Strengthening buyer power on the EU gas market: Import caps and supply diversification Svetlana Ikonnikova∗ and Gijsbert T.J. Zwart† March 2009
Abstract We study how various policies on natural gas trade between EU importers and the large non-European suppliers, e.g. Russia, may improve EU member states’ aggregate and individual buyer power on the market. We describe the distribution of market power in this bilaterally oligopolistic market using the Shapley value. We show that commitments to reduce EU aggregate imports from any particular supplier do not increase buyer power. In contrast, importers can benefit and strengthen their bargaining position if the EU restricts bilateral gas trade between individual member states and individual suppliers. We also find that investments in import capacity for new suppliers improves the bargaining position of all EU importers, even those that cannot directly trade with the new supplier. This can provide a rationale for EU support for new investment projects, like the Nabucco pipeline. We explore the relevance of these observations in a stylised model of the EU gas market. Keywords: Natural gas markets, buyer power, Shapley value. JEL classification: L12, L41, C71, Q48.
∗ Center
for Energy Economics, Bureau of Economic Geology, University of Texas at Austin, University Station, Box X Austin, Texas 78713-8924, e-mail:
[email protected]. † TILEC, Tilburg University, and Netherlands Bureau for Economic Policy Analysis, CPB, PO Box 80510, 2508 GM The Hague, The Netherlands, e-mail:
[email protected].
1
1
Introduction
In its recent Strategic Energy Review (European Commission, 2008c) the Commission stresses the importance of diversification of natural gas supplies to Europe. In the face of rising demand and declining indigenous gas production, many observers voice concerns over growing dependence of European natural gas consumers on a limited number of supplying countries. Geographical diversification of supplies is proposed as an answer to the ensuing concentration of technical and political risks (see e.g. CIEP, 2004, Stern, 2006). The limited number of large suppliers creates concerns for seller market power, too. Though less emphasized in official EU publications, diversification may also be viewed as a means to reduce such seller power (see e.g. Helm, 2007). Natural gas dependence is in fact interdependence. While European buyers are concerned about supply security, Russian, Norwegian and Algerian suppliers are equally worried about security of demand: large investments in pipelines tie both producers and consumers together. This implies that not only do producers possess seller power, but also buyers may exercise buyer power. Diversification may also be seen in this light: by expanding the set of sellers, buyers hope to create a better position in negotiations. We focus on the question how national governments or the EU may affect this buyer power. Current decentralized negotiations limit the amount of coordination that the EU may be able to achieve. Clearly, it is not the EU itself, but individual firms within the member countries that negotiate trades with the large state-owned foreign producers. And even though member states often have substantial influence on decisions of their national gas incumbents (in particular concerning decisions on new infrastructure capacity), interests of individual member states often diverge. The controversy surrounding the NordStream pipeline, a planned cross-Baltic pipeline directly connecting Russia with Germany, is a case in point. Whereas this pipeline reduces German exposure to risks of transit problems, countries east of Germany see their roles, and bargaining power towards Gazprom, diminished. It seems unlikely that the EU will be able to replace current bilateral negotiations between individual buyers and producers and form an effective single purchasing block towards all producer countries to increase buyer power. Are there other less intrusive mechanisms that can be decided on jointly by member states and are jointly beneficial? We focus on two types of policies. One potential policy measure to ensure diversification of supplies is to cap quantities purchased from any individual country. Such a measure has in fact been implemented in Spain, a country which relies on Algerian gas for a large part of its supplies. Spain’s Hydrocarbons Sector Law, dating back to 1998, stipulates that any gas marketer should limit its supplies sourced from any single country to 60% of its total portfolio. This forces market players to turn to other (perhaps more expensive) producing countries and helps to diversify Spanish supplies. Whereas in 2000 Algerian gas supplies to Spain almost reached the 60% mark, currently only a third of total supplies originates from that country (CNE, 2008). 2
A second policy to encourage diversification is to financially and politically stimulate building infrastructure to new source countries. A current example is the EU’s backing of various large pipeline projects, in particular the Nabucco project that aims at bringing Caspian gas to Europe, bypassing Russia (Stern, 2006). We analyse the potential impact on buyer power of such policy measures using a bargaining theory approach. We describe the relations of producers and consumers using cooperative game theory. The multilateral negotiations of gas supply contracts by producers and buyers are depicted by a game in characteristic function form. The game assumes that bargaining among the market players is efficient: supply contracts are non-linear and complete with respect to prices and quantities. This assumption of non-linear contracts in the international wholesale gas market is not unrealistic in practice, given the prevalence of take-or-pay contracts for fixed quantities1 . These closely resemble simple (non-linear) quantity forcing contracts (as also pointed out in Smeers, 2008). To solve the game we apply a general solution concept – the Shapley value. Here we follow a bargaining approach adopted in a series of works, including Gul (1989), Stole and Zwiebel (1996), Inderst and Wey (2003). Shapley’s solution concept has both cooperative and non-cooperative foundations and can be defined axiomatically. This allows one to avoid assumptions on a specific bargaining protocol of the game, of which little is known. Besides, assuming that all agreements can be renegotiated before any plans are executed, Stole and Zwiebel (1996) show that only the Shapley payoffs are renegotiation-proof. All this makes the Shapley value an appealing measure of power in supply chains. Caps on contract quantities between individual suppliers and buyers can compromise efficiency. We find that caps on capacities contracted from individual suppliers cannot raise buyer surplus if all buyers form a single purchasing block, effectively behaving as a monopsonist. The case is different when buyers are fragmented: under oligopsony, constraining the contract space available to each buyer-seller pair can shift bargaining power to rival buyers negotiating with the same seller. If this external effect dominates the efficiency effect, buyers may agree, to their mutual advantage, to impose such constraints. We show that constraints on bilateral trades can make all buyers better off. A simple sufficient condition for existence of such surplus-improving constraints is that the number of buyers exceeds the number of sellers. Conversely, constraints on aggregate buyer dependence on single sellers never improves joint buyer surplus. Diversification by investing in connections with smaller or new suppliers, such as the Caspian states, may well be individually efficient for buyers, as it can allow for more efficient production as well as improve the buyer’s individual bargaining power vis-` a-vis its old suppliers. However, also here an external effect is at play. Since a seller’s outside options become less valuable if the rival buyer can source gas from elsewhere, other buyers also enjoy increased bargaining 1 Typically natural gas contracts take the form of long-term contracts specifying a given volume and a given payment (which often is linked to some other index price), and only very little variation in supply around this quantity is allowed. See Asche et al. (2002).
3
power. This may call for buyer cooperation to increase capacities beyond their individual optimum. Concerns about increasing European dependence on external gas producers, as well as disputes about the best path for liberalization prompted a substantial body of research on gas market issues. A large part of the literature, starting with the contributions of Mathiesen et al. (1987), and Golombek et al. (1995, 1998) describes producers as Cournot players, treating European buyers as price-takers. Haurie et al. (1987) extended the analysis to allow for stochastic demand. An advantage of the structure of these models is that they can be straightforwardly extended to incorporate more detailed description of transport and storage markets. Gabriel and Smeers (2006) provide a survey of the earlier literature, some recent examples focussing on the European market include Boots et al. (2004), Egging and Gabriel (2006), Holz et al. (2008) and Lise et al. (2008). Inclusion of resource rents considerations is addressed in Zwart and Mulder (2006) and Zwart (2009). A major disadvantage of this approach though is that it puts all the bargaining power in the hands of the producers, an assumption which as pointed out seems invalid in practice (see also Smeers, 2008, for a similar critique). Our approach in the current paper deviates from the assumption of unilateral market power and instead develops a more balanced description of buyer and seller power. Here we follow the line of research adopted in Hubert and Ikonnikova (2003, 2004) and Hubert and Suleymanova (2006), in modelling the gas market bargaining game using a cooperative game theory approach. We first discuss the bargaining framework, in terms of the Shapley value, that we use to model bilateral negotiations between buyers (importing countries) and suppliers (exporting countries). We then turn to the issue how coordination between buyers may improve their bargaining position. Although closer cooperation in purchasing might bring even more benefits to buyers, we focus on the more realistic policy of restricting importers’ dependence on individual exporting countries. We provide some general results on when such policies might help, and when they are ineffective. Then we turn to the issue of expanding opportunity for trade with new suppliers. We show that also buyers who do not directly trade with the new supplier may benefit from such expansion. Finally we explore the theoretical considerations in a stylised numerical model that roughly captures supply and demand structure on the (future) EU gas market.
2
Basic bargaining framework
To describe the trade between European gas buyers and non-European sellers, we adopt a cooperative game theory framework. We describe bargaining over supply contracts between buyers and sellers as a game in characteristic function form. To solve the game and determine the trade surplus earned by each market player we use the Shapley value concept. Let the set of all the buyers be B with its elements denoted as bi , the set of sellers be S with sj ∈ S. We denote the set of all the players N = B ∪ S. Buyers may be thought of as the large gas
4
incumbents in Europe, such as GdF, Eni, RWE and E.On-Ruhrgas. By sellers we mean the major natural gas exporters to Europe, such as Russian Gazprom, Algerian Sonatrach and Norwegian StatoilHydro.2 Value, or trade surplus, is created if at least one buyer contracts, or forms a coalition M , with at least one seller. In general, a coalition consists of multiple players who are bound by contracts. To each possible coalition M ⊂ N we assign a trade surplus v (M ) equal to the sum of the producers’ S ∩ M and buyers’ B ∩ M surpluses. The function v is also called a value or characteristic function. Buyer surplus might be thought of as a monopoly rent of a gas supply incumbent, or as the consumer welfare generated by the incumbent’s supply of gas to his country end-users, or something in between, depending on the effects of state influence on the (often state-owned) buying firm. Seller surplus may more readily be identified with profits, although here, too, other objectives (e.g. political) may be valued by the controlling government. We solve the game (v, N ) applying the Shapley value φ, which defines each player’s expected payoff taking into account that every buyer and seller is aware of the bargaining between all other players and eventually the grand coalition forms. The latter means that all buyers negotiate with all sellers. The Shapley value is calculated as a weighted sum of a player’s contributions to all possible coalitions: X φj (v) = P(M ) [v(M ∪ j) − v(M )] (1) M :j ∈M /
where P(M ) = |M |! (|N | − |M | − 1)!/|N |! could be interpreted as the probability of coalition M preceding player j in a random order bargaining setting. A player’s contribution can also be expressed in terms of a difference operator: 4j v(M ) = [v(M ∪ j) − v(M )]. The Shapley value if computed in relative terms, as a share of the grand coalition profit, can be interpreted as a player’s bargaining power.3 The motivation for using this value function approach is its robustness to the specification of the negotiation process, of which in principle little is known. The chosen solution concept - the Shapley value - is the unique solution satisfying a plausible set of axioms. The advantage of the Shapley solution is also that it can be derived as an outcome of a non-cooperative bargaining game in which producers and buyers bilaterally negotiate over supplies, as demonstrated by Inderst and Wey (2003) (see also Gul, 1989, Stole and Zwiebel, 1996). Our analysis crucially assumes that buyers do not compete for end-users4 . Each buyer contracts gas for consumers in its own, usually national, market. Though liberalization challenges this view it seems fair to say that competition between European gas incumbents is limited. As the EU Commission notes in 2 As a rule these supplying companies are state-controlled; therefore if there is no confusion we will often refer to them by country names, e.g. say Russia instead of Gazprom. 3 See more on the subject in Hubert and Ikonnikova (2003). Thus, we are able to analyze the changes in the players’ bargaining or market power in terms of changes in the value function v. 4 Otherwise this would introduce contract externalities on supply contracts, and prevent us from applying the Shapley value as a solution concept.
5
its 2007 benchmarking report, “With very few exceptions, electricity and gas markets in the EU remain national in economic scope with limited competition”, whereas the HHI in production and imports for most national markets within the EU is larger than 5000 (European Commission, 2008a). This lack of competition may be due to lack of transport infrastructure between markets, to reluctance to switch supplier by end-users, or to anticompetitive agreements between the major players, as alleged by the European Commission.5 In the remainder of this paper we will be interested in European policy measures that may enhance the bargaining power of the buyers vis-`a-vis the sellers. We imagine a two stage procedure. In the first stage a regulatory authority, e.g. the European Commission, imposes some measures (trade limitations, subsidies, etc.) affecting the buyers’ negotiation strategy. In the second stage, buyers bargain with the suppliers subject to these pre-agreed policies. We want to explore whether there are particular measures or restrictions on the allowed contracts that may benefit all buyers.
3
Policies towards increasing buyer power
The rationale for buyer coordination lies in spill-over effects among buyers: one buyer’s freedom of negotiation with a supplier may affect another buyer’s bargaining power with the same supplier. Below we will derive the intuition for this based on the Shapley value. Loosely speaking, how much value a given buyer brings to a coalition may affect the contribution of another buyer joining the coalition. As a result coordination among rival buyers that reduces the scope for negative externalities may lead to an increase in the aggregate buyers’ bargaining power. Segal (2003) studies various types of coordination among substitutable players – competing buyers in our case – and the effects of such coordination on complementary players (sellers). In particular, he considers contracts which give some players veto power over other players’ transactions, so-called ”exclusive contracts”. In terms of equation (1), if buyer b1 has an exclusive contract with buyer b2 , a seller s cannot negotiate with b2 if b1 is not a member of coalition M ∪ s ∪ b2 . Segal (2003) demonstrates that exclusive contracts between buyers reduce sellers’ shares of total surplus, while total surplus stays the same. As a consequence, the aggregate buyer surplus increases.6 Although this type of buyer coordination may therefore be a fruitful approach to raising buyer power, intrusive coordinating measures such as these seem hardly compatible with the desire to achieve competition in the internal market for end-users, and moreover, politically unrealistic7 . Therefore, in this section we will turn our attention to two less drastic measures that seem closer 5 In case COMP/39.401, the Commission issued a “Statement of Objections to E.ON AG (E.ON), E.ON Ruhrgas AG and Gaz de France (GDF), concerning their alleged agreement or concerted practice to keep out of each other’s home market for the supply of gas”. 6 Segal (2003) also considers full collusion between buyers. Unlike exclusive contracts, full collusion (e.g. through merger) need not always increase buyer power. 7 Note however that in case of Caspian gas, the EU itself brings up the suggestion to form purchasing blocks, European Commission (2008b)
6
to current views on policies stimulating diversification of supplies and that may result in buyer surplus gains. We first discuss the imposition of limits to the dependence on individual suppliers. Next we explore the rationale for subsidizing investment in capacity to new suppliers.
3.1
Caps on import dependence
Does a commitment to a cap on total imports from a single source improve a buyer’s bargaining position? We first show that such a measure may only reduce buyer surplus if all buyers already purchase cooperatively, forming a single monopsonist8 . Under the constraint, (i) the profit of the grand coalition may only decrease, and (ii) the contributions of the unified buyer to coalitions with the restricted seller can only fall. We state this result as Proposition 1 If all buyers cooperate , forming a monopsony, any constraints on contracted capacities with a particular seller may only decrease aggregate buyer surplus. Proof. The payoff to the monopsonist buyer b is given by the Shapley value X φb = P(M ) [v(M ∪ b) − v(M )] (2) M :b∈M /
where the sum is over all coalitions M ⊂ S consisting of sellers. Since any coalition without the single buyer generates a zero surplus, the sum only involves positive terms. Constraining the feasible allocations can never increase each surplus v(M ∪ b). Note that individual suppliers may benefit from constraints on their rivals’ quantities. More interesting is the situation when buyers constitute an oligopsony, with each buyer negotiating bilaterally with each seller, and when there are collectively agreed constraints on the bilateral trade volumes. Such a situation is more relevant to the present situation, where some argue for limiting imports in particularly from Russia, pointing to the increasing dependence on this single supplier. If such constraints on trade are introduced, firstly any individual buyer will suffer from constraints on his own trade, as in the monopsony case. However, now the restrictions on one buyer may make the other buyers better-off: restricted sellers’ bargaining power drops because they are more vulnerable in negotiations with non-constrained buyers. If each buyer’s loss from the former effect is smaller than the positive externalities from other buyers’ constraints then a policy of bilateral trade restrictions is beneficial. 8 Suppose indeed that gas importers would establish a joint purchasing authority, which would purchase all natural gas for Western Europe. We look at how this central purchaser’s pay-off changes if for example supply from Russia were restricted to 30% of the total nonEuropean gas supply.
7
It depends on the type of constraints that are imposed whether such buyer power increasing effects can occur. We first demonstrate that jointly agreeing on maximum aggregate imports from any given supplier negatively affects aggregate buyer surplus. This extends proposition 1 to the case with multiple buyers. Proposition 2 Restricting the aggregate imports from an individual supplier can never increase aggregate buyer surplus. Proof. Consider any subset M which includes the restricted seller, say s1 . Obviously, a constraint on s1 ’s export can only decrease v(M ). We now note that this decrease in value is always at least as large for v(M ∪ bi ) as it is for v(M ): the shadow price of the constraint can only increase as demand is increased. As a result, φbi can only decrease as a result of such a constraint. Again, such a policy will have the effect of reducing the affected supplier’s surplus. The only beneficiaries of this measure, if any, are the rival suppliers who gain in bargaining power as a result of the deterioration of the buyers’ outside options. Calls for collective restrictions to reduce European dependence on any of Europe’s suppliers, and to force diversification, are therefore ill-advised from this bargaining position point of view. On the other hand, individual restrictions on bilateral negotiations may be more useful, as the next proposition states. Proposition 3 If the number of buyers B exceeds the number of sellers S, there generically exist constraints on bilaterally traded quantities that (i) improve aggregate buyer surplus (ii) make all buyers individually better off. Proof. We will first show that any constraint on traded quantities between supplier s1 and buyer b1 that only affects v({s1 , b1 }) improves total buyer surplus P bi ∈B φbi provided that B > S. Note that in φb1 the term v({s1 , b1 }) appears with coefficient P (1), whileP in φbi , i > 1, it appears with coefficient −P(2). The coefficient of v({s1 , b1 }) in bi ∈B φbi therefore equals 1 1 2 (B − 1) P(1) − (B − 1)P(2) = − (3) S + B S + B − 1 (S + B − 1) (S + B − 2) (S + B − 3)! = (S − B) (4) (S + B)! Since any constraints can only affect v({s1 , b1 }) negatively, and since all other terms are by assumption not affected by the constraint, the result follows. To see that such constraints only affecting v({s1 , b1 }) generically do exist, denote ∗ the traded quantity optimising v({s1 , b1 }) by q11 . Note first that under standard conditions on supply and demand functions, in no other coalition can the optimal ∗ traded quantity be higher than q11 : this would yield marginal buyer surplus P ∗ ∗ ∗ lower than P (q11 ) and marginal supplier costs C 0 higher than C 0 (q11 ) = P (q11 ), 8
violating the first order conditions. For similar reasons, provided that in other coalitions including s1 and b1 , either or both have non-zero trade with other members of those coalitions, the volume of trade between s1 and b1 , q11 , in ∗ those coalitions will be strictly lower than q11 . To prove the second part, note that for each buyer bi we can pick a supplier si that has a non-zero trade with this buyer in the coalition {si , bi }. We can also find an interval of constraints on this pair’s trade that only affects v ({si , bi }), by a (negative) amount δvbi , and keeps all other surpluses unchanged. For the set of these B constraints, we have X Abi bj δvbj δφbi = bj
where A is a matrix consisting of entries P(1) the diagonal, and −P(2) for all off-diagonal entries. It is easily checked that A is invertible, with inverse −1 −1 given by (P(1) + P(2)) (P(1) − (B − 1) P(2)) times a matrix with entries P(1) − (B − 2) P(2) on the diagonal, and P(2) of the diagonal. By equation (3) the prefactor is negative, and each of A−1 ’s rows sums to P(1) + P(2) times this prefactor. Hence A−1 maps the vector (δφ, ..., δφ) with δφ > 0 to a strictly negative vector δv. By choosing δφ sufficiently small, δv is realised through a set of constraints that only affects the two-player coalitions. While proposition 3 shows that B > S is a sufficient condition, it is by no means necessary. Consider for instance a situation where B > S and add to the set of suppliers a number of small, capacity constraint (fringe) suppliers so that the total number of suppliers S 0 exceeds the number of buyers B. Evidently, as the importance of these suppliers diminishes, we should retrieve the result we had without the fringe sellers, the case B > S. Indeed, suppose we have one very small seller sn , where very small means that v(M ∪ sn ) ≈ v(M ). Now, similar to the proof of proposition 3, focus again on constraints affecting only v({s1 , b1 }) and v({s1 , sn , b1 }) = v({s1 , b1 }) + ε (in the limit of very small sn these are identical). In φb1 the term v({s1 , b1 }) now enters with coefficient P(1)+P(2), while in rival buyers’ values it has coefficient −P(2)−P(3). Adding up v({s1 , b1 }) terms in all buyers’ valuations now leads to X
φbi
=
(P(1) + P(2) − (B − 1)(P(2) + P(3))) v({s1 , b1 }) + other terms
=
(S − 1 − B)
bi ∈B
(S − 1 + B − 3)! (S − 1 + B)!
i.e. in the limit that ε → 0 we duly obtain the result we would have without the fringe seller. For small but finite ε, we may now have that a sufficiently small constraint decreases buyer value, but that a slightly higher constraint increases it. We showed that when B > S there exist constraints which increase total buyer value. We used constraints which only bite off-equilibrium (i.e. not in 9
the grand coalition) to prove this. These constraints therefore do not reduce aggregate buyer and seller surplus. This does not mean that optimal constraints will not affect total surplus either. We next explore a specific two buyer, two seller example to illustrate that optimal constraints may have real effects on the outcome. Example 4 For the situation of two buyers and two sellers, our proposition provides no guidance: the coefficient of the v ({si , bj }) term in aggregate buyer surplus is zero. In fact joint buyer surplus in this case is given by φb,joint =
1 1 1 1 1 v + v−s1 + v−s2 − v−b1 − v−b2 2 6 6 6 6
where v is shorthand for grand coalition surplus, and v−x denotes the surplus of the coalition consisting of all players except x. We explore the special case in which both buyers and sellers are symmetric, and supply and demand functions are linear: assume seller costs Csi (q) and buyer surplus Bbi (q) are given by 1 2 aq 2 1 = q − q2 2
Csi (q)
=
Bbi (q)
We first compute optimal traded quantities q for each pair, for each of the relevant coalitions. Then we assess the impact of introducing symmetric constraints. First note that optimal bilaterally traded quantities are not unique: if suppliers s1 and s2 make a swap leaving total supplies to each buyer as well as total production from each supplier equal, total welfare v is invariant. Here we focus on a symmetric equilibrium where each supplier sells the same amount q to each buyer. This minimizes the maximum bilateral trade. We can then easily compute optimum bilateral quantities q for each coalition, as well as total surplus q=
1 2+2a
q−si = q−bi =
v = 2q
1 1+2a 1 2+a
v−si = q−si v−bi = q−bi
1 So, in the grand coalition each supplier sells q = 2+2a to each buyer, so total sales for each supplier are 2q, and total consumption for each buyer equals 2q. Joint surplus of all four players equals 2q. The other rows can be interpreted analogously. We observe that q < q−si and q < q−bi , while q−bi > q−si if a > 1. If production costs rise steeply, a > 1, then bilateral trades are largest in the coalitions with one buyer and two sellers. Since picking constraints that only bind in the v−bi coalitions raise joint buyer surplus, we see that for a > 1 such constraints exist.
10
We now proceed to compute total surplus for each coalition in the presence of a constraint on trade q ∗ . Provided the q ∗ constraint binds, we find
v v−si v−bi
= = =
4q ∗ − 4(1 + a)q ∗2 2q ∗ − (1 + 2a)q ∗2 2q ∗ − (2 + a)q ∗2
Optimization of φb,joint then straightforwardly yields that optimal symmetric constraints exist only if a > 1. Optimal constraint q ∗ is given by 3 5 + 7a In this case, the optimum constraints also impact the grand coalition. For a ≤ 1, constraints always decrease aggregate buyer welfare. q∗ =
3.2
Coordination of investment
If a buyer expands the number of its suppliers it increases its gross surplus through two channels. First, it increases total surplus by enlarging the production set, allowing for cheaper production of the gas. And second, it decreases rival sellers’ bargaining power by expanding its own outside options. The new supplier will also win surplus, and both will invest in creating an opportunity to trade if their joint gross surplus gains exceed the investment costs. Introduction of a new trading partner by one buyer b1 also creates spill-overs on the rival buyers bi>1 who cannot trade with the new supplier. Again there is an efficiency effect and a strategic effect: the entry of the new supplier reduces the production burden on the old suppliers, reducing their marginal costs of supplying the other buyers bi>1 . And furthermore, the old suppliers’ outside option to shift sales to b1 becomes less valuable, so that buyers bi>1 have an improved bargaining position. We formalise these ideas in the next proposition. Proposition 5 Suppose several buyers bi share a number of suppliers sj . A buyer b1 establishing a connection with a new supplier sS+1 that will not supply any of the other buyers bi>1 increases both the surplus of b1 and the surplus of the other buyers bi>1 . Already existing sellers’ surplus is reduced. Proof. Introduction of a new seller sS+1 into the game changes the game from an N -player game to an N + 1-player game. To compare Shapley values for both games, observe that for coalitions of M players the probabilities P(M ) for ˜ the N -player game and P(M ) for the new N + 1-player game satisfy ˜ ˜ P(M ) = P(M ) + P(M + 1).
(5)
The pay-off to the buyer b1 enjoying the new connection to sS+1 is now given by.
11
φ˜b1
=
X
X
˜ P(M )∆b1 v(M ) +
M ⊂N
= φb1 +
˜ P(M + 1)∆b1 v(M ∪ sN +1 )
M ⊂N
X
˜ P(M + 1)∆sS+1 b1 v (M )
(6)
M ⊂N
by (5). Here ∆sS+1 b1 v (M ) = v(M ∪ sS+1 ∪ b1 ) − v(M ∪ sN +1 ) − v(M ∪ b1 ) + v(M ) is the second difference operator. The additional term is positive since v(M ∪ sN +1 ∪ b1 ) ≥ v(M ∪ b1 ) (with strict inequality at least for some M ), and v(M ∪ sN +1 ) = v(M ) if M does not include b1 . For other buyers bi>1 , we similarly compare φbi and φ˜bi and by again using (5) we find that the only differences in expressions occur for those subsets that include b1 : φ˜bi − φbi
X
=
˜ P(M )∆bi v(M )
M ⊂N :b1 ∈M
+
X
˜ P(M + 1)∆bi v(M ∪ sS+1 )
M ⊂N :b1 ∈M
−
X
P(M )∆bi v(M )
M ⊂N :b1 ∈M
=
X
˜ P(M + 1)∆sS+1 bi v (M )
M ⊂N :b1 ∈M
and again this is positive since ∆sS+1 bi v (M ) > 0 for M including b1 : there are larger gains from attracting an additional buyer when part of b1 ’s demand is fulfilled by sS+1 . Similarly, existing sellers’ surplus is reduced: X ˜ φ˜s − φs = P(M + 1)∆s s v (M ) . i
i
S+1 i
M ⊂N :b1 ∈M
where ∆sS+1 bi v (M ) ≥ 0. As a result, we find that the total buyers’ increase in value is not fully captured by the direct beneficiaries of the new link, b1 and sS+1 . From the point of view of total buyer welfare, the new pipeline will have too small capacity, or is not built in spite of achieving positive net buyer gains. This provides some rationale for other buyers to support building the new link, even if they are no direct beneficiaries. Note that from a social point of view, this may well lead to too large pipeline capacity: part of the gains come at the expense of existing suppliers.
4
Numerical Example
In addition to the theoretical results we find it interesting to have a look at empirical results. In this section we provide numerical estimates of the losses 12
and benefits of coordination and policies considered in the theoretical part of the paper. Using available market data and experts’ projections for the future EU15 demand for natural gas we calculate the Shapley value under various assumptions with respect to buyers’ cooperation and import constraints. We focus on the Western European countries, namely the EU 15. To keep the picture clear and avoid computational complexities, we divide the EU 15 countries into three major gas importing regions, we call them France, Germany, and Italy. The first one includes besides France itself: Belgium, Luxembourg, Spain, Portugal, and unless we say otherwise, the UK. The region entitled Germany includes Germany, Austria, Finland, Sweden, Switzerland, and the Netherlands. The latter however, can also be excluded: Dutch production will continue to meet Dutch demand; we focus only on net importers. The last region referred to as Italy insists only of Italy and Greece.9 As for exporters, we again apply some aggregation and distinguish three major non-EU natural gas exporting regions: Russia, Algeria, and Norway. Russia represents Russian and Former Soviet Union Republics’ supply. Algeria is the collective name for the North Africa natural gas and liquified natural gas suppliers, e.g. Lybia and Egypt, which deliver almost all of their LNG to the EU. The calculated Shapley values illustrate the findings of our theoretical model and provide the intuition.
4.1
Calibration
We calibrate the model to describe projected natural gas trade in Europe around 2030. We assume that European importers first consume all projected domestically produced gas, and hence focus on imports from outside Europe to meet residual gas demand. Demand We estimate the parameters of residual demand for gas from Algerian, Russian, and Norwegian regions using the projections from IEA (2005), OME (2004), and British Petroleum (2008). We assume that demand for gas in each of the importing regions is growing at an average rate of 1.6% per year. Hence, the regions of France, Germany, and Italy will demand up to 40% more gas by 2030. Aggregating residual demand (correcting for European production projections) we find that the region of France will buy about 86 bcm/a from exporters to the EU, 26bcm/a of which are for the UK, the region of Germany will import around 110 bcm/a, and the region of Italy will demand roughly 96 bcm/a. For simplicity we assume that any demand not met by our strategic sellers – Norway, Algeria and Russia – will be obtained from an outside backstop source at an exogenously fixed price. Such sources may include demand response, switching to other fuels or imports from a global LNG market. Importers will be willing to buy their residual demand from the strategic sellers at a variable price up to this exogenous price ceiling. Although it is difficult to 9 We also do not include Ireland into the list of involved countries, because its present and the expected future consumption of imported natural gas is negligible.
13
make projections about the price for gas in the next twenty years, we follow IEA (2005) and assume a low case price to be 250 euro per thousand cubic meters and a high level price to be 300 tcm/a. Supply To calculate supply surplus for exporters we estimate linear total longrun marginal cost functions, which include marginal production, transportation and investment costs. For our calculations we use the following marginal cost formula: mcA (q) = 160 + 0.25q mcR (q) = 127 + 0.35q mcN (q) = 148 + 0.35q with q in bcm/a, and costs in euro/tcm. Details of our assessment are provided in the appendix.
4.2
Results
Now, following our theoretical study, we describe numerically how the importers’ coordination, export constraints, and investments affect trade benefits and distribution of market power. We report the results in tables 1, 2, and 3. For each country under the given scenario we provide its absolute payoff and its relative share in the total trade surplus in that scenario. The latter we also interpret as the bargaining or market power of a player. The first column in table 1 describes the “base” case when there is no cooperation or any type or coordination among the importers and there are no constraints on how much can be purchased from a particular exporter. In table 1 we provide the results of our calculations for the low price case. Table 1: Impact of coordination and aggregate constraints on payoffs and bargaining power Base bn %
EU bn
%
EU, kR ≤ 70 bn %
kR ≤ 70 bn %
Germany France Italy
6.17 4.82 5.37
23 18 20
} 16.80
64
} 14.80
59
5.30 4.18 4.64
21 16 18
Algeria Norway Russia
2.43 2.69 4.90
9 10 20
2.50 2.50 4.62
9 9 18
4.00 3.76 2.63
16 15 10
3.77 3.92 3.40
15 16 14
In the second column we look at how the payoffs and the distribution of the bargaining power changes if the buyers purchase jointly, acting as a single monopsonist. We refer to a proxy player for the buyers as “EU”. In principle, as discussed in the theoretical part of the paper, the result of the buyers’ unionisation may be either beneficial or detrimental for the buyers. In our case, 14
we observe that through establishing a purchasing block buyers win, their aggregate profits increase by around 440 mln. Total welfare remains the same as in the base case, and accordingly buyers also improve their strategic position, increasing their bargaining power by 3 percentage points. The exporters in turn lose 3 percentage points. The third column reports the results for the case when the European incumbents continue to play as a single entity and a regulator caps maximum supply coming from Russia at 70 bcm/a. Given total demand of around 292 bcm/a, this means that Russia is limited to supply not more than about a quarter of total imports. In line with proposition 1 we observe that the proxy player “EU” loses in absolute as well as in relative terms. Buyer bargaining power decreases by 5 percentage points and turns out to be even lower than it was before the cooperation. In total the EU 15’s surplus declines by 2 bn or by over 12 percent. The target of the restriction, Russia, gives up its dominant exporter position. The producer loses almost a half of its bargaining power and becomes the weakest market player. At the expense of Russia the other producers are getting stronger thanks to the redistribution of the demand. Norway gains over one billion in comparison to the “EU” case. Algeria, now the largest producer, enjoys 1.56 bn increase of its payoff and 7 percentage points growth in its relative share. These findings are in line with the observation that capping one supplier’s aggregate market share cannot benefit the EU region. The fourth column, with the heading kR ≤ 70, describes the situation when the importers act as individual players but obey the restriction not to buy more than 70 bcm/a from Russia in aggregate. The impact of the restriction is stated in proposition 2. First, we note that the constraint harms the buyers, they lose in terms of profits and in terms of bargaining power. The intuition behind this result is the following. When Russia can cover only a limited part of the buyers’ demand the other suppliers, Algeria and Norway, become more valuable. Hence, we obtain that while Russia loses its rivals strengthen their position and earn higher payoffs. As follows from the table the buyers with the highest demand, Germany and Italy, suffer most. The payoff of the former region declines by 870 mln and the payoff of the latter decreases by 730 mln, that is each loses roughly 14% of their consumer surplus. The absolute loss of the French region is slightly smaller, 620 mln. Besides, the strategic position of every incumbent weakens by about 2 percentage points. Altogether buyers weaken their position versus sellers by 6 percentage points. Looking at the numerical results for the sellers we notice that Russia’s bargaining power drops by almost a third from 20% to just 14%. Thanks to the constraint all the sellers become almost equally competitive with a minor advantage for Algeria and Norway. In the next table 2 we present the numerical results for the situation when the EU authority imposes constraints on bilateral trade in order to restrict seller power and bring benefits to buyers. For comparison we also list the results of the ”base” scenario. As suggested by proposition 3 the positive effect of bilateral trade constraints can with certainty be obtained in the case when the number of sellers is smaller than the number of buyers. So for the results in table 2 we split the “France” region into the UK and France. Following the demand data 15
Table 2: Impact of bilateral constraints on payoffs and bargaining power Base, p=250 bn %
4 regions bn %
Constraints bn %
Base, p=220 bn %
Constraints bn %
Germany France Italy UK
6.17 4.82 5.37
23 18 20
6.16 3.36 5.37 1.45
23 13 20 5
6.21 3.37 5.40 1.46
24 13 20 6
6.39 3.48 5.58
26 14 23
6.40 3.48 5.58
26 14 23
Algeria Norway Russia
2.43 2.69 4.90
9 10 20
2.44 2.70 4.92
9 10 20
2.48 2.78 4.72
9 11 18
2.15 2.42 4.46
9 10 18
2.29 2.68 4.06
9 11 17
we take that the UK requires about 26 bcm/a by 2030 to cover its needs, with 60 bcm/a left for French demand. Since we keep the aggregate demand and the aggregate supply the same, the value of the grand coalition, or the sum of all the payoffs stays the same. In the second column, entitled “4 regions”, we list the payoffs for the four buyers and three sellers case. The split of the France region leads to the division of its payoff corresponding to the demand values. Otherwise the payoffs and relative shares of the players hardly change: the buyers lose insignificant amounts of their profits in favour of suppliers who benefit as the competition between the buyers becomes more aggressive. The last column in the first part of table 2, named “Constraints”, illustrates the conclusions of proposition 3. We select the constraints on the bilateral trade so as to make the importers better off. We assume that the large buyers – Germany, Italy, and France – are constrained to buy not more than about two thirds of their total needs in bilateral trade from Russia, 80 bcm/a, 60 bcm/a, and 40 bcm/a respectively. In addition, France should not negotiate buying more than 40 bcm/a from Algeria in any coalition. As for the UK we have to pick up somewhat stronger constraints for this small buyer to exert positive externalities on the large buyers covering the negative effects from their constraints. Namely, we limit the bilateral trade for the UK with any seller by a half of its demand, 13 bcm/a. As follows from the figures, in this situation all buyers enjoy some increase in the payoffs and Germany, the strongest buyer gain 1 percentage point in its bargaining power. As a result of the constraints Russia loses 10 percents of its bargaining power. Algeria and Norway benefit and become stronger competitors. The same result is more difficult to establish for the case when the number of sellers is the same or greater than the number of buyers. We are able to select buyer surplus improving bilateral trade constraints for the three by three case as well, if we assume a slightly lower willingness to pay. In the second part of table 2 we present the resulting figures. We assume a willingness to pay of 220 euro/tcm. In the first of the two columns we report the results for the “base” variant and for the situation when there are constraints on the bilateral trade. Once again we choose the constraints so as to ensure that the buyers are not 16
worse off and that Russia’s position on the market is weakened. Though in this example the change in figures is only insignificant, we still are able to illustrate the effect of the policy. Table 3: Impact of investments and entry on payoffs and bargaining power Base bn %
No contract bn %
Contract bn %
Germany France Italy
6.17 4.82 5.37
23 18 20
6.51 5.10 5.69
24 19 21
6.88 4.95 5.53
25 18 20
Algeria Norway Russia Caspian
2.43 2.69 4.90
9 10 20
2.17 2.43 4.48 1.04
8 9 16 4
2.24 2.51 4.61 0.71
8 9 17 3
In the last table we give the results for the situation when the European importers decide upon investment to help new producers to enter the market, say producers from the Caspian region. Importers may coordinate or make non-cooperative decisions. Here we avoid the difficulty of solving the investment problem to find the optimal level of investments, but rather investigate how investment as such will affect the payoffs of the market players and the distribution of market power. We illustrate the situation discussed in section 3.2. We consider a non-trivial case when the new exporter introduced into the market is not the most efficient one, but let us say the second best and look at two possible situations. First, we assume an investing region pays for field development and transportation infrastructure to ensure the new producer has incentives to enter the market. In other words, the buyer bears all the entry sunk costs. The results are given in the second column titled “no contract”. In the second scenario, we assume that to cover its investment costs the importer signs a long-term contract with the new producer10 , which obliges the latter to sell its gas only to the investor. The results for this case are given in the last column. In both scenarios we assume that one pipeline with capacity of 30 bcm/a is built, which is the expected capacity for the Nabucco pipeline, which we have in mind as an example. Again for comparison we put into the first column the payoffs and the bargaining power values of the situation when there are three buyers and three sellers and no coordination or specific constraints imposed on trade. Comparison of the figures from the first and the second columns demonstrates that the entry improves the buyers’ bargaining position and with this their payoffs. Even Russia which is still the dominant producer loses both in profit and in bargaining power. It gives up 4 percentage points of its relative share to the newcomer – Caspian, which obtains 420 mln or almost 9 % in relative terms. The new 10 Or, alternatively, the pipeline is completely relation specific and geography prevents any sales without the importer’s consent.
17
rival undermines other exporters’ positions as well. Algeria and Norway lose 1 percentage point in bargaining power and 260 mln in profits. For the second column we assume that the new capacities are not contracted and can be used by any buyer. It means that the non-investors free-ride on the investment. Let’s assume that it is Germany, the buyer with the highest demand, who invests. Then, Germany earns 350 mln in annual profits which must cover the annual investment costs.11 Further, we see the pay-offs for other buyers to participate in investment. Thus, France gains 280 mln from the entry and Italy 320 mln. The aggregate increase in buyers’ payoffs is definitely large enough to cover the investment costs. Finally, we explore the situation, when Germany invests under the condition that Caspian producers make a commitment only to sell to Germany. In this situation Germany increases its profit and bargaining power further, by 710 mln and 9% respectively. Interesting, through Germany the other buyers are still able to benefit from the entry, though less than in the previous case. France obtains 130 mln more than in the base case and Italy 160 mln, though neither improves its bargaining position in relative terms. The commitment with Germany makes Caspian exporters less competitive on the market; yet, the presence of the new seller still harms the payoffs and the positions of the well-established producers, like Russia. Hence, we find a third way, in addition to cooperation and export constraints, to limit the power of sellers so that buyers will benefit in absolute terms as well strengthen their bargaining position.
5
Conclusions
EU natural gas import dependence is likely to grow substantially over the coming decades. This may increase market power for the large exporters to Europe. An increasing number of modelling studies of the European natural gas market explores the effects of these developments on the future gas prices in the EU. Usually, in such studies all the bargaining power is assigned to the suppliers. However, the assumption that only exporters can behave strategically, e.g. as Cournot players, is less obvious when there is a limited number of concentrated buyers. We argue that one should take into account that EU buyers may also play strategically and exercise their buyer power to affect the market equilibrium. In this paper we use a cooperative game theory description of these sellerbuyer interactions and derive the market players’ strategic positions and bargaining power. We study how the distribution of bargaining power can be changed by EU policies. We consider how regulation towards diversification of supply routes and restrictions on trade, which are discussed in practice, can help in strengthening EU buyer power. One observation frequently made in policy discussions is that the EU should reduce its dependence on exports from the large suppliers, in particular from 11 This figure likely is too low to cover investment, unless the project is sponsored by other parties, e.g. the producer itself.
18
Russia. Restrictions imposed on aggregate EU imports (from a particular producer) might serve this purpose. Such a policy would reduce exposure to technical or political risks, assuming markets will not be sufficiently developed so that these risks can be properly internalised by individual purchasers contracting the gas. We show that aggregate import caps are ineffective as a tool to improve EU buyer power. The limitations imposed on one producer will make the other suppliers stronger, redistributing power among suppliers rather than benefiting buyers. Instead, we derive and confirm numerically that restrictions on bilateral trade of individual member states with individual producers can have a positive effect both on the individual buyers’ bargaining power and on buyers’ trade surpluses. In the second part of our study we investigate how diversification of supply routes, or entry of new producers into the European market, may improve EU buyers’ strategic positions. We consider investments in pipeline capacity enabling entry of new suppliers. The EU actively promotes various investment projects both politically and financially, for instance the Nabucco pipeline. We show that entry is beneficial and strategically favorable for all EU importers, not only for those that directly contract with the entrant. Hence, even if diversification is beneficial, a free-riding problem among importers may occur.
References F. Asche, P. Osmundsen, and R. Tveteras. European market integration for gas? Volume flexibility and political risk. Energy Economics, 24:249–265, 2002. M. Barina. Future of Natural Gas Markets: LNG. Working paper, 2005. M.G. Boots, F.A.M. Rijkers, and B.F. Hobbs. Trading in the Downstream European Gas Market: A Successive Oligopoly Approach. The Energy Journal, 25:73–102, 2004. British Petroleum. Statistical Review of World Energy 2008. BP Annual Report, 2008. CIEP. Study on energy supply security and geopolitics. Study for DG-TREN, 2004. TREN/C1-06-2002. CNE. Report on national gas consumption in 2007. Comisi´on Nacional de Energ´ıa, 2008. R. Egging and S. A. Gabriel. Examining market power in the European natural gas market. Energy Policy, 34:2762–2778, 2006. European Commission. Progress in creating the internal gas and electricity market, 2008a. COM(2008) 192 final.
19
European Commission. Green Paper: Towards a secure, sustainable and competitive European energy network, 2008b. COM(2008) 782 final. European Commission. Second Strategic Energy Review, 2008c. COM(2008) 781 final. S. Gabriel and Y. Smeers. Complementarity problems in restructured natural gas markets. In A. Seeger, editor, Recent Advances in Optimization, Lecture Notes in Economics and Mathematical Systems 563. Springer, 2006. R. Golombek, E. Gjelsvik, and K.E. Rosendahl. Effects of Liberalising the Natural Gas Markets in Western Europe. The Energy Journal, 16:85–111, 1995. R. Golombek, E. Gjelsvik, and K.E. Rosendahl. Increased Competition on the Supply Side of the Western European Natural Gas Market. The Energy Journal, 19:1–18, 1998. F. Gul. Bargaining foundations of Shapley value. Econometrica, 57:81–95, 1989. A. Haurie, G. Zaccour, J. Legrand, and Y. Smeers. A stochastic dynamic NashCournot model for the European gas market. Technical Report G-86-24, GERAD, Ecole des Hautes Etudes Commerciales, Montr´eal, Qu´ebec, Canada, 1987. D. Helm. European energy policy: meeting the security of supply and climate change challenges. EIB Papers, 12:30–49, 2007. F. Holz, C. von Hirschhausen, and C. Kemfert. A strategic model of European gas supply (GASMOD). Energy Economics, 30:766–788, 2008. F. Hubert and S. Ikonnikova. Strategic Investment and Bargaining Power in Supply Chains: A Shapley Value Analysis of the Eurasian Gas Market. Humboldt University Discussion paper, 2003. F. Hubert and S. Ikonnikova. Hold-Up, Multilateral Bargaining, and Strategic Investment: The Eurasian Supply Chain for Natural Gas. Humboldt University Discussion paper, 2004. F. Hubert and I. Suleymanova. Strategic Investment in International Gas Transport Systems: A Dynamic Analysis of the Hold-Up Problem. Humboldt University Discussion paper, 2006. IEA. World Energy Outlook. OECD/IEA, Paris, 2005. R. Inderst and C. Wey. Bargaining, mergers, and technology choice in bilaterally oligopolistic industries. RAND Journal of Economics, 34:1–19, 2003. W. Lise, B. F. Hobbs, and F. van Oostvoorn. Natural gas corridors between the EU and its main suppliers: Simulation results with the dynamic GASTALE model. Energy Policy, 36:1890–1906, 2008. 20
L. Mathiesen, K. Roland, and K. Thonstad. The European Natural Gas Market: Degrees of Market Power on the Selling Side. In R. Golombek, M. Hoel, and J. Vislie, editors, Natural Gas Markets and Contracts. North-Holland, Amsterdam, 1987. OME. Future natural gas supply options and supply costs for Europe. Report to Madrid Forum, 2004. J. Perner and A. Seeliger. Prospects of gas supplies to the European market until 2030 – results from the simulation model EUGAS. Utilities Policy, 12: 291–302, 2004. I. Segal. Collusion, exclusion and inclusion in random-order bargaining. Review of Economic Studies, 70:439–460, 2003. Y. Smeers. Gas models and three difficult objectives. CORE Discussion Paper 2008/9, 2008. J. Stern. The new security environment for European gas: worsening geopolitics and increasing global competition for LNG. Oxford Institute for Energy Studies, NG 15, 2006. L. Stole and J. Zwiebel. Intra-firm bargaining under non-binding contracts. Review of Economic Studies, 63(3), 1996. G.T.J. Zwart. European natural gas markets: resource constraints and market power. The Energy Journal, 2009. G.T.J. Zwart and M. Mulder. Natgas: a model of the european natural gas market. CPB memorandum 144, 2006.
A
Assessment of costs of supply
We assume that marginal production costs are linear mc(q) = c + mq. These costs are also referred to as the price at source. All three producing countries expect a considerable rise in their production costs. Norway and Russia will switch to new more remote fields in difficult terrain: the former to Norwegian and Barents sea fields, the latter to Yamal peninsular permafrost and Shtokman fields. Harsh conditions of the terrain, distance, and peculiarity of the field formation make the producers use switch to LNG (liquified natural gas) technology for supply. Algeria is already a large supplier of LNG to Western Europe, in particular Spain, France and Italy. It will pursue further developments. To this end more LNG liquefaction capacities will be build along with the new field exploration. Based on Perner and Seeliger (2004) and OME (2004) we assess average production costs for Russia, Norway, and Algeria with respect the current levels of production as mcA (q) = 100.0 + 0.2q, mcR (q) = 85.4 + 0.35q, and mcN (q) = 102 + 0.35q. 21
By the next step we integrate the transportation cost into our cost formula. We average between the results obtained in Hubert and Ikonnikova (2003) and presented in OME (2004) which was shown to provide a good estimate.Finally, we account for investment costs. Exporters, in particular Norway and Russia, have to invest significant amount of money to replace the current depleting fields with the new ones, build infrastructure to connect the new fields with the transit grid, and invest in extension of export pipeline system or LNG fleet. We take the prospect figures for investments in pipelines and field development, annualise them and take per capacity. We use a common approach for the annualisation of I·r investment and find the annual payments as (1−(1/(1+r) T )) , where r = 0.15 is the 12 real interest rate. To obtain the final cost figure we assume that each exporter counts for a minimum 25% mark-up or countries’ royalties when reporting costs. We adjust the total marginal cost functions to obtain the predicted increase in production capacities. Thus, given the price of 250 euro per tcm under the nonlinear contracts, which imply efficient bargaining, we obtain that total supply to France, Italy, and Germany of about 300 bcm/a. It is the exact amount we estimate for the demand functions. Finally, for our calculations we use the following marginal cost formula: mcA (q) = 160 + 0.25q mcR (q) = 127 + 0.35q mcN (q) = 148 + 0.35q
12 Data on LNG production, liquefaction, and shipping costs can be found e.g. in Barina (2005).
22
