Chapter 1. European Banks Unfettered
- Tamim Bayoumi
- Published Date:
- October 2017
The massive expansion of the European banking system between 1985 and 2002 is a crucial, and previously underestimated, aspect of the North Atlantic financial crisis. Over this period European bank assets doubled in relation to the economy. The growth was spearheaded by the emergence of a small number of major banks referred to as “national champions” that were mainly located in the northern core of the Euro area—Germany, France, the Netherlands, and Belgium. The assets of these banks ballooned even as their capital buffers—the reserves they held in case their loans went bad—thinned. The upshot was the emergence of a small number of northern mega-banks that were quickly becoming too big to save even as they became less sound in the face of shocks.
In the subsequent boom over the 2000s these trends took on a life of their own. The mega-banks in the Euro area core continued to expand, providing much of the funding for the housing price bubbles in the United States and financial bubbles in the Euro area periphery. When the bubbles burst the losses whipsawed back onto the core of the Euro area, widening the banking crisis across the North Atlantic. Indeed, every banking system in the Euro area experienced a crisis in 2008 except for Finland.1 The northern European mega-banks were thus major protagonists that helped drive the North Atlantic banking crisis rather than hapless victims, as they are often portrayed.
How the northern European banks were allowed to expand so far, so fast, is a cautionary tale of unanticipated spillovers from financial regulations. Three decisions, directed at differing goals and taken by distinct groups, created the environment for the mega-banks to thrive. In chronological order they were: The 1986 Single European Act, driven by the European Commission; the 1992 Maastricht Treaty, primarily planned by central bankers; and the 1996 market risk amendment to the Basel Accords, under the purview of bank regulators. Together, these three decisions drove an unsustainable expansion of the major European banks that exploited lax regulation and supervision. This business strategy was so successful that by the early 2000s just twelve banks in the Euro area owned one-quarter of all bank assets. Most of these mega-banks were located in northern Europe since this is where the investment banking operations that were central to this strategy were most established. Examples include Deutsche Bank in Germany, BNP Paribas in France, and ING in the Netherlands. Understanding the transformative nature of these changes it is useful to first survey the state of the banking system in the early 1980s.
European Banking in the 1980s
In the early 1980s there was no real concept of an integrated European banking system. Rather, the European Community’s banks operated in a series of fragmented national markets.2 German banks took deposits from Germans in Deutsche marks and lent to Germans, a pattern that was repeated throughout the Community—just 4 percent of bank assets in Germany, Europe’s largest economy, came from foreign institutions. Bond and equity markets were also split along national lines, and, in any case, were much less important than in the United States. Banks were the dominant financial intermediaries in Europe, shuffling money from savers to borrowers.3
The national focus of European banking is vividly illustrated by the nationalization of the French banking system. In 1981, a socialist government was elected under François Mitterrand on a platform to nationalize major industrial groups and private banks. Accordingly, in February 1982 two conglomerates with major banking interests (Paribas and Suez) and thirty-six individual banks were nationalized, bringing over 90 percent of deposits under state control.4 Strikingly, this takeover of virtually the entire banking system of Europe’s second largest economy was accomplished using legislation crafted for domestic companies. It did not require the complex procedures and negotiations involved in nationalizing foreign-owned firms, such as the French subsidiaries of ITT and Hoechst.5 European banks were nationally run and nationally owned. As a result, a government takeover, as in France, was easily accomplished.
This was not how the European banking system was supposed to be evolving. The Treaty of Rome that founded the (then) European Economic Community in 1957 had pledged an “ever closer union” across its members that encompassed an ambitious program of economic integration, including a unified banking system. The Treaty had envisaged a gradual process as barriers to trade were removed and firms were given the right to establish themselves in other members of the Community. This plan was reinforced in the early 1970s, when the European Council directed that banks from across the members of the Community should face the same regulation and supervision.
On the ground, however, it was clear that these provisions were not creating a European-wide banking system. This reflected the existence of separate currencies as well as national policies. The costs involved in swapping one currency for another discouraged cross-border banking. While the charges for a single transfer between (say) the Deutsche mark and the French franc were only a few percentage points, they added up quickly if the same funds had to be transferred back and forth several times as would typically occur in a truly integrated cross-border bank. Capital controls added further headaches by disallowing some transactions and adding time, cost, and paperwork to others. Such controls were particularly prevalent in countries suffering from relatively high inflation, including major European Community countries such as France, Italy, and the United Kingdom. These controls reflected the strains these members faced in maintaining international competitiveness given the commitment to limiting exchange rate fluctuations contained in the European Exchange Rate Mechanism. Finally, every Community country except the United Kingdom required foreign-owned banks to raise local capital, adding costs that further fragmented the European banking system.
Irritating as the costs of different currencies, capital controls, and local capital levies were for banks, the most permanent barrier to cross-country banking would turn out to be the informal barriers created by national regulators. Each country had its own set of banking regulations with which any foreign venture had to comply. Since any entry into the domestic market by a foreign bank required the approval of national governments, regulators could easily block foreign ventures. As noted by an independent report to the European Commission in 1988, while “there are no overt barriers to the establishment of foreign banks” the costs of compliance could be considerable and there was “control over the acquisition of domestic banks by foreign entities in all the countries”.6 Regulators ensured that European banks were stopped at the border, but changes were about to happen. Indeed, 1986 marked a pivotal point in European banking history as the first of the transformative decisions was passed.
* * *
The Road to Hell is Paved with Good Intentions
Three major policy decisions, all aimed at integrating the European banking system, culminated in the long banking boom that ended abruptly with the North Atlantic crisis. Each decision made individual sense, but unforeseen interactions turned these individual palliatives into a poisonous mixture. The first of these was the Single European Act that was the brainchild of the European Commission.
The Single European Act
The institution that is responsible for driving European economic integration is the European Commission, set up in 1957 by the Treaty of Rome. Like most institutions, the fortunes of the Commission have waxed and waned with the quality of its leadership. As a general rule, the leaders of the Commission have tended to be relatively ineffective, in part because weak leadership in the European Commission enhanced the relative power of major members such as France and Germany. In 1985, however, the presidency was given to Jacques Delors, an energetic, ambitious, and efficient former French minister. This choice was no accident. Rather, it signaled the importance that President Mitterrand of France had decided to give to accelerating European economic integration after he turned away from his earlier go-it-alone socialist policies.
Jacques Delors immediately set about re-energizing the push towards European economic integration and a single market. The first fruit of this new sense of leadership and purpose was the passage of the Single European Act in 1986, which committed the members of the Community to the creation of “an area without internal barriers in which free movement of goods, persons, services and capital is ensured” by the end of 1992. As a finishing touch, the act renamed the European Community as the European Union to further emphasize the longer-term commitment to greater economic and political integration.
For the banking industry, the crucial element was the free movement of services and capital. Given the existing constraints on integrated banking, the Commission focused on opening capital markets, imposing a single banking model, and breaking down regulatory barriers to Europe-wide banking. The opening of capital markets was achieved smoothly. Details on how to comply with the Single European Act were provided to the members of the Union by directives crafted by the European Commission. One such directive asked member countries to liberalize all capital movements in the Community and between Community countries and third countries by July 1990.7
Looking forward in time, the wider objective of eliminating the costs of exchanging currencies was achieved by Jacques Delors in his most striking accomplishment, the Maastricht Treaty on Economic Union. The Treaty, agreed in December 1991 and signed in March 1992, established the path to European Monetary Union (EMU). It specified that by the start of 1999 wholesale financial transactions would be carried out in a common currency and hence would not involve transaction costs (Euro notes and coins were introduced in 2002). Since the membership of the monetary union was not to be determined until 1998, there remained uncertainty as to which countries would be admitted and hence the group across which transaction costs would be eliminated. However, it was clear from the start that the currency union would incorporate a wide swathe of Europe, almost certainly including the northern core of Germany, France, the Netherlands, and Belgium. European bankers could confidently make plans on the assumption that transaction costs would be eliminated across a much of the continent by the end of the 1990s.
The Single European Act was equally successful in synchronizing European banking models. The Second Banking Directive, issued by the Commission in December 1989, defined what services a bank could perform.8 Crucially, the proposal encompassed both commercial banking (loans to companies and households) and investment banking (supporting financial markets through such services as brokerage, equity issuance, and arranging mergers and acquisitions). This “universal” banking model (universal because banks could engage in both commercial and investment banking instead of having to specialize in one or the other) was typical in much of continental Europe, including Germany. It contrasted with the approach taken in the United States, where commercial and investment banking activities were required to be performed by different firms, as well as the United Kingdom, which also had a tradition of separate commercial and investment banks (the latter called merchant banks because of their origins in financing trade).
Competition across national regulators within the Union ensured that this universal banking model was rapidly adopted. This was because the Commission’s Second Directive allowed a bank to offer all services approved by its national supervisor throughout the rest of the Union. As a result, any country that retained a narrow banking model risked putting its banks at a disadvantage by limiting the services its banks could offer compared to foreign competitors. Depositors might switch from a domestic to a foreign-owned universal bank because the latter (say) could provide brokerage services that allowed clients to invest money. This risk provided a powerful incentive for countries to rapidly adopt the wide definition of banking proposed by the Commission. The universal banking model was a crucial development that helped drive the subsequent expansion of core Euro area mega-banks into investment banking. Before turning to that part of the story, however, it is necessary to explain why regulatory competition foiled the most important part of the Commission’s plan to create an integrated European banking system.
At the heart of the Single European Act was a plan to overcome national resistance to cross-border ventures through a new and innovative process called “mutual recognition”. Mutual recognition replaced the earlier approach that encouraged integration by ensuring that all European Union firms faced the same regulations within a country regardless of their nationality. Countries had found it easy to thwart this approach since, for example, German regulators could and did use their responsibility to approve new firms to discourage entry by foreign entities, including banks. Under mutual recognition, by contrast, if a French firm wanted to set up a new venture in Germany, it would only be subject to French regulations (as long as the French rules complied with EU-wide standards). Since the foreign entrant was not subject to German rules, German regulators had no basis to reject the entry of a French firm. And the same logic applied to a German firm entering the French market—hence the descriptor “mutual” recognition. In a single stroke, the act promised to speed up the process of economic integration by lowering barriers to cross-border firms. Indeed, it promoted rapid integration in many industries, but not in banking.
Explaining why mutual recognition failed to integrate banking requires an understanding how the rule interacted with the incentives of national supervisors. In the Second Banking Directive, the European Commission introduced mutual recognition through a “passport” system and the principle of source country control. European Union banks were issued a single passport that allowed them to do business anywhere in the Union. Source country control meant that foreign bank branches—parts of the domestic bank set up in a foreign country—operated under the rules and supervision of the source country rather than the host country. Hence, for example, if Deutsche Bank set up a branch in Paris, the branch was allowed to follow German banking law and was supervised by the German bank regulators rather than French ones. By contrast, if Deutsche Bank set up a subsidiary—a bank incorporated in France—it was under French supervision.
This approach was completely different from the traditional role played by foreign bank branches and subsidiaries that still operates in the rest of the world. Typically, in (say) the United States, a branch of a German bank is subject to US regulations. However, since the branch is only allowed to offer limited services such as support for traveling clients, it operates under less stringent rules than a full US bank. If a German bank wanted to offer a wider range of services in the United States then it has to set up a subsidiary bank, fully incorporated in the United States and thus subject to full US regulation. By contrast, mutual recognition meant (and still means) that within the European Union there is no difference between the services that are offered by a branch and a subsidiary. Rather, the difference is who supervises the bank. The Germans supervise all branches of German banks elsewhere in the European Union such as France. By contrast, a French subsidiary of a German bank is supervised by the French.
In practice, however, this elaborate scheme to create an integrated EU banking system via bank branches was stymied by informal barriers set up by national supervisors. Since banking rests on relationships with clients, expansion almost always occurs through acquisition rather than starting a new bank from scratch, since buying an existing bank automatically transfers its clients. Any attempt by a foreign bank to acquire an existing domestic bank, however, involves negotiations with the existing owners and, importantly, with the current national bank supervisor. So, for example, if a German bank wanted to buy a French bank, the deal would need to get the blessing of the French supervisor. National supervisors used this leeway to resist foreign entry. As one commentator put it in the early 2000s:
Despite legislation on freedom of entry, rumors abounded of public intervention to deter entry of foreign banks in the case of the sale of CIC in France and of Generale Banque in Belgium, an (unsuccessful) attempt to prevent the sale of a bank of the Champalimaud group to Banco Santander in Portugal, and of a desire by the Central Bank of Italy to keep large banks independent.9
As a result, cross-border mergers and acquisitions were lower in the banking industry than in other sectors. In Germany, for example, despite the Single European Act, foreign banks owned just 7 percent of all banking assets by the end of 2002. And Germany was not an outlier. Foreign bank assets represented under one-tenth of all assets in every major Euro area country except France, where they squeaked in just above that threshold.
Another telling sign of the failure of “passporting” and the importance of national bank supervisors in resisting foreign entry was that the vast majority of the cross-border bank mergers involved the creation of locally owned and supervised subsidiaries rather than foreign-supervised branches.10 For example, between 1985 and 2002 assets of foreign subsidiaries in the German banking system quintupled to 5 percent of output while assets in foreign bank branches remained stuck at around 2 percent, a pattern that was repeated across the Union. There were clear reasons for regulators to encourage cross-border entry through subsidiaries rather than branches. National supervisors had a strong interest in overseeing as much of the domestic banking system as possible, as problems in foreign ventures could spill over onto local banks and markets. Given the risk that problems in one bank could rapidly spread and undermine confidence in the rest of the system, it was eminently logical to discourage foreign branches. The underlying weakness of the bank passport system was that, for instance, a French branch of a German bank was supervised by Germans but the main costs of any problems were likely to be incurred in France.
Equally telling is the exception to this pattern of limited foreign entry occurring largely through subsidiaries, namely the United Kingdom. The UK banking system had a very different structure from the rest of the Union as it hosted a large foreign banking presence focused on the highly internationalized UK capital markets. These investment banks were incorporated as branches rather than subsidiaries so as to integrate market support across the major international financial centers, especially London and New York, while the interaction with UK clients was limited. Such international links were much more easily achieved using a branch that was run out of headquarters than by an independent subsidiary. The importance of investment banking in London led to a relaxed attitude to foreign entry. Almost half of all UK bank assets were in branches of foreign banks—London was the melting pot of the international banking system. The gulf with the rest of the EU is equally striking. By the end of 2002, the assets of EU branches in the United Kingdom were twice the size of such branches in the much larger Euro area. Put another way, while more than 20 percent of UK bank assets came from cross-border EU branches, for the Euro area this ratio was only 3 percent. In addition, the UK harbored numerous bank branches from the rest of the world, a type of foreign entry that barely registered in the neighboring Euro area. This example of the impact of a relaxed attitude to foreign entry underlines the central role played by other EU supervisors in blocking integration of the banking system.
Why Was National Bank Supervision Retained in the Maastricht Treaty?
The second crucial policy decision for European banking actually involved no decision—rather retaining the status quo. In the 1992 Maastricht Treaty that determined the structuring of the future currency union, European leaders decided to keep responsibility for supervising and rescuing banks at the level of individual member countries rather that centralizing at the union level, the arrangement typical in other monetary unions such as the United States. The decision to retain national supervision was crucial as it undermined the European banking system in two ways. Competition across national supervisors meant that they increasingly became supporters for their own major banks. This led to the creation of large “national champions” and promoted a supervisory “race to the bottom” as regulators looked to boost the competitive position of their own champions by not being too intrusive. In addition, the fact that in the end almost all entry into other European countries was through subsidiaries meant that the costs of the eventual crisis were bottled up in individual countries. The quip from former Bank of England governor Mervyn King, that banks live internationally but die nationally, was especially true within the Euro area. This had far-reaching consequences. Regulators and governments in the crisis countries were overwhelmingly responsible for the losses, helping to meld their banking and fiscal problems. On the other hand, because a significant part of the financing for the bubbles came from elsewhere in the Euro area, often through loans to local banks, the potential knock-on costs of bank failures to the rest of the Union produced strong incentives across the Euro area to fudge the assessment of the resulting bad loans.
Given the important role played in destabilizing the Euro area banking system by the decision in the Maastricht Treaty to keep bank supervision with member states, it is worth probing its background more thoroughly. The decision fits with the treaty’s broad philosophy of subsidiarity, which means leaving issues that did not need to be centralized at the national level. As a result of this objective, the Maastricht Treaty left most major responsibilities outside of monetary policy with individual members, such as fiscal policy and structural reforms.11 The general philosophy was to focus on creating a monetary union within the existing institutional structure of the Union rather than adding further federal bells and whistles. In the case of banking, the broad regulatory structure was already defined centrally by the European Commission through Union-wide directives, including on capital controls, the services a bank could perform, and the size of capital buffers. The issue, therefore, was whether day-to-day supervision of these rules should remain a national responsibility.12
This question provoked a lengthy discussion in the Delors Committee, the group largely comprised of central bankers that provided the blueprint for the eventual Maastricht Treaty. Intriguingly, in the discussion of bank supervision, the traditional roles of the German and the British officials in discussions about European integration were reversed. The Germans, who were generally sympathetic to a more federated structure for the Union, took a strong view that bank supervision should remain at the national level. This reflected the Bundesbank’s concern that the newly formed central bank should be truly independent and not subject to outside pressure. The fear of the German negotiators was that if banking supervision was elevated to the level of the Union then problems in the banking system would also be dealt with by the center. Given the small size of the Commission’s budget, this would likely imply a need for the European Central Bank (ECB) to provide support for troubled banks. Such responsibilities would be a distraction from the bank’s central objective of maintaining price stability and would encourage political lobbying about its decisions. To avoid this risk, Karl Otto Pöhl, President of the Bundesbank, pushed strongly for national supervision in the Delors Committee. By contrast, the British, who were generally the most skeptical of any move toward federation and obtained an opt-out from the single currency, were sympathetic to a Union-wide regulator. This reflected the size of the UK banking system, which was large even by European standards and therefore potentially costly to rescue. In addition, the British were keenly aware that their supervisors held no effective oversight over the massive assets in foreign branches, especially those from other European Union banks where the Second Banking Directive explicitly cut them out of the process. Accordingly, Robert Leigh-Pemberton, the Governor of the Bank of England, argued for centralized supervision.
In common with most issues in the Maastricht Treaty, the German approach prevailed, albeit with a nod to the British view.13 The text of the Delors Report stated that the European System of Central Banks (ESCB) would “participate in the coordination of banking supervision policies”. Over the Maastricht negotiations, the ESCB’s role in financial supervision was further downgraded to “contributing” to the policies pursued by “competent authorities”.14 This clearly left the supervision of banks to national regulators. However, the Treaty also included a get-out-of-jail-free card that allowed supervision to be centralized. More specifically, Article 105.6 authorized the European Council to let the ECB take on bank supervision if the Council acted “unanimously on a proposal from the Commission and after consulting the ECB and after receiving the assent of the European Parliament”. Unsurprisingly, given the requirement for unanimity across all member states, this option was not invoked until 2012, four years after the financial crisis first broke.
By hard-wiring regulatory competition into the European financial system, the Maastricht Treaty reduced the incentives of European supervisors to look carefully at the behavior of major domestic banks. That was because their banks competed with banks from other countries whose supervisors might be cutting them more slack.15 It also meant that there was no institution examining the evolution of the European banking system as a whole. With nobody minding the shop, the mega-banks were able to use changes in the rules on capital buffers to become larger and less safe. This opening came through the Basel Committee rules on international bank capital.
The Rise of Bank Internal Risk Models
The third policy decision that drove the expansion in Euro area banking involved international regulations promulgated through the Basel Committee on Banking Supervision. The Basel Committee, whose decisions played a crucial role in the evolution of the European banking system and the North Atlantic crisis, was formed by the Group of Ten (unintuitively comprising eleven major advanced economies) in 1974 to improve supervisory quality and understanding worldwide.16 It is housed in the Bank for International Settlements in Basel and, while it has no formal legal standing or permanent staff, it remains the main driver behind regulation for internationally active banks. One of its major roles is to craft internationally consistent rules on capital buffers.
The push for consistent capital standards originated from concerns that differing national rules were providing some banks with a competitive advantage compared to their international competitors by allowing them to hold thinner capital buffers. Bank capital is intrinsically risky as owners are the first to lose their money in the case of financial distress. Accordingly, investors demand a higher rate of return on capital compared to safer forms of borrowing such as bonds or deposits. The concern was that competition across supervisors was creating a regulatory “race to the bottom” in which each country tried to make their banks more competitive by diluting the requirements on their expensive capital buffers, leading to inappropriately thin buffers across the board.17 The risk had been underlined by the international repercussions of major bank failures, such as that of Continental Illinois Bank in 1984. Even so, while national regulators were putting increasing emphasis on rules on bank capital, in 1985 there was “considerable variation in the mode and details of the capital regulation … and little apparent interest in most supervisors in harmonizing their capital regulations”.18
A crucial breakthrough came in 1986 when, rather to their surprise, negotiators from the US and UK rapidly settled on a common set of capital standards.19 The two countries were particularly influential members of the Basel Committee, reflecting both the size of their economies and the importance of the New York and London financial markets for international banking. In addition, they obtained an agreement in principle to adopt their new standards from Japan, another important member of the Committee. The core of the US/UK proposal comprised a consistent definition of what types of assets could be included in bank capital, the relative riskiness of different types of commercial loans, and the amount of capital that had to be held against such “risk-weighted” assets. The proposal focused on commercial loans since in both countries investment banking was mainly performed outside of the main banking sector.
The US/UK proposal became the basis for negotiations throughout 1987 on a uniform definition of bank assets and capital across the members of the Basel Committee. The discussions were most contentious around which assets should be classified as capital, with different countries supporting definitions that included less effective buffers that their own banks already held so as to minimize additional capital demands. So, for example, Japanese negotiators supported the inclusion of unrealized capital gains on bank equities and those from the United States advocated the inclusion of certain types of preferred stock. The final outcome was the first set of uniform international capital adequacy standards, known as the Basel 1 Accord. The rules were agreed in 1988 and were to be enforced in all member states by the end of 1992 so as to allow national supervisors the time to amend their existing frameworks.
The preamble to the agreement explained that the objective was “to secure international convergence of supervisory regulations governing the capital adequacy of international banks”.20 The new standards were aimed at strengthening the “soundness and stability of the international banking system” and at “diminishing an existing source of competitive inequality among international banks”.21 Basically, each bank loan was to be placed in one of five possible buckets with risk weights varying from 0 to 100 percent.22 So, for example, loans to governments of advanced countries were considered extremely safe and were given a risk weight of zero. At the other end of the scale, corporate loans were seen as highly risky, and were given a weight of 100 percent, while mortgages attracted a middling value of 50 percent. These buckets were then summed to get total “risk-weighted” assets. Required capital buffers were then calculated as a percentage of these risk-weighted assets. Reflecting the lack of agreement on what should be included in bank capital, Basel 1 included two levels of minimum capital buffers. “Core” (Tier 1) capital, that was largely equity and retained profits, had to be at least 4 percent of risk-weighted assets, while the sum of core and “supplementary” (Tier 2) capital that included less effective buffers such as subordinated debt (subordinated because in the event of a bankruptcy it would be written down before more senior debt) had to be at least 8 percent of risk-weighted assets.
Since the Basel Committee included seven of the ten members of the European Union, the Basel 1 Accord rapidly unified European bank capital rules. The non-participants were Ireland, Denmark and Greece, who had little alternative but to adopt the rules agreed by the larger members. The accord also covered the major international competitors to European Union as the other members of the Committee were Canada, Japan, Sweden, Switzerland, and the United States. However, as will be discussed in the next chapter, the US commercial banks were also under additional US-specific capital standards. This overlay, which effectively meant that the US banks were under a more stringent capital regime, was to have far-reaching consequences since it provided US banks with incentives to sell loans to the European banks over the boom of the 2000s.
The 1996 Market Risk Amendment
The rapid adoption of the universal banking model in Europe as a result of the European Commission’s Second Banking Directive created issues for the Basel 1 Accord. The focus of the existing Basel rules was commercial loans to firms or individuals. The large European banks, however, were increasingly expanding into investment banking and hence held increasing amounts of market assets such as equities and bonds. This involved a different type of “market” risk—losses coming from a fall in the price of these assets. This contrasted with the United States where the Depression Era Glass–Steagall Act continued to separate commercial banking from investment banking (other members of the Basel Committee fell between the wide universal banking model in Europe and the narrower commercial banking one in the United States).
In response to the increase in investment banking activities in its banks, the European Union decided that it needed rules on capital buffers for market risk. The Basel Committee became concerned that this move could lead to disjointed international capital rules, with different countries adopting different rules on market risk. Accordingly, in 1993 the Committee proposed an amendment to the Basel 1 risk weights to cover market risks, closely modeled on the European Commission’s European Capital Adequacy Directive issued the previous month.23 The essence of the proposal was to divide bank activities into traditional commercial loans, held in the “banking” book, and holdings of securities, held in the “trading” book. Assets in the banking book would be subject to the standard Basel 1 credit risk weights. By contrast, securities held in the trading book would be subject to new weights based on market risk. Securities would be classified on standardized measures of the risk of large changes in prices—basically the same “bucket” approach that already applied to credit risk.
To the surprise of the Committee, this proposal ran into strong criticism. This came mainly from the large banks, who argued that the proposed buckets were much less sophisticated than their own rapidly evolving “value-at-risk” models that calculated the risk to the value of an entire portfolio by taking into account not simply the volatility of individual asset prices but also the correlations across such prices. They noted that adopting the Committee’s proposal would lessen incentives to continue to develop their own internal risk models. Instead, the large banks asked to be allowed to use their own models to calculate the capital buffers needed for the trading book. The assumption of the large banks, which turned out to be correct, was that internal risk models would allow them to save on capital buffers and provide them with a competitive edge over their smaller competitors.
Disconcerted that banks found the proposed methodology “old-fashioned”, the Committee set up a Models Task Force to examine the banks internal models. The task force, led by Christine Cummings from the Federal Reserve Bank of New York, reported back to the Committee that they were impressed with the “obvious sincerity and expertise” of the banks.24 The report concluded that there were compelling arguments to adopt internal risk models, including “greater precision, avoidance of duplication and incentives to develop adequate systems”. Accordingly, the revised proposal on market risk, published in 1995, allowed large banks to use their internal risk models as the basis for calculating capital buffers for market risk.25
The decision to allow market risk to be calculated using internal risk models was made despite evidence that results differed widely across banks. The Models Task Force had earlier asked fifteen banks to calculate risks on several dummy portfolios. It reported that half of all banks had estimates of underlying risk that differed from other banks by over 50 percent.26 In response, the task force proposed “carefully structured safeguards to minimize the risk of abuse”. The report went on to the optimistic observation that “a moderate amount of supervisory guidance as to acceptable risk measurement practices could substantially reduce the dispersion of these results”. In response to these concerns the Basel Committee proposed a series of safeguards to reign in the internal models by setting a range of quantitative parameters as well as qualitative standards. In addition, the Committee proposed to top up the results from the risk models with a “multiplication factor”, whereby the final capital buffer would be three times that calculated by the internal risk model to compensate for the limitations of even the best models to predict shocks.
In a taste of things to come, these safeguards faced a barrage of criticism from the large banks and the Committee started down the path of negotiating the detail of the rules with them. As a result of these discussions, the final version of the amendment provided some additional flexibility for the banks (although the controversial multiplicative factor was retained).27 Crucially, however, there was no follow-up to check whether these changes made the results from the internal risk models more similar. Whatever the intentions of the Committee, the answer appears to have been no. Major differences in results were again found in the early 2000s, when the properties of internal risk models were next examined in detail in the run-up to the switch to the Basel 2 capital rules. Indeed, the Basel Committee has still not solved this problem.
The market risk amendment was a watershed moment that involved three crucial changes to the philosophy of the Basel Committee. At a very basic level, the Committee accepted the need for rules on international capital buffers to move beyond the traditional concern of bank regulators, namely the risk of commercial loans to individuals or firms going sour, and to provide a benchmark for market risks to bank balance sheets coming from buying and selling securities, in other words investment banking. In tandem, the Committee also accepted the principle of using internal bank risk models in the calculation of capital buffers. The Committee tacitly accepted that the large banks understood the risks from market trading better than the supervisors. Finally, as a corollary to that change, the Committee engaged in detailed discussions on bank capital regulations with the banking industry, in particular the major international banks.
The important question is why the Basel Committee decided to farm out the calculation of capital buffers to large banks. Here, a crucial role was played by the growing belief that market discipline would constrain risk-taking. This was largely driven by views coming from the United States. The major US banks had been at the forefront of developing value-at-risk models. In addition, Alan Greenspan, Chair of the Federal Reserve, and the Fed staff in charge of bank regulation were strong believers in the power of market discipline in controlling the risks taken by investment banks. This is made clear in Greenspan’s memoirs, sent to the printers just before the North Atlantic crisis and titled (with unintended irony) The Age of Turbulence, where he wrote about his experience at the Fed that:
Since I was an outlier in my libertarian opposition to most [financial] regulation, I planned to be largely passive in such matters and allow the Federal Reserve governors to take the lead. … Taking office, I was in for a pleasant surprise. … What I had not known about was the staff’s free market orientation, which I now discovered characterized even the Division of Bank Supervision and Regulation. … So while the staff recommendations at the Federal Reserve Board were directed to implementing congressional mandates, they were always formulated with a view toward fostering competition and letting markets work. … The staff also fully recognized the power of counterparty surveillance [i.e., market discipline] as the first line of protection against overextended or inappropriate credit.28
The Federal Reserve made a clear distinction between the appropriate regulation of commercial banking and investment banking. Depositors in commercial banks that lent to individuals and firms were protected by deposit insurance. They had few incentives to monitor the risks the bank was taking as they were protected from losses should the bank fail and were in any case relatively unsophisticated investors. Government regulation was needed to ensure that banks did not exploit this lack of attention by taking excessive risks. By contrast, investment banks were funded by sophisticated investors through large, uninsured “wholesale” deposits. Such investors could monitor the behavior of the investment banks and stop them from taking excess risks by threatening to withdraw their money. In the Fed’s view, government regulation was not needed and could, indeed, be counterproductive since it could lessen the incentives for investors to monitor the investment banks. Far from reinforcing the discipline that markets exerted on banks, government regulation could undermine it.
These views explain the why the Federal Reserve put its very considerable weight behind the disastrous decision to use internal risk models for market risk that left the banks in charge of calculating the buffers needed for their investment banking operations. While regulators were supposed to monitor the adequacy of these models, in practice such checks were always going to be difficult given large differences in results from models across banks. However, in the view of the Fed such checks were largely irrelevant given “the power of counterparty surveillance as the first line of protection against overextended or inappropriate credit”. Internal risk models were not really an instrument for bank regulation, but rather were a rejection of the need for such regulation. This also explains why a Basel task force headed by a Federal Reserve official concluded that there were “compelling arguments” in favor of internal risk models based on information provided by bankers with “obvious sincerity”.
The naïve decision to allow major international banks to use internal risk models to calculate their capital buffers for investment banking generated a cascade of destructive incentives in the Euro area banking system. It gave large banks a competitive edge over smaller ones as they could expand into investment banking on the cheap using thin cushions of expensive capital. For similar reasons, it also advantaged banks with more aggressive risk models, as they could operate with thinner capital buffers than their competitors. This incentivised large banks to tweak their internal risk models to be more aggressive, so as to save on capital and become more competitive compared to their peers. All of this helped drive the emergence of the small number of mega-banks that dominated European banking on the eve of the North Atlantic financial crisis. By using internal risk models to calculate the amount of (expensive) capital a bank needed, the market risk amendment changed the nature of the internal models from useful tools for risk management to a constraint on the size of the firm’s balance sheet. They became a way of lowering costs rather than assessing the level of risk. As in quantum physics, observation changed the nature of the experiment. Once risk models switched from an internal tool to assess risks to an external measure that determined the needed level of capital, banks altered the way in which these models were used.
Internal risk models also made the banking system inherently more cyclical. The reason for this is that internal risk models use market prices as a signal of the riskiness of an asset. If the economy is booming and asset prices are rising, the models interpret this as meaning that the asset is safer and hence that the buffers can be reduced. The capital thus saved could be redeployed to expand the bank’s lending, an increase in loanable funds that reinforced the boom. Of course, the same process works in reverse as became painfully obvious after the North Atlantic crisis. If an unsustainable boom in asset prices turns into a bust, banks’ internal models force them to shrink lending which exacerbates the resulting downturn. This contrasted with the standardized approach, where risk weights do not vary with cyclical fluctuations in asset prices. The cyclical nature of internal risk models was a major driver of the boom-bust that started in the late 1990s and reversed course after the crisis.
The irony here is that even though the market risk amendment was largely driven by the US Federal Reserve, it had a much larger impact on the European banks than the US ones. This was because the EU’s Second Banking Directive ensured that the European banks were at the forefront of the move toward universal banking involving mega-banks with large investment banking operations. The competitive advantages provided by internal models resulted in an industry increasingly driven by a small number of large and largely national banks that, particularly in the Euro area core, exploited the market risk amendment by expanding into investment banking. By contrast, in the United States investment banking was largely performed by the already lightly regulated independent investment banks.
The Poisonous Mixture
The confluence of the Single Market initiative, the single currency, and the Basel capital rules aimed to create an integrated and competitive European Union banking system. In practice, however, unanticipated spillovers meant that it generated a rapidly expanding and over-banked system that remained overwhelmingly national and was increasingly dominated by ever-larger mega-banks with thin capital buffers that were increasingly involved in investment banking. Key roles were played by the universal banking model coming from the Single European Act, by continuing competition across national supervisors as a result of the Maastricht Treaty, and by the Basel Committee’s 1996 market risk amendment. In particular, the market risk amendment allowed major banks to economize on capital buffers in their expanding investment banking operations. The banks used the newly freed capital for further expansion, often plowing it back into investment banking since the associated capital charges were so low. European regulators did little to prevent this process as competition between national regulators provided few incentives to reign in the activities of national banks. A vivid illustration of the influence of regulatory competition in eroding bank supervision was the move to “light touch” regulation in the United Kingdom and Ireland in the early 2000s, which left their banks relatively free from supervision.29 In the end, the decision by the Basel Committee to allow banks to use internal models to calculate capital buffers for market risk generated exactly the kind of diminution in capital standards on investment banking in Europe that the Committee had originally been formed to avoid.
* * *
Charting the Transformation of European Banking 1985–2002
The transformation of the European banking system from the 1980s through to 2002 can be clearly seen in the numbers. This narrative opens with a broad-brush summary of developments through the late 1990s, followed by a more detailed snapshot of the system in the early 2000s when more information on European banking becomes available with the advent of the Euro.
Expansion after 1985
The European banking system expanded enormously between 1985 and 1999 as restrictions on bank activities were lifted and universal banking took hold. Assets almost doubled as a ratio to output for the future members of the Euro area for which data are readily available (Figure 4). (Assets are reported as a ratio to output so as to abstract from the natural increase in the size of the banking sector as the economy gets larger.) Almost all of the increase came from the northern core of the future Euro area—Germany, France, the Netherlands, and Belgium. Indeed, the size of the core banking system caught up with that of the United Kingdom, traditionally the largest system in Europe because of its international orientation. By contrast, the size of banks in the periphery of the Euro area (measured as a ratio to output) hardly changed. By the turn of the century the core banking system was twice as large as that of the remaining “periphery” when compared to the size of the economies, having been almost equal in 1985.
Figure 4.Euro area banking boomed after 1985. Assets as a percentage of GDP.
Source: Dermin (2002) B49 country table except Netherlands in 1999 which comes from the ECB.
Notes: For Belgium the 1999 column is for 1999/2000. Euro area data are weighted using 1999 GDP in euros.
The 1996 Basel market risk amendment was a major driver of the increase in the core banking system. Figure 5 reports quarterly data on the banking systems of Germany and the Netherlands, the two countries for which such data are readily available. Assets expanded rapidly as a ratio to output in both banking systems, particularly after 1996. The fact that most of the expansion went into investment banking can be seen by the much faster increase in the darker lines that show the path of total loans compared with the lighter lines which show the amount of commercial loans (i.e., loans to firms, households, and governments). The expansion into investment banking involved buying overseas ventures. In particular, Germany’s largest bank, Deutsche Bank, acquired investment banks in London (where they bought Morgan Grenfell in 1989) and New York (where they bought Bankers Trust in 1999). In the Netherlands, the ING Group bought Bank Brussels Lambert in 1997.
Figure 5.Investment banking in the Euro area core expanded rapidly after 1996.
Source: National Banking Statistics.
The pattern of mergers and acquisitions in European banking clearly shows how the resistance of Euro area supervisors to cross-border takeovers created large national banks with overseas investment banking operations. In Figure 6 every European bank takeover in the 1990s is defined along two dimensions. Do the banks have the same business model within the industry (e.g., is a commercial bank acquiring another commercial bank) or different business models across the industry (e.g., is a commercial bank acquiring an investment bank)? And is the merger domestic or cross-border? The results show a stark contrast in takeovers between banks with similar business models and those involving banks with different models. Over three-quarters of takeovers of similar banks were domestic (the light grey segments). Basically, commercial banks bought local commercial banks in a process that helped to create nationally orientated banks. On the other hand, the overwhelming number of transactions involving banks with different business models were cross-border (the black areas). This is the dynamic that created universal banks with investment banking operations in market centers such as London. The overseas expansion into investment banking reflected the adoption of a universal banking model in Europe combined with the permissive attitude of the UK regulators to foreign entry, especially after the “big bang” deregulation of the City of London in the late 1980s. Indeed, a significant part of cross-border activity involved the dismemberment of the UK merchant banks (the local name for independent investment banks). By 2002, thirteen major UK merchant banks had been acquired by overseas institutions; seven by European Union banks (including Deutsche Bank’s purchase of Morgan Grenfell), two by Swiss ones, and four by US banks.30 As will be discussed in the next chapter, the independent US investment banks proved much more resilient.
Figure 6.Most European bank mergers involved domestic agglomeration.
Source: Demine (2002) Table 12.
The Banking System 1999–2002
By the early 2000s, the expansion of the Euro area banking system was switching from the core to the periphery of Europe. The rate of increase in the banking systems in the core countries (Germany, France, the Netherlands, and Belgium) were slowing as a proportion of output, while those in the periphery were starting to accelerate, albeit from a much lower base. This buoyancy in the periphery reflected the fall in interest rates as a result of European Monetary Union, which was accompanied by accelerating inflation in housing and other asset prices, a trend that would continue through the 2000s.
There were also major differences in the underlying business models between these two halves of the Euro area. The banks in the core were much more focused on investment banking. For example, commercial loans represented only some 30 percent of bank assets in France and Belgium. By contrast, such loans comprised more than half of the assets in the Italian, Spanish, and Portuguese banks. As a result, the assets of the core banks were more diversified internationally. By 2002, around 10 percent of all bank assets in the core Euro area were held in the United States and United Kingdom, the countries that housed the international investment banking hubs of New York and London. For the periphery this ratio was less than 5 percent.
In another trend that was destined to continue in the 2000s, the banking systems started to take off in Ireland and the United Kingdom, the two EU countries that adopted “light touch” supervision. In Ireland, this represented an attempt to repeat the earlier successful strategy of using government policy to attract foreign manufacturing multinationals (in that case through lower tax rates), an approach that had turned Ireland from a poor country into the “Celtic tiger”. In the case of banking, the attractions of low taxes were reinforced by a permissive “light touch” approach to supervision. As a result, Ireland’s banking system ballooned from two-and-a-half times the economy in 1997 to an enormous five times the economy by 2002 as foreign banks entered the market. The new business was largely in investment banking, and Ireland’s ratio of commercial loans to assets fell rapidly to 25 percent, the lowest in the Euro area. A similar transformation occurred in the UK banking system, but it was less dramatic because the United Kingdom started with a much larger foreign investment banking presence.
Despite the push to create an integrated banking system, banking in the Euro area remained mainly domestic (Figure 7). For the region as a whole, less than 10 percent of all assets came from banks based in other members of the Union. This was only modestly higher than the ratio seen in 1999, with an intuitive pattern in which cross-border assets were lower in large economics than smaller ones. Furthermore, despite passporting, the vast majority of such cross-border assets resided in subsidiaries rather than branches as a result of pressure from national regulators.31 Underlining the importance of such pressure, the much more open UK banking system was the most internationalized in the European Union, with fully half of bank assets owned by foreign banks (equally split between EU and non-EU banks). Boosted by light touch regulation, Ireland was the second most internationalized.
Figure 7.Euro area banks were largely national in 2002.
Source: ECB Structural Banking Statistics.
The Creation of the Euro Area Mega-Banks
A wave of bank mergers and acquisitions in the 1990s increased the role of a small number of large banks in the Euro area. While this can be measured in the aggregate—the share of assets in the largest five banks rose in every Euro area country between 1998 and 2002 with the exception of Finland—the more important part of the story involved the rapid rise of a small number of national mega-banks, tellingly called “national champions”. By 2002 just twelve Euro area banks already held one-quarter of regional assets and their role continued to expand through to the eve of the crisis in 2007.32 The increasing dominance of a small number of rapidly expanding mega-banks reflected the competitive advantage conferred on large banks by the ability to use internal risk models to calculate capital buffers on investment banking operations. In some cases, national regulators also supported the creation of mega-banks as a defensive reaction to larger banks elsewhere. This was particularly the case in the south of Europe, which had a tradition of smaller and more local banks.
Strikingly, all of the twelve mega-banks already existed in 2002, with six having been formed in the late 1990s. Given the distinct differences in business models, it is useful to treat the core Euro area banks separately from those in the southern periphery. The largest Euro area mega-banks were universal banks from the core Euro area comprising Germany, France, the Netherlands, and Belgium (Figure 8). They were generally older and more established banks, several of which were expanding into investment banking through acquisitions in the United Kingdom or the United States (Figure 9 provides a potted history of the main mergers and acquisitions that brought these banks into being and expanded them). For example, Deutsche Bank bought Morgan Grenfell in the UK and Bankers Trust in the US, ING bought Barings in the UK, and Société Générale of France bought TCW in the US. The other three core Euro area mega-banks, which also had major investment banking operations, were BNP Paribas and Credit Agricole of France and Commerzbank of Germany. The smallest two mega-banks in the core were more specialized. Natixis was a pure investment bank, while Dexia focused on financing local governments, a business model that was heavily dependent on the Basel Committee’s decision to give advanced country government debt a zero risk-weight.
Figure 8.Euro area mega-banks were already becoming too big to fail.
Source: Scientific Committee of the European Systemic Risk Board (2014).
Figure 9.Mergers in the late 1990s completed the European mega-banks.
Source: Corporate websites.
The four mega-banks in the southern Euro area periphery, UniCredit and Intesa in Italy and Santander and BBVA in Spain, had a very different profile. They were all relatively late arrivals, being created by mergers of smaller commercial banks in 1998 and 1999, often with the support of regulators. They were smaller than most of their rivals in the core and were still basically commercial banks. In many ways, the southern Euro area mega-banks are best seen as a defensive agglomeration of commercial banks in response to the rapid expansion of their northern counterparts.
The rapidly expanding mega-banks were already becoming too large to be managed by the corresponding national governments. The largest three (Deutsche Bank of Germany, ING Group of the Netherlands, and BNP Paribas of France) each had assets of almost 10 percent of Euro area output. They were much larger as a ratio of their national economies. Indeed, the assets of ING handily surpassed the annual output of the Netherlands. Problems in such massive banks were obviously going to have major spillovers to the rest of the economy, underlining that they were too large to fail. Indeed, given that bank support was provided at the national level, they were arguably also becoming potentially too large to save.
At the same time, the northern mega-banks in particular were using internal risk models to save on capital buffers. Figure 10 shows on the vertical axis the capital buffers using equity against total assets, which did not involve risk models, and on the horizontal axis using the Basel capital buffers against risk-weighted assets, which did use them.33 In 1996, before the market risk amendment, these two measures of bank soundness were positively related so that a high regulatory capital ratio also signaled a proportionately high ratio of equity to assets. A Basel regulatory capital of 5 percent, for example, corresponded with an equity–asset ratio of about 3 percent, and 10 percent corresponded with about 6 percent. In 2002, this positive relationship still held for the mega-banks in the periphery which focused on commercial banking but had vanished for the mega-banks in the Euro area core with a universal banking model. Indeed, these universal banks showed, if anything, a negative relationship between the size of their equity cushion as a proportion of assets and the regulatory capital ratio based on risk-weighted assets. While the Basel regulatory ratio showed capital buffers to be comfortably above the 4 percent minimum, equity cushions were thinning as a ratio to assets in universal banks such as Deutsche Bank, Commerzbank, Société Générale, and ING. The contrast with the commercial banks as well as with their own behavior in 1996 makes it clear that this change was driven by the corrosive effect of internal risk models.
Figure 10.Internal risk models led to thin capital buffers.
Source: Scientific Committee of the European Systemic Risk Board (2014).
The loss in correspondence between unweighted and risk-weighted measures of the capital cushion would not have mattered if the internal models were a good measure of balance sheet risk. After all, the main reason for introducing risk weights was to better differentiate the level of risk across types of assets. However, while admittedly crude, the Basel 1 risk weights for commercial banking were consistent across banks and were broadly sensible. Lending to firms, for example, was generally riskier than (say) lending for household mortgages, just as the risk weights implied. By contrast, internal risk models generated bank-specific results that encouraged downward pressure on capital cushions that was to continue into the 2000s. Tellingly, in the North Atlantic crisis, the cruder measure of equity to total assets turned out to be a better predictor of banking stress than more complex risk-weighted ones used by regulators.34
In addition, risk weights created opportunities for transatlantic purchases of assets based on differences in regulations rather than economic returns. This was because the regulated US banks were generally (effectively) under a simple leverage ratio that took no account of risk while the European banks were under loose risk-weighted capital rules. Having the two halves of the North Atlantic economy under different capital standards provided incentives for US banks to sell low-risk loans to European banks through securitized assets. The US financial system will be discussed in detail in the next chapter, but the significant holding of US assets by European banks in 2002 with only a limited reciprocal relationship suggests that this factor was already starting to matter.
* * *
The European Banking System in 2002
By 2002, the European banking system was showing evidence of maturing. After a period of rapid expansion of the northern European banks brought about by the single market program and the 1996 Basel market risk amendment, the ratio of assets to output was stabilizing in core banking systems of Germany, France, the Netherlands and Belgium. Growth was continuing in the periphery, partly fed by loans from the core, but banking was generally a smaller part of their economies so some catch-up could be seen as appropriate. At two-and-a-half times the economy, the assets of the banking system were large by international standards, but at least part of this difference was because banks performed more investment banking, services that in many other countries were carried out by specialized non-bank institutions.
At the same time three trends were sources for concern, all of which were linked to the behavior of national supervisors. The first, and most obvious, was the failure to create an integrated European banking system. As discussed earlier, this reflected a strong preference of national supervisors for domestic rather than foreign mergers. This contrasts with the smooth road to national banks in the United States, discussed in the next chapter.
The second trend was the continuing rapid increase in the relatively large banking systems in the United Kingdom and Ireland that specialized in investment banking and where supervisors embraced “light touch” regulation. This continuing dynamism contrasted with the relative stagnation of the northern Euro area core in the early 2000s. Light touch regulation eroded the authority of other national supervisors, as their banks could argue that they also needed lenient supervision so as to level the playing field. National supervisors who insisted on maintaining a tough approach in the face of such regulatory competition also ran the risk that domestic banks would move major operations to UK and Irish subsidiaries, reducing national supervisors of the ability to observe the full activities of their banks.
The final trend was the increasing importance of a small number of mega-banks from the northern core in the EU banking system (Figure 11). This largely reflected the use and abuse of internal risk models. This is clearly shown by the increasing dissonance between the comfortable Basel regulatory capital ratios—based on risk-weighted assets using internal models being reported to supervisors and markets—and the thinning of unweighted measures of capital buffers in the northern European universal banks. The fact that many of these mega-banks were labeled “national champions” is a telling recognition of the importance of regulatory competition in shaping the European banking landscape. In a truly integrated system, the nationality of the bank would be relatively irrelevant—champions would be defined by competitive advantage rather than by nationality.
Figure 11.Structure of Euro area banking in 2002.
None of these trends would have been impossible to solve. However, such a response would have first required an acknowledgement that competition across national regulators was creating unwelcome strains in Euro area banking and that strong central supervision was needed. Such a reassessment was difficult, as it would have required reopening the decentralized structure of financial supervision that had been carefully negotiated in the Maastricht Treaty. In the event, this reassessment did not happen until after the crisis. It would also have required pushback in the Basel Committee on internal risk models, a move which would have met with resistance from the banks themselves and from the US Federal Reserve. Instead, thin capital buffers became central to the business models of the core Euro area mega-banks. Tellingly, after the crisis it was the US and UK regulators that championed tougher bank capital buffers, while French and German regulators supported the continued use of internal risk models that had become intrinsic to the business strategies of their mega-banks.