscal quarters scaled by total assets in the last quarter before the takeover announcement.
R&D is the target company’s R&D expense over the last four fiscal quarters scaled by total assets in the last
quarter before the takeover announcement. Intangibles is the target company’s total intangible assets scaled
by total assets in the last quarter before the takeover announcement. S &P 500 is the annualized return on
the S&P 500 over the last fiscal quarter of the target before the takeover announcement. Credit Spread is the
spread between Baa rated corporate bonds and the rate on 10-year U.S. treasuries on the day before the
merger announcement

48
 Table 4: Determinants of Bidder Type for Losing Bids

Bidder Type = Strategic

(1) (2)
Constant 2.608∗∗∗
(0.292)
Size -0.059∗ -0.100∗∗∗
(0.032) (0.034)
Q-ratio 0.068∗ 0.095∗∗
(0.039) (0.040)
Leverage -0.260 -0.047
(0.198) (0.205)
Cash Flow -1.015∗∗∗ -1.046∗∗∗
(0.296) (0.301)
Cash 0.334 0.531∗∗
(0.228) (0.248)
Intangibles -0.275 -0.104
(0.170) (0.204)
R&D 6.060∗∗∗ 3.934∗
(2.069) (2.051)
S&P 500 -0.205 -0.240
(0.286) (0.298)
Credit Spread -17.981∗∗∗ -19.087∗∗∗
(5.911) (6.067)
Log(# of bidders) -0.651∗∗∗ -0.657∗∗∗
(0.043) (0.044)
Financial Winner -0.984∗∗∗ -0.965∗∗∗
(0.070) (0.073)
Winning Bid -0.006∗∗∗ -0.006∗∗∗
(0.002) (0.002)
Industry fixed effects No Yes
Observations 4,659 4,659
Pseudo R2 0.164 0.180
Notes: This table summarizes the estimation results of a logit model that estimates the determinants of
whether a losing bid is from a strategic firm or a financial sponsor. The sample of bids used to estimate this
logit model is those losing bids where the bidder type is known. All variables are defined in Table 3. We
cluster standard errors by target industry. Levels of significance are presented as follows: ∗ p<0.1; ∗∗ p<0.05;
∗∗∗ p<0.01.

49
 Table 5: Estimation Results

Strategic Financial Strategic Financial

Intercept 0.047 -0.019 -0.090 -0.108
(0.017) (0.018) (0.059) (0.070)
𝜃 2.518 2.288
(0.249) (0.203)
𝜎 0.183 0.237 0.173 0.219
(0.017) (0.022) (0.011) (0.015)
Size 0.004 -0.006
(0.007) (0.008)
Leverage 0.071 0.085
(0.116) (0.119)
Leverage2 0.078 0.014
(0.175) (0.155)
Q-ratio 0.005 -0.021
(0.011) (0.008)
Cash Flow -0.169 0.043
(0.081) (0.040)
Cash 0.026 0.063
(0.050) (0.050)
R&D 0.804 1.178
(0.679) (0.466)
Intangibles -0.035 -0.088
(0.039) (0.052)
S&P 500 0.116 0.094
(0.076) (0.078)
Credit Spread 3.527 3.979
(1.360) (1.613)
Notes: This table summarizes the estimation results from estimating the model of bid-
ders fundamental valuation of bidders as specified in Equation 12. Columns 1 and 2
(3 and 4) report estimates without (with) control variables. Columns 1 and 3 are pa-
rameter estimates for strategic bidders. Columns 2 and 4 are parameter estimates for
financial bidders. Intercept is the coefficient for the intercept. 𝜃 is the 𝛽 parameter from
the beta distribution that determines the distribution of strategic investors’ sensitivi-
ties to accounting. Size is the coefficient on the target’s log of total assets, measured
the quarter before the takeover announcement. Leverage and Leverage2 are the coeffi-
cients on the target’s leverage and leverage squared, measured the quarter before the
takeover announcement. Q-ratio, Cash Flow, Cash, R&D, and Intangibles, are the coef-
ficients on the target’s Tobin’s Q, operating cash flow over the previous four quarters
before the takeover announcement, cash balance, R&D expense over the previous four
quarters before the takeover announcement, and total intangibles. These amounts are
scaled by total assets and measured the quarter before the takeover announcement.
S&P 500 is the coefficient on the annualized market return during the last fiscal quar-
ter before the merger announcement. Credit Spread is the coefficient on the spread
between Baa-rated corporate bonds and the 10-year U.S. treasures, measured the day
before the merger announcement. 𝜎 is the variance of the bidders’ valuation. Stan-
dard errors are block-bootstrapped by deal over 500 draws.

50
 Table 6: Counterfactuals: Alternative Accounting Regimes

Valuation Effects Distributional Effects

Runner-up ΔM&A Probability ΔPE
Premium Valuation
% Failed Dealvalue PE Winner Assets

Impairment 44.91% 1.378 1.80% — 29.61% —

Cash 31.70% 1.253 2.69% -9.94% 54.47% 65.67%
Amortization with Impairment
5 Years 37.15% 1.306 2.07% -5.62% 40.74% 29.86%
10 Years 38.93% 1.322 1.97% -4.29% 37.35% 20.73%
15 Years 40.19% 1.334 1.95% -3.40% 35.28% 15.10%
20 Years 41.16% 1.343 1.94% -2.73% 33.99% 11.66%
30 Years 42.39% 1.354 1.89% -1.83% 32.51% 7.79%
40 Years 43.02% 1.360 1.83% -1.33% 31.62% 5.36%
Amortization
5 Years 37.21% 1.307 2.07% -5.57% 40.60% 29.47%
10 Years 39.27% 1.325 1.96% -4.05% 36.78% 19.18%
15 Years 40.89% 1.341 1.94% -2.91% 34.32% 12.53%
20 Years 42.21% 1.352 1.89% -1.95% 32.68% 8.21%
30 Years 44.05% 1.369 1.81% -0.60% 30.48% 2.32%
40 Years 45.20% 1.380 1.79% 0.22% 29.30% -0.83%

Notes: This table shows counterfactual results for coporate takeover auctions under different accounting
treatments of goodwill. It compares M&A market outcomes for the current impairment-only regime to a
cash-accounting regime, an amortization-only regime and a regime that requires both amortization and an-
nual impairment testing. Results are reported for amortization periods ranging between 5 and 40 years.
Premium is the average premium of the takeover price over the market value of the target based on an as-
cending auction with random jump bids. Runner-up Valuation is the average valuation of the bidder with the
second highest valuation. % Failed is probability of a takeover failing, i.e., no bidder has a valuation above
the market value of the target. ΔDealvalue is the percentage difference in aggregate dealvalue under an al-
ternative accounting regime and the current impairment regime. This calculation takes into account both
the change in average deal value and the proability of failing deals. Probability PE Winner is the percentage
of auctions with a financial bidder winning.
ΔPE Assets is the percentage difference of total assets acquired by financial acquirers under an alternative
accounting regime compared to the impairment regime.

51
 Table 7: Counterfactuals for select industries

Panel A: Business Equipment

Valuation Effects Distributional Effects

Runner-up ΔM&A Probability ΔPE
Premium Valuation
% Failed Dealvalue PE Winner Assets

Impairment 43.65% 1.366 2.14% — 29.44% —

Amortization
10 Years w/I 36.93% 1.303 2.47% -5.00% 38.65% 24.72%
10 Years 37.28% 1.306 2.45% -4.74% 37.93% 22.74%
Panel B: Oil and Gas

Valuation Effects Distributional Effects

Runner-up ΔM&A Probability ΔPE
Premium Valuation
% Failed Dealvalue PE Winner Assets

Impairment 42.58% 1.358 1.69% — 34.58% —

Amortization
10 Years w/I 39.62% 1.330 1.76% -2.14% 39.61% 12.09%
10 Years 39.81% 1.332 1.75% -2.00% 39.31% 11.40%

Notes: This table shows how the counterfactual results vary by select Fama-French-12 industry. It compares
the current impairment regime to regimes that require a 10-year amortization period with and without an-
nual impairment testing (w/I). Panel A summarizes the results for Business Equipment targets. Panel B
highlights the Oil and Gas industry. Premium is the average premium of the takeover price over the market
value of the target based on an ascending auction with random jump bids. Runner-up Valuation is the aver-
age valuation of the bidder with the second highest valuation. % Failed is probability of a takeover failing,
i.e., no bidder has a valuation above the market value of the target. ΔDealvalue is the percentage difference
in aggregate dealvalue under an alternative accounting regime and the current impairment regime. This
calculation takes into account both the change in average deal value and the proability of failing deals. Prob-
ability PE Winner is the percentage of auctions with a financial bidder winning. ΔPE Assets is the percentage
difference of total assets acquired by financial acquirers under an alternative accounting regime compared
to the impairment regime.

52
 Table 9: Counterfactual: Changing Competitive Environment

Valuation Effects Distributional Effects

Runner-up ΔM&A Probability ΔPE
Premium Valuation
% Failed Dealvalue PE Winner Assets

Competition as observed
Impairment 44.91% 1.378 1.80% — 29.61% —
Amortization w/I 38.93% 1.322 1.97% -4.29% 37.35% 20.73%
Amortization 39.27% 1.325 1.96% -4.05% 36.78% 19.18%
Adding one financial bidder
Impairment 46.68% 1.392 0.83% 2.22% 33.42% 15.37%
Amortization w/I 40.98% 1.339 0.88% -1.80% 42.17% 39.85%
Amortization 41.32% 1.342 0.87% -1.55% 41.49% 37.95%
Increasing financial bidder proportions
Impairment 41.98% 1.350 2.32% -2.54% 38.76% 27.58%
Amortization w/I 37.13% 1.305 2.48% -6.02% 46.87% 48.76%
Amortization 37.37% 1.308 2.48% -5.86% 46.21% 46.92%

Notes: This table compares counterfactual takeover outcomes of alternative goodwill accounting treat-
ments under changing competitive environments. Competition as observed imposes the empirically observed
distribution of bidder competition. Adding one financial bidder increases the bidder pool in each auction by
one financial bidder. Increasing financial bidder proportions keeps the total number of bidders constant but
increases the incidence of a financial bidder. The considered accounting alternatives are the current impair-
ment regime, an amortization-only regime, and an amortization regime with annual impairment testing
(Amortization w/I). For brevity results are only reported for a 10-year amortization period. Premium is the
average premium of the takeover price over the market value of the target based on an ascending auction
with random jump bids. Runner-up Valuation is the average valuation of the bidder with the second highest
valuation. % Failed is probability of a takeover failing, i.e., no bidder has a valuation above the market value
of the target. Probability PE Winner is the percentage of auctions with a financial bidder winning. ΔDealvalue
is the percentage difference in aggregate dealvalue under an alternative accounting regime and the current
impairment regime. This calculation takes into account both the change in average deal value and the proa-
bility of failing deals. ΔPE Assets is the percentage difference of total assets acquired by financial acquirers
under an alternative accounting regime compared to the impairment regime. Aggregate changes are com-
pared to the impairment regime under the emprically observed competitive environment.

53