Does environmental regulation create merger incentives? Working Paper

This paper studies merger incentives for polluting Cournot Örms under a competitive tradable emission permits market. We Önd that when Örms are symmetric and marginal costs are constant, an horizontal merger is welfare enhanicing if e¢ciency gains are high enough for the merger to take place. The presence of a competitive (or monopolistic) outside market that also trades in the permits market makes proÖtable a merger that would not happen otherwise. When Örms are vertically related in an input-output chain, an horizontal merger in one of the markets increases proÖts in the other market due to the permits price decrease. Finally we consider an oligopoly-fringe model in which Örms di§er in their marginal production costs. A merger between the dominant oligopolistic Örms decreases the permits price and is always proÖtable. Such setting is relevant to assess the observed mergers between power generators in several market for permits, like the Regional Greenhause Gas Initiative (RGGI), allowing us to derive some policy recommendations.


Introduction
Many papers have analyzed the welfare consequences of horizontal mergers in the presence of imperfect competition (see Motta, 2004 for a complete mapping of the literature). How mergers incentives and welfare properties are modiÖed in the presence of an environmental externality, regulated by tradable emission permits, is a less explored issue to which we contribute. The paper closest to our analysis is Hennessy and Roosen (1999), who show that in a perfectly competitive industry, incentives coming from the permits market may motivate a merger that wouldnít take place otherwise. The last section of Hennessy and Roosenís paper brieáy mentions how the previous result could be modiÖed under Cournot competition in the output market, suggesting that it is ambiguous. We study in depth which are the mechanisms behind that ambiguity and its welfare implications. With this purpose, we consider symmetric Örms that compete ‡ la Cournot in a polluting industry subject to an environmental regulation based on a perfectly competitive market for permits. ! UniversitÈ Paris Dauphine, PSL Research University, LEDA CGEMP, 75016 Paris, France. E-mail: anna.creti@dauphine.fr y EPEE (EA 2177), UniversitÈ d¥Evry-Val-dëEssonne and DÈpartement d ¥Economie, Ecole Polytechnique, France. E-mail: eugenia.sanin@univ-evry.fr we account for the impact of a horizontal merger. We develop Örst the case of symmetric Örms with constant marginal costs before and after the merger, and second the case where the merger generates e¢ciency gains. In Section 3 we analyze the proÖtability of a horizontal merger in the case where several sectors are subject to the market for permits. In particular, in Subsection 3.1 we consider the case where two di §erent sectors belong to the same permit market and, in Subsection 3.2, the case in which the two sectors subjects to environmental regulation are vertically related (either downstream or upstream). In Section 4 we relax the assumption of symmetric Örms before the merger by analyzing a case with an oligopolistic dominant group of Örms and a competitive fringe. Finally, in Section 5 we analyze the extent to which our results could be applied to understand merger incentives in the power sector under the RGGI market and eventually in the EU-ETS market. In Section 6 we conclude and derive some policy implications.

The importance of e¢ciency gains
Consider a market for a product where the inverse demand function for a quantity Q ! 0 is given by P (Q). Assume that P 0 (Q) < 0; P 00 (:) " 0 or P 00 (:) ! 0 not too large for all Q. N symmetric Örms compete ‡ la Cournot in this market. The production technology exhibits constant returns to scale and c denotes the marginal cost of production.
Production generates pollution: for each unit of the good ( > 0 units of a pollutant are emitted. Pollution is socially costly and Örms are required to buy as many pollution permits as their individual emissions. Assume also that the total number of pollution permits E > 0 supplied by the authority, is Öxed. Our main assumptions relate to the functioning of tradable pollution permits. Assumption 1. The market for pollution permits is perfectly competitive.
Assumption 1 implies that Örms have no market power on the permits market. Therefore, the equilibrium price for permits + is such that the total demand for permits, (Q, equals the total supply of permits, E.
(Q = E: The previous assumption can be justiÖed by the fact that the environmental regulation covers many states in the case of the RGGI and also many sectors in the case of the EU-ETS. This assumption is quite usual in environmental literature (see for example the seminal paper of Sartzetakis, 1997 andSartzetakis, 2004).
We focus on the interesting case in which the environmental regulation e §ectively constrains Örmsí decisions, i.e., + > 0. A su¢cient condition for this to be the case is that the total supply of permits is smaller than the total pollution that would be generated by a hypothetical monopolist: This assumption ensures that the price of permits + is positive after a merger of any size. We consider a game in which Örms choose their production levels of the Önal good and the market for permits clears simultaneously.
A horizontal merger. Let us derive the conditions under which a merger is proÖtable, from a private and a social point of view.
In the absence of merger, Örm i solves the following problem: where Q = P N i=1 q i . Indeed, for each unit of the Önal good produced, Örm i emits %q i units of pollution that must be covered with permits. Consider an interior solution of the previous optimization problem in which Örm iís quantity is characterized by the corresponding Örstorder condition: Equation (3) is su¢cient provided that the demand is concave or not too convex, which holds under our assumptions. An equilibrium of the game is a production proÖle fq i g i=1;:::;N and a permits price & such that, for each Örm i = 1; :::; N , Equation (3) is satisÖed and the permits market clears, i.e.: The crucial feature implied by the market clearing condition (1) on the permits market is that the total quantity of good produced only depends on the exogenous supply of permits E and the pollution factor %. Using Equations (3) and (1), the proÖt of each Örm at the symmetric equilibrium is given by (see derivation in Appendix): Suppose now that M Örms, N > M > 2, decide to merge. Such a merger reduces the number of Örms from N to N ! M + 1. The proÖt of each Örm at the symmetric equilibrium is thus given by: Denote by -& $ & post ! & pre the di §erence between the permits price before and after the merger. Using the Örst-order conditions in Equation (3), we obtain: Simple computations show that the variation in the revenue of the regulator associated to the sale of permits exactly corresponds to the change in the industry¥s proÖt (-. $ . post ! . pre ): Our results from a welfare perspective can be summarized as follows.
Proposition 1. In a symmetric Cournot case with constant marginal costs, any merger is welfare neutral and does not a §ect consumers surplus. The permits price decreases, which implies that the regulator¥s revenue from permits sales is redistributed to Örms.
Proof. See Appendix.
Since the quantity produced is constant, a merger is welfare neutral. This would be the case even in the extreme case of full monopolization i.e. N = M .
For the merger to take place, M has to satisfy the following condition. Condition 1 In a symmetric Cournot case with constant marginal costs, a merger of M Örms is proÖtable if and only if: This condition holds for any demand function and is similar to the one derived by Salant et al. (1983) who looks at the proÖtability of a merger in a constant marginal cost and symmetric Cournot oligopoly with a linear demand function. A discussion on whether the previous condition is more or less restrictive than in the case where there is no market for permits is provided in the Appendix. In particular, we Önd that if demand is linear, the permits market relaxes the proÖtability condition of mergers.
E¢ciency gains. Assume that the merger creates e¢ciency gains, that is, after the merger takes place the participants have a lower marginal production cost: c m = c!%; with % " 0 and where m stands for a Örm participating in the merger. In the pre-merger game, individual proÖts are the ones in Equation (4). From proÖt maximization of the merged Örm and of the symmetric non-merged Örms we get, respectively, the following equilibrium values: where i stands for the outside Örms. Substituting the above equilibrium quantities inside the proÖt function we get proÖts for the merged Örms and the individual N ! M outsiders: , The previous condition implies that e¢ciency gains must be large enough to give Örms incentives to merge.
It is worth noting that the merger is never proÖtable for the outsiders, i.e. it can be easily shown that , post i < , pre i : From the FOCs we can calculate the change in permits price after the merger that after simpliÖcation can be expressed as compared to the case without e¢ciency gains as: The second term may be stronger than the Örst one and the price of permits may increase after the merger. This is the case for a su¢ciently high e¢ciency gain 3 i.e.
This increase is due to the fact that, after the merger, the merged Örms proÖting from an e¢ciency gain have incentives to increase their production. Since total quantity produced is Öxed by the supply of permits, this simply results in a higher demand for permits that, with a Öxed supply, produces an increase in permits price. The overall industryís proÖt increases after a merger for any $: Consequently, since consumer surplus is una §ected due to the fact that total quantity at the industry level remains constant, the merger is welfare enhancing. 4 The previous results can be summarized in the following proposition.
Proposition 2. In a Cournot symmetric case, a merger is welfare enhancing if it generates production e¢ciency gains for the merged Örms. The e¢ciency gain must be high enough for the merger to take place. The merger is never proÖtable to outsiders.
Proof. See Appendix.
The result is particularly relevant to analyze mergers in the power sector where part of the included Örms are using low cost technologies that could also be less polluting.
So far we have assumed that only the Örms that are present in the Önal market participate in the permits market. We relax this assumption in the following Section 3 and show that a merger in one market a §ects the allocation of production in all the markets that cover their emissions with permits.
3 When markets are linked by the permits market

The presence of an outside output market
Consider that on top of the product market considered so far, there is a demand for permits coming from another industry denoted by ëoí for ëoutsideí, i.e.: A merger between M of the N Örms is now proÖtable if and only if: We observe that this condition di §ers from the the Condition 1 by a multiplicative positive factor Q(#pre ) 2 > 0 that can be higher or lower than 1, making the proÖtability condition more or less binding than in the case without the presence of an outside market.
Indeed, any merger now a §ects the total quantity produced by the industry in which the merger takes place: the permits price variation due to the merger produces a change in the demand for permits coming from the outside industry, which impacts the proÖtability of the proposed merger. As an example, suppose that the merger leads the merged Örm to contract its output. The mergerís competitors respond by expanding their production due to the strategic substitutability of quantities but, overall, the total quantity produced by the industry is likely to decrease, leading to a decrease in permits price. This decrease in permits price reinforces the proÖtability of the merger, although it also tends to give producers more incentives to expand their output, in particular to the outside industry.
Under our assumptions on P (:), we have !P 0 (Q)Q increasing in Q. Therefore, to understand whether a merger is more or less proÖtable with respect to the case with no outside industry, we must analyze (i) its impact on permits price and (ii) how the equilibrium production Q(:) varies with the permits price. Regarding the second point, since Q(%) and Q o (%) are linked through the market clearing condition, analyzing Q(:) boils down to analyzing how the quantity of the outside industry varies with the permits price. Let us distinguish three distinct cases: " Perfectly competitive outside industry. In that case, the price on the outside market is equal to the marginal cost, which includes the permits price. Therefore: " Monopolistic outside industry. In that case, a standard argument yields: " Imperfectly competitive outside industry. The analysis of this case is less immediate. Indeed, following Myles (1987), whether the (total or individual in the symmetric case) quantity decreases with an increase of the marginal cost of all the Örms depends on the slope of the elasticity of the inverse demand function. Similarly, what happens in a non-competitive outside market could impact the permits price (Sanin and Zanaj, 2011).
For simplicity let us consider the linear case Q o (%) = C ! D%, i.e. Q 0 o (%) < 0, and linear demand in the output market P (Q) = a ! bQ: In such conÖguration the industry production varies with the permits price according to the following expression: which implies Q 0 (%) = %oD % > 0. We can then derive the following proposition.
is high enough) a merger that would not take place without the presence of the outside market now becomes proÖtable.
Proof. See following reasoning.
FOCs before and after the merger are in this case: After substituting Equation (15) inside Equations (16) we can compute (% for this case: A su¢cient condition for Equation (17) to be negative after the merger is: 5 Looking at Equation (18) and comparing it with Equation (15), we see that it is a condition on the relative size of the sector compared to the outside one. In particular, the outside sector needs to be small enough for the total demand for permits to decrease after the merger provoking a decrease in permits price. In particular, this condition can be interpreted as a condition on the polluting intensity of the outside sector: ! o < E C : Regarding the condition of merger proÖtability in the presence of an outside market, there is an additional aspect to consider. In the case of linear demand !P 0 (Q(% post )) = !P 0 (Q(% pre )) = b and from the FOCs in Equations (16) ; respectively. Then, the proÖtability condition in Equation (14) becomes: Notice that, if the permits price increases after the merger, the LHS is lower than 1 and the proÖtability condition in the presence of an outside market becomes more restrictive than condition in Equation (8). Instead, if the permits price decreases the LHS can be higher or lower than 1 and the condition can be more or less restrictive than in the case without the presence of the outside market. It is particularly interesting to study wether it is possible that a merger that is not proÖtable (i.e. not verifying Equation (8)) becomes proÖtable solely due to the existence of an outside market (i.e. it veriÖes Equation (19)) that absorbs the demand change in the permits market after the merger. This is the case when the number of merged Örms M satisÖes the following condition: To study the Örst inequality let us call A " (20) is satisÖed for high values of A: 6 The inequality on the RHS in (20) is satisÖed for the following values: 7 The intersection of these conditions is non-empty. This means that the existence of an outside market makes the merger proÖtable. The mechanism is simple: the decrease in the permits price due to the merger is strong enough for A to be big. Moreover, condition (21) ensures that the number of merged Örms is small enough.
The previous results proves to be particularly relevant in the case of the EU-ETS market. The fact that several industries are linked through the permits market regulation changes the proÖtability of a merger inside each industry. In this sense, the fact that for example the cement industry is also under the EU-ETS regulation could change the proÖtability of a merger in the power sector.

A horizontal merger in an input-output structure
We continue exploring the analysis of horizontal mergers but in the context of vertically related industries, which is particularly relevant in the power sector. An upstream industry produces an input which is then used (on a one-to-one basis) by a downstream industry to produce a Önal good. The production of both the input and the output generates pollution, although possibly with di §erent intensities. This is the case in the EU-ETS where polluting input sectors (e.g. electricity or gas) sell their goods to polluting downstream sectors (e.g. steel, cement, paper), both subject to an environmental regulation. We build on the so-called double Cournot model of Salinger (1988), or more recently Gabszewicz et al., (2013). The downstream industry, denoted with ëdí, has N d Örms competing ‡ la Cournot on a Önal market in which the inverse demand function is given by P d (Q d ) and where the total quantity of Önal good produced is Q d . Downstream Örms are symmetric, the marginal cost being c d and the pollution factor $ d . The upstream industry, denoted by ëuí, has N u Örms competing ‡ la Cournot on the intermediate good market. Upstream Örms are symmetric, the marginal cost being c u and the pollution factor $ u .
Both upstream and downstream Örms are assumed to be price-takers on the permits market. The timing of the game is as follows: upstream Örms decide their production levels; downstream Örms choose their quantities; the permits market clears. 6 Note that the inequation is a third-degree polynomium in M that has 3 roots. For high values of A the polynomium is always positive in the range 1 < M < N. 7 Notice that this polynomium has 3 positive roots: one is equal to 1, another one higher than N and only one equal to N ! 1 2 p 4N + 1 + 1 2 in the relevant set of parameters.
Downstream equilibrium. Downstream Örms i = 1; :::; N d maximize their proÖts as follows: where p u is the price paid for the upstream good. Since one unit of input is required for each unit of output produced, it must be the case that Q d = Q u . The previous FOC then represents the inverse demand curve in the input market.
Upstream equilibrium. Each input producer j = 1; :::; N u maximizes its proÖt: Permits market equilibrium. Since we assume Q d = Q u " Q; and since agents are price-takers in the permits market we can write: Equilibrium proÖts. The proÖt of downstream and upstream producers are given by: After substituting Equation (24) in Equations (25) we get:

Downstream horizontal merger
The constraint in Equation (24) ensures that the equilibrium output both in the downstream and the upstream market respects the environmental constraint. Then, the proÖtability of a downstream merger is simply the condition in Equation (8).
Increasing the concentration of the downstream sector has an indirect e §ect on the upstream market. By using Equation (26) and deÖning ( = ( d + ( u ; the di §erence between post-merger and pre-merger proÖts in the upstream industry is shown to be positive. This is the case because the quantity produced in both market stays the same. Therefore, Equation (24) holds and the price of the upstream good is also unchanged. As in Equation (6), however, the merger in the downstream sector decreases the permits price. This e §ect beneÖts both upstream and downstream sectors. The proÖts of the downstream sector increases since fewer Örms produce the quantity Q, thus leading to a decrease permits demand and price. Therefore, Equation (7) still holds. The di §erence in the proÖts post and pre-merger for the vertically related industry can be written as follows: (27) meaning that an industry proÖts from the decrease in permits prices produced by the merger in another vertically related industry.

Upstream horizontal merger
Since the produced quantity remains unchanged, there is a direct e §ect of the merger on permits price that beneÖts directly the upstream Örms and indirectly the downstream Örms. All the previous results remain the same simply by exchanging the subindex d by u and vice versa.
Proposition 4. When a merger takes place in one industry, the decrease in permits price proÖts the vertical related industry.
Proof. In Appendix.

Pollution reallocation when Örms are heterogeneous
Consider now that Örms di §er in their marginal costs of production. With this purpose we slightly change our setting and consider the following ëoligopoly -fringe modelí that could describe the power market in countries like France or in some American states.
Oligopoly-fringe model. There are two types of Örms producing in the Önal market: # D dominant Örms, denoted by i = 1; :::; D, whose production technology exhibits constant returns-to-scale with marginal cost c d .
# A fringe composed of F competitive Örms i = 1; :::; F , whose production technology exhibits decreasing returns-to-scale with total cost C f (q The fringe is assumed to be a price-taker on the product market. Both types of Örms simultaneously decide their quantities and the permits market clears. The fringe e¢ciency, summarized by c f , a §ects the relative size of the proposed merger: for instance, when c f is low (large), the fringe produces a large (small) quantity and the merger between dominant Örms involves a small (large) fraction of the total output. Absent any merger, an (interior) equilibrium is a proÖle of quantities ffq i;d g i=1;:::;D ; fq i;f g i=1;:::;F g and a permits price " such that each dominant Örm i = 1; :::; D, maximizes its proÖt according to the following FOCs: (28) where the second line shows the competitive fringeís FOC. The cap on emissions determines that: This condition ensures that the market share of dominant Örms is positive both before and after the merger. It can also be interpreted as a restriction on the fringeís size: the part of output E * produced F has to be su¢ciently low in order to verify the constraint on output due to the existence of a pollution cap.
Let us also assume for simplicity that inverse output demand is linear P = a # bQ.
After summing the FOCs of the fringe over F , solving the system of FOCs and the permits equilibrium condition we get the proÖts of the dominant Örms before merging, i.e.: A similar reasoning must be done to calculate proÖts after the merger: Since in equilibrium all q i;d are the same, and considering the permits price equilibrium, the proÖtability condition 1 " Dpos > M1 " Dpre can be rewritten as: and the subindex t can take the value pre or post merger. As usual, the proÖtability condition is related to the change in permits prices due to the merger, which in this case is always negative when Equation (30) is satisÖed: 8 which in turns determines a positive proÖt di §erential for dominant Örms: (35) From the previous reasoning we derive the following proposition.
Proposition 5. In the oligopoly-fringe model, a merger between the oligopolistic Örms is always proÖtable and accompanied by a permits price decrease.

Proof. See Complementary Material available upon request.
With respect to the benchmark case of Cournot competitors (Section 2), here the merger is always proÖtable, no matter the size of the merged entity. Dominants restrict output knowing that the fringe absorbs the production di §erential. As a consequence, given that the total output is constant in the pre and post cases, the demand for permits always decreases and so does the corresponding price.
The non-strategic behavior of the fringe also has some straightforward implications on the equilibrium permits price and quantities produced. In terms of comparative statics, we have: An increase of the number of dominants who merge leads to a decrease in permits price, an output contraction by the dominants perfectly compensated by an increase of production by the fringe. As a consequence, dominants always have an incentive to become a unique entity after the merger. We can then establish the following corollary: Proof. See Appendix. 8 Notice that permits prices are calculated by solving the system of FOCs, substituting the market for permits clearing condition and rearranging in the case with and without a merger, respectivelly.

Case Studies
Big mergers have taken place after the creation of tradable emission permits both in the US and in Europe. In this section we argue that the use of tradable emission permits may be one of the mergers driving reasons. To this end, we give several examples that illustrate the validity of our results.

An application to the RGGI
The Regional Greenhouse Gas Initiative (RGGI) was put in place in 2009 (with a Örst auction on the 25th of September 2008) as the Örst mandatory cap-and-trade program to limit CO2 emissions in the United States. 9 Nine states currently participate in the RGGI: Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New York, Rhode Island, and Vermont. Several states and Canadian provinces act as observers: Pennsylvania, QuÈbec, New Brunswick, and Ontario. New Jersey formerly participated, but it withdrew in 2011.
The market works as follows. One hundred sixty three electric power generators are required to obtain a number of permits (or allowances) equal to the number of tons of CO2 emitted. 10 11 The prices in the RGGI market have continuously increased since the end of 2012. Such increase is mostly due to a tighter cap that came into e §ect in 2013. Due to this update on the market cap we are unlikely to observe a decrease in permits price after each merger takes place but only a deceleration in the price increase to which all these mergers may have contributed.
Given the results underlined in this paper we think it is not a coincidence that the most important mergers in the RGGI area have happened after this great tightening of the market cap in 2014, which are all summarized in Table 1.

INSERT FIGURE 2
Among the mergers that have taken place between Örms with predominant presence in the states covered by the RGGI 12 , the biggest one has been the merger of Exelon Corporation and Pepco Holdings Incorporated. Such merger was announced on April 2014 for more than $12 billion. At the merger time, Exelon served 6.7 million costumers through its subsidiaries in Illinois, Pennsylvania and Maryland and intervened in all stages from generation to delivery of electricity. On the other side, Pepco Holdings served 1.9 million customers in the District of Columbia and Maryland. Pepco Holdings also owns Delmarva Power in Delaware and Maryland. This merger was presented as a horizontal one, even if Exelon holds all activities at a time, being an example of the type of mergers mentioned in Section 3.2. The U.S. Energy Information Agency explains that, according to Exelon, cost reductions available through increased scale and geographic proximity were the primary reasons for the merger (see Today in Energy 13 report in October 2014).
Another merger among Örms in the RGGI, announced around the same dates, is the NRG Yield Inc. acquisition of Alta Wind Energy Center of Terra-Gen Power LLC and of ArcLight Capital Partners LLC, for an estimated $2.5 billion. NRG Yield Inc. is a Delaware unit of NRG Energy Inc., Terre-Gen Alta Wind is a New York-based provider of renewable energy and ArcLight is a Boston-based equity Örm. This merger 14 could be an example of the case modeled in Section 2.2 since the inclusion of a renewable source will most probably generate e¢ciency gains in the context of environmental regulation.
The prospect of a decrease in permits price after the merger could indeed, as shown by our model, be a cost reduction argument that increases merger incentives. Indeed, the average price of CO2 allowances during the Öst quarter of 2014 (i.e. the quarter before the two mentioned mergers took place) went from around $3.50 at the beginning of the quarter to almost $4.50 at the end of the quarter averaging $3.87. As compared to the previous quarter it increased of 26 percent and 37 percent as compared to the Örst quarter of 2013. 15 Many other mergers have taken place among Örms belonging the RGGI and many U.S. investor-owned utilities have indeed consolidated in recent years. 16 In the table below we show all the mergers that took place during 2014, the year when prices started to steadily increase due to the tightening of the cap.

Could our argument also apply to the EU-ETS?
The European Union Emissions Trading Scheme (EU-ETS) started in 2005 as the worldís Örst and biggest international trading system for CO2 emissions. As in the case of the RGGI, Örms under regulation are required to obtain a number of permits equal to the number of tons of CO2 emitted. During the Örst years permits where allocated to Örms for free using a method called grandfathering (a percentage reduction of your installationís emissions in 1990) but nowadays most of permits are auctioned. 17 The EU-ETS now covers emissions from more than 11,000 heavy energy-using installations (including power stations and industrial plants) and airlines operating between participating countries (all 28 EU countries plus Iceland, Liechtenstein and Norway). In total it covers around 45% of the EUís greenhouse gas emissions and accounts for over 3/4 of international carbon trading. 18 Power and CHP plants represented around half of the allowances allocated in Phase 2 (2008-2012) of the EU ETS. Di §erently from the RGGI where the electricity market and the permits market are in almost perfect correspondence, the link between the two markets is most probably weaker. In Section 3.1 we show the case in which other sectors also participate in the permits market. 13 See EIA website. 14 For a comparison of the two mergers mentioned see the PwC Quarterly Reports. 15 See RGGI Market Report Q1, 2014. 16 For more information on this see The Public Power Report. 17 For a description of the auctioning mechanism and results see the Common Auction Platform report. 18 For a full description of the EU-ETS regulatory framework and coverage see ETS Handbook.
We think that a merger in the EU-ETS case could be described by the conditions detailed in that section.
Several Örms concentrations have taken place since the EU-ETS has been established. For instance, The Economist of the 26th of February 2009 explained how the frenetic merger trend present since 2006 has made the European market evolve into an oligopoly 19 dominated by a few-cross border giants, such as EDF from France, E.ON and RWE from Germany and Enel from Italy. 20 Just to cite the most important mergers we are referring to, let us mention Enel¥s acquisition of Endesa that started at the beginning of 2006 (when Enel bought 67% of Endesa bringing in a Spanish partner called Acciona to appease local feeling) and ended at the 20th of February 2009 when Enel bought Acciona¥s stake for 11.1 billion euros. 21 Few days before the Spanish government had accepted the merger of Gas Natural with Union Fenosa. 22 That same week Vattenfall, the largest Nordic utility, bought the Dutch energy Örm called Nuon for 8.5 billion euros. 23 Back in January of that same year the German RWE also bought the Dutch Essent for 9.3 billion euros. 24 The reasons behind the previous mergers are, most certainly, multidimensional. In any case such mergers have coincided with a systematic decrease in CO2 prices in accordance with our results. Indeed, the Örst merger cited coincides with the most changing year of the EU-ETS (in that period, the price of emissions permits tripled in the Örst six months of Phase I and then collapsed by half in a one-week period in 2006). Since then, a systematic price decrease has characterized the CO2 market. Our model shows that the number of important mergers in the power sector may have contributed to the price slowdown, due to other economic factors.

Conclusion
This paper extends Hennessy and Roosen (1999)¥s result on the merger incentives created by the existence of a tradable emission permits market to the case in which the polluting Örms are non-competitive, as it is the case in most energy markets and in particular in the power sector. We Önd that, if Örms in the sector are symmetric and competing ‡ la Cournot, an horizontal merger is welfare improving but there is a critical size for it to occur and, in the presence of a perfectly competitive (or monopolistic) outside market that also participates in the same tradable emission permits market, under some conditions a merger that would not be proÖtable without it now takes place. The Örst result suggests that, as prices go up in the RGGI, merger incentives coming from the CO 2 market will continue to increase. Similarly, the second result says that in the case of the EU-ETS, where more than one industry is under the same CO 2 market, incentives for mergers due to the sole existence of this market are even greater. We also study the case of vertically related sectors, which is particularly relevant in power markets, Önding that an horizontal merger in one of two markets that are vertically linked increases proÖts in both markets due to the provoked decrease in permits price. we Önd that a merger between oligopolistic Örms in a market where there is a competitive fringe is always proÖtable as opposed to the symmetric Cournot case were is a critical size needed.
Our results regarding the modiÖcation of the merger proÖtability condition in the presence of a tradable emission permits market provide new insights for understanding many mergers that have taken place both under the RGGI and the EU-ETS. The policy implications from these results are quite straightforward: there is a trade-o § between promoting a high CO 2 price to reduce the environmental externality and promoting competition in energy markets, in particular in the power sector where Örms are also vertically related with other Örms covered by the same tradable emission permits market. The regulator should then counterbalance pro et contra before Öxing (or tightening) the cap on emissions and should work closely with competition authorities to avoid putting too much pressure on consumers. These considerations add to the considerations in De Feo et al. (2013) regarding the decrease in competition that could arise in the power sector when a permits market is in place.
Since Q = E % by hypothesis we can rewrite and since we are in a symmetric equilibrium Substituting this FOC inside the proÖt function we get that simpliÖes to equation (4).

Proof of Proposition 1
The quantity produced does not change after the merger due to the equilibrium condition on the permits market: Q = E " : From the FOC of symmetric Örms we Önd that: and consequently the di §erence is which can be simpliÖed as the expression in the main text.
On the other hand the sum of proÖts before and after the merger are and consequently the di §erence is The previous equations shows that what is earned by the industry after the merger perfectly compensates the government loss due to the decrease in the permits price.

Discussion on Condition 1
Under the assumption of linear demand a ! bQ and without a permits market, the Salant (1983)ís condition for merger proÖtability, i.e. , post > M, pre ; simply depends on the number of Örms in the industry after the merger takes place, that is: The proÖtability condition in our model is given by Condition 1 that is: Notice that Condition 1 holds for any demand function. In our model total production is Öxed by the availability of permits E and therefore mergers proÖtability only depends on the number of Örms after the merger.
Condition 1 can be more or less restrictive than the case without permits market. In fact, solving the system of the previous two equations we Önd the following quadratic function of N : that has two real roots, one of them that is always positive i.e.: For a number of merging Örms M > r 1 , the sole existence of a permits market relaxes the proÖtability condition.

Proof of Condition 2
When there are e¢ciency gains the cost after the merger takes place decrease as follows: c m = c ! ); with ) " 0: Firms that do not merge maximize while merged Örm get together and maximize max qm!0 Cournot system of FOCs is then 25 : Since non merged Örms are symmetric the system can be written as: Di §erently from what happened when the merger does not create a cost asymmetry when there are e¢ciency gains the full monopolization case cannot be treated simply as a particular case of the general resolution formula since full monopolization makes one type of Örm (the ones less e¢cient) to disappear (note that N !M in the equation appears in the denominator). 25 In the case of full monopolization the second equation becomes P 0 (Q) P N "M 1 qi = 0 simply because there is only the now merged Örm that stays in the market.
In the case of non full monopolization we then solve the system by equalizing to Önd : ProÖts for non-merged Örms are then ProÖt for the merged Örm is now The merger is proÖtable if: ) ! m;post > M) pre Note that in the case of full monopolization the previous condition boils down to the condition in equation 8: giving 1 > 1 N which is always true. Let us go back to the case where there is no full monopolization, i.e. from now on N > M > 2 : Since !P 0 ( E % ) > 0 we can simply write the proÖtability condition as: that is a perfect square of the form a 2 ! b 2 = (a + b) (a ! b) : The condition for positivity is then Which condition is more binding depends on the sign of x y Since N is always higher than 1, the expression is always positive and the binding condition is

Proof of Proposition 2
Part 1: Impact of merger in permits price From the FOCs 2) ! P 0 ( E " ) E N" + P ( E " ) ! c ) any ! is: Consumer surplus is una §ected due to the fact that total quantity at the industry level remains constant. The industry gain more than compensates the loss of the government (that is equal to !!" " E). The merger is then welfare enhancing: 26