Skip to main content
Log in

Post-merger internal organization in multitier decentralized supply chains

  • Published:
Journal of Economics Aims and scope Submit manuscript

Abstract

This paper analyses the effects upon firms and consumers of horizontal mergers in a multitier decentralized supply chain with a finite number of players in each tier, when firms may opt for two different post-merger internal organization forms: multidivisional, in which separate divisions are kept, or traditional, with cost synergies. To this effect, we develop and solve a formal game theory-based Cournot model. The main results are: independently of the tier in which the merger takes place, higher synergies do not always lead to higher consumer welfare; despite the fact that the proposal of a traditional merger reveals significant cost savings consumer welfare may still decrease with the merger; traditional downstream mergers tend to be more profitable than traditional upstream ones; multidivisional mergers are always profitable.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. As opposed to a centralized supply chain, in which a central authority is responsible for decision making.

  2. For recent examples of mergers between auto component makers and also of mergers involving automobile manufacturers see, for instance, https://www.bloomberg.com/news/articles/2019-10-29/honda-hitachi-to-merge-four-car-parts-makers-yomiuri-says and https://www.wardsauto.com/news-analysis/who-are-winners-all-these-automotive-mergers.

  3. A firm’s merging strategy can also be driven by risk reducing purposes which is something we do not consider. For a recent study of the impact of mergers and acquisitions on return on assets volatility see, e.g., Wu and Chiang (2019).

  4. Note that we focus exclusively on the price effects of a merger and do not consider any eventual effects on investment which may benefit consumers. In other words, we focus exclusively on the static effects of a merger and do not address its dynamic effects. For a recent discussion of the trade-off between these two effects in the case of the wireless telecommunication industry, see Grajek et al. (2019).

  5. Other proposed solutions include product differentiation and price competition instead of quantity competition (Deneckere and Davidson 1985), quadratic cost functions (Perry and Porter 1985), non-Cournot conjectures (Kwoka 1989), general demand function forms (Cheung 1992), among others.

  6. Lambertini and Pignataro (2019), following Ziss (1998), study divisionalization under product differentiation, when single division firms can, at the outset, choose to build up a new second division that sells the same product but is independently managed. The level of multidivisionalization in the industry will depend on the fixed cost of creating additional divisions and on the degree of product differentiation among firms.

  7. Creane and Davidson (2004) present the examples of Seagram’s, Procter and Gamble, Mitsubishi, Estée Lauder Cosmetics and JWP Inc. as firms that promote internal competition, and of Hewlett Packard and the auto industry as cases of firms that treat their divisions differently. Huck et al. (2004) present evidence of mergers that induce a partial shift of production from one division to another, making the distribution of output across plants more asymmetric. The examples include the US meat-packing industry in the 80’s, the Pepsi and Quaker 2001 merger and the Volvo–Ford merger.

  8. A recent article that addresses the effects of horizontal mergers in a two tier industry in the presence of vertical integration is Pinopoulos (2020). In this article, a vertically integrated firm faces competition from an independent competitor downstream, firm \(D_{2}\) who is supplied by upstream firm \(U_{2}\). The effects of a merger between the vertically integrated firm and \(U_{2}\) are analyzed.

  9. The authors give as example the existence of many PC assemblers, but of only a few chip manufacturers, whereas, on the contrary, the automotive industry presents few final assemblers, but a lot of manufacturers for most parts.

  10. For a review of game theoretic applications to supply chain management, see Leng and Parlar (2005).

  11. See Brito and Catalão-Lopes (2019) for a comprehensive explanation of the traditional merger cost reduction, and of excluding the possibility of realizing the multidivisional gains without a merger, simply by splitting a firm into two divisions.

  12. The expressions for \(\overline{z_{i}}\) are presented in Appendix 2.

  13. As Creane and Davidson (2004) show, although it is possible to have prices set in sequence by the divisions of the merged firm, this is not profitable because there will be no first mover advantage.

  14. We also make the assumption of two-firm mergers for tractability. If we allowed for mergers between a higher number of firms, several additional questions would need to be answered: How would a multidivisional merger be organized when the number of insiders is larger than two? How do the cost savings in a traditional merger depend on the number of insiders? Although it is possible to answer these questions (for instance, we conjecture that Creane and Davidson’s Propositon 3, p. 965, “The merged firm’s profit per division is maximized when there is a single firm producing at each stage in the production process” applies to this case), allowing for different numbers of insiders would lead to a multiplication in the number of parameters needed to characterize the equilibrium: one would need to know how many mergers of x insiders took place, of each type an in each tier.

  15. Note that the decrease in the average marginal cost \(\overline{v_{1}}\) depends on the number of downstream active firms. If this number is large, the impact on the average marginal cost of one of them having lower costs will be small.

  16. The reason is the following: in a Cournot model, equilibrium price is the average of the demand intercept and the firms’ marginal costs. When \(n_{2}\) is large the demand intercept has a lower impact in the average. Recall that downstream cost savings increase the demand intercept, thus increasing \(P_{2}\) in a way that depends on the number of terms in the average.

  17. Recall that the demand intercept of the upstream firms is \(A_{2}=A_{1}-\overline{v_{1}}\), which is not affected by these mergers. The demand slope is affected but it has no impact on price.

  18. The same possibility was also pointed out in Brito and Catalão-Lopes (2019) for the case of a single tier industry.

  19. Conclusion (v) holds in the absence of traditional mergers, i.e., \(t_{1}=t_{2}=0\) and (vii) holds if \(t_{1}=0\).

References

  • Adida E, DeMiguel V (2011) Supply chain competition with multiple manufacturers and retailers. Oper Res 59(1):156–172

    Article  Google Scholar 

  • Brito D, Catalão-Lopes M (2019) Are larger merger synergies bad news for consumers? Endogenous post-merger internal organization. Scand J Econ 121(4):1728–1756.

    Article  Google Scholar 

  • Carr SM, Karmarkar US (2005) Competition in multiechelon assembly supply chains. Manag Sci 51(1):45–59

    Article  Google Scholar 

  • Cheung F (1992) Two remarks on the equilibrium analysis of horizontal merger. Econ Lett 40:119–123

    Article  Google Scholar 

  • Cho S-H (2014) Horizontal mergers in multitier decentralized supply chains. Manag Sci 60(2):356–379

    Article  Google Scholar 

  • Cho S-H, Wang X (2017) Newsvendor mergers. Manag Sci 63(2):298–316

    Article  Google Scholar 

  • Corbett CJ, Karmarkar US (2001) Competition and structure in serial supply chains with deterministic demand. Manag Sci 47(7):966–978

    Article  Google Scholar 

  • Creane A, Davidson C (2004) Multidivisional firms, internal competition, and the merger paradox. Can J Econ 37(4):951–977

    Article  Google Scholar 

  • Deneckere R, Davidson C (1985) Incentives to form coalitions with Bertrand competition. Rand J Econ 16:473–486

    Article  Google Scholar 

  • Faulí-Oller R (2002) Mergers between asymmetric firms: profitability and welfare. Manch School 70(1):77–87

    Article  Google Scholar 

  • Grajek M, Gugler K, Kretschmer T, Mişcişin I (2019) Static or dynamic efficiency: horizontal merger effects in the wireless telecommunications industry. Rev Ind Organ 55:375–402

    Article  Google Scholar 

  • Heywood JS, McGinty M (2007) Convex costs and the merger paradox revisited. Econ Inq 45(2):342–349

    Article  Google Scholar 

  • Huck S, Konrad K, Müller W (2004) Profitable horizontal mergers without cost advantages: the role of internal organization, information and market structure. Economica 71(284):575–587

    Article  Google Scholar 

  • Kwoka JE (1989) The private profitability of horizontal mergers with non-Cournot and Maverick behavior. Int J Ind Organ 7(3):403–411

    Article  Google Scholar 

  • Lambertini L, Pignataro G (2019) On the social (sub)optimality of divisionalization under product differentiation. J Econ 128(3):225–238

    Article  Google Scholar 

  • Leng M, Parlar M (2005) Game theoretic applications in supply chain management: a review. INFOR Inf Syst Oper Res 43(3):187–220

    Google Scholar 

  • Lin Y-T, Parlaktürk AK, Swaminathan JM (2014) Vertical integration under competition: forward, backward, or no integration? Prod Oper Manag 23(1):19–35

    Article  Google Scholar 

  • McAfee RP, Williams MA (1992) Horizontal mergers and antitrust policy. J Ind Econ 40(2):181–187

    Article  Google Scholar 

  • Mialon SH (2008) Efficient horizontal mergers: the effects of internal capital real location and organizational form. Int J Ind Organ 26(4):861–877

    Article  Google Scholar 

  • Netessine S (2009) Supply webs: managing, organizing, and capitalizing on global networks of suppliers. In: Kleindorfer P, Wind YJ (eds) The network challenge: strategy, profit, and risk in an interlinked world. Wharton School Publishing, Philadelphia Chap. 13

    Google Scholar 

  • Perry MK, Porter RH (1985) Oligopoly and the incentive for horizontal merger. Am Econ Rev 75(1):219–227

    Google Scholar 

  • Pinopoulos IN (2020) Upstream horizontal mergers involving a vertically integrated firm. J Econ 130(1):67–83

    Article  Google Scholar 

  • Salant S, Switzer S, Reynolds R (1983) Losses from horizontal merger: the effects of an exogenous change in industry structure on Cournot-Nash equilibrium. Q J Econ 98(2):185–199

    Article  Google Scholar 

  • Tombak M (2002) Mergers to monopoly. J Econ Manag Strategy 11(13):513–546

    Article  Google Scholar 

  • Wu C-H, Chiang H-E (2019) Impact of diversified mergers and acquisitions on corporate risk. J Econ Manag 15(1):93–115

    Google Scholar 

  • Zhou D, Karmarkar US, Jiang B (2015) Competition in multi-echelon distributive supply chains with linear demand. Int J Prod Res 53(22):6787–6808

    Article  Google Scholar 

  • Ziss S (1998) Divisionalization and product differentiation. Econ Lett 59(1):133–38

    Article  Google Scholar 

Download references

Acknowledgements

M. Catalão-Lopes and D. Brito acknowledge financial support from Fundação para a Ciência e a Tecnologia, repectively through grants UIDB/GES/00097/2020 and UIDB/04007/2020.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Margarida Catalão-Lopes.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix 1

Proof of Lemma 1

Making use of the results in Brito and Catalão-Lopes (2019) and observing that unitary costs in the downstream tier are equal to \(v_{1}+P_{2}\), we know that after \(t_{1}\) traditional mergers and \(m_{1}\) multidivisional mergers the equilibrium quantity in tier 1 for the outsiders, \(q_{1i}(P_{2})\), insider to a traditional merger, \(q_{1t}(P_{2})\), and leader and follower in a multidivisional merger (\(q_{1L}(P_{2})\) and \(q_{1F}(P_{2})\), respectively) are:

$$\begin{aligned} q_{1i}(P_{2})&=q_{1F}(P_{2})=\frac{A_{1}-v_{1}-P_{2}-t_{1}s_{1}}{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) }\\ q_{1L}(P_{2})&=\frac{2\left( A_{1}-v_{1}-P_{2}-t_{1}s_{1}\right) }{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) }\\ q_{1t}(P_{2})&=\frac{A_{1}-v_{1}-P_{2}+s_{1}\left( n_{1}+m_{1} -2t_{1}+1\right) }{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) } \end{aligned}$$

from where total output is obtained:

$$\begin{aligned} Q_{1}(P_{2})&=(n_{1}-2t_{1}-2m_{1})q_{1i}(P_{2})+t_{1}q_{1t}(P_{2} )+m_{1}(q_{1F}(P_{2})+q_{1L}(P_{2}))\\&=\frac{\left( n_{1}+m_{1}-t_{1}\right) \left( A_{1}-v_{1}-P_{2}\right) +t_{1}s_{1}}{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) } \end{aligned}$$

Knowing that \(Q_{1}=Q_{2}\) we can write \(Q_{2}\) as

$$\begin{aligned} Q_{2}(P_{2})=\frac{\left( A_{1}-v_{1}\right) \left( m_{1}+n_{1} -t_{1}\right) +s_{1}t_{1}}{b_{1}\left( m_{1}+n_{1}-t_{1}+1\right) } -P_{2}\frac{m_{1}+n_{1}-t_{1}}{b_{1}\left( m_{1}+n_{1}-t_{1}+1\right) } \end{aligned}$$

or, inverting,

$$\begin{aligned} P_{2}=\frac{\left( m_{1}+n_{1}-t_{1}\right) \left( A_{1}-v_{1}\right) +s_{1}t_{1}}{m_{1}+n_{1}-t_{1}}-\frac{m_{1}+n_{1}-t_{1}+1}{m_{1}+n_{1}-t_{1} }b_{1}Q_{2}. \end{aligned}$$

Thus, the demand in tier 2 is given by \(P_{2}=A_{2}-b_{2}Q_{2}\), with \(A_{2}=\frac{\left( m_{1}+n_{1}-t_{1}\right) \left( A_{1}-v_{1}\right) +s_{1}t_{1}}{m_{1}+n_{1}-t_{1}}\) and \(b_{2}=\frac{m_{1}+n_{1}-t_{1}+1}{m_{1}+n_{1}-t_{1}}b_{1}\).

Replacing into the equilibrium expressions for the equilibrium outputs of the different types of firms in the second tier, we obtain

$$\begin{aligned} q_{2i}&=q_{2F}=\frac{A_{2}-v_{2}-t_{2}s_{2}}{b_{2}\left( n_{2} +m_{2}-t_{2}+1\right) }\\ q_{2L}&=\frac{2\left( A_{2}-v_{2}-t_{2}s_{2}\right) }{b_{2}\left( n_{2}+m_{2}-t_{2}+1\right) }\\ q_{2t}&=\frac{A_{2}-v_{2}+s_{2}\left( n_{2}+m_{2}-2t_{2}+1\right) }{b_{2}\left( n_{2}+m_{2}-t_{2}+1\right) } \end{aligned}$$

which simplifies to:

$$\begin{aligned} q_{2i}&=q_{2F}=\frac{\left( m_{1}+n_{1}-t_{1}\right) \left( A_{1} -v_{1}-v_{2}-s_{2}t_{2}\right) +s_{1}t_{1}}{b_{1}\left( m_{1}+n_{1} -t_{1}+1\right) \left( m_{2}+n_{2}-t_{2}+1\right) }\\ q_{2L}&=2\frac{\left( m_{1}+n_{1}-t_{1}\right) \left( A_{1}-v_{1} -v_{2}-s_{2}t_{2}\right) +s_{1}t_{1}}{b_{1}\left( m_{1}+n_{1}-t_{1} +1\right) \left( m_{2}+n_{2}-t_{2}+1\right) }\\ q_{2t}&=\frac{\left( m_{1}+n_{1}-t_{1}\right) \left( A_{1}-v_{1} -v_{2}+s_{2}\left( m_{2}+n_{2}-2t_{2}+1\right) \right) +s_{1}t_{1}}{b_{1}\left( m_{1}+n_{1}-t_{1}+1\right) \left( m_{2}+n_{2}-t_{2}+1\right) } \end{aligned}$$

As for aggregate output,

$$\begin{aligned} Q_{2}=\frac{\left( n_{2}+m_{2}-t_{2}\right) \left( A_{2}-v_{2}\right) +t_{2}s_{2}}{b_{2}\left( n_{2}+m_{2}-t_{2}+1\right) }, \end{aligned}$$

we obtain after inserting the expressions for \(A_{2}\) and \(b_{2},\)

$$\begin{aligned} Q_{2}= & {} \frac{\left( n_{2}+m_{2}-t_{2}\right) \left( m_{1}+n_{1} -t_{1}\right) }{b_{1}\left( n_{2}+m_{2}-t_{2}+1\right) \left( m_{1} +n_{1}-t_{1}+1\right) }\\&\left( \left( A_{1}-v_{1}-v_{2}\right) +\frac{t_{1}s_{1}}{\left( m_{1}+n_{1}-t_{1}\right) }+\frac{t_{2}s_{2}}{\left( n_{2}+m_{2}-t_{2}\right) }\right) \end{aligned}$$

Given that \(P_{2}=A_{2}-b_{2}Q_{2}\) and replacing for \(A_{2}\), \(b_{2}\) and \(Q_{2}\) we get

$$\begin{aligned} P_{2}= & {} \left( \frac{(A_{1}-v_{1})}{\left( n_{2}+m_{2}-t_{2}+1\right) }+\frac{\left( n_{2}+m_{2}-t_{2}\right) v_{2}}{\left( n_{2}+m_{2} -t_{2}+1\right) }\right. \\&\quad \left. +\frac{t_{1}s_{1}}{\left( n_{2}+m_{2}-t_{2}+1\right) \left( n_{1}+m_{1}-t_{1}\right) }-\frac{t_{2}s_{2}}{\left( n_{2} +m_{2}-t_{2}+1\right) }\right) \end{aligned}$$

Replacing \(P_{2}\) into the outputs of the firms in the first tier we obtain

$$\begin{aligned} q_{1i}&=q_{1F}=\frac{q_{1L}}{2}\\&=\frac{\left( m_{2}+n_{2}-t_{2}\right) \left( m_{1}+n_{1}-t_{1}\right) \left( A_{1}-v_{1}-v_{2}\right) }{b_{1}\left( m_{1}+n_{1}-t_{1}+1\right) \left( m_{2}+n_{2}-t_{2}+1\right) \left( m_{1}+n_{1}-t_{1}\right) }\\&\quad + \frac{s_{2}t_{2}\left( m_{1}+n_{1}-t_{1}\right) -s_{1}t_{1}\left( \left( m_{2}+n_{2}-t_{2}+1\right) \left( m_{1}+n_{1}-t_{1}\right) +1\right) }{b_{1}\left( m_{1}+n_{1}-t_{1}+1\right) \left( m_{2} +n_{2}-t_{2}+1\right) \left( m_{1}+n_{1}-t_{1}\right) }\\ q_{1t}&=\frac{\left( A_{1}-v_{1}-v_{2}\right) \left( m_{2}+n_{2} -t_{2}\right) -\frac{t_{1}s_{1}}{\left( n_{1}+m_{1}-t_{1}\right) } +t_{2}s_{2}+s_{1}\left( n_{2}+m_{2}-t_{2}+1\right) \left( n_{1} +m_{1}-2t_{1}+1\right) }{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) \left( n_{2}+m_{2}-t_{2}+1\right) } \end{aligned}$$

For total output, \(Q_{1}=\frac{\left( n_{1}+m_{1}-t_{1}\right) \left( A_{1}-v_{1}-P_{2}\right) +t_{1}s_{1}}{b_{1}\left( n_{1}+m_{1}-t_{1} +1\right) }\), we obtain the same expression as for \(Q_{2}\):

$$\begin{aligned} Q_{1}=\frac{(A_{1}-v_{1}-v_{2})\left( n_{1}+m_{1}-t_{1}\right) (n_{2} +m_{2}-t_{2})+t_{1}s_{1}(n_{2}+m_{2}-t_{2})+t_{2}s_{2}\left( n_{1} +m_{1}-t_{1}\right) }{b_{1}\left( n_{1}+m_{1}-t_{1}+1\right) \left( n_{2}+m_{2}-t_{2}+1\right) } \end{aligned}$$

so \(P_{1}=A_{1}-b_{1}Q_{1}\) yields

$$\begin{aligned} P_{1}=\frac{A_{1}\left( n_{1}+m_{1}-t_{1}+n_{2}+m_{2}-t_{2}+1\right) +(v_{1}+v_{2})\left( n_{1}+m_{1}-t_{1}\right) (n_{2}+m_{2}-t_{2})-t_{1} s_{1}(n_{2}+m_{2}-t_{2})-t_{2}s_{2}\left( n_{1}+m_{1}-t_{1}\right) }{\left( n_{1}+m_{1}-t_{1}+1\right) \left( n_{2}+m_{2}-t_{2}+1\right) }. \end{aligned}$$

Let us denote \(n_{i}+m_{i}-t_{i}\) by \(r_{i}\) (\(i=1,2\)). Then we can simply write:

$$\begin{aligned} Q_{1}&=Q_{2}=\frac{(A_{1}-v_{1}-v_{2})r_{1}r_{2}+t_{1}s_{1}r_{2} +t_{2}s_{2}r_{1}}{b_{1}\left( r_{1}+1\right) \left( r_{2}+1\right) }\\ P_{1}&=\frac{A_{1}\left( r_{1}+r_{2}+1\right) +(v_{1}+v_{2})r_{1} r_{2}-t_{1}s_{1}r_{2}-t_{2}s_{2}r_{1}}{\left( r_{1}+1\right) \left( r_{2}+1\right) }\\ P_{2}&=\frac{r_{1}\left( A_{1}-v_{1}+r_{2}v_{2}-t_{2}s_{2}\right) +t_{1}s_{1}}{r_{1}\left( r_{2}+1\right) } \end{aligned}$$

We now address the conditions needed for positive quantities. As we only deal with a single two-firm merger, we will only consider this case. When there is a multidivisional merger, all outputs are symmetrically divided across firms except for the leader division of the merged firm, that produces twice as much as any other. Hence, if aggregate output is positive (which is always the case), so will be the output of each individual firm. When there is a traditional merger in tier i, then the same happens in tier j: firms are symmetric in the tier in which the merger did not occur and if total output is positive, so will be the output of each individual firm in tier j. In tier i, however, the outsiders will have a lower output because they do not benefit from synergies. Thus we only need to ensure that the outsiders to a traditional merger have a positive equilibrium output. If the merger occurs in the first tier, we need that \(q_{1i}(n_{1},0,1)>0\) and if the merger occurs in the second tier, we need that \(q_{2i}(n_{2},0,1)>0\). Using the expressions above:

$$\begin{aligned}&\frac{n_{2}\left( n_{1}-1\right) \left( A_{1}-v_{1}-v_{2}\right) -s_{1}\left( n_{1}-n_{2}+n_{1}n_{2}\right) }{b_{1}n_{1}\left( n_{1}-1\right) \left( n_{2}+1\right) }>0\Leftrightarrow z_{1}<\frac{n_{2}\left( n_{1}-1\right) }{\left( n_{1}-n_{2}+n_{1}n_{2}\right) }\\&n_{1}\frac{A_{1}-v_{1}-v_{2}-s_{2}}{b_{1}n_{2}\left( n_{1}+1\right) } >0\Leftrightarrow z_{2}<1 \end{aligned}$$

\(\square \)

Proof of Lemma 2

If the multidivisional merger occurs in tier 1, the variation in the downstream price is

$$\begin{aligned} \Delta P_{1}&=\frac{A_{1}\left( n_{1}+n_{2}+2\right) +n_{2}\left( v_{1}+v_{2}\right) \left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }-\frac{(n_{1}+n_{2}+1)A_{1}+n_{1}n_{2}(v_{1}+v_{2})}{(n_{1}+1)(n_{2}+1)}\\&=\frac{-n_{2}\left( A_{1}-v_{1}-v_{2}\right) }{\left( n_{1}+2\right) \left( n_{1}+1\right) \left( n_{2}+1\right) } \end{aligned}$$

If the multidivisional merger occurs in tier 2, the variation in the downstream price is

$$\begin{aligned} \Delta P_{1}&=\frac{A_{1}\left( n_{1}+n_{2}+2\right) +n_{1}\left( v_{1}+v_{2}\right) \left( n_{2}+1\right) }{\left( n_{1}+1\right) \left( n_{2}+2\right) }-\frac{(n_{1}+n_{2}+1)A_{1}+n_{1}n_{2}(v_{1}+v_{2})}{(n_{1}+1)(n_{2}+1)}\\&=\frac{-n_{1}\left( A_{1}-v_{1}-v_{2}\right) }{\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+1\right) } \end{aligned}$$

Price drops more when the merger occurs in tier 1 if and only if\(\ n_{2} >n_{1}{:}\)

$$\begin{aligned}&\frac{n_{2}\left( A_{1}-v_{1}-v_{2}\right) }{\left( n_{1}+2\right) \left( n_{1}+1\right) \left( n_{2}+1\right) }-\frac{n_{1}\left( A_{1}-v_{1} -v_{2}\right) }{\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+1\right) }\\&\quad =-\left( n_{1}-n_{2}\right) \frac{\left( n_{1} +n_{2}+2\right) \left( A_{1}-v_{1}-v_{2}\right) }{\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+2\right) \left( n_{1}+1\right) } \end{aligned}$$

\(\square \)

Proof of Lemma 3

  1. (a)

    If the traditional merger occurs in tier 2, the final price change is

    $$\begin{aligned} \Delta P_{1}&=\frac{A_{1}\left( n_{1}+n_{2}\right) +n_{1}\left( v_{1}+v_{2}\right) \left( n_{2}-1\right) -s_{2}n_{1}}{\left( n_{1}+1\right) n_{2}}\\&\quad -\frac{(n_{1}+n_{2}+1)A_{1}+n_{1}n_{2}(v_{1}+v_{2} )}{(n_{1}+1)(n_{2}+1)}\\&=n_{1}\frac{A_{1}-v_{1}-v_{2}-s_{2}\left( n_{2}+1\right) }{n_{2}\left( n_{2}+1\right) \left( n_{1}+1\right) } \end{aligned}$$
  2. (b)

    If the traditional merger occurs in tier 1, the final price change is

    $$\begin{aligned} \Delta P_{1}&=\frac{A_{1}\left( n_{1}+n_{2}\right) +n_{2}\left( v_{1}+v_{2}\right) \left( n_{1}-1\right) -s_{1}n_{2}}{n_{1}\left( n_{2}+1\right) }\\&\quad -\frac{(n_{1}+n_{2}+1)A_{1}+n_{1}n_{2}(v_{1}+v_{2})}{(n_{1}+1)(n_{2}+1)}\\&=n_{2}\frac{A_{1}-v_{1}-v_{2}-s_{1}\left( n_{1}+1\right) }{n_{1}\left( n_{2}+1\right) \left( n_{1}+1\right) } \end{aligned}$$
  3. (c)

    Price falls more when the merger occurs in tier 1 if and only if

    $$\begin{aligned}&\frac{A_{1}\left( n_{1}+n_{2}\right) +n_{2}\left( v_{1}+v_{2}\right) \left( n_{1}-1\right) -s_{1}n_{2}}{n_{1}\left( n_{2}+1\right) }\\&\quad <\frac{A_{1}\left( n_{1}+n_{2}\right) +n_{1}\left( v_{1}+v_{2}\right) \left( n_{2}-1\right) -s_{2}n_{1}}{\left( n_{1}+1\right) n_{2}}\\&\qquad \Leftrightarrow \frac{\left( n_{1}-n_{2}\right) \left( n_{1}+n_{2}\right) \left( A_{1}-v_{1}-v_{2}\right) -n_{1}^{2}s_{2}\left( n_{2}+1\right) +n_{2} ^{2}s_{1}\left( n_{1}+1\right) }{n_{1}n_{2}\left( n_{2}+1\right) \left( n_{1}+1\right) }>0\\&\qquad \Leftrightarrow z_{1}>\frac{n_{1}^{2}}{n_{2}^{2}}\frac{n_{2}+1}{n_{1}+1}z_{2}-\left( n_{1}-n_{2}\right) \frac{\left( n_{1}+n_{2}\right) }{n_{2}^{2}\left( n_{1}+1\right) } \end{aligned}$$

\(\square \)

Proof of Corollary 1

  1. (a)

    For an equal number of firms in both tiers, \(n_{1}=n_{2}\), the downstream price is lower with the traditional merger that involves the largest cost reductions. This follows from Lemma 3, making \(n_{1}=n_{2}:\) \(z_{1}>z_{2}\).

  2. (b)

    For an equal degree of cost savings, the price downstream is lower when the merger takes place at the downstream tier if and only if \(n_{1}>n_{2}\) and \(z<\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}\), or \(n_{1}<n_{2}\) and \(z>\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}\). This follows from Lemma 3, making \(z_{1}=z_{2}=z\):

    $$\begin{aligned} z\left( n_{1}-n_{2}\right) <\left( n_{1}-n_{2}\right) \frac{n_{1}+n_{2} }{n_{1}+n_{2}+n_{1}n_{2}} \end{aligned}$$
  3. (c)

    In the absence of synergies, \(z_{1}=z_{2}=0\) and the expression in Lemma 3 becomes

    $$\begin{aligned} 0>0-\left( n_{1}-n_{2}\right) \frac{\left( n_{1}+n_{2}\right) }{n_{2} ^{2}\left( n_{1}+1\right) }\Leftrightarrow n_{1}>n_{2}. \end{aligned}$$

\(\square \)

Proof of Lemma 4

Consumers prefer a traditional to a multidivisional downstream merger if and only if

$$\begin{aligned} n_{2}\frac{\left( n_{1}-1\right) +z_{1}}{n_{1}\left( n_{2}+1\right) }>\frac{n_{2}\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }\Leftrightarrow z_{1}>z_{1tm}:=\frac{2}{n_{1}+2} \end{aligned}$$

Consumers prefer a traditional to a multidivisional upstream merger if and only if

$$\begin{aligned} n_{1}\frac{\left( n_{2}-1\right) +z_{2}}{n_{2}\left( n_{1}+1\right) }>\frac{n_{1}\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }\Leftrightarrow z_{2}>z_{2tm}:=\frac{2}{n_{2}+2} \end{aligned}$$

\(\square \)

Proof of Proposition 1

The merger that leads to higher consumer surplus is the one that leads to higher output. Recall that the equilibrium outputs in the four different mergers are:

Equilibrium Output

Merger

Upstream

Downstream

Multidivisional

\(Q_{M2}=n_{1}\frac{\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }\)

\(Q_{M1}=n_{2}\frac{\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }\)

Traditional

\(Q_{T2}=n_{1}\frac{\left( n_{2}-1\right) +z}{n_{2}\left( n_{1}+1\right) }\)

\(Q_{T1}=n_{2}\frac{\left( n_{1}-1\right) +z}{n_{1}\left( n_{2}+1\right) }\)

(a) Consider initially the case of \(n_{1}-n_{2}>0\). Then:

$$\begin{aligned}&Q_{M2}>Q_{M1}\Leftrightarrow \frac{n_{1}\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }>\frac{n_{2}\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }\\&\quad \Leftrightarrow \frac{\left( n_{1}-n_{2}\right) \left( n_{1}+n_{2}+2\right) }{\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+2\right) \left( n_{1}+1\right) }>0\\&Q_{M2}>Q_{T2}\Leftrightarrow \frac{n_{1}\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }>n_{1}\frac{\left( n_{2}-1\right) +z}{n_{2}\left( n_{1}+1\right) }\\&\quad \Leftrightarrow n_{1}\frac{2-z\left( n_{2}+2\right) }{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }>0\Leftrightarrow z<\frac{2}{n_{2}+2}\\&Q_{M2}>Q_{T1}\Leftrightarrow \frac{n_{1}\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }>n_{2}\frac{\left( n_{1}-1\right) +z}{n_{1}\left( n_{2}+1\right) }\\&\quad \Leftrightarrow \frac{2n_{2}+n_{1}^{2} +n_{2}^{2}-zn_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }{n_{1}\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+1\right) }>0\Leftrightarrow z<\frac{2n_{2}+n_{1}^{2}+n_{2}^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }\\&Q_{T2}>Q_{T1}\Leftrightarrow n_{1}\frac{\left( n_{2}-1\right) +z}{n_{2}\left( n_{1}+1\right) }>n_{2}\frac{\left( n_{1}-1\right) +z}{n_{1}\left( n_{2}+1\right) }\\&\quad \Leftrightarrow \left( n_{1}-n_{2}\right) \frac{-\left( n_{1}+n_{2}\right) +z\left( n_{1}+n_{2}+n_{1}n_{2}\right) }{n_{1}n_{2}\left( n_{2}+1\right) \left( n_{1}+1\right) }>0\Leftrightarrow z>\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}} \end{aligned}$$

It is easy to show that

$$\begin{aligned} \frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}<\frac{2}{n_{2}+2}<\frac{2n_{2}+n_{1}^{2}+n_{2}^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) } \end{aligned}$$

This results from the following inequalities:

$$\begin{aligned}&\frac{2n_{2}+n_{1}^{2}+n_{2}^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }-\frac{2}{\left( n_{2}+2\right) } =\frac{\left( n_{1}-n_{2}\right) ^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }>0\\&\frac{2}{n_{2}+2}-\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}} =\frac{n_{2}\left( n_{1}-n_{2}\right) }{\left( n_{2}+2\right) \left( n_{1}+n_{2}+n_{1}n_{2}\right) }>0 \end{aligned}$$

So, for different values of z the output ranking is as follows:

\(z<\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}\)

\(\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}<z<\frac{2}{n_{2}+2}\)

\(\frac{2}{n_{2}+2}<z<\frac{2n_{2}+n_{1}^{2}+n_{2}^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }\)

\(z>\frac{2n_{2}+n_{1}^{2}+n_{2}^{2}}{n_{2}\left( n_{2}+2\right) \left( n_{1}+1\right) }\)

\(Q_{M2}>Q_{T1}>Q_{T2}\)

\(Q_{M2}>Q_{T2}>Q_{T1}\)

\(Q_{T2}>Q_{M2}>Q_{T1}\)

\(Q_{T2}>Q_{T1}>Q_{M2}\)

(b) Consider now the case of \(n_{1}-n_{2}<0\). Then:

$$\begin{aligned}&Q_{M1}>Q_{M2}\Leftrightarrow \frac{n_{2}\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }>\frac{n_{1}\left( n_{2}+1\right) }{\left( n_{2}+2\right) \left( n_{1}+1\right) }\\&\quad \Leftrightarrow \frac{-\left( n_{1}-n_{2}\right) \left( n_{1}+n_{2}+2\right) }{\left( n_{2}+2\right) \left( n_{2}+1\right) \left( n_{1}+2\right) \left( n_{1}+1\right) }>0\\&Q_{M1}>Q_{T2}\Leftrightarrow \frac{n_{2}\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }>n_{1}\frac{\left( n_{2}-1\right) +z}{n_{2}\left( n_{1}+1\right) }\\&\quad \Leftrightarrow \frac{2n_{1}+n_{1}^{2} +n_{2}^{2}-zn_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) }{n_{2}\left( n_{1}+2\right) \left( n_{1}+1\right) \left( n_{2}+1\right) }>0\Leftrightarrow z<\frac{2n_{1}+n_{1}^{2}+n_{2}^{2}}{n_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) }\\&Q_{M1}>Q_{T1}\Leftrightarrow \frac{n_{2}\left( n_{1}+1\right) }{\left( n_{1}+2\right) \left( n_{2}+1\right) }>n_{2}\frac{\left( n_{1}-1\right) +z}{n_{1}\left( n_{2}+1\right) }\\&\quad \Leftrightarrow n_{2}\frac{2-z\left( n_{1}+2\right) }{n_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) }>0\Leftrightarrow z<\frac{2}{n_{1}+2}\\&Q_{T2}>Q_{T1}\Leftrightarrow n_{1}\frac{\left( n_{2}-1\right) +z}{n_{2}\left( n_{1}+1\right) }>n_{2}\frac{\left( n_{1}-1\right) +z}{n_{1}\left( n_{2}+1\right) }\\&\quad \Leftrightarrow \left( n_{1}-n_{2}\right) \frac{-\left( n_{1}+n_{2}\right) +z\left( n_{1}+n_{2}+n_{1}n_{2}\right) }{n_{1}n_{2}\left( n_{2}+1\right) \left( n_{1}+1\right) }>0\Leftrightarrow z<\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}} \end{aligned}$$

As before,

$$\begin{aligned} \frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}<\frac{2}{n_{1}+2}<\frac{2n_{1}+n_{1}^{2}+n_{2}^{2}}{n_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) } \end{aligned}$$

So, for different values of z, the output ranking is as follows:

\(z<\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}\)

\(\frac{n_{1}+n_{2}}{n_{1}+n_{2}+n_{1}n_{2}}<z<\frac{2}{n_{1}+2}\)

\(\frac{2}{n_{1}+2}<z<\frac{2n_{1}+n_{1}^{2}+n_{2}^{2}}{n_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) }\)

\(z>\frac{2n_{1}+n_{1}^{2}+n_{2}^{2}}{n_{1}\left( n_{2}+1\right) \left( n_{1}+2\right) }\)

\(Q_{M1}>Q_{T2}>Q_{T1}\)

\(Q_{M1}>Q_{T1}>Q_{T2}\)

\(Q_{T1}>Q_{M1}>Q_{T2}\)

\(Q_{T1}>Q_{T2}>Q_{M1}\)

\(\square \)

Proof of Lemma 5

  1. (a)

    A multidivisional downstream merger is profitable if and only if

    $$\begin{aligned}&\frac{2n_{2}^{2}}{\left( n_{1}+2\right) ^{2}\left( n_{2}+1\right) ^{2} }+\frac{n_{2}^{2}}{\left( n_{1}+2\right) ^{2}\left( n_{2}+1\right) ^{2}}>\frac{2n_{2}^{2}}{(n_{2}+1)^{2}(n_{1}+1)^{2}}\\&\quad \Leftrightarrow \frac{n_{2}^{2}\left( -2n_{1}+n_{1}^{2}-5\right) }{\left( n_{1}+2\right) ^{2}\left( n_{1}+1\right) ^{2}\left( n_{2}+1\right) ^{2}} >0 \end{aligned}$$

    A multidivisional upstream merger is profitable if and only if

    $$\begin{aligned}&\frac{n_{1}}{\left( n_{1}+1\right) \left( n_{2}+2\right) ^{2}} +\frac{2n_{1}}{\left( n_{1}+1\right) \left( n_{2}+2\right) ^{2}}>\frac{2n_{1}}{(n_{1}+1)(n_{2}+1)^{2}}\\&\quad \Leftrightarrow \frac{n_{1}\left( -2n_{2}+n_{2}^{2}-5\right) }{\left( n_{2}+2\right) ^{2}\left( n_{2}+1\right) ^{2}\left( n_{1}+1\right) } >0 \end{aligned}$$

    We have that \(\left( -2n_{i}+n_{i}^{2}-5\right) >0\) for any \(n_{i}\ge 4\).

  2. (b)

    A traditional upstream merger is profitable if and only if

    $$\begin{aligned} n_{1}\frac{\left( 1+z_{2}\left( n_{2}-1\right) \right) ^{2}}{n_{2} ^{2}\left( n_{1}+1\right) }>\frac{2n_{1}}{(n_{1}+1)(n_{2}+1)^{2} }\Leftrightarrow z_{2}>z_{2TN}:=\frac{\left( \sqrt{2}-1\right) n_{2} -1}{n_{2}^{2}-1} \end{aligned}$$
  3. (c)

    A traditional downstream merger is profitable if and only if

    $$\begin{aligned}&\frac{\left( n_{2}\left( n_{1}-1\right) +z_{1}\left( n_{2}+n_{1}\left( n_{2}+1\right) \left( n_{1}-2\right) \right) \right) ^{2}}{n_{1} ^{2}\left( n_{2}+1\right) ^{2}\left( n_{1}-1\right) ^{2}}>\frac{2n_{2} ^{2}}{(n_{2}+1)^{2}(n_{1}+1)^{2}}\\&\quad \Leftrightarrow z_{1}>z_{1TN}:=n_{2}\frac{n_{1}-1}{n_{1}+1}\frac{\left( \sqrt{2}-1\right) n_{1}-1}{n_{1}\left( n_{2}+1\right) \left( n_{1}-2\right) +n_{2}} \end{aligned}$$

\(\square \)

Proof of Proposition 2

A pair of firms in any tier will propose the merger that is more profitable.

  1. (a)

    A multidivisional downstream merger is more profitable than a traditional downstream merger if and only if

    $$\begin{aligned}&\frac{2n_{2}^{2}}{\left( n_{1}+2\right) ^{2}\left( n_{2}+1\right) ^{2} }+\frac{n_{2}^{2}}{\left( n_{1}+2\right) ^{2}\left( n_{2}+1\right) ^{2}} \\&\quad >\frac{\left( n_{2}\left( n_{1}-1\right) +z_{1}\left( n_{2} +n_{1}\left( n_{2}+1\right) \left( n_{1}-2\right) \right) \right) ^{2} }{n_{1}^{2}\left( n_{2}+1\right) ^{2}\left( n_{1}-1\right) ^{2}}\\&\qquad \Leftrightarrow z_{1} <z_{1TM}:=\frac{n_{2}(n_{1}-1)\left( \left( \sqrt{3}-1\right) n_{1}-2\right) }{n_{1}(n_{1}^{2}(n_{2}+1)-3n_{2}-4)+2n_{2}} \end{aligned}$$
  2. (b)

    A multidivisional upstream merger is more profitable than a traditional upstream merger if and only if

    $$\begin{aligned}&\frac{n_{1}}{\left( n_{1}+1\right) \left( n_{2}+2\right) ^{2}} +\frac{2n_{1}}{\left( n_{1}+1\right) \left( n_{2}+2\right) ^{2}} >n_{1}\frac{\left( 1+z_{2}\left( n_{2}-1\right) \right) ^{2}}{n_{2} ^{2}\left( n_{1}+1\right) }\\&\quad \Leftrightarrow z_{2} <z_{2TM}:=\frac{\left( \sqrt{3}-1\right) n_{2}-2}{\left( n_{2}+2\right) \left( n_{2}-1\right) } \end{aligned}$$

\(\square \)

Proof of Proposition 3

The result follows directly from Lemma 4 and Proposition 2. The intervals always exist:

$$\begin{aligned} z_{1TM}-z_{1tm}=n_{1}\frac{2\left( n_{1}-2\right) +\left( 3-\sqrt{3}\right) n_{2}\left( n_{1}-1\right) }{\left( n_{1}+2\right) \left( -n_{1}^{2}\left( n_{2}+1\right) +n_{1}\left( 2n_{2}+2\right) -n_{2}\right) }<0 \end{aligned}$$

and

$$\begin{aligned} z_{2TM}-z_{2tm}=\frac{\left( \sqrt{3}-3\right) n_{2}}{\left( n_{2} -1\right) \left( n_{2}+2\right) }<0. \end{aligned}$$

Note that \(\left( -n_{1}^{2}\left( n_{2}+1\right) +2n_{1}\left( n_{2}+1\right) -n_{2}\right) \) is maximized at \(n_{1}=1\) and at \(n_{1}=2\) takes value \(-n_{2}<0\). Thus, it is always negative for \(n_{1}\ge 4\). \(\square \)

Proof of Proposition 4

The proof follows from combining the results in Lemma 3 and Proposition 2. The intervals for \(z_{i}\) are non-empty for all parameter values because

$$\begin{aligned} z_{1tn}-z_{1TM}&=\frac{1}{n_{1}+1}-\frac{n_{2}(n_{1}-1)\left( \left( \sqrt{3}-1\right) n_{1}-2\right) }{n_{1}(n_{1}^{2}(n_{2}+1)-3n_{2}-4)+2n_{2}}\\&=n_{1}\frac{\left( 2-\sqrt{3}\right) n_{2}\left( n_{1}-1\right) \left( n_{1}+2\sqrt{3}+5\right) +\left( n_{1}-2\right) \left( n_{1}+2\right) }{\left( n_{1}+2\right) \left( n_{1}+1\right) \left( n_{2}+n_{1}\left( n_{2}+1\right) \left( n_{1}-2\right) \right) }>0\\ \frac{1}{n_{2}+1}-z_{2TM}&=\frac{1}{n_{2}+1}-\frac{\left( \sqrt{3}-1\right) n_{2}-2}{n_{2}^{2}+n_{2}-2}\\&=\frac{\left( 2-\sqrt{3}\right) n_{2}\left( n_{2}+2\sqrt{3}+5\right) }{\left( n_{2}-1\right) \left( n_{2}+1\right) \left( n_{2}+2\right) }>0 \end{aligned}$$

\(\square \)

Appendix 2

In this appendix we present most of our results for the case in which a given number of traditional mergers or multidivisional mergers may have occurred in any tier before the additional merger under analysis. We start with four useful remarks that include expressions that will be relevant in the proofs, as well as parameter constraints.

Remark 1

For any \(n_{i}\) we have \(r_{i}=n_{i}+m_{i}-t_{i}\). Given that we are assuming that at the current industry structure it is possible to have an additional two-firm merger and that there is at least one surviving outsider after the merger, the maximum number of previous multidivisional or traditional two-firm mergers is \((n_{i}-3)/2\). So \(n_{i}-(n_{i} -3)/2<r_{i}<n_{i}+(n_{i}-3)/2,\) where the lower bound corresponds to all previous mergers being traditional and the upper bound corresponds to all previous mergers being multidivisional. As we assume \(n_{i}\ge 4\), we have that \(r_{i}>4-(4-3)/2=3.5\). As the number of firms is an integer, we assume \(r_{i}\ge 4\). After any number of traditional mergers in tier i, it is possible that there is at least an additional merger between previously non-merged firms and at least a single outsider. So, \(r_{i}>t_{i}\).

Remark 2

From Lemma 1, the equilibrium output and prices are given by:

$$\begin{aligned} Q_{1}(r_{1},r_{2},t_{1},t_{2})&=Q_{2}(r_{1},r_{2},t_{1},t_{2})=\frac{r_{1}r_{2}+r_{2}t_{1}z_{1}+r_{1}t_{2}z_{2}}{\left( r_{1}+1\right) \left( r_{2}+1\right) }\\ P_{1}(r_{1},r_{2},t_{1},t_{2})&=\frac{A_{1}\left( r_{1}+r_{2}+1\right) +(v_{1}+v_{2})r_{1}r_{2}-t_{1}s_{1}r_{2}-t_{2}s_{2}r_{1}}{\left( r_{1}+1\right) \left( r_{2}+1\right) } \\ P_{2}(r_{1},r_{2},t_{1},t_{2})&=\frac{r_{1}\left( A_{1}-v_{1}+r_{2} v_{2}-t_{2}s_{2}\right) +t_{1}s_{1}}{r_{1}\left( r_{2}+1\right) } \end{aligned}$$
(2)

where \(r_{i}=n_{i}+m_{i}-t_{i}\) and \(s_{i}=z_{i}(A_{1}-v_{1}-v_{2})\) (\(i=1,2\))

Individual outputs, divided by \(\frac{A_{1}-v_{1}-v_{2}}{b_{1}}\) are:

$$\begin{aligned} q_{1i}&=q_{1F}=\frac{q_{1L}}{2}=q_{1t}-z_{1}=\frac{r_{1}\left( r_{2}\left( 1-t_{1}z_{1}\right) +t_{2}z_{2}\right) -r_{1}t_{1}z_{1} -t_{1}z_{1}}{r_{1}\left( r_{2}+1\right) \left( r_{1}+1\right) }\\ q_{2i}&=q_{2F}=\frac{r_{1}\left( 1-t_{2}z_{2}\right) +t_{1}z_{1} }{\left( r_{1}+1\right) \left( r_{2}+1\right) }\\ q_{2L}&=2\frac{r_{1}\left( 1-t_{2}z_{2}\right) +t_{1}z_{1}}{\left( r_{1}+1\right) \left( r_{2}+1\right) }\\ q_{2t}&=\frac{r_{1}\left( 1-t_{2}z_{2}\right) +t_{1}z_{1}+r_{1} z_{2}\left( r_{2}+1\right) }{\left( r_{1}+1\right) \left( r_{2}+1\right) } \end{aligned}$$

We need \(t_{2}z_{2}<1\) to ensure a positive \(q_{2i}=q_{2F}\) in the absence of downstream traditional mergers \((t_{1}=0)\). This implies that all outputs in tier 2 are positive, regardless of the numbers of firms or previous mergers.

So, we only need to ensure that \(q_{1i}>0,\) that is:

$$\begin{aligned} r_{1}\left( r_{2}\left( 1-t_{1}z_{1}\right) +t_{2}z_{2}\right) -r_{1} t_{1}z_{1}-t_{1}z_{1}>0\Leftrightarrow z_{1}<\frac{r_{1}\left( r_{2} +t_{2}z_{2}\right) }{t_{1}\left( r_{1}+r_{1}r_{2}+1\right) } \end{aligned}$$

It is easy to check that this threshold increases with \(r_{1}\) and \(r_{2}\), meaning that additional multidivisional mergers make this condition easier to verify.

With an additional traditional downstream merger the condition becomes

$$\begin{aligned} z_{1}<\frac{(r_{1}-1)\left( r_{2}+t_{2}z_{2}\right) }{\left( (r_{1} -1)+(r_{1}-1)r_{2}+1\right) (t_{1}+1)} \end{aligned}$$

which is a stronger condition. With an additional traditional upstream merger the condition becomes

$$\begin{aligned} z_{1}<\frac{r_{1}\left( (r_{2}-1)+(t_{2}+1)z_{2}\right) }{\left( r_{1}+r_{1}(r_{2}-1)+1\right) t_{1}} \end{aligned}$$

So, the following constraint on cost savings ensures that all outputs are positive:

$$\begin{aligned} z_{1}&<\overline{z_{1}}:=\min \left\{ \frac{(r_{1}-1)\left( r_{2} +t_{2}z_{2}\right) }{\left( (r_{1}-1)+(r_{1}-1)r_{2}+1\right) (t_{1} +1)},\frac{r_{1}\left( (r_{2}-1)+(t_{2}+1)z_{2}\right) }{\left( r_{1} +r_{1}(r_{2}-1)+1\right) t_{1}}\right\} \\ z_{2}&<\overline{z_{2}}:=1/t_{2} \end{aligned}$$

Note that these imply

$$\begin{aligned} t_{1}z_{1}&<\frac{r_{1}\left( r_{2}+t_{2}z_{2}\right) }{\left( r_{1}+r_{1}r_{2}+1\right) }<1+\frac{t_{2}z_{2}}{r_{2}}\\ t_{2}z_{2}&<1+\frac{t_{1}z_{1}}{r_{1}} \end{aligned}$$

Remark 3

Individual profits, divided by \(\frac{\left( A_{1} -v_{1}-v_{2}\right) ^{2}}{b_{1}}\), are:

$$\begin{aligned} \pi _{1i}(r_{1},r_{2},t_{1},t_{2})&=\frac{\left( -r_{1}r_{2}+t_{1} z_{1}\left( r_{1}+r_{1}r_{2}+1\right) -r_{1}t_{2}z_{2}\right) ^{2}}{r_{1}^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}}\\ \pi _{1m}(r_{1},r_{2},t_{1},t_{2})&=3\left( r_{1}r_{2}\left( t_{1} z_{1}-1\right) +t_{1}z_{1}+r_{1}t_{1}z_{1}-r_{1}t_{2}z_{2}\right) \\&\quad *\frac{-r_{1}r_{2}+\left( r_{1}+r_{1}r_{2}+1\right) t_{1}z_{1}-r_{1} t_{2}z_{2}}{r_{1}^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}}\\ \pi _{1t}(r_{1},r_{2},t_{1},t_{2})&=\frac{\left( z_{1}\left( r_{1} ^{2}\left( r_{2}+1\right) -r_{1}\left( t_{1}-1\right) \left( r_{2}+1\right) -t_{1}\right) +r_{1}\left( r_{2}+t_{2}z_{2}\right) \right) ^{2}}{r_{1}^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}}\\ \pi _{2i}(r_{1},r_{2},t_{1},t_{2})&=\frac{\left( -r_{1}-t_{1}z_{1} +r_{1}t_{2}z_{2}\right) ^{2}}{r_{1}\left( r_{1}+1\right) \left( r_{2}+1\right) ^{2}}\\ \pi _{2m}(r_{1},r_{2},t_{1},t_{2})&=\frac{3\left( -r_{1}-t_{1}z_{1} +r_{1}t_{2}z_{2}\right) ^{2}}{r_{1}\left( r_{2}+1\right) ^{2}\left( r_{1}+1\right) }\\ \pi _{2t}(r_{1},r_{2},t_{1},t_{2})&=\frac{\left( r_{1}+r_{1}z_{2} +t_{1}z_{1}+r_{1}r_{2}z_{2}-r_{1}t_{2}z_{2}\right) ^{2}}{\left( r_{1}+1\right) \left( r_{2}+1\right) ^{2}r_{1}} \end{aligned}$$

Remark 4

Throughout the proofs it will be necessary to solve several inequations of the type \(f(x)=ax^{2}+bx+c>0\) with \(x\in \left[ 0,\overline{x}\right] ,\) where the sign of a is uncertain, \(c<0\) and \(b>0\). Note that under these circumstances \(f(0)=c<0\) and \(\frac{\partial f}{\partial x}(x=0)=b>0\). If \(f({\overline{x}})>0\) the solution is: if \(a>0\): \(x>\max \left\{ \frac{-b+\sqrt{b^{2}-4ac}}{2a},\frac{-b-\sqrt{b^{2}-4ac}}{2a}\right\} =\frac{-b+\sqrt{b^{2}-4ac}}{2a}\). If \(a<0,\) \(x>\min \left\{ \frac{-b+\sqrt{b^{2}-4ac}}{2a},\frac{-b-\sqrt{b^{2}-4ac}}{2a}\right\} =\frac{-b+\sqrt{b^{2}-4ac}}{2a}\). So, whenever we have this type of inequality, the solution is \(x>\frac{-b+\sqrt{b^{2}-4ac}}{2a}\).

Lemma 2B

A multidivisional merger decreases final consumer price, independently of the tier where the merger takes place. \(\square \)

Proof of Lemma 2B

Downstream price is given by

$$\begin{aligned} P_{1}(r_{1},r_{2},t_{1},t_{2})=\frac{A_{1}\left( r_{1}+r_{2}+1\right) +(v_{1}+v_{2})r_{1}r_{2}-t_{1}z_{1}(A_{1}-v_{1}-v_{2})r_{2}-t_{2}z_{2} (A_{1}-v_{1}-v_{2})r_{1}}{\left( r_{1}+1\right) \left( r_{2}+1\right) } \end{aligned}$$

Let \(\Delta P_{1}^{mi}\) denote the change in downstream price after an additional multidivisional merger at tier i.

An additional multidivisional two-firm merger at tier i increases \(r_{i}\) by 1. Thus, the change in downstream price is given by \(\Delta P_{1}^{mi} =P_{1}(r_{i}+1,r_{j},t_{i},t_{j})-P_{1}(r_{i},r_{j},t_{i},t_{j}).\)

After a merger in tier 1, the change in price is then

$$\begin{aligned} \Delta P_{1}^{m1}=-\left( A_{1}-v_{1}-v_{2}\right) \frac{r_{2}\left( 1-t_{1}z_{1}\right) +t_{2}z_{2}}{\left( r_{1}+2\right) \left( r_{1}+1\right) \left( r_{2}+1\right) }<0 \end{aligned}$$

as \(r_{2}\left( 1-t_{1}z_{1}\right) +t_{2}z_{2}>0.\)

After a merger in tier 2, the change in price is then

$$\begin{aligned} \Delta P_{1}^{m2}=-\left( A_{1}-v_{1}-v_{2}\right) \frac{r_{1}\left( 1-t_{2}z_{2}\right) +t_{1}z_{1}}{\left( r_{2}+2\right) \left( r_{2}+1\right) \left( r_{1}+1\right) }<0 \end{aligned}$$

as \(r_{1}\left( 1-t_{2}z_{2}\right) +t_{1}z_{1}>0\). \(\square \)

Lemma 3B

(a) In the presence of a traditional upstream merger, the downstream price falls if and only if \(z_{2}>z_{2tn}:=\frac{r_{1}+t_{1}z_{1} }{r_{1}\left( r_{2}+t_{2}+1\right) }\). (b) In the presence of a traditional downstream merger, the downstream price falls if and only if \(z_{1} >z_{1tn}:=\frac{r_{2}+t_{2}z_{2}}{r_{2}\left( r_{1}+t_{1}+1\right) }\). (c) The downstream price is lower after the traditional downstream merger than after the traditional upstream merger if and only if \(z_{1}>z_{2} \frac{r_{1}^{2}\left( r_{2}+1\right) +t_{2}\left( r_{2}+r_{1}^{2}\right) }{r_{2}^{2}\left( r_{1}+1\right) +t_{1}\left( r_{1}+r_{2}^{2}\right) }-\left( r_{1}-r_{2}\right) \frac{r_{1}+r_{2}}{r_{2}^{2}\left( r_{1}+1\right) +t_{1}\left( r_{1}+r_{2}^{2}\right) }\). \(\square \)

Proof of Lemma 3B

Likewise, let \(\Delta P_{1}^{ti}\) denote the change in downstream price after an additional traditional merger at tier i. A traditional merger in tier i means increasing \(t_{i}\) by 1 which decreases \(r_{i}\) by 1. Thus,

$$\begin{aligned} \Delta P_{1}^{ti}=P_{1}(r_{i}-1,r_{j},t_{i}+1,t_{j})-P_{1}(r_{i},r_{j} ,t_{i},t_{j}). \end{aligned}$$

(a) If \(i=2\) (upstream merger), the change in price is given by

$$\begin{aligned} \Delta P_{1}^{t2}=\left( A_{1}-v_{1}-v_{2}\right) \frac{r_{1}+t_{1} z_{1}-r_{1}z_{2}\left( r_{2}+t_{2}+1\right) }{r_{2}\left( r_{2}+1\right) \left( r_{1}+1\right) } \end{aligned}$$

which is negative if and only if

$$\begin{aligned} z_{2}>\frac{r_{1}+t_{1}z_{1}}{r_{1}\left( r_{2}+t_{2}+1\right) } \end{aligned}$$

(b) If \(i=1\) (downstream merger), the change in price is given by

$$\begin{aligned} \Delta P_{1}^{t1}=\left( A_{1}-v_{1}-v_{2}\right) \frac{r_{2}+t_{2} z_{2}-r_{2}z_{1}\left( r_{1}+t_{1}+1\right) }{r_{1}\left( r_{2}+1\right) \left( r_{1}+1\right) } \end{aligned}$$

which is negative if and only if

$$\begin{aligned} z_{1}>\frac{r_{2}+t_{2}z_{2}}{r_{2}\left( r_{1}+t_{1}+1\right) } \end{aligned}$$

(c) This follows from

$$\begin{aligned}&P_{1}(r_{1},r_{2}-1,t_{1},t_{2}+1)>P_{1}(r_{1}-1,r_{2},t_{1}+1,t_{2} )\\&\quad \Leftrightarrow z_{1}>z_{2}\frac{r_{1}^{2}\left( r_{2}+1\right) +t_{2}\left( r_{2} +r_{1}^{2}\right) }{r_{2}^{2}\left( r_{1}+1\right) +t_{1}\left( r_{1}+r_{2}^{2}\right) }-\left( r_{1}-r_{2}\right) \frac{r_{1}+r_{2}}{r_{2}^{2}\left( r_{1}+1\right) +t_{1}\left( r_{1}+r_{2}^{2}\right) } \end{aligned}$$

\(\square \)

Corollary 1B

  1. (a)

    For an equal industry structure in both tiers, \(r_{1}=r_{2}\) and \(t_{1}=t_{2}\), the downstream price is lower after the traditional merger (upstream or downstream) that involves the largest cost reductions.

  2. (b)

    For an equal degree of cost savings, \(z_{1}=z_{2}=z,\) the downstream price is lower after the traditional downstream merger than after the traditional upstream merger if and only if \(\left( r_{1}-r_{2}\right) \left( r_{1}+r_{2}+r_{1}r_{2}\right) +t_{2}\left( r_{2}+r_{1}^{2}\right) -t_{1}\left( r_{1}+r_{2}^{2}\right) >0\) and \(z<\frac{\left( r_{1} -r_{2}\right) \left( r_{1}+r_{2}\right) }{\left( r_{1}-r_{2}\right) \left( r_{1}+r_{2}+r_{1}r_{2}\right) +t_{2}\left( r_{2}+r_{1}^{2}\right) -t_{1}\left( r_{1}+r_{2}^{2}\right) }\) , or

    \(\left( r_{1}-r_{2}\right) \left( r_{1}+r_{2}+r_{1}r_{2}\right) +t_{2}\left( r_{2}+r_{1}^{2}\right) -t_{1}\left( r_{1}+r_{2}^{2}\right) <0\) and \(z>\frac{\left( r_{1}-r_{2}\right) \left( r_{1}+r_{2}\right) }{\left( r_{1}-r_{2}\right) \left( r_{1}+r_{2}+r_{1}r_{2}\right) +t_{2}\left( r_{2}+r_{1}^{2}\right) -t_{1}\left( r_{1}+r_{2}^{2}\right) }\).

  3. (c)

    In the absence of synergies, the downstream price is lower after the traditional downstream merger than after the traditional upstream merger if and only if there are more “equivalent firms” downstream \((r_{1}>r_{2})\).

\(\square \)

Proof of Corollary 1B

  1. (a)

    Making \(r_{1}=r_{2}\) and \(t_{1}=t_{2}\) in the expression of Lemma 3(c) it results immediately that the downstream price is lower after the traditional downstream merger than after the traditional upstream merger if and only if \(z_{1}>z_{2}\).

  2. (b)

    This follows from Lemma 3(c), making \(z_{1}=z_{2}=z\).

  3. (c)

    This results immediately from Lemma 3(c), making \(z_{1}=z_{2} =0\).

\(\square \)

Lemma 4B

Consumers prefer a traditional to a multidivisional merger in tier i if and only if \(z_{i}>z_{itm}:=\frac{r_{j}+t_{j}z_{j}}{r_{j}} \frac{2}{r_{i}+2t_{i}+2}\). \(\square \)

Proof of Lemma 4B

This happens when

$$\begin{aligned} P_{1}(r_{i}-1,r_{j},t_{i}+1,t_{j})-P_{1}(r_{i}+1,r_{j},t_{i},t_{j})<0 \end{aligned}$$

If \(i=1\) (downstream merger), this is equivalent to:

$$\begin{aligned}&\left( A_{1}-v_{1}-v_{2}\right) \frac{2\left( r_{2}+t_{2}z_{2}\right) -r_{2}z_{1}\left( r_{1}+2t_{1}+2\right) }{r_{1}\left( r_{2}+1\right) \left( r_{1}+2\right) }<0\\&\quad \Leftrightarrow z_{1}>\frac{r_{2}+t_{2}z_{2}}{r_{2}}\frac{2}{r_{1}+2t_{1}+2} \end{aligned}$$

If \(i=2\) (upstream merger), this is equivalent to:

$$\begin{aligned}&\left( A_{1}-v_{1}-v_{2}\right) \frac{2\left( r_{1}+t_{1}z_{1}\right) -r_{1}z_{2}\left( r_{2}+2t_{2}+2\right) }{r_{2}\left( r_{2}+2\right) \left( r_{1}+1\right) }<0\\&\quad \Leftrightarrow z_{2}>\frac{r_{1}+t_{1}z_{1}}{r_{1}}\frac{2}{r_{2}+2t_{2}+2} \end{aligned}$$

\(\square \)

Lemma 5B

(a) A multidivisional merger in any tier is profitable. (b) A traditional upstream merger is profitable if and only if \(z_{2}>z_{2TN} \). (c) A traditional downstream merger is profitable if and only if \(z_{1}>z_{1TN}\). \(\square \)

Proof of Lemma 5B

(a) If the multidivisional merger takes place at tier 1, it is profitable if

$$\begin{aligned} \pi _{1m}(r_{1}+1,r_{2},t_{1},t_{2})-2\pi _{1i}(r_{1},r_{2},t_{1},t_{2})>0 \end{aligned}$$

which is equivalent to

$$\begin{aligned} \frac{\left( \begin{array}{c} \left( r_{2}^{2}r_{1}^{2}\left( r_{1}^{2}-2r_{1}-5\right) -r_{2}\left( -2r_{1}^{4}+2r_{1}^{3}+20r_{1}^{2}+16r_{1}\right) +\left( -4r_{1}+r_{1} ^{2}-2\right) \left( r_{1}+2\right) ^{2}\right) z_{1}^{2}t_{1} ^{2}\\ -2r_{1}\left( r_{1}^{3}\left( r_{2}+1\right) -r_{1}^{2}\left( 2r_{2}+1\right) -r_{1}\left( 5r_{2}+10\right) -8\right) \left( r_{2}+t_{2}z_{2}\right) z_{1}t_{1}\\ +r_{1}^{2}\left( r_{2}+t_{2}z_{2}\right) ^{2}\left( r_{1}^{2} -2r_{1}-5\right) \end{array} \right) }{r_{1}^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}}>0 \end{aligned}$$
(3)

The coefficient of \(z_{1}^{2}t_{1}^{2}\) is always positive: It is a U-shaped parabola in \(r_{2}\) (because \(r_{1}^{2}\left( r_{1}^{2}-2r_{1}-5\right) >0,\) \(\forall r_{1}\ge 4)\) which, evaluated at \(r_{2}=4,\) takes value \(-88r_{1}-174r_{1}^{2}-40r_{1}^{3}+25r_{1}^{4}-8>0\). It is also increasing at that point because

$$\begin{aligned}&2*4*r_{1}^{2}\left( r_{1}^{2}-2r_{1}-5\right) -\left( -2r_{1} ^{4}+2r_{1}^{3}+20r_{1}^{2}+16r_{1}\right) \\&\quad =2r_{1}\left( -30r_{1}-9r_{1} ^{2}+5r_{1}^{3}-8\right) >0 \end{aligned}$$

which proves that

$$\begin{aligned}&\left( r_{2}^{2}r_{1}^{2}\left( r_{1}^{2}-2r_{1}-5\right) -r_{2}\left( -2r_{1}^{4}+2r_{1}^{3}+20r_{1}^{2}+16r_{1}\right) \right. \\&\quad \left. +\left( -4r_{1}+r_{1} ^{2}-2\right) \left( r_{1}+2\right) ^{2}\right) >0 \end{aligned}$$

and hence the numerator in (3) is a U-shaped parabola in \(t_{1}z_{1}\).

Evaluating the numerator of (3) at \(t_{1}z_{1}=\frac{r_{1}\left( r_{2}+t_{2}z_{2}\right) }{\left( r_{1}+r_{1}r_{2}+1\right) }\) we obtain \(\frac{3r_{1}^{2}\left( r_{2}+t_{2}z_{2}\right) ^{2}}{\left( r_{1} +r_{1}r_{2}+1\right) ^{2}}>0\). Moreover, (3) is decreasing at that point: the derivative is \(-6r_{1}^{2}\left( r_{2}+t_{2}z_{2}\right) \frac{r_{1}+r_{2}+r_{1}r_{2}+2}{r_{1}+r_{1}r_{2}+1}<0.\ \)Thus, (3) is always positive for \(z_{1}t_{1}\in \left[ 0,\frac{r_{1}\left( r_{2} +t_{2}z_{2}\right) }{\left( r_{1}+r_{1}r_{2}+1\right) }\right] \).

If the multidivisional merger takes place at tier 2, it is profitable if

$$\begin{aligned} \pi _{2m}(r_{1},r_{2}+1,t_{1},t_{2})-2\pi _{2i}(r_{1},r_{2},t_{1},t_{2})>0. \end{aligned}$$

This is equivalent to

$$\begin{aligned} \frac{\left( r_{2}^{2}-2r_{2}-5\right) \left( -r_{1}-t_{1}z_{1}+r_{1} t_{2}z_{2}\right) ^{2}}{r_{1}\left( r_{2}+2\right) ^{2}\left( r_{2}+1\right) ^{2}\left( r_{1}+1\right) }>0 \end{aligned}$$

which is always true.

(b) If the traditional merger takes place at tier 2, it is profitable if

$$\begin{aligned} \pi _{2t}(r_{1},r_{2}-1,t_{1},t_{2}+1)-2\pi _{2i}(r_{1},r_{2},t_{1},t_{2})>0 \end{aligned}$$

which is equivalent to

$$\begin{aligned} \frac{\left( \begin{array}{c} r_{1}^{2}\left( 2t_{2}+2r_{2}t_{2}^{2}-2r_{2}^{2}t_{2}-2r_{2}^{3}t_{2} -2r_{2}^{2}+r_{2}^{4}+t_{2}^{2}-r_{2}^{2}t_{2}^{2}+2r_{2}t_{2}+1\right) z_{2}^{2}\\ +2r_{1}\left( t_{2}\left( -2r_{2}+r_{2}^{2}-1\right) +r_{2}^{2}-r_{2} +r_{2}^{3}-1\right) \left( r_{1}+t_{1}z_{1}\right) z_{2}\\ +\left( 2r_{2}-r_{2}^{2}+1\right) \left( r_{1}+t_{1}z_{1}\right) ^{2} \end{array} \right) }{r_{1}r_{2}^{2}\left( r_{2}+1\right) ^{2}\left( r_{1}+1\right) } \end{aligned}$$

At \(z_{2}=0\) this is negative and increasing, respectively because \(\left( 2r_{2}-r_{2}^{2}+1\right) <0\) and

$$\begin{aligned} t_{2}\left( -2r_{2}+r_{2}^{2}-1\right) +r_{2}^{2}-r_{2}+r_{2}^{3}-1>0. \end{aligned}$$

At \(z_{2}=\frac{1+\frac{t_{1}z_{1}}{r_{1}}}{t_{2}}\) (the maximum admissible value), it is positive (\(\left( r_{2}-1\right) ^{2}\frac{\left( r_{1} +t_{1}z_{1}\right) ^{2}}{r_{1}r_{2}^{2}t_{2}^{2}\left( r_{1}+1\right) }>0\)).

Then, from Remark 4,

$$\begin{aligned} z_{2}>z_{2TN}:=\frac{r_{1}+t_{1}z_{1}}{r_{1}}\frac{\left( r_{2}-1\right) \left( r_{2}+1\right) \left( r_{2}(\sqrt{2}-1)-1\right) -t_{2}\left( r_{2}^{2}-2r_{2}-1\right) }{t_{2}^{2}\left( 2r_{2}-r_{2}^{2}+1\right) -2t_{2}\left( r_{2}-1\right) \left( r_{2}+1\right) ^{2}+\left( r_{2}-1\right) ^{2}\left( r_{2}+1\right) ^{2}} \end{aligned}$$

(c) If the traditional merger takes place at tier 1, it is profitable if

$$\begin{aligned} \pi _{1t}(r_{1}-1,r_{2},t_{1}+1,t_{2})-2\pi _{1i}(r_{1},r_{2},t_{1},t_{2})>0 \end{aligned}$$

which is equivalent to

$$\begin{aligned} \frac{\left( \begin{array}{c} \left( \begin{array}{c} t_{1}^{2}\left( 2r_{1}^{2}r_{2}+6r_{1}^{3}r_{2}-2r_{1}^{4}r_{2}+5r_{1} ^{2}+2r_{1}^{3}+r_{2}^{2}-r_{1}^{4}-4r_{1}^{2}r_{2}^{2}+4r_{1}^{3}r_{2} ^{2}-r_{1}^{4}r_{2}^{2}-6r_{1}r_{2}-2\right) +\\ -2t_{1}\left( r_{1}+1\right) ^{2}\left( r_{2}\left( r_{1}-1\right) ^{2}+r_{1}\left( r_{1}-2\right) \right) \left( r_{1}-r_{2}+r_{1} r_{2}\right) +\\ \left( r_{1}+1\right) ^{2}\left( -2r_{1}+r_{2}+r_{1}^{2}r_{2}+r_{1} ^{2}-2r_{1}r_{2}\right) ^{2} \end{array} \right) z_{1}^{2}\\ +2\left( r_{2}+t_{2}z_{2}\right) \left( r_{1}-1\right) \left( \begin{array}{c} t_{1}\left( r_{2}+r_{1}^{2}r_{2}\left( r_{1}-3\right) +r_{1}^{2}\left( r_{1}-2\right) +r_{1}\left( r_{2}-3\right) \right) \\ +\left( r_{1}+1\right) ^{2}\left( r_{2}+r_{1}\left( r_{1}r_{2}-2\right) +r_{1}\left( r_{1}-2r_{2}\right) \right) \end{array} \right) z_{1}\\ +\left( 2r_{1}-r_{1}^{2}+1\right) \left( r_{1}-1\right) ^{2}\left( r_{2}+t_{2}z_{2}\right) ^{2} \end{array} \right) }{r_{1}^{2}\left( r_{1}-1\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}} \end{aligned}$$

At \(z_{1}=0\) this is negative and increasing, respectively because \(\left( 2r_{1}-r_{1}^{2}+1\right) <0\) and

$$\begin{aligned} \left( \begin{array}{c} t_{1}\left( r_{2}+r_{1}^{2}r_{2}\left( r_{1}-3\right) +r_{1}^{2}\left( r_{1}-2\right) +r_{1}\left( r_{2}-3\right) \right) \\ +\left( r_{1}+1\right) ^{2}\left( r_{2}+r_{1}\left( r_{1}r_{2}-2\right) +r_{1}\left( r_{1}-2r_{2}\right) \right) \end{array} \right) >0. \end{aligned}$$

At the maximum \(z_{1}\) (\(z_{1}=\frac{r_{1}\left( r_{2}+t_{2}z_{2}\right) }{\left( r_{1}+r_{1}r_{2}+1\right) t_{1}}\)) this is positive: \(\left( r_{2}+t_{2}z_{2}\right) ^{2}\frac{\left( -t_{1}+r_{1}^{3}r_{2}-2r_{1} ^{2}r_{2}-2r_{1}^{2}+r_{1}^{3}+r_{1}r_{2}\right) ^{2}}{r_{1}^{2}t_{1} ^{2}\left( r_{1}-1\right) ^{2}\left( r_{2}+1\right) ^{2}\left( r_{1}+r_{1}r_{2}+1\right) ^{2}}>0.\)

Then, from remark 4,

$$\begin{aligned} z_{1}&>z_{1TN}\\&:=\left( r_{2}+t_{2}z_{2}\right) \left( r_{1}-1\right) \\&\quad \times \frac{ \begin{array}{c} -\left( \begin{array}{c} t_{1}\left( r_{2}+r_{1}^{2}r_{2}\left( r_{1}-3\right) +r_{1}^{2}\left( r_{1}-2\right) +r_{1}\left( r_{2}-3\right) \right) +\\ \left( r_{1}+1\right) ^{2}\left( r_{2}+r_{1}\left( r_{1}r_{2}-2\right) +r_{1}\left( r_{1}-2r_{2}\right) \right) \end{array} \right) \\ +\left( r_{1}+1\right) \left( -t_{1}-2r_{1}^{2}r_{2}+r_{1}^{3}r_{2} -2r_{1}^{2}+r_{1}^{3}+r_{1}r_{2}\right) \sqrt{2} \end{array} }{\left( \begin{array}{c} t_{1}^{2}\left( 2r_{1}^{2}r_{2}+6r_{1}^{3}r_{2}-2r_{1}^{4}r_{2}+5r_{1} ^{2}+2r_{1}^{3}+r_{2}^{2}-r_{1}^{4}-4r_{1}^{2}r_{2}^{2}+4r_{1}^{3}r_{2} ^{2}-r_{1}^{4}r_{2}^{2}-6r_{1}r_{2}-2\right) \\ -2t_{1}\left( r_{1}+1\right) ^{2}\left( r_{2}\left( r_{1}-1\right) ^{2}+r_{1}\left( r_{1}-2\right) \right) \left( r_{1}-r_{2}+r_{1} r_{2}\right) \\ +\left( r_{1}+1\right) ^{2}\left( -2r_{1}+r_{2}+r_{1}^{2}r_{2}+r_{1} ^{2}-2r_{1}r_{2}\right) ^{2} \end{array} \right) } \end{aligned}$$

In order to discuss in which tier a traditional merger is more likely to be profitable, we now compare the conditions for merger profitability when tiers are symmetric, that is, when \(z_{i}=z\), \(t_{i}=t\) and \(r_{i}=r\).

In this case, the condition that for a traditional upstream merger to be profitable, \(z_{2}>z_{2TN},\) is equivalent to

$$\begin{aligned} z&>\frac{r+tz}{r}\frac{\left( r-1\right) \left( r+1\right) \left( r(\sqrt{2}-1)-1\right) -t\left( r^{2}-2r-1\right) }{t^{2}\left( 2r-r^{2}+1\right) -2t\left( r-1\right) \left( r+1\right) ^{2}+\left( r-1\right) ^{2}\left( r+1\right) ^{2}}\Leftrightarrow \\ z&>\frac{-r+r^{2}\left( \sqrt{2}-1\right) }{\left( r-1\right) \left( r+r^{2}+\left( \sqrt{2}-1\right) t\left( r-\sqrt{2}-1\right) \right) } \end{aligned}$$
(4)

and the condition that for a traditional downstream merger to be profitable, \(z_{1}>z_{1TN}\), is equivalent to

$$\begin{aligned} z&>\left( r+tz\right) \left( r-1\right) \\&\quad \times \frac{-\left( t\left( r\left( -r-2r^{2}+r^{3}-2\right) \right) +\left( r+1\right) ^{2}r\left( -r+r^{2}-1\right) \right) +\left( r+1\right) \left( -t+r^{2}\left( -r+r^{2}-1\right) \right) \sqrt{2} }{\left( t^{2}\left( 4r^{3}+r^{4}+2r^{5}-r^{6}-2\right) -2t\left( r+1\right) ^{2}\left( r\left( -r+r^{2}-1\right) \right) r^{2}+\left( r+1\right) ^{2}\left( r\left( -r+r^{2}-1\right) \right) ^{2}\right) }\Leftrightarrow \\ z&>\frac{\left( \sqrt{2}-1\right) r\left( r-1\right) \left( r-\sqrt{2}-1\right) }{\left( -r-2r^{2}+r^{4}+\left( \sqrt{2}-1\right) t\left( -r^{2}\left( \sqrt{2}+2\right) +r\left( \sqrt{2}+2\right) +r^{3}-2\sqrt{2}-3\right) \right) } \end{aligned}$$

As

$$\begin{aligned}&\frac{\left( \sqrt{2}-1\right) r\left( r-1\right) \left( r-\sqrt{2}-1\right) }{\left( -r-2r^{2}+r^{4}+\left( \sqrt{2}-1\right) t\left( -r^{2}\left( \sqrt{2}+2\right) +r\left( \sqrt{2}+2\right) +r^{3}-2\sqrt{2}-3\right) \right) }\\&\quad -\frac{-r+r^{2}\left( \sqrt{2}-1\right) }{\left( r-1\right) \left( r+r^{2}+\left( \sqrt{2}-1\right) t\left( r-\sqrt{2}-1\right) \right) }\\&\quad =-\frac{r\left( r-\sqrt{2}-1\right) \left( r+1\right) }{\left( r+r^{2}+\left( \sqrt{2}-1\right) \left( r-\sqrt{2}-1\right) t\right) \left( r-1\right) }\\&\qquad \times \frac{\left( \sqrt{2}-1\right) r\left( r-2\right) +\left( 3-2\sqrt{2}\right) \left( r-\sqrt{2}-2\right) t}{-r-2r^{2}+r^{4}+\left( \sqrt{2}-1\right) \left( r^{3}-r^{2}\left( \sqrt{2}+2\right) +r\left( \sqrt{2}+2\right) -2\sqrt{2}-3\right) t}<0 \end{aligned}$$

because all terms in brackets are positive, we conclude that there are values of z for which only the downstream merger is profitable and not the opposite. \(\square \)

Proposition 2B

(a) Downstream firms propose a traditional merger for \(z_{1}>z_{1TM}\) and a multidivisional one otherwise. (b) Upstream firms propose a traditional merger for \(z_{2}>z_{2TM}\) and a multidivisional one otherwise. \(\square \)

Proof of Proposition 2B

(a) Downstream firms will propose the most profitable merger. It is the traditional merger if and only if

$$\begin{aligned} \pi _{1t}(r_{1}-1,r_{2},t_{1}+1,t_{2})>\pi _{1m}(r_{1}+1,r_{2},t_{1},t_{2}) \end{aligned}$$

which is equivalent to:

$$\begin{aligned} \frac{\left( \begin{array}{c} z_{1}^{2}\left( \begin{array}{c} r_{1}^{8}\left( r_{2}+1\right) ^{2}-2r_{1}^{7}\left( r_{2}+1\right) ^{2}\left( t_{1}-1\right) -r_{1}^{6}\left( r_{2}+1\right) \left( 5r_{2}+8t_{1}+2r_{2}t_{1}^{2}+2t_{1}^{2}+6r_{2}t_{1}+7\right) \\ +2r_{1}^{5}\left( r_{2}+1\right) \left( -4r_{2}-t_{1}+2r_{2}t_{1} ^{2}+2r_{2}t_{1}-8\right) +\\ r_{1}^{4}\left( 18r_{2}+28t_{1}+32r_{2}t_{1}^{2}+20r_{2}^{2}t_{1}+11r_{2} ^{2}+22t_{1}^{2}+8r_{2}^{2}t_{1}^{2}+46r_{2}t_{1}+8\right) \\ -2r_{1}^{3}\left( -18r_{2}-20t_{1}-2r_{2}t_{1}^{2}+r_{2}^{2}t_{1}-5r_{2} ^{2}-12t_{1}^{2}+4r_{2}^{2}t_{1}^{2}-13r_{2}t_{1}-16\right) \\ -r_{1}^{2}\left( 8r_{2}-16t_{1}+28r_{2}t_{1}^{2}+22r_{2}^{2}t_{1}+11r_{2} ^{2}+8t_{1}^{2}+10r_{2}^{2}t_{1}^{2}+32r_{2}t_{1}-16\right) \\ +4r_{1}r_{2}\left( t_{1}+1\right) \left( -r_{2}-2t_{1}+r_{2}t_{1}-4\right) +4r_{2}^{2}\left( t_{1}+1\right) ^{2} \end{array} \right) \\ +2z_{1}\left( r_{1}-1\right) \left( r_{1}+1\right) \left( r_{2} +t_{2}z_{2}\right) \left( \begin{array}{c} r_{1}^{5}\left( r_{2}+1\right) +r_{1}^{4}\left( 2t_{1}+3\right) \left( r_{2}+1\right) \\ -r_{1}^{3}\left( r_{2}+2t_{1}+4r_{2}t_{1}+2\right) -r_{1}^{2}\left( 7r_{2}+14t_{1}+6r_{2}t_{1}+12\right) \\ +4r_{1}\left( -t_{1}+r_{2}t_{1}-2\right) +4r_{2}\left( t_{1}+1\right) \end{array} \right) \\ +2\left( 2r_{1}-r_{1}^{2}+2\right) \left( r_{1}-1\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+t_{2}z_{2}\right) ^{2} \end{array} \right) }{r_{1}^{2}\left( r_{1}-1\right) ^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{2}+1\right) ^{2}} \end{aligned}$$

At \(z_{1}=0\) this is negative and increasing, respectively because \(\left( 2r_{1}-r_{1}^{2}+2\right) <0\) and

$$\begin{aligned}&r_{2}\left( r_{1}-1\right) \left( r_{1}+1\right) \left( 2t_{1}\left( -2r_{1}+r_{1}^{2}-2\right) +3r_{1}^{2}+r_{1}^{3}-4\right) \\&\quad +r_{1}\left( r_{1}+2\right) \left( 2t_{1}\left( r_{1}^{2}-3r_{1}-1\right) +r_{1}\left( r_{1}-4\right) +r_{1}^{3}-4\right) >0. \end{aligned}$$

At \(z_{1}=\frac{r_{1}\left( r_{2}+t_{2}z_{2}\right) }{\left( r_{1} +r_{1}r_{2}+1\right) t_{1}}\) this is equal to

$$\begin{aligned} \frac{\left( r_{2}+t_{2}z_{2}\right) ^{2}\times \left( \begin{array}{c} r_{2}^{2}r_{1}^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}-1\right) ^{4}+\\ 2r_{2}r_{1}\left( r_{1}-1\right) ^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}^{3}-2r_{1}^{2}-t_{1}\right) +\\ 2t_{1}^{2}\left( -r_{1}^{4}+6r_{1}^{3}+5r_{1}^{2}+6r_{1}+2\right) +\\ r_{1}^{3}\left( r_{1}-2\right) \left( r_{1}^{2}-2r_{1}-2\right) \left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}+\\ 2r_{1}^{2}\left( r_{1}-2\right) \left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}-t_{1}\right) \end{array} \right) }{r_{1}^{2}\left( r_{1}-1\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}+2\right) ^{2}\left( r_{2}+1\right) ^{2}t_{1}^{2}\left( r_{1}+r_{1}r_{2}+1\right) ^{2}} \end{aligned}$$
(5)

If \(-r_{1}^{4}+6r_{1}^{3}+5r_{1}^{2}+6r_{1}+2>0\) then this is clearly positive. If \(-r_{1}^{4}+6r_{1}^{3}+5r_{1}^{2}+6r_{1}+2<0\) then the second term in the numerator of (5) is larger than (replacing \(r_{2}\) by 4 in the first line and \(t_{1}^{2}\) by \(r_{1}^{2}\) in the third line)

$$\begin{aligned} \left( \begin{array}{c} 4^{2}r_{1}^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}-1\right) ^{4}+\\ 2r_{2}r_{1}\left( r_{1}-1\right) ^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}^{3}-2r_{1}^{2}-t_{1}\right) +\\ 2r_{1}^{2}\left( -r_{1}^{4}+6r_{1}^{3}+5r_{1}^{2}+6r_{1}+2\right) +\\ r_{1}^{3}\left( r_{1}-2\right) \left( r_{1}^{2}-2r_{1}-2\right) \left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}+\\ 2r_{1}^{2}\left( r_{1}-2\right) \left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}-t_{1}\right) \end{array} \right) \end{aligned}$$

which can be written as

$$\begin{aligned}&r_{1}^{2}\left( -36r_{1}-110r_{1}^{2}+232r_{1}^{3}+180r_{1}^{4}-152r_{1} ^{5}-89r_{1}^{6}+34r_{1}^{7}+17r_{1}^{8}+68\right) \\&\quad +2r_{1}^{2}\left( r_{1}-2\right) \left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}-t_{1}\right) \\&\quad +2r_{2}r_{1}\left( r_{1}-1\right) ^{2}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}^{3}-2r_{1}^{2}-t_{1}\right) >0. \end{aligned}$$

Then, from Remark 4,

$$\begin{aligned} z_{1}&>z_{1TM}\\&:=\left( r_{1}-1\right) \left( r_{1}+1\right) \left( r_{2}+t_{2}z_{2}\right) \\&\quad \times \frac{\left( \begin{array}{c} \left( \begin{array}{c} -\left( r_{1}+1\right) \left( r_{1}+2\right) ^{2}\left( r_{1}\left( r_{2}+1\right) \left( r_{1}-2\right) +r_{2}\right) \\ -2t_{1}\left( -2r_{1}+2r_{2}-3r_{1}^{2}r_{2}-2r_{1}^{3}r_{2}+r_{1}^{4} r_{2}-7r_{1}^{2}-r_{1}^{3}+r_{1}^{4}+2r_{1}r_{2}\right) \end{array} \right) \\ +\left( r_{1}+2\right) r_{1}\left( \left( r_{1}+1\right) \left( -2r_{1}+r_{2}+r_{1}^{2}r_{2}+r_{1}^{2}-2r_{1}r_{2}\right) -2t_{1}\right) \sqrt{3} \end{array} \right) }{\left( \begin{array}{c} -2t_{1}^{2}\left( \begin{array}{c} r_{1}^{6}\left( r_{2}+1\right) ^{2}-2r_{1}^{5}r_{2}\left( r_{2}+1\right) -r_{1}^{4}\left( 16r_{2}+4r_{2}^{2}+11\right) \\ +2r_{1}^{3}\left( 2r_{2}+3\right) \left( r_{2}-2\right) +r_{1}^{2}\left( 14r_{2}+5r_{2}^{2}+4\right) \\ -2r_{1}r_{2}\left( r_{2}-2\right) -2r_{2}^{2} \end{array} \right) \\ -2t_{1}\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( r_{1}\left( r_{2}+1\right) \left( r_{1}-2\right) +r_{2}\right) \left( r_{1}-r_{2}+r_{1}r_{2}\right) \\ +\left( r_{1}+2\right) ^{2}\left( r_{1}+1\right) ^{2}\left( -2r_{1} +r_{2}+r_{1}^{2}r_{2}+r_{1}^{2}-2r_{1}r_{2}\right) ^{2} \end{array} \right) } \end{aligned}$$

(b) Upstream firms will propose the most profitable merger. It is the traditional merger if and only if

$$\begin{aligned} \pi _{2t}(r_{1},r_{2}-1,t_{1},t_{2}+1)>\pi _{2m}(r_{1},r_{2}+1,t_{1},t_{2}) \end{aligned}$$

which is equivalent to:

$$\begin{aligned} \frac{\left( \begin{array}{c} r_{1}^{2}\left( -2t_{2}^{2}\left( r_{2}^{2}-2r_{2}-2\right) -2t_{2}\left( r_{2}-1\right) \left( r_{2}+2\right) ^{2}+\left( r_{2}+2\right) ^{2}\left( r_{2}-1\right) ^{2}\right) z_{2}^{2}+\\ 2r_{1}\left( -4t_{2}+2r_{2}^{2}t_{2}+3r_{2}^{2}+r_{2}^{3}-4r_{2} t_{2}-4\right) \left( r_{1}+t_{1}z_{1}\right) z_{2}+\\ 2\left( 2r_{2}-r_{2}^{2}+2\right) \left( r_{1}+t_{1}z_{1}\right) ^{2} \end{array} \right) }{r_{1}r_{2}^{2}\left( r_{2}+2\right) ^{2}\left( r_{1}+1\right) } \end{aligned}$$

At \(z_{2}=0\) this is negative and increasing, respectively because \(-2\left( -2r_{2}+r_{2}^{2}-2\right) <0\) and

$$\begin{aligned} 2r_{1}\left( 2t_{2}\left( r_{2}^{2}-2r_{2}-2\right) +3r_{2}^{2}+r_{2} ^{3}-4\right) >0. \end{aligned}$$

At \(z_{2}=\frac{1+\frac{t_{1}z_{1}}{r_{1}}}{t_{2}}\) it is positive: \(\left( r_{2}-1\right) ^{2}\frac{\left( r_{1}+t_{1}z_{1}\right) ^{2}}{r_{1} r_{2}^{2}\left( r_{1}+1\right) t_{2}^{2}}>0.\)

Then, from remark 4,

$$\begin{aligned} z_{2}>z_{2TM}:=\frac{r_{1}+t_{1}z_{1}}{r_{1}}\frac{\left( \begin{array}{c} -2t_{2}\left( -2r_{2}+r_{2}^{2}-2\right) -\left( r_{2}-1\right) \left( r_{2}+2\right) ^{2}\\ +r_{2}\left( r_{2}+2\right) \left( r_{2}-1\right) \sqrt{3} \end{array} \right) }{\left( \begin{array}{c} -2t_{2}^{2}\left( r_{2}^{2}-2r_{2}-2\right) -2t_{2}\left( r_{2}-1\right) \left( r_{2}+2\right) ^{2}\\ +\left( r_{2}+2\right) ^{2}\left( r_{2}-1\right) ^{2} \end{array} \right) } \end{aligned}$$

\(\square \)

Proposition 3B

In tier i, final consumers’ and merging firms’ interests diverge for \(z_{iTM}<z_{i}<z_{itm}\). \(\square \)

Proposition 4B

In tier i, the most profitable merger lowers consumer welfare for \(z_{iTM}<z_{i}<z_{itn}\). \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Catalão-Lopes, M., Brito, D. Post-merger internal organization in multitier decentralized supply chains. J Econ 132, 251–289 (2021). https://doi.org/10.1007/s00712-020-00723-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00712-020-00723-7

Keywords

JEL Classification

Navigation