Merger Spillovers in Collaborative Partnerships

Merger Spillovers in Collaborative Partnerships
Olubunmi Faleye
D’Amore-McKim School of Business
Northeastern University, Boston, MA 02115
[email protected]; (617) 373-3712
Tiantian Gu
D’Amore-McKim School of Business
Northeastern University, Boston, MA 02115
[email protected]; (617) 373-2932
Anand Venkateswaran
D’Amore-McKim School of Business
Northeastern University, Boston, MA 02115
[email protected]; (617) 373-7873
March 2017
Abstract
We examine whether a firm in a collaborative partnership experiences spillover effects from the merger of
its partner with a third firm. We find that merger announcement return to the non-merging partner is
positive. Its post-merger abnormal sales growth and operating profitability are also positive. Spillover
effects are larger when the merger creates additional complementarities for the non-merging partner, such
as when the partnership is intra-industry and the merger is with an intra-industry firm or when the
partnership provides for technology transfer or R&D collaboration and the merger is with an R&D intensive
firm. Other potential sources of spillover effects including the relaxation of financial constraints and an
increased likelihood that the non-merging partner subsequently becomes a takeover target appear to be less
important.
JEL classification: G34, L22, L24
Keywords: Spillover effects; Joint ventures; Strategic alliances; Mergers

We gratefully acknowledge comments and suggestions by Gordon Phillips, Gerard Hoberg, Toni Whited, and
seminar participants at Northeastern University, University of Saskatchewan, and University of Massachusetts Lowell.
We are grateful to Gerard Hoberg and Gordon Phillips for providing their data on product similarity industries.
1.
Introduction
Collaborative partnerships such as joint ventures and strategic alliances are interfirm contracts that
allow each partner continued access to specified resources (e.g., technical skills, production facilities, and
intellectual properties) of the other without repeated open-market transactions or an outright merger. These
partnerships are pervasive,1 and one of their basic properties is that they create linkages that magnify the
potential for evolving circumstances at one firm to affect the prospects of another. Yet not much is known
about the nature and sources of such spillover effects. Our goal is to fill this gap by studying the effects on
a firm of a merger involving its joint venture or strategic alliance collaborators.
Mergers provide an interesting setting for studying these spillover effects. They are major events
that often transform the strategic landscape of merging firms, which increases the likelihood that firms
connected to them are affected by the transaction. Relatedly, prior work (e.g., Shahrur, 2005) shows that
mergers have significant effects on the combining firms’ rivals, customers, and suppliers. Given the
complexities of interfirm relations, it is important to understand whether these effects differ for firms that
are partners of the merging firms compared to those that are non-partners. Mergers are also relatively more
common compared to other significant corporate events such as bankruptcies and restructurings. As a result,
firms in collaborative partnerships are more likely to have their partners merge with other firms than be
involved in other significant events.
We obtain data on collaborative partnerships and mergers from Thomson Reuters’ Securities Data
Company (SDC) Platinum database. These data include 2,870 joint ventures and strategic alliances initiated
by 2,158 unique publicly traded U.S. firms between 1986 and 2013 and 1,354 mergers involving these firms
during the same period. Our results suggest that, on average, mergers involving one partner have positive
spillover effects on the other. In particular, cumulative abnormal returns (CAR) over the five-day window
(-2, +2) surrounding the merger announcement average a significant 0.16% for the full sample of portfolio
of partners not involved in the transaction. As we show later, CARs are much larger in subsamples where
1
Data from SDC show that 76% of S&P 500 firms initiated at least one partnership between 1986 and 2013 and 33%
initiated six or more partnerships during the same period.
1
theory predicts stronger spillover effects.
Mergers, of course, are not random. It is therefore possible that firms consider the likely effects of
proposed mergers on their collaborative partners before entering into these transactions. This implies that
our results could be confounded by the possibility that observed mergers involving firms in partnerships
are those the participants expected not to have a deleterious effect on their partners. We perform additional
tests to alleviate this concern. Each focuses on transactions where the merging firm’s decision is plausibly
mostly influenced by factors other than the merger’s potential effect on its collaborative partners.
First, we focus on mergers occurring during merger waves because previous research (Mitchell and
Mulherin, 1996; Shleifer and Vishny, 2003; Harford, 2005) suggests that the primary drivers of such
transactions are exogenous regulatory and/or technological shocks and temporarily inflated equity prices.
Second, we analyze a subset of these mergers that occurred in industries affected by specific regulatory
events, including communications (elimination of state regulation of cellular telephone rates by the
Omnibus Budget Reconciliation of 1993), banking (Riegle–Neal Interstate Banking and Branching
Efficiency Act of 1994), and transportation (Interstate Commerce Commission Termination Act of 1995).
Finally, we restrict the sample to deals where the merging firm is a target (rather than a bidder) since targets
are less likely to initiate deals (Roll, 1986; Schwert, 1996) and highly likely to be acquired once a bid is
made (Schwert, 2000).2 In each setting, we find results that are similar to those for the full sample.
Next, we evaluate sources of the observed spillover effect. Based on prior work on the origins and
benefits of collaborative partnerships (e.g., McConnell and Nantell, 1985; Johnson and Houston, 2000;
Allen and Phillips, 2000) and sources of gains in other interfirm relations (Shahrur, 2005; Bena and Li,
2014; Erel, et al., 2015), we hypothesize that the spillover effect is attributable to any or a combination of
three potential sources: synergies arising from increased asset complementarity, relaxation of financial
constraints, and a higher likelihood that the non-merging partner itself subsequently becomes a takeover
2
The literature on mergers routinely models the process as beginning with a bidder searching for a target but targets
sometimes do initiate bids. Yet it is not likely that a target’s decision to auction itself is materially influenced by the
potential effect of the transaction on its collaborative partners because sell decisions are often associated with other
first-order events such as exit of a major investor, the desire for a white knight, or activist demands (Faleye, 2004).
2
target. To evaluate the synergy hypothesis, we note that prior research suggests that the potential for asset
complementarity varies with the nature of the relationship between collaborating partners on the one hand
and the merging parties on the other. Johnston and Houston (2000) show that complementarities are higher
for horizontal (i.e., intra-industry) relationships while Gomes-Casseres et al. (2006), Allen and Phillips
(2000), and Bena and Li (2014) suggest similar effects for technologically related firms. Thus, we examine
patterns of spillover effects for different types of relationships to see if spillover effect increases with the
potential for asset complementarity as the synergy hypothesis predicts.
We find that spillover effects are larger for firms in horizontal partnerships whose partners merge
with another firm in the same industry (1.15% vs. 0.00% for other firms). Spillover effects are also greater
for firms with similar products to their partners (using definitions and data from Hoberg and Phillips (2010,
2015)) when the partner merges with a third firm that also offers similar products (1.18% vs. 0.07% for
other partnership-merger combinations). Since the merger of a same-industry (or similar-product) partner
with a third firm within the same product market increases asset complementarity across the three firms,
we interpret these results as consistent with the synergy hypothesis. We also find that mergers likely to
generate greater technology complementarity and increase the flow of technological information to the nonmerging partner create stronger spillover effects. In particular, spillover effects are greater if the partnership
includes a technology transfer or R&D collaboration agreement and the merger is with an R&D intensive
third firm (0.48% vs. 0.08% for other partnerships) or when the partners share significant technology
overlap and the merger is with a third firm whose technology also overlaps with the collaborating partners
(0.44% vs. 0.11% for others).
Increased synergies should eventually manifest in improved operating performance. Therefore we
analyze the non-merging partner’s abnormal operating performance in the year following the merger as a
further test of the synergy hypothesis. We follow Barber and Lyon (1996) and define abnormal operating
performance as the match-adjusted change in operating variables from the year before to the year after the
merger. We find that the non-merging partner experiences a 4.4% and 4.7% abnormal increase in revenue
growth and operating profitability, respectively. Such increases are also higher for those in horizontal
3
partnerships whose partners merge with other firms in the same industries and those who are significantly
technologically related with their merging partner and its merger counterparty.
While these results are consistent with the synergy hypothesis, we recognize that they are also
consistent with other explanations. For example, they may simply reflect the positive spillover effects of
horizontal mergers on firms that share product-market relationships with merger participants as shown by
Fee and Thomas (2004) and Shahrur (2005). They may also reflect gains to the non-merging partner from
increased market power arising from reduced product-market competition following the merger. To test
these alternative explanations, we focus on intra-industry mergers and examine the stock price reactions of
non-partner industry firms, as well as corporate customers and suppliers of the partnering firms. Consistent
with prior work, we find that partner and non-partner rival firms experience positive stock price reactions
at the announcement of horizontal mergers. However, the effect is stronger for the former (1.15% vs.
0.47%). In contrast, customers and suppliers of partnering firms experience no significant spillover effects
from the merger. Since the market power hypothesis predicts negative spillover effects on customers and
suppliers (Eckbo, 1983), we conclude that the evidence is inconsistent with it.
Next, we examine the role of financial synergies in explaining merger spillover effects. First, we
construct the Whited and Wu (2006) index of financial constraints and examine post-merger changes in this
index for non-merging firms relative to a match group established on the basis of industry, firm size, and
index scores in the year preceding the merger. For the full sample, the non-merging partner experiences no
abnormal change in financial constraints following the transaction. In contrast, financially constrained nonmerging partners experience a significant reduction in financial constraints in the year after relative to the
year before the merger. We also find that financially constrained counterparties experience a significant
stock price reaction of 0.87% at the merger announcement while merger announcement returns are
insignificant for financially unconstrained non-merging partners. Furthermore, announcement returns are
larger for financially constrained partners that experience the largest reduction in financial constraints
4
(relative to those that experience smaller reductions in financial constraints). Taken together, these results
are consistent with the financial constraints hypothesis.3
We test the takeover hypothesis by first examining the effect of a merger involving one partner on
the likelihood that the other partner subsequently becomes an acquisition target. We find that the probability
of becoming a target increases significantly with the proportion of a firm’s partners involved in a merger
during the previous three years. We then examine how the counterparty’s stock price reaction at deal
announcement varies with the predicted probability of its becoming a target in a subsequent deal. We find
that counterparties with greater increases in the likelihood of becoming a subsequent target experience a
more positive stock price reaction, which is consistent with the takeover hypothesis.
Overall, our results suggest that firms in collaborative partnerships are affected by the decisions of
their partners to engage in mergers, that such spillover effects are generally positive, and are partially
explained by increased opportunities for operating and financial synergies as well as a higher likelihood
that the counterparties themselves subsequently become acquisition targets. We evaluate these explanations
jointly by estimating regressions of announcement period CARs on proxies for each source while
controlling for firm and partnership characteristics. Proxies for asset complementarity are positive and
significant in these regressions while those for financial constraints and takeover likelihood are
insignificant. Thus, we conclude that increased operating synergies play the most meaningful role in
explaining merger spillover effects.
This paper joins Boone and Ivanov (2012) as the first to study the spillover effects of collaborative
partnerships. Boone and Ivanov focus on the spillover effects of corporate failures on strategic alliance
partners and show that the bankruptcy of one partner causes negative abnormal stock returns for the nonbankrupt partner and reduces its future profitability. Our results complement theirs by demonstrating that
the spillover effects of collaborative partnerships are positive when one partner is involved in a merger.
Our results also provide additional evidence on merger synergies and the extent to which asset and
3
Results are similar when we use the Hadlock and Pierce (2010) index as an alternative measure of financial constraint.
5
technology complementarities influence the spillover effects of mergers. Finally, we contribute to the
literature on interfirm linkages and the spillover effects of shocks to one firm on interconnected firms.
Cohen and Frazzini (2008) shows that firms’ revenue and operating income are significantly more
correlated when they are linked as customer–supplier relative to periods when they are not linked. Hertzel
et al. (2008) show that distress leading to bankruptcy is associated with negative stock price effects for
suppliers of affected firms. Chang et al. (2015) show that rivals and suppliers of credit-downgraded firms
experience significant credit spread increases during such downgrades. We extend this literature by
documenting and explaining the spillover effects of mergers on firms connected to the merging entities
through a collaborative partnership.
2.
Prior Studies and Primary Hypotheses
The motives and benefits of interfirm partnerships have been subjects of several prior studies. These
studies suggest that partnerships are formed when the net benefits of cooperation exceeds those from
repeated open-market contracts or a merger. Building on property rights theory of the firm (Grossman and
Hart, 1986; Hart and Moore, 1990), Gattai and Natale (2013) use contract-level data to show that firms
form partnerships when they interact repeatedly over time and across projects to undertake complementary
investments in physical assets that are easily usable outside the relationship. Robinson (2008) shows that
alliances arise to govern contracts that are not enforceable when written within firms but that are enforceable
between firms, especially when such contracts involve diversifying and/or riskier activities
Existing work broadly concludes that collaborative partnerships create value for partnering firms.
McConnell and Nantell (1985) show that joint venture participants experience positive abnormal returns
around formation of the partnership. Chan et al. (1997) report similar results for strategic alliances and also
show that partnering firms achieve better cash flow, return on assets, and return on equity. Johnson and
Houston (2000) find that horizontal partnerships create gains for both partners while vertical partnerships
create wealth mainly for suppliers. Lastly, Allen and Phillips (2000) show that alliances and joint ventures
6
in conjunction with block equity ownership create significant increases in targets’ stock prices and
operating profitability.
By definition, collaborative partnerships create interfirm dependencies that potentially tie fortunes
of partnering firms to evolving circumstances at each other. Yet, with the exception of Boone and Ivanov
(2012), the literature is silent on the spillover effects of partnerships on collaborating firms. We seek to fill
this gap by analyzing the effects on a firm of a relatively common but significant strategic event (i.e., a
merger) involving its partner. The previous studies referenced earlier show that interfirm partnerships create
value by generating synergies, providing operating flexibility, easing financing constraints, and alleviating
contracting costs. Our basic premise is that the extent to which a merger involving one partner creates
spillover effects for the other depends on whether and by how much the merger magnifies or attenuates
these effects.
First consider synergies. The literature largely accepts that synergies are created through
complementarities that cause the combined value of independent firms to be greater than the sum of their
individual values. Johnson and Houston (2000) suggest that such complementarities include scale
economies in production and distribution, cost savings from adjacent production facilities and/or reduced
inventory transportation costs, and the shifting of control over resources to managers of superior ability. In
general, realizing synergies requires the (total or partial) combination of erstwhile freestanding firms. Thus,
a merger involving one firm in a collaborative partnership can increase complementarities with the other
partner by bringing a third firm into the relationship. We term this the synergy hypothesis. This hypothesis
implies that the extent of such complementarities and thus the potential for spillover effects depend on the
firms’ product-market relationships and their asset characteristics. We evaluate it by testing for greater
spillover effects in relationships with greater potential for complementarities.
Intuitively, opportunities for complementarities are greatest in horizontal relationships because
horizontally related firms produce and sell similar products in similar markets. Consistent with this,
Johnston and Houston (2000) find that synergies are higher for horizontal partnerships (relative to vertical
partnerships) while Fee and Thomas (2004) and Shahrur (2005) show that efficiency gains explain the
7
wealth effects of horizontal mergers. Bernile and Bauguess (2011) use management forecasts of mergerrelated synergies to show that merger announcement returns and post-merger operating performance
increase with forecasted synergies. Based on these results, our first test of the synergy hypothesis focuses
on whether merger spillover effects are stronger when one of the firms in a horizontal partnership merges
with a third horizontally related firm.
Several studies on interfirm relations focus on complementarities arising from the flow of
technological information. Chan et al. (1997) show that alliances that involve the transfer or pooling of
technical knowledge create more value than those that do not. Gomes-Casseres et al. (2006) show that
collaborative agreements enhance the technological capability of partnering firms by promoting the sharing
of technological knowledge between partners. They also show that the flow of technological information is
greatest between units of an integrated firm and least between pairs of non-allied entities. Bena and Li
(2014) show that firms with overlapping technologies are more likely to merge and that such mergers create
greater innovation output.
Overall, these studies suggest that gains and complementarities are higher in relationships with
greater technological overlap. Thus, our second test of the synergy hypothesis focuses on whether merger
spillover effects are stronger when one of a pair of partnering firms with overlapping technologies merges
with a third firm in the same technology space as the collaborating partners. As detailed later, we use two
proxies to measure technology overlap between collaborating partners. The first is an indicator variable that
equals 1 if the partnership agreement explicitly provides for technology transfer or R&D collaboration
between the parties. This is based on the intuition that such a formal provision makes sense only if there is
some overlap in the partnering firms’ technologies. The second is the technology overlap measure of Bena
and Li (2014) and is based on correlations between patent portfolios of the partnering firms.
Erel et al. (2015) show that mergers help to relieve financial constraints at target firms while Berg
and Friedman (1978) report a survey of managers showing that capital acquisition is a primary reason for
establishing collaborative partnerships. Similarly, Johnson and Houston (2000) show that collaborative
agreements create value for partnering firms when one partner is subject to information asymmetry
8
problems that limit its ability to raise external capital. These studies suggest that a merger involving a firm’s
collaborative partner may result in favorable spillover effects for that firm by easing the merging partner’s
financial constraints or enhancing its ability to alleviate financial constraints at the non-merging partner.
We term this the financial constraints hypothesis. We test it by evaluating changes in the non-merging
partner’s financial constraint around the merger and whether spillover effects are greater when the nonmerging partner experiences greater reductions in financial constraints.
In addition to the foregoing, a merger involving a firm’s collaborative partner may result in positive
stock price spillover effects because it causes investors to increase the likelihood of the firm itself becoming
a takeover target. This is our takeover probability hypothesis. To evaluate it, we first examine the effect of
a merger involving one partner on the likelihood that the other partner subsequently becomes an acquisition
target. We then examine whether the non-merging partner’s stock price reaction at deal announcement
varies positively with the predicted probability of its becoming a target in a subsequent deal.
In spite of our hypotheses on the spillover benefits of a merger involving a firm’s partner, we
recognize that such a merger may nevertheless have negative wealth effects on the non-merging partner. It
could, for example, divert the merging firm’s attention and resources away from the non-merging partner
to its merger partner in such a manner as to reduce synergies with the former especially because it may
signal a strategic shift that marginalizes existing partnerships. Furthermore, practitioners and textbooks
often propose that a collaborative agreement is in many cases a precursor to the merger of partnering firms.
Thus, the decision of a firm’s partner to merge with another firm implies the loss of a potential merger
partner with whose operations, culture, and management it is quite familiar. This may signify that the
merging partner considers the partnership as not beneficial or significant enough to pursue to its logical
conclusion of a complete integration and could portend reductions in any financial flexibility obtained by
the non-merging partner from the partnership. Finally, integrating the combining businesses may place a
significant demand on the talent and time of the merging partner’s top management, thereby reducing its
ability to maintain its commitment and contributions to the partnership.
9
3.
Sample Construction and Variable Descriptions
We start with all joint ventures and strategic alliances announced between January 1, 1986 and
December 31, 2013 involving two public firms headquartered in the United States. For tractability, we
restrict our sample to partnerships with exactly two partners though each firm may have more than one
bilateral partnerships. These conditions result in an initial set of 5,722 firm-pairs, representing 93% of all
partnerships between U.S. public firms in the SDC database. Next, we obtain a list of all completed mergers
announced between January 1, 1986 and December 31, 2013 from SDC’s mergers and acquisitions (M&A)
database. As with partnerships, we require that both the acquirer and target be U.S. public firms. We also
exclude M&A transactions with a deal value of less than $1 million. These requirements produce an initial
sample of 5,271 deals.
Next, we identify all partnerships where one partner is either a target or an acquirer. For ease of
exposition, in the partnership between A and B, A is the non-merging partner and B is either the target or
acquirer in the merger between B and a third company, C. Our objective is to evaluate the effect of the
merger between B and C on A, in part by examining A’s abnormal stock return around the merger
announcement date. To mitigate potential contaminations due to deals that are close to each other, we
require merger events affecting each non-merging partner to be separated by at least 90 days. This yields
6,603 instances where one partner from a collaborative partnership is involved in a merger. It includes 1,354
merger deals and 2,870 unique bilateral alliances. These represent about 26% of all M&A transactions
(5,271) and 50% of all partnerships (5,722) in our initial sample. Table 1 provides the industry distribution
of the partnerships in our final sample, based on Fama-French 10 industries. It shows that the business
equipment industry has the largest number of partnerships at 50.1% of partnerships, followed by the
healthcare, medical equipment, and drugs industry at 15.5%.
3.1.
Classifying partnerships and mergers
We classify each partnership or merger as horizontal or non-horizontal for the purpose of testing
the synergy hypothesis. A horizontal partnership is one where both firms are in the same historic 4-digit
10
standard industrial classification (SIC) code industry in the year immediately preceding the merger. We
define horizontal mergers analogously. All other partnerships and mergers are non-horizontal.
Hoberg and Phillips (2010, 2015) argue that SIC codes represent groupings based on inputs to a
firm, whereas an industry classification that relies on firm outputs may be more appropriate in certain cases,
such as when the goal is to capture interfirm complementarities. Consequently, they develop a text-based
algorithm that classifies firm-pairs as operating in “similar” output markets based on product descriptions
in corporate filings. Using their dataset, we classify partnerships as similar-product partnerships if both
partners operate in a similar output market in the year of the merger or either of the two years preceding it;
all other partnerships are non-similar-product partnerships. We use an analogous procedure to classify
mergers into similar-product mergers and non-similar-product mergers, based on acquirer-target pairs.
These data are available only for 1996-2013. Thus, tests using them are limited to this period.
Panels A and B of Table 2 present summary statistics for non-merging partners and merging
partners, respectively. Given our sampling requirements, each non-merging partner can have up to four
instances per year where one of its partners engages in a merger. 4 To avoid duplication, we report firm
characteristics for non-merging partners at the firm-year level. The non-merging partner is typically smaller
in terms of total assets and revenue vis-à-vis the merging partner. Average (median) assets of non-merging
partners amount to $14.0 billion ($513 million), compared to $42.1 billion ($16.4 billion) for the merging
partner. Revenues display a similar pattern. Non-merging partners are also younger than merging partners
in terms of the number of years since each firm first appeared on Compustat, with an average (median) age
of 18 (14) years relative to merging partners whose average (median) age is 28 (33) years.
Panel C of Table 2 reports partnership characteristics. The median firm has 3 partnerships and the
median partnership has existed for 4.5 years before the merger announcement. About 20% of partnerships
are horizontal and approximately 33% of partnerships specify a transfer of technology or R&D
collaboration between the two partners.
4
About 63% of non-merging partners appear once, 27% twice, 9% three times, and 1% four times per year.
11
3.2.
Measures of spillover effects
Our primary measure of the spillover effect of a merger on collaborative partners of the merging
firm is the cumulative abnormal stock return to the collaborative partners around the merger announcement.
Because each merging firm typically has several partnerships (we restrict our sample to partnerships with
only one other firm but each firm may have several bilateral partnerships), we measure the spillover effect
as the cumulative abnormal return (CAR) to the equally-weighted portfolio of partners over the three [-1,
+1] and five [-2, +2] days surrounding the merger announcement. We calculate CARs as residuals from
market model regressions estimated over days [-256, -46] relative to the merger and require a minimum of
100 usable returns during the estimation window.
As discussed later, we also evaluate spillover effects using abnormal operating performance,
defined as match-adjusted incremental sales growth and operating profitability (OIR, i.e., the ratio of
operating income after depreciation to sales). To construct these variables, we begin with all Compustat
firms that were not involved in a merger or partnership over our entire sample period. We then use a
Mahalanobis distance matching procedure based on values in the pre-merger year to identify up to five
potential matches per non-merging partner. Each potential match is in the same 2-digit SIC code industry
as the non-merging partner and is one of the five closest firms to it based on firm size (natural log of
revenue) and either sales growth or OIR, depending on the relevant performance metric. The final match
firm is the closest of the five potential matches that has data on sales growth and OIR for years t-1, t, and
t+1, where t is the merger year, and whose sales lie between 25% and 200% of the non-merging partner’s
sales. Of the 5,266 non-merging partner-years, we are able to match 4,362 (representing 1,126 out of 1,354
mergers).5 We define incremental operating performance as the difference between years t+1 and t-1 values
5
We examine the validity of our matching procedures by testing for significant differences in our outcome variables
(sales growth and OIR) between portfolios of match firms and non-merging partners in the year before the merger.
Neither the mean nor the median of each variable is significantly different across the two samples. We also estimate
a logit regression where the dependent variable is an indicator variable that equals 1 for non-merging partners and 0
for match firms. The explanatory variables in this regression are sales growth, OIR, and firm size, all in the year before
the merger. Neither sales growth nor OIR is significant in this regression but firm size is positive and significant. Thus,
our test firms are larger than their matches, which is not surprising since we allow matching firms to be within 25%
and 200% of test firms.
12
for each firm on each variable and calculate each non-merging partner’s abnormal operating performance
metric by subtracting its match firm’s incremental operating performance value from its own incremental
operating performance value. We perform our tests using sales-weighted portfolios of non-merging
partners’ abnormal operating performance measures.
4.
Empirical Results
4.1.
Partnership formation announcement returns
We start with an examination of the market’s response to the original formation of our sample
partnerships by calculating CARs for both partners over the 3- and 5-day windows surrounding the
partnership formation date. The first row of Table 3 shows that partnership formations attract positive and
significant stock price reactions, with average CAR[-1, +1] and CAR[-2, +2] of 1.35% and 1.43%,
respectively. These numbers are consistent with those in prior studies (e.g., McConnell and Nantell, 1985;
Johnson and Houston, 2000; Allen and Phillips, 2000) and suggest that the partnerships in our sample are
similar to those examined in earlier work.
The other rows in the table reveal interesting contrasts in partnership announcement CARs for
subsamples of partnering firms based on whether they participate in subsequent mergers and their roles in
such deals. First, non-merging partners experienced a higher CAR at partnership announcement than
partners that eventually become merger participants even though mean partnership announcement CARs
are positive and significant for both sets of firms (CARs[-1, +1] and [-2, +2] of 1.97% and 2.04% vs. 0.54%
and 0.61%, respectively). More importantly, this result is driven by partners that eventually become bidders.
For these firms, average partnership announcement CARs are 0.32% and 0.35% for the [-1, +1] and [-2,
+2] windows, compared to 2.29% and 2.46% for their non-merging partners. In contrast, average
partnership formation CARs for eventual targets are 0.99% and 1.10%, statistically indistinguishable from
the 1.13% and 0.85% for the non-merging partners of eventual targets. Assuming that partnership
announcement CARs capitalize each firm’s expected future benefits from the partnership, these results
suggest that partners that derive fewer benefits from collaborative agreements tend to seek out other
13
strategic opportunities by becoming bidders in subsequent mergers. In contrast, firms that benefit more
from collaborative agreements pursue no other strategic deals unless they are sought out as targets by other
firms.
As shown in Table 2, non-merging partners are smaller than those that eventually become merger
participants. Since CARs are percentage returns, this raises the question of whether the differences reported
above are simply mechanical size effects. We evaluate this by focusing on the subsample of large partnering
firms, defined as those in the top one-third of the distribution of total assets for all Compustat firms in the
partnership formation year. If the results in Table 3 are mechanical, then we should find no differences in
partnership formation CARs between eventual merger participants and non-participants in this sample. As
it turns out, this is not the case. Average partnership announcement CARs for large eventual bidders are
0.15% and 0.22% for the [-1, +1] and [-2, +2] windows, respectively. These are significantly smaller than
the corresponding CARs of 0.63% and 0.92% for large non-merging partners of these firms. In contrast,
there are no differences in partnership formation CARs for large eventual targets (3- and 5-day CARs of
0.58% and 0.71%) and their large non-merging partners (3- and 5-day CARs of 0.59% and 0.76%). Results
are similar when we define large firms as those in the top quartile or top half of the distribution of total
assets. Thus, even among the largest firms, partners with smaller formation returns become bidders while
those with larger partnership returns either do not engage in a subsequent merger or do so only because
they are sought out as targets.
4.2.
Stock price spillover effects
Next, we examine the spillover effect of a merger involving a firm’s partners by analyzing the non-
merging partner’s stock price reaction to the announcement of the merger. The first row of Table 4 presents
full sample average CARs for portfolios of non-merging partners at merger announcement: 0.05% and
0.16% for the [-1, +1] and [-2, +2] windows, respectively. The former is statistically insignificant while the
latter is significant at the 5% level. Thus, partners of merging firms appear to experience positive spillover
effects at merger announcement. As we show later, these effects are much stronger when the potential for
complementarities are greater.
14
A plausible concern with this result is that mergers are not random. In particular, a firm’s decision
to merge with another firm may not be independent of its existing partnerships. If firms consider the
potential effect of a merger on their partners, observed mergers will be those where merging firms expect
a non-negative spillover effect on their collaborative partners and our result would be a spurious aftermath
of this self-selection. We address this concern by performing additional tests where we focus on transactions
that are likely primarily motivated by considerations other than their potential spillover effects on the
merging firms’ partners.
In the first test, we focus on transactions occurring during merger waves based on findings in prior
studies (e.g., Mitchell and Mulherin, 1996; Shleifer and Vishny, 2003; Harford, 2005) that suggest such
transactions are largely motivated by factors external to the merging firms, such as an exogenous regulatory
or technological shock, stock market misvaluation, and relatively cheap external financing. We follow the
methodology of Harford (2005) to identify industry-years experiencing merger waves. The benefit of doing
so is that it allows us to identify a relatively large sample of transactions that plausibly occurred in response
to industry shocks. However, the identification is not completely clean because it is based on a statistical
analysis of the concentration of industry mergers.
As a result, our second test uses a subset of these merger wave transactions that occurred in
industries affected by significant regulatory events. These industries and regulatory events are
communications (Omnibus Budget Reconciliation Act of 1993 eliminated state regulation of cellular
telephone rates), banking (Riegle–Neal Interstate Banking and Branching Efficiency Act of 1994 codified
at the national level the elimination of branching restrictions at the state level), and transportation (Interstate
Commerce Commission Termination Act of 1995 eliminated regulation of trucking freight rates).6 We also
include merger wave transactions in the computer and business services industries following the widespread
adoption of the internet in 1998 since the internet was a major shock to those industries. Because these
6
Several other regulatory events occurred during our sample period but are not included because they did not
precipitate merger waves in affected industries.
15
shocks are industry-specific, we require both partners to be in the same affected industry to be included in
this sample.
In the third test, we restrict the sample to cases where the merging partner is the target in the merger
based on the premise that targets are less likely to initiate the merger process (Roll, 1986; Schwert, 1996)
and are highly likely to be acquired once a bid is made (Schwert, 2000). Since bids are usually made at
substantial premiums and target directors owe a fiduciary duty to their shareholders, it is less likely that a
consideration of the potential effect of a received takeover offer on the target’s collaborative partners plays
a significant role in whether the bid results in a merger. Also, while targets sometimes initiate the acquisition
process by putting themselves up for sale, such decisions are often motivated by first-order strategic events
such as a corporate restructuring, activist demands, exit of a major investor, or the desire for a white knight
(Faleye, 2004). Finally, our earlier result showing that eventual targets earn similar partnership formation
returns as their partners not involved in a merger is consistent with the argument that factors related to the
partnership are less likely to affect whether a firm becomes a target in a subsequent merger.
The last three rows of Table 4 present results of these tests. For non-merging partners of targets,
portfolio CARs average 0.35% and 0.56% for the [-1,+1] and [-2,+2] windows, respectively. Both are
significant at the 5% level. For partners of firms merging in a merger wave, average portfolio CARs are
positive but not significant at conventional levels. As we show later, portfolio CARs are significant for
subsamples of these firms in our tests of the synergy hypothesis. Finally, 3- and 5-day portfolio CARs
average 2.57% and 2.23% for industry partners of merging firms in industries affected by specific
regulatory events (the former is significant at the 1% level and the latter at the 10% level). Overall, we
conclude that these results suggest that our primary result is not likely an artefact of a selection problem.
Next, we evaluate our hypotheses on the sources of these effects, starting with the synergy hypothesis.
4.3.
Evaluating the synergy hypothesis
Panel A of Table 5 reports spillover effects for non-merging firms in horizontal partnerships whose
partners engage in a horizontal merger, that is, where the merging and non-merging partners are in the same
historic 4-digit SIC code industry in the year immediately preceding merger announcement and the non-
16
partner merger counterparty is also in the same 4-digit SIC code industry. For these firms, portfolio CARs
average 1.14% and 1.15% for the [-1, +1] and [-2, +2] windows, both significant at the 1% level. In contrast,
as the table shows, average portfolio CARs (-0.12% and 0.00%) are insignificant for other non-merging
partners (i.e., those in non-horizontal partnerships and those in horizontal partnerships whose partners
merge with a third firm in a different industry). Furthermore, average CARs for the former group are each
significantly larger than the corresponding figure for the latter at the 1% level. Results of analogous tests
based on classifications using Hoberg and Phillips’ product similarity measures are shown in Panel B of
Table 5. As the table shows, results are similar to those based on SIC code industries. Spillover effects are
positive and significant for non-merging partners whose similar-product partners merge with a third firm
with similar product offerings but insignificant for other non-merging partners (average 3- and 5-day CARs
of 0.91% and 1.18% vs. 0.02% and 0.07%, respectively). The former are each statistically larger than the
latter at the 5% level.
Panels C and D of Table 5 present comparisons of spillover effects based on the extent of
technology relatedness between partnering firms and the non-partner merger counterparty. On our first
proxy for technology overlap (i.e., partnerships that explicitly provide for technology transfer or R&D
collaboration), we presume that the potential for overlap with the merger counterparty is greater if it is R&D
intensive (i.e., its ratio of R&D expenses to total assets exceeds the sample median). For our second proxy,
we classify firms as technologically related based on correlations between their patent portfolios using the
method of Bena and Li (2014).
In both cases, spillover effects are stronger in relationships with greater technology overlap.
Average 3- and 5-day portfolio CARs for non-merging firms in partnerships that provide for technology
transfer are 0.50% and 0.48% when their partners merge with an R&D intensive firm. Both are significant
at the 1% level. Corresponding figures for other non-merging partners are -0.08% and 0.08%, respectively.
Neither is statistically significant. In addition, the 3-day CAR is statistically smaller than the corresponding
CAR for non-merging firms in the first category (p-value = 0.013) while the 5-day is statistically marginally
smaller with a p-value of 0.123. Similarly, average CARs are positive (0.19% and 0.44%) and significant
17
for non-merging firms that are technologically related to their partners in the Bena-Li sense when such
partners merge with another technologically related firm. For other non-merging partners using this proxy,
average spillover effects are insignificant.
These results are consistent with the synergy hypothesis. Combinations of firms in similar product
markets increase the potential for complementarities across the combining firms. As our results show, such
transactions generate more positive stock price reactions for the non-merging partner. Likewise, deals
involving firms with similar technologies offer a greater potential for complementarities in the technologies
and operations of affected firms. Our results show that such transactions produce better spillover benefits
for collaborative partners of the direct participants.
As discussed earlier, a potential concern is that our tests may suffer from a self-selection bias where
only partnering firms that expect a merger to not harm their partners engage in mergers. While we believe
that the pattern of spillover effects reported above is inconsistent with such bias, we nevertheless perform
additional tests to ameliorate this concern. In particular, we repeat our tests using the sample of transactions
where the merging partner is a target and those that occur during merger waves (we do not perform these
and other subsample tests for industries affected by specific regulatory events because of the small sample
size). As shown in Table 5, results are qualitatively similar to those for the full sample.7 Overall, we
conclude that our findings are consistent with the synergy hypothesis.
4.3.1.
The synergy hypothesis: alternative explanations
The above tests of the synergy hypothesis rely on differences in spillover effects across different
types of partnerships/mergers and firm asset characteristics. Although results of these tests are consistent
7
As columns (2) and (3) of Table 5 show, horizontal partners of targets in horizontal mergers earn average CARs of
1.16% and 1.03% for the [-1, +1] and [-2, +2] windows, compared with 0.20% and 0.46% for partners of other targets.
Average portfolio CARs for partners offering similar products as targets in similar-product mergers are 1.14% and
0.84%, compared with 0.33% and 0.56% for other partners. Furthermore, firms in technology transfer partnerships
experience average spillover effects of 0.78% and 0.50% when their partner is the target in a merger with an R&D
intensive bidder, compared with 0.21% and 0.64% for all other cases. Finally, patent-based technologically related
partners of targets in mergers with technologically related bidders experience average merger spillover effects of
0.69% and 1.07%, versus 0.30% and 0.40% for partners of other targets. Table 5 reports similar patterns for firms
whose partners engage in mergers during a merger wave.
18
with the pattern predicted by the synergy hypothesis, we recognize that some of the results are also
consistent with other alternative explanations. For instance, it is possible that the result for firms in
horizontal relationships simply reflects the spillover effects of intra-industry mergers since prior studies
such as Fee and Thomas (2004) and Shahrur (2005) show that horizontal mergers create positive spillover
effects for other firms in the merger industry.
We address this concern by examining spillover effects on other firms in the same industry as the
non-merging partner but who are not collaborative partners of the merging firms. For horizontal mergers
involving one of the firms in a horizontal partnership, CARs for non-partner same-industry firms are
significant and average 0.28% and 0.47% for the 3- and 5-day event windows (not tabulated). This is
consistent with the positive intra-industry spillover effects reported in prior studies. However, the spillover
effect is stronger for collaborative partners of the merging firms. As reported earlier, average 3- and 5-day
CARs for these firms are 1.14% and 1.15%. Differences in CARs between partner and non-partner industry
firms are significant at the 5% level. Similarly, while non-partner firms in the same product space as a
merging firm and its collaborative partner experience positive CARs averaging 0.25% and 0.48% for the [1, +1] and [-2, +2] windows, spillover effects are stronger for the collaborative partner at 0.91% and 1.18%,
with the differences being significant at the 10% level. These results suggest that the spillover effects on
the collaborative partner extend beyond those experienced by other firms who merely share a product
market relationship with the merging entities and are consistent with the explanation that the merger
provides more direct benefits for the non-merging partner.
Another alternative explanation for our results is that the non-merging partner experiences positive
spillover effects because it gains from increased market power stemming from reduced product market
competition following the merger. We perform two sets of test to evaluate this alternative explanation. First,
we examine changes in market concentration following the merger and evaluate whether observed spillover
effects vary with those changes in a manner consistent with what would be expected under the market
power explanation. Second, we rely on predictions from prior studies on the effect of market power on
customers and suppliers and examine whether customers and suppliers of the firms in our sample exhibit
19
spillover effects that are consistent with those predictions. We restrict these tests to transactions where all
three firms are in the same industry (i.e., those where the partnership as well as the merger is horizontal)
because market power and industry concentration issues are most meaningful in intra-industry transactions.
Our industry concentration tests use sales-based Herfindahl-Hirschman indices (HHI) calculated
using three different industry definitions. These are 2-digit SIC code industries, 4-digit SIC code industries,
and Hoberg-Phillips product similarity industries. Panel A of Table 6 shows changes in industry
concentration from the year before to the year after the merger. Mean and median change in 2-digit SIC
code HHI are 0.9 and 5.2 points, respectively. Neither is statistically different from zero and both are
economically immaterial relative to the 100-point threshold used by the Federal Trade Commission in
deciding if to investigate or challenge proposed intra-industry mergers. Similarly, mean and median change
in HHI for product similarity industries are 38.5 and 48.8 points, respectively. Again, neither is statistically
different from zero and both are obviously less than 100. In contrast, 4-digit SIC code HHI increases by
159.1 points on average, with a median of 149.2 points. Both are statistically significant. Thus, taken
together, the evidence is mixed on whether industry concentration increases after the intra-industry mergers
in our sample.
Next, we analyze merger announcement CARs for portfolios of non-merging firms split into three
groups based on changes in industry concentration. This allows us to examine whether spillover effects
increase with increases in industry concentration as the market power explanation implies. As Panel A of
Table 6 shows, portfolio CARs for non-merging firms in 2-digit SIC code industries with the lowest
increases in concentration (i.e., those in the bottom third of the distribution of HHI changes) average 1.20%
and 1.98% for the [-1, +1] and [-2, +2] windows. Corresponding numbers for those in the middle group are
0.31% and -0.39%, respectively, while those in the top tercile average 1.54% and 1.49%. This pattern is
not consistent with a market power explanation. The remaining rows in the table show similar patterns in
the distribution of portfolio CARs for different industry concentration levels using 4-digit SIC code
industries and the Hoberg-Phillips product-similarity industries. Results are also similar when we split the
sample into quartiles or halves instead of terciles (not tabulated). In all, we conclude that our market
20
concentration tests do not support the notion that the positive spillover effects on the non-merging partner
in horizontal relationships is due to gains from a merger-induced increase in market power.
Eckbo (1983, p. 245) suggests that “one could discriminate between the collusion and efficiency
theories by examining the abnormal returns to the merging firms’ corporate customers and suppliers.”
Following this, both Fee and Thomas (2004) and Shahrur (2005) formulate predictions for suppliers and
customers of firms in intra-industry mergers that allow differentiation between market power and efficiency
explanations (see their respective Table 1). In summary, a market power explanation predicts negative
effects on suppliers and/or customers, depending on whether firms in the merger industry gain market
power in input or output markets or both, because increased market power enables rent extraction from
firms in upstream and/or downstream industries. In contrast, an efficiency explanation predicts zero to
positive effects on customers and suppliers, depending on whether efficiency gains are shared with suppliers
and/or customers.
Panel B of Table 6 presents results of our analysis of spillover effects on customers and suppliers.
In Panel B1, we follow Shahrur (2005) and define customer and supplier firms based on industry
relationships using the Benchmark Input-Output tables from the U.S. Bureau of Economic Analysis (BEA).
Specifically, customer (supplier) firms are all firms in an industry that buys (sells) at least 1% of its input
(output) from (to) the merger industry. As Table 6 shows, spillover effects on the full sample of customers
and suppliers are always positive but mostly insignificant. At the least, this is inconsistent with the market
power explanation. We obtain similar results for different subsets of customers and suppliers defined on
the basis of post-merger changes in merger industry concentration, as Table 6 shows.8
In Panel B2, we follow Fee and Thomas (2004) and identify significant corporate customers and
suppliers of the partnering firms based on each firm’s FAS 131 disclosures for the current year and each of
the preceding two years. For all customer firms, average spillover effect is positive and marginally
significant. In addition, all but one of the CARs for subsamples of customers based on post-merger changes
8
Table 6 reports results for industry concentration using HHI based on 4-digit SIC code industries. Results are similar
using other definitions of industry concentration but are not reported in order to conserve space.
21
in merger industry concentration are positive even though they are mostly insignificant. Spillover effects
are also insignificant for the full sample and various subsets of suppliers. Since we do not find significant
negative spillover effects on customers and/or suppliers of the partnering firms, we conclude that our results
are inconsistent with a market power explanation of the positive spillover effect on the non-merging partner.
Rather, they suggest that the merger of a firm’s partner with a third firm provides additional efficiency
gains for the non-merging partner. We explore this further in the next section by analyzing abnormal
operating performance of the non-merging partner following the merger.
4.3.2.
Abnormal operating performance
Panel A of Table 7 presents results for match-adjusted incremental sales growth. For the full
sample, mean and median abnormal sales growth rates are 4.4% and 2.8%, respectively. Each is significant
at the 1% level. Thus, relative to comparable firms in the same industry, collaborative partners of merging
firms increase their revenue by 4.4% on average from the year before to the year after the merger.
Furthermore, this result does not appear to be attributable to self-selection in merger decisions. In particular,
mean and median abnormal revenue growth rates are positive and significant for partners of targets, firms
whose partners merge during a merger wave, and industry partners of merging firms in industries affected
by specific regulatory events, as the last three rows of Panel A of Table 7 show.
As in our event study tests, we also evaluate differences in abnormal revenue growth based on the
potential for complementarities between partnering firms and the merger counterparty. In Panel A of Table
8, mean and median abnormal revenue growth rates are 6.5% and 8.0% for horizontal partners of firms in
horizontal mergers, compared with 4.1% and 2.4% for other non-merging partners. Likewise, mean and
median abnormal growth rates are higher for firms in similar-product partnerships whose partners merge
with firms operating in similar product-markets than for other firms (7.3% and 7.4% vs. 4.1% and 2.5%,
respectively). Results are comparable when we consider technology complementarities. Firms in
technology transfer partnerships experience mean and median abnormal revenue growth rates of 4.3% and
5.3% when their partners merge with R&D intensive firms, compared with 3.4% and 2.0% in other cases.
22
Similarly, mean and median abnormal revenue growth rates are 5.9% and 5.7% for non-merging firms when
their technologically related partners merge with other technologically related firms, compared with 3.2%
and 1.9% in other cases. These differences are consistent with the synergy hypothesis. However, only those
for technologically related non-merging firms and those in technology transfer partnerships are statistically
significant. Results are qualitatively similar when we restrict the sample to partners of targets and those in
merger waves, as columns (2) and (3) show.
Panel B of Table 7 and Table 9 present results of analogous tests for match-adjusted incremental
operating profitability. In Panel B of Table 7, mean and median abnormal OIR are 4.7% and 0.6% for the
full sample; each is significant at the 1% level. Thus, partners of merging firms significantly increase their
post-merger operating profitability relative to similar firms in the same industry. Furthermore, such
profitability gains are higher when the merger creates greater complementarities for the non-merging
partner. In Table 9, mean and median abnormal OIR are 9.6% and 2.6% for horizontal partners of firms in
horizontal mergers. Each is statistically significant and significantly larger than the corresponding figure
(3.9% and 0.5%, respectively) for firms in other kinds of relationships. Likewise, abnormal OIR is larger
for firms in similar-product partnerships whose partners merge with other firms operating in similar
product-markets (mean and median abnormal OIR of 8.9% and 3.6% vs. 4.8% and 0.7% for other firms).
Abnormal operating profitability is also larger when the merger increases the proximity of the non-merging
partner to a third firm in the same technology space, either because the partnership provides for technology
transfer and the third firm is R&D intensive or because the three firms share significant technology overlap.
Results are qualitatively similar for partners of targets and those of firms merging during merger waves.
Based on these results, we conclude that our tests support the synergy hypothesis. Mergers that
increase complementarities with the non-merging partner generate significantly higher abnormal returns.
Such mergers also are generally associated with larger increases in revenue growth rates and operating
profitability. In the next two sections, we evaluate our hypotheses on other potential sources of spillover
effects.
23
4.4.
Evaluating the financial constraints hypothesis
Our primary measure of financial constraints is the Whited-Wu index.9 Following prior literature,
we classify firms in the bottom tercile of scores for all Compustat firms in the year preceding the merger as
financially unconstrained and those in the top tercile as financially constrained. Next, we examine abnormal
changes in index scores, defined as the difference in match-adjusted index scores from the year before to
the year after the merger. We select match firms using the same procedures as for sales growth and operating
profitability with the modification that we substitute index scores for the operating performance variables.
For the full sample, mean and median abnormal change in index scores are insignificant (not tabulated).
Furthermore, financially unconstrained non-merging firms experience a slight increase in index scores in
the year after the merger. In contrast, mean and median abnormal index scores are negative and significant
for financially constrained non-merging firms, indicating that these firms experience a significant reduction
in financial constraints following the merger.
Panel A of Table 10 presents merger announcement CARs for financially constrained and
unconstrained non-merging partners. For the former, mean CARs are 0.27% and 0.87% for the [-1, +1] and
[-2, +2] windows. Both are significant at the 5% level. In contrast, mean CARs for financially unconstrained
non-merging partners are 0.06% and 0.01% for the 3- and 5-day event windows, neither of which is
significant. In addition, average CAR[-2, +2] is significantly larger for financially constrained non-merging
partners. Thus, these firms experience larger spillover effects from the merger of their collaborative
partners. Since financially constrained firms gain from the easing of financial constraints while
unconstrained firms do not, these results are consistent with the financial constraints hypothesis. We obtain
similar results using the Hadlock-Pierce index, as Table 10 shows.
We also examine variations in stock price spillover effects for financially constrained non-merging
partners based on the extent to which their financial constraint relaxes following the merger. Mean CAR[1, +1] and CAR[-2, +2] for portfolios of financially constrained non-merging partners with above-median
9
As a robustness check, we also use scores on the Hadlock-Pierce index.
24
post-merger reductions in financial constraint are 0.87% and 1.03%, respectively, compared with -0.05%
and 0.47% for financially constrained non-merging partners with below-median reductions in financial
constraint. The former are statistically significant (while the latter are not) and statistically larger than the
latter at the 5% level. Table 10 shows similar patterns when we use the Hadlock-Pierce index. Once again,
these results are consistent with the financial constraint hypothesis.
In addition, we analyze abnormal sales growth and OIR for non-merging partners based on their
financial constraint status. Panel B of Table 10 shows that mean and median abnormal sales growth rates
are positive and significant for firms in both categories, suggesting that financially constrained and
unconstrained firms experience significant increases in match-adjusted revenue growth rates following
mergers of their partners. Nevertheless, median abnormal revenue growth rate is significantly higher for
financially constrained firms (5.2%) than for those that are financially unconstrained (2.0%).10 Similarly,
abnormal sales growth rates are positive and significant only for financially constrained non-merging
partners with greater post-merger reduction in financial constraints.
Panel C of Table 10 presents results for abnormal operating profitability. Mean and median
abnormal OIR are 7.5% and 1.6% for financially constrained non-merging partners, compared with 2.0%
and 0.4% for those that are financially unconstrained. Both sets of statistics are significant but only the
means are statistically different. Also, financially constrained partners with greater reduction in financial
constraint do not experience statistically larger abnormal operating profitability than those with smaller
reductions. Overall, we conclude that these results are mostly though not fully consistent with the financial
constraints hypothesis. Next, we evaluate whether the market updates its assessment of the likelihood of
the non-merging partner becoming a takeover target and if such changes are a channel for the merger’s
spillover effects on the partner.
10
Mean abnormal revenue growth rates are similar: 3.7% and 3.8% for financially constrained and unconstrained
firms, respectively.
25
4.5.
Evaluating the takeover probability hypothesis
First, we estimate a logit regression that predicts the probability of a firm becoming a takeover
target as a function of standard variables in the literature (including leverage, return on assets, size,
market/book ratio, and asset tangibility) as well as the number of the firm’s collaborative partners involved
in a merger within the preceding three years. We are primarily interested in the latter variable as it allows
us to estimate the incremental effect of the merger of a firm’s collaborative partners on the likelihood that
the firm itself becomes a takeover target. Results (not tabulated) shows that this variable is positive and
significant at the 1% level. Economically, an increase of one standard deviation in the number of a firm’s
partners that are involved in mergers increases the probability of the firm itself subsequently becoming a
takeover target by 7.2 percentage points when other variables are evaluated at their sample means.
Next, we sort non-merging partners into quartiles based on the predicted marginal effect of our
variable of interest and classify those in the top and bottom quartiles as high and low takeover probability
firms, respectively. We then examine CARs and abnormal operating performance metrics for the two
categories. Results are summarized in Table 11.
For non-merging partners with the largest increases in the likelihood of becoming a takeover target
following the merger of their partners, Panel A shows that merger announcement returns are positive in
both the 3- and 5-day windows and significant in the former (CARs[-1, +1] and [-2, +2] of 0.16% and
0.20%, respectively). In contrast, merger announcement returns are insignificant for non-merging partners
experiencing the least increase in the probability of becoming a target. These results are weakly consistent
with the hypothesis that the spillover effects of the merger of a firm’s partner are partly explained by
changes in the market’s assessment of the likelihood of the firm itself becoming a merger participant.
We explore this further by analyzing abnormal sales growth rate and abnormal OIR for high- and
low-takeover probability firms. As Panel B of Table 11 shows, results are only partially consistent with the
takeover probability hypothesis. For example, abnormal sales growth rate is insignificant for the high- and
low-takeover probability samples as would be expected if observed stock price spillover effects are strictly
in response to a reassessment of the likelihood of takeover. In contrast, Panel C of Table 11 shows that
26
abnormal OIR is significant for high-takeover probability firms, which is inconsistent with the takeover
probability hypothesis.
In all, results in this section suggest that the likelihood of a firm becoming a takeover target
increases when its collaborative partners are involved in mergers. They also suggest that this higher
probability partially explains the spillover effects of the merger on the non-merging partner. However, the
weak operating performance results imply that other factors play important roles in determining the
occurrence and magnitude of these spillover effects. As shown earlier, these channels include improved
opportunities for synergy and the relaxation of financial constraints. In the next section, we evaluate all
channels concurrently within a regression framework.
4.6.
Regression analysis
Table 12 presents results of regressions where the dependent variable is the cumulative abnormal
return to the non-merging partner over the 3-day window surrounding the merger announcement.
Explanatory variables in column (1) are basic characteristics of the non-merging partner, including size
(natural logarithm of sales), asset tangibility (ratio of net property, plant, and equipment (PPE) to total
assets), return on assets, R&D intensity, leverage, and capital expenditures. All variables are as of the fiscal
year preceding the merger and are winsorized at the 1st and 99th percentiles. The results indicate that
spillover effects are more positive for R&D intensive firms and those whose assets consist mostly of
tangible PPE. In contrast, spillover effects decline with spending on capital expenditures.
Column (2) includes several characteristics of the collaborative partnership in addition to the firm
attributes in column (1). These additional variables are an indicator variable that equals 1 for joint ventures
and 0 for strategic alliances, duration of the partnership at merger announcement, the total number of
partnerships in which the non-merging partner participates, the number of its partners involved in mergers,
and whether the partnership includes an explicit supplier provision. As column (2) shows, the three firm
characteristics that are significant in column (1) remain significant. In contrast, none of the partnership
attribute variables is significant.
27
Columns (3)–(5) present results of separate regressions evaluating synergies, relaxation of financial
constraints, and increased takeover probability as channels for spillover effects after controlling for the
basic firm and partnership attributes in columns (1) and (2). In column (3), our proxies for improved synergy
opportunities are indicator variables that equal 1 for non-merging firms in horizontal partnerships whose
partners announce horizontal mergers, those in technology transfer partnerships whose partners merge with
R&D intensive firms, and those with high patent-based technology overlap with their partners who merge
with other technologically related firms.11 Consistent with our univariate results, the horizontal relationships
variable is positive and significant at the 5% level. Its coefficient indicates that merger announcement
returns are higher by 73 basis points for non-merging firms in horizontal partnerships whose partners
engage in a horizontal merger. Similarly, the indicator variable for firms in technology transfer partnerships
whose partners merge with R&D intensive firms is positive and significant at the 10% level. As the table
shows, such firms experience a cumulative abnormal return of 41 basis points after controlling for other
firm and partnership characteristics. Our final proxy for synergies is positive but statistically insignificant.
These results suggest that improved operating complementarities provide a significant explanation
for the spillover effects of mergers on collaborative partners of the merging firms even after controlling for
basic firm and partnership characteristics. In contrast, columns (4) and (5) show that neither the relaxation
of financial constraints nor an increase in the likelihood of takeover offers a significant explanation once
we include controls for firm and partnership attributes. In particular, the indicator variables for financially
constrained and high-takeover-probability non-merging partners are both insignificant in columns (4) and
(5), respectively. In column (6), we report results of a final regression where we include proxies for
synergies, relaxation of financial constraints, and increased takeover probability as well as basic firm and
partnership variables. As the column shows, results are quite similar to those in the other columns where
we focus on each channel independently of the others.
11
We do not include an indicator variable for our fourth proxy for complementarities (i.e., the Hoberg-Phillips product
market relatedness measure) because it is highly correlated with the horizontal-partnership-horizontal-merger variable
(ρ = 0.8). If we include it, it is not significant but other variables maintain their magnitude and level of significance.
28
5.
Summary and Conclusion
This paper asks two main research questions. First, does a firm experience spillover effects when
its collaborative partner merges with a third firm? Second, if so, what are the sources of such spillover
effects? The increasing importance of partnerships as an integral component of the strategies of a large
number of firms underscores the significance of these issues.
We explore our questions by analyzing a sample of 1,354 mergers involving the collaborative
partners of 2,158 unique firms in 2,870 partnerships between 1986 and 2013. Our results are consistent with
the notion that collaborative partnerships create significant inter-dependencies between the partnering
firms. Average merger CAR to the portfolio of non-merging partners is positive. Their match-adjusted
incremental sales growth and operating income in the year following the merger are both positive as well.
The pattern of significance of both abnormal stock returns and abnormal operating performance
suggests that the spillover effects arise from increased complementarities created by the merger for the nonmerging partner. Spillover effects are positive and significant in instances where the merger provides the
non-merging partner with additional opportunities for synergy (e.g., when the partnership is horizontal and
the merger is with a third firm in the same industry and when the partnership explicitly provides for
technology transfer and the merger is with an R&D intensive third firm). In contrast, spillover effects are
insignificant when incremental synergy opportunities are likely minimal. Other potential sources of merger
spillover effects (such as an increase in the likelihood that the non-merging partner will subsequently
engage in a merger and the relaxation of financial constraints) appear to be of less importance once we
account for firm and partnership attributes.
29
References
Allen, J. W., Phillips, G. M., 2000. Corporate equity ownership, strategic alliances, and product market
relationships. Journal of Finance 55, 2791–2815.
Barber, B. M., Lyon, J. D., 1996. Detecting abnormal operating performance: The empirical power and
specification of test statistics. Journal of Financial Economics 41, 359–399.
Bena, J., Li, K., 2014. Corporate innovations and mergers and acquisitions. Journal of Finance 69, 1923–
1960.
Berg, S. V., Friedman, P., 1978. Joint ventures in American industry Part II: Case studies of managerial
policy. Mergers and Acquisitions 13, 9–17.
Bernile, G., Bauguess S. W., 2011. Do merger related synergies exist? Working paper. Singapore
Management University.
Boone, A. L., Ivanov, V. I., 2012. Bankruptcy spillover effects on strategic alliance partners. Journal of
Financial Economics 103, 551–569.
Chan, S. H., Kensinger, J., Keown, A., Martin, J., 1997. Do strategic alliances create value? Journal of
Financial Economics 46, 199–221.
Chang, J. H., Hung, M. W., Tsai, F. T., 2015. Credit contagion and competitive effects of bond rating
downgrades along the supply chain. Finance Research Letters 15, 232–238.
Cohen, L., Frazzini, A., 2008. Economic links and predictable returns. Journal of Finance 63, 1977–2011.
Eckbo, B. E., 1983. Horizontal mergers, collusion, and stockholder wealth. Journal of Financial
Economics 11, 241–273.
Erel, I., Jang, Y., Weisbach, M. S., 2015. Do acquisitions relieve target firms’ financial constraints?
Journal of Finance 70, 289–328.
Faleye, O., 2004. Cash and corporate control. Journal of Finance 59, 2041–2060.
Fee, C. E., Thomas, S., 2004. Sources of gains in horizontal takeovers: Evidence from customer, supplier,
and rival firms. Journal of Financial Economics 74, 423–460.
Gattai, V., Natale, P., 2013. What makes a joint venture: Micro evidence from Sino-Italian contracts.
Review of Financial Economics 22, 194–205.
Gomes-Casseres, B., Hagedoorn, J., Jaffe, A., 2006. Do alliances promote knowledge flows? Journal of
Financial Economics 80, 5–33.
Grossman, S. J., Hart, O. D., 1986. The costs and benefits of ownership: A theory of vertical and lateral
integration. Journal of Political Economy 94, 691–719.
Hadlock C. J., Pierce, J. R., 2010. New evidence on measuring financial constraints: Moving beyond the
KZ index. Review of Financial Studies 23, 1909–1940.
Harford, J., 2005. What drives merger waves? Journal of Financial Economics 77, 529–560.
Hart, O. D., Moore, J., 1990. Property rights and the nature of the firm. Journal of Political Economy 98,
1119–1158.
Hertzel M. G., Li Z., Officer M. S., Rodgers K. J., 2008. Inter-firm linkages and the wealth effects of
financial distress along the supply chain. Journal of Financial Economics 87, 374–387.
Hoberg, G., Phillips, G., 2010. Product market synergies and competition in mergers and acquisitions: A
text-based analysis. Review of Financial Studies 23, 3773–3811.
Hoberg, G., Phillips, G., 2015. Text based network industries and endogenous product differentiation.
Journal of Political Economy, forthcoming.
Johnson, S., Houston, M., 2000. A reexamination of the motives and gains in joint ventures. Journal of
Financial and Quantitative Analysis 35, 67–86.
McConnell, J., Nantell, T., 1985. Corporate combinations and common stock returns: The case of joint
ventures. Journal of Finance 40, 519–536.
30
Mitchell M. L., Mulherin, H. J., 1996. The impact of industry shocks on takeover and restructuring activity.
Journal of Financial Economics 41,193–229.
Robinson, D. T., 2008. Strategic alliances and the boundaries of the firm. Review of Financial Studies 21
649–681.
Schwert, G. W., 1996. Markup pricing in mergers and acquisitions. Journal of Financial Economics 41, 153–
192.
Schwert, G. W., 2000. Hostility in takeovers: In the eyes of the beholder? Journal of Finance 55, 2599–2640.
Shahrur, H., 2005. Industry structure and horizontal takeovers: Analysis of wealth effects on rivals, suppliers,
and corporate customers. Journal of Financial Economics 76, 61–98.
Shleifer, A., Vishny, R. W., 2003. Stock market driven acquisitions. Journal of Financial Economics 70,
295–311.
Whited, T. M., Wu, G., 2006. Financial constraints risk. Review of Financial Studies 19, 531–559.
31
Table 1: Sample Distribution by Industry
The sample period is from 1986 to 2013 and includes all partnerships where one of the partners in a joint venture or
strategic alliance is involved in a merger. An event is the merger date for each non-merging partner. All three firms
associated with an event, that is, the non-merging partner (A), the merging partner (B), and the corresponding
target/bidder (C) are separate entities and U.S. public firms. The industry membership distribution is for the number
of events, mergers, partnerships and unique non-merging partners.
Events
Mergers
Partnerships
Fama-French 10-Industry
#
Consumer non-durables
Consumer durables
Manufacturing
Oil, gas, and coal
Business equipment
Telephone and television
transmission
Wholesale, retail, and some
services
Healthcare, medical equipment,
and drugs
%
%
#
%
#
%
175
2.7%
47
3.5%
87
3.0%
66
3.8%
95
1.4%
30
2.2%
50
1.7%
30
1.7%
556
8.4%
135 10.0%
210
7.3%
133
7.6%
83
1.3%
49
1.7%
35
2.0%
3,394 51.4%
49
3.6%
485 35.8%
1,439 50.1%
736 42.1%
247
3.7%
53
3.9%
116
4.0%
64
3.7%
324
4.9%
89
6.6%
152
5.3%
112
6.4%
1,068 16.2%
Utilities
75
1.1%
Others
586
8.9%
6,603
100%
Total
#
Unique nonmerging
partners
238 17.6%
43
445 15.5%
3.2%
40
1.4%
185 13.7%
282
9.8%
2,870
100%
1,354
100%
302 17.3%
34
1.9%
237 13.6%
1,749
100%
32
Table 2: Merging Partner, Non-merging Partner, and Partnership Characteristics
Panel A reports summary statistics for total assets, net sales, and firm age for non-merging partners. Panel
B reports the corresponding statistics for merging partners. Panel C reports summary statistics for
partnership characteristics. The sample period is from 1986 to 2013. In Panels A and B, we aggregate the
sample data to firm-year observations to avoid duplication in the case of multiple transactions per firm in
any given year. The sample consequently, reduces from 6,603 to 5,266 observations. We compute summary
statistics for partnership characteristics that are time invariant by aggregating observations to the level of
the partnerships. There are 2,870 unique partnerships in our sample. Sample size for partnership duration
at merger (Panel C) is at the transaction rather than partnership level because firms in a partnership can
engage in multiple mergers at different points in time. Other partnership characteristics in Panel C are at
the partnership level. Firm age is the number of years since a firm’s first appearance in Compustat. Techtransfer agreement equals 1 if the partners share either a technology transfer or R&D collaboration
agreement, 0 otherwise. Horizontal partnership equals 1 if both partners are in the same historical 4-digit
SIC code industry in the year immediately preceding the merger, 0 otherwise. P5 and P95 are the 5th and
95th percentiles, respectively.
Variable
Sample
Mean
Median
P5
P95
Total assets ($ million)
5,266
13,979
513
15
49,996
Net sales ($ million)
5,266
6,660
378
5
31,620
Firm age (years)
5,266
18
14
3
45
Total assets ($ million)
5,266
42,075
16,385
78
114,799
Net sales ($ million)
5,266
26,953
13,047
49
96,293
Firm age (years)
5,266
28
33
4
50
Sample
Mean
Median
P5
P95
Duration at merger (years)
6,603
5.76
4.50
0.42
15.50
# of Partners
2,870
6.45
3
1
26
Tech-transfer agreement
2,870
0.33
0
0
1
Horizontal partnership
2,870
0.20
0
0
1
Panel A: Non-merging Partner
Panel B: Merging Partner
Panel C: Partnerships
33
Table 3: Partnership Formation Cumulative Abnormal Returns
This table reports mean cumulative abnormal return (CAR) around partnership announcement dates for
2,227 distinct partnerships announced between January 1, 1986 and December 31, 2013. CARs are
estimated using a market model. Return windows are centered on the partnership formation announcement
date. T-statistics are for the null hypothesis that the standardized prediction errors are equal to zero. *, **,
and *** indicate significance at 10%, 5%, and 1% levels, respectively.
Description
All events with available formation CARs
No. of Events
4,454
Non-merging partner
2,648
Merging partner
2,648
CAR [-1,1]
(t-stat)
1.35%***
(9.05)
1.97%***
(8.87)
0.54%***
(3.74)
CAR [-2,2]
(t-stat)
1.43%***
(8.14)
2.04%***
(8.01)
0.61***
(3.30)
2.29%***
(8.38)
0.32%*
(1.86)
2.46***
(7.95)
0.35%*
(1.67)
1.13%***
(3.33)
0.99%***
(3.92)
0.85%**
(2.51)
1.10%***
(3.28)
Merging partner is eventual bidder:
Non-merging partner
1,927
Merging partner
1,927
Merging partner is eventual target:
Non-merging partner
838
Merging partner
838
34
Table 4: Stock Price Spillover Effects
This table reports mean cumulative abnormal return (CAR) for portfolios of non-merging partners in 1,354
mergers involving the other party in a collaborative partnership between 1986 and 2013. CARs are for
equal-weighted portfolios of non-merging partners. We exclude all events where a non-merging partner is
associated with more than one merger event in a 90-day period centered on the merger announcement date.
Partner is target equals 1 for non-merging firms whose partners are targets in the merger transaction
according to SDC. Transaction in merger wave equals 1 for firms whose partners merge during an industry
merger wave determined using the methodology of Harford (2005). Regulatory event transactions equals 1
for firms in industries whose merger waves are tied to a clear regulatory event or emergence of the internet.
Test statistics are shown in parentheses. ***, **, and * indicate significance at 1%, 5%, and 10% levels,
respectively.
Full sample
No. of
Mergers
1,354
Partner is target
593
Transaction in merger wave
483
Regulatory event transactions
75
CAR [-1, 1]
CAR [-2, 2]
0.05%
(1.30)
0.35%**
(2.51)
0.08%
(1.21)
2.57%***
(2.72)
0.16%**
(2.08)
0.56%**
(2.57)
0.01%
(0.71)
2.23%*
(1.93)
35
Table 5: Evaluating the Synergy Hypothesis Based on Patterns of Stock Price Spillover
This table reports mean cumulative abnormal return (CAR) for portfolios of non-merging partners stratified
into subsamples based on relations between pairs of partnering firms and between pairs of merging firms.
In Panel A, partnerships and mergers are horizontal if both firms in the relationship are in the same historical
4-digit SIC code industry in the year immediately preceding the merger. In Panel B, partnerships and
mergers are similar-product relations if both firms in the relationship operate in a similar output market in
the year of the merger or either of the two years preceding it, based on Hoberg-Phillips product similarity
measures. In Panel C, a partnership is a technology transfer partnership if it contains an explicit agreement
to transfer technology between the two partners. A merger partner is R&D intensive if its ratio of R&D
expenses to total assets exceeds the sample median. In Panel D, a high overlap partnership is one where the
correlation between the two partnering firms’ patent portfolios exceeds the sample median. A high overlap
merger partner is one where the correlation between the merging firms’ patent portfolios exceeds the sample
median. Correlations between patent portfolios are calculated following Bena and Li (2014). Results in
column (1) are for the full sample. Results in column (2) are for firms whose partners are targets in the
merger transaction. Results in column (3) are for firms whose partners merge during an industry merger
wave determined using the methodology of Harford (2005). Test statistics are shown in parentheses. *, **,
and *** indicate significance at 10%, 5%, and 1% levels, respectively. Entries in the “Difference” rows are
differences between CARs for portfolios of non-merging partners in the two categories above each row.
Levels of significance for this row are based on one-tailed tests because the null hypothesis is directional.
36
Table 5 continued
(1) Full sample
N
[-1, +1]
[-2, +2]
(2) Partner is target
N
[-1, +1]
[-2, +2]
(3) Deal in merger wave
N
[-1, +1]
[-2, +2]
A: Horizontal relations based on SIC codes
1.14%***
(3.45)
Other partnership–merger combinations
1,261 -0.12%
(0.18)
Difference
1.26%***
(3.97)
B: Horizontal relations based on Hoberg-Phillips product similarity
Horizontal partners, horizontal mergers
Similar-product partners, similar-product
mergers
Other partnership–merger combinations
1.15%***
(2.85)
0.00%
(1.17)
1.15%***
(2.75)
127
0.91%**
(2.34)
0.02%
(1.03)
0.89%**
(2.27)
1.18%**
(2.49)
0.07%
(1.42)
1.11%**
(2.13)
77
0.50%***
(2.66)
-0.08%
(-0.04)
0.58%**
(2.24)
0.48%***
(2.89)
0.08%
(0.75)
0.39%
(1.16)
190
0.19%*
(1.66)
0.06%
(0.63)
0.13%
(0.50)
0.44%***
(3.17)
0.11%
(0.57)
0.33%
(0.97)
216
279
169
1,114
Difference
524
478
1.16%**
(2.46)
0.20%*
(1.80)
0.96%*
(1.76)
1.03%
(1.48)
0.46%**
(2.05)
0.57%
(0.76)
101
1.14%**
(2.61)
0.33%**
(2.30)
0.81%
(1.18)
0.84%
(1.49)
0.56%**
(2.24)
0.28%
(0.30)
68
0.78%**
(2.42)
0.21%
(1.56)
0.57%
(1.14)
0.50%
(1.53)
0.64%*
(1.92)
-0.14%
(-0.20)
174
0.69%**
(2.35)
0.30%
(1.59)
0.39%
(0.82)
1.07%**
(2.55)
0.40%
(1.33)
0.67%
(1.05)
148
447
454
1.47%**
(2.56)
-0.19%
(0.25)
1.66%***
(2.80)
1.43%**
(2.00)
-0.32%
(-0.13)
1.75%**
(2.22)
1.24%**
(2.40)
-0.13%
(0.53)
1.37%**
(2.04)
1.74%**
(2.52)
-0.22%
(0.11)
1.95%**
(2.15)
0.25%
(1.41)
0.05%
(0.95)
0.20%
(0.40)
0.14%
(0.98)
0.07%
(0.40)
0.07%
(0.11)
0.15%
(1.23)
0.12%
(1.02)
0.03%
(0.06)
0.39%
(1.64)
-0.10%
(0.09)
0.49%
(0.74)
C: Technology-transfer partnerships
Technology-transfer partnership, R&D
intensive merger partner
Other partnership–merger combinations
521
1,226
Difference
491
438
D: Patent-based technology overlap
High overlap partnership, high overlap
merger partner
Other partnership–merger combinations
Difference
494
1,151
454
420
37
Table 6: Evaluating the Market Power Hypothesis
These tests are based on the sample of firms in horizontal partnerships where one partnering firm engages
in a horizontal merger with a third firm. Panel A presents changes in industry concentration from the year
before to the year after the merger and associated merger announcement CARs for non-merging firms
grouped into terciles based on the change in market concentration for their industries. Industry
concentration is the revenue-based Herfindahl-Hirschman index (HHI) calculated using all firms in
Compustat. Two-digit and four-digit SIC industries include all firms sharing the same two-digit and fourdigit SIC codes, respectively. Hoberg-Phillip industries are from Hoberg and Phillips (2010) and are
available starting in 1996. Panel B presents merger announcement CARs for corporate customers and
suppliers of the partnering firms. In Panel B1, customers (suppliers) are firms in industries that buy (sell)
at least 1% of their input (output) from (to) the merger industry using Benchmark Input-Output tables from
the BEA. In Panel B2, customers and suppliers are identified from each firm’s FAS 131 disclosure of
important customers and suppliers. Portfolios for different industry concentration changes in Panel B are
based on HHI calculated for four-digit SIC code industries. *, **, and *** indicate significance at 10%,
5%, and 1% levels, respectively.
Panel A: Post-merger Change in Industry Concentration
Portfolio CARs
HHIt+1 – HHIt-1
Industries
Mean
Median
Two-digit SIC industries
551
Top tercile ΔHHI
190
Middle tercile ΔHHI
151
Bottom tercile ΔHHI
210
0.85
(0.10)
156.05***
(11.90)
2.34
(1.22)
-140.69***
(-11.63)
159.09***
(7.04)
711.34***
(9.05)
146.02***
(22.75)
-272.74***
(-9.39)
38.45
(0.56)
190.75
(1.30)
-20.14
(-0.22)
-73.09
(-0.62)
5.18
(0.66)
54.84***
(11.57)
6.34
(0.58)
-89.25***
(-11.96)
149.24***
(11.24)
380.81***
(9.19)
149.24***
(14.61)
-113.29***
(-9.92)
48.82
(1.59)
169.30**
(2.20)
51.30**
(1.97)
-187.90
(-1.53)
Four-digit SIC industries
551
Top tercile ΔHHI
112
Middle tercile ΔHHI
305
Bottom tercile ΔHHI
134
Hoberg-Phillips industries
303
Top tercile ΔHHI
100
Middle tercile ΔHHI
140
Bottom tercile ΔHHI
63
Portfolios CAR[-1, 1] CAR[-2, 2]
256
94
65
97
256
61
133
62
152
50
64
38
1.10%***
(3.32)
1.54%***
(2.72)
0.31%
(0.13)
1.20%**
(2.43)
1.10%***
(3.32)
0.62%
(0.94)
1.41%***
(2.87)
0.90%
(1.60)
0.73%*
(1.87)
1.48%**
(1.97)
0.80%
(1.35)
-0.35%
(-0.69)
1.20%***
(2.90)
1.49%*
(1.89)
-0.39%
(-0.95)
1.98%***
(3.38)
1.20%***
(2.90)
0.34%
(0.62)
1.64%***
(2.67)
1.10%
(1.50)
1.02%**
(2.06)
0.44%
(0.63)
2.50%***
(2.97)
-0.69%
(-1.08)
38
Table 6 continued
Panel B: Merger Announcement CARs for Customers and Suppliers
B1: Based on BEA Input –
Output Tables
Customers
Portfolios CAR[-1,1] CAR[-2,2] Portfolios CAR[-1,1] CAR[-2,2]
Full sample
247
Top tercile ΔHHI
60
Middle tercile ΔHHI
128
Bottom tercile ΔHHI
59
B2: Based on Firm-level
FAS 131 Disclosures
Suppliers
0.04%
(0.65)
0.22%
(1.23)
-0.04%
(-0.14)
0.03%
(0.16)
0.07%
(1.02)
0.27%
(1.17)
-0.02%
(0.28)
0.06%
(0.39)
253
61
131
61
Customers
0.02%
(0.73)
0.22%**
(2.09)
-0.02%
(-0.43)
-0.08%
(-0.35)
0.06%*
(1.82)
0.40%***
(3.23)
-0.03%
(0.16)
-0.08%
(-0.11)
Suppliers
Portfolios CAR[-1,1] CAR[-2,2] Portfolios CAR[-1,1] CAR[-2,2]
Full sample
144
Top tercile ΔHHI
31
Middle tercile ΔHHI
81
Bottom tercile ΔHHI
32
0.41%
(1.51)
0.01%
(0.36)
0.70%
(1.23)
0.04%
(0.84)
0.68%*
(1.77)
0.13%
(0.31)
1.34%*
(1.84)
-0.45%
(0.46)
126
35
59
32
-0.23%
(-0.68)
0.96%
(0.59)
-0.93%
(-1.15)
-0.24%
(-0.34)
-0.64%
(-0.09)
-0.16%
(0.18)
-1.58%
(-0.61)
0.55%
(0.63)
39
Table 7: Revenue and Profitability Spillover Effects
Our measure of revenue spillover effects is the non-merging partner’s abnormal sales growth (ASGR),
defined as its match-adjusted change in sales growth rate from the year before to the year after the merger.
The match firm for each non-merging partner is the closest same-industry firm in terms of sales and sales
growth rate in the year prior to the merger. ASGR is winsorized at the 1st and 99th percentiles. Our measure
of profitability spillover effects is the non-merging partner’s abnormal operating income ratio (AOIR),
defined as its match-adjusted change in operating income ratio from the year before to the year after the
merger. The match firm for each non-merging partner is the closest same-industry firm in terms of sales
and operating income ratio in the year prior to the merger. Operating income ratio is operating income after
depreciation divided by sales. AOIR is winsorized at the 1st and 99th percentiles. Partner is target equals 1
for non-merging firms whose partners are targets in the merger transaction according to SDC. Transaction
in merger wave equals 1 for firms whose partners merge during an industry merger wave determined using
the methodology of Harford (2005). Regulatory event transactions equals 1 for firms in industries whose
merger waves are tied to a clear regulatory event or emergence of the internet. Test statistics are shown in
parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
No. of
Mergers
Full sample
1,126
Partner is target
428
Transaction in merger wave
399
Regulatory event transactions
63
A: Revenue (ASGR)
Mean
0.044***
(6.21)
0.071***
(3.92)
0.042***
(3.15)
0.133**
(2.35)
Median
0.028***
(4.81)
0.041***
(3.44)
0.019**
(2.24)
0.055*
(1.78)
B: Profitability (AOIR)
Mean
0.047***
(5.38)
0.042**
(2.27)
0.060***
(3.92)
0.077
(1.13)
Median
0.006***
(3.18)
0.009**
(2.15)
-0.001
(0.40)
0.017
(1.13)
40
Table 8: Evaluating the Synergy Hypothesis Using Patterns of Revenue Spillover
This table reports abnormal sales growth for portfolios of non-merging partners stratified into subsamples
based on relations between pairs of partnering firms and between pairs of merging firms. In Panel A,
partnerships and mergers are horizontal if both firms in the relationship are in the same historical 4-digit
SIC code industry in the year immediately preceding the merger. In Panel B, partnerships and mergers are
similar-product relations if both firms in the relationship operate in a similar output market in the year of
the merger or either of the two years preceding it, based on Hoberg-Phillips product similarity measures.
In Panel C, a partnership is a technology transfer partnership if it contains an explicit agreement to transfer
technology between the two partners. A merger partner is R&D intensive if its ratio of R&D expenses to
total assets exceeds the sample median. In Panel D, a high overlap partnership is one where the correlation
between the two partnering firms’ patent portfolios exceeds the sample median. A high overlap merger
partner is one where the correlation between the merging firms’ patent portfolios exceeds the sample
median. Correlations between patent portfolios are calculated following Bena and Li (2014). Results in
column (1) are for the full sample. Results in column (2) are for firms whose partners are targets in the
merger transaction. Results in column (3) are for firms whose partners merge during an industry merger
wave determined using the methodology of Harford (2005). Test statistics are shown in parentheses. *, **,
and *** indicate significance at 10%, 5%, and 1% levels, respectively. Entries in the “Difference” rows are
differences between abnormal sales growth for portfolios of non-merging partners in the two categories
above each row. Levels of significance for this row are based on one-tailed tests because the null hypotheses
are directional.
41
Table 8 continued
(1) Full sample
N
ASGR
ASGR
Mean
Median
(2) Partner is target
N
ASGR
ASGR
Mean
Median
(3) Deal in merger wave
N
ASGR
ASGR
Mean
Median
89
83
A: Horizontal relations based on SIC codes
0.065**
(2.41)
Other partnership–merger combinations
1,033 0.041***
(5.56)
Difference
0.023
(0.94)
B: Horizontal relations based on Hoberg-Phillips product similarity
Horizontal partners, horizontal mergers
Similar-product partners, similar-product
mergers
Other partnership–merger combinations
217
139
925
Difference
0.080**
(2.42)
0.024***
(4.19)
0.056
(1.42)
368
0.073**
(2.18)
0.041***
(5.32)
0.032
(1.05)
0.074**
(2.29)
0.025***
(4.10)
0.049
(1.35)
0.043***
(2.96)
0.034***
(4.01)
0.009
(0.56)
0.052***
(3.44)
0.020***
(3.29)
0.033*
(1.69)
135
0.059***
(5.44)
0.032***
(3.34)
0.027*
(1.85)
0.057***
(5.23)
0.019***
(2.61)
0.038**
(2.29)
157
58
333
0.032
(0.65)
0.070***
(3.52)
-0.039
(-0.75)
0.055
(0.67)
0.037***
(3.00)
0.019
(0.46)
-0.008
(-0.14)
0.050**
(2.42)
-0.058
(-0.95)
0.054
(0.48)
0.029*
(1.87)
0.025
(0.019)
0.066
(1.65)
0.061***
(2.98)
0.006
(0.13)
0.053*
(1.90)
0.038***
(3.09)
0.015
(0.28)
140
0.079**
(2.42)
0.043**
(1.99)
0.035
(0.93)
0.064**
(2.37)
0.028*
(1.95)
0.036
(0.96)
121
367
57
374
0.111***
(2.73)
0.035**
(2.49)
0.076*
(1.85)
0.073**
(1.99)
0.016*
(1.88)
0.057
(1.36)
0.135***
(2.68)
0.037***
(2.67)
0.098**
(2.02)
0.081**
(2.24)
0.017*
(1.90)
0.064*
(1.79)
0.045*
(1.67)
0.029*
(1.83)
0.016
(0.54)
0.062**
(2.26)
0.016
(1.62)
0.046
(1.41)
0.061***
(3.18)
0.023
(1.31)
0.039
(1.40)
0.081***
(2.77)
0.017
(1.45)
0.065
(1.32)
C: Technology-transfer partnerships
Technology-transfer partnership, R&D
intensive merger partner
Other partnership–merger combinations
421
990
Difference
351
355
D: Patent-based technology overlap
High overlap partnership, high overlap
merger partner
Other partnership–merger combinations
Difference
423
929
317
349
42
Table 9: Evaluating the Synergy Hypothesis Using Patterns of Profitability Spillover
This table reports abnormal operating profitability for portfolios of non-merging partners stratified into
subsamples based on relations between pairs of partnering firms and between pairs of merging firms. In
Panel A, partnerships and mergers are horizontal if both firms in the relationship are in the same historical
4-digit SIC code industry in the year immediately preceding the merger. In Panel B, partnerships and
mergers are similar-product relations if both firms in the relationship operate in a similar output market in
the year of the merger or either of the two years preceding it, based on Hoberg-Phillips product similarity
measures. In Panel C, a partnership is a technology transfer partnership if it contains an explicit agreement
to transfer technology between the two partners. A merger partner is R&D intensive if its ratio of R&D
expenses to total assets exceeds the sample median. In Panel D, a high overlap partnership is one where the
correlation between the two partnering firms’ patent portfolios exceeds the sample median. A high overlap
merger partner is one where the correlation between the merging firms’ patent portfolios exceeds the sample
median. Correlations between patent portfolios are calculated following Bena and Li (2014). Results in
column (1) are for the full sample. Results in column (2) are for firms whose partners are targets in the
merger transaction. Results in column (3) are for firms whose partners merge during an industry merger
wave determined using the methodology of Harford (2005). Test statistics are shown in parentheses. *, **,
and *** indicate significance at 10%, 5%, and 1% levels, respectively. Entries in the “Difference” rows are
differences between abnormal operating profitability for portfolios of non-merging partners in the two
categories above each row. Levels of significance for this row are based on one-tailed tests because the null
hypothesis is directional.
43
Table 9 continued
(1) Full sample
N
AOIR
AOIR
Mean
Median
(2) Partner is target
N
AOIR
AOIR
Mean
Median
(3) Deal in merger wave
N
AOIR
AOIR
Mean
Median
A: Horizontal relations based on SIC codes
0.096***
(2.63)
Other partnership–merger combinations
1,033 0.040***
(4.58)
Difference
0.057**
(1.90)
B: Horizontal relations based on Hoberg-Phillips product similarity
Horizontal partners, horizontal mergers
Similar-product partners, similar-product
mergers
Other partnership–merger combinations
217
139
925
Difference
0.026**
(2.58)
0.005**
(2.34)
0.021**
(2.20)
89
368
0.089**
(2.13)
0.048***
(4.95)
0.041
(1.10)
0.036***
(2.74)
0.007***
(3.09)
0.029**
(2.49)
0.068***
(2.88)
0.035***
(4.16)
0.032*
(1.61)
0.006
(1.20)
0.009***
(4.09)
-0.003
(-0.61)
135
0.070***
(3.79)
0.036***
(4.20)
0.034**
(1.95)
0.012***
(2.61)
0.006***
(2.65)
0.005
(0.87)
157
58
333
0.094
(1.62)
0.026
(1.36)
0.069
(1.33)
0.036*
(1.72)
0.007
(1.29)
0.029
(1.42)
0.087
(1.11)
0.036*
(1.82)
0.051
(0.82)
0.044
(1.51)
0.007
(0.94)
0.036
(1.60)
-0.025
(-0.54)
0.061***
(3.11)
-0.086**
(1.98)
0.000
(-0.84)
0.013***
(3.33)
-0.012**
(-2.24)
140
0.097**
(2.39)
0.014
(0.73)
0.083**
(2.15)
0.033***
(2.89)
0.002
(0.20)
0.031***
(2.61)
121
83
367
57
374
0.156**
(2.27)
0.050***
(3.18)
0.106**
(2.03)
0.020
(1.41)
-0.003
(-0.03)
0.023
(1.62)
0.073
(1.31)
0.062***
(3.67)
0.011
(0.19)
0.052
(1.51)
-0.004
(-0.19)
0.057**
(2.13)
0.101**
(2.18)
0.023
(1.61)
0.077**
(2.10)
-0.004
(0.24)
0.005
(0.83)
-0.009
(-0.33)
0.045
(1.63)
0.073***
(3.72)
-0.028
(-0.84)
-0.003
(0.00)
0.004
(0.87)
-0.007
(-0.39)
C: Technology-transfer partnerships
Technology-transfer partnership, R&D
intensive merger partner
Other partnership–merger combinations
421
990
Difference
351
355
D: Patent-based technology overlap
High overlap partnership, high overlap
merger partner
Other partnership–merger combinations
Difference
423
929
317
349
44
Table 10: Tests of the Financial Constraint Hypothesis
Non-merging partners are financially constrained if they score in the top tercile of the Whited-Wu (or
Hadlock-Pierce) financial constraints index in the year preceding the merger and financially unconstrained
if they score in the bottom tercile. Financial constraint is relaxed if a financially constrained firm’s index
score declines by greater than the sample median decline in the post-merger year. Panel A reports mean
CARs for equally-weighted portfolios of non-merging partners. Panels B and C report abnormal sales
growth rate (ASGR) and abnormal operating income ratio (AOIR) for the same portfolios. ASGR and AOIR
are calculated relative to same-industry match firms with the closest sales growth or operating income to
the non-merging partner in the year before the merger and are the match-adjusted change in each variable
from the year before to the year after the merger. Test statistics are reported in parentheses. ***, **, and *
indicate significance at the 1%, 5%, and 10% levels, respectively. Observations for “Constraints relaxed”
and “Constraints not relaxed” subsamples add up to more than observations for “Financially constrained”
because some merging firms have separate non-merging partners that fall into both subsamples in the same
year.
Whited – Wu Index
Panel A: CARs
Financially constrained
Sample
466
Sample
Mean
Median
533
0.037**
(2.33)
0.038*
(1.94)
0.004
(0.18)
0.038***
(3.90)
Mean
Constraints not relaxed
362
Financially unconstrained
903
Constraints relaxed
311
Constraints not relaxed
348
Financially unconstrained
689
Panel C: AOIR
Financially constrained
Sample
0.87%**
(2.19)
1.03%*
(2.20)
0.47%
(0.43)
0.01%
(0.59)
364
Financially constrained
CAR[-1, +1] CAR[-2, +2]
0.27%**
(2.14)
0.87%**
(2.72)
-0.05%
(0.31)
0.06%
(1.01)
622
Constraints relaxed
Panel B: ASGR
Hadlock – Pierce Index
Sample
533
Constraints relaxed
311
Constraints not relaxed
348
Financially unconstrained
689
***
0.075
(3.24)
0.040
(1.43)
0.078**
(2.29)
0.020***
(2.79)
CAR[-1, +1] CAR[-2, +2]
0.14%
(1.20)
0.74%*
(1.92)
-0.69%
(-0.88)
0.07%
(0.97)
1.03%**
(2.06)
1.53%**
(2.28)
-0.42%
(0.13)
0.05%
(0.67)
Sample
Mean
Median
0.052***
(2.90)
0.038***
(2.70)
0.021
(0.54)
0.020***
(3.52)
393
-0.029
(-1.27)
0.066**
(2.13)
-0.126**
(-4.11)
0.040***
(5.11)
-0.026
(-1.29)
0.086**
(2.39)
-0.109***
(-3.52)
0.025***
(4.63)
Median
Sample
Mean
Median
393
0.023
(0.73)
0.085**
(2.07)
0.019
(0.41)
0.038***
(4.79)
-0.005
(-0.37)
0.007
(0.85)
-0.056
(-1.43)
0.007***
(3.95)
**
0.016
(2.45)
0.019*
(1.89)
-0.003
(0.82)
0.004**
(2.38)
251
236
989
246
250
769
246
250
769
45
Table 11: Tests of the Takeover Probability Hypothesis
Non-merging partners are high takeover probability firms if they are in the top quartile of predicted
probabilities of becoming a target based on the number of their collaborative partners involved in a merger
in the preceding three years. They are low takeover probability firms if they are in the bottom quartile.
Panel A reports mean CARs for equally-weighted portfolios of non-merging partners. Panels B and C report
abnormal sales growth rate (ASGR) and abnormal operating income ratio (AOIR) for the same portfolios.
ASGR and AOIR are calculated relative to same-industry match firms with the closest sales growth or
operating income to the non-merging partner in the year before the merger and are the match-adjusted
change in each variable from the year before to the year after the merger. Test statistics are reported in
parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
Panel A: CARs
Sample
CAR[-1, +1]
CAR[-2, +2]
High takeover probability
521
Low takeover probability
626
0.16%*
(1.73)
0.07%
(-0.10)
0.20%
(1.47)
0.18%
(0.37)
Sample
Mean
Median
High takeover probability
463
Low takeover probability
391
0.0019
(0.13)
0.0229
(1.39)
0.0324
(1.30)
0.0237
(1.60)
Sample
Mean
Median
High takeover probability
463
Low takeover probability
391
0.0416***
(3.84)
0.0416
(1.64)
0.0136***
(3.34)
0.0079
(1.38)
Panel B: ASGR
Panel C: AOIR
46
Table 12: Explaining Merger Spillovers – Regression Results
The dependent variable is the 3-day merger announcement cumulative abnormal return to the non-merging
partner in a collaborative partnership whose partner merges with a third firm. Firm size is the natural
logarithm of sales. Tangible assets is the ratio of net PPE to total assets. Leverage is long-term debt plus
short-term debt, divided by total assets. R&D is research and development expenditures divided by total
assets. Capex is capital expenditure divided by total assets. ROA is ratio of operating income before
depreciation to total assets. Each variable is for the non-merging partner and is measured at the end of the
fiscal year preceding the merger. Joint venture equals 1 for partnerships listed in SDC as JVs, 0 for others.
Duration is the natural log of the number of years since partnership formation. Supplier equals 1 if SDC
indicates that the partnership provides for a supplier agreement, 0 otherwise. Partnerships is the natural log
of the number of bilateral partnerships in which the non-merging partner is a member. Merging partners is
the natural log of the number of the non-merging partner’s bilateral partners involved in a merger. H-H
relation equals 1 for non-merging partners in horizontal partnerships whose partners enter a horizontal
merger with a third firm, 0 otherwise. Tech-transfer, high R&D equals 1 for non-merging partners in
technology transfer partnerships whose partners merge with a R&D intensive third firm, 0 otherwise.
Technology overlap equals 1 for non-merging partners whose patent portfolios are highly correlated with
their partners’ when those partners merge with another firm whose patent portfolios are highly correlated
with theirs, 0 otherwise. Financial constraints relaxed equals 1 for non-merging partners in the top tercile
of the Whited-Wu index in the pre-merger year whose index scores declined by greater than the median
decline in the post-merger year, 0 otherwise. High takeover probability equals 1 for non-merging partners
in the top quartile of predicted takeover probabilities based on the number of partner firms in a merger over
the preceding 3 years, 0 otherwise. Test statistics based on robust standard errors clustered at the firm level
are in parentheses. Each regression includes year fixed effects. Continuous variables are winsorized at the
1st and 99th percentiles. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
47
Table 12 continued
Firm size
Tangible assets
Leverage
R&D
Capex
ROA
(1)
(2)
(3)
(4)
(5)
(6)
-0.0005
(-1.14)
0.0183***
(2.68)
0.0035
(0.66)
0.0217**
(2.14)
-0.0362**
(-1.98)
0.0008
(0.13)
-0.0005
(-0.84)
0.0186***
(2.67)
0.0031
(0.57)
0.0213**
(2.07)
-0.0376**
(-2.06)
0.0011
(0.17)
0.0005
(0.23)
-0.0017
(-1.58)
0.0012
(0.29)
-0.0006
(-0.34)
0.0012
(0.44)
-0.0003
(-0.63)
0.0174**
(2.55)
0.0037
(0.68)
0.0174*
(1.72)
-0.0370**
(-2.05)
0.0002
(0.04)
0.0008
(0.32)
-0.0018*
(-1.71)
0.0018
(0.43)
-0.0011
(-0.56)
0.0013
(0.46)
0.0073**
(1.97)
0.0041*
(1.73)
0.0002
(0.09)
-0.0004
(-0.67)
0.0183***
(2.68)
0.0032
(0.58)
0.0200**
(2.00)
-0.0364**
(-2.01)
0.0013
(0.21)
0.0005
(0.22)
-0.0017
(-1.57)
0.0012
(0.30)
-0.0006
(-0.31)
0.0011
(0.40)
-0.0004
(-0.74)
0.0194***
(2.82)
0.0028
(0.51)
0.0206**
(2.04)
-0.0376**
(-2.08)
0.0006
(0.10)
0.0005
(0.23)
-0.0017
(-1.58)
0.0013
(0.31)
-0.0006
(-0.31)
0.0011
(0.39)
-0.0001
(-0.21)
0.0181***
(2.64)
0.0035
(0.64)
0.0175*
(1.73)
-0.0359**
(-1.99)
-0.0002
(-0.03)
0.0008
(0.33)
-0.0019*
(-1.71)
0.0018
(0.44)
-0.0012
(-0.61)
0.0012
(0.44)
0.0071*
(1.92)
0.0042*
(1.79)
0.0005
(0.19)
0.0024
(0.63)
0.0032
(1.32)
0.008
6,352
Joint venture
Duration
Supplier
Partnerships
Merging partners
H-H relation
Tech-transfer, high R&D
Technology overlap
0.0026
(0.68)
Financial constraints relaxed
0.006
0.006
0.007
0.006
0.0033
(1.35)
0.006
6,355
6,352
6,352
6,352
6,352
High takeover probability
R-squared
Observations
48