Regulating National Firms in a Common Market∗ Sara Biancini† December 2008

Abstract The paper studies the regulation of national firms in a common market. Regulators can influence the production of national firms but they incur in a positive cost of public funds. First, we show that market integration is welfare improving in both countries if and only if the efficiency gains compensate for the negative impact of business stealing. Then, we show that supranational competition can have very different consequences on the rent seeking behavior of firms, depending on cost correlation and ex-ante technological risk. Finally, we characterize the global optimum and show how it can be sustained in a decentralized bargaining solution. JEL codes: L43, L51, F15. Keywords: regulation, competition, market integration, asymmetric information, cost of public funds.

∗

I would like to thank Emmanuelle Auriol for constant support and advice. I also thank Jacques Cr´emer for many helpful discussions, Bernard Caillaud, Gianni de Fraja and Bruno Jullien for thoughtful comments, and seminar participants at JEI 2006 Nantes, EEA 2006 Vienna, PET07 Nashville, ADRES 2007 Paris, Bocconi University and THEMA (U Cergy-Pontoise). All remaining errors are mine. † Sara Biancini, Universit´e de Cergy-Pontoise, THEMA, F-95000 Cergy-Pontoise. Email: [email protected].

1

1

Introduction

Historically, monopoly regulation has been a response to market failures, such as increasing returns to scale and externalities. In most countries, government intervention took the form of the creation of public monopolies. More recently, the poor performance of public enterprizes, associated with soft budget constraints and lack of incentives, has motivated widespread reforms introducing partial privatization and liberalization. These interventions were aimed to stimulate productivity and to decrease prices through the creation a more competitive environment. Concerning privatization, econometric studies give mixed results about its effects on efficiency and prices.1 For liberalization, there is some more evidence that it increases productive efficiency, but the effect on prices is still debated.2 One of the problems is that, even after liberalization, high concentration persists, due to residual economies of scale and barriers to entry. In markets where the national leaders stay dominant, the competitive pressure can be increased through market integration. When the boundaries of the market are enlarged, competition can take place even if the efficient scale of production is large. Moreover, market integration can increase welfare in various way: increasing efficiency reallocating production toward the more efficient producers, enhancing the profit possibility of efficient firms and, possibly, limiting the rent seeking opportunities of national monopolies. Supranational institutions in developed and less developed regions are actively engaged in enhancing the development of a common markets. However, the development of supranational competition in regulated industries remains challenging, raising special concerns which are the object of our analysis. A crucial point is that market integration removes barriers to trade, while regulation acts at the national level: competition thus reduces the ability of the regulator to control the national market and to induce the preferred Ramsey-type tariffs. Conflict of interests and lack of coordination between national policies can jeopardize the success of market integration. To address these concerns, a broad welfare analysis of the impact of market integration in regulated market is needed. The European experience offers several examples of regulated markets which have been progressively exposed to competition and in particular to foreign competition. In telecommu1 For instance, Boylaud and Nicoletti (2000), studying reforms in telecommunications, find no evidence that the change in ownership structure matters. Domah and Pollit (2001) and Zhang, Parker, and Kirkpatrick (2002), studying the reforms in electricity in developed and developing countries respectively, reject the hypothesis that privatization per se leads to increasing efficiency or decreasing prices. 2 The idea that the efficiency gains related to reforms are not necessarily transmitted into price have been confirmed by several empirical works on the liberalization of the electricity sector. For instance, the panel data analysis of Green and Newbery (1998), Domah and Pollit (2001) and Hattori and Tsutsui (2003) find that reforms have been associated with increasing prices.

2

nications, the process of liberalization and market integration is probably the most advanced. The larger providers operate at the European level and they have reciprocally challenged their monopoly position in the home country. Some of the main players are public or mixed-public firms, others are completely privatized. Even when privatization and liberalization are introduced, asymmetric restrictions of the activities of the national leader are common: regulation is often asymmetric, placing additional requirements on incumbent or dominant suppliers in order to correct the consequences of their market power (especially when the incumbents are engaged in both competitive and non competitive activities). In addition, direct government intervention seems to be the rule in case of crisis, sometime conflicting with the general antitrust and non discrimination policies.3 Other regulated industries are by far less competitive than telecommunications, but market integration and the removal of barriers to trade put some competitive pressure on the incumbents. An example is the postal service. In recent years, many UPS have bought private parcels operators to consolidate their presence in other member states. In the same way, energy markets are progressively integrated. Market integration is particularly developed in Northern countries, independently of the more general process of integration at the EU level. Norway and Sweden have liberalized their markets, allowing neighbor operators to enter the national market. The incumbent public monopolies have been privatize to a very small extent, but foreign competitors are allowed to serve the market. The situation is similar for transports. In this case, every national leader has market power in its country. For railways, public ownership and government funding are widespread, due to the social value of the industry and the persistent economies of scale. For airlines, the process of privatization is more pervasive. However, government direct participation remains.4 Moreover, the recent crisis of the industry in the early 2000s has shown that, direct government intervention takes place whenever the national carrier encounters a major threat. Even in the United States, government officers are usually in favor of rescuing airlines, creating barriers to exit and soft budget constraints (the government does not allow firms to fail). Finally, in spite of the attempts to make the market more competitive in the last years we have assisted to an increase of concentration at the EU level, with important mergers such as British Airways-Iberia or Airfrance-KLM. The market appears to be concentrated at the European level. This calls for public intervention in order to reduce distortion related to market power. 3

An example is the intervention of the French government in favor of France Telecom, which has been under scrutiny of the EU Commission under the legislation on State aid. 4 For instance, the French government controls 44 percent of Airfrance, which represents 81 percent of the merged entity Airfrance-KLM.

3

Beside the European experience, other processes of regional integration have interested regulated market. One important example is the creation of regional markets for electricity, such as the African Power Pools (West, East and South Africa Power Pools), the regional market of the Greater Mekong Subregion and several initiatives in Central and South America. In all these cases, market integration is expected to generate high efficiency gains, allowing to exploit the economies of scales related to large infrastructures (e.g. dams) and through the creation of a more competitive environment. However, the possible conflicts arising from the lack of coordination of government interventions represents a source of concern and a potential brake on development of the regional markets. The present work studies of the optimal regulation of national firms in a common market. In a common market, the regulation of the former monopoly becomes regulation of the “national champion”. We adopt a model of regulation ` a la Baron and Myerson (1982) in which each regulator sets menu of contracts setting the quantity of the regulated firm5 and a regulatory instrument (tax or transfer), depending on the firm’s production cost (type). In this paper, we do not allow the regulator to contract with the foreign firm. This is equivalent to assuming that there is asymmetric (or incomplete) regulation in each jurisdiction.6 For simplicity, one can think to the case in which the national firm is public. Even in the case of privatized firms, asymmetric regulation is used in practice in many liberalized market, to correct for the consequences of market power. Asymmetric regulation consists in placing additional requirements on incumbent or dominant suppliers. It concerns particularly former monopolies engaged both in regulated and unregulated, which could restrict the development of competition through cross subsidization of activities. Asymmetric regulation is widely used in telecommunications (see for instance Flacher and Jennequin, 2008). In electricity, the presence of many public and mixed firms and the high levels of concentration makes this assumption quite natural7 For instance, European Union allows National Regulators to impose regulations on those operators with significant market power in all in markets that are not effectively competitive. Similarly, asymmetric regulation has been taken as a model in the liberalization processes of most developing countries. Finally, asymmetric regulation can arise from the fact that competitors enter in unregulated segments of the market or in unregulated markets producing substitutes goods8 5 It could be argued that it is more common to regulate prices than quantities. Quantity regulation allows to simplify the analysis. However, this is not crucial for the nature of the results, as explained in Section 2. 6 This assumption is discussed and relaxed in Section 3.1.1 7 For instance, during the California deregulation experiment, a ceiling was imposed on the retail price of incumbent suppliers. 8 For instance, trains and trucks compete in freight transportation and in general the truck industry is not

4

Beside of being common in practice, asymmetric regulation has also been already analyzed in the literature (for instance in Caillaud, 1990 and Biglaiser and Ma, 1995 for the case of entry of unregulated national producers). In the present paper we concentrate on the effect of supranational competition on the contract between the regulator and the national firm. In this context, under complete information, when the cost of public funds is positive, competition is welfare enhancing if and only if the variable costs of the two firms are different enough. In this case, the high cost country benefits from a price reduction and the low cost country from export revenues. When the costs are close, the (negative) business stealing effect prevails. Competition is not very beneficial to consumers (small price effect) and it harms the national firm, and hence tax payers, through business stealing. These welfare effects are robust: under asymmetric information the main trade off between efficiency gains and business stealing still holds. However, under asymmetric information, market integration has an additional impact on efficiency through the seeking behavior of the regulated firm. We show that cost correlation and the level of ex ante technological risk are crucial to determine the nature of this impact. If cost are subject to idiosyncratic shocks, the information rent captured by regulated firms increases at least for low cost types. If cost correlation is high, the rent generally decreases, except possibly for very inefficient types. As a consequence, the welfare impact of market integration under asymmetric information depends both on cost correlation and on the level of ex ante technological risk. Once we have shown the potential welfare reducing effect of market integration, we consider the possibility of cooperation between regulators. In a process of regional integration, regulators can try to achieve collective gains. As a benchmark, we first look to the global first best solution. This is the utilitarian solution which maximizes the sum of the two national welfare. At the globally optimal allocation of production, the country with the less efficient technology is in general a loser of the integration process, even if its own consumers enjoy a lower price. For this reason the efficient solution cannot emerge in a non cooperative framework without side transfers. Each decentralized cooperative solution has to repay the negative impact of business stealing on public finance and the costs related to restructuring. This idea that it could be necessary to sustain the losers of the liberalization process is consistent, for instance, with the practical experience of the introduction of the National Competition Policy (NCP) in Australia. NCP was introduced in 1995: at the time, the government commissioned a public enquiry on the impact of the new Policy on the different communities and social groups. This regulated while railways are heavily regulated. Similarly, high speed railways compete with airlines (subject to lighter regulation) on some routes. In telecommunications, the heavily regulated fixed lines operators are increasingly exposed to competition from mobiles and internet services.

5

was explicitly aimed to evaluate the need for structural adjustment policies towards the losers of the liberalization process. In the context of the European Union, the Structural Funds are the instrument used to reduce disparity in development and in particular “developing infrastructure, (...) targeting the development of trans-European networks in the area of transport, telecommunications and energy” (EC 1260/99). Our results suggest that cooperation in the form of transfers should be used in order to provide funding for infrastructure and restructuring policies. Similarly, for the case of less developing countries, international institution such as the World Bank and international donors should take into account the distributive effects of integration when targeting the distribution of resources for investment and aid.

1.1

Related literature

The literature concerning the interactions between regulation and market integration is not very developed. The strategic trade policy literature, starting from the seminal paper of Brander and Spencer (1983)9 , concentrates on the strategic effects of trade subsidization policies. In these models, market power per se is not detrimental to welfare, since it is exerted only on foreign consumers. Brainard and Martimort (1996, 1997) introduce asymmetric information in a Brander and Spencer framework, showing how the interaction of regulatory policies can reduce the agency cost associated to subsidization policies and mitigate the inefficiencies related to market share rivalry. Combes, Caillaud, and Jullien (1997) develop Brainard and Martimort’s adding and national consumer surplus concerns. They use a common market model, in which governments may subsidize domestic producers. The regulatory instrument is a quantity subsidy (associated with a lump sum tax on profits). They also look at the strategic effect of subsidy policies and find that it is optimal to allow for subsidies in these market (as opposed to the general rule which prevents state aids to firms). However, they don’t consider the fiscal effect of competition, which arises whenever the public funds are costly. As Collie (2000) shows, when public funds are costly and lump sum transfers are not allowed, there exists a full range of the value of the cost of public funds for which it is welfare improving to forbid any subsidization policy. However, even in the absence of export subsidies, the public finance aspect of regulation is an important one, especially when considering regulated industries, as opposed to trade policies. Although state aid might be forbidden, the profits of national firms are valuable to governments. The first reason is that incumbent providers are often vertically and/or horizontally integrated with firms operating in non competitive segments of the market. 9

For more details about the strategic trade policy literature, see also Brander (1997).

6

In this context, the operating profit of the regulated provider might help satisfying the participation constraint in the non competitive segments (e.g. network activity and/or universal service obligations). A second reason is that profits of the national champions are more easily taxed than the profits of foreign firms and can help to finance public policies. The model in the present paper, although not capturing all the complexity of actual regulated industries, is a reduced form description of a situation in which the profit of the national firm are valuable because they are more easily taxed away than foreign ones. Indeed, taxation by regulation through monopoly Ramsey pricing has constituted the traditional way of raising funds in order to cover fixed investment or to cross-subsidize consumption of less favored groups of consumers. As Armstrong and Sappington (2005) notice, competition can “complicate the regulatory policy undermining preferred tax structures”. Similarly, Laffont and Tirole (2000), discussing procompetitive reforms in telecommunications, argue that competition, limiting the scope for cross subsidization and taxation by regulation, may induce an increase in the total transfers paid to the industries. As we will see, including public finance concerns, market integration is shown to have complex welfare implications. The present paper also relates to the work of Calzolari (2004) Calzolari and Scarpa (2007). The former looks at the interaction between the policies of different regulators. In this model, there is only one multinational firm operating in two different countries. The paper deals with the capacity of multinational firm to benefits from lack of coordination of the regulators of different countries. In contrast, we look at the interaction between regulators of different national firms which compete in a common market. Calzolari and Scarpa (2007) considers the optimal regulation of a firm which is a monopoly at home but competes abroad with a foreign firm. It is a model of regulation with transfers, but public funds are not costly. For this reason, if the marginal costs are constant (no externality of the foreign production of the regulated firm on production for the internal market), the pricing rule in the regulated market does not change with competition. The regulatory policy is affected only if there are economies (or diseconomies) of scale. Interestingly, the model shows that allowing a private firm to operate in a foreign market increases the price distortion related to asymmetric information. As the firm also operated in the foreign market, it can earn an additional rent on foreign activities. This model does not consider the case in which the regulator has to deal with entry of a foreign operator in the home market. Yet economic integration is a process of reciprocal opening of the market to foreign competitors. Adding this aspect to the picture, we give different insights on the impact of market integration and in particular on the behavior of the information rent. 7

1.2

Plan of the paper

The paper will proceed as follows. In Section 2 the basic model is presented. Section 3 analyzes the case of complete information: it characterizes the equilibrium of the model and the impact of market integration on welfare. Section 4 considers the case of asymmetric information. Section 5 presents the global maximizing solution, which would be imposed by a welfare maximizing supranational regulator. It also shows how this globally efficient solution can be obtained as a decentralized cooperative solution through Nash-Bargaining between regulators. Section 6 concludes.

2

The model

There are two countries, i ∈ {1, 2}. The demand in each country is given by: Qi = ϕi (d − p) Where Qi is the home demand in country i. In a closed economy, Qi corresponds to qi , the quantity produced by the national monopoly of country i. Moreover, ϕi the size of country i and p the market price. Then, inverse demand in country i writes: p(Qi ) = d −

Qi ϕi

In our specification, ϕi can be interpreted as the size of the country in terms of population and/or the level of development (which affects the level of demand). The national firm is regulated following and incentive contract, determining the quantity10 and a regulatory instrument (tax or transfer). In the integrated markets, each of the two regulators maximizes the home welfare, given by the sum of national consumer surplus and the profit of the national firm. Regulator of country i is not allowed to contract with firm j (asymmetric regulation).

2.1

Closed economy

As a benchmark, we consider the closed economy case (Qi = qi ). We use a standard monopoly regulation framework ` a la Baron and Myerson (1982). The regulator maximizes the expected 10

It could be argued that it is much more common to regulate prices than quantities. However, quantity regulation helps us to avoid Bertrand-type paradoxes. Indeed, under price regulation the results would be qualitatively similar when considering closely substitute goods (e.g. defining consumer surplus S(Q) = d(q1 + q2 ) − 21 q12 − 21 q22 − cq1 q2 , with c sufficiently small). In particular, the welfare analysis is qualitatively the same as well as the impact of market integration on information rents.

8

welfare under close economy, WiC , subject to the participation and incentive constraints of the firm. Welfare is given by:

WiC = S(qi ) − p(qi ) qi + (1 + λi ) ti + Πi

(1)

S(qi ) is gross consumer surplus. Πi is the profit of the national producer (net of the fixed costs, which we consider as sunk). Finally, ti the regulatory instrument (tax or transfer). Πi = (p(qi ) − θi )qi − ti + Ui θi is the constant marginal cost of firm i.11 . We can assume that firms also sustain a fixed cost K, which measures the economies of scale in the industry. This cost is considered as sunk and it does not affect the participation decision of firms nor the chosen policies. Ui is the information rent left to the firm under the optimal contract. Under asymmetric information, regulators do not observe the cost of the regulated firm and they leave some rent to the firm in order to maintain incentives. The participation constraint takes the form:

Πi ≥ 0

(2)

We consider a direct revelation mechanism in which firm i reports its cost and Regulator i offers a menu of contracts {qi (θi ), t(θi )} to firm i (the revelation principle assures this is without loss of generality). The regulator maximizes expected welfare under participation and incentive compatibility constraints of the firm. The solution of this problem is standard (Baron and Myerson, 1982; Laffont and Martimort, 2002). To illustrate the characteristics of the optimal solution, we denot the virtual cost of firm i as: θiv = θi + γi Where γi is the distortion related to asymmetric information. Assumption 1 The marginal cost θi follows a cumulative distribution function F (θi ), i ∈ {1, 2}, on the support [θ, θ]. The distribution is known by the regulator. The rent of the regulated firm can be written: 11

The underlying assumption is that firms have a constant marginal cost θi .In our linear specification, expanding or reducing production in response to market integration has no impact on the unit cost. The effects of different specifications of the marginal are explored in Auriol and Biancini (2008)

9

UiC

=

Z

θ

qi (θ)dθ

(3)

θi

Under Assumption 1 and taking into account (3), at the optimal contract γi takes the following form: C

γi = γ =

   0,  

under complete information;

λi F (θi ) 1+λi f (θi ) ,

under asymmetric information.

We also make the following standard assumption, which ensures that the solution is monotone in θi :

Assumption 2 (Monotone hazard rate property) The hazard rate

F (θi ) f (θi )

is non decreasing in θi .

The optimal price and quantity are a function of the virtual cost θiv . qiC =

ϕi (d − θiv )(1 + λi ) 1 + 2λi

p(qiC ) = d −

(4)

qi λi d + θiv (1 + λi ) = ϕi 1 + 2λi

Assumption 3 Production is always socially desirable: v

d > θi

When public funds are not costly (i.e. λ = 0), the regulator maximizes consumer gross surplus net of production cost. The result is marginal cost pricing at the virtual cost. When λi > 0 the price is raised above the cost with a rule which is proportional to the elasticity of demand εi : p(qiC ) − θiv λi 1 = C 1 + λi εi p(qi ) and εi =

(1+λi )θiv +λi d (d−θiv )(1+λi ) .

This corresponds to a Ramsey-type tariff. The regulated monopoly price

is set above marginal cost. This means that the equilibrium t is positive (lump sum tax on profits).

10

2.2

Common market

We now suppose that the market is open. For simplicity, we take a perfectly integrated market in which demand is given by: P =d−

Q ϕ1 + ϕ2

Where Q = q1 +q2 is the total quantity produced in the common market. The assumption of perfect integration turns out not to be crucial for the results, which are robust when considering transportation costs or other forms of persisting segmentation12 . Each of the two regulators maximizes home welfare. Regulator i maximizes the surplus of national consumers plus the profits of the national firm. Welfare of country i is thus given by: WiO = S(Qi ) − P (Q)Qi + (P (Q) − θi )qi + Πi + (1 + λi )ti

where Qi =

ϕi ϕi +ϕj Q

(5)

2

Q i i and S(Qi ) = d ϕiϕ+ϕ Q − 12 ϕiϕ+ϕ . Regulator i maximizes the expected j j ϕi +ϕj

welfare of country i. The participation constraint takes the form:

Πi = (P (Q) − θi ) qi − ti + Ui ≥ 0

(6)

At the optimal solution (6) will be binding. Solving this condition with respect to ti and replacing the result in (5) we obtain:

WiO = S(Qi ) − P (Q)Qi + (1 + λ)(P (Q) − θi )qi − λUi

(7)

Equation (7) makes the role of the parameter λ apparent. The regulator puts a weight 1 + λ on the national firm’s gross operating profit πi = (P (Q) − θi )qi . On the contrary, the information rent Ui is socially wasteful: the rent seeking behavior of the regulated firm is costly to society. The model captures in a reduced form the idea that the operating profit of the national leader is socially valuable, because it increases the total taxes levied from the industry (and thus can help to cross subsidize fixed investment or other public policies). On the other hand, rent seeking introduces a distortion which is socially wasteful. Without loss of generality, we adopt the following notation: Assumption 4 ϕ1 + ϕ2 = 2, ϕ1 = x, ϕ2 = 2 − x, 0 ≤ x ≤ 2. 12

For a similar model including transportation costs, see Auriol and Biancini (2008). A model with segmented markets and iceberg transportation costs or fixed entry costs also gives qualitatively similar results.

11

In the symmetric case x = 1. In the asymmetric case, x > 1 means that Country 1 is the larger one.

3

Common market under complete information

In this section we consider the equilibrium in the case of complete information (i.e. Ui = 0, γi ≡ 0, θiv = θi ). Regulator i maximizes national welfare under the participation constraint of the firm Πi ≥ 0. We start studying the symmetric case (i.e. λ1 = λ2 = λ and x = 1), which gives the main insights. In Section 3.2, we assess the impact of asymmetries (heterogeneity between countries).

3.1

The symmetric case

We derive the the optimal quantities and price in the symmetric case. By convention, we denote ∆ the cost difference between producer 2 and producer 1, ∆ = θ2 − θ1 . If ∆ > 0, country 1 has the more efficient technology. When |∆| ≤ qiO =

2(d−min{θ1 ,θ2 }) , 3+2λ

both firms produce and:

2(1 + λ)(d − θi + ∆(1 + 2λ)) 2 + 3λ

(8)

Then, for any level of ∆ and λ, the lowest cost firm has the largest market share. When |∆| >

2(d−min{θ1 ,θ2 }) 3+2λ

shut down of the less efficient firm occurs. In this case, the market is

covered by the more efficient firm, producing a quantity equal to:

QO =

4(1 + λ)(d − min{θ1 , θ2 }) 3 + 4λ

(9)

Using equations (4), (8) and (9) we can compare quantities under closed and open economy. When λ = 0, the quantity produced by firm i is reduced with respect to closed economy whenever the foreign firm is more efficient (θj < θi ). In this case, the former monopoly leaves some space to the more efficient competitor and consumers enjoy lower prices. If λ > 0, the business stealing effect is costly to society. Regulator i reacts reducing less often the quantity produced by the regulated firm. In particular, the quantity produced by firm i decreases with respect to a closed economy if and only if the competitor is strictly more efficient than the national firm: θj − θi ≤ −

λ(d − θi )(1 + λ) < 0, ∀ λ > 0 1 + 2λ

12

The business stealing effect induces the regulator to expand the quantity produced when confronted with a more efficient foreign competitor. Similarly, the behavior of the price is closely related to the value of the cost of public funds. At the interior solution, the price takes the form: P (Q) =

2 dλ + 2(1 + λ) θ1 +θ 2 2 + 3λ

When λ = 0 the price is equal to the average marginal cost. Then, compared to the closed economy case, the price is higher for the low cost country and smaller for the high cost one. When λ > 0 the price may decrease even if the competitor is less efficient than the domestic firm. In particular, the price decreases whenever: θj − θi < Then, for |∆| <

λ(d−min{θ1 ,θ2 }) 3+2λ

λ(d − θi ) 1 + 2λ

the price decreases in both countries. In this case, consumers

would oppose market integration, while the national operating profit increases. On the contrary, in the less efficient country consumer would enjoy a lower price, but the operating profit (and thus the lump sum tax) decreases. Market integration has distributive effects and may generate winners and loser in both countries. For this reason consumers may oppose market integration even in countries in which total welfare increases with integration. Conversely, a welfare maximizer regulator would promote market integration in cases in which consumers loose from it. The net welfare effect of market integration is more difficult to asses. When a firm is relatively efficient, it gains form market opening, due to export profits. On the other hand, if λ is positive and the difference in marginal cost is small, the negative business stealing effect outweighs the efficiency gains. Proposition 1 Under complete information, for λ = 0, market integration increases welfare in both countries. For any λ strictly positive, market integration increases welfare in both countries if and only if the difference in the marginal costs is large enough.

Proof: See Appendix 1. When λ, the regulators can expand the production of the national firm at no cost (the lump sum tax/subsidy to the firm is a pure transfer). Then, a foreign competitor would expand its 13

market share with respect to closed economy if and only if it brings efficiency gains. However, when λ > 0 business stealing becomes more costly to society. When expanding the quantity produced by the national firm, the lump sum tax is decreased, which has a negative impact on total welfare. Figure 1 illustrates the welfare gains of country 1 for λ = 0 and λ > 0 respectively. For λ = 0, the welfare gains are non negative and symmetric with respect to ∆. The reduction in the domestic operating profits is compensate by the increase in consumer surplus. For λ > 0 the welfare gains shift downwards and to the left. As a result, the intercept (corresponding to ∆ = 0) is negative. Since ∆ = θ2 − θ1 , the welfare gains of country 2 are symmetric with respect to the vertical axis. Then, if θ1 = θ2 both countries loose from integration. Moreover, the welfare gains of the two countries are asymmetric. For the most efficient one the gains are strictly increasing in |∆|. For the less efficient they have U-shape. The welfare gains are first decreasing and then increasing in |∆|. Eventually, for |∆| large enough, the welfare gains are positive in both countries. Figure 1: Welfare Gains: WiO − WiC , Complete Information.

Countries with large cost differences (i.e. |∆| >> 0) should be in favor of market integration, which is mutually beneficial. One possible example is the integration of electricity markets between France (low cost region) and neighbor countries (Italy, Spain). The production cost is very different between countries and both can benefit from market integration.13 Similarly, high 13

For the Nord Pool the situation is apparently different, since there is not a big difference in generation costs, at least on average. Nevertheless, as Ward, Allen and Davis (2002) notice, the success of the Pool is strictly related to the complementarity of fuel sources: the significant hydro capacity of Norway (100 % ) and Sweden (50 %) can, in wet years, provide cheap electricity beneficial to other markets; the significant thermal capacity of

14

gains from trade are expected in the South and East African Power Pools. High cost countries like Angola and Burundi are expected to benefit from lower generation costs (and prices) and low cost ones such as Ethiopia and Democratic Republic of Congo expect high export revenues. Recent estimates shows that these gains might be very relevant, varying from 2% up to 9% GDP. The same line of reasoning applies to the other international experiences mentioned in Section 1.

3.1.1

Removing the assumption of asymmetric regulation

The potential welfare losses are related to the market share rivalry which pushes regulators to induce an increase in the regulated quantity in order to avoid business stealing from the foreign rival. The welfare reducing impact of market share rivalry and business stealing plays a similar role in the “trade and competition” literature (see discussion in Section 1.1). However, there are important differences. In the case of export subsidies, the prisoner’s dilemma related to uncoordinated policies can be avoided committing to “laissez-faire”. Forbidding trade policies can increase welfare solving the prisoner’s dilemma arising from the market share rivalry. However, in the case of regulation policies, there is no obvious policy which can avoid the welfare losses. Indeed, if a “laissez-faire” regime is introduced (i.e. forbid to asymmetrically regulate the national firm, to avoid the distortion related market share rivalry), the situation would not be improved as long as the fiscal asymmetry persist (i.e. national profits can be taxed away more easily and thus countries put a weight on national operating profit). To see this, let us assume that the firms are completely deregulated and they produced their unregulated Cournot quantities in the common market. These are given by:

qi =

1 q1 + q2 (d − ) + θj − θi 2 2

(10)

Replacing quantities (10) in the welfare function (5), it can be easily verified that the results in Proposition 1 are preserved.14 Full deregulation does not solve the problem because the fiscal effect of business stealing (reducing the national operating profit and thus tax revenue) remains. Another possibility would be to extend the same form of regulation to the foreign firm. This could be difficult for practical reasons. Sometimes regulators are too weak to discipline large foreign multinationals (especially in less developed countries). In other cases, foreign providers Denmark (85 %) and Finland (55 %) can provide “dry-year” reserve for the hydro countries. 14 The proof is obtained using the same steps of the proof of Proposition 1 (see Appendix 1).

15

enter in deregulated segments of the market or in separated industries producing substitute goods: extending regulation would require to put in place new regulated markets, with possibly high costs and additional distortions.15 More importantly, symmetric regulation would also encounter difficulties related to the conflict of interest of national regulators. As explained above, for |∆| small each regulator would like to reduce the quantity produced by the foreign firm in order to reduce the business stealing effect. However, this kind of intervention (limiting the scale of production or the market share) is generally allowed only in order to avoid abuses of market power of dominant firms, typically the national champions (through divestiture, price regulation, etc.) and not toward entrants (foreign or national). Policies aimed to limit the market share of foreign competitors (or increase their prices) would in practice constitute an unlawful barrier to trade in many jurisdictions (e.g. in the EU). In our context, the possible role of supranational cooperation takes the form of policy coordination, which is the object of Section 5.

3.2

Asymmetries between countries

We now turn to case of asymmetric countries. We first consider the impact of the size of the country on the gains from trade. For simplicity, we restrict the attention to the interior solution, where both quantities are positive.16 The following Result holds. Result 1 Let λ1 = λ2 = λ and consider an increase in x. Then ∂ (W1O − W1C ) ∂ (W2O − W2C ) < 0 and >0 ∂x ∂x The welfare gains are decreasing in the size of the country. The smaller country gains more (or loses less) from market integration. Proof: see Appendix 2. Gains from trade are larger for the country with the smallest internal market. When the market is integrated, firms have access to foreign demand. The larger is the size of the foreign market, as compared to the domestic one, the higher are the potential gains from trade. This result was also present in Combes, Caillaud, and Jullien (1997), which concentrate on the case λ = 0. We 15

Moreover, even when extending regulation to the competitor is possible at no exogenous cost, under asymmetric information incomplete regulation could be preferred to full regulation when the cost of public fund is very large, see Biancini (2007). 2(d−θi )(ϕi (2+λi +λj )+2(1+λi )λj ) 16 This holds whenever |∆| ≤ , ϕi ∈ {x, 2 − x}, θi > θj . 2(ϕi −2(1+λi ))(1+λj )

16

have shown that it holds for all levels of λ. However, the impact of the cost of public funds is also relevant. The welfare gains are a non linear function of both λi and λj . It is then more difficult to establish a general result. However, we have the following local Result: Result 2 Let x = 1, ∆ = 0, λ1 = λ2 = λ and consider a local increase in λi . Then, the gains from trade in country i decrease and the gains of country j decrease:

∂ (WiO − WiC ) <0 ∂ λi λi =λj ,∆=0

and

∂ (WjO − WjC ) >0 ∂ λi λi =λj ,∆=0

Moreover, the two derivatives are increasing in ∆. Then, the negative impact on the welfare gains of country i is lover when ∆ increases and the positive impact on j is higher.

Proof: see Appendix 3. The gains from economic integration decrease with the size of the country but decrease (at least locally) with the cost of public funds (i.e. the weight of operating profits in the welfare function). This sheds some light on the impact of the cost of public funds. An increase in λi decreases the gains from trade of country i. This is likely to be relevant especially in developing countries, where taxation by regulation is an important source of revenue because other forms of taxation are not very developed (see Laffont, 2005). When the operating profit of the firm decreases, the loss of revenue decreases the tax levied on the industry (or increases the transfer). For this reason, market opening may be harmful if not accompanied by some policies which can compensate for the loss of revenue of the national firms.

4

The impact of asymmetric information

When firms have private information about production costs, the second best regulation requires the payment of an information rent to the regulated firms. In this context, market integration and competition could constitute an instrument to improve the performance of regulated firms by reducing the burden of the rents paid by society to the regulated monopolies. As this section shows, this is not always the case. The results crucially depend on cost correlation. We assume that the distribution of θi is common knowledge. The realization of the costs, on the contrary, is private information of the firms. For simplicity, we assume that each firm is

17

informed about the effect of the shocks on the two marginal costs.17 The incentive compatibility constraint is modified. If costs are correlated, overstating its cost the domestic firm knows that the competitor is more efficient than it is believed by the regulator. Mimicking a higher cost the national firm is confronted with a higher anticipated response of the competitor. For this reason, the gains from mimicking a high cost are reduced. We assume that the national regulator and the national firm cannot write contracts contingent on the realization of foreign variables. This can depend on the fact that foreign variables are not verifiable.18 For simplicity, we restrict the attention to the symmetric case. From the first order condition of firm i, we obtain:19 1 ∂qj ∂Πi = −(1 + )qj ∂θi 2 ∂θi Where: ∂qj ∂qi ∂qj dθj ∂qj = + ∂θi ∂qi ∂θi ∂θj dθi ∂q

The first term in the RHS is equal to 0 (Cournot model). ∂θjj (i.e. the impact of an increase of dθ the own marginal cost) is negative. Finally. the term j is greater than 0 in case of positive dθi correlation. In this case, the slope of the information rent is reduced and the price is closer to efficiency. The regulator maximizes the expected welfare: WiAI = Eθ1 ,θ2 [S(Qi ) − P (Q) Qi + Πi + (1 + λi ) ti ] subject to the constraints (6) and:

  1 ∂qj ˙ Ui = − 1 + qi 2 ∂θi The information rent can be written:

Ui =

Z

θ

θi



1 ∂qj 1+ 2 ∂θi

17



qi dθi

(11)

This particular assumption does not influence the result. If the cost of the competitor was not known, the reaction function of the firm will depend on the expectation. 18 If foreign variables were verifiable, competition could also open the possibility of yardstick competition. In this case, the regulator could write contracts contingent on the performance of the foreign firm. In this paper, we don’t consider this possibility. With yardstick competition and perfect correlation, the regulator could extract all the information rent (Cr´emer and McLean, 1985; Cr´emer and Mc Lean, 1988). However, in the case of partial cost correlation the slope of the information rent would be our qualitative results preserved (see also Biancini, 2007). 19 For details on the solution technique, see for instance Laffont and Martimort (2002).

18

The first order condition of this problem is:

Eθj |θi [4(d − θiv ) − qi (3 + 4λi ) − qj (1 + 2λi )] = 0

(12)

Where:   λ 1 ∂qj F (θi ) γi = γ = 1+ 1+λ 2 ∂θi f (θi ) O

Thus the reaction function of Regulator i to the policy chosen by Regulator j is: qiAI =

4d − Eθj |θi [θiv − (1 + 2λi )qj ] 3 + 4λ

In order to get explicit results, we consider the two limit cases of uncorrelated costs and perfect correlation. These limit cases approximate the more general ones of high or low correlation between the variable production costs.

4.1

Uncorrelated costs

We start considering the case of uncorrelated marginal costs. More precisely, we assume that costs are distributed over the same support and have equal mean, but they are subject to idiosyncratic shocks. In this case, market opening has no direct impact on the rent extraction ∂q

problem ( ∂θji = 0). The slope of the information rent is unaffected and still depends on the hazard rate of the characteristic of the regulated firm. However, the quantity is not the same under complete and asymmetric information. In particular, the presence of a competitor allows the regulator to reduce the quantity of the regulated firm with a lower impact on the price with respect to monopoly (because the quantity produced by the competitor is not affected by a reduction of the regulated quantity). This may reduce the social burden of the rent, giving a new value to competition. From the system of the two first order condition, we obtain: qiO,AI =

2(1 + λ)(d − θiv + ∆e (1 + 2λ)/(3 + 4λ)) 2 + 3λ

(13)

∆e is the expected difference in virtual costs (∆e = θiv − Eθ θjv , where Eθ θjv is the expected virtual cost). When comparing the quantities and price under market integration with the closed economy benchmark, the results are similar to the ones obtained in Section 3. They hold here for the expected quantities and price. The slope of the rent is not affected by the presence of competition, due to the hypothesis of independence of the marginal costs. Then, the total rent can increase or decrease, depending on the behavior of the optimal quantity. In order to solve analytically for the value of the information rent, we make the following Assumption. 19

Assumption 5 The marginal cost θi is uniformly distributed over [θ, θ]. The following proposition holds: Proposition 2 Under asymmetric information and uncorrelated marginal costs distributed as in Assumption 5, if λ = 0 the rent decreases for all types. If λ > 0 there are two cases: • If

θ−θ d−θ

≤

• If

θ−θ d−θ

>

λ(1+λ)(3+4λ) 1+3λ(2+λ(5+4λ)) , λ(1+λ)(3+4λ) 1+3λ(2+λ(5+4λ)) ,

the rent increases for all types. there exists a threshold θˆ ∈ (θ, θ) such that the rent increases

ˆ Thus, the rent increases for efficient types for types θi ≤ θˆ and decreases for types θi > θ. and decreases for the less efficient ones.

Proof: see Appendix 4. The result in Proposition 2 can be interpreted as follows. In our model, the variance of the shocks on marginal costs is proportional to (θ − θ). Moreover, when d − θ is large, the size of the market is large with respect to cost variability. We thus refer to

θ−θ d−θ

as to the level of ex ante

technological risk. If λ > 0, the rent decreases for all types if and only the ex ante technological risk is low. If not, the regulator distorts the quantities of the less efficient types downwards and thus the rent abandoned to these producers is reduced. Indeed, whenever integration induces the regulator to increase the regulated quantity, the rent is also increased. This is similar to Calzolari and Scarpa (2007) in which a rent is also paid on the “foreign” activity of the national firm (in our case, the quantity sold to foreign consumers). However, contrarily to Calzolari and Scarpa (2007), we show that the increase in the rent is not only caused by the existence of “foreign” activities: it could be the case that the rent paid on the quantity sold in the domestic market also increases, whenever the regulator is willing to expand the produced quantity in order to reduce the market share of the competitor (on this point, see also Biancini, 2007). On the other hand, taking into account that in the case of market opening the foreign producer could also serve national consumers, we show the rent can decrease for the less efficient types (which market share is reduced in the common market). In Figure 2 the information rent under closed and open economy is represented for the case of high and low technological uncertainty respectively. The dotted line represents the rent under closed economy and the solid line the information rent in the case of an integrated market. Computing now the welfare effect of market integration in the case of uncorrelated marginal costs, we find the following result.

20

Figure 2: The Information Rent: Uncorrelated Costs.

Result 3 Under asymmetric information and uncorrelated marginal costs, market integration decreases welfare for θi = θjv . Welfare increases in both countries if and only if |θi − θjv | is large enough. Proof: See Appendix 5. The difference in welfare is negative whenever the difference between the national marginal cost and the foreign virtual cost is small. As in the case of complete information, market integration increases welfare only in the case in which the two technologies are different enough.20 We now turn to the case of correlated shocks.

4.2

Perfectly correlated costs

We consider the opposite limit case of perfect correlation θi = θj = θ. We have: γi = γ

PC

  λ q˙ F (θ) = 1+ 1+λ 2 f (θ)

This is the case in which the rent reducing impact of competition is maximized. Also in this case, we solve for the case of θ distributed as in Assumption 5. The first order condition becomes: 20

One may notice that, in order for the technologies to be different enough, it is necessary that the ex ante technological risk is also large.

21

2d(1 + λ) − 2θ(1 + 2λ) − λ(θ − θ)q˙ + 2λθ − q(2 + 3λ) = 0 This differential equation has a linear solution of the form:

q=

2d(1 + λ) − θ(2 + 3λ) + λθ 2 + 3λ

(14)

so that q˙ = −1. The following result holds. Proposition 3 Under asymmetric information and perfectly correlated marginal costs distributed as in Assumption 5, if λ = 0 the rent decreases for all types. If λ > 0 there are two cases: • If

θ−θ d−θ

≤

• If

θ−θ d−θ

>

λ(1+λ)(3+4λ) 1+3λ(2+λ(5+4λ)) , λ(1+λ)(3+4λ) 1+3λ(2+λ(5+4λ)) ,

the rent decreases for all types. there exists a threshold θ˜ ∈ (θ, θ) such that the rent decreases

˜ for types θi ≤ θ˜ and increases for types θi > θ.

Proof: see Appendix 5. When shocks are perfectly correlated, competition, reducing the slope of the information rent, allows to reduce the quantity distortion required for maintaining incentives. In the case of very inefficient types, for which under monopoly the downward distortion of the second best quantity is large, the rent, which is depends on the produced quantity, can increase with respect to regulated monopoly. Figure 3 shows the difference in the information rent under closed and open economy for the case of large and small ex ante technological uncertainty respectively. The dotted line represents the rent under closed economy and the solid line the information rent in the case of an integrated market. Computing the difference between welfare in a closed economy and under market integration we obtain the result illustrated in the following Proposition. Result 4 Under Assumption 5, for λ = 0 welfare in the two countries is not affected by integration. For λ > 0, there are two cases: • For 0 < λ ≤ 6, there exists a threshold r1 (λ) such that welfare increases for all types θ ∈ [θ, θ] if and only if

θ−θ d−θ

> r1 (λ). 22

Figure 3: The information rent: correlated costs.

• For λ > 6, there exist two thresholds r1 (λ) and r2 (λ), r1 (λ) < r2 (λ), such that welfare increases for all types θ ∈ [θ, θ] if and only if r1 (λ) <

θ−θ d−θ

< r2 (λ).

Proof: see Appendix 5. When the ex ante technological risk is small (i.e.

θ−θ d−θ

< r1 (λ)), the gains associated to rent

reduction are also small. Because costs are perfectly correlated (i.e. they are the same), market integration does not bring any other efficiency gain. For this reason, welfare decreases with integration at least for some types. On the contrary, welfare increases for all types when ex ante technological risk large, except if λ is very high (λ > 6). In the latter case, for very high levels of the ex ante technological risk, the negative business stealing effect may prevail on the rent reducing effect of competition (for

θ−θ d−θ

> r2 (λ)), welfare decreases for some types). It can be

shown that this occurs for very efficient types (i.e. θ ≃ θ). If we interpret the parameter λ as the shadow cost of public funds, only the case 0 < λ < 6 is empirically relevant and we can conclude that market integration increases welfare for all types if and only if the ex ante technological risk is large enough. Very large values of λ would mean that the government is putting an extremely high weight on producer’s profits, as in the case of capture (corrupted regulator). In our analysis we are primarily interested in the discussion of the regulation opportunities and constraints of a benevolent regulator in a common market and we thus mainly concentrate on the case λ < 6. 23

4.3

Concluding remarks on the impact of asymmetric information

Competition is in general thought to put constraints on the regulated firm and to limit its capability of capturing information rents. The analysis above shows that this is not always the case and the direction of the effects depends crucially on the stochastic structure considered. When shocks are uncorrelated, the information rent tends to increase, at least for the more efficient types. On the contrary, with high correlation, the rent generally decreases (though it may increase for very inefficient firms). Both scenarios could be empirically relevant, depending on the industry considered. Variations of the information rent would transmit to the tax extracted (or the transfer paid) to the regulated firm. Moreover, firms may have an impact on policy determination (i.e. lobbying against/pro market opening): for this reason, the distributive effect on rents may become very important in practice. The analysis above shows that, when the costs are not correlated the qualitative results of Section 3 apply. When correlation is high, market integration is more valuable and welfare can increase for all possible realizations of the firm types, depending on the size of the ex ante technological risk.

5

Cooperation between regulators

We are now interested in the possibility of solving the problem arising at the non cooperative equilibrium from the lack of coordination between the the two regulators. The welfare reducing effect of market opening is related to the fact that each regulator does not take into account the impact of its policy on foreign consumer and taxpayers. When considering a process of regional market integration, we can imagine that some cooperation will emerge among regulators or that the creation of supranational regulatory bodies will mediate among the conflicting interests reaching a coordination of the domestic policies. In such a situation, considering only the Nash-Cournot solution is restrictive, since it rules out any possible role for cooperation between institutions. We now consider the case of cooperation between countries. For simplicity, we restrict the attention to the symmetric case and we assume complete information (as explained in Section 4, the welfare analysis would be qualitatively similar for the case of uncorrelated idiosyncratic shocks). As a first step we focus global welfare maximizing solution. A supranational social planner has no national preferences and maximizes the total welfare of the integrated market. This theoretical benchmark describes a process of integration in which the two countries are

24

fully integrated. The operating profits of the two firms enter symmetrically in the welfare function (with weight 1 + λ). For a concrete example one can think to German reunification. The East and West economic systems have been unified under the same government (full unification of regulatory bodies and fiscal system). We consider the solution in the complete information case. As we have seen, the results are qualitatively similar to the case in which costs are subject to idiosyncratic shocks. After characterizing the global optimum, we will move to the decentralized cooperative equilibrium.

5.1

Global welfare maximizing solution

The supranational utilitarian social planner maximizes the sum of welfare of the two countries. In this linear model, the global optimum prescribes shut down of the less efficient firm. Then the optimal solution has the following characteristics: • Only the most efficient firm produces. • If, without loss of generality, we assume that firm 1 is the most efficient firm (θ1 < θ2 ), total quantity is equal to:

q ∗ = 2 q1C (θ1 )

(15)

Where q1C (θ1 ) is the regulated monopoly quantity of country 1 in the case of closed economy. The less efficient firm shuts down and the more efficient covers all the market. This result arises because of the simplifying assumption of no segmentation and constant marginal costs. In the case of decreasing returns to scale, the quantity reduction would be less drastic, but always larger than the one observed at the equilibrium solutio (see Auriol and Biancini, 2008). Naturally, at the globally optimal solution total welfare of the integrated market is larger than the one obtained at the decentralized equilibrium. In fact, at the equilibrium solution both regulators suffer from the fact that they take uncoordinated decision and they could do better sharing the gains form coordination. A supranational welfare maximizing social planner, would also share the surplus generated with production equally among the taxpayers. In the global maximization problem a central benevolent government imposes a policy to the unified market. This captures somehow the kind of integration which occurred in the case 25

of the German reunification. In the process of reunification, two regions with an important productivity gap have been merged under the same government. At the beginning of the reunification process, the physical productivity of East Germany was estimated to be about 1/3 of that in the West (Czarnitzki, 2005). As R¨ oller and Hirchhausen (1996) point out, the particularity of the East German case was that restructuring and privatization were managed by the same institution. State aid has accompanied the restructuring process. The provision of public goods and governmental services were just redistributed within the state sector: eastern Germany railways and telecommunications became part of the western German counterparts. As Siegmund (1997) notices, the budget constraint of the privatization agency “could be made politically soft because mainly Western German taxpayers were paying and will pay for the losses”. This framework does not seem particularly suitable to describe other cases of regional market integration, in which each country has an independent regulator and cooperation has to mediate among the possibly diverging objective. Indeed, the globally optimal allocation of production differs from the equilibrium one because it generates welfare losses in at least one of the countries. For this reason, as we show in the following, allowing for transfers between the formerly separated regions is a way to reduce the inefficiencies generated by market integration.

5.2

Decentralized solution: Nash Bargaining

As seen in Section 3, the global welfare maximizing solution does not emerge from the decentralized decisions of the two regulators. One of the reasons is that profits are not shared between countries. Moreover, in general countries cannot commit ex ante to a certain profile of production. In this case, the globally efficient solution has to rely on ex post bargaining between the two countries. In the real world, there is in general no court to punish deviation from an agreement of this kind.21 Nevertheless, regulators can cooperate in order to reduce the negative impact of each other policies. To consider this possibility, we compute the cooperative equilibrium in which countries bargain on the gains from coordination. The global optimum can be obtained as a cooperative Nash bargaining solution sustained with a side transfer T . By definition, any Nash bargaining solution maximizes with respect to T the Nash product:

(W1∗ − W1N )α (W2∗ − W2N )β

(16)

21 An important exception for the case of the EU is agriculture, where agreement on “quotas” of production are enforced with fines to producers. This is indeed an exception, in other markets it is difficult to imagine the creation of a EU policy with quotas of production for regulated markets.

26

Where:

W1∗ = W1 (q1∗ , q2∗ ) − (1 + λ)T W2∗ = W2 (q1∗ , q2∗ ) + (1 + λ)T The following result holds: Proposition 4 Let T be the optimal transfer from country 1 to country 2. Then, for λ sufficiently large, T is always positive: the most efficient country has to compensate the less efficient one. Proof: see Appendix 6. When λ = 0, the regulator maximizes net consumer welfare. She just cares about finding the cheapest provider (for country 2, the foreign firm). The problem is that the regulator in the high cost country cannot control production of the more efficient firm in order to induce it to internalize the effect of its policy on the foreign consumers. The regulator is willing to pay a transfer in order to induce the foreign firm to increase the production (this transfer is indeed a substitute for the possibility of paying a subsidy to the foreign firm for increasing the quantity). The increase in consumer surplus obtained this way is greater than the one it gets financing losses of an inefficient national firm (as it happens at the Cournot Nash equilibrium, as shown in Section 3). For λ > 0, the public finance effect intervenes. The regulator is not just interested in getting the service at the minimal marginal cost. For λ large enough, the transfer T is always positive. The country with the more efficient technology pays a transfer to the less efficient one in order to compensate for the the damage deriving from restructuring (shutting down the national firm or drastically reducing its market share). This seems to be the more relevant case. Since national production is valuable for the regulator, market restructuring which reallocates production between countries on the basis of efficiency has to be accompanied with some transfer to the countries which suffer from restructuring. As an example, we consider the symmetric Nash bargaining solution in which α = β. The following Result holds: Result 5 At the symmetric Nash bargaining solution, the optimal transfer from country 1 to country 2 is given by: 27

T∗ = τ −

W1N − W2N 2 (1 + λ)

where τ = 12 t1 is half of the operating profit of firm 1 at the global maximizing solution. The transfer T is equal to τ reduced by a term proportional to the difference in the outside option of the two regulators, represented by the Nash-Cournot equilibrium. Substituting for the values of WiN and τ : T∗ =

1 + 2λ 1 + 2λ λ(1 + λ) (d − θ1 )2 ∆2 − ∆(d − θ1 ) + 2(2 + 3λ) 2 + 3λ (1 + 2λ)2

For λ = 0, T ∗ is always negative: 1 θ1 + θ2 T ∗ |λ=0 = − (d − )∆ 2 2 Moreover, the side transfer T ∗ is increasing in λ. In fact, we have: 2 d − θ1 +θ (d − θ1 )2 ∂T ∗ 2 = − ∆ ∂λ (1 + 2λ)3 (2 + 3λ)2

This expression is always positive under Assumption 3. Even in the presence of the transfer T , the total tax revenue of the less efficient country (the international transfer plus the tax to the former national producer) decreases with respect to a closed economy. Relying on a more efficient foreign competitor may reduce prices and increase overall efficiency, but it has a negative impact on national operating profits and thus on taxation by regulation.

5.2.1

Alternative specification: bargaining on the gains from integration

In the Nash bargaining solution computed in the section above, we have assumed bargaining on the gains from coordination, taking as given the fact that markets are perfectly integrated. This describes a situation in which countries have already committed to market integration. An alternative specification could take the closed economy welfare as the non cooperative benchmark. In this case we have WiN = WiC in the Nash bargaining problem described in Equation (16). Here we allow countries to bargain over the gains from integration. This specification fits the case in which countries can stick to statutory monopoly (delaying or opposing the realization of a common market) if the have no gains from integration. The result of Proposition 4 is preserved (see Appendix 6). If we compute the symmetric Nash bargaining solution we have: 28

T′ =

1+λ 1+λ λ(1 + λ) ∆2 − ∆(d − θ1 ) + (d − θ1 )2 4(1 + 2λ) 2(1 + λ (1 + 2λ)2

This transfer T ′ is higher than T ∗ . In fact we have: T′ − T∗ =

2 (d − θ1 +θ 2 )(3 + 5λ)∆ >0 2(2 + λ(7 + 6λ))

Then, if countries can oppose integration, the transfer to the least efficient one is bigger than in the case in which integration is taken as given. This also means that, if countries commit to share equally the full benefit from integration, the compensation to the country with the inefficient technology is larger.

6

Conclusion

The present paper analyzes the interaction between market integration and national regulatory policies. This constitutes a way to look at the issues arising in many international context, in which national regulators have to deal with firms operating on a supranational market. Adopting a two firms and two regulators model, we show that market integration may decrease welfare in one or both countries. Market integration can be welfare reducing because of its impact on the operating profit of the regulated firms (business stealing effect). When production costs are similar, market share rivalry induces both countries to inefficiently expand the quantities. The level of taxes decreases for one or both countries with respect to closed economy (or the total transfers increase) and this loss is not compensated by significant efficiency gains. The net welfare impact of market integration might be negative in both countries. The paper also shows that the impact of supranational competition on the rent seeking behavior of firms can go both ways, depending crucially on the level of cost correlation and the degree of ex ante technological risk. The effect of market integration on the agency problem of the regulator may thus be different in different industries. The negative welfare result arising at the equilibrium is obtain under the assumption of benevolent regulators. The equilibrium policies are optimal from the point of view of national regulators, although globally inefficient. In the last part of the paper, we show that the globally efficient allocation of production can be reached in a decentralized framework allowing for Nash bargaining between the regulators. In this case, side transfers are paid. When the cost of public funds is an issue, the less efficient country has to be compensated for the loss related to shutting 29

down national production. The public finance aspects of regulation is shown to be important to determine the optimal regulatory policy. It may be necessary to accompany market integration with transfers to the losers of the integration process harmed by the process of restructuring triggered by supranational competition.

References M. Armstrong and D. Sappington. Recent developments in the theory of regulation. Handbook of Industrial Organization, 3, 2005. E. Auriol and S. Biancini. Market integration, regional regulation and investment in regulated markets. Mimeo, Toulouse School of Economics, 2008. D. Baron and R. Myerson. Regulating a monopolist with unknown costs. Econometrica, 50(4): 911–930, 1982. S. Biancini. Incomplete regulation, competition and entry in increasing returns to scale industries. Mimeo, Toulouse School of Economics, 2007. G. Biglaiser and C.A. Ma. Regulating a dominant firm, unknown demand and industry structure. RAND Journal of Economics, 26:1–19, 1995. O. Boylaud and G. Nicoletti. Regulation, market structure and performance in telecommunications. OECD Working Papers, 2000. S.L. Brainard and D. Martimort. Strategic trade policy design with asymmetric information and public contracts. Review of Economic Studies, 63(1):81–105, 1996. S.L. Brainard and D. Martimort. Strategic trade policy with incompletely informed policymakers. Journal of International Economics, 42(1):33–65, 1997. J. Brander. Strategic Trade Theory. Handbook of International Economics, 3, 1997. J.A. Brander and B.J. Spencer. International R&D Rivalry and Industrial Strategy. Review of Economic Studies, 50(4):707–22, 1983. B. Caillaud. Regulation, competition and asymmetric information. Journal of Economic Theory, 52:87–100, 1990.

30

G. Calzolari. Incentive regulation of multinational enterprises. International Economic Review, 45(1):257, 2004. G. Calzolari and C. Scarpa. Footloose Monopolies: Regulating a National Champion. CEPR Economic Papers, 2007. D.R. Collie. State aid in the European Union: The prohibition of subsidies in an integrated market. International Journal of Industrial Organization, 18(6):867–884, 2000. P.P. Combes, B. Caillaud, and B. Jullien. Common market with regulated firms. Annales d’Economie et Statistique, 47, 1997. J. Cr´emer and R.P. Mc Lean. Full extraction of the surplus in bayesian and dominant strategy auctions. Econometrica, 56(6):1247–1258, 1988. J. Cr´emer and R.P. McLean. Optimal Selling Strategies under Uncertainty for a Discriminating Monopolist when Demands are Interdependent. Econometrica, 53(2):345–362, 1985. D. Czarnitzki. The extent and evolution of productive deficiency in Eastern Germany. Journal of Productivity Analysis, 24:211–231, 2005. P.D. Domah and M.G. Pollit. The restructuring and privatization of the regional electricity companies in England and Wales: a social cost-benefit analysis. Fiscal Studies, 22(1):107– 146, 2001. D. Flacher and H. Jennequin. Is telecommunications regulation efficient? An international perspective. Telecommunications Policy, 2008. R. Green and D.M. Newbery. Competition in the electricity industry in england and wales. In D.R. Helm and T.J. Jenkinson, editors, Competition in Regulated Industries. Oxford University Press, 1998. T. Hattori and M. Tsutsui. Economic impact of regulatory reforms in the elettricity supply industry: a panel data analysis for oecd countries. Energy Policy, 2003. J.J. Laffont. Regulation and Development. Cambridge University Press, 2005. J.J. Laffont and D. Martimort. The theory of incentives: the principal-agent model. Princeton University Press, 2002. J.J. Laffont and J. Tirole. Competition in telecommunications. MIT Press, 2000. 31

L.H. R¨ oller and C. Hirchhausen. State aids, restructuring and privatization in the New German L¨ ander. WZB Discussion Papers, 1996. U. Siegmund. Was Privatization in Eastern Germany a Special Case? Some Lessons from the Treuhand. William Davidson Institute, 1997. C. Zhang, D. Parker, and Y. Kirkpatrick. Electricity sector reforms in developing countries: an econometric assessment of the effects of privatization, regulation and competition. University of Manchester, 2002.

Appendix 1: Welfare Gains Consider country 1. The welfare gains are computed replacing quantities (4), (8) and (9) in the welfare functions (1) and (5). Developing computations we obtain:

W1O − W1C = Λ1 ∆2 + Λ2 ∆(d − θ1 ) + Λ3 (d − θ1 )2 Where:

Λ1 =

Λ2 =

Λ3 =

   

2(1+λ)2 (3+4λ)2 , (1+λ)2 (3+4λ) , 2(2+3λ)2

   0,  4(1+λ)2    (3+4λ) ,          

2) if ∆ < − 2(d−θ 2+3λ ;

2) if − 2(d−θ 2+3λ ≤ ∆ ≤

2(d−θ1 ) 2+3λ ;

2(d−θ1 ) 2+3λ . 2) if ∆ < − 2(d−θ 1+2λ ; 2 (1+λ) (3+4λ) 2(d−θ1 ) 2) , if if − 2(d−θ 2+3λ ≤ ∆ ≤ 2+3λ ; 2(2+3λ)2 1) 0, if ∆ > 2(d−θ 2+3λ . 2 (5+16λ(1+λ)) 2(d−θ2 ) − (1+λ) 2(1+2λ)(3+4λ)2 , if ∆ < − 2+3λ ; 2 (3+4λ) 2(d−θ1 ) 2) − (1+λ) , if − 2(d−θ 2+3λ ≤ ∆ ≤ 2+3λ ; 2(2+3λ)2 (1+λ)2 (1+4λ) 1) if ∆ > 2(d−θ 2+3λ . 2(1+2λ)(3+4λ) ,

if ∆ >

When firm 2 shuts down, welfare in country 1 always increases with integration. When firm 2 produces, Because Λ1 ≥ 0, the welfare gains are an upward sloping parabola, with a 1) minimum in ∆ = − Λ2 (d−θ . For Λ = 0, the minimum is attained in ∆ = 0, and W1O − W1C is 2Λ1 1) increasing with |∆|. For Λ > 0 the minimum is attained in ∆ = − λ(d−θ 1+λ < 0 and in ∆ = 0 the 2

(3+4λ) welfare gains are equal to − (1+λ) < 0. The U shape then insures the behavior described 2(2+3λ)2

in Proposition 1, taking into account the condition |∆| ≤ d (i.e. that the maximum admissible value of ∆ is obtained when θ2 = d and θ1 = 0).

32

Appendix 2: asymmetric demand We have γ1 = x, γ2 = 2 − x. Consider country 1. Then:     x x q1 + q2 2 q1 + q2 O W1 = d (q1 + q2 ) − − d− q1 − θ1 q1 − λt1 2 2 2 2 t1 =



 q1 + q2 d− − θ1 q1 2

Moreover, at the optimal solution: q1 =

2(d − θ1 )(x + λ) + ∆(2 − x + 2λ) 2 + 3λ

q2 =

2(d − θ2 )(2 − x + λ) − ∆(x + 2λ) 2 + 3λ

Welfare in the case of closed economy can be written: W1C =

x(d − θ1 )2 (1 + λ) 2(1 + 2λ1 )

Substituting the in the welfare functions and developing computations, we obtain:

W1O − W1C = Λi1 ∆2 + Λi2 ∆(d − θ1 ) + Λi3 (d − θ1 )2 Where:

(1 + λ)2 (4(1 + λ) − x) 2(2 + 3λ)2 λ(1 + λ)(4(1 + λ) − x) = (2 + 3λ)2 2 λ (1 + λ)((5 + 9λ)x − 4(1 + 2λ)) = − (2 + 3λ)2

Λi1 = Λi2 Λi3 Then,

∂(W1O − W1C ) ii 2 ii 2 Λ1 ∆ + Λii 2 ∆(d − θi ) + Λ3 (d − θi ) ∂x Where:

(1 + λ)2 2(2 + 3λ)2 λ = − (2 + 3λ)2 λ2 (1 + λ)(5 + 9λ) = − 2(1 + 2λ)(2 + 3λ)2

Λii = − 1 Λii 2 Λii 3

33

∂(W1O −W1C ) ∂x (d−θ1 )2 λ2 to − 2(1+λ)

The derivative where it is equal

1 )λ is a concave function of ∆, with a maximum in ∆ = − (d−θ 1+λ ,

< 0. Then: ∂(W1O − W1C ) <0 ∂x

Conversely, with the same steps and taking into account that the size of country 2 is 2 − x, we obtain: ∂(W2O − W2C ) >0 ∂x .

Appendix 3: asymmetric λ When λi 6= λj , the reaction function of regulator i depends only on λi : 4d(1 + λi ) − (3 + 4λi )qi − (1 + 2λi )qj − 4θi = 0 i 6= j The equilibrium quantities are computed from the system if the two first order condition: qi =

2((d − θ1 )(2 + λi + 3λj + 2λi λj ) + ∆(1 + 2λi )(1 + 2λj )) 4 + 5(λi + λj ) + 6λi λj

Now consider country 1. Substituting the quantities in the welfare functions and developing computations we obtain: ∂(W1O − W1C ) iii 2 iii 2 Λ1 ∆ + Λiii 2 ∆(d − θ1 ) + Λ3 (d − θ1 ) ∂λ1 Where:

2(1 + λ1 )2 (1 + λ2 )2 (3 + 4λ1 ) (4 + 5(λ1 + λ2 ) + 6λ1 λ2 )2 2(1 + λ1 )(1 + λ2 )(λ1 + λ2 + 2λ1 λ2 )(3 + 4λ1 ) = (4 + 5(λ1 + λ2 ) + 6λ1 λ2 )2 3 λ (17 + 4λ2 (7 + λ2 )) + λ21 (13 + 2(7 − 8λ2 ))λ2 − 3λ22 − λ1 λ2 (6 + 19λ2 ) = − 1 2(1 + λ1 )(4 + 5(λ1 + λ2 ) + 6λ1 λ2 )2

= Λiii 1 Λiii 2 Λiii 3

We consider the case ∆ = 0. In this case,

∂(W1O −W1C ) ∂λ1

=

2 ∂(Λiii 3 (d−θ1 ) ) . ∂λ1

We have:

∂(W1O − W1C ) (d − θ1 )2 λ(5 + λ(24 + λ(40 + λ(25 + 3λ)))) |λ1 =λ2 ,∆=0 = − <0 ∂λ1 (1 + 2λ)(2 + 3λ) ∂(W2O − W2C ) (d − θ1 )2 λ(3 + 4λ) |λ1 =λ2 ,∆=0 = >0 ∂λ1 (2 + 3λ) 34

Then, increasing λ1 decreases the gains from trade of Country 1 and increases the gains from Country 2. It is easy to show that, when ∆ locally increases from 0, the derivatives increase. Then, the negative impact of an increase in λ1 is more severe if country 1 is relatively inefficient (the result is local and holds for ∆ small enough. Similarly, the positive impact of an increase in λ1 on country 2 is higher if ∆ increases.

Appendix 4: uncorrelated costs The information rent in the case of closed economy UiO and common market UiC are computed replacing respectively (8) and (4) in (11) and (3). Solving the inequality UiC − UiO ≥ 0 with respect to θi gives: λ(3 + 4λ)(2 d (1 + λ) − θ(1 + 2λ)) + 2θ(1 + 3λ(2 + λ(5 + 4λ))) θi ≥ θˆ = (1 + 2λ)(2 + 3λ)(1 + 4λ) For λ = 0 this is always satisfied. For λ > 0, θˆ is lower than θ if and only if: θ−θ λ(1 + λ)(3 + 4λ) ≥ 1 + 3λ(2 + λ(5 + 4λ)) (d − θ) . Substituting for the values of the quantities and the information rents in the welfare functions (5) and (1) we can compute the value of the difference in welfare along the line θi = θjv . For λ=0 WiO − WiC |θi =θj = −

1 (θi − Eθ v )2 ≤ 0 36

Moreover, when θi 6= θjv , WiO − WiC is strictly positive whenever |θi − θj | is large enough. To show this, let θi = θj + ∆. For λ = 0, we have that W O − W C > 0 if and only if: 1 |∆| > | (θi − Eθ)| 4 For λ > 0, more tedious computations show that the same kind of result holds for θi = θjv and θi = θjv + ∆.

Appendix 5: perfectly correlated costs The information rent UiO is computed using (11) and (14). Solving the inequality UiO − UiC ≥ 0 with respect to θi gives: 35

θi ≥

λ(3 + 4λ)(2d(1 + λ) − θ(1 + 2λ)) + 2θ(1 + 3λ(2 + λ(3 + 5λ))) (1 + 2λ)(2 + 3λ)(1 + 4λ)

This is smaller than θ if and only if: θ−θ λ(1 + λ)2 ≥ λ(3 + 4λ) d−θ Substituting for the values of the quantity q(θ) and the information rent in the welfare function we can compute the value of the difference in welfare W O − W C . This gives:

WiO − WiC =

1 2 d(2(1 + 2λ(2 − λ)2 )) + θλ(3 + 2λ(3 + λ)) θ −θ 4 2(1 + λ)(2 + 3λ) +

θ(4d(1 + λ)2 + 2θλ(3 + 4λ) − θ(2 + λ)(7 + 6λ)) 4(1 + λ)(2 + 3λ)

This is a U shaped function of θ with a minimum in θ =

2d(1+λ)−λθ(5+6λ) . (1+λ)(2+3λ)

(17) Replacing this

value in (17), it can be verified that W O − W C is positive for all types if and only if: θ−θ d−θ θ−θ d−θ

√ √ (1 + λ)( 6(1 + 2λ)(2 + 3λ) − 4 λ(1 + λ)(3 + 2λ)) √ ≥ r1 = λ(6 + λ(2λ(9 + 8λ) − 3)) √ √ (1 + λ)( 6(1 + 2λ)(2 + 3λ) + 4 λ(1 + λ)(3 + 2λ)) √ ≤ r2 = λ(6 + λ(2λ(9 + 8λ) − 3))

By Assumption 3,

θ−θ d−θ

≤

1+λ λ .

(18) (19)

Then, for all λ ≤ 6, the relevant inequality is (18) and welfare

increases for all types if and only if ex ante technological risk is small enough. For λ > 6 welfare also increases for all types when ex ante technological risk is very high (i.e.

θ−θ d−θ

≥ r2 ).

Appendix 6: cooperative Nash Bargaining To prove Proposition 4, it is sufficient to establish that Country 2 is a net loser in the absence of side transfers. If this is the case, each Nash bargaining solution has to insure at least the disagreement payoff, given by the welfare at the non cooperative solution. Without operating profit redistribution across countries, W2 (q1 = 2q1C , q2 = 0) is smaller W2O if and only if λ >

∆ d−θ2 −∆ .

The latter inequality is satisfied if and only if λ is large enough. In this case,

Country 1 has to pay a positive transfer to Country 2 in order to insure the level of welfare obtainable at the non cooperative solution. If the bargaining power of country 2 is increased,

36

the transfer T will be larger.

When considering bargaining on the gains from integration, W2 (q1 = 2q1C , q2 = 0) is smaller than W2C if and only if λ >

θ +θ

∆(d− 1 2 2 ) (d−θ2 )

> 0. Then, the transfer to the less efficient country has

to be positive if λ is large enough.

37