مقررات انتقال بهینه بازار یکپارچه انرژی

The capacity of the transmission network determines the extent of integration of a multi-national energy market. Cross-border externalities render coordination of network capacity valuable. Is it then optimal to collect regulatory powers in the hands of a single regulator? Should a common system operator manage the entire network? I show that optimal network governance depends on (i) whether the centralized regulatory agency is able to balance the interests of the different countries; (ii) asymmetries across countries in the gains from market integration; (iii) network characteristics (substitutability versus complementarity); and (iv) the social cost of operator rent.

The European Commission (2007) views the completion of an integrated European energy market as essential for ensuring competitiveness, sustainability and security of energy supply in Europe. Market integration depends crucially on the transmission network connecting the member states being capable of reliably transporting energy from power plants in one country to consumers in another. Increasing shares of solar and wind energy place additional requirements on the grid as production and energy flows become more volatile. The transmission network is considered a natural monopoly in many countries. In a monopoly setting, establishing an efficient transmission network boils down to implementing a well designed regulatory policy. Most liberalized electricity markets have been restructured one country at a time. Owing to the national scope of liberalization, transmission regulation has also been national in scope. National regulatory agencies govern national system operators who own and manage the national transmission networks. The question is whether transmission governance along national borders is still optimal in a multi-national energy market. In an integrated energy market, improvements in grid capacity at home have implications even abroad because the removal of each transmission bottleneck affects energy flows and prices across the entire market. With too narrow a focus on domestic effects, national regulatory agencies run the risk of ignoring externalities abroad when devising regulatory policy for the national system operator. Two examples from the Nordic electricity market illustrate the influence of narrow national interests over transmission management. The Nordic electricity market constituted the world's first multi-national liberalized electricity market and now spans Denmark, Finland, Norway and Sweden.1 In the spring of 2008, a number of transmission lines connecting southern Norway and southern Sweden broke down, severely limiting export capacity to Sweden. According to the Norwegian regulator, the line failures were largely due to insufficient maintenance by the Norwegian system operator, Statnett. Admitting that the repairs were taking an unusually long time, Statnett emphasized that the security of supply for Norwegian consumers was never jeopardized. Meanwhile, the consumers in southern Norway had been enjoying comparatively low electricity prices. The effects on consumers and producers in Sweden (or elsewhere) seem to have been absent from the Norwegian discussion. In 2010, the European competition authority warned the Swedish system operator Svenska Kraftnät that the practice of alleviating domestic congestion problems by limiting exports to Denmark could be illegal. By cutting the outflow of electricity, Svenska Kraftnät was able to offset excess demand in southern Sweden, thereby achieving its objective of a uniform electricity price across Sweden. Danish interests, concerned with higher electricity prices in Denmark, filed a complaint with the EU accusing Svenska Kraftnät of abusing its monopoly position as the sole provider of Swedish transmission capacity. Svenska Kraftnät has subsequently decided to solve Swedish congestion problems through means other than reducing export capacity.2 With the cross-border externalities in mind, would it not be better to establish a common regulatory agency responsible for the entire transmission network? And should the national system operators be merged into a common system operator? In this paper I formally analyze these horizontal aspects of network governance. The discussion has, so far, centered around the costs and benefits of vertical separation of transmission operation from production; see e.g. Cremer et al. (2006) for an analysis and Pollitt (2008) for an account of the arguments. For the fear of integrated utilities discriminating against competitors and investing inadequately in their networks, the EU recommends full ownership unbundling of transmission and production assets ( EU, 2009b). However important vertical structure may be, overall network performance depends crucially on the incentives induced by the regulatory policies adopted by the different member states. The newly established Agency for the Cooperation of Energy Regulators (ACER) reflects this concern. ACER is furnished with the task of coordinating transmission regulation across the EU member states and deciding on the terms and conditions for access to cross-border infrastructure in case of national disagreement. On the system level, The European Network of Transmission System Operators for Electricity (ENTSO-E) is a collaboration of the system operators in the EU with the objective of coordinating and promoting system operator interests. I consider a two-country energy market with interconnected networks. Network reliability is a measure of market integration and increases with maintenance spending in both networks. Gains from energy trade render network reliability valuable, but maintenance and transfer payments are costly. The purpose of regulation is to provide the system operators with the appropriate incentives for network maintenance while minimizing maintenance cost and transfer payments. First-best optimal spending occurs at the point at which the marginal benefit of network reliability equals the marginal social maintenance cost.3 Network governance is a question of both how many regulators there should be and the optimal number of system operators (transmission owners). Therefore, a taxonomy of network structures needs to be compared with one another. The Nordic electricity market exemplifies the governance structure labeled Separation in Table 1. Separation constitutes the most decentralized network structure: Every country has its own national system operator (NSO) regulated by a national regulatory agency (NRA). An advocated contender is full centralization, here labeled Integration, where the responsibility for managing the entire transmission grid is merged in a common system operator (CSO), supervised by a common regulatory agency (CRA). Common regulation constitutes a compromise between Separation and Integration and features a set of NSOs jointly regulated by a CRA. An example of Common regulation is Great Britain, where Ofgem regulates the three transmission owners National Grid Electricity Transmission, Scottish Power Transmission Limited and Scottish Hydro-Electric Transmission Limited. To complete the picture, Common agency describes a situation where multiple national regulatory agencies independently regulate a single CSO. In practice, proponents of a single system operator typically envision a complementary coordination of regulatory policies. For example, an investigation of the desirability of a single Nordic system operator concluded that national governments should simultaneously be forced to relinquish some (regulatory) autonomy, otherwise interference from the national governments would create inefficiencies in system operation ( EMG, 2008). I therefore skip a detailed analysis of Common agency at this stage, although one might want to consider it for the sake of completeness. To compare welfare under the different structures, I assume that the common regulator selects the (for him) optimal regulatory policy. Under Separation, on the other hand, the two national regulatory agencies (NRAs) play a non-cooperative game against each other: Each NRA chooses its regulatory policy to maximize national welfare given the choice of policy by the other NRA. A benevolent common regulator who can commit to complete long-term contracts can always replicate any set of contracts implemented by the national regulatory agencies and can potentially do better. Centralized regulation is always optimal in this case. Therefore, decentralized regulation can be optimal only if (i) the regulator is non-benevolent; or (ii) the regulator has commitment problems; or (iii) there are problems of contractual incompleteness at the centralized level. This paper analyses regulation from a political economy standpoint. 4 The countries differ in their valuation of market integration because of cross-country differences in the gains from trade. The common regulator maximizes a weighted average of national welfare in the two countries, where the weights are meant to capture the political influence of the respective countries over the design of the common regulatory policy. The trade-off between centralized and decentralized regulation is between internalizing cross-border externalities of market integration and tailoring regulatory policies to each individual country to reflect differences in the valuation of market integration. National regulatory agencies (NRAs) provide insufficient incentives for network maintenance because they only consider the domestic and not the foreign gains from market integration. In addition, total maintenance spending is suboptimally distributed across the network because of a lack of coordination between the NRAs. Establishing a common regulatory agency (CRA) takes care of the coordination problem. However, total maintenance spending can be too high or too low under centralized regulation depending on the weight of the different countries in the objective function of the common regulator. If, for example, a country with very little to gain from market integration controls the CRA, maintenance incentives are vastly insufficient because the CRA grossly understates the value of market integration. In this case, regulatory decentralization is preferable to centralization. The key to establishing a well-functioning common regulatory agency thus lies in ensuring a balanced political influence across countries. With sufficiently equal distribution of political power, no country can exert enough influence over the regulatory policy to tilt it in one's own favor. The externality/bias trade-off is classical in studies of political integration and dates back at least to Oates (1972). Ellingsen (1998) notes how asymmetric gains from integration favor decentralization. The importance of political balance for the desirability of centralization has gone relatively unnoticed, as far as I understand (although the result is straightforward), possibly because most models assume majority voting. Laffont and Pouyet (2003) are an exception. They analyze the costs and benefits of decentralized policies in a multi-national procurement model. Unlike in the present paper where political conflict is between countries (inter-jurisdictional conflict), Laffont and Pouyet (2003) assume that political conflict is between shareholders and non-shareholders (intra-jurisdictional conflict). The centralized buyer places less weight on consumer surplus than firm rent if shareholders are in majority, but cares nothing about firm rent if shareholders are in minority. Opposite to this paper, centralized procurement is found to welfare dominate decentralized procurement if and only if influence is asymmetrically distributed between shareholders and non-shareholders. This result can be traced to a peculiar specification of the objective function of the centralized buyer in their model: Under shareholder majority consumer surplus weighs more the larger is shareholder majority. Consumer surplus and firm rent have near equal weights in the limit when almost everybody is a shareholder, in which case the centralized buyer acts almost as the benevolent social planner. Consider next the optimal number of system operators. An informational asymmetry is the source of an agency problem between the regulator(s) and the system operator(s). The regulator has insufficient information to assess whether network performance is inferior for exogenous reasons (low productivity) or endogenous reasons (insufficient maintenance). By understating the productivity of the network, a system operator can secure itself excessive transfer payments relative to the cost of maintaining the network. All else equal, the regulator would prefer to minimize this informational rent because transfer payments are costly to society. Network complementarity means a higher marginal value of maintenance spending in one part of the network, the higher is productivity in the other part of the network. An NSO understating productivity then imposes a negative informational rent externality on the foreign NSO because of reduced maintenance spending and lower transfer payments in the foreign network. By merging system operation in a single CSO, the regulator forces the network owners to internalize this negative rent externality—thereby reducing overall informational rent. In the opposite case of network substitutability, a CSO may have an incentive to underreport productivity in one part of the network in order to increase the marginal value of maintenance spending and informational rent in the other part of the network. It is optimal to split system operation between two NSOs to prevent the CSO from internalizing this positive informational rent externality. However, network substitutability is only a necessary, but not sufficient condition for a positive informational rent externality to arise under a single CSO. A cost complementarity works in the opposite direction. The agency problem is exacerbated in the presence of a CSO because all information about the network is centralized in the hands of a single agent. The superior ability of the CSO to jointly understate productivity in all parts of the network renders it particularly costly to maintain high reliability in a network of uniformly low productivity, because of the additional informational rent. Thus, a CSO understating productivity (i.e. exaggerating maintenance cost) in one part of the network simultaneously exaggerates the (virtual) marginal maintenance cost in the other part of the network. Nevertheless, cost complementarity is weak relative to network substitutability if the social cost of transfer payments is low. In this case, a common regulator would optimally split system operation between multiple national system operators. Dana (1993) and Gilbert and Riordan (1995) are the first to observe that a negative informational rent externality stemming from complementarities in production favors monopoly production. Mookherjee and Tsumagari (2004) generalize these findings by showing that multiple agents are optimal under a positive informational rent externality. Severinov (2008) explores the technological foundations of rent externalities. In his model, the rent externality is negative if inputs are weak complements or weak substitutes, and positive if inputs are asymmetric and strong complements or substitutes.5 I extend Severinov's analysis further and demonstrate that the economic environment, and not only the technological environment, determines the magnitude of the rent externality. The rent externality could be negative, and maintaining a single CSO could be optimal, even under network substitutability, provided transfer payments are costly to society. More generally, the present paper contributes to the literature on multi-contracting. Multi-contracting describes a situation where one or several principals contract with one or several agents. This is the first attempt, as far as I know, to analyze the welfare implications of changing the number of principals (here: regulators) as well as the number of agents (here: system operators) in a unified framework. Starting from a situation with multiple regulators and system operators, centralizing regulation and system operation is socially optimal in case of balanced political power and network complementarities. Conversely, if a country with little to gain from market integration possesses a dominating influence over the common regulatory agency, it is better to fully decentralize regulation and split system operation than to maintain centralized regulation and system operation. The existing literature is more partial in addressing either the optimal number of agents assuming a single principal (see Armstrong and Sappington, 2007 for a survey), or analyzing the optimal number of principals assuming a single agent (see Martimort, 2007 for a survey of such common agency models).

No network governance structure does uniformly better and no governance structure performs uniformly worse than all others in this model. Rather, optimal governance depends on (i) political factors, i.e., how well the common regulatory agency is set up to balance the economic interests of the different countries; (ii) technological factors, i.e., network substitutability versus complementarity; (iii) economic factors, i.e., how the gains from energy market integration vary across countries, and the social cost of operator rent. Having a common regulatory agency is better from a welfare perspective than maintaining national regulatory agencies on the proviso that political influence is sufficiently balanced across the countries participating in the common regulatory agency. With an equal distribution of political power, no country can exert enough influence over regulatory policy to tilt it in one's own favor. The importance of balanced political influence is well understood by the European Union. The Board of Regulators of the Agency for the Cooperation of Energy Regulators (ACER) resides under instructions to act independently from any government of a member state. Only one representative per member state may be admitted to the Board of Regulators, and the board members have one vote each (EU, 2009a). With balanced political influence, the preferences of the regulator are aligned with those of the benevolent social planner. Under those conditions, a common system operator is socially optimal when there are negative rent externalities associated with under-performance, whereas splitting network operation is socially optimal if the rent externalities are positive. For an illustrative example of a positive rent externality, recall the capacity problems between Norway and Sweden described in the Introduction. In May 2008, precisely when the interconnections between southern Norway and southern Sweden failed, the NorNed cable connecting southern Norway and the Netherlands went operational. The longest submarine power cable in the world, NorNed was a prestige project for its owners Statnett and TenneT (the Dutch system operator). Reduced export capacity from southern Norway to southern Sweden, combined with a large electricity surplus that year, pushed down prices in southern Norway and increased the value of electricity trade between the Netherlands and southern Norway. Being the only transmission line directly connecting the two markets, NorNed could sell its transmission capacity at a vastly higher price than projected (Statnett, 2008). Statnett probably internalized part of the increased profitability of the NorNed cable resulting from reduced capacity on its Norwegian–Swedish interconnection. The perceived value of repairing the Norwegian–Swedish connection might have been higher had NorNed instead been fully owned by TenneT, or had the interconnection broken down on the Swedish instead of the Norwegian side of the border. Whether the repairs would have been carried out more expediently under different ownership of the interconnections, remains a matter of speculation.