Report No. 20657-TU Turkey Country Economic Memorandum Structural Reforms for Sustainable Growth (In Two Volumes) Volume 1: Main Report September 15, 2000 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank Turkey-Country Economic Memorandum Currency Eq uivalents (Exchange Rate Effective September 15, 2000) Currency Unit = Turkish Lira TL I = US$ 0.0000015 US$ I = TL 668,229 Government Fiscal Year January I - December 31 Weiehts and Measures Metric System ABBREVL4TIONS AND ACRONYMS AMS - Aggregate Measure of Support PAYG - Pay-As-You-Go ASCU - Agricultural Sales Cooperative PSE - Producer Subsidy Equivalent Union BK - Bag-Kur QR - Quantitative Restriction BO - Built Operate RB - Restructuring Board BOO - Built Operate Own SDIF - Saving Deposit Insurance Fund BOT - Built Operate Transfer SIS -. State Institute of Statistics BRSA - Banking Regulatory and Supervisory SOE - State Owned Enterprises Authority CAYKUR - Turkish Tea Company SPO - State Planing Organization CEM - Country Economic Memorandum SRS - Sustained Reform Scenarios CPI - Consumer Price Index SSK - Social Security Institution DB - Defined Benefit TAS - Telecommunications Authority of Singapore DC - Defined Contribution TEAS - Turkish Electricity Generation and Transmission Company EMS - Equivalent Measurement of Support TEDAS - Turkish Electricity Distribution Company ERSP - Earnings-Related Pension TEKEL - Turkish Tobacco Monopoly ES - Emekli Sandigi TFP - Total Factor Productivity FFE - Full-Time Farmer Equivalent TMO - Turkish Grain Office GHIS - General Health Insurance System TOOR - Transfer of Operating Rights GNP - Gross National Product TRS - Truncated Reform Scenarios LNG - Liquid Natural Gas TSFAS - Turkish Sugar Factories MARA - Ministry of Agriculture and Rural TZOB - Turkish Farmers Association Affairs MOH - Ministry of Health VAT - Value Added Tax NA - Notional Account WTO - World Trade Organization NDC - Notional Defined Contribution WUO - Water Users' Organization PA - Privatization Administration Vice President: Johannes Linn (ECAVP) Country Director: Ajay Chhibber (ECC06) Sector Director: Pradeep Mitra (ECSPE) Sector Leaders: Samuel Otoo (ECSPE) Ataman Aksoy (DECPG) Team Members: Ismail Arslan (team leader), James Parks, Mediha Agar, Rossana Polastri, Pinar Baydar (ECSPE); Craig Burnside (DECRG); Esperanza Lasagabaster, Alan Thompson, Shiyan Chao (ECSHD); Anita Schwarz (HDNSP); Jan Alexander, Timothy C. Irwin (PSDPP); John Nash (ECSSD); Biagio Bossone (FSP); James S. Moose (ECSEG); A. Halis Akder, Erol H. Cakmak, Hasan Ersel, Haluk Kasnakoglu, Guven Sak, Ercan Uygur (Consultants) TURKEY: COUNTRY ECONOMIC MEMORANDUM STRUCTURAL REFORMS FOR SUSTAINABLE GROWTH Table of Contents EXECUTIVE SUMMARY ......................................................................1I CHAPTER 1: MACROECONOMIC FRAMEWORK AND FISCAL SUSTAINABILITY ...............................I I INTRODUCTION ............. 2 THE MACROE CONOMIC BACKGROUND ............2 3 GROWTH IN TURKEY .4 4 PUBLIC SECTOR DEBT AND PUBLIC SECTOR SOLVENCY .12 5 ACHIEVING F ISC AL SUS TAINABILITY ..........; 16 6 CREDIBILITY, SUSTAINABILITY AND THE STABILIZATION EFFORT .18 7 THE NEW ECONOMIC PROGRAM .22 CHAPTER 2: SOCIAL SECURITY REFORM .................................... 28 1I INTRODUCTIO.28 2 PROBLEMS IN THE PRE-REFORM SYSTEM .30 3 THE 1999 REFORM: AVERTING A FISCAL CRISIS .32 4 SECOND STAGE OF REFORMS: BUILDING A SOUIND OLD AGE RETIREMENT INCOME SYSTEM .35 5 LONGER-RUN NEXT STEPS ...................... 39 6 HEALTH INSURANCE .44 7 UNEMPLOYMENT INSURANCE .49 8 RECOMMENDATIONS. 5 1 CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS ......................................................... 54 1I INTRODUCTIO.54 2 ELECTRICITY.56 3 GAS.61 4 TELECOMMUNICATIONS.64 5 RECOMMENDATIONS .69 CHAPTER 4: AGRICULTURAL AND RURAL DEVELOPMENT .................................................... 71 1I INTRODUCTIO.71 2 THE CHALLENGE OF GRoWTH. 7 1 3 OVERVIEW OF SUPPORT INSTRUMENTS .72 4 AGRICULTURAL STATE ECONOMIC ENTERPRISES .78 5 OVERALL IMPACT OF THE AGRICULTURAL SUPPORT SYSTEM. 78 6 REFORMS IN OTHER COUNTRIES-HOW HAVE OTHER COUNTRIES ADDRESSED THESE PROBLEMS? . 80 7 NEW REFORM PACKAGE .81 8 THE GOvERNMENT'S NEW P.O'.E IN AGRICULTURE .84 9 LONGER TERM REFORMS: RECOMMENDATIONS .85 CHAPTER 5: BANKING SECTOR DEVELOPMENTS AND REFORM ........................................................ 87 1I INTRODUCTIO.87 2 MAJOR STRUCTURAL CHANGES AND DISTORTED INCENTIVES STRUCTURE .88 3 FINANCIAL RISKS AND RISK MANAGEMENT .91 4 THE STATE BANKS .94 5 NEW REFORM PACKAGE .97 CHAPTER 6: MEDIUM-TERM PROSPECTS ................................... 107 1 INTRODUCTION.107 2 PUBLIC SECTOR ADJUSTMENT.107 3 INFLATION AND GROWT .109 4 EXTERNAL BALANCE AND FINANCING .111 5 RISKS FACED BY THE REFORM PROGRAM . 12 REFERENCES ......................117 TECHNICAL APPENDICES ..................... 121 TABLES EXECUTIVE SUMMARY TABLE 1: ALTERNATIVE FISCAL ADJUSTMENT SCENARIOS TO STABILIZE DEBT/GNP RATIO CHAPTER 1: MACROECONOMIC FRAMEWORK AND FISCAL SUSTAINABILITY TABLE 1.1: GROWTH VOLATILITY IN EMERGING ECONOMIES TABLE 1.2: VOLATILITY OF GROWTH IN TURKEY, 1990-1999 TABLE 1.3: CONSOLIDATED PUBLIC SECTOR BALANCE TABLE 1,4: NET DEBT OF THE CONSOLIDATED PUBLIC SECTOR TABLE 1.5: DEBT ACCUMULATION, 1994-1999 TABLE 1.6: ALTERNATIVE FISCAL ADJUSTMENT SCENARIOS TO STABILIZE DEBT/GNP RATIO TABLE 1.7: SHORT-RUN DYNAMICS OF THE REQUIRED PRIMARY SURPLUS TABLE 1.8: NOMINAL INTEREST RATES AND INFLATION SINCE MID-1999 TABLE 1.9: FISCAL ADJUSTMENT TARGETS FOR 2000 CHAPTER 2: SOCIAL SECURITY REFORM TABLE 2.1: PENSION SUMMARY DATA PRIOR TO THE 1999 SOCIAL SECURITY REFORM TABLE 2.2: COMPARATIVE DATA ON THE PENSION SYSTEM PRE- AND POST-1999 REFORM TABLE 2.3: ADVANTAGES AND DISADVANTAGES OF LONG-RUN OPTIONS TABLE 2.4: SELECTED BASIC HEALTH INDICATORS TABLE 2.5: HEALTH SERVICES PERFORMANCE INDICATORS COMPARISON CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS TABLE 3.1: COMPARATIVE ENERGY DATA TABLE 3.2: ELECTRICITY DEMAND AND SUPPLY PROJECTIONS TABLE 3.3: TELECOMMUNICATIONS COMPARATORS TABLE 3.4: SALE OF EQuITY IN TURK TELEKOM CHAPTER 4: AGRICULTURAL AND RURAL DEVELOPMENT TABLE 4.1: TOTAL TRANSFERS TO AGRICULTURE IN TURKEY AND OECD COUNTRIES TABLE 4.2: FISCAL IMPACT OF THE REFORM PROGRAM, 2000-2003 CHAPTER 5: BANKING SECTOR DEVELOPMENTS AND REFORM TABLE 5.1: RATIO OF BANK ASSETS TO GNP FOR SELECTED YEARS TABLE 5.2: BANKING STRUCTURAL INDICATORS 1990-95: INTERNATIONAL COMPARISON TABLE 5.3: VARIOUS INTEREST RATES TABLE 5.4: CAPITAL RATIOS OF COMMERCIAL BANKS TABLE 5.5: OPEN FOREIGN-EXCHANGE POSITIONS OF BANKS TABLE 5.6: STATE-OWNED BANKS: SELECTED FINANCIAL RATIOS TABLE 5.7: DUTY LOSSES OF STATE BANKS FIGURES EXECUTIVE SUMMARY FIGURE 1: HISTORICAL GROWTH & INFLATION RATES FIGURE 2: PUBLIC SECTOR BORROWING REQUIREMENT AND INTEREST PAYMENTS FIGURE 3: GNP & PER CAPITA GROWTH RATES IN EMERGING ECONOMIES FIGURE 4: PROJECTED IMPROVEMENT IN THE DEFICIT OF THE SSI SYSTEM FIGURE 5: INDICATIVE CONTINGENT LIABILITIES IN POWER SECTOR FIGURE 6: AGRICULTURAL GROWTH AND TRANSFERS FIGURE 7: DUTY LOSSES OF STATE BANKS FIGURE 8: PROJECTED GROWTH & INFLATION FIGURE 9: PER CAPITA INCOME CHAPTER 1: MACROECONOMIC FRAMEWORK AND FISCAL SUSTAINABILITY FIGURE 1.1: GNP GROWTH, 198 8-1999 FIGURE 1.2: PSBR AND INTEREST PAYMENTS FIGURE 1.3: CPI INFLATION FIGURE 1.4: BASE MONEY GROWTH AND INFLATION, 1990-1999 FIGURE 1.5: PRIVATE INVESTMENT GROWTH BY TYPE, 1990-1999 CHAPTER 2: SOCIAL SECURITY REFORM FIGURE 2.1: PROJECTED IMPROVEMENT IN THE DEFICIT OF THE SSI SYSTEM, 1998-2050 FIGURE 2.2: PROJECTED AGING OF TURKISH POPULATION FIGURE 2.3: PER CAPITA HEALTH EXPENDITURE OF THREE SCHEMES CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS FIGURE 3.1: INDICATIVE PROJECTED LOSSES FOR TEAS WITHOUT REFORM CHAPTER 4: AGRICULTURAL AND RURAL DEVELOPMENT FIGURE 4.1: AGRICULTURAL GROWTH AND TRANSFERS FIGURE 4.2: COST OF AGRICULTURAL CREDITS FIGURE 4.3: SHARES OF CATEGORIES OF TRANSFER CHAPTER 5: BANKING SECTOR DEVELOPMENTS AND REFORM FIGURE 5.1: PSBR AND BANK PROFITS FIGURE 5.2: DUTY LOSSES OF STATE BANKS CHAPTER 6: MEDIUM-TERM PROSPECTS FIGURE 6.1: PUBLIC SECTOR BORROWING REQUIREMENT FIGURE 6.2: PUBLIC SECTOR PRIMARY BALANCE FIGURE 6.3: PUBLIC SECTOR DEBT AND INTEREST PAYMENTS FIGURE 6.4: INFLATION FIGURE 6.5: REAL GNP GROWTH FIGURE 6.6: REAL PER CAPITA INCOME FIGURE 6.7: SUSTAINED REFORM: GNFS EXPORTS AND IMPORT GROWTH AND CURRENT ACCOUNT DEFICIT/GNP FIGURE 6.8: TRUNCATED REFORM: GNFS EXPORTS AND IMPORT GROWTH AND CURRENT ACCOUNT DEFICIT/GNP BOXES CHAPTER 1: MACROECONOMIC FRAMEWORK AND FISCAL SUSTAINABILITY Box 1.1: EMPLOYMENT AND GROWTH Box 1.2: CREDIBILITY AND STABILIZATION EFFORTS: LESSONS FROM ELSEWHERE CHAPTER 2: SOCIAL SECURITY REFORM Box 2.1: PENSION FUNDS AND CAPITAL MARKETS: LESSONS FROM CHILE Box 2.2: REVIEW OF TWO OECD MODELS WITH MULTIPILLAR MANDATORY RETIREMENT INCOME SYSTEMS CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS Box 3.1: EU DIRECTIVE FOR ELECTRICITY BOX 3.2: INTERNATIONAL REGULATORY PRINCIPLES & CAPABILITIES CHAPTER 5: BANKING SECTOR DEVELOPMENTS AND REFORM Box 5.1: BANK PRIVATIZATION: SOME RELEVANT INTERNATIONAL EXPERIENCES CHAPTER 6: MEDIUM-TERM PROSPECTS Box 6.1: FISCAL ADJUSTMENT UNDER A CRAWLING PEG TURKEY: COUNTRY ECONOMIC MEMORANDUM STRUCTURAL REFORMS FOR SUSTAINABLE GROWTH EXECUTIVE SUMMARY 1. Turkey faces three major economic challenges. First, it must improve macroeconomic fundamentals via a durable fiscal adjustment aimed at taming high inflation and breaking the grip of high real interest rates that constrain growth. Second, it must raise productivity by eliminating costly incentive distortions and improving the quality of institutions and rules that govern markets in order to promote competition and encourage private investment in productive activities. Third, it must address disparities in economic opportunity and the social issues critical to realizing its full development potential and ensuring that the benefits of economic growth are broadly distributed. The first two challenges are the main themes for the CEM. The third challenge is discussed in depth in a recent Bank report entitled "Turkey: Economic Reforms, Living Standards and Social Welfare Study." Fiscal imbalances are the key to understanding Turkey 's high inflation and volatile growth 2. Despite impressive achievements over the past two decades, Turkey's economy continues to operate under a cloud of vulnerability-plagued by persistent fiscal imbalances, chronically high inflation, and sharp swings in the business cycle. Over the past two decades, several attempts to stabilize the atteonom hv falle sho Figure 1: Historical Growth & Inflation Rates economy have fallen short, and high growth has never been Growth X Inflation sustained for long. Inflation 5 70 was higher and growth was 60 lower, on average, in the 1990s 4 50 than in the 1980s. Annual 4 inflation averaged 44 percent in 2 30 the 1 980s before climbing to 76 12 percent in the 1990s (figure 1). 10 In parallel, average annual 1090 1990-99 198-90 growth fell from over 5 percent I to under 4 percent. Volatility Source: SIS and World Bank. approximately doubled over the same period as the standard deviation of GDP growth increased from 2.7 percent to 5.5 percent. 3. This report shows that fiscal imbalances are key to understanding Turkey's inflation problem and its volatile growth. Its findings suggest that Turkey's inability to sustain high growth can be closely linked, among many factors, to the lack of macroeconomic stability. Unsustainable fiscal policy has put repeated pressure on the lira, and led to chronic and high inflation. When crises have hit, fiscal policy has been unable to act as a smoothing influence on the business cycle-instead contractionary policies have been implemented to achieve monetary stability, worsening the real impact of shocks. 4. The findings also suggest that previous attempts at stabilization have failed precisely because they did not address the structural sources of the fiscal deficit. The fiscal situation improved somewhat in the late 1990s as the primary deficit of the public sector came down. But the government achieved little credibility with these efforts. Inflation remained high, and nominal interest rates reached record levels in 1998-99. So, the success of the current round of economic reform depends critically on the government's ability to correct these structural imbalances. Figure 2: Public Sector Borrowing 5. When running a large fiscal deficit Requirement and Interest Payments (figure 2), a government faces two choices: (in % of GNP) finance the deficit by accumulating debt or printing money, or reduce the deficit through 25 fiscal adjustment. Obviously, there are long- 20 run constraints on debt financing if the public Inters PaymentsP debt is growing faster than national income. PB Money creation is feasible up to a limit, but has high costs in terms of macro instability and inflation. Turkey has relied on a mix of borrowing and money creation, rather than making the structural changes needed to 1990 1991 1992 1993 1994 1995 1 ;96 1997 1998 1999 address the deficit. At times the total stock of - 1 public debt has been stable in GNP terms, but Source: Treasury and World Bank. it has tended to ratchet upward in discrete jumps corresponding to external crises. 6. To help close its financing gap, the government has resorted to the inflation tax to generate 2-3 percent of GNP in seignorage annually-and has increasingly turned to issuing non- cash debt. This non-cash debt has been used to stabilize the balance sheets of public sector enterprises, especially state banks that make losses on subsidized lending programs. High inflation has eroded the demand for money and impeded deepening of the domestic financial sector. This, in turn, has raised the cost of servicing the public debt and lowered its maturity, propagating a vicious circle of large deficits and chronic inflation. 7. Standard fiscal sustainability calculations suggest that Turkey does not need an enormous improvement in its long-run primary fiscal balance to stabilize its public debt level and eliminate the inflation tax (table 1). A primary Table 1: Alternative Fiscal Adjustment Scenarios to surplus of about 1.3 percent of GNP Stabilize Debt / GNP Ratio would be needed to maintain the public (Required Primary Surplus) debt at its current level in GNP terms in Rea InflAon Rate an environment of 10 percent inflation, GNP 10% 5% 10 percent real interest rates, and 5 Growth Real |inre Rate Real fN t Rate percent growth. From 1996 through | 5% 10% 1 5% 10| % 1999, the average primary deficit for the 1 3% -0.1 2.6 D.3 3.1 consolidated public sector was about 1.3 5% -1.2 1.3 -0.9 1.8 percent of GNP, so this represents a Source: World Bank. correction of some 2.6 percent of GNP. 8. To establish credibility and meet debt service costs, higher primary surpluses are required in the short run until interest rates come down. It may also be necessary for the government to make an up-front cut in its overall debt stock to break with the past trend of upward ratcheting ii a'hd thereby lower inflationary expectations. One-time revenues from privatization can be instrumental in this regard: a reduction in debt of 1 percent of GNP requires an equivalent amount of revenue. Growth has remained below its potential due to macroeconomic instability, insufficient investment in human capital, and low productivity growth 9. Despite having the potential to be a leader among emerging markets, Turkey has not achieved the high growth of middle-income countries in East Asia (figure 3). On average, the economy grew at just under 4 percent per year over the past decade-respectable, but well Figure 3: GNP & Per Capita Growth Rates below the top middle-income countries. More in Emerging Economies (1965-1998) worrying, average growth over the decade was - below that of the 1980s, 3.9 percent compared 1i10 with 5.3 percent. A cross-country comparison 8 suggests that countries that grew faster than 6 l-I_ Turkey did so partly because they accumulated 4 physical capital faster-but more importantly L t El because they accumulated human capital more quickly and because they had faster total factor v productivity growth. _ 3 4SP 3 Per Capita Inore 10. Turkey's investment rate in recent S times, although not comparable to the exceptionally high levels of Korea, has been on the order of 25 percent of GDP, suggesting that the quantity of investment is not a problem. But the distribution and quality of investment are problematic. Driven by excessive government borrowing and the inflation tax, real interest rates have averaged 20 percent over the decade, well above any reasonable measure of the return on capital investment and a clear signal of the distortions affecting the Turkish economy. Over a third of Turkey's fixed investment is devoted to residential construction, a classic hedge against inflation. Although Turkey has made impressive improvements in education and literacy since the 1 980s, its secondary school enrollment rate of 51 percent is low for a country with a per capita income above US$3,000. 11. This Country Economic Memorandum (CEM) emphasizes the influence of the policy environment on total factor productivity growth as a key explanatory factor. The less distorting the policy, the more efficiently all factors of production will be used. Openness, sustainable fiscal policy and low inflation have been shown to be important factors in generating an enabling environment for growth. Other key policies that affect economic efficiency include the extent of direct government intervention in the real sector, the enabling environment for private investment and the decree of competition in among economic agents. 12. Turkey has made some progress in implementing policy reforms. A series of trade and capital market reforms in the early 1980s transformed the country into an open economy by the end of the decade. But these reforms did little to control inflation and correct fiscal imbalances. They also left incomplete the enabling environment for the private sector, particularly in the iii agriculture, infrastructure and financial sectors. Much of the public sector reform agenda has remained untouched. The current reforms have a good chance of succeeding 13. The current reform program launched in mid-1999 may succeed because it addresses many of the structural problems at the heart of Turkey's fiscal imbalances and disappointing growth performnance. The new program emphasizes second generation structural reforms dealing with economic regulation, structural fiscal reforms, finance, infrastructure and privatization. The main elements of the program are: (i) a significant up-front fiscal adjustment, (ii) a nominal exchange rate anchor via a crawling peg regime, (iii) forward-looking incomes policy in the public sector, (iv) structural reform of the social security system and public expenditure management, as well as in the agriculture, infrastructure and financial sectors, and (v) accelerated privatization. Turkey's new status as a candidate for the European Union, agreed in December 1999, provides additional impetus for economic, social, and political reforms. Over the longer term, Turkey is now more likely to follow up the current program of disinflation and structural reform with actions to further liberalize the economy, tackle deeper institutional reforms of the public sector, and address broader governance issues central to the country's future development. 14. The current momentum for structural reform can permanently lower the fiscal deficit and curb inflation. Critically, the government has moved to reform the social security system. The significance of this reform cannot be overstated. The system was on a rapidly deteriorating path toward insolvency with a deficit of about 3 percent of GNP in 1998-projected to reach 5 percent of GNP by 2010 and 10 percent of GNP by about 2030. Now, the deficit has been contained buying critical time for the government to prepare the way for a full multi-pillar social security system over the longer term. Agricultural reform will also contribute to sustaining fiscal adjustment. The current support policies based on price and credit subsidies cost the budget 2-3 percent of GNP annually, in addition to a further 5 percent of GNP borne by consumers. Under the reform program, credit subsidies have already been eliminated and price supports will be phased out in favor of direct income support to farmers. Accelerating privatization, notably in telecommunications and energy, is central to the adjustment process. Privatization revenue of 8 percent of GNP is targeted over the 2000-02 period. Deregulation of the energy sector to replace the current centralized model with competitive markets for electricity and gas will promote private participation without government guarantees, thereby removing a key source of contingent liabilities for the budget. Financial sector reforms will tighten prudential regulation, delink the state-owned banks from budget subsidies and set the framework for their privatization. These reforms will improve the soundness of the banking system, addressing another important source of contingent liabilities through the deposit insurance fund. 15. Equally important, the government's structural reforms will raise productivity by improving the incentive framework, stimulating competition, and encouraging productive private investment. Actions to strengthen the enabling environrnent for the private sector cut across the program. The social security reform opens the door to private sector participation through voluntary private pension schemes. The agriculture reform program explicitly targets a reversal of the steady decline in agriculture productivity. Agriculture price and credit subsidies, and the state enterprises required to implement them, trap farmers in dead-end crops, stifle competition, iv and block private initiative. The shift to direct income support will eliminate these distortions while accompanying measures to privatize agricultural state-owned enterprises (SOEs) will promote private participation in processing and marketing activities. Reforms of the telecommunications and energy sectors are paving the way for greatly increased private participation and steadily increasing competition in line with WTO and EU standards. Financial sector reform will contribute directly to increased productivity by raising the efficiency of financial intermediation and improving resource allocation. Credibility requires decisive and bold actions 16. To succeed, the government must maintain and enhance the credibility of the reform program through clear, decisive actions. Two key issues stand out. First, the fiscal adjustment launched in 2000 on the basis of revenue increases, many of which are temporary, must be sustained over the medium term. This will require a more balanced mix of permanent tax increases, structural fiscal reforms, and expenditure measures. Research on fiscal adjustment shows that it is more likely to be successful when it involves credible measures to contain expenditure. To make sustainable long-run expenditure changes the government has initiated an ongoing study to review public expenditure and institutions. In this, the government needs to address the limited credibility and transparency of the budget process. There are too many off- budget items. The use of non-cash debt to prop up public sector enterprises and growing reliance on contingent liabilities lead to an undervaluation of the state's true obligations. 17. Second, the government must capitalize on its early success and continue to implement the structural components of its reform program. Despite the good progress so far in social security, banking, and elsewhere, the most difficult structural challenges clearly lie ahead. Specific goals and timetables for reform should be publicly announced and then adhered to. International experience, most recently highlighted by the collapse of Brazil's Real Plan, suggests that fiscal adjustment underpinned by prospective reform has a tendency to fall apart when planned reforms are not undertaken. With Turkey's record of failed reforms, actions will speak louder than words. Figure 4: Projected Improvement in the Deficit of the SSI Significant progress has been System, 1998-2050 made in reforming the social 0% __ _ _ __ __ - security system -2 % 18. A major policy reform in % - the social security system, which 10% - will contain but not eliminate the -12% deficit, was enacted in August -14% 1999 (figure 4). The reforn- -m - - raises the minimum retirement 1998 2003 200B 2013 2018 2023 2028 2033 2038 2043 2048 age for a full pension to 58/60 - B -Treasury Base-Line (female/male) immediately for new entrants, and to 52/56 for Source: Treasury and World Bank. existing contributors with a 10- year transition period. It also increases the minimum contribution period to 25 years for new entrants to the ES (Emekli Sandigi) and BK (Bagkur) programs covering civil servants and the v self employed. For the SSK (Social Security Institution) program covering workers, the minimum period is increased to 7,000 days for new entrants. For existing contributors to SSK, the minimum contribution period is increased from 5,000 to 6,000 days with a 1 0-year transition period. 19. Successful implementation of this policy reform depends on improvements in administration, particularly accurate and up to date information on the wages earned and contributions paid for the entire work history of each contributor. The authorities are now moving on a second stage of reform centered on administrative and institutional measures. In addition to meeting the information requirements of the August policy reform, this initiative will help improve coverage and compliance of the social security system and administratively separate the pension and health insurance programs. The second stage also encompasses introduction of a legal framework for voluntary funded private pensions, an important step towards a full multi-pillar system. Further changes to the social security system will be needed 20. Over the longer run, Turkey will need to implement deeper structural reforms to the social security system-to respond to the requirements of an aging population with increasing life expectancy and to eliminate the residual deficit. These reforms will also reinforce the system's contribution to the development of domestic savings and growth. The longer run reform responds to the following objectives: * Provide adequate income security for the elderly and disabled who have lost their capacity to work. * Achieve viability over the medium term and greater longer-term sustainability. * Deter inequities within the system. X Reduce labor market distortions and evasion. * Build a more efficient administrative structure to support the pension policy framework. These ambitious policy objectives will entail a substantial revision of the existing mandatory insurance system. There are three options for reforming the system: The first is to reform the PAYG scheme and maintain its defined benefit (DB) structure. * The second is to reform the PAYG scheme but transform it into a Notional Account scheme. The third is to implement a greater downsizing of the PAYG scheme accompanied by the introduction of a funded, defined contribution pillar- managed by the private sector. Social security reform may catalyze a wider reform of the health sector 21. The momentum of the social security reform has the potential to initiate wider reform of the health sector. Health reform is a wide ranging undertaking that must go well beyond the administrative overhaul of the health insurance programs run by the social security institutions. Even so, the current initiative may provide an important spark. In the short term action is needed to broaden administrative reform, including separating health expenditure accounts from pension accounts and improving the health insurance data base. Establishing a basic health care package vi is another key step. Longer-term structural reforms should move to integrate health service related functions horizontally. Part of this vision is a Ministry of Health that sets policy priorities and monitors outcomes and quality. Another aspect is a social security system that finances and purchases a basic package from a competitive provider market. And third is a well regulated private sector that plays a role in offering supplementary insurance and competes in the provider market by selling services in accordance with contractual agreements with insurance agents. Unemployment insurance could strengthen the social security system 22. The unemployment insurance scheme introduced as part of the August 1999 policy reform may strengthen the social insurance system. But further steps will be needed to monitor its fiscal impact and ensure a balance between worker protection and disincentives to work. The new unemployment insurance scheme constitutes the third tier of Turkey's social insurance system. The scheme stipulates an annual budgetary contribution of 2 percent of the covered wage bill and a commitment to cover any eventual deficit. It is not expected to pose a burden on the budget in the near term as the existing compulsory savings program which includes a two percent government contribution was terminated when the unemployment insurance scheme became effective in June 2000. 23. But the longer-term fiscal consequences of unemployment insurance are unclear and will have to be carefully monitored as the system becomes fully operational over the next two years. With unemployment insurance in place, the justification for mandated severance payments is no longer present. The combination of both systems offers more than 100 percent replacement-a clear distortion. If mandated severance is phased out, accrued rights should be grand-fathered and this could bridge the transition to full reliance on unemployment insurance as the only mandated program. In the future, private firms and workers should be freely allowed to negotiate severance payments within the overall compensation package to the worker. Infrastructure reform is a cornerstone of Turkey 's future growth and fiscal stability. 24. The government is tackling long-standing problems in infrastructure with an ambitious reform program focused on energy and telecommunications. The demand for infrastructure investment (an estimated US$3 billion a year for electricity alone) risks outpacing supply, creating bottlenecks to growth. Turkey pioneered the build-operate-transfer (BOT) model and made several attempts to launch privatization in the key infrastructure sectors with the objectives of keeping pace with technological change, shifting the burden of investment to the private sector and reducing the public debt. But too little attention was paid to the legal and regulatory framework and the need to promote competition. In addition, state monopolies have continued to dominate energy and telecommunications, while private investors have hesitated to commit to Turkey without government guarantees. As a result, a paradoxical situation has developed where the state's contingent liabilities under take-or-pay contracts have grown rapidly (figure 5), while insufficient private investment may lead to power shortages in the next two years. 25. The government's reform program combines up-front measures to establish modern, transparent regulatory environments, with action to restructure state monopolies, introduce market forces, and accelerate privatization. This difficult and complex agenda will take several vii years to complete. Once the requisite structural reforms are in place, the policy emphasis will shift to further liberalization of the energy and telecommunications sectors over the longer term in the context of Turkey's WTO commitments. This liberalization will also support Turkey's future EU accession process. Power sector reforms should be accelerated 26. A major restructuring under way in the electricity sector aims to ensure that Turkey can meet its growing energy demand at a sustainable cost to the budget and economy. The program centers on the introduction of competitive, regulated markets for gas and electricity to replace the current centralized model that relies on state monopolies and government guarantees. These guarantees have been necessary in the past largely because the legal and regulatory framework didn't provide private investors with adequate safeguards that their investment would earn a competitive rate of return. But the Treasury has been taking almost all the market risk associated with the build-operate-transfer (BOT) and related build-own-operate (BOO) schemes that have been used to attract private financing for power plants. 27. Reliance on government-guaranteed BOT projects, not granted through a competitive bidding process, has led to very high costs for new energy supplies in Turkey. These costs have contributed to a significant decline in the financial condition of TEAS (the state-owned power generation and transmission company). These guarantees also represent large contingent liabilities for the public sector. If TEAS is unable to meet its obligations under Figure 5: Indicative Contingent Liabilities in Power IS unabte to meet IS obugatlons unaerSector (US$ million) the power purchase agreements, the Sector_(US$_million) Treasury is required to meet the costs (figure 5). Based on stylized 1,500 8,000 assumptions, without real price increase's 11200 ----6,000 900 - --- or other positive action, TEAS will face 600 4,000 mounting costs and consequent losses 300 2,000 that could reach some US$1.5 billion a year by 2010. °° @ P 28. Faced with the choice of raising U CurnWative n Annual tariffs to unacceptably high levels, La_Cumulative_@l_Annual__ cutting other expenditure to finance ever Source: World Bank. larger subsidies from the budget, or undertaking fundamental reform of the energy sector, the government has opted for the third choice. It has decided to introduce a regulated market model: the so-called bilateral model. Under this model the government will largely withdraw from the electricity generation and distribution businesses. Electricity generation plants will sign contracts for power directly with distribution companies without government guarantees. The government's role will be refocused on determining sector policy, owning the transmission system and setting up an independent regulatory commission to make sure that the rules are respected and that prices are competitively determined. The bilateral market model, successfully implemented in Latin America, is fully compatible with EU energy directives. As part of the reform, TEAS is being restructured into separate generation, transmission and trading companies. State-owned thermal generation plants and distribution companies will be divested, either under the current transfer of operating rights viii (TOOR) program or through outright sale of the companies after 2000. Similar reform plans are being prepared for the gas sector. 29. Managing the transition to the market model requires careful attention to address potential power shortages. Production under existing contracts (including additional capacity planned under TEAS's own investment program) is expected to be sufficient to meet demand (projected to grow by some 9 percent annually) through at least 2005. But small shortages of up to 2-3 percent of demand may arise during 2001-02, if the current, below-average hydropower conditions persist. The government is considering issuing limited guarantees for additional projects that could be realized in time to cover this potential shortfall. Based on international experience, it is expected that private investors will finance adequate expansion of electricity capacity after 2005 under the market model without government guarantees. Telecom reforms are progressing well, but serious challenges lie ahead 30. There has been good progress with reforms in telecommunications, and decisive steps will be required to sustain the momentum. Legislation to reform the telecommunications sector was approved by Parliament in January 2000. The new legislation, comprising amendments to several existing laws, clears the way for opening the capital of Turk Telekom to private sector participation. It also establishes an independent regulatory authority for telecommunications. And it enables private provision of all value-added and wireless services. 31. The new telecoms legislation initiates the process of deregulating the sector over the medium term. It sets out a broad market structure for telecommunications in Turkey and defines a four-year transition to this new market model. Building on this progress, the government successfully tendered a third GSM license in April, raising an unexpectedly large US$2.5 billion in fees and a further US$400 million in VAT. Two further GSM licenses are planned, including one for Turk Telekom. The next step is to sell a 20-percent stake in Turk Telekom to a strategic investor. This sale, expected to be completed in late 2000, will be followed by a public offering and other share sales, bringing the state's holding down to 51 percent. 32. These structural measures are setting the stage for broader liberalization of the telecoms sector. Turk Telekom' s exclusivity in long-distance and international telephony will be lifted by the end of 2003 in line with Turkey's commitments to WTO. Further steps will be needed to ensure full compatibility with EU norms-establishing clear criteria for approving licenses, and clarifying tariff policy, the interconnection regime, and universal service obligations. The current system of agriculture support policies is holding back growth 33. Most observers agree that Turkey's agriculture sector has tremendous potential, but this has gone largely unrealized in part because of distortionary and outmoded support policies. The agriculture growth rate has been trending downward for the last decade. There is some evidence that agricultural growth has been falling due to stagnant or even declining technical efficiency measured by total factor productivity. This is a startling conclusion considering the enormous advances in agricultural efficiency worldwide. The implications for Turkey are far reaching. Agriculture accounts for some 14 percent of GNP and a much larger 45 percent of employment, although this has fallen considerably from around 65 percent in 1970. The decline in agricultural ix growth is all the more startling given the very high-and increasing-transfers to agriculture. According to official OECD estimates, total transfers, including from consumers, were US$12.1 billion in 1997 and US$15.7 billion in 1998, or well over half of agricultural GDP (figure 6). Total transfers have increased, as a percentage of GDP, from 4.8 percent in 1986-88 to about 6 percent in 1996-98. At the same time, the corresponding figure for all OECD countries has fallen from 2.5 percent of GDP to 1.3 percent. 34. A key problem is the structure of Figure 6: Agricultural Growth and Transfers agriculture support which has traditionally - been channeled through a complex maze of 10 18 input and credit subsidies, and support prices 8___l __16 for selected crops. This system has 4 14 _ necessitated extensive government 2- 12 E involvement in agriculture processing and 0 marketing, managed through a network of 4-4 SOEs and state controlled cooperatives. The - [ ' system has stifled private initiative in ,0 J 2 agriculture and placed a heavy burden on 1988 1990 1992 1994 1996 1998 consumers as they bear the majority of the - Ag Support -Growth rate Trend (Growth) costs (2/3 or more). This burden falls disproportionately on low- and middle-income Source: SIS Treasury and World Bank estimates. consumers, including the landless poor in rural areas, who spend a greater proportion of their incomes on food. Many other countries have also pursued input subsidy and output price support policies similar to those used in Turkey. But in recent years, most developed countries and many developing countries have moved away from these policies. A major agriculture reform is underway 35. As part of its reform agenda, the government has developed a strategy to phase out the current agriculture support mechanisms and replace them with "decoupled" direct income payments, which would not be contingent on input use or output production decisions. This reform is designed to ensure some stability for farmers' incomes while freeing them to make more efficient production decisions. Although import protection will remain significant, it will be reduced, and domestic agriculture prices will be linked to world prices, allowing them to reflect more accurately underlying economic values. 36. The direct income support system is being piloted in 2000 with the objective of rolling it out at the national level in 2001-02. Payments under the system will be only moderately targeted at first, based on a cap on the total amount per farmer. But they could become more targeted over time toward poorer farmers. Under the reform, the fiscal cost of existing agricultural policies is projected to drop from about US$5 billion in 1999 to US$400 million by 2002. How much of the savings will be channeled to direct income payments has not yet been decided. This choice will have to balance affordability, political acceptability, equity, and other considerations. The payment amount should not be set so low that there is a sharp up front decline in farmers' incomes. It should be reduced over time as agricultural growth recovers and direct income support becomes more targeted or is merged with the general social protection system. x 37. The government has started the support policy rationalization. The fertilizer subsidy, held constant in nominal terms since 1997, will remain so through 2001. In December 1999, the govermnrent issued a decree to phase out credit subsidies during 2000. Under the decree, nominal interest rates on subsidized credit are to be held constant until they equal the three- month rolling average of the 12-month t-bill rate plus a 500 basis point markup. This point spread will be maintained thereafter. Given the rapid fall in interest rates in early 2000, the credit subsidy has been effectively eliminated. On agricultural price supports, the government has announced a set of policy changes that introduce links to relevant world market prices for grain and initiate a phase-out of government subsidies for price supports by 2002. Price supports for grains in 2000 have been set at levels that reduce the premium over world prices to no more than 35 percent. Import tariffs on grains will be reduced as well. 38. The agriculture reform program encompasses commercialization and privatization of state assets in the sector ,vith the medium-term objective of withdrawing the state from a direct role in agricultural and agro-industrial production. In the first phase in 2000 the government is imposing a hard budget constraint on agricultural SOEs, including specific enterprise-by- enterprise limits on Treasury loan guarantees, equity injections and budgetary transfers. In parallel, privatization proceedings will be initiated for agricultural SOEs, including those active in tobacco, tea, and sugar beets. A law was passed in May 2000 giving complete operational and financial autonomy to the agriculture sales cooperative unions (ASCUs), which dominate marketing of many cash crops, including eliminating any budget support or other subsidies and government role in their operation. The law establishes a restructuring board to oversee the process of transform the ASCUs into true private cooperatives over four years. Longer term rejorms will be needed to help farmers adjust to the new environment 39. Turkey needs to confront several other agricultural issues over the medium term to help farmers thrive in the new incentive environmnent. One issue is irrigation policy. Many more irrigation schemes have been started than there are funds to complete within a reasonable timeframe. An average completion time of 11 years lowers the economic rate of return on irrigation investment to unacceptable levels. The government should desist from starting new schemes, identify the most cost-effective projects- and reallocate available funds to complete these schemes. 40. A second area for reform is forestry. While the current approach is not generating large explicit fiscal costs, institutional and policy deficiencies are limiting the ability of the sector to provide sustainable income generation for forest communities and revenue for the budget. Needed are a restructuring of the forestry-related organizations, privatization of commercial activities, devolution of some responsibilities to local communities, and a refocus of central government responsibiiities on legitimate functions that require national coordination. 41. The government should also consider institutional changes to the Ministry of Agriculture and Rural Affairs (MARA) to improve its ability to carry out its responsibilities in the new policy environment. This would likely involve changes in the way research is undertaken along the lines piloted in the Agricultural Research Project-supported by the Bank. Restructuring extension services is also called for-to improve links with research and to promote private sector provision of these services in line with international best practice. xi A sound banking sector will support stability and growth 42. Recognizing the importance of an efficient banking sector and acknowledging the risks that the disinflation program poses for the weaker banks, the government has launched a broad based financial sector reform program. This program has three core components: * Modernizing the legal, regulatory and institutional framework for banking supervision in accordance with EU and world standards. * Resolving problem private banks. * Reforming the state banks. 43. The potential of Turkey's emerging banking sector to support stability and growth has been severely constrained by distorted incentives arising from longstanding macroeconomic imbalances and a weak supervisory regime. The incentive structure for banks has suffered from chronically high and erratic inflation, increasing public sector borrowing requirements, and the use of tax breaks to favor placement of government debt over private sector borrowing. This enviromnent has also encouraged fragmentation of the sector-with the proliferation of small banks dependent on high real returns from government securities. At the same time, limitations in the legal and regulatory framework for banking supervision have allowed a number of private banks to accumulate weaknesses in their asset portfolios. The financial ratios of the state banks, which account for some 40 percent of all bank assets, have also deteriorated. The main state banks have built up large and growing stocks of illiquid claims on the Treasury, the so-called "duty losses", as a result of government-mandated programs and subsidized loans. Important actions to strengthen the financial sector are underway 44. The financial sector reform has started strongly-with enactment of a new banking law, up front action on private bank resolution, and steps to staunch the flow of duty losses by phasing out subsidized credit programs. The new banking law adopted in June 1999 and subsequent amendments adopted in December have created the legal foundation for modernizing the banking sector. Building on the new law, updated prudential regulations in line with Basle and EU standards have been enacted which provide for a phase-in period for the banks to reach full compliance. A new bank regulatory and supervisory agency (BRSA) has been established and is expected to be independent from political interference. 45. In late 1999 the government used the authority provided under the new law to transfer five insolvent banks to the Deposit Insurance Fund. The resolution of all eight banks currently under the Fund is underway. But much more is needed to overcome the structural problems in the banking sector accumulated over decades of macroeconomic instability. Consolidating the new legal and regulatory framework, overseeing the transition process for the private banks, and carrying out the restructuring and privatization of the state banks top the list of priorities. xii Consolidating the new frameworkfor the banking sector 46. The BRSA is taking over bank regulation and supervision responsibilities fulfilled by the Treasury and Central Bank. The previous supervisory regime was fragmented and contained several serious flaws that contributed to the emergence of a relatively large group of problem banks. So, unifying the separate banking regulatory and supervisory functions under one agency will be a major achievement of the new banking law, as long as BRSA can establish the requisite autonomy from the political sphere. In this respect, it is crucial that BRSA earn a strong reputation. Indeed, credibility, high-quality assessments, and broad visibility will be essential for BRSA's recommendations and actions to carry the necessary weight. 47. In addition to changing banking legislation, important regulatory actions have been taken including: * A much more stringent loan loss provisioning rule in line with international best practice. * Large exposure/connected lending limits which will be consistent with applicable EU directives once fully phased in. * A new capital adequacy rule allowing its application on a consolidated basis. * A much lower foreign exchange exposure limit. Revisions of accounting and disclosure standards and regulations on risk management standards are also planned. 48. The next step is to implement fully these new regulations and to monitor carefully compliance by the banks as the new standards are phased in. The new incentive framework emerging from the disinflation process, together with the new, tougher regulatory environment, will generate significant pressures on the banking sector to change, including the likelihood of significant mergers and shakeouts. The result will be a much stronger, more competitive and resilient banking sector. But the transition will require active oversight by the authorities. In this process, some resistance from the banks could arise, and the authorities will have to work closely with the leaders of the banking community to Figure 7:_Duty Losses of State Banks ensure that the shift to the new environment is as - - -- smooth as possible. 14 12 Reforming the state banks t 8 49. Reforming the state bank sector is essential a 6 to improving the strength and efficiency of the 4 m banking system as whole. The predominance of 2 m state banks in Turkey's financial system is a legacy 1995 199 1997 of the past. Some of these banks are not only 1998 1997 99 19 plagued by duty loss problems and overstaffing, but _E their governance and capital structure do not allow Source: Treasury and World Bank. them to make the financial and human capital investments necessary to compete and thrive in increasingly competitive domestic financial markets. The state bank reform process will have to be sequenced, well timed, and decisively implemented. The objective is to privatize the four state-owned commercial banks (Vakif Bank, xiii Halk Bank, Emlak Bank and Ziraat Bank) over the medium to longer term. As a first step, legislation to privatize Vakif Bank has been adopted. Framework legislation for the commercialization and privatization of the other three banks is under preparation. For some of the state banks-notably Halk and Ziraat- which have functioned in large part as government agencies-commercialization will be a complex process. It must encompass resolution of the duty losses which have reached some US$20 billion on paper (figure 7). The Government expects Vakif Bank, Emlak and Halk to be privatized by 2002. Privatization of Ziraat- will likely take longer, given its extensive involvement in the support system for agriculture. Medium-term outlook 50. Medium-term projections show that a period of sustained fiscal adjustment will be required if Turkey is to attain its disinflation and growth targets. Supported by the IMF and World Bank, the stabilization program targets a ' . . . ..................Figure 8: Projected Growth & Inflation decline in CPI inflation to single digits by 2002, compared with the 69 percent in 1999 90 - - 8 (figure 8). In parallel, real interest rates are 70 projected to fall below 1O percent. In tandem 60- with falling inflation and interest rates, the 0_50___I____ government's macro policies aim to pull the 30 (2e economy out of recession. The program targets 20 - _4 - ) 10 ~~~~~~~~~~~~~~~~~~~~~(6) annual growth in the range of 5-6 percent over (8) the 2000-03 period, as compared to the drop in 1999 2000 2001 2002 2003 output of 6.4 percent recorded in 1999. r3 oustaned Reform Truncatrd Reform -Sustained Reform -Truncated Reform 51. Attaining these targets requires Source: World Bank. continued tough actions to address structural problems in areas such as public expenditure management, state enterprises (including state banks), social security, the regulatory framework, subsidy programs and privatization. Under a Sustained Reform Scenario, the pace and Figure 9: Per Capita Income (US$) quality of fiscal adjustment would be sufficient -- to yield a permanent reduction in the PSBR and stabilize the public debt (chapter 6). An 3 --- aggressive adjustment effort will engender a 3,N robust private sector response and would be 3,2X0 ._, - l consistent with a significant improvement in 3,1_l external credit-worthiness. By 2003, per capita 2-,- income, in constant dollars, is projected to be Li2, l[ 18 percent higher than in 1999 (figure 9). 22_l 1900 2000 2001 22 23 52. A halfway approach to adjustment will i a 9t-Re 9 TrutaMR l not work, for it would pose serious risks of renewed financial crisis in the near term and would not be viable in the medium term. Turkey's creditworthiness would deteriorate, and the economy would remain highly exposed to external and internal shocks. Economic growth would stay well below potential and inflation would remain dangerously high. Mired in stagflation, the xiv government's ability to fund Turkey's social priorities and other development needs would be severely constrained. 53. Under a Truncated Reform Scenario, where economic reform would lose momentum by the last quarter of 2000, the key conditions for sustainable fiscal adjustment could not be met. Revenues would be expected to shrink gradually and expenditures to grow with expansionary fiscal policy. The PSBR would deteriorate and progress in taming inflation would unravel. Faced with an acute trade-off between inflation and growth in a difficult macro environment, policies would likely favor the latter. But stop-go policies, high and volatile inflation rates, and likely balance of payments difficulties would lower growth rates and increase volatility. There would be a severe risk that inflation would climb back into triple digits. The return of inflation and high interest rates would squeeze out private investment. GNP growth in 2000-03 would average less than 3 percent. 54. The key structural policy recommendations to achieve sustainable growth in the medium to long term constitute a six point agenda for the government: * Establish a stable macroeconomic environment that is pro-growth, generating a primary surplus of at least 3 percent of GNP over 2000-03 to underpin macroeconomic stabilization (chapter 1). * Complete reviews of expenditure programs and fiscal institutions and implement policy and institutional reform, so as to ensure the quality of the fiscal adjustment and protect public investments in human capital'. * Give priority to social security and social assistance. Turkey has made good progress with a well-designed policy reform. The government is also initiating an administrative reform to improve coverage and compliance of the PAYG system and introducing a legal framework for voluntary funded private pensions. Over the longer run, further structural change to the social security system is needed to respond to the requirements of an aging population with increasing life expectancy and contribute to the development of domestic savings and growth (chapter 2). • Avoid bottlenecks to growth and an unsustainable build up of contingent liabilities-by deregulating the energy and telecommunication sectors in line with international best practice and EU norms. This requires measures to create an enabling legal and regulatory framework, restructure state monopolies and accelerate privatization. Proper structural reform will pave the way for effective liberalization of the energy and telecoms sectors to promote private investment and competition (chapter 3). * Carry through with the current ambitious agricultural reform agenda to modernize support polices and withdraw the state from a direct role in production and marketing. Round out The govemment is carrying out a Public Expenditure and Institutional Review (with the assistance of the World Bank) and recently completed a Fiscal Transparency Review (with the assistance of the IMF). The importance of quality fiscal adjustment has been highlighted in recent research by Alesina and Perotti (1995) and Easterly (1999). xv the reform agenda with subsequent action to address fundamental structural problems in irrigation and forestry, and reorient state institutions (chapter 4). * Pursue financial sector reform to improve the efficiency of financial intermediation and promote stability. Build on the impressive progress achieved to consolidate the new legal and regulatory framework for bank supervision, oversee the adjustment of the private banking sector to the new incentive framework and regulatory environment, and reform the state banks (chapter 5). xvi CHAPTER 1: MACROECONOMIC FRAMEWORK AND FISCAL SUSTAINABILITY 1 INTRODUCTION 1. Despite GNP growth averaging just under 4 percent a year in the 1990s, Turkey's economy operates under a cloud of vulnerability, plagued by persistent fiscal imbalances, chronically high inflation, and sharp business cycle fluctuations. This chapter shows that fiscal imbalances are the key to understanding Turkey's inflation problem and its volatile growth performance. In the past two decades, several attempts to stabilize the economy have produced mixed results. High growth has never been sustained for very long and earlier stabilization efforts did not address the structural sources of the fiscal deficit. The chapter suggests that the lack of sustained high growth can be closely linked, among other factors, to poor macroeconomic policy in the form of inflation and fiscal deficits. It also argues that previous attempts at stabilization have failed because they did not eliminate the structural sources of the fiscal deficit. The success of the current round of economic reform in Turkey depends critically on addressing these structural imbalances. Equally important, structural reforms will raise productivity by improving the incentive framework, stimulating competition, and encouraging productive private investment. 2. The next section of this chapter provides a brief summary of Turkey's macroeconomic experience in the 1 990s. It illustrates clearly that high growth has never been sustained over long periods and that inflation has remained stubbornly high despite numerous stabilization efforts. The section highlights the fact that by 1999 a major policy change was needed. 3. The third section investigates sources of economic growth and the linkages between growth and fiscal policy. It addresses how the government could influence growth through policy reform aimed at eliminating fiscal imbalances and reducing inflation. The question of whether capital accumulation and productivity growth have been sufficient to sustained rapid growth is also discussed. The text argues that human capital accumulation and total factor productivity growth have been especially sluggish. Next, attention shifts to the volatility of growth. Growth and economic activity could have been much more stable in the 1 990s if Turkey had implemented fiscal policy differently. The section concludes with a discussion of how the on-going stabilization effort will affect economic growth. 4. Since this chapter argues that fiscal policy will play the key role in realizing Turkey's policy objectives, the fourth section looks at how public debt and finances have evolved. It documents that Turkey was indeed plagued by a substantial structural fiscal deficit in the 1 990s. Reliance on significant amounts of seignorage revenue allowed the total stock of public debt to remain relatively stable through 1998 despite the large fiscal deficit. As a result, inflation stayed high throughout the decade . Had the government foregone seignorage revenue, the level of debt would have grown rapidly. In 1999 the inflation rate fell slightly causing seignorage revenue to decline. This, combined with a recession, led to a sharp increase in the level of public debt as a share of GNP. ' It is also possible that the financing mix of the government affected the inflation rate especially in the late 1990s when the government shifted its borrowing substantially to domestic debt. High real interest rates on domestic debt would have exacerbated the inflationary impact of the deficit. 5. The fifth section estimates the fiscal adjustment needed for sustainability-the current reform program appears to be consistent with the estimates-and underscores the importance of credibility. The logic of the government's budget constraint implies that today's inflation rate is a function of the government's current and expected future fiscal policies. Rapid inflation can result from current fiscal imbalances or from the anticipation of future fiscal imbalances. A government that has consistently relied on seignorage to keep its debt in check faces a credibility problem when it tries to control inflation: even if it runs a fiscal surplus today, uncertainty will remain regarding its future decisions. So, the key to reducing inflation in Turkey is for the government to ensure the adequacy and credibility of its fiscal reform effort. 6. The key to credibility, as the sixth section argues, is to have an anti-inflation policy that stresses the importance of long-run structural reform over short-term fiscal and monetary measures. In this regard, the new program may succeed where others have failed. Although some of its short-term fiscal and monetary measures are temporary, the medium and long run fiscal adjustments will come mainly from structural reform of the public sector. These reforms are under way so the program has already gained longer run credibility than previous efforts. Turkey's current political stability, which compares favorably with the prior experience of weak coalition governments, strengthens the prospect for continued structural reform. 7. The seventh and final section of the chapter provides further details of the government's new reform program. It argues that it is designed in line with the need for structural reform. Furthermore, the initial stages of its implementation have been quite successful: inflation and nominal interest rates have declined sharply in recent months. 2 THE MACROECONOMIC BACKGROUND 8. Growth has been volatile and below potential. Over the past decade, Turkish economic growth has been erratic and, on average, well below the levels of the best performing emerging markets. Turkey had a moderate average growth rate of 3.9 percent p.a. during 1990-99. But it Figure 1.1: Turkey: GNP Growth, 1988-1999 should be kept in mind that population growth 12.0 -- - ------ - -- - averaged about 1.8 percent per year over the same 100 period so that average per capita GNP growth was 8.0 6.0 only 2.1 percent. Periods of economic expansion 4.0 have alternated with periods of equally rapid 2.0 J - ____ decline (figure 1.1). Following a year of severe -t (2.0) recession in 1994, the economy went through a (4 0) boom period of above trend growth between 1995 (6.0) and 1997. This was largely due to strong export (8.0) L performance stimulated by the real depreciation of 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 the lira in 1994 and new markets in the CIS Source: SIS and World Bank. countries-the volume of exported goods and services grew by 16 percent each year in 1994-97. There was also a strong boost to business confidence and investment stemming from the 1996 customs union with the EU. However, the economy was badly hit by the Russia crisis in 1998 and output shrank by 6.4 percent in 1999. 2 9. Fiscal imbalances have been large. Figure 1.2: Turkey: PSBR and Interest Unsustainable fiscal policies have typically driven Payments (in % of GNP) periods of macroeconomic instability in Turkey 2S (figures 1.2). The financial crisis in early 1994 coincided with a public sector borrowing 20 requirement (PSBR) consistently over 10 percent of PSBR GNP in the previous three years. After a sharp 15 positive adjustment, a result of the 1994-95 oi stabilization package, the fiscal position weakened anew in 1996-97, reflecting a sharp increase in 5 public sector wages, a widening social security deficit, generous hikes in agricultural support prices, 1990 1991 1992 1993 994 1995 1996 1997 1998 1999 and transfers to prop up financially weak state banks. - With access to external financing constrained and Source: Treasury and World Bank. large external debt repayments scheduled, the government was forced to increase its reliance on monetization and domestic financing. This fueled Figure 1.3: CPI Inflation (%) inflation (figure 1.3) and a rapid accumulation of 130 ---- - ----- domestic debt. As the government's domestic debt 120 110 burden increased, the social consequences of 100 persistently high inflation and large fiscal deficits became more obvious with growing income 90 inequality and constrained public spending on social |0 70 and economic infrastructure. 60 10. With inflation accelerating to more than 100 50 percent by the beginning of 1998, the government 'o1< o .:of &?' o_ N >3 , N launched yet another stabilization program. __ However, fiscal policy became expansionary in the Source: SIS and World Bank. second half of 1998 and the first half of 1999 when early elections were called and economic activity was slowing. The expansion was expenditure- driven, with an increase in public sector wages in 1999 from 7.2 to 8.4 percent of GNP being a major contributor. Transfer payments and tax revenue were both hit by the economic downturn. Although the fiscal expansion slowed toward the end of 1999, the achievements of 1998 were nearly wiped out by the end of the year as the consolidated public sector PSBR reached a record 23 percent of GNP. The cost of the fiscal expansion was a rapid increase in public debt. With the elimination of the public sector primary surplus, the stabilization program lost credibility. Having shrugged off the turmoil in Asia, the economy proved vulnerable to the emerging-market crisis that followed the Russian default. The second half of 1998 was particularly difficult-with massive capital outflows, rising real interest rates, and declining economic activity. 11. Through much of 1999 nominal interest rates remained high despite somewhat lower inflation, reflecting the market's belief that the mix of loose fiscal and tight monetary policies could not be sustained. As a result, the real interest burden of the public sector jumped from 5.4 3 Real interest rates, defined as the interest rate on 3-month time deposits minus a 3-month moving average of current and lagged monthly (annualized) inflation rates, peaked at almost 33 percent in July 1999. 3 percent of GNP to 10.9 percent of GNP. This combined with the deterioration of the primary balance of the public sector from 0.8 percent to -2.7 percent of GNP to drive up the operational deficit of the public sector by 9 percentage points from 4.6 to 13.6 percent of GNP. Public sector debt rose from 44.5 percent of GNP to 58 percent of GNP. Two major earthquakes in August and November 1999, made the fiscal problems and economic outlook even worse, and the economy contracted by 6.4 percent during 1999. 12. Policy clearly could not continue on this path for long, and the new government installed following elections in April 1999 formulated a new economic program founded on fiscal adjustment and deep structural reform. The government reached agreement with the IMF in late 1999 on a macroeconomic framework for 2000-02. 3 GROWTH IN TURKEY 13. This section examines economic growth in Turkey and addresses several questions. The first subsection analyzes the sources of economic growth. It begins by discussing how the government could influence growth through policy reform by eliminating fiscal imbalances and reducing inflation. It also asks whether capital accumulation and productivity growth have been sufficient to sustain rapid growth. It argues that human capital accumulation and total factor productivity growth have been sluggish. The next subsection looks at the volatility of growth. It argues that growth and economic activity could have been much more stable in the l 990s if Turkey had corrected its fiscal imbalances. In particular, fiscal policy in the 1990s was largely procyclical, providing stimulus during expansions and exacerbating cyclical downturns. The final subsection concludes with a discussion of how the ongoing reform effort will impact on economic growth. 3.1 Sources of Long-Run Growth 14. This subsection focuses on the fundamental determinants of growth: capital accumulation and total factor productivity (TFP). It argues that these determinants of growth are, in part, a function of the macroeconomic policy environment and that this is important to explaining Turkey's poor growth performance. It also argues that the levels of human capital and TFP are especially low for a country of Turkey's income level so that structural reform may be best focused on policies that will raise human capital and TFP, rather than on those aimed at increasing the rate physical capital accumulation. 3.1.1 The Impact of Macroeconomic Policies 15. Macroeconomic policies have a direct impact on economic growth through TFP, and an indirect effect through the rate of capital accumulation. This is illustrated in Burnside and Dollar's (2000) recent work on the effects of foreign aid on growth. They find that openness, sustainable fiscal policy and low inflation are all conducive to higher growth. Their research proposes a policy index that summarizes the growth impact of openness and a country's macroeconomic policies. In particular, the index is given by: (1) policy index = 1.26 + 6.4 x budget surplus - 1.4 x inflation rate + 2.2 x openness 4 where the budget surplus is measured relative to GDP, and the openness measure is from Sachs and Warner (1995). The policy index measures the per capita growth rate that a country's economic policies would produce if all other country characteristics were equal to their averages in a cross-country database., The quantitative findings reflected in the coefficients in equation (1) are that macroeconomic policies can have a significant impact on growth. 16. Turkey has become an open economy since 1980, but there is plenty of room for further policy improvement. At the beginning of the 1980s, Turkey would have rated very poorly according to the Burnside-Dollar index-for three reasons. First, Turkey was a relatively closed economy (Sachs and Warner give it a "closed" rating until 1989) at the beginning of the decade. Second, Turkey suffered fiscal imbalances for most of the 1980s. The central government budget deficit averaged about 5 percent of GDP over the period. Third, the inflation rate hovered near 40 percent. Together these facts imply a policy index near 0.4 in the 1980s, well below the sample average of 1.2 across the 56 developing economies in the Burnside-Dollar data set, most of them less developed than Turkey. However, starting in the early 1980s, a number of trade reforms were initiated and the government carried out a deliberate policy of real exchange rate depreciation. Restrictions on imports eased beginning in 1984 and there was a major tariff reduction in 1989. By 1990, almost all import restrictions had been removed. The capital account was also liberalized. Foreign currency deposits were legalized in 1984, and by 1989 capital flows were unrestricted. Overall, Turkey was a fairly open economy by 1990. By becoming more open, Turkey expanded its long-run growth potential substantially-by 2.2 percentage points per year according to the Burnside-Dollar index. 17. To take full advantage of the impact of increased openness on growth, Turkey must embark on a new round of reform aimed at restoring sound fiscal policy and reducing inflation. In the mid to late 1990s, Turkey suffered high budget deficits and inflation rates that worked against the improvement in the policy index resulting from greater openness. Because of deteriorating fiscal policy and inflation, Turkey's policy index toward the end of the 1990s was barely above that of the average developing country. A similar conclusion is drawn by another study that concludes) that per capita output growth in Turkey could have been 1-1.5 percentage points higher in the 1990s if inflation had been in single digits., 3.1.2 Structural Factors 18. This subsection turns to a growth accounting exercise to analyze more deeply where the shortfall in growth in Turkey is coming from. This analysis is meant to shed light, in particular, on those structural elements of government policy that could foster growth. Should the government encourage investment in physical or human capital through direct incentives, or should its efforts be directed towards improving TFP? These questions are not easily answered because there is no complete consensus among economists about the relative importance of each 'The coefficients in the index are obtained from a panel regression where real GDP per capita is the left-hand side variable, and the right-hand side variables include a variety of institutional variables and development indicators as well as the three policy variables mentioned above. See Burnside and Dollar (2000) for details. This study, A. Ghosh and S. Phillips, "Warning: Inflation May be Harmful for Your Growth," IMF Staff Papers, December 1998. controls for determinants of growth other than inflation. 5 determinant of growth., The discussion below explores whether Turkey's rates of investment in physical and human capital are unusually low compared with other emerging countries, or whether its productivity growth is relatively slow. Box 1.1: Employment and Growth Output per capita is given by: (2) output output workers working age population (2) = x x- population workers working age population population The ratio of the working age population to the overall population is largely determined by demographic factors that presumably move slowly. Althog governments may not have much influence on this ratio they need to care about it because demographic trends can have an enormous effect on government budges especially through social programs. But im terns of grt the other two terms are the main focus here. The analysis in the main text uses data on output divided by the working age population but intepret it as if it were a measure of output per worker. If employable workers leave the workforce, productivity growth will appear to slow down even if true producfivity is growing quite quickly. A recent World Bank study shows that Turkey has had disappointing per capita income growth for two reasons.' Output per worker is rising but not as fast as in countries, such as Spain and Korea, which have experienced episodes of rapid growth.8 This could be explained, as in the main text, by relatively slow TFP growth or capital accumulation. But Turkey has also seen the fraction of its working a oulation that is employed fall dramatically. According to the same study, the fracton of Turkey's potential labor fobrce at was economically active fell from about 69 percent in 1975 to about 50 percent in 1997. This put Turkey second from the top in the OECD in 1975 only to fall to second bottom in the OECD in 1997. What makes thesefigures all the more remarkable and important is that the pattern is preserved if one looks exclusively at prime-age (25-54 years old) individuals whose decisions regarding entering the labor force are less likely to be affetW by health and education trends. Why is the employed fraction of the employable population shrinking? Is it supply drivendoes it depend on supply factors other than the real wage, such as a socio-demographic change in favo of home work? Or is it driven by some sort of labor market inieffciency The imovement of people to the cities may have led to a change in labor force participation rates; higher incomes in the cities may have led to families substituting toward less measured employment. But excessively high real wages in the public sector may have led to declining employment in the private sector (rather than falling wages) for standard efriciency-wage reasons. 19. A simple identity indicates that output per capita is output per worker times the number of workers per capita. The discussion here focuses mainly on output per worker while box 1.1 discusses employment creation. To discuss the sources of growth in output per worker, this I An ongoing debate in economics concerns the explanation of cross-country differences in income levels and growth. Some economists argue that the important determinant of growth is capital accumulation both physical and human, while others argue that more important, but less understood, is total factor productivity. Mankiw, Romer and Weil (1992) argue that over three- quarters of the cross-country variation in income levels can be attributed to cross-country variation in stocks of physical and human capital. Klenow and Rodriguez-Clare (1997) argue that the correct fraction is more like one-third, when human capital is measured more carefully. They also argue that almost all of the cross-country variation in growth rates is due to variation in total factor productivity growth. Easterly and Levine (2000) reinforce their findings by showing that empirical testing soundly rejects economic models where capital accumulation is the driving force behind growth. See World Bank (2000), chapter 1. In Spain's period of rapid growth (1964-90), output per worker rose almost 8 percent per year in the industrial sector. In Korea, between 1965 and 1989 output per worker in industry grew at more than 16 percent per year. In Turkey, between 1981 and 1997 the comparable number was less than 4 percent per year. 6 section takes a production function approach that decomposes growth into three components: the part due to the accumulation of physical capital, the part due to human capital accumulation, and the part due to increases in total factor productivity (TFP). The appendix provides the theoretical details of this decomposition., For 1960-85, the estimate of the growth rate of output per worker for Turkey from the Penn World Tables was about 3.2 percent a year. A simple decomposition of this growth into its sources suggests that 1.3 percentage points of it were due to the accumulation of physical capital, 1.0 to human capital accumulation, and just 0.9 to TFP growth. 20. Klenow and Rodriguez-Clare (1997) compare income levels in 1985 across 98 countries. According to their data set, at that time, output per worker in Turkey was 21 percent of its level in the United States.', But physical and human capital per worker were 17 and 8 percent, respectively, of their U.S. levels. Using Klenow and Rodriguez-Clare's assumptions about the shares of capital and labor in national income, this means the combined capital gap relative to the U.S. explains about 74 percent of the income gap, while the remaining 26 percent is explained by a TFP shortfall., 21. Klenow and Rodriguez-Clare's data set also compares income growth rates from 1960 to 1985 across the same 98 countries. They find that GDP per worker in Turkey grew at an average rate of 3.2 percent a year over this period, ranking it 26th of 98 countries. Turkey's growth rates of physical and human capital per worker over the same period were 4.4 and 3.6 percent, respectively. This implies that TFP growth was still 25th out of 98 countries.,2 Of the 25 countries that grew faster than Turkey, 19 had faster productivity growth, 17 had more rapid physical capital accumulation, but all of them had more rapid human capital accumulation. 22. The literature that uses regressions to identify factors that influence long run growth seems to confirm these findings. An authoritative paper in this field is Levine and Renelt (1992) in which the authors identify robust explanatory variables for growth. Using their analysis Polastri (1999) shows that over the last two decades Turkey's growth could have been higher if it had had a higher secondary school enrollment rate and a higher ratio of investment to GDP. This follows from Levine and Renelt's baseline equation, which implies that: (3) per capita growth rate = other factors + 3.17 x secondary school enrollment rate + 17.5 x investment rate. 9 The growth accounting done here assumes a neoclassical production function with constant returns to scale. The analysis in this section, while sensitive to these assumptions, is meant to be suggestive rather than conclusive. Statements about relative growth rates of output, physical and human capital are invariant to the model. Statements about TFP and the decomposition of growth into its three sources, however, are sensitive to the modeling assumptions. 0 Unfortunately their data set defines the number of workers as the number of people in the working age population so it does not allow for relative efficiency of the labor market in different countries. " Klenow and Rodriguez-Clare assume that the shares of physical and human capital are 0.3 and 0.28 respectively. They measure productivity as labor-augmenting so their measure of the productivity gap is larger, at about 52 percent. 2 Klenow and Rodriguez-Clare's estimate of productivity growth is 2.0 percent because they measure the labor-augmenting level of technology rather than total factor productivity. 7 so that growth is enhanced with higher secondary school enrollment rates and greater investment. Polastri (1999) presents evidence that over the last two decades Turkey has fit this regression line rather well. 23. Despite its very real achievements in raising human capital, it is instructive to compare Turkey with other countries at similar income levels. The most recent figures for Turkey suggest a gross secondary enrollment rate of 56 percent compared with enrollment rates as low as 30 percent in the mid- 1970s. Clearly this implies a significant acceleration in human capital accumulation. Turkey's GDP per capita in PPP US$ was about US$6,350 in 1997 according to World Bank figures. Consider the 13 countries in the World Bank data set for which enrollment data are available and for which GDP per capita in 1997 was between US$5,000 and US$7,700. For these countries the average gross enrollment rate was about 69 percent in 1994. Turkey's enrollment rate was 13 percentage points lower at that time. Had it simply maintained the average for countries at about the same income level its annual per capita growth rate would have been about 0.4 percentage points higher according to Levine and Renelt's analysis. Looking at the countries that have grown very rapidly in recent decades-such as Korea-reveals much higher enrollment rates. Korea's gross enrollment rate was already close to 100 percent in 1994. If Turkey improved its secondary enrollment by a further 20 percentage points, its growth rate would rise by 0.6 percentage points a year according to Levine and Renelt's estimates. It is not only through secondary schooling that Turkey's level of investment in education could be brought more in line with that of its competitors. Total spending on all forms of education shrunk by roughly half, as a percentage of the government budget, between 1992 and 1998. The current literacy rate of 87 percent, though higher than in 1990, could still be substantially improved. 24. Turkey has also made inroads in improving human capital in other dimensions besides education, but much more could be done. The 1990s have seen significant declines in infant mortality as the number of physicians and hospital beds per capita have risen. Social and health insurance cover a larger fraction of the population than in 1990. However, the pension system covers less than 70 percent of the labor force and payroll taxes are punitively high, yet the social security system is financially unsound due to overly generous benefits. While the health insurance system is financially in surplus, it is underused and health outcomes remain unsatisfactory for a country with Turkey's income. Less than 70 percent of the population is covered by health insurance. 25. For physical capital, data from the State Planning Organization show that Turkey had an investment rate of about 24.5 percent of GDP in the 1990s. For countries at roughly its income level, the investment rate averaged 23 percent of GDP according to World Bank figures, slightly less than the value for Turkey. Turkey's additional investment effort relative to the average was worth about 0.25 percentage points of growth according to equation (3). This combined with the evidence from Klenow and Rodriguez-Clare, suggests that Turkey's main emphasis need not be on more rapid capital accumulation. It would appear than human capital and productivity growth are more important areas for attention. 26. The remaining factor explaining growth, TFP, itself has a number of determinants, some of which are probably exogenous and outside the government's ability to influence. But TFP is surely affected by the policy environment within a country. The reasoning is straightforward- 8 government policies often introduce distortions into the economic environment that affect the efficiency with which factors of production and resources are allocated. Cross-country differences in TFP will reflect cross-country differences in how distorted the economic environment is. Therefore, the analysis suggests that TFP can be significantly increased and growth performance enhanced by structural reforms to remove distortions and improve the environment for private investment, in combination with credible macroeconomic stabilization policies. 3.2 Sources of Volatility 27. Turkey experienced highly volatile Table 1.1: Growth Volatility in Emerging Economies growth during the 1 990s. Real GNP growth 1 1980s was as high as 9 percent in 1990, and as low Sd Nimn CoV Sd n GoV as -6 percent in 1994 and 1999. The Thailand 6.4 5.2 1.2 3.2 7.5 0.4 standard deviation of GDP growth over the Turkey 5.5 3.9 1.4 2.7 4.8 0.6 decade was 5.5 percent (or 5.9 percent for Argentina 5.4 4.2 1.3 5.4 -1.0 5.4 GNP). An international comparison of Malaysia 5.2 6.9 0.8 3.5 5.6 0.6 volatility is provided in table 1.1, which Korea 5.0 6.3 0.8 2.4 9.1 0.3 presents statistics on the standard deviation, Chile 3.8 6.4 0.6 7.2 3.3 2.2 mean and coefficient of variation of annual Mexico 3.5 3.3 1.1 4.1 1.5 2.7 growth (GDP) across a group of 1O Philippines 2.4 2.8 0.9 5.2 1.6 3.3 emerging economies. In the 1980s, Turkey Brazil 2.3 2.3 1.0 4.2 2.4 1.8 had the third lowest volatility as measured India 2.2 5.7 0.4 1.9 5.9 0.3 by the standard deviation of its growth rate. Sd: Standard Deviation By the 1990s, it had the second highest CoV: Coefficient of Variation volatility while its mean growth rate fell. Source: Uygur (2000). 28. Higher volatility does not lead to higher growth. At least in this small sample, countries with higher volatility tended not to grow faster. If decade averages are removed from the data the correlation between the standard deviation of the growth rate and the mean of the growth rate is -0.29, and is significantly different from zero.l3 Countries in the sample that have achieved high growth have done so without accepting higher volatility-the coefficient of variation declines sharply as the mean growth rate rises. 29. On the production side, both growth in agriculture and services in Turkey are somewhat less variable than output growth as a whole, while industrial production is slightly more variable than output (table 1.2). Agricultural output growth is the least highly correlated with output as a whole, with a correlation of just 0.49. As examples of this lack of correlation, total output fell 5.5 percent in the crisis year 1994, while agricultural output only fell by 0.5 percent. And in 1995-97 output as a whole grew by more than 7 percent in each year, while agricultural output grew at an average rate of just 1.4 percent, less than its average over the decade. Then in 1998, as industrial and service output growth slowed, agriculture had its best year of the decade, expanding 7.6 percent. "'There is a strong and significant negative correlation between the coefficient of variation and the mean growth rate. Because the standard deviation and the mean are roughly uncorrelated, this negative correlation follows from the construction of the coefficient of variation, which is the standard deviation divided by the mean. When decade averages are not removed from the data, the correlation between the standard deviation and the mean is -0.28 and remains statistically insignificant. 9 30. On the expenditure side, the volatility of the various demand components resembles that observed in several open industrialized Table 1.2: Volatility of Growth in Turkey, 1990-1999 economies. At first glance it appears that peranntincome theories of - Sd permanent inoetere fRelative Correlation consumption would not explain Turkish Sd to GNP with GNP data, since private consumption growth GNP 5.9% 1.00 1.00 is about as volatile as GNP. But it is Production Account important to decompose consumption Agriculture 4.3% 0.71 0.49 into its durable and nondurable Services 5.8% 0.97 0.61 components because spending on durable Industry 6.1% 1.03 0.96 goods is better thought of as investment. Expenditure Account Private Consumption 5.8% 0.97 0.91 It is possible to make this distinction for Excluding Durables 3.1% 0.52 0.83 Turkey as the national accounts data Durables 13.8% 2.32 0.84 separately account for durable, semi- Public Consumption 3.5% 0.59 0.54 durable, and other consumption Private Investment 16.1% 2.71 0.83 purchases. With this distinction, the Public Investment 18.2% 3.06 0.63 patter predicted by permanent income Exports 8.5% 1.43 0.38 ptteorne is observed. Nondurable Imports 19.1% 3.21 0.90 theories iS observed. Nondurable Sd StnadDeito Sd: Standard Deviation consumption growth is only half as Source: World Bank. variable as GNP growth, while purchases of durables and semi-durables are twice as volatile as GNP growth. 31. With regard to the other expenditure components, public sector consumption growth, 40 percent less volatile than output growth, is not highly correlated with output growth as a whole, with a correlation coefficient with GNP growth of 0.54. The public sector wage and salary bill represents more than half of this category of spending but contributes very little to its volatility or cyclicality. The standard deviation of the growth in wage and salary payments is just 1.8 percent, its correlation with GNP growth is about zero. But other public consumption purchases, probably more discretionary, are 45 percent more volatile than GNP and are more procyclical. Investment spending by both public and private sectors is highly volatile, with private investment growth being 2.7 times as volatile as GNP growth and public investment growth 3.1 times as volatile. Private investment is more strongly procyclical than public investment for two reasons: private investment is more sensitive to other cyclical factors, while public investment purchases are sensitive to the needs of fiscal restraint. For example in 1995 when GNP and private investment grew annually at rates of 8 and 18 percent, public sector investment declined by 12 percent as public sector finance continued to bear the brunt of the 1994 crisis. Exports of goods and services, about 40 percent more volatile than output as a whole, are not very highly correlated with output. This shows that export sales can benefit from real currency depreciations that take place in crisis years, and that exports are sensitive to foreign demand shocks. Imports of goods and services, in growth terms, are more than three times as volatile as GNP and strongly procyclical. This probably reflects the volatility and procyclicality of spending on fixed investment and durable goods that together are a significant fraction of imports. 32. Why is there so much volatility in the Turkish economy? One hypothesis is that the economic environment is very uncertain-characterized by volatile real interest rates- so that firms and households find long-run planning to be difficult. In times of economic crisis or heightened uncertainty, they postpone important investment and consumption decisions until 10 stability returns, adding to the effects of any economic shock. A second explanatory factor is that the policy response to shocks tends to magnify, rather than dampen, their cyclical effects. In the 1994 crisis, for example, fiscal policy responded by moving rapidly to a primary surplus. But this was achieved through drastic cuts in public investment and non-wage current expenditures, magnifying the output effects of the currency crisis. Finally, by opening its capital account completely by 1989, Turkey exposed itself to additional external volatility. As with many countries that liberalized their capital account around the same time, Turkey experienced rapid credit expansion along with significant capital inflows. This left it vulnerable to capital flow reversals and bank failures, both of which played significant roles in the 1994 crisis. Indeed, contingent liabilities to governments that arise from banking sector crises can cause currency crises and inflation in otherwise healthy economies. 14 Of course, Turkey also suffered a significant real shock in the early 1 990s as a result of the Persian Gulf war and a smaller real shock in 1998 when the Russia crisis led to a contraction in exports to the region. 33. To support the hypothesis that uncertainty has played an important role in volatility, consider the paths of base money growth, inflation, and the money market interest rate (figure 1.4). In the early 1990s, base money growth, initially slower than the Figure 1.4: Base Money Growth and Inflation, 1990-1999 inflation rate, was accelerating rapidly. In 1994, it jumped to 120 160 percent and inflation rose, in Money Market anticipation, to an average annual 140 Rate P,° rate of more than 100 percent.4 120 BaseMoney Nominal interest rates also rose in CPowth 1994, with money market rates i00 above 130 percent. It is difficult to g 80o assess whether firms and households anticipated that inflation would 4 Ia R continue to rise, or whether they 40 Inflation Rate believed that it would fall from that 20 point forward. What would have i been clear to households and firms is that, at the inflation rates of the time, 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 real interest costs would have been - _-- substantial and that borrowing had a Source: Central Bank and World Bank. lot of uncertainty associated with it. A similar situation arose in 1999 when inflation fell but nominal interest rates and base money growth remained high (figure 1.4). b 34. As predicted by the uncertainty hypothesis, most categories of private investment declined in 1994 and 1999 (figure 1.5). In general, machinery growth, purchases of consumer durables, and building construction are strongly negatively correlated with the differential 1 See, for example, Burnside, Eichenbaum and Rebelo (1999, 2000) and Corsetti, Pesenti and Roubini (1998). s The money growth rate is calculated from end-of-year to end-of-year, whereas the inflation rate is the percentage change in the average annual GDP deflator. Hence the timing of the inflation series lags that of the money growth series. 16 An interesting and related hypothesis is that uncertainty about inflation, and hence high real interest rates, are also a function of public sector price setting behavior. See Berument (2000). 11 between the interest rate and the inflation rate (the ex-post real interest rate), with correlation coefficients of -0.69, -0.58, and -0.79 respectively.,' The correlation for housing construction is weaker at -0.17. Figure 1.5 suggests that this is a function of a housing boom in the early l 990s that occurred despite high interest rates. If one were to consider the basic shape of the boom to be a trend, it would appear that deviations of housing investment from that trend were negatively correlated with the interest-inflation differential. 35. To support the view that fiscal policy has Figure 1.5: Private Investment Growth by Type been instrumental in creating and exacerbating 1990-1999 cycles, consider the fact that during 1990-93 the - s_ == primary deficit averaged about 5 percent of GNP. s Except 1991, this was a period of rapid growth at an average rate of more than 6 percent meaning 40 that fiscal stimulus was applied during an D0 ierent7'al, expansion. This generated high and persistent inflation, increasing reliance on seignorage, and -. D-.D the buildup of a large stock of debt, the effects of -20 - which are still being felt today. In 1994 the - Machinery buildup of debt led to a crisis in which the money 1990 1991 1992 1993 1994 1995 1996 997 1998 1999 supply had to be increased sharply and rapid fiscal = - 7 -r = restraint had to be applied to the point that a 35 - - - - 30 J primary surplus of more than 1 percent of GNP 25 ni was achieved. The sharp increase in inflation, 20 Differential along with the fiscal restraint, caused a decline in 15 Building GNP of 6.1 percent. Fiscal restraint continued to Buildings be applied into 1995 but, with strong economic o - -- ---- - recovery until 1997, most components of -5 3 government spending rebounded strongly during -15 1995-97. The public sector wage bill remained -20 - 1 -9 -- 1995 almost unchanged in real terms in 1996 and 1997, but all other components of spending rose faster Source: World Bank. than GNP. In particular, other forms of current public sector consumption spending rose 31 percent in real terms in 1996 and 1997. Public sector investment, sharply curtailed in 1994 and 1995, rose almost 50 percent in real terms. So, apart from keeping wage payments under control, the pattern of procyclical spending continued. In 1998 the onset of the Russian crisis forced the government to modify fiscal policy in the face of an external shock. The vulnerability of the government's spending program to an entirely external shock clearly reinforces the urgency of implementing long-lasting reforms. 4 PUBLIC SECTOR DEBT AND PUBLIC SECTOR SOLVENCY 36. This section looks in detail at fiscal management in Turkey over the most recent 6-year period. The first subsection demonstrates that the public sector was on average in a position of approximate primary balance, but ran significant operational deficits in 1994-99. Until 1999 Related to the fact that investment is negatively correlated with the real interest rate is the fact that private savings rates are typically positively correlated with the real interest rate as established in a recent World Bank study: Loayza, Schmidt-Hebbel and Serven (2000). 12 operational deficits resulted in modest debt accumulation because seignorage revenues were significant and the economy was growing. In 1999 the debt stock rose sharply because the economy fell into a recession, and fiscal policy became expansionary while monetary policy tightened. Turkey was on an unsustainable path in the 1990s because it relied heavily on seignorage revenue to control its stock of debt. Even if the level of public sector debt stabilizes, significant sources of contingent liabilities to the government raise the urgency of sustained adjustment. 4.1 Measuring Public Sector Deficits and Debt 37. Many issues in the measurement of the public sector's debt and deficit come up in the analysis of Turkish data. First, there is the definition of the public sector. Beyond central and local governments, there are the fiscal accounts of state-owned enterprises and extra-budgetary funds. There is a significant state-owned banking system to account for, as well as the central bank. Different definitions of the public sector require different degrees of account consolidation. Second, there is the effect of inflation and currency depreciation on the valuation of stocks of debt measured in terms of nominal currency units. 38. This section works with figures from government agencies along with IMF estimates of certain quasi-fiscal activities. Figures forthen oeaiubiscal sctorities. banes Table 1.3: Consolidated Public Sector Balance for the overall public sector balance 04ofGP14195961719 19 and its components as a percentage of (% of GNP) 1994 1995 199 1997 1s99 1999 GNP show a fairly clear picture of Total Expenditures 28.7 25.4 30.7 31.8 33.1 40.5 GNP show a fairly clear picture of Total Revenues 21.0 20.7 21.8 23.3 24.2 26.2 rising budget deficits after the crisis PSBR 7.7 4.8 9.0 8.5 8.9 14.3 and failed stabilization of 1994 (table Quasi-fiscal Deficit 1.3 0.4 4.2 4.5 6.8 10.4 1.3). Most important are the Adjusted PSBR 9.0 5.2 13.1 13.1 15.8 24.7 consolidated budget and unpaid duty Net Interest Payments 10.1 9.1 11.9 10.9 16.2 21.9 losses, which account for most of the Primary Balance 1.0 3.9 (1.3) (2.2) 0.5 (2.9) overall deficit. Real Interest Payments 4.8 4.9 6.0 10.2 10.9 10.3 Operational Balance (4.2) (0.3) (7.1) (2.9) (4.8) (14.5) 39. The inclusion of unpaid duty Source: Treasury, IMF and World Bank. losses reflects the notion that these are an important contingent liability likely to be a realized in the near future. Since 1994 the state banks, Ziraat and Halk, have made big losses on their quasi-fiscal activities, especially in subsidized lending. To maintain the balance sheets of these two institutions these losses have been covered by the issuance of non-transferable claims on the government called unpaid duty losses. While these claims are obviously held entirely within the public sector, it can be argued that the net stock of these claims should be included in the public sector's debt for two reasons. First, if the government had provided direct subsidies to farmers and small business, as Ziraat and Halk do, these subsidies would have shown up directly in the government's budget figures. Second, in order to privatize Ziraat and Halk under its reform program, the government will need to cover the accumulated stock of duty losses. 40. The conventional budget balance is a measure of limited usefulness because changes in it can be dominated by the magnitude of interest payments in a high-inflation economy. More useful measures are the primary and operational balance. To measure the primary balance, net interest payments are subtracted from expenditures before the balance is computed. One subtlety 13 that arises in doing this for Turkey is that interest on deposits received by the government is recorded in the official accounts as general revenue rather than as negative interest expenditure. Figures on interest payments in table 1.3 reflect an adjustment to take this into account." 41. The primary deficit is useful in measuring the current fiscal stance of the public sector, which has changed sharply from year to year since 1994 (table 1.3). In 1995 and 1998, fiscal policy was driven by austerity and the public sector had an overall primary surplus. But restraint has never been sustained for more than a year at a time as witnessed by the primary deficits of 1996, 1997, and 1999. On the other hand the primary deficit has averaged close to zero over the 6-year period. 42. The operational deficit-the primary deficit plus real interest payments-is useful in tracking the evolution of debt stocks over time. Real interest payments are simply nominal interest payments corrected to take into account capital gains that can accrue on nominal debt in inflationary environments. The operational balance, in a significant deficit position in every year since 1995, was on a rapidly deteriorating path in 1999 (table 1.3). Relatively loose fiscal policy since the Russian crisis, along with tight monetary policy to keep inflation in check, led to very high ex-post real interest rates and significant real interest payments for the government. This, in turn, led to the rapid accumulation of debt in 1999. The government has improved the consistency of its fiscal and monetary policies in recent months, and real interest rates appear to be falling rapidly. Table 1.4: Net Debt of the Consolidated Public Sector 43. The level of debt grew rapidly in (percent of GNP) 1993 1994 1995 1996 1997 1998 1999 1994 and remained roughly stable through most of the subsequent years Domestic debt 9.4 14.1 12.2 20.5 20.4 24.2 38.6 despite significant operational deficits in Extemal debt 25.7 30.6 29.1 26.0 22.5 20.3 19.4 1996, 1997 and 1998 (table 1.4). But in Total debt 35.1 44.7 41.3 46.5 42.9 44.5 58.0 1999 the level of debt shot up once more to a projected 58 percent of GNP. The change in the ratio of debt to GNP can be decomposed into three components as: (4) change in debt/GNP = operational deficit - seignorage - growth effect. The growth effect takes into account the fact that if the operational deficit and seignorage are both zero, so that real debt is not increasing, the ratio of debt to GNP will fall as long as there is real growth in the economy. There are two reasons debt has not accumulated faster in Turkey (table 1.5). First, the government has managed consistently to raise seignorage revenue of about 2.5 to 3 percent of GNP. Second, real growth has been fairly rapid, so the growth effect has typically reduced the stock of debt by 2 or more percentage points of GNP. In periods of 18 The adjustment has no effect on the measures of the overall budget balance because interest receipts are deducted both from current revenue and from current expenditure inclusive of interest. The adjustment only affects the measures of the primary balance. 19 Seignorage is the change in the monetary base or reserve money expressed as a percentage of GNP. The growth effect is computed as bgl(l+g) where b is the debt to GDP ratio at the beginning of the period, and g is the real growth rate within the period. 14 recession such as 1994 and 1999, debt stocks have spiked upward as the growth effects have worked in the opposite direction.2" 4.2 Public Sector Solvency Table 1.5: Debt Accumulation, 1994-1999 (percent of GINP) 194 1995 1996 1997 1998 1999 44. The solvency of the public est( sector depends on its ability to Change in Debt/GNP ratio 9.6 -3.4 5.2 -4.4 1.6 13.5 indefinitely continue to service its Operational Deficit 4.2 0.3 7.1 2.9 4.6 13.6 obligations. There are a variety of S-ignorage 3.3 3.0 2.4 2.9 21. 2.9 interpretations of the public Growth Effect sector's ability to service its debt in Source: Treasury, IMF and World Bank. the long term. The concept of solvency used here is one that requires the level of debt to remain stable relative to the level of nominal GNP.21 This concept allows the level of debt to grow over time, but not at a rate faster than the nominal growth rate of GNP, at least in the long run. 45. Given the solvency criterion that public debt should not grow relative to GNP, it would seem that Turkey's public sector in the early 1990s was on an unsustainable path. After 1994, stability was partially restored, but after four years of relative stability, debt began to grow in 1999. From 1990 through 1994, the debt grew from 29 percent of GNP to about 45 percent of GNP. It was still about 45 percent of GNP in 1998, but rose sharply to 58 percent of GNP in 1999. In 1990, 80 percent of the public sector's debt was held externally. By 1994, this was down to 69 percent, although the amount of external debt had risen from 23 percent of GNP to 31 percent of GNP. From the end of 1994, Turkey gradually paid down its external debt until it was below 20 percent of GNP at the end of 1999. Meanwhile, domestic debt has accumulated rapidly. Although domestic debt represented just 14 percent of GNP in 1994, it represented about 39 percent of GNP in 1999. The increase in domestic debt is mostly accounted for by the more than 100 percent increase in central government cash debt and the rapid accumulation of unpaid duty losses in the Ziraat and Halk banks. High real interest costs on domestic cash debt have exacerbated the fiscal sustainability problem. 46. There are other causes for concern: * First, while the public sector's external debt has declined since 1994, the banking sector and the private non-bank sector have accumulated foreign debt at an unprecedented pace. These sectors had a combined external debt of just 7.7 percent of GNP in 1990, but this rose to almost 30 percent of GNP at the end of 1998. Certain private sector obligations could end up as contingent liabilities of the public sector and deserve special attention given the role of public sector bailouts, especially of banks and other financial institutions, in the recent Asian financial crisis. * Second, the government faces important contingent liabilities such as those in the energy sector.22 Privatization in the energy sector could leave the government with an important 11 The change in debt/GNP in table 1.5 is not the sum of the three components shown in table 1.5 because the operational deficit does not take into account revaluation effects on external debt due to real exchange rate movements. 21 Other concepts of solvency are discussed in the appendix. 2' These are discussed further below and in chapter 3 of this report. 15 fiscal burden in the medium run, eroding potential gains from stabilization in the next few years. * Third, there are longer-term issues connected with the social security system. Prior to the August 1999 reform, annual deficits of the social security system would have reached about 4 percent of GNP within a decade as demographics began to work against the system's solvency. With the reform, the system should reduce its deficit to about 1 percent of GNP over the medium term. However, current World Bank projections suggest that the deficit will gradually increase over the long term and could reach 5 percent of GNP by 2050. 5 ACHIEVING FISCAL SUSTAINABILITY 47. The evidence presented above suggests that fiscal policy has been unsustainable since 1990. While there have been periods in which debt has been stable as a share of GNP, each period of recession and crisis has coincided with an upward ratcheting of the debt stock. Furthermore, any stability of the public debt that has been achieved has been gained through high inflation.23 To address whether Turkey's current policies are sustainable, this section looks at whether the public sector can stabilize its debt under its growth and inflation targets. One way to answer this question is to: (i) determine the level of seignorage consistent with Turkey's growth and inflation targets, (ii) assume a desired long-run level of debt relative to GNP, and then (iii) determine the size of primary surplus consistent with these goals. The conclusion from this simple analysis is that Turkey's current reform effort is consistent with long-run and short-run fiscal sustainability. 5.1 Long-run Sustainability 48. The starting point of the analysis is to determine the long-run level of seignorage the government could hope to raise given different inflation targets. In Appendix I a simple model of the long-run demand for base money is described which assumes that the monetary base as a fraction of GNP is a downward sloping function of the nominal interest rate, denoted m = L(r + ,r), where m is base money relative to GNP, r is the real interest rate and ;r is the inflation rate. The level of seignorage that can be raised by the government relative to GNP is then given by (Tr + g)m / (I + ;T + g), where g is the growth rate of real GNP. Appendix 1 shows that the steady state budget constraint of the government is given by: (5) s (r - g) b _ E+ g m l+±r+g l+fr+g where s is the long-run primary surplus as a fraction of GNP and b is the some constant level of debt relative to GNP. The second term on the right hand side of equation (1) is seignorage. If the real interest rate and the growth rate of real GNP are equal, the public sector can run a primary deficit equal in magnitude to its seignorage revenue. 2': A recent IMF simulation suggests that if the public sector had foresworn seignorage revenue in the 1990s, its debt level would now have reached almost 100 percent of GNP. 16 49. To get a benchmark estimate of the limits on seignorage revenue, a long-run money demand function was estimated for Turkey. The estimate is given by L(r + ,) = 0.107exp[-O.861(r + 7r)] . This implies that a decline in the inflation rate of about 1 percentage point causes a roughly 0.9 percent increase in real balances held by the public. Using this estimate of the long-run money demand function, the level of seignorage relative to GNP is given by (sr + g)L(r + or) / (1 + if + g), which is a function of three parameters, the inflation rate, the real interest rate and the real growth rate. 50. Table 1.6 presents a few scenarios for long-run steady states calibrated for Turkey. The table assumes different values of the real Table 1.6: Alternative Fiscal Adjustment Scenarios to growth rate of the economy and the real Stabilize Debt / GNP Ratio interest rate, and assumes that the inflation (Required Primary Surplus) rate achieves a long-run target of either 10 R percent or 5 percent per year. The level of GNP 1 Iai a 5% debt is assumed to stay constant at its Growth Rel Interest Rhe Real Interest Rate current level of 58 percent of GNP. With 5 5% I 700 5% 11/0 percent real growth, Turkey could achieve a 3% -0.1 2.6 0.3 3.1 stable level of debt even with a primary 5% -1.2 1.3 -0.9 1.8 deficit, but only if the real interest rate was Source: World Bank. low. Since 1994 the effective real interest rate on public sector debt has averaged over 10 percent, and in 1999 was about 25 percent. 5.2 Short-run Dynamics 51. The calculations in the previous section are valid for long-run steady states, but Turkey is not in a long-run steady state. It is undertaking a major stabilization effort in order to achieve its long -run goals of single digit inflation and sustained high growth. Appendix 1 shows that the government's debt relative to GNP evolves in the short run approximately according to: (6) Ab, = rb,-l - s, - a, - g,b,-l where r, is the ex-post real interest rate on the public sector's outstanding debt, b, is the level of debt as a percentage of GNP at the end of period t, s, is the primary surplus as a percentage of GNP, a, is seignorage revenue as a Table 1.7: Short-Run Dynamics of the Required percentage of GNP, and g, is the growth rate Primary Surplus (percent of GNP) of real GNP. 2000 2001 2002 Long-Run Real GNP Growth 5.5% 5% 5% 5% 52. Suppose the government wishes to Inflation 42% 15% 7.5% 5% keep its level of debt stable while Real Interest Rate 20% 15% 10% 7.5% implementing its reform plan. At the Primary Surplus 3.7 3.5 1.6 0.4 beginning of the transition to a long-run Source: World Bank. steady state, more revenue will be generated from seignorage than in the long-run, because inflation will likely take time to fall. So seignorage will help keep the level of debt in check in the short run. But real interest rates may remain high in the short-run, which would tend to raise the level of debt. A simple scenario for 17 the short-run computes the primary surplus required to keep the level of debt at 58 percent of GNP in each of the next three years (table 1.7). It assumes that the growth rate of real GNP will be about 5 percent in each year; that the GNP deflator's inflation rate will drop from 42 percent in 2000 to 15 percent in 2001 to 7.5 percent in 2002 and a long run level of 5 percent thereafter; and that the real interest rate will drop from 20 percent in 2000 to 15 percent in 2001 to 10 percent in 2002 and a long run level of 7.5 percent thereafter. For simplicity, the long-run is assumed to be achieved by 2003 at which point the real growth rate, the interest rate, the inflation rate and the debt to GNP ratio are all assumed to remain constant forever. 53. Under this scenario very sharp increases in the primary surplus are needed in the short run. The main reason for very large surpluses in the short run, but not in the long run, is that in the simulated scenario the program is assumed to lack credibility in the short run. Real interest rates are assumed to remain quite high. This is consistent with events in 1999. Why were real interest rates so high in 1999? In an inflationary environment, where the future behavior of fiscal policy and inflation rates is uncertain, investors will demand a premium to hold government debt, especially domestic debt. As domestic debt's share of the public debt has increased this risk premium has become more and more important. Turkey has tried stabilization before, and it has failed-so it is not surprising that the announced reforms did not have immediate credibility. It is also reasonable to assume that real interest rates may fluctuate in the near future, as the government attempts to maintain the credibility that the reforms have gained. What is clear is that the sooner lasting credibility is obtained, the less pressure there will be on the budget. To break the inflationary cycle, the government needs to keep sending clear signals that it will carry out its reform program and permanently slash its reliance on seignorage revenue. 6 CREDIBILITY, SUSTAINABILITY AND THE STABILIZATION EFFORT 6.1 Credibility 54. Ultimate success of the current reform effort depends on its generating sufficient credibility. That is, agents operating in the Turkish economy must somehow be convinced that there is a high likelihood that significant reforms will actually be carried out. If not, the behavior of these agents will undermine the reform process by raising its short-term cost to the point that the government might choose to abandon it. One needs to look only at Turkey's prior experience with inflation stabilization-or at international evidence-to be convinced that credibility is important. Turkey never succeeded in controlling inflation because it never made a sustained effort towards fiscal adjustment. Fiscal contractions, often associated with periods of crisis, were not sustained once better times returned. At the same time, previous stabilization efforts were thwarted by the failure of real interest rates to come down, a sure sign that credibility was lacking. 55. The experience of Brazil is instructive and should be of particular interest to policy makers in Turkey. In 1994, after many attempts at controlling inflation, Brazil introduced the Real Plan, a comprehensive program for controlling inflation that was based on a verifiable nominal anchor (a tightly managed crawling-peg exchange rate regime) and a large fiscal adjustment to be achieved through significant prospective reforms of the public sector. The plan was extremely successful at first and inflation fell rapidly. Arguably this happened because the 18 government appeared to be very serious about its plans for reform, unlike in previous Brazilian attempts at disinflation. But progress on actual reform implementation was slow and the required fiscal adjustment was not achieved. In the meantime, Brazil was exposed to external shocks, namely the East Asian and Russian crises. Facing reduced credibility and the possibility that a recession induced by tighter monetary policy would be the cost of maintaining the exchange rate target, Brazil abandoned the plan in early 1999 (box 1.2). 56. The lesson for Turkey could not be clearer: short-run success is no guarantee of long-run results. Building on this lesson, the danger for Turkey with its current reform program is that the up-front fiscal adjustment is based in large part on temporary measures, not the long-term structural reforms intended by the government. Furthermore, most of the fiscal adjustment is taking place on the revenue side of the budget, not the expenditure side. Alesina and Perotti's (1995) research has shown that fiscal adjustment is most likely to be successful when it comes on the expenditure side. In the absence of up-front savings from long-term reform, the government must at least continue to carry out the longer-run reforms to signal that long-run expenditure savings will be forthcoming. As Easterly (1999) argues, the quality of fiscal adjustment is crucial to its success. This means that fiscal savings must be generated through sustainable changes in fiscal policy, as opposed to drastic and unsustainable cuts in public investment and non-wage current expenditures such as those used to restore public finances in 1994-95. The ongoing Public Expenditure and Institutional Review will be an important input into these long- run reforms. 57. Another lesson from the Brazilian experience regards the speed with which inflation declined under the Real Plan. It is often argued that expectations of inflation are very sluggish and do not respond immediately with the implementation of an anti-inflation policy. The Brazilian economy had been plagued by extremely high inflation and had a history of failed stabilization, yet under the Real Plan inflation fell quite quickly. Presumably this had something to do with the plan's initial credibility. On the other hand, there is a widely held view that it is more difficult to lower inflationary expectations in economies like Turkey that have experienced rapid but not hyper-inflation. The possibility that inflation and inflationary expectations in Turkey might respond slowly to fiscal adjustment and a well defined nominal anchor is a significant risk factor that deserves careful study. Deeper knowledge of the micro-structure of product and labor markets in Turkey is also needed in order to better understand price-setting behavior in the economy. 58. To ensure credibility, specific goals and timetables for reform should be clearly stated and strictly adhered to. In this regard, the hiatus in reforms that followed the Russian crisis and led up to the 1999 elections was problematic. The earthquake in the Marmara region in August 1999 also raised doubts that the government would continue its reform effort. Fortunately, the government renewed its commitment to reform. To signal continued commitment, the government must avoid slippage on structural reform. Some slippage has occurred in the recent past. For example, after tax reform legislation was enacted in 1998, widening the tax base and reducing the lag between tax accrual and payment, proposals for simplification of VAT through a tax on luxury goods were not adopted. After the Marmara earthquake, the government temporarily reversed some of the tax reforms effective in 2000, although the government has reinstated these reforms and implemented further measures. Looking forward, many aspects of the reform process, though planned, have yet to be carried out, e.g., regulatory reform in the 19 infrastructure sectors, administrative reform of the social security system, accelerated privatization and banking sector reform. These reforms are discussed further below. Box 1.2: Credibility an St zabilitio iifr Lemns frm ElseWhere T0see whythe Credibility of fiscal reform is so important o netoly considerteational experience with stabilization. Credibilityisestablishedwhenthegoer nconrtcs prvate ants t the ikeAihd th itwill deviate from its announced plans is extrmely low. vWhen a govern t devate in an important exhag rat basetddisifaion, t:his can co;mein two forms: (i) a patia dfuln? usom7e ofitsobigartions00 00 engineeired by an immediate deviation is tr d path fr thxa d ti or float) or (i a0 revisio off 6 teure targeted path for tehg ethatiatio fro thepolicyfraewok ofiscaldb ad would beanychange t d t a d theresn v f isve inthprsntp valueoseigraethrugh higeinlato(if osible). his fllows frm the imutbl logi of the governme tlifetimebudgeta constrainte whc sae tiha Value of government debt present value af futur rmaysrpue Take, as an example. Brazil. Since 1986 Brazi hamplemnted si tabilization Plan otyt onrlifain The first, four Pans-CruzaO (986), Bresser (1987) Summer(99,adCllr11 >ue smlrmtos Thtey tried to fix or conitro the exchang rate and. chane nlatoayepcain hog aeadpiefezs In om caesfiscal~ reform was discussed but i was rarely ipentdThCoorIPlan ven went toth extrardiary lengt of frezing bank accdounts. Why i h eom al h ulcwsnvrcnicdta fiscal refomwould be undertaken-tis wa why c otoshdt eue.Ee hnifainsbie h public remained unconvinced. As reut iquk iditydreupadottsatdflin.Whuthealtyo convince theipublic of its sincerity, and, unaleto, tolerte recession, the oeneti ec aeaadoe t ln leadingto upward spikes in prics The resultin decli'e ithevaueofgvrendbtmeuporls seigoae Thefithstailzaio efor, heCo61o II Plan of February 199 1, did not relyo rc-iig nta it usedtight montay oliy omine wththe prOMise of future f 'isclefrm The pan was.succssfulinbining *inflatio dow fo aoutthre onts,butprices beanto rieainn mid19.Tegvrmn rpsdfsa esrs Brazil's sixth stblzto efr-h ealPlpan imnplemented in July 1994-iitially wa extraordinarily succesfl and credible. The goernmentdinot free waes or pri Turkey, it announced significant fiscal mea;sures some~ of whichi it carried ouxt. Its efforts were so1 crdble ta tatal maintaine a raid MOney growh rto efst 6 months of the pa without generating inflati nttight revefinue Unfo r unate h gotve come i set'pertio deficits wereinhe of 3.8-75 pe entofD in the period 1995- Byt end of 1998, the publc sector's nt debt had risen to nearly 50 percentf Of GD? fro a60 lof 27 percent ini 1995. Then Brazil Was hit by a series of external shocks beginning wit Thailand's float and ending With Russia's devahluation in 1998. Faced Wihmonigdb and increase uncertain as to its intentions on refrm, ith govenmnt had achoice: tighten monetary policy and risk aecessionrabadonthe poan. Te grnmnhsehlat te course.T e urrency w floated, depreciating by6 ecn niay199 n nlto eund Brazil's experienc is not unique. Ih Lin Ara Aretina's experience piior to the intodion of a currency I board, and Peru's experience under various admiistrations duing this perdiodreJ classic examsples offail ed staiiation caused by a lack of sutined fisc adjustment. Turkeys disinflatio attempts ofihe iarly 190s afret 20 59. International experience and past history in Turkey argue in favor of a rapid move toward fiscal balance and stabilization of the debt stock, together with deep structural reforms in order to establish strong credibility early on in the reform process. At the same time, convincing the public that reforms will be sustained is as important as rapidly carrying out the first stage of the program today. This requires a transparent and clear political decision to carry out reform, careful design of each reform measure, establishment of a timetable, and steady implementation of reforms over the medium run in accordance with this timetable. 60. Gaining credibility is the key to moderating real interest rates and reducing the real interest burden on the budget. At a consolidated level of debt of 58 percent of GNP, each percentage point of real interest saved represents an extra contribution to debt reduction of more than 0.5 percent of GNP. In 1999, the market-clearing price of government debt revealed that investors were still very uncertain about the path of future policy. In late 1999 and early 2000, investors apparently gained some confidence. Nominal interest rates fell. For example, the 3- month time deposit rate fell from 76 percent in October to about 40 percent in the first 3 months of 2000 despite the fact that annualized monthly CPI inflation only fell to 40 percent by March 2000 (table 1.8). So real interest rates have fallen. If the govermnent makes credible plans for the future and starts carrying them out, the risk premium on government debt will stay down. If it backtracks on planned reforms, real interest rates will surely rise once more and raise the overall fiscal cost of the reform effort. Table 1.8: Nominal Interest Rates and Inflation Since Mid-1999 (percent) 1999 2000 Jul. Aug. Sep. Oct. Nov. Dec. Jan. Feb. Mar. Interest rate 81 81 76 76 72 60 38 38 40 CPI inflation 57 64 101 109 64 100 77 55 41 Note: The interest rate is the average 3-month time deposit rate in each month. CPI inflation is measured as the ratio of the CPI in the current month CPI to the CPI 3 months previously. It is converted to an annual rate. Source: Central Bank, SIS and World Bank. 61. A starting point for credibility is a longer-term approach to expenditure and revenue management. Without long-term planning a broad consensus in favor of economic reform will be difficult to come by. Without consensus, the political costs of reform may make it untenable. Budget transparency is also a problem. Too many transactions, including a substantial stock of contingent liabilities arising from Treasury guarantees, are treated as "off-budget" or below the line in the public sector accounts. State-owned financial institutions hold a substantial quantity of non-cash domestic debt in return for quasi-fiscal losses generated by subsidized lending that does not bear a market determined interest rate and is not marketable. None of these transactions shows up in the standard fiscal accounts. A more open and accurate accounting of public sector finances is critical to the public easily being able to monitor the performance of the stabilization program. Claims of fiscal success need to be backed by credible figures that support the claims. While the government has tied its anti-inflation policy to an easily monitored nominal anchor (a targeted path for the exchange rate) international experience has repeatedly shown that this does not guarantee the success of fiscal stabilization. One of the keys to success is the credibility of the government's fiscal plan. 21 6.2 How Will the Reforms Affect Sustainability? 62. The reform program envisioned by the government can make the public debt stock more sustainable by leading to lower real interest rates in the long run and by contributing to a larger primary surplus in the short run. Maintaining the social security reforn is most important to the improvement of the primary surplus given that the pre-reform system was running a deficit of 3 percent of GNP, while the new system reduces that to about 1 percent of GNP. Banking sector reform is also critical to sustainability. The government has already taken important steps in this regard by upgrading prudential regulations and by gradually reducing the size of open foreign exchange positions that banks are allowed to hold from 50 to 20 percent of net worth. This will reduce the risk of the government facing a banking crisis caused by an external shock in other emerging markets. The banking system's liquidity is an ongoing concern for the government while it remains heavily indebted, because the banks hold 70 percent of the government's outstanding domestic debt. Reforms in the agriculture and energy sectors will also contribute to fiscal sustainability by reducing budgetary transfers and lowering contingent liabilities. 63. As the next section points out, the initial market reaction to the reform program has been positive. Real interest rates and inflation have fallen. Continued action on reform will be needed. The reform program is ambitious as the details in the next section make clear. It has the potential to correct Turkey's chronic inflation and slow growth. With Turkey's record of previous failed reforms, actions will speak louder than words-prompt implementation of the program is crucial. 7 THE NEW ECONOMIC PROGRAM 64. The key elements of the government's economic program, which is being supported by the IMF and World Bank, are: * A significant up-front fiscal adjustment and tighter fiscal policy, designed to put public finances on a sustainable path. * A nominal exchange rate anchor starting with a 12-month pre-announced crawling peg designed to lower expected inflation and interest rates. * A "quasi" currency board arrangement where all base money is created through capital inflows without any room for sterilization. * A supportive, forward-looking incomes policy in the public sector. * Structural reforms in the core sectors-social security, agriculture, infrastructure, and financial sector. * Acceleration of the privatization program. 65. The IMF approved a multi-year Standby arrangement in December 1999. The Bank approved an Economic Reform Loan in May 2000 and a Financial Sector Adjustment Loan is in the advanced stages of preparation. 66. The stabilization program targets a decline in inflation to single digits by 2002, compared with CPI inflation of 69 percent in 1999. In parallel, real domestic interest rates are projected to 22 fall from an estimated average of 40 percent (ex-post) in 1999 to about 12 percent by 2002. In tandem with falling inflation and interest rates, the government's program aims to pull the economy out of recession. The program targets growth in the 5-6 percent range in 2000-03 as compared to a drop in output of 6.4 percent in 1999. 67. To achieve these macroeconomic targets, the program centers on restoring public finances to a sustainable path, which is critical to easing inflationary expectations and lowering real interest rates. The stock of total public sector debt, including losses of the state banks on subsidized credits (duty losses) increased from 44 percent of GNP in 1998 to 58 percent in 1999, an unsustainable increase. A key factor was the very high real domestic interest rate in 1999. The program aims to stabilize the public debt as a share of GNP in 2000 at or near the level of end- 1999 and then to begin reducing the public debt stock over the medium term. Special emphasis is put on reducing the so-called cash debt component, which is comprised of government securities that are particularly sensitive to domestic interest rates. The outcome will depend not only on short-term deficit reduction but also on a rapid and sustained fall in real interest rates. This will require sufficient structural and institutional changes in the public sector to convince the markets that public finances are now on a sustainable path. 68. The stabilization program combines strong and sustained fiscal adjustment with a nominal exchange rate anchor and tighter, forward-looking incomes policy. Fiscal policy in 2000 targets an underlying primary surplus for the consolidated public sector of 3.7 percent of GNP (table 1.9) excluding earthquake expenditures of ] .5 percent of GNP. Given the trend at unchanged policies, the 3.7 percent surplus implies a fiscal adjustment of over 7 percent of GNP. Fiscal policy in 2001-02 would be geared to maintaining the primary surplus at the level of 2000 in order to allow for a modest reduction in the public debt to GNP ratio. Table 1.9: Fiscal Adjustment Targets for 2000 (% of GNP) Consolidated Rest of Consotid*ted Budget Public Sector Public Sector 1. Primary Balance before Measures -2.0 -3.3 -5.3 2. Measures (3+4) 6.6 0.9 7.5 3. Revenues 5.0 Fiscal package adopted by Parliament 3.5 Supplementary revenue package 1.1 Quarterly advance income tax payment 0.4 4. Expenditures 1.6 Social security reform 0.5 Agriculture reform 0.2 0.9 1.1 Current Expenditure cuts 0.9 5. Primary Balance with Measures (1+2) 4.6 -2.4 2.2 6. Earthquake expenditures 1.5 1.5 7. Primary balance excluding earthquake exp. (5+6) 6.1 -2.4 3.7 Source: IMF and World Bank estimates. 69. The government has taken essentially all of the fiscal measures underpinning the 2000 fiscal targets. The Parliament has approved a major revenue package expected to generate at least 3.5 percent of GNP, including 1.2 percent in windfall profit taxation on interest from government bonds issued before December 1, 1999. A further I percent in windfall taxes is 23 expected from a provision in the 2000 budget to revise the method for computing the taxable profits of corporate holders of government securities. The government has approved by decree a supplementary package of measures expected to generate an added 1.1 percent of GNP in revenues in 2000. This second package includes an increase in the standard VAT rate from 15 to 17 percent and in the rate for consumer durables from 23 to 25 percent, as well as increases in the VAT on tobacco and alcoholic beverages. 70. On the expenditure side, the government is committed to generating about 2.5 percent of GNP in fiscal savings through a series of measures including social security and agricultural reforms, forward indexation of civil service salaries and limits on replacement hires in the public sector. The 2000 budget sets a limit of 80 percent on replacement hiring in the civil service, and the government has issued a decree imposing a 15 percent limit for state economic enterprises in the Treasury's portfolio. 71. The second element of the program is a substantial shift in financing the public sector borrowing requirement toward external financing and privatization. This will ease pressure on domestic debt markets and interest rates. Acceleration of the privatization program will play a key role in meeting the PSBR and reducing the stock of public debt with a target of generating cash revenues of some US$7.6 billion in 2000 and US$18 billion over 2000-02. Equally important will be increased external financing to accommodate a shift to a more balanced mix of internal and external borrowing compared with the nearly exclusive reliance on domestic markets in recent years. The program targets US$5.3 billion in net external borrowing by the consolidated public sector in 2000 and some US$14.5 billion over, the 2000-02 period. The medium-term external financing plan includes US$4 billion under the IMF Standby and US$3 billion in quick-disbursing support from the World Bank. 72. The third key element of the program is the exchange rate regime. In early December 1999, the Central Bank announced a crawling peg exchange rate regime designed to provide a nominal anchor for the disinflation program. The rate of crawl for 2000 has been announced as 20 percent for the year, declining from 2.1 percent per month during the first quarter to 1.0 percent per month over the last three months. At the end of each quarter, the Central Bank will announce the rate of crawl for an additional three months without changing the schedule already announced. Starting in mid-2001, a symmetrical band will be introduced around the central parity rate, with the width of this band to be increased at a rate of 15 percentage points per year. 73. During the first 18 months of the program, monetary policy will approximate the rules of a currency board with no permanent increase in net domestic assets of the Central Bank beyond the level of end-1999. This policy implies that external capital flows will not be sterilized, thus maximizing their impact on domestic interest rates and minimizing upward pressure on the value of the lira. 74. Another key element in the stabilization program is incomes policy. Incomes policy will support the nominal anchor, although the commitments for 2000 reflect a difficult compromise for the government between the requirements of the disinflation program and social pressures. The 2000 budget law incorporates a forward indexation mechanism for 2.6 million civil servants (who represent 11 percent of the labor force). Following an initial 15 percent increase granted in January, the government limited total civil service wage increases for the first half of 2000 to 19 24 percent, equal to cumulative inflation plus a 2 percent bonus in accordance with the budget law. The increase for the second half of 2000 has been set at 10 percent, which will result in a 31 percent increase for the year. The government agreed with stakeholders to increase the minimum wage in line with targeted inflation. At the same time, salaries in the broader public sector (excluding civil servants) will increase sharply given the large nominal increases approved before the April 1999 elections. Several private companies have announced that they will follow the government's lead in setting wages in 2000. While a forward-looking incomes policy is a welcome improvement, the various compromises made so far represent weak points to be closely monitored. 75. The new stabilization program has a better chance of success because it involves long- lasting and sustainable reforms to public expenditure as well as tax reform.14 Of the reforms so far, tying public sector wages to inflation targets, rather than to past inflation is a major first step. The ongoing privatization of state-owned enterprises that represent a drain on public finances is another important step-not so much because privatization receipts reduce the government's current borrowing requirement, but because the ongoing drain on the public coffer is removed.21 Tax reforms have been implemented which reduce the time lag between tax accrual and tax payment. Tax rates have been restructured, and the tax base broadened. To avoid the possibility of contingent liabilities arising from the financial sector, banking supervision has been strengthened. 76. As part of its economic program, the government has launched an ambitious structural reform agenda. The structural reform program aims to underpin macroeconomic stability and create the conditions for sustaining high growth. The structural reform agenda focuses on five critical actions: * Structural fiscal reforms to support fiscal adjustment. * Policy and institutional reforms to ensure an equitable and financially sustainable social security system. * Structural reforms to deregulate the energy and infrastructure sectors to promote private sector participation. * Rationalization of agricultural support policies and privatization of agricultural SOEs to promote agricultural growth and rural income generation. * Financial sector reform to ensure stability and improve the efficiency of financial intermediation. 77. With regard to fiscal adjustment, the government has two structural objectives. The first objective is to sustain the fiscal adjustment in 2001 and beyond by replacing once-and-for-all measures with long-term fiscal measures. The second is to consolidate the shift in the fiscal balance over the medium term through institutional reforms to improve public sector management; i.e., ensure the quality of adjustment and deeper structural reforms in key areas which have been the source of fiscal and quasi-fiscal deficits in the recent past, including the social security system as well as the energy, financial and agricultural sectors. Within this 24 Recent work by Easterly (1999) suggests that fiscal restraint that involves unsustainable cuts to vital government expenditure programs is unlikely to be sustained. Alesina and Perotti (1995) argue that fiscal reforms based on revenue enhancement, rather than expenditure reduction, are less likely to succeed. 2 This is true whenever the opportunity cost of selling a public sector enterprise later rather than sooner is positive. 25 context, the government is improving the transparency of its budget process. Credit subsidies by state banks have been included in the 2000 budget for the first time. The process has begun of closing down the wide variety of budgetary and extra-budgetary funds that are not subject to the scrutiny of the regular budget process. A fiscal transparency review has been completed and a public expenditure and institutional review is underway. 78. Reforming the social security system is essential to restoring its fiscal solvency over the medium term and to ensuring an adequate and sustainable level of benefits. Just one week after the August 1999 earthquake, a major policy reform of the social security system was enacted. Its importance cannot be overstated: the system was on a rapidly deteriorating path towards insolvency with a deficit of about 3 percent of GNP in 1999 projected to reach 5 percent of GNP by 2010 and 10 percent of GNP by about 2030. In its first phase, social security reform has restructured the eligibility and benefits of the public pension system to reduce the deficit to about 1 percent of GNP. A second phase of reform currently underway focuses on strengthening the administrative structure, extending the system's coverage and eliminating contribution arrears. The second phase includes introduction of a framework for supplementary individual pension schemes (chapter 2). 79. The provision of adequate and efficiently operated infrastructure services is a prerequisite to the establishment of sustainable economic growth and acceptable living standards. Public resources are insufficient to fully finance the required investment in infrastructure, particularly in the energy sector. Under the current structure, the government is building up large contingent liabilities, particularly in the energy sector, which threaten future fiscal sustainability. Privatization and private sector participation will be accelerated overall, but especially in the areas of telecommunications and energy. This effort will generate privatization revenue of 3.5 percent of GNP. More importantly, from a long-run perspective, these sectors will be deregulated and opened to competition, fostering growth not only within the sectors but also across all sectors of the economy (chapter 3). 80. Agricultural reforms are important in that the government's current price and credit subsidies are both highly distortionary and very costly to the budget. Under the program, input and credit subsidies are to be phased out. Agricultural sales cooperatives will be given full autonomy from the government and most SOEs in the agriculture sector will be privatized. The current costly and inefficient system of agriculture price supports will be replaced by a direct income support system that will create new income generation opportunities by improving incentives. The direct income support system will also improve the targeting of budget support to smaller farmers (chapter 4). 81. The authorities prepared a three-pronged approach to financial sector reform involving: * Strengthening the legal and regulatory framework (banking infrastructure). * Developing an effective and orderly strategy to deal with problem banks and support for restructuring and consolidation in the private banking sector. * Taking steps to commercialize, restructure and privatize the state banks. 82. These reforms are aimed at restoring soundness to the banking system, thus eliminating an important source of financial instability and contingent liabilities to the government. They are 26 also intended to make transparent the financial relations between the budget and the state banks as part of the preparation for their privatization. Currently, state bank balance sheets are dominated by a non-transparent system of claims on the government for past subsidies paid by the banks in the form of "duty losses" (chapter 5). 27 CHAPTER 2: SOCIAL SECURITY REFORM26 1 INTRODUCTION 83. By the late 1990s, the financial imbalances of the social insurance system in Turkey had become the most urgent fiscal problem. A fiscal crisis was looming. The financial imbalances of the system were impairing growth, crowding out other social and infrastructure investments, and generating a highly inequitable reallocation of resources within and across generations. A major policy reform enacted in August 1999 will shrink but not eliminate the deficit, thereby buying time for the formnulation of further structural reform. The authorities are now moving on a second stage of social security reforms centered on an administrative reform of the public pay-as-you-go (PAYG) system to improve coverage and compliance, and the introduction of a legal framework for voluntary funded private pensions. Over the longer run, Turkey will need to implement deeper structural reforms of the social security system to respond to the requirements of an aging population with increasing life expectancy and contribute to the development of long-term domestic savings and growth. Longer-term policy options for pension reform include further modification of the PAYG system, more innovative PAYG reforms based on notional accounts, or the introduction of a full multi-pillar system. Structural reforms of the health insurance programs and measures to better integrate the unemployment insurance scheme introduced under the August reform with existing severance payment programs round out the longer-term policy agenda for social insurance in Turkey. 84. The provision of income security is one of the hallmarks of developed societies and modern nation states as documented in a well known World Bank report27. The framework for old-age security in almost all high-income countries and in most developing countries, including Turkey, traditionally involves state-managed pension schemes that pay an earnings-related defined benefit financed on a pay-as-you-go basis. But experience shows that PAYG pension schemes that have tried to combine the three functions of social security-saving, redistribution, and insurance- have often produced sizable fiscal deficits, costly factor market distortions, and perverse redistribution to higher income groups, without providing adequate income security for the old. Turkey's pension system has faced the same problems. 85. Turkey's social security system consists of three distinct institutions-Sosyal Sigortalar Kurumu (SSK), Emekli Sandigi (ES), and Bag-Kur (BK)- covering different areas of the labor market. SSK covers public and private sector workers excluding civil servants, while ES covers civil servants and BK covers the self-employed and farmers. Although the three social security administrations differ and have specific issues related to their individual legislative framework, they share many problems. Most important, they were all severely affected by the elimination of the minimum retirement age in 1992, which led to average retirement ages as young as 47 years of age in SSK and 48 years of age in ES, the lowest in the world. Clearly, no social security system can thrive when people are spending more time collecting benefits than they spend contributing to the system. Without reform, the discrepancy between the time spent contributing and the time spent collecting benefits would grow even further as life expectancy increases. 26 This chapter draws on background papers by Esperanza Lasagabaster and others (1999) and Guven Sak (2000) It also benefits from input from by the Bank team working on pension reform in Turkey. 21 "Averting the Old Age Crisis: Policies to Protect Old and Promote Growth", World Bank Policy Research Report, Oxford University Press (1994). 28 86. The new government has moved quickly to prepare a two-phase reform strategy to correct the growing deficits and improve the social security system's operational effectiveness. The first phase, based on legislation enacted in August 1999, has focused on policy reforms to the public PAYG pension system to stem fiscal deficits approaching 3 percent of GNP per year. One of the most important achievements is the re-introduction of the minimum retirement age for all three schemes. The new law reduces the fiscal deficits considerably but does not eliminate them. The second phase of reform, planned for 2000-01, focuses on strengthening the administrative and institutional underpinnings of the system, harmonizing the three schemes, extending the system's coverage, separating the accounts of the pension system from associated health and unemployment insurance, and eliminating accumulated contribution arrears. This phase will also introduce a framework for supplementary private pension schemes. These reforms will set the stage for a transition over the longer term to a broader, financially sound and administratively articulate social security system. With the August 1999 reform, the government has some fiscal breathing room in which to consider next steps. 87. The longer term social security reform agenda must go further and address a series of complex structural issues. First, while the August reform will result in major fiscal improvement, the PAYG system will continue to run a residual deficit of 1.5 percent of GNP per year over the medium term. Moreover, this deficit is projected to increase to some 4 percent of GNP by 2035. The projected reappearance of large deficits is attributable to the expected increase in life expectancy. Internationally, a balanced PAYG system requires contribution periods 2-3 times longer than the benefit receipt period. Turkey still is far from this norm. To maintain financial balance as life expectancy increases, either retirement ages have to rise, benefits must be cut, or contribution rates must rise. This recurrent fiscal problem is an on-going issue for all countries that rely exclusively on PAYG pension systems. 88. Coverage and compliance represent a second set of structural problems. The less than full coverage of the Turkish system can be attributed both to a lack of compliance and the voluntary coverage for certain sectors of the labor force. The administrative reform under preparation, together with the provision in the August 1999 legislation that pensions be calculated based on the full earnings history of the beneficiary, will help address this issue. But coverage and compliance are likely to remain problems until the underlying incentives to participate and contribute to the system are strengthened, complemented by a well conceived implementation strategy. The limited coverage and recurrent deficits together pose a further issue of equity both between and within generations in Turkey. Since much of the working age population is not covered by any of the three social security schemes and benefits remain generous compared to contributions, government revenue collected more broadly is being used to cover deficits in in the pension system for the less than 50 percent of the employed covered by the schemes. Moreover, covered workers generally have higher incomes than uncovered workers. 89. A final set of structural issues involves the health insurance and unemployment insurance components of the system. Key issues include extending the coverage of the health insurance schemes and improving the incentive structure of the unemployment insurance. 90. Based on a detailed diagnostic of the current social security system including an evaluation of the recent legislative reform, this chapter provides policy guidance for developing the longer-term reform agenda. First, the performance of the PAYG pension system to date is 29 reviewed, and the impact of the August 1999 reform analyzed. Then, the main outlines of the second phase of reform are presented including the administrative reform agenda and the new framework for private pension schemes. Building on this background, the chapter examines the challenging choices and trade-offs that the government will face in ensuring the pension system's medium- and long-term sustainability while guaranteeing income security to the elderly. Various policy options for achieving these objectives are presented. They range from a limited program based on further rationalization of the PAYG scheme to a full-scale reform involving rationalization of the PAYG scheme accompanied by introduction of a privately managed funded pillar that will allow a broader diversification of retirement income risks. 91. The focus then shifts to the health insurance system. Any change to the mandatory pension system will affect health insurance due to strong financial and institutional linkages. While reform of health insurance should be pursued in the context of social security reform, Turkey must also address broader changes to the public health system, an undertaking that surpasses the scope of this Report. The chapter also briefly reviews the newly introduced unemployment insurance scheme and proposes measures to improve its incentive structure. 2 PROBLEMS IN THE PRE-REFORM SYSTEM 92. Prior to the August 1999 reform, all three social security institutions were running substantial deficits, as much as 3 percent of GNP per year in the late 1 990s. While the schemes suffered from the inherent design flaws listed below, these flaws were exacerbated by subsequent measures imposed by Parliament (see table 2.1). 93. Lack of a minimum retirement age. Prior to reform, Turkey was the only country in the world that simultaneously lacked a minimum retirement age and had a low minimum contribution period for collecting a pension, as low as 10 years for some contributors to SSK (see table 2. 1). Other countries with no minimum retirement age, such as Brazil, also face fiscal problems, but their minimum contribution period for men is 35 years and for women 30 years. The combination in Turkey of no minimum retirement age and a low minimum contribution period led to people being allowed to retire as early as age 35. 94. PA YG disequilibrium. In the long run, PAYG equilibrium requires that the ratio of the average benefit to the average contribution equal the ratio of the contribution period to the retirement period. Since in Turkey the contribution period was shorter than the retirement period, the affordable benefit should have been lower than the contribution rate. Contribution rates in the three systems ranged from 15 percent in Bag-Kur for farmers to 35 percent in ES, while benefit rates ranged from 54 percent in SSK for those contributing only ten years to 75 percent in ES for basic pensions. Clearly, Turkey could not sustain a system in which the contribution period was in some cases only half the retirement period, as in SSK, but benefit rates were more than double contribution rates. 95. Lack of automatic indexation of contribution ceiling. Given the very high inflation rates experienced in Turkey over the past decade, the lack of indexation of the contribution ceiling in SSK played further havoc with the system. The ceiling on wages subject to social security contributions in the fall of 1995 actually fell below the minimum wage. While statutory contribution rates in Turkey are relatively high, the failure to index the contribution ceiling has 30 caused effective contribution rates to fluctuate from high to low resulting in fluctuating revenues. Pensions, which are tied to the same ceiling as wages, have also fluctuated. This fluctuation has generated a great deal of income uncertainty during retirement, the avoidance of which is the chief aim of pension systems. As a result, in each of the three systems, the average pension is high relative to the contribution rate, but paradoxically low relative to average income. Table 2.1: Pension Summary Data Prior to the 1999 Social Security Reform Fund Contributors Contribuiion ratw Minimum Benefits determination and beneficiies contribution and reference period (thousands)R period ___ SSK Contributors: 6,376 33.5 percent First option First option Private sector 20 percent for 13.9 years plus 25 60 percent for 13.9 years plus employees and o/w male: 5,574 pensions; years of SSK I percent for each additional workers from state 12 percent for membership (male) or 240 days enterprises Total beneficiaries: 2,759 health care. and 20 (female) 1.5 percent for other Second option Old-age pensioners: 1,792 benefits Second option 54 percent plus I percent for 10 years each additional 240 days and if age 55 (male) or 1% for each year above 50 (female) minimum age Reference wage: Last 5 years Emekli Sandigi Contributors: 1,995 35 percent 25 years - male 75 percent plus I percent for Civil servants combined for 20 years - female each additional year and 1% o/w male: 1,582 pensions and for each year above ages of health care 55/50 Total beneficiaries: 1, 14 Reference wage: Last year Old-age pensioners: 679 Bag-Kur Contributors: 2,805 32 percent First option First option Self-employed 20 percent for 25 years - male 70 percent plus 1 percent for olw male: 2,442 pensions 20 years - female each additional year (15 percent for Total beneficiaries: 877 farners); Second option Second option 12 percent for 15 years 60 percent plus I percent for Old-age pensioners: 591 health care if age 55 (male) or each additional year and 1% 50 (female) for each year above the minimum age. Reference wage: Last year Source: World Bank. 96. Weak link between contributions and benefits. While pre-reform pensions in SSK were linked in parts to wages paid in the last five years, in ES and Bag-Kur only the last year's wages were used to calculate pensions. These policies resulted in severe underreporting of wages, so that contribution revenues in SSK and Bag-Kur were lower without a commensurate reduction in pensions. Furthermore, for both SSK and Bag-Kur, Parliament added an unfunded social assistance benefit to be paid to all pensioners, which at one point accounted for as much as 80 percent of the average payment to pensioners. Since this amount was flat and given to all pensioners, the link between contributions and benefits actually received was almost completely destroyed. The social assistance payment was frozen in nominal terms in 1996, restoring some link between contributions and benefits even prior to the 1999 reform. 97. Incentives to contribute for too short a period The benefit formulae in all three systems promoted short contribution histories. In SSK, for example, an individual received a 60 percent 2 Information on contributors and beneficiaries based on data from 1997. 31 benefit for 14 years of service, an accrual rate of more than 4 percent per year of service, with only 1.5 percent more for each additional year of service. 98. In summary, Turkey's pension scheme has suffered from serious weaknesses. Its primary objective of old age income security has been distorted over the years. The system has become a major fiscal burden, impairing Turkey's macroeconomic stability and economic growth. Deficits have resulted in poorly targeted subsidies while crowding out other social investments. The system has fostered high intergenerational inequities, as younger participants have contributed to a non-sustainable system. Evasion has been encouraged by high replacement rates for short contribution periods and other administrative weaknesses. Finally, the system's structural ftagmentation by economic sectors has impaired coverage expansion and introduced inequities among participants. 3 THE 1999 REFORM: AVERTING A FISCAL CRISIS 99. Over recent years, a series of pension reform proposals to address the looming fiscal crisis were discussed. The proposals suggested gradually introducing a minimum retirement age and rationalizing benefits. Political consensus, however, was insufficient to retrench existing entitlements. In the meantime the fiscal burden of the system continued to grow. Aware of the imminent fiscal crisis, the new government that took office in May 1999 identified pension reform as a priority within the economic agenda and moved quickly to prepare and build consensus around a new reform proposal. The August 1999 reforn represented an enornous effort to undo some of the worst problems in the pension system. There are four options to reformn a PAYG system: (i) increasing the contribution period, (ii) reducing the benefit collection period through a retirement age increase, (iii) reducing benefits paid to retirees, and (iv) increasing contribution rates. The August 1999 reform included components of all four of these measures in an attempt to bring the system back towards balance. 100. The main features of the reformed pension system are presented below and summarized in table 2.2. * Retirement age and minimum contribution period The August 1999 reform introduced a minimum retirement age of 58 (female) and 60 (male) for contributors entering the reformned system. These retirement ages are still low by international standards and insufficient to achieve long-term sustainability. Current contributors are allowed a gradual transition period, starting with a minimum retirement age of 38 (female) and 43 (male) for contributors who are less than two years away from retirement and increasing to 52 (female) and 56 (male) for those who are more than ten years away from retirement. The minimum contribution period for new SSK entrants is established at 20 years and for current SSK participants is gradually raised from 13.9 years to 16.7 years. * Benefit rules. The benefit formnula for new entrants to SSK and BK provides for a 3.5 percent accrual rate for the first ten years of contributions, a 2 percent accrual rate for the next 15 years, and a 1.5 percent accrual rate for each year thereafter. Thus the replacement rates for 10, 25, and 40 years of contributions are 35 percent, 65 percent, and 82.5 percent respectively as a percentage of the reference wage. For current participants, benefits will be a weighted average of the old and new benefit formulae. 32 * Reference wage period The reform provides for gradual expansion of the reference wage period to the full contribution history. This measure will improve linkages between contributions and benefits. Basing a pension on lifetime earnings has many advantages. The most important is that it removes incentives to game the system; as noted above, people had incentives under the old system to underreport earnings early in their careers to minimize contributions, but to report correctly or even over report earnings during the years used in the pension calculation to maximize pensions. The return to the individual is maximized, but the cost is borne by the pension system. Basing pensions on lifetime earnings makes the base on which contributions are paid identical to the base on which pensions are calculated, removing incentives to distort earnings. The second advantage of lengthening the wage base is that it often reduces the benefits, since the career average wage is usually below the average of the past five years. * Pension indexation. Discretionary pension indexation generally based on civil service wage increases was replaced by a transparent and financially conservative rule-pensions will now be indexed to the consumer price index (CPI). The ceiling on SSK contributions was raised to three times the minimum contribution base at the time of the reform, then it was raised to four times in June 2000 and can be further increased by executive decree up to five times.29 The ceiling will be automatically indexed to the CPI and real GDP growth rate. Figure 2.1: Projected Improvement in the Deficit of the SSI System, 1998-2050 0% -7- ---f -2% -6% Z -8%/ - 0 -10% - -12% - -14% - -16% -18% 1998 2003 2008 2013 2018 2023 2028 2033 2038 2043 2048 Year - WB T Treasury Base-Line Source: Treasury and World Bank. 2 The ceiling was 60 percent higher than the minimum wage at the time of the reform. 33 Table 2.2: Conmparative Data on the Pension System Pre- and Post-1999 Reform Fund Minimum Minimum Benefits determination Reference period retirement age contribution period for benefits Pre-reforml Post-reforn Pre-reibrm Post-rejir,n Pfre-reform Post-reform Pre-reformn Post-refortm SSK None 60 - male First option First option First option First option Last 5 years: Full Private sector 58 - female 13.9 years plus 25 20 years 60 percent for 3.5 percent last 10 years contribution employees and years of SSK 5,000 days plus I accrual rate for for those period" workers from membership if male percent for each the first 10 years; paying at the state or 20 years additional 240 2 percent for the contribution enterprises membership if days next 15 years; 1.5 ceiling" female percent for each additional year 55 - male 60 - male Second option Second option Second option Idem Last year Last year 50 - female 58 - female 10 years if age 55 12.5 years 54 percent plus I (male) or 50 percent for each (female) additional 240 days plus 1% for each year beyond minimum age Emekli Sandigi None 60 - male 25 years - male 25 years- 75 percent plus I Remains Last year Civil servants 58 - female 20 years - female percent for each unchanged additional year Bag-Kur 55 - male 60 - male First option 25 years First option Same as SSK Full Self-employed 50-female 58-female 25 years-male 70 percent plus I contribution 20 years - female percent for each period. additional year None 60 - male Second option 15 years Second option Same as SSK 58 - female 15 years if age 55 60 percent plus I (male) or 50 percent for each (female) additional year plus I% for each year beyond minimum age Source: World Bank. 34 101. The impact of the reform is quite dramatic, as shown in figure 2. 1.3° The bulk of the impact is due to the retirement age change and the increase in the contribution ceiling. However, the reform is clearly not adequate in the short run to eliminate deficits and in the longer run the deficits are projected to rise again as the population continues to age and benefits rise due to the delayed impact of the increase in the ceiling on wages. 102. The broad scope of the August 1999 reform notwithstanding, simulation results indicate that the reformed pension system will not attain financial balance in the medium term due to insufficient adjustments in benefits and the generous eligibility criteria applied to the transition generation. The system's deficit is projected to decrease until 2015. Thereafter, the trend will is projected to reverse with projected deficits surpassing 5 percent of GDP by 2050. The increasing deficits reflect in part the adverse impact of future increases in longevity. After 2050, the deficit is projected to increase further. By then, it is assumed that coverage expansion will have stabilized. However, the relative share of retirees will continue to increase as a result of the coverage expansion observed in earlier years. 4 SECOND STAGE OF REFORMS: BUILDING A SOUND OLD AGE RETIREMENT INCOME SYSTEM 4.1 Administrative Reform 103. Now that the initial policy phase of the pension reform has been enacted, the government is turning its attention to a second phase of less high profile, but equally important, administrative reform. The administrative reform faces two challenges. The first is to support effective implementation of the PAYG policy reform. The August 1999 policy reform has shifted the base for calculation of pension benefits to the full working life of the contributor. This requires reorganization of the pension administration, in particular an increase in information sharing among the three pension funds, and a major upgrade of their management information systems. The second challenge is to improve the coverage and corhpliance of the PAYG pension system. The present administrative structure is unable to identify and enforce compliance from the nearly 40 percent of the working-age population who do not contribute to the pension system. Improving coverage and compliance will not only strengthen the social protection system in Turkey, it will also contribute to improving the financial balances of the pension system. 104. The government is well aware that meeting these challenges will necessitate clear separation of the management of pension and health insurance, and progressive integration of the three pension funds under a single administration over the longer term. It has also drafted a time- bound strategy for the administrative reform of the social security system and legislation to reorganize the key institutions involved has been submitted to Parliament. 'J Figure 2.1 was produced using the World Bank PROST model. Treasury estimates using an earlier ILO model are more optimistic. While the Treasury model is more detailed, the optimistic Treasury projections can be attributed to 3 sets of assumptions: (i) no improvement in life expectancy after 25 years, (ii) continuous employment and contribution from the time of entry ir the labor force until retirement, death, or invalidity, and (iii) high levels of employment and coverage growth arising from specifying an endogenous relationship between real GDP growth and wage bill growth with real GDP growth assumed at 6% in the long run. 35 105. Implementing the policy reform. Administrative reform is needed in order for the institutions managing Turkey's public pension system to cope with the additional administrative burden imposed by the new pension legislation. In addition to collecting revenue and paying pensions, the pension institutions must now maintain detailed personal work histories on each contributor to determine initial benefit levels upon retirement. A key objective of lengthening the wage history is to create stronger incentives not to underreport wages. However, in practice, this will depend crucially on the ready availability of data and accurate tracking of the wage histories of individual contributors. The three pension institutions will need to integrate their existing databases into a shared informnation system which is both comprehensive and readily accessible. Such an integrated management information system could be expanded to incorporate data sharing with the voluntary private pension funds as these develop over time. 106. Improving coverage and compliance. The administrative reform should also improve the ability of the pension institutions to expand the coverage of their programs and ensure compliance. Both BK and SSK have been unable to collect all due revenues. For example, SSK collects only 85 percent of the revenues declared by emplovers. In addition, many employers simply do not register with the pension institutions. Others that do register their companies fail to register all workers, underreport hours worked or underreport wages paid. Ensuring full collection of declared payments and eliminating accumulated arrears will require additional staff resources for inspection. Identifying the full wage base will require considerable change in the overall administrative approach. As a first step, information sharing among SSK, ES, and BK will improve the accuracy of data about known payers and prevent individuals from collecting benefits from multiple agencies. However, this will not provide information on people outside the system. Improving coverage will require a new approach to data collection, establishment of a unique identifier number used by all three agencies, and a single database of persons of working age that will be readily accessible to all three agencies. This new database could be used both to help identify non-contributors and to provide contributors and pensioners with improved information on their individual positions with the PAYG pension system. New internal incentives, reorganization of management structures and intensive training will ensure that the new integrated information systems are used effectively. 107. Separating pensions and health insurance. The three pension institutions will need to separate completely pension administration from health care provision and financing. Currently, the management, finances and accounting of pensions and health insurance are effectively merged in a manner which inhibits efficient and transparent operation. The administrative separation of pensions and health insurance will help improve service to contributors and beneficiaries for both pensions and health insurance. It will also ensure that an accurate and transparent picture of pension and health insurance finances is available at all times. By improving service provision and financial transparency, this administrative separation of pensions and health insurance will inspire more trust in the public social security system, which in turn will facilitate efforts to improve compliance. The separation will also facilitate deeper structural reform of the public health insurance system which is the largest provider of health services in Turkey. 108. Integration under single administration. Progressively integrating the three pension institutions under a single administration is a longer-term goal. At present, the three institutions develop their own administrative practices and their own solutions to administrative problems. 36 The lack of harmonization and the development of individual, often incompatible, solutions to administrative problems cannot be addressed from inside the individual institutions. Achieving uniform norms and standards across the agencies will require guidance and monitoring from one central government structure. Currently, ES is under the Ministry of Finance and the other two agencies are under the Ministry of Labor. Such an integration could logically take place under the Ministry of Labor. An administrative unit would have to be set up within the Ministry of Labor capable of guiding and monitoring the administration of the three funds and harmonizing their activities. This unit could serve as the policy development center for all three public pension funds and take responsibility for monitoring the impact of future pension reforms. It could also maintain the joint database. 4.2 Voluntary Funded Private Pensions 109. As part of the second phase of social security reform, the government has prepared the legal framework for introducing a privately funded pillar in the pension system. The new private pension schemes will complement the existing PAYG system. Participants in the new schemes will continue to be required to contribute to one of the PAYG programs. Through this reform, the government is aiming at three objectives. The first is to provide a comprehensive regulatory framework for private pension schemes which already exist to some extent in Turkey. The second objective is to expand the scope of the pension system, given the low level of benefits currently provided, without affecting the financial balances of the PAYG programs. The third objective is to expand the range of long-term savings instruments in the country. Development of voluntary private pension schemes under the new framework will prepare the ground for the possible transition to a full multi-pillar pension system over the longer term. Draft legislation for the new regulatory framework has been submitted to Parliament together with a separate law on the tax treatment of the private pension schemes. Both laws are expected to be enacted before the end of 2000. 110. The need for a new regulatory framework stems in part from the fact that the market for private pension schemes is already developing, albeit in a limited way. The regulation and monitoring of the existing products has been relatively weak.3 As the market for private pensions will expand, the government is right to introduce more comprehensive regulation before the market shapes itself. The new regulatory framework will help to reorganize the current private pensions market and bring about qualitative changes such as better portfolio management and a broader asset spectrum. Decisions will need to be made regarding the status of the existing products as many of them may not qualify under a more rigorous regulatory regime. 111. The promotion of voluntary private pension schemes is designed to expand the scope and attractiveness of Turkey's pension system without impacting the financial bottom line of the PAYG system. Given its current fiscal situation and that of the three public pension agencies, the government has chosen not to divert contributions from the public agencies to private pension funds at this stage. The government has also been reluctant to add a mandatory funded component to the existing contribution structure, given the already high contribution rates for most salary levels. Its short-term solution is to promote the development of voluntary private pensions to complement the public pension programs in an attempt to achieve some of the This section draws on a background paper by Sak 2000. 37 benefits of a funded system without incurring most of the costs. The new schemes can also contribute to improved coverage and compliance of the PAYG system as participation in the latter will be a prerequisite. This incentive will operate as long as the administrative reformn proceeds in parallel and the private pension schemes are incorporated into the integrated information system. 112. The government's efforts to promote the development of private pension schemes stems as well from a view that these schemes can be an important vehicle to enlarge the size and depth of the financial system in Turkey over the medium term. The new private schemes will contribute to increased savings and capital market development while providing opportunities for better pensions for individuals under a regulated setting. Currently the level of financial market development in Turkey appears to be adequate for fostering private pensions. There are financial institutions equipped to operate pension schemes and appropriate financial instruments are available in the market. The figures indicate that the capital market would have the capacity to absorb the additional financial savings flowing from private pensions. 113. Designing the right incentives. It should be noted at the outset that voluntary pension systems that complement a public pension system tend to appeal to higher-income individuals who already are saving a portion of their income. This saving is often transferred to the voluntary pension schemes from some other savings vehicle. The voluntary pension system may therefore lead to little increase in aggregate savings, at least in the short run. Furthermore, some of the demand for a private pension system also stems from the perception among workers that there is a gap in their earnings that needs to be filled. Despite the new pension law, which establishes minimum retirement ages of 60 and 58 for new entrants, workers expect to quit their jobs or to be laid off in their upper forties as they are now, and so perceive a gap between the time when their jobs end and the time when their pensions begin. The perception is that the voluntary pension could fill that gap, allowing those workers who choose to contribute extra effectively to continue to retire at age 47, the current average age of retirement, with little loss of income. To avoid the resultant negative impact on the PAYG system of losing these contributors, it is important that the new private pension schemes provide sufficient disincentives to dissuade workers from retiring too early. The current proposal which would result in loss of tax exemptions if money is withdrawn before the retirement age is in line with international practice. 114. Turkey already has relatively high labor costs and high payroll taxes. The effective rates are being raised even further by the August 1999 reforms, as discussed above. In order to induce workers to join the system and part with even more of their salaries, the system has to be attractive, flexible, and transparent. Putting excessive restrictions on the accounts will reduce their value to workers and make them less attractive. In this light, the voluntary pension system should not require disability or death insurance or annuitization upon retirement. Disability and death insurance is expensive, and since most participants in the voluntary scheme would be covered by one of the public schemes and would be covered in the event of disability or death, duplication in the voluntary scheme is unnecessary. The current proposal for the voluntary pension system does not include mandatory annuitization or disability and death insurance. 115. Tax treatment. The tax treatment of voluntary private pensions should be consistent with international practices while ensuring competitiveness with existing long-term savings products in Turkey. Typically either: (i) the contributions that finance the plan and the current investment 38 return on the saved assets are exempt, while tax is imposed on the entire value of the fund when it is withdrawn (the "E/E/T" structure); or (ii) the contribution at the time it is made is taxed, but then the investment return and that part of the payout on maturity which represents repayment of the original taxed contributions are exempt (the "T/E/E" structure). In the draft tax law, contributions up to 10 percent of wage earnings with a ceiling of 50 percent of the annual minimum wage are tax exempt. The investment returns on the fund will also be tax exempt and the full benefits will be subject to income tax. Early withdrawals will be subject to tax penalties. This is in effect a form of "E/E/T". 116. Turkey should consider a broader reform of financial sector taxation which will move the entire tax structure closer to international norms and ensure a set of tax advantages which provides clear incentives in favor of longer-term savings instruments. Once the new law is enacted, broadly equivalent tax treatment will be provided to pension funds, life insurance products and health insurance products. However, there will still exist short-term financial instruments in Turkey which have more generous tax treatment than the pension funds, although these exemptions are due to expire in 2001. Mutual funds and equities held more than three months are fully tax exempt at both the buying and selling points, including investment earnings. Moreover, the lack of tax differentiation between pension and life insurance products will leave incentive distortions in place given that pension products are restricted by the minimum retirement age when a person can receive benefits, while life insurance products can be cashed in after 10 years. In addition to addressing these incentive issues, the broader overhaul of financial sector taxation should ensure that favorable tax treatment is given only to products from financial institutions which are supervised and regulated as such. 5 LONGER-RUN NEXT STEPS 117. The objectives and scope of the social security reform agenda need to be broadened substantially over time to properly address the remaining structural flaws and financial problems Figure 2.2: Projected Aging of Turkish that will continue to affect the pension system. Population (1997, 2020, 2075) The longer-run reform encompasses the following 70Y ------- objectives: (i) provide adequate income security 60% for the elderly and disabled; (ii) achieve financial o 50%r balance over the medium term and ensure greater C 40%= longer-term sustainability; (iii) deter inequities 30 within the system; (iv) reduce labor market _ 0% distortions and evasion; and (v) build a more s efficient administrative structure to support the pension policy framework. 1% 20 1997 2030 2050 2075 Years 118. Based on the analysis thus far, it is clear a .17 a 1-60 o 61+ that the August 1999 reforms made major Source: World Bank estimates. progress toward a fiscally sustainable pension reform including imposing a minimum retirement age. These steps were politically difficult but absolutely critical. Going back to the four potential reform measures for a PAYG system- increasing the contribution period, reducing the benefit period, raising the contribution rate, and reducing the benefit rate-the August 1999 reform made enormous progress particularly in the 39 last three. As noted above, while some gains have been achieved in lengthening the contribution period, these are less dramatic than the changes in the other measures. It will be necessary to revisit these issues in the future and to revise further the reformned pension system. Importantly, as life expectancy continues to rise, the number of elderly will increase (see figure 2.2 ), with each pensioner collecting pensions for a longer period of time, resulting in a higher stream of pension payments. As fertility continues to decline, fewer young people will join the labor force, resulting in stagnant or declining revenues at the same time that the stream of pension payments is increased. This will once again put financial pressure on the pension system. 119. The ambitious long-term policy objectives outlined above will entail a substantial revision of the existing mandatory insurance system. This section lays out different options for reforming the system: (i) reform the PAYG scheme and maintain its defined benefit (DB) structure; (ii) reform the PAYG scheme but transform it into a notional account scheme; or (iii) implement a greater downsizing of the PAYG scheme accompanied by the introduction of a funded, defined contribution pillar managed by the private sector. Specific examples of each of these options are shown in annex II. 120. The various reform paths will result in retirement income systems that differ in terms of their financing structure and level of political, demographic, and investment risks. But irrespective of the path followed, the reform will need to address a series of social and economic choices. First, to achieve financial viability, the reform will need to revise the targeted benefit level (i.e., replacement rate) and the targeted number of beneficiaries (i.e., minimum retirement age). Trade-offs will need to be made between these two parameters. These choices assume that contribution rates will not be increased as a means to improve the financial situation. Overall social security contribution rates (i.e., 36.5 percent statutory rate for private sector workers) are already high and surpass the OECD average. Second, Turkey's pension system redistributes resources to participants with short contribution periods, penalizing workers with long contribution periods. Third, despite its short contribution requirement, the system has not addressed the broader needs of low income workers. Low income workers might not be able to accumulate enough savings under the mandatory retirement system to provide them with an adequate income level during their old age, even with long working careers and contribution periods. To protect low income workers, the reform should examine whether a minimum pension level should be guaranteed. This guarantee could be satisfied through a redistribution of resources within the pension system or could be financed from general tax revenues. 121. Fourth, the reform will need to separate the health and pension schemes, preventing further cross-subsidies, and examine the current fragmentation of the pension system into four separate schemes. A unified pension scheme will better address the needs of a flexible and more integrated labor market in the future. Although benefits and contribution rates could be equalized without integrating the schemes, a unified scheme will more easily avoid political pressures to raise benefits for special interest groups, which is unjustified on equity grounds. Furthermore, since coverage expansion will vary across schemes, the financial evolution of the schemes will differ over time even if contributions and benefits are harmonized. Unifying the schemes will allow better management of financial resources within the overall pension system. 122. Reform option 1 conventional PAYG reforms. One option available to the government in the longer term is to modify the same set of parameters adjusted by the August 1999 reform: 40 raising contribution periods, lowering benefit periods, raising contribution rates, and lowering benefit rates. Little scope remains to raise contribution rates, given the level in the rest of the OECD. There is some potential, although small, for lowering benefit rates further. This can be accomplished primarily through changing the indexation of the wage base and rationalizing the benefit formula. By the time new entrants begin to retire, the retirement ages will most likely be too low, with people spending over 20 years in retirement, so there will be some scope for reducing the benefit period. Increasing the contribution period has large potential because even with the retirement age for new entrants set at 60 for men and an average age of 22 for starting a work career, there is time to fulfill double the SSK minimum contribution periods. 123. The international evidence suggests that subsequent and continuous PAYG system reforms are time consuming and become more and more difficult politically as the credibility of the reforms are threatened by the constant pace of change. One approach is to follow the example of Ecuador, where legislation is currently being reviewed by Congress that explicitly allows the retirement age to be revised automatically every five years based on changes in life expectancy. This provides some flexibility within the law without the need to revisit the legislation each time life expectancy rises. 124. Reform option 2: applying an innovative approach to PAYG schemes-the NA model. A relatively new paradigm has appeared that seeks to avoid some of the problems associated with repetitive PAYG reforms while at the same time avoiding the costs involved in a transition to a funded system. These notional account systems have been adopted in Sweden, Latvia, and Poland, with a variant adopted in Italy, often coupled with a funded system. These systems adopt the vocabulary of fully funded defined contribution systems, whereby each contribution of the individual is registered in a notional account and interest is accumulated on the balance in these accounts each year. At retirement, the balance in the account is converted into an annuity, based on life expectancy at retirement. This portion of the system is completely analogous to a funded defined contribution system. However, the system is still run on a PAYG basis. The contributions are recorded in these notional accounts, but the money is not set aside or invested. Instead it is used immediately to pay pensioners, although a small reserve may be accumulated by the government to cover future deficits. The system automatically reduces pensions as life expectancy rises. However, there is also a potential poverty impact. The reduction in pensions as longevity increases with no change in retirement behavior will result in some individuals receiving pensions inadequate to live on. At that stage there undoubtedly will be demands to introduce minimum pensions or even revise the benefit formulae. 125. Reform option 3: the multi-pillar reform. Thus far, the discussion has focused on the establishment of a more viable retirement income system through the rationalization of the PAYG scheme. The demographic and political risks that afflict PAYG schemes have been pointed out and the advantages of the NA model over the conventional PAYG model in trying to mitigate the political risks have been outlined. However, the NA model remains a PAYG system and cannot escape the demographic issues inherent in all PAYG systems. Growing international awareness of PAYG risks and the need to ensure adequate income security for future generations have prompted a wave of further pension reforms that have led to the establishment of privately managed funded pillars --primarily defined contribution [DC] within the mandatory retirement income system (see box 2. 1). 41 126. The fiscally sustainable rate of return on a PAYG system in the long run is only equal to the rate of wage growth, while the sustainable rate of return on a DC system is equal to the real rate of interest. The real rate of interest in the long run usually exceeds the rate of real wage growth, often by 1 to 2 percentage points, suggesting that the same contribution can lead to higher benefits in a DC system. The DC system also potentially can raise the level of national savings and deepen the domestic capital market. At the same time, DC systems tend to reduce labor market distortions even more than NA systems, making the labor market function better. However, the pure DC system cannot redistribute toward the poor, so that some PAYG component is needed to undertake the redistributive function if such is desired. Empirical evidence from Chile's privatization of its pension system in 1981 indicates that the reform had a strong impact on improving the efficiency of existing savings ( Holzmann 1998). Box 21. FPension Funds and Capittal Markets: Lessons from Chile The0 Chilean pension rfrm was undertaken in 1981 and has served as the prototype for other Latin American countries and OECD niiddle-income countries (e.g.. Poad and Hngary). The reform tiransformd Chile's PAYG scheme into a defined contribution scheme. prvatly manged. Workers contribute 10 percent of their earnings, subject to a ceiling for old age pensions. An additional 3 percent (on aa is allocaed topay for disabilityand survivors insurance as well as administrative costs. Each pension fund management company subcontracts group life and disability insurance for its affiliates with authorized private insurance companies. At theoutset of the refomChile's capitalmakets werepoorl developed. Stock market capitalization (1981-1985) and total value traded were 15 percent and 5 percent of GDP, repectively. To ensure the safety of workers' assets and adequate returns on their investments, the newly created pension system estblished tighter regulations and oversight than those observed in countries with more developed financial mre: The "tiist' and irustoee figures for pension plas and their manager, frequently observed in developed financi markets were considered insufficient and wecre rerinfrced through the creation of pension uinds and pension fund administrators as etdisinct le ntities. Investment regulaions werecarefully devis gand rous investent limits were initially applied. In 1981, investment limits were 100 perentrof goveramnt securitis, 80 ptent tgae bonds, and 70 percent of bank liabilitics (reduced to 40 percent in 1982). Equity investment depicting highr vo tlit risk,ereinitially barred. In 1985, the ceiling on govemment securities was lowered to 50 perent. a qutyi i er ldupt30percenit fulations have continued to evolve gradually over the years. permitting investmets in riskier instrumernts. As of toa ivetment lms in equtiies e nearly 40 percent. Investment limits in foreign securities. prsently equivalent to 12 percentofthe po were onl grually introduced, limiting opportunities for diversiftcation. The portfolio of Chile's pension Ainds has eld along wit a it opents and investment rules, as the table below indicates. The overzall ceof the Chile e fund system has been very positive. By the end of 1997, pension funds in Chile managed asets valued at44 perent f(DP. Rea riates f reumrn. austed fir aministrative costs surpassed 6 percent over the 1981-1997 period. Pension funds have been amar driving force for innovaion and rwth in Chile's capital markets. Today, pension funds are the largest holders ofmortgao corporate bonds, and equity securities. CHLEANPENSO FUNDS: PORTLIO COMPOSITION _________ ______:X a of toth a l assetXi ,- iTY FASSET 1982 1985 1990 1919 1997 :bwmnrnt s estimates that new entrants with 40 years of work and retiring at 65 will be able to receive a combined replacment rate of 60 percent of last salary. Following the reform, redistribution toward lower- income workers will be supported from general tax revenues. The laster will cover the differnce between the combined pension from the two pillars and pre-established minimum pension guarantee. The regulatory structure applied in Poland is more rigorous than that of the UK in line with the lesser development of Polish capital markets. The Polish mandatory funded pillar is restricted to the DC structure. Pension funds are operated as legal entities and are administrated by specilised pension fund management companies. In addition, pension funds are subject to investment restrictions. Source: Esperanza Lasagabaster and others (1999). 128. What should be the size of the public and private pillars? Within the multi-pillar framework, there is a wide range of international experiences and policy options. At one end of the spectrum are countries that are about to introduce small DC pillars, financed with contributions equivalent to 2-3 percent of wages (e.g., Sweden 2000). At the other end there are countries that have privatized their mandatory retirement systems and minimized the public sector role to the provision of a minimum pension guarantee financed from general tax revenues (e.g., Chile or Mexico). The PAYG schemes in these latter countries were closed to new entrants. A third set of countries has pursued greater balance between the size of the funded and unfunded 43 pillars and the private and public sector roles (e.g., Switzerland, the United Kingdom, Argentina, Poland, and Hungary). Box 2.2 describes the multi-pillar systems of the United Kingdom and Poland and highlights similarities and differences. This third approach permits greater diversification among the sources of retirement income and the investment and demographic risks associated with each one of the pillars.12 129. A balanced multi-pillar option could be a way for Turkey to establish a diversified and more secure mandatory retirement income system. The PAYG pillar would be smaller than it is currently in terms of both revenues and benefits. The new funded pillar would be established from a diversion of current contributions allocated to the PAYG pillar. Table 2.3: Advantages and Disadvantages of Long-run Options LOn RAun Refor"n q0!ion Advantakps Disadvantages More Conventional PAYG reforms No transition costs Requires repeated reforms which become politically difficult Encourages evasion as social security pensions become less valuable and as rules change constantly Notional Accounts PAYG Automatic link between reduced Still may not be fiscally sustainable benefits and increased life without further reforms expectancy which does not require Reduced benefits can lead to greater policy interventions poverty among the elderly No transition costs Multipillar with Mandatory Funded Better benefits with lower Transition costs in the short run Defined Contribution Component contribution rates Better fiscal sustainability Higher national savings Better labor market linkages Diversification between public and private benefits Source: World Bank. 6 HEALTH INSURANCE 6.1 Structure and Performance to Date 130. The social security network is the largest health insurance provider in Turkey. About 35 million people are covered under the three health insurance schemes of the social security system This is about 85 percent of the total insured population. The remaining 15 percent of insurees (about 6 million) are active civil servants and their dependents whose health care benefits are funded directly from general revenues. Private health insurance coverage in the country is small, reaching only 500,000 people. Private health insurance is largely indemnity insurance. It is currently not profitable and in most cases in need of cross-subsidies from other insurance packages, mostly life insurance that is offered in parallel. Some private clients are also insured under one of the social insurance schemes but choose double insurance to get access to better 32 In some countries, the choice to balance the two pillars has also been shaped by the initial fiscal costs associated with transforming the implicit debt of the PAYG system into explicit debt. The Turkish pension system would be confronted with high fiscal transition costs if a large funded pillar were established. 44 quality services from private health providers.,, About 21 million people are not covered by public health insurance and by default fall under the responsibility of the service network of the Ministry of Health. Among them, about 4.8 million poor receive subsidies for health care through the "green card" scheme operated by MOH, which entitles them to health care without any co-payments. 131. Premiums paid by employees and employers constitute the main source of financing for health care provided by the social security system. Health insurance premium rates in SSK are 12 percent, out of which 1 percent is allocated to maternity leave compensation. In BK, health insurance premiums have been increased to 15 percent following the recent amendments to the social insurance law. A common statutory contribution rate of 35 percent is paid to cover pension and health expenditures in ES, but no allocative formula exists to separate the two. Besides premiums and small co-payments, SSK health facilities receive some funding from fees paid by non-members using SSK services. The SSK health insurance scheme has maintained a surplus, which has been used to subsidize pension payments. BK health insurance has run relatively small deficits in recent years. In ES, health expenditures have accounted for 15 percent of total expenditures on average during the 1996-1997 period. On the surface, the health insurance schemes seem to have fewer financial problems than the pension schemes. Health expenditures in BK, for example, have contributed thus far only to a limited extent to total BK deficits. Rising deficits, however, are a growing concern. The BK scheme faces difficulties in regular premium collection. Irregular income and low financial capacity result in poor compliance rates. The system's capacity for collection enforcement is also weak. 132. The health system in Turkey clearly has been under funded for a long time. Total health expenditures by the three social insurance funds have accounted for 1 percent of GDP and only 17 percent of total health expenditure. In the last two decades, total health expenditure in Turkey has been around 3.7 percent of GNP, while the average for OECD countries has been 7.5 percent. Annual per capita health expenditure has been in the range of US$100 in recent years.34 Table 2.5: Health Services Performance Indicators Comparison Argentina Chile Russia Turkey UK EU average Beds per 1000 population 4.6 3.1 2.4 2.3 4.6 Physicians per 1000 population 2.6 1.1 3.8 0.9 1.7 2.9 Average distance to PHC facility (km) 17 3.7 20.1 Average length of stay 14.52 7.9 6.9 12.2 Average length of stay (acute hospitals) 9.14 7.0 5.7 4.8 8.7 Occupancy rate 83.37 72.6 55.6 76.7 Inpatient admission rate 21.6 5.9 15.2 17.6 Acute inpatient admission rate 5.7 11.7 14.3 Source: Argentina and Chile - WB LAC statistics, Ministry of Health (1995-97); other HFA indicators, Europe. 133. Overall health system performance. An examination of the overall performance of the Turkish health care system reveals conflicting results. On the one hand, the three publicly financed social security schemes offer rather generous benefit packages and cover more than 50 percent of the total population. Yet health outcomes in Turkey are unsatisfactory, the worst in the OECD and below comparable middle-income countries. Infant and maternal mortality are among :' Country Report, Ministry of Health 1997. 1 "Health Sector Reforms in Turkey," Ministry of Health 1999. 45 the highest in middle-income countries (tables 2.4 and 2.5). Life expectancy is nearly ten years below the OECD average. There are large gaps in health status between urban and rural areas, and between eastern and western provinces. The poor carry the highest burden of disease, premature death, and preventable Table -2.4: Selected Basic Health Indicators disabilities. Argentina Chile Turey UK EU 134. Inadequate population Life Expectancy at 75.7 77.4 71.7 79.5 80.7 Birth (females) coverage. Low coverage of health Life Expectancy at 68.6 70.2 66.5 74.1 74.0 insurance affects people's access Birth (males) to health care. About one third of Infant Mortality 24.3 15.6 39.9 6.2 6.1 the total population are not Rate covered by any insurance. GNP per capita 8,970 4,810 3,160 covered by any msurance. ($) Furthermore, even though on paper, about 35 million people Note: Data for Argentina and Chile are 1996 and Turkey is 1997, other are covered under the three countries, 1995. Source: Data of GNP per capita of Argentina, Chile and Turkey are from insurance schemes, the actual Development Economics (Country at a Glance) 9/99, WHO HFA/EURO coverage may be much lower. database. 135. Low utilization patterns. The overall utilization of health services in Turkey is relatively low as compared to international standards. The average bed occupancy is around 64 percent in SSK hospitals while only 54 percent in MOH hospitals. Hospital admissions per 10,000 population are fewer than 700, as compared with the OECD average of 1639 admissions per 10,000. Average annual doctor contacts are 3.5, which is 37 percent below the OECD average of 5.5. This low utilization rate may be due to financial constraints of the population and low quality of services. 136. Another reason for low utilization patterns is poor service quality. Inefficient use of limited resources and poor management by both the SSK and MOH have resulted in a largely run-down public provider system. Long waiting Figure 2.3: Per Capita Health Expenditure times and lack of drugs and qualified staff of Three Schemes (in US$) discourage people from seeking care even when _ covered by insurance. Some hospital and 200 _ _ 100 medical payments are delayed up to 12 months 160 due to excess bureaucracy. 140 '0120 137. Equity concerns. The overall distribution s of health benefits is inequitable. Even among the 60 40 insured population, benefits are unevenly 20 distributed. Per capita health expenditure among i °62 1993 19t4 1995 1996 the three schemes shows significant variation Yea, (figure 2.3). Per capita health spending of SSK V SSK E--E-S BK decreased from US$58 in 1992 to US$41.2 in Source: World Bank. 1996 while that of BK increased from US$ 19 to US$23.2 in the same period. Only ES doubled its expenditure during the period (from US$97 to US$188). Differences in benefit coverage partly result in variations in expenditure levels. SSK and BK insurance benefits are supposed to cover all outpatient and inpatient diagnosis and treatment. SSK members are served in the institution's own health facilities and referral to 46 outside services is limited. The high per capita health expenditure in ES is due partly to its coverage being limited to retired people, a higher-risk population that has greater needs of medical care, and partly to lack of utilization control. 6.2 Reforming Health Insurance: Short-term Actions and Medium-term Agenda 138. Health insurance under the current social security system cannot meet the needs of health care financing for the entire population. About 34 percent of the population does not have health insurance, and escalating costs under the current system have made an expansion of coverage difficult. Successive governments have attempted to introduce reforms aimed at providing universal coverage to the population. However, little progress has been made so far. More than five years ago, the government announced its intention to establish a General Health Insurance System (GHIS), which aimed to cover the uninsured population. To date, GHIS has not become a reality. In fact, the proposed draft law was not considered by Parliament. 139. Under the second phase of social security reform, the Government intends to separate the health insurance schemes from the pension system and address health insurance and pension issues separately. Thus, the immediate concern should be how best to separate health insurance from the pension system. This issue requires careful assessment of whether to keep three separate health insurance schemes for SSK, BK, and ES, establish one regulatory structure (super-indendency), or create one consolidated national health insurance scheme. In recent years, there has been no shortage of studies and consultant reports on various health insurance models.35 The debate on which health financing model Turkey should develop is ongoing. Ultimately, the selection of an arrangement that fits Turkey's situation is as much a political as a technical decision. 140. Short-term Actions. The momentum created by social security reform has the potential to impel a wider reform of the health sector to address the concerns of technical efficiency and long-term fiscal sustainability of publicly financed health insurance. The following important actions can be launched in the short term: D Separate health expenditure accounts from pension accounts. Under the current schemes, income and expenditures for health insurance are not clearly separated from pension accounts. Assessing the costs of health insurance, including costs of administration, is difficult within the current accounting and recording systems. Separating the accounts for pension and health would increase transparency, provide comparative information, and facilitate actuarial calculation for long-term sustainability of the health system. * Improve the information database. ES has the most advanced information system of the three systems. SSK has relatively reliable data on its pension coverage, but not on its health insurance coverage. Information on pensioners' dependents is not recorded in its database. BK needs to improve information on insurers and dependents. Better data collection is the first step to improve management of these schemes. "Health Financing Report." Health Insurance Commission of Australia 1995, and "Social Security and Health Insurance Reform Project Study Reports" discussed various models. 47 * Maintain a unified premium collection system. Separating the health scheme from the pension scheme does not necessarily mean separating the premium collection function, since contributions are collected from the same employees and employers. * Establish a basic health care package. Wide disparities between services provided and expenditure levels among the three insurance schemes create inequitable access to care. Establishing a unified basic benefit package, accessible to all social security insured members, would increase equity and efficiency in the sector. * Create an independent health insurance superintendency. This structure will be entrusted with establishing and monitoring the basic package, guiding purchasing decisions, resolving consumer grievances, monitoring overall service quality and accountability, translating health priorities into purchasing decisions, and supporting operational research. In addition, this new structure would regulate and supervise the private (currently largely unregulated) private health insurance market. * Target public expenditures to protect the poor. To achieve greater value-for-money for public spending and to reduce inequity in access, subsidies must be targeted to the most vulnerable groups. Under the current system, general revenue subsidizes the deficits of the social security institutions. The poorest are currently not covered under SSK, BK, or ES. Instead of such subsidies, the government should gradually expand coverage by subsidizing full cost for low-income populations. * Initiate sector analysis. A review is needed to shed light on the wider sector implications of social security reform, in particular as it relates to the reform of the MOH and its service network (mandate, capacity, hospital autonomy, financing, and assistance to the poor). Besides reviewing the MOH institutional reform, this study should explore how to increase insurance coverage beyond the formally employed (e.g., subsidized family insurance for those who have the ability to pay a subsidized insurance quota and free access to the basic package for the indigent-current "green card" holders). 141. Medium-term Development Agenda. The long-term goal is to provide all Turkish citizens with access to quality health services on equitable terms in order to improve the health outcomes of the population at large and to spend public resources in an efficient manner. Currently the system is organized around social groups (MOH covering the poor, Social Security covering the middle class, and the private sector covering the wealthier) rather than functions (policy making and evaluation, financing and payments, provision of care). 142. Longer-term development should move toward integrating these functions horizontally. This integration would involve (i) a MOH that sets policy priorities and monitors outcomes and quality; (ii) a social security system that finances and purchases a basic package from a pluralistic provider market; and (iii) a well-regulated private sector that plays a role in offering supplementary insurance and competes in the provider market by selling services under contractual agreements with insurance agents. 143. This long-term goal requires careful sequencing of reform steps that go beyond social security and pension reform. The Health Financing Policy Options Study"6 exarnined various options for Turkey. The study indicates that the most optimistic projection would not extend insurance coverage much beyond 60 percent of the population ten years from now. To achieve 36 "Health Financing Report," Health Insurance Commission of Australia 1995. 48 universal coverage would require significant changes in policies, legislation, and organizational structure. The government's commitment to social security reform opens a unique window of opportunity to initiate this process. 7 UNEMPLOYMENT INSURANCE 7.1 Introducing Unemployment Insurance 144. The severance payments program has traditionally been the primary form of protection against involuntary unemployment in the formal labor market. The scheme also provides benefits upon retirement.37 Benefits are equivalent to a month of salary per year worked up to 12. months. Over recent years, several proposals were prepared to replace the severance payments scheme with the unemployment insurance scheme given that the latter constitutes a better insurance instrument. 145. An unemployment insurance scheme was introduced in the context of the August 1999 reform. The new unemployment insurance scheme became effective in June 2000, but no payments will be made before February 2002. Participation in the unemployment insurance scheme is mandatory for SSK workers. The system will be financed with a 7 percent contribution rate on the wage base (2 percent, 3 percent and 2 percent paid by the employee, employer, and government, respectively).38 Using 1999 as a reference period, collections are estimated to reach 0.7 percent of GDP, including the government subsidy equivalent to 0.2 percent of GDP, whereas expenditures are estimated to be 0.5 percent of GDP (assuming 600,000 workers benefit from the scheme). Based on these estimates, the program would produce a net surplus of 0.2 percent of GDP." 146. Contribution rates surpass those observed for similar programs in other OECD countries. The government will not only subsidize the system through a 2 percent contribution rate but will also assume any deficits that the system might incur. The unlimited government guarantee will not encourage financial discipline and could foster pressures to expand the amount and duration of benefits. 147. Benefits will equal 50 percent of the average of the last four months of net insurable salary, a reasonable amount. However, an alternative phased-down schedule should be explored to encourage greater job search during the early months of unemployment, as discussed below. Moreover, benefits will be provided for long periods even for workers with short contribution periods. For example, workers contributing for 1.5 years will receive benefits for as long as six months. Workers with more than three years contribution history will receive benefits for up to ten months. 148. Moreover, the mandatory severance payments scheme was not eliminated along with the introduction of the unemployment insurance scheme, as earlier proposals had recommended. As 3 Women also received severance payment benefits upon leaving employment due to marriage. 3 The current contributions to the Compulsory Savings Scheme will be reallocated to the new Unemployment Insurance Scheme. Thus, overall payroll taxes for SSK workers will not be raised excluding workers in small enterprises. 3D The number of registered unemployed workers for April 1999 and April 1998 was 1.7 million and 1.5 million, respectively. ("Results of Household Labor Force Survey," State Institute of Statistics 1999) These figures include workers who are not covered by the social insurance system, e.g., young workers and the long-term unemployed. 49 a result, workers who are involuntarily laid off will receive a total sum of severance and unemployment benefits higher than their initial salary. This could encourage shirking on the part of some workers in the anticipation of combined unemployment and severance benefits. 7.2 Improving Incentives of the New Unemployment Insurance Scheme 149. Unemployment insurance can be welfare enhancing if related moral hazard effects are minimized through appropriate policy design and if benefits derived from consumption smoothing outweigh the potential increase in the average unemployment spell. The framework of the newly introduced unemployment insurance in Turkey could benefit from policy changes that will produce a better balance between risk reduction and moral hazard incentives and will prevent additional fiscal costs. First, severance payments have become redundant iristruments to protect workers against the risk of being laid off. Second, the program should be actuarially balanced and its sources of financing should be limited to employers and employees. Third, the structure of benefits needs to be rationalized to minimize moral hazard effects. Policy reforms need to be accompanied by a strong administrative structure to deter fraud and facilitate and monitor job search. 150. With the introduction of unemployment insurance, a government-mandated severance payments scheme is no longer justified. In lieu of such a scheme, private firms and workers should be allowed freely to negotiate severance payments and supplementary retirement plans within the overall compensation package for the worker. Firms may choose to offer supplementary retirement plans to encourage lower turnover of workers and fully internalize benefits derived from investments in worker training. 151. The most critical question is how to address accrued rights of workers under the existing severance payments scheme. Accrued rights should be respected if the duration of unemployment benefits for short contribution periods is reduced, as recommended below. In this case, the severance payments scheme would act a bridge and ease the transition into the new unemployment scheme.,, 152. Although publicly mandated, the program should seek to be actuarially balanced and avoid imposing additional fiscal costs. Over short periods, the program will face surpluses and deficits in line with business cycles. Based on other country experiences, a premium rate of 5 percent (paid solely by employers and employees) should be adequate to finance the program, provided benefits are not overly generous and the program is not subject to fraud. Government estimates of net program costs (using 1999 as a reference point) suggest that a lower contribution rate could be viable, particularly if the benefit structure is tightened. 153. The duration of benefits should be curtailed. Provision of benefits over long periods increases costs and creates serious moral hazard effects, discouraging job search and adjustments to the reservation wage. Empirical evidence supports the view that the duration of the benefit 1O If the duration of unemployment benefits for short contribution periods were reduced to two-three months, individuals who are involuntarily laid off in the immediate years following the introduction of the unemployment insurance scheme would receive limited unemployment benefits. This shortfall could be compensated by rights accumulated under the forner severance payments scheme. 50 affects the unemployment duration and that the escape rate from unemployment increases close to the termination of the benefit.41 154. Instead of a flat 50 percent benefit, unemployment benefits could be scaled down over time. For example, for the first three months, the benefit could amount to 70 percent and could be reduced thereafter to 50 percent and 30 percent. The scaled-down benefit would ensure a minimum protection level but would reduce the expected utility associated with unemployment and encourage a more intense job search in earlier months. 155. The success of the unemployment insurance program will hinge largely on the support of a strong administrative structure. The key functions of the administrative structure include premium collection, registration of the unemployed, benefit calculation, and assistance with employment services and job search. The administrative structure should not only assist with job search but also have the capacity to monitor it and prevent fraud by individuals who have found employment but continue to collect benefits. 8 RECOMMENDATIONS 156. The August 1999 pension reform has gone a long way toward correcting the gross imbalances in the Turkish pension system. Previously, benefit periods were too long, contribution periods too short, and benefit rates too high with respect to the effective contribution rates. Benefit periods have been reduced substantially for new entrants; benefit rates have been cut, while effective contribution rates have been raised. Contribution periods have been raised slightly, but are still short relative to global norms and relative to the levels required for fiscal sustainability. This issue is compounded by the incentives toward shorter working careers in the benefit formula. Despite achieving major fiscal savings, the reform was insufficient to cope with the changing demographics of Turkey. The system will continue to receive important fiscal subsidies due to the gradual transition awarded to current contributors. 157. The success of the reforms depends critically on improvements in administration as the wages earned and contributions paid will need to be recorded for the workers' entire work histories in SSK and BK. Furthermore, increased efforts to enforce compliance will be required to promote the legitimacy and fairness of the pension funds. Efforts to establish a voluntary pension system need to be integrated with the existing pension funds. The voluntary pension system needs to be transparent, flexible, and attractive to workers, with a view toward eventually providing the basis for a funded component. These efforts reflect the government's commitment to continue improving the pension system. 158. The scope of the pension reform will need to be broadened over the longer term in order to guarantee adequate income security for future generations. The achievement of these policy objectives requires a long-term perspective. Small steps can extend the financial viability of the system for some years. But continued minor reforms weaken the system's credibility, particularly among younger cohorts who repeatedly observe the government devaluing pension promises. Because pension reform leads to a redistribution of resources, it is difficult and requires strong political support. However, further reform is necessary. Society at large will "' See Meyer 1990 for an analysis of the impact of unemployment insurance on the duration of unemployment using data from the US over the period (1978-1993). 51 benefit from greater macroeconomic stability, higher economic growth, and improved labor market incentives. The gains to society at large are compelling motives for pursuing further reform. 159. A sounder mandatory system can be built on the basis of the existing Defined-Benefit PAYG scheme, but alternative policy models for mandated retirement income can also be explored. An example is a rationalization of the PAYG system accompanied by the establishment of a mandatory funded pillar serviced by the private sector. This model allows a diversification of risks between the public and private sector pillars. Other OECD and non- OECD countries (including middle-income countries) have already implemented funded pillars reforms with positive results. In addition, the establishment of a mandatory private pillar could improve the efficiency of national savings. 160. The longer-term reform agenda will have to address a series of social and economic issues. First, to attain medium-term viability and greater longer-term sustainability, the reform will need to revise the targeted level (i.e., income replacement rate) and the targeted number of beneficiaries (i.e., minimum retirement age). Second, Turkey's pension system has not addressed the needs of low income workers who are covered by the system but might not be able to accumulate enough savings to provide them with adequate income during old-age. The reform should examine whether a minimum pension level should be guaranteed and, if so, determine its financing source -social security contributions or general tax revenues. Third, the fragmentation of the system by economic sectors creates pressures for reinstating privileges to special economic groups and is less suitable to a more integrated labor market in the future. Moreover, it will impair redistribution of resources from lower-income workers to higher- income workers within the system. Fragmentation across economic sectors also affects redistribution within the health insurance scheme. 161. Health sector reform is a large undertaking that encompasses the overall public health system. The complex issues of health care financing and provision, and the need to improve the health status of the population, particularly the poor, will require detailed sector studies and careful sequencing of reform steps. Nonetheless, the three health insurance schemes under the umbrella of the social security system share the same source of financing, contribution collection, and administrative system with the respective pension schemes. Any change in the pension reform will therefore directly or indirectly affect health insurance. Health insurance reform should be pursued along with pension reform. 162. In the long term, health sector reform should seek to provide all Turkish citizens with access to quality health services on equitable terms and promote a more efficient spending of public resources. Presently, the system is organized around social groups rather than functions. Longer-term development should move toward integrating these functions horizontally and restructuring the respective roles of the MOH, the social security system, and the private sector. 163. The new unemployment insurance scheme constitutes the third tier of Turkey's social insurance system. Its policy framework should be revised to prevent the social insurance system from placing additional burdens on fiscal resources and to improve the balance between worker protection and moral hazard incentives. With the introduction of unemployment insurance, mandated severance payments are no longer justified. Accrued rights under the severance 52 payment system could be preserved and act as a bridge until the new unemployment insurance scheme becomes fully operative. In the future, private firms and workers should be allowed freely to negotiate severance payments within the overall compensation package to the worker. 53 CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS42 I INTRODUCTION 164. Infrastructure is a cornerstone of Turkey's future growth and fiscal stability. The demand for infrastructure investment (an estimated US$3 billion per year for the electricity sector alone) risks outpacing supply, thereby creating bottlenecks to growth. In an effort to attract private participation and investment, Turkey pioneered the build-operate-transfer (BOT) model and made several attempts to launch privatization in the key infrastructure sectors with the objectives of keeping pace with technological change, shifting the burden of investment to the private sector, and reducing the public debt. However, insufficient attention was paid to the legal and regulatory framework and the need to promote competition. State monopolies have continued to dominate energy and telecommunications, while private investors have hesitated to commit to Turkey without government guarantees. As a result, a paradoxical situation has developed in which the state's contingent liabilities under take-or-pay contracts have grown rapidly on the one hand, while insufficient private investment has led to electricity shortages on the other. The government has decided to tackle these problems with an ambitious package of reforms focused on the energy and telecommunications sectors. The program combines up-front measures to establish modern, transparent regulatory environments with action to restructure state monopolies, introduce market forces and accelerate privatization. This is a difficult and complex agenda which will take several years to complete. Once the requisite structural reforms are in place, the policy emphasis will shift over the longer term to further liberalization of the energy and telecommunications sectors in the context of Turkey's WTO commitments and EU accession process. 165. Turkey was one of the first countries to champion involving the private sector in the provision of utility and infrastructure services in the mid- I 980s, but the overall results have been disappointing so far. An incomplete legal framework, lack of independent regulators, and limited use of competitive bidding have lowered private investment and generated high costs for the budget through long-term government guarantees for take-or-pay contracts and forgone privatization revenues. The lack of transparency and competition have encouraged those private companies that have invested to demand a high rate of return. TEAS, the state-owned electricity generation and transmission company, is facing major losses in large part because of the high cost of power purchased from independent producers. In telecommunications, there are now two thriving private mobile operators, and the govermnent is finalizing the sale of a third GSM license for US$2.5 billion (excluding VAT) to an international consortium led by Telecom Italia. Turk Telekom will also be awarded a GSM license on the same terms and conditions. However, privatization of Turk Telekom has been repeatedly delayed and progress in attracting private investors to value-added services has been limited. The gas sector remains largely a public monopoly with government ownership of transmission system (BOTAS) and two of the five distribution systems. The other three gas distribution systems are municipally owned. 166. A large part of the problem has been the lack of a comprehensive, well articulated reform strategy. Government policy fosters efficient operation of infrastructure and utility services when 42 This chapter draws on a background paper by Ian Alexander et al. 1999 and from input from the Bank team working on infrastructure sector reform in Turkey. 54 it is based on a set of benchmark policies, namely (i) privatization: the transfer of commercial risks to the private sector (referred to as economic privatization) rather than the simple transfer of ownership (title privatization); (ii) competition (structural regulation): the creation of markets where competition is possible; and (iii) conduct regulation: regulation covering areas such as pricing and customer service for those parts of the infrastructure and utility industries that remain monopolistic. International experience has demonstrated the importance of a proper sequencing of infrastructure reforms. A first priority is to establish an adequate regulatory environment for private sector participation (conduct regulation). Equally important is to restructure public monopolies before they are privatized in order to protect competition (structural regulation). These structural measures can then set the stage for full economic privatization involving the transfer of commercial risks as well as ownership. 167. The recent economic reform program has begun to address some of the key aspects of infrastructure reform in a more systematic manner. In August 1999, Parliament approved a long- awaited set of constitutional amendments that (i) create a constitutional basis for privatization and allow the government to determine by law which investments and services carried out by public entities can be contracted out or transferred to private agents; (ii) provide for international arbitration to settle disputes arising from concession contracts and other legal agreements concerning public services that involve a foreign element, i.e., a foreign investor with either a direct or a financial interest; and (iii) limit the Council of State (Danistay)'s function to an advisory role with respect to concession contracts and other legal agreements concerning public services. 168. On the basis of the constitutional amendments, Parliament in December 1999 approved changes to the BOT Law that (a) include electricity generation, transmission, distribution, and trading operations within the scope of the law, and (b) subject BOT projects-including those in the electricity sector-to commercial law. Parliament also approved in December amendments to the Danistay law to reflect its new advisory role and set time limits on its decision-making process. Further legislation was passed in January to provide for international arbitration for concession contracts and to give the government the authority to extend the right to international arbitration retroactively to existing contracts on a case-by-case basis. In parallel, telecommunications reform has gained momentum. Legislation to reform the sector was approved by Parliament in January 2000. Comprising a set of amendments to several existing laws, the new legislation: (i) enables the conversion of Turk Telekom into a joint-stock company subject to the commercial code, clearing the way for opening its capital to private sector participation; (ii) establishes an independent regulatory authority for telecommunications; and (iii) authorizes private provision of all value-added and wireless services. 169. The government must now build on this early momentum to complete its structural reform agenda in the infrastructure sectors and pave the way for full liberalization over the longer term. This chapter outlines the structure and problems of Turkey's energy and tele- communications sectors. It summarizes what has happened in terms of reform to date and what the government has set out to accomplish over the medium term. The chapter also identifies issues and potential obstacles that the government will face in implementing its reform program including market structure, privatization, competition, pricing and regulatory framework. It concludes with a set of recommendations. 55 2 ELECTRICITY 170. Turkey presently has an installed power generation capacity of about 26,100 MW (40 percent hydro and 60 percent thermal) and an extensive transmission and distribution network that provides access to electricity to almost 100 percent of the population. Electricity consumption has been growing at an annual average of 9 percent over the last decade, and the power system is capable of meeting peak power demand of about 19,000 MW. The demand for electricity is forecast to grow at an annual average of 8 percent over the next ten years, as a consequence of economic growth and the low level of per capita electricity consumption in Turkey (1,840 kWh in 1999 compared to the approximately 5000-7000 kWh typical of Western Europe). This growth will require annual investment of about US$3 billion in generation, transmission, and distribution. This level of investment cannot be mobilized by the public sector given Turkey's present fiscal conditions. 171. Market structure. Turkey's energy sector, including electricity, has traditionally been state owned, with the government influencing the management and decisions of the SOEs. As a result of reforms since 1984, Turkey's electricity industry now has many players, but at the operating level the sector is still dominated by SOEs. The two largest firms are TEAS and TEDAS, the state-owned distribution company. There are now many other privately owned firns in generation. These companies have entered the industry through BOT or auto-generator schemes. They currently account for about 21 percent of electricity generation. More recently, five large build-operate-own (BOO) contracts were competitively bid and this additional generation capacity (amounting to nearly 6,000 kilowatts) is expected to come fully on line by 2003. In many of these cases, the extent of risk transfer, i.e., economic privatization, has been limited. In a typical BOT or BOO generation project, the private owners bear construction and operating cost risks. However, the private operator signs a Table 3.1: Comparative Energy Data long-term power purchase Generatio T&D Iosses Consumption Total agreement with the state- Capacity as % of per capita natural ga owned TEAS in which the outpUt kwh consumption latter commits itself to buy the 3 Mc output of the plant for a period Turkey 21 16 1,364 0.35 Germany 110 5 6,426 3.30 of, say, 20 years at a fixed France 113 6 6,992 1.37 price in foreign currency, and UK 69.7 7 5,704 3.09 often also indirectly Italy 613 7 4,739 2.00 guarantees fuel supplies. This Mexico 37.63 14 1,3052 1.20 contract, backed by a / 1998, 2/ 1995, 3/ 1997 government guarantee from Source: EIA (USDOE), OECD. the Treasury, typically assures the private investors that the project will be profitable, irrespective of the future demand for its power, the price at which that power can be sold, the exchange rate, or even fuel supplies. As a result, the government retains most of the commercial risks that in other countries and other industries in Turkey are borne by the private sector4". `1 The broader implications of this retention of risk, apart from the possible losses faced by TEAS, are discussed in the background paper 56 172. There are four private distribution companies. Two private distribution companies were created as a result of the 1984 reforms: AKTAS serves the Asian side of Istanbul and the KAYSERI company serves the province of the same name. In addition, two distribution companies, CEAS and Kepez, have been private for many years. The distribution system currently served by TEDAS has been divided into 25 regions. Long-term transfer of operating rights contracts (TOORs) have been awarded to private investors for 8 thermal power plants and 14 distribution regions. 173. Technical performance. Considerable problems have plagued operations of the existing power infrastructure. In generation, the plant availability factor has remained low, owing to insufficient water for the hydro plants and operating problems with the thermal plants. The losses in the transmission and distribution network have risen over time to about 22 percent of generation. Transmission losses are about 3 percent and distribution losses rose to about 19 percent in 1998 from around 16 percent earlier in the decade. These losses, however, vary greatly and in some regions are reported to be higher than 60 percent. While better than that of many developing countries, the operating performance of the system leaves considerable room for improvement (table 3.1). 174. Financial performance. TEDAS, the government-owned distribution company, is profitable. On the other hand, TEAS reported accounting losses in 1998 and 1999 of US$94 million and US$145 million, respectively. If the electricity tariff for TEAS is not increased and no other actions are Figure 3.1: Indicative Projected Losses for TEAS without Reform taken, these losses will continue to grow in the -1000 near future. The piecemeal approach 60 - 9 that has been followed I , so far in the energy 50 ,X sector has pushed 60 TEAS into losses and also created r30I / snowballing contingent l* / liabilities for the 2- 4 0 300 budget. As more BOT - , - 2000 projects come on line, 10 - -- i10 TEAS is buying an I w-o--- -= - 0 increasing amount of 19 _ 1922 - 2 _ __ -M XX 5 2010 power from private generators. The prices - _ _ it pays to private _ _ _ dtca_ : --- - TdLsT( I __ _ generators include an Source: World Bank projections. amount to cover operating and capital costs, as well as a premium due to the lack of competition in the generation market. On average, these purchase prices are well above the price at which TEAS is authorized to sell power to TEDAS. Thus the shift to private power producers tends to bring to light the problem of uncompetitive generation costs, as well as the problem of low retail tariffs which do not fully cover costs. A comparison of the prices paid by TEAS to private generators with the 57 prices it receives from TEDAS makes the problem clear. TEAS is currently selling electricity for about 3.8 US cents per kWh including transmission (end-1999), but is paying private generators prices in the range of 6-9 US cents per kWh. Without real increases in the price TEAS charges and the introduction of market competition, together with complementary steps to cut costs and improve efficiency with TEAS itself, the company's financial position is likely to deteriorate significantly over the medium term. Figure 3.1 is based on stylized assumptions but serves to reinforce the point that without real price increases and other reform action, TEAS will face mounting costs and consequent losses (in the figure these losses are shown as almost US$1.5 billion per year by 2010, with a cumulative value of over US$8 billion)44. The potential magnitude of the problem is large. 175. Government policies and proposals. To address the serious problems accumulating in the electricity sector, the government is moving to introduce a new market model that will (i) transfer most of the task of supplying and distributing electricity and the associated market risks to the private sector; (ii) eliminate the need for additional state-guaranteed power purchase agreements; (iii) minimize costs through competitive pressure on producers and distributors; and (iv) ensure consistency with recent developments in the electricity market worldwide, in particular in the European Union. Under the so-called bilateral model, the government will largely withdraw from the electricity generation and distribution businesses45. Electricity generation companies will sign contracts for power directly with distribution companies without government guarantees. The government's future role will be largely confined to determining sector policy, owning the transmission system, and setting up an independent regulatory body to make sure that the rules are respected and that prices are competitively determined. 176. The government is committed to honoring its obligations under existing guarantees and these would be transferred to a new state trading company (see below). The bilateral market model is fully compatible with the energy directives of the European Union (box 3. 1), and has been successfully introduced in Latin America. As long as it is implemented correctly, the market model will provide an appropriate framework for meeting Turkey's electricity needs. Very recently some foreign and Turkish companies have indicated to the government that they are willing to make major investments in new power plants without government guarantees even before the new bilateral model is in place. These investments would use the existing scheme for auto-generation. 177. Preparations for the introduction of the new electricity market are well underway. A directive has been issued splitting TEAS into a transmission company (which will remain state owned), a generation company (much of which will be privatized), and a trading company to assume the existing long-term supply contracts that TEAS has with the private power companies. A financial recovery plan for TEAS has been adopted by the government which aims to restore the company to financial health by the end of the year. An electricity markets law is under preparation which will establish the bilateral model, create a regulatory authority, and allow the full privatization of state-owned electricity generation and distribution assets. The new law is expected to be enacted before the end of 2000 and the new market rules would be operational by " Other plausible scenarios leading to even greater losses are described and documented in annex III and in the background paper. Most of the currently installed hydropower will remain with the government since these plants are multi-use (power generation, irrigation and flood control) and often considered strategic. 58 2002. The government's aim is for new power generation coming on line by 2005 to operate solely on the basis of market contracts. 178. Privatization in the electricity sector. As part of its program, the government is looking to accelerate the pace of privatization in the electricity sector and place new emphasis on economic privatization by selling assets outright. Over the past three years, the government awarded a number of transfer of operating rights contracts for state-owned thermal generation and distribution assets to private investors. These TOOR contracts are similar to the BOTs in that the degree of economic privatization is severely limited with most financial risks remaining with the government. T o date only a few of these TOOR contracts have been finalized because the private investors have not been able to make the required up-front payments. A number of investors holding TOOR contracts have applied for international arbitration and these applications are under consideration by the government. In the event, the prospects of financing most of the TOOR contracts remain uncertain. The government has stated that those TOOR contracts which do not reach financial closure by the end of the year will be allowed to lapse46. After they lapse, the government is expected to initiate a new privatization approach based on outright sale of the underlying generation and distribution assets. Box 3.1: EU Directive for Electricity From February 19, 1999, customers equivalent to about 30 percent of demand in each EU country have been allowed to choose their electricity suppliers. The EU Directive originally was passed by the EU Council of Ministers in 1996. In fact, a much larger percentage of customers (as much as 60 percent of demand) now is able to choose its suppliers as a result of the fully opened markets in the U.K., Finland, Sweden, Germany, and Spain. The EU Directive ordered a minimum 30 percent of each domestic market to be opened up to competition by February 19- the customers within this top 30 percent being labeled "eligible" customers. Those countries that have not accelerated the opening of their market as per the directive, such as France, Austria, and Italy, will open up progressively until 33 percent of total electricity demand is in the competitive arena. Also, the directive requires disaggregated accounting for each of the functions of the industry (generation. supply, distribution, and transmission). It also outlaws cross-subsidization between the different functions. Importantly, the directive insists on non-discriminatory access to networks for all new entrants and existing suppliers, allowing them to supply newly acquired end-customers. Key provisions within the directive can be summarized as follows: P*rogressive supply liberalization in the EU market, culminating in 33 perment of electricity demand being open to competition by 2003; * Immediate non-discriminatory access to networks for all players, * An impartial licensing procedure for all new generation plants. Source: Standard & Poors Credit Week February 23, 2000. 179. Industry structure-introducing competition. A fundamental objective of the electricity reform is to unlock competition. Turkey already makes use of limited competitionfor the market in electricity. Private generators competed to win the BOO contracts, encouraging them to offer the best terms possible (most of the BOT contracts were not competitively bid). However, Turkey has not yet made much use of competition in the electricity market. That is, electricity '6 The government is considering the possibility of limited extensions of this deadline on a case-by-case basis where administrative delays are attributable to the authorities. 59 purchasers, including distribution companies, generally do not yet have a free choice of supplier; rather they must buy from TEAS or TEDAS (depending on the voltage at which they take power). Similarly, generation companies do not have a choice of buyer; they generally must sell to TEAS. The proposed new market model will change this situation by allowing distribution companies to buy from any generator and generators to sell to any distribution company. The model should generate substantial competition among generators. Further, as envisaged in the draft law, competition for final consumers will be introduced in a phased manner in line with the EU directive. 180. Competitive markets elsewhere in the world. Many countries in the world have now introduced some form of competitive market in which at least large purchasers have a choice of supplier. In several countries, even households can choose which company to buy power from. In Europe, the United Kingdom and the Scandinavian countries were the first to develop competitive markets. More recently, the European Union agreed that all its member countries had to phase in competition. Elsewhere in the OECD, competitive markets have been introduced in the United States, Canada, and Australia. Most developing countries have not yet introduced competitive markets, but there are several exceptions. In South America, Chile and Argentina were the first to create markets, and others in the region have since followed. Now Southeast Asian countries such as the Philippines have quite advanced plans to introduce competitive power markets. The effectiveness of competitive markets in power depends in part on whether generators will compete vigorously with each other or whether some may have significant market power. This in turn depends on factors such as the number of independent generators and wholesale buyers in the market. In Turkey, the size of the system (26,100 MW and growing rapidly) and the number of potential generators and distribution companies suggest a very substantial potential for competition to develop. 181. Keeping competitive options open. Some of the choices the government makes now will have a critical effect on the ease, and cost, with which competition can be introduced. The distribution TOORs are a case in point. The TOOR contracts for distribution, TOOR (D), give the private investors the exclusive right to distribute power in their franchise areas for 30 years- an issue questioned by the Competition Board. In principle, therefore, they preclude retail competition-that is, choice by final power customers. The government is considering including incentives in the electricity market law to encourage the TOOR(D) investors to accept competition within their franchise areas. However, it remains unclear how effective such provisions could be in practice. To date, none of the TOOR(D) contracts has been finalized. Allowing the TOOR(D) contracts which do not reach financial closure by the end-2000 deadline to lapse will leave the way clear for introducing phased retail competition through the option of outright sale of the distribution companies concerned in accordance with the government's strategy and the relevant EU directive. 182. Conduct regulation. The government intends to establish an independent regulatory agency for the electricity sector. Similar to the Competition Board and existing agencies for the banking and telecommunications sectors, the electricity regulatory agency will operate autonomously with its own budget. The agency is expected to have full authority over tariff policy and a clear mandate to ensure competitive practices and a level playing field in the electricity sector. The agency is expected to be created through the new electricity markets law which will also codify its authority. Establishment of an effective and truly independent 60 regulator will set the cornerstone for the future electricity market in Turkey. The need for a complementary authority for the gas sector has been accepted by the government, although this could be combined with the electricity agency into an broader energy regulator, if not immediately, then possibly over the medium term. 183. Managing the transition. Managing the transition to the market model will require careful attention. Due to delays in the previous program (BOOs and especially BOTs) a power shortage (about 2 percent of demand) is possible in 20001 and a somewhat larger shortage (3 percent of demand) in 2002. These shortages may not occur, since hydropower conditions may be better than assumed, demand may not be as strong as forecasted, and some ongoing generation projects may be completed earlier than expected. The government is considering providing a limited number of guarantees for generation projects that can be in operation (i.e., commissioned) by 2002 in order to avoid the potential electricity shortages. Parallel actions to adjust tariffs, improve payment discipline, and reduce technical losses and theft of electricity will also help close the potential short-term demand/supply gap. The authorities project that, based on existing projects under construction and additional projects for which government guarantees have already been issued, supply will be sufficient to meet market demand in the period 2003- 2004 (table 3.2) and no further take-or-pay guarantees are expected for projects coming on stream after 2002. Table 3.2: Electricity Demand and Supply Projections (MWh) 199 2000 2001 2002 2003 2004 Electricity Demand 117,588 126,800 138,810 151,430 165,200 180,225 Electricity Supply 117,588 126,776 136,441 147,213 197,435 215,666 1. Current Generation Capacity 115,601 124,171 127,559 127,559 127,559 127,559 2. Projects Under Construction 0 0 4,332 8,448 16,760 25,220 3. New BO and BOT Projects 1/ 0 0 450 5,806 47,716 57,487 4. Net Import 1,986 2,605 4,100 5,400 5,400 5,400 Electricity Gap 0 -24 -2,369 -4,217 32,235 35,441 Memo item: Demand growth rate 7.20 7.83 9.47 9.09 9.09 9.10 Source: Government. 1/ Not under construction yet. 3 GAS 184. The gas sector in Turkey, while significantly smaller than the electricity industry, is also growing fast. Reasons for this growth are (i) the growing role of gas as a fuel for electricity generation, (ii) expanding domestic and industrial gas usage, and (iii) the potential for gas transit across Turkey from Central Asia to Europe. These and other factors make focusing on the reform of the gas sector important, not only for supporting the liberalization of the electricity industry but also because gas is a rapidly growing source for domestic energy consumption. 185. Market structure. The current market structure is based on monopoly provision and consequently (i) BOTAS, a 100 percent state-owned enterprise, has a monopoly on import of gas--both natural gas through pipelines and liquefied natural gas, or LNG, through a terminal close to Istanbul; (ii) BOTAS is the only transmission company; and (iii) distribution and supply 61 is provided by local companies that are owned either by the municipality (as is the case in Istanbul, Ankara, and Izmit) or by BOTAS (as is the case in Bursa and Eskisehir). Gas is imported on the basis of long-term take-or-pay contracts, with BOTAS entering into a series of major new contracts due to the expected significant increase in gas usage, especially for electricity generation. Although Russian gas dominates at the moment, LNG from Algeria and Nigeria currently are supplied and gas from Iran is due to start flowing into Turkey next year. Thanks to its monopoly position, BOTAS has been a significant profit maker and had a sales margin of over 30 percent and a return on assets of just over 32 percent in 1998. 186. Government policies andproposals. The government's plan is to establish a competitive market and to encourage private sector participation through a phased policy. This reform plan will be anchored by a set of amendments to the Petroleum Law to create the framework for a liberalized, competitive gas market. The plans encompass (i) splitting BOTAS into four separate companies, including a gas/oil transmission company and a gas trading company that will take responsibility for all existing long-term gas purchase/import and sale contracts and two gas distribution companies (see item v); (ii) removal of BOTAS's monopoly on the import of gas; (iii) establishment of an independent regulatory body; (iv) actions to encourage private sector participation in all aspects of the gas market (including the possible privatization of the supply role), and (v) privatization of two state-owned gas distribution companies in two provinces, namely in Bursa and Eskisehir. These actions are planned to be undertaken in 2000-01. Draft amendments to the Petroleum Law have been prepared and are expected to be submitted to Parliament in the coming months. 187. Competitive markets elsewhere. Gas liberalization has, to some extent, lagged behind electricity liberalization. One of the primary reasons for this is that countries are all too often dependent on just one or two sources for gas supplies (in the case of Turkey this source was traditionally Russia), which often resulted in the inability of wholesalers to differentiate themselves in any meaningful way. However, several markets have liberalized, including Australia, New Zealand, the US, the UK, and Argentina and Mexico. Consideration is also being given to this subject in many other countries. The EU Gas Directive will lead to the introduction of competition, as will the plans being discussed in Thailand. 188. Keeping competitive options open. There are several competitive options for the gas market, and these are not dissimilar to the types of competitive market seen in electricity. Within the gas industry the two elements that are closest to natural monopolies are those involved with the pipelines, i.e., transmission and distribution. Other aspects, such as production, wholesale and retail can all be competitively provided. As such, the models of possible industry structure include, (i) a wholesale competitive market with regional monopolies providing retail services, and (ii) both wholesale and retail markets that are competitive. The latter approach is- the preferred option. However, where markets still need to be developed, it may be necessary to adopt the former model although any long-term exclusivity should be limited to the distribution function (not retail). A phased and clearly time-bound opening of gas retail services to competition might be a suitable transition model. "A good summary of the options for the design of the gas market is provided in The Emergence of Markets in the Natural Gas Industry. Andrej Juris. Policy Research Working Paper number 1895. The World Bank. 62 189. Three issues need to be considered when thinking about the emergence of competitive markets in the gas industry. The first is a range of suppliers. It is necessary to have a range of suppliers who are willing to provide wholesale gas at a range of prices and quality (here quality may reflect the nature of the supply, for example, an interruptible contract or discounts for bulk purchasers). Reliance on a single supplier in the majority of cases will not allow the development of competitive markets. The second issue is take-or-pay contracts. Few, if any, gas markets have developed without the use of these contracts which provide the supplier with a degree of security with respect to the demand for its product since it generally must invest a significant amount in large "sunk" pipeline projects. However, as markets develop, the need for take-or-pay contracts starts to diminish, especially if some of the liability can be linked to a take-or-pay contract for major consumers (e.g., glass or fertilizer factories that are dependent on the gas supply and therefore able to commit, or an independent power producer that has some form of power purchase agreement), thus allowing more of the output to be sold through competitive markets. The third issue is bypass, when a company is allowed to build its own links to a customer, bypassing the existing distribution network. 190. Conduct regulation. The government is planning to establish an independent regulator for the gas industry. This authority will determine transmission and distribution access rules and tariffs and the methodology for the regulation of retail prices. Turkey does, however, have the option of considering a single eneygy regulatory body. Given the key linkages between the gas and electricity industries (both in terms of inter-fuel competition and gas's position as a major fuel source for generation), a combined energy regulator would appear to yield significant benefits. The potential economies of scale and scope of merging the gas and electricity regulators should not be underestimated". There has been discussion of moving the Directorate of Petroleum Affairs out of the Ministry of Energy and Natural Resources to form the basis of the gas regulator. While this may be a practical solution to the establishment of a regulatory body, it is one that should be handled carefully since it will be essential to establish a clear separation, both de jure and de facto, between the ministry and the independent economic regulator. 191. In conclusion, Turkey is in a good position to establish a competitive gas market, particularly given the expected diversification of supply and the relatively limited size of the sector at present. The country is at a crossroads with respect to the gas industry. Significant growth in the sector can be expected over the next decade. Rapid action to introduce a competitive market will facilitate this growth by attracting more private investment, while helping to ensure that the gas sector develops in a cost effective and efficient manner. Maintaining the current monopolistic structure will likely lead to a replication of the problems facing the electricity industry (contingent liabilities, high prices, shortage of investment, etc.). Reform of the gas sector is also critically important for successful development of competition in the electricity sector given the growing role of gas as a fuel source for electricity generation. Therefore the government should actively seek to introduce gas sector reforms as fast and as deeply as possible. 4 Information on the international choice of single industry or single sector regulatory bodies is provided in the background paper. 63 4 TELECOMMUNICATIONS 192. Restructuring Turkey's telecommunications sector is at the heart of the reform process. The sector is expected to attract significant international attention over the coming years owing to its large size, up-to-date infrastructure technology, and respectable quality of services. The reforn aims to increase private sector participation, promote competition and prepare for the liberalization of fixed line services after 2003 in line with Turkey's international commitments. The immediate Table 3.3: Telecommunications reform priorities are (i) to modernize the legal and Comparators regulatory framework, and (ii) increase private sector Telephone Mobile participation and generate cash for the budget by opening main lines telepbones the capital of Turk Telekom and selling additional mobile per 1,100 per 1,000 licenses. The longer term objective is to increase the people' people' availability of high-quality telecommunications services Turkey 224 13 to every segment of society at a reasonable cost and Germany 564 42 ensure that Turkey benefits fully from the new global UK 528 122 information economy. Italy 440 112 Mexico 95 11 193. Industry structure. The telecommunications '/ 1996 industry is currently dominated by Turk Telekom, a Source: World Development Report national monopoly that has exclusive rights to all fixed 1998/99. line voice operations. Turk Telekom established the first mobile company (analogue) and local cable television operations, and is responsible for Turkey's radio and television transmitters. Recent reforms since the early 1990s have led to the introduction of several private sector players: (i) two GSM 900 mobile telephone companies established initially under 15 year revenue-sharing agreements subsequently converted into 25 year license agreements (including 15 percent revenue sharing with the Treasury), and (ii) a series of companies providing value-added services such as internet access and cable television under revenue-sharing agreements. Currently, the government is proceeding with the sale of three GSM 1800 licenses including one to Turk Telekom. Table 3.3 provides information on international telecommunications comparators. 194. Financial performance and pricing issues. Turk Telekom is currently in a strong financial position, having made a pre-tax profit of over US$1 billion each year since 1997. These figures represent an approximately 50 percent margin on sales. This is a healthy result for a telecommunications company although Turkish accounting rules may overstate the results relative to international accounting standards. However, the results may hide a potentially problematic situation: tariffs are above cost for international and long-distance traffic, and below cost for local services. Further, there is no direct fixed line rental charge. The high international and long-distance charges are cross-subsidizing local services and generating the high profit margins for the company49. Turk Telekom is able to sustain these high prices through its monopoly. There has even been an attempt to outlaw international call-back services to further 9 The price difference between international calls and local services was reduced significantly in 1999. 64 defend the company's position"0. Without this dominant position Turk Telekom's financial performance might significantly less impressive. As long as tariffs are unbalanced market players are not making rational economic decisions since they are basing their decisions on false price signals. Consequently, there are macroeconomic and misallocation issues. 195. Government policies and proposals. Legislation to reform the telecommunications sector was approved by Parliament in January 2000. The new legislation, comprised of a set of amendments to several existing laws, initiates the process of deregulating the sector over the medium term. It sets out a broad market structure for the future of telecommunications in Turkey and defines a four year transition to this new market model, including establishing a regulatory authority and opening the capital of Turk Telekom to private participation. The new regulatory board has been appointed and the decree setting up the authority was published in August. While Turk Telekom will retain its monopoly on fixed line services through 2003, these services will be liberalized thereafter in line with Turkey's commitments under WTO. 196. Privatization. The legislation transforms Turk Telekom into a joint-stock company in order to open its capital to private participation. The authorities have prepared an action plan for implementing the conversion including a plan for the company's civil servants. The Table 3.4: Sale of Equity in Turk Telekom government has planned to open the Phase Action % of shares to be sold capital of Turk Telekom in stages, starting I Transfer to Postal company 10 with the sale of a 20 percent stake to a Sale to strategic investor 20 strategic investor. The strategic investor, if 2 Initial Public Offering 14 strategic ~~~~~~~~~~~New York 1 it is a consortium, must be at least 51 Turkey 5 percent owned by a reputable international 3 Remaining Government telephone operator. Subsequent stake, possibly to be sold at 51 transactions would include sale of 5 a later date percent of the shares to the company's Source: Privatization Agency. workforce and local investors and a further 14 percent of the shares to be sold through an international public offering (table 3.4). The tender for the block sale of 20 percent of Turk Telekom was announced by mid-June. However, given limited interest in the tender among potential strategic investors, this 20 percent share may be increased. 197. Opening the capital of Turk Telekom responds to several factors in the government's broader telecommunications strategy. First is the phased introduction of competition for fixed line services. There is a formal proposal to the WTO that fixed-line liberalization will occur by the end of 2003 and this timetable is confirmed by the new telecommunications legislation. The grace period from the opening the capital of Turk Telekom to the liberalization of the market is to allow Turk Telekom to further restructure and prepare adequately for full competition. The second factor, linked to the first, is a phased removal of the cross-subsidy supporting local ; Call back is legal in a large number of countries. Call back is just one way of by-passing the facilities of the incumbent . There will always be incentives to by-pass where there are expensive service charges that are substantially above costs. Call back requires the participation of an operator in another country to facilitate the service. It has been agreed by the membership of the ITU that where a particular country has made call back illegal, that country can enlist the assistance of other countries (effectively the regulator in that country) in enforcing its ban on call back by instructing the participating operator in that country to desist. This is frequcntly very difficult to enforce.' 65 services. Further, a line rental charge is to be introduced. This will mean that cost reflective tariffs and leased line access will be achieved and competition can occur on a truly level playing field. Third, mobile and value-added services are being made fully competitive by the shift from revenue sharing agreements to licenses. Fourth, Turk Telekom will be granted a GSM license and will keep its cable business, although both will be shifted into separate subsidiaries. 198. The tender process for two of the new GSM 1800 licenses was conducted during the first half of 2000. Negotiations on the sale of one of the licenses to a consortium including Telecom Italia and Is Bankasi for US$2.53 billion (before VAT) are at an advanced stage. The tender of the second license was unsuccessful since no bidder was willing to meet the reserve price set as the winning bid for the first license". A third GSM 1800 license will be sold to Turk Telekom at the reserve price of US$2.53 billion (excluding VAT). 199. Competition. Private sector involvement and the creation of competitive markets are not new to the Turkish telecommunications market. Limited competition already exists for mobile services. However, this competition is limited because (i) Turk Telecom has a monopoly on the key elements related to fixed-line connections, (ii) Turk Telekom also has a monopoly over the provision of international calls, (iii) prices for fixed-line services are such that local calls are cross-subsidized by long-distance and international calls52, and (iv) until 1998 all additional services were provided through revenue-sharing agreements with Turk Telekom so that the new entrants were "partners" of Turk Telekom as much as competitors. Moreover, Turk Telekom owns the only real fixed-line alternative to its own service: the fibre-optic based cable television system, Although other utility companies with their own distribution systems such as electricity and water may have the ability to offer alternative infrastructure, they are not authorized to enter the telecoms market. Turk Telekom also operates its own mobile (analogue) service and is expected to be granted a new GSM license later this year. Under these constraints, only very restricted competition can be expected until the monopoly on fixed line services is lifted at the end of 2003. Given the limited scope of the current plans to restructure Turk Telekom, competition is likely to be limited after 2003 as well. 200. Competitive markets elsewhere in the world. Experience throughout the world is that the telecommunications market, owing to the rapid pace of technical innovation, is the easiest of the utility and infrastructure markets in which to introduce competition. For example, the long- distance fixed wire business is now proving easily replicated in a competitive market. Electricity companies can string wires along their pylons at little cost and both gas and rail companies have their own telecommunications systems (for signaling, linking operations offices, etc). These wires, especially when they are upgraded to fibre-optic, provide greater capacity than the companies actually require, so that the surplus can be rented out to telecommunications companies. 201. In a recent study, Wallsten (1999) finds that structural reform is a key driver for improving performance in telecommunications. The study considers information from 30 countries in Africa and Latin America from 1984 to 1997. Using a range of physical indicators, s' The second place bidder for the first license had offered USS1.35 billion. S This tariff structure impacts on competition since any mobile competitor to Turk Telekom at the local level faces the problem of fixed-line calls being subsidized while the competitor is unable to earn the above cost fees from international calls, owing to Turk Telekom's monopoly, meaning that competition must focus on quality and availability rather than price. 66 such as main line penetration, telecom employees per main line, and the price of a three-minute local call, the study assesses the impact of competition, privatization, and conduct regulation. The findings of this research are that (i) competition (the extent of which is determined by structural regulation) has a tangible impact on performance; (ii) privatization by itself has a negative impact on performance, which reinforces the importance of the role of the regulator in enforcing transitional rules while effective competition is established; and (iii) privatization combined with the establishment of conduct regulation has a positive impact on performance. International best practice has shown that the development of certain, credible and independent regulatory processes are the foundation for competitive growth in the sector itself and the adjacent activities of the web economy.Iq 202. Keeping competitive options open. Turkey has several options for the establishment of effective competition: (i) establishing the mobile operators as real competitors by allowing them to be alternative infrastructure providers--this would give service providers a choice and place competitive pressure on Turk Telekom; (ii) splitting the operation of the fixed network from the provision of telephony services-preferably through the complete separation of the businesses- which ensures that the infrastructure provider has no incentive to favor one operator over another; (iii) separating the cable television operation from the main fixed-line system so that real competition at the local loop level can be provided by the local cable companies offering telephony services over their fiber-optic systems; (iv) removing the near unenforceable ban on call-back services as soon as possible so that effective external competition is created, (v) adopting international best practice for the regulatory environment, particularly with respect to interconnection and licensing; and (vi) developing a regulatory framework that would allow tariffs to be determined by the market. Some of these reforms would need to be phased with appropriate milestones so that the commitment to the creation of a fully competitive market is clearly shown. However, the government--particularly through its restructuring plans for Turk Telekom--appears to have precluded some of the options for rapidly expanding competition in the sector. 203. Conduct regulation. The first utility regulatory agency in Turkey was created with passage of the new telecommunications legislation. The Telecommunications Board has already been established, as required by the January legislation, and the five-person commission has been appointed. Full legal establishment of the regulatory body took place in August 2000 when the Telecommunications Authority was formally established by government decree. The principles of minimum international best practice for utility regulation are described in box 3.2. A legal and regulatory framework that enhances the security of investments in the context of liberalisation will generate new products, services, exports, and employment. While the January 2000 law took a significant step towards establishing these conditions, by itself it is not sufficient. Further legislation may be required to address some of these areas, but the detail will depend on the precise EU accession agreement reached between Turkey and the EU. The nascent regulator, already faced with a multi-operator environment, will have immediate short- term and longer term capacity building requirements, and its actions will need to be supported by appropriate implementing regulations. Areas that will need to be addressed include: * Greater transparency, including publication of criteria, especially of the licensing system; A recent EU Directive has addressed this issue. 67 * Clarification of the licensing system for value-added-services, including the determination of the need for notification, class licences, and individual licences; and * Enhanced structural separation to ensure fair and open competition, particularly for access to fibre-optic cable infrastructure and mobile access to the fixed network. Box 3.2: International Regulatory Principles and Capabilities The minimum international best practice is based on a set of elements found in the WTO Regulatory Reference Paper, which has been agreed to by more than 70 countries. These elements are: * An independent regulatory authority * Transparency in general * Competitive safeguards of the type that would flow from a competition or anti-monopoly law * Objective, non-discriminatory, and transparent allocation of scarce resources (e.g., frequencies, numbers, rights of way) * Public availability of licensing critenra * An equitable interconnection regime. WTO requirements should be seen as a minimum stepping stone for Turkey. The EU Directives on telecommunications set a higher hurdle for the regulatory system. These principles, however, are not in themselves sufficient. The principles have to be implemented effectively and the appropriate regulatory bodies require a set of capabilities including the following. ^ A modem licensing procedure and the means of revoking or modifying licenses ^ Procedures to monitor and enforce timely compliance with legislative and license provisions * A detailed, timely, effective, and practical means of resolving network interconnection issues, including technical, capacity, compatibility, routing, price, billing, and other related matters * A capability and methodology to assess the costs of the provision of telecommunications services provided to the public and to other operators for tariff approval purposes * A radio spectrum (frequency) monitoring and management capability, an allocation plan, a radio authorisation policy for public and private networks, and a radio authorisation process ^ A telephone numbering plan and procedures for the allocation of blocks of numbers * A capability for specifying standards for different classes of telecommunications equipment and for testing compliance with these standards d Appropriately defined Universal Service Obligations and costing of such obligp,ations ^ Procedures to monitor service quality and resolve disputes related to quality between operators and between operators and customers * Criteria whereby the tariffs for particular services gain "unregulated" status. Source: World Bank. 204. In conclusion, creating a competitive telecommunications industry is a strategic objective for Turkey. The government has taken a number of key decisions in the design and early implementation of its telecommunications reforn program, while others are still under preparation. The central lessons from international experience deserve careful consideration as the reform program moves ahead. With respect to structural issues, the mains lessons are: (i) competition is the only way to ensure efficient operation and sufficient technological innovation to keep up with the pace of global change; (ii) clear and credible regulation, especially of issues such as access to fixed networks, is vital; and (iii) strong measures to restructure the incumbent monopoly and restrict its market power, while often difficult, are necessary for real competition and effective conduct regulation. Turkey is starting to build a base for effective competition 68 through independent mobile operators and the new telecommunications legislation has set the framework for modem conduct regulation. There is a clear and urgent need for the process of tariff re-balancing to be accelerated. Looking forward, the government should capitalize on this new momentum, make the tough restructuring decisions required to secure future competition, and carry through with its privatization efforts. It is essential that Turk Telekom's GSM license be held by a legally separate subsidiary to ensure transparency and a level playing field. Moreover, the cable television operations ideally should be shifted out of Turk Telekom, but if this is not feasible, these operations should be placed again into a legally separate subsidiary. Third, the exclusivity period for fixed line services should be kept as short as possible. Finally, the Telecommunications Authority should become fully operational as soon as possible. 5 RECOMMENDATIONS 205. The government is making impressive progress in reforming Turkey's infrastructure sectors, notably energy and telecommunications. The private sector is already playing a significant role in both energy and telecommunications, notwithstanding considerable constitutional and other legal barriers, and this role will expand a great deal under the government's reform strategy. Government action is underway on a number of outstanding policy and financial issues-some of which stem from the short-term, partial solutions that were implemented in the past while constitutional concerns and other issues were being resolved"4. In the near term, the government must press on with a complex structural reform agenda in order to establish an efficient operating environment for the private sector while providing consumers with adequate protection. This will set the stage for Turkey to realize the full benefits from increased competition as the infrastructure sectors are progressively liberalized over the medium term. 206. The following considerations will help guide future reform actions: * Economic privatization. The privatization process needs to be strengthened so that privatization is economic rather than merely legal (i.e., with little or no de facto transfer of risks to the private sector); * Structural regulation. Policy reforms should strive for real competition where possible- ensuring that decisions are not taken that could be hard to reverse (e.g., granting 30 year exclusivity deals, failing to unbundle state monopolies, etc.). Where the introduction of competition is phased, the phases should be kept as short as possible and commitment to clear, timebound milestones established; and S In this context, the government will be faced with the particularly difficult issue of managing expensive government guaranteed agreements which are not competitive in the new market environment. Of course, the government is bound by Turkey's legal commitments and it is essential to maintain investor confidence. However, the financial costs to the budget of these agreements could be very large indeed. Options for managing such uncompetitive agreements which other countries have used successfully are discussed in the background paper (Ian Alexander et al. 1999). 69 Conduct regulation. Regulation is going to be especially important for those sectors in which either there is a transition to competition or real "head-to-head" competition is not possible. As such, establishing effective and credible regulatory bodies is essential. Although the government has made a good start, much remains to be done especially in bringing about consistency, ensuring operational effectiveness, and aligning Turkey with international best practice. 70 CHAPTER 4: AGRICULTURAL AND RURAL DEVELOPMENT 5 1 INTRODUCTION 207. Turkey's tremendous potential for rural growth has gone largely unrealized. This poor performnance has enormous implications for living standards given that agriculture still accounts for some 45 percent of total employment. It also has generated major fiscal costs as successive governments have pumped large subsidies into the sector in an effort to maintain rural incomes. Reliance on inefficient agriculture price support policies has kept food prices high, placing a heavy burden on low-and middle-income consumers. Past attempts at reform have been piecemeal and ineffective, and Turkey's international commitments under the WTO and Customs Union with the EU have had little impact on agriculture policies. The current government has taken a much more comprehensive approach to agriculture reform based on a two-track strategy. First, the anachronistic structure of subsidies for inputs, credit and price supports is being gradually eliminated, to be replaced with a system of non-distortionary direct income support for farmers. Second, the state is withdrawing from direct intervention through a combination of autonomy for the agriculture sales cooperative unions (ASCUs) and privatization of most state-owned enterprises in agriculture. Beyond the current reform program, scheduled to be completed by 2002, the longer-term policy agenda focuses on broader issues of rural development including a more rational approach to irrigation investment, reform of forestry and livestock policies, and support for off-farm income generation as the percentage of the workforce employed in agriculture continues to decline over time. 2 THE CHALLENGE OF GROWTH 208. While overall GNP grew at around 4 percent per year over the past decade, agriculture grew at about a third that rate. As a result, the sector shrank as a share of the economy from 36 percent to 15 percent. Agriculture's share of exports has fallen from around 60 percent to Figure 4.1: Agricultural Growth and Transfers about 10 percent. The agriculture growth rate has been trending downward (figure 4.1). a 16 There is some evidence that the falling 6 - t 4 1 agricultural growth is due to stagnant or even S 2 12 E declining technical efficiency in Turkish 3 C In agriculture as measured by total factor 2 V2 8 productivity from 1990 to 1996. This is a -6 X L 6 startling conclusion, considering the enormous t8 4 advances in agricultural technical efficiency in 198° 01 1 1 1996 2 the world at large 6. The sector still accounts 198B 1990 1992 1994 1996 1998 for a very large share of employment, though 1 Z Ag Suppor - Growth rate-Trend (Growth) this also has fallen considerably. Unlike in Source: SIS, Treasury and World Bank estimates. many other countries, the proportion of the rural population in Turkey that is properly classified as poor is about the same as this proportion in urban areas. Nonetheless, the fact that so much of the total population is agricultural means ss This chapter heavily draws on background papers by John Nash (1999), Erol Cakmak (1999) and Haluk Kasnakoglu (1999). 56 See H.Akder, H. Kasnakoglu and E. Cakmak" Turkey: Sources of Growth in Turkish Agriculture", April 2000. 71 that quite a large fraction (about 42 percent) of the poor population in Turkey is employed in agriculture, with 11 percent of agricultural workers classified as poor, a higher percentage than any other employment category, except construction. The importance of rural development is amplified by the fact that rural areas show lower levels of human development as measured by an index of education, literacy, and income. i 209. Development of rural areas and agriculture in particular has been impeded by extensive government intervention in the sector. Many government initiatives have been counterproductive. Trade controls, government procurement, heavy government involvement in marketing, input subsidies (especially credit and fertilizer), and large-scale investment in irrigation infrastructure on a largely subsidized basis have created a net inflow of resources from the government to agriculture, but have had many negative effects on the sector and the economy at large. They have discouraged production of products in which Turkey has a comparative advantage, squeezed out private sector marketers, and subsidized inefficient production technologies. The benefits of the subsidies have gone mainly to larger, wealthier farmers. In addition, some of the subsidies have been so large as to have serious macroeconomic effects. These policies also create barriers to Turkey's closer integration with the EU. While previous governnents have long recognized some of the problems created by inefficient agriculture policies, reform has proven difficult in the face of political opposition. 210. In late 1999, the Government announced a new reform program designed to promote more efficient resource use and agricultural productivity growth while reducing the heavy burden on the budget and consumers. This program comprises three elements. First, the introduction of a unified national program of direct income support will improve the access of smaller farmers to budget support and create new income generation opportunities by improving incentives. In parallel, the government intends to phase out the unsustainable and distortionary system of subsidies for fertilizer, credit and price supports--which disproportionately benefit large farmers and regressively tax consumers--and link prices to world market prices. Third, privatization of most state enterprises in agriculture--including the quasi-governmental sales cooperative unions-will reduce government involvement in the marketing and processing of agricultural products. 211. This chapter provides a historical perspective on the evolution of agricultural support policies and analyzes their adverse effects. It then describes the government's reform program that is now underway and concludes with a suggested longer-term agenda for agricultural policy. 3 OVERVIEW OF SUPPORT INSTRUMENTS 212. In the crops sector, government interventions have been primarily in the form of price supports, augmented by quantitative restrictions (QRs) on imports in the past and high tariffs recently. In the livestock sector, QRs and tariffs have been the main mechanism to support prices. Price controls and export taxes have been employed to protect consumers. Input subsidies and credit are provided to farmers to improve yields and income and to counterbalance the implicit protection given to domestic input industries through border measures. Production of a few important crops has been administratively controlled. 5, See "Turkey: Economic Reforms, Living Standards, and Social Welfare Study", gray cover report of World Bank, January 2000. 72 213. Output Price Support has been the most widely used instrument of agricultural policy in Turkey, starting in 1932 with wheat. It has always been at the center of policy discussions. Until the 1960's, agricultural support purchases were limited to 8-10 cereals, opium, tobacco and sugarbeet. In the 1960's and 1970s, this climbed to 22 crops. The number of products covered started to decline in the 1980s and in 1990 only 10 were covered. However, in 1991 coverage again began to increase, and reached 26 commodities-the most ever-- in 1992. This was significantly reduced in 1994, and since then formal support purchases have been limited to cereals, tobacco, tea and sugarbeet. Technically, tea has no official procurement price, but rather has a price supported above world market levels by trade policy. The volatility of the number of covered crops illustrates the political nature of the system and how much it has been driven by short-term considerations, rather than any coherent long-term strategy. 214. Support prices are announced by government decree each year. Related SOEs and agricultural sales cooperative unions (ASCUs) are commissioned to buy at the announced floor prices5". Crops can also be sold to independent buyers. For some crops, a system of "deficiency payments" or premia was introduced in place of floor prices, whereby a target price is announced together with a low (world price-based) intervention price. Farmers selling their crop to ASCUs or commodity exchanges receive the difference between the price obtained and the target price as a payment directly from Ziraat Bank, which is then reimbursed by the Treasury. This system has been used for cotton and in 2000 will apply to oilseeds as well. 215. Support prices set above international prices for these commodities require import restrictions or tariffs. The degree to which these policies distort domestic prices and impose fiscal costs depends on the support price vis a vis the world price and the price at which the SEE sells its purchases. When world prices are low or political pressure is high, there is little to constrain increases in procurement prices and tariff levels, since the tariff rate for wheat bound under WTO agreements is 180 percent. Support prices and tariff levels were significantly increased in 1997 and kept high in 1998 and 1999. Reform of this system is a key objective of the new economic program, and the government has already taken some steps in this direction (Section 4.6). 216. In addition to the formal support price system, procurement effectively financed by the government is carried out by the 16 Agricultural Sales Cooperative Unions (ASCUs). These organizations are nominally apex organizations of farmer-owned cooperatives, but in reality are controlled by the government through its appointment of management and its provision of finance from the price support and stabilization fund to cover their losses. Thus, the pricing and purchasing decisions of many of the ASCUs have been driven by a political imperative to provide high prices for growers, as well as employment in the marketing and processing sections of the ASCUs themselves, and sometimes in unrelated businesses owned by them. Turning these into true member-owned and operated organizations along the lines of cooperatives in other countries is another key objective of the government's new agricultural strategy, and beginning with the 2000 harvest, ASCUs will begin a restructuring process to make them financially and managerially autonomous. 5 After 1994, ASCUs did not formally have automatic responsibility to buy at official support prices. Nonetheless, they may request permission, and have continued to do so, running up large losses which must be funded from the government budget. 73 217. Trade policies have formed an important element of the support system, since it is impossible to administer support prices significantly different from border prices without some kind of trade policy intervention. After 1980, there were significant changes in the trade policies in the direction of elimination of licenses, state monopolies and reduction of duties. Turkey started implementing the World Trade Organization (WTO) agreement on agriculture in 1995. This was followed in 1996 by the formation of a Customs Union (CU) between Turkey and the European Union (EU). However, the CU agreement has very few concrete implications for the Turkish agricultural policy as of now, because agricultural products were exempted from the CU. An eventual full membership of Turkey in the EU would require the harmonization of agricultural policies with the EU. But the EU's own agricultural policy is evolving and its exact configuration cannot be anticipated at this stage. It is however clear that EU farm policies are reducing reliance on commodity price supports and moving towards direct income support. Grain support prices are now approaching world prices and are likely to approach closer in the future, under the policy scenario suggested in the Commission's Agenda 2000. 218. Turkey's tariff binding commitments under the WTO on agricultural products were registered at such high levels - 180 percent for wheat, 225 percent for bovine and sheep meat, 135 percent for sugar, etc. They will cause little impact on existing policies. Turkey's tariff bindings by 2004 will almost all be above the EU final bound levels and Turkey would have to conform to the lower EU levels under the future access process. A striking feature of the current applied tariffs on agricultural products is their level and dispersion. Considering that it is a middle income country where the weight of food in consumers spending is much higher than in industrial countries, Turkey currently (2000) applies extraordinarily high tariffs for food products. This high level of protection implies a substantial implicit tax on low and middle income consumers. 219. Except for technical standards and sanitary and phytosanitary controls, under the current trade regime Turkey does not apply formal quantitative restrictions (quotas, import permits and licenses) on agricultural imports. This is a very positive characteristic of trade policy, which is consistent with the Uruguay Round Agreement on agriculture and with the overall goals of a transparent and rule based trade regime. It is important to preserve this. However, in recent years, there have been some mis-uses of sanitary controls to restrict imports on an ad hoc basis, for example, a sudden application of zero tolerance for chemical residue to restrict wheat imports. These should be avoided in the future. 220. Turkey has undertaken export subsidies commitments with WTO for 44 agricultural products, although in recent years, the govemrnent provided subsidies for only a few. Sugar and wheat export subsidies in recent years are potential problems considering Turkey's zero export commitments for export subsidies. Under the WTO agreement, domestic support policies subject to commitments include price support, input subsidies, supply controls, direct payments. As all developing countries, Turkey is exempted from public expenditure reduction commitments for both the so-called "green box" policies (considered undistorting policies such as research, extension, enviromnental programs) and for rural development programs and some input subsidies. 74 221. On the other hand, Turkey has bound "de minimus " support for all products. This means that domestic product-specific support policies (for example wheat) are not allowed to exceed 10 percent of that product's value of production and that non-product specific policies are not allowed to exceed 10 percent of the value of total agricultural production. Under the Agreement there are some options regarding the method of calculating the actual level of domestic support (the so-called aggregate measure of support or AMS and the equivalent measurement of support or EMS). It is the judgment of some Turkish specialists that for some products the current level of price support could be challenged. Alternative calculations concluded that the actual AMS are higher than those reported to the WTO (although not higher than those reported to the OECD), exceeding the 10 percent maximum allowed under the minimus rule. The overall picture that emerges is that WTO commitments on export subsidies, and the de minimus domestic support policies may expose Turkey's current policies to challenge under WTO rules. This danger will be eliminated by the reform program currently underway (section 4.6). 222. Input subsidies constitute a second important component of agricultural support policies. The most important have been credit and fertilizer subsidies, though both are being eliminated under the reform program. Figure 4.2: Cost of Agricultural Credits 223. Credit subsidies. Short-term and investment credit for agriculture has long been 180 4,000 160 -3.500 subsidized by the government at interest rates 140 3,000 well below inflation and commercial rates. 120 2.500 Interest rates on loans from Ziraat Bank-either -100 -- 2,000 to ASCUs for crop purchase, directly to farmers, 1,500 or through the intermediation of credit coops- 60 1,000 have been significantly negative in real terms. 20 _00 In addition, unpaid loans of the ASCUs have , - _ 1 _ _ 1994 1950 1996 19 198 1999 been routinely covered as "duty losses" of the -. 1 Treasury, so the Ziraat Bank has had little Cost Credits ( mron) -4-Avg. Lending Rate , incentive to evaluate creditworthiness of - _ =- = - - - Source: Treasury, Ziraat Bank and World Bank borrowers or collect overdue loans. Losses on estimates. its loans to other borrowers have produced liquidity problems for Ziraat, forcing it to borrow heavily on interbank markets and pay high rates to attract deposits. Subsidized credit has distorted economic decisions on how and what to produce, encouraged investment in activities with low rates of return, promotes rent-seeking, and imposed a severe fiscal burden as well (figure 4.2). 224. Credit subsidies have been an ongoing drain on the whole financial system. On paper, bad debts in agriculture appear to be small.59 Lending to farmers both directly and through agricultural credit co-operatives (but unlike loans to ASCUs) seems generally sound and made with adequate safeguards. However, it is possible that loans may be repaid now simply to ensure continued access to more cheap loans, in which case defaults may rise once subsidies are curtailed. 5 As of end-August, 1999, about 13 percent of Ziraat Rank's agricultural portfolio of TL 1,676.4 billion was overdue. Of the amounts overdue, about 70 percent was loans to state economic enterprises and ASCUs. None of the loans to credit cooperatives were overdue. 75 225. Notwithstanding the good repayment rate, Ziraat Bank has steadily lost money. Although Ziraat Bank claims that profits from its other businesses have kept its agricultural lending rates low, profits on non-agricultural business in a competitive banking system cannot be enough to subsidize agricultural lending at high levels. The erosion in Ziraat Bank's equity was easily hidden - from the public, the government, and (given the inadequacies of inflation accounting) perhaps even from the bank itself. Ziraat Bank's illiquidity was more readily apparent. The government has from time to time reimbursed Ziraat Bank for its losses in the form of budgetary cash transfers and government paper. Although similar in purpose to "duty losses" for other agricultural subsidies, such reimbursements have not been based directly on the value of the current year's losses and have generally been paid only after significant delays up to several years. The resulting liquidity squeeze has forced Ziraat Bank to borrow heavily on interbank markets and pay high rates to attract deposits. High borrowing costs have in turn exacerbated the bank's liquidity problems and eroded its real capital base. 226. The government has never examined how effective credit subsidies have been in stimulating agriculture production. Studies in other countries have shown that money borrowed ostensibly for agricultural production is often used for other purposes, and anecdotal evidence indicates that the same is true in Turkey. Indeed, the Farmers' Association (TZOB) has stated that one-quarter to one-third of borrowers of subsidized credit may not be farmers at all. Furthermore, some farmers borrowed ostensibly for livestock production (since the interest rate is lower) and used the credit for crops. Thus, much of the credit was used for non-agricultural activity, and some of what was used for agriculture was likely used for purposes other than that for which it was intended by the government. As a result, cheap credit has probably not stimulated agriculture production very much. 227. It is certain that the cheap credit has had the same negative effects in Turkey as it has in other countries. It encourages overuse of funds, impedes the integration of rural financial markets with the rest of the financial system, and creates serious fiscal difficulties for Ziraat Bank and ultimately for the budget. As long as Ziraat Bank and the credit cooperatives (which borrow from Ziraat Bank) remained the conduits for credit subsidies, the rural sector was not able to develop the links to the rest of financial sector that would in the long run improve its access to financial services. As part of its new economic program, the government has introduced measures which effectively phased out the credit subsidies in early 2000 (see section 4.7). 228. Fertilizer subsidies. Subsidies have been given to the domestic manufacture and consumption of fertilizers since 1961, and for most of this time, both ex-factory and farmer prices for all types of chemical fertilizers were set by government decree. Fertilizer retail prices were decontrolled in 1986, and the government's monopoly on imports and domestic trade was replaced by an arrangement under which both the government and domestic producers could engage in trade. However, the restricted access to import licenses raised domestic prices so much that the reforms were not successful in moving domestic prices, adjusted for domestic freight and marketing costs, towards international levels, subsidies notwithstanding. Further opening of domestic trade in 1988 in the absence of greater import liberalization was also unsuccessful in giving farmers access to fertilizer at international prices. 76 229. In 1994, the government opened imports to any firn (producer or trader) with a minimum capital, eliminated duties on imports from the EU, and reduced import duties on imports from other countries to EU range (5-11 percent). It also began to pay subsidies directly to farmers. These reforms were immediately successful in moving domestic prices, exclusive of subsidy, which had been over 50 percent higher than international border prices in 1993 and 25 percent in 1994, almost to parity with these prices. This greatly benefited farmers. The reforms also made the subsidy more transparent. However, significant delays in subsidy payments to manufacturers in 1994 led to temporary plant closures and inadequate fertilizer supplies in that year. Delays in cooperative and farmer reimbursement have continued. 230. The fact that the rate of subsidy and the delay in payment of the subsidy is not predictable by farmers, fertilizer producers, or importers has impeded their ability to make rational long-term plans. The subsidy was set until recently as a percentage of market price, with the percentage varying considerably over the years. In 1996 and 1997 the subsidy was about 40- 50 percent of the market price, depending on the type of fertilizer. In November 1997, the government decided to fix the fertilizer subsidy in nominal TL per kg. This shift in policy has reduced the fertilizer subsidy substantially in real terms, as inflation has steadily eroded its value. The subsidy in 1999 was equivalent to about 25- 30 percent of the price. The highest subsidy as a fraction of the price (31 percent) is given for urea. 231. The fiscal cost of the fertilizer subsidy was approximately US$500 million annually between 1990 and 1998, and then fell around US$275 million in 1999 as a result of the change in the support program. The fiscal cost depends on several factors, including amount of fertilizer used by farmers, which in turn depends on the prices of fertilizer and crops, as well as weather conditions each year. Since all of these factors change from year to year, and the percentage of the subsidy has been subject to change as well, it has not been possible for Treasury to predict with certainty the amount needed in the budget. 232. Fertilizer subsidies no longer have any economic justification. At some time in the past, fertilizer subsidies may have had some justification in farmers' lack of familiarity with the benefits of their use and in the need to compensate farmers for domestic fertilizer prices that were kept far above world prices to protect the domestic fertilizer industry. However, neither of these justifications is still valid. Farmers are well aware of the economic benefits as they have been using fertilizers for many years. In recent years there has been a slow-down in consumption growth, in spite of high subsidy levels. Since the mid-1980s, it is difficult to detect any consistent growth in use. This pattern indicates a mature market, where farmers are well acquainted with fertilizer use. As for the second rationale for the subsidy, previous reforms in fertilizer import and marketing regulations (especially those in 1994) have been successful in bringing domestic prices (without subsidy) to world levels. Therefore, the subsidy is no longer needed. Furthermore, little of the subsidy reaches the population that would benefit most-- the smaller, poorer farmers. A simple estimate assuming fertilizer use is proportional to land use shows that 37 percent of the subsidy goes to the 5 percent of farmers with the largest farms, while only 22 percent goes to the 67 percent of farmers with less than 5 hectares. 233. Supply control measures have been used to a limited extent to control the fiscal cost of support policies. Tobacco (since June 1986), hazelnuts (nominally since June 1983) and tea (since June 1987) are under area or production control. Sugarbeet output is indirectly controlled 77 to some extent by the state-owned sugar company (Seker) through contracts with growers. Payments are given to tobacco farmers to compensate for the area controls and to tea producers to compensate for lost production from pruning. The hazelnut area reduction program has never been made effective because of lack of funding. 4 AGRICULTURAL STATE ECONOMIC ENTERPRISES 234. Agricultural SOEs carry out a variety of functions, including price controls on the producer and consumer levels, setting of procurement quotas, direct procurement of targeted commodities, and mechanisms to curtail production of surplus commodities. As input prices are kept high to support producers of commodities, and prices for outputs held down to subsidize consumers, SOEs incur substantial "duty losses." Further, SOEs are tools for the pursuit of social and political objectives, particularly to boost employment. This increases operating costs of SOEs, and requires periodic debt and equity injections from Treasury to cover liabilities. These policies impose significant and unsustainable costs on Treasury. Between 1994 and 1999 losses (duty losses and equity injections) for the main SOEs in agriculture (Caykur, Seker, Tekel, and TMO) amounted to US$6.2 billion (figure 4.3). The numbers vary greatly from year to year, reflecting whether procurement prices are high or low, and whether major equity injections are made to cover accumulated losses. Tekel and TMO have been the largest sources of losses. In the absence of fundamental reforms, there would remain a great potential for these losses to "balloon" in the future. For this reason, SOE reform and privatization will be a key part of the government's strategy (section 7). 5 OVERALL IMPACT OF THE AGRICULTURAL SUPPORT SYSTEM 235. The current system of agricultural support policies is fiscally expensive and economically inefficient. It encourages waste and abuse, and is not a cost-effective way to address policy objectives such as alleviation of rural poverty and regional development. The "duty loss" system of administration -- by which losses of organizations directly responsible for giving the subsidies are reimbursed by the Treasury, often after long delays -- has burdened these organizations with debt, impeded planning, and obscured the actual magnitude of the subsidies. 236. Overall, agricultural production in Turkey is protected; the protection is paid partly by consumers and partly by the govemment (taxpayers). Measuring the protection quantitatively is tricky, and the estimates vary significantly, though they agree on the broad conclusions. According to the official OECD estimate, total transfers to the agricultural sector, including from taxpayers and consumers, are estimated to have been around US$12.1 billion in 1997 and US$15.7 billion in 1998, or well over half of agricultural GDP, although it is not clear how much of this actually reached farmers.,, Of this total, the amount transferred from taxpayers is G The OECD estimate is derived using a consistent methodology for all OECD countries by taking the actual transfers for 13 crops and livestock, which are the same across all OECD countries, and then "grossing up" the figures to derive estimates for agriculture as a whole. Kasnakoglu and Cakmak ("The Fiscal Burden and Distribution of Costs and Benefits of Agricultural Support Policies in Turkey", September 1999), using different adjustment procedures for Turkey in the years 1993-96, derive estimates of US$3.3 billion and US$8.8 billion for fiscal and total transfers in 1997, and US$3.0 billion and US$6.5 billion in 1998. These are lower than the official estimates, but still huge. While the grossing up procedure may raise the estimate of the transfer. the OECD methodology understates the true figure to the extent that it does not include transfers to SEEs via government bonds and writeoffs. To preserve comparability in comparisons with other countries and over several decades in Turkey, the current paper uses official OECD estimates in Table 4. 1. 78 estimated to have been US$3.2 billion in 1997 and US$4.5 billion in 1998. Total transfers have increased as a percentage of GDP in Turkey from 4.8 percent in 1986-88 to about 6 percent in 1996-98. At the same time, the corresponding figures for all OECD countries has fallen from 2.5 percent of GDP to 1.3 percent. 237. The burden on consumers in Turkey was estimated to be US$11.8 billion in 1998 and has been steadily increasing since 1994. This burden has fallen disproportionately on low-and middle-income consumers, including the landless poor in rural areas, who spend a greater proportion of their incomes on food. It is estimated that the poorest 20 percent of the population in Turkey spends 51 percent of its income on food, beverage, and tobacco, while the richest 20 percent spends only 37 percent. Overall, the support programs impose an implicit tax of about 8 percent on the non-farm population. 238. Two conclusions emerge from table 4.1. First, Turkey's transfers have been very large, as a percentage of GDP. Other OECD countries give larger transfers per farmer, but they have very few farmers, and very high income levels, compared to Turkey. Thus this high level of transfers per farmer is much less fiscally sustainable in Turkey. Second, in other OECD Table 4.1: Total Transfers to Agriculture in Turkey and OECD Countries 1993 194 1995 1996 1997 1998 Total Transfers/GDP (%) Turkey 6.6 5.9 6.6 5.9 6.3 5.9 OECD 1.6 1.6 1.5 1.3 1.2 1.4 Total Transfers/Capita ($) Turkey 171 130 180 171 189 198 OECD 378 373 375 333 312 362 Total Transfers/FFE ($) Turkey 1,423 1,083 1,521 1,797 2,032 2,109 OECD 15,651 15,440 15,955 14,493 14,600 16,343 Total Transfers/Agricultural Land ($/ha) Turkey 259 196 283 330 373 388 OECD 284 280 284 254 256 287 Sources: OECD (1997), OECD (1996), OECD (1998), OECD (1999) and Kasnakoglu - Cakmak Study. Note: FFE is "full-time farmer equivalent" Figure 4.3: Shares of Categories of Transfer countries, the transfers have been declining since the mid- 1980s, while in Turkey they may even have even 1% increased. 239. While the relative importance of each instrument 13% used for support changes substantially from one year to the next, over the four-year period 1995-98, the most 4% important instrument by far was the credit subsidy, which accounted for about 55 percent of all support expenditures. (figure 4.3) Transfers to SOEs, including duty losses, equity injections, and write-offs, accounted for about 15 a*Expenetre on mdterial budget U Input subsidies percent, and input subsidies (mostly fertilizer) about 13 ilOutputand deasteSb ayeintscO percent. It should be noted that the distinction between r _ output payments and payments to the SOEs is not Source: Kasnakoglu and Cakmak, using OECD methodology. 79 economically meaningful, since both serve as support to production of the major commodities. The output payments classification includes some small items, such as milk premiums and disaster payments that are not administered through the SOEs. 6 REFORMS IN OTHER COUNTRIES-HOW HAVE OTHER COUNTRIES ADDRESSED THESE PROBLEMS? 240. Many countries have pursued input subsidy and output price support policies similar to those used in Turkey. However, few-- especially among developed countries-- have had programs with such high levels of support and fiscal cost, relative to the size of the economy or the agricultural sector. In the last few years, most developed countries and many developing countries have embarked on reform programs. For OECD countries, this is reflected in the low and declining figures in table 4.1 as a percentage of GDP. These reforms are driven by a recognition that these policies are not effective ways of supporting farmers, as well as by fiscal pressure and the need to comply with obligations under international agreements such as the WTO and the NAFTA. 241. In the US, the deficiency payment program (and production controls) has been largely ended for the most important crops. In the EU, Common Agricultural Policy (CAP) reforms since 1992 have brought the prices of many important agricultural products (particularly cereals) much closer to world prices, and reforms are continuing. Subsidy payments in the EU are increasingly linked to farmers' measures to conserve resources or to regional development. Profound reforms in their agricultural commodity policies towards less reliance on price support programs have also taken place in recent years in Canada, New Zealand, and Australia, as well as in many middle-income developing countries, especially in Latin America. In Mexico, for example, support through pricing and subsidies (measured as Producer Subsidy Equivalent, or PSE) fell from 30 percent in 1992-94 to only 5 percent in 1995, and was replaced by direct income support. By way of comparison, Turkey's overall PSE in 1998 was estimated by the OECD to be 42 percent. 242. Reforming governments have recognized that the objectives for which the inefficient policies were designed can be better met by using other instruments. One of the most important of these is the direct income support subsidy payment, which can substitute for price supports or input subsidies. The principle underlying the direct payment is that it should be based on something over which the producer does not have control. For example, it cannot be based on how many tons of wheat the farmer produces in the current year. Otherwise, it is equivalentto offering a higher price for his wheat. For this reason, direct support cannot be used to encourage production of specific crops or use of specific inputs. The principle of providing subsidies through lump sum payments instead of through prices is called "decoupling" or "delinking" support from prices. In the US, the EU, and Mexico, the reduction of price supports has been accompanied by full or partial compensation to farmers through de-coupled payments based on production in some baseline period in the recent past. Direct payments each year stay the same or are reduced by a pre-determined formula. They do not change from year to year based on that year's input use or output production. 243. The advantages of substituting direct income payments for other subsidies are manifold: (i) efficiency of resources devoted to agriculture increase as farmers are encouraged to make 80 production and input use decisions on the basis of market prices; (ii) consumer prices of some agricultural products are reduced; (iii) budget planning and execution improve since the amount of subsidy amount no longer depend on the level of input use or output produced; (iv) farmer income and the total farm budget are more predictable; (v) stable and predictable direct income payments can be pledged as collateral for loans; (vi) Focused distribution of government support to fanners most in need of assistance becomes possible since direct payments can be targeted at the small, poorer farmers. These farmers devote a great part of production to their own consumption and use few purchased inputs, so they do not benefit much from input subsidies and support prices. This latter point is relevant to Turkey where budgetary resources are limited and the agricultural population is not disproportionately poor; and finally (vi) direct income payments lower the distribution cost of subsidies. 7 NEW REFORM PACKAGE 244. As part of its reform agenda, the government has developed a strategy to phase out the current support mechanism and to introduce "decoupled" direct income payments. The direct income support program is being tested on a pilot basis in 2000 and a national farmer registry is being prepared. Rollout of the program at the national level will start in 2001 and be completed in 2002. The payments will be based on a fixed amount per hectare subject to cap on the total payment per farmer. Therefore, the program will be only moderately targeted at first. But the payments could become more targeted over time toward smaller farmers and possibly integrated into the broader social safety net. 245. The government is starting the support policy rationalization with agricultural inputs and credit. As mentioned earlier, the fertilizer subsidy has been held constant in nominal terms since 1997, resulting in a reduction of the unit subsidy from approximately 45 percent of the total price at the end of 1997 to approximately 31 percent by August 1999. The fertilizer subsidy will remain constant in nominal terms in 2000-01. The unit subsidy should fall to close to 20 percent of the unit price by the end of 2000. 246. In December 1999, the governrment introduced a program for phasing out agriculture credit subsidies channeled through Ziraat Bank over the course of 2000. The program involves: (i) holding the nominal interest rate on agricultural credits constant until it is equal to the three- month rolling average of the twelve-month t-bill rate plus 500 basis points and then holding this spread constant; (ii) introducing a variable rate loan option for farmers; and (iii) providing cash compensation from the budget to Ziraat Bank for any subsidy accrued in 2000. As t-bill rates fell sharply in early 2000, the subsidy element has been effectively eliminated already. 247. The government has announced a set of policy changes which introduce a link between support prices and relevant world market prices and initiate a phase-out of government subsidies for support prices by 2002. Support prices for grains in 2000 were linked to appropriate world reference prices and set at levels which reduce the premium over these world prices to no more than 35 percent. Import tariffs on grains will be reduced as well. The sales price for grain of TMO, the state grain purchasing company, will be no less than the lower of: (i) the purchase price of TMO plus storage costs incurred up to the date of sale including imputed interest charges on stocks, or (ii) the tariff-inclusive import parity price for grain of equivalent quality. 81 This pricing policy is intended to reverse the past incentives of traders and millers to refrain from buying grain during the harvest season and leave all storage costs to be absorbed by TMO. 248. An auction for tobacco will be introduced for the 2001 season and prices for tobacco not sold at auction will be set at a discount below the lowest auction price. The discount will be increased in future auctions to discourage the production of low-quality, unsaleable tobacco. Additional measures will be introduced in 2001, including: (i) further reduction in the premium over world grain prices and import tariffs on grains; (ii) reduction of premium paid on oilseeds and cotton; (iii) reduction of payments under the tea pruning program; and (iv) reform of pricing mechanisms for sugar beets to reduce fiscal costs and make these prices more market determined. Savings from these further reform measures in 2001 will be channeled back to farmers through the direct income support program. 249. The agriculture reform program encompasses commercialization and privatization of state assets in the sector with the medium-term objective of withdrawing the state from a direct role in agricultural and agro-industrial production. As a first step, the government has imposed a hard budget constraint on agricultural SOEs for 2000, including specific enterprise-by-enterprise limits on Treasury loan guarantees, equity injections and budgetary transfers. In parallel, restructuring and privatization process will be initiated. 250. Three new laws necessary to reform Tekel are expected to be passed during 2000. One will introduce auctions to replace the current system of support prices at which Tekel must purchase tobacco. The second will de-monopolize the production of alcoholic spirits (one of Tekel's monopoly divisions), thereby allowing private entry into the industry. The third law will enable the privatization of Tekel's production facilities for spirits, salt, and tobacco products. The government plans to complete divestiture of 40-50 percent of Tekel's commercial assets in 2001, and all remaining assets by the end of 2002. 251. In 2000, a new Sugar Law is expected to be passed that will eliminate the support pricing system and allow TSFAS and private companies to negotiate prices directly with growers. The government intends to phase out budget transfers to TSFAS over the next two years (as the direct income support system is put in place) to encourage TSFAS to operate on a commercial basis and to negotiate prices with beet farmers consistent with elimination of its losses. The government plans to eliminate budgetary subsidies by 2002. TSFAS will be broken up and its various units will either be sold or liquidated over the next three years. The privatization process is expected to start in 2001 and be completed by 2002. The goal is to privatize all viable factories. The non-viable factories will be liquidated following successful implementation of the direct income support system and the establishment of appropriate social safety net provisions. An inventory and evaluation of the sugar plants of TSFAS will begin in 2000 with the objective of preparing their privatization. Before the privatization begins, the basic medium-term strategy would be announced to phase out budget support for sugar, avoid export subsidies and lower the sugar tariff. The current sugar tariff of 141 percent is being reduced by 1.5 percentage points per year through 2004 in accordance with Turkey's WTO commitments. 252. Caykur has already announced its new policy of buying only high-quality tea from pruned areas. Caykur's purchases will be limited to 600,000 tons in 2000 in line with the Government's fiscal program. Caykur has increased nominal prices by no more than 25 percent 82 to encourage growers to sell to the private sector factories. Caykur plans to further decrease its purchases in 2001 while the production of alternative crops such as kiwi increases. Profitable factories of Caykur will be privatized and non-viable factories will be liquidated by the end of 2002. 253. TMO will be restructured and down-sized in 2001 to preserve only its assets needed to carry out its minimal level of purchases and storage. In the longer term, most of its assets will be privatized as TMO restrict its grain purchases to strategic reserves. Consequently, TMO will retain ownership of only limited warehouse space and divest the remainder. As of 2002, TMO will make purchases only from the commodity exchange market. The agricultural input supply state economic enterprise, TZDAS, has been transferred to the Privatization Agency, and its staff reduced from 3,300 to 1,500. The privatization of its remaining assets will be finished by the end of 2000, with almost all of its assets sold or liquidated, and all employees terminated or re- assigned. 254. In early 2000, legislation was passed giving complete autonomy to ASCUs. The government will take over the ASCUs' existing debts, in accordance with concrete benchmarks in their restructuring programs, and set aside in its budget some TL 160 trillion for providing working capital and restructuring. A restructuring board (RB) will be set up to oversee the restructuring process including debt forgiveness and disbursement of restructuring funds. The board will be composed of seven members appointed by the Council of Ministers. The RB will carry out its functions through an executive body that will be composed of specialists and consultants who are highly experienced particularly on issues of planning, organizational restructuring and/or cooperative development. This executive body will have full autonomy to function efficiently within the framework of its work program as determined by the board. 255. Predicting the impact of this comprehensive reform programn is difficult and there is much uncertainty. Nonetheless, some rough estimates of the fiscal impacts over the next few years have been developed by the Bank and Treasury. These are given in table 4.2. The estimates of the cost of the direct income Table 4.2: Fiscal Impact of the Reform Program, 2000-2003 support are based on an assumed US$80 per hectare (USs milin) No &form 2000 2001 2002 2003 Credit Subsidy 1,400 655 0 0 0 payment which is somewhat Fertilizer Subsidy 245 245 221 0 0 higher than that given in TMO 865 844 364 0 0 Mexico under a similar reform Tobacco 538 472 192 196 0 prograrn. Preliminary Cotton 269 269 202 0 0 program. Preliminary QOilseeds 51 51 38 0 0 estimates suggest that this ASCUs 860 243 432 185 0 would not be sufficient to Sugar 811 711 569 0 0 compensate many Turkish Subtotal 5,039 3,490 2,018 381 0 farmers for income lost as a Direct Income result of the removal of Payment 0 2 576 1,152 1,152 exst . Total 5,039 3,492 2,594 1,533 1,152 existing subsidies. The per hectare subsidy could be Source: Treasury and World Bank. raised, as table 4.2 suggests, while still realizing a net saving compared to the pre-reforn policies. Nonetheless, it is not recommended that an attempt be made to fully compensate farmers for all subsidies being removed. One reason is that farmers will be able to adjust their decisions based on the new price 83 structure in such a way that their actual fall in real income (excluding the direct payments) will be smaller than would be suggested by calculations made assuming they will continue to use the same inputs and grow the same crops as before. Another reason is that attempts to compensate individual farmers based on their own actual losses would require a complex and non-transparent system, obviating much of the clarity of the direct income approach. 256. It is clear that some farmers will lose a large amount of subsidies under the refortn, while some will lose a little, and some-those who grow crops that were not heavily supported and did not use much credit or fertilizer-will gain. Livestock producers will unambiguously gain from cheaper feed grain prices. Small-scale producers, who grow mainly for self-consumption, will benefit. Simple estimates of losses of farmers growing highly supported crops are inherently too high since they assume that farmers will keep growing the same crops when their relative prices fall, whereas in reality many will change to more profitable crops or other activities, thereby minimizing the drop in income. These estimates also depend greatly on the base year as support prices vary from year to year 61 Estimates show large losses for wheat farmers, for example, when using the very high support price in 1999 as the reference point. Also, it is clear that tobacco farmers who grow unsaleable tobacco will either upgrade their production so it can be sold at auction or switch to another activity. There will be strong incentives from the program for farmers in areas not well suited to wheat, tobacco, and sugar to switch to crops such as cotton, sunflower, horticultural crops, and forage, and to livestock. 257. The government intends to provide additional assistance to help affected farmers make the transition from products which currently are highly supported (such as tobacco and hazel nuts) to other, more marketable crops and rural off-farm activities. This farmer transition program will concentrate on regions where the existing policies have resulted in uneconomical production of the supported crops. The program is expected to encompass financial support to purchase inputs and make the necessary investments to carry out the transition. 8 THE GOVERNMENT'S NEW ROLE IN AGRICULTURE 258. Freed from responsibility for the costly subsidy programs and commercial activities, the government should concentrate on those activities that it can truly carry out more effectively than the private sector. Some of the budget released by subsidy reduction and privatization and liquidation of SOEs could be re-channeled to compensatory payments, research and extension, and other means of helping farmers adjust to the new environment. Some legitimate public sector fumctions--such as research and extension--which are currently carried out by certain SOEs. These could be transferred to the Ministry of Agriculture and Rural Affairs. Research and extension workers from Tekel, Caykur and Seker could be employed by farmers associations with matching grants from the government, to assist farmers to become more competitive in growing their traditional crops or to develop alternative activities in agriculture or rural industry. 61 The estimates of losses per hectare from removing input subsidies, support prices, and premia (based on farm budget information from 1999) are the following: dryland wheat- US$158 per ha. (37 percent fall), irrigated wheat-- US$284 per ha. (32 percent fall), cotton-- US$69 per ha. ( 6 percent fall), sunflower-- US$27 per ha. (4 percent fall), tobacco (unsaleable support grade)-- US$1,136 per ha. (51 percent fall). 84 9 LONGER TERM REFORMS: RECOMMENDATIONS 259. Beyond the most urgent objective of reforming support policies, the government will need to confront several other agriculture problems in the medium term. One is irrigation policy where there are problems in two dimensions. The most important dimension is investment policy. Many irrigation schemes have been started, some of them years ago, but are funded annually at such low levels that it will still be many years before any benefits are realized. This problem is likely to be exacerbated in the current period of budgetary stringency. What is needed is to desist from starting new schemes, to identify the most cost-effective projects and focus funding to complete them within a reasonable timeframe. This will require not only a more focused irrigation investment strategy, but also better coordination and perhaps a rebalancing of investment expenditures between DSI and GDRS, since the work of both of these agencies is needed to bring projects to the point where water is actually produced for farmers. The second dimension of irrigation policy requiring action is the transfer of irrigation infrastructure to water users' organizations (WUOs). Turkey's current policy is to transfer responsibility for operating and maintaining parts of the irrigation infrastructure to WUOs. This has been a very successful program which has reduced DSI's operational costs significantly, while empowering the users to make many of their own decisions and assume responsibility for issues that affect them directly. In some respects, Turkey's policy toward WUOs is a model for developing countries. However, the program is constrained by the fact that WUOs have no legal standing in Turkey as they do in developed countries. Legislation is needed that will allow WUOs to attain the status of autonomous legal entities, thereby enabling them to assume ownership of irrigation infrastructure. increasingly large portions of the infrastructure could then be transferred to individual WUOs and associations of WUOs. This would give them greater decision-making authority and incentives to make long-term investments. These incentives are lacking as long as WUOs have only uncertain usufruct rights as is currently the case. 260. A second area where reforms will be needed is the forestry sector. While the current framework is not producing highly visible fiscal costs, institutional and policy deficiencies are limiting the ability of the sector to provide sustainable income generation for forest communities and revenue for the budget. What is needed is a restructuring of the forestry-related organizations, hiving off commercial activities, devolving some responsibilities to local communities, and focusing central government responsibilities on legitimate functions that require national coordination. More detailed recommendations are contained in the Forestry Sector Review currently being discussed with the government. 261. Third, the government should determine what institutional changes in MARA might improve its focus on its remaining responsibilities in the new policy environment. This would probably involve larger scale reforms in the way research is funded and carried out, along the lines piloted in the Bank's Agricultural Research Project. It would also call for restructuring of extension to encourage better linkages with research and more private sector provision of these services , as is best practice in other countries. But MARA's responsibilities in these areas would need to be expanded to cover some of the crops over which it has traditionally played a limited role, as the associated SOEs (tea, tobacco, sugar) are privatized. 262. Finally, in the longer term, many actions will need to be taken in preparation for EU accession. This will require organizational changes in the agricultural institutions, 85 harmonization of much legislation and regulations (including veterinary, sanitary, and phytosanitary standards), improvements in agricultural statistical systems (including both farner and livestock registration), and upgrading of processing plants. These are of course not easy tasks, but Turkey will benefit from the experience of the other accession countries. Emphasis should be placed on the more important and challenging goal of meeting the general criterion for EU accession-that new members must have economies that are able to compete in the unified market. This puts the focus squarely on measures to improve productive efficiency in primary agriculture and agro-marketing and processing, including reforming support policies, providing an environment conducive to the development of marketing infrastructure (e.g., commodity markets), and improving farmer services (research, extension, and veterinary services), as complements to the ongoing reform of support policies. 86 CHAPTER 5: BANKING SECTOR DEVELOPMENTS AND REFORM6Z 1 INTRODUCTION 263. Turkey's dynamic and increasingly market-oriented banking sector can contribute to high and stable economic growth. But that potential has been severely constrained by the distorted incentives from the country's long-standing macroeconomic imbalances. The incentive structure for banks has suffered from chronically high inflation, large public sector borrowing requirements, and use of tax breaks to favor placement of government debt over private sector borrowing. Although total bank assets have doubled as a share of GNP over the past two decades, banks have progressively reduced the maturities of their operations as savers have overwhelmingly opted for bank deposits and money-market instruments while abnormally high real interest rates on short-term government paper have offered easy profits. This environment has encouraged fragmentation with the proliferation of small banks. At the same time, limitations in the legal and regulatory framework for banking supervision have allowed many banks to accumulate weaknesses in their asset portfolios. The financial ratios of the state banks, which account for some 40 percent of all bank assets, have deteriorated. The main state banks have built up large and growing stocks of illiquid claims on the Treasury, as a result of government-mandated subsidy and preferential credit programs. 264. Recognizing the importance of an efficient banking sector and acknowledging the risks that the disinflation program poses for the weaker banks, the government has launched a broad based program to reform the financial sector. This program has three core components: * Modernizing the legislative, regulatory and institutional infrastructure for the banking sector in accordance with EU and world standards. * Resolving problem private banks. * Reforming the state banks. The program has had a strong start with the enactment of a new banking law, quick action on private bank resolution, establishment of an independent Banking Regulatory and Supervisory Agency (BRSA), and steps to staunch the flow of duty losses by phasing out subsidized credit programs. But much more will be needed to overcome the structural problems accumulated over decades of macroeconomic instability. Restructuring and privatizing state banks, managing the adjustment of private banking to the new incentive framework are at the top the list of priorities on the longer term policy agenda. 265. This chapter briefly analyzes the major banking developments, the incentive structure for banking and its impact on resource allocation--with an emphasis on medium-term implications including risk vulnerabilities. It then focuses on the state bank sector and discusses the problems currently affecting the state owned banks, in particular their large losses deriving from quasi- fiscal activities. The chapter also reviews the new banking reform program and offers recommendations on the structural policies needed to increase banking sector's perfornance and stability. 68 This chapter draws from the background papers by Bossone (1999) and Ersel (2000). The chapter also benefits from the inputs and the information made available by the Bank team for financial sector reform in Turkey: (i) Banking Sector Review-I, November 1998; (ii) Banking Sector Review - 11, November 1999. 87 2 MAJOR STRUCTURAL CHANGES AND DISTORTED INCENTIVES STRUCTURE 266. Over the two decades since the liberalization took effect, the Turkish banking system has grown considerably (table 5.1) and increased its deposit base. At the same time, banks have shifted their financing activity from lending to securities investment (largely government debt) and fixed assets. Banks have also diversified their activity into foreign currency denominated transactions and off-balance-sheet products. Chronically high and erratic inflation, as well as large asset price Table 5.1. Ratio of Bank Assets to volatility, have altered the time and liquidity GNP for Selected Years preferences of the banks, shortening their time horizon (Values in USD billion) and inducing large portfolio shifts toward greater Year Bank GNP Ratio liquidity. The average loan maturity of the TL loans has 1980 A18.6t 69.7 26.7 declined significantly. 1985 27.0 68.2 39.6 1990 57.9 150.1 38.6 267. The degree of competition and openness of 1995 68.4 170.1 40.2 Turkey's banking sector has grown considerably. The 1998 117.8 205.8 57.2 size of the industry relative to the economy has 1999 133.5 187.4 71.3 increased. In 1999, Turkey had 81 banks, of which 52 Source: Calculations on data from Banks were privately-owned, including 19 development and Association of Turkey. investment banks. Over the past 20 years, sector concentration has diminished with the asset share of the ten largest banks decreasing by some 13 percentage points to 68 percent in 1999. The state presence in the banking sector has contracted, although it is still large and the largest commercial bank in the country is state owned. Table 5.2: Banking Structural Indicators Quasi-fiscal activities of state banks, including 1990-95: International Comparison preferential and subsidized credit and Net InAerest Overhead Concer&atior : Marink Costs index* agricultural commodity support programs, (%ofoass) (0/oofassels) have also been reduced. The presence of Turkey 0.10 0.06 0.44 foreign banks in Turkey has increased. Argentina 0.07 0.10 0.50 foreign ~~~~~~~~~~~~ ~~~Brazil 0.11 0.11 0.60 Chile 0.04 0.03 0.47 268. International comparison of banking Greece 0.03 0.04 0.77 structural indicators for the 1990-95 period Hong Kong 0.02 0.02 0.72 Ireland 0.01 0.01 0.74 (table 5.2) shows that Turkish banks had a Korea 0.02 0.02 0.31 relatively higher net interest margin, close to Malaysia 0.03 0.02 0.49 average overhead costs, and a relatively low Mexico 0.05 0.05 0.58 concentration index. The apparent profitability Source: A. Demirguic-Kunt A and R. Levine, 1999, Bank- of Turkish banks has been on the rise, both in Based and Market-Based Financial Systems: Cross-Country nominal and real terms.,3 This trend has been Comparisons, mimeo. * Measured as share of the assets of the three largest banks especially marked for the private commercial in total banking sector assets. banking sector which has recorded steadily increasing average profit ratios. 269. The country's experience with financial sector reform shows the negative consequences of undertaking financial liberalization without macroeconomic stabilization. The Turkish 69 Caution must be exercised in using profitability indicators for Turkish banks due to accounting standard inadequacies, among which the lack of inflation accounting and insufficient loan provisioning, particularly before 1996. There have been improvements in accounting standards and loan loss provisioning (see section 5). 88 experience demonstrates that the lack of a policy framework to redress the macroeconomic imbalances prior to undertaking financial liberalization may hinder the very purpose of the latter to improve the allocation of resources and to support higher and more stable output. 270. The uncertain economic environment was responsible for many of the changes in banking activity observed in Turkey. But other important incentive effects have been at work as well to affect resource allocation decisions. Financial asset prices and the lending Table 5.3: Various Interest Rates decisions have been heavily distorted Financial by the government's increasing Instruments 1993 194 1995 1996 197 1998 199 borrowing requirements. The heavy Deposits public sector borrowing s requirements.f Demand 11.1 5.1 13.5 19.3 26.8 30 27.4 public sector borrowings and the use of Time I mn 52.9 61.8 83.7 76.1 78.3 80.5 72.1 tax advantages to make public debt 3 mn 64 77.3 83.9 79.7 83.2 82.8 59.5 instruments attractive to investors have 6 mn 69.1 81.1 83.4 84.6 91.5 85.6 48.3 provided banks with a rather I yr 74.8 95.6 92.3 93.8 96.6 94.8 46.7 comfortable way to earn income (at a USD Dep. I yr 4.1 4.8 6.5 8 9.3 10.9 12.3 G-bonds 85 137 108 115 111 106 106 low perceived risk) simply by T-Bills 3 mn 86 170 133 141 111 119 97 purchasing and holding high interest- yielding public securities, while cutting Inflation back on lending and investment CPI 66.1 106.3 96 80.4 85.7 84.6 64.9 financing activities. No comparably WPI 58.4 120.7 86 75.9 81.8 71.8 53.1 high and apparently secure gains could Source: Central Bank, Treasury and Banks Association of be obtained from lending to enterprises Turkey. with the possible exception of short- term working capital loans to firms with strong and stable cash-flow. The high interest rate premiums paid by the government on its debt have made for a large share of bank profits (table 5.3 and figure 5.1). The government has enhanced the market attractiveness of public Figure 5.1: PSBR and Bank Profits debt securities through generous tax credits and exemptions. Moreover, by requiring banks to hold some fixed share of liabilities in Treasury 3 - - - ---_ 1 bills as liquid reserves, the government has secured a predictable demand component for v. 2 A 12z domestic public debt. 2 ' 271. As the interest rate premiums on 7 - X / 8 government debt have exceeded the rate of domestic currency depreciation expected by the o -4-- 4 market, banks have found it profitable to fund 1991 1992 1993 1994 1995 1996 1997 1998 government bond purchases with foreign --4CorDraalBankProf6s -- PSR(rbtIp) borrowing. This has led them to incur large open foreign-exchange positions. Overall, banks Source: Banks Association. seem to have underestimated the risks inherent in overly extending investments in government paper and in opening FX positions. Their concern with risks may have been weakened by the generous deposit insurance coverage-which until June 2000 protected 100 percent of saving deposits. This full coverage made it easier for all banks to raise deposits from the public regardless of the quality of their balance sheets. 89 272. The explanation for the abnormally high interest rates persistently recorded in Turkey over the 1 990s may rest primarily on the effects of the uncertain dynamics of domestic inflation,64 exacerbated by a monetary policy oriented to stabilize the real exchange rate which left prices without an anchor. It also partially reflects the strong competition for funds in the money market between the government and the state-owned banks' refinancing needs. Inflation uncertainty has kept the liquidity preference of the public very high, inducing individual savers to hold large volumes of short-term deposits (more than half of which are in foreign exchange) and to invest in instruments-such as repo's-with a much shorter maturity than government debt. This has translated into persistently high interest rate spreads between government securities and money- market instruments, which banks have found profitable to arbitrage. 273. Distorted financial sector incentives inevitably are transmitted by banks to the agents across the economy, eventually resulting in demand-supply imbalances of investable funds. Turkey's investment-to-GNP ratio, capital productivity, and output growth--although remarkably high relative to the volatile economic environment--are lower than in comparable economies.65 Only the largest industrial companies have direct access to international financial markets for funding investment, while most other enterprises are forced to rely on internally generated savings. Therefore, Turkey may be missing important development opportunities especially in technologically advanced sectors that require long-term risk capital. Also, smaller firms may not have the means to grow in size and productivity. Support from external financial sources for domestic investment has been limited as well.6, Deposits and credits from abroad have been dominated by short-term flows, and investments in securities have mostly gone into government debt financing. In the second half of the 1990s, foreign loans made directly to non-financial enterprises increased. While foreign lending may have become an increasingly important source of finance for large domestic firms, informal evidence indicates that most of this lending is short- term and does not necessarily finance export-oriented businesses. 274. While correcting financial sector incentives rests on the success of the stabilization and structural reform program, such a program is not devoid of risks for the banking sector. As stabilization is pursued, banks will most likely face a difficult transition before they regain the ability to compete under restored incentives. There are reasons to believe that the banks' prolonged disintermediation from real sector financing activities has weakened their skills to select and monitor good long-term business projects for funding. Bank information capital may have eroded, together with the banks' ability to evaluate and manage real sector risks appropriately. Also, in the absence of strong real-sector lending skills and in a context in which banks often belong to large conglomerate groups, lending to parent companies and connected parties is perceived by some banks as a safe practice, while in fact it may lead to inferior quality assets. 275. With both inflation and public sector financing requirements reduced drastically, bank profits from investing in government securities and from financial arbitraging will no longer be as attractive. Also, weakened skills to select new good borrowers and projects may induce banks 64 This hypothesis is supported by the recent study by H. Beremut and K. Malatyali, Determinants of Interest Rates in Turkey, Research Department Discussion Paper 9902 (Ankara: The Central Bank of the Republic of Turkey). ' See Chapter 1. See 0. Celasun, C. Denizer, and D. He, Capital Flows, Macroeconomic Management, and the Financial System: The Turkish Case, 1989-97, IMF Working Paper, Feb. 1999 (Washington DC: International Monetary Fund). 90 to take on excessive risks and affect their capital position. Difficulties in risk evaluation will be further exacerbated as industrial enterprises, too, have to adjust to a low-inflation economic environment. Banks will have to identify the potential winners of the new game on the basis of limited information and new parameters. During and after stabilization, bank competition will become more intense and riskier as much as unavoidable and desirable. At the same time, banks will face increased pressure on profits and capital from comprehensive upgrades to the prudential regulations. While this process will allow a better and more efficient banking industry eventually to emerge, it is possible that in the interim period some weak banks will not be able to survive in the new environment. Others will be strong enough to succeed and others still will have the potential to do so but will require rehabilitation and, possibly, external assistance. 3 FINANCIAL RISKS AND RISK MANAGEMENT 276. As much as the incorrect incentives was discussed above produce asset misallocation, they also largely deterrnine the risks to which banks are currently exposed. Financial market developments in the second half of 1998 and in the course of 1999 highlighted the critical sensitivity of Turkish banks to changes in market confidence. Such sensitivity makes bank funding volatile and dependent on market expectations as to the country's economic developments and policy stance. Also, to the extent that incentives have led banks to engage heavily in financial arbitraging, opening large foreign exchange positions and maturity gaps, bank exposures to price changes-- most notably, sudden interest rate hikes and exchange rate devaluation--represent considerable risks. Although banks and the banking authorities have taken steps to strengthen the sector's capital base, concerns remain on the quality of risk management at the bank level and the relative priority given to risk issues by bank supervisors. 277. The risk attitude of Turkish banks cannot be assessed properly without considering the impact of deposit insurance on their risk taking behavior. The Turkish authorities recognize that market discipline has been weakened by the government decision, taken during the 1994 banking crisis, to extend insurance coverage to 100 percent of saving deposits in domestic branches. This has made it easier even for the weakest banks to raise funds from depositors. In some cases, low- quality banks have engaged in practices of charging extra-high deposit rates and lending to over- risky projects in the hope to grow out of their liquidity/solvency problems67. A plan for gradual reduction of full deposit insurance introduced in June 2000. 278. Capital adequacy. Since the 1994 banking crisis, private commercial banks have increased their share capital and reserves in proportion to assets, while the capital position of state banks has further weakened (table 5.4). The authorities have taken decisive steps in strengthening bank capital. Banks in Turkey are required to maintain an 8 percent minimum capital adequacy ratio and to report on their capital adequacy position quarterly. The banking sector has consistently stayed above the minimum required. However, some banks in the system have capital ratios which are well below the prudential limits set by the supervisory authorities. The state banks are less capitalized than the private institutions, and the domestic banks remain far less capitalized than the foreign ones. Moreover, capital adequacy requirements didn't incorporate market risks and consolidated capital ratios, until 2000. 6 A plan for gradual reduction of full deposit insurance was put into operation in June 2000 (see section 5.2). 91 279. Credit risk. The short-term orientation of activity and the relatively low propensity to lend to the real sector have limited the banks' overall exposure to the risk of borrower default. Moreover, bank capitalization appears to be comfortable relative to potential credit losses. However, credit risk may be a problem for some Table 5.4: Capital Ratios of Commercial Banks individual banks that have Capital ratios by IBank 1"3 194 1995 19% t1997 199 1999 followed overly Categories aggressive growth Total Capital/Risk-weighted assets strategies with insufficient Commercial State owned 14.9 8.1 8.3 8.5 11.7 8.6 11.8 concern for the quality of Privately owned 9.1 9.7 14.8 13.3 11.9 12.0 14.7 assets and counterparts Foreign 13.1 20.2 21.7 20.8 13.5 21.7 23.4 financed. The fall in loan Development and Investment 11.0 3.6 7.9 14.2 17.6 24.4 24.0 repayment in 1998, Net Working Capital' /Total assets especially to private Commercial State owned -1.0 -4.2 -5.2 -4.1 -1.7 -2.6 -2.2 commercial banks, may Privately owned 3.7 3.7 5.5 4.7 4.5 1.7 -6.5 be a harbinger. Bank Foreign 7.8 13.3 9.6 9.5 7.5 9.0 8.4 management across the Development and Investment 5.3 -1.6 1.9 6.6 9.4 13.3 8.8 industry should act to Source: Central Bank, Treasury and Banks Association of Turkey. strengthen loan evaluation ' Defined as (shareholders' equity + current and last year's profit) - and monitoring (non-performing) assets. procedures, as well as loan provisioning. This becomes even more important as bank competition for real sector business intensifies. A new regulation was issued on loan classification and loss provisioning, the new rules were still emphasizing actual--as opposed to potential--losses and determined provisions according to the length of time a loan has remained in default and the type of collateral provided for by the borrower.,, These problems were resolved in 2000 (see section 5.1). 280. The need for enhanced practices for credit risk evaluation and monitoring and for stronger provisioning rules should be considered also in the light of three additional credit risk factors. The first indirectly derives from the exposure of some large bank borrowers to foreign exchange risk. This originates from a sizeable portion of foreign currency loans being extended to companies which do not engage in foreign trade related business69. As these companies' debt repayment capacity is vulnerable to exchange rate depreciation, lending banks are indirectly exposed to default risk. The second risk factor derives from banks carrying increasing off- balance sheet risk positions in the form of guarantees on loans. The third factor involves risks from connected lending. There are reported instances of insider lending practices where the terms and conditions of the loans are not arms-length, and of breach of prudential limits through cross lending of group banks to related enterprises. 281. Liquidity risk. Since the 1994 crisis, the overall stability of the banking sector's liabilities as well as the quality of liquidity management and supervision have improved. Large banks in Turkey take liquidity risk seriously. Their willingness to invest in retail franchise and international reputation, their experience with liquidity markets and products as a source of 6 The new regulation has abolished the "special follow up" loan classification, which sought to allow banks to classify loans according to perceived potential problems. 6 This is against the prudential regulation. 92 business, and their proven attitude to managing liquidity in a highly uncertain environment make them relatively less exposed to this kind of risk than to price risks. Concerns arise from two sources. First, the sector is vulnerable to shocks. As banks are similarly positioned on the securities and foreign exchange markets, major external or policy shocks might lead banks simultaneously to seek to shield their portfolio at a time when they might also be facing capital flight and deposit withdrawals. This could make their liquidity prospects much less safe that anticipated. Another major source of concern for systemic liquidity problems derives from Turkey's overall economic policy environment. As Turkey's economy is increasingly integrated in the international financial markets and Turkish banks have expanded their funding from abroad, capital flows have become an important component of the country's financing.-, 282. Exchange rate risk Most Turkish banks, especially in the private sector, run larger exchange rate risk positions than would normally be considered prudent. Although taking such risks has so far proven to be profitable under the previous incentives structure, there remains a significant chance that adverse exchange rate movements might seriously challenge the financial stability of the banking sector, as virtually all risk-taking banks are positioned much in the same direction, keeping open short-term FX positions and holding longer-term Lira instruments. As the same banks are also the dominant players in both the foreign exchange and government debt markets, all would likely move in unison to shield their portfolios in response to major shocks. The open foreign-exchange position of the banking sector widened in the early 1 990s and again after 1996. This position also has grown in relation to assets, especially for the private commercial banking sector, reaching almost 10 percent in 1999 (table 5.5). These figures do not include the increasing off-balance sheet foreign-exchange commitments that banks would be called on to honor in case of default by the original borrowers, and that would obviously increase in the event of currency depreciation. Since 1995, the Central Bank monitors an net open foreign-exchange position indicator which includes both forward and Table 5.5: Open Foreign-Exchange Positions of Banks' indexed transactions. As a ratio to Values in USD billions 1995 1996 1997 1998 1999 capital base, the net open position of Open Positions (OP) 3.1 2.5 4.9 8.4 13.2 the whole sector has more than tripled Net Open Positions (NOP)2 0.4 1.2 1.8 2.6 2.8 between 1995-98 and has continued to Ratios (in %) widen in early 1999. OP/Total Assets 4.5 3.0 5.3 7.2 9.9 widen in early 1999. OP/Net Worth 50.8 33.8 56.2 81.0 168.4 NOP/Total Assets 0.6 0.1 0.2 0.3 2.1 283. Turkish banks focus closely on NOP/Capital Base 10.5 29.9 31.4 33.0 35 their currency risk. They measure it Source: Central Bank, Treasury and Banks Association of reasonably well and devote Turkey. considerable management attention to 1 Positive values indicate that FX liabilities exceed FX assets. it. Internal exposure limits range 2 Defined as (Total Assets + Forward widely. Most banks accept the Central Purchases) - (Total Liabilities + Forward Sales) widely.mMostobankssacceptpthe Cventra in FX, including exchange indexed transactions. Bank limit, other self-impose even tighter constraints. Still others hold much larger position than sector average. Bank treasurers have ample access to, and familiarity with, on- and off-balance sheet instruments to hedge their risks and to adjust their position quickly, and markets are deep enough to assure that individual banks can change their risk position with no delay. Of course, the same level of comfort would evaporate if most or all banks were to move in unison in response to shocks. 70 See Celasun et al., cit. 93 284. Regulation now requires that banks keep their net open foreign-exchange position below 20 percent of capital. This limit, which was generously set at 50 percent in 1997, was reduced to 20 percent by September 1999, as agreed by the Turkish authorities in their letter of intent to the Fund. This reduction will bring Turkey closer to the international best practice level, ranging between 10-15 percent. 285. Interest rate risk. As a result of their large and increasing securities portfolios, banks in Turkey run considerable risks in terms of potential capital or income losses from changes in interest rates. A simple maturity gap analysis shows the significant capital losses that the banking sector may incur from given interest rate increases.'! To the extent that a bank can re-price its interest sensitive assets easily when its liability prices change--that is, by being able to liquidate the securities any time prior to maturity at the going price--the maturity gap tends to overestimate the true risk. However, for a bank to be able to realize its securities promptly, securities markets must be sufficiently liquid and deep. This is not so if all the banks try to shield their portfolio by selling at the same time, as is typically the case in anticipation of major shocks when banks have similar portfolio structures and share common expectations. 4 THE STATE BANKS 286. Structural Aspects Notwithstanding the expansion of private banks in Turkey since the inception of financial liberalization in 1980, public ownership in the banking sector still remains unduly large with one third of total banking assets in the hands of the state. Three of the largest five banks (Ziraat, Halk, and Emlak) are still state owned, covering together more than 29 percent of the sector's assets, Ziraat Bank alone, with its 16 percent total sector asset share, is large enough to be regarded as a benchmark by the other players in the market.72 287. State bank operations in Turkey are subject to the state-owned enterprise legal framework, which is inadequate to allow state banks to operate as commercial entities and leaves them especially vulnerable to political interference (even in daily operations). Profitability is not an operational objective for state banks and the continuing political intrusion in their business deprives them of the flexibility needed to adapt their structure and business practice to market changes. As in many countries, state banks in Turkey are an important instrument for public intervention in the economy. The Government uses them for a number of non-commercial objectives such as agricultural support, income redistribution, and industrial, urban, and physical infrastructural development. 288. The characteristics of the three main state banks are as follows: * Ziraat Bank provides most of its lending to farming and agro-industry activities (with more than 60 percent of loans extended at preferantial rates). With its nationwide and capillary branch network, the bank has a large deposit base and carries out activities on behalf of the Treasury such as budgetary transfers, agricultural commodity support programs, tax See the background paper by Bossone (1999) on the banking sector in this study. Vakiflar Bank is also a state bank, and the seventh largest bank in Turkey. However, it enjoys a rather unique status. It wvas originally established to manage the funds of the Turkish foundations; it was not given a mandate to support specific industries, and was granted full flexibility to operate as a private bank under private law. Vakiflar Bank engages in a large range of activities spanning from retail banking to investment financing, capital markets services, and investment banking. The state-owned banking sector includes also three development and investment institutions whose relative importance has declined over the decade. 94 collection, and paymaster services. Its management plans to expand commercial lending (including the growing consumer-credit segment) to 35 percent in 2000 from 20 percent in 1997. * Halk Bank is mandated by the Government to provide subsidized financing to small- and medium-size enterprises, craftsmen and tradesmen, and their cooperatives. About 27 percent of Halk's loans are at subsidized terms, and the bank's management has until recently remained committed to preserving the institution's unique role as a vehicle for preferential and subsidized credit to small business (especially in the remote and economically weaker regions of the country). Halk's large branching network allows the bank to access a large pool of depositors. Halk has also been used in recent years by the Government as a workout agency for troubled state banks. By end-1 998, Halk's duty losses represented 51 percent of its total assets. * Emlak Bank-the eighth largest Turkish bank in terms of assets-has traditionally operated as the real-estate state agency to finance construction and housing for low-income population groups at no profit, and to disburse funds provided by international development institutions. During the 1 990s, Emlak has diversified into foreign trade financing and personal banking, the latter providing a source of low-cost funding and low-risk earnings. In recent years, a large stock of unsold houses has produced heavy losses, weakening the bank's earning potential, and wiping out its free capital for future operations. The future of the bank is thus linked to cash injections from the public sector or to privatization. 289. Quasi-fiscal activities distort the profit orientation and resource allocation decisions of the state banks (and, indirectly, of the private banks as well, via the competitive forces). If preferential credit is excluded, the ratio of loans to assets of the state commercial banks has been considerably less than the private banks' ratio during Table 5.6: State Owned Banks:' Selected Financial Ratios the whole decade (16 Ratios 1993 1994 1995 1996 197 1998 1999 percent as compared to 40 Net Income/Average Total Assets 2.4 0.0 0.2 0.7 0.6 0.7 1.5 percent on yearly average). Net Working Capital 2/Total Assets -1.0 -4.2 -5.2 -4.1 -1.7 -2.6 -2.2 During the 1990s the Liquid Assets/Total Assets 37.6 36.9 32.5 35.2 28.1 24.0 28.4 profitability of the state Source: Banks Association of Turkey. commercial banking sector 'Excluding Development Banks has been low and 22Defined as shareholders' equity + current and last year's profit - permanent assets considerably below that of private banks, with only a 0.9 percent yearly average net income-to-assets ratio, against 3.4 percent of private commercial banks.;3 The sector's net working capital has remained negative throughout the decade and capital adequacy has diminished since 1995 (from 13 percent to 11.7 percent). State banks suffer from serious liquidity problems that have become especially critical in recent years, with a more than 7 percentage points drop in the liquidity ratio recorded since 1996 (table 5.6). `As will be clarified below, the true income generation capacity of the state banks is much weaker than what appears from the banks' income statement. Such weakness is largely concealed by the high capitalization of receivables from Treasury, which has provided both Ziraat and Halk banks with an artificial income support mechanism and has progressively eroded their financial discipline. 95 290. The Problem of "Duty Losses " Ziraat and Halk banks face unrecovered costs from duties carried out on behalf of the Government, including agricultural commodity and small business support, loan write-offs from farmers, disaster relief operations, and rescue of insolvent banks.,, The stock of accumulated Table 5.7: Duty Losses of State Banks receivables of Ziraat and Halk Bank combined (also called duty losses) 199s 196 1997 1998 1999 TL Trillion reached almost TL 11 quadrillion in Ziraat Bank 92.7 409.6 945.6 2,395.6 6,123.8 1999 or 13 percent of GNP, from 2.2 Halk Bank 76.5 221.8 570.5 1,586.6 4,232.0 percent in 1995. Such a rapid growth Total 169.2 631.4 1,516.0 3,982.3 10,355.8 is due to the snowball effect of US$ million interest rate capitalization on the stock Ziraat Bank 1,518.0 3,809.6 4,618.1 7,660.6 11,338.3 Halk Bank 1,252.7 2,063.6 2,786.2 5,073.7 7,835.6 outstanding (see below). Ziraat Bank Total 2,770.7 5,873.2 7,404.3 12,734.3 19,173.9 accounts for 60 percent of the total Share in GNP duty-loss claims (table 5.7 and figure Ziraat Bank 1.18 2.73 3.22 4.48 7.83 5.2). Halk Bank 0.97 1.48 1.94 2.96 5.41 Total 2.15 4.22 5.16 7.44 13.24 291. Partial payments were made Source: Treasury and World Bank estimates. by the Government to Ziraat and Halk from 1995-1999, adding up to 10 percent of the total receivables. In 1999, 15 percent of the stock outstanding at end-1998 was converted into 5-year maturity non-marketable securities paying annual coupons equal to the CPI plus a 10 percent premium, with coupons to be paid in cash. In 2000, the principal of the Figure 5.2: Duty Losses of State Banks securities issued will be rolled over ; and a new securities issue will take 14 place against 15 percent of the stock of duty losses outstanding at 12 end- 1999. 110 ,Z 8 292. The handling of the duty- loss claims has produced perverse a 6 incentives that have weakened the 4 capital generation capacity and the operational and financial discipline 2 of the banks involved, resulting in a deterioration of their efficiency 1995 1996 1997 1998 1999 level. Since the duty losses were * , ,- , * {3~~~~~~~~~~ Ziraat Bank V Halk Bank incurred, the corresponding claims __ _ Bank = akn on the Government have been reporthe Govn t han baln Source: Treasury and World Bank. reported on Ziraat's and Halk's balance sheet as performing assets accruing interest income. The interest rate on such claims has been set periodically by government decree, at levels beyond market reference rates. 74 According to Turkish law, state banks are to be compensated for costs sustained on behalf of the Government; however, since the Government does not provide for ex-ante budgetary appropriations to cover for such costs, bank claims for expenses related to quasi-fiscal duties need to be audited. The Govemment's repayment obligations become open-ended commitments and compensations are paid with several year delays. 96 293. The mechanism for interest rate determination on the duty losses has led to a situation in which the costs of inefficient bank operations have been concealed, and where new inefficiencies could originate unchecked and be covered with new government IOUs. Thus, the mechanism has de facto removed the banks' budget constraint. The banks have used it to expand their deposit liabilities at will, and to finance their operations with no concern for the associated costs and risks. The state banks have been in a position to access the deposit and interbank markets and afford to pay virtually any interest rate premium since the resulting expenses would always be recovered under the duty loss reimbursement mechanism (see section 5.4 for further discussion). 5 NEW REFORM PACKAGE 294. The government launched its financial sector reform program with approval by Parliament of a new banking law in June 1999. The law established the Banking Regulatory and Supervisory Authority (BRSA) as an independent agency that will combine supervisory functions which are currently split between the Treasury and the Central Bank. However, the law did not provide the agency with complete independence over the entry and exit of banks, or over changes to the prudential regulations. In order to correct these weaknesses, as well as establish clear procedures for problem bank resolution, Parliament approved a series of amendments to the banking law in December 1999. The amendments also tighten restrictions on insider and connected lending. The overall objective of the legislative and regulatory reform effort has focused on bringing the regulatory and supervisory regime for the Turkish banking sector up to the level of international best practice in line with Basle and EU standards. 295. The financial sector reform program encompasses consolidation of the private banking sector and restructuring of the state banks leading to their privatization over the medium term. Immediately following enactment of the banking law amendments, the government launched a bank resolution operation involving the takeover of five insolvent private banks by the Deposit Insurance Fund and the liquidation of one small investment bank. Through this operation, the government is moving to correct weaknesses in the private banking system and to send a clear signal that the regulatory framework will be vigorously enforced from now on. The state banks will be operationally restructured and their financial position strengthened in order to improve efficiency and put management on a commercial basis. The restructuring plan will cover a workout with Treasury of the stock of accumulated duty losses. 5.1 Strengthening Banking Regulatory and Supervisory Infrastructure 296. Prior to the recent banking reform, the primary responsibility for prudential regulation and supervision of banks rested with: i) the Undersecretariat of the Treasury, in charge of drafting regulations and conducting off-site bank monitoring; ii) the Board of Sworn Auditors, responsible for on-site examinations; and iii) the Central Bank of the Republic of Turkey, supervises the financial positions of banks through its off-site surveillance system. Regulators and supervisors from all three institutions are technically competent and professionally capable, and enjoy a high reputation across the banking community. They are familiar with international practices and have succeeded in advancing Turkey's compliance with international standards of banking supervision. The previous arrangements, however, and in particular the fragmentation of responsibilities did not allow for best use of human and institutional resources. Also, the effectiveness of supervisory action was severely affected by the large degree of political 97 interference. Political involvement was a major obstacle to timely and necessary action on difficult banks and to the appropriate resolution of problem banks. 297. Creation of a new independent regulatory and supervisory authority The new banking law mandated the creation of a new independent Banking Regulation and Supervisory Agency (BRSA), governed by a seven member board. The BRSA takes over the bank regulation and supervision responsibilities previously fulfilled by the Treasury, Board of Sworn Auditors and Central Bank 75. Unifying the separate banking regulatory and supervisory functions under a single agency is a major milestone, provided that the BRSA will be able to achieve and maintain the necessary autonomy from the political sphere. In this respect, it will be crucial for the BRSA to earn a strong reputation. The higher its credibility, the quality of its assessments, and the visibility of its views, the more difficult it will be for politicians to resist the BRSA's recommended action. The new institution will have to recruit quality staff and fill the large gap in manpower resources currently affecting the supervisory offices, especially at the Treasury. 298. Upgradedprudential regulation The government has taken action to improve prudential regulation. The previous regulatory regime for banks contained several serious flaws that almost certainly have contributed to the emergence of a relatively large group of problem banks in the banking system. The government has taken steps to correct serious flaws concerning the excessively weak loan loss provisioning rule and the overly lenient large exposure and connected lending limits. Strong remedial action in these two areas is a precondition for the success of the overall banking sector reform program. Other related areas where additional regulatory action is underway include: (i) capital adequacy; (ii) foreign exchange exposure; (iii) risk management; and (iv) accounting standards applicable to banks for prudential reporting and financial disclosure purposes. 299. Loan loss provisioning The government issued a revised loan classification and specific loan loss provisioning rule in December 1999 that puts an end to the previous practice of regulatory forbearance in response to real sector pressures. The new rule will require banks to classify their exposures primarily on the basis of creditworthiness considerations, although debt service track record considerations will also be taken into account. The banks are given a period of four years to build up the increased loan loss provisioning balances mandated by the new rule for all loans outstanding prior to the date of the latest amendment to the banking law (i.e., late December 1999). The upgraded provisioning requirements will apply immediately to all new and rescheduled loans. 300. The new loan loss provisioning regulation has also altered the necessary supporting documentation, including: (i) revised regulatory report formats; (ii) revised disclosure requirements, mandating immediate disclosure of the full amount of specific provisions required, whether already set aside or not, for banks' total loan portfolios, and (iii) a communique detailing borrower creditworthiness criteria and providing practical guidance to the banks on the new loan classification structure. Building on these results, the authorities should seek further progress in the area of loan classification and provisioning rules to (i) introduce a 5 percent specific provisioning requirement for watch loans; (ii) lift the exemption for agricultural support loans; (iii) include a specific provisioning requirement for equity exposures; and (iv) introduce The Central Bank will continue to regulate and supervise reserve and liquidity requirements and foreign exchange activities. 98 more prudent collateral classification. Furthermore, full tax deductibility should be gradually introduced for the provisioning of loan losses in line with international best practice standards. 301. Connected / insider lending limits Another important regulatory area that needed considerable strengthening was the control and monitoring of large and connected lending. With the December amendments, tighter limits were imposed on both on- and off-balance sheet commitments to related parties--especially companies belonging to the same group. However, these limits should be further tightened and gradually brought in line with applicable EU directives. Also, a clear mandate should be assigned to banking supervisors to monitor the evolution of this type of risk in individual banks and to evaluate the lending criteria underlying risk concentration observed vis-a-vis related parties. 302. Capital adequacy andforeign exchange exposure rules on a consolidated basis The Turkish capital adequacy rules in recent years were substantively in line with BIS standards except for the following: (i) there was no capital adequacy requirement applicable on a consolidated basis (e.g., for banks and their financial subsidiaries combined), meaning that the regulatory authority was not in a position to assess the capital adequacy of financial groups and the potential of any weaknesses in group capital to affect parent banks; and (ii) there were no capital charges for market risks (securities price fluctuation risk, foreign exchange risk, and interest rate risk). In December 1999, the government issued a first revision to the capital adequacy rule allowing its application on a consolidated basis (as of July 1, 2000). But in view of the sophistication required on the part of the banks to comply with new rules for calculating market risk charges, the government intends to introduce such charges later during the year 2000. The consolidated financial reporting requirement should be amended further to: (i) allow quarterly verification of banks' compliance with the consolidated capital adequacy requirement and (ii) extend coverage of consolidated reporting requirement to horizontal conglomerates. 303. Correcting the incentives for banking in Turkey also requires improving the current safety nets. The full deposit insurance coverage created some imprudent banking practices and competitive distortions as a result of the moral hazards involved, and it encouraged weaker banks to expand their deposit base by offering above market interest rates. The government postponed the reduction of this coverage while it was taking over the weak banks in order to avoid a serious erosion of depositor confidence, and possibly system-wide bank runs. Finally, a plan for gradual reduction of deposit insurance was put into operation in June 2000. Full deposit insurance is now limited to balances of TL 100 billion (equivalent to US$160,000) or less. The coverage will be reduced to TL 50 billion (US$80,000) in 2001. The deposit insurance limit will be brought into compliance with applicable EU directives by 2002. There should be a strong commitment by the government not to override the new limit. This commitment is needed to eradicate the public's perception that, in the event of a crisis, the government would be ready to raise the deposit limit on full insurance coverage. 304. It is important to ensure maximum coordination between the overseer of the securities clearing and settlement system, the Capital Markets Board, and the banking supervisory authorities, both for the design and development of the oversight framework and for the day-to- day monitoring of the clearing and settlement process. Securities settlement in Turkey could be further strengthened by having market participants enter into liquidity- and loss-sharing arrangements that would help the system prevent or withstand the delivery failures from large 99 participants, thus reducing the systemic risk. Also, consideration should be given to giving the BRSA the opportunity to have a role in the oversight of the securities clearing and settlement system, in close coordination with the Capital Markets Board. Precisely because of the dominant role of banks in all financial market segments in the country, it is unclear whether separate institutions for the supervision of banking and capital markets represents the optimal solution for the country. 305. Cooperation also will be necessary between the BRSA and the Central Bank-if necessary, through a protocol or memorandum of understanding-in so far as the oversight of the domestic payment system requires well-established and fluid channels of communications and information exchange between the two institutions. 5.2 Resolution Strategies for Problem Banks 306. The government has started to develop an effective and orderly strategy to deal with problem banks in order to maintain confidence in the banking system and ensure that the problems will not escalate into a systemic crisis, with serious consequences for the entire economy. In the 1 990s, the Turkish banking system experienced several bank failures. In 1994, three small banks were closed and sent into liquidation. From 1997 to early 1999, ownership of three more banks was transferred to the Saving Deposit Insurance Fund (SDIF), thus fully protecting depositors and creditors. A large number of additional banks also faced financial difficulties as a result of macroeconomic volatility, hardening of external conditions, and regulatory and supervisory tolerance. 307. The banking law amendments of December 1999 enhanced the legal and supervisory framework for problem bank resolution. To ensure a smooth resolution process, the new amendments gives the BRSA and SDIF the authority to take over all insolvent banks. The amendments also provides the two institutions authority and responsibility to restructure a problem bank to facilitate its sale in full or in part, or to liquidate the remainder based on existing laws. Immediately after the banking law amendment was approved by Parliament, the authorities intervened in five deeply insolvent banks and transferred these banks to the SDIF for resolution using the enhanced resolution mandate under the new rules, and revoked the license of a sixth bank to be liquidated. 308. One crucial regulatory area that needs improvement is pre-emptive action. The BRSA board should define operational guidelines that would identify bank performance criteria on the basis of which bank management would have to take corrective measures within a given period of time and be subject to penalties for inaction. It is important that action be taken while bank capital is still positive so as to prevent problems from becoming worse at later stages and to fend off possible systemic repercussions. 309. The BRSA should develop and implement a special supervisory regime for problem banks. This regime would include intensified supervisory efforts to bring about the corrective actions necessary to rectify problem banks before they deteriorate into insolvency. These actions should include: (i) timely and formal identification of the banks as problem banks; (ii) more and frequent off-site and on-site examinations to follow-up on identified weaknesses; (iii) regular meetings with the banks' boards of directors; (iv) frequent reporting from the banks to the bank 100 supervisors regarding the status of identified weaknesses and corrective actions; and (v) the use of enforcement measures to establish discipline and incentives for appropriate corrective action. 310. Failure resolution and the role of the SDIF. The amended banking law established a clearer framework for resolving failed banks. Previously, the role of the SDIF under the law was institutionally very complex and very broadly defined as the primary liquidity provider to illiquid banks, capital provider to insolvent banks, and generally to act as the banking rehabilitation agency. This definition has led to the undesirable outcome where it was rather easy for a commercial bank to enter the SDIF with all the incentives supporting such an entry, but where no clear and quick exit strategy for the banks to be removed from the SDIF existed. Thus the banks tended to continue to stay in the SDIF till such time a buyer could be found. 311. The SDIF should undertake the resolution of banks under the principle of "least cost resolution", that is at least cost to the SDIF in order to ensure the protection of the insured depositors as envisaged under the banking law. Under Article 14 of the banking law (after its amendment), the losses suffered will be applied immediately to the shareholders' capital and the shareholder rights will be revoked in all cases. 312. Further improvement is needed in the area of bank resolution. The SDIF should-be granted further operational flexibility to liquidate insolvent banks in the speediest and least costly manner, for example, through purchase and assumption transactions and insured deposit transfers. Still not addressed by the new banking law is the absence of a bank liquidation law separate from the bankruptcy law, which would enable the SDIF unilaterally to exercise options such as insured deposit transfers and purchase and assumption transactions. These options would considerably improve the bank loss resolution process. 313. The role of lender of last resort should appropriately be the role of the Central Bank, as envisaged under the current Central Bank law, whereby the Central Bank can provide up to two times the capital of a bank in temporary liquidity difficulties. Banks that remain illiquid in spite of the liquidity support provided by the Central Bank should be declared insolvent and transferred for resolution as "failed banks" to the SDIF. The December 1999 banking law amendment incorporated these principles in the law. 314. Resolving the present pipeline of problem banks The cost of full rehabilitation of all the taken-over banks has been estimated to be US$6-7 billion, whereas the SDIF has available only about US$1 billion. At the same time, it is critical to move quickly as the total cost is growing monthly. On the basis of detailed audits of each bank, the SDIF is expected to set forth a time- bound plan for the resolution using the least cost of method. It is likely that the government will issue securities to cover the cost of the resolution plan. 5.3 Reforming the State Bank Sector 315. Reforming the state bank sector in Turkey must be a priority if the country's banking system is to gain the necessary strength and efficiency. The reform process will have to be sequenced, well phased, and steadily implemented. The reform should combine resolution of the duty-loss problem with the commercialization and privatization of the state banks. Solving the duty-loss problem alone would not accomplish much if the governance of the banks was not 101 reformed as well. To achieve this latter objective, steps are needed to be taken to enable the state banks to operate under commercial incentives and to build up enough franchise to attract private buyers. Box 5.1: Bank Privatization: Some Relevant International Experiences Relevant examples of bank privatization in emerging economies include the cases of Argentina, Hungary, and Poland. In Argentina, bank privatization preceded restructring. The basic strategy adopted in 1994-98 to privatize eighteen of the twenty-six provincial banks was first to place the active assets in the entity to be privatized and then to match those assets with liabilities, up to the point where the entity's net worth met Argentina's capital requirements. The remaining assets and liabilities were transferred to a separate agency. As residual assets were not enough to cover residual liabilities, the provinces were enabled to borrow the needed extra cash from a government fiduciary fund, developed with the assistance of the Bank and the Inter-American Development Bank. In this way, short-erm obligations could be converted to longer terms. Also, as political opponents to the privatization program succeeded in having branching and labor restrictions imposed on the new banks, these were granted concessions including long-tern contracts to provide banking services to the provinces and guarantees on the quality ofthe acquired assets. Althbough the post-privatization period is still short, initial indications show that the situation has improved in most provinces. Privatized banks still largely rely on their service contracts with the provinces to generate a high share of their income while the transition to commercial lending is proving difficult. Finally, although the sustainability of the new banks will be uncertain concem for yet some time to come, privatization has lifted a considerable fiscal burden from the provinces. In Hungary, bank restructuring preceded privatization, albeit in a lengthy and - under many respects - inadequate way. In April 1992, the new bankruptcy legislation led more than ten thousand companies to begin restructuring or liquidation procedures (an almost ten-fold increase over the number of filings in 1991). Since these loans were to be classified as bad, the bankruptcy act triggered a rise in provisions that put banks under heavy strains. In December 1992 the Government opted for a portfolio clean-up operation whereby banks could swap bad loans for state financial instruments at a discount price. Once removed from the bank balance sheets, the bad loans were to be placed with a government agency that would arrange for their workout. To finance the operation, the Government issued long-term (20 year-) consolidation bonds. But the consolidation did not help much, as only about one-third of the face value of the bad debts was replaced. It took too long to resolve bad loans issue. The Government then decided for a "bank conciliation" program, involving bank recapitalization in exchange for banks submitting a medium-term restructuring plan, including a debtor consolidation component, and a privatization plan. In fact, enterprise indebtedness failed to be addressed as the main issue of bank conciliation, since banks lacked incentives to engage in debt restnrcturing, and the vast majority of debts had to be written off. Yet, banks eventually got rid of most non- performing loans and their portfolio improved, opening the way to privatization. A much more efficient, privately run, and open domestic banking system has emerged as a result In Poland, bank restructuring and the form:idable problem of bad loans were dealt with simultaneously during the early nineties. The authorities designed a bank-led bad-loan workout program, involving funds for bank recapitalization and a three-year strategy for privatizing state banks. A large part of the USD I bn fund provided by international donors in 1990 for the stabilization of the zloty was made available to finance bank recapitalization. State bank managers and employees were given the right to buy a certain amount of shares at half the market price if the banks were privatized. The program was'designed to reserve a prinmary role for banks in controlling enterprise governance, avoid unloading debt onto the government, and resolve debt issues without triggering unnecessary liquidation. Banks were given temporary quasi-judicial powers to lead out-of-court conciliation agreements. In this regard, Poland adopted a decentralized approach as opposed to a centralized approach centered on a govemment agency. As a result of the comprehensive reformn program, the main regional banks have been privatized, the commuercial banks participating in the program have a more than adequate capital base, and problem loans are fully provisioned. Also, workouts were actively pursued, competition in the banking industry has since been increasing, and the industry has gained in efficiency and profitability. Source: Bossone (1999) background paper for the CEM. 316. While Vakif and Emlak banks can be privatized in a short time with little or no intervention in their structure, Halk Bank and especially Ziraat Bank require significant 102 reorganization before being brought to the market.76 They should first be removed from the state- owned enterprise legislation and transferred to the governance of the Treasury. With the necessary assistance, the Treasury should then undertake the restructuring needed to prepare the two banks for privatization. In particular, their portfolio should be re-evaluated, their capital base strengthened, and their business strategies re-profiled on a commercial basis. Also, their internal systems of accounting, loan assessment, and risk management should be improved, and their staffs and branch networks rationalized. The banks should charge fully commercial fees for any services provided to the government or to the public on the government's behalf. Legislation to allow for privatization of Vakifbank has already been enacted. The government is currently working on a legal framework for commercialization and eventual privatization of the other three state-owned banks (Ziraat, Halk and Emlak). 317. International experience shows that there are two main prerequisites for successful privatization of the two major Turkish state banks. One is to identify strategic partners that would take a significant stake in the privatized institution and endow it with know-how, new managerial capacity, and a new business vision. In most cases, this is accomplished by attracting foreign institutions that have a strong interest in entering the domestic market. The second is to design an incentive structure that would align the interests of the management and staff of the would-be private institution with the success of privatization -- for instance, by issuing stock options that could be exercised at privatization (box 5.1). 5.4 Resolving the Duty-loss Problem 318. The effect of the duty-loss problem on the quality of the asset portfolio and efficiency of Ziraat and Halk Banks is such that their future viability and the prospects of successful privatization depend critically on resolving the problem. The resolution plan must have two components: preventing new duty losses aside from interest accruing on past duty-loss claims, and managing the stock of past claims outstanding. The first component can be addressed by ensuring that all new quasi-fiscal tasks executed by the banks on behalf of the government be fully and transparently funded through the consolidated budget. In this regard, the government has already taken action to phase out credit subsidies and has committed to fully budgetize the flow of duty losses associated with subsidized lending by Ziraat and Halk in 2002 As part of the its commercialization, Ziraat Bank should start charging appropriate fees for duties performed on behalf of the Treasury. 319. The stock of duty-loss claims should be converted into marketable government securities bearing a market interest rate as the second component of the resolution plan. This conversion into marketable securities will break the open-ended commitment of the government to cover the banks' capital imbalances through ad-hoc duty-loss reimbursements. It will also break the banks' dependence on high cost borrowing in the inter-bank market for liquidity. As the duty- loss claims are transformed into marketable securities, the current interest rate mechanism should be replaced with a fully market-determined interest rate. Currently, the yield on the duty-loss stock is computed quarterly based on the average of monthly treasury bill rates plus a different " Emlak's balance sheet is still loaded with a stock of unsold houses. Yet, the problem is manageable and does not constitute a major obstacle to Emiak sale. 103 spread for Ziraat Bank and Halk Bank.,, This ex-ante determination of the interest rate margins moves some distance toward introducing more discipline (or a "less soft" soft budget constraint) for both banks. Moving to fully market determined interest rates will complete the process and impose a truly hard budget constraint. Once implementation of the duty-loss resolution plan is complete, no extra interest margin should be paid to the banks over and above the securities' market interest rate 320. The conversion of the duty-loss claims into market securities should satisfy certain objectives. In particular, the securities issued should: (i) carry conditions and incorporate options that would give the banks enough flexibility to convert part of the debt into liquidity as necessary to fund their operations and based on their liquidity/return preferences; and (ii) carry enough incentives to induce the banks to hold the securities in their portfolio as a profitable source of income allowing the government to extend the debt maturity as much as possible. Such features could be incorporated in a long-term callable government bond, earning an appropriate premium over comparable (international) benchmark rates. The bond should be fully marketable and give the banks the possibility to raise cash when needed. At the same time, the interest on the bond should give the banks enough incentive to economize on their liquidity needs and to retain the bond in their portfolio. 321. The conversion of the duty-loss claims will generate strong positive incentives for both institutions and contribute to the restructuring and privatization prospects of the two banks by improving the quality of their assets, liquidity, and overall performance in terms of income/liquidity mix. The conversion into securities with market interest rates would eliminate the perverse interest income effect incorporated in the current duty-loss debt management practice and force the banks into stronger financial discipline. Following the conversion, Ziraat and Halk would have to manage their securities holdings based on their liquidity needs and income targets. They would have the option to use the new available liquidity to reduce their deposit rates and even to refund depositors and shrink the size of their balance sheet. The new instrument could become a market benchmark for securities pricing and could eventually contribute significantly to the development of Turkey's long-term securities market. 5.5 Improving Risk Management 322. Addressing the macroeconomic imbalances will reduce the high profits that currently attract banks to financial trading and arbitraging, and will encourage them to take on larger credit risk positions. In a more stable economic environment, banks will have a greater need to diversify their activities and will re-intermediate themselves in the real sector financing business". In that event, some of the current excesses could eventually be reabsorbed, although new types of risks could arise. 323. In the meantime, the authorities should encourage banks--if necessary, through regulatory measures--to diversify the composition of their portfolio liquidity so as to avert the risks associated with large and symmetric concentration of investments in government debt. Bank portfolio should include a broader range of liquid instruments such as deposits with The spread over treasury bill rates was 35 percent for Ziraat in the first half of 2000. It has been lowered to 26 percent. The same spread for Halk Bank was 21 percent in the first half of 2000. It has been increased to 27 percent. 7 Ersel (2000) background paper for the CEM. 104 international primary custodians, OECD government bonds with high rating, and standbys issued by foreign (highly-rated) banks. 324. The authorities should also ensure that individual banks strengthen their management policies and frameworks in the areas of credit and market risks. Establishing internal systems for risk measurement, monitoring, and control, as well as for risk analysis and reporting, will have to be a priority. Risk evaluation and reporting will have to be carried out on a consolidated basis. The supervisory authorities should give high priority to improving the quality of risk measurement reported by the banks. Finally, the supervisory authorities--supported, if needed, by external assistance--will have to ensure that the technical framework adopted by each bank is complemented by an adequate internal organizational setup for the effective implementation of risk management and operational procedures and policy decisions. The inflow of long-term capital in the domestic financial sector should be strongly encouraged and domestic banks should be encouraged to open their capital to the participation of foreign partners. 5.6 Adjustment to a New Incentives Regime 325. Macroeconomic reforms are changing banking sector incentives from investment in risk free government securities to real risk based banking. Incentives correction primarily rests on adjusting and stabilizing the economy. Clearly, the distortive effects of high real interest rates on bank portfolio decisions can only be cured if inflation, inflation uncertainty, and the fiscal deficit are all credibly and perrnanently reduced. While macroeconomic stabilization will require time to bring about acceptable results, its short-term impact could raise considerable financial stability problems for the banking system. Banks might likely have to face a difficult transition period before they could gain the ability to compete successfully under restored incentives. Once the economy is stabilized, bank competition will become more intense. This will allow for a better and more efficient banking industry eventually to emerge7". 326. In a CEM background paper, Ersel (2000) investigates the response of banks to changes in some fundamental macroeconomic variables in a simple framework. In this context, it is assumed that the stabilization program will affect: (i) the volatility of inflation, which stands as a proxy for overall uncertainty, (ii) the pressure exerted by public sector borrowing on the financial system; and (iii) the rate of growth of GNP. Although a stabilization program will influence all these three macroeconomic variables almost simultaneously, for analytical purposes, they are separately treated in the study. The purpose of this exercise is to shed some light on bank behavior for the initial stage of the implementation of a stabilization program, i.e. during the period when banks rely more on their past experiences. The major findings of the paper can be summarized as follows: 327. A decline in inflation uncertainty will induce banks to increase the share of credits in their assets at the expense of securities portfolio, which is consistent both with the theory and also with the experiences of the countries that implemented similar programs. In Turkey, the credit market, in contrast to the securities market, is dominated by private sector, therefore this finding indicates that a decline in inflation uncertainty will induce banks to divert from their 7 See Selassie (2000), 'Implications of Disinflation for Banks Profitability in Turkey", in Turkey: Selected Issues and Statistical Appendix, IMF and Van Rijckeghem (1997), "The Political Economy of Inflation: Are Turkish Banks Potential Losers from Stabilization ?" Istanbul Stock Exchange Review. 105 current credit rationing practice. On the other hand, such a decline in inflation uncertainty will also enable banks to change their liability structure by increasing the share of non-deposit funds borrowed. Since the major source of such funds is international financial markets, this result is not surprising, since foreign banks consider inflation uncertainty as a major negative factor in evaluating country risk. 328. A decrease in the public sector borrowing can be interpreted in various ways. Here, such a decrease is considered significant if it reduces the pressure exerted on the financial system and measured by the ratio of the change in public sector debt to the change in the broad money supply. In other words, this variable stands for the credibility of fiscal policy rather than a quantity restraint. Therefore, a positive development in this ratio is interpreted as moving into a better environment, and banks increase their exposures both to the public sector (increase in securities portfolio) and to the private sector (an increase in credits). Developments in the opposite direction induce banks to stay liquid, notably through increasing their deposits in foreign banks. 329. As can be expected, banks interpret an increase in the growth rate of the economy as a positive development and react by expanding the share of their credits in their assets. On the other hand, since foreign banks interpret such a development similarly, an increase in the growth rate of the economy enables Turkish banks to substitute deposits with non-deposit funds mostly borrowed from foreign banks. Finally, under such circumstances banks are able to expand their banking services, which is approximated by the off-balance sheet items/total assets ratio considered in the study. 330. The study confirms that banks will respond to successful disinflation by increasing the share of credits in their assets at the expense of their securities portfolio. This result is consistent both with the theory and also with the experiences of countries that implemented similar programs. On the other hand, such a decline in inflation uncertainty will also enable banks to change their liability structure by increasing the share of non-deposit funds. The new incentive framework emerging from the stabilization and structural adjustment , together with the new, tougher regulatory environment, will generate significant pressure on the banking sector to change including the likelihood of significant mergers and shakeouts. The result will be a much stronger, more competitive and resilient banking sector in the future. However, this transition will require active oversight by the authorities. 106 CHAPTER 6: MEDIUM-TERM PROSPECTS 1 INTRODUCTION 331. This chapter provides a quantitative framework for assessing Turkey's medium-term macroeconomic prospects under two broad sets of assumptions for economic policies. One, the Sustained Reform Scenario, assumes continued implementation of the current economic reform program including: (i) sustained fiscal adjustment to attain the inflation and growth targets and significantly improve Turkey's international creditworthiness; and (ii) decisive actions to address structural problems in such areas as public administration, social security, infrastructure, state enterprises (including state banks), and subsidy programs. Under the second scenario, the Truncated Reform Scenario, key policies are reversed starting in late 2000. 332. The core policies of the medium-term framework for the Sustained Reform Scenario are: * Sustaining the fiscal adjustment to yield a primary surplus of about 3 percent of GNP (chapter 1). * Continuing the aggressive disinflation program with appropriate action to moderate nominal price stickiness in the economy. * Fully implementing coherent and comprehensive reforms in public expenditure and administration, social security, agriculture, and finance (chapters 1, 2, 4, and 5). * Establishing credible legal and regulatory frameworks to promote private investment in energy and infrastructure (chapter 3). * Maintaining the strong privatization effort by selling majority stakes in major state owned enterprises and utility companies (chapter 1 and 3). 333. The key policy assumptions for the Truncated Reform Scenario are: The political and social consensus for strong economic program would weaken in the last quarter of 2000. * FFiscal policy would become expansionary with large salary and wage increases granted to civil servants and public sector workers, high agricultural prices granted to farmers, delays in public sector price adjustments; as a result the nominal exchange rate anchor would become nonviable. * Structural reforms in agriculture would lose momentum, as would the second phase of reform of the social security system; financial sector and infrastructure sector reforms would also go off-track. * The commitment to tackle the structural fiscal problem would fade quickly, causing the risk premium on government borrowing and real interest rates to increase once again, resulting in sharply higher interest payments and debt/GNP ratios. 2 PUBLIC SECTOR ADJUSTMENT 334. The challenge for public sector adjustment in Turkey has three aspects: (i) to reduce substantially the sector's burden on domestic financial savings; (ii) to improve the economy's efficiency and competitiveness; and (iii) to enhance the social development impact of public sector activities. Much of the recent deterioration in the fiscal position reflects the mounting 107 interest cost of meeting virtually all of the public sector borrowing requirement (PSBR) from fairly shallow domestic financial markets. To Figure 6.1: Public Sector Borrowing bring inflation and real interest rates down to Requirement as % of GNP moderate levels, Turkey needs to sustain a primary (non-interest) fiscal surplus of about 3 percent of GNP. It also needs to engineer a ___-_---_--_-----_----_-----_- --_-l__ sizable shift in the pattern of deficit financing ;2 toward privatization proceeds and external 2 sources, at least temporarily (chapter 1). ,. 15 335. Sustained reform scenario. Continuation s of the aggressive adjustment effort initiated in late - * 1999 is assumed in the Sustained Reform 19% ' 200 2W2 Scenario. This likely will have a substantial -,-_-itor_T--m __ payoff in direct financial benefits and greater Source: World Bank. credibility of the government's efforts. These, in turn, would engender a more robust private sector response. Overall, the pace and quality of fiscal adjustment would be expected to yield a permanent reduction in the PSBR, projected to decline by more than 4 percent of GNP in the first year of program. Further reductions in the PSBR Figure 6.2: Public Sector Primary Balance would bring it down to 4 percent of GNP by 2002. as % of GNP Structural reforms would proceed along a broad front, with the authorities moving rapidly in areas 4 _ ._-..__.__ -_ where the reforms are already under way 3 A l including agricultural support policies, 2 deregulation, privatization, social security, financial sector and public expenditure management. The sizable transitional costs of restructuring-as for social safety net programs (3) --__--__ and severance payments-would have to be 9 _ accommodated in the fiscal program. Public L _ __ ;; rM sector revenues would be expected to increase by Source: World Bank. about 4 percent of GNP in 2000, with most of the increase coming from personal and corporate income taxes. Indirect tax revenues would also increase, but rather moderately. Non-interest current expenditures would be unlikely to decrease, largely due to restructuring costs and the provision of higher quality services. Public investment would rise only gradually in the initial years of the program, as the public investment program was rationalized and efforts were made to strengthen implementation capacity. But it would increase more rapidly thereafter. 336. The viability of the fiscal program under the Sustained Reform Scenario depends on significant external financing and privatization proceeds in the next 2-3 years. Net public sector financing from abroad is assumed to be about 2.6 percent of GNP in 2000-01. With this financing mix and a sustained primary surplus of about 3 percent of GNP, domestic interest payments and the overall PSBR would decline steadily in GNP terms. As the domestic financial markets stabilized, there would be a gradual shift back to domestic financing, but the domestic debt to GNP ratio would not increase due to real GNP growth. 108 337. Truncated reform scenario. Under this scenario the government would be unlikely to achieve two key goals: a sustained primary surplus of 3 percent of GNP and a sizable shift in deficit financing toward privatization proceeds and external sources. Revenue enhancement would be expected to shrink gradually because temporary taxes introduced in late 1999 as part of the current stabilization program would not be replaced. Total revenues would decline from 27 percent of GNP in 2000 to about 24 percent in Figure 6.3: Public Sector Debt and Interest 2003, with most of the decline coming from Payments as a % of GNP corporate and personal income taxes. Indirect taxes would stay at about the same level for most of the period, with some decline toward the 'I *_---------- ------- 00 end due to a slowdown in economic activity. 10 I 338. On the expenditure side, recurrent __ outlays would rise sharply as real wages and salaries recovered some of the erosion suffered I 1 # 20 202 m00 in 2000. With the sale of some of the more - profitable SOEs (such as Telekom and TUPRAS I IReincit l --Thiabd Rebm krus0Pemen* refineries), budget transfers for the remaining --___________________________ _ sectors would rise to cover the continued losses Source: World Bank. of chronic deficit entities such as TCDD (railways) and PT (postal services). The social security system would not be enhanced with administrative reforms. Capital outlays would also be expected to decline slightly from their current low levels. Limited progress on the structural fiscal problem would boost the risk premium on government borrowing and real interest rates back to the levels of the late 1990s. The PSBR would deteriorate sharply (figures 6.1, 6.2, and 6.3). 3 INFLATION AND GROWTH 339. Under the Truncated Reform Scenario, inflation and growth performance would likely be disappointing. Without a large and sustainable Figure 6.4: Inflation fiscal adjustment, the conditions for sizable __ disinflation would not materialize. With limited program credibility and a continuing need for 8D significant domestic financing of the public 70 __ sector, inflation would soon accelerate. Faced with an acute tradeoff between inflation and ___ growth in a difficult political environment, 30 financial policies would likely continue to favor growth. But stop-go policies, high and volatile _______ inflation rates, and the threat of balance of 999 2960 2001 2002 payments difficulties would lower growth rates _ __ and make them more volatile. There is a severe Source: World Bank risk that inflation would climb back into triple digits (figure 6.4). 340. In an environment of high interest rates and inflation, private investment would increase only modestly, fueled largely by housing. Managing the balance of payments would probably 109 lead to periodic abrupt adjustments in overall macro policies, resulting in stop-go cycles in economic activity. As a result, growth would be curtailed sharply. Projected GNP growth of 2 percent in 2001-03 (figure 6.5) would be insufficient to prevent a further increase in Turkey's already high unemployment rate. Figure 6.5: Real GNP Growth 341. Under the Sustained Reform Scenario, conditions for a rapid reduction in inflation would be much more favorable. Credible fiscal policies backed by sustained structural reform 40- would facilitate continuation of the tight forward looking incomes policies introduced in 2000. (2.0: With a much-improved fiscal situation, more 43) + X availability of medium- and long-term external 2001 finance, and rapid and sizable privatization, 139 2DD 2001 2002 2043 there would also be much greater scope for continued use of the exchange rate as a nominal Wr ank. anchor. These conditions would help ensure a smooth shift to inflation targeting as the exchange rate band widens after mid-2001 and a reduction in inflation to single digits by 2002. 342. The economy would be expected to recover under the Sustained Reform Scenario, with growth of 5-5.5 percent projected for 2000 and 5-6 percent over the medium term.80 Fiscal adjustment, tight incomes policies and more stringent prudential regulation in the banking sector are likely to have a dampening affect on aggregate demand, but this would be balanced by declining interest rates and higher capital inflows. Specific factors underlying strong recovery under sustained reform include: * A sustained decline in real interest rates- Figure 6.6: Real per Capita Income ($US) which would stimulate consumption and _ _ _- __ investment demand. * Greater confidence in the policy - -- __-- framework, higher external inflows, 3300 including sizable external support from the 3A IFIs for the reform program and for 3 _ earthquake reconstruction. Z.4. The stimulus from earthquake 2,7I4 reconstruction. 33193 2004 20a1 2002 2003 6~~~~~~~~ Snobrd Reixm I Tmvaed RefDrm 343. Sustaining high growth over the medium ---- ---- term will depend on successful disinflation and Source: World Bank. rapid progress in implementing structural reforms. The proposed financial reforms are very important to sustaining a strong supply response, allowing the resources liberated by the fiscal adjustment to be channeled into productive activities. Faster GNP growth would permit both 80 It should be noted that the strong recovery already observed in 2000 is partly due to the low starting point from the recession of 1999. Growth in 2001 will be subject to in part to the specific disinflation path that the economy follows and could be lower than the 5-6 percent target. However, once the objective of single digit inflation is achieved after 2001, the conditions for sustaining high growth in the targeted range would be in place under the Sustained Reform Scenario. 110 more rapid growth in private consumption and a higher saving rate. By 2003 per capita income in constant dollars would be 18 percent higher than in 1999 and 13 percent higher than under the Truncated Reform Scenario (figure 6.6). 4 EXTERNAL BALANCE AND FINANCING 344. The primary challenges on the external front are (i) to manage external debt service payments of about US$75 billion in 2000-03, (ii) to finance the substantial current account deficits Figure 6.7: Sustained Reform: GNFS Exports projected under the Sustained Reform Scenario, and Imports Growth and Current Account and (iii) to ease the pressure on domestic finance by DP_ ___ __ __ mobilizing net external resources to finance the 10 _ ____ _ 3 budget. Under the Sustained Reform Scenario, -2 i growth would be higher than in the Truncated 4 - 1 Scenario. This would be supported by greater r0) i investment in productive activities and a higher , 2) (1) m inflow of export-oriented foreign direct investment. (4) -(2) Stronger export performance would support faster ( __ k (3) import and output growth. Due to modest (10) (4) exchange rate appreciation under the nominal 199 200 2X1 202 2 anchor and the demand-side effects of strong output Ex_ _0 in-mr- Imp" -CAB recovery, the trade deficit would widen and the Source: World Bank. current account would register substantial, but declining deficits over the medium tern (figure 6.7). The credibility of sustained reform would ensure that these deficits are financeable. 345. Under the Truncated Reform Scenario, the return to high inflation would likely cause the exchange rate anchor to become nonviable. Export growth would decelerate. After a sharp increase in Figure 6.8: Truncated Reform: GNFS Exports tend~~~ tosablz and Imports Growth and Current Account 2000, import growth would also tend to stabilize Deficit/GNP with macro policies geared to containing the trade ___ ___ deficit as the availability of external financing 3 30 declines in response to the loss of reform s 2.0 credibility. As a result, the current account deficit 4 would likely stay at or below around one percent of L f It GNP, after an initial increase in 2000 (figure 6.8). s2) .2 (1.0) Under the Sustained Reform Scenario export (4) ----- growth would be higher than in the Truncated ft1 -- (20) Scenario. The reason is that domestic demand (1O) - _ . 2 30) would be more constrained in the Sustained 10 2001 2003 Reform, and there would be a greater inflow of ___ __ - export-oriented foreign direct investment. Stronger Source: World Bank. export performance would permit faster import and output growth. Due to the (secondary) demand-side effects of output expansion, the trade and current account deficits would be somewhat larger than under the Truncated Reform Scenario. 111 346. Even with limited or no current account deficits, Turkey's medium-term gross external financing requirements will be very considerable. In the Truncated Reform Scenario, gross inflows totaling about US$70 billion would have to be mobilized in 2000-03. Net financing from official sources would likely remain negligible under this scenario, and raising this volume of funds solely through commercial credit would presents the authorities with severe challenges and risks. In the last few years, an important channel for financing the fiscal deficit has been inflows of short-term money from abroad into the banking system, with subsequent placement of these funds in government paper. Turkey's vulnerability to crises of confidence created by these short-term flows was highlighted in the 1998 Russian crisis, with outflows of US$6-7 billion. Without reform, much smaller fiscal deficits and lower inflation, the incentives for this sort of risky external borrowing by the banking system will remain high. 347. A credible external financing plan is envisaged under the Sustained Reform Scenario. The simulations call for gross inflows of about US$84 billion, with substantial fast-disbursing resources available from official sources during the initial years of the adjustment effort. With official financing in place and financial sector reforms under way to enhance the efficiency and stability of domestic financial markets, the commercial side of Turkey's external financing structure would be expected to improve rapidly. Strong foreign participation in the privatization process would be potentially a key instrument-as shown by Latin America's recent example. Foreign direct investment would be expected to increase rapidly. Normal access to international debt markets would be gradually restored and Turkey's sovereign rating would return to investment grade over the medium term. 348. In summary, a truncated reform in Turkey would risk a renewed financial crisis in the near termn and is not viable in the medium term. Under such a scenario, Turkey's creditworthiness would remain impaired, and the economy would be highly vulnerable to external payments difficulties. Economic growth would stay well below potential. Inflation would be dangerously high. The government's ability to reorient spending toward social and other development needs would be increasingly constrained. Conversely, the gains to the economy from comprehensive and robust disinflation and reform effort would be considerable and long-lasting. Turkey's dynamic private sector has repeatedly shown that it will respond to an improved macroeconomic environment. The prospect of closer ties with the EU has also created favorable business expectations. So there is every indication that a determined effort to address the root causes of the macroeconomic instability would elicit a very positive private sector response and set the stage for broad-based and rapid growth of incomes and employment. 5 RISKS FACED BY THE REFORM PROGRAM 349. This section discusses the key sources of risk facing the government's reform program. Its overall message is straightforward-the faster deep structural reforms are achieved, the better these sources of risk can be managed and eventually eliminated. The main risk to the program, as has been emphasized above, is that the reform effort might falter along the way. The government might decide to slow the implementation of scheduled reforms, or back away from them entirely. As has been highlighted in this report, such backtracking has proven fatal to previous reform efforts in Turkey and elsewhere. In the end, this boils down to a question of political will and the strength of the social consensus for reform. The governing coalition has 112 demonstrated its cohesion and reforn credentials over the past year. Continued political stability and reform leadership are linchpins of ultimate success of the program. 350. Of course, political support is conditioned in large part by social factors. These factors arise from the tight incomes policy and range of socially sensitive reforms included in the program such as the pension reform and the reform of agricultural support policies. In addition, stabilization, privatization and market deregulation measures may impact on established interests or adversely affect certain segments of the population. A group particularly at risk are workers in enterprises to be privatized. While these workers are, in general, not among the poorest segments of Turkish society, the living standards of their families could be severely disrupted by the reforms. As a result, privatization could be delayed and/or a more general backlash against the reform program might develop. Other social categories could be adversely affected as well, for example, households with fixed incomes and small farmers. The authorities have undertaken measures to strengthen the social safety net including the introduction of unemployment insurance and the direct income support program in agriculture. However, more up-front action may be needed to bolster the social safety net and maintain the social consensus for reform. 351. Assuming that the political and social foundation for the reform remains in place, several key economic risks come to the fore including the critical importance of ensuring the quality and sustainability of fiscal adjustment. While this point may seem obvious, it needs emphasis because there are plenty of examples of countries achieving numerical targets for fiscal reform through unsustainable adjustment.91 A simple example is reduced expenditure on maintaining fixed public investments. Eventually expenditure cuts in this area must be reversed or, if they are not undone, they will lead to increased capital spending requirements, or a lower return on investment and hence lower growth. Another example particularly relevant to Turkey is the possibility that broad expenditure cuts will fall disproportionately on social spending for health and education. This would damage the economy's future prospects by limiting investment in people which is a core determinant of growth (chapter 1), as well as by weakening social support for the reform effort. While the 2000 fiscal package relies heavily on short-termn measures and across-the-board spending cuts, the government intends to shift to more permanent revenue measures in 2001-02 combined with more fundamental measures to improve expenditure management. As has been argued above, international evidence suggests that fiscal adjustment is most successful when governmnents make progress on improving the expenditure side of their budgets.82 This suggests that Turkey's ongoing effort to design public expenditure reforms is on the right track.83 Over the longer-term, the strong emphasis on structural reforms will help ensure that the expenditure side adjustments are sustainable. 352. Another key risk arising from the stabilization program is that posed by the choice of nominal anchor, in Turkey's case the crawling-peg exchange rate regime. The exchange rate regime has been designed so as to provide a clear reference for price and interest expectations, while avoiding the drawbacks experienced by other countries that pursued exchange-rate based stabilizations. As with any exchange rate based stabilization the crawling peg implies a certain 81 A number of cases are presented in Easterly (1999) "When is Fiscal Adjustment an Illusion". 82 See Alesina and Perotti (1995) " Fiscal Expansion and Adjustments in OECD Countries" . 83 The govemment has initiated a reform of expenditure management and has requested World Bank support for a Public Expenditure and Institutional Review to assist in the development of an agenda for expenditure reform and improved public sector govemance. 113 loss of control over monetary policy. The advantage, of course, is that the exchange rate is an easily monitored signal of the government's commitment to its macroeconomic policy framework as a whole. The disadvantage is that, in order to meet exchange rate targets without excessive widening of the current account over the short run, the government may be forced to take additional fiscal tightening measures that go beyond what has been initially programmed with negative implications for growth. The present government's impressive track record on structural reform helps alleviate this risk by bolstering the credibility of Turkey's disinflation effort. Continued rapid progress in implementing structural reforms will counterbalance pressures on the exchange rate. It will also facilitate the envisaged transition to a more flexible exchange rate regime when a gradually widening band is introduced around the central parity rate starting in mid-2001. This pre-announced transition is an important risk-mitigating feature that will help Turkey avoid problems that other countries have faced with exiting from a nominal exchange rate anchor. Such a step is often interpreted by financial markets as a signal of policy reversal. By announcing its exit strategy up-front, the government intends to make sure that the policy of eventually returning to a more flexible exchange rate regime is clearly understood in advance. For more discussion of global experience with fiscal adjustment and exchange rate anchors, see box 6.1. Box 6.1: Fiscal Adjustment Under a Crawling Peg Stabilization comnes in a number of forms, usually classified as populist, orthodox-money-based, orhodox- exchange-rate-based, and heterodox.,, Tuy's macroeconomic famework would probably put it in the orthodox- exchange-rate-based category, meaning that the program is based on a combination of fiscal adjustment along, with a nominal anchor in the form of a targ band for the exchange rate. It has a touch of heterodoxy, in that the program includes a forward looking indexation scheme for public sector wages, but less emphasis is put on this aspet of the program. The experience of other countries with exchange-rate-based stabilization highlights some of the advantages of Turkey's chice as well as some of the risks that come with it. Classic examples arethe refrms of 1978 in the Southern Cone countries. Chile, Uruguay and Argentina. More recet examples of exchange-rate-based stabilization include Mexico in 1987 and Brazil's Real Plan of 1994. From the experiences of these countries and oters, which used alternative adjustment schemes, the overwhelming evidencet is that inflation cnot be stabilized without successful fial adjustment. So sustained fiscal aidjustment remains the key element required for success. Countries such as Turkey, with high, but not hypei-, inflation have a special problem in that inflation inertia due to indexed contracting is endemic in the economy. In countries with similar initial levels of inflation this has e*iqty resulted in th slow convergence of inflation to its ultimate targeted level, and, hence, real exchange rate appreciation. This real appreiation, in tur, has hurt comeitiveness and has led to expectations that the exchange rate regime would be aba in some way. These expctations, in turn, make the exchange rate regime hard to sustain. Policymakers should be aware of this risk. But they should also be aware that most failed exchange-rate-based stabilizations also involved slippage on fiscal targets. 353. A further source of stabilization risk involves the dynamic between wages and prices. Clearly, disinflation is not possible without wage restraint, which depends once again on the strength of the social consensus. In chronically high inflation countries, the problem is generally exacerbated by the formation of strong backward linkages between wages and prices, reinforced in many cases by explicit indexation mechanisms. While Turkey does not have a tradition of " This discussion is based on the one to be found in Pierre-Richard Agenor and Peter J. Montiel (1996) Development Macroeconomics, Chapter 8. 114 formal wage indexation, informal indexing has become deeply ingrained. The current disinflation program incorporates a shift to forward-looking indexation of public sector wages. Turkey's previous experience with such a scheme in the 1980s was mixed. Initial success was overturned when inflation surprises eroded real wages-union support for the scheme was lost. While such erosion is not inherently part of the program's design, there is the risk that if the inflation targets are not met, popular support for the overall reform effort could slip and unions could press for larger nominal wage increases. The private sector has been reluctant to follow the government's lead in shifting to forward indexation and this could also affect the pace of disinflation. To address this issue and establish a permanent forum for dialogue with its social partners, the government is relying on the Economic and Social Council which it plans to formalize with legislation later in 2000. 354. The structural reform program is subject to its own specific risks. In addition to the social dimensions mentioned earlier, these risks range from overburdening the capacity of the public administration to carry out reforms to underestimating the changes required to attract sufficient amounts of private capital and fulfill Turkey's growth potential. International experience over the past two decades has demonstrated conclusively the importance of comprehensive structural reforms and the limited impact of halfway measures. As this report has shown, Turkey's present economic problems derive in large part from unfinished or partial reforms in strategic sectors such as energy, and the reluctance of previous governments to tackle deep rooted structural problems in agriculture. While these issues are complex and difficult, Turkey can benefit from the international experience of other emerging markets that have undertaken deep reforms and are now enjoying high levels of private investment and growth. The experience of middle income countries such as Mexico in reforming their agriculture sectors can be a useful guide. As with stabilization, the risks on the structural side will diminish as the reform progresses, the supply response strengthens and the program gains credibility. A sustained increase in output will go a long way to maintaining and consolidating public support. 355. Finally, the degree of external support for the program is a critical risk factor. First and foremost, Turkey needs expanded markets for its exports and increased inflows of foreign direct investment. The current reform program will help integrate Turkey into the global economy and it is in the interest of the country's external partners to support this process. The Sustained Reform Scenario envisages capital inflows on the order of US$84 billion over the 2000-03 period. The lion's share of this is expected to come from private sources and will depend on sustained reform to raise external creditworthiness and attract foreign investment. At the same time, a significant percentage is expected to come from official sources, including Turkey's main bilateral and multilateral partners in addition to the World Bank and IMF. If provided in a timely manner and well sequenced with the pace of reform, an official financing will play an essential catalytic role in supporting the government's program and paving the way for the high levels of private capital inflows required to support the disinflation effort and achieve Turkey's full growth potential. 356. The government must closely monitor the risks to the program's success, but more importantly, it should focus on reform implementation-the more it achieves on reform the more maneuvering room it will have for macroeconomic management. The government is carefully tracking its short-term fiscal and monetary targets in coordination with the IMF. At the same time, it is closely monitoring interest rates and inflation, both of which act as signals of public 115 confidence. The government should also be vigilant for signs of overvaluation in the real exchange rate and in the current account. But most importantly, the government should concern itself with reform implementation-the more it achieves on reform the more maneuvering room it will have for macroeconomic management. 116 REFERENCES Akder, Halis, H. Kasnakoglu and E. H. Cakmak (1999) " Turkey Sources of Growth in Turkish Agriculture," CEM Background Paper, The World Bank. Alesina, Alberto and Roberto Perotti (1995)" Fiscal Expansions and Adjustments in OECD Countries," Economic Policy, 21 (October), pp 205-48. Alexander, Ian T. Irwin and L.Bertolini (1999), " Turkey Reforming the Infrastructure Sectors," CEM Background Paper, The World Bank. Atiyas, Izak, Yasemin Bal Gunduz, Ferhat Emil, Can Memduh Erdem, and Didem Bahar Ozgun (1999) " Fiscal Adjustments in Turkey: The Role of Quasi-Fiscal Activities and Institutional Reform," manuscript, Treasury. Berument, Hakan (2000) "Public Sector Pricing Behavior and Inflation Risk Premium in Turkey?" mimeo, Bilkent University. Berument, Hakan and K. Malatyali (1999) " Determinants of Interest Rates in Turkey," Research Department Discussion Paper 9902, The Central Bank of the Republic of Turkey, Ankara. Bossone, Biagio (1999) "Turkey: The Banking Sector, " CEM background paper, World Bank. Burnside, Craig (1997) " Government Debt and Policy Sustainability in Brazil," mimeo, Policy Research Department. Burnside, Craig and David Dollar (2000) "Aid, Policies and Growth," forthcoming American Economic Review, 90. Burnside, Craig, Martin Eichenbaum and Sergio Rebelo (1999) "Prospective Deficits and the Asian Currency Crisis," World Bank Policy Research Working Paper 2174. Burnside, Craig, Martin Eichenbaum and Sergio Rebelo (2000) "On the Fundamentals of Self- Fulfilling Speculative Attacks," NBER Working Paper 7554. Celasun, Oya, C. Denizer, and D. He (1999) " Capital Flows, Macroeconomic Management and the Financial System: Turkish Case, 1989-97", IMF Working Paper, February 1999. Corsetti, Giancarlo, Paolo Pesenti and Nouriel Roubini (1998) "What Caused the Asian Currency and Financial Crisis?" mimeo, New York University. Demirguc-Kunt, Asli and R. Levine (1999), " Bank-Based and Market-Based Financial Systems: Cross Country Comparisions, mimeo, The World Bank. 117 Easterly, William R. (1999) " When is Fiscal Adjustment an Illusion," Economic Policy, 28 (April), pp 57-86. Easterly, William R. and Ross Levine (2000) " It's Not Factor Accumulation" Stylized Facts and Growth Models," mimeo, The World Bank. Ersel, Hasan (2000)" Bank Response To Changes In Macroeconomic Variables In Turkey," CEM Background Paper, The World Bank. Gosh, A. and Phillips, S. (1998) " Warning: Inflation May Be Harmful for Your Growth," IMF Staff Papers, IMF Health Insurance Commission of Australia, (1995), "Health Financing Report," unpublished manuscript. Juris, Andrej (1997) " The Emergence of Markets in the Natural Gas Industry" Policy Research Working Paper number 1895, The World Bank. Kasnakoglu, Haluk and E. Cakmak (1999), "The Fiscal Burden and Distribution of Costs and Benefits of Agricultural Support Policies in Turkey", CEM Background Paper, The World Bank. Klenow, Peter and Andres Rodriguez-Clare (1997) " The Neoclassical Revival in Growth Economics" Has it Gone too Far?," NBER Macroeconomics Annual 1997, 12, pp 73-103. Lasagabaster, Esperanza, Alan Thompson and Shiyan Chao (1999) " Turkey Social Insurance Reform, " CEM Background Paper, The World Bank. Levine, Ross and David Renelt (1992) "A Sensitivity Analysis of Cross-Country Growth Regressions," American Economic Review, 82(4), September, pp 942-63. Loayza, Norman, Klaus Schmidt-Hebbel and Luis Serven (2000) "What Drives Private Saving Across the World?" forthcoming, Review of Economics and Statistics. Mankiw, N. Gregory, David Romer and David N. Weil (1992) " A Contribution to the Empirics of Economic Growth," Quarterly Journal of Economics, 107 pp 407-37. Meyer, DB.D. (1990), "Unemployment Insurance and Unemployment Spells," Econometrica, 58, 757-782. Ministry of Health of Turkey (1995-1996-1997) Country Report Ministry of Health, (1999), Health Sector Reforms in Turkey, mimeo, Ankara. Nash, John, (1999) " Turkey- Supporting Rural Development: Options for Refom" CEM Background Paper, The World Bank. 118 OECD (1996). Agricultural Policies, Markets and Trade in OECD Countries, Paris. OECD (1997). Agricultural Policies, Markets and Trade in OECD Countries, Paris. OECD (1998). Agricultural Policies, Markets and Trade in OECD Countries, Paris. OECD (1999). Agricultural Policies, Markets and Trade in OECD Countries, Paris. Queiser, M. (1998), "The Second-Generation Pension Reforns in Latin America," Working Paper, OECD. Polastri, Rossana (1999), "The Links Between Economic Fluctuations And Growth In Turkey," CEM Background Paper, The World Bank. Sak, Guven (2000) " Developing Private Pension Plans In Turkey: Institutional And Regulatory Framework, Supervision, Taxation And The Absorptive Capacity Of Capital Markets," CEM Background Paper, The World Bank. Selassie, Abe Aemro (2000), 'Implications of Disinflation for Banks Profitability in Turkey", in Turkey: Selected Issues and Statistical Appendix, IMF Social Security and Health Insurance Reform Project Study Reports. State Institute of Statistics of Turkey (1999), " Results of Household Labor Force Survey. Superintendencia de Administradoras de Fondos de Pensiones (1995), "El Sistema Chileno de Pensiones", Santiago de Chile. Uygur, Ercan (2000), "Erratic Growth And Private Investment Behaviour In Turkey," CEM Background Paper, The World Bank. rd The Financial Times (2000), April 23rd. The World Bank (1994) " Averting the Old Age Crisis: Policies to Protect Old and Promote Growth," The World Bank Policy Research Report, Oxford University Press. The World Bank (1998/99) "Knowledge for Development", World Development Report. The World Bank (2000) "Turkey: Economic Reforms, Living Standards and Social Welfare Study," Poverty Reduction and Economic Management Unit, Europe and Central Asia Region, Grey Cover Report. 119 Van Rijckeghem, Caroline (1997), "The Political Economy of Inflation: Are Turkish Banks Potential Losers from Stabilization ?" Istanbul Stock Exchange Review. 120 TECHNICAL APPENDICES CHAPTER 1: MACRO-ECONOMY Production Functions and the Sources of Growth 1. A simple production function approach writes output, Y , as a function of the stock of physical capital, K, employment, N , human capital, H, and factor-augmenting technology, A, say of the form Y = Af (K, H, N) .a5 If this function is constant returns to scale it can be written as y = Af (k,h,l) where y is output per worker, k is capital per worker and h is human capital per worker. Totally differentiating this last equation one gets (1) dy = f(k,h,l)dA + Afk(k,h,l)dk + Afh(k,h,l)dh. Dividing through by y one gets (2) dy dA + Afk(k,h,l)k dk + Afh(k,hjl)h dh y A y k y h If we make the further assumption that physical and human capital are paid their marginal products (i.e. factor markets are competitive) then (3) dy d dk dh y A Kk Hh where sK and sH are the shares of physical and human capital in national income.B1 Letting dz / z represent the growth rate of z notice that this means that the growth rate of output per worker can be written as a linear combination of total factor productivity growth, and the growth rates of physical and human capital per worker. Concepts of Sustainable Debt 2. The public sector's flow budget constraint implies that the change in the nominal value of its outstanding monetary and debt obligations will be the sum of its interest payments on these obligations minus its primary surplus. This is illustrated by equation (4) (4) B, - B,_1 + M, - M, = n,1 B,-l - S, where M, is the monetary base at the end of period t, B, is the public sector's outstanding debt obligations at the end of period t, n, is the nominal interest rate on these obligations, and S, is the nominal value of the primary surplus at date t. b' The functional form in this equation ignores the possibility that output can change because of cyclical fluctuations in factor utilization. As a result, the equation is best thought of as arising from a long run production function. e" An altemative interpretation of equation 3 is that sK and SH are the parameters on K and H in a Cobb-Douglas production function. 121 3. The first solvency condition described here requires that the primary surplus should finance the public sector's interest payments in the long run. That is (5) B, -B,1 + M, -Mt, =0 or S, = n_ B,-, in the long run. In other words, the public sector eventually receives no net outside financing. Turkey never satisfied this solvency criterion in the 1990s. The consolidated primary surplus was less than interest payments in every year from 1990 through 1998. In fact, as a fraction of GDP, interest payments exceeded the primary surplus by an average of almost 10 percent of GDP from 1990-94, but this gap narrowed to about 8 percent of GDP from 1995-98. 4. Requiring that the stock of debt plus nominal obligations not grow is probably too strict a solvency criterion. More generally one might consider steady states where the nominal debt grows at some constant rate, given by g8, and in which the nominal interest rate is given by a constant n, so that (6) St + M, - Mt, = =(n -9B)BI-l The first solvency condition simplifies to S, = nB,-l or M, - M, = -g3B,-, in such steady states. We will examine more general steady states using the second solvency criterion. 5. To determine the second solvency condition, start from some date 0 and iterate forward on the flow budget constraint until the following constraint is obtained (7) Bo = l B, ESi + Mi - Mii fl (I +ni) i=' f |(I +nj) i=O j=0 The second solvency condition is that the limit of the first term on the right-hand side of equation (7) is 0. In other words the present value of future primary surpluses and seignorage is pledged to completely finance the current level of debt. In this case (8) B E S, + M, -M,1 '=' (I + n,) i=O 6. Notice that the solvency condition amounts to a condition on the growth rate of the outstanding stock of debt. The geometric average of the growth rate of the nominal debt over time must be less than the geometric average of the nominal interest rate in order for the limit of the first term to be 0. 7. In one sense the second solvency condition is more restrictive than the first. This can be seen by again considering steady states in which the nominal debt grows at some constant rate gB and in which the nominal interest rate is n. It is clear that n > g, for the second solvency condition to be satisfied whereas the first solvency condition puts no constraint on the interest rate. On the other hand, there is a sense in which the second solvency condition is weaker than 122 the first because it allows the level of nominal obligations to rise. This arises from the fact that in such steady states the second condition only amounts to the flow budget constraint (6), while the first condition amounts to the more restrictive S, = nB,-l. 8. The second solvency condition can be reinterpreted as follows. Rewrite the flow budget constraint in a long-run steady state as (9) -S, =M -M_- l (g -n)B,_l Whenever the second solvency condition is violated, so that gB > n, this means that the primary deficit is so large that it cannot be financed by seignorage. Additional debt must be accumulated beyond what is needed to roll interest payments over into new debt. 9. To verify whether the second solvency condition holds is difficult because it requires information about expectations of future debt and interest rates in Turkey. However, it is straightforward to observe actual realizations of the growth rate of debt and the nominal interest rate. And, on average, the second solvency condition requires that the nominal interest rate exceed the growth rate of the debt. In the first half of the 1 990s, except 1990, the debt grew more rapidly than the nominal interest rate, with the growth rate of the debt averaging about 91 percent per annum, while the money market interest rate averaged about 84 percent per annum for 1991-94. In the second half of the 1990s, the debt has grown at a rate slightly closer to the nominal interest rate. From 1995-98 the debt grew at an average rate of 78 percent, while the market interest rate averaged about 73 percent. So Turkey appears to have moved toward a slightly more sustainable debt position according to the second criterion. 10. The third concept of solvency is that the public sector debt should not grow in size, relative to GNP, in the long run. This condition is more (less) restrictive than the previous one whenever the long run nominal interest rate exceeds (is less than) the long run nominal growth rate of GNP. Since the last concept of solvency will play a greater role in the analysis, it is useful to express the quantities in equation (4) relative to GNP: (I 0 bi bt- +Mt- M-] n-b-S,_st ( )~ ~ ' I+N,y I+g' I+v l+g "' where b, is the debt to GNP ratio at time t, m, is the monetary base relative to GNP at time t, s, is the primary surplus as a fraction of GNP at time t, and gN, is the growth rate of nominal GNP at time t. 11. Requiring a stable level of debt relative to GNP simply means that in the long run b, = b,, = b. Assume, further, a constant long-run nominal interest rate, n, and a stable growth rate of GDP, gN -Suppose the demand for nominal money balances is given by MID = P,Y,L(n, ), where Y, is real GNP, P, is the GNP deflator and L' < 0. This demand 123 function assumes a unit elasticity of money demand with respect to output. Then in the long-run steady state, money balances relative to GNP will be given by m = L(n), and equation (10) will imply (11) s+ mN m = (ngHN)b 1 +g 9N +gN where the primary surplus as a fraction of GNP is constant by implication. Specification of Money Demand 12. The money demand function used in the analysis is (12) Inln(,) = a + 0 ln(P,Y, ) - 77n,- In equilibrium money demand is equated to the supply of base money, M, . (12) was estimated by least squares by imposing a unit income elasticity of demand for money, i.e. a = 1. The series M, /P,Y was measured using the ratio of the monetary base to nominal GNP, and was deseasonalized using the multiplicative Xl I procedure in Eviews. It was then regressed on a constant and the nominal interest rate using data from 87Q1-98Q2 with two outlying observations excluded (94Q1 and 94Q2). The resulting estimates of a and q7, are a = -2.237 and q= 0.8609. 13. In the debt sustainability analysis it is assumed that the nominal interest rate is given by n, = r + E,fr,+,, where it+l is the future inflation rate. In this setting the equilibrium price level is given by (13) P,+ 17 (17 + E, [ln(M,+1 / ,+j )+ 7r -a] so that the equilibrium inflation rate is given by (14) , = l1[Et u, - g,+j )+ (AE, )ln(M,+ l / Y;+i)], where p, is the growth rate of base money and g, is the growth rate of real GNP. 124 CHAPTER 2: SOCIAL SECURITY REFORM 1. As discussed in the text, in the longer run, the Turkish Government will need to propose further reforms to the pension system to make it viable. The various options Table 1: Long-Term Reform Options include further reforms to the PAYG Option Par*uIar variant Of pton system, a move to a notional accounts Mild 2015 - retirement age is raised for new system, or a move to a combined system, PAYG entrants to 65 for both men and women; with some PAYG components and some Benefits unchanged; funded components. The pros and cons of Aggressive 2015 - retirement age raised for those under PAYG the age of 35 at the rate of I year each 2 years each of these further reforms can best be to reach the age of 65 for both men and illustrated by comparing simulations for women for men under the age of 25 and each of these options. It should be noted women under the age of 23; that each of these options have a countless 2015 - base benefits are gradually reduced for those under the age of 45 over a 20 year number of variations, and the particular period to 20% for the first 10 years instead of forms shown here are not necessarily the 35% variants that the Government may consider Notional 2015 - new entrants moved to notional at some future date or that the World Bank Accounts accounts system, indexed to GDP g rowth would necessarily advise the Government No required retirement age change to adopt. They are merely being used to Multipillar 2015 - retirement age is raised for new entrants to 65 for both men and women; illustrate the differences between the 2015 - 100% of people under the age of 45 options. switched to the new system, but with the old benefit formula applied proportionally to 2. The particular characteristics of the those years of service prior to 2015 four options are shown in the table below: 2015 - new system - 1/2 of contribution applied to funded system; reduce PAYG a mild PAYG reform, a more aggressive benefits for the switchers to 1% accrual for all PAYG reform, a notional accounts reform, _I years including the first and a PAYG reform with a funded pillar. Source: World Bank. The Government has indicated a desire to postpone any further reform until 2015. All the variants shown here take that concern into account. 3. The results shown in Table Table 2: Deficits in SSK under Different Long Run Policies 2 only look at SSK. Similar results (as % of GDP) Year current MMl Aggvesive Notional Multi- can be produced for all three Yskw PAYG PAYG Accounts pillar systemsmPl. PAG PY cons pla 2000 -1.0 -1.0 -1.0 -1.0 -1.0 4. As can be seen, the mild 2010 -.04 -.04 -.04 -.04 -.04 2020 -.5 -.5 -.5 -.4 -1.4 PAYG reform does not show any 2030 -2.0 -2.0 -2.0 -1.9 -2.6 impact until 2060, the first year 2040 -3.6 -3.6 -3.4 -3.6 -3.3 that the retirement age would be 2050 -3.6 -3.6 -1.7 -3.5 -2.6 affected. After a brief interruption, 2060 -4.8 -4.3 -2.4 -3.7 -2.3 where the deficit comes down, the 2070 -5.7 -3.8 -3.3 -2.5 -1.8 deficit begins to rise again, in 2075 -5.9 4.3 -3.6 -2.1 -2.0 response to the aging of the Source: World Bank. population. The more aggressive PAYG reform shows its impact as early as 2040 because the benefit changes start taking hold by 125 then. Once the retirement age increase begins, between 2040 and 2050, there is a sharp drop in the deficit. However, once again, due to longevity increases, the deficit begins to rise once again. 5. The notional accounts reform begin to take effect between 2040 and 2050, when the new entrants in 2015 and after begin to retire. Because the notional accounts reform adjust benefits to longevity increases, once this form takes hold, the deficit continues o fall. The benefits under this reform end up at only 37.5 percent of the original benefits for men and 35.7 percent for women by 2075. These benefits will continue to fall in response to the continued longevity increases. Since there are no actual investments taking place, the government has to decide how the interest rate to be provided on the balances will be determined. The selections so far range from Italy's growth rate of GDP to Poland's 50-50 weighted combination of wage growth and inflation, with Sweden's rate of wage growth and Latvia's rate of growth of the wage bill falling in between. It was assumed in these simulations that the national interest rate would equal the GDP growth rate in line with the current indexation of the wage base. 6. Furthermore, the government must decide how the pensions are to be determined. In all cases, the life expectancy at the age of retirement is used in the calculation of the pension. As life expectancy increases, the pension for individuals retiring at the same age with similar accumulated balances will automatically fall. The hope is that as individuals see the low level of their pensions, they will respond by deferring their retirement and continuing to work and contribute until their pension reaches more acceptable levels. This should occur without requiring further reform efforts, which is one of the chief advantages of this approach, like the Ecuador approach. Should people fail to respond to these incentives, the pensions will be automatically reduced, bringing some fiscal relief, again without the need for further reform efforts. Of course, from the perspective of the system, it is far preferable for individuals to delay their pensions because the system both collects more revenue and reduces expenditure, while if pensions are simply reduced, there is no positive impact on revenue. However, the notional account system relies exclusively on incentives to implement this policy rather than explicitly forcing individuals to delay retirement. 7. The multipillar reform, by contrast to the other reforms, initially entails higher costs as the contributions to the PAYG pillar are partially diverted to the funded pillar. Eventually expenditures also fall as those who contributed less to the PAYG system begin to retire and receive lower pensions from the PAYG system, but there is a lag between when the revenues to the PAYG system fall and when the expenditures fall. However, in the longer run, it results in the lowest costs of all the systems. These low costs are also achieved with benefits which average 66.3% of the former men's benefits and 64.6% of the former women's benefits. Since there is a PAYG component not indexed to longevity, eventually the costs will begin to rise again, but benefits could be cut again and still end up higher than the notional accounts benefits or could be formally linked to longevity increases. The multipillar results are not surprising: they basically illustrate the greater pensions possible from the same contribution in a funded system than in a PAYG system as the demography of a country matures. 8. Structure ofthe PAYGpillar. The public pillar under a multipillar system can be structured in a variety of ways. Its structure should take into account the level of protection to be granted to low income workers and workers who might have suffered adverse investment 126 returns leaving them with low pensions. Protection against these risks can be provided through a minimum pension guarantee financed from general tax revenues or through the introduction of a redistributive component within the PAYG pillar as proposed earlier. The smaller-sized PAYG can also become more fully redistributive as in the case of Argentina and the UK. What is being proposed here is a 50-50 split of the contribution, but no increase in its redistribution. Again, this is simply for illustrative purposes. 9. Transition costs. Analogous to the monopillar PAYG reform, the multipillar reform will need to review promised benefits to the transition generation and identify needed adjustments in order to reduce financial costs and manage intergenerational transfers. The introduction of the funded pillar raises an additional set of issues that concern the transition generation and the financing of the explicit debt resulting from the diversion of contributions to the funded pillar. Even if benefits of current contributors are considerably reduced, as recommended, the explicit debt could be significant in the initial reform years until the cumulative impact of those reforms is felt. 10. Countries have developed several approaches to dealing with transition costs, most using a combination of policies. The first is to reduce the liabilities before the multipillar reform is introduced. Clearly, the August 1999 reform made a major contribution in this direction, but with long transition periods. Perhaps more can be done in this regard, either through further conventional PAYG reform or through introduction of notional accounts. The second approach is to allow individuals to remain within the reformed old system or to choose the new system. If the choice is voluntary, full compensation for those who choose the new system may not be necessary. The third approach is to establish a multipillar system but only for younger generations. This approach results in reducing the diversion of current contributions to the funded system in the short term, but increases the length of the transition period, i.e., the government has smaller deficits, but for a longer period of time. Only the mild retirement age is being included in the multipillar reform package here and individuals are being forced to move to the new system if they are under 45 years of age in 2015. 127 CHAPTER 3: REFORMING THE INFRASTRUCTURE SECTORS Contingent Liabilities I. Many sales and operation rights transfers that have taken place to date have required the government to provide guarantees for the private sector. While these guarantees may not lead to any direct costs to the government, they do represent potential costs, especially if issues such as the pricing of electricity, which could be significant, are not addressed. This section investigates these contingent liabilities and proposes some actions to address them. 2. The potentialfiscal advantages ofprivatization. Privatization is often advocated as a way of improving the government's fiscal position. And indeed it can help the government's fiscal situation. It can generate up-front cash revenues that permit the government to retire debt. Although the government loses an asset of similar value, the transaction may improve the government's balance sheet by reducing leverage. Moreover, since the private owner is likely to operate the business more efficiently than the government, the sales price received by the government is likely to exceed the value to the government of the asset it has sold. Similarly, the private financing of new investments can help the government's fiscal position by permitting investment without additional government borrowing -or taxation to finance it. 3. The fiscal effects of BO(T)s. The fiscal benefits of privatization depend, however, on the amount of risk transfer they effect, not on the transfer of legal title. In some cases, therefore, the apparent fiscal benefits of privatization exceed the real benefits. The BOT electricity projects discussed earlier illustrate this issue, but the same problems may arise in Turkey in sectors such as toll roads or gas pipelines. The issue has already arisen in the water sector, where risks that were not passed to the private sector are now leading to the central government's having to make payments to meet obligations accepted by a municipality. 4. In some-although not all-respects, BO(T) power generation plants are more like traditional public projects, financed by borrowings from foreign lenders, than like typical private companies in other industries in Turkey or power generation projects in most industrialized countries. Another way of seeing this is to compare the BO(T) projects with projects financed by TEAS, financed with foreign borrowing guaranteed by the Treasury. In both cases, TEAS and the government will do well financially if, and only if, there is sufficient demand for the output of the plants and that output can be sold at a cost-covering price. In both cases, the providers of the finance-the BO(T) company and its financiers in the one case, the foreign lenders in the other- will do well so long as TEAS or the government has sufficient resources to meet its financial obligations. The main difference, from this point of view, between the BO(T) and the public project, is that the power purchase agreement and associated documents take the place of the loan agreement. 5. This discussion highlights the similarities between "privately" financed BO(T) projects and traditional public finance. There are, of course, important differences. One is that the BO(T) scheme makes explicit the cost of equity financing in a project, because the BO(T) project company insists more than most governments on being fully compensated for the cost of the equity capital it has invested. Another is that the BO(T) framework ensures that the construction and operating risks are transferred to the private sector and in such a way that one company is 128 concerned to minimize the sum of construction and operating costs. BO(T) projects therefore have some advantages over traditional public projects, but they do not significantly reduce the fiscal burden of financing new investments. 6. Financial liabilities and contingent liabilities. One of the reasons that BO(T) projects appear to have large fiscal benefits is that they tend to disguise the fiscal costs of projects. They can do this in either or both of the following substitution effects: (i) off-balance-sheet financial liabilities for on-balance-sheet debt; and (ii) contingent for actual liabilities. The fiscal liabilities created by BO(T) projects often go just by the name of "contingent liabilities." However, it is useful to distinguish between liabilities for the government that are truly contingent and those that are actual but disguised. 7. Off-balance-sheetfinancial liabilities. The energy BO(T) illustrates the difference. TEAS's liabilities are actual, not contingent. TEAS must pay the private generators so long as their plants are working. The Treasury's liabilities (viewed narrowly at least) are contingent. It must pay only if TEAS fails to meets its obligations. But from the entire government's point of view, the liabilities are really actual, not contingent. TEAS is wholly owned by the government, so the financial obligations of TEAS are financial obligations of the government. From the perspective of the government's fiscal position, therefore, the energy BO(T) is principally an issue of actual versus off-balance-sheet liabilities.87 8. Contingent liabilities. The Izmit water BOT illustrates the case of a liability that from the government's point of view was purely contingent. The Izmit municipal purchaser, which is not owned by the central government, had the actual (off-balance-sheet) liability. The government's liability was truly contingent, in that it had to pay only if the municipality defaulted, which is the case. 9. Tackling the problem. Off-balance-sheet and contingent liabilities are not necessarily undesirable-just as ordinary on-balance-sheet debt and government expenditure are not necessarily undesirable. The problem with them is that they are difficult to see. Because the BOT contracts do not lead to increases in the debt recorded on TEAS's balance sheet, for example, TEAS's management and the government will tend to be slower in recognizing a build-up of the company's effective debt levels. Similarly, the government may not be aware of the real cost of the contingent liabilities it incurs through its guarantee program until long after the liability is created. As a result, when the government undertakes BO(T)-type investments it may be misled, for a while, into thinking that these privatizations have had a significant positive fiscal impact. The challenge for the government with regard to these liabilities is twofold: (i) to improve its systems for monitoring, budgeting, and managing its contingent liabilities and off-balance-sheet debt; and (ii) to move, in the medium term, beyond BO(T)-style projects to fuller privatization, in order to realize its full potential benefits for the economy and the budget. 10. The government's ability to move to full privatization (defined as shifting most economic risk to the private sector) will depend largely on (i) the legal ability to transfer ownership of I The energy BO(T)s do also create contingent as well as actual liabilities for TEAS and for the government as its owner. The actual net off-balance-sheet liability for TEAS can be estimated, given assumptions about the future price of fuel, the exchange rate, the value of the electricity received, the value of the plant after 20 years. etc. The contingent liabilities faced by TEAS then relate to possible changes in the variables, such as the fuel cost and the exchange rate, that determine TEAS's liability. 129 assets; (ii) structural regulation that permits and facilitates the operation of fully competitive markets where they are possible; and (iii) conduct regulation of the remaining monopolies that assures investors that they will be able to sell their services to customers for cost-covering prices (while also protecting consumers from the abuse of monopoly prices). 11. The remainder of this section addresses the first challenge of improving the monitoring, budgeting for, and management of the risks inherent in off-balance-sheet financial liabilities and contingent liabilities. 12. Improved accounting The first step in tackling the potential problem created by off- balance-sheet financial liabilities and contingent liabilities associated with infrastructure projects is to improve the government's monitoring of existing liabilities. The best approach is one that considers the whole range of the government's liabilities, including those associated with general municipal borrowing and pension systems; the issue is not specific to infrastructure. This chapter on infrastructure, therefore, provides only a brief summary of the options. The background paper provides an explanation of how to estimate the off-balance sheet debt in a BO(T) project. 13. Different approaches to monitoring are possible: (1) simple listings, in the budget or public accounts, of (a) contingent liabilities and (b) off-balance-sheet liabilities (which accountants sometimes call commitments), without any statement of the values Qf the liabilities; (2) tables of contingent liabilities that include estimates, where possible, of the maximum amount the government stands to lose and tables of off-balance-sheet liabilities (commitments) that include estimates of the amounts of the debt; and (3) more sophisticated attempts, using techniques developed to price financial derivatives, to value the cost of the risk-bearing assumed by the government when it issues guarantees or enters into contracts creating off-balance-sheet liabilities. 14. There is, of course, a trade-off between ease of making estimates and the usefulness of information. Option 3 provides very useful information but is costly to implement. The best approach is likely to be one that combines options 1, 2, and 3, roughly as follows: (i) identifying and disclosing the existence of all significant liabilities; (ii) providing quantitative estimates of maximum losses and the amounts of the off-balance-sheet liabilities for larger, quantifiable liabilities; and (iii)valuing risk bearing in only the most important cases. 15. Controlling: budgeting and debt management. After improving monitoring, the second step is to improve budgeting procedures and the management of the liabilities. Again, the issue is broader than infrastructure, so some options are simply listed below. * Imposing a budget limit on the maximum exposure that can be incurred through guarantees of all types-thus extending the government's existing policy of limiting the value of municipal debt that can be guaranteed by the government, to cover other guarantees. * Imposing limits on the off-balance-sheet liabilities that Government agencies and businesses can contract (or incorporating off-balance-sheet liabilities in any limits on their ability to contract ordinary debt). * Requiring budget appropriation of the value (calculated under option 3 above) of guarantees issued. 130 * Incorporating the existence of off-balance-sheet liabilities and contingent liabilities in the debt-management strategies used by the government's debt-management group (or asset- and-liability management office). 131 CHAPTER 4: AGRICULTURAL AND RURAL DEVELOPMENT Assumptions for Fiscal Impact Estimation of Reform I. State-Owned Enterprises 1. TMO TMO No Reform 2000 Reform 2001 2002 2003 Purchases (ton) 4,500,000 4,500,000 2,250,000 0 0 Purchases price 192 188 162 0 0 (US$/ton) Fiscal Impact 865 844 364 0 0 (million US$) Wheat purchase 4,5 million ton is expected volume in 2000 for both scenarios, half of previous year in 2001. Wheat support price US$192 is per ton in 2000 (no reform scenario). Based on 42,5 percent increase over TL 80,000 per kg in 1999, US$188 per ton is for 2000 (reform scenario) consistent with Pw=US$120 fob, US$19 cif conversion and 35 percent markup. US$162 per ton is in 2001 (reforrn scenario), based on Pw=US$l35fob, US$19 cif conversion and 5 percent markup. 2. TEKEL TEKEL No Reform 2000 Rform 2001 2002 2003 Purchases (ton) 160,000 160,000 80,000 80,000 0 Purchases Price 3,364 2,951 2,400 2,450 0 (US$/ton) Fiscal Impact 538 472 192 196 0 (million US$) Tobacco purchases of 160,000 ton is expected volume in 2000 for both scenarios because auction system will be applied in 2001. Half of previous year will be purchased in 2000 (80,000 ton) and 2002. Tobacco support price is US$ 3,364/ton in 2000 (no reform scenario), based on 42,5 percent increases over TL 1,400,000 per kg. US$2,951/ton is in 2000, based on 25 percent increases over TL 1,400,000 TL. US$2,400/ton is in 2001, based on 25 percent discount from export price of US$3,200/ton). US$2,450/ton is for 2002 consistent with projected increases in Pw. 3. TSFAS TSFAS No Reform 2000 Reform 2001 2002 2003 Purchases (ton) 12,500,000 12,500,000 10,000,000 0 0 Purchases price 65 57 57 0 0 (US$/ton) Fiscal Impact 811 711 569 0 0 (million US$) 132 Sugarbeet purchases 12,5 million ton is expected volume in 2000 for both scenarios because contracts have already been completed. 10 million ton is expected volume after TSFAS has autonomy in making contracts. Sugarbeet support price is US$65 per ton in 2000 (no reform scenario), based on 42,5 percent increases over TL 27,000 per kg. US$57 is in 2000 (reform scenario), based on 25 percent nominal increase. US$57 is in 2001 (reform scenario), based on 12 percent increase. II. Input Subsidies and Premium Payments Fertilizer Budget allocation for year 2000 is TL 145 trillion, which corresponds to US$245 million. This figure is the same for both no reform and 2000 reform cases, and it decreases to US$221 million in year 2001 considering about 10 percent depreciation in TL. Cotton Cost of cotton premium in 2000 is approximately TL 160 trillion, which corresponds to US$269 million. The premium amount is reduced by 10 percent in 2000, this will de reduced the cost US$202 million. Oilseeds Cost of oilseeds premium in 2000 is TL 30 trillion, which corresponds to US$51 million. The premium amount is reduced by 25 percent in 2001. The cost declines to US$38 million. ASCUs ASCU cost assumes a budget of TL 410 trillion in no reform scenario. The cost is projected as US$ where 70 percent of the reminder of (860-410) for 2001 and 30% for 2002. III. Direct Income Support Assumes subsidy of US$80 per hectare per year for the pilot US$50 per hectare and a maximum of 20 hectares per owner to be subsidized. Half of all owners are beneficiaries from direct payments in 2001, but all owners in 2002 and 2003. 8,000 farmers are considered in calculations in 2000. 8 million hectares are used in calculations in 2001 which 90 percent of this area is under 20 hectares. 16 million hectares are used in calculations in 2002 and 2003 (90% of this area is under 20 hectares). 133