Printer Friendly

Natural Gas Transits and Market Power: The Case of Turkey.


The Southern Gas Corridor (SGC) consists of planned pipeline projects that connect the natural gas producers in the Caspian region and the Middle East (Azerbaijan, Turkmenistan, Iran, Iraq and Israel) with the natural gas markets of the European Union (EU). The EU promotes the SGC for two reasons: (1) it would like to diversify its natural gas supplies and (2) it aims to close its growing supply gap that arises due to decreasing indigenous production. Turkey has a key role in realizing the SGC, since Turkey's geographical location is between the producing countries and the EU. This crucial role of Turkey is widely discussed in the literature. (1) Compared to Ukraine, which is a single-source transit country for Russian gas only, Turkey has the potential to become a multi-source transit country fed by several suppliers from the Caspian region and the Middle East or Russia. The goal of the Turkish government, however, is not only to aim for a pure transit role for Turkey, i.e. allow upstream producers the access to the Turkish transmission network and to the EU downstream market, but rather to use its multi-source advantage for actively trading in the natural gas markets, as is outlined in Skalamera (2016):
Turkey, however, bargained hard against a straightforward transit role,
intending instead to take over the role of a hub, which means that it
would buy gas arriving at its borders, consume what it needs, and sell
on the balance at profit.

However, this perception is far away from the economic definition of an energy hub. (2) In economic terms, the Turkish perception means that Turkey wishes to use its geographical location to exercise market power in the European natural gas market (transit market power). If the natural gas producers have market power themselves, Turkey's plans would give rise to double marginalization (Tirole, 1988). This perspective is missing within the current discussion about the SGC although it could potentially eliminate the economic benefits of the entire project.

Hence, the research objective of this paper is to investigate possible implications of Turkey's strategic behavior for the EU natural gas markets and for the economic feasibility of the SGC project. The global natural gas market model COLUMBUS (Hecking and Panke, 2012) is extended and applied in order to simulate strategic behavior of transit countries like Turkey. (3) In a simulation for the year 2030, a case with Turkish market power is compared to competitive Turkish transits, i.e. a scenario in which upstream producers have to pay only transportation costs to ship gas through Turkey to European markets. Besides varying the Turkish behavior, different market structures in the European upstream market are considered, i.e. an oligopolistic upstream market and a competitive upstream market, in order to derive a comprehensive understanding of Turkey's role in the SGC.

The structure of the paper is as follows: In Section 2, a review of literature that is is relevant for the analysis is given. A stylized theoretical model to discuss the problem of Turkish transits is developed in Section 3. Subsequently, in Section 4, the global natural gas market model COLUM-BUS and its inputs are described. Afterwards, the model calibration is discussed. Based on the calibration, Section 5 focuses on the model results and discusses the implications of Turkish transit market power for the EU. Finally, Section 6 concludes.


There are four different streams of literature to which this work is related to: (a) literature about gas market modeling based on non-cooperative game theory, (b) literature about natural gas transits, (c) publications about Turkey's energy relations, and (d) literature focusing on double marginalization.

The first literature stream is based on simulation models that are programmed as mixed complementarity problems (MCP). As the COLUMBUS model that is used within this work, MCPs allow the simulation of market behaviour and thus to consider different forms of competition on different stages of the value chain. An early study is provided by Boots et al. (2004) in which gas producers are represented as oligopolists in a static model called GASTALE. The model considers downstream traders that act either oligopolistically or competitively. The study shows that successive oligopolies in gas markets lead to high prices - similar to the case of successive oligopolies in the SGC in this study. Later on, a dynamic version of GASTALE is developed by Lise and Hobbs (2009) that consider the SGC producers Azerbaijan, Iran and Iraq as potential suppliers for Europe. A further early work is Gabriel et al. (2005a). It also considers the natural gas supply chain as a MCP in which the traders marketing gas of the producers had market power. Several existence and uniqueness results are provided as well as illustrative numerical results. Gabriel et al. (2005b) considers more in-depth numerical simulation of a version of this model for the North American natural gas market. In a later contribution by Holz et al. (2008), a static model named GASMOD is applied to analyze the European gas markets with regard to their market structure. Using data of 2003 they analyze different combinations of competition in upstream and downstream markets and come to the conclusion that Cournot competition in both markets (double marginalization) is the most accurate representation to model the European gas market. In Section 4.3, a similar calibration exercise is done for the years 2014 and 2016. In later research, Holz (2009) extends the static GASMOD model into a dynamic version.

Within the stream of literature that focuses on gas transits, Yegorov and Wirl (2010) analyze games that appear in the context of gas transits. They distinguish between games with a transit country as a net gas exporter (such as the case of Turkem gas transits through Russia) and with a transit country as a net gas importer (such as Turkey). They conclude that the game structure arising from a transit problem is not absolute but depends on geography and international law. Furthermore, von Hirschhausen et al. (2005) analyze Ukrainian market power for Russian gas exports to Central Europe. They focus on the effects of an alternative Russian export route to Central Europe, the Yamal pipeline via Belarus and how cooperation between Ukraine and Russia could have made the investment into the Yamal pipeline unnecessary. Dieckhoner (2012) analyzes Ukrainian transits from a security of supply perspective discussing potential diversification options for Europe like the Nabucco pipeline. Later, Chyong and Hobbs (2014) introduce a strategic European natural gas market model to analyze a gas transit country. They apply their model to investigate the case of the South Stream gas pipeline. The question of Ukrainian transit market power is hereby important for the profitability of this offshore pipeline. Transit market power is represented by a conjectured transit demand curve approach. However, the conjectural variations of the transit country are chosen as a calibration parameter and vary between 0 and 1. This approach is common in natural gas market modeling but also often criticized, e.g. by Perry (1982), Dockner (1992) and Smeers (2008). Within the literature about transit problems, there are further cooperative game theory approaches: Hubert and Ikonnikova (2004), Hubert and Suleymanova (2008), and Hubert and Ikonnikova (2011), for instance, analyze market power of transit countries within the Eurasian supply chain. Furthermore, they examine strategic investments into alternative infrastructure projects to bypass the transit countries and reduce their market power. However, the above-mentioned works focus all on Ukraine, a single source transit country fed by Russian gas only. In the study at hand, the potential multi-source transit country Turkey that would not be dependent on a single dominant exporter is in the focus of investigation.

Within the literature about Turkey's energy relations, there are geopolitical and economic contributions. Cagaptay (2013) discusses geopolitical factors associated with different potential gas suppliers for Turkey. Skalamera (2016) finds that there are many obstacles for Turkey to become a gas hub. Furthermore, Berk and Schulte (2017) show that Turkey's potential to become an important transit country for the European natural gas market is strongly restricted. Moreover, they quantify different drivers that could increase Turkish transit volumes and therefore its importance as a transit country.

Apart from a specific gas market context, there are works that discuss options to avoid double marginalization. Joskow (2010) analyses different factors that impact the decision of companies to either rely on markets to source supplies or to integrate vertically. Double marginalization would be a neoclassical factor favoring vertical integration. In the context of this study, competitive access for upstream gas producers to the Turkish transmission grid would lead to the same shipment quantities through Turkey that vertically integrated companies, i.e. upstream producers owning pipelines through Turkey, would choose. Besides the neoclassical double marginalization approach, which focuses on the implications of market power for market efficiency, Joskow (2010) also mentions Transaction Cost Economics (TCE), which focuses on the implications of market power for firms' efficiency. According to TCE, firms could rely on different contractual relations that minimize other firms' bargaining power, e.g. vertical integration, joint ventures, long-term contracts. The conditions of those contractual relations depend on the degree of asset specific investments required for a given market and the extent to which firms are locked-into already binding contractual relations. In the context of our study, which has a market efficiency perspective, contractual relations could avoid a double marginalization structure and transfer a part of the upstream producers' rent to Turkey or European consumers. Therefore, it is important to note that the simulated configurations (competitive transits and double marginalization) are two extreme outcomes, and bargaining about the rents could also lead to a solution in between.

The value added to the literature of this analysis is two-fold. Firstly, it considers the specific case of Turkey and quantitatively examines its potential to exercise transit market power in the EU gas market. Secondly, a double marginalization approach (successive oligopolies) is used to describe a multi-source transit country like Turkey. (4)


Tirole (1988) describes double marginalization in the most basic setting, the succession of two monopolies in a vertical integrated value chain. In this section, an extended version of this text-book model is introduced to describe a market structure with a multi-source transit country potentially giving rise to double marginalization and suppliers that are not dependent on the transit country. Therefore, a setup with 4 players, 3 producers and the multi-source transit country, is considered in order to obtain insights into the functioning of transit market power. It is assumed that the transit country and the producers are not vertically integrated. Producer 1 can sell volumes [q.sub.1] directly to the final market representing a value chain without double marginalization. Producer 2 (respectively producer 3) is dependent on the transit country and thus can only sell volumes [q.sub.2] (respectively [q.sub.3]) to the transit country that then resells the volumes [q.sub.T]=[q.sub.1] + [q.sub.3] to the final market. Figure 1 illustrates the stylized model. The assumption that all the volumes entering the transit country are resold corresponds to the assumption that no domestic market of the transit country needs to be served (in the absence of indigenous production of the transit country).

The final market has a price [P.sub.F] and a total supply Q = [q.sub.1] + [q.sub.T]. The inverse demand function of the final market is assumed to be linear with an intercept [alpha] and a slope [beta]:

[P.sub.F] (Q) = [alpha]-[beta]Q

The profit-maximization problem of producer 1 with her marginal cost [C.sub.1] is given by:

max([[product].sub.P1]) with [[product].sub.P1] = ([P.sub.F] (Q) - [C.sub.1]) * [q.sub.1] subject to [q.sub.1] [greater than or equal to] 0 (1)

The corresponding first-order conditions with a conjectural variation [r.sub.1] = [partial derivative][q.sub.T]/[partial derivative][q.sub.1], which takes on the value of 0 for Cournot behavior and -1 for competitive behavior of the producer, are:

[C.sub.1] - [P.sub.F] + [beta]*(1 +[r.sub.1])* [q.sub.1] [greater than or equal to]0 [perpendicular to] q [greater than or equal to] 0 (2)

Producer 2 (respectively producer 3) produces gas at marginal cost [C.sub.2] (respectively [C.sub.3]) and sells it to the transit country at the price [P.sub.E]. The problems of the producers 2 and 3 are given by:

max([[product].sub.Pi]) with [[product].sub.Pi] = ([P.sub.E] ([q.sub.i]) - [C.sub.i]) * [q.sub.i] subject to [q.sub.i] [greater than or equal to] 0 for i=2,3 (3)

The corresponding first-order conditions are:

[C.sub.i] -[P.sub.E]-[[partial derivative][P.sub.E]/[partial derivative][q.sub.i]] *[q.sub.i][greater than or equal to]0[perpendicular to][q.sub.i] [greater than or equal to] for i = 2, 3 (4)

The inverse demand function [P.sub.E] ([q.sub.T]) is found by considering the transit country's profit maximizing problem and its first-order conditions with the conjectural variation [r.sub.T] = [partial derivative][q.sub.1]/[partial derivative][q.sub.T]. The transit country's profit is determined by the difference between the price of the final market [P.sub.F] (Q) and the price for which the transit country can buy volumes from the upstream producer [P.sub.E]:

max([[product].sub.TR]) with [[product].sub.TR] = ([P.sub.F](Q) -[P.sub.E]) * [q.sub.T] subject to [q.sub.T] [greater than or equal to]0 (5)

The first-order conditions are given by:

[P.sub.E] -[alpha] + [beta][q.sub.1] + [beta][q.sub.T] + * (1 + [r.sub.T])* [q.sub.T] [greater than or equal to]0 [perpendicular to] [q.sub.T] [greater than or equal to]0 (6)

If [r.sub.T] has the value -1, transits are modeled as competitive, whereas the value of 0 corresponds to a situation in which the transit country exerts market power (Cournot conjecture). If [q.sub.T] > 0 is fulfilled, the equation (6) can be rewritten as:

[P.sub.E] = [alpha]-[beta][q.sub.1] -[beta][q.sub.T] * (2 + [r.sub.T]) (7)

With [q.sub.T] = [q.sub.2] + [q.sub.3], this can be plugged into equation (4). With [r.sub.2] = [partial derivative][q.sub.3]/[partial derivative][q.sub.2] and [r.sub.3] = [partial derivative][q.sub.2]/[partial derivative][q.sub.3], this yields:

[C.sub.i] -[P.sub.E] +[beta]* (1 + [r.sub.i]) * (2 + [r.sub.T]) * [q.sub.i] [greater than or equal to] 0 [perpendicular to] [q.sub.i] [greater than or equal to] 0 for i = 2, 3 (8)

Equations (2) and (8) define the mixed complementarity problem for the stylized model. The important insight is that the inverse transit demand function can be included in the first-order conditions of producer 2 and producer 3. Turkey's inverse transit demand function is implemented in the global gas market model COLUMBUS accordingly as described in detail in the appendix.


4.1 Model Description & Overview

In order to analyze the double marginalization induced by a multi-source transit country within a more complex market, the global natural gas market model COLUMBUS (cf. Hecking and Panke (2012), Growitsch et al. (2014), Hecking et al. (2016), Berk and Schulte (2017) as well as Berk et al. (2017)) is extended and applied. It is an intertemporal partial equilibrium model. Formulated as an MCP, it is able to account for strategic behavior of the upstream sector. Inputs are assumptions about production capacities, demand and gas infrastructure. COLUMBUS is a dynamic model which means that demand for investment into gas production and infrastructure are determined endogenously based on exogenously given economic factors such as investment costs and discount rates.

In its standard version, the COLUMBUS model is only able to consider strategic behavior of the vertical integrated suppliers defined "as a trading unit associated with one or more production regions" (Hecking and Panke, 2012). Transit countries, as in the focus of this study, are not associated with their own production region but can buy gas at their border from the neighboring countries. Therefore, the model is extended by introducing transit countries such as Turkey as profit-optimizing exporters. Technical details of the model extensions as well as a detailed technical description of the existing standard version of the COLUMBUS model can be found in the appendix.

The COLUMBUS model is calibrated with the data described in Section 4.2. Two calibrations with different conjectural variations are considered, one calibration to the year 2014 and one calibration to the year 2016. Both years are relevant because we aim on having calibrated configurations for an oligopolistic and a competitive upstream sector. As shown in Section 4.3, the European gas market of 2014 fits better to an oligopolistic setup, whereas the European gas market of 2016 fits better to the competitive assumption.

4.2 Input Data and Assumptions

4.2.1 Market Characteristics

In line with political and regulatory targets of the EU (5) (ACER, 2015b), further integration of the natural gas markets until 2030 is assumed. The EU market is aggregated into two clusters of countries: (1) a Northern & Western European (NWE) market and (2) a South Eastern European (SEE) market. The respective countries of each cluster are shown in Figure 2. The SEE market consists of the Balkan peninsula and Italy. Italy will be connected to the SGC by the TAP pipeline that is planned to become operational in 2018. The NWE market is composed of the remaining EU countries. The countries of each cluster are assumed to form an integrated market. Integration means that only one entry tariff (respectively exit tariff) has to be paid in order to ship gas into (respectively out of) the integrated market area. (6) A prerequisite for such a market design are investments in pipeline connections between the countries of the market area to reduce the risk of structural congestion. (7) An integrated market implies that there is only one gas price within each market area.

While the NWE market is already today characterized by a high degree of market integration in terms of sufficient infrastructure, competitive hub pricing and a high number of supply sources, the SEE market currently lacks connecting infrastructure and is dominated by Russian gas supply and oil-indexed long-term contracts (ACER, 2015a). However, there are various infrastructure projects (e.g. the CESEC initiative) (8) and regulatory incentives (e.g. agreements between the European Commission and Gazprom about destination clauses and pricing issues in LTCs (9) that aim on increasing the market integration within the SEE region. Hence, the assumption of an integrated market in SEE in 2030 is in line with the EU's long-term energy strategy. The modeling of two segments of the EU gas market allows a differentiation of effects of imports via the SGC on the NWE and SEE markets.

4.2.2 Demand

The model is based on linear demand functions as in Lise et al. (2008). Inputs for each demand region are a reference demand, reference price and point elasticity of demand. (10) The fundamental data source for the historical reference demand is the Natural Gas Information 2017 (IEA, 2017). The future development of the reference demand is based on the projections of the Medium Term Gas Market Report 2015 (IEA, 2015a) and the New Policies Scenario of the World Energy Outlook 2015 (WEO) (IEA, 2015b). Hence, a nearly constant demand development in the EU is considered in this analysis. The point elasticities of demand are chosen in line with Growitsch et al. (2014) and Egging et al. (2010). Thus, for instance, for Europe a price elasticity of -0.25 is assumed. The European reference price is based on the Title Transfer Facility (TTF) price for the history, whereas the future development of reference prices is in line with (IEA, 2015b).

4.2.3 Production

The indigenous production of the EU is modeled exogenously, i.e. the respective EU reference demand is reduced by indigenous production. However, all external natural gas suppliers relevant for the EU such as Norway, Russia, suppliers from North Africa, but also potential suppliers from the SGC as Azerbaijan, Turkmenistan, Iran, Iraq and Israel as well as global players that are able to supply LNG to the EU, are modeled endogenously. The input data about production capacities, operational and capital costs is based on a comprehensive literature research of current and historic upstream projects. Data has been obtained from Seeliger (2006), Aguilera et al. (2009), Hecking et al. (2016), publications of the Oxford Institute for Energy Studies, current press notifications about new field discoveries / developments, and by exchange with industry experts.

4.2.4 Infrastructure

The COLUMBUS model encompasses the major elements of the global gas infrastructure including pipelines and LNG terminals. Some projects that have reached the financial investment decision (FID) status are exogenously given to the model (e.g. LNG terminals in the USA and Australia). The data for the existing pipeline infrastructure in Europe is based on the capacity map and the Ten Year Network Development Plan (TYNDP) of the European Network of Transmission System Operators for Gas (ENTSOG, 2015). In Turkey, the existing pipeline connections from Russia (Blue Stream), Georgia (Southern Gas Pipeline) and Iran (Tabriz-Ankara Pipeline) are modeled. In addition, the first stage of the Trans Anatolian Pipeline (TANAP) and the Trans Adriatic Pipeline (TAP) are considered in the model with commissioning in 2018 and 2020. Information regarding LNG liquefaction and regasification capacities has been gathered from publications of Gas Infrastructure Europe (GIE) (GIE, 2015) and from the LNG Industry Report 2015 which is published by the International Group of Liquefied Natural Gas Importers (GIIGNL, 2016). Facts about gas storage originate from reports of Gas Storage Europe (GIE, 2015) and the Natural Gas Information 2017 (IEA, 2017).

Besides investment costs, short-run marginal transport costs are relevant for the market equilibrium. As already mentioned in Section 4.2.1, for the two considered European market areas, uniform capacity weighted entry/exit tariffs based on ACER (2014) for 2014 and on ACER (2016) for 2016 and 2030 are used. (11) The Ukrainian entry/exit tariffs are from Interfax (2015). (12) Transport costs for the SGC, for the South Caspian Pipeline (SCP), for the TANAP and for the TAP are based on a detailed analysis by Pirani (2016). A distance-based approach is applied to derive transport costs for other non-European world regions for which no detailed cost data is available.

The analysis is based on a pure economic rationale. This means that if not explicitly stated no political constraints are considered. Such constraints could be for example limited pipeline investment options between countries hostile to each other, or limited production capacities in countries that are politically unstable. While we know that political factors should be taken into account for a comprehensive analysis of Turkey's role in the SGC, we think that the economic perspective is helpful to understand drivers of all relevant stakeholders in gas markets including political actors. (13) Furthermore, the model does not consider discrete investment choices. Therefore, the simulation may also identify small capacity demands for investment into infrastructure that would not take place in reality.

4.3 Model Calibration

The calibration results are shown in Figure 3 and Figure 4. Figure 3 illustrates modeled and historical EU natural gas supply by source in 2014 and 2016. The respective bar in the middle depicts historical data from IEA (2017). The left bar illustrates the COLUMBUS simulation results if the upstream sector behaves oligopolistically, and the right bar is the result for competitive behavior. For 2014, it becomes clear, that the oligopolistic case matches history better than the results with the competitive assumption. In the competitive case, about 5% more gas would have reached the EU gas markets compared to the historical imports. According to the model results, especially Russia withheld gas volumes in 2014. In 2016, there is an opposing picture. Figure 3 shows that the simulation of competitive behavior of the upstream producers matches reality better than oligopolistic behavior. In a market with oligopolisitc behavior, 6% less gas would have been consumed in 2016 compared to the actual consumption. However, when comparing both behaviors, it becomes clear that Russia was able to deter additional LNG imports by offering its gas at more competitive prices.

Furthermore, Figure 4 shows the historical average European import natural gas prices of 2014 and 2016. (14) It depicts also the price results of the COLUMBUS simulation, differentiated for the NWE and the SEE market as well as for an oligopolistic and a competitive upstream behavior for each respective year. Again, it can be seen that in 2014 the simulation of oligopolistic suppliers fits the reality best. For 2016, historic prices match better with a simulation of competitive suppliers.

However, it is hard to predict today if the upstream producers will behave oligopolistically or competitively in 2030. Therefore, both potential developments are considered in the following analysis.


5.1 Turkish Transit Market Power in an oligopolistic European gas market

In order to analyze the effects of Turkish transit market power in an oligopolistic EU gas market (based on the conjectural variations for the model calibration to 2014), two different scenarios are investigated: (1) a scenario with competitive Turkish transits, i.e. the SGC upstream producers (15) can access the Turkish transmission system and have to pay only the transport costs, and (2) a scenario with Turkish transit market power, i.e. the SGC upstream producers have no own access to the Turkish transmission system and need to sell the volumes to an Turkish exporter (for an overview of all considered scenarios in this analysis cf. Table 3 in Appendix C). (16)

Initially, a scenario with competitive Turkish transits is considered. The left bar of Figure 5 illustrates the simulated EU supply mix with competitive Turkish transits in 2030. Due to exhausting resources, the EU natural gas production declines from 125 bcm in 2016 to 98 bcm in 2030. For similar reasons, Norwegian imports are diminished from 115 bcm in 2016 to 65 bcm in 2030. In addition, Russian imports decrease from 149 bcm in 2016 to 112 bcm in 2030 in the oligopolistic scenario due to the withholding of quantities. The LNG market, which is assumed to be competitive, partly fills the resulting supply gap. Another part is filled by imports from the SGC via Turkey. On this route 45 bcm reach the EU market in 2030. Assuming that 10 bcm/a of SGC capacity is already financed in the TAP project and will be realized, this means that an additional pipeline capacity investment into the SGC of 35 bcm/a would be economically viable according to the model results. Obviously, Turkey and the SGC producers could benefit from an oligopolistic EU market situation with high prices in 2030. Hence, the share of EU's gas consumption that arrives via the SGC could be about 9%.

Besides the scenario with competitive transits, a scenario in which Turkey acts as a Cournot player, buying gas from the neighboring SGC producers and reselling it to the EU with a profit margin, is considered (a behavior called Turkish transit market power in the following). (17) The SGC producers are assumed to have pipeline access to the EU market via Turkey only. Because the SGC producers are modeled as Cournot players as well, this implies successive oligopolies with double marginalization as described in Section 3. (18) Pipeline investments on Russian territory by non-Russian actors are thereby excluded. The assumption that SGC producers are not able to deliver gas via Russia to the EU is relaxed in Appendix 10.2. The simulation results of the scenario when Turkey exerts market power are shown in the right bar of Figure 5. If Turkey exerts market power, Turkish re-exports would be much lower than in the competitive case at 13 bcm/a or additionally to the TAP capacity 3 bcm/a in 2030. For the EU this would mean higher gas prices and thus a slightly lower demand (-10 bcm/a). However, most of the gas that would originally be imported via Turkey could be replaced by higher LNG imports (+10 bcm/a) as well as higher direct imports from Russia (+7 bcm/a).

The effect of Turkish transit market power on the EU gas market prices in 2030 is shown in Figure 6. (19) The figure again compares a situation with (left bar) and without (right bar) the exertion of Turkish transit market power. Additionally, due to the differentiation of the EU markets into a NWE and a SEE market, regional prices in Europe can be analyzed. In the competitive scenario, prices are lower in SEE than in NWE in 2030. This is opposed to today's situation in which prices in South Eastern Europe are the highest on the continent. As already illustrated in Figure 4 in Section 4.3, the calibration results also show higher prices in SEE in 2014 and 2016 than in NWE. This can be explained with the fewer number of exporters that offer gas in the SEE market compared to NWE. If the SGC producers enter the market as new suppliers via Turkey, competition increases and leads to lower prices. However, if Turkey would exert market power, the positive effect of further market entries diminishes resulting again in higher prices in SEE. It can be observed that by the exertion of Turkish transit market power prices in NWE would be 4.3% higher, while prices in SEE would be 6.9% higher than in a situation with a competitively behaving Turkey. This points out that in an oligopolistic European gas market structure the strategic behavior of Turkey would have a significant economic impact, in particular on the SEE market.

Figure 7 shows the implication of Turkish transit market power on the profits of Turkey, Russia and the SGC producers. Additionally, the figure points out the impact on the EU consumer surplus. It shows the differences in profits and consumer surplus between a competitively acting Turkey and when Turkey exerts transit market power. In the competitive case, the Turkish profits are by definition 0. Thus, if Turkey exerts market power, it earns profits of 1.8 billion EUR in 2030. Due to less SGC gas within the EU gas markets, more Russian gas is exported to the EU in the transit market power case which leads to higher Russian profits of 2.5 billion EUR. However, profits of the SGC producers are 13.1 billion EUR lower in 2030. The EU suffers a loss of consumer surplus by 6.6 billion EUR. (20)

The results discussed so far have focused on transits of the SGC producers via Turkey to the EU. However, it is also important to look at the domestic Turkish gas market. Within this study it is assumed that Turkey would not exert market power in its domestic market. This is in line with a policy of the Turkish government to aim on low domestic gas prices that support economic growth. Thus, the domestic market can be directly supplied by all connected exporters. In the scenario with competitive Turkish transits, Turkey's modeled gas demand grows to 63 bcm in 2030 from 46 bcm in 2016. If Turkey exerts transit market power, its domestic demand is expected to amount to 65 bcm in 2030 according to the model results. In the market power case, the SGC producers have an incentive to ship gas to the Turkish domestic market instead of using the expensive transit option to the EU. Hence, the competition in the Turkish domestic market increases leading to 5% lower gas prices and hence to 1.1 billion EUR additional Turkish consumer surplus compared to the case with competitive Turkish transits. Thus, Turkey benefits twice by exertion of transit market power: (1) by profits from transits and (2) by a higher consumer surplus in its domestic market.

Figure 8 shows the origin of the gas exports of Turkey to the EU in 2030. It compares the transits for both considered scenarios (with and without the exertion of Turkish transit market power). If Turkey behaves competitively, about two thirds or 30 bcm of Turkish transits to the EU is Azerbaijani gas from the Shah Deniz field in the Caspian Sea. Since no Iranian sanctions are considered (pure economic rationale), an additional 11 bcm of Iranian gas would reach the EU market via Turkey in 2030. This figure seems to be quite small compared to the fact that Iran has the world's largest natural gas reserves (BP, 2016). Nevertheless, according to the model results, Iran supplies other markets than the EU such as Pakistan, India or the global LNG market. (21) Furthermore, about 4 bcm of expensive Israeli off-shore gas from the Mediterranean Sea would reach the EU. Turkmenistan and Iraq would not transit gas via Turkey to the EU due to comparably low price signals and the far distance. They would only supply the Turkish domestic market (both would deliver about 7 bcm). Whereas Turkmenistan would supply gas to Asian customers, the exports from Iraq are limited due to the increasing indigenous demand (mainly gas demand from the crude oil production).

On the contrary, if Turkey exerts market power, nearly all of the 13 bcm gas transits that would reach the EU would be from Azerbaijan. The reason lies in the country's missing alternative demand sinks and thus the strong Azerbaijani dependence on Turkey compared to Iran that can ship gas to the above mentioned alternative markets. Gas from Israel, however, would be too expensive and not exploited. Again, Turkmenistan and Iraq would deliver the Turkish domestic market only (Turkmenistan: 12 bcm, Iraq: 10 bcm). Besides that, Turkmenistan would focus on non-European markets. Against this background, Appendix D.2 considers a sensitivity analysis in which Azerbaijan can ship gas through Russian territory to the EU. In such a setup, market power would earn Turkey profits in the range of of 0.4-0.5 billion EUR instead of 1.8 billion EUR in the case without Russian transits.

5.2 Impact of Turkish Transit Market Power in a competitive European gas market

In the next step, the effects of Turkish transit market power in a competitive EU market is investigated. Hereto, we use the same conjectural variations as for the model calibration for 2016 (cf. Section 4.3). However, as already shown by Berk and Schulte (2017), (22) the chance of Turkey to become an important transit country for the EU is quite limited under competitive market conditions. There is only a minor demand for expensive gas from the SGC in a competitive EU gas market setting with a nearly constant future gas demand development. Similar results are found in this study.

If Turkey behaves competitively and SGC producers have to pay only the current TANAP transit fees, 23 bcm of gas would pass through Turkey to the EU in 2030. Nearly 18 bcm would come from Azerbaijan and approximately 5 bcm form Iran. Gas from Israel would be too expensive to reach the EU markets. However, even in this situation Turkey would be able to exert transit market power. Hereby it would earn profits of 0.5 billion EUR. Nonetheless, if Turkey would exert market power in such a competitive environment, the potential of the SGC project to diversify the EU gas markets is negligibly small. Approximately 5 bcm from Azerbaijan would reach the EU gas markets. That means that even the capacity of the already financed first stage of the SGC would be oversized. That underlines the minor importance of the SGC under competitive market conditions.


The results of the study illustrate that Turkey has the potential to exert market power in the EU natural gas markets if an oligopolistic market structure (similar to the historical gas market in 2014) is assumed. If Turkey behaves competitively in this market environment, 45 bcm of Turkish transit volumes would arrive in Europe in 2030 according to the model outcome. In such a situation, gas prices in the SEE region could be lower than in the NWE region because the SGC producers would increase the competition, in particular in the SEE region. In the case of Turkish transit market power, however, the transits through Turkey would be reduced to 13 bcm in 2030, illustrating a big potential to withhold quantities from the markets. According to the model outcome, gas prices in the NWE region would be 4.3% higher in this setting in 2030 compared to a situation with competitive Turkish transits. However, SEE would be most significantly affected by 5.9% higher prices if Turkey exercises market power. The consumer surplus of the EU would be 6.6 billion EUR lower compared to the case in which Turkey behaves competitively. If Turkey would only withhold quantities to the European markets and not to its domestic market, lower gas prices in Turkey would be the consequence. Hence, Turkey could increase its consumer surplus (by 1.1 billion EUR) besides earning profits from transits (1.8 billion EUR) making it attractive for Turkey to use the market power option.

However, in a competitive future gas market setting (similar to the historical gas market in 2016), gas imports via Turkey and the SGC would be only of minor importance, even if Turkey behaves competitively. Hence, also the Turkish potential of pursuing transit market power is limited.

Our analysis illustrates that the economic raison d'etre for the SGC is only given for an EU gas market that is characterized by oligopolistic natural gas suppliers. However, in this oligopolistic environment, Turkey could benefit from exerting market power and hereby eliminate the potential benefits of the SGC for the EU. As a policy implication, the EU could prefer direct connections between supply and demand avoiding new dependencies on transit countries. Other potential policy measures would be the harmonization of Turkey's energy laws with EU directives that guarantee a non-discriminative access to transmission grids, and policies incentivizing contractual relations between the SGC producers, Turkey and European importers. Additionally, the EU as well as the SGC countries could make concessions in other sectors of the economy if Turkey allows competitive transits given that low or no tariffs to access foreign markets can be negotiated with exchange deals in the same or other sectors of the economy.


We thank Marc Oliver Bettzuge, Steven A. Gabriel, Christian Growitsch, Harald Hecking, Stefan Lorenczik, Volkan Ozdemir as well as the participants of the 40th IAEE International Conference in Singapore in June 2017 for helpful discussions. Additionally, we are thankful to three anonymous referees. This paper benefited from the project "Turkey's Potential as Future Energy Hub - Economic Developments and Political Options", which was funded by the Mercator foundation.


ACER (2014). "Annual Report on the Results of Monitoring the Internal Electricity and Natural Gas Markets in 2013." Technical report Agency for the Cooperation of Energy/ Council of European Energy Regulators Ljubljana.

ACER (2015a). "Annual Report on the Results of Monitoring the Internal Electricity and Natural Gas Markets in 2014." Technical report Agency for the Cooperation of Energy/ Council of European Energy Regulators Ljubljana doi:10.2851/14037.

ACER (2015b). "European Gas Target Model Review and Update." Technical report Agency for the Cooperation of Energy/Council of European Energy Regulators Ljubljana.

ACER (2016). "Annual Report on the Results of Monitoring the Internal Electricity and Natural Gas Markets in 2015." Technical report Agency for the Cooperation of Energy/ Council of European Energy Regulators Ljubljana.

Aguilera, R.F., R.G. Eggert, G.C. Lagos, and J.E. Tiltonf (2009). "Depletion and the Future Availability of Petroleum Resources" The Energy Journal 30(1): 141-174.

Berk, I., E. Cam, and S. Schulte (2017). "An Economic Assessment of Turkey's Future Role in European Oil and Gas Supply Security (Forthcoming)" in M. Schroder, M. O. Bettzuge, and W. Wessels (editors), Turkey as an Energy Hub? Contributions on Turkey's Role in EU Energy Supply, Chapter 3 NOMOS, Baden-Baden 1st edition. https://doi. org/10.5771/9783845282190-141.

Berk, I. and S. Schulte (2017). "Turkey's Future in Natural Gas - Becoming a Transit Country?" EWI Working Paper 17/01.

Boots, M.G., F.A.M. Rijkers, and B.F. Hobbs (2004). "Trading in the downstream European gas market: A successive oligopoly approach." The Energy Journal 25(3): 73-102.

BP (2016). "Statistical Review of World Energy 2016." Technical Report June BP London.

Cagaptay, E. (2013). "The Geopolitics of Natural Gas - Turkey's Energy Policy and the Future of Natural Gas." Technical report Baker Institute Houston.

Chyong, C.K. and B.F. Hobbs (2014). "Strategic Eurasian natural gas market model for energy security and policy analysis: Formulation and application to South Stream." Energy Economics 44: 198-211.

Dieckh[delta]ner, C. (2012). "Simulating Security of Supply Effects of the Nabucco and South Stream Projects for the European Natural Gas Market." The Energy Journal 33(3): 1-28.

Dockner, E.J. (1992). "A Dynamic Theory of Conjectural Variations." The Journal of Industrial Economics 40(4): 377-395.

Egging, R., F. Holz, and S.A. Gabriel (2010). "The World Gas Model. A multi-period mixed complementarity model for the

global natural gas market." Energy 35(10): 4016-4029.

Egging, R., C. von Hirschhausen, F. Holz, and S.A. Gabriel (2009). "Representing GASPEC with the World Gas Model" The Energy Journal 30(Special Issue: World Natural Gas Markets and Trade): 97-117.

ENTSOG (2015). "Ten Year Network Development Plan 2015." Technical report European Network of Transmission System Operators for Gas Brussels.

Gabriel, S.A., S. Kiet, and J. Zhuang (2005 a). "A Mixed Complementarity-Based Equilibrium Model of Natural Gas Markets" Operations Research 53(5): 799-818.

Gabriel, S.A., J. Zhuang, and S. Kiet (2005b). "A large-scale linear complementarity model of the North American natural gas market." Energy Economics 27: 639-665.

GIE (2015). "GSE Storage Map."

GIIGNL (2016). "The LNG Industry: 2016 Edition" Technical report International Group of Liquefied Natural Gas Importers Paris.

Growitsch, C., H. Hecking, and T. Panke (2014). "Supply Disruptions and Regional Price Effects in a Spatial Oligopoly-An Application to the Global Gas Market." Review of International Economics 22(5): 944-975.

Heather, P. (2015). "The evolution of Europeantraded gas hubs." OIES Paper NG 104.

Hecking, H. (2015). "Rethinking Entry-Exit: Two New Tariff Models to Foster Competition and Security of Supply in the EU Gas Market."

Hecking, H. and T. Panke (2012). "COLUMBUS - A global gas market model." EWI Working Paper 12/06.

Hecking, H., S. Schulte, A. Vatansever, and S. Raszewski (2016). "Options for gas supply diversifiaction for the EU and Germany in the next two decades." Technical report ewi ER&S/ EUCERS Cologne/London.

Holz, F. (2009). Modeling the European Natural Gas Market - Static and Dynamic Perspectives of an Oligopolistic Market Ph.D. thesis.

Holz, F., C. von Hirschhausen, and C. Kemfert (2008). "A strategic model of European gas supply (GASMOD)." Energy Economics 30(3): 766-788.

Hubert, F. and S. Ikonnikova (2004). "Hold-Up, Multilateral Bargaining, and Strategic Investment: The Eurasian Supply Chain for Natural Gas." Working Paper.

Hubert, F. and S. Ikonnikova (2011). "Investment Options and Bargaining Power: The Eurasian Supply Chain for Natural Gas." The Journal of Industrial Economics LIX(1): 85-116.

Hubert, F. and I. Suleymanova (2008). "Strategic Investment in International Gas-Transport Systems: A Dynamic Analysis of the Hold-up Problem." DIW Discussion Papers 846.

IEA (2015a). "Medium-Term Gas Market Report 2015" Technical report International Energy Agency Paris doi:10.1787/oilmar-2015-en.

IEA (2015b). "World Energy Outlook 2015." Technical report IEA Paris.

IEA (2017). "Natural Gas information 2017."

Interfax (2015). "NKREKU ustanovila "Ukrtransgazu"tarify na tranzit gaza dlja tochek vhoda i vyhod."

Joskow, P. (2010). "Vertical Integration." Prepared for "Issues in Competition Law and Policy."

Lise, W. and B.F. Hobbs (2009). "A Dynamic Simulation of Market Power in the Liberalised European Natural Gas Market." The Energy Journal 30(Special Issue): 119-135.

Lise, W., B.F. Hobbs, and F. van Oostvoorn (2008). "Natural gas corridors between the EU and its main suppliers: Simulation results with the dynamic GASTALE model." Energy Policy 36(6): 1890-1906.

Perry, M.K. (1982). "Oligopoly and Consistent Conjectural Variations." The Bell Journal of Economics 13(1): 197-205.

Pirani, S. (2016). "Azerbaijan's gas supply squeeze and the consequences for the Southern Corridor." OIES Paper NG 110.

Pirani, S. and K. Yafimava (2016). "Russian Gas Transit Across Ukraine Post-2019: pipeline scenarios, gas flow consequences, and regulatory constraints." OIES Paper NG 105.

Seeliger, A. (2006). Entwicklung des weltweiten Erdgasangebots bis 2030: eine modellgestutzte Prognose der globalen Produktion, des Transports und des internationalen Handels sowie eine Analyse der Bezugskostensituation ausgewahlter Importnationen. Dissertation, University of Cologne, Cologne.

Skalamera, M. (2016). "The Russian Reality Check on Turkey's Gas Hub Hopes." Belfer Center Policy Brief.

Smeers, Y. (2008). "Gas models and Three Difficult Objectives." CORE Discussion Paper.

Tagliapietra, S. (2014a). "The EU-Turkey Energy Relations after the 2014 Ukraine Crisis: Enhancing the Partnership in a Rapidly Changing Environment." Fondazione ENI Enrico Mattei Working Paper Series.

Tagliapietra, S. (2014b). "Turkey as a Regional Natural Gas Hub: Myth or Reality? An Analysis of the Regional Gas Market Outlook, beyond the Mainstream Rhetoric." Fondazione ENI Enrico Mattei Working Paper Series.

Tirole, J. (1988). The Theory of Industrial Organization. MIT Press, Cambridge, Massachusetts.

von Hirschhausen, C., B. Meinhart, and F. Pavel (2005). "Transporting Russian Gas to Western Europe - A Simulation Analysis." The Energy Journal 26(2): 49-68.

Wigen, E. (2012). "Pipe Dreams or Dream Pipe? Turkey's Hopes of Becoming an Energy Hub." The Middle East Journal 66(4): 598-612.

Winrow, G. (2013). "The Southern Gas Corridor and Turkey's Role as an Energy Transit State and Energy Hub." Insight Turkey 15(1): 145-163.

Yegorov, Y. and F. Wirl (2010). "Gas Transit, Geopolitics and Emergence of Games with Application to CIS Countries." US-AEE-IAEE 10-044.


The following model description is based on Hecking and Panke (2012). COLUMBUS is a spatial model consisting of vertices and edges. Vertices can be either sources (production facilities) or sinks (demand). Pipelines and LNG shipping routes are connected with edges. Table 1 gives an overview of all sets, parameters and variables in the model.

A.1 Notation: Sets, Variables and Parameters
Table 1: Sets, Dual Variables, Parameters

1. Sets

n,n1 [member of]N              all model nodes
c [member of]C                 cost levels (steps of piecewise
                               linear supply function)
t [member of] T                months
y [member of]Y                 years
p [member of] P [member of] N  producer / production regions
e [member of] E [member of] N  exporter / trader
d [member of]D [member of]N    final customer / demand regions
r [member of] R [member of] N  regasifiers
l[member of]L [member of]N     liquefiers
s [member of]S [member of] N   storage

2. Primal Variables

[pr.sub.p,c,t]                 produced gas volumes
[fl.sub.e,n,n1,t]              physical gas flows
[tr.sub.e,d,t]                 traded gas volumes
[st.sub.s,t]                   gas stock in storage
[si.sub.s,t]                   injected gas volumes in stroage
[sd.sub.s,t]                   depleted gas volumes from storage
[dr.sub.p,c,y]                 depleted resources
[ip.sub.p,c,y]                 annual investment into
                               production capacity
[it.sub.n,n1,y]                annual investment into pipeline
                               transport capacity
[is.sub.s,y]                   annual investment into storage capacity
[ilng.sub.y]                   annual investment into LNG
                               transport capacity
[ir.sub.r,y]                   annual investment into regasification
[]                    annual investment into
                               liquefaction capacity
[mdo.sub.e,d,f]                minimal delivery obligation

3. Dual Variables

[[lambda].sub.p,c,t]           marginal costs of physical gas supply by
                               exporter e to node n in time period t
[[sigma].sub.s,t]              (intertemporal) marginal costs of
                               storage injection
[[alpha].sub.p,c,y]            marginal value of resources in node n
                               at cost level c in year y
[[beta].sub.d,t]               marginal costs / price in node n in
                               time period t
[[mu].sub.p,c,t]               marginal benefit of an additional
                               unit of production capacity
[[PHI].sub.n,n1,t]             marginal benefit of an additional
                               unit of pipeline capacity
[[epsilon].sub.s,t]            marginal benefit of an additional
                               unit of storage capacity
[[PSI].sub.s,t]                marginal benefit of an additional unit of
                               storage injection capacity
[[theta].sub.s,t]              marginal benefit of an additional unit of
                               storage depletion capacity
[l.sub.t]                      marginal benefit of an additional unit of
                               LNG transport capacity
[[gamma].sub.r,t]              marginal benefit of an additional unit of
                               regasification capacity
[[zeta].sub.l,t]               marginal benefit of an additional unit of
                               liquefaction capacity
[[chi].sub.e,d,t]              marginal costs of delivery obligation

4. Parameters

[dem.sub.d,t]                  final customer's demand for
                               natural gas
[Cap.sub.n,t/n,n1t/n,c,t]      monthly infrastructure capacity
[res.sub.n,c,y]                maximum resources
[trc.sub.n,n1,t]               transport costs
[prc.sub.n,c,t]                production costs
[opc.sub.n,t]                  operating costs
[inc.sub.n,y/n,n1,y/n,c,y]     investment costs
[dist.sub.n, n1]               distance between node n and
                               node n1 in km
[LNG.sub.cap]                  initial LNG capacity
speed                          speed of LNG tankers in km/h
[Cf.sub.s]                     conversion factor used for storage
                               inj. and depl. capacity
elt                            economic life time of an asset
[slope.sub.d,t]                slope of the linear demand function
                               in node d
[cv.sub.e]                     conjectural function of exporter e;
                               market power level

The COLUMBUS model is based on profit optimization problems of the different players (exporters, producers, transmission system operators, liquefiers, regasifiers). Each profit optimization problem has corresponding first order conditions. Together with the market clearing conditions, the first order conditions define the model.

A.2 KKT Conditions

A.2.1 Exporters

-[[beta].sub.d,t]+(c[v.sub.e]+1)*[slope.sub.d,t]*[tr.sub.e,d,t] - [[chi].sub.e,d,t] + [[lambda].sub.e,d,t][greater than or equal to]0 [perpendicular to] [tr.sub.e,d,t][greater than or equal to]0 [for all]e,d,t. (9)

-[[lambda].sub.e,n,n1,t] + [[lambda].sub.e,n,t] + [trc.sub.n,n1,t] + [opc.sub.n,t] +[[empty set].sub.n,n1,t] + [[zeta].sub.l,t] + [[lambda].sub.r,t] + [l.sub.t]*2*[dist.sub.l,r][greater than or equal to]0 [perpendicular to] [fl.sub.e,n,n1,t] [greater than or equal to] 0 [for all]e,n,n1,t. (10)

A.2.2 Producers

[mathematical expression not reproducible] (11)

-[[alpha].sub.p,c,y+1] - [[alpha].sub.p,c,y] [less than or equal to]0 [perpendicular to] [dr.sub.p,c,y][greater than or equal to]0 [for all]p,c,y (12)

[in.sub.c,p,y] - [summation over (t[member of]T(y)] [[mu].sub.p,c,y] [greater than or equal to]0 [perpendicular to] [ip.sub.p,c,y][greater than or equal to]0 [for all]p,c,y (13)

A.2.3 Transmission System Operators

[mathematical expression not reproducible] (14)

A.2.4 Liquefiers

[mathematical expression not reproducible] (15)

A.2.5 Regasifiers

[mathematical expression not reproducible] (16)

A.2.6 LNG Shippers

[mathematical expression not reproducible] (17)

A.2.7 Storage

[[beta].sub.d,t] + [[sigma].sub.s,t] + [[theta].sub.s,t] [greater than or equal to]0 [perpendicular to] [sd.sub.s,t][greater than or equal to]0 [for all]s,t (18)

-[[sigma].sub.s,t] + [[beta].sub.s,t] + [[rho].sub.s,t] [greater than or equal to]0 [perpendicular to] [si.sub.s,t][greater than or equal to]0 [for all]s,t (19)

[[epsilon].sub.s,t] = [DELTA][[sigma].sub.s,t] = [[sigma].sub.s,t+1]-[[sigma].sub.s,t] [less than or equal to]0 [perpendicular to] [st.sub.s,t][less than or equal to]0 [for all]s,t (20)

[mathematical expression not reproducible] (21)

A.2.8 Market Clearing Conditions

The market clearing conditions are given by the following equations:

[n.summation over (c[member of]C)][pr.sub.p,c,t]-[tr.sub.e,d,t] + [n.summation over (nl[member of](n1,n)[member of]A)][fl.sub.e,n1,n,t] - [n.summation over (nl[member of](n, n1)[member of]A)][fl.sub.e,n, n1,t] = 0 (22)

[n.summation over (c[member of]E)][tr.sub.e,d,t] + [mdo.sub.e,d,t] + [sd.sub.s,t] + [si.sub.s,t]-[dem.sub.d,t] = 0 [perpendicular to] [[beta].sub.d,t] free [for all]d,t. (23)

Equation (22) must be fulfilled for each exporter e [member of] E that is active at the node n [member of] [N.sub.e]. Additionally, the equation ensures equality of traded volumes and physical flows. Equation (23) defines the gas balance at demand nodes d in month t making sure that the final demand is met.


Equation (9) defines the first-order conditions of the exporter's problem. (23) This problem is re-formulated to optimize profits of exporters that sell volumes to a transit country with the transit country's conjectural variation [] nd the slope of the final demand region with the function [slpe.sub.dem,t]. (24)

-[[beta].sub.d,t]+([cv.sub.e+1])*(2+[])*[slope.sub.dem,t]*[tr.sub.e,d,t]-[[chi].sub.e,d,t]*[[lambda].sub.e,d,t]*[greater than or equal to]0 [perpendicular to] [tr.sub.e,d,t] [greater than or equal to] 0[for all]e,d,t (24)

Equation (24) has the structure of equation (8). The transit country can be modeled as competitive (conjectural variation [] = -1) or as a Cournot player (conjectural variation [] = 0). The exporters that are supplying to a final market (including the transit country itself) have still first-order conditions of the form of equation (9).

Furthermore, the market clearing conditions given by equations (22) and (23) need to be extended. The volumes bought by the transit country [transit.sub.t] need to be included in those market clearing constraints for the nodes at the border of the transit country where it buys the transit volumes n [member of] [N.sub.TR]:

[n.summation over (c[member of]C)][pr.sub.p,c,t] + [transit.sub.t] - [tr.sub.e,d,t] + [n.summation over (n1[member of](n1,n)[member of]A)][fl.sub.e,n1,n,t] -[n.summation over (n1[member of](n,n1)[member of]A)][fl.sub.e,n,n1,t]=0 [perpendicular to] [[lambda].sub.e,n,t]free [for all]e,n[member of][N.sub.TR],t (25)

The volumes bought by the transit country transitt are included in the second market clearing constraint as follows:

[n.summation over (c[member of]E)][tr.sub.e,d,t] + [sd.sub.s,t] - [si.sub.s,t] - [transit.sub.t]= 0 [perpendicular to] [[beta].sub.d,t] free [for all]d,t (26)

Table 2: Data and sources


Reference demand      Natural Gas Information 2017 (IEA, 2017),
                      Medium-Term Gas Market Report 2015
                      (IEA, 2015b), World Energy Outlook 2015
                      (NPS) (IEA, 2015a)
Price elasticities    Growitsch et al. (2014) and Egging et al. (2010)
Reference price       Based on TTF 2014
Production costs      Hecking et al. (2016), Seeliger (2006) and
                      Aguilera et al. (2009)
Existing pipeline     Ten Year Network Development Plan
infrastructure        (ENTSOG, 2015)
LNG facilities        Capacity Map GIE (2015), LNG Industry
                      Report GIIGNL (2016)
Storage facilities    Gas Storage Map GIE (2015), Natural Gas
                      Information 2017 (IEA, 2017)
Transportation costs  ACER Market Report 2014 ACER (2015b),
                      Interfax (2015) and Pirani and Yafimava (2016)

Table 3: Scenario Overview

Number  Section      Upstream Sector  Turkish        Further Scenario
                                      Behavior       Characteristics

1.1     Section 5.1  oligopolistic    competitive    -
 .2     Section 5.1  oligopolistic    oligopolistic  -
 .1     Section 5.2  competitive      competitive    -
 .2     Section 5.2  competitive      oligopolistic  -
A.1     Appendix     oligopolistic    oligopolistic  transits from
        D.2                                          Azerbaijan and
                                                     Turkmenistan via
                                                     Russia possible
A.2     Appendix     oligopolistic    oligopolistic  cartel of Russia,
        D.2                                          Azerbaijan and


D.1 Sensitivity on Turkish behavior in the oligopolistic setup

Figure 9 illustrates how the Turkish transits (by origin) vary if the conjectural variation of Turkey is varied between -1 and 0 in the oligopolistic gas market configuration. It becomes clear that Israel and Iran are very sensitive on Turkey's transit behavior. If Turkey decides to exert market power, transits of these countries via Turkey to Europe are not competitive. Azerbaijan, however, is less sensitive because it is only able to export gas via Turkey (cf. Section 10.2 for a sensitivity in which this assumption is relaxed).

Figure 10 illustrates how the Turkish conjectural variation affects natural gas prices in SEE and NWE. While the SEE price is below the NWE price with competitive Turkish transits, this interrelation changes when market power is exerted: For conjectural variations larger than -0.8, the SEE price is larger than the NWE price.

The market situations with Turkish conjectural variations between -1 and 0 illustrated in the Figures 9 and 10 are cases in which both the EU and Turkey would benefit from the SGC, i.e. Turkey would earn some profits from transiting, and the EU would enjoy lower gas prices compared to a situation with double marginalization. Such a market situation could be e.g. the result of a bargaining process between gas consumers and the transit country (similar to the bargaining between upstream producers and the transit country mentioned in Section 2). However, it is important to note that such a bargaining solution could become obsolete if a competitive upstream market structure is assumed instead of an oligopolistic market because fewer volumes would pass through the SGC in the competitive setup.

D.2 Caspian Gas via Russia in the Oligopolistic Setup

In Section 5, it was assumed that Azerbaijan would be able to deliver gas via Turkey only to reach the EU market. Besides the EU, Azerbaijan could solely sell its gas to Georgia or Turkey. However, the Turkish and Georgian demand for Azerbaijani gas is relatively small and accounted for only 8 bcm in 2015 (Pirani, 2016). Looking into the past, Azerbaijan delivered up to 2 bcm of gas to Russia in 2012 (Pirani, 2016). Since then supplies have declined to zero in 2015. As of 2016, Azerbaijan is even importing about 2 bcm/a from Russia. (25) The main reason is the increasing domestic demand and the underdeveloped production of the Shah Deniz field. This situation may change when the Shah Deniz stage 2 will come online. Then, Azerbaijan would be able again to export gas also to Russia or even via Russia into the EU.

The following sensitivities are implemented within an oligopolistic upstream market structure. Therefore, the results from Section 5.1 are the relevant reference to compare the sensitivities to. In a first sensitivity, it is assumed that Azerbaijan and also Turkmenistan are able to deliver gas competitively via Russia into the EU while Turkey is exerting transit market power. As a consequence, both countries would not deliver any gas via Turkey and total Turkish transits to the EU would only be at 6.7 bcm in 2030. As shown in Figure 11, these 6.7 bcm of natural gas that would reach the EU are Iranian gas. Due to reduced competition in the first-stage oligopoly (SGC producers competing about the transits through Turkey) compared to a situation in which Azerbaijan and Turkmenistan are part of this oligopoly, the remaining SGC producers can exercise more market power when selling gas to Turkey. Hence, it becomes more profitable for Iran to export gas via Turkey to the EU. The EU, however, benefits from Azerbaijani and Turkmen gas supplies via Russia. While EU prices would be 1.5% (0.9%) lower in SEE (NWE) compared to the scenario "Turkish market power" without outside options of Azerbaijan and Turkmenistan, EU's consumer surplus would be 0.1 billion EUR higher. As can be seen in Figure 11, due to lower natural gas transits, Turkey's profit would be 0.5 billion EUR if Azerbaijan and Turkmenistan can circumvent Turkey instead of previously 1.8 billion EUR. But because of lower European gas prices and stronger competition with Azerbaijan and Turkmenistan in its key markets, Russia would also lose 0.7 billion EUR revenues as well as 0.2 billion EUR of profits by allowing transits on its territory compared to the case in which Turkey exercises market power and no SGC producer can ship through Russia. Thus, a situation in which Russia would allow Azerbaijan and Turkmenistan to use its infrastructure to bring additional gas amounts into the EU seems to be not likely. Therefore, this is not a viable solution for a more competitive European upstream gas market.

Another possible scenario would be that Russia buys gas from Azerbaijan and Turkmenistan and resells it to the EU instead of allowing competitive transits - similar to Turkey's assumed behavior. However, it is questionable if double marginalization would be the appropriate approach to describe this setting, since Russia has a huge indigenous gas production with comparably low production costs. Hence, Azerbaijan and Turkmenistan are not in a good position to exert market power against the Russian exporter. (26) Therefore, the scenario in which Russia buys gas from Azerbaijan and Turkmenistan is modeled as a cartel situation in which the three countries offer their gas amounts jointly as one player. (27) Together, these countries are in a strong position to act strategically. Thus, compared to the scenario in which all SGC producers have to sell gas to an oligopolistic Turkey in order to deliver gas to European markets, gas prices are higher in both modeled EU market areas (SEE and NWE) by about 2.8%. This leads to an EU welfare loss of 1.8 billion EUR compared to the Turkish market power scenario with all SGC producers selling to Turkey. Nonetheless, as illustrated in Figure 11, even if Russia and the Caspian producers Azerbaijan and Turkmenistan would form a cartel, still 5.7 bcm of mainly Iranian natural gas would reach the EU markets via Turkey. Turkey could earn 0.4 billion EUR of profits.

Concluding, if Azerbaijan und Turkmenistan can ship gas through Russia (either competitively or by cooperation forming a cartel with Russia), the volumes that Turkey could resell to Europe would be below the already financed TAP capacity of 10 bcm/a. Nevertheless, Turkey could still earn profits of 0.4-0.5 billion EUR from the transits. (28)

Simon Schulte (*) and Florian Weiser (**)

(*) Institute of Energy Economics, University of Cologne, Vogelsanger Strasse 321a, 50827 Cologne, Germany. simon., +49 (0)221 27729 229

(**) Corresponding author. Institute of Energy Economics, University of Cologne, Vogelsanger Strasse 321a, 50827 Cologne, Germany.

(1.) See for instance Berk et al. (2017), Tagliapietra (2014a), Tagliapietra (2014b), Winrow (2013), Wigen (2012) or Lise et al. (2008)

(2.) Heather (2015), for instance, identifies five important requirements for an energy hub: a high level of (1) liquidity, (2) volatility and (3) anonymity as well as (4) market transparency and (5) traded volumes. Furthermore, a physical hub is a location where several pipelines coming from and going to different directions converge and enable physical trade and competition. The Turkish perception of becoming a hub rarely fulfills those requirements. For a deeper discussion of this topic see also Berk et al. (2017).

(3.) In reality, besides buying gas volumes upstream and reselling them downstream, a transit country could exert market power by inducing high transit fees or imposing taxes for gas transits on its territory. Those measures would result in a mark-up increasing the price of gas deliveries through the transit country and hence have a similar effect for the final customers as a policy of the transit country to explicitly buy and resell gas.

(4.) In contrast to Chyong and Hobbs (2014), the conjectural variation of a transit country takes on either the value of the Cournot conjecture or the competitive conjecture. Thus, the critique of arbitrary conjectural variations is not relevant for this analysis.

(5.) Within this study the EU includes the United Kingdom, Switzerland, Norway and all states of former Yugoslavia.

(6.) Uniform entry/exit tariffs are assumed that are calculated as a capacity weighted average of historical tariffs from the ACER market monitoring reports (ACER (2014) and ACER (2016)). Basing the analysis on historical tariffs implies that the costs of further investments into the natural gas infrastructure would be regained at the exit points to the customers. For an interesting discussion of how to derive entry/exit fees in an integrated European market cf. Hecking (2015).

(7.) Persistent congestion within a market area would lead to high redispatch costs that would have to be distributed to the gas customers within the market area.



(10.) The general level of the demand is an input to the model as the reference demand. However, given the fact that the model is an equilibrium model, the equilibrium demand is an output of the model and can deviate marginally from the input demand path.

(11.) Due to the fact that we consider only two market regions within Europe changes of Entry/Exit tariffs have a minor impact only.

(12.) The assumed Ukrainian tariffs from 2015 imply that the Ukrainian route is the most expensive Russian export option to Europe. So despite not modeling the Ukrainian market power with respect to transit volumes endogenously, the Ukrainian market power is reflected in the exogenous tariff assumption.

(13.) Additionally, there is literature that imposes similar restraints with respect to political factors, e.g. Berk and Schulte (2017).

(14.) Average import border price as reported by the World Bank. Applied exchange rate: 1.32 EUR/USD (2014), 1.10 EUR/USD (2016)

(15.) SGC producers are potential suppliers from the Caspian region as Azerbaijan and Turkmenistan or from the Middle East as Iran, Iraq and Israel, that are also assumed to act strategically. Hence, they potentially withhold quantities to generate higher prices.

(16.) In Appendix D.1, a sensitivity analysis on the conjectural variation of Turkey with values between -1 (competitive) and 0 (Cournot behavior) is considered.

(17.) In terms of the theoretical model described in Section 3 this means Turkey has a conjectural variation of 0.

(18.) Russian transits through Turkey are still assumed to be competitive. Russian volumes are not bought by the Turkish Cournot player but can be sold to the European markets through Turkey directly by the Russian exporter that pays competitive transit fees. Turkey is not in the position to force Russia into a double marginalization structure as long as Russia has alternative channels to supply the European markets. Russia's direct investment options to Europe are not restricted and Russia rather prefers such direct routes to the EU as Nord Stream 2 due to lower costs compared to the Turkish transit option.

(19.) Prices are in real terms based on EUR 2014.

(20.) Summing up the differences of all rents shown in Figure 7, Turkish market power leads to a net welfare loss of 15.4 billion EUR compared to a scenario with competitive Turkish transits given an oligopolistic upstream market. This welfare loss could be avoided by contractual relations. In a setup with market power by Turkey, especially long term contracts with minimum take-or-pay obligations between European importers and the SGC producers could lead to fixed volumes flowing through the SGC. Additionally, a transit contract with Turkey could be signed. Another possible contractual relation would be a joint-venture of the SGC producers, Turkish transmission operators and European importers. Besides neoclassical approaches, future research could consider transaction cost based theories for a comprehensive analysis of the most suitable contractual relations in the SGC.

(21.) For a more detailed discussion about Iranian exports see Berk and Schulte (2017).

(22.) A further study that investigates the role of the SGC under competitive market conditions is Hecking et al. (2016).

(23.) Growitsch et al. (2014) use a different convention of conjectural variations. This explains the difference between equation (11) in Growitsch et al. (2014) and equation (9).

(24.) In the study at hand this is in particular the slope of the linear demand function of the EU market which is modeled in two regions. A more detailed description of the regions is given in Section 4.2. It is based on each country's linear demand function that are aggregated for the respective EU regions. The parameters of the EU demand functions determine the demand function for Turkish transit gas.


(26.) Turkey, on the other hand, does not have many options to buy gas from different producers.

(27.) For modeling a cartel the same modeling approach as in Egging et al. (2009) is chosen.

(28.) In reality, it is possible that the Caspian countries and Russia could find a form of cooperation between competitive transits and the cartel. In principle, a transit problem can also be seen as a bargaining problem in which cooperation (cartel) and Cournot competition among the respective producers would be extreme outcomes (cf. the discussion in Section 2 about options to avoid double marginalization). However, both considered scenarios with respect to the relations between the Caspian countries and Russia have similar implications for the SGC, i.e. if Azerbaijan and Turkmenistan ship through Russia, the volumes coming through the SGC are diminished.
COPYRIGHT 2019 International Association for Energy Economics
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Schulte, Simon; Weiser, Florian
Publication:The Energy Journal
Article Type:Report
Geographic Code:7TURK
Date:Mar 1, 2019
Previous Article:Understanding Dynamic Conditional Correlations between Oil, Natural Gas and Non-Energy Commodity Futures Markets.
Next Article:From Residential Energy Demand to Fuel Poverty: Income-induced Non-linearities in the Reactions of Households to Energy Price Fluctuations.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters