Unilateral E ects Screens for Partial Horizontal Acquisitions: The Generalized HHI and GUPPI

Transcription

1 Unilateral E ects Screens for Partial Horizontal Acquisitions: The Generalized HHI and GUPPI Duarte Brito, António Osório, Ricardo Ribeiro and Helder Vasconcelos Autoridade da Concorrência Seminar October 2015 BORV (2015) Screens for PHA October / 61

2 Introduction Partial acquisitions do not completely and permanently eliminate competition among the rms involved, but may nevertheless present signi cant competitive concerns As a consequence, competition agencies have taken an increased interest in assessing their anti-competitive e ects Some renown examples include: UK Competition Commission assessment of ITV s proposed 17.9% acquisition by rival BskyB The European Commission assessment of Premiere s proposed 25% acquisition by rival News Corporation DoJ assessment of Wilkinson Sword s proposed 23% nonvoting acquisition by rival Gillette FTC assessment of Kinder Morgan s proposed buyout by the Carlyle Group, a partial owner of rival Magellan Midstream BORV (2015) Screens for PHA October / 61

3 Introduction In this paper, we propose a generalization to a partial horizontal acquisition setting of the two most traditional indicators used to screen the unilateral anti-competitive e ects... The Hel ndahl-hirschman Index (HHI), typically suitable for Cournot homogeneous-product industries The Gross Upward Price Pressure Index (GUPPI), typically suitable for Bertrand di erentiated-product industries...a generalization that does not increase the data requirements associated with a typical noti cation to a competition agency BORV (2015) Screens for PHA October / 61

4 Introduction The proposed generalized indicators (i) can deal with acquisitions that give rise to: Cross-ownership structures Acquisitions by owners that are internal to the industry, rival rms Ex: ITV s proposed 17.9% acquisition by BskyB Common-ownership structures Acquisitions by owners that are external to the industry, but participate in more than one rival rm Ex: Kinder Morgan s proposed buyout by the Carlyle Group BORV (2015) Screens for PHA October / 61

5 Introduction The proposed generalized indicators (ii) can deal with acquisitions of: Financial interest rights Acquisitions that give the owner the right to receive a share of the the stream of pro ts generated by the operations and investments of the acquired rm Ex: Wilkinson Sword s proposed 23% nonvoting acquisition by Gillette Corporate control rights Acquisitions that give the (partial) owner the right to in uence the decisions of the acquired rm Ex: ITV s proposed 17.9% acquisition by BskyB BORV (2015) Screens for PHA October / 61

6 Introduction The proposed generalized indicators (iii) can deal with acquisitions of: Direct nancial interest or corporate control rights Indirect nancial interest or corporate control rights An owner has an indirect nancial interest or corporate control right in rm B if it holds a nancial interest or corporate control right in rm A and, in turn, rm A holds a nancial interest or corporate control right in rm B This issue is particularly important for antitrust purposes because indirect (partial) acquisitions may constitute a way of evading competition rules The proposed generalized indicators (iv) can deal with full mergers BORV (2015) Screens for PHA October / 61

7 Literature Review There have been other proposed generalizations of the two screening indicators, HHI and GUPPI, to partial horizontal acquisition settings in the literature However they tend to focus on a subset of partial acquisition scenarios To the best of our knowledge our proposal constitutes the rst generalization that can cope with acquisition settings involving all types of owners and rights Owners that can be internal and external to the industry Rights that can involve nancial interest and corporate control, can be direct and indirect, can be partial or not BORV (2015) Screens for PHA October / 61

8 Literature Review Proposed HHI Generalizations* Cross Common Both Ownership Ownership Types Panel A: HHI Direct Reynolds and FI Snapp (1986) Direct Bresnahan and O Brien and _ FI and CC Salop (1986) Salop (2000) Direct and Indirect Flath (1992) FI DSV (2000) Direct and Indirect Current FI and CC Paper * FI and CC denote Financial Interest and Corporate Control, respectively. DSV (2000) denotes Dietzenbacher et al. (2000). Bresnahan and Salop (1986) model direct FI and CC, but can not address all types of CC arrangements. Flath (1992) models direct and indirect FI, but does not propose an indicator. BORV (2015) Screens for PHA October / 61

9 Literature Review Proposed "GUPPI" Generalizations* Cross Common Both Ownership Ownership Types Direct _ O Brien and _ FI and CC Salop (2000) Direct and Indirect DSV (2000). FI Direct and Indirect Current FI and CC Paper * FI and CC denote Financial Interest and Corporate Control, respectively. DSV (2000) denotes Dietzenbacher et al. (2000). O Brien and Salop (2000) model direct FI and CC and propose a PPI, a measure which lead afterwards to the GPPUI, but was never generalized to partial horizontal acquisition settings.. DSV (2000) model direct and indirect FI, but do not propose an indicator. BORV (2015) Screens for PHA October / 61

10 The Theoretical Framework The Setup The general setting is adapted from Brito et al. (2014) There are N single-product rms, indexed by j 2 = f1,..., Ng There are also K owners, indexed by k 2 Θ f1,..., N,..., K g, which implies that we allow owners to be: Internal to the industry (those from the subset =) External to the industry (those from the subset Θn=) BORV (2015) Screens for PHA October / 61

11 The Theoretical Framework The Setup The total stock of each rm j is composed of voting stock and non-voting (preferred) stock φ k,j 2 [0, 1] denotes the nancial interest of owner k in rm j It captures the owner s holdings of total stock in the rm, regardless of whether it be voting or non-voting stock γ k,j 2 [0, 1] denotes the corporate control of owner k in rm j It is typically a non-linear function of the owner s holdings of voting stock in the rm BORV (2015) Screens for PHA October / 61

12 The Theoretical Framework Firm s Operating Pro t The operating pro t of rm j is generated by the rm s own operations and can be written as follows: where: p j price of rm j π j = (p j mc j ) q j c j mc j (assumed constant) marginal cost of rm j (can be relaxed) q j quantity of rm j c j xed cost of rm j BORV (2015) Screens for PHA October / 61

13 The Theoretical Framework Firm s Aggregated Pro t We model nancial cross-ownership settings by distinguishing between a rm s operating and aggregated pro t The aggregated pro t of rm j in such settings includes... The operating pro t The returns on the cross-holdings of rm j in all other rival rms...which implies that it can be written as follows: Π j = π j + g 2=nj φ j,g Π g Since there are N rms in the industry, we have N aggregate pro t equations (like the one above), which constitute a system of N equations that we can solve to determine the aggregate pro t of each rm BORV (2015) Screens for PHA October / 61

14 The Theoretical Framework Manager s Objective Function In a nancial cross-ownership setting, barring any market imperfections, owners will typically agree (and give the appropriate incentives) that the manager should maximize aggregate pro ts However, in other partial ownership settings, owners may have con icting objectives and, therefore, may not agree on the best course of action for the rm For example, in a nancial common-ownership setting, an external owner of rm j who also has a large nancial interest in rival rm g typically wants rm j to pursue a less aggressive strategy than the strategy desired by an external owner with no nancial interest in rm g BORV (2015) Screens for PHA October / 61

15 The Theoretical Framework Manager s Objective Function We model corporate control cross-ownership, nancial interest common-ownership, and corporate control common-ownership settings as follows: We assume that managers weight the con icting objectives of the di erent owners according to each owner s corporate control over the decision-making within the rm We assume that the objective of each owner is captured by the returns of the nancial interests he/she holds over the aggregate pro t of the di erent rival rms BORV (2015) Screens for PHA October / 61

16 The Theoretical Framework Manager s Objective Function This implies that the manager of rm j should maximize the following objective function: ϖ j = k2θ/j γ k,j R k where: R k return of owner k 2 Θ In the paper, we show that, under the above assumptions, maximizing ϖ j is entirely equivalent to maximizing a counterpart objective function ϖ 0 j BORV (2015) Screens for PHA October / 61

17 The Theoretical Framework Manager s Objective Function The counterpart objective function ϖ 0 j consists of a weighted sum of the operating pro ts of all the rms in the industry over which the owners of rm j have (direct or indirectly) a nancial interest and/or a corporate control stake on: where: ϖ 0 j = g 2= w j,g π g w j,g weight that the manager of rm j attributes to rm g s operating pro t A weight that depends on the nancial interest and corporate control rights that the owners of rm j (direct or indirectly) hold over rm g BORV (2015) Screens for PHA October / 61

18 The Theoretical Framework Manager s Objective Function Let N N matrix W stack the weights attributed by the di erent managers to each rm s operating pro t: w 1,2 w 1,N w 2,1 1 w 2,N W = w N,1 w N,2 1 In the paper, we show that these weights can be straightforwardly computed from four matrices, that stack the nancial interest and corporate control rights of the di erent owners (internal and external) over each rm BORV (2015) Screens for PHA October / 61

19 The Theoretical Framework Manager s Objective Function A N N matrix F that captures the nancial interest rights of internal owners φ 1,2 φ 1,N F φ 2,1 0 φ 2,N = φ N,1 φ N,2 0 A (K N) N matrix F that captures the nancial interest rights of external owners 2 3 φ N +1,1 φ N +1,2 φ N +1,N φ N +2,1 φ N +2,2 φ N +2,N F = φ K,1 φ K,2 φ K,N BORV (2015) Screens for PHA October / 61

20 The Theoretical Framework Manager s Objective Function A N N matrix C that captures the corporate control rights of internal owners γ 1,2 γ 1,N C γ 2,1 0 γ 2,N = γ N,1 γ N,2 0 A (K N) N matrix C that captures the corporate control rights of external owners 2 3 γ N +1,1 γ N +1,2 γ N +1,N γ N +2,1 γ N +2,2 γ N +2,N C = γ K,1 γ K,2 γ K,N BORV (2015) Screens for PHA October / 61

21 The Theoretical Framework Manager s Objective Function Having stacked all the nancial interest and corporate control rights information into the above four matrices, we can straightforwardly compute the N N matrix W as follows: where: W = diag (L) 1 L L = (I (BC ) ) 1 C F (I F ) 1 B denotes a N N diagonal matrix with elements, b j,j, given by: 1/aj,j if a b j,j = j,j > 0 1 if a j,j = 0 A = diag (C F) denote a N N diagonal matrix (with elements a j,j ) formed by substituting zeros for all o -diagonal elements of C F diag (L) is the N N matrix formed by substituting zeros for all o -diagonal elements of L BORV (2015) Screens for PHA October / 61

22 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI In a Cournot homogeneous-product industry, the manager of rm j 2 = solves: max q j ϖ 0 j = g 2= w j,g π g = g 2= w j,g ((p (Q) mc g ) q g c g ), where: p 1 =... = p j =... = p N = p = p (Q), for all j 2 = p (Q) denotes the inverse market demand function Q = j2= q j denotes the aggregated industry output level BORV (2015) Screens for PHA October / 61

23 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI The Nash equilibrium, for an interior solution, for all j 2 =, is characterized by: (p (Q ne p (Q ) mc j ) + ne ) g 2= w j,g Q qne g = 0, which after some rearranging becomes: p (Q Lerner = ne ) mc j j2= p (Q ne sj ne ) where: = 1 η j2= g 2= w j,g sg ne sj ne, sj ne = qj ne /Q ne denotes the output share of rm j 2 = η = ( Q ne / p (Q)) (p (Q) / Q ne ) denotes the (absolute value of the) price-elasticity of demand BORV (2015) Screens for PHA October / 61

24 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI This result establishes that the (output) weighted margins to price ratio of all the rms in the industry is a function of: 1 The price-elaticity of demand η 2 A measure of concentration that constitutes our proposed generalized HHI pre-acquisition... GHHI = j2= g 2= w j,g s ne g s ne j,...and which incorporates the di erent rights (direct and indirect, involving control or not, partial or not) of all types of owners (internal and external to the industry) BORV (2015) Screens for PHA October / 61

25 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI Consider now an hypothetical acquisition Let = denote the subset of rms (direct and indirectly) involved in the acquisition Independently of the particulars of the acquisition, it will de nitely impact the weights w j,g of all j, g 2 = Let w j,g denote the post-acquisition weights, with w j,g = w j,g for any j, g 2 =n = and w j,g 6= w j,g for any j, g 2 = BORV (2015) Screens for PHA October / 61

26 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI Assuming a setting of no e ciency gains (as in the standard HHI), the interior Nash equilibrium post-acquisition, is characterized by:! p Q Le erner = ne mc j j2= p Q ne s ne j = 1 η j2= g 2= w j,g s g ne s j ne. The idea behind our proposed generalized HHI is to use information local to the pre-acquisition Nash equilibrium to predict the directional price impacts of acquisitions To do so, we assume that Q ne is the only variable that re-equilibrates after the acquisition, i.e., we ignore the re-equilibration across the output shares BORV (2015) Screens for PHA October / 61

27 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI Under the no re-equilibration assumption, we have that: s j ne j 2 = = s ne j This implies that the interior Nash equilibrium post-acquisition can be re-written as: for p Q Le erner = ne mc j j2= p Q ne! s ne j = 1 η j2= g 2= w j,g sg ne sj ne. As a consequence, the di erence between the post- and the pre-acquisition (output) weighted margins to price ratio is given by: Le erner Lerner = 1 η G fhhi GHHI. BORV (2015) Screens for PHA October / 61

28 The Proposed Generalized Indicators Cournot Homogeneous-Product Industries: The Generalized HHI This establishes that the industry upward pricing pressure, gross of e ciency gains, is a function of: 1 The price-elaticity of demand η 2 The change in GHHI, our concentration measure For any given price-elasticity of demand, we have that: The higher the increase in the GHHI, the greater the unilateral e ects impact of the acquisition on the (output) weighted margins to price ratio And, as a consequence, the greater the likelihood that competition agencies should decide to issue a second request to conduct a more detailed analysis of the acquisition BORV (2015) Screens for PHA October / 61

29 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI In a Bertrand di erentiated-product industry, the manager of rm j solves: max p j ϖ 0 j = g 2= w j,g π g = g 2= w j,g ((p g mc g ) q g (p) c g ), where: q g (p) is the quantity demanded for the product of rm g p is the vector of prices for all the products available in the industry BORV (2015) Screens for PHA October / 61

30 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI Under this setting, the Nash equilibrium pre-acquisition, for an interior solution, for all j 2 =, is characterized by: q j (p ne ) + p ne j mc j q j (p ne ) p j + g 6=j2= w j,g p ne g mc g q g (p ne ) p j = 0, which after some rearranging, yields that: p ne j = mc j q j (p ne ) q j (p ne ) / p j + g 6=j2= w j,g p ne g mc g DRgj, where: qg (p DR gj = ne )/ p j q j (p ne denotes the diversion ratio from rm j s )/ p j product to rm g s product BORV (2015) Screens for PHA October / 61

31 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI Consider now an hypothetical acquisition Let = denote the subset of rms (direct and indirectly) involved in the acquisition Independently of the particulars of the acquisition, it will de nitely impact the weights w j,g of all j, g 2 = Let w j,g denote the post-acquisition weights, with w j,g = w j,g for any j, g 2 =n = and w j,g 6= w j,g for any j, g 2 = BORV (2015) Screens for PHA October / 61

32 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI Assuming a setting of no e ciency gains (as in the standard GUPPI), the interior Nash equilibrium post-acquisition for all j 2 = is characterized by: p j ne q j (~p ne ) = mc j q j (~p ne + ) / p j g 6=j2 = w j,g p g ne + g 6=j2=n = w j,g p ne g mc g DRgj mc g DRgj The idea behind our proposed generalized GUPPI is to use information local to the pre-acquisition Nash equilibrium to predict the directional price impacts of acquisitions BORV (2015) Screens for PHA October / 61

33 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI To do so, we assume that the price of rm j s product is the only variable that re-equilibrates after the acquisition We ignore the re-equilibration of the remaining variables, i.e., all the other prices, all the quantities and all the price-e ects Under the no re-equilibration assumption, we have that: p ne g = p ne g for g 6= j 2 = q j (p ne ) = q j (~p ne ) for j 2 = q g (~p ne ) / p j = q g (~p ne ) / p j for all g 2 = BORV (2015) Screens for PHA October / 61

34 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI This implies that the interior Nash equilibrium post-acquisition for all j 2 = can be re-written as: p j ne q j (p ne ) = mc j q j (p ne + ) / p j g 6=j2 = w j,g pg ne mc g DRgj + g 6=j2=n = w j,g p ne g mc g DRgj. The above result implies that the di erence between the post- and pre-acquisition price of the product of rm j 2 = is given by: p ne j pj ne = g 6=j2 = ( w j,g w j,g ) p ne g mc g DRgj BORV (2015) Screens for PHA October / 61

35 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI This establishes that rm j s product upward pricing pressure, gross of e ciency gains, is a function of: 1 The change in the weights w j,g of all j, g 2 = 2 The pre-acquisition price-cost margins of all g 2 = 3 The diversion ratios of all j, g 2 = I.e., it is a funtion of information referent solely to the products in =, i.e., to the products of the rms involved in the acquisition BORV (2015) Screens for PHA October / 61

36 The Proposed Generalized Indicators Bertrand Di erentiated-product Industries: The Generalized GUPPI This establishes our proposed generalized GUPPI for product j 2 =: GGUPPI j = p ne j pj ne /p ne j = g 6=j2 = ( w j,g w j,g ) p ne g mc g DRgj /p ne j The higher the level of GGUPPI for each product involved in the acquisition, the greater the unilateral e ects impact of the acquisition on its price And, as a consequence, the greater the likelihood that competition agencies should decide to issue a second request to conduct a more detailed analysis of the acquisition. BORV (2015) Screens for PHA October / 61

37 Empirical Application We now present an empirical application of the GHHI and the GGUPPI to a partial acquisition in the wet shaving industry, with the objective of providing a step-by-step illustration of how to compute the two proposed indicators On December 20, 1989, the Gillette Company, contracted to acquire 22.9% of the nonvoting equity shares of Eemland Management Services BV (Wilkinson Sword s parent company) for about $14 million At that time, consumers in the United States annually purchased over $700 million of wet shaving razor blades at the retail level BORV (2015) Screens for PHA October / 61

38 Empirical Application Five rms supplied all but a nominal amount of the those blades: The Gillette Company, which had been the market leader for years American Safety Razor Company BIC Corporation Warner-Lambert Company Wilkinson Sword Inc. BORV (2015) Screens for PHA October / 61

39 Empirical Application The acquisition made Gillette a large owner of a competitor The DoJ allowed the acquisition after being assured that this stake would be passive However, even when the acquiring rm cannot in uence the conduct of the target rm, the partial acquisition may still raise antitrust concerns BORV (2015) Screens for PHA October / 61

40 Empirical Application The Weight Matrix In order to apply the two proposed indicators to the above setting, we begin by calculating the pre-acquisition weights (in matrix W) To do so, we require information on both the nancial interest and the corporate control structure of the ve rms in the industry pre-acquisition We make use of public information, that we supplement with two assumptions The corporate control right an owner has over the decision making of a rm is equal to the share of voting rights she owns in the rm Minority external owners do not engage in common-ownership So that we can aggregate without loss of generality the nancial and control rights of each rm s minority external owners into a single ctitious external owner BORV (2015) Screens for PHA October / 61

41 Empirical Application Financial Interest and Corporate Control Rights Pre-Acquisition* ASR BIC WL WS G j/k F CC F CC F CC F CC F CC Panel A: Internal Owners 01 American Safety Razor Company BIC Corporation Warner-Lambert Company Wilkinson Sword, Inc The Gillette Company Panel B: External Owners 06 Equitable (a) Allsop Venture Partners III, LP Goldman Sachs Group, LP Scudder Stevens and Clarck Leucadia-Mezzanine (b) Grantham Mayo Van Otter ASR Minority Owners Bruno Bich BIC Minority Owners WL Minority Owners Eemland Management Services BV Berkshire Hathaway, Inc State Street Bank and Trust Co G Minority Owners * Figures are in percentage points. Adapted from Schedule 14A (proxy statement) information and 93/525/EEC decision. F denotes each external owner s holdings of total stock. CC denotes, under Assumption 5, each external owner s holdings of voting stock. ASR, B, WL, WS and G denote American Safety Razor Company, BIC Corporation, Warner-Lambert Company, Wilkinson Sword, Inc., and The Gillette Company, respectively. (a) Equitable denotes the cumulative ownership of Equitable Capital Partners, LP, Equitable Deal Flow Fund, LP, Equitable Capital Partners (Retirement Fund), LP, and The Equitable Life Assurance Society of the United States. (b) Leucadia-Mezzanine denotes the cumulative ownership of Leucadia Investors, Inc. and Mezzanine Capital and Income Trust 2001 PLC. BORV (2015) Screens for PHA October / 61

42 Empirical Application The Weight Matrix Having described the nancial and the corporate control structure of the ve rms, we can begin to convert that information into the four matrices discussed above A 5 5 matrix F that captures the nancial interest rights of internal owners F = BORV (2015) Screens for PHA October / 61

43 Empirical Application The Weight Matrix A 14 5 matrix F that captures the nancial interest rights of external owners F = BORV (2015) Screens for PHA October / 61

44 Empirical Application The Weight Matrix A 5 5 matrix C that captures the corporate control rights of internal owners C = BORV (2015) Screens for PHA October / 61

45 Empirical Application The Weight Matrix A 14 5 matrix F that captures the corporate control rights of external owners C = BORV (2015) Screens for PHA October / 61

46 Empirical Application The Weight Matrix Having constructed the above four matrices, we have all the necessary information to compute matrix A =diag (C F) A = BORV (2015) Screens for PHA October / 61

47 Empirical Application The Weight Matrix...matrix B, which denotes a diagonal matrix with elements, b jj = 1/a jj if a jj > 0 and b jj = 1 if a jj = B = BORV (2015) Screens for PHA October / 61

48 Empirical Application The Weight Matrix...matrix L = (I (BC ) ) 1 C F (I F ) L = BORV (2015) Screens for PHA October / 61

49 Empirical Application The Weight Matrix...and the pre-acquisition weight matrix W = diag (L) 1 L: W = This result implies that pre-acquisition (i.e., absent cross- and common-ownership), and barring any market imperfections: Owners agree (and give the appropriate incentives) that the manager should maximize own-operating pro ts BORV (2015) Screens for PHA October / 61

50 Empirical Application The Weight Matrix Gillette s acquisition of 22.9% nonvoting equity interest in Wilkinson Sword gives rise to a nancial interest cross-ownership structure in the industry Eemland, an external owner, fully controls Wilkinson Sword, but shares the nancial interest in the rm with Gillette, an internal owner What are the implications to the weights in matrix W? Matrices C and C remain unchanged, since the rights transacted involve no voting, and thus involve no change in corporate control Matrices F and F change since nancial interests rights were transacted BORV (2015) Screens for PHA October / 61

51 Empirical Application The Weight Matrix These changes induce the following post-acquisition weight matrix ~W: ~W = This result implies that the manager of Wilkinson Sword should maximize own-operating pro ts The reason being that while Wilkinson Sword has two owners, only one of them, Eemland, has control over the manager And Eemland only cares about the returns of the equity it holds in Wilkinson Sword BORV (2015) Screens for PHA October / 61

52 Empirical Application The Weight Matrix The above result also implies that the manager of Gillette should maximize a weighted average of Gillette and Wilkinson Sword s operating pro ts The reason being that Gillette s aggregate pro t, which determines the equity returns of Gillette s owners, includes not just the own- rm s operating pro ts, but also the 22.9% share in Wilkinson Sword s aggregate pro ts Which, in this case, coincide with Wilkinson Sword s operating pro ts since the rm does not engage in cross-ownership Overall, this suggests that partial acquisitions of nonvoting interests change the incentives of the acquiring rm, but not of the acquired rm BORV (2015) Screens for PHA October / 61

53 Empirical Application The Generalized HHI We now address the computation of our proposed generalized market concentration measure, the generalized HHI In order to apply the generalized HHI to a particular setting, we require information about: The weight matrix W (discussed above) The pre-acquisition output shares of the rms in the industry BORV (2015) Screens for PHA October / 61

54 Empirical Application The Generalized HHI Firm Pre-Acquisition Output Shares* j output share 01 American Safety Razor Company 1% 02 BIC Corporation 20% 03 Warner-Lambert Company 14% 04 Wilkinson Sword, Inc. 3% 05 The Gillette Company 50% * Figures adapted from DoJ (1990). BORV (2015) Screens for PHA October / 61

55 Empirical Application The Generalized HHI Using the above information, we can compute the generalized HHI pre- and post-acquisition The pre-acquisition industry has a generalized HHI of: 3, 106 = (1) 2 + (20) 2 + (14) 2 + (3) 2 + (50) 2 This result makes clear that in the absence of cross- and common-ownership, the GHHI reduces to the standard HHI It also suggests that the wet shaving industry was highly concentrated even before December 20, 1989 BORV (2015) Screens for PHA October / 61

56 Empirical Application The Generalized HHI The post-acquisition industry with a generalized HHI of: 3, 140 (1) 2 + (20) 2 + (14) 2 + (3) (50) (3) + (50) 2 This result suggests that the acquisition involved an increase in concentration of less than 100 points, which implies that it was also unlikely to have adverse competitive e ects This seems to validate the decision of DoJ not to challenge the operation BORV (2015) Screens for PHA October / 61

57 Empirical Application The Generalized GUPPI We now address the computation of our proposed generalized upward pricing pressure measure, the generalized GUPPI In order to apply the generalized GUPPI to a particular setting, we require information about: The weight matrix W pre- and post-acquisition The pre-acquisition prices, margins, and diversion ratios for the rms which weights exhibit changes pre- and post-acquisition BORV (2015) Screens for PHA October / 61

58 Empirical Application The Generalized GUPPI The di erence between ~W and W is given by: ~W W = , I.e., In our application, only the weight of Gillette s manager on Wilkinson Sword s operating pro ts does change This implies that, in our application, we require information solely on the pre-acquisition prices, margins, and diversion ratios involving the two rms BORV (2015) Screens for PHA October / 61

59 Empirical Application The Generalized GUPPI Prices, Margins and Diversion Ratios* j WS G Panel A: Prices and Margins ($) Price Margin Panel B: Diversion Ratios 04 Wilkinson Sword, Inc The Gillette Company * Price and margin gures are in USD. WS and G denote Wilkinson Sword and Gillette, respectively. All information computed using the demand and cost estimates in Brito et al. (2014). BORV (2015) Screens for PHA October / 61

60 Empirical Application The Generalized GUPPI Since only the weight of Gillette s manager on Wilkinson Sword s operating pro ts does change, solely Gillette s products will exhibit an upward pricing pressure The acquisition s GGUPPIs are thus given by: GGUPPI j = 0 for j = f1, 2, 3, 4g GGUPPI 5 = (0.229) (0.375) (0.225) / (4.036) = 0.479% This result suggests that the acquisition was unlikely to have adverse competitive e ects, since it involved an upward pricing pressure of only 0.479% in Gillette s products This seems to validate the decision of DoJ not to challenge the operation BORV (2015) Screens for PHA October / 61

61 Conclusion In this paper, we propose a generalization to a partial horizontal acquisition setting of the two most traditional indicators used to screen unilateral anti-competitive e ects: the HHI and the GUPPI The proposed generalized indicators can deal with all types of acquisitions that may lessen competition in the industry We also provide an empirical application to several acquisitions in the wet shaving industry as a step-by-step illustration of how to compute the two indicators BORV (2015) Screens for PHA October / 61