April 1999 CONSEQUENCES FOR GREECE AND PORTUGAL OF THE OPENING-UP OF THE EUROPEAN BANKING MARKET Patrick Honohan Development Economics Group, The World Bank Summary Committed, as they were by membership of the European Union, to the removal of exchange controls and to allowing entry through cross- border provision of banking services, Greece and Portugal (the least prosperous and most peripheral countries of the Union) have experienced financial liberalization on a wide front. The result has been a huge expansion in the domestic financial sector. Interest rate spreads increased at first, but competition has reversed that initial trend and consumer benefits have been substantial. Although the trigger for these changes was internationalization, the changes have been largely domestic: penetration of foreign-owned banks has been remarkably limited. _____________________________________________________ An earlier version was prepared while the author was with the Economic and Social Research Institute, Dublin. He is grateful to Stijn Claessens, Jordi Gual and Ross Levine for very helpful comments. CONSEQUENCES FOR GREECE AND PORTUGAL OF THE OPENING-UP OF THE EUROPEAN BANKING MARKET CONTENTS Summary 1 Introduction and Overview 2 The Background: European Financial Liberalization 2.1 Removal of Exchange Controls and the Exchange Rate Mechanism 2.2 The European Single Market Project 2.3 Liberalization of domestic controls on financial intermediaries 3 Impact on the Peripheral EU Countries 4 Greece: Dismantling Pervasive Government Involvement 4.1 General Economic Background 4.2 Structural Reform in the Financial Sector: Overview 4.3 Concentration and ownership; foreign entry 4.4 Foreign business of the banks 4.5 Interest Rates and Bank Margins 4.6 Financial Liberalization and Macroeconomic Stability 4.6 Policy lessons 5 Portugal: Nationalization and Reprivatization 5.1 General Economic Background 5.2 Structural Reform in the Financial Sector: Overview 5.3 Concentration and ownership: foreign entry 5.4 Foreign Business of the Banks 5.5 Interest Rates and Bank Margins 5.6 Financial Liberalization and Macroeconomic Stability 5.7 Policy Lessons 6 Concluding Remarks Annex: Some Illustrative Econometric Results on the Dynamics of the Spread CONSEQUENCES FOR GREECE AND PORTUGAL OF THE OPENING-UP OF THE EUROPEAN BANKING MARKET 1 Introduction and overview Should developing countries fear or welcome internationalization of their banking systems? The experience of the European Union throws light on this question. Not only have EU member states removed all exchange controls, but a "single passport" system allows banks to operate freely across borders within the Union. Of course the experience of the larger and more sophisticated of the EU financial systems is of some relevance to developing countries. But it is especially instructive to see how Greece and Portugal have fared, given that they started from a position in the mid-1980s that was not too dissimilar to that faced by some middle income countries today. Both countries have experienced extensive financial liberalization in recent years, yet there are still considerable contrasts between their experiences. Portuguese banking has been undergoing a huge ownership transformation as the banks that were nationalized in 1975 have been returned to private Portuguese ownership. On the other hand, the continued dominance of State-controlled banks in Greece makes that financial system still rather inward looking, despite considerable liberalization. Membership of the EU has implied for these countries a long-term commitment to the inevitable process of increased financial integration in parallel with increased economic integration more generally in Europe. Though the impact of this on the financial system has been indirect, in that it is the domestic liberalization measures which have had the most dramatic effect, nevertheless it is clear that this liberalization was rendered inevitable by the EU-wide trends. The benefits of the reforms have been notable: an improved quality of banking services and lower costs. At the same time both countries have escaped severe macroeconomic instability and the impact on the budget has been absorbed through tighter fiscal policy. Banking profitability was 1 improved by the removal of quasi-fiscal impositions, but the initial strong improvement has been partly reversed by increased competition. So far it does appear that the newly liberalized banks have not succumbed to the errors of poor management or euphoria that have caused banking crises in other countries. There is greater awareness on the part of the authorities of the need to focus bank regulation on prudential issues, underpinned by the EU-wide adoption of Basle-type capital adequacy rules. Although entry of foreign banks is no longer being restricted, the share of the foreign banks in total banking remains small and rather specialized. Admittedly, the access of foreign banks to acquire stakes in the banks being privatized in Portugal was restricted, but overall this experience should do much to assuage fears of a rapid domination of banking by foreign interests when banking markets are opened to foreign competition. 2 2 The Background: European Financial Liberalization Domestic and international financial liberalization have proceeded in parallel in the European Union. From the late 1970s steps have been taken in most EU countries to liberalize domestic financial markets. At the same time two important strands of legislation at EU level forced the pace of internationalization. The first was the progressive removal of exchange controls within the EU, the second was the single passport approach to freeing-up international trade in banking and other financial services. Finally, the prospect of a common currency, which duly came into effect for eleven of the fifteen member states on 1 January 1999, has been an additional force making for an increasingly seamless financial market in Europe ­ even though the reality falls somewhat short of the goal. 2.1 Removal of Exchange Controls and the Exchange Rate Mechanism Although all EU members have long accepted the conditions of Article VIII of the IMF Articles of Agreement and as such do not restrict current payments either within the EU or with the rest of the World, all of them retained restrictions on capital movements to some extent or other until at least the late 1970s. The 1979 abolition of all capital controls in Britain under the Thatcher administration was a landmark here. Other member states retained capital controls throughout most of the 1980s, and these were seen by many as a largely necessary part of the authorities' toolkit in dealing with speculative attacks against the ERM. This perspective appeared to have been confirmed in September 1992 when a huge wave of speculation drove sterling and the Italian lira out of the ERM. Of course, to some extent the problem lay in a perception that macro-policies being conducted by these countries were not consistent with the exchange rate peg. Nevertheless, the experience reinforced the drive towards a common currency.1 1 From August 1993 the ERM adopted extremely wide plus-or-minus 15 per cent margins, and these remain in effect under the so-called "ERM-II" arrangement for Greece. These margins were not tested by speculative attack, presumably because the pegs were regarded as tenuous and because would-be speculators were conscious of the risk of a large movement in the opposite direction. 3 A major effect of the abolition of exchange controls is to remove the ability of the authorities to fix bank interest rates at any rate other than that determined by market forces which tend to induce a convergence of risk-adjusted rates of return on different currencies. The potential for taxing interest income from bank deposits also declines as such tax can often be avoided through capital outflows or other forms of asset substitutions (as became evident when German depositors responded to the imposition of a withholding tax on deposit interest by moving their funds to Luxembourg, thereby forcing repeal of the tax). For the banking sector the removal of exchange controls has the effect of introducing an important degree of competition in that funds can be moved or borrowed abroad to take advantage of banking services provided there. For large, established and well-known enterprises borrowing in major world currencies this route is important. Nevertheless, exchange rate risk and information and transactions costs continued to provide barriers for smaller enterprises and individuals, especially to the extent that they would require banking services to be denominated in less widely used local currencies. The single banking passport and the single currency were seen as the vehicles for removing these barriers. 2.2 The European Single Market Project Although long-described as a "Common Market", the EEC (subsumed in the EU from 1993) provided for freedom of trade only in goods, with services being largely governed by national regulation which, in the case of banking and many other financial services, in practice entailed significant restrictions or at least costs to entry. The result of this segmented EU market in banking services was a wide range of prices for different banking services, reflecting differing efficiency, competitive structure and regulatory costs. The 1988 Price Waterhouse study of eight countries showed very substantial differentials for some services and calculated that if these differentials (above the lowest cost country for each service) 4 could be substantially removed by effective competition the gains in terms of consumer surplus would be as much as 1½ per cent of EU GDP. Of course this begged the question as to whether effective competition would indeed remove all such inefficiencies, as the costs of acquiring information could make it difficult for new entrants to undercut incumbents even in the high-cost countries ­ and subsequent analysis suggests that the full benefits envisaged remain far from being achieved ­ but it does indicate the scale of the economic inefficiencies that existed.2 One possible approach to removing the barriers to trade would have been to establish a supranational regulator of financial services. However, in the Single European Act, adopted in 1986, the EU adopted an alternative and much more effective approach, namely the principle of mutual recognition. According to this principle, often known as the "single passport" approach, each member state must recognize the regulatory provisions of any other as equivalent to its own. Banking was one of the first sectors in which the approach was implemented, mainly through the Second Banking Directive, adopted in 1989 and in effect since 1993. As well as laying down basic principles for banking regulation, this Directive defines a broad list of banking activities (including investment banking). Any credit institution authorized to conduct any of the listed activities in its home country is allowed to conduct the same activities in all member states. Such business can be carried out by a branch of the bank located in the host country, or across international frontiers; it is not necessary to incorporate a subsidiary in the host country. The mutual recognition principle avoids the need for detailed central legislation, but it does raise the risk of regulatory competition within the basic principles established by the Second Directive. Increasing awareness of the prudential risks of under-regulation has thus led to further Directives strengthening minimum standards, and in particular implementing for the EU the Basle principles. 2Useful references among the large literature covering European banking liberalization include European Commission (1988, 1997), Begg (1992), Edey and Hviding (1995), Gual and Neven (1993), Thomson and Taylor (1994), Vives (1991). 5 Bankers responded in a variety of ways to the threats and opportunities afforded by the single passport. Anticipatory action by several banks involved acquisition of strategic cross-border stakes; another approach, implemented by Credit Lyonnais, was to make more aggressive acquisitions with a view to establishing an important retail presence in most EU markets. But, as had been predicted by experts, cross-border provision of banking services, as opposed to changes in shareholding, remained largely confined to the wholesale end of the market, or the traditional form of accepting deposits reflecting funds trying to avoid the costs of financial repression at home, as well as in search of better quality financial services. Undoubtedly the high cost for newcomers of information acquisition as well as of brand-name and confidence development has remained a significant barrier, albeit a natural one, not imposed by government action. The natural barrier arising from the heavy fixed costs involved in establishing a network of branches may also be placed in the same category.3 It is widely believed that introduction of the single currency from 1999 will have significantly reduced barriers to cross-border banking in participating countries. Although the substantial information technology costs of the changeover to the new currency will offer an opportunity for large banks to introduce software that may help them provide services across borders, the general expectation is that this will remain a gradual development. But there will be new opportunities to exploit economies of scale in financial markets more generally, offering the prospect of markets gaining ground over banks (notably as a result of securitization) in what is still an "over-banked" continent (Prati and Schinasi, 1997). The new wave of large-scale banking mergers that got under way in 1999 has so far been more between banks in a single currency rather than involving cross- border activity. 2.3 Liberalization of domestic controls on financial intermediaries 3There is an additional fiscal barrier to cross-border lending into Portugal in the form of a withholding tax (Folkerts-Landau et al., 1997) 6 Although it has occurred in parallel with the removal of exchange controls and the introduction of the single passport, the process of liberalization of domestic monetary controls has been essentially a domestic programme, and not one determined by EU-wide legislation or policy. In the 1950s and 1960s, most European countries had heavily regulated banking markets. Restrictions on the lines of business which banks could conduct and on the opening of branches resulted in national banking markets which were heavily segmented. Interest rate controls, either government imposed or operated by officially sponsored cartels, limited the ability of banks to engage in price competition. Credit ceiling restricted the expansion of credit, while selective credit controls often favoured the provision of credit to priority sectors. While these controls may have helped ensure the financing of sectors seen as having high priority in post-War reconstruction in Europe and may have protected the profitability of banks, the authorities in many countries became concerned about the stifling effects of such controls on competition, and also about their long-term impact on efficiency. But above all, it was the emergence during the 1960s and 70s of new financial institutions, not subject to the full rigour of controls, which rang the death-knell for controls. These new near-banks, often created precisely to bypass the restrictive regulations, filled gaps in the provision of financial services (such as real estate development finance and consumer credit), but they undermined the competitiveness of the controlled banks and ultimately made those controls unsustainable. The growth of offshore banking centres such as the Cayman Islands and Luxembourg, and the increasing ease of communications with them were additional factors. It was actually Germany that was the first to begin dismantling banking controls, but it was soon followed by others. Liberalization progressed rapidly in the UK, which already had the most developed financial market in Europe at mid-Century. The turning point there was "Competition and Credit Control" (1971) a policy which, by moving away from bank-by-bank credit ceilings and 7 from the setting of bank lending rates on a quasi-cartelized basis, aimed to improve the competitiveness of the banking system and its ability to innovate. Although in outline the process of deregulation appears quite similar from one country to another, structural differences persisted in EU banking markets, partly reflecting traditional business practices and partly the state of development of the non-bank financial sector. Thus, though UK banks added mortgage lending to their traditional focus on short-term self-liquidating commercial loans, the sophistication of the UK equity and bond market meant that UK banks made comparatively little inroads into longer-term business finance, whereas German and French banks lost little of their previous appetite for such activity. Further convergence of the style and strategy of the major European banking systems has presumably awaited the completion of the single market process in non-bank financial markets4 and the arrival of a single currency. We are thus very far from the emergence of any new standard pan-European banking structure, and the fears of some that a few dominant institutions would quickly assume a continent-wide role have not materialized. Indeed, although the process of liberalization has removed or reduced explicit and implicit taxation on banks, and has contributed to a lowering of their cost base, it is clear that it has also reduced bank profitability on a long-term basis by increasing competition. The size of the balance sheet of the banking and financial system has expanded greatly in response to liberalization. This reflected the fact that banks did not just increase intermediation as between different firms and individuals, but also provided greatly increased maturity and risk transformation services for firms and individuals tailoring the maturity and risk characteristics of their portfolio to their needs. 4 Taxation differences and regulatory impediments, as well as certain ambiguities in the interaction of different legal structures are among the remaining barriers. 8 Among the threats presented by the liberalization of banking are macroeconomic destabilization resulting from a surge of credit, especially to the household sector, and microeconomic problems of bank insolvency. European macroeconomic instability of the 1970s was primarily caused by the interaction of non-banking factors, especially the two major oil-price rises and the collapse of the Bretton Woods system of fixed exchange rates. But there is also evidence of an initial surge in aggregate credit following liberalization as banks responded to a pent-up demand. This pattern is by now well understood, and can be offset by monetary policy actions using the indirect means of control still available to the authorities. The microeconomic threat of bank insolvency also needs to be taken very seriously. Since the 1970s all EU countries have experienced at least some increase in the incidence of bank insolvency. Certainly the pre-liberalization era was not so vulnerable to overt bank failure, though the costs of inefficient banking were substantial and largely hidden. Early problems in the Netherlands and Germany were associated with the worldwide jump in interest rates in the late 1970s. There have also been a number of instances where connected lending in industrial-financial groups caused failure. But most episodes of widespread bank distress in Europe have been associated with unwise herd behaviour in lending for property development during the boom phase of an unsustainable boom-and-bust cycle. Although common patterns can be detected, it is noteworthy that international spillovers of solvency crises have been absent. Even the very similar Nordic banking crises were linked more through a demonstration effect (whereby banks and regulators in neighbouring countries each imitated the others'errors) than through any direct contagion. In those countries where there has been substantial state and regional government ownership in banking liberalization has typically been accompanied by a trend towards privatization. Likewise, although mutual ownership of banks and near-banks remains important in several EU countries, the 9 trend here (faster in some countries than in others) has been towards corporatization. In this, the banking sector is following a wider move towards less government intervention, rather than setting the trend. In particular, there are no EU requirements to privatize banking institutions. Although growing internationalization, the removal of exchange controls and the single passport have influenced the development of domestic banking policy, overall, the transformation of European banking in the past quarter century has owed much more to domestic liberalization than to coordinated action at EU level. 3 Impact on the Peripheral EU Countries Most EU countries are in the highest income bracket, with per capita GNP upwards of 60 per cent of the US figure. This tends to limit the relevance to developing countries of their experience with financial market liberalization. However, the two least prosperous members, Greece and Portugal, joined the EU with inherited banking systems that had many points in common with those in developing countries. It is therefore of considerable interest to track the evolution of their systems during the past decade of liberalization and internationalization. In both countries banking groups retain dominance in the financial system, both directly and through their control of other financial entities in insurance and mutual funds. Accordingly the main stories can be learnt by examining the liberalization process as it affected banking. In 1984, the banking systems of both countries operated under a regime of high reserve requirements and binding quantitative credit controls. These had the effect of channelling a substantial fraction of the resources mobilized by the financial system either to the Government itself or to preferred categories of borrower, at low interest cost. Most of the banks were Government-owned or controlled, and in the case of several of these a large volume of doubtful and non-performing assets had accumulated in the loan portfolio. Interest rates were also subject to administrative control, contributing to the net effect that the banking system as a whole was 10 unprofitable and weak in its capital structure. Both financial systems operated behind exchange controls, which indeed were a necessary condition for persistence of such distorting regulatory regimes. The rapid dismantling of the regulatory banking controls, the liberalization of entry and, at least in the case of Portugal, the privatization of most of the Government-owned banks have brought the Greek and Portuguese financial systems much closer to the norm of more market-driven financial systems which had earlier become established in the rest of the EC. Adaptation of the systems to the new regulatory environment is still under way. At present the most dramatic developments for Portugal have been the ownership restructuring. In Greece the largest banks are still Government- controlled, and, while entry by new private participants has been vigorous, modernization and technological convergence rather than changes in ownership structure and competition have been the key features of the Greek financial system changes. With capital adequacy requirements substantially harmonized by legislation (directive) on an EU- wide basis, and with the inevitability of a degree of regulatory competition, there are now but relatively minor differences in the regulatory structure for banking in both countries (cf. Barth et al., 1997). Both have pursued the universal bank model, with permission both for banks to make investments in non-financial firms and for investment by non-financial firms in banks (though in this respect Greece remains more restrictive than Portugal). Both allow securities and insurance activities, but there are line-of-business restrictions inhibiting real estate activities. Both have recently introduced deposit insurance schemes on a compulsory basis: that in Greece covering the first ECU 20,000, that in Portugal fully covering the first ECU 15,000 and a portion of the next ECU 30,000 per depositor. Prudential supervision is the responsibility of the central bank in each country. These institutions have moved to bring their prudential regulations into line with evolving EU standards, and to modernize their supervision practices. Both conduct on-site examinations, generally every 2-3 years 11 in Greece, annually in Portugal. Enforcement actions are made public in Portugal, but not in Greece. Overall, there has not been much monetary deepening - indeed, as shown in Figure 4, in many respects the initial post-liberalization loss of financial depth, measured by private sector bank credit or by money, has not yet been fully reversed.. (But financial deepening, defined more widely to include non-bank financial liabilities did increase in both countries. To some extent, this pattern reflects a declining share of banking relative to other intermediation, including the use in Greece of tax-driven near-bank instruments, a topic beyond the scope of the present paper). The following sections review the experience of liberalization in Greece and Portugal.5 Beginning with a note on the general economic background, for each country we provide (1) an overview of the structural reform process, followed by (2) an examination of trends in ownership, showing the very limited role of foreign ownership. Then we discuss (3) the degree - so far limited - to which liberalization has contributed to an expansion of the banks' international business. Turning to interest margins and profitability we show (4) that both countries have experienced a transitory increase in bank margins and profits, followed by some narrowing, under increased competitive pressure. Liberalization has made achieving macro stability more difficult, but this task has been achieved (5). We conclude (6) with overall remarks about policy issues arising. 6 5This can be modeled formally though, given the uncertainties, such modeling could hardly overturn the interpretation advanced here. 6In addition to the quoted references, and reports of the national central banks, etc., the account draws on extensive interviews conducted by the author with senior bankers in Athens and Lisbon during 1995. 12 Greece 4 Greece: Dismantling Pervasive Government Involvement 4.1 General Economic Background Greece is the poorest of the EU countries; by 1995 its per capita GNP was put at about 42 per cent of the US level, a figure which it had not advanced in a decade. Greece joined the EU in 1981, shortly after the restoration of a democratic Government following the period of military rule. The population of Greece is just over 10 million. With exports of goods accounting for less than 13 per cent of GDP, it may be considered a relatively closed economy; nevertheless, tourist revenues and invisible earnings from shipping are extremely important, and go a long way towards making up for the low export of goods, though there has been a current account deficit in the balance of payments through most of the 1990s. The only EU country without a land border connecting it to the rest, Greece's economic and financial structure too has remained somewhat apart from that of the Union. Inflation in Greece has long been the highest in the EU, exceeding 20 per cent as recently as 1991; 1995 was the first recent year which saw inflation back into single figures, and the subsequent trend has been good with inflation under 5 per cent in 1998. Until quite recently Government borrowing remained stubbornly high - over 16 per cent of GDP in 1990 and 14 per cent in 1993. However, a determined effort to meet the budgetary criteria for EMU membership is at last bearing fruit, and the borrowing was below 3 per cent of GDP in 1998. 4.2 Structural Reform in the Financial Sector: Overview The dramatic structural changes in the Greek financial system since the mid-1980s have transformed what was a most distorted regime, with an elaborate structure of controlled interest rates, compulsory redeposits and lending coefficients, into what is now a fairly transparent and market-driven system. (Cf. Antoniades and Kouzionis, 1989, Papademos, 1991, 1992). 13 Greece As an illustration of the regulatory changes, in 1985 portfolio and reserve requirements restricted the allocation of 78 per cent of bank deposits, a figure which by 1993 was down to 9 per cent reserve requirement at the central bank (with average remuneration on those deposits at about 10 percentage points below the market cost of funds).7 Interest rates are now completely freely determined by the banks, many of the restrictions on lending rates having been removed as early as 1987, with the final restriction (a floor on the savings account rate) removed in March 1993. Despite deregulation, the system is still dominated by a heavily-concentrated banking system (the subsidiaries of whose main players are also important in the insurance and fund management businesses). The largest institutions are still state-controlled, and though they have been gradually losing market share, they still retain an enormous dominance with about four-fifths of the market of both deposits and lending as late as end-1993, and about 60 per cent in 1998. They were said in the mid-1990s to be still subject to some government pressures on their activities, limiting their autonomy.8 These state-controlled banks inherited a heavy burden of non-performing loans and over-valued equity holdings, resulting from bad credit decisions, often effectively directed by government. Their market dominance into the 1990s allowed them to charge high lending margins in the 7More recently a tightening of monetary policy in 1995-97 has involved an increase in reserve requirements, though only to 12 per cent. Because of the below-market remuneration of required reserves, they impose a reserve asset penalty equivalent to about 100 basis points on credits. In addition, an annual 1.2 per cent commission is payable to the Bank of Greece on most lending (recently extended to include foreign currency lending). A further addition of another 100 basis points or so to the cost of credit is imposed by the turnover tax on bank interest receipts (known as EFTE) which is added by the banks to the quoted interest rate. The rate of EFTE was been reduced in 1997 from 4 per cent to 3 per cent of interest payments. 8These banks labour under a rigid salary structure, and senior management is politically appointed (even to the extent of the top management changing with a change in Government). The strength of the bank employees'union has made it difficult to achieve efficiencies. In addition heavy pension burdens resulting from very generous pension plans (now reformed) add to the operating expenses of these banks. 14 Greece deregulated environment keeping them afloat despite the impaired portfolio. The new entrants, individually and collectively much smaller, expanded rapidly, the high interest margins of the market leaders allowing them to make very profitable lending in niche markets without aggressively challenging the leaders in terms of market share. Foreign banks played only a small part in this process.9 (). It was only during the 1990s that the authorities acted to recapitalize the main banks, which it did through a combination of bond issues and debt assumption totalling a cumulative 7½ per cent of one year's GDP by 1997. While the largest bank is now restored to full capital adequacy, at least one other large bank will still need financial injections, to meet the cost of writing-off unrecoverable loans, mainly to non-financial state-owned enterprises. Operating efficiency at the state-controlled banks is still generally acknowledged to be much lower than at the private banks,10 and considerable retrenchment is still envisaged. Financial liberalization (especially of nonbanks) has helped the financial sector to expand. The explosion in the range and volume of assets and transactions is one measure of this, but perhaps even more telling is the growth in real activity as hinted at by employment. Although, at 2.3 per cent in 1994 (including nonbanks), employment in the sector still represents a smaller share of the total than in any other EU country, this share has grown steadily since the early 1980s, when it was as low as 1.5 per cent. This expansion in the face of liberalization certainly suggests that the Greek financial system is making an increased contribution to the economy. Although interest margins widened at first, a recent narrowing also confirms that the banks' customers are benefiting. 4.3 Concentration and ownership; foreign entry 9 With the exception of EFG-Eurobank which, controlled by expatriate Greek interests (Latsis family), might be seen as quasi-domestic 10Indeed, it is mainly in profitability, and in such measures as the volume of private lending per staff member that the relative efficiency of private banks has been measured. Return on equity for State- controlled banks was in the region 8-10 per cent in the late 1990s, compared with 30-40 per cent for listed private banks. The overhead expenses of the private banks expressed as a share of total assets is not notably or consistently lower, presumably because of the different mix of business that they conduct. 15 Greece State control of the largest institutions has had the effect of giving Greece a very highly concentrated banking system. The banking system has traditionally been subdivided into a commercial banking sector and the specialized credit institutions, the former accounting for about two-thirds of assets, and rather less than three-quarters of deposits (Table 3). Three of the four largest commercial banks are state-owned,11 and the four-bank concentration ratio in this sector at end-1993 was 82 per cent, if measured on deposits, and around two-thirds if measured on assets or branches. But because the state controls the largest institutions,12 effective concentration is considerably higher than indicated by these ratios. The market share of state-controlled banks declined only gradually from 88 per cent in 1985 to 79 per cent in 1993 (measured by deposits). The decline was largely due to the financial difficulties of some of these state-owned institutions, which has inhibited them from competitive pricing and resulted in a gradual but steady drain of business to established and to more than a dozen new private Greek-owned banks. These have doubled their market share since 1985 to about 13 per cent, and most have proved to be extremely profitable, a situation which is evidently a corollary of the high margins being charged by the state-owned banks. Privatization of some smaller institutions and a further loss of market brought the state-controlled 11 National, Commercial and Ionian. The last-named is a subsidiary of the Commercial, but will soon be privatized.. The Alpha-Credit and EFG Eurobank are the largest of the private banks, the former being an old-established institution, the latter a new-comer which recently absorbed inter alia what had been a larger private bank, the Ergo. In early 1999 it was announced that the Alpha-Credit would acquire the privatized Ionian. EFG Eurobank is part of the wider international group controlled by the Latsis family; in 1998-99 it sold a 10 per cent stake to Deutsche Bank, and also launched an IPO on the Greek exchange. 12 A peculiarity of the ownership structure of the banking system is that the controlling interest of the state in so many banks derives largely from its ability to vote the shares of the social security and pension funds who are the registered owners of many of the shares. These shares are quoted and actively traded on the Athens Stock Exchange, and recent liberalization of pension fund investment restrictions apparently open the door to a more flexible management of these funds. Although the existing fund managers have no intention of selling, it may very well be that the days of state-control of the largest banks are numbered even without any formal privatization. 16 Greece banks' share to about 60 per cent by 1998. Further privatizations of medium-sized banks are in train. The apparent persistence of high banking margins might have been expected to result in substantial entry and expansion by foreign banks. In the event, the number of foreign banks established as subsidiaries or branches in Greece remained little changed at around 20 from the mid-1980s to the mid-1990s. It is generally supposed that, as elsewhere, information barriers have prevented the foreign-owned banks from making greater inroads. Furthermore, the comparative lack of foreign interest in the privatizations is said by market participants to reflect foreign banks'reluctance to deal with labor market rigidities. Most of the foreign banks are specialized in traditional international bank business.13 Despite the relative decline of the state-controlled banks, Bank of Greece data (Table 3) indicate that foreign banks enjoyed by 1997 only a modest increase to 9 per cent in market share of deposits, with the same figure for loans. Foreign banks'share in total assets (over 15 per cent by 1997) has always been larger, apparently reflecting their activity in intermediating external funds, and placing them in the interbank market. 4.4 Foreign business of the banks Greek banks are not very active in foreign lending. Indeed, their foreign asset holdings, fluctuating between 4 and 7 per cent of GDP, have been relatively smaller than those of any other EU country, and almost all of the lending is interbank. On the other hand, Greek banks do have a much larger volume of foreign deposits. These reached a peak of 26.5 per cent of GDP in 1989 and slipped only slightly thereafter. Over 90 per cent of these deposits were from nonbanks, with the result that Greece ranks third in the EU, after Belgium-Luxembourg and the UK, for cross-border deposits by nonbanks. 13 For example, Barclays target the shipping sector. Only Citibank is really active in retail segments, including mortgages. 17 Greece Two of the state-controlled banks, the National and the Commercial, have branches in foreign locations with a large expatriate Greek population (e.g. France, Germany, South Africa, UK, US); these branches are aimed at the Greek element, and have very little involvement in the foreign markets as such. Several of the new private banks have also entered this market (e.g. London). Another new area for expansion is in the former planned economies of nearby Eastern Europe. Subsidiaries or affiliates of Greek banks have already opened in several countries in the region, notably Romania and Bulgaria. An interesting illustration was the offer by the National during the Kosovo crisis of 1999 to acquire a majority stake in the largest bank in FYR Macedonia, a bank distressed from depositor outflows.. 4.5 Interest Rates and Bank Margins Quoted rates Spreads between quoted deposit and lending rates in Greece are higher than anywhere else in the EU. Various reasons may be enumerated. First, the burden of fiscal and quasi-fiscal impositions on the banking system.14 Second, the fact that quoted lending rates may be misleading in that they do not take account of discounts for prompt loan servicing, a phenomenon which is not commonly encountered in other EU banking markets. Third (as mentioned above) the pricing policy of some of the large state-controlled banks in maintaining high lending rates on performing loans in order to ensure adequate cash flow. 14 These have been greatly reduced but, as mentioned above, still remain relevant. As late as early 1993, the effect of taxes, reserve requirements and other quasi-fiscal impositions on the banking system was estimated at 600 basis points, though a part of this was not included in the quoted interest rates used in official statistics. 18 Greece The largest component of the banks'funding sources are savings accounts,15 and therefore the rate on savings deposits is the most representative deposit rate. Short-term lending by the banks is rather larger than long-term lending and so the short-term lending rate is taken as the most relevant lending rate. (Table 2) The quoted spread between the savings deposit rate and the short-term lending rate jumped from a regulated 5.5 per cent in the mid-1980s to 8.0 per cent at the end of 1987 and to between 10.4 per cent and 11.5 per cent in the early 1990s (exceptionally it exceeded 13 per cent during the currency crisis of 1994). Subsequently, the gap fell back to 8 per cent by the end of 1995.16 Figure 2 shows the spread relative to the time deposit rate, revealing the same pattern. Although the sharp contraction in spreads during 1995-96 was not fully maintained, 1998 quoted spreads were well below the earlier peaks Note that the substantial holdings of Treasury bills and bonds implies that rates on these instruments are also important. The relative stability of nominal rates on Treasury bills as compared with other rates can be explained by the Government's reluctance to borrow domestically at very high rates. Accordingly there has been a tendency for the Government to switch its borrowing abroad rather than see Treasury bill rates rise. At first, therefore, financial market liberalization appears to have resulted in higher, rather than lower, quoted interest spreads. Nevertheless, heavy reliance on foreign currency by corporate borrowers since mid-1994 represented the beginnings of a change. With the removal of exchange controls aggressive competition between banks seeking to onlend foreign exchange to the best 15 They represent 73 per cent of private deposits in M3. Other important sources of deposit-type funds outside M3 are repos and foreign-exchange denominated deposits; some of the latter are redeposited with the Bank of Greece where, until recently, they enjoyed a Government exchange rate guarantee. 16 The interbank rate reacts more quickly to market forces than either the deposit or lending rates, and we therefore find a negative correlation between the spread of lending rate over interbank and the spread of deposit rate below interbank. 19 Greece borrowers resulted in a dramatic narrowing of spreads.17 Effectively competition was happening through the back door of foreign currency-denominated lending intermediated by domestic banks. It was not long before the spreads on domestic lending shrank in line. Although the influence of business cycle, the general level of interest rates, and capital flow surges can strongly influence quoted interest rate spreads, our interpretation of a long-swing in competitive conditions following liberalization with underlying spreads first widening, then shrinking, is strongly suggested by the data (and is not out of line with the opinion of market observers). We hope in the near future to report econometric results throwing more light on the matter. Bank margins and profitability Quoted interest rates do not indicate the overall average net interest margins earned by the banks. For this we must turn to bank income and expenditure accounts, though these too may be subject to question, especially in regard to the accrual of unpaid interest on non-performing loans, a practice which was prevalent up until recently in Greece.18 The OECD statistics provide a long-time series for the largest commercial banks (Table 1, Figure 1). These show a general decline in net interest margins until 1987, followed by a sharp rise to 1991, thereby confirming the evidence from quoted interest rates of greatly enhanced interest margins since liberalization. Non-interest earnings also grew vigorously, while staff and other operating expenses declined both in relation to the balance sheet total and as a share of GDP. This allowed a substantial increase in before-tax profits, despite higher provisions. Subsequently profitability fell back sharply in 1992-94, though not to the low levels recorded in the late 1980s. 17For a while, bank spreads on domestic currency loans were far higher than those on foreign exchange loans. The same borrower could be offered an interest rate on a foreign currency loan embodying a spread above cost of funds some 300 basis points lower than for a drachma loans. (Note that this 300 basis point gap is on the spread above cost, and thus not explicable by exchange rate risk.) 18 Here the data is complicated by the accounting treatment of special bank recapitalizations that took place to meet the consequences of under-provisioning. 20 Greece Most of the private banks are highly profitable, a fact that is illustrated by the stock market valuation of the private Alpha-Credit bank being - by the mid-1990s - higher than that of the National bank despite the latter having a balance sheet some six times larger. 4.6 Financial Liberalization and Macroeconomic Stability The removal of exchange controls was accomplished more smoothly than had widely been anticipated. Under EU law, like Portugal, Greece had a derogation to delay the final removal of exchange controls until 1995, though all but short-term capital movements had been freed by May 1993. During 1994, outward capital movement accelerated substantially, apparently for fear that the final removal of exchange controls would be accompanied by a downward adjustment in the exchange rate. In the event, the removal of exchange controls was brought forward to 16 May 1994 without any change in exchange rate policy This decisive action confirmed the authorities' commitment to stability and openness and the outflows were soon reversed. Indeed, capital inflows have since been flowing to the extent that they have become a destabilizing factor threatening to expand liquidity against the authorities'attempts to reduce inflation. Much of this inflow reflects foreign currency-denominated borrowing by Greek firms and households. Evidently, the more stable exchange rate and the much lower nominal interest rates in foreign currency has encouraged these inflows. This has presented the authorities with an increasingly acute dilemma. How to discourage excessive inflows while maintaining sufficiently high interest rates to maintain downward pressure on inflation. In the event, a number of technical measures have been adopted, but these have only been partially successful in limiting inflows. Although the inflows have been partly sterilized and have boosted official external reserves, the substantial shift into the use of foreign currency-denominated lending by banks could begin to present some prudential risks for the banks in the event of a change in exchange-rate policy (if borrowers proved unable to service foreign currency debt that had appreciated in value). 21 Greece In the event, the depreciation of the exchange rate slowed considerably since 1994 - to less than 2 per cent per annum in effective terms - at the cost of some loss of competitiveness, as the fall in inflation has been rather slow. Export growth appears to have been hit by this, and competitiveness worried made the drachma a target of contagious speculation in October 1997. Subsequently a managed depreciation on entry into the ERM in March 1998 was seen as restoring the loss of competitiveness, thereby paving the way for early membership of EMU. Incomplete liberalization also created difficulties of monetary control in another dimension as a result of differential taxation of different deposit-type instruments (a factor which contributed to limiting monetary depth as conventionally measured). Institutional and depositor response to after- tax differentials made such differential taxation increasingly unsustainable. At the same time growing use of complex derivative products also complicated the assessment of market risk assumed by intermediaries and firms. Nevertheless, research has shown that underlying monetary demand behaviour remained stable and predictable, implying that sufficient scope for the authorities remains to achieve adequate monetary control. (Ericsson and Sharma, 1996; Papaioannou and Gatzonas, 1997) 4.6 Policy lessons For long, successive Greek governments used the domestic banking market to bail-out insolvent companies.19 This was possible as long as the system remained substantially closed, with few independent Greek banks and a small foreign presence concentrated on shipping finance. But once Greece had committed to the EU single market process and removal of exchange controls the days of this regime were evidently numbered. 19For some details see, for example, the OECD Economic Survey - Greece (1998) 22 Greece Though the adaptation was long-drawn out, and has not yet been fully completed, the worst fears of the sceptics have been confounded. Removal of exchange controls was successfully advanced and the speculative flurry it induced soon died away - though there was another episode of stress in 1997. A vibrant private banking sector was the chief beneficiary of the liberalization process, with foreign banks remaining relatively passive despite the lure of high interest margins.20 Recapitalization and reform of the state-controlled banks - also entailed by commitment to EU rules on capital adequacy - has been costly, but in a real sense the costs had already been incurred, albeit not openly reflected in the banks'accounts. Interest rate margins and bank profitability rose at first following liberalization, but subsequently fell back. This pattern is often observed: removal of interest rate controls allowing banks to build up profits before competitive forces get properly under way. With the fall in margins less pronounced than in Portugal (discussed below), a higher fraction of the economic efficiency gains from the liberalization will have come in the form of better credit allocation. The bail-outs, documented above, are but one (imperfect and partial) way of measuring the scale of this prior misallocation. Macroeconomic and monetary management has been complicated, with the authorities having to cope with unwanted inflows as well as guarding against the risk of triggering speculative outflows. Yet at last fiscal correction and disinflation seem well under way thanks to government commitment and strong leadership by the central bank. 20Restrictions on the access of foreign banks to the domestic real estate lending market may also have been a discouraging factor (Folkerts-Landau et al., 1997). 23 Greece Implementation of prudential supervision is said to have been strengthened, and undoubtedly the introduction of EU legislation, although not yet fully implemented, has given a clear policy direction away from seeing the banks as an arm of government policy and towards a recognition of the need for them to be financially autonomous. Until very recently the Central Bank was far from independent, with a change in senior management automatically accompanying a change in government. Legislation was enacted at end-1997 to bring Greece into compliance with the requirements of the EMU so far as Central Bank independence is concerned. Though directed chiefly at the achievement of macro stability, such independence also serves to further demarcate the banking system from government interference. 24 Portugal 5 Portugal: Nationalization and Reprivatization 5.1 General Economic Background Portugal joined the EU along with Spain in 1986, some ten years after the democracy had been restored following the Salazar years. Portugal's economic performance since accession to the EU has been quite strong, though it faltered in 1993-4. Its per capita GNP jumped from 41 per cent of the US figure in 1987 to 48 per cent a decade later, easily surpassing Greece. The population of Portugal is 10 million. It is characterized by the lowest wage rates in the EU combined with high labour force participation and low unemployment. Exports of goods have accounted for a rapidly increasing share of GDP which reached 24 per cent in 1997. Textiles represent a large but declining share of these exports. Traditional economic links with Brazil and with former African colonies provide further contrasts between the Portuguese economy and that of the core EU countries. Nevertheless, the EU accounts for some four-fifths of exports, up from about two-thirds in 1986. Inflation in Portugal has been much higher than in the core EU countries, reaching over 13 per cent in 1990; however, it has been brought under control in recent years and was about 2 per cent in 1997, the qualifying year for the Maastricht criteria. The general government deficit has shrunk from 7 per cent in 1993 to well under the Maastricht ceiling of 3 per cent in 1997. The current account of the balance of payments has fluctuated around approximate balance over the past decade. 5.2 Structural Reform in the Financial Sector: Overview 25 Portugal Apart from three foreign-owned banks,21 the entire Portuguese banking system was nationalized in 1975 and operated for the next decade or so in a regime of directed credit and binding credit ceilings, controlled interest rates, heavy reserve requirements and no entry (cf. Borges, 1992, 1993, Macedo 1990, 1993). In addition to the control dimension, the effect of many of the regulations was to impose a heavy quasi-tax on financial intermediation (which had the effect of reducing the need for explicit government borrowing). The past decade has seen this regulated regime almost entirely replaced by one which approximates that in most other EU member states in respect to openness and deregulation, and where the state's ownership share declined rapidly from more than 90 per cent to around one-quarter by 1997 (Table 5). The main stages in this liberalization have been (i) the admission of new banks from 1983 (this required a revision of the constitution), beginning with four private Portuguese-owned banks and six foreign owned. (ii) privatization of the nationalized banks (requiring a further constitutional revision) which began in 1989; (iii) elimination of interest rate controls (which were removed on lending by 1988 and on deposits by 1992); of credit ceilings in 1990; and of other controls such as on branching and on new products, and the progressive phasing-out of reserve requirements.. The response of the re-privatized financial sector has been very vigorous. The number of banks jumped from 18 in 1984 to 44 ten years later and to 62 by 1997; the number of bank branches more than doubled (from less than 1500 in 1984 to 3400 ten years later), though bank staff numbers only increased modestly. Indeed, as a share of GDP, staff costs declined quite sharply at first, but recovered somewhat in the 1990s, reflecting the conflicting pressures of increased efficiency and increased services. New products and new technology have been introduced, and the banks 21 These were Credit Lyonnais, Bank of London and South America (Lloyds), and Banco do Brasil. 26 Portugal expanded credit liberally - arguably too liberally, boosting property prices especially in Lisbon in the late 1980s (Decressin and Mauro, 1998, discuss the credit boom noting that it went mainly to households). The banks remain the dominant players in an increasingly diverse financial sector which overall, despite evidence of overstaffing in many institutions, continued to increase its share of total employment, at least until 1994. While the rapid growth of the Portuguese economy during 1986 to 1992 (along with relatively rapid inflation) helped the reprivatized banks put any undue legacy of bad and doubtful loans behind them, they still inherit excessive staff levels and costs. The recent wave of mergers (discussed below) was seen as a basis on which economies can be effected. 5.3 Concentration and ownership: foreign entry From 1983 the possibility of establishing new private banks (foreign or domestic) was reopened. Since the programme of re-privatization began in 1989 the ownership structure and concentration of the banking system has been in rapid flux. By 1995, almost two-thirds of the State's banking holdings had been sold, and by 1997 only the largest bank CGD remained in the State's hands.22 In measuring concentration in Portuguese banking it is important to distinguish between concentration as measured relative to distinct banks and concentration of control in groups. At first, after the privatizations, concentration was not high by European standards. In 1994 the market leader CGD had about 23 per cent of the deposit market and the four-bank ratio was about 51 per cent, unchanged from 1987. Looking at the consolidated position, the four-group ratio was 62 per 22 The CGD was already State-owned before 1975, when it was primarily a savings bank (there is no postal savings bank). 27 Portugal cent.23 At that stage only one of the newly established private banks24 was in the top five, the other three25 being re-privatized banks. However, a series of mergers among large institutions in 1994-95 resulted in two large private banking groups challenging the CGD's position as the largest bank. After the merger, the four- group concentration ratio jumped close to 80 per cent (based on deposit-type resources - Table 5).26 There has been some discussion of the possibility of further mergers leading to just two large private banking groups dominating. Although there is almost a score of foreign-owned banks now operating in the market, their aggregate share by 1995 had reached only about 9 per cent of loans and about 6 per cent of deposits, and they achieved very little further advance by 1998 (Table 5). Some of the foreign banks were reportedly rather unsuccessful at first, with bad loan-loss experience. Some are active in retail banking, despite the fact that rather high charges were imposed on new banks wishing to open branches. None has more than about a 1½ per cent market share. Most of the foreign banks are EU-based, the largest being the Spanish Banco de Santander; US-based banks have about a one per cent market share between them. Some observers have expressed the opinion that the authorities would resist the sale of a large Portuguese bank to a foreign concern, and it is certainly true that 23 Whether this is larger or smaller than in 1987 depends on whether one consolidates the whole of the government's ownership into one group. Looking at the separate legal entities, C4 (based on loans) was 53 per cent. But at that time most of the banks were government owned, so if the government-owned banks are taken as a single group, C4 for 1987 jumps to 97 per cent. 24 The fifth-ranked BCP, which acquired a controlling interest in second-ranked BPA during 1995. 25 BPA, BTA and BESCL in second, third and fourth place respectively ranked by consolidated balance sheet aggregates. As noted, BPA came under the control of fifth-ranked BCP, while seventh-ranked BPSM acquired (from the Spanish bank Banesto) a controlling interest in BTA. 26 And C3 rose to 70 per cent. 28 Portugal concern was expressed when it was revealed that Banesto had surreptitiously acquired control of the second largest bank, BPA. Indeed, Banesto subsequently had to divest itself of this control, as it was held to be in violation of the rules established for the reprivatization of banks. An alternative interpretation, however, is that the political opposition in that case was solely to the acquisition by a foreign concern at reprivatization, and that no such reaction would occur in the case of a market transaction. 5.4 Foreign Business of the Banks Before the removal of exchange controls in 1993, Portuguese banks had not been very active in foreign lending: their foreign assets had hovered around 10 per cent of GDP. But there has been a very substantial jump since then. The foreign presence of Portuguese banks takes four distinct forms. First, there is the network of branches operated by the larger banks in traditional emigrant destinations, especially France, where CGD alone has some 43 branches, but also in South Africa and Canada Second, a new expansion into Spain: CGD have acquired three banks with a total of 160 branches, mostly in parts of Spain near the Portuguese border; BESCL have also acquired a Spanish bank with a retail network. Third, traditional foreign banking activities (corporate and foreign exchange) in major centres including London, New York, Brussels and Geneva. Fourth, a relatively modest, though growing presence in former colonies in Africa, and Latin America, and in Macau. 5.5 Interest Rates and Bank Margins The main stylized facts about Portuguese bank interest rates and margins are easily interpreted in terms of the process of liberalization and increased competition. Following the liberalization of bank lending rates, combined with retention of binding credit ceilings in 1988, lending rate margins began to expand, a process which accelerated as the bank privatization process continued. After the lifting of credit ceilings in 1991, allowing competition between banks for lending opportunities, the 29 Portugal margins dropped rather sharply, and have continued to drift lower. Product innovation, including commercial paper and syndicated loans priced on an auction basis, has lowered borrowing costs for certain low-risk customers; household borrowers in the mortgage market have also benefited. These trends can, at least partly, be documented by reference to available statistics on quoted interest rates and on average bank margins (Table 4 Quoted rates Figure 3 displays the gap between the main bank deposit and lending rates 1987-98. The qualitative features described above are clearly evident in the plot. The gap between the lending and deposit rates in the figure reached over 1000 basis points during 1990-91, before declining sharply to the vicinity of what was still a high 600 basis points during 1992-96, and then again to about 350 basis points by late 1998. In addition to the regulatory changes already mentioned, bank interest rates will have been influenced by the cost of reserve requirements. Calculating the effect of reserve requirements on break-even lending rates is complicated by the fact that the requirements amounted to a two-tier system with a base quantity plus a fraction of deposits. At the end of 1994, the 17 per cent reserve requirements were replaced by a 2 per cent unremunerated reserve requirement. The remaining 15 per cent were converted to bonds, one-third of which bear no interest, the remainder being remunerated at a rate of interest close to market. Assuming a marginal cost of funds in the region of 9 per cent, the implicit tax amounts to about 80 basis points added on the interest that must be earned on other assets. In addition, a stamp duty, recently reduced to 7 per cent of the interest paid, 30 Portugal together with a smaller, one-off percentage on the capital value of the loan is payable by the borrower (over and above the quoted interest rate).27 Though quoted spreads may respond to the business cycle and to the level of wholesale interest rate levels, it is not easy in this instance to explain away the long swing in quoted interest margins from the mid-1980s to the late 1990s as a product of either of these factors. Both the frequency and the phase of the macroeconomic and interest rate cycles are quite different to that of the quote spreads, and no obvious significant regression relation can be detected. Bank margins and profitability Available data on bank profitability confirms the pattern of interest rate movements discussed already (Table 1). As shown in Figure 1, there was a definite improvement in profitability and a widening of average margins from 1985 to 1990/91. Indeed, average interest margins widened by about 2 percentage points in this period. Since then, there has been a narrowing of interest margins. There is now (as elsewhere) a move to unbundle costs and to make separate charges, which means that some of the decline in interest margins is being compensated for by an increase in other income. However, as a share of assets, profits too have fallen back from the record levels achieved in 1991. However in interpreting figures expressed as a ratio of total assets, it is important to be aware that the denominator (total assets) has been swollen in recent years by a disproportionate increase in gross interbank claims (some arising from intragroup transactions); these peaked in 1994 at over 31 per cent of the total balance sheet, but fell back to under 27 per cent by 1996. Certainly, non-interest sources of profit have been boosted: the share of net interest in operating income fell from 84 per cent in 1989 to 65 per cent in 1997. 27A fully comprehensive analysis of the time-varying effects of various quasi-taxes would take us too far afield, but would be desirable in a more comprehensive assessment of the determinants of interest spreads. 31 Portugal Non-price competition There has been a very strong increase in branching and in bank spending on advertising in the post- liberalization period. For example, branch numbers almost doubled in the four years 1988-92, from 1566 to 2839, and reached 3766 by 1996. Evidently this reflects increased non-price competition (Barros, 1995). The use of branching as a competitive device reflects the strong spatial nature of retail banking competition, and the importance of distance and other geographical aspects (Cabral and Majure, 1993). The relatively large number of significant banks and the public availability of considerable detail about their operations has allowed the recent emergence of a substantial body of econometric research on detailed aspects of the competitive behaviour of Portuguese banks. Though tentative, the conclusions of this work point to competition having increased more in the deposits market than in the market for loans (Barros and Leite, 1994) They also uncover evidence that, beyond a certain minimum scale, size continues to confer market power (Pinho, 1995). Thus, the banking market has remained far from perfectly competitive. The conclusions of the econometric research are thus generally consistent with the evidence from aggregate spreads and margins.28 5.6 Financial Liberalization and Macroeconomic Stability 28One interesting aspect of the research is the distinctions which are drawn between the behaviour of different classes of banks, notably as between foreign-owned banks and the rest. For example, foreign- owned banks appear to pay higher interest rates and higher non-interest costs in the attempt to establish an adequate market share (Pinho, 1995). Another curious finding is that private banks have expanded branches more in rural areas, whereas the remaining state-owned banks in the late 1980s expanded more in urban areas. 32 Portugal The quasi-taxes on the Portuguese financial system represented an important, though implicit, support to the budget over the years. Their removal made the task of attaining budgetary balance more difficult. Nevertheless it has been accomplished, and faster than many expected. The removal of exchange controls had the effect of exposing Portugal to the risk of substantial speculative movements, and its membership of the ERM from 1992 exacerbated this vulnerability. In fact, the high interest rates being maintained in Portugal (and in Spain) in the years immediately following EU membership - and before exchange controls were removed - did result in substantial capital inflows which risked over-heating the economy. The generalized ERM currency crisis of 1992-93 signalled a reversal in these capital movements and, although Portugal was never under the sustained speculative pressure experienced by other ERM members, it did devalue twice (along with Spain) in this period. Because the exchange controls were finally removed in December 1992 during the currency crisis, it is hard to be sure what the net effect of the removal was on capital movements, but it is generally held that the removal enhanced the credibility of the government's macroeconomic strategy and helped ensure a resumption of portfolio inflows when the crisis was over (Macedo, 1995). 5.7 Policy Lessons Reprivatization of the main Portuguese banks triggered a struggle for market share and a substantial modernization and expansion of the banking system. Control of banks was seen by many as representing the major investment opportunity in the economy, leading in due course to the emergence of strong industrial-financial conglomerate groups.29 The prudential risks that might be associated with such expansionism based on interconnected lending may yet prove to have been underestimated (and it remains to be seen whether the policy response in terms of supervision is 29 There was also the desire of several families or individuals to reacquire banks they had owned before the nationalization. 33 Portugal adequate), but to date, the reprivatized Portuguese banking system has been a commercial success story. Blocked from significantly participating in the privatization, foreign banks have played only a small role in this revitalization, though acquisitive Spanish banks may become an increasingly strong force in the Portuguese market. As in Greece, interest rate margins and bank profitability rose at first following privatization but subsequently fell back under the combined pressures of increased competition and economic recession in 1993-94. The subsequent economic recovery has not seen a recovery of margins, suggesting that increased competitive forces have been the dominant effect. Based on the narrowing of margins since the early 1990s, and the decline in profits, the prima facie conclusion is that consumer benefits have been substantial. While a more complete analysis could quantify a number of qualifications (the role of reduced taxation; the suitability of total assets as a denominator, etc.), these are unlikely to overturn this interpretation. Macroeconomic management has been complicated by the removal of exchange controls and the fiscal adjustment process made more difficult by the removal of the quasi-fiscal taxes, but Portugal has weathered these difficulties and has resumed a satisfactory growth path with a sustainable fiscal position. Indeed, the credibility of the macroeconomic adjustment is likely to have been strengthened by the parallel commitment to liberalization and removal of exchange controls. Prudential supervision is said to have been strengthened (with a greater emphasis being given to risk-based as opposed to rule-based supervision), and it is noteworthy that the central bank has been streamlining its operations, reducing staff levels dramatically (by over one third in half-a-decade). 34 Portugal On-site inspections have been conducted as a matter of routine for many years. Inevitably, the true quality of the supervision will only become known when the system is placed under pressure. 30 While the adoption of a universal banking model and the abolition of exchange controls has placed Portugal very much at the liberal end of the banking spectrum (Barth et al., 1997), it is worth noting the sequencing that was employed. Thus the strengthening of the legislative framework (as part of the process of adopting EU standards) and the privatization of banks was accomplished before the most drastic of the regulatory liberalizations (Decressin and Mauro, 1998). This will have contributed to the stable evolution that has been experienced. 30See Decressin and Mauro (1998) for an account of the upgrading of prudential regulation, together with an up-beat discussion of the state of capital adequacy and loan-loss provisioning. 35 6 Concluding remarks Committed, as they were by membership of the European Union, to the removal of exchange controls and to allowing entry through cross-border provision of banking services, Greece and Portugal have experienced financial liberalization on a wide front. The result has been an expansion in the domestic financial sector in both countries in terms of activity and services, but there have been important differences. Greek banking is still dominated by state-controlled institutions whose slow emergence from an inherited overhang of bad debt has both presented profitable opportunities for newcomers and delayed the emergence of a fully competitive and efficient banking regime. The Portuguese state banks were reprivatized - though foreign access to this process was limited - and there has since been a scramble for market share resulting in a tendency to increased concentration. Interest rate spreads increased at first following liberalization, as the administered restraints of the past were removed, but competition has reversed that initial trend and consumer benefits appear to have been substantial. Both countries have seen some macroeconomic instability in the post-liberalization years. Greece in particular is still in the process of managing a tricky disinflation which, as in many other countries, has been accompanied by somewhat unwelcome capital inflows fuelling foreign currency borrowing by domestic firms and enterprises. Portugal also saw volatile capital inflows and suffered from exchange rate policy uncertainty in 1992-93, but there have been no bank failures. Although the trigger for the financial liberalization was internationalization in one form or another, the changes have been largely domestic: penetration of foreign-owned banks has been remarkably limited. Of course their entry was not encouraged, or even facilitated by the authorities beyond the implementation of what was required by the EU; for example, foreign entities were formally 36 precluded from playing a major role in the Portuguese privatizations, and there are still a few tax impediments. Also, the still-dominant role of the state-controlled banks in Greece will have been a non-market deterrent. But the experience does serve to show how resilient the domestic banking industry can be in the face of the threat of entry from abroad. When their position is contested, local financial intermediaries can react by improving their efficiency and competitiveness. 37 References Antoniades, D. and D. Kouzionis (1989), Financial Distress of Industrial Firms in Greece and their Impact on the Greek Banking System, World Bank, Background Paper for 1989 World Development Report, Washington D.C. Barros, P.P. (1995), "Post-entry Expansion in Banking", Universidade Nova de Lisboa, Working Paper 242. Barros, P.P. and A.P.N. Leite (1994), "Conduct and Competition in Portuguese Commercial Banking", Universidade Nova de Lisboa, mimeo. Barth, J.R., D.E. Nolle and T.N. Rice (1997), "Commercial Banking Structure, Regulation and Performance: An International Comparison", US Comptroller of the Currency Economics Working Paper 97-6. Begg, I. (1992), "The Spatial Impact of Completion of the EC Internal Market for Financial Services", Regional Studies, 26, 333-347. Borges, A. (1992), "Portugal, Monetary and Financial System", in P. Newman, M. Millgate and J. Eatwell, eds. The New Palgrave Dictionary of Money and Finance (London: Macmillan). Borges, A.M. (1993), "Portuguese Banking in the Single European Market", in Dermine, ed., op. cit. Cabral, L. and R. Majure (1993), "A Model of Branching, with an Application to Portuguese Banks", Bank of Portugal, Working Paper 5-93. Claessens, C. and T. Glaessner (1997), "Internationalization of Financial Services in Asia", World Bank, mimeo. Decressin, J, and P. Mauro (1998), "The Portuguese Banking System: Feeling Its Pulse on the Eve of EMU Membership", in IMF Country Report 98/127 Edey, M. and K. Hviding (1995), "An Assessment of Financial Reform in OECD Countries", OECD Working Papers 154. Ericsson, N.R. and S. Sharma (1996), "Broad Money Demand and Financial Liberalization in Greece", IMF Working Paper 96/62 38 European Commission (1988), Research on the Costs of Non-Europe: Basic Findings, Vol. 9, Price Waterhouse. European Commission (1997), The Single Market Review: Credit Institutions and Banking. (London: Kogan Page). [ERE Report] Folkerts-Landau, D., D. Mathieson and G.Schinasi, eds. (1997), International Capital Market: Developments, Prospects and Key Issues (Washington, D.C.: International Monetary Fund). Gardener, E.P.M., P. Molyneux and B. Moore (1997), "The Impact of the Single Market Programme on EU Banking: Select Policy Experiences for Developing Countries", University of Wales Institute of European Finance, mimeo. Gual, J. and D. Neven (1993), "Banking" in European Economy, Social Europe No. 3 (Market Services and European Integration. The Challenges for the 1990s). Macedo, J.B. de (1990), "External Liberalization with Ambiguous Public Response: The Experience of Portugal", in C.Bliss and J.B. de Macedo, Eds., op. cit. Macedo, J.B. de (1995), "Convertibility and Stability 1834-1994: Portuguese Currency Experience Revisited", Universidade Nova de Lisboa, Working Paper No. 239. Organization for Economic Cooperation and Development (1998), Costs and Profitability of Banking (Paris: OECD). Organization for Economic Cooperation and Development (Various years), Economic Surveys: Greece and Portugal. (Paris: OECD). Papademos, L. (1991) "Monetary Policy and Financial Markets in the 1990s" in T.S. Skouras, ed. The Greek Economy: Economic Policy for the 1990s (London: Macmillan) Papademos, L. (1992), "Greece, Monetary and Financial System", in P. Newman, M. Millgate and J. Eatwell, eds. The New Palgrave Dictionary of Money and Finance (London: Macmillan). Papaioannou, M.G. and E.K. Gatzonas (1997), "Financial Innovation Involving the Greek Drachma", IMF Working Paper 97/14. Pinho, P.S. de (1995), "The Impact of Deregulation on Price and Non-Price Competition in the Portuguese Deposits Market", Universita Nova de Lisboa, mimeo. 39 Prati, A. and G.J. Schinasi (1997), "European Monetary Union and International Capital Markets: Structural Implications and Risks", IMF Working Paper 97/62. Story, J. and I. Walter (1997), The Political Economy of Financial Integration in Europe (Manchester University Press). Thomson, G. and M. Taylor (1994), "The Developing Single Market in Financial Services", Bank of England Quarterly Bulletin, November, 341-346. Vives, X. (1991), "Banking Competition and European Integration", in Giovannini and Mayer eds., op. cit. Wyplosz, C. (1999) "Financial Restraints and Liberalization in Postwar Europe", mimeo (forthcoming in volume being edited by Caprio, Honohan and Stiglitz). 40 Table 1: Greek and Portuguese Banks: Elements of Profit and Loss Account % average balance sheet 82 85 86 87 88 89 90 91 92 93 94 95 96 Greece (Commercial banks) Net interest margin 2.02 1.35 1.27 0.68 0.75 1.26 1.63 2.2 1.61 1.57 1.36 2.02 1.87 Non-interest 1.39 1.77 1.78 2.16 1.96 1.85 2.05 2.52 2.20 2.18 2.86 2.22 2.33 Gross margin 3.41 3.12 3.05 2.84 2.71 3.11 3.69 4.71 3.81 3.75 4.22 4.24 4.20 Operating expenses 2.52 2.45 2.29 2.22 2.20 2.32 2.36 2.42 2.33 2.35 2.51 2.73 2.86 o/w staff 2.04 1.98 1.82 1.71 1.71 0.78 1.32 2.29 1.48 1.40 1.71 1.59 1.52 Provisions (net) 0.45 0.33 0.32 0.25 0.22 0.33 0.45 0.69 0.32 0.34 0.39 0.26 0.55 Profit before tax 0.43 0.34 0.44 0.37 0.29 0.46 0.87 1.61 1.16 1.06 1.31 1.26 0.79 Portugal (All banks) Net interest margin 2.47 2.37 2.76 3.41 3.66 4.12 NA 4.97 4.07 3.37 2.78 2.28 2.08 Non-interest 1.25 0.75 0.61 0.76 0.80 0.80 NA 1.16 1.34 1.32 1.04 0.72 0.94 Gross margin 3.72 3.13 3.37 4.17 4.46 4.92 NA 6.13 5.41 4.69 3.82 3.00 3.03 Operating expenses 2.06 2.17 2.24 2.26 2.27 2.30 NA 2.75 2.78 2.51 2.22 1.95 1.95 o/w staff 1.66 0.95 1.13 1.91 2.18 2.61 NA 3.38 2.63 2.18 1.6 1.05 1.08 Provisions (net) 1.01 0.62 0.84 1.35 1.43 1.59 NA 1.77 1.54 1.17 0.94 0.40 0.39 Profit before tax 0.64 0.33 0.29 0.56 0.75 1.02 NA 1.60 1.09 1.01 0.66 0.65 0.69 Table 1 (continued): Greek and Portuguese Banks: Elements of Profit and Loss Account % GDP 82 85 86 87 88 89 90 91 92 93 94 95 96 Greece (Commercial banks) Net interest margin 0.98 0.76 0.74 0.41 0.40 0.78 0.98 1.26 0.95 0.97 0.83 1.21 1.17 Non-interest 0.68 1.00 1.03 1.29 1.05 1.14 1.24 1.44 1.30 1.35 1.75 1.33 1.46 Gross margin 1.66 1.76 1.77 1.70 1.46 1.92 2.22 2.70 2.26 2.33 2.59 2.53 2.64 Operating expenses 1.23 1.38 1.33 1.33 1.18 1.44 1.42 1.39 1.38 1.46 1.54 1.63 1.80 o/w staff 0.43 0.38 0.44 0.37 0.27 0.49 0.80 1.31 0.88 0.87 1.05 0.90 0.84 Provisions (net) 0.22 0.19 0.19 0.15 0.12 0.20 0.27 0.39 0.19 0.21 0.24 0.15 0.34 Profit before tax 0.21 0.19 0.25 0.22 0.16 0.28 0.53 0.92 0.69 0.66 0.81 0.75 0.50 Portugal (All banks) Net interest margin 3.33 3.34 3.21 3.93 4.16 4.68 5.55 5.87 5.29 5.03 4.54 4.34 4.07 Non-interest 1.68 1.06 0.71 0.88 0.91 0.91 1.29 1.37 1.75 1.98 1.70 1.37 1.84 Gross margin 5.01 4.40 3.92 4.82 5.07 5.59 6.85 7.24 7.03 7.01 6.24 5.70 5.91 Operating expenses 2.77 3.06 2.61 2.61 2.59 2.62 2.85 3.25 3.62 3.76 3.64 3.70 3.80 o/w staff 2.23 1.34 1.31 2.21 2.48 2.97 3.99 3.99 3.42 3.26 2.61 2.00 2.11 Provisions (net) 1.36 0.87 0.98 1.56 1.63 1.81 2.35 2.09 2.00 1.75 1.54 0.76 0.75 Profit before tax 0.87 0.47 0.33 0.65 0.86 1.16 1.64 1.89 1.41 1.51 1.07 1.24 1.36 Note: Break in Portuguese data for total assets in 1990; Greece: 4 large banks only up to 1988. Source: OECD Bank Profitability, 1998, IMF: International Financial Statistics. 42 Table 2: Greece - Interest Rates Interbank rate Time deposits Short-term Treasury bills (overnight) (3-12 month) lending (3-month) 1998 Dec. 11.9 10.0 17.5 11.1 Sept. 11.7 10.8 18.2 - June 13.4 10.7 18.6 11.8 Mar. 13.2 11.0 19.3 12.8 1997 Dec. 11.0 11.2 19.1 12.9 Sept. 11.0 9.5 18.4 8.4 June 11.0 9.6 18.3 8.5 Mar. 10.4 10.1 19.3 9.2 1996 Dec. 12.6 11.9 20.2 10.2 Sept. 13.3 12.6 20.6 11.5 June 14.1 13.9 21.2 12.4 Mar. 13.8 14.6 21.2 12.4 1995 Dec. 14.4 14.5 21.1 - Sept. 15.6 14.7 21.7 13.4 June 15.9 15.4 23.1 14.5 Mar. 17.0 17.3 25.3 15.3 1994 Dec. 17.5 17.1 26.4 15.8 Sept. 18.4 18.7 27.4 18.0 June 27.6 20.2 29.3 - Mar. 19.8 18.4 27.0 16.0 1993 Dec. 19.9 19.2 28.4 18.0 Sept. 25.0 19.4 28.3 18.0 June 21.3 19.2 28.7 17.8 Mar. 26.5 19.5 28.9 19.0 1992 Dec. 28.7 19.4 28.7 19.0 Sept. 30.7 19.3 28.6 17.5 June 22.3 19.5 28.7 17.0 Mar. 24.1 20.6 28.8 17.5 1991 Dec. 28.6 20.6 29.2 18.0 Sept. 20.2 20.6 29.5 19.0 June 17.8 20.5 29.5 19.0 Mar. 21.2 20.7 29.5 19.0 1990 Dec. 23.3 20.5 29.5 19.0 Sept. 19.1 20.4 28.9 19.0 June 22.2 19.1 26.8 19.0 Mar. 17.1 18.0 25.3 18.0 1989 Dec. 19.3 17.8 25.4 17.0 Sept. 17.5 23.7 - June 16.9 22.3 - Mar. 16.5 22.4 - 1988 Dec. 20.9 17.1 22.7 16.0 Sept. 17.0 22.8 - June 17.6 23.0 - Mar. 17.6 23.0 - 1987 Dec. 16.0 17.8 23.0 17.5 1986 Dec. 15.8 20.5 - 1985 Dec. 15.8 20.5 - Source: Bank of Greece 43 Table 3: Greece - Share of Banking Market % of total Deposits Assets Loans Govt. Securities 1980 State-controlled banks 87.3 81.8 92.2 86.2 o/w specialized banks 28.5 34.2 48.7 0.5 Private Greek-owned banks 5.9 5.8 4.1 8.3 Branches of foreign banks 6.8 12.3 3.7 5.5 1985 State-controlled banks 87.7 80.3 92.6 88.2 o/w specialized banks 28.2 31.6 54.7 5.3 Private Greek-owned banks 6.6 6.6 4.6 9.0 Branches of foreign banks 5.8 13.1 2.8 2.7 1990 State-controlled banks 84.8 82.3 90.1 83.0 o/w specialized banks 28.8 34.2 55.1 16.3 Private Greek-owned banks 8.3 7.7 5.9 11.9 Branches of foreign banks 6.9 10.0 4.1 5.1 1993 State-controlled banks 78.9 77.4 82.5 85.6 o/w specialized banks 27.4 32.3 49.2 33.2 Private Greek-owned banks 13.0 11.3 11.3 10.7 Branches of foreign banks 8.0 11.3 6.2 4.2 1997 State-controlled banks 73.2 67.1 69.7 84.8 o/w specialized banks 27.7 28.6 33.4 32.5 Private Greek-owned banks 17.8 17.5 21.4 11.7 Branches of foreign banks 9.1 15.5 8.9 3.5 Source: Bank of Greece 44 Table 4: Portugal - Interest Rates Deposits Commercial Lending rates (91-180 days) paper* Discount** Loans*** 1998 Dec 3.3 3.3 8.3 6.0 Sep 3.8 4.2 9.3 7.0 Jun 3.9 4.3 9.5 7.2 Mar 4.2 4.4 10.3 8.1 1997 Dec 4.6 5.0 10.7 8.4 Sep 4.8 5.2 11.1 8.2 Jul 4.7 5.5 11.5 8.0 Mar. 5.3 6.6 12.2 9.7 1996 Dec. 5.5 6.4/ 6.8 12.1 11 Sept. 6.2 7.6 12.9 11.2 June 6.4 7.6 13.6 10.8 Mar. 7.2 8.3 14.4 12.3 1995 Dec. 8.1 9.2 14.6 12.7 Sept. 8.3 9.2 15.6 13 June 8.9 9.6 15.9 13.5 Mar. 8.9 11.0 16.2 13.8 1994 Dec. 9.3 10.3 16.3 14.7 Sept. 9.5 10.7 17.3 15.2 June 9.3 14.4 17.4 15.4 Mar. 9.2 10.0 17.9 15.0 1993 Dec. 10.2 12.0 18.2 15.7 Sept. 10.6 11.9 19.3 15.8 June 12.6 13.1 20.4 16.6 Mar. 12.7 15.7 20.6 17.1 1992 Dec. 14.0 16.7 20.8 18.9 Sept. 14.5 15.5 21.3 19.5 June 16.3 16.8 22.5 19.8 Mar. 17.0 17.3 23.8 21.5 1991 Dec. 18.2 19.0 23.9 21.2 Sept. 17.1 18.6 24.2 21.1 June 17.3 19.2 24.5 21.8 Mar. 13.9 20.2 25.0 22.1 1990 Dec. 11.8 22.0 24.9 21.8 Sept. 11.5 21.6 24.5 21.5 June 9.4 18.7 24.0 22.1 Mar. 10.9 19.2 23.2 21.5 * Short-term commercial paper; before 1991: CRISTAL loans. ** Discount of commercial bills: non-financial private enterprises, 91-180 days. ***Loans and advances; non-financial enterprises, 91-180 days. Source: Bank of Portugal, Statistical Bulletin, Table B.6.1, B.6.2, Table B.8.1.6. 45 Table 5: Market shares in Portuguese banking, 1984, 1987 and 1994 Deposits Assets Loans 1987 State-owned 95.2 92.1 92.9 Foreign 2.6 4.1 4 Top four banks 57.6 53.9 53.9 1994 State-owned (two groups) 35.0 31.7 31.7 Foreign 6.2 9.3 8.7 Top four banks 50.5 Top four groups 62.2 60.6 62.4 Top five groups 71.8 68.8 69.7 1997 State-owned 28.9 23.4 Foreign (June 1998) 6.6 11.2 9.0 Top five groups 82.8 78.0 77.1 Source: Based on data provided by Associacao Portuguese de Bancos, CGD, and bank annual reports. 1997/98: Bank of Portugal. Note: Comparability of figures is approximate. 46 Figure 1: Greece and Portugal: Bank Income Net interest margin, 1985-96 6.00 5.00 4.00 3.00 2.00 1.00 0.00 85 86 87 88 89 90 91 92 93 94 95 96 Profit before tax, 1985-96 2.00 1.50 1.00 0.50 0.00 85 86 87 88 89 90 91 92 93 94 95 96 Greece Portugal 47 Figure 2: Greece: interest spread, 1985-98 Short-term lending less time deposit 10 9 8 7 nts 6 poi 5 entagec 4 Per 3 2 1 0 85 87 88 89 90 91 92 93 94 95 96 97 98 48 Figure 3: Portugal Lending Interest Spread, 1987-98 Above Deposit Rate 12 10 8 points 6 entage Perc 4 2 0 87 88 89 90 91 92 93 94 95 96 97 98 49 Figure 4: Greece and Portugal - Financial Depth Greece 100 90 80 70 DCP 60 M2 GDP 50 DCP(w) % 40 30 M3 20 10 0 80 85 90 95 Portugal 140 120 100 P 80 M2 GD 60 DCP % 40 20 0 80 85 90 95 Note: M2, M3 represnet measures of monetary holdings; DCP and DCP (w) are measures of bank credit to the private sector, both expressed as percentages of GDP. Source: IFS lines: M2= 34+35; M3=M2+44+45; DCP= 32d; DCP(w)=DCP+ 42d. 50 Annex Annex: Some Illustrative Econometric Results on the Dynamics of the Spread The discussion of interest spreads in the text observes that liberalization was followed by an initial upward swing in interest rate spreads, followed by a gradual reversal. We suggest that this is attributable to a particular post-liberalization dynamic whereby the removal of controls at first results in an increase in spreads reflecting a lack of competition, but that competition builds in response to the high spreads and has the effect of reversing them. It would be desirable to back-up this informal discussion with an econometric analysis, in particular to assess whether the pattern can easily be explained by other factors such as the state of the business cycle, or movements in the level of wholesale rates. This annex sketches a somewhat superficial analysis of this type using annual data on the Portuguese spreads, 1984-97. We find that several features of a model of the competitive dynamics do emerge from the econometrics, and are not overturned by the inclusion of variables representing cyclical factors, interest rate levels and taxation of the financial system. Although it needs to be borne in mind that the data series is very short, and that the data representing explanatory variables is quite imperfect, the results do provide some support for the theory presented in the text. Consider, then, the following simple model of the dynamic response of bank spreads to changing competitive conditions, as well as cyclical and transitory factors. Let the equilibrium spread between bank deposit and lending rates at time t be: f st* = f(ct, zt), with c < 0. where c represents the degree of competition in the market, with c0 = 0 representing a monopoly or cartel, and c = UHSUHVHQWLQJ SHUIHFW FRPSHWLWLRQ and z is a vector of other factors affecting the interest spread. We propose a bivariate dynamic model of the evolution of the actual spread st and the degree of competition ct based on the following simple updating rules: st - st-1=(st - st ) + ut -1 ct - ct-1= 1(st - 0) + vt -1 In words, the spread follows a partial adjustment model to the equilibrium spread, while the competition responds positively to last period's spread (over a normal spread). A priori, we may think of these as fast and slow dynamics respectively, with spreads likely to adjust fairly quickly to the equilibrium, conditional on the state of competition, and competition evolving slowly. The other explanatory variables z could include measures of the state of the business cycles, the level of interest rates and taxation (explicit and implicit) affecting spreads. If z is stationary, this specification suggests that the long-term equilibrium spread is 0. When the spread is higher than this, competitors enter, and vice-versa. The state of competition variable c is a latent variable, and needs to be substituted out before estimation. Solving the second equation successively we obtain: ct = 1( t-1st - t ) + t-1vt + c0. 0 0 0 Thus the key latent variable c may be proxied by a weighted sum of the spreads from time zero and a time trend, with a negative sign on the latter. Our focus is chiefly on exploring the longer-term competitive dynamics, but it is necessary to condition on these 'nuisance'variables. Here we use GDP growth, nominal interbank rate, and the ratio of bank reserves to bank deposits. (This last-mentioned variable is admittedly a very imperfect measure of the taxation, explicit and implicit, of bank deposits.) The source for each of these is IFS. For the spread, we use net interest income as a share of total assets, data from OECD Bank Profitability to 1990, from Bank of Portugal thereafter). We experimented with linear and negative exponential specifications of the function f, and with a variety of additional stochastic dynamics. In practice, a linear specification gave the best fit. The two variable proxy31 for c works rather well, with an estimated equilibrium spread of about 3.1 percent (Equation A). Addition of the reserve ratio and growth rates, along with first and second order autocorrelation coefficients and the lagged spread given an excellent fit, as shown in Table A1 (Equation B) and Figure A.1 (though note the relatively few degrees of freedom). The level of the nominal interest rate does not appear to be significant (not shown). Equation A is incomplete in terms of the model, because the lagged dependent variable is not included. Looking then at Equation B, we infer point estimates as follows: f = 0.22; 0 = 3.3; 1 = ­ 0.20 c The estimated value of is rather low for what we have described as fast dynamics, but otherwise these key estimates are not out of line with prior reasoning. Signs and magnitudes of the coefficients on nuisance variables are reasonable. For example, 5 per cent growth widens spreads by 50 basis points as compared with zero growth; a ten percentage point increase in reserves corresponds to a 37 basis point increase in spread. The results provide suggestive, though not conclusive, evidence of the proposition that liberalization first increased spreads, as the cartelized banking system exploited monopoly power, but then longer term forces of increased competition started to dominate. 31 An alternative proxy (the total number of banks) performed poorly. 52 Table A1: Portugal: Regression results Annual change in bank interest spread Eqn A Eqn B Variable coeff (t-stat) coeff (t-stat) Constant 0.055 (0.3) -0.178 (2.0) Reserve requirement 0.037 (4.2) GDP Growth 0.102 (4.0) Sum of spreads (-1) -0.211 (4.7) -0.202 (4.2) Time trend (1984) 0.664 (4.1) 0.664 (3.8) Spread (-1) -0.218 (3.8) AR(1) -1.634 (3.1) AR(2) -0.757 (1.5) RSQ / Mean depvar 0.877 / 0.013 0.985 / -0.066 Adj RSQ / S.D. depvar 0.852 / 0.455 0.951 / 0.450 S.E.R. / A.I.C 0.175 / -0.450 0.100 / -1.614 S.S.R. / Schwarz 0.306 / -0.319 0.030 / -1.324 Log-Likhd / F-stat 5.92 / 35.6 16.9 / 28.7 DW / Prob-F 2.74 / 0.000 2.22 / 0.001 Sample / d.f. 1984-97 / 10 1984-97 / 5 Figure A.1: Portugal - Change in Spread: Fit for Equation B 1.0 0.5 0.0 -0.5 0.15 0.10 -1.0 0.05 0.00 -0.05 -0.10 -0.15 86 87 88 89 90 91 92 93 94 95 96 97 Residual Actual Fitted 53