MPIfG Working Paper 97/4, April 1997 

Regulatory Competition and International Cooperation

by Philipp Genschel and Thomas Plümper[1]

Abstract

Recent research has shown that regulatory competition does not necessarily lead to downward pressures on regulation, but may at times also push the level of regulation upwards. Extending David Vogel's "California effect" argument, this paper shows that such upward pressure may not only result directly from the dynamics of the competitive process but also from international cooperation. Evidence from two case studies on international capital market regulation is used to identify the conditions under which cooperation in the shadow of regulatory competition is likely to succeed or fail. The successful multilateral standardisation of banking capital requirements in the BIS is compared to failed attempts to harmonise interest taxation across EC member states.

1. THREE DYNAMICS OF REGULATORY COMPETITION

In recent research it has become common to argue that the globalization of markets leads to downward pressures on regulation. Regulation, according to this view, is costly to the regulatees. Mobile targets of regulation reduce their regulatory cost by moving to jurisdictions with low levels of regulation. Governments in turn offer easy regulation in order to attract larger shares of mobile factors and production processes with the result that regulation spirals down to ever lower levels (Scharpf 1995: 10). A well-publicised early example of such a deregulatory spiral is the American experience with corporate chartering. In the US, corporate charters are granted by the individual states. Yet, since all states are required to recognise each others' charters, they competed for incorporations by offering corporation-friendly chartering requirements. In the course of this competition, the level of protection for shareholders, employees, customers, and the general public has been progressively lowered: a race to the bottom. Since the race was won by Delaware, the notion that regulatory competition develops a deregulatory dynamic has been called the "Delaware effect" (Cary 1974; Vogel 1995: 5).

Yet, as David Vogel has shown in a recent book, economic integration and regulatory competition may also at times push the level of regulation upwards (Vogel 1995). Strict regulation need not be a competitive disadvantage. High standards may even favour domestic producers, because it is often easier for them to comply than for foreign competitors. Hence, industry may occasionally even lobby for higher levels of protection. Once a rich country with a large market has adopted a higher standard, foreign producers are forced to adapt if they do not want to lose market access. Foreign governments may react by raising their own level of regulation, thus starting a regulatory race to the top. Think of California as an example: In recent decades, California has been the pacesetter in environmental regulation, both nationally and globally. Environmentally conscious Californians demanded progressively higher levels of protection regarding car emissions and other fields, and the Californian market was large enough to pull other states along. Hence, the competitive ratcheting upwards of regulatory standards has been called the "California effect" (Vogel 1995: 6).

However, upward pressure on regulation may not only result from competitive dynamics. It can also be caused by international cooperation. Sometimes, competing states manage to stop their competition through collective action. The deregulatory spiral is then countered by a cooperative turnaround. The purpose of this paper is to analyse the factors which determine whether such turnarounds occur and are successful. The analysis is based on two case studies. The first story recounts the successful standardisation of capital adequacy requirements in international banking. It demonstrates that multilateral cooperation among nation-states can stop a deregulatory downward spiral and turn it around into a race to the top. The second case shows that this strategy is not always available. The failure of the EC to counter tax competition by agreeing on a common withholding tax on interest payments gives some clues as to when cooperative turnarounds are likely to fail.

The plan for the paper is as follows: The cases are presented in section 2. Section 3 compares the evidence, and section 4 offers an structural explanation for the observed variance. From this base, section 5 develops two potentially very general propositions about the structural determinants of successful cooperation. Section 6 summarises the results and speculates on their significance for international relations research at large.

2. INTERNATIONAL COOPERATION IN CAPITAL MARKET REGULATION

The banking case and the taxation case were selected for basically two reasons. First, they both concern the regulation of financial markets. Due to the high transborder mobility of financial assets, regulatory arbitrage is particularly strong in these markets, and the peculiar problems of regulatory competition correspondingly easy to observe and analyse. Second, the cooperation success in banking regulation was achieved within the comparatively "weak" institutional setting of the Bank for International Settlements (BIS), whereas the cooperation failure in tax coordination occurred in the institutionally much "stronger" arena of the European Union (EC). Hence, success and failure cannot simply be attributed to institutional factors. This rules out an institutional explanation and thus renders the structural account, suggested by this paper, more plausible. It is differences in the underlying constellation of interests rather than differences in the institutional environment which explain the divergent outcomes of the two cases.

2.1 A Case of Success: Standardising Capital Adequacy Requirements in International Banking

After the second world war, the integration of the world financial markets had reached a historical low. National markets operated largely independent of each other. They were self-contained and inward looking. Only during the 1960s did they begin to open up. Capital controls were lowered, Euromarkets developed in London and elsewhere, and the volume of international financial transactions increased. The collapse of the Bretton Woods system in 1973 and the first oil shock added further momentum to the re-emergence of international finance. The 1980s reinforced the trend through rapid innovations in communication technology, radical innovations in financial services, and political moves towards market liberalisation of the 1980s (Cerny 1993: 63-65).

As the regulatory and technical barriers between national markets eroded, the potential for regulatory arbitrage increased. Borrowers and lenders gained significant leeway to reduce their net regulatory burden by evading from high-regulation jurisdictions to places where standards and regulations were lower. Tough standards turned into a locational disadvantage. Fear of large-scale capital flight decreased the autonomy of national regulators to implement their preferred level of regulation. The risk of a deregulatory spiral increased (Bryant 1987: 139).

The effects of market integration and regulatory competition were ambiguous. The efficiency of the financial system grew, but so did its vulnerability (Bryant 1987: 87-91; Peccioli 1983: 11-12). The inherent risks of financial integration were exposed by a couple of spectacular bank breakdowns - Herstatt, Franklin National Bank, Banco Ambrosiano - which showed that given globalized financial markets the illiquidity of even a single bank could lead to chain reactions, which infect banks and banking systems all over the world. At the same time, the ability of banks to withstand shocks declined, because the intensified international competition forced them to operate on a smaller basis of bank capital. Banks became more fragile while the potential damage of a breakdown of the global financial system increased.

The concern for the solvency of banks reached a crisis level when in 1982 Mexico announced its inability to meet upcoming interest payment obligations. Politicians and bankers feared that the debt crisis might spread to other Latin American borrowers, and endanger a large number of international banks which had only very limited reserves to absorb losses from unpaid debt (Kapstein 1994: 87).

"Governments in the major industrial countries became increasingly uneasy about a general decline in the capital strength of their banks" (Herring/ Litan 1995: 107). Central bankers at the BIS - the "club" of the world's thirteen most important central banks[2] - declared that a "further erosion of capital ratios should be resisted and (...) supervisors should not allow the capital resources of their major banks to deteriorate from the present levels." (Reinicke 1995: 161) However, unilaterally keeping capital ratio requirements stable or even raising them threatened to undermine the international competitiveness of national banks. The defence of capital ratios, therefore, required coordinated, multilateral action. To this end, regulators from the most important financial centres started negotiations on a common standard under the auspices of the BIS.

The negotiations turned out to be difficult. Part of the difficulty was technical. National definitions and measurements of bank capital differed widely among the BIS member states, making agreement on what constitutes capital and how much capital is enough for international banking hard to achieve. The institutional differences between anglo-saxon and continental banking further increased the conflict, suggesting different approaches to capital adequacy. Germany and Switzerland took the most extreme position, arguing that their national regulations were up to the task, and that only the other countries should act to stabilise the banking systems (Reinicke 1995: 162).

The technical problems were eventually resolved when in 1984 a special working group developed a methodology for the cross-country comparison of capital levels. Ironically, however, this methodology added to the political problems and made them more difficult to solve because it showed just how divergent capital-asset ratios were across the BIS member states: While Swiss banks had a capital-asset ratio of about 7 percent, the same ratio in France was below 2 percent. Awareness of these divergencies made the conflict as to how to regulate bank capital requirements even more intractable. The disheartened American Federal Reserve Bank (Fed) reported to Congress that the chances for agreement on an international standard appeared slim (Kapstein 1994: 108), and in early 1986 participants of a BIS meeting openly admitted that the introduction of standard capital requirements seemed unlikely (Reinicke 1995: 166).

The deadlock in Basle was particularly painful for the Fed. In May 1984 one of the largest banks of the country - the Continental Illinois - had collapsed, highlighting the fragility of the US banking system. Congress asked for safer regulations in reaction, but the influential American Bankers Association warned that it would oppose any attempt to raise safety standards unilaterally. Boxed in between both positions, the Fed looked for a regulatory strategy which was less risky economically than a unilateral move, but less politically demanding than collective action under a BIS agreement.

As a first step the Fed approached the Bank of England in July 1986 to suggest a bilateral agreement on a common capital adequacy standard. The regulatory systems of both countries were sufficiently similar to make a common understanding easy. Moreover, given the importance of London and New York as financial centres, a bilateral agreement would be a powerful signal which other countries could not easily ignore (Reinicke 1995: 168).

The British immediately accepted the American plan, and after only six month of negotiation a bilateral Anglo-American agreement was announced in January 1987. Since the negotiations had been conducted in great secrecy, the announcement came as a surprise to the other BIS member states. Most of them reacted with outrage. While the agreement explicitly invited the rest of the BIS to join, its technical details were controversial, and the secret process by which it had been reached was perceived as breaching rules of behaviour that had developed in the BIS over the years (Sebenius 1992: 345). Yet, despite their anger, the other countries found it impossible to simply dismiss the Anglo-American move. The agreement carried the implicit threat that foreign bank operations in US and UK markets could be made contingent on the adoption of the new bilateral standard. No international bank could abandon these markets without losing competitiveness (Kapstein 1989: 341).

In order to regain the initiative, the BIS reinvigorated its own negotiations. The Fed participated but also opened a separate bargaining track with the Japanese in order to bring them into the Anglo-American agreement (Sebenius 1992: 345). Japan was a large factor in international finance - seven of the ten largest banks in the world were Japanese. If it supported the agreement, the rest of the BIS would find it very hard to continue to resist the Anglo-American plans. In a sense, however, Japan was a particularly unlikely coalition partner because its banks supposedly gained a strong competitive edge from comparatively low capital ratios. Nevertheless, Japan was under considerable pressure to acquiesce to American demands. Involved in a calamitous trade row, it did not want to risk a second conflict over banking regulation. The Fed's and the Bank of England's tacit threat that they might make access to their markets contingent on compliance with their capital standard further increased the stakes for the Japanese government.

Eventually, Japan gave in and accepted the Anglo-American proposal with some revisions. When the BIS reconvened in late summer 1987, it was faced with a fait accompli. The bargaining problem had changed dramatically. After the world's three most important financial centres had agreed on a common capital standard, the only choice left to the other member states was to either join the "superpowers" or forego the benefits of a globally harmonised regulation altogether. Even Switzerland and Germany finally yielded to the US-UK-Japanese standard since their banks would find it difficult to escape from complying with it anyway. Agreement in the BIS was suddenly easy. Only six months later, in December 1987, it presented a first draft for a common standard. The final draft was passed in Basle in July 1988.

Today, the so-called Basle accord is widely acknowledged as providing a good starting point for assessing the soundness of international banks (Kapstein 1994: 106). Rating agencies such as Standard & Poor's or Moody's use it as a baseline to assess the solvency of their customers. As a consequence, interbank lending has become more expensive for undercapitalised banks. A low capital ratio is no longer a competitive advantage. Moreover, the Basle accord has pushed up capital requirements not only in BIS member states but also in non-member states. Since the standard was adopted by the rating agencies, a capital base as defined by the accord has become the precondition for international banking. As a result, the definitions of the Basle accord spread California-effect-like to various countries outside the BIS. Australia, Finland, Greece, New Zealand, Norway, Austria, Portugal, Spain and Turkey were among the first to adopt the standard (Pecchioli 1989: 345-378). Others including Hongkong, China and Singapore have followed. Hence, it seems fair to say that the regulation of bank capital is an example case of a deregulatory spiral turned around into a race to the top. The following case study will show that the success of cooperative turnarounds cannot be taken for granted.

2.2 A Case of Failure: Harmonising Withholding Taxes in the European Union

In 1987 the European Commission proposed a directive eliminating all capital controls within the EC. The free movement of capital was generally welcomed as an important step towards a single European capital market. But it also aroused fears of capital tax competition and fiscal degradation. Some economists worried that the single market would transform the European Community "into a single (large) tax haven" (Giovannini/ Hines 1991: 172). France and Italy were particularly concerned that the liberalisation of capital movements would undermine their fiscal position.

In order to calm these fears, an additional paragraph was added to the capital movements directive before it was finally passed in June 1988[3] which instructed the Commission to make proposals how to prevent tax competition, and committed the Council to decide on these proposals by mid-1989 (Helleiner 1994: 158). In keeping with this task, the Commission in February 1989 proposed the introduction of a common 15 percent withholding tax on interest income from savings and bonds. Some member states already levied withholding taxes - Belgium, France, the UK, Ireland, Italy, Portugal, Spain (Walz 1988: 85) - but the tax rates differed widely, and some member states, most notably Luxembourg, did not levy any withholding taxes at all. A common tax rate seemed necessary to prevent free capital movements and tax arbitrage from creating a Delaware-like downward spiral in interest taxation.

The response of member states to the Commission's proposal was mixed. The British government flatly denied the necessity for any tax coordination, and criticised the withholding tax plan as a disguised retreat from the full consequences of free capital movements. Luxembourg complained that the proposed tax was "anti-European" and would drive money away from the EC to financial centres elsewhere. If there was to be any tax coordination at all it should be arranged on a higher plane, such as the OECD, where the most important non-EC financial centres are also involved.

The rest of the member states looked more favourably upon the idea of tax coordination. They agreed that the establishment of a single capital market necessitated common measures against tax competition. But they disagreed with the Commission's approach. The Netherlands, for example, doubted that the harmonisation of withholding taxes was the best way to combat tax evasion. In 1987 the Dutch government had introduced a so-called automatic reporting system which obliges banks to routinely communicate financial information to the tax authorities. Rather than a common withholding tax, the Dutch wanted to see this system extended to the rest of Europe.

As the controversy on the Commission's proposal unfolded, Germany's position turned out to be pivotal. In January 1989 the German government had introduced a national withholding tax on savings and bonds to combat excessive tax fraud in capital income taxation (Köpf/ Walz 1988: 85). The similarity of the new German tax to the Commission's proposal, and the near coincidence of its introduction to the proposal's presentation worked as a powerful reinforcement. In April 1989, however, in an unexpected reversal of policy, the German government decided to abolish the withholding tax, supposedly because it had done serious harm to German capital markets. According to the Bundesbank, the announcement of the tax had led many German investors to buy DM Eurobonds and other tax-free offshore assets instead of lending in the domestic market. This had contributed to a record outflow of funds in 1988. The DM exchange rate came under pressure, domestic bond yields rose, and it became more expensive for German residents, including the government, to borrow in DM than for non-residents. Germany lost attractiveness as a location for financial services (IMF 1990: 64; Schlesinger 1990).

Since Germany joined the ranks of the dissenters in EC negotiations as well, the chances of winning over Luxembourg or the UK were reduced to nothing. In May 1989, during a meeting of EC finance ministers, the plan for a common European withholding tax was quietly laid to rest (Financial Times 22 May 1989: 4). The problem of tax competition did not go away, however, and forced the hand of at least some member state governments. Upon capital liberalisation, a couple of member states lowered their withholding tax rates or extended tax exemptions in order to prevent an outflow of funds. As the Belgian example shows, the fiscal price of these moves was sometimes high. In early 1990, the government slashed the withholding tax rate from 25 percent to 10 percent to stem the drain of Belgian funds to neighbouring withholding-tax-free and secretive Luxembourg. According to official estimates, the net budgetary loss amounted to 0.9 percent of GDP in the first year and to 0.6 percent of GDP even in the fifth year after the rate cut (Defeyt 1992: 65).

Quite understandably, therefore, the Belgian government continued to look for ways to put effective limits on tax competition. In May 1990, it called upon the International Monetary Fund to suggest appropriate rules of behaviour for tax policy, arguing that tax competition was a truly global problem. Yet, the IMF (International Monetary Fund) refused to get involved, and the Belgian government refocused its ambitions on the EC. In 1993 it started a major initiative together with Germany to relaunch the plan for a common European withholding tax. Germany's - somewhat surprising - (re-)conversion to a withholding tax supporter was caused by the Federal Constitutional Court. In 1991, the Court had ruled that the safeguards against tax fraud in the taxation of interest income were unconstitutionally low, leaving the government with the choice of either introducing additional safeguards or abandoning the interest tax base altogether.

Struggling to finance German unification, the government was not prepared to give up the taxation of interest income. Nor did it want to provoke large-scale capital flight. Yet, when in the autumn of 1992 the Federal parliament decided to introduce a withholding tax of 30 percent as of January 1993[4], massive outflows of German funds were already underway and continued through 1993. The main beneficiaries were Luxembourg and the German banks doing business there. The main loser was the Federal treasury. In 1993 it took in DM 11 billion of gross revenues from the new withholding tax instead of DM 24 billion as initially projected (Bundesbank 1994: 49-55). As German money flocked across the border, the German government joined forces with the Belgians to plug the "loophole" Luxembourg. It was agreed that Belgium would make withholding taxes a priority of its EC presidency in the second half of 1993, and that Germany would see the resulting legislation through the Council under its presidency one year later (Frankfurter Allgemeine Zeitung 8 April 1993: 15).

The British government was most outspoken in its criticism of the bilateral initiative. It argued that a common withholding tax would drive business away to New York and Tokyo, and called for a free market in taxation. Luxembourg was equally opposed to the tax, reiterating that the EC was the wrong forum for tax coordination. Common rules and rates should be introduced on the OECD level rather than just among the EC member states (Financial Times 12 July 1994: 3). Criticism also came from other member states which found fault with the technical details of the withholding tax plan. The withholding tax supporters went out of their way to meet this criticism. But in the end they were unable to break the opposition. The re-run of the withholding tax plan failed like its predecessor.

3. DIFFERENCES AND SIMILARITIES

A casual reading of the two cases reveals strong similarities. Both start with the erosion of fences separating national capital markets. Governments feared that increased regulatory arbitrage would create downward pressures on regulation and taxation precipitating a regulatory race to the bottom. International institutions moved in to prevent such a race. In the banking case, the BIS set out to protect sound banking practices by developing multilateral rules for capital adequacy. In the taxation case, the EC attempted to rule out tax competition through the harmonisation of withholding taxes.

The EC and the BIS are very different institutions. But with respect to our two cases, they show important similarities. The number of member states is about equal, 13 in the BIS, 12 (later 15) in the EC. The relationship between the members and non-members of the institutions shows a similar pattern. The BIS includes representatives from the most important but by no means all international financial centres. Countries such as Singapore or Hongkong are missing. Likewise the EC includes the most important but not nearly all jurisdictions competing for European investment. For example, Switzerland and the Channel Islands are not included. Finally, both institutions operate under the same decision rule. The central bankers in the BIS decide by unanimous agreement, and so does the EC council of ministers with respect to matters of taxation.

Unanimous agreement is difficult to reach unless negotiators start from perfectly aligned preferences and perceptions. Unsurprisingly, therefore, both cases passed through an extended episode of conflict, and deadlock. Being blocked by disagreement, BIS representatives in 1986 admitted that multilateral agreement on a common capital adequacy standard was elusive, and the EC decided in 1989 to end discussions on the Commission's plan for withholding tax harmonisation.

Beyond this impasse, however, the two stories diverge. In the banking case, the stalemate in the BIS triggered Anglo-American bilateralism. The extension of the bilateral agreement to include Japan, and the tacit threat to implement it no matter what the rest of the world does, did apparently change the strategic situation for the rest of the BIS member states so much that they gave up their former reservations, and accepted a multilateral standard along Anglo-American-Japanese lines. Thus the stalemate was broken, and the Basle accord did introduce reasonably effective safeguards for international banking. Moreover, once in place, the accord turned out to be so popular that even countries which are not members of the BIS decided to copy and implement it. As of now, about 100 states including all important financial centres nowadays adhere to the rules of the accord, thus bearing witness of a largely successful cooperative turnaround.

Now contrast this with the taxation case. After the negotiations on the withholding tax plan got deadlocked in the Council of Ministers, none of the supporters of the plan made any attempt to push it through by minilateral action. To be sure, Belgium, and Germany relaunched it on the EC level where - quite unsurprisingly - it ran again into the conflicts which had stopped it the first time around. Belgium also tried to initiate maxilateral action on the global level but was rebuffed by the IMF. Even if the IMF had not refused to get involved, the odds are that negotiations in the IMF would have suffered from even more severe haggling problems than the negotiations at the EC level. In the end, multilateral tax coordination was not forthcoming. Tax competition continues unchecked, undermining - as some observers see it - the foundations of European welfare states.

4. A STRUCTURAL EXPLANATION

Why did two political processes which set out in similar fashion lead to opposite outcomes? The reason is structural: Underneath the apparent similarity there lies a fundamental strategic difference which accounts for the variance.

The cases are similar in two respects. First, both deal with a collective action problem. In the banking case, national regulators in the US and elsewhere tried to maintain sound banking standards but found themselves constrained by the risk of foreign competition. In taxation, similarly, EC members such as France or Belgium that wanted to maintain high taxes on interest income were constrained by the risk of being undercut by competing states. Second, in both cases, countries disagreed about how to deal with the collective action problem. As discussions in the BIS and the EC Council of Ministers quickly revealed, differences in domestic institutions and economic posture suggested different approaches to cooperation in banking and interest taxation respectively. In the ensuing negotiations, the heterogeneity of national interest gave rise to conflict and deadlock.

The intersection of collective action problems and problems of interest heterogeneity[5] is difficult to deal with because both problems call for solution strategies which are partly contradictory. In order to solve a problem of collective action it is useful to extend the range of cooperators: as more and more actors join the cooperation, there are less and less actors left who could potentially free ride (Martin 1993: 98-100). In order to resolve problems of interest heterogeneity, by contrast, it is useful to limit the range of cooperators. If cooperation remains confined to a small group of like-minded actors, agreement on a common approach is relatively easy. Thus, interest heterogeneity can be kept low by the exclusion of dissenters. But as the number of cooperators grows, interest heterogeneity will eventually grow as well, and the risk of deadlock and stalemate will increase (Kahler 1993; Martin 1993: 99).

In order to solve both problems at once - the collective action problem and the interest heterogeneity problem - the actors have to find a viable balance between inclusiveness and exclusiveness. They have to form a coalition that is large enough to benefit from cooperation even if everybody else defects, but small enough to keep interest heterogeneity within manageable limits (Genschel/ Plümper 1996; Keohane 1990).

Whether or not such a group can be put together is partly a question of strategic and social skill (Lax/ Sebenius 1991: 163-166). More fundamentally, however, it depends on the difficulty of the underlying problems. If the collective action problem is so severe that only a very large group can benefit from cooperation, and if the degree of interest heterogeneity is so high that only very small groups are able to agree on a solution, it will be comparatively difficult to organise cooperation. If, however, even a fairly small group can gain from cooperation, and if interest heterogeneity is limited to a conflict between a large homogenous majority and a small dissenting faction, then it will be comparatively easy to put together a viable[6] group of cooperators.

4.1 Initiating cooperation

The fundamental difference between the banking and the taxation case, and the reason behind the different outcomes of the two, is that the collective action problem is much more severe in the latter than in the former, and leaves less room for reducing interest heterogeneity through minilateralism. In the banking case, states acted on the assumption that a trilateral coalition of the US, the UK, and Japan was large enough to profit from cooperation all by itself: Even if everybody else should free ride, these three countries would be better off cooperating than in a state of universal non-cooperation. Given this expectation, constructing a minilateral coalition became a worthwhile pursuit. Not that minilateralism could eliminate interest heterogeneity. But it reduced the problem to coaxing Japan into agreement with the Anglo-Saxon states. That was no mean challenge, but it was manageable.

In the taxation case, by contrast, no one believed that, say, Germany, Belgium, France and Italy could gain from coordination merely among themselves. Rather, the common wisdom held that tax coordination had to include at least Luxembourg but preferably also tax havens outside the EC to be profitable for the cooperators. This perception of the collective action problem left no space to reduce interest heterogeneity through minilateral initiatives. There were even doubts that the EC as a whole would be large enough to constitute a viable coalition for tax coordination.

These differences in perception have a very real base in the different economic logics of the cases. On the face of it, the taxation case and the banking case are both about the regulation of banking operations. Yet the taxation of interest income and the requirement to hold certain capital ratios pertain to very different banking activities. Tax competition determines where capital is fed into the international banking system (deposit taking), while the regulation of capital asset ratios determines how capital is processed within the system (lending). Deposit taking is a fairly simple service which does not require much infrastructure or expertise. It can be relocated quite easily to new and hitherto undeveloped jurisdictions. Hence, even countries which have not been important banking centres in the past can potentially be turned into tax havens. Lending, by contrast, is a very complex service presupposing a rich supply of liquidity, expertise, and technical infrastructure. Thus, being spatially close has positive externalities. This is why banks tend to concentrate their lending activities in a few large international financial centres, and why, in contrast to deposit taking, it is very difficult to shift lending to greenfield sites abroad.

This difference in the underlying economic logic explains why minilateralism was a viable strategy in banking but not in taxation. In the banking case, regulatory competition was largely limited to competition among established financial centres, and not all competing centres were perfect substitutes. Regulators correctly anticipated that no bank could afford to exit the three most important locations - New York, London, Tokyo - at once. Hence, cooperation among the regulators of these three markets was sufficient to ensure that the gains from cooperation in terms of increased stability outweighed the possible loss in terms of decreased market share. In the taxation case, by contrast, EC member states were near perfect substitutes for deposit taking. Hence, minilateral tax coordination was insufficient to block the exit routes of tax evaders. In fact, even if all EC states agreed to cooperate, tax evaders could still deposit their savings in European but non-EC jurisdictions such as Switzerland or the Channel Islands at little additional cost. Hence, it was doubtful that even an EC-wide tax harmonisation would be sufficient to protect interest tax revenues from being eroded by tax competition.

The short of it is that the banking case constituted a less difficult collective action problem because fewer actors were required to put together a viable nucleus of cooperation. Hence it was possible to initiate cooperation through minilateral action, and to circumvent problems of interest heterogeneity by excluding potential dissenters from the minilateral core.

4.2 Spreading Cooperation

Note, moreover, that the banking case is remarkable in another respect as well. Not only was cooperation initiated but it also spread from its small minilateral beginning to eventually include virtually all relevant countries. This California-effect-like snowballing of cooperation had not initially been anticipated. Originally, the Basle accord had been conceived as a safety standard. Since safety standards increase production costs without adding value, it was not expected that other governments outside BIS would voluntarily copy it. As it turned out, however, this expectation was wrong. The reason was that the standard was adopted by the banks themselves as a means to simplify transactions in the interbank market.

Considerable transaction costs in the interbank market stem from the problem of assessing the "soundness" of potential borrowers - a fuzzy concept that is hard to observe. The capital adequacy rules of the Basle accord improved the intersubjectivity and observability of soundness by offering a standardised measure for market risk. As credit rating agencies started to apply this measure, banks were under pressure to comply with it even if this was not a formal requirement in their home country. Being able to signal compliance quickly and credibly turned into a competitive advantage. Since this signalling was more easy to do if compliance was certified by national authorities, many non-BIS governments adopted the rules of the accord to help domestic banks prove their solvency.

Obviously there is no equivalent to this pattern of spontaneous diffusion of cooperation in the tax case: since no nucleus of cooperation emerged in the first place, it was impossible for the cooperation to spread. But, just for the sake of argument, assume that a minilateral coalition for tax coordination had formed. Would cooperation have spilled out to further countries?

We argue that this is highly unlikely. Spontaneous diffusion implies that cooperation is self-stimulating, i.e. that the incentive to join grows as the number of cooperators increases. In the tax case, however, cooperation is likely to be self-limiting. That means that as the number of cooperators grows in a hypothetical instance of tax coordination, the incentive for outsiders to join will decrease. The reasoning is simple: Being a tax haven in a world where every other state is also a tax haven is not very profitable, but being the sole tax haven in an otherwise tax-haven-free world is potentially very profitable. Hence, the contagion process will eventually stop as more and more states join the coalition for tax coordination. We may conclude that even if a minilateral nucleus of cooperation had emerged in the tax case, it would not have grown into a full-fledged cooperative turnaround.

In way of summary, we may say that tax coordination was more demanding than concerted banking regulation for two reasons: it was more difficult to initiate, and it would not have spread spontaneously.

5. THE LOGIC OF COOPERATIVE TURNAROUNDS

The foregoing analysis suggests that given a certain level of interest heterogeneity the feasibility of a cooperative turnaround depends on two structural factors: the size of the smallest possible coalition that can gain from cooperation all by itself, and the external effects of cooperation on non-cooperators. A cooperative turnaround is relatively easy if, as in the banking case, the minimum-sized coalition is fairly small (opponents can be excluded), and if the cooperation encourages outsiders to also switch to a cooperative strategy. A cooperative turnaround is relatively difficult if, as in the tax case, the minimum-sized coalition is fairly large (opponents have to be included), and cooperation makes non-cooperation more attractive to outsiders.

The logic behind this empirical intuition can be summarised in a Schelling diagram (Schelling 1978: 216-225), a graphical representation of multi-actor games that indicates how the utility payoff of a given actor varies as a function of his or her own behaviour and the behaviour of other actors. For simplicity, the range of behavioural alternatives is restricted to the binary choice between cooperation and defection. The behaviour of other actors is marked on the horizontal axis: at point 0 no other actor cooperates; at point n the whole population of actors cooperates. The C and D curves[7] indicate the utility payoff of cooperation and defection to the individual actor. Now compare figures 1 and 2 representing the fundamental structural aspects of the banking and the tax case respectively.

Figure 1: The Banking Case

Figure 2: The Tax Case

In both figures, the C and D curves are upward sloping. This means that in either case the utility from cooperation and defection increases as the number of cooperators increases. Hence, all states prefer all other states to cooperate. However, each state individually prefers to defect as long as the D curve is above the C curve. Non-cooperation is the dominant strategy, and universal non-cooperation is the collective equilibrium outcome.

The main usefulness of the Schelling diagrams is in describing the two fundamental differences between the collective action problem in the banking case and the taxation case. First, the discrepancy in the position of point k illustrates why actors found it easier to initiate cooperation in the former than in the latter. K indicates the threshold for self-sustaining cooperation. If there are at least k cooperators, each of them will be at least as well off as in a state of universal non-cooperation.[8] That k is small in figure 1 means that it takes only a fairly small number of states to put together the minimum-sized coalition that can gain from cooperation all by itself. States in favour of cooperation can go ahead without soliciting outside support. The likelihood that cooperation is prevented by interest heterogeneity is correspondingly low. Conversely, that k is large in figure 2 implies that it takes a fairly large number of states to make cooperation self-sustaining.[9] The proponents of cooperation depend on additional allies in order to form a viable coalition. The leeway to avoid conflict of interest through the exclusion of opponents shrinks. Therefore, the emergence of cooperation is less likely.

Second, the different slopes of the C and D curves illustrate why cooperation was self-stimulating in the banking case but self-limiting in the tax case. In figure 1, representing the former, the C curve is steeper than the D curve. This implies that the gains from cooperation grow faster than the gains from defection as the number of cooperators increases. Every state switching from defection to cooperation reduces the advantage of the remaining defectors. Hence it takes less and less sidepayments or coercion to convince the remaining outsiders to join the club. Eventually, as the C curve crosses the D curve, the outsiders join voluntarily. To the right of this intersection, cooperation is the dominant strategy. The Prisoners' Dilemma mutates into either an Assurance or a Harmony game. The number of cooperators snowballs until virtually all relevant states cooperate.

This is in sharp contrast to the taxation case as represented by figure 2. Here the C and D curve are diverging. The gains from defection are rising faster than the gains from cooperation as the number of cooperators increases. With every additional actor who joins the cooperation it becomes less attractive for the remaining outsiders to also join. Cooperation is self-limiting. Even if enough states agree to constitute a k-group, the odds are that the cooperation will not spread to very many other states.

6. A TYPOLOGY OF REGULATORY COMPETITION

Political economy folklore assumes that regulatory competition leads to downward pressures on regulation. However, as David Vogel has demonstrated, regulatory competition may also at times push the level of regulation upwards. Extending Vogel's argument, this paper shows that such upward pressure may not only result from the dynamics of the competitive process but also from negotiated agreements on international cooperation.

The analysis of the two case studies suggests that the feasibility of such a cooperative turnaround depends on two structural factors: the size of the smallest possible coalition that can gain from cooperation all by itself (k-group), and the external effect of cooperation on non-cooperators (diverging or converging C and D curves). A cooperative turnaround is relatively easy if the k-group is small (relative to the largest group of like-minded actors favouring cooperation) and if cooperation reduces the temptation to defect; it is difficult if the k-group is large and if cooperation is self-limiting.

Figure 3 combines the two factors k-group size and external effect of cooperation to construct a 2x3 matrix. This matrix helps not only to distinguish six different scenarios of regulatory competition with respect to their amenability to cooperation, but it also shows that the likelihood of a California or a Delaware effect depends on the same set of structural factors which also determine the likelihood of a cooperative turnaround. The three dynamics of regulatory competition all flow from the same structural source.

If the structure of regulatory competition is such that a single country can profit from raising its regulation above the level of competing states (k group < 1), there is no competitive pressure to deregulate and, hence, no collective action problem. Some countries will regulate upwards unilaterally, and if the C and D curves are converging, others will unilaterally switch to high regulation as well (complete California effect). If, however, the curves are diverging, the odds are that the contagion process will be self-limiting. Some countries will refuse to switch. The California effect remains incomplete.

The situation is different if k is larger than one (k group > 1). Then no single country can resist competitive deregulation unilaterally. Downward pressures will be more intense, and it will require cooperation to counter them. States which want to regulate upward (or do not want to deregulate) have to team up and act in concert. This is comparatively easy to do if k is small because then the pro-cooperation states are often strong enough to initiate cooperation without any outside support. It is more difficult if k is large and the degree of interest heterogeneity is high, so that the pro-cooperation states have to recruit the support of dissenting states in order to put together a viable coalition. But even if a k-group can be put together, there is still the chance that cooperation remains incomplete. If the utility of defection grows faster than the utility of cooperation as the number of cooperators increases, that is if the C and D curves are diverging, there is some probability that the cooperation will not spread to all relevant states.

While the distinctions offered by figure 3 were derived from the analysis of the specific problems of regulatory competition, their potential range of application is much larger and extends to the whole range of cooperation problems in international relations. In recent years, there has been much debate between liberal institutionalists and neorealists about the preconditions of successful international cooperation. The positions of both camps differed widely, but their conception of cooperation was amazingly similar in at least one particular respect. Both treated cooperation as a dichotomous variable, as an all-or-nothing proposition where either all relevant states cooperate or none do (Baldwin 1993). In the real world, however, it is rare to find total cooperation or total non-cooperation. More often there is a mix where some states cooperate while others refuse to follow. Hence, in order to understand real-world cooperation it is not enough to know if cooperation will emerge. One also needs to know how many states will participate. Our typology offers hunches for both kinds of question.

NOTES

1   This paper is accepted for publication by the Journal of European Public Policy. Earlier versions of this article were presented at the conference on
"Problem Solving Capacity of Transnational Governance Systems" held at the Max-Planck-Institut für Gesellschaftsforschung in Cologne, 8-9 November 1996, and the International Studies Association Annual Convention, Toronto, 18-22 March 1997. We would like to thank Fritz W. Scharpf for valuable comments. Discussions with Philip Cerny, Vivek Dehejia, Peter Hall, Virginia Haufler, Adrienne Héritier, John Odell and Günther Schulze have also been extremely helpful.
2   The membership includes the central banks of Belgium, Canada, Denmark, France, Germany, Italy, Japan, Luxembourg, the Netherlands, Sweden, Switzerland, the UK, and the USA.
3   Council directive 88/361/EC of 8 July 1988 for the implementation of Article 67 of the EC Treaty.
4   Technically, the new tax is not a true withholding tax. It is not withheld directly at the source, i.e. by the borrower, but in between borrower and lender by the financial intermediary transmitting the interest payment.
5   Garrett and Weingast have modelled such a confluence of a collective action problem and a problem of interest heterogeneity as a battle of the sexes game nested within a Prisoner's Dilemma (Garrett/ Weingast 1993: 184; see also Genschel/Plümper 1996: 242).
6   This assumes that the remaining dissenters do not have a veto to stop willing cooperators.
7   As a matter of formal consistency, assume that the C curve indicates the utility of actor m while the D curve indicates the utility of actor m+1, where m is the number of cooperating actors.
8   The size of k depends on two fundamental factors, the cost of cooperation for the first cooperator (intersection of C curve and y-axis) and the rate of increase of net benefits for the cooperators with each additional cooperator (slope of the C curve).
9   Indeed, in the extreme case of zero transaction co
sts, perfect capital mobility, and perfect substitutability of jurisdictions with respect to deposit taking, k may be equal to n. Under these circumstances, tax coordination among even a large group of states makes as much sense as trying to fill a bathtub with an unplugged drain.

REFERENCES

Baldwin, David A. (ed.) (1993): Neorealism and Neoliberalism: The Contemporary Debate, Columbia University Press, New York.

Bryant, Ralph (1987): International Financial Intermediation, Brookings, Washington.

Bundesbank (1994): Aufkommen und ökonomische Auswirkungen des steuerlichen Zinsabschlags, in: Deutsche Bundesbank, Monatsbericht Januar, Frankfurt a.M., 45-58.

Cary, William L. (1974): Federalism and Corporate Law: Reflections upon Delaware, in: Yale Law Review 83: 4, 663-705.

Cerny, Philip (1993): The Deregulation and Re-regulation of Financial Markets in a more Open World, in: Cerny, Philip (ed.), Finance and World Politics, Edgar Elgar, Aldershot, 51-85.

Defeyt, Philippe (1992): Tax Structure Harmonisation in 1992: The Belgian Perspective, in: Winckler, Georg (ed.), Tax Harmonisation and Financial Liberalisation in Europe, St. Martin's Press, New York, 57-70.

Garrett, Geoffrey/ Barry R. Weingast (1993): Ideas, Interests, and Institutions: Constructing the European Community's Internal Market, in: Goldstein, Judith/ Robert O. Keohane (eds.), Ideas, and Foreign Policy. Beliefs, Institutions, and Political Change, Cornell University Press, Ithaca, NY, 173-206.

Genschel, Philipp/ Thomas Plümper (1996): Wenn Reden Silber und Handeln Gold ist: Kooperation und Kommunikation in der internationalen Bankenregulierung, in: Zeitschrift für internationale Beziehungen 3:2, 225-253.

Giovannini, Alberto/ James R. Hines Jr.(1991): Capital Flight and Tax Competition: Are there Viable Solutions to Both Problems?, in: Giovannini, Alberto/ Colin Mayer (eds.), European Financial Integration, Cambridge University Press, Cambridge, 172-220.

Helleiner, Eric (1994): States and the Reemergence of Global Finance: From Bretton Woods to the 1990s, Cornell University Press, Ithaca.

Herring, Richard J./ Robert E. Litan (1995): Financial Regulation in the Global Economy. Washington D.C.: Brookings.

IMF (International Monetary Fund), (1990): International Capital Markets. Developments and Prospects. Washington D.C.: IMF.

Kahler, Miles (1993): Multilateralism with Small and Large Numbers, in: Ruggie, John Gerard (ed.), Multilateralism Matters: The Theory and Praxis of an Institutional Form, Columbia University Press, New York, 681-708.

Kapstein, Ethan B. (1989): Resolving the Regulator's Dilemma: International Coordination of Banking Regulation, in: International Organization 43, 323-347.

Kapstein, Ethan B. (1994): Governing the Global Economy, Harvard University Press, Cambridge.

Keohane, Robert O. (1990): Multilateralism: An Agenda for Research, International Journal 45, 731-764.

Lax, David A./ James K. Sebenius, (1991): Thinking Coalitionally: Party Arithmetic, Process Opportunism, and Strategic Sequencing, in: Young, H. Peyton (ed.), Negotiation Analysis, University of Michigan Press, Ann Arbor, 153-193.

Martin, Lisa L. (1993): The Rational State Choice of Multilateralism, in: Ruggie, John Gerard (ed.), Multilateralism Matters: The Theory and Praxis of an Institutional Form, Columbia University Press, New York, 91-121.

Pecchioli, Rinaldo M. (1983): The Internationalisation of Banking: The Policy Issues, OECD, Paris 1983.

Pecchioli, Rinaldo M. (1989): Bankenaufsicht in den OECD-Ländern: Entwicklungen und Probleme, Nomos, Baden-Baden.

Reinicke, Wolfgang H. (1995): Banking, Politics and Global Finance: American Commercial Banks and Regulatory Change, 1980-1990, Edgar Elgar, Aldershot.

Scharpf, Fritz W. (1995): Negative and Positive Integration in the Political Economy of European Welfare States. Jean Monnet Chair Papers 28, Florence: European University Institute.

Schelling, Thomas C. (1978): Micromotives and Macrobehavior, Norton, New York.

Sebenius, James K. (1992): Challenging Conventional Wisdom of International Cooperation: Negotiation Analysis and the Case of Epistemic Communities, in: International Organization 46, 323-366.

Vogel, David (1995): Trading Up: Consumer and Environmental Regulation in a Global Economy, Harvard University Press, Cambridge, Mass.

Walz, Hartmut, (1988): Quellenbesteuerung von Zinseinkünften, Wirtschaftswissenschaftliches Studium 17, 85-87.

Copyright © 1997 Philipp Genschel and Thomas Plümper

No part of this publication may be reproduced or transmitted without permission in writing from the authors.
Jegliche Vervielfältigung und Verbreitung, auch auszugsweise, bedarf der Zustimmung der Autoren.

MPI für Gesellschaftsforschung, Paulstr. 3, 50676 Köln, Germany