1. Empirical investigation of the divestment options embedded in takeover premia paid by U.S. public acquirers
P. Alexakis, National & Kapodistrian University of Athens
M. I. Chondrokouki, Athens University of Economics & Business
A. E. Tsekrekos, Athens University of Economics & Business
14th Summer School in Stochastic Finance

2. Introduction
Global investment in mergers and acquisitions has reached unprecedented levels in recent years (Barkema and Schijven, 2008). Paralleling this practical importance, M&A activity has increasingly become a focus of study in academic research.
Between , 864 papers (i.e. 48 per year on average) have “merger, merge, acquisition, acquire” or “M&A” as a keyword in the Top-20 Econ/Fin/Acc/Man/Soc journals (Haleblian et al., 2009).
A strong and recurring research finding that has been established early on is that on average,
“Acquirers pay too much”: Acquisitions do not enhance acquiring firm value (Asquith, 1983; Jarrell and Poulsen, 1989; Malatesta, 1983; Travlos, 1987)
“Target shareholders fare well”: targets accounted for the majority of those gains (Asquith & Kim, 1982; Datta et al., 1992; Hansen and Lott, )

3. Introduction
M&As are often treated as independent events , and the premium paid is treated as-if it is the solution of an “one-shot game”:
Barkema and Schijven (2008): Most M&As are actually a component of a broader acquisition strategy. They demonstrated that the performance implications of a single acquisition are dependent on that acquisition’s position within the acquirers’ acquisition sequence. Thus, there are acquirer gains that are often overlooked when examining single acquisition events.
Moreover, are divestitures/divestments of acquired assets part of an optimal acquisition strategy/sequence?
Haleblian et al (2009): There is potential value in examining whether, how, and under what conditions firms can extract valuable resources from acquisition assets that are subsequently divested. Restructuring often leads to the divestiture of capital assets, product lines, and businesses that are considered to be strategic misfits, and thus, creates higher synergy benefits.

4. Introduction
Several papers have investigated the view that acquisition and divestment decisions are potentially a sequence of deals in a firm’s strategy (e.g. Mitchell and Lehn, 1990; Kaplan and Weisbach, 1992; Allen et al., 1995; Fluck and Lynch, 1999; Hanson and Song, 2003; Marquette and Williams, 2007).
Two alternative explanations:
The divestiture is a correction of a previous acquisition that was mistake.
The divestiture was the original intent prior to the acquisition and the acquiring firm extracts value from the acquisition- divestiture combination.
A recent stream of theoretical papers has employed the parallel between acquisitions and ‘real options’ to examine the optimal timing and terms of takeovers
Lambrecht (2004); Lambrecht and Myers (2005, 2007); Alvarez and Stenbacka (2006) among others.

5. Example
In May 2002, KONE (a Finnish engineering and services company, 4th larger manufacturer of elevators, escalators and cargotec worldwide) acquired PARTEK at a price of 1.45 billion Euros.
PARTEK was a conglomerate with a wide range of activities (container and cargo handling, production of forestry and agricultural machinery, mining technology, etc.).
Late in 2003, KONE divested most of the acquired activities outside its core business, at an estimated cash inflow of 1.1 billion Euros.
The container and cargo handling activities that were maintained by KONE yielded synergy value benefits in the region of 960 million Euros.

6. Motivation
Takeovers promote economic efficiency if the bidder is able to reorganize the assets of the target so as to create synergy gains (i.e. exploitation of complementarities or economies of scale/scope).
Acquisitions are typically sizeable, (largely) irreversible investment decisions characterized by substantial uncertainty.
Acquisitions create ‘real options’ whose value and optimal ‘exercise’ incorporate a substantial degree of implementation uncertainty regarding:
The timing and magnitude of the synergy benefits that might materialize (time and effort needed to make different corporate cultures, governance systems and product brands fit together)
The magnitude of acquired assets with no synergetic benefits, and the time and effort needed to search for their most efficient use (including the possibility of their divestment or liquidation)
The lower this uncertainty, the more willing the firm to undertake the acquisition.

7. Contribution
We provide empirical evidence in favour of the predictions of a real options model (Alvarez and Stenbacka, 2006) that considers “acquisitions and subsequent divestments of non-core operations” as a compound investment opportunity, with both ‘legs’ of the transaction adding value to the acquiring firm.
Empirical tests of existing models in the ‘real options’ literature have been few and primarily focus on natural resource investments (Paddock et al., 1988; Moel and Tufano, 2002) and urban land valuation (Quigg, 1993; Grovenstein et al., 2011).
Many researchers model several aspects of M&A deals in a ‘real options’ framework (e.g. Lambrecht, 2004; Lambrecht and Myers, ), but our focus is narrow: We only test the predictions of one specific theoretical model, not for ‘real options’ considerations in the timing and terms of M&As in general.
We add to the existing empirical literature on M&As that considers sequences of “acquisitions and subsequent divestments”; our approach differs in that we directly test a theoretical model that predicts optimal timing behavior to explain such sequences, instead of taking an empirically-centered view of looking for empirical evidence in line with any of the two proposed explanations.

8. Methodological Approach
Theoretical model proposed by Alvarez and Stenbacka (2006)
Three firms:
Acquiring firm, A: faces the opportunity to acquire a target firm,
target firm, T, and
outside firm, O: the firm to which the acquiring firm A (after the acquisition of T) can sell out any acquired activities that are outside its core business.
It is assumed that post-acquisition of firm T, firm A has the option to divest the activities outside its core business to firm O that can make a more efficient use of these non-core resources.
The acquiring firm’s initial option to buy out the target firm is a compound real option: the acquisition option exercise gives firm A another option, to sell out the non-core activities of the target firm to another firm.

9. Methodological Approach
The firms operate in an environment where the market uncertainty is captured by an exogenous state variable following a geometric Brownian motion
𝑑 𝑋 𝑡 =𝜇 𝑋 𝑡 𝑑𝑡+𝜎 𝑋 𝑡 𝑑 𝑊 𝑡 , 𝑋 0 =𝑥,
with 𝜇 and 𝜎>0 exogenous constants and 𝑊 𝑡 a standard Brownian motion.
Prior to the takeover, the business activities of firms A and T generate profit flows 𝜋 𝛢 (𝑥) and 𝜋 𝑇 𝑥 , respectively.
Through the takeover, firm A acquires the right to reorganize the assets of T so as to create synergy gains of the form
𝜋 𝑐 𝐴 𝑥 + 𝜋 𝑑 𝐴 𝑥 ≥ 𝜋 𝛢 𝑥 + 𝜋 𝑇 𝑥 (1)
where the post-acquisition profit flow is decomposed into 𝜋 𝑐 𝐴 𝑥 , a component associated with the core business of firm A, and 𝜋 𝑑 𝐴 𝑥 , a component associated with the activities outside the core competence of firm A, and which is optimally divested.

10. Methodological Approach
The outside firm O can make better use of the non-core resources, which is formalised by assuming that
𝜋 𝑂 𝑥 ≥𝜋 𝑑 𝐴 (𝑥) (2)
Inequalities (1) and (2) imply that the acquiring firm A can capture synergy gains via the acquisition, and in addition to the primary benefit it receives an embedded option that represents potential disinvestment gains.
Crucial assumption: after the acquisition of T, there is a consolidation phase of random length that firm A must endure before the synergies captured by (1) can be realized.
The randomness in the length of this consolidation phase (implementation uncertainty), captures the uncertainty of whether corporate cultures, governance systems and/or product brands of firms A and T are compatible enough so that the synergy gains in (1) can be quickly realized.
Similarly, there is a search phase of random length before the divestment gains of inequality (2) can be realized, since firm A must find an outside firm O that can make more efficient use of the non-core assets.

11. Methodological Approach
The solution
Expected cumulative present value of cash flow 𝑓(𝑥)
𝑅 𝑟 𝑓 𝑥 ≔ 𝑬 𝑥 0 +∞ 𝑒 −𝑟𝑡 𝑓( 𝑋 𝑡 ) 𝑑𝑡
Takeover price of firm T by firm A that optimally anticipates the subsequent divestment of non-core assets
𝑃 ∗ 𝑥 = 𝑅 𝑟 𝜋 𝛵 𝑥 + 1− 𝛽 𝛢 𝛽 𝑑 𝑅 𝑟 𝜋 𝑂 − 𝜋 𝑑 𝛢 𝑥 − 𝑅 𝑟+ 𝜆 2 𝜋 𝑂 − 𝜋 𝑑 𝛢 𝑥 + 𝑅 𝑟 𝜋 𝑐 𝛢 + 𝜋 𝑑 𝛢 𝑥
− 𝑅 𝑟+ 𝜆 1 𝜋 𝑐 𝛢 + 𝜋 𝑑 𝛢 − 𝜋 𝛢 − 𝜋 𝛵 𝑥 − 𝛽 𝑑 𝜆 2 𝑅 𝑟+ 𝜆 1 𝑅 𝑟+ 𝜆 2 𝜋 𝛰 − 𝜋 𝑑 𝐴 𝑥 − 𝑅 𝑟 𝜋 𝛵 𝑥 − 𝑅 𝑟 𝜋 𝐴 𝑥
𝜕𝑃 ∗ 𝜕 𝜆 1 𝑥 =− 1− 𝛽 𝛢 𝜕 𝑅 𝑟+ 𝜆 1 𝜋 𝑐 𝛢 + 𝜋 𝑑 𝛢 − 𝜋 𝛢 − 𝜋 𝛵 𝜕 𝜆 1 𝑥
− 𝛽 𝑑 𝜆 2 1− 𝛽 𝛢 𝜕 𝑅 𝑟+ 𝜆 1 𝑅 𝑟+ 𝜆 2 𝜋 𝛰 − 𝜋 𝑑 𝐴 𝜕 𝜆 1 𝑥 >0
𝜕𝑃 ∗ 𝜕 𝜆 2 𝑥 =− 𝛽 𝑑 𝜆 1 1− 𝛽 𝛢 𝜕 𝑅 𝑟+ 𝜆 1 𝑅 𝑟+ 𝜆 2 𝜋 𝛰 − 𝜋 𝑑 𝐴 𝜕 𝜆 2 𝑥 >0
𝜕𝑃 ∗ 𝜕𝜎 𝑥 ≶0

12. Methodological Approach
Alvarez and Stenbacka (2006) predict that a firm will find it more valuable to follow a compound strategy of acquisition, followed by divestment of non-core business resources if
it possesses higher bargaining power and/or
anticipates lower implementation uncertainty in realizing the benefits of the strategy.
The empirical testing hinges on:
correctly accounting for the non-randomness of firms following such a compound (acquisition and divestment) strategy and
on proxies for the acquiring firm’s bargaining power and the implementation uncertainty time lags that capture the uncertainty in implementing the strategy.

13. PREMIUM i = β 0 + β 1 X i + β 2 FLIP i + e i (3)
Empirical Framework
The appropriate empirical methodology should control for the inherent endogeneity of the choice to follow a (compound) “takeover and subsequent divestment of non-core assets” firm strategy.
Treatment selection model (Campa and Kedia, 2002):
PREMIUM i = β 0 + β 1 X i + β 2 FLIP i + e i (3)
where X i is a vector of independent variables affecting the takeover premium, FLIP i is a dummy variable that takes the value of 1 if the takeover was followed by a divestiture and 0 otherwise, β= β 0 , β 1 , β 2 is a vector of parameters to be estimated and e i is an error term.
We assume that a firm’s decision to implement this compound strategy is determined by
FLIP i ∗ =β Z i + μ i (4)
FLIP i ∗ =1, if FLIP i ∗ >0
FLIP i ∗ =0, if FLIP i ∗ <0,
where FLIP i ∗ is an unobserved latent variable, Ζ i is a vector of independent (instrumental) variables affecting the choice of the FLIP strategy (i.e. the firm’s latent choice/regime) and μ i is an error term.

14. Empirical Framework
Following Campa and Kedia (2002), we jointly estimate equations (3) and (4) in a simultaneous equations framework.
Key determinants of a firm’s decision to implement a 𝐹𝐿𝐼𝑃 or not (and thus determine the optimal premium to offer):
The acquirer’s bargaining power
The implementation uncertainty involved in realizing the benefits of the choice
The economic uncertainty in the target’s profit generating ability
In equation (4) the instrumental variables affecting the latent choice should include proxies for the acquirer’s bargaining power ( 𝐵𝐴𝑅𝐺𝐴𝐼𝑁 𝑖 𝐴 ), for the strategy’s implementation uncertainty ( 𝐼𝑈𝑃 𝑖 ) and for the target’s economic uncertainty ( 𝑈𝑁𝐶𝐸𝑅𝑇𝐴𝐼𝑁𝑇𝑌 𝑖 𝑇 ):
𝑍 𝑖 = 𝐵𝐴𝑅𝐺𝐴𝐼𝑁 𝑖 𝐴 , 𝐼𝑈𝑃 𝑖 , 𝑈𝑁𝐶𝐸𝑅𝑇𝐴𝐼𝑁𝑇𝑌 𝑖 , 𝐶𝑂𝑁𝑇𝑅𝑂𝐿𝑆 𝑖 (5)

15. Data
All merger and acquisition announcements from January 1999 to December 2013 for U.S. listed firms [Reuters Thomson One]
Divestitures by acquiring firms subsequent to the acquisitions for the same period are considered [Reuters Thomson One]
Stock market data and accounting data [CRSP, Compustat]
Restrictions:
We exclude minority stake purchases, acquisitions of remaining interest, leveraged buyouts, repurchases, self-tenders, exchange offers and joint-ventures.
We exclude divestitures due to bankruptcy, ‘fire sales’, spin-offs to shareholders and divestitures required by regulatory authorities.
We require data availability for variables (presented below).
Event (“flip”): An elaborate transaction which consists of an acquisition and a subsequent divestiture of part of the acquired assets (Marquette and Williams, 2007). SEC EDGAR filings (8-K, 10- Q) were employed to manually link acquisitions-divestitures.

16. Data Event identification procedure All Acquisitions 20,510
All Divestitures
19,682
Potential Flips
Criterion 1:
Acquisition Acquirer is Divestiture Target
Criterion 2:
Acquisition Acquirer is Divestiture Target Parent
Criterion 3:
Acquisition Target is Divestiture Target
Criterion 4:
Acquisition Target is Divestiture Target Parent
1,730
3,482
181
393
Actual Flips 18 71 22
107

17. Example Alcoa acquires Cordant Technologies
Extract of SEC EDGAR filing for the initial acquisition:
“PITTSBURGH and SALT LAKE CITY--March 14, 2000
Alcoa Inc. and Cordant Technologies Inc. today announced they have entered into a definitive agreement under which Alcoa will acquire all outstanding shares of Cordant for $57.00 per share payable in cash.
Alcoa will commence the transaction with a cash tender offer for 100%, but not less than a majority, of the outstanding Cordant shares on a fully diluted basis. If a majority of the outstanding shares are purchased in the tender offer, any remaining Cordant shares will be exchanged for cash in the amount of $57.00 per share in a merger of Cordant and Alcoa's acquisition subsidiary.
The transaction is valued at approximately $2.9 billion based on 40 million fully diluted shares of Cordant common stock on March 13, 2000 and the assumption of $685 million in debt…”
Extract of SEC EDGAR filing for the subsequent divestiture:
“Pittsburgh and Minneapolis, Jan. 31, Alcoa Inc. and ATK (Alliant Techsystems Inc.) announced today that they have reached a definitive agreement under which ATK will acquire Alcoa's Thiokol Propulsion business for $685 million in cash. Thiokol, a leading supplier of rocket propulsion systems for space launch vehicles with sales of approximately $570 million, became part of Alcoa with the acquisition of Cordant Technologies in May ATK is a supplier of aerospace and defense products, including conventional munitions and propulsion rocket motors.
The transaction, which has received all necessary corporate approvals of both companies, is subject to customary regulatory approvals. It is expected to close by the end of the second quarter of 2001.
Commenting on the announcement, Alcoa Chairman and CEO Alain Belda said, "This transaction is beneficial to all parties involved. It delivers value to Alcoa shareholders, and Thiokol becomes a strategic part of ATK, a company in the propulsion business and positioned to help Thiokol realize the full value of its products and technologies."…”

18. Data
Control sample: It comprises of firms that made acquisitions but did not make subsequent divestitures during the same period ( ).
All acquisitions that were not followed by divestitures (no matching).
Propensity score matching technique: We perform matching according to the year the acquisition took place, the number of the acquirer’s and the target’s 4-digit SIC codes, and dummies that indicate whether the acquirer and the target are conglomerates.
Caliper matching algorithm: It specifies a maximum propensity score distance (‘caliper’) by which a match can be made (0.01).
Nearest neighbour matching algorithm: Each acquisition from the original sample is matched with the 3 nearest (closest in terms of propensity score) acquisitions of the control sample.

19. Variables Endogenous treatment equation:
Dependent variable, 𝐹𝐿𝐼𝑃 𝑖 : dummy variable that takes the value of 1 if the acquisition was followed by a divestiture, 0 otherwise.
Bargaining power, 𝐵𝐴𝑅𝐺𝐴𝐼𝑁 𝑖 𝐴 , proxies:
Acquirer’s size, 𝑆𝐼𝑍𝐸 𝑖 𝐴 (Ahern, 2012): logarithm of total assets (predicted + sign).
Asymmetry of information, 𝐼𝐴 𝑖 𝐴 , regarding the acquisition target (Mantecon, 2008; Chung and Zhang, 2014): bid-ask spread of the target’s stock price (predicted + sign).
Implementation uncertainty, 𝐼𝑈𝑃 𝑖 , proxies:
Number of days, 𝐿𝑛 𝐷𝐴𝑌𝑆 𝑖 : One plus the logarithm of the number of days between an acquisition’s announcement and effective dates (predicted – sign).
Conglomerate target, 𝑁𝑜 𝑜𝑓 𝑆𝐼𝐶𝑠 𝑖 𝑇 : the number of the target’s 4-digit SIC codes (predicted + sign).
Economic uncertainty in the target’s profit generating ability, 𝑈𝑁𝐶𝐸𝑅𝑇𝐴𝐼𝑁𝑇𝑌 𝑖 𝑇 , proxy:
Historic volatility of the target’s stock price returns, 𝑉𝑜𝑙 𝑖 𝑇 : calculated over the −100,−60 day-window (and different time windows) relative to announcement day (predicted +/- sign).

20. Variables Endogenous treatment equation (control variables):
Acquirer’s leverage, 𝐿𝐸𝑉 𝑖 𝐴 (Graham and Harvey, 2001; Bancel and Mittoo, 2004; Brounen et al., 2004; Marchica and Mura, 2010; Yung et al., 2015): ratio of long-term debt to total assets. (predicted + sign).
Same industry, 𝑆𝑎𝑚𝑒𝑆𝐼𝐶𝑠 𝑖 (Porter, 1987; Ravenscraft and Scherer, ; and Kaplan and Weisbach, 1992): dummy variable that takes the value of 1 if the acquirer and the target belong to the same industry, 0 otherwise (predicted – sign).
Conglomerate acquirer, 𝐶𝑜𝑛𝑔𝑙𝑜𝑚 𝑖 𝐴 (Porter, 1987): dummy variable that takes the value of 1 if the acquirer is a conglomerate, 0 otherwise (predicted + sign).
Acquirer’s ratio of 𝐶𝐹𝑂𝑡𝑜𝑇𝐴 𝑖 𝐴 (Jensen, 1986): Operating cash flows to total assets.

21. Variables Premium equation:
Dependent variable, 𝑃𝑅𝐸𝑀𝐼𝑈𝑀 𝑖 (Schwert, 1996): the percentage difference between the final acquisition price and the share price of the acquired target 102 days (and different time windows) before the announcement of the acquisition.
𝑅𝑢𝑛𝑈𝑝 𝑖 (Schwert, 1996; Dionne et al., 2010): the cumulative abnormal return of the target’s share price over a one-month period (and different time windows) before the announcement (predicted + sign).
Target’s size, 𝑆𝐼𝑍𝐸 𝑖 𝑇 (Comment and Schwert, 1995; Schwert, 2000): logarithm of total assets (predicted – sign).
Target’s past performance, 𝑆𝐺𝑅 𝑖 𝑇 (Schwert, 2000; Dionne et al., 2010): sales growth (predicted +/- sign).
Cash payment, 𝐶𝑎𝑠ℎ 𝑝𝑒𝑟𝑐𝑒𝑛𝑡 𝑖 (Eckbo and Langohr, 1989; Slusky and Caves, ; Comment and Schwert, 1995 and Betton et al., 2008): percentage of cash used in the acquisition (predicted + sign).
Number of bidders, 𝑁𝑜 𝑜𝑓 𝑏𝑖𝑑𝑑𝑒𝑟𝑠 𝑖 (Flanagan and O’Shaughnessy, 2003; Slusky and Caves, 1991; Gondhalekar et al., 2004): number of actual bidders involved in the initial acquisition (predicted + sign).

22. Empirical Results: Endogenous treatment equation – no matching
Endogenous treatment: 𝐹𝐿𝐼𝑃 𝑖 =1 or 𝐹𝐿𝐼𝑃 𝑖 =0 is the dependent variable
 Variable
Pred. Sign
Specification (1)
Coefficient
Specification (2)
Coefficient 
Specification (3)
Specification (4)
Intercept
-1.442***
(-5.40)
-1.872***
(-6.58)
-1.345***
(-5.89)
-1.113***
(-6.33)
Economic uncertainty
𝑉𝑜𝑙 𝑖 𝑇
+/-
***
(-7.17)
***
(-6.03)
-7.013***
(-2.94)
-5.280**
(-2.27)
Acquirer’s bargaining power
𝑆𝐼𝑍𝐸 𝑖 𝐴 +
0.040**
(2.08)
0.078***
(3.49)
𝐼𝐴 𝑖 𝐴
11.855***
(5.64)
9.871***
(4.91)
Implementation uncertainty
𝐿𝑛 𝐷𝐴𝑌𝑆 𝑖 -
-0.173***
(-4.53)
-0.110***
(-3.84)
𝑁𝑜 𝑜𝑓 𝑆𝐼𝐶𝑠 𝑖 𝑇
0.109***
(4.03)
0.071***
(3.54)
Controls
𝐶𝑜𝑛𝑔𝑙𝑜𝑚 𝑖 𝐴  +
-0.096
(-0.87)
0.004
(0.03)
0.056
(0.50)
0.116
(0.88)
𝐿𝐸𝑉 𝑖 𝐴 + 
1.061***
(3.74)
1.356***
(4.54)
1.096***
(3.76)
1.262***
(4.16)
𝐶𝐹𝑂𝑡𝑜𝑇𝐴 𝑖 𝐴
0.442
(1.04)
0.644
(1.51)
0.816*
(1.71)
0.934**
(2.11)
𝑆𝑎𝑚𝑒𝑆𝐼𝐶𝑠 𝑖  -
0.046
(0.70)
0.109
(1.60)
-0.024
(-0.37)
0.022
(0.33)
Year Indicators
Yes 
Wald 𝜒 2 test for:
𝐻 0 : 𝜌 𝜇𝑒 =0
24.86###
[0.000]
25.95###
21.37###
19.45###
No. Obs.
1,518
1,551
1,486
1,517

23. Empirical Results: Premium equation – no matching
Premium equation ( 𝐹𝐿𝐼𝑃 𝑖 is endogenous): PREMIUM is the dependent variable
 Variable
 Pred. Sign
Specification (1)
Coefficient
Specification (2)
Specification (3)
Specification (4)
Intercept
0.644***
(8.17)
0.749***
(9.10)
0.725***
(9.05)
0.768***
(9.44)
𝑅𝑢𝑛𝑈𝑝 𝑖 +
0.648***
(6.58)
0.642***
(6.57)
0.653***
(6.60)
0.657***
(6.66)
𝑆𝐼𝑍𝐸 𝑖 𝑇 -
-0.049***
(-5.73)
-0.059***
(-6.76)
(-6.93)
-0.063***
(-7.53)
𝐹𝐿𝐼𝑃 𝑖
0.809***
(10.99)
0.780***
(11.01)
0.832***
(10.63)
0.796***
(9.65)
𝑆𝐺𝑅 𝑖 𝑇
 +/-
-0.056**
(-2.23)
-0.055**
(-2.24)
-0.023
(-0.85)
(-0.86)
𝐶𝑎𝑠ℎ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡 𝑖  +
0.001**
(2.57)
(2.44)
(2.59)
(2.47)
𝑁𝑜 𝑜𝑓 𝐵𝑖𝑑𝑑𝑒𝑟𝑠 𝑖
0.005
(0.14)
0.007
(0.19)
-0.003
(-0.08)
0.000
(0.01)
Year Indicators
Yes
𝐿𝑜𝑔𝐿
Wald 𝜒 2 test for: 𝐻 0 : all coefs=0
336.85###
[0.000]
337.88###
320.68###
298.87###
No. Obs.
1,518
1,551
1,486
1,517
***, **, * denote significance at the 1%, 5% and 10% level, respectively, ### denotes significance at the 1% level.

24. Empirical Results: Endogenous treatment equation – caliper matching
Endogenous treatment: 𝐹𝐿𝐼𝑃 𝑖 =1 or 𝐹𝐿𝐼𝑃 𝑖 =0 is the dependent variable
 Variables
Pred. Sign
Specification (1)
Coefficient
Specification (2)
Coefficient 
Specification (3)
Specification (4)
Intercept
-1.415***
(-5.17)
-1.843***
(-6.46)
-1.325***
(-5.77)
-1.121***
(-6.29)
Economic uncertainty
𝑉𝑜𝑙 𝑖 𝑇
+/-
***
(-6.99)
***
(-5.49)
-7.004***
(-2.86)
-5.264***
(-2.27)
Acquirer’s bargaining power
𝑆𝐼𝑍𝐸 𝑖 𝐴 +
0.043**
(2.19)
0.076***
(3.35)
𝐼𝐴 𝑖 𝐴
11.406***
(5.45)
10.028***
(5.06)
Implementation uncertainty
𝐿𝑛 𝐷𝐴𝑌𝑆 𝑖 -
-0.180***
(-4.61)
-0.117***
(-4.02)
𝑁𝑜 𝑜𝑓 𝑆𝐼𝐶𝑠 𝑖 𝑇
0.085***
(3.41)
0.081***
(3.28)
Controls
𝐶𝑜𝑛𝑔𝑙𝑜𝑚 𝑖 𝐴 + 
-0.099
(-0.91)
-0.037
(-0.27)
0.056
(0.48)
0.109
(0.82)
𝐿𝐸𝑉 𝑖 𝐴  +
1.054***
(3.71)
1.314***
(4.19)
1.117***
(3.76)
1.248***
(4.13)
𝐶𝐹𝑂𝑡𝑜𝑇𝐴 𝑖 𝐴
0.503
(1.17)
0.712
(1.56)
0.831*
(1.74)
1.005**
(2.26)
𝑆𝑎𝑚𝑒𝑆𝐼𝐶𝑠 𝑖 - 
0.048
(0.73)
0.086
(1.19)
-0.020
(-0.31)
0.016
(0.24)
Year Indicators
Yes
Yes 
Wald 𝜒 2 test for:
𝐻 0 : 𝜌 𝜇𝑒 =0
25.75###
[0.000]
24.01###
20.63###
20.65###
No. Obs.
1,474
1,449
1,466

25. Empirical Results: Premium equation – caliper matching
Premium equation ( 𝐹𝐿𝐼𝑃 𝑖 is endogenous): PREMIUM is the dependent variable
  Variable
Pred. Sign
Specification (1)
Coefficient
Specification (2)
Specification (3)
Specification 4)
Intercept
0.681***
(8.32)
0.803***
(9.56)
0.727***
(9.03)
0.777***
(9.43)
𝑅𝑢𝑛𝑈𝑝 𝑖 +
0.642***
(6.42)
0.635***
(6.11)
0.654***
(6.50)
0.662***
(6.56)
𝑆𝐼𝑍𝐸 𝑖 𝑇 -
-0.052***
(-5.95)
-0.067***
(-7.22)
-0.060***
(-6.90)
-0.065***
(-7.44)
𝐹𝐿𝐼𝑃 𝑖
0.824***
(11.46)
0.801***
(10.67)
0.834***
(10.53)
0.818***
(10.30)
𝑆𝐺𝑅 𝑖 𝑇
 +/-
-0.056**
(-2.11)
-0.031
(-1.13)
-0.024
(-0.91)
-0.021
(-0.79)
𝐶𝑎𝑠ℎ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡 𝑖  +
0.001**
(2.39)
(2.06)
(2.43)
(2.32)
𝑁𝑜 𝑜𝑓 𝐵𝑖𝑑𝑑𝑒𝑟𝑠 𝑖
0.007
(0.18)
0.001
(0.03)
0.003
(0.07)
0.002
(0.05)
Year Indicators
Yes
𝐿𝑜𝑔𝐿
Wald 𝜒 2 test for: 𝐻 0 : all coefs=0
341.16###
[0.000]
321.46###
318.75###
308.83###
No. Obs.
1,474
1,449
1,466
***, **, * denote significance at the 1%, 5% and 10% level, respectively, ### denotes significance at the 1% level.

26. Empirical Results: Endogenous treatment equation – nearest neighbour
Endogenous treatment: 𝐹𝐿𝐼𝑃 𝑖 =1 or 𝐹𝐿𝐼𝑃 𝑖 =0 is the dependent variable
 Variable
Pred. Sign 
Specification (1)
Coefficient
Specification (2)
Coefficient 
Specification (3)
Specification (4)
Intercept
-1.113***
(-3.56)
-1.860***
(-5.65)
-1.362***
(-4.12)
-0.920***
(-3.68)
Economic uncertainty
𝑉𝑜𝑙 𝑖 𝑇
+/-
***
(-4.26)
-8.694***
(-3.43)
-6.632*
(-1.95)
-6.362
(-1.61)
Acquirer’s bargaining power
𝑆𝐼𝑍𝐸 𝑖 𝐴 +
0.039
(1.53)
0.101***
(3.86)
𝐼𝐴 𝑖 𝐴
8.377***
(2.87)
5.153*
(1.82)
Implementation uncertainty
𝐿𝑛 𝐷𝐴𝑌𝑆 𝑖 -
-0.114**
(-2.26)
-0.076**
(-2.17)
𝑁𝑜 𝑜𝑓 𝑆𝐼𝐶𝑠 𝑖 𝑇
0.035**
(2.46)
0.030*
(1.78)
Controls
𝐶𝑜𝑛𝑔𝑙𝑜𝑚 𝑖 𝐴 + 
0.253**
(2.03)
0.270**
(2.33)
0.357***
(3.05)
0.478***
(3.13)
𝐿𝐸𝑉 𝑖 𝐴
1.073***
(2.92)
1.230***
(3.67)
0.964***
(2.68)
1.391***
(2.94)
𝐶𝐹𝑂𝑡𝑜𝑇𝐴 𝑖 𝐴
0.033
(0.07)
0.570
(1.24)
0.591
(1.28)
1.386**
(2.36)
𝑆𝑎𝑚𝑒𝑆𝐼𝐶𝑠 𝑖 - 
0.110
(1.23)
0.131*
(1.72)
0.069
(0.81)
0.137
(1.32)
Year Indicators
Yes 
Wald 𝜒 2 test for:
𝐻 0 : 𝜌 𝜇𝑒 =0
19.35###
[0.000]
28.23###
19.87###
8.90###
[0.003]
No. Obs.
483
494
469
477

27. Empirical Results: Premium equation – nearest neighbour
Premium equation ( 𝐹𝐿𝐼𝑃 𝑖 is endogenous): PREMIUM is the dependent variable
 Variable
 Pred. Sign
Specification (1)
Coefficient
Specification (2)
Specification (3)
Specification (4)
Intercept
0.425***
(3.03)
0.741***
(6.37)
0.603***
(4.36)
0.655***
(4.60)
𝑅𝑢𝑛𝑈𝑝 𝑖 +
0.792***
(4.51)
0.663***
0.593***
(3.17)
0.635***
(3.49)
𝑆𝐼𝑍𝐸 𝑖 𝑇 -
-0.045***
(-2.85)
-0.075***
(-5.43)
-0.061***
(-3.82)
(-3.75)
𝐹𝐿𝐼𝑃 𝑖
0.790***
(7.64)
0.813***
(10.38)
0.858***
(8.14)
0.713***
(4.75)
𝑆𝐺𝑅 𝑖 𝑇
 +/-
-0.029
(-0.50)
-0.003
(-0.06)
-0.033
(-0.53)
-0.009
(-0.14)
𝐶𝑎𝑠ℎ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡 𝑖  +
0.002***
(2.87)
0.001
(1.57)
0.001*
(1.87)
(1.54)
𝑁𝑜 𝑜𝑓 𝐵𝑖𝑑𝑑𝑒𝑟𝑠 𝑖
-0.030
(-0.80)
0.010
(0.26)
-0.035
(-0.82)
(-0.67)
Year Indicators
Yes
𝐿𝑜𝑔𝐿
Wald 𝜒 2 test for: 𝐻 0 : all coefs=0
140.25###
[0.000]
218.17###
149.25###
75.62###
No. Obs.
483
494
469
477
***, **, * denote significance at the 1%, 5% and 10% level, respectively, ### denotes significance at the 1% level.

28. Conclusions
We test the hypotheses of a real options model, proposed by Alvarez and Stenbacka (2006), which establishes in a perfectly rational-expectation framework the optimal timing of sequences of acquisitions and subsequent divestitures.
Acquisitions can be thought as a compound (real) option: The acquiring firm has the additional option to sell activities outside its core business to a third firm, which can make more efficient use of these resources.
Takeover completion makes it possible to exploit synergy gains and it also incorporates the embedded (real) option represented by the potential divestment gains.
This paper empirically tests the predictions of this real option theory directly, by manually collecting a sample of identified “acquisitions and subsequent divestment of non-core operations” sequences by US listed firms between 1999 and 2013.
We find strong evidence that acquisition/divestment decisions are indeed affected by proxies for the acquirer’s bargaining power, the target’s cash flow uncertainty and possible exogenous implementation risks, and also that a firm’s decision to follow this compound strategy positively affects the acquisition premium.

29. Thank you for your attention