Report No. 39771- TJ Tajikistan Programmatic Public Expenditure Review June 22, 2007 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank NCPS NetCreditto the PrivateSector NDS NationalDevelopmentStrategy NFA NetForeignAssets NPV NetPresentValue OIN Other ItemsNet OTL OverheadTransmission Lines PARS Public AdministrationReformStrategy PDPG Programmatic Development Policy Grant PEFA Public Expenditureand FinancialAssessment PEIR Public Expenditureand InstitutionalReview PEP PrivateEnterprisePartnership PETS Public ExpenditureTracking Surveys PFM Public FinancialManagement System PFMMP Public FinancialManagementModernizationProject PIP Public InvestmentProgram PIT PersonalIncomeTax PPER Programmatic PublicExpenditureReview PPF ProductionPossibilityFrontier PPG Public and PubliclyGuaranteed PPIAF Public-PrivateInfrastructureAdvisory Facility PRGF PovertyReductionand Growth Facility PRSP PovertyReductionStrategy Paper PSRP Public Sector Reform Project QFD Quasi fiscal deficit RM Reserve Money SAC StructuralAdjustment Credit SECO Office of the Secretary of State for Economics SFCC State FinancialControlCommission SME Small and MediumEnterprises SURE Seemingly UnrelatedRegressionEquations SWAps Sector-wideApproaches TA TechnicalAssistance TFP TotalFactorProductivity TIN Taxpayer IdentificationNumber TOE Tons of Oil Equivalent UNICEF UnitedNationsChildren's EducationFund USAID UnitedStates Agency for InternationalDevelopment VAT Value addedtax Country Director: Annette Dixon Table of Contents PREFACE ................................................................................................................................................ 1 EXECUTIVE SUMMARY ..................................................................................................................... 1 1. INTRODUCTION ........................................................................................................................ 1 A BACKGROUND 1 B THEDEVELOPMENT c. MAINFINDINGSAND .. ................................................................................................................................ CONTEXTINTAJIKISTAN ................................................................................ 2 RECOMMENDATIONS OF THE 2003/04 PEIR 4 D KEYCHALLENGES FORPUBLIC EXPENDITURE .. REFORM ................................................................. .................................................. 6 F. SEQUENCINGTHE WORK OFTHE PPER E THE THEMESAND CONTENTS OFTHE PROGRAMMATIC PUBLIC EXPENDITUREREVIEW.................12 .......................................................................................... 8 2. ECONOMIC GROWTH: CONSTRAINTS, CHALLENGESAND IMPLICATIONS FOR PUBLICEXPENDITURE ................................................................................................................... 13 A INTRODUCTION . ............................................................................................................................. B. EXPLAINING TAJIKISTAN'SECONOMICGROWTHPERFORMANCE 13 c. ................................................... 14 CHALLENGES AND CONSTRAINTS TO SUSTAINABLE FUTUREGROWTH ........................................... 19 D COMPARATIVEADVANTAGE POSSIBLEGROWTH AND TRAJECTORIES E MODELING THEGROWTH EFFECTSOF PUBLIC EXPENDITURE ......................................................... .......................................... 24 28 30 3. F ...CONCLUSIONS AND IMPLICATIONS FOR PUBLIC EXPENDITURE POLICIES ...................................... FISCALSPACE INTAJIKISTAN ........................................................................................... 34 A. INTRODUCTION ............................................................................................................................. 34 B FISCAL POLICY SINCEINDEPENDENCE ........................................................................................... 35 c. FISCALSPACE . ................................................................................. D. DOMESTIC REVENUES FOR PRIORITY EXPENDITURES ................................................................................................................... 39 E. BUOYANCY INCOME ELASTICITYOF REVENUES 39 AND 41 F CHANGES INTAXPOLICY 42 G TAXADMINISTRATION .. ..................................................................... ................................................................................................................. .............................................................................................................. H. GOVERNMENT 44 . BORROWING .......................................................................................................... ................................................. 49 J. RESOURCE ENVELOPE IFISCALIMPLICATIONSOFREFORMSINTHEELECTRICITYSECTOR 56 64 K THEEFFICIENCY OF GOVERNMENT 66 L CONCLUSIONS .. PROJECTIONS .............................................................................................. EXPENDITURE ............................................................................................................................... ....................................................................... 67 4. HOW CANTHE MTEF BE INTRODUCEDIN TAJIKISTAN? ......................................... 72 A INTRODUCTION .. ............................................................................................................................. 72 B MOTIVATION INTRODUCING THE MTEF c. FOR ................................................................................ 73 74 D CONSTRAINTSTO INTRODUCING SECTOR BASED 75 E WHAT CHANGES ARE REQUIREDTO SUPPORTTHE INTRODUCTION OF THE MTEF? .. WHAT H A S BEEN ACHIEVED SO FAR? ............................................................................................. PLANNING ........................................................ ....................... F. HOW SHOULDTHE INTRODUCTION OF THE MTEFBE SEQUENCED OVER TIME? 79 86 G STRATEGY FOR PHASED IMPLEMENTATION OF THE MTEF . ........................................................... ............................. H. CONCLUSION ................................................................................................................................ 90 93 APPENDIX1 MEDIUM-TERMBUDGET .......................................................................... APPENDIX2: ACTION PLANOF STEPTO IMPLEMENT MTEF INPILOT SECTOR 2006-2007.................... FRAMEWORK 94 96 APPENDIX 3: REGIONAL LESSONS FROM THE DEVELOPMENT OF MTEFS .......................................... 98 5. PUBLICEXPENDITUREAND FINANCIAL ACCOUNTABILITY (PEFA) ASSESSMENT ................................................................................................................................... 105 A INTRODUCTION . ........................................................................................................................... B. INTEGRATED ASSESSMENT OF PFMPERFORMANCE .................................................................... 105 c. ASSESSMENTTHE .................................................................. 106 D. PROSPECTSFORREFORM PLANNINGAND IMPLEMENTATION OF IMPACT OF PFMWEAKNESSES ...................................................... 110 112 Annexes ANNEX1. THEIMPLICATIONSOF PUBLICEXPENDITUREFOR ECONOMICGROWTH TAJIKISTAN .......1 ANNEX11. EXPENDITURES THE STATEBUDGETOF THEREPUBLICOF TAJIKISTAN2005-2010 ........ 156 IN 19 OF PREFACE The ProgrammaticPublic Expenditure Review (PPER) was launched by the World Bank in May 2006, at a workshop with officials o f the Government of Tajikistan (GOT),under the leadership of the Minister o f Finance. At that workshop, it was also agreed to undertake a Public Expenditureand FinancialAssessment (PEFA) which is a comprehensive diagnostic assessment of the state of the Public Financial Management (PFM) system based on a comparison with internationally agreed benchmarks. The PPER is supported by some of the development partners of Tajikistan, notably the UK Department for InternationalDevelopment(DFID) and the Office of the Secretary of State for Economics (SECO) of Switzerland. The PPER is intended to be closely aligned with ongoingreformsto PFM, such as the introduction of the Medium Term Expenditure Framework: (MTEF), and the technical assistance required to support these reforms. The PPER will support these PFM reformsthroughanalytical and diagnosticwork. The PPER has been designedto cover five themes which representthe key challenges for fiscal policy and PFMreforms in Tajikistan. The themes are as follows: Theme 1: Public Expenditure, Fiscal Space and Growth; Theme 2: Policy-based Budgeting; Theme 3: Efficiencyof Public Expenditure in the Social Sectors; Theme 4: Fiduciary Risks; and Theme 5: Inter-governmentalfiscal relations. The PPERwill be undertakenover three fiscal years (2006/07-2008/09)and published inthree volumes, of which this representsthe first. The work inVolume 1covers two of the issues under Theme 1 - the implicationsof public expenditures for economic growth and long-term fiscal space for priority expenditures - which respond to the need to analyze how scarce budget resources can best be allocated to meet the Government's developmental objectives for growth and poverty reduction. Volume 1 also includesan outline o f the strategy for implementingthe MTEF, which responds to a decisiontaken by the Government in 2006 to begin implementingthe MTEF in 2007, and a summary of the PEFA results. This report was written by Sudharshan Canagarajah (TTL) and Martin Brownbridge (Consultant). The PPER team include: Sudharshan Canagarajah, William Dillinger, Jariya Hoffman, Utkir Umarov, Shuhrat Mirzoev (ECSPE); Ernest0 Cuadra, Dina Abu-Ghaida, Anne Bakilana, Peyvand Khaleghian, Saodat Bazarova, Vlaidmir Kolchin (ECSHD); Santiago Cornejo, Logan Brenzel (HDNHE); Waly Wane (DECRG); PascaleKervyn(ECSPS); Aziz Khaidarov(ECCTJ); MartinBrownbridge, NathanSmith, Taras, Pushak, and Victor Murinde (consultants). The peer reviewers are Anand Rajaram (PRMPS), Mary Betley (external - DFID) and Elena Nikulina (ECSPE). Overall guidance for this report was provided by Carlos Felipe Jaramillo and Cheryl Gray. Takhmina Jumaeva and Damika Somasundaram assisted the team. EXECUTIVE SUMMARY Chapter 1: Introduction 1. The ProgrammaticPublic ExpenditureReview 2007-2009 (PPER) aims to provide analysis, diagnosis and policy recommendationsin respect of the key medium and long term challenges facing fiscal policy and Public Financial Management (PFM) in Tajikistan. The PPER is programmatic in the sense that its work program extends over three years and encompasses a planned three volumes, of which this i s the first. The extended time scale of the work program reflects the broad scope and depth of the PFM issues which need to be covered as well as the intentionto align the work program with ongoingtechnical assistance (TA) to support PFMreforms and, therefore, to sequence the work appropriately. 2. The PPER was launched in May 2007 at a workshop in Dushanbe with senior Governmentof Tajikistan (GOT)officials, where the substantive themes to be coveredin the PPER were discussedand agreedupon. A public expenditureworking group, chaired by the Minister of Finance, was set up and provides the formal link between the Bank (and other donors) and the GOTfor purposes of planning the PPER. Development partners, notably DFID (UK) and SECO (Switzerland) have contributedto the financing of the PPER. The PPER builds on the Public Expenditure and Institutional Review 2003/04 (PEIR) which provided the first comprehensive overview of public expenditure issues inTajikistan(World Bank, 2005D). 3. The PPER is structured around five themes which represent the main challenges for PFM reform in Tajikistan and which cover the main PFM and fiscal policy issues which have been prioritized by the GOTand by the donors who are assisting the GOT with PFM reforms. These themes are: Theme 1: Public Expenditure, Fiscal Space and Growth; Theme 2: Policy-basedBudgeting; Theme 3: Efficiency of Public Expenditure inthe Social Sectors; Theme 4: FiduciaryRisks; and Theme 5: Inter-governmentalfiscal relations. 4. Theme 1 responds to questions of fiscal policy design, especially related to the composition of expenditure and the magnitude o f external borrowing, which are becoming increasingly important in Tajikistan and which are likely to have a profound impact on the long term development of the economy. The PPER takes a public finance perspective to evaluate how fiscal policy can best contribute to long term sustainable economic growth and how fiscal space can be created for priority public expenditures, taking account of the scope to mobilize domestic revenues, attract donor grants, access sustainable borrowing and improve expenditure efficiency. The analysis of growth oriented fiscal policy, presentedunder theme 1, is an approachwhich was recommended by the DevelopmentCommitteeo fthe Bank inApril 2006. i 5. Theme 2 responds to the decision taken by the GOTin2006 to reinvigorate efforts to implement the Medium-term Expenditure Framework (MTEF) - which had begun in 2002 under the Bank's IBA2 project - and to seek financial and technical support from development partners for this endeavor. The PEIR had identified the lack o f linkages betweenbudget allocations and strategic policy formulation as a major weakness of the budget process; a weakness which the introduction o f the MTEF aims to address. The analysis undertaken under Theme 2 i s primarily intended to support the TA for implementing the MTEF, by diagnosing the main constraints and making feasible recommendations which can facilitate the gradual introduction o f the MTEF. 6. The social sectors, notably, education, health and social protection, will play a key role in achieving the Millennium Development Goals (MDGs), but the quality o f social services provided to the population i s impeded not just by shortages o f budgetary resources but also by serious weaknesses in the allocative and technical efficiency o f expenditures in these sectors. Theme 3 focuses on expenditure efficiency in the social sectors, an issue which has not hitherto been subject to rigorous analysis, using methodologies which include efficiency frontier analysis and Public Expenditure Tracking Surveys (PETS). 7. Previous diagnostic studies o f the PFM system, including the Bank's 2003 Country Financial Accounting Assessment (CFAA), the International Monetary Fund's (IMF's) 2003 Safeguards Assessment and the DFID's 2005 Fiduciary Risk Assessment, identifiedserious weaknesses in fiduciary controls. These weaknesses also deter donors, who could potentially provide more aid to the government budget, including budget support, from doing so (currently a large share o f donor aid i s actually disbursed off budget, for example to Community Based Organizations). Theme 4 focuses on fiduciary issues and includes a comprehensive evaluation o f the PFM system using the internationally accepted best practice methodology o f the PEFA assessment. 8. Inter-governmental fiscal relations, betweenthe various layers o f Government, are both complex (as explained in the PEIR) and evolving in Tajikistan. They impact on various aspects o f the PFM system, notably through the Local Budget (the budget for local governments) which accounts for the bulk o f social sphere expenditures and in the collection o f tax revenues. Reform o f the PFM system, including the introduction o f the MTEF, must therefore take account of inter-governmental fiscal relations. Theme 5 will provide advice on how the structure o f inter-governmental fiscal relations can be reformed to best accommodate the requirements o f the MTEF, other PFM reforms and the GOT'SPublic Administration Reform Strategy (PARS). 9. This volume o f the PPER focuses mainly on Themes 1 and 2, with chapters on how public expenditure can best support economic growth, and on fiscal space for priority expenditures (under Theme 1) and the introduction o f the MTEF (Theme 2). It also includes a summary o f the PEFA, the main report o f which will be published separately. The work under Themes 3 and 5 requires thore time to complete and will be published in subsequent volumes. It is anticipated that further work under Theme 2, as 11 which the economy needs most to generate higher growth rates i s not investment in public capital such as infrastructure, because the economy is already quite well endowed with this type of capital, but investment for the production of marketable goods and services, for which the private sector i s much better suited than the public sector. Third, the private sector, drivenby the profit motive and operating incompetitive markets, faces much stronger incentives to generate improvements in factor productivity than does the public sector. 15. Private sector investment rates are very low, at 5-6% o f the Gross Domestic Product (GDP), despite the recovery o f the economy. One o f the main reasons for the lack o f private investment, as found in surveys o f the private sector. such as the 2005 EBRD-World Bank Business Environment and EnterprisePerformance Survey (BEEPS), i s the poor institutional climate for investors, characterized by multiple complex and untransparent regulations administered by many different Government agencies which give scope for discretion on the part o f public officials and offer opportunities for corruption. In reality, there is no level playing field for private investors; instead each firm makes its own individual arrangements with officials, a situation which is extremely damaging for efficient resource allocation and is anathema to a predictable and stable businessclimate. 16. Consequently, a major policy priority should be the urgent implementation o f measures to improve the institutional climate for private investors, by reforming the systems for permits and licenses and tax administration, so that a more simple, predictable and transparent regulatory framework is put in place, to create a clear set o f rules which apply to all investors. The Government has begun to implement reforms in this area, by streamlining licensing procedures, for example, but much more needs to be done to strengthenthe investment climate. 17. Tajikistan appears ta have a comparative advantage in the generation o f hydroelectric power for export to neighboring countries, and a number o f hydropower projects (HPPs) are currently under construction or beingplanned. The PPER argues that the Government should allow private investors to play the leading role in owning and managing all o f the new HPPs for two reasons. First, given the huge scale o f investment resources required (several billions o f dollars for all o f these projects combined), the Government does not have the financial resources or borrowing capacity to fund these investments without jeopardizing external debt sustainability. Second, energy projects can be operated on a commercial basis and are, therefore, more suitable for private sector investment'which can provide more efficient management o f these projects than can the public sector. However, while the generation and export o f hydropower i s potentially profitable, if long term export contracts can be secured, it will create very few jobs once the construction phase o f these projects has beencompleted. It should not comprise the sole engine o f growth for the economy; and while HPPs can contribute to long-term development o f the Tajik economy, they should not be viewed as a panacea. 18. Given that budgetary resources are scarce and that public sector borrowing capacity is very constrained, the careful prioritization o f Government expenditures i s iv essential. The PPER cautions against large public new investments in infrastructure, even if these have the attributes of public goods (e.g., roads) and will not, therefore, be supplied by the private sector. Tajikistan already has a large stock of public capital, relative to the size of its economy, which is a legacy of heavy investment during the Sovietera, but the Governmentlacks the budgetaryresources to maintainthis adequately. Moreover, surveys of the private sector do not indicate that deficiencies in the public infrastructureare a major constraint to private sector business. A World Bank regional study o f transport and constraints to trade in Central Asia, including Tajikistan, concluded that the road networks in the region are relatively extensive and largely sufficient to meet the needs of users, although roads are in poor condition because of inadequate maintenance (World Bank, 2004B). Consequently it is unlikely that investment in new public capital assets is a priority for the economy. Instead, priority should be givenwithin the budget to funding adequate maintenance of the core networks of the existingpublic infrastructure; expenditureswhich are likely to offer higher rates of returnthan investmentinnew capitalassets. 19. Labor productivity is low but can be improved over the long term through better public educationand healthservices. Therefore, expanding access to, and improvingthe quality of, public education and health services should be a priority for budget policy, with larger budget allocations to these sectors alongside measures taken to improve expenditure efficiency. Public provision of these basic social services is essential because, given low per capita incomes, privateprovisionis not affordable for most of the population. 20. To summarize, this chapter of the PPER advocates a development strategy in which priority in the Government budget is placed on expanding and improving the provision of the education and health services needed to boost human capital development, alongside the essential maintenance of the existing core infrastructure networks. New capital projects should not be a priority for the budget, because it is not evident that an expansion of the public capital stock is really essential to support economic growth and because they would inevitably crowd out funding for essential public servicesand maintenance. Governmentshouldleave the private sector to invest in commerciallyviable infrastructureprojects as well as a broadrangeof other, more labour intensive, goods and services, but increased private investment will require a marked improvementinthe institutional climate for business. Such a strategy offers the prospect of sustainable economic growth with employment creation, rising labor productivity which will raise real wages, and improvements inpublic services which will enhance the nonincomeaspects of humanwelfare. 21. There is a danger that poor public policy decisions, influenced by an excessively optimistic view of the benefits to be gained from large infrastructure projects, lock Tajikistan into an inferior development trajectory. Such policy choices would include heavy public investment in infrastructure, funded by external borrowing, the neglect of education and health services and the failure to improve the institutional climate for private investment. The outcome of such policies is likely to be capital intensive growth V which generates few jobs and, therefore, does little to reduce poverty, and inthe long run may be derailed by an unsustainablepublic debt burden. Chapter 3: Fiscal Space in Tajikistan 22. The concept of fiscal space refers to the additional budgetary resources which could be mobilised for priority public expenditures. This chapter adopts the fiscal diamond framework, which was proposed in a paper presented to the Bank's Development Committee Board in April 2006, to evaluate fiscal space over a 10 year time frame. The fiscal diamond entails analysis of the scope for mobilising resources from four sources: domestic revenues, donor grants, sustainable borrowing and expenditure efficiency improvements, while ensuring that fiscal policy i s consistent with macroeconomic stability. This chapter also includes analysis of the budgetary implications o f reforms in the electricity sector to reduce the QFD (estimated at 9% of GDP in2006) because this is critical for long term fiscal sustainability. 23. The fiscal space projections offer grounds for guarded optimism, but also raise important warnings about fiscal sustainability if key policy reforms are not implemented. Fiscal space equivalent to 8% of GDP per annumcouldpotentially be mobilized by 2016, but this is contingent on reforms to tax administration and PFM systems. Without these institutional reforms, very little fiscal space will be created. The dangers to fiscal sustainability arise from excessive external borrowing and the large QFD. 24. Domestic revenues could potentially be increased by 5-6% of GDP (from nearly 19% of GDP in2006 to over 24% of GDP in2016). This i s the most important potential source of fiscal space. Tax administration reforms (discussed in paras 51-53 below) contribute the bulk of the increase. Given that there is substantial scope for improving . the efficiency of the existing tax administration, the PPER projects that, if tax administration reforms recommended by the IMF and the Asian Development Bank (ADB) are implementedeffectively (which will take time), tax revenues could rise by about 20% over 10 years, which would raise additional tax revenue of more than 3% of GDP per annum after 10years. 25. There is only limited scope to raise more revenue through tax policy reforms for two reasons. First because rates are already quite high by international standards for two of the most important taxes; 20% for value added tax (VAT) and 25% for corporate income tax, and so further increases in these rates is not advisable to avoid distortions. Second, although customs duty rates are relatively low, most imports enter Tajikistan duty free under regional trading arrangements with the Former Soviet Union (FSU) countries; raising customs tariffs will not generate much additional revenue. The most promising option for increasing tax rates is provided by the personal income tax (PIT), because current rates are low, with the top marginal rate being only 13%. Less than 5% of labor income was paid as PIT in 2004. The PPER recommends raising PIT rates, including raising the top marginal PIT rate to 25%, which is the same level as that of the corporateincome tax (which would remove a distortion and an avenue for tax avoidance). Higher PIT rates could potentially double the share of labour income paid as tax and vi generate revenue of about 1% of GDP. Higher PIT rates are alsojustified on the grounds that they would make the tax system more progressive. 26. Revenues from new HPPs, comprising both taxes or royalties on the export of electricity and returns to the GOT equity stake in these projects, could contribute around 1% of GDP to revenue by 2016, although this will depend on factors which are not yet certain, such as the price paid by neighboring countries ,for electricity and the cost of transmission to export markets. 27. Donor grants currently constitute a relatively small share (about 5%) of the budget resource envelope. The bulk of donor grants are disbursedoff-budget, mainly to CBOs, which partly reflects donors' concerns about fiduciary weaknesses in the Government budget. In 2005, donors disbursed at least US$109 million in grants off-budget, comparedto only US$18 million inthe form of on-budget grants. Fiscal space would be created by an increase in donor grants to the budget, arising from an overall increase in grant aid to Tajikistan, in line with commitments made by the G-8 countries to increase their aidto the poorest countries, and/or by donors shifting a larger proportion of their aid to the Government budget. The latter will require improvements in public financial managementto strengthen and make more transparent the budget process and to improve the allocation of expenditures, especially inthose sectors which can contribute to meeting the MDGs. 28. It is very hard to predict how donor aid to Tajikistan will evolve over a 10 year period, and the range of possible outcomes i s large. Tajikistan might receive no increase in grants, because donors remain unconvinced that fiduciary standards are improving sufficiently and/or because Tajikistan contracts more loan finance from non traditional lenders, thereby reducing the perceived need for more grant aid. At the other extreme, grants could increase several fold as donors both increase grant aid to Tajikistan as part of a general expansion of aid to low income countries and channel a larger share of their grant aid to the government budget. The projection made in the PPER falls between these two extreme scenarios. Grants to the budget are projected to rise by 10% a year in nominal dollar terms, using the 2007 budget projection as a base. This will raise grants from 1.2% of GDP in2006 to 2.2% of GDP in2016. 29. Tajikistan's external public and publicly guaranteed (PPG) debt was reduced to sustainable levels as a result of debt relief from bilateral creditors and the IMF and by the Government restricting borrowing for the Public Investment Programme (PIP) to 4% of GDP per annum during the 2000s. However, in 2006, GOT contracted new loans of US$604 million from China, as a result of which external loan disbursements will jump to an average of 14% of GDP during 2007-2009 (see Chart 1 below). Consequently, the Debt Sustainability Analysis (DSA), undertaken in early 2007 by the Fund and Bank, found that Tajikistan is at a high risk of debt distress and will breach two of the thresholds for external debt sustainability stipulated in the IMF/IDA Low Income Countries DSA Framework. vii Chart 1: External loan disbursementsas percent of GDP and the implied pathof the NPV of PPG externaldebt to exports, 2005-2016 18% 180% 16% 150% 14% n \ Exp 8 12% J 120% Y z W f9 10% g +Loan Disbursements to GDP - RHS cE 90% 8% E zz 0 P 6% 60% 6 -1 4% 30% 2% 0% 0% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2018 30. The approach taken inthe PPER to calculate the amount of fiscal space that could be created by external borrowing is to assume that, after the peak in new loan disbursements during 2007-2009, further borrowing will be constrained so as to reduce the NPV of Public and Publicly Guaranteed (PPG) external debt to the thresholds of 150% of exports and 40% of GDP which are deemed sustainable under the IMFIIDA Low Income Countries DSA Framework by 2016. This implies that after 2009, external loan disbursements will have to fall back to an average of around 5% of GDP if the external debt is to be sustainable, as shown inChart 1, whichprojects loandisbursements as a percent of GDP andthe associatedpath of NPV of PPG external debt to exports from 2005 to 2016. 31. Over the long term, average annual disbursements of external public debt should be restrained to 5 6 % of GDP at most if external debt sustainability is not to be jeopardized. Although this i s higher than the level in2006 (by 2.5% of GDP), it is much lower than the amount of loan disbursements projected over the medium term (2007- 2009), a level which is clearly not sustainable over the long term. These projections do not include non-PPG debt such as cotton debt. Ifthe GOTwere to assume the cotton debt or borrow to fund debt relief for the cotton farmers, the ratios of NPV of external debt to exports and GDP would be higher and so the scope for new borrowing after 2009 would be correspondingly lower if the external debt is to remain sustainable.There is no room for domestic financing of the budget because the domestic financial market is very shallow; i s too heavily dollarized, and is undiversified (the non-banking sector is very small).If Government were to borrow domestically, control of the money supply would be jeopardized (as happened in the 1990s) leading to high inflation, and private sector borrowers would be crowded out of credit markets. The long term sustainable rates of domestic borrowing should be slightly negative to allow monetary growth to be ... Vlll controlled, the National Bank of Tajikistan (NBT) to accumulate foreign exchange reservesand for private sector credit to grow. 32. The scope for efficiency improvements of public expenditure is hard to quantify because of data limitations. However, major allocative and technical inefficiencies characterize public expenditures, so there must be the potential for improvingboth types of efficiency through reforms to PFM systems, such as the introduction of the MTEF, which over the long term should create fiscal space. Given that Government final consumption and investment expenditures (i.e., excluding transfer payments and interest payments) amountedto 17%of GDP in2006, evena relatively modest 10%improvement in expenditure efficiency over the long term would generate fiscal space amounting to 1.7%of GDP. 33. Electricity sector reformswill have a profoundimpact on the budget over the long term. The sector incurs a QFD estimatedat 9.3% of GDP in2006, which is measuredas the difference betweenthe revenue which Barki Tajik, the state ownedelectricity utility, requires to operate and properly maintain the existing infrastructure for the supply of electricity, and the revenue which it actually collects in cash. Sub-optimal electricity tariffs, which are only one quarter of the level needed to cover the long run average incremental cost (LRAIC) of domestic supply, contribute the largest part of the QFD, supplementedby technical and commercial losses and poor cash collectionof electricity bills. Although the QFDis not currently a cash burdenon the Government budget, it will become so if no measures are taken to eliminate it, because the consequences of the Government not funding the QFD will be that essential maintenanceand rehabilitationof the electricity transmission and distribution system will be neglected, and hence electricity supply will deteriorate, with adverse consequences for the economy. 34. The most important reformneededto reducethe QFD is an increase inthe average domestic tariff to the LRAIC of supply. Higher electricity tariffs will have a negative impact on some components of the budget, reducing tax revenue paid by the aluminum smelter, and increasingthe costs to the budget of Government's own electricity bills and o f transfer payments to compensate low income electricity consumers under the energy compensation mechanism (ECM). The aggregate fiscal impact of electricity sector reforms, including the revenues mobilized from HPPs, is overwhelmingly positive, generatingnet gains of almost 10% of GDP, most of which constitutes the reductionof the QFD. 35. Table 1 below summarizes the fiscal space which could be created in 2016, compared to 2006, if the reforms identified above, tax administration, tax policy, PFM and the electricity sector, are implemented. Additional budgetaryresourcesamountingto 8.6% of GDP could be mobilized, mainly from domestic revenues, but additional expenditures amounting to 1.9% of GDP would be required to fund higher interest payments (because of higher external debt), higher Government utility bills and ECM payments and to fund the remaining QFD in the electricity sector (reduced to 0.6% of GDP). Expenditure efficiency gains would add 1.7% of GDP to fiscal space; hence net fiscal space would amount to 8.4% of GDP per annum by 2016. ix Table 1. Fiscal Space in2016 compared to 2006; Yo of GDP 2016, change over 2006 BudgetResource Envelope 8.6 Domestic revenues 5.6 Grants 1.o I Financing I 2.0 I Net External Financing 2.5 Net Domestic Financing -0.5 Additional resources required for: 1.9 Interest Davments 0.4 Government electricity bills 0.5 Energy Compensation Mechanism 0.4 FundingremainingQFD 0.6 Gains inExpenditure Efficiency 1.7 (assumes 10%efficiency gain) X Chapter 4: How can the MTEF be introduced in Tajikistan? 36. This chapter of the PPER responds to the GOT'Sdecision taken in 2006 to implement a MTEF and to seek technical assistance for this from donors. The introductionof the MTEF also reflectsthe desire of Tajikistan's development partners to develop meaningfulsector budget strategies, especially in the social sphere, around which they can provide aid to the budget. The chapter provides a broad strategy for implementing the MTEF in a gradual and phased basis, building on the strategy presentedat a roundtable organised by the MOF in May 2006. 37. The motivation for introducing the MTEF is to improve budget resource allocation and to link this with the strategic policy objectives of the government, includingthose set out in the National Development Strategy (NDS) and the Poverty Reduction Strategy Paper (PRSP). Previous efforts to introduce an MTEF under the Bank's IBTA2 project, had not gone further than the preparation of a rudimentary fiscal framework which provides a hard budget constraint for aggregate expenditure and, therefore, is a usefultool for macroeconomicmanagement, but does not play any meaningfulrole as a budget planningtool at a lower level of expenditurethan that of aggregateexpenditure. 38. The major constraintsto introducingan MTEF which can play a meaningful role as a tool for medium-termsector planningare two-fold. First, the State budget is very fragmented, with more than 100 Key Budget Organizations (KBOs); each prepare budgets and deal separately with the Ministry of Finance (MOF) inthe budget process. In some of the major sectors, such as education, responsibilityfor preparing and implementing budgets is disbursed among multiple KBOs, including local governments and various types of central spending agencies (there are more than 40 KBOs in the education sector). The line ministries have the responsibility for preparingsector policiesbut are effectively marginalizedfrom preparingthe budgets for their sectors; hence there is a disconnect between policy formulation and budgeting, especially in the social sectors where the bulk of expenditures are channelled through the Local Budget and bypass the line ministries. In addition, the current and capital expenditures are treated separately in the budget preparation process, with the former being the responsibility of the Ministry of Economy and Trade. 39. The second major constraint to introducing an MTEF is the very weak capacities for budget planning in the MOF, line ministries and other KBOs. Familiaritywith modern budget techniques and concepts is very scarce among public officials. The line ministries have very small budget planning departments because they have not previously had to undertake these functions. Budgets are essentially prepared by KBOs in a purely mechanistic way, following rules and formulae stipulated by the MOF in the budget circulars. Introducinga system of policy based budgetingsuch as the MTEF requires puttingin placethe skills neededfor forecasting demand for public services, prioritisingand costingexpenditures and making choices betweencompetingexpenditure demands; these skills are virtually non-existentat the moment. 40. The MTEF strategy involves giving the line ministries a much greater role in the budget preparationprocess, so that budget allocations can be better aligned with xi sector strategies. This will require changes to the budget process and a major strengtheningof capacities for budgetingin the line ministries. Ineach sector, the line ministry responsiblefor policy formulation will be designated the Lead Line Ministry (LLM). At the start of the budget process, the LLMs will provide guidance to all KBOs in their sectors to ensure that budget submissions accord with sector policy. The KBOs will then channel their budget submissions for expenditures in each sector to the relevant LLM, each of which will collate all the submissions for its sector and prepare a consolidated sector budget to be submitted to the MOF. In addition, the MTEF strategy will integratethe capital and current budgets. 41. The severity o f the capacity constraints means that implementation of the MTEF must of necessity proceed cautiously, with the introduction of new budget processes being matched to the building of technical capacities for budgeting in the MOF and line ministries, supported by effective donor TA projects. Consequently, the MTEF will be implemented in phases, beginning with a pilot sector; education. Education has been chosen as a pilot sector in part because donors are willing to provide a significant amount of TA to the sector. The pilot phase, which is expected to last for two years, will involve: a) Strengtheningthe capacities of the MOF to produce a medium-term fiscal framework, through TA for revenue forecasting and macroeconomic forecasting; b) Building a genuine sector budget process in the education sector, with the support of TA for the Ministry of Education from the Bank, the European Union(EU) and the UnitedNations Children's EducationFund(UNICEF). 42. Additional sectors will be brought into the MTEF after the pilot phase has been completed, and the lessons learned in the pilot phase should help expedite this. As noted above, capacity building will be a prerequisite for success of the MTEF. Fortunately donors have prioritised TA for capacity building so funding for these activities should not be a serious constraint, but effective coordination and harmonisation of donor TA efforts will be essential. In this regard, a joint donor missionon supportingintroductiono fthe MTEF, which was held inApril 2007, was a usefulstep forward. xii Chapter 5 PEFA Assessment 43. This chapter provides a summary of the PEFA assessment carried out between October 2006 and April 2007. The purpose of the PEFA assessment is to provide the GOT with an objective assessment of the country's PFM systems, support a better understanding of the overall fiduciary environment of the budget and assist in identifying those parts of the PFM systems most in need of reform. The PEFA assessment following the standard PEFA methodology which grades 28 different aspects of the PFM system and 3 relatedto donor practises, with each indicatorgraded on a scale from A (highest) to D (lowest). Table 2 presents the grades for each indicator as evaluated by the PEFA assessment. 44. Tajikistan's overall performanceagainst the PEFA indicators is poor. Out of the 28 PFM indicators, only two receivedA grades (aggregate revenue outturn [which has consistently over-performed against the budget] and the comprehensiveness of budget information); there were only four B grades, compared to 11 C grades and 7 D grades. Four indicators were not assessed because of lack of information. Each of the three indicators for donor practices received D grades, implying that the performance of donors is also poor. 45. Hence there is considerable room for improvement in PFM systems as well as with donor practices. Areas which scored badly included:classificationof the budget; public access to information; multi year perspective for budget planning; predictability of the availability of funds for expenditure commitments; quality and timeliness of financial statements, external audit and the proportion o f aid which is managed through national budget procedures. However, some o f these deficiencies are being tackled through ongoing PFM reforms supported by TA from donors, such as the introduction of the MTEF, support for the external audit agency and reform o f the internal audit system. Policy Recommendations of the PPER 46. This section brings together the policy recommendations from each of the different chapters of the PPER as well as the preliminary findings of ajust concluded Bank missionto Tajikistan to analyse the sequencing and content of reforms to PFM following on from the PEFA assessment. The policy recommendations of this volume of the PPER are discussed under three headings (which are not mutually exclusive); (i)institutional reforms; (ii)composition of public expenditures; and (iii) threats to the sustainabilityof public finances. xiii Table 2: Overview of PEFA Assessment Results Indicator I I Issue covered IRating A. PFMOut-Turns: Budget Credibility I PI-17 Recordingand management of cash balances, debt and guarantees C+ [pending finalization] PI-18 Effectivenessof payroll controls not assessable/furtherdata required PI-19 Competition, value for money and controls in procurement not assessablelfurtherdata reauired 47. Major institutional reforms are essential, to create fiscal space and are also a prerequisite for translating this fiscal space into realizable benefits for the economy, public services and poverty reduction. Institutional reforms are needed in three areas in particular; PFM systems, tax administration and the business climate for private investment. xiv 48. PFM reforms will create fiscal space, by improving the efficiency of expenditures and thereby freeing up resources for other budget priorities and by enhancingthe credibility of the budget as a tool for povertyreduction,which will help the Government to mobilize more donor grants for the budget. In addition PFM reforms to improve the allocative and technical efficiency of expenditureare needed to strengthen the delivery of public services. Key PFM reforms which are currently being implementedor planned are: (i) the introductionof the MTEF, beginningon a pilot basis in the education sector; (ii)the implementationof wage reform in the civil service and health sectors, to strengthen incentives for critical staff; (iii)the introductionof automated Treasury systems at the local level, which are linkedto the MOF and which will eventually be followed by the introduction of a computerised FinancialManagement Information System (FMIS; (iv) the development of a modern internalcontroland audit system; (v) the alignmentof accountingmethodologieswith international public sector accounting standards; and (vi) the introduction of a new budget classificationsystem. 49. The PFM reform agenda is ambitious, especially given the severe capacity constraints in the public service. Hence the PFM reforms should not be hurried and should be sequencedto evolve in parallel with the design of methodology,enactment of new legislationand capacity building. A comprehensivecapacity buildingprogram is required to train staff throughout the public service, and especially in the line ministries, inmodern budgetingand accountingmethodologies. 50. The success of PFM reformswill depend to a large extent on the details of their implementation. Therefore it is essential that each of the PFM reforms is well planned and realistic. Implementationshould focus on key aspects of the reforms which are essential for improving PFM systems in Tajikistan, rather than attempting to raise the whole of the PFM system to internationalstandards, which is not feasible. A system-wide perspective should be taken to help determine the best program of reforms in terms of outcomes and risks. This means attendingto the weak links in the PFM system and focusingresources on the areas of highest potentialgain. 51. Tax administration reforms are needed for two reasons. First, to create fiscal space by strengthening revenue mobilisation: tax administration reforms could potentially generate tax revenue equivalent to more than 3% of GDP per annum by 2016. Second, to improve the business climate for the private sector by making the tax system more transparent, fair and less open to abuse: the BEEPS reported that 80% of businessespay bribesto tax inspectors(World Bank, 2005A: 16-17). 52. A strategy for tax administration reform has been recommended by the IMF (Harrison et al, 2005) and the ADB. The priority for reform is to introduce modern tax administration practices facilitated by computerization. Self assessment by taxpayers, backed up by selective risk basedaudits, should replacethe current system whereby tax officials attempt to verify manually every tax return. Over the longer term, the structureof the tax department should be re-organizedto reducethe number o f smalltax offices in the regions which play a marginalrole intax collection. Large taxpayers should be handled by the large Taxpayer Unit and model tax inspectorates should be established in each of the three regions to handle medium sized taxpayers. The tax police should be abolished, as this unit is not compatible with modem tax administrationpractices. xv 53. Over time, the closure of small offices, together with computerization and changes in working practices, will allow staff numbers in tax administration to be reduced. This will create resources for raising staff salaries substantially, so as to strengthenincentivesto attract and retainhigher caliber staff and reduce incentivesfor corruption. Tax officials also require substantialtraining, especially inauditing. 54. As well as tax administrationreform, improving the institutional climate for the private sector to attract more private investment will require cleaning up the system for business permits and licences. The aim should be to reduce the absolute number of permits and licences required and the number of different agencies involved in processing them and to put in place a system which is transparent, predictable and does not offer opportunities for arbitrary decision making on the part of public officials. 55. The importance of the institutional reforms discussed above cannot be over- emphasised. These reforms are essential to improve economic governance. World Bank research on the transition economies of Europe and Central Asia (ECA) found that the impact on economic growth of the size of the Government budget depends on the quality o f governance (Varoudakis, 2006). In countrieswith poor governance, an increase in the size of the Government budget impedes growth, because a large tax system characterisedby poor governance seriously distorts resource allocation while poor governance reduces the positive benefits of potentially productive public expenditures. It is only in countries with good governance that creating fiscal space for productive expenditureswill actually be translated into higher growth. Tajikistan scores badly on most measures of governance; it is ranked second worse among 27 ECA countries in terms of governmenteffectiveness. Consequently, improvementsin governance are a prerequisite if increased Government expenditures in the priority sectors are to bring about higher economic growth rates; improved governance must precedean expansionof the budget. 56. The PPER makes clear recommendations about how the allocation of expenditures within the Government budget should be shifted to support sustainable economic growth and poverty reduction. Increased public investment, whether in commercial energy projects or in infrastructure projects which have the character of public goods, shouldnot be a priority for the budget. Commercialenergy projectscan be owned and financed by the private sector, without the (or with minimal) need for direct Government funding and economic growth does not depend on new public infrastructure projects. Moreover, the huge funding requirements for public investment projects will inevitably crowd out resources for priority current expendituresand/or undermineexternaldebt sustainability. 57. Over the medium-term (2007-2009), slightly over 50% of the entire Government budget will be allocatedto capital expenditures, leavingvirtually no room for any real increases in the funding for current expenditures. Given the constraints on current budget expenditures, it is not even clear how the new public assets created by this surge in capital spending can be properly maintained over the long term. Once this surge in capital projects is completed (in 2009), the budget should be rebalanced towards recurrent expenditures, which is where more funds are neededthe most. xvi 58. Instead of new capital projects, priority in the Government budget should be accordedto the key social sectors, healthand education, which can strengthen human capital formation. These are services which will not be provided optimally by the private sector, if only because most people in Tajikistan cannot afford private provision, so the Government must ensure comprehensive provision o f basic health and education services. Allocatingmore budgetary resources to healthand education, alongside PFM reforms to improve the efficiency of expenditures in these sectors, is necessary to expand access to and improve the quality of health and education services. Inturn this will enable the quality of the workforce to improve and thereby boost labour productivity, which will be an important source of future economic growth. 59. There are problems with the existing public infrastructure, not least because a lack of funds has prevented most of it from being properly maintained since independence. Therefore the priority in the budget should be to ensure that sufficient funds are made available for maintainingadequately the core networksof the existing infrastructure. This will generate higher returns to spending than investment in new infrastructureassets. 60. The potentialbenefits to be derivedfrom institutionalreformsand the creationof fiscal space for priority expenditures will be undermined if the long-term sustainability o f public finances is put in jeopardy. There are two main threats to fiscal sustainability in Tajikistan, excessive external borrowing and the QFD of the electricity sector. 61. Tajikistanfaces a highrisk o f externaldebt distress as a result of large new loans contracted in 2006 for infrastructure projects. External loan disbursements during 2007-09 will average 14% of GDP, a level which is not sustainable. To ensure long term external debt sustainability it will be necessary to restrict new loan disbursements after 2009 to an average of around 5% of GDP per year. The argument sometimes made tojustify heavy externalborrowing,that the projectsfinanced by the loans are economically viable and will generate large benefits, are not tenable. External debt sustainabilityis a macroeconomicconcept and the external debt burden must be evaluated in relationto macroeconomic variables, such as the size of GDP, exports and Government revenues, irrespective of the benefits yielded by individual projects. Key internationally recognized thresholds for external debt sustainability, such as the Net PresentValue (NPV) of debt to exports and GDP, will be breachedby very wide margins ifexternalborrowingis not reducedsharply after 2009. 62. The policy recommendation of the PPER is that the GOT should set clear annual limits on the level of new external borrowingfor the medium- and long-term, which are consistent with bringing the NPV of external public debt back into line with internationally accepted thresholds for debt sustainability in a low income country, such as those defined in the IMF/IDA Low IncomeCountries DSA Framework. This will give a clear signalto the private sector and Tajikistan's development partners that the Governmentwill implement sound debt managementand ensure that a future debt crisis is avoided. 63. The electricity sector's QFD will inevitablybecome a burdenon the budget over the long term unless it is eliminated through electricity sector reforms, because the xvii consequences for the economy of not properly maintaining and rehabilitating the transmission and distribution network will be too damaging. Therefore, the Government budget will have to fund these essential expenditures if Barki Tajik cannot afford to do so. Furthermore, the large size of the QFD, estimated at more than 9% of GDP in 2006, meansthat it will severely squeeze the fiscal space available for priority expenditures and/or force Governmentto incur unsustainableborrowing. 64. The electricity sector reforms needed to eliminate the QFD are not in dispute. Average electricity tariffs must be increasedto around four times the current level. In addition, Barki Tajik must reduce its commercial losses by improvingits billing and collectionsystems and reduce technical lossesthrough proper maintenance and repair of the distribution system. These reforms will however be politically painful, as consumers who currently enjoy subsidised electricity will have to pay much higher tariffs. 65. As noted above, this is the first of a planned three volumes of the PPER. Subsequentvolumes will make further policy recommendations. There will be further recommendations relatedto the implementationof the MTEF, based on the practical experience of those working on the ground to provide TA and build capacity in the MOF and line ministries. Recommendations will also be made to improve expenditure efficiency in the health and education sectors. In addition, a comprehensive PFM reform strategy, which is agreed with the GOT and can be supportedby development partners, will be set out. xviii 1. INTRODUCTION A. BACKGROUND 1. The PPER aims to deepen understanding of key analytical issues related to fiscal policy and PFM in Tajikistanand to support the ongoing PFM reform program in the country. The fiscal policy and PFM issues addressed in the PPER includethe need to focus fiscal policy on supporting sustainable economic growth and poverty reduction, the introduction of policy based budgeting, improving the efficiency of public expenditures and strengthening fiduciary controls. The analysis and diagnostic work in the PPER will provide the basis for the preparationof a government owned PFM reform strategy and action plan which will guide the policy dialogue and technicalassistanceon PFMreforms. 2. The PPER adopts a programmatic approach with the work taking place over three years (2006/7-2008/9). The PFM reform agenda in Tajikistan is wide and challenging and the volume of work to be covered is accordingly large; hence the three-year time-frame of the PPER. The sequencing of the work program of the PPER is guided by dialogue with the GOTwith the aim of supporting the ongoing PFM reforms being undertaken in the country. This volume represents the findings from the first year's work of the PPER. 3. A workshop to launch the PPER and discuss the three year agenda was held in Dushanbe on 15 May 2006 and attended by senior GOT officials. The views expressed at the workshop guided the design of the PPER and the associated work program. To facilitate Government ownership of the PPER, a Public Expenditure WorkingGroup, which includes representativesfrom the MOF, other key government agencies and donors supportingPFM reforms, was establishedto coordinatethe work program for the PPERand meetsregularly. 4. A Public Expenditureand InstitutionalReview (PEIR), undertaken in 2003/04, providedthe first comprehensive overview of public expenditure issues in Tajikistan (World Bank, 2005D). Together with a 2003 Country Financial Accountability Assessment (CFAA) and 2003 Country ProcurementAssessment Report(CPAR), the PEIR identified critical weaknesses in the processes for allocating budget resources, in budget executionand PFM. However, implementationto date of PEM and PFM reforms has been at best slow and uneven. It is evident that neither the GOT nor the donors has fully comprehendedthe complexitiesof implementingthese reforms in the Tajik context: on the GOT side it is unlikely that the reform steps have been fully owned or supported, while the donors have not collectively analyzed the institutional characteristics of the budget processes in Tajikistanand the difficultieswhich this has posedfor the implementationof reforms. However,in 2006, the GOT took a number of steps to move the PFM reform agenda forward, including the introduction of a 1 MTEF and the drafting of a PARS. The PPER responds to, and supports, the GOT'S ommitmentto its own PFM reformagenda. 5. The PPER is organized around five themes which represent the critical challenges for budget system reform in Tajikistan. The rationalefor these five themes i s that they cover the main areas of fiscal policy and PFM reforms which have been prioritizedby the GoTand by the donors who are assistingthe Go with PFM reforms. These five themes are: (i)strategic fiscal policy issues: aligning fiscal policy and public expenditures with the requirements to support sustainable economic growth and poverty reduction; (ii)reform of the budget process to support the introductionof policy based budgeting; (iii)efficiency of public spending in the key social sectors of health, education and social protection; (iv) public financial management, especially internal controls to prevent corruption, reporting of budget outturns, the internal and external audit functions and procurement; and (v) intergovernmentalfiscal relations, especially relations between the center and local governments (local hukumats). These themes are very similar to the issues that have been identified as critical for improvingPFM inother Commonwealthof IndependentStates (CIS) countries 6. The recommendations arising from the PPER will be brought together to comprise a PFM reform strategy and action plan. This action plan will provide the main policy document of the GOT to guide PEM and PFM reforms and will also serve to guide the technical assistance programs of interested donors (e.g., World Bank, European Commission (EC), DFID, ADB, and the United States Agency for International Development (USAID). The Public Expenditure Working Group will oversee implementationof the PFM reform strategy and action plan. 7. The rest of this introductionis organized as follows. Section 2 briefly reviews the developmental context in Tajikistan. Section 3 summarizes the findings and recommendations of the 2003/04 PEIR. Section 4 outlines the most critical challenges for public expenditurereforms. Section 5 outlines the objectives,activities and outputs of each of the five themes o f the proposed PPER. Section 5 also notes which of the reports under each of the themes are included in this volume of the PPERandwhich will be includedin subsequent volumes. B. THEDEVELOPMENT CONTEXTINTAJIKISTAN 8. Tajikistan is a low income country which is recovering from civil war and economic collapse and where economic and budget management is impeded by severe institutional weaknesses. The World Bank, alongside other donors, is providing support to policy reforms through a series of Programmatic Development Policy Grants (PDPG) and technical assistance projects such as the Public Sector Reform Project (PSRP) and Public Financial Management Modernization Project (PFMMP). Reforms to public financial management(PFM) and public administration reform (PAR) are important components of the overall economic reform program of the GOT, supported by the Bank in the Country Partnership Strategy (CPS) and throughthe PDPGseries. 9. Tajikistan gained independence from the Soviet Union in 1991. During the Soviet era, what is now Tajikistan was a state-controlled command economy dominated by the cotton and aluminum industriesand heavily subsidized by transfers 2 from the rest of the Soviet Union. The Soviet era left two important legacies for the independent state of Tajikistan. First, much of the productionstructure was very rigid and ill suited to the demands of a market economy. Second, Tajikistan inherited a relativelywell developedeconomic and social infrastructure. 10. Following the break up of the Soviet Union, Tajikistan suffered economic collapse as a result of three factors: the disruption to integratedtrade and production systems within what had been the Soviet Union, the ending of subsidies from the Soviet Union and a civil war which raged from 1992to 1997. Real GDP fell by more than 70% before it beganto recover at the end of the 1990s. The restorationof peace and security throughoutthe country combined with the bringing back into production of underutilized capacity allowed for a rapid recovery of output, with real GDP growth averaging just under 10% per annum during 2000-2005. Nevertheless, aggregate real output is still about 30 percent lowerthan its level in 1990. 11. Moreover the living standards of the Tajik population suffered both from the fall in real GDP and the ending of subsidies from the Soviet Union. Tajikistan now has the status of a low income country, with a per capita GDP of around US$360 per annum. With employment opportunities severely limited inTajikistan, approximately one third of the working age population is estimated to migrate abroad for work, either on a seasonal or more permanent basis, with the majority of migrants working in Russia. The incidence of income poverty was estimated at 64 percent in 2004, although the country fares better with respect to some of the indicators of human development. ' 12. The economic collapse and ending of Soviet subsidies also had profound implications for the real value of government's budget resources, which fell dramatically. Prior to independence, about 40% of the budget for what is now Tajikistan, was subsidized by transfers from other parts of the Soviet Union. The Government sector inherited by the independent Tajikistan is far too large to be funded adequately by the much reduced budget resources available to the Government, even though these budgetresources are now supplemented by donor aid, mainly in the form of project support, amountingto around 4% of GDP. 13. Because there was no real effort to cut the size of public employment, real wages in the Government sector have plummeted to levels far below that which is required to provide even a minimum subsistence level of income; while funding available for the operationaland maintenanceexpenditures of government agencies is only a fraction of that required to deliver public services or maintain the public infrastructure. The very low salary levels are an inducement to corruption among public servants. 14. Since the end of the 1990s, the GOT has implementedeconomic reformswhich have begun to yield positive results. Macroeconomic management in the 2000s, supported by a Poverty Reductionand Growth Facility (PRGF) programagreed with the IMF, has improved,with inflationreducedto single digit levels in 2004 and 2005. As noted above, the economy has recovered more than half of the fall in production ' Human development indicators are generally better than the average for low income countries. For example, life expectancy at birth is 66 years, compared to the low income country average of 59 years. Gross secondary school enrolment is 82% compared to an average of 46% for low income countries. 3 suffered in the 1990s. The external public debt of Tajikistan has been reduced as a result of debt relief agreements negotiated with Russia and other bilateral creditors and more recently the Multilateral Debt Relief Initiative (MDRI) with IMF.2 Tajikistan has accumulated balance of payments surpluses in the last few years, enabling the country's foreign exchange reserves to be gradually rebuilt. Hence the economy does not suffer from serious domestic or external macroeconomic imbalances. 15. Structuraland institutional reformsare now the major challenges for Tajikistan as it seeks to develop its economy and reduce poverty. The priorities for reform efforts are: firstly, the promotion of private investment to provide employment and gradually shift the economy away from the rigid capital intensive production structures built up during the Soviet era and secondly, the restructuring of the public sector so that it can focus scarce public resources on the efficient provision of essential public goods and services. The latter issue is discussed in more detail in the following sections. c. MAINFINDINGSANDRECOMMENDATIONSTHE2003/04PEIR OF 16. The 2004 PEIR focused on the following issues: Strategic Policy Formulation; Budget Management and Execution; the Centre-Sub-National Relations; and the Public Investment Planning and Process. Since the 2004 PEIR was the first such activity for Tajikistan, it focused on providing an overview of all the PFM issues. The mainfindings ofthe PEIR are as follows. The budget is very poorly aligned to strategic policy formulation, mainly because the public administration is still organised along the lines of the old Soviet model, in which the budget was subordinated to central planning. The roles and functions of different institutions within public administrationare often unclear or overlap. Fiscal relationships between central and local governments are highly complex and impede rational budget preparation linked to clear national priorities. Public administration is hampered by a severe lack of capacity, very low wages and corruption. Capitalexpenditures are fragmentedand not integratedwith the current budget. The delivery o f public services has deteriorated because of budget resource constraints exacerbated by the weaknesses in budget management. 17. Some of the recommendations of the 2004 PEIR have led to important reform measures being implemented such as the establishment of a National Treasury, a budget audit agency and the enactment of a public finance law. However, the overall recordof implementingbudget reforms has beenpoor due to the very weak capacities In 2000, external public debt had risen to 128% of GDP but this was reduced to just under 40% of GDP in 2004 as a result of debt cancellationby creditors andthe growth in GDP (IMF, 2005). In 2006, , the stock o f debt increasedwhen the GOT contractedloans of US604 million from China. 4 within the public service for understanding and managing reforms, a lack of strong ownership from the GOTand a lack of political will and specific technicalknow-how with respect to how to implementbudget reforms. The sections below providefurther details on the actions implementedunder each category reviewed in the PEIR. The remainingchallenges for budget system reform are discussed in detail in Section 4: these are areas which require additional analysis which is undertaken as part of the PPER. (a) Strategic Policy Formulation and Policy Management 18. Tajikistanhas retained most of the organizationalfeatures of government from the Soviet era, which are illsuited to the effective formulationand implementationof policy in a market economy. The policymaking and budget processes are highly fragmented, without a clear demarcation of responsibilitiesfor different agencies and with several parallelresourceallocationmechanisms. Although some steps to address these problems have been taken, notably the endorsement of the PARS, the establishment of a civil service register, the submission to Parliament of a civil service law and the endorsement of a civil service wage reform strategy, the major institutional reforms required to facilitate policy based budgeting, such as the implementationof the MTEF, have yet to be implemented. (b) Budget Managementand Execution 19. Giventhe institutionalweaknesses referredto in the above paragraph, it is very difficult to translate strategic policies into coherent budget allocations. This has had profoundconsequences for the budget, because with the budget sufferinga substantial decline in real resources, a strategic re-allocation of budget resources is essential. Instead Government has attempted to maintain all existing budget structures even though there are insufficient budget resources to fund them adequately. Some important budget reforms have been implemented, such as the enactment of a Public Finance Law, the establishment of a National Treasury and establishment of an external audit agency. However the major reforms to the organizationalstructure of the budget which are necessary to facilitate policy based budgeting have yet to be undertaken. These reforms, focused around introduction of the MTEF, are at center of the budget reform strategy to be implemented over the medium-term and are supported under Theme 2 of the PPER (c) Public Investment Program 20. Public investment expenditures are fragmented between the externally funded PIP and the CentralState InvestmentProgram (CSIP) funded by the State budget, are not properly integrated into the rest of the budget and are not guided by a single development strategy. Also, there is no formalized methodologyfor the selectionof investment -projects according to transparent criteria. The Government has taken some actions to improve the PIP, which is now prepared on a three-year rolling basis and updated annually, while the external borrowingrequirement associated with the PIP must be submitted to Parliament for approval. The priorities for reform in this area are to integrate planning of investment expenditures with the rest of the budget and to adopt rigorous formal procedures for evaluating and selecting projects for inclusionin the budget. 5 (d) IntergovernmentalFiscalRelations 21, Sub nationalunits of administrationplay a major role inthe provisionof public services, includingeducation, healthcare, and housingand communityservices. The system used to finance these services continues to bear the imprint of the Soviet era, requiring lengthy negotiationsbetween tiers of government over spending needs and revenue prospects. As a result, sectoral ministries have little influence over the allocation of resources at the local level. The PEIR discussed several options for reform. At the strategic level, these includeda shift of budget makingauthority from sub national units to the sectoral ministries. The Government has opted for this approach, which is a prerequisitefor introducinga sector-based MTEF (see Chapter 4), but is still working out the proceduraldetails of how to do this. The PEIR also recommendedthe introductionof moretransparent criteria into the allocationof fiscal transfers. With the exception of the experiment in capitation-based education financing, no progresshas beenmadeon this front. D. KEY CHALLENGES FORPUBLICEXPENDITUREREFORM 22. The main challenges for reformof public expenditures, which are addressed in the PPER, relate to the allocationof scarce budget resources to meet strategic policy objectives, the efficiency with which budget resourcesare used for service delivery, a range of fiduciary issues and the budgetary relationships between different tiers of government. The identification of these challenges was guided by the priorities identified in strategic planning documents of the GOT,such as the NDS, the views expressed by GOT officials, the findings of the PEIR and other diagnostic work undertakenby the Bank and other donors (e.g., the CFAA). They also reflect lessons drawn from fiscal reform programs among other CIS countries which have faced similar problems in reformingSoviet model public administrationsystems. 23. While there has been a marked improvementin the macroeconomic aspects o f fiscal policy management since the late 1990s, important challenges are emerging which may threaten long term fiscal sustainability and macroeconomic stability. There is pressure from both within and outside Tajikistan to increase government spending on MDG related programs, although the budgetary resources that is available to fund an expansion of such programs, from domestic revenues or donor aid, are likely to be quite limited. The Government has also prioritized large infrastructureprojects in the energy and transport sectors and is seekingto fund these with externalborrowing.It contracted loans of US$604million from Chinain 2006 to fund road rehabilitationand the construction of an overheard power line. It seems probable that Tajikistan may be able to access further large external credits to fund infrastructureprojects,but the Debt Sustainability Analysis (DSA) undertaken by the Bank and Fund in early 2007 (IMF and World Bank, 2007) found that Tajikistan faces a highrisk of debt distress becauseof the sharp rise in new borrowing. 24. Any substantialincrease in Government expenditures also raise concerns over the absorptive capacity of Government, the burden imposed on the state budget to fund counterpartcontributionsto foreign financed capital projects and the impact on inflation through increased aggregate demand. After having been broughtdown into single figures through prudent fiscal and monetary policies in 2004 and 2005, 6 inflation rose to more than 12% in 2006, buoyed by a surge in aggregate demand as well as supply side shocks. 25. The allocation of budget resources involves two distinct but related problems: first, the process by which budget resources are allocated(the institutionalrelations of the budget process) and secondly, the substance of the budget allocations (i-e., how much does Government spend and what does it spend its resources on). The main developmental priority is to promote rapid and sustainable economic growth so that real per capita incomes can be raised. This implies that a key priority for government spending should be to support rapid economic growth. The robust economic growth recorded since 1998 was largely driven by the rebound from economic collapse during the 1990s, with existing productive capacity (e.g., in the cotton and aluminum sectors) brought back into production. But, as the economy moves back towards its production possibility frontier, it does not seem likely that these high rates of economic growth can be sustained for much longer without attracting new private investment in commercial enterprises and without public expenditures which can prevent the deterioration of existing infrastructures and/or build new infrastructure which complements private investment. In this context, how best to allocate budget resources to support economic growth is a key policy question. 26. The size of public sector - 385,000 employees - and the scope of its activities are too large given the very limited budget resources available to efficiently fund these activities. To avoid drastic cuts in public employment, real salaries have been cut to levels that are often too low to meet subsistence needs. The Government has also inherited from the Soviet period an extensive portfolio of capital assets (including economic and social infrastructure) which it cannot afford to operate or maintainproperly. 27. Therefore, the public sector needs to scale back its activities and focus scarce budgetary resources on areas where market provision of goods and services is not feasible. However, even within the general area of public goods and services, rigorous prioritization will be necessary because legitimate demands far outstrip potential budget resources. There are important choices to be made between competingexpendituredemands; for example, between human resource development, social protectionthrough transfer payments to vulnerable individuals, maintainingthe existing infrastructure and constructing new infrastructure. The trade offs in these situations are not always clear or easy to evaluate. The analytical work of this PPER is intendedto facilitate understandingof these tough choices. 28. Fiduciary issues have moved to the center of policy discussions on financing government programs for economic growth and poverty reduction. Although Tajikistan moved from "post conflict" to a "gradual developer" three years ago, donor projects are heavily focused on social service delivery, and a majority o f these projectsare funded off budget and bypass local institutionsdue to capacityconstraints and fiduciary concerns. This has led to a lack of coherence between the Government budget and donor projectsresultingin a fragmentationand duplicationof local service delivery. Unfortunately, independent assessments have identified serious concerns related to fiduciary risk in the budget and for this reason donors are reluctant to provide more of their aid in the form of budget support. In order to address these donor concerns, Government needs to undertake periodic reviews of fiduciary risks 7 which will assess the progress and recommend corrective measures to address these challenges. 29. Tajikistan's intergovernmental fiscal relations are an important challenge for improvingPFM. Because a majority of socialexpenditures are implementedthrough the local government budgets, it will not be possible to implementan MTEF based on sector planning in the key sectors without reforming intergovernmental fiscal relationships. As such, the recently adopted MTEF action plan includes changes in budget planning in terms of the role of line ministries and local Hukumatswhile per capita financing requires changes in national-sub national relations and resource allocationmechanisms. E. THETHEMES AND CONTENTSOF THEPROGRAMMATICPUBLICEXPENDITURE REVIEW 30. Given the challenges outlined above, the PPER provides analysis to assist the Government to design and implementreformsunder the following five themes. Theme 1: Public Expenditure,FiscalSpace and Growth 3 1. How best to allocatebudgetresourcesto support economic growth is addressed in Theme 1 of the PPER, through a DFID-financed study on the determinants of economic growth and their implicationsfor public expenditure. The study examines the factors behind Tajikistan's recent strong economic recovery and assesses what will be required to sustain economic growth over the long term. It identifies which sectors of the economy are likely to be the drivers of growth over the long term, in light of the country's potential comparative advantage, and reports the results of an endogenous growth model to examine econometrically the causal links between different types of public expenditure and growth in Tajikistan and other CIS countries.The findings of this study are presented in Chapter 2 of this volume of the PPER. Further work under this theme will examine the implications for poverty reductionof differenttrajectoriesof economic growth. 32. There are important questions pertaining to the potential size of the budget resourceenvelope and therefore the scope for expanding expenditures inpriority areas to support growth and povertyreduction: for example, what scope is there for growth of domestic revenues?; what is a sustainable level of public sector borrowing?;how much more grants from foreign aid donors can be mobilized?;can resources be freed up through improvements in expenditureefficiency? These questions are covered in Chapter 3 of this volume of the PPER which includes long term projectionsof fiscal space utilizingthe analyticalframeworkproposed in a recentjoint IMFBank report to the Development C~mrnittee.~ Theme 2: Policy-Based Budgeting 33. Improvingthe allocationo f budget resources will require institutional changes to the budget process. The existingbudget process has been shaped by the structures "Fiscal Policy for Growth and Development:An Interim Report", Backgroundreport by the IMF and World Bank for the DevelopmentCommittee Meeting, April 2006. 8 inherited from the Soviet era, in which the Government budget was a component of the broader system of economic central planningrather than having its own separate functions and objectives,and by the decentralizationwhich took placeafter the end of the civil war, which established a system of dual subordination which left the role of line ministries unclear. The budget process is extremely fragmented and complex, with over 100KBOs each havingtheir own budget and dealingdirectly with the MPF in the budget process. Many of these KBOs have responsibility for expenditures covering multiple functions. The line ministries have responsibility for policy formulation but their direct control over budgets in their respective sectors is highly circumscribed by the fragmentation of the budget. The structure of the existing budget system providesno institutionalmechanismsfor allocatingbudget resourcesin a coherent and transparent manner to meet strategic policy objectives, hence it is very difficult to link the strategy outlined in the NDS or PRSP to the budget. Instead, annual budget allocations are determined by a combinationof rigid formulae linking allocationsto existingstructures and bargainingbetweenthe central institutions(such as the MOF) and KBOs. In addition, the current and capital budgets are essentially determined separately, with the capitalbudget controlledby the Ministry o f Economy and Trade. 34. Introducing policy based budgeting in Tajikistan will require major institutional changes to the budget process. These changes will entail reorganizing the budget on a sectoral basis and strengthening the role of line ministries in the budget allocations for their respective sectors together with the integration of the current and capital budgets. These issues are covered in Theme 2 of the PPER. The studies prepared under this theme include an analysis of the major changes in the budget process and in relations between different budget organizationswhich will be necessary to enable the MTEF to be implemented in Tajikistan, which is presented in chapter 4 of this volume of the PPER. Subsequent volumes of the PPER will include analytical work related to the specific needs of the education sector, which is pilot sector which will spearhead the implementation of the MTEF. This will include analysis of how the line ministry will relate to the myriadbudget organizationswhich implementeducation expenditures. Theme 3: Efficiency of Public Expenditurein Social Sectors 35. Theme 3 examines the efficiency of expenditures in the social sectors. The focus is on the social sectors, rather than the budget, because these sectors will play a crucialrole in the country's effortsto achieve the MDGs. Donorsare keento provide additional financing for these sectors (e.g., The Education for All-Fast Track Initiative), but are concerned about how efficiently funds will be spent, and also because it'is not feasible, in terms of resources, to employ the methodologies proposed to evaluate efficiency(e.g., ExpenditureTracking Surveys) across the whole budget. 36. The social sectors (mainly health, education and social protection) comprise approximately half o f all non-PIP government expenditures. As noted above, relatively well developed but expensive and structurally rigid educational, health and social protectionsystems were put in place duringthe Soviet era but the sustainability of these systems and the quality of service delivery has been severely underminedby the contraction of the Government's budget resource envelope over the last decade 9 and a half. These problems are compounded because the limited evidence which is available indicates that the resources which are channeled to the social sectors are not allocated in an optimal manner and that operational efficiency is poor because of weak management systems, lack of appropriate incentives and corruption. Hence, improving allocative and technical efficiency in the social sectors is essential if strategic public expenditure objectives in these sectors are to be achieved. However more detailed diagnostic analysis of the allocation and use of resources in the social sectors are requiredto inform recommendationsfor policy reform inthis area. 37. Work carriedout under this theme will includean examinationof the technical efficiency of expenditures in the education and health sectors which utilizes the efficiency frontier methodology. 38. PETS in the education and health sectors will be conducted in collaboration with the Ministriesof Educationand Healthand local government^.^ The findings of the PETS will be reported in subsequent volumes of the PPER. The PETS will provide a complete pictureof resource flows in the education and health sectors (e.g., how much is spent and on what?how do different KBOs obtaintheir funds? how are budget decisions made? and how are funds managed in the KBOs?). They will identifythe supply side factors that break the chain betweenthe spendingof resources and its transformation into service delivery. Equity of access to educational and health services will also be covered by PETS and household survey analysis. The PETS will establish base line indicatorsfor monitoringthe use of resources inthe key social sectors. In addition, the findings of the PETS will help to improve transparency and accountability in budget execution and the resource utilization in both sectors. Furthermore, it will provide important inputs and recommendations for policy reform necessary for the development of sector wide approaches (SWAPS)in the education and health sectors. Although the PETS will be launched in FY 2007, the analysis and dialogueon the recommendations is scheduledthrough FY 2008 and FY 2009. This theme will also include a study of social protection which will be launchedin FY 2008. Theme 4: Fiduciary Risks 39. The Bank and other donors have supported several diagnosticstudies - of the PFM system, including the Bank's 2003 Country Financial Accounting Assessment (CFAA), IMF's 2003 Safeguards Assessment and DFID's 2005 Fiduciary Risk Assessment, which have concluded that serious weaknesses in the system of PFMs are a major impediment to improving the efficiency of budget resource use in Tajikistanand deter aid donors from providinga greater share of their aid in the form of budget support rather than the less efficient project aid (although there are circumstances where project aid is a more suitable vehicle for delivering aid, for example when technicalassistance or externalknowledgeis required,the composition of donors' aid to Tajikistan is very heavily weighted towards projects, which comprise 96% of projected aid to the Government in 2007, with budget support comprisingonly 4%). See Annex 2 for an expanded concept note, `Public Expenditure Tracking Surveys in Education and Health' which examines the rationale for, and objectives of, the proposed PETS in more detail 10 40. The complexity and scale of problems in fiduciary systems has meant that implementation of the PFM reform agenda has so far been slow and difficult. Nevertheless it is imperativethat PFM reforms do not allow faltering, hence Theme 4 o f the proposed PPER, covers the PFM issues, using the PEFA methodology. The rationalefor includingPFM issues in the PPER is not to simply add further diagnoses to those which have already been undertaken, but to specificallyfocus on what needs to be done to facilitate implementationof PFM reforms. 41. The major objectives of Theme 4 are to review the progress made in implementingthe PFM reforms recommendedby previousdiagnostic studies such as the CFAA and CPAR, to assess whether, and to what extent, the recommended reforms have been implementedunder on-going World Bank, IDF and other donor TA projects, to identify the obstacles to implementingthese reforms, to set out the remainingagenda for PFM reforms and to provide guidance for its implementation. The accepted framework of PEFA is used as the methodologyfor benchmarkingthe 2003 CFAA, CPAR and PEIR assessments and then to review the fiduciary status of public finances in 2006. T he PEFA framework includes a government self assessment of the PFM system which is intended to ensure maximum government participationin and ownership of the process. Key issues which are examined under Theme 4 include:the adequacy of PFM systems, measuresto preventcorruptionand misuse of funds, the reporting o f fiscal data to the public and the quality and transparency of these data, the adequacy of internal and external audit systems, the effectiveness of the State Financial Control Commission (SFCC) and procurement reform'. A summary of the PEFA Report, which includes all 31 of the PEFA ratings, is presented inChapter 5 of this volume of the PPER. Theme 5: Inter-Governmental Fiscal Relations 42. Reform of intergovernmentalrelations cannot be avoided if the Government is to implement a sector-based MTEF, as proposed by the MOF at its MTEF Round Table in May 2006. Hence theme 5 of the PPER covers intergovernmental fiscal relations. It examine the process of allocating budget resources and the roles played by central ministries, line ministries and local governments (oblast, rayon, jamoat), the arrangements for revenue sharing between the center and local governments, and the implications which these arrangements have for resource allocation. It also examines the implications for local government budgets of introducing per capita financing in the health and education sectors, and more broadly the PAR strategy. The objective of this analysis is to generate recommendations for streamlining the local government system consistent with the PAR strategy and integrating it into a budget system which can support policy based budgeting. The work in Theme 5 is closely linked to the MTEF implementationaction plan and the recommendations in the PAR strategy: as such, the priority in the first year of the PPER is to provide guidance on restructuring intergovernmental relations in the budget preparation processfor the education sector, which will be the pilot sector for the MTEF. Further work under this theme will be determined by the needs of the other sectors which are subsequently brought into the MTEF as well as by the lessons learned from the httD:liwww.Defa.org provides a description of the methodologyand the steps involved in carrying out a PEFA assessment, 11 education sector beginning FY08. The work under this theme will be reported in subsequent volumes of the PPER. F. SEQUENCING THEWORK OFTHEPPER 43. The PPER will be carried out over the course o f three fiscal years (2006/07- 2008/09) and publishedinthree volumes, of which this is the first. The sequencingof the work has been guided in part by the requirements for supporting the ongoing program o f PFM reforms which is being implemented by the GOTand by the needto respond to pressing demands for analytic work on how best to align the limited resources of the budget with the Government's strategic objectives for growth and poverty reduction. Hence Chapters 2 and 3 of this report, which address the issue of how best government expenditures can support economic growth and the potential fiscal space available to fund priority expenditures, were identified as priority areas for analysis and have been includedinthe first volume of the PPER. 44. The analysis in Chapter 4 o f the first volume, outlining a strategy for implementingthe MTEF, responds to the decisiontaken by the Government in 2006 to implement an MTEF, beginning in 2007, and to seek TA from donors to support this. Further work under Theme 2 will build on this strategy to support the MTEF reforms in the pilot sector and other sectors. This work has already begun in the education sector and will be includedin subsequent volumes. Inadditionto the sector work, the next set of issues which are pertinent to the implementationof the MTEF involvethe inter-governmentalfiscal relations which are covered under Theme 5 and these will also be addressed in subsequent volumes of the PPER. 45. The PEFA assessment is also a priority for the PPER for two reasons. First, because it providesessential diagnostic analysis on which to begindevelopinga PFM reform strategy and secondly, because a comprehensive fiduciary assessment is a prerequisite for many donors to consider providing budget support to Tajikistan, which is essential to create more fiscal space for priority expenditures. Hence, the PEFA assessment was conducted in the first year of the PPER and a summary is includedin this volume of the PPER. 46. The work under Theme 4, which examines the efficiency of public expenditures in the social sectors, was started in the first year o f the PPER but, because of the time requiredto design, prepare and carry out the PETS and analyze the data as well to complete the frontier efficiency analysis, the results will be reported insubsequent volumes of the PPER. 47. Drawingon the analysis o f the PPER, and especiallythe findings o f the PEFA, a PFM reform strategy and action planwill be drawn.This will identify the priorities for PFM and set out a sequenced action plan for achieving these reforms. The PFM reformstrategy and action planwill be includedina subsequent volume ofthe PPER. 12 2. ECONOMIC GROWTH: CONSTRAINTS, CHALLENGESAND IMPLICATIONSFOR PUBLIC EXPENDITURE A. INTRODUCTION 48. Sustained and rapid economic growth will be a necessary, but not sufficient, condition for the eradication of mass poverty in Tajikistan. Although the recent growth performance of the economy has been very strong, this largely represents a recovery from the very steep economic contractionwhich took place in the first half of the 1990s, which was made possible by bringing back into production under- utilized factors of production. Growth will inevitably subside once the scope for further increases in the capacity utilization of the existing capital stock is used up, unless a combination of new capital investments and improvements in factor productivity provides new sources of growth by expanding aggregate supply. The role of fiscal policy in supportingthe recoveryof the economy has primarily involved its role in macroeconomic stabilization, but future growth will depend far more on a strengtheningof the supply side of the economy, critical elements of which, notably public infrastructureand humancapital, will dependon budgetary policies. 49. The contention of this chapter is that fiscal policy, and especially public expenditurepolicies, will not only be important for supporting sustainable long term economic growth, but will also shape the pattern of economic growth and development in Tajikistan. It is possible to envisage alternative long-term growth trajectories for Tajikistan, distinguished in part by whether public expenditure priorities are focussed on large infrastructureinvestments, or on social services and human capital development, and the extent to which government improves the investment climate for the private sector and economic governance more generally. These trajectories have markedly different implications for poverty reduction and humandevelopment. 50. This chapter argues for a balanced development strategy in which the priorities for public expenditure are: (i)the strengthening of human capital, mainly through the improvement of basic education and health services, and (ii)the maintenance of the core networks of the existing infrastructure (especially,those attract private investment). . Such a strategy offers the prospect of economic growth components of the infrastructurewhich comprise public goods which are unlikely to combinedwith employment creation, poverty reduction and improvements in human welfare. 51. One of the motivations for this analysis is that developing countries, which have followed a strategy of heavy public sector-led investments in infrastructureand capital intensive projects funded by external borrowing often failed to achieve sustainable long-termgrowth or povertyreductionand eventually ran into an external 13 debt crisis which had extremely damaging economic and social consequences. It is clearly importantfor Tajikistanto avoid makingthese mistakes. 52. The chapter is organized as follows. Section 2 reviews the performance of the economy since independence, analyzes the factors which explainthe trends in growth and assesses the implications of this for future growth. Section 3 evaluates the constraints to growth in Tajikistan, drawing on the findings of several recent World Bank studies. Section 4 examines the potentialcomparativeadvantage of Tajikistan, what this implies for possible long-term economic growth trajectories and how such trajectories will be shaped by economic and budgetary policies. Section 5 provides some econometric evidence on the links betweendifferenttypes of public expenditure and growth in a sample of transition economies which includesTajikistan. Section 6 concludes and examines the implications for public expenditure policies of the analysis of the constraints to, and prospectsfor economic growth in Tajikistan. B. EXPLAININGTAJIKISTAN'S ECONOMICGROWTHPERFORMANCE 53. Tajikistan's economy contracted steeply in the six years following independencein 1991. The low point in output occurred in 1996, by which time the cumulative fall in GDP was 68%. In the next few years the economy began to recover, slowly at first, with growth acceleratingin the 2000s, averaging nearly 10% per annum during 2000-2004, before subsidingto 6.7% in 2005. By 2005, the Tajik economy had recovered 30% of the fall in GDP which had taken place in the 1990s, so that real GDP in 2005 was about 60% of the level in 1990. The decline in agriculture was steeper than that of the GDP and its recovery was not as rapid. The agriculture's share in GDP fell by 15 percentage points over the 1991-2005 period, whereas the share of industry in the economy is now at approximatelythe same level as it was at independence. 54. Tajikistan's economy followed a similar path to those of the other CIS countries, all of which suffereda decline after the break up of the Soviet Unionat the start of the 1990s (dubbed the "great contractions" by De Broeck and Koen 2000) followed by a recovery beginning later in the 1990s. However, the magnitude of the initial drop in GDP, its duration and the pace of recovery, differed among these economies. Tables 2.1 and 2.2 compare Tajikistan's GDP growth with that of other low and middle income CIS economies since the break up of the Soviet Union. In five countries (Armenia, Azerbaijan, Kazakhstan, Uzbekistan and possibly Turkmenistan, [althoughdata after 2001 are not available], realGDP hadrecoveredto more than the 1990 level by 2005. The economy of the Kyrgyz Republic had only attained 80% of its 1990 level by 2005, while those of Moldova, Georgia and Tajikistan were still between 40% and 50% lower than in 1990. Those countries whose output has now surpassed its 1990 level all suffered a smaller initial contractionof GDP than Tajikistan, and begantheir recovery earlier than Tajikistan. 55. In the decades prior to the break-up of the Soviet Union, the economy of the USSR had undergone extensive economic growth; i.e., growth was generated by increasingthe supply of factor inputs, especiallycapital. Consequently, the economy was characterizedby high capitaVoutput ratios. The capital/outputratiowas officially estimated at around three in 1986but some Western estimates put it higher, at around five (Easterly and Fischer, 1994: 32). The elasticity o f substitution between capital 14 and labor also appears to have been low, which together with the high capital/output ratio, implies diminishing returns to capital, which can explain the long term slowdown in economic growth in the Soviet Union (Easterly and Fischer, 1994).6 In the Tajikistan Republic of the USSR during 1971-1990, the average growth in the capital stock was 5.2% per annum, similar to the average of 5% for the USSR as a whole. Output growth in Tajikistan during 1971-90was 2.6% per annum, similar to the rate of 2.3% for the whole o f the USSR. However, the growth in the labor force in the Tajikistan Republic was 3% per annum, much higher than the average o f 1% for the USSR (De Broeck and Koen, 2000: 15). This would imply that the capital/outputratio in Tajikistan was similar to that of the average for the USSR but that its capital/laborratio was lowerthan the average for the USSR. 56. The problems in the economy were evident long before the break-up of the Soviet Union. In the 20 years before the break-up (1970-1990), its growth of output per worker was worse than that o f the Soviet Union as a whole, which was itself on a downwardpath. Inthis period, annual growthper worker in Tajikistanwas only 1%, compared to 2.8% for the USSRq7Growth in all of the Central Asian republicswas below the average for the USSR in this period. The worst performing sector in Tajikistan was agriculture, with average growth per worker during 1970-1990 of negative 1.8% per year (Easterly and Fischer, 1994: 44, 45). The poor performance of agriculturecharacterized all of the CentralAsian republics in this period, possibly becauseofenvironmentalproblemscausedby over-investmentin cotton. 57. Several researchers have used a growth accountingframework to analyze the sources of economic contraction and growth in the CIS countries (De Broeck and Koen, 2000), low income CIS countries (Loukoianovaand Unigovskaya, 2004) and Tajikistan(Matovu, 2005), following their independencefrom the Soviet Union. The growth accounting framework disaggregates GDP growth into three components: those attributable to the growth in the capital stock, the growth in the labor force (and labor productivity), and the growth in total factor productivity (TFP). The latter is essentially a residual which picks up all of the influences on the productivity of factors of production, as well as measurement errors: TFP does not just reflect exogenous technological progress (De Broeck and Koen, 2000). These growth accounting exercises assume a constant return to scale production function with elasticitiesof outputwith respect to labor andcapital of 0.7 and 0.3 respectively. 'Note the negativeTFP growth rates for the periods 1981-85 and 1986-90shown in Table 2.4. These growth rates refer to Net Material Product (NMP), which was the definition of output in the Soviet Union. The growth rates of NMP per worker reported in Easterly and Fischer (1 994: 44) are higher than those implied by the data in De Broeck and Koen (2000: 15); the latter imply negative growth rates of output per worker in Tajikistan, given that annual rates of labour force growth (3 percent%) were higher than output growth (2.6 percent%), but both sources indicate that growth rates of output/NMP per worker were lower in Tajikistan than the average for the USSR. 15 0 P 0 hl m 0 0 hl hl 0 0 hl 7 0 0 hl 0 0 0 hl Q) Q) Q) 7 03 Q) Q) r b Q) Q) 7 to z 0) In Q) Q) 7 z t Q) m Q) Q) 7 3 fi 9) 3m 4m L 9) E E .I 0 E .I I c, 2: h I sa d .I v1 z.. u E s 9) F 58. Matovu (2005) found that TFP growth was the main cause of the collapse and then recovery inoutput inTajikistan. Matovu's results are shown in Table 2.4. TFP growth was strongly negative in 1991-95, slightly positive during 1996-2000, and very strong during 2001-04, mirroring the trends in output. The capital stock was virtually stagnant throughout the period since independence, although there was a slight recovery (growth of 1.1% per annum) in 2001-04. The labor force declined in the 1990s by about 1% per annum, but then grew by 3% per annum in 2001-04. The results for Tajikistan in the 1990s reportedby Loukoianova and Unigovskaya and De Broeck and Koen are consistent with those of Matovu. The recovery in GDP in Tajikistan was, therefore, driven mainly by a recovery in TFP, and to a lesser extent, growth in the labor force. Matovu estimates that TFP growth contributed about three quarters of the GDP growth during 2001-2004, boosting GDP growth by an average of 7.2% per annum. 59. The trend in TFP in Tajikistan was similar to that of most other CIS countries. Throughout the CIS region, there was a decline inTFP associatedwith the contraction in output, while TFP growth was the main contributor to the subsequent recovery in output. Except in Azerbaijan, capital investment made no more than a very minor contribution to the output recovery in the low income CIS countries (Loukoianova and Unigovskaya (2004). 1981-85 1986-90 1991-95 .1996-2000 2000-04 Capital 4.6 2.9 0.2 -0.1 1.1 Labor 3.1 2.9 -0.9 -1.1 3.0 TOP -2.4 -1.6 -15.5 1.3 7.2 output 1.2 1.3 -16.1 0.5 9.7 60. The contraction in output in Tajikistan was caused by three factors. First, the Tajik economy was integrated closely into that of the USSR, but the break-up of the Soviet Union severely disrupted trade patterns and production, and led to higher prices of imported materials and fuel which had previously been subsidized. Secondly, the economy suffered a huge fiscal shock when transfers from the Soviet Union, which had accounted for almost half of the Government budget during the Soviet era, ceased at independence (IMF, 2006B: 4). Thirdly, the civil war which afflicted the country between 1992 and 1997 severely disrupted the economy. As a consequence of the output collapse, much of the country's capital stock was under- utilized in the 1990s. As output fell much faster than capital (which can only adjust very gradually), TFP also fell. As such, the economy was operating far below its Production Possibility Frontier (PPF) by the mid 1990s. The output collapse was accompanied by severe macroeconomic imbalances, with huge fiscal deficits largely financed through money creation, hyperinflation which exceeded 2000% in 1995 and 18 a rapid build up of external debt, mainly in the form of trade credits from other CIS countries which exportedgoods to Tajikistan.* 61. The economy began to recover in 1997, with growth accelerating in 2000. Growth appears to have been driven mainly by a recovery in aggregate demand, especially private consumption, which in turn was boosted by an expansionof official capital inflows and workers' remittances. Given that the economywas operatingwell within its PPF in the mid 1990s, an expansionof supply capacity was not requiredfor a recovery of output; all that was required was to bring back into production the existing under or un-utilized productive capacity, as occurred in the aluminium industry, for example. T he movement back towards the PPF is reflected in the recovery of TFP which is found inthe growth accountingestimates. 62. The economic recovery was boosted by two factors. First, the restoration of peace and security in 1997, and secondly, the fiscal adjustment which began in the second half of the 1990s and allowed the macroeconomyto be stabilized. As noted above, the pattern of economic contraction and recovery in Tajikistan was similar to that of other CIS#economies. Segura-Ubiergo, Simone and Gupta (2006) analyze the relationship between fiscal adjustment and growth in the CIS countries during the period 1992-2001and argue that fiscal adjustmentboostedeconomic growth because the reduction in the monetization of fiscal deficits and the enhanced credibility of macroeconomic management helped to stimulate an inflow of donor inflows and private capital (as well as remittances)which contributedto the recovery of aggregate demand. c. CHALLENGES AND CONSTRAINTSTO SUSTAINABLE FUTUREGROWTH 63. It is unlikely that Tajikistan can rely on the same sources of growth which dominated the recovery of the economy during the 2000s, notably the high rates of TFP growth made possible by a shift back towards the economy's PPF, driven by an expansion of aggregate demand, for much longer. This is because most of the spare capacity which characterizedthe economy in the mid 1990s has now been brought back into production and very little new capacity has been added, because rates of investment have been low (Table 2.3). In addition, it is possible that the PPF has actually shifted inwards since independence because of a decline in the size and quality of the labor force, caused by the migration of workers to seek work outside Tajikistan, although we do not have accurate and up to date statistics to substantiate this. 64. To explore whether the economy has now returned to close to its PPF, Chart 2.1 displays two estimates of capacity utilization which differ in the assumed average depreciation rate of the capital stock. To estimatecapacity utilization, we apply to an initial estimate of the capital stock the reported rates of gross fixed capital formation (GFCF) and assumptions of the depreciation rate. Tajikistan's peak output was in 1988. In the absence of data on the capital output ratio for Tajikistan, we make the assumptionthat the capital output ratio in the peak year for output was three, which is in line with the official estimate for the USSR as a whole in 1986, but lower than Tajikistan had no external debt at independence, but by 1996, its public and publicly guaranteed external debt had reachedUS$867 million, equivalentto 84% of GDP. 19 Western estimates. It is not clear whether the Tajik economy was more or less capital-intensive than that o f the USSR as a whole, given that the former comprised sectors which were probably much more capital intensive than the average for the USSR (hydropower and aluminum smelting) and sectors which were less capital intensive than average (agriculture). 65. We assume that there was full capacity utilization in the peak year o f output, and that the real value of the capital stock was the same in 1990 as in 1988. Real GDP in 1990 was 94% of the peak year level (De Broeck and Koen, 2000: 9), which implies that capacity utilization in 1990 was 94%. For depreciation of the capital stock, we make two separate assumptions. T he first is the 3% estimate used by Loukoianova and Unigovskaya (2004) for all low income CIS countries. However, this estimate of a 3% depreciation rate only includes the physical deterioration of capital; it does not take account of obsolescence of capital (also called economic deterioration). To incorporate economic deterioration o f capital, we also include a higher, 5% depreciation rate. We use estimates of GFCF in constant prices, from the World Bank country database for Tajikistan. 66. As shown in Chart 2.1, if the depreciation rate was only 3% throughout the 1990-2005 period, the real level o f the capital stock would be at virtually the same level in 2005 as it was in 1990; Le., gross fixed investment matched depreciation on average throughout this period. In that case, capacity utilization in 2005 would still be only just under 70%, despite it having doubled since the mid 1990s. Hence there will still be room for further increases in capacity utilization which could generate more economic growth. However, if the depreciation rate, including economic deterioration, is assumed to be 5%, then the real value o f the capital stock in 2005 would be only 80% of the 1990 level (Le., gross investment has been insufficient to offset depreciation) and capacity utilization would be almost 90% in 2005. In that case, the scope for further increases incapacity utilization will runout quite quickly. 20 Chart 2.1. Estimated Utilisation of the Capital Stock with 3 percent and 5 percent Depreciation Rates, 1990-2005 30% - --i-5%depreciation 20% -. -3% Depreciation 10% - 0% 7. Source: see paras 64-66. 67. The higher rate o f depreciation is probably more realistic than the lower one, for two reasons. First, it is likely that physical depreciation rates in Tajikistan were higher than the average for low income CIS countries because some of the capital stock in Tajikistan was damaged in the civil war in the 1990s. Secondly, economic deterioration cannot be ignored because some of the public enterprises established during the Soviet era are not economically viable in a market economy, and the capital stock of these enterprises cannot readily be used in other sectors of the economy.' Consequently it appears likely that the economy is close to the point where capacity utilization cannot be increased any further and is, therefore, close to its PPF. 68. Ifthe economy is now operating close to its PPF, future growth in Tajikistan will require an expansion of aggregate supply (a rightward shift in the PPF), through higher rates of capital investment and improvements in TFP, including labor productivity. Factor productivity could be improvedby strengtheningcompetition in the economy, by attracting more foreign direct investment which could provide modern management skills and, over the long term, by strengthening human capital through better educationand training which would enhance labor productivity. Given that average per capita incomes are very low in Tajikistan, the scope for mass private sector provision of education is limited and hence the public sector must provide the bulk of educationservices. Labor productivity could also be raised ifthe healthstatus of the workforce could be improved through better health care. Compared to other For example, many of the machine producing factories in Tajikistan set up during the Soviet era relied on inputs brought in from elsewhere in the Soviet Union. These factories became unprofitable after the break up of the Soviet Union because of increasesin the prices of their inputs and the cost of transportation(Umarov and Repkine, 2004). 21 CIS countries, labor productivity in Tajikistan is very low (only 70% of the level in the Kyrgyz Republic andhalfthat inArmenia and Georgiafor example) althoughthis cannot be attributable solely to the quality of the labor force; the high costs of doing business make an important contribution to low labor productivity (World Bank, 2005B: 3). 69. Increased private investmentwill be crucial to an expansionof the supply side of the economy for three reasons. First, given budgetary constraints, the scope for funding much higher levels of public investment on a sustainable basis, without pre- empting budgetary resources for essential current expenditures, is very limited. Secondly, the type of investment which the economy needs most to generate higher growth rates is not investment in public capital such as infrastructure, but investment for the production of marketable goods and services, for which the private sector is much better suited than the public sector. Thirdly, private sector investors, operating under competitive conditions and facing the incentives to make profits, are more likely to generate improvementsin factor productivity than would be the public sector which lacks the same incentives. 70. Private investment has been very low in Tajikistan in recent years, averaging only just over 5% of GDP in the 2000s (see Table 2.5). Given the rapid recovery in aggregate demand since the late 1990s, together with the restoration of macroeconomic stability, one might have expected a stronger response from private investorsto take advantage of the expandingdomestic market, as well as possibly to export to neighboringcountries, some of which are also enjoying buoyantgrowth. Table 2.5: Private Investment,Percent of GDP: 2001-2006 2001 2002 2003 2004 2005 2006* Private 4.2 5.4 5.5 5.4 6.0 6.0 investment 71. One of the reasons for the paucity of private investment is that the institutional climate for the private sector is poor. Surveys by the World Bank have found that the business climate is characterized by unpredictability, a lack of transparency, an uneven playing field for businesses and corruption. Most damagingly, instead of a clear set of rules for all businesses to ensure fair competition, individual firms make their own special arrangements with Government officials (World Bank, 2005B: 22-24). There is a plethoraof complex regulations which act as a barrier to entry by start-up businesses and are a burden for existing firms. Businesses need to obtain many permits from different government agencies; the procedure for obtaining permits is complex, inefficient, not transparent, and costs businesses both money and time. Corruption is pervasive; the findings of the BEEPS indicate that private firms pay bribesto tax and customs inspectors, to local authorities, to obtain permits and to access utilities. The BEEPS reported that 80% of businesses paid bribes to tax inspectors (World Bank, 2005A: 16-17). The operation of'tax administration, which involves the tax liabilities of companies being computed by tax officials rather than through self assessment by taxpayers backed up by risk based audit, is also a 22 hindrance to the private sector, because it allows too much scope for discretionary decisions and predatorybehavior on the part of tax officials andthe tax police. 72. Improvingthe institutional climate for private investors, by radically curtailing the number of licenses and permits required, enhancing transparency in the implementationof regulations, ensuringa level playing field for all investors, curbing opportunities for corruption, and introducing modern concepts of tax administration, will be a prerequisite for an increase in genuine private sector investment. Some measures have already been taken to address these problems. Parliamentenacted and amended legislation in 2005 and 2006 to streamline licensing procedures and reduce the number of activities which are subject to licensing, but there is still much to be done to improvethe investmentclimate. 73. While higher rates of private investment in marketable goods and services is clearly essential for long term growth, whether the economy requires more public investment is less certain. The country inherited a large stock of public capital from the Soviet era, to the extent that the capacity to supply utilities is "over-dimensioned'' for the size of the economy (World Bank, 2006C), althoughsome of the public capital assets are not especiallywell suitedto a marketeconomy (for example inthe transport sector, investmentin the Soviet era favoured the railways over the road network) and some assets are now in a poor state of repair because adequate maintenance has not been carried out since the country became independent. 74. Two sectors are particular pertinent for any discussion of the country's public capital requirements: electricity and transport infrastructure. As discussed below, Tajikistan has a comparative advantage in the production of hydroelectricity, but this does not mean that the supply o f electricity for domestic consumption needs to be expanded. Tajikistan already consumes a much higher level of electricity per capita than the average for a developingcountry." Moreover, domestic electricity demand i s likely to stagnate when electricity tariffs are raised to cost recovery levels as planned by the Government." Therefore, it is unlikely that an expansion of electricity supply to the domestic market is critical to support future growth of the economy. Expendituresare requiredto rehabilitateand maintainthe transmissionand distribution system, but when electricity tariffs are raised to cost recovery levels it should be possible for Barki Tajik, the electricity utility, to fund these expenditures without recourse to the Governmentbudget. 75. Investment in hydro electricity generation for export to neighboringcountries i s potentially commercially viable and offers major investment opportunities, but these opportunitieswill be best exploitedby private investorswith their own capital at risk, provided that Governmentcan provide credible guarantees against expropriation and other non-commercialrisks.Public investment should not be required to develop hydroelectricity generation for export, and indeed is not desirable because this will 10 Tajikistan's annual per capita consumption of electricity is 2,236 KWh, compared to the average for a middle income country of 1,422 KWh, even though its GDP per capita is only around one fifth of the average for a middle income country. Excluding the electricity consumption(about 40 percent%of the total), Tajikistan still consumes around 1,300 KWh per annum (World Bank, 2006C: 36). " Thecurrentaveragelevelofelectricity tariffs isaround0.6 centsperKWh. TheGovernmentintends to raise tariffs to the level required to cover the long run average incrementalcosts of supply, which i s 2.3 cents per KWh in current prices. 23 pre-empt budgetary resourceswhich are requiredfor essential public services such as education. 76. It is also unlikely that major investments in the transport infrastructure are necessary to support long term economic growth, because there is little evidence to indicatethat the inadequacies of the transport infrastructureare a major constraintto private investment and growth. In the 2005 BEEPS, only 2% of businesses in Tajikistan,citedtransportation as a major obstacle for the growth and operation of their business, while only 11% of businesses in Tajikistan cited transportation as a moderate obstacle (World Bank, 2006C: Annex 1). A World Bank regional study of Transport and Constraintsto Trade in Central Asia, including Tajikistan, concluded that the road networks in the region are relatively extensive and largely sufficient to meet the needs of users, although roads are in poor condition because of inadequate maintenance. It argued that CentralAsian countries should focus on the maintenance of existing roads, focusing on the core road networks that can be maintained on a sustainable basis within the resources available ineach country (World Bank, 2004B). 77. To summarize, Tajikistan economy's investment requirements to support sustainable long-term growth consist mainly of investment in the production of marketable goods and services, including electricity for export, rather than investments in infrastructureto provide public goods and services. Hence the private sector rather than the public sector is much better suited to take responsibility for financing, owningand managingthese investments. D. COMPARATIVEADVANTAGE AND POSSIBLEGROWTH TRAJECTORIES 78. Tajikistan's future development will be shaped to some extent by its comparative advantage, as well as by governmentpoliciesand regionaldevelopments. A country's comparativeadvantage on world markets is closely linkedto its resource endowments, or at least those resource endowments which are immobile between countries, in line with Heckscher-Ohlin trade theory. Wood and Mayer (2001) demonstrate empirically that relative resource endowments explain 40-60%of the composition of exports in a large sample of 111 countries, in which exports are classified as manufactures, processed primary products and unprocessed primary products and resourceendowments comprise land, unskilledlaborand skilled labor. 79. On the basis o f its resource endowments, where should Tajikistan's comparative advantage lie? The methodologyemployed by Wood and Mayer (2001) involves comparingtwo indicatorsof resource endowments: land area per worker and years of schooling per worker. Countries with a higher than average land area per worker and with an average low (high) years of schoolingper worker export primary (processed) commodities. Those countries with low land area per worker and with low (high) average years of schooling per worker export labor (capital) intensive manufactures. Alongside land area per worker we also consider arable land per worker, because only a small share of Tajikistan's total land area is arable. We also include energy endowments, because these are important for Tajikistan. Energy endowments are measured as tones of oil equivalent (TOE) per worker per annum obtainable from proved recoverable reserves of crude oil and natural gas and from hydropower potential. Hydropower potential is defined as the economically 24 exploitable hydropower capacity.'* Hydropower potential is expressed in Terawatt hours per annum and then converted into TOEs per annum.I3 Proved recoverable reserves of crude oil and natural gas are the volume occurring in known natural reservoirswhich can be recovered in the future under present and expected economic conditionswith existingavailabletechnology. To convert recoverable reserves into a flow of energy per year, it is assumed that these reserves will be recovered over a 20 year time horizon: hence the annual recoverable volume is the total recoverable reservesdividedby 20. Table 2.6 shows Tajikistan's resource endowments compared to those of seven major regional groups of countries and two individual large countries. 80. What do the data in Table 2.6 suggest about the nature of Tajikistan's comparativeadvantage?It is unlikely that Tajikistan has a comparative advantage in agricultural exports, because its arable land area per worker is no higher than the world average. The sectoral patterns of growth in Tajikistan since 1970also suggests that the country's comparative advantage does not lie in agriculture because, despite the fact that the Soviet Government had invested heavily in agriculture in the Tajik Republic, the growth rate of the sector has consistently lagged behindthat of the rest of the economy for the last three and a half decades. Nevertheless, there may be exceptions to this; for example, Tajikistan may have a comparative advantage in cotton production given that this crop has historically been economically viable. Tajikistan does not have the relative resource endowments of an exporter of labor intensive manufactures - low land area per worker and low skill levels and so is - unlikely to have a comparative advantage in these products. Tajikistan's landlocked geography also impedes its ability to export labor intensive manufactures, because transport costs are more important for manufactured exports as a result oftheir import content generallybeinghigherthan that of other types o f exports. This does not mean that no agriculture or manufacturingcan be viable in Tajikistan. There are parts of the country, such as the Ferghana Valley, which are very fertile and can support economically viable agriculture, such as cotton or food crops for the domestic market. It is also possible that some manufacturing activities, especially those involving the processing of locally produced raw materials, such as cotton, can be viable. But neither agriculture nor manufacturing, with the exception of the energy intensive aluminum smelting, are likely to be the leadingsectors of growth in the traded goods sectors. I 2This is the amount of gross theoretical capacity that can be exploited within the limits of current technology under present and expected local economic conditions. Gross theoretical capacity is the maximum power which could be generated if all natural water flows were turbined with 100 percent efficiency to sea level or the country's borders. The definition is from the WEC Survey of Energy Resources. ''One Megawatt hour is equal to 0.086 TOEs. 25 Table 2.6: Resource endowmentsof Tajikistan and other countries and groups of countries Land area Arable land Years of Energy sq kilometres sq kilometres schooling TOE per per 100workers per 100workers worker p.aI MENA 4.0 0.3 5.5 15.7 SSA 6.0 0.4 3.0 0.9 LAC 5.6 0.4 6.1 2.9 EA&P 1.1 0.1 6.4 0.3 SA 0.6 0.2 4.6 0.1 AdEcon 4.2 0.4 8.8 0.7 EE&FSU 8.7 0.9 9.5 2.1 China 1.o 0.1 6.4 0.4 USA 4.3 0.8 12.1 1.2 Tajikistan 4.8 0.3 10.1 9.1 World 3.2 0.3 6.7 1.4 MENA is Middle East and North Africa; SSA is Sub-Saharan Africa; LAC is Latin America and the Caribbean; EA & PC is East Asia and the Pacific, excluding China; SA is South Asia; Adcom is Advanced Economies,excluding USA and EE&FSU is EasternEurope and the Former Soviet Union. Sources: Years of schooling; Barro and Lee (2001) and Wood and Mayer (2001) Population; World DevelopmentIndicators Land area and arable land areadata are from the Nationmaster.comwebsite. Energy data are from the website ofthe WEC Survey of Energy Resources. 81. The resources which Tajikistan is relatively well endowed with are energy supplies and skilled labor, defined in terms of the average years of schooling of the workforce, although this indicator may overstate the skill levels of the workforce because of both a deterioration in the quality of education since the break up of the Soviet Union and migrationof workers. Tajikistan is very well endowed with energy resources, with 9 TOESof energy per worker per annum, compared to the world average of 1.4. Almost all of Tajikistan's energy resource endowment derives from its hydropower potential. Only one country in the rest of the world, Iceland, has a greater hydropower potential per worker than Tajikistan and there are only five countries in the world which have larger potential hydropower resources even in absolute terms, all of which are much larger countries than Tajiki~tan.'~Moreover, Tajikistan is geographically close to countrieswhich are both large and energy scarce, notably China and the countries of South Asia, which can provide potential export markets for electricity. New hydropower projects in Tajikistan could generate electricity at a cost of only 2.5-3 cents per KWh, and transmit power at around 0.5 14 These are the Democratic Republicof Congo, Brazil, Canada, the USA and the RussianFederation. 26 cents per KWhto export markets, comparedto the costs inpotentialexport marketsof around 5 cents per KWh (World Bank, 2004C). Hence Tajikistan has a strong comparative advantage in the energy sector, specifically the generation of electricity from hydropower. 82. The generation of hydroelectricity for export is, therefore, likely to be a major growth industry in Tajikistan over the longterm. However, this is not an unmitigated blessing. Countries endowed with natural resources, including those among the transition economies, have tended to experience slower growth than those countries which are less well endowed. Kronenberg(2004) argues that the prime reasons for this "resource curse" in the transition economies were corruption and the neglect of basic education.Inaddition, hydropowergenerationis very capital intensiveands will create very few jobs, other than in the construction phase of the projects. Hence Tajikistan cannot depend upon hydropower generationas the sole engine of growth if that growth is to contribute to employment creation and poverty reduction. Whether or not the exploitation of Tajikistan's hydropowerpotentialturns out to be a curse or a blessing for the country will depend on the nature of its economic and fiscal policies. It is possible to envisage two quite different, and admittedly extreme, long term growth trajectories in Tajikistan, with the differences between these two trajectories essentially attributable to differences in the Government's budgetary policies and its policies towards private investment. We set out each of these two possible trajectories below. Growth Trajectory I:Growth withjobs and human development 83. The first of these two possible growth trajectories would lead to a beneficial outcome in terms of human development and welfare. It would involve private sector investmentin hydropower generation, mainly for export, with government restricting its own role in the industry to providing a sound regulatory framework, reforming the domestic electricity market to ensure the long run commercial viability of domestic supply, and providing credible guarantees for private investors against non- commercialrisk such as expropriationor major changes in the tax regime. 84. Government would also implement fundamental reforms to improve the institutional climate for private investment outside the electricity sector. This is essential, because ifthe economy is to combine growth with employmentcreationand therefore poverty reduction, it must attract private investment into sectors other than electricity generationwhich are more labor intensive. Giventhat the growth o f export earnings is likely to be dominated by electricity exports, the domestic market for goods and services will provide the main focus for private investment outside o f the electricity industry. 85. Under this trajectory, Government would focus its budgetary resources on strengtheningthe provision of basic public services to enhance human capital and fight poverty, especially by improving the public provision of education, and providing other essential public services such as the maintenance of law and order, protection of property rights and regulation of markets where this essential to protect the public. Over the long term, this will facilitate an improvement in labor productivity and thereby complement increased private investment in boosting sustainable economic growth. Improved labor productivity will also enable real 27 wages to rise. Consequently this growth trajectory offers the prospects of three important development gains: (1) rapid and sustainable economic growth, driven by private investment in hydropower generation for export and in other more labor intensive industries mainly serving the domestic market, combined with increased labor productivity arising from better education and training and from factor productivity improvements brought about by more competitive markets; (2) income poverty reduction brought about by employment creation and rising real wages as a result of private investment in the non-electricity sectors and better education and training; and (3) reduction in the non-income aspects of poverty as a result of improvements in essentialpublic services such as educationand healthcare. Growth Trajectory 2: Growth withoutjobs and human development 86. Investmentin hydropower generation for export also plays a major role in the second growth trajectory but unlike in the first trajectory, the Government itself would undertake a large share'of the investment, using external loans to supplement its own resources. The second difference with the first trajectory is that the overall investment climate remains fundamentally unreformed with the consequence that private investment does not increase and, therefore,job creation is minimal. 87. With the capital investment requirements of the energy sector, and perhaps large capitalprojects in other sectors such as transport, taking up a major share of the Government budget, the fiscal space for funding basic public services will be squeezed, foreclosing the possibility of achieving major improvements in the provision of education and health services. In addition, the external borrowing required to fund the Government's capital investments may threaten external debt sustainability, which would further reduce the fiscal space for funding basic public services. Consequently, this trajectory offers the prospects of economic growth, althoughat lower rates than the first trajectory because of lower private investment in the non electricity sector and lower investment inhuman capital, but it will create few jobs and do little to reduce poverty, whether income poverty or the non income aspects of povertysuch as health. 88. The basic differencesinterms of public policy betweenthe two trajectoriesare twofold. First, the priorities for public expenditure in the first trajectory are basic social services, especially education, while in the second trajectory they are capital investments. Secondly, fundamental reformsare enactedto strengthen the investment climate inthe first trajectorybut not inthe second. E. MODELINGGROWTHEFFECTSOFPUBLICEXPENDITURE THE Public Expenditure and Economic Growth: What Does Endogenous Growth Theory Tell Us? 89. There is now a large literature dealing with endogenous growth theories and the empirical modellingand testingof these theories. The causal factors of economic growth in exogenous growth theory are the accumulation of physical and human capital, alongside various institutionalfactors (macroeconomic stability, trade policy, etc) which, inter alia, can affect the efficiencywith which capital is used. Exogenous growth models, therefore, allow for a possible role for fiscal policy (as well as other 28 economic policies) in the growthprocess. Fiscalpolicy could boost economic growth through the impact of public expenditures on physical or human capital, whereas it might depress growth if fiscal deficits crowd out private investment or lead to macroeconomic instability. Endogenous growth theory suggests that the type of public expenditures are likely to be important for growth; not all public expenditures may be beneficial and some may have negative affects, depending on the specific circumstances in individualcountries. 90. There have been many empirical studies analysing the impact of public expenditures on economic growth, but the findings are not unanimous. The research o f Aschauer (1989), Barro (1991) and Easterly and Rebelo (1993) support the argument that government expenditures on physical infrastructurehave a significant positive impact on growth through their affects on private-sectorproductivity. Other researchers have argued that the relationship between government expenditure and economic growth tends to be sector-specific. For example, Bose, Haque and Osborn (2003) examine the growth effects of government expenditure for a panel of 30 developing countries in the 1970s and 1980s, with a particular focus on sectoral expenditures, finding that the share of government capital expenditure in GDP is positively and significantly correlated with economic growth, but that current expenditure is insignificant. They also find that, at the sectoral level, government investment and total expenditures on education are the only outlays that are significantly associated with growth once the budget constraint and omitted variables are taken into consideration. 91. Because the circumstances of each country differ, the results obtained from a large sample of countries may not necessarily be applicable to every individual country. To support the analysis of the links between fiscal policy and economic growth in Tajikistan, an endogenous growth model, similar to that of Bose, Haque and Osborn (2003), which included different types of government expenditures, including capital and current expenditures, and education and health sector expenditures, alongside other explanatory variables such as indicators of the level of human capital, the initial level of GDP, and the fiscal surplus and private investment, was estimated with data from a sample of eight transition economies in Central Asia and the Caucasus, including Tajikistan. This study was carried out with DFID financing as a background paperto this chapter and is attachedas an annex. 92. The methodologywas applied in two steps. The first step was to usethe model to study the growth effects of government expenditure in the eight sample countries for the period 1996 - 2004 and identify the patterns of government expenditurethat are associated with economic growth. The second step was to use Seemingly Unrelated Regression Equations (SURE) techniques to isolate the coefficients for Tajikistan specifically and hence identify the types of public expenditure which are most likely to support sustainedeconomic growth inTajikistan. 93. Although there are a number of drawbacks with this type of econometric analysis - the sample size is small, the growth relationshipsacross all of the sample countries may not be identical, the growth effects of expenditures may have long gestation periods and may not be readily captured in annual data, reverse causation, from growth to public expenditure, cannot be ruled out, etc - the results are in line with those o f many other empirical studies and do provide useful insights into the 29 type o f relationships between fiscal policy and economic growth which exist in Tajikistan. 94. The evidence from this study supports the prevalent view in modern growth theory that education is an important factor in economic growth, with the level of government expenditures on education having a significant and robust correlation with growth. The study also found that the share of government capital expenditure in GDP is positively and significantly correlated with economic growth. This result provides the basis for government investments in sectors such as transport and communication, where it may be difficult to attract the optimal levels of private investment because of the public good characteristics of these investments. However, the impact of public investment seems to be conditional on the non-government investment variable (which is a proxy for private investment) in Tajikistan. Non- government investment is associated with economic growth in a significant and positive manner, especially in the SURE estimates for Tajikistan. There is also strong evidence to support human capital variables in the eight transition countries and in Tajikistan. In particular,the initial human capitalvariable, expressedas enrolment in primary and secondary schools is consistentlysignificant in the regression estimates, as is life expectancy. Finally, the Government budget surplus has a significant and positive impact on economic growth (or alternatively, a larger fiscal deficit depresses growth); moreover the magnitude of the impact is quite large: on average, a one percentage point increase in the budget surplus is associated with an increase in real per capita growthrate of about 0.6 percentage points. F. CONCLUSIONSAND IMPLICATIONS FORPUBLICEXPENDITURE POLICIES 95. Tajikistansuffereda dramatic contractionof its economy in the first half of the 1990s as a result of the disruption to trade and production patterns caused by the break up of the Soviet Union, the loss of fiscal transfers from the Soviet Union and the civil war. The economy began to recover in the second half of the 1990s with strong rates of growth, averaging9% per annum, being achieved in the 2000s. The recovery was largely the result of a rebound in aggregate demand, facilitated by the gradual stabilizationof the economy and inflows of remittances and donor aid, which enabled under-utilized production capacity to be brought back into production. Although by 2005, realoutput was still approximately40% belowthe levelat the start of the 1990s, before the break-up of the Soviet Union, the limits to further rapid growth based solely on increasing the utilization of existing capacity will soon be reached. The economy's aggregate supply capacity has probably fallen since independence, because gross investment rates have not been sufficient to offset the depreciation of existing capital and because many skilled workers have left the country, either permanently or as migrant workers. 96. Sustainable future growth will depend on expanding the supply side o f the economy, through higher rates of capital investment and improvements in factor productivity, rather than on simply boosting aggregate demand. The critical requirements for long term future growth are a major rise in private investment in marketable goods and services, investment in human capital to raise labor productivity and in increase in competitionand the removal of market distortions to enhance incentivesfor strengtheningtotal factor productivity. To attract more private investment it will be necessary to improve the poor institutional environment for 30 private sector business, which is characterized by excessive, complex and untransparent regulationsand corruption. 97. An expansion of public investment is not so pressing a need for the economy, because it is already well endowed with public capital assets as a result of high investment rates during the Soviet era. Some of the existing infrastructure systems, such as electricity supply, are over dimensioned for a low income economy. The responses to business surveys do not suggest that infrastructuralinadequacies are the major constraint to private sector businesses. Moreover, the budget lacks adequate funds to maintain properly even the existing public capital assets, and hence it is likely that allocating more funds to maintenance will generate higher social rates of returnthan investment innew publiccapitalassets. 98. Tajikistan has enormous hydropower generation potential and borders large energy scarce countries which can provide future export markets for electricity generated in Tajikistan. The capital costs of new hydropower developments are enormous (the combined costs o f two projects currently under consideration are in excess of 100% of GDP) and hence it is essential to attract private capital into these investments. Private investment in hydropower generation will also help to ensure that they are managed efficiently and profitably, but to attract private investment it will be necessary for government to reform the domestic electricity market so as to make electricity supply commercially viable in the long run, and to provide a secure regulatory and tax structure for the electricity sector which can provide private investors with guaranteesagainst noncommercialrisk. 99. Tajikistancannot rely solely on hydropowergeneration as the engine of growth if it is to benefit from equitable growth with job creation and poverty reduction, because hydropowergeneration is very capital intensiveand hence private investment in more labor intensive industrieswill be necessary to createjobs. The exploitationof natural resources has not always benefited developing and transition economies. Transition economies endowed with natural resources have experienced slower rates of growth than those transition economies that are less well endowed. The relatively poor performance of natural resource abundant transmission economies has been attributedto the neglect of education and corruption. 100. To examine the causal links between different types of public expenditure and economic growth, we used an endogenous growth model on a sample of eight transition economies includingTajikistan. The regressors includeda vector of budget variables alongside other variables which are hypothesised to affect growth in endogenous growth theory, such as human capital. Because there were insufficient observationsto estimate the modelon Tajikistan alone, we usedthe Zellner procedure of seemingly unrelated regression equations (SURE) to isolate the coefficients for each country. 101. The evidence from the regressions supports the prevalent view in modem growth theory that education is an important driver of economic growth. We also found that the share o f government capital expenditure in GDP is positively and significantly correlated with economic growth. Non-governmentinvestment is also associated with economic growth in a significant and positive manner, especially in the Zellner estimates for Tajikistan. Also, we find strong evidence for the role of 31 human capital variables in the growth process; in particular, the initial human capital variable, expressed as enrolment in primary and secondary schools was consistently significant, as was life expectancy. 102. What are the Government expenditure implications of the analysis in this chapter for sustainable economic growth in Tajikistan? Government expenditures on investment and human capital development, especially education, would boost economic growth in the long run. However, if Tajikistan is to follow a development strategy, which combines economic growth with human development, it should prioritize expenditures on education, health and other forms of human capital development, rather than investment, for two reasons. First, because a shortage of public capital is not a major constraint to economic growth, given that the public infrastructure is already relatively well developed, and second, because the scale of investments required to develop major infrastructure projects, such as hydropower generation for export, are far greater than can be accommodated within the Government budget unless the Government were to embark on heavy external borrowingwhichwouldjeopardizefuture externaldebt sustainability. 103. Given that budgetary resources are inevitably constrained by the imperativeof maintainingdebt sustainability, an increase in government capital expenditures will squeeze the fiscal space available for current spending on educationand other forms of humancapital development and on the essential maintenance needs of the existing infrastructure. A growth strategy focussed on heavy public investment in energy and other large infrastructure projects would be detrimental to the balanced long run development of the country, especially if it were combined with a failure to reform the institutional climate for private investment, because hydropower generation will create very few jobs and because of the adverse budgetary consequences for educationspending. 104. Instead it would be preferableto shift the emphasis of budgetary allocations towards human capital development, especially education and health, and to the essential maintenance requirements of the core assets of the existing public infrastructure,while leaving investments in large commercially viable infrastructure projects such as hydropower generation to the private sector. Improving the education and health systems and maintainingthe existing infrastructureassets are clearly public goods which will not be undertaken by the private sector, and hence government must take the lead in providing these goods. In contrast, the private sector can, in principle, provide the capital needed to develop major infrastructure projects provided that the risks involved can be controlled by putting in place the appropriate regulatory framework. Tajikistandoes not spend enough on public health and education services. Aggregate budgetary spendingon healthand education is low, at 1.3%and 4% of GDP respectivelyin 2006. Healthexpenditures commandjust 7% of the State budget, less than halfofthe internationaltarget of 15%." 105. A development strategy in which Government focuses on expanding and improving the provision of the key public services needed for human capital IsComparedto other CIS countries, Tajikistan spends about the same proportion of GDP on education but only just over half as much on health(Lorie, 2003: 26), although it is arguable that most of the CIS countries need to spend more on these sectors to reverse the decline in human capital which has occurredsince the break up o f the Soviet Union. 32 development, offers the prospect of sustainable growth with employment creation. Rising labor productivity will raise real wages, and improvements in public services will enhance the non-incomeaspects of human welfare. 33 3. FISCAL SPACEIN TAJIKISTAN'~ A. INTRODUCTION 106. The design of fiscal policy in Tajikistan poses major challenges for policymakers. The real value of budget resources available to fund government expenditure has contracted sharply since independence, because of the economic collapse and the ending of transfers from the Soviet Union which subsidized service provision during the Soviet era, and as result. the quality of public services has deteriorated. An expansion in the budgetary resources allocated to public services such as education, healthcare and the maintenance of public infrastructure, combined with improvements in the efficiency of government expenditures, is essential if Tajikistan is to achieve its developmental objectives and especially its goals for poverty reduction. However, if an expansionof the budget resource envelope is to be consistent with the promotion of sustainable economic growth, it is crucial that any increases in the tax burdendo not distort incentives for saving, investment and trade, and that Government borrowing does not jeopardize macroeconomic stability, crowd private sector borrowers out of credit markets, or threaten government solvency. Moreover, even if an expansion of the Government budget allows expenditures on priority sectors to be increased, improvements in economic governance, such as public financial management, will also be needed to ensure that this generates higher growth and poverty reduction(Varoudakis, 2006). 107. This chapter examines the fiscal space which is potentially available to Tajikistan over the next 10 years. The concept of fiscal space refersto the additional budgetary resources which can be mobilized for priority public expenditures through increases in the budget resource envelope and improvements in the allocative and/or technical efficiency of government expenditures. We employ the framework of the fiscal diamond to analyze fiscal space; the four points of the fiscal diamond represent the four potential sources of additional resources for priority public expenditures: domestic revenues, grants from donors, sustainable borrowing and improvements in the efficiency of public expenditure^.'^ The quasi fiscal deficits in the energy sector are also included in the analysis, because they will drastically reduce fiscal space available for priority expenditures unless they are eliminated through energy sector reforms, and these energy sector reforms will also affect the budget. 108. The chapter is organized as follows. Section 3.2 reviews fiscal policy in Tajikistan since the 1990s, highlightingthe progress made in recoveringfrom the dire fiscal position of the mid 1990s but also the remainingchallenges. Sections 3.3 to 3.7 utilize the framework of the fiscal diamond to analyze the scope for expanding the l6This chapter draws heavily on a Tajikistan case study done for the DevelopmentCommittee Paper on "Fiscal Policy for Growth and Poverty Reduction" presentedto the World Bank board in April 2007. 17The concept of the fiscal diamond as a tool for analyzing how the scope for fiscal space differs across countries was recommendedin a paper preparedby the Bank for the Development Committeeof the Bank and the Fund; "Fiscal Policy for Growth and Development: An InterimReport", April 2006. 34 budget resource envelope in Tajikistan without jeopardizing fiscal sustainability or macroeconomic stability. This involvesanalyzingthe potentialto mobilize additional budget resources from domestic revenues (section 3.4), grant aid (section 3.5) and borrowing(section3.6). The measuresneededto tackle the quasi fiscal deficits inthe energy sector, and the fiscal impact of these measures, are discussed in section 3.7. Section 3.8 summarizes the impact on the aggregate budget resource envelope of the projections for domestic revenue, grants, borrowing and the quasi fiscal deficit. Section 3.9 examines the scope to create fiscal space by improvingthe efficiency of government expenditures. Section3.10 providesa conclusion. 109. The main conclusion o f this chapter is that Tajikistan could expand its fiscal space substantially in 10years time, potentiallyby around 8 percent of GDP annually, but only if the Government implements major reforms to tax administration, public expendituremanagement and electricity pricing. Reformsto tax administrationcould potentially add almost 3.2% of GDP to the fiscal resource envelope. Even more important are electricity sector reforms needed to cut the sector's quasi fiscal deficit that amounted to 9.3% of GDP in 2006. If these institutional and electricity sector reforms are not implemented, not only will there be very little growth in the budget resource envelope and no gains from improved expenditure efficiency, but the Government will eventually have to allocate substantial fiscal resources to the electricity sector to fund essential rehabilitationand maintenance of the transmission and distribution system, which will drastically cut the fiscal space available for priority expenditures. Moreover, even if additional budget resources could be mobilized without reforms to public expenditure management, especially to improve the allocation and technical efficiency of expenditures, it is doubtful whether additional government expenditure would have a positive impact on economic growth, 110. Hence the main policy message of this chapter is the urgency of reforms to improve governance in public financial management, intax administrationand in the electricity sector. These reforms are a prerequisite for both mobilizing fiscal space for priority expenditures and for ensuring that any expansion of the budget made possible by fiscal space translates into higher growthand poverty reduction. B. FISCALPOLICY SINCEINDEPENDENCE 111. Tajikistan gained independence in 1991, but the country suffered a civil war between 1992 and 1997 and an economic collapse. The country also suffered a massive fiscal shock in the early 1990s, with a dramatic shrinking o f budget revenue which may have been by as much as 80% in real terms. The fiscal shock arose from two sources. First, duringthe Soviet era, Tajikistan had been a recipientof transfers from the Soviet Union which accounted for almost half of the Government budget, but these transfers ceased at independence (IMF, 2006B34). Secondly, the tax base contracted with the economic collapse. The fiscal shocks led to huge fiscal deficits which worsened the macroeconomic instability. 112. Although the fiscal data for the first half o f the 1990s, when the country still usedthe Russian ruble, are not comparable with data for subsequent years. It is clear that very large fiscal deficits were incurred in this period, in the region of 20% of GDP on average. The large fiscal deficits incurred in the first half of the 1990swere 35 almost entirely financed by credits from the central bank, as government's domestic borrowing requirement could not be met by the very shallow domestic financial system. Centralbank financingof the fiscal deficit, combined with the collapse of the economy, fuelled hyperinflation, which reached 2,100% in 1995. 113. In addition, quasi fiscal deficits led to a sharp build up of government and government guaranteed external debt, caused mainly by the Government financing imports o f commoditiesfrom Russia and other former Soviet Union(FSU) countries. Tajikistan, which at independence had no external public debt, had by 1995 accumulated over US$SOO million of external debt (equivalent to 134% of GDP), of which the major share was bilateral debt contracted from FSU countries. By 1994, Tajikistan was already experiencing serious difficulties in servicing its debt and reschedulingagreementswere signedwith three FSUcountries inthe mid 1990s. 114. The Government began implementing an IMF ESAF program in 1998, a key element of which was the reductionin the fiscal deficit and the eliminationof central bank financing of the deficit. The overall fiscal deficit was reduced from 11.2%of GDP in 1995, to an average of just under 3% of GDP in the final four years of the 1990s, and domestic financing of the fiscal deficit to an average of 1% of GDP in this period (see Chart 2.1).`* Consequently the broad money growth rate was brought down in the second halfof the 1990sand inflationwas cut to 30% by 1999. However, Government and government guaranteed external debt rose by a further US$lOO million inthe second halfof the 1990s, tojust under US$900 million in 1999. '*These deficits are on a cash basis. There were substantial expenditure and revenue arrears, but the lack o f a proper treasury system impededthe accurate reportingof these arrears. 36 Chart 2.1 Fiscal Aggregates as Percentof GDP: 1995-2005 30% 25% 20% 15% 10% 5% 0% -5% -10% -15% - 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 [-Expenditure `Domestic Revenue - - Domestic Financing -Fiscal Deficit] Source:M F 115. The cut in the fiscal deficit in the second half of the 1990s was achieved by cutting total government expenditures sharply in real terms: between 1995 and 1999 the real value of expenditurescontractedby 40%. Cuts in breadsubsidies, which had amounted to nearly 10% of GDP in 1995, accounted for a large part of the cut in expenditures." Expenditures on salaries and wages also fell sharply, from 7.4% to 3.6% of GDP between 1995 and 1999, because of both cuts inthe real value of wages and in the size of the payroll. Revenues actually declined by over 10% in real terms in this period, mainly because the Government rationalized and reduced the heavy rates of taxationwhich hadbeen imposedon the cotton and aluminum export sectors, and also becausethese sectors suffereddeclines inoutput. 116. Real GDP growth, which had begun to recover in 1997, accelerated in the 2000s, averaging 9% per annum during 2000-2005. Strong economic growth and better tax administrationboosted government revenues, which rose by more than 6% of GDP from 1999 to 2005 (see Section 3.4). This allowed government expenditures to increasewhile maintainingthe overall fiscal deficit at between2% and 4% of GDP, with the exceptionof 2001 when it temporarilyjumped to 5.6% of GDP, and avoiding virtually any resort to domestic financing of the deficit. Without the need to finance the fiscal deficit, the Central Bank was able to exert tighter control over the money supply and inflation; inflation was brought down to below 15% in 2001 and to below 10%by 2004, although it subsequentlyjumped to over 12% in 2006. 19Breadsubsidies were replacedby cash transfers, but substantialarrears were accumulated on these transfer payments. 37 Chart 2.2 External Debt, Percent of GDP: 1993-2005 140-c 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Source: IMF 117. The recovery in government expenditure was very pronounced. Primary government expenditures rose from 13% of GDP in 1998 (the lowest level since independence) to 21.8% of GDP in 2005. Given that GDP also recovered strongly over this period, this implies an increase in the real value of government expenditure (using the GDP deflator as a proxy for the price index for government services) of approximately 180% between 1998 and 2005. Basic public services have been among the major beneficiaries of the recovery of the Government budget. For example, expenditures on education in the State Budget (Le., excluding the PIP) rose from 2% of GDP in 2000 to 4% of GDP in2005, while expenditures on health in the State Budgetrose from 0.7% of GDP to 1.5% of GDP in the same period. 118. The external debt burden has been reduced since 2000, when the nominal dollar value of external debt (including non government guaranteed debt) peaked at US$1,227 million (128% of GDP). External debt management was also tightened after 2000: new borrowing was restrictedto concessionalloans and a ceiling of 4% of GDP (approximately US$30 million in 2001) was imposed on new external borrowing to fund the Public Investment Program (PIP). Public and publicly guaranteed (PPG) external debt continuedto rise, albeit slowly, until 2004 when debt cancellation, debt for asset swaps and debt restructuring agreements were reached with Russiaand Pakistan. In 2006, Tajikistan was granted debt relief on its IMF debt under the Multilateral Debt Relief Initiative (MDRI) worth US$97.5 million in nominal terms. By 2005, the nominal value of PPG external debt had fallen below 40% of GDP. However, the external debt indicators began to deteriorate again in 2006, because GovernmentcontractedUS$600millionof loans from Chinato finance infrastructureprojects (see Section3.6 for more details). 38 119. Despite the progress made in strengthening fiscal policy, serious problems remain. The Government has inherited from the Soviet era a structure of public institutionsand physicalassets which is large too large to be operated and maintained properly given the availablebudgetary resources. Consequentlythe inputs into public services cannot be funded adequately and are spread too thinly over many different activities. For example, salaries of government employees are far too low to provide even a subsistence wage (the average monthly wage was around 60 Somonis, equivalent to US$20 in 2004). There are insufficient funds for critical non-wage operational inputs, such as medicines in the health service: government expenditure on health amounts to the equivalent of only US$5 per capita. The physical infrastructurecannot be maintainedproperly. Moreover,while the Government'sown finances are relatively sound, there are serious problems elsewhere in the public sector, in particular in the electricity sector, which incurred a quasi fiscal deficit (QFD) of 9.3% of GDP in 2006. 120. To summarize, important progress has been made in several key aspects of fiscal policy since the late 1990s. First, there has been a major fiscal consolidation which has been a critical factor in stabilizing the macro economy, with the virtual elimination o f domestic financing o f the fiscal deficit. Secondly, there has been a recovery in domestic revenues which in turn has boosted the budget resource envelope. Thirdly, the expansion in the budget resource envelope has enabled the funding of social services to be increased substantially. Fourthly, the external public debt burden has been reduced from unsustainable levels through bilateral debt relief and the MDRI, although the risk to future debt sustainability is high because of the recent contractingof large new loansfrom China. c. FISCALSPACE FOR PRIORITY EXPENDITURES 121. The following sections of this Chapter examine the scope for creating fiscal space for additional priority public expenditures over the next 10 years from the four points of the fiscal diamond; domestic revenues, grants, sustainable borrowing and improvements in the efficiency of expenditure. We also analyze the budgetary implications of reforms to the electricity sector which are needed to eliminate the QFD of the sector and the potential budgetary impact of hydropower investments. The base from which our long term fiscal projections are made is the preliminary fiscal outturn for 2006. The budgetary resource envelope for primary expenditures amounted to 21.3% of GDP in 2006. This was composed o f domestic revenues of 18.8% of GDP, grants of 1.2% of GDP (this excludes the MDRI grant of US$97.5 million which has also been excluded from net domestic financing), and financing of 1.8% of GDP. The latter consisted of net external financing of 1.6% of GDP, net domestic borrowingo f -0.3%o f GDP and privatizationreceipts of0.6% of GDP. D. DOMESTICREVENUES 122. There are three potential sources of growth in domestic revenues as a share of GDP; first, the income elasticity of some taxes, secondly, changes in tax policy (tax rates, thresholds, exemptions etc), and thirdly, improvements in tax administration. This section explores the scope for raisingdomestic revenues from each of these three sources over the next 10years. 39 123. Table 3.1 compares the tax revenue collected in Tajikistan with that of other CIS countries.20 Although Tajikistan had the second lowest revenue to GDP ratio among the CIS countries, this is at least partly explained by Tajikistan having the lowest GDP per capita income in the CIS. Lorie (2003), investigating tax performance in CIS countries, regressed the tax revenue to GDP ratio on PPP GDP per capita for a large sample of developed, transitional and emerging market economies, with an additional regression which included a dummy variable for the OECD, graduating transition and CIS countries. The predicted tax revenue to GDP ratio for Tajikistan was 13.6% and 18.1% in the regressions with, and without, a dummy respectively. Tajikistan's tax revenue of 16.6% of GDP in 2006 was not, therefore, out of line with what would be expected based on the country's per capita income. Table 3.1. Tax Revenue/GDPRatiosin CIS Countries,2004 GDP per capita US Tax Revenueas percent dollars of GDP Armenia 1060 17.2 Azerbaiian 940 16.9 Unweighted average 1270 24.1 Unweighted average of 701 20.2 low income CIS countries 124. Domestic revenues amounted to 18.8% of GDP in 2006, of which 2.2% of GDP was non tax revenue (see Table 3.2).21 Domestic revenue relies heavily on income and profit taxes, which include a payroll-based social tax, which contributed 25% of revenue, and VAT, which contributed almost 40% of revenue in 2006. Customs duties contribute a relatively small share of revenue (6%) mainly because large shares.of imports are accorded duty free entry under bilateral trade agreements with CIS countries. 'O Data for 2004 for Turkmenistan are missing. Lorie (2003: 37) reports data on tax revenue for Turkmenistanof23 percent of GDP in 1998. "Thisincludesextrabudgetaryfundsof1.3percentofGDPreportedinIMFfiscaldata. 40 Table 3.2. Domestic Revenues as % of GDP 2000 2001 2002 2003 2004 2005 2006 Domestic revenues 13.9 14.6 16.6 16.2 16.8 19.3 18.8 Tax revenues 12.9 14.0 14.8 15.0 15.2 16.6 16.6 personal income tax 1.2 1.3 1.2 1.1 1.1 1.1 1.2 profit tax from corporations 0.6 0.6 0.6 0.5 0.6 0.5 0.6 minimum tax on enterprise profits 0.0 0.0 0.0 0.0 0.0 0.3 0.4 social taxes 1.6 1.6 1.8 1.7 1.8 2.1 2.0 tax on real estate 0.0 0.0 0.0 0.0 0.2 0.1 0.2 land tax 0.4 0.4 0.3 0.3 0.2 0.1 0.1 single tax from agric producers 0.0 0.0 0.0 0.0 0.0 0.8 0.5 sales tax on exports 3.3 2.5 1.9 1.9 1.4 1.5 0.6 cotton fibre 2.1 1.7 1.2 1.3 1.1 0.8 0.6 aluminum 1.2 0.8 0.6 0.6 0.3 0.7 0.0 tax on retail sales 0.0 0.0 0.0 0.0 0.3 0.4 0.4 VAT 2.5 3.6 4.8 5.2 5.7 6.3 7.4 domestic VAT 1.5 1.5 1.3 1.3 1.3 1.4 1.9 external VAT 1.o 2.1 3.4 4.0 4.4 4.9 5.5 road user tax 0.4 0.5 0.6 0.7 0.6 0.5 0.7 simplified tax 0.0 0.0 0.0 0.0 0.1 0.1 0.1 excise tax 0.5 0.6 0.9 1.o 0.9 0.8 0.7 customs duties 1.4 1.8 1.7 1.5 1.6 1.4 0.9 processed product tax 0.4 other taxes 0.9 1.1 1.1 1.2 0.6 0.6 0.6 Non tax revenues 1.o 0.6 1.8 1.2 1.7 2.7 2.2 Sources: Ministryo f Finance, M F E. BUOYANCY INCOME ELASTICITYOFREVENUES AND 125. Domestic revenues have been quite buoyant since 2000, rising by 4.9% of GDP in six years (Table 3.2). Because there have been many changes in tax policy since 2000, with some taxes abolished, new ones introduced and the coverage and/or rates of other taxes changed, it is difficult to disentangle precisely all of the reasons for the buoyancy of tax revenues between 2000 and 2006. However, VAT on imports has played the major role, adding 4.5% of GDP to tax collections. The growth in VAT was largely attributable to a change in the methodology, effected in the early 2000s, for calculating the VAT from the origin to the destination principle for trade with CIS countries (IMF, 2006: 10); Le., the VAT on goods which are traded between Tajikistan and other CIS countries is now levied by the importer (destination) instead of the exporter (origin) which was previouslythe case. As Tajikistan has a largetrade deficit with other CIS countries, the change in methodology for applying the VAT served to greatly expand the VAT base.22 In 2006 VAT was further boosted by very buoyant import growth of 38%, which was at least partly driven by the strong growth of remittances. 22Levying VAT on the basis of the origin of imports was common to all CIS countries for their intra- CIS trade, but most have now switched to the destination basis. 41 126. Growth in social taxes and excise taxes also made modest contributions to the buoyancy of tax revenues during the 2000-06 period. In 2005, tax collections were boosted by a number of tax policy changes made that year, such as the increase in the aluminum tax rate (from 1% to 3% of exports, which was then replaced by the processed product tax in 2006) and the introduction of a minimum tax on enterprises and a single agriculturaltax. 127. While it is difficult to assess the income elasticity of the major tax handles based on their historical performance because of the changes in tax policy, their observed buoyancy does provide some guide as to whether or not these tax handles are income elastic. In fact, as noted.above, only VAT on imports has displayedmuch evidence of buoyancy during the 2000s. It does not appear likely that income or profit taxes will be income elastic inthe future. In the case of profit taxes the formal enterprisesector of the economy largely comprises enterprises set up inthe Soviet era which are not very dynamic, and hence are unlikely to expand rapidly as a share of GDP, while the bulk of formal sector employment is in the Government sector, and hence the scope for increasingthe tax base is also highly constrained, given that any increase in the Government wage bill faces budget constraints. Because of the structure of the economy and employment, it is not surprisingthat neither individual incometax nor profit taxes have displayed muchbuoyancy over the last five years. 128. Will VAT on imports continue to increase as a share of GDP? This seems unlikely. Imports are already high as a share o f GDP (48% in 2006), especially for a landlockedcountry, and so will probably not rise further. The main factor that led to the buoyancy displayed by VAT on imports during the 2000s, the change in the methodology for computing the VAT, is a one off effect which will not be repeated. Consequently it seems prudent to conclude that natural income elasticity of the tax system over the mediumterm will not be a significant source of additional revenue. E". CHANGES IN TAXPOLICY 129. What scope is there for raising more revenue through tax policy changes without creatingdistortionsto resource allocationwhich would impede growth, which effectively means that tax rates should be moderate and not out of line with international levels. This subsection discusses the scope for generating additional revenues by raising the tax rates on the tax handles which are major contributors to tax revenues. Corporate Profits Tax 130. Corporateprofits are currentlytaxed at 25% (the rate was reducedfrom 30% in 2005). The current rate is in line with the unweightedaverage corporate profit tax for CIS countries and the transition countries of Eastern Europe (Lorrie, 2003: 38) which implies that if Tajikistan wants to be competitive in creating a business friendly tax system, it should not reverse the 2005 tax rate reduction. Moreover, the tax base for corporateincometax is small, given the rate of 25%, taxable corporate income is only about 2.5% of GDP. Raising the corporate profit tax rate would not generate much additionaltax revenue. 42 PersonalIncomeTax 131. Personal income tax (PIT) rates are low. There are three rates; O%, 8% AND 13%. The current top marginalrate of 13%is much lowerthan the equivalentinother CIS countries, which ranges from 20% to 35% (Summers and Baer, 2003: 7). Less than 5 percent of labor income was collectedin personal income taxes in 2004, which is attributableto boththe schedule of low rates and the fact that the incomes of a large share of workers fall below the thresholdat which tax begins to be paid. Nevertheless it should be possible to increase the average PIT collection to about 10% of labor income by raising the top marginal rate to 25% and levying a rate of 10-15% on average incomes, which would generate additional revenue of about 1% o f GDP. Raising PIT rates could also make the personal income tax more income elastic, because the schedule of taxes will become more progressive. The top marginal PIT should be set at the same level as that of the corporate income tax rate, in line with international best practice, to remove the opportunity for the owners of small businessesto reduce their tax liabilities by shifting income from profits to their own salary. There is also a case on equity grounds for raisingthe top marginalrate of PIT. Social Security Taxes 132. Social security taxes make a substantial contribution to revenue but the applicable rates are already high - enterprises pay 25% of their monthly wage bill - and this is, in effect, a tax on the hiring of labor. Hence raising the tax rate would discourage employment. Moreover, these taxes are earmarked for pensionand social security benefits so there is little point in raising the contributions unless this is neededto meet a gap in the funding of benefits. VAT 133. The VAT rate is already 20% (and is effectively 22% because of the road user tax which is also leviedon the VAT base at a rate of 2%). This is at the high end of VAT rates worldwide, so an increase in the rate is unadvisable. Customs Duties 134. Customs duties rates are relatively low: there are seven rates with a range of between 0% and 15%. However, although imports are high as a share of GDP, the customs tax base is much narrower because around three quarters of all imports originate in the CIS countries, with many of which, including Russia, Tajikistan has signed free trade agreements. Although the average customs duty rate is about 7%, the effective tariff rate during 2001-2005, calculatedas customs duties collectedas a percentage of the value of total imports, was only 2.5%, implying that about 65% of importsenter Tajikistantariff free. Hence any increase incustoms tariffs would apply to a minority of imports which would limit its potential for revenue generation, as well as distortingresourceallocation. 43 Excise Duties 135. There could be scope for raising excise duty rates, but this is also probably limited. Almost two thirds of excise duties comprise taxes on petroleum products. Petrol is already taxed at a high rate (about US$0.5 per liter) which limits the scope for further increases, although diesel is taxed much more lightly. There may also be scope for raisingtaxes on alcohol. Summary of Tax Policy Measures 136. Raising the PIT rate (with the top marginal rate aligned with the corporate income tax rate) offers the most promising avenue for mobilizing more revenue through tax policy changes without adversely affecting incentives for economic growth. T his would also improve the progressivity of the tax system and make it more income elastic. Apart from higher excise duty rates on alcohol and possibly fuel, there is limited scope for raising more revenue from indirect taxes, either because tax rates are already high (e.g., VAT) or because the tax base is narrow (customs duties). G. TAXADMINISTRATION 137. Improvementsin the tax administration, which is implementedby the Ministry of State Revenuesand Duties (MSRD) through its tax and customs departments, offer the greatest scope for enhancing domestic revenue. Reforms to tax administration have been underway for several years, and some progress has been made. The tax department has been restructuredalong functional lines and a Large Taxpayer Unit (LTU) set up. A new tax and customs codes were introduced in 2005. Nevertheless, major weaknesses intax administrationhave still to be rectified. 138. Although the tax code is sound in most respects and in line with international standards (Summers and Baer, 2003: 8), it has not yet been fully implemented. Working practices do not accord with the norms of a modern tax administration, notably in the lack of self assessment by taxpayers. The tax department is badly organized, with resources spread over far too many offices (74) geographically, of which the majority deals with relatively few taxpayers whose contributionto total tax is marginal. The setting of monthlytargets for eachtax district also impedeseffective tax collection at the national level. Although all taxpayers have a Taxpayer IdentificationNumber (TIN), the law does not require that the TIN be quoted on all invoices, which is an impedimentto preventingfraud inthe payment of indirect taxes. The registrationof taxpayers is chaotic. There is virtually no computerizationof tax administration.The quality of tax audits is very poor because there are too few tax officers with the requisiteskills. Tax administrationsuffers from a shortage of funds for operational activities and office buildings are poorly maintained and lack basic equipment. Tax collection is also seriously hampered by corruption, which is partly the resultofthe low salaries paidto tax officers. 139. The collections of some indirect taxes appears to be too low given the size of the tax base, which would indicatethat substantial scope exists for raisingtax revenue by improving compliance through better tax administration or by tightening up on exemptions. For example, the amount of VAT collected in 2006 is consistent with 44 VAT having been levied on only about a third of private consumptionexpenditures, which implies that either compliance is poor or that a large share of consumption itemsare exempt from VAT. 140. The priority for tax administration reforms is to introduce modern tax administration practices backed up by comp~terization.~~Tax officials should stop attemptingto verify manually every tax return, and instead adopt the principle under which taxpayers are expected to comply voluntarily, through self assessment, with their returns beingsubject to audits on a selective basis.The auditingsystem needs to be strengthened, with better trained auditors undertakingaudits selected on the basis of greatest risk. The law should be revised to enforce the quoting of a TIN on all invoices.Muchwork needsto be done to enable the tax code to be fully implemented, such as drafting and ratifying supportingregulations. Efficiency intax administration can be enhanced by the introduction of IT systems for taxpayer registration and accountingand core administrationfunctions. 141. Over the longer term, the structure of the tax department should be re- organized to reduce the number of small tax offices. in the regions which play a marginal role in tax c~llection.~~Large taxpayers should be handled by the LTU, which needs strengthening. Model tax inspectorates should be established in each of the three regions to handle medium sized taxpayers. The closure of small offices, computerization, plus a change in working practices, will then allow the number of officials employed by the tax department to be reduced by at least 50% (Le., from the current level of around 1,600 to 750). This would enable salaries for the remaining staff to be raised substantially which would provide better incentives to attract and retain higher caliber staff and reduce incentives for corruption. In addition the tax police, which is a separatetax enforcement agency within the MSRDcomprising20% of its staff, should be abolished and the resources which are thus saved used to strengthen tax administration. The tax police are not compatible with modern tax administrationpractices which are basedon self assessment: insteadthe existence of a separate tax agency which has close contact with taxpayers encourages corruption (Summers and Baer, 2003: 24). 142. The Customs Department can also be strengthened through the upgrading of facilities and infrastructure at border posts, developing application systems and procedures to support modern customs practices and training customs officers (Harrisonet al, 2005). 143. How much additional revenue could tax administration reforms generate? Although it is difficult to be precise, based on the experience of other developing countries which implemented major reforms to tax administration, an increase in tax revenue of around 20% should be attainableover the longterm; i.e, tax revenue could be raisedfrom the current levelof around 16%to 19-20%o fGDP. 23 These paragraphs are based on the recommendationsfor reform of an ADB funded tax administration reform project and the recommendationsin Harrison et al (2005). 24 The geographical fragmentation of the tax system reflects the system of financing local government budgets, whereby a proportion of taxes collected locally are retained by local authorities (and usually supplemented by transfers from the center). The rationalization of tax offices, therefore, needs to be linked to changes in the financing of local government budgets. 45 Long-term Revenue Projections 144. It does not appear likely that the income elasticity of the tax system will make a significantcontributionto revenue growth in the future, for the reasons discussed in Section 3.4.1 above. The scope for raisingtax rates is also limited because the rates of several of the most important tax handles, such as corporate income taxes and VAT, are already at levels which are in line with comparable rates in similar economies, while the scope for raising more revenue from customs duties is constrained by Tajikistan's free trade agreements with its maintrading partners. The exceptions are personal income taxes and excise duties, where raisingtax rates could mobilize at least another 1% of GDP in tax revenue. Improvingtax administration offers the best prospects for enhancing tax revenue. The tax system is characterized by serious weaknesses which, in principle, can be rectified. Introducingthe practices of modern tax administration could generate additional tax revenue of over 3% of GDP. Although it will not be easy to reformtax administration, it should be possible to implementmajor reformsover the 10year time horizonof this paper. 145. Table 3.3 provides 10year projectionsfor domestic revenues as a share o f GDP. The projectionsincorporatethe following assumptions. All existingrevenue handlesgrowth in linewith nominalGDP except for the following. Basedon the assumptionthat marginalPIT ratesare raised, with the top PIT rate set at 25%, the tax to 8% by 2010, and by a further 0.25 percentage points of labor income peryear. . collectedfrom PIT as a share of labor income rises from the current level of over 4% Formalsector labor income is assumed to be a constant 25% of GDP. The VAT projectionsincorporatethe gains from VAT leviedon the higher cost of electricity, which is explained in section 3.7. However,we also makean offsettingadjustment to take into account the likelihoodthat higherelectricitytariffs will depress demand for spendingon consumer goods other thanelectricity,and hence depress the VAT collectedfrom sales ofthese goods. The assumption is that total privateconsumption expenditure will be inelastic with respectto increases in electricitytariffs; Le., the marginalpropensityto consume is not affected. Therefore, consumptionof goods and services, other than electricity, falls by the same absolute amount as the increase in utility bills for privateconsumers, which we proxy by takingthe total increase in electricity bills minusthat of government and TALCO. Consequently, almost three quarters of the gain inVAT from higherelectricity bills is offset by lower VAT paymentson goods and services other than electricity. The processedproduct tax, which is leviedon TALCO, is reduced by the amount of subsidy requiredto offset the impactof higherelectricitytariffs on TALCO's input costs, so that by 2009 TALCO no longer pays this tax, as explainedin section 3.7. 46 146. To project the cotton export sales tax, we make use of the modelingof cotton sector development scenarios made in World Bank (2005B: 56) which examines the potential impact of reforms to raise efficiency in the sector, including reforms to establish secure land rights, resolve the problem of farm debt, abandon the setting of productiontargets, introducecompetitionin input supply, establish alternativesources of financing for cotton farms and liberalize trade in raw cotton and cotton fiber. Efficiency enhancing reforms could raise cotton yields per hectare by a third, to 2.5 tons per hectare, which was the lower bound of yields obtained during the Soviet period, and couldraise the efficiencyof ginningby 19%. Total cottonincomes would therefore rise byjust over 50%, to almost US$330 million a year. Assumingthat the export tax rate remains at lo%, export tax revenue could be raised to US$33 million, compared to the tax collection of US$16 million in 2006 (when production was adversely affectedby drought). Our projectionsassume that this levelof export taxes will be reached at the end of the 10 year forecast period, in 2016, after a series of equal increases ineachofthe precedingyears. 147. The gains from tax administration reform are projected at a cumulative 2 percent o f total tax revenue per year for 10 years, but only beginningin 2008 because time will be required to strengthen the institutional capacity of tax administration; hence by 2016 tax administrationreform boosts tax revenues by 18%. The revenue projections also include the projected revenue earned from government's capital investment in the hydropower projects, and the surplus in the form of taxes or royalties which can be earned from electricity exports, which is explained in section 3.7. 148. Domestic revenues increase by 5.6% of GDP, from 18.8% of GDP in 2006 to 24.4%in2016, as a result of the following factors. The increase in individualincome tax, together with the elasticity of that tax handle, generates gains of 1.2% o f GDP. VAT increases by 0.3% of GDP as a result of the net impactof electricitytariff rises. However,the processedproducttax revenues amountingto 0.4% of GDP in2006 are foregone because o f the impact of higher electricity tariffs on the cost structure of TALCO. Tax administration reforms yield additional revenue of 3.2% of GDP by 2016, contributingalmost 60% of the growth in revenues over the next 10 years. A further 1.3% of GDP in revenue is provided by income from capital investments in HPPs and royaltiesand/or taxes from exports ofelectricity. Grants 149. Donor grants to the Government budget have beenlow, althoughprojectgrants may have been under-reported. Excluding the MDRI debt relief received in 2006 (which is recorded as a grant in fiscal data) grants to the budget amountedto 0.7% of GDP in 2005 and 1.2% of GDP on 2006, but grants should jump to around 2% of GDP in 2007, boosted by the receipt of US$9.2 million from the Educationfor All- Fast Track Initiative. The majority of donor grants are disbursed to agencies outside the Government budget, such as community based organization^.^^ Future levels of 25An estimated US$l09 million of grant aid from donor organizations was disbursed outside the budget in2005 comparedto US$17million of grants disbursedto the budget inthe form of both budget support and project aid. The estimate for the non budget grants is a minimum because data were not available from all donors. 48 grant aid to the budget, in the form of budget support and/or project aid, could be boosted if donor governments increase their overall aid to Tajikistan, in line with commitments made at international fora to expand aid in real terms to the poorest countries, and/or if a larger share of donor aid were to be channeled to the Government budget. The latter will depend on the perceived quality of public financial management, especially with regard to aid in the form o f budget support. Concerns about fiduciary weaknesses have deterred donors from providingmore aid in the form of budget support (the summary of the PEFA assessment in Chapter 5 gives detailsof fiduciaryweaknesses). 150. It is very difficult to make any quantitative projections of the volume of budget grants over the longterm, becausethe range of possible outcomes is large. At the low end of the range, Tajikistan might receive no increase in grants, because donors remain unconvinced that fiduciary standards are improving sufficiently and/or because Tajikistan contracts more loan finance from non traditional lenders, thereby reducingthe perceived need for more grant aid. At the other end of the range, grants could increase several fold, possibly to around 5% of GDP, as donors both increase grant aid to Tajikistanas part of a more general expansion of development assistance to low income countries and channel a much larger share of their grant aid to the Government budget. The latter will clearly require improvements in public financial management to strengthen and make more transparent the budget process and to improve the allocation of expenditures, especially in those sectors which can contribute to meetingthe MDGs. The assumption we have used here is that donor grants will increase by 10% a year in nominal dollar terms, using the 2007 budget projection as a base, which is consistent with the commitment made by the G8 countriesin 2005 to increase their aid to poor countries by 60% by 2010. We assume that the 10%growth ingrants will be sustainedthroughto 2016. 151, Projecteddonor grants for the budget as a percent of GDP are shown at the bottom of Table 3.5. They rise from 1.2%in 2006 to 2.0% of GDP in 2007, and then slowly increase to 2.2% of GDP by 2016. Consequently, the projected increase in grants makes only a very modest contribution to fiscal space over the long term. However, these projections are unavoidably highly speculative and depend to a large extent on decisionswhich are outside the controlofthe Tajikistangovernment. H. GOVERNMENTBORROWING Recent Trends in Borrowing 152. Net borrowing to fund the Government budget has been relatively modest in the last six years. During 2001-2006, net external financing averaged 2.8% of GDP and net domestic borrowing averaged negative 1% of GDP. The former was constrained by the need to restore external debt sustainability, which led to government imposing a constraint on external borrowing to fund the PIP of 4% of GDP per annum. Domestic borrowing was constrained by government's policy priority to stabilizethe macro economy which requiredthat Government should avoid financing its deficit by borrowing from, and instead accumulate savings with, the Central Bank, to allow for a deceleration of money supply growth and the accumulation of foreign exchange reserves by the Central Bank. The very shallow nature of the domestic financial system offers very little scope for Government to 49 borrow either from the commercial banks or from non bank financial institutions. Consequently, the macroeconomic imperatives dictated that domestic financing should be negative. Privatizationreceipts added an average of 0.5% to the budget's financing resources during 2001-2006. Table 3.4 below providesdetails of financing for the Government budgetduring2001-2006. Table 3.4: Government Financing 2001-2006 % of GDP Other external financing includes arrears and debt rescheduling Net borrowingfrom the NBT excludes the receipt of the MDRIdebt relief in 2006 (equivalentto 3.5 percent of GDP). Other domestic borrowingincludes nonbank borrowingandthe discrepancy Source: Ministry of Finance,LMF 153. The following subsections evaluate the magnitude of government borrowing which couldbe incurredina sustainable manner over the next 10years. External Borrowing 154. Government's net external financing will increase dramatically over the medium term as a result o f its contractingof a US$604 million loan for infrastructure projects from Chinaand its intentionto contract a loanof US$400 million from China for a HPP. Net external financing is projectedto rise to around 14%of GDP in 2007 and 2008 before falling to 11%of GDP in 2009. The magnitude of externalfinancing over the next three years will lift the NPV of Tajikistan's PPG external debt over at least two o f the five thresholds at which debt is deemed sustainable under the IMF/IDA Low Income Country Debt Sustainability Assessment Framework (see Tables 3.5 and 3.6). 50 Table 3.5: IMF/IDA LIC DSA External Debt SustainabilityThresholds for a Medium CPIA Performer Indicator Threshold (%) NPV of externaldebt to GDP 40 NPV of externaldebt to exports 150 NPV of external debt to government revenue 250 I Externaldebt service to exPorts I 20 I 51 23U 23U E 3 U 1 3 U 1? 3 U -- 3 U 2 3 U D 3 3 U 0 3 3 U c 3 3 U s 3 U m 155. Our projections for external financing are derived using the following methodology. For the next five years (2007-201l), we adopt the projections made in the most recent DSA, which was completed in January 2007 (IMF & World Bank, 2007). External loan disbursements and amortization in this period mainly reflect the implementation of loan agreements already signed, such as the US$604 million Chinese infrastructure loan. We then make our own projections for 2012-16. After 2011, there is more scope for loan disbursements to vary, as these will involve loan contracts not yet signed. As the purpose of this chapter is to assess the scope for fiscal space which is consistent with sustainable borrowing, our projections are based on the assumption that the Government will aim to bring back, over the long term, its external debt indicatorsto levelswhich do not exceed the thresholdsat which external debt is deemed unsustainable, by restricting new external borrowing. Hence, given the projected loan repayments and the grant element of the loans, we construct a profile of loan disbursements which ensures that all sustainability indicators are met by 2016. 156. The MF/IDA Low Income Country Debt Sustainability Framework identifies five thresholds for external debt sustainability, which are conditional on the CPIA scores of the debtor. Higher CPIA scores imply stronger economic and fiscal management and this allows the sustainable level of external debt to be higher. Tajikistan was judged a medium CPIA performer in 2006 (after being previously scored as a low performer). We assume that Tajikistan will be able to maintain its status as a medium CPIA performer over the long term, and as such, its external debt sustainability will be evaluated on the basis of a medium CPIA performer, the external debt thresholdsfor which are set out in Table 3.5 157. The projected net external financing over the mediumterm will lift the NPV of debt to both GDP and exports above the sustainability thresholds, although the indicators will start to decline after 2009. The NPV of debt to exports is the most problematic, as this exceeds the threshold by the largest margin. Under our projections, this indicator falls back to the threshold of 150%by 2016. The NPV of debt to GDP returns to the threshold earlier, by 2013. We assume that all external PPG debt is used to finance the Government budget,26that the average grant element of the outstanding debt stock in 2007 (about 23%) will remain the same throughout the 10 year period of the projections and that average interest payment on external debt will rise to 1.75%. Repayments after 2011 are projectedon the assumptionthat the average loan maturity is 25 years with an average grace period of five years, and that, on average, one third of the portfolio in each year is still under its grace period. 158. Table 3.6 presents the profile of external loan disbursements and repayments, together with the implied trends inthe debt indicators, over the next 10 years. As can be seen from the table, to restore external debt sustainability, it is necessary for external loan disbursements to fall sharply after 2009, by which time the bulk of the Chinese loans will have been disbursed. Disbursements must average not more than about 4% o f GDP during 2009-13 and then about 5% of GDP in the following three years. 26This may not necessarily be valid, for two reasons. m First, ifTajikis tan contracts new credits from the IMF under the PRGF, the financing will not be used for the budget. Secondly, some of the external loans may be used by public enterprises, or by the Central Bank. 53 159. Consequently, once the projected increase in external loan disbursements during 2007-09 has been absorbed, there is very little scope left for any further increase in external borrowing without breachingthe debt sustainability ceilings. The sustainable level of gross public external borrowing is only 4-5% of GDP, and even this level of borrowing would be unsustainable if the country were to suffer a major shock to GDP growth or export growth. This is one of the key policy messages of this paper: the Government cannot solve its budgetary resource constraints through higher external borrowing on a sustainable basis, even if external lenders are willing to extend it more credit in the medium term, because that will jeopardize long term debt sustainability. Domestic Borrowing 160. The main constraint on government domestic borrowing is the need to maintain macroeconomic stability without crowding out private sector borrowing from the banking system. Government's domestic debt stock is small, at around 4% of GDP, and this debt carries a fixed interest rate of about 5%; hence the sustainability of the domestic debt stock is not the binding constraint on domestic borrowing. Given the low level of domestic debt, Government could issue more domestic debt and remain solvent, but the issuance o f more debt faces three constraints: the very narrow domestic monetary system (broad money was only 8.5% of GDP at the end of 2006), the high level o f polarization of the banking system (foreign currency deposits were 69% of total bank deposits at the end of 2006), and the lack of significant non-bank financial institutions.The poorly developed domestic financial sector also constrains the use of alternative macroeconomic instruments to create fiscal space. IfGovernment were to fund a more expansionary fiscal policy by borrowing from the Central Bank, the latter would then face the need to sterilize the reserve money thus created, but its ability to do so is constrained by the same characteristics of the domestic financial system which restrict the scope for governmentborrowingfrom the domestic market. 161. To derive projections of domestic borrowing consistent with macroeconomic stability, we assume that only borrowing from the Central Bank and commercial banks is possible. This is because the non bank financial sector is too small to hold any significant amount of government debt. Sustainable borrowing from the Central Bank is determined with reference to its balance sheet; whereby the increase in reserve money (RM) is identical to the sum of the changes in its net foreign assets (NFA), net credit to government (NCG), net credit to the private sector (NCPS) and other items net (OIN). ARM ANFA + ANCG + ANCPS + AOIN 162. We assume that the velocity of circulation of reserve money will fall at 2.5% per year; hence reserve money will grow at the annual rate of nominal GDP growth plus 2.5%, NFA is the sum o f foreign exchange reserves and foreign exchange liabilities. We assume that the former increase at the same rate as imports (which grow at the same rate as GDP), thereby maintaining the foreign reserve to import cover constant, while foreign exchange liabilities, valued in current dollars, remain constant. We assume no net lending by the Central Bank to the private sector and a zero change in its other items net. With these assumptions, the CentralBank's lending 54 to the Government can be derived as the residual. After 2007 (for which we take the projection from the Government budget) this is negative up to 2016 and averages 0.4% per annum (Table 3.6). 163. The constraints on government borrowingfrom the commercial banks are the growth of their liabilities and the extent to which government wants to avoid competingwith the private sector and public enterprises for the very limited pool of commercial bank credit. Commercialbank liabilities are mainly deposits, but more than two thirds of deposits are foreign currency denominateddeposits and we assume that commercial banks will not use foreign currency deposits to fund purchases of government debt because of the currency mismatch that this would entail. Consequently, the resources available for funding government borrowing are restrictedto domestic currency deposits. We assumethat Government will restrict its net borrowing from commercial banks to 30% of the increase of domestic currency deposits, thereby leavingthe remaining70% to fund lendingto the private sector and public enterprises, as well as holdings of cash reserves and other assets. We assume that the velocity of circulationof Somoni deposits will fall by 5% per annum. As can be seen from Table 3.6, government borrowingfrom the commercialbanks is positive but very small, averaging only 0.1% of GDP per year during 2007-2016. Net domestic borrowing,the sum of net borrowingfrom the CentralBank and commercial banks, is slightly negative over the next 10years, averaging 0.2% of GDP per annum. This is slightly larger than the average so far during the 2000s (Le,, the savings are slightly less) butthe amount of fiscal space created is very small. PrivatizationReceipts 164. The Government budget received modest inflows from privatization receipts, averaging 0.5% GDP during 2000-2006 (see Table 3.4). This was the result of a program of privatizing mainly small scale public enterprises, which is now almost complete. AI though there are still several large-scale enterprises remaining in the public sector, such as TALCO, the aluminum smelter, we have not made projections of privatizationreceipts for the next 10 years because it is not yet clear what further plansthe Governmenthas for privatizingthese public enterprises. InterestPayments 165. To make projections for external interest payments, we use the DSA projections for the period 2007-2011 and in subsequent years we assume that the average interest rate on the outstandingstock of debt is 1.75%. For domestic debt, the interest rate on the 300 million Somonis of existing government debt, which is non- marketable, is 5%. We assume that new debt will be issued on the money market, througha competitiveauction, and that it will, therefore, pay a market interestrate. It is likely that new marketable debt will be short term, given that the money market is not liquid. Hence we assume that the interest rate paid on this debt will be equal to that paidon short term time deposits. To projectthis interest rate, we take the average deposit rate for time deposits with a maturityof 1 monthto 3 months over the last five years, which was 11.5%. 166. The projected external and domestic interest payments, as a percent of GDP, are shown in Table 3.6. As can be seen, they increaseby 0.4%of GDP duringthe 10 55 with all of the increase attributable to higher interest payments on external debt. . year period of the projection, from 0.5% of GDP in 2006 to 0.9%of GDP in 2016, Interest costs remain moderate because the bulk of the Government's debt stock is externaldebt contracted at concessionalrates of interest. I.FISCAL IMPLICATIONS OF REFORMS INTHE ELECTRICITYSECTOR 167. Government'sfiscal positionover the longterm will be affected profoundly by developments in the electricity industry. The electricity industry, which is almost entirely in public ownership, incurs substantial QFDs in servingthe domestic market, mainly because of selling electricity at below its long run marginal cost. However, the future prospects for the industry and for the Government budget are brighter if Government fully implements agreed reforms. It should be possible to reduce the QFDs through electricity tariff reformand investments in system rehabilitation,while major new investments in hydropower generation will create the potential for substantial export earnings and government revenue, althoughthere will also be costs for Government. This section examines the long-termfiscal implicationsof reforms and investments in the electricity sector. It begins with an explanationof the QFDs, briefly outlines the main planned and potential investments in the sector, and then makes 10 year projections of the fiscal impact of electricity sector investments and reforms. Quasi FiscalDeficits 168. The electricity sector is the main source of QFDs in Tajikistan. World Bank (2005E) estimatedthe QFD in the sector as the difference betweenthe actual revenue collected in cash by Barki Tajik, the state owned electricity utility, and the revenue which would be collected if it set electricity tariffs at levels sufficient to cover the longrun average incrementalcost (LRAIC)27and reduced its technicaland collection losses to levels which are in line with internationalnorms. The LRAIC is computed as the sum of the costs required for capital investment in rehabilitation of the electricity system plus the incrementaloperational and maintenance expenditures on that system, divided by incrementalsales of electricity, all evaluated over a 20 year time horizon. 169. The electricity sector QFD for 2003 was estimated in World Bank (2005E) as US$273 million. We have used this methodologyto update these estimates to 2006.28 The QFD arises from three sources. First, Barki Tajik incurs technical and commercial losses which in 2006 amount to 17% of electricity supplied to the domestic market, as a result of loss of electricity in the distribution system, theft of electricity, defective metering and inappropriate use of norm-based billing for customers without meters. The norm for technical and commercial losses is set at 10% o f electricity generated, hence Barki Tajik's excess technical and commercial losses, which contribute to the QFD, amount to 7% of electricity supplied in 2006. Evaluated at the LRAIC, the excess technical and commercial losses amount to US$25 million in 2006. The second source of QFDs is the difference between the average tariff charged by Barki Tajik and the LRAIC. The LRAIC is estimated at 2.3 21The LRAIC is used as a proxy for long run marginal cost. 28This methodology is also employed in Ebinger (2006) where the QFDs are referred to as hidden costs. 56 cents per KWh in 2006 prices (2.1 centsKWh in 2003 prices), compared to the average tariff charged by Barki Tajik of 0.6 cents per KWh in 2006. The losses arising from chargingaverage tariffs which are not much more than a quarter of the LRAIC are projected to reach US$210million in 2006. Thirdly, Barki Tajik incurs losses because only 67 percent of the electricity bills it issues to its customers are actually paid in cash, compared to the internationalnorm of 98 percent. These losses amount to a further US$27 million. The total QFD, therefore, is estimated at US$262 million, or 9.3% of GDP, in 2006. 170. What are the fiscal implications of the QFDs in the electricity sector? In most years, the QFDs are not subsidized directly through the Government budget; an exceptionwas 2001 when Barki Tajik defaultedon tax payments because of liquidity constraints. However, some form of government subsidy will eventually become inevitable unless the QFDs are eliminated, because Barki Tajik will not be able to fund the investment and associated operational and maintenance costs which are requiredto rehabilitatethe electricity system and keep it functioning at an adequate standard. Consequently either the costs will have to be borne from the Government budget, thereby reducing fiscal space by an equivalent amount, or the electricity system will deteriorate, reducingthe supply and reliability of electricity to customers, which will inevitably have a negative impact on economic growth. To eliminate the QFD in the electricity sector, it will be necessary for Barki Tajik to raise the average tariff to the LRAIC and to reduce its technical and commercial losses and its billing losses, through better meteringand enforcement of payment of bills, for example. 171. The gas utility, Tajikgas, also incurred QFDs as a result of technical and commercial losses (especially incorrect billing), but these were much lower in magnitude than those incurred in the electricity industry; in the region of US$13 million, or 0.8% of GDP in 2003. Tariff reforms implemented since 2003 have largely eliminated the QFD in the gas industry. It is not clear whether there are significantQFDs outside of the energy sector because no comprehensive investigation of this issue has been undertaken,. There are potential QFDs in the cotton sector, where cotton farmers have debts which have not been fully verified but which could amount to as much as US$300 million. These debts are unpayable, in that they greatly exceedthe capacity of cotton farmers to service them out of their income from cotton, and Government may face pressure to provide some form of bail-out for the farmers. FutureHydropower Investments 172. Major investments in the electricity sector will have an impact on the Government budget over the longterm, through their capital costs ifundertaken with public funds and through the fiscal revenues which could be generated by taxing exports of electricity. Table 3.7 lists major investments which are ongoing or proposed. These include rehabilitationof the existing generation, transmission and distribution system, for which investments amount to US$199 million and investments in overhead transmission lines (OTLs) of US$244 million. Most of the investments in rehabilitation and OTLs (US$223 million) will be made by Government, largely through on lending foreign loans to Barki Tajikq2' If these 29One of the Chinese loans referred to in section 3.6 will fund the construction of an OTL. 57 investments are completed over a five year period (2006-20lo), Government's contribution will amount to an average of 3% of GDP per year. Table 3.7: Capital Costs of Major Ongoing,Planned and PotentialElectricity Sector Investments IProjectsand Nature of IMillionscapital ' 1 of US dolllars Total GOT 1 Source of GOT Investments cost contribution contribution Rehabilitationof generation, 199 199 Mainly loans transmissionand distribution from ADB, Kwf, systems IDA Overhead TransmissionLines 244 224 Chinese loans Sangtuda 1HPP 600 100 From existing assets Sangtuda 2 HPP 220 Rogun 2900 800 From existing assets Other HPP 54 Total 4317 1331 173. At least two large investments in HPPs are currently being planned or under consideration. The most advanced in terms of project preparationare the Sangtuda I and I1 projects, with a total cost of US$820 million, of which Government's contribution is US$lOO million in the form of existing assets.30 An even larger project is the Rogun HPP, which will cost about US$3 billion, and for which Governmentwill contribute existing assets which it claims are worth US$800million. The HPPs will expand the generating capacity of Tajikistan and thereby provide power surpluses for export to countries such as Pakistan, Afghanistan, Iran and Russia. The Sangtuda HPP could be commissionedin 2009 and attain full production of TWh 2.67 per annum in2013.3' Roguncould be commissionedin 2014 and reach full capacity of TWh 14.2 per annum in 2020. Together with the loss reductions attained through rehabilitation of the transmission and distribution network, these investmentscould add more than TWh 17 per annum by 2020 to the current capacity of TWh 12 million. By 2016, the time horizon for this paper, power sector investmentscould add almostTWh 8 per annum to the existingcapacity. 174. The Government should leave the bulk of the investment in new hydropower projects to the' private sector, except for the assets which it already owns (the construction which took place during the Soviet period). The combined investment costs ofthe Sangtudaand Rogunprojectsamounts to around 100%of GDP, which far outstrips government's borrowing capacity (World Bank, 2005B: 46). Moreover, the generation and export of electricity is not a public good which needs to be provided 30 These are assets built during the Soviet era, when the projects were started but not completed. 3 'A TeraWatt hour (TWh) is one billion Kilowatt hours (KWh). 58 by the Government; rather it is a commercial enterprise which can be undertaken by the private sector which is more likely to provide efficient management than Government. 175. If the HPP investments listed in Table 3.7 are implemented, the Government budget will benefit through two channels. First, government will acquire a share in the capital structure of the projects, in the form of equity and debt because it will contribute to the projectsthe value of assets constructedduring the Soviet era. Hence Government will earn a return on capital in the form of interest and dividends. Secondly, the export of electricity could generate a surplus over and above the financial costs of generation and transmission, part or all of which could accrue to Government in the form of taxes or royalty payments. Projectionsof the FiscalImpact of Electricity Sector Investmentsand Reforms 176. The reforms to the electricity sector include electricity tariff reforms, to raise average domestic tariffs to the level of the LRAIC by 2010, which together with the investmentsoutlined above, will affect the fiscal position through seven channels: (i) reduction of the QFD; (ii)increase in the cost of government's own electricity consumption; (iii) increase in VAT paid on electricity consumption; (iv) higher payments through the Energy Compensation Mechanism; (v) a return on Government's capital investment in the Sangtuda and Rogun HPPs; and (vi) a share of the surplus earned by exports generated by the Sangtuda and Rogun HPPs. Table 3.8 provides 10 year projectionsof the fiscal impact of the electricity sector reforms. The projections for the dollar values of the variables are all expressed in 2006 prices, unlessotherwise stated. QFDs 177. The projected QFDs take account of the scheduled reduction in technical and commercial losses by Barki Tajik from 17%of gross supply to the domestic market in 2006 to 10%by 2016, and a rise in cash collections from 67% of payments in 2006 to 90% in 2016. The main source of QFD reduction is the proposed increase in the average tariff from the 2006 level of 0.6 centsKWh to the LRAIC level of 2.27 centsKWh, in constant 2006 prices in 2010.32Together these measures cut the QFD from US$262 million (9.3% of GDP) in 2006 to US$30 million (0.7% of GDP) in 2016. Government Utility Payments 178. Government is projected to pay US$2.8 million for its own electricity consumption in 2006. We assume that tariffs for government consumption will be increased in line with the proposalmade by the World Bank33,tariffs should be raised to 4.58 cents/KWh, in constant 2006 prices, by 2010. This will push up 32 The tariffs on which the projections inthis sub-section are made are those set out in the "World Bank Comments and Suggestions on the Proposal of the Government of Tajikistan to Adjust Electricity Tariffs to Reach Financial Viability Levels by 201O", May 2006. 33 These are set out in the "World Bank Comments and Suggestions on the Proposal of the Government of Tajikistan to Adjust Electricity Tariffs to Reach Financial Viability Levels by 2010", May 2006. 59 Government's electricity bills to US$29 million by 2016, comparedto US$3.5 million if tariffs were to remain at the 2006 levels. Hence an increase of US$26 million, or 0.6% o f GDP by 2016, can be attributed to the proposed tariff increases. VAT on Electricity 179. Electricity is subject to a 20% VAT. With the proposed tariff increases and the projected electricity demand, VAT paid on all electricity consumption is projected to rise to US$82 million in 2016, compared to a projected US$22 million if tariffs were to remain at 2006 levels.34 The gain in VAT payments as a result of the tariff increase is US$62 million, or 1.2% of GDP, in 2016. 34The projected US$20 million in VAT payments in 2016 which assumes no tariff increase does not take account of the higher domestic demand that would result from lower prices. This is because it is very unlikely that higher demand could actually be met given the levels of domestic supply, especially as the Sangtuda HPP, which will contribute to domestic supply from 2009 onwards, will not be economically viable iftariffs do not increase. 60 Table 3.8: FiscalImpact of ElectricityReformsand Investments 2006-2016; % of GDP 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 1, Quasi Fiscal Deficit -9.3 -7.4 -5.6 -3.3 -0.4 -0.Q -0.8 -0.7 -0.7 -0.7 -0.6 1.a from excess technical -0.B -0.7 -0.6 -0.4 -0.3 -0.2 -0.1 0.0 0.0 0.0 0.0 and commercial losses 1.b from pricing below L M I C -7.5 -5.5 -3.8 -1.8 0.7 0.0 0.0 0.0 0.0 0.0 0.0 1.cfrom excess non cash payments -0.Q -1.1 -1.2 -1.1 -0.8 -0.8 -0.7 -0.7 -0.7 -0.7 -0.6 2. Increase in govi electricity payments 0.1 0.2 0.3 0.6 0.6 0.6 0.6 0.6 0.5 0.5 2.1 Govt payments at 2006 tariffs 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 2.2 Govi payments at projected tariffs 0.1 0.2 0.3 0.4 0.7 0.7 0.7 0.7 0.6 0.6 0.6 3. Gain in VAT 0.3 0.8 1.0 1.4 1.4 1.3 1.3 1.3 1.2 1.2 3.1 VAT at 2006 tariffs 0.6 0.6 0.6 0.5 0.5 0.5 0.5 0.5 0.5 0.4 0.4 3.2 VAT at projected tariffs 0.6 0.9 1.2 1.5 1.9 1.Q 1.8 1.8 1.7 1.7 1.6 4. Required TADAZ subsidy 0.0 0.2, 0.4 0.6 0.6 0.5 0.5 0.5 0.5 0.4 5. ECM 0.1 0.1 0.3 0.5 0.5 0.5 0.5 0.4 0.4 0.4 5.1 ECM at 2006 tariffs 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 5.1 ECM at projected tariffs 0.1 0.2 0.3 0.4 0.6 0.6 0.6 0.6 0.5 0.5 0.5 6. Return on HPP investments 0.1 0.1 0.2 0.2 0.2 0.5 0.7 6.1 Sangtuda 0.1 0.1 0.2 0.2 0.2 0.2 0.2 6.2 Rogun 0.0 0.0 0.0 0.0 0.1 0.3 0.5 7. Share of HPP surplus 0.1 0.2 0.2 0.4 0.6 7.1 Sangtuda 0.1 0.2 0.1 0.0 0.0 7.2 Rogun 0.0 0.0 0.1 0.4 0.6 8. Aggregate Fiscal Impact -9.3 -7.2 -5.4 -3.3 -0.6 -1.1 -0.8 -0.6 -0.5 0.1 0.5 Source: See Text Item"4. RequiredTADAZ subsidy" is factored in for illustrationpurposes.This may be true for low case scenario. TALCO 180. Raising the electricity tariff will have a negative impact on the finances of TALCO, which consumes around6.5 TWh of electricity peryear. We assume that the tariff paid by TALCO will rise in line with the proposal made by the World Bank, which is that it will increase to 1.36 centsKWh, in 2006 prices, by 2010. With the electricity tariff raised to this level, TALCO electricity costs will rise to US$lll million a year (inclusive of VAT) by 2016, an increase of US$42 million over its electricity costs ifthe tariff was to remainat the 2006 level. 181. It is not clear whether TALCO could absorb suchan increase in input costs and remains solvent, given that its current profits amount to only US$20 million per year; lower profitability of TALCO is mostly driven by tolling arrangement that prevents the companyto benefit highworld market pricesfor aluminum.The financial viability of TALCO is also constrainedby the very high costs of transporting its material input (alumina) and output (aluminum) overland: these costs are around 4.5 times higher, per ton of aluminum, than comparable costs for a competitive developed country aluminum smelter, which are only partly offset by lower labor costs (World Bank, 2005B: 48). A more detailed analysis of the long term commercial viability of TALCO is required to enable a more definitive judgment to be made about the financial capacity of TALCO to absorb higher electricity costs. Given that the purpose 61 of this paper is to make long term fiscal projections, rather than recommendations specific to TALCO, we think it is justified to adopt a conservative view, for the purposes of fiscal planning, of the capacity of TALCO to absorb higher electricity prices, with the proviso that we are not expressing the view that TALCO should receive subsidies from the Government budget; it would clearly be preferable if the efficiency of TALCO could be improved sufficiently to allow it to absorb higher electricity costs without recourse to government subsidies. 182. Consequently, we assume that TADAZ can only absorb electricity cost increases up to a maximum of US$20 million per year, and that all cost increases above this level will have to be absorbed by the budget, through, for example, government foregoing taxes owed by TADAZ.35Therefore the cost to the budget of subsidizing TADAZ's electricity consumption in 2016 will be US$22 million, or 0.4 percent of GDP. We assume that this subsidy will be affected mainly through lower tax payments by TADAZ. Energy CompensationMechanism 183. Under the ECM, Government makes payments through the budget to cover a fixed monthly volume of electricity consumption of eligible households whose aggregate householdincome falls below the average monthly salary for the applicable rayon. After implementing planned revisions to the ECM, it is expected that it will cover annual electricity consumption of KWh 1,800 per householdand that 312,000 households will benefit.36 We assume that the number of households covered by the ECM will rise in line with population growth of 2% per year. At 2006 tariffs for residential consumption of less than 250 KWh per month, this would imply that the budgetary costs of electricity subsidies under the ECM would be US$4.2 million in 2016, but the costs will rise to US$24.5 million if these tariffs rise from the current level of 0.5 cents/KWh to 2.93 centsKWh (in 2006 prices), in line with the proposal of the World Bank. Hence electricity tariff reform will raise the costs of the ECM by US$20million, or 0.4% of GDP, by 2016. 3 5The main source o f taxation of TADAZ was the export sales tax, but this was replaced in 2006 by the processedproducts tax. TADAZ paid US$lO million in processedproducttax in 2006, much less than the US$29million projected inthe Governmentbudget. 36 These data are from World Bank (2006B). 62 Return on GovernmentInvestmentin Sangtuda and Rogun 184. To project the return which will accrue to Government from its share of the capital inthe HPPs, we multiply Government's percentageshare of the capitalof each HPP by the estimated financialcost of the HPPexpressed incentsKWh of output and by the projectedannual output of the project. The financial costs of the projects and their projected outputs are taken from World Bank (2004C: 31, 57 and 58). The financialcosts of Sangtudaand Rogun Phase 1 (which is relevantfor the time horizon considered here) are estimated at 2.44 centsKWh and 2.93 KWh respectively. Government's share ofthe capital of Sangtuda 1 & 2 is assumedto be 12%, while that of Rogun is assumedto be 20%..37Revenuesdo not beginto accrue until 2010, when Sangtuda is projectedto begin generating electricity. By 2016, by which time the Sangtuda HPP will have reached its maximum output, it is projectedthat it will earn an annual return of US$8 million for Government, while Rogun, which will have reached only a third of its maximum output, will earn an annual return of US$27 million. The combinedreturns from the two HPPs will amount to 0.7% of GDP in 2016. urplusfrom ElectricityExports 185. Electricity could be exported at prices above the financial cost of generation and transmission, thereby earning a surplus, all or part of which could be captured by the Government in the form of taxes or royalties. T o project the potential budget revenue from electricity exports, we assume that exports will equal the amount to which domestic supply of electricityexceeds domestic demand, and that the supply of exports will be met first by Rogun, if available and sufficient, and then by Sangtuda. We assume that the export price will be 5 centsKWh. Transmission costs are estimated at 0.53 cents per KWh.38We assume that halfof the surplus earned by the Sangtuda and Rogun HPPs can be captured by Government; the distribution of the surplus will be determined through a process of strategic bargaining between government and its co-investors. The share of the surplus which accrues to Government rises from US$l million in 2012, the first year of exports, to US$29 million in2016, equivalentto 0.6% of GDP. However, the surplus is very sensitiveto the export price. If this were lower at 4 centsKWh, exports would still be viable because the price would still exceedthe financialcosts of generation andtransmission (although only just in the case of exports from Rogun), but the surplus would be drastically cut. Net FiscalImpact of ElectricityReforms 186. The net fiscal impact of the electricity sector is derived by summingtogether all the components discussed above. We are concerned primarily with the change in the net impact over a 10 year time horizonwhich amounts to a positive gain o f 9.8% 37The investment cost o f Sangtuda 1 and 2 is USIS820 million, of which government will contribute US$IOO million. The investment cost of the Rogun project is estimated at about US$3 billion. Government claims that the value .ofexisting assets, which will comprise government's contribution, is US$800 million, but as this may be disputed by its investment partners we have assumed that government's share will be valued at around US$600 million. This is the average of the transmission costs to Iran, Pakistanand Afghanistan given in World Bank (2004C: 36). 63 of GDP in 2016 compared to 2006. The QFD reduction provides by far the largest component; a gain of 8.7% of GDP. The tariff rises which yield most of the QFD reduction also boost VAT payments by 1.2% of GDP, but raise Government electricity bills by 0.5% of GDP and ECM subsidies by 0.4% of GDP, while forcing Government to subsidize TALCO, probably by foregoing taxes, by 0.4% of GDP. Government earns a return on its investments in the HPPs of 0.7% of GDP and a surplus on their electricity exports of 0.6% of GDP. while the net impact of electricity sector reforms is positive, the main benefit of the reforms is to eliminate most of the QFD, rather than to generate significant amounts of additional budgetary resources which could be used to fund an expansion of public services. The on- budget impact of the reforms and investments in the electricity sector is a net gain of 1.1% of GDP in 2016 versus 2006; tax and non-tax revenues increase by 2.5% of GDP while Government expenditureson utility bills, ECM and the TALCO subsidy rise by 1.3%of GDP. J. RESOURCE ENVELOPEPROJECTIONS 187. This section brings together the projections in the previous four sections to derive a budget resource envelope for the next 10 years. The budget resource envelope is shown in Table 3.9 above and comprises all of the budget resources which are available to fund government expenditures, under the State Budget and the PIP. It is the sum of domestic revenues, grants, net external financing (disbursements of loans minus amortization) and net domestic borrowing. We also show the budget resource envelopefor primary expenditures; i.e., excluding interest payments, and the projected additional requirementsfor government electricity bills and the ECM. The quasi fiscal deficit for the electricity sector is also includedinTable 3.9. 188. From a level ofjust under 22% of GDP in 2006, the budget resource envelope expands rapidly in the medium term, to over 35% of GDP in 2007 and 2008, due to the large disbursements of external loans for infrastructure projects. However, as discussed in section Habove, this level of borrowing is not sustainable and hence the budget resource envelope falls back sharply to around 26% of GDP during 2010- 2013, before climbing to around 30% of GDP in 2016. By 2016, the budget resource envelope will have risen by 8.6% of GDP comparedto the level in 2006. The rise in domestic revenues contributesjust under a third. Grants contribute an additional 1% of GDP to the budget resource envelope but the contribution of net domestic financing declines by 0.5% of GDP, as we have not made any projections for privatization receipts. 189. Interest payments increase by only 0.4 percentage points of GDP between2006 and 2016, hence the primary budget resource envelope expands by just over 8% of GDP in this period. However, budget resources amountingto a further 1% of GDP will be required to fund higher government electricity payments and the ECM, and there will also still be a small quasi fiscal deficit of 0.6% of GDP remaining in the electricity sector in 2016, which may have to be covered from budget resources. Therefore the increase in budget resources for priority expenditures, after taking account of higher interest payments, government electricity bills, the ECM and the needto cover the remainingQFD inthe electricity sector, is around6.5% of GDP. 64 5 N $ 3 N 2 N n 5 N N r 3 N r 5 N 2 3 N W VI m 3 N 3 x) 3 N 3 \ 3N s N 3 K. THEEFFICIENCYOFGOVERNMENTEXPENDITURE 190. In principle, fiscal space can be created by improving the allocative and technical efficiency of government expenditures, thereby allowing more outputs in terms of public goods and services to be produced for a given amount of budgetary resources. Howeverit is very difficult to quantify ina rigorousmanner the magnitude of expenditure inefficiency and the potential scope for creating fiscal space through efficiency improvements. 191. Allocative efficiency of government expenditures in Tajikistan is likely to be low because budget allocations are still heavily influencedby those which prevailed duringthe Sovietera. These allocationswere heavily biasedtowards capital intensive technologies, highly centralized service provision and very rigid expenditure structures. For example, the health service is overly specialized, segmented and hierarchical with a bias in favor of secondary and tertiary health care and against primary health care. The number of health facilities in the country is around 2,800, which is excessive, and as a result many of these facilities cannot be adequately staffed, supplied with necessary inputs such as medicines, or properly maintained.In the education sector, employee compensation consumes 83% of the recurrent expenditure in the local budgets, from which most schools are funded, leaving insufficient funds for teachingmaterialsand essential classroom maintenance. Many of the schools are in a very poor state of repair and are often unusable in winter because heatingsystems have brokendown. Expenditureson repair and maintenance of all government assets from the State budget amount to only 1.5%o f GDP (World Bank, 2005D: 43),39 which given that the value of the public capital stock must be in excess of 100% of GDP, cannot be anywhere near sufficient. The weaknesses in the budget process for allocating expenditures, discussed in Chapter 4, are a constraintto improvingthe allocativeefficiency of government expenditures. 192. It is also likely that technical efficiency is low in Tajikistan, given that many aspects of the country's public expenditure management system are weak. For example, accountability for the spending of public funds is weak. In principle, spending units face a dual accountability, to both the local authority and the applicable line ministry, but in practice this simply means that there is no effective accountability for the spending of funds. Control of the payroll is impeded because there is no complete civil service register with data on all staff in all budget units. Internal audit is deficient as it focuses mainly on the legal compliance of expenditures, its operatingprinciplesfall short of internationalstandards and there are no internal audit functions in spending agencies. There are weaknesses in public procurement arising from the lack o f a clear legal framework for open competitive bidding and potential conflicts of interest in the roles of the Procurement Agency. (Radev et al, 2005) 193. Methodologies exist for quantifying the technical efficiency of government expenditures. Herrera and Pang(2005) estimate the technical efficiency of education and healthexpenditures in a sample of 140 countries, usingthe Free Disposable Hull (FDH) and the Data Envelopment Analysis (DEA) methodologies: these 39This may underestimate total government expenditure on repair and maintenance because it is possible that some of the PIP expenditures, although classified as investment, are actually for repair and maintenance. 66 methodologies involve estimatingan efficiency frontier, which representsthe level of efficiency reached by the most efficient country in the sample, from a production function linking inputs (sectoral government expenditures) to sectoral outputs (indicators of health status and educational attainments). Herrera and Pang report results for Tajikistan which indicate that the technical efficiency of education spending is between 65% and 78%, and that of health expenditure is between 65% and 88% of the respective efficiency frontiers. However, there are serious weaknesses in data which impede any comparison of expenditure efficiencies in Tajikistan with those of other countries. In particular, the measures of outcomes in key areas such as health and education are unreliable. For example, while official government statisticsshow an infant mortality rate (IMR) of 27.9 deaths per 1000 live births, estimates of the IMR from living standards and demographic surveys range between 78 and 94.5 (World Bank, 2005C). In addition, the production frontier methodology takes no account of factors of production other than government expenditure which could have a significant impact on the output indicator. Government expenditures comprise only a small share of total expenditure in the health sector, approximately 16% in 2003; it is difficult to distinguish the impact of government health spending on health outcomes from that of health expenditures from other sources. Furthermore, the efficiency frontier methodologiesonly apply to technicalinefficiency, not to allocativeinefficiency. Giventhe serious deficienciesin the budget process which impede the alignment of budget allocations with strategic policy objectives, together with the dearth of basic skills for budget preparation, such as the prioritization and costing of expenditure proposals in spending agencies (Chapter 4), allocative inefficienciesare probablyat least as, if not more, importantas a source of expenditureinefficiencythan technical inefficiencies. 194. As such, it is probably not realistic to make any quantification of the magnitude of expenditure inefficiencies in Tajikistan. At best we can acknowledge that major allocative and technical inefficiencies exist and that there is potential for improvingboth types of efficiency through reforms to the PEM systems, which over the long term should create fiscal space. Given that government final consumption and investment expenditures (i.e., excludingtransferpayments and interestpayments) amounted to 17% of GDP in 2006, even a relatively modest 10% improvement in expenditure efficiency over the long term would generate fiscal space amountingto 1.7%of GDP by 2016. L. CONCLUSIONS 195. The major longterm challenges for fiscal policy in Tajikistan are to restructure government expenditures to improvethe quality and efficiency of public services in a manner which can support economic growth and provide poverty reducing public services, while maintaining macroeconomic stability and reducingthe large QFD in the electricity sector. 196. Fortunately, the findings of this study suggest that there is potentially large fiscal space for expanding government expenditures, but realizing this fiscal space will require major reforms by Government. Moreover, failure to implementeffective reforms to public financial management, tax administration and the electricity sector would pose serious risks to fiscal sustainabilityand macroeconomic stability over the medium-to long-term. 67 197. Even before taking into account the scope for improving the efficiency of expenditures, fiscal space amounting to 6.6% of GDP could potentially be realized over the next 10 years, comprising an increase in the budget resource envelope of 8.6% of GDP, minus the additional budget resources required for higher interest payments (0.4% of GDP), higher government electricity payments (0.5% of GDP), higher payments to the energy compensation mechanism for low income electricity consumers (0.4% of GDP) and the funding of the remaining QFD in the electricity sector (0.6% of GDP). If the implementation of public expenditure management reformscouldraise the efficiencyof government expenditureon the provisionof final consumptionand investment expenditures by 10%over 10years (which is a relatively modesttarget), additionalfiscal space of about 1.7%of GDP could be created, raising the combined fiscal space from budget resource mobilization and expenditure efficiency improvements to more than 8%of GDP by 2016. Table 3.10 summarizes the projections of fiscal space in 2016, the final year for the projections in this study, compared to 2006. 68 2016, change over 2006 Budget Resource Envelope 8.6 Gains in ExpenditureEfficiency 1.7 (assumes 10%efficiency gain) Net fiscal space 8.4 198. Increased domestic revenues, potentiallyamountingto 5.6%o f GDP make the largest contribution to fiscal space over the next 10 years; however most of these gains are contingent upon major institutional reforms to the tax and customs administration. Tax administration reforms, which include internal restructuring of the tax department, introductionof modern systems of tax administrationfocused on voluntary taxpayer compliance enforced by selective risk based audits, computerizationand better trainingand incentivesfor tax and customs officials, could potentially generate additional tax revenue of 3.2% of GDP. Most of the additional gains to domestic revenue, on top of those generated by tax administration reforms, are also contingent on policy reforms. Increasing the marginal rates of personal income tax could contribute gains of up to 1.2% of GDP, while revenues amounting to 1.3% of GDP couldbe generatedfrom the HPP investments, inthe form of returns to government's own investment in the HPPs and the tax or royalties levied on their earnings. The viability of the HPPs is also dependent upon reforms in the electricity sector, particularlyincreasingdomestic tariffs to commerciallyviable levels. 199. The scope for expanding the budget resource envelope through additional borrowing is very limited. Disbursements during 2007-2009 of the external loans, which have already been contracted, will raise the NPV of PPG external debt above the GDP and export thresholds at which it is considered sustainable under the IDA/IMF Low Income Countries Debt Sustainability Framework. Consequently, while external debt disbursements are projected to rise sharply during 2007-09 because of the disbursement o f Chinese loans, they must subsequently fall back to around 4-5% of GDP ifthe NPV of externaldebt is to be broughtback in line with the sustainabilitythresholds. 200. The main constraint to higher domestic borrowing is not the sustainability of government's domestic debt, which is very small, but the macroeconomic 69 consequences of domestic borrowing from very shallow and highly dollarized domestic financial markets. Broad money is the equivalent of only 8% of GDP, foreign currency deposits comprise more than two thirds of all bank deposits and the non bank financial system barely exists. Consequently there is very little scope for issuinggovernment debt to the commercialbanks without crowdingout private sector borrowers from credit markets. There are no non-bank financial institutions which can purchase debt, and any significant amount of government borrowing from the Central Bank would jeopardize the latter's efforts to control reserve money. The improvementinthe economy achieved since the late 1990s has beenmade possible in part because of the restoration of macroeconomic stability, which in turn is attributable to the reduction of the fiscal deficit. The macroeconomic stability achieved over the last few years will be put in jeopardy if Government resorts to domestic borrowing or if external debt again becomes unsustainable because of excessive externalborrowing. 201. Public expenditure reforms will also play a role in mobilizing more donor grants for the budget. The Government budget has benefited from only a small fraction of the donor grants disbursed, although this situation improved in 2006. Increased grant inflows, driven by both higher levels of donor aid to low income countries in general and a greater willingness on the part of donors to channel their aid through the Government budget, which is contingent upon reforms which can assure donors that their funds are efficiently and transparently spent, couldpotentially yield additional budget resources in the form of grants equivalent to 1% o f GDP by 2016. 202. The largest single threat facing the budget is the QFD of the electricity sector, estimated at 9.3% of GDP in 2006. The QFD is the difference between the revenue which the state owned electricity utility requires operatingand properly maintaining the existing infrastructurefor the generation and supply o f electricity andthe revenue which it actually collects in cash. The main cause of the QFD is suboptimal electricity tariffs, supplemented by technical and commercial losses and poor cash collectionof electricity bills. Althoughthe QFD is not currently a cash burden on the Government budget, it will become so if no measures are taken to drastically reduce it. The consequences of the Government not funding the QFD will be that essential maintenance and rehabilitationof the electricity transmissionand distribution system will not be undertaken, and hence electricity supply will deteriorate. Fortunately there are measures, albeit politically painful ones, which can be taken to slash the QFD, principally by raising electricity tariffs to the level equivalent to the long run average incremental cost of electricity supply (which requires an almost fourfold increase in average tariff levels from those prevailing in 2006). Higher electricity tariffs will have a negative impact on some components of the budget, reducingtax revenue paid by TALCO, the aluminum smelter, and increasing the costs to the budgetof government's own electricitybills and transfer paymentsto compensate low income electricity consumers. But the aggregate fiscal impact of electricity sector reforms is overwhelmingly positive, generating net gains of almost 105 of GDP. 203. The fiscal space detailed in these projections, if realized, would allow government expenditure to be raised to around 30% of GDP over the long term, compared to 22% o f GDP in 2006. While this would make the funding of priority expenditures much more feasible, it will not be sufficient to generate higher growth 70 and poverty reductionwithout substantial improvements in the quality of governance, and especially in PFM systems. The preliminaryfindings of the World Bank's ECA RegionalFiscalStudy indicatethat, in countries with poor governance (which include Tajikistanwhich is ranked second worse among 27 EasternEurope and CentralAsian countries in terms of government effectiveness), potentially productive public expenditures (defined as education, health, housing and economic affairs, including transport) have little positive impact on economic growth, whereas unproductive public expenditures have a negative impact. Consequently, in countries with poor governance, large total public expenditures reduce growth, probably because they require high levels of taxation which distort incentives for work, investment and saving; hence growth is maximized by keeping the size of government small (Varoudakis, 2006). 204. The implications for Tajikistan are that major improvements in the quality of governance will be a prerequisiteif increasedgovernment expenditures in the priority sectors are to have a positive impact on economic growth. Improved governance must precede an expansion of the budget. Mobilizing more tax revenues to fund an expansion of the budget in the context of poor economic governance, including an untransparent tax administration system, may depress economic growth through distortions to resource allocation, while the benefits of an increase in government expenditurewill be limited if governance is not improved. Moreover, many of the measures identified in this chapter to expand the budget resource envelope, such as tax administrationreforms, will take many years to generate more resources, whereas there is an urgent need to ensure that the existingbudgetary resources are used more efficiently to deliver better quality public services. 205. Consequently the message that emerges clearly from this Chapter, as well as other Chapters,, is that institutional reforms to improve the quality of public expendituremanagement, government fiduciary systems and tax administration, and reforms in the electricity sector, as well as reformsto the business climate for private investors, are essential if Tajikistanis to bothgenerate fiscal space over the longterm and to benefit, in terms of economic growth and poverty reduction, from the higher government expenditures madepossible by the fiscal space. 71 4. HOW CAN THE MTEFBEINTRODUCEDIN TAJIKISTAN?40 A. INTRODUCTION 206. The MTEF is a tool for sector-based medium-termbudget planningwhich has been introduced in a wide range of developing and transitional countries,` often with the support of donor agencies who view reforms to public expendituremanagement (PEM) as critical to the achievement ofthe MDGs. The majorobjectiveof the MTEF is to facilitate policy-based budgeting (Le., linking budget allocations to strategic policy objectives) within a sound fiscal framework. The MTEF is now regarded as "the sine qua non of good PEM" (Le Houerouand Taliercio, 2002). 207. In May 2006, the Ministry of Finance (MOF) organized a Round Table, attended by government officials and representatives of donor agencies, to present a proposalfor introducing an MTEF in Tajikistan on a pilot basis, beginningwith the 2007 budget process. The MOF subsequently circulateda Government Resolutionto obtainpolitical authority for the introductionof the MTEF and the radicalchanges to the budget process that this will entail. This is not the first effort to implement an MTEF in Tajikistan. The MOF had begun work on an MTEF (termed the Medium Term Budget Framework [MTBF]) in 2000 and actually presented the MTBF at the 2002 Consultative Group Meeting. However, this previous effort, while providing some useful information on projected medium term expenditures, broken down by functional classification,hadno more than a marginalimpact on budget planning. 208. The problems experienced by Tajikistan in introducing an MTEF (and in implementing PEM reforms in general) are not unique to this country especially in comparison with the other CIS which share a similar history and institutional structures. A recent World Bank study (Andrews and Moon, 2003) explored the implications of weak PEM capacity for developing the PRSP approach (for which policy based budgeting is integral) in the PRSP countries of Eastern Europe and CentralAsia (ECA). One of the mainfindings of this study is that PEMcapacities are even weaker in this group of countries than in those low income countries which qualify for HIPC debt relief; that capacity for budget formulation is the weakest aspect of PEM, and that efforts to introduce PRSPs in ECA countries have run far ahead of the PEM capacity required to support effective policy based budgeting. Moreover, of the 11 ECA countriesevaluated in this study, Tajikistan was judged to have the weakest PEMcapacity. 40This chapter is based on the non lending technical assistance for launching the MTEF in Tajikistan which began in January 2006 and is managed by the World Bank and DFID.DffD. The strategy for implementing the MTEF was presented at a Round Table organized by the Ministry of Finance in Dushanbein May 2006 and subsequently endorsedthrough a GovernmentResolutionin October 2006. 72 209. The purpose of this chapter is to understandwhy the MTEF proved so difficult to introduce in Tajikistan, examine the obstacles to its introductionand to map out a set of feasible measureswhich will allow for a gradual introductionof the MTEF over the next few years. While there is now strong political will to implementan MTEF in Tajikistan, there are major constraints especially in relation to the existing institutional relationships of the budget process and the very weak (almost non- existent in some agencies) capacities for implementingmodern budget process in the public service. Hence it is necessary to be realistic about the pace at which an MTEF can be introduced and to be cautious in implementingreforms, especially in relation to aligning the introduction of new budget processes to the building o f technical capacities in the relevant agencies, as well as promoting a broader understandingof the rationale for the MTEFwithin the public service and in civil society. A dogmatic approach to MTEF implementationshould be avoided; there is no blue print for the MTEF which can be applied uniformly to every country. Insteadthe introductionof MTEF reforms must be done in a pragmatic matter, with each step in the reform process carefully designed to ensure that it is compatible with the technical and institutionalcapacities which are necessary for itseffective implementation. 210. The chapter is organized as follows. Section 2 explains the motivation for introducing the MTEF. Section 3 outlines the extent of progress made so far in developing the MOTIF. Section 4 discusses the critical constraints and obstacles to introducingan MTEF in Tajikistan. Section 5 outlines a strategy for the introduction of an MTEF, in a phased manner, and Section 6 sets out the measures which will be requiredto implementthis strategy. B. MOTIVATION INTRODUCINGTHE MTEF FOR 211. The primary motivation for introducing the MTEF is to improve budget resource allocation and to link this with the strategic policy objectives of the Government, including those set out in national development strategies such as the PRSP. Rationalbudget allocations, in which expenditures are prioritisedto best meet the budget's objectives is not possible under the existing budget system for two reasons. The first is the fragmentation of the budget among more than a hundred KBOs and the separate determination of current and capital budgets, which is discussed in Section 4, and the second is the norm-basedsystem of allocating budget resources, whereby expenditures are determined by formulas linked to structures in place (such as schools or hospitals). 212. The existing structure of facilities and staffing is highly suboptimal and inefficient: it was inherited from the Soviet era when budget resources were much larger, and even if it was optimalthen (which is debatable) it is certainly no longer so given the steep decline in available budget resources. For example, there are approximately 2,800 health facilities in the country, including hospitals in villages, but many of these facilities are little more than crumblingshells: there are no patients in the beds, no drugs and very few staff in attendance. There are also large numbers of authorized staff posts, reflecting the norms and the facilities in place, but only a fraction are filled because the salaries are far below even subsistence level. Rationalising health facilities and staffing levels is essential to improve allocative efficiency but the existing system provides neither the institutional mechanisms nor the incentivesfor doing this. The most appropriate institutionsfor allocating budget 73 resources in line with strategic policy objectives are the line ministries, which have responsibilityfor formulatingsector policies, but under the existingbudgetsystem the line ministries have only limited influence over sectoral budget allocations because they directly controlonly a small share of these allocations; hence policy formulation is divorced from budget allocations. The most important change to the budget process, with the introductionof the MTEF. will bring about is the centralizationof responsibilityfor formulatingsector budgets inthe relevantlineministry. 213. Although the main motivation for reformingthe budget allocation system is to improve the allocative efficiency of the budget, it will also provide additional benefits, not least because by linking budget allocations more clearly to policy objectives in a transparent manner, it will encourage donors to channel more of their aid directly to the Government budget in the form of budget support. This is essential if the severe constraints on funding for essential current expenditures, such as operational and maintenance expenditures in the health and education sectors are to be alleviated. c. WHAT HASBEENACHIEVED SO FAR? 214. The MTBF prepared by the MOF consists of a set of three year fiscal projectionscoveringthe main revenue heads, expenditures and financing. The 2006 MTBF, which includes the 2006 budget plus the projections for 2007 and 2008, is shown in Appendix Table 1. The Government budget consists of two separate components: the state budget and the PIP. The MTBF breaks down the state budget expenditures into sectors, which correspond to broad functional classifications:e.g., law enforcement structures, education and health. There is no breakdown of the PIP into sectors. The state budget is composed of three main institutional components: the Republican budget (comprising national institutions such as line ministries); the local government budgets and special funds (such as pensions). This breakdown is not shown inthe MTBF. 215. The main value of the existing MTBF is that it provides a medium-term aggregate fiscal framework which is both internally consistent (there are no financing gaps) and consistent with macroeconomic objectives, in particular Government's projecteddomestic borrowingrequirementis consistent with the Bank o f Tajikistan's planned monetary program. Up to 2006, the MTBF closely reflectedthe three year PRGF programme which the Government negotiatedwith the IMF (the PRGF ended in 2006 and no successor programme has been implemented) The fiscal framework , of the MTBF provides a hard budget constraint for aggregate expenditure and, therefore, is a useful tool for macroeconomic management. However, the existing MTBF does not play any meaningfulrole as a budget planningtool at a lower level of expenditurethan that o f aggregate expenditure. 216. The projections of sector expenditures contained in the MTBF have the appearance of sector expenditure ceilings, but they do not play this role in the budget process. This is because no budget institutionactually receives the ceiling for each sector and has the responsibility for drawing up a sector budget within that ceiling. As discussed in the following section, responsibility for budgeting is fragmented among a huge number of KBOs, bothwithin, and spanning, different sectors. In fact, because of the fragmentation of budget institutions, there is no effective sector 74 planninglinked to budget allocations in Tajikistan. Instead of acting as a hardbudget constraint for sector expenditure planning, the sector expenditure projections are simply the aggregation, across all KBOs, of their projected functional expenditures relatedto each sector: e.g., the projectededucationsector expenditureis the sum of all expenditures on education by each o f the 40 plus KBOs which undertake education expenditures. At best these projections provide some limited data on projected expenditures by functional classification, but even that is somewhat tenuous, because in the absence of hard budget constraints at the sector level, there is no mechanism within the budget process to ensure that, once future budget estimates are prepared, they will not deviate from the sectoral breakdowncurrently shown in the MTBF. 217. It is useful to view an MTEF as consisting of budget planning at three levels, differentiatedby level of aggregation. 218. Level 1 is the determination of a budget resource envelope derived from a medium term fiscal framework, consistent with macroeconomic objectives. The budget resource envelope defines how much Government can allocate to all of its expenditures in aggregate. 219. Level 2 is the division of the budget resource envelope into a set of sector ceilings. A sector ceiling should represent a hard budget constraint under which a sector should plan and prioritize its expenditures: Le., the sector ceiling is not simply a forecast or a projectionof a sector's expenditures; it is a crucial planningtool of the MTEF. T he allocation of budget resources into sector ceilings should be determined by inter-sectoralbudget priorities which are set out in national planning documents, such as the PRSPandNDS. 220. Level 3 is the allocation o f the expenditureceiling for each sector into a sector wide budget. This allocation should be driven by the intra-sectoralpriorities set out in the medium-term sector expenditureplans of each sector. 221. Viewed from this perspective, the MTBF in Tajikistan fulfills the requirements o f Level 1 of MTEF budget planning, but not Levels 2 and 3. This is also reflected in the respective roles which different budget institutions have played in preparing the MTBF. T he line ministries make almost no input into preparation of the MTBF, which instead is preparedalmost entirely by the MOF. 222. The priority for PEM reform should be to try and gradually extend the MTEF to take on boardlevels 1and 2, consistingof the inter and intra sectoral allocations of expenditures, without jeopardizing the successes achieved in the area of macroeconomic management brought about largely through a sound fiscal framework. The following section examines the constraints and obstacles to extendingthe MTEF to encompass sector expenditureplanning. D. CONSTRAINTSTO INTRODUCINGSECTORBASED PLANNING 223. Sector-based budget planning faces three major constraints in Tajikistan: the fragmentation of the budget, the separation of current and capital budgets and the lack of capacity for budget planningwithin budget institutions. 75 Fragmentationofthe Budget 224. There are hundreds of different budget organizations(Le,, organizations which receive and spend funds from the State budget). A critical distinction is between KBOs and other budget organizations (non-KBOs). KBOs deal directly with the MOF in preparing budget submissions and negotiating with the MOF during the budget process. There are 118 KBOs comprising line ministries, The Presidential Administration, the Majlisi Oli, the courts, sundry commissions, committees, academies and universities and local governments (three Oblasts, 13 Rayons of RepublicanSubordination(RSS) and Dushanbe.Non-KBOs do not deal directly with the MOF; instead they are subordinated to a KBO. For example, the rayons, except for the RSS, are subordinated to the Oblast in which they are situated. There are several hundred non-KBOs. The fragmentationof the State budget is partly a legacy o f the Soviet era, when Republican budgets in the Soviet Union were typically fragmented among many budget organizations without there being dominant line ministries, and partly the consequence of the decentralization implemented after the end of civil war inthe 1990s. 225. The problem for sector-based budget planning is not the large number of non- KBOs - in most countries there is a plethora of budget organizationswhich actually implement expenditures - but the large number of KBOs. With 118 KBOs, many of which, including all of the 17 local governments, have responsibilities which span several sectors, the number of KBOs in most sectors runs into double digits. In the education sector, there are around 40 KBOs. In most o f the sectors, there is a line ministry which has responsibility for formulatingpolicies, but that line ministry only controls a fraction of the expenditures within its sector. For example, the Ministry o f Educationdirectly controlsonly 9% of the total education budget:the remaining91% is split among the 40 plus KBOs in the sector, with the 17 local governments accountingfor almost all of this. The Ministry of Healthcontrols only about 10% of the health sector budget. As each KBO submits its budget requests directly to the MOF and then negotiates with the MOF to finalize its budget estimate, the line ministries are effectively bypassed in the determination of expenditure allocations. The line ministries can influencethe content of the budget instructionswhich are sent by the MOF to KBOs at the start of the budget process, but they have very little influenceover the actualallocationof expenditures except for their own expenditures (whichare largely their administrativeexpenditures and the expenditures of those non KBOs which are subordinatedto them). 226. Fromthe standpoint of sector budget planning, the fragmentationof the budget among over 100 KBOs has several adverse consequences. First, there is no single budget institutionin each sector to which the sector ceiling can be applied and which can take responsibility for prioritizing sector-wide expenditures according to sector policies. Secondly, the connection between sector policy formulation and budget planning is very tenuous, because there is no mechanism through which policies formulated by the line ministries can be translated directly into budget allocations. Thirdly, in terms of the volume of expenditures, the most important KBOs are the local governments, which control about 30% of all of the expenditures in the State budget, excludingthe special funds. The focus of the local authorities is not meeting nationally determined sectoral objectives; it is to meet the specific needs of their communities. There are legitimate reasons for giving local authorities control over 76 budgets, but for them to control the major part of the budgets of key sectors such as health and education is not compatible with a budget based on nationally formulated sector budget policies. It would still be possible to develop an MTEF in a decentralized system, in the sense that medium-term expenditure plans could be drawn up, but it would be an MTEF devoid of sector-based expenditure allocations and sector-based expenditure plans and, therefore, an MTEF which would be very difficult to align with strategic policy objectiveswhich are also based around sectoral objectives, such as a typical PRSP. Separation of Current and Capital Budgets 227. In addition to the fragmentation of the budget among the numerous KBOs, the budget is fragmented betweenthe current budget andtwo separate capital budgets; the Central State Investment Program (CSIP), which is part of the State budget, and the PIP which is fully funded by external donors. The CSIP and the PIP fall under the aegis of the Ministry of Economy and Trade (MOET). The selection of projects for the CSIP and PIP involves a process of negotiationbetween the KE30 promoting the project, usually in collaboration with the external funding agency if the project is in the PIP, and the MOET. Although the CSIP, unlike the PIP, is formally integrated into the State budget, what matters for budget planningis that the process of making allocative decisions in relation to the CSIP is distinct from that relating to the current expenditures in the State budget. 228. It is not clear how the selection of projects relates to expenditure ceilings, if it does at all, and there is no relationship between ceilings given to KBOs for current expenditures and capital expenditures. Therefore, sectors or KBOs cannot trade off capital expenditures for current expenditures, or vice versa, and so cannot formulate a sector-wide budget which is fully comprehensive and covers current and capital expenditures. In fact it would appear that KBOs have every incentive to try and maximize their share of the CSIP and PIP, because this has no negative impact on the allocation of budget resources that they will receivefor current expenditures, but may boost their allocation in the future because these allocations are largely driven by spending norms linked to structures (therefore the more structures that are built now, the larger will be the current budget allocation inthe future). 229. The separation of the PIP from the current budget is, of course, typical in many developing countries mainly because PIP projects are driven by the availability of funds from donors, and hence decisions to undertake these projects are not easy to integratewithin the normal budget process. However, it is more unusualto also treat separately that part of the capital budget which is funded from the Government's own budget resources. 230. The consequence of the fragmentation of the budget among over 100 KBOs and the separation of the current and capital budgets in the budget process is that meaningful sector planning, whereby sector wide expenditures are prioritised and allocated according to a coherent medium term sector expenditure plan, is rendered impossible. The sector expenditures shown in the MBTF are not the product of a coherent sector planning process, they are simply the amalgamation o f numerous atomised allocative decisions taken largely in isolationfrom each other. Furthermore, the "sector ceilings" shown in the MBTF can play no meaningful role in the budget 77 planning process; their only value is to show, once the budget process is complete, how much resourceshave been allocatedto each sector. Capacities for Budget Planning 231. Capacities for budget planning throughout government are very weak. Even basic concepts of strategic budget formulation and prioritization are not well understood, partly because the budget system which was inherited from the Soviet era, and which is still largely intact, was very different from that o f a modernwestern budget system, in which a set of clearly defined government activities are costed and prioritized. The Soviet budget system was essentiallya mechanicalexercise in which the budget was subordinatedto a broader five year economic plan. The methodology for determining budget allocations involved the use of input norms, whereby allocations reflected structures in place, rather than any systematic attempt to match resources to policy objectives. Normative budgeting is still the dominant methodology for determining budget allocations in Tajikistan. In addition, the Soviet budget system did not entail a central role for strong line ministries in budget management. Therefore these ministries did not develop a body of expertise in managingthe budget in their particular area. Instead, as was noted above, the budget system was highly fragmented with numerous small budget organizations having responsibility for expenditures, which prevents economies of scale in activities such as budget planningbeingrealised. 232. Line ministries have very few staff, if any, with expertise in formulating budgets, such as expertise on costing activities, undertaking quantitative project appraisals, or prioritising expenditures. Within line ministries, only a small share o f the staff is devoted to budget formulation, which is not surprising given that the line ministries themselves have such a limited role in the budget process. For example, in the Ministry of Education, the planning and economy department which deals with budget formulation (inthe 44 budget organizationsdirectly subordinatedto the MOE) has only seven staff members. Tajikistan's systemic weaknesses in budget formulationare shared with most other CIS countries(Andrews and Moon, 2003). Summary 233. The analysis above suggests that the lack of progress achieved so far in implementing the MTEF in Tajikistan is not simply the result of a lack of "political will". Rather it reflects major constraints arising from the structure of the budget process and severe capacity weaknesses. The budget system inherited from the Soviet era could hardly be less conducive to the development of policy based budgetingfocussed on sectoral budget allocations and sector planning, which is the essence of an MTEF. Yet this budget system has changed in few fundamental respects in the 15 years since independence, and the major change which has occurred, the introduction of "ad hoc" decentralizationprocesses and the shifting of most social expenditures into the local budget, is a further barrier to budget formulation driven by sector budget policiesformulatedat the national level. 78 E. WHAT CHANGESARE REQUIREDTO SUPPORT THE INTRODUCTIONOF THE MTEF? 234. To implement fully the MTEF in Tajikistan, it will be necessary to effect radical restructuring of the relationships between different budget institutions in the budget process, to restructure line ministries and build up their capacity for budget formulation, to integratethe state capital investment program and, at a later date, the PIP, with the current budget, to lengthen the budget timetable and to centralize tax payments. Radical institutional restructuring of budget processes would be difficult even in public administrations characterized with strong administrative and technical capacities and experience of implementing institutional reforms, neither of which characterizes Tajikistan. Consequently, it will be necessary to proceed with the reforms cautiously, with the introduction of new budget processes being matched to the building of technical capacities for budgeting in the MOF and line ministries, supported by TA from donors. Chart 1 ExistingBudgetInstitutionalArrangementsin Each Sector KBOs Ministry uIDepartment 1 OtherKBO Budget instructionsfrom MOF to KBOs Expenditurerequestsfrom KBOs to MOF Bilateral negotiationsbetweenMOF and KBOs Final budget allocationsfrom MOF to KBOs Note: Some KBOs budget for expendituresin more than one sector RestructuringBudgetRelations 235. The most important changes required to implement the MTEF relates to the organization of the budget into sectors. E ach sector, includes many KBOs, but one 79 budget organization in each sector must take the lead in coordinating the KBOs in the sector and formulating budget policies ifa comprehensivesector-wide budget is to be prepared. The only feasible KBOs which can assume this task are the line ministries. Although some sectors contain one or more line ministries, only one of these has responsibility for policy formulation in that particular sector (e.g., the Ministry of Education has responsibility for education policy, although there are other line ministries which operate educational establishments). We will refer to a line ministry with policy responsibility as a lead line ministry (LLM). The critical reform needed to enable effective sector budgeting to be carried out is to make the LLMs responsible for formulating the entire budget in their respective sectors, so that budget allocations can be aligned with sector policies. 236. The required restructuring will entail all KBOs in a given sector channelling their budget submissions through the relevant LLM for that sector, with the latter having the authority to revise the budget submissions of KBOs to ensure that they accordwith the sector's priorities and can be accommodatedwithin the overall sector expenditure ceiling. This will meanthat most KBOs will no longer make their budget submissions directly to the MOF. Insteadthe MOFwill receive a consolidatedbudget submission covering the entire sector from the LLMand will negotiate with that LLM to finalize the sector's budget allocation. Ineffect KBOs which are not LLMs will be subordinated to the LLMs for the purpose of budget planning and will become more like non-key budget organisations. In the Republican budget, this switch will be straightforward, at least in theory, because most KBOs other than line ministries are relatively small and have a single area of responsibility: hence they will be subordinated to one LLM for all of their expenditures. The local governments, however, have expenditure responsibilities which span several sectors, and hence they will be subordinatedto several different LLMs. Chart 2 MTEFBudget InstitutionalRelationshipsin Each Sector Other KBOs Other .',' 4 b ........................ b Budget instructions, including sector ceilings Allocation of sector expenditures from MOF to LLMs Negotiations betweenLLMand Submissionof sector budgetestimatesfrom other KBOs LLMsto MOF Bilateral negotiationsbetween MOF and L%s to finalise budget allocations 237. Chart 1 show the budget relationshipsunder the existing budget system: each individual KBO deals directly with the MOF, receivingbudget instructions from the MOF, making a budget submissionto the MOF and then negotiatingwith the MOF. Chart 2 shows the budget relationships required for a sector-based MTEF. In each sector, the LLMs intermediatebetweenthe MOF and all other KBOs inthe sector. 238. It will not be possible to restructure the entire budget around sectors led by a LLM, for two reasons. First it is unlikely to be politically feasibleto subordinate the budgets of politically powerful KBOs such as the Apparatus of the President or the Majlisi Oli to a LLM. These KBOs will continue to deal directly with the MOF outside of any sectoral structures which are put in place to cover public administration. Secondly, while it may be possible to create a sector covering domestic law enforcement institutions, it is likely that the defense will be treated separately, for security reasons. This is not a fatal omission for the MTEF however. In most countries where the MTEF has been implemented, there are budget organizations which still operate outside of sectors, for political or security reasons. What is important is to organize as large a share of the budget as possible into sectors, and especially those components of the budget which are most closely linked to the policy objectives inthe PRSP, such as the social sectors. 239. Table 4.1 shows the five best candidates for creatingsectors, together with the LLMfor each sector and some of the other KBOs which will become subordinate to the LLM. As can be seen from the table below, these five sectors comprise almost half o f all expenditures in the State budget (and slightly more than half of primary expenditures). The main omissions are public administration bodies and foreign economic relations, and law enforcement structures, which together account for 35% of the State budget. As mentioned above, it would be desirable to create a sector for domestic security, encompassing the Ministry of Internal Affairs, police, prisons and the law courts. Sector Lead Line Ministry Subordinated KBOs Share o f State budget (%) Education Ministryof Education Committee on Sport and 21.8 Physical Education, univ., Local governments Health Ministry o f Health Local governments 6.9 Social Protection Ministryo f Labour and Social Protection Fund 12.4 Social Welfare Agriculture MinistryofAgriculture Ministryof Environment 3.2 Protection and Forest Economy, Ministryof LandImprovement and Water Economy Academy o f Agricultural Sciences Forest Economic Production Association Transport IMinistry o fTransport ILocal governments 4.9 81 240. Inaddition to the KBOs, there are hundreds of budget organizationswhich are non-key budget organizations and which are subordinated to KBOs in the existing budget system. In most cases, these relationships will not change when the budget system is restructuredaround sectors. 241. The required restructuring of the budget process to move towards a more centralized structure for preparing a sector-based budget will clearly not be easy to implement and will take several years to complete; both to reorganize the budget process and build capacities for budgeting in line ministries. Hence the strategy recommended in this Chapter is to move gradually, beginning with one pilot sector, and then bring other sectors into the MTEF at a later date. The pilot sector is the education sector, hence in the short term (1-2 years), the aim should be to create a sector budget process in this sector, with the Ministry of Education as LLM. In the medium term (3-5 years), another two sectors can be brought into the MTEF. The lessons learned from introducingthe MTEF in the pilot sector should prove valuable when the MTEF is introducedin other sectors. 242. With a sector-based MTEF, the role of the LLM will be central to budget planning in each sector. The LLM will receive a single expenditure ceiling for its sector from the MOF at the start o f the budget process. The LLM will take responsibility for allocating that ceiling among all the KBOs in its sector and for ensuringthat each KBO's expenditure is allocated accordingto the sector's priorities. It will send budget instructionsto all the KBOs in its sector, receivebudget requests, and hold negotiationswith them as part of the budget process. The sector itself, led by the LLM, should have the discretion to allocate expenditures among the main economic classifications of the budget provided that the overall sector ceiling is not breached and that statutory requirements, unavoidable expenditures such as utility payments and government directivesare fully taken into account. Having prepared a sector-wide budget estimate in consultation with the KBOs in its sector, the LLM should then submit this estimate to the MOF, the aggregate of which should not exceed the sector ceiling. There will then be negotiations between the LLM and the MOF to finalize the sector's budget allocation. The MOF itself should not revise the sector's estimates provided that they comply with some basic budget requirements, such as fully providing for unavoidableexpenditures. In effect, the role of the MOF in the sector's budget is to ensure that the sector ceiling is not breached and that budget implementationwill not bejeopardizedbecause insufficient budget allocations have been made for unavoidable expenditures, such as statutory expenditures and utilities. Responsibility for budgeting within the sector should be delegated to the sector itself, under the direction of the LLM. 82 The Roleof LocalGovernments in the BudgetProcess 243. The local authorities will continue to implement expenditures in various sectors but the introductionof an MTEF will entail major changes in their role in the budget process. In the current system, the local authorities, which are KBOs, are responsible for receivingbudget submissions from subordinated budget organizations and then making a budget submission to the MOF. To be consistent with a sector- based MTEF, the local authorities, which are KBOs, should make their budget submissions to the relevant LLMs and consequently they will no longer deal directly with the MOF in the budget process, at least for those areas of expenditure which are organized into sectors. 244. Centralizing budget formulation in line ministries will clearly impinge on the authority of local governments, although the discretion of local governments is heavily circumscribed under the existing budget system. It will be desirable, especially from the standpoint of promoting greater local accountability and autonomy, for Governmentto allocate a portionofthe localbudgetto local authorities to be spent entirely at the discretionof the local authoritiesthemselves. For example, these expenditures could cover local authority administration costs and specific projects importantto local needs. This should be accompanied by measures to build capacity inlocal governments for budgetplanning, so that the budget funds which are availableto be spent at their discretionare utilisedin an optimalmanner to meet local needs. In the medium-term, each Oblast and RSS should be encouragedto prepare a LocalGovernmentBudget FrameworkPaper, settingout the local priorities for public expenditure and how local government budget allocations will be used to meet these priorities. Integrationof Capital and Current Expendituresin Sector Budgeting 245. With the sectors themselves having responsibility for all of their expenditures, including the allocation between the different economic classifications of expenditures, there will be no need for a separate Central State InvestmentProgram. All capital investment projects will instead be fully incorporated into the sectoral budgets, and planned entirely at the sector level, and the decision to undertake a proposed capital investment project will be made by the relevant sector, taking into account whether the project is in accord with its expenditure priorities, whether it is affordable under the medium-term sector ceiling and the implications of the project with respect to current budget costs in the future, once the investment is completed. Each sector, led by its LLM, will have the discretion to determine how much is allocated to current expenditures and how much to capital expenditures, within the overall sector expenditure ceiling. The aggregate amount of capitalexpenditurein the budget will, therefore, not be determined centrally: it will simply be the outcome o f the sum of decisions taken at the sector level. Therefore, the role of the MOET in planningcapitalexpenditures in the State budget will become largelyredundant. 246. Eventually the PIP should also be incorporated into sector ceilings, although this is more complicated because o f the separate funding arrangements for the PIP and also becausethe levelof externalborrowingshould be a budget constraint. 83 StrengtheningCapacities in Line Ministries 247. The greatly enhanced role of line ministries in the budget process will not be possible without a major strengthening of their capacities for budget planning; including capacities for costing expenditure proposals, linking activities to policy objectives, cost benefit appraisalof projects, and the longterm forecastingof demand for services. Line ministries will need to be restructuredto allocate many more staff to budget planning functions. They will also require substantialtechnical assistance in budget planning, especially in critical sectors such as the social sectors, and this should be a priority for the TA programs of donors. 248. Sector Working Groups (SWGs) should be set up in each of the sectors to act as a forum for all of the major KBOs to discuss both sector policy and budget allocations. The SWGs should be chaired by a senior official of the LLM of that sector. Their principle tasks will be to: (i)prepare a mediumterm expenditure plan, settingout the policy objectivesof the sector, the prioritization of expendituresto best meet those objectives, and the costing of expenditures; (ii)provide a forum for discussions within the sector on the allocation of that sector's budget resources; and (iii)monitorthe implementationof the sector's expenditures and progresstowards meetingthe sector's policy objectives. The Processfor DeterminingSector ExpenditureCeilings 249. The allocation of the budget resource envelope into sector expenditure ceilings is not simply a technical exercise; it has a political dimension because the allocation of budget resources involves political choices between competingstrategic demands. Therefore to ensure that the MTEF can command widespread political support, there must be a political mechanism for determining inter-sectoral budget allocations. In effect, this means that the final decision over these allocations should be made by a body comprisingkey political actors, such as Ministers, even though such a body will require technical advice from the MOF to help it reach decisions. The National Budget Commission (NBC) could play this role. Whichever body is given this role will needto draw up criteria for guiding inter-sectoralbudget allocations. The Budget Timetable 250. The institutional arrangements required to implement an MTEF also have important implications for the timetable of the budget process, because some steps in the budget process must logically precede others ifthe MTEF is to be implementedin a meaningful way. In particular, resource envelopes must be determined before resources are allocated; this applies both at the level of aggregate budget and at the sector level. In the existing budget process, the determination of the resource envelope and the allocation o f expenditures are conducted simultaneously, because the budget submissions from KBOs include both expenditure requests and revenue projections. 251. There are five distinct steps in the budget process which must be followed in sequence ifthe MTEF is to be implementedproperly. 84 Step 1:The MOF determines the budget resource envelope. Step 2: The budget resource envelope is allocated into sector expenditure ceilings and the sectors are informedof their respectiveceilings. Step 3: The sectors, led by the LLM, determine detailed budget allocations within their ceilings and then, through the LLM, submit budget requeststo the MOF. Step 4: MOF holds bilateral negotiations with the LLM in each sector to finalize sectoral budgets. Step 5: MOF submits the budget to Cabinet for approval, followed by submission to the Majlisi Namoyanagon. 252. The current budget process starts towards the end of April or the beginningof May. Introducing the MTEF will require a longer budget process, hence an earlier starting date, in order to allow sufficient time for stages 2 and 3 in the above sequence; and especially stage 3 which is likely to be time consuming. Distributionof Taxes 253. Under the existing budget system, the collection of taxes is decentralized to offices of the Ministry of State Revenues and Duties in the local government areas. During the budget process, revenue projections are made for each local government area. The expenditures under the local budget are then funded through a combination of the retention o f a predeterminedpercentage o f the taxes (such as VAT and excise tax) collected in the local area and transfers from the central treasury. Under this system, the local governments bear the risks and benefits of any deviation in taxes collected at the local level relative to the budgeted levels. This system provides perverse incentives for local governments to under budget for revenues; because this allows them to retain more of the revenue which is actually collected and to spend this without proper accountability (e.g., it is spent on items which have never been appropriatedby the National Parliament). 254. Because an MTEF requires that most expenditure allocations should be determinedcentrally, there is no reason for taxes collected by tax offices in the local authoritiesto be retainedby the local authorities. Insteadall taxes should be remitted to the national Treasury which should then distribute funds to the local authorities, in line with the budget estimates and the dictates of cash management. The exception would be those taxes specifically earmarked for spending at the discretion of local authorities (such as land taxes) which would be retained in full by the local authorities. Centralizing tax collections would also mean that the risk of revenues under performing would be transferred to the center, because expenditures on the local budgets would no longer be tied directly to taxes'collected in the same local area. Consequently there would be no need for separate revenue projections for each local government area. Instead tax administration could be carried out in a more centralizedmanner, with many o f the local offices, which collect a negligible share of total tax revenue, closed down, which would enhance efficiency in tax administration. 85 This should be implemented over the medium-term, in line with the reforms to strengthen tax administration. F. HOW SHOULDTHEINTRODUCTIONOFTHE MTEFBESEQUENCEDOVERTIME? BasicPrinciples 255. Experience shows that there is a need to avoid being too dogmatic about MTEFs; in practice they can take a variety of forms aimed at strengthening the link betweengovernment policies and budgetary allocationsover time. At the same time, building such a link can take significant time and the steps should be built organically. Stakeholders should bewareof overly ambitious expectations. Often, the beginnings of a framework can be put in place with the substance for the framework beingimprovedover time. 256. In terms of sequencing of the development of the MTEF process itself, experience inthe region4'indicates: 0 Focussing initially on the top-down (macro-fiscal framework and initial sector ceilings) rather than on the bottom-up process (the costing and content of detailedsector expenditure strategies) hastended to work most effectively. 0 In the design of the strategic budgeting process, simplicity in programmes works most effectively. It is importantthat information requested from sector ministries is not overly detailed, time-consumingor unlikely to be credible. In particular, three-year allocations for detailed activities are unhelpful when the first priority shouldbe to get broad resourceallocations more in line with sector strategies. 0 Once there is a basic strategic budgeting framework in place, Government should work to ensure that existing elements of the strategic budgetingprocess are made stronger before introducing new elements (such as performance indicators). 0 Finally, Governments should program the link with wider budget systems (particularly,Treasury systems) explicitly into the reformprogramme. 257. The essential prerequisite for meaningful sector work is that the Ministry of Finance has established its central budget discipliningrole and can control aggregate fiscal di~cipline.~~Moreover detailed sector analyses are wasteful of staff time if basic expenditure control does not exist and if reliable expenditureinformationis not available. 41See Appendix 3 of this chapter. 42This is not a hard-and-fastrule as sector expenditure analyses can help sector ministries to prioritize their intra-sectoral expenditures and thereby assist with sector reforms. In general, however, if the Ministry of Finance has not established fiscal discipline in relation to sector ministriesthen there is the risk of detailedand well-plannedexpenditureplans being undermined.atthe Ministry of Finance. 86 Phasesof MTEFDevelopment 258. Lessons from experiences of introducingMTEFs have highlighteda number of phases of relative development of MTEFs (see Table 4.2). The process of working through the phases is not necessarily linear, and movement across phases should be expectedto take considerabletime and sufficientinstitutionalcapacity development. 259. During the preparatory phase for introducing an MTEF, the basics of the legal and institutionalframework (e.g. ensuring that a modern Budget Systems Law43 i s in place, establishment of a Government MTEF Steering responsibility for coordinatingthe MTEF given to a department in the Ministry of Finance such as the Budget Department, etc.) have been established or are underway. In some countries, the legal and institutional framework is the foundation for the beginning of a two- stage strategic budgetingprocess. What is important in the legislationis reference to a two-stage process (rather than necessarily specifically introducingan MTEFper se) whereby there is a stage prior to the start of the annual budget process where Government reviews the fiscal framework. Whilst not having the legal and institutional framework in place does not necessarily prevent countries from beginningsuch a process, it appears to be an importantlegitimisingfirst step. 260. In the primary MTEF process, once the basic legal and institutional framework has been established, countries starting implementation of a two-stage strategic budgeting process begin with the development of a basic multi-year macro/fiscalframework, largely based around an externally-developed,e.g. financial programming,framework. At this stage, there is hited policy analysis of the overall resource framework, and there are no, or only limited, sector expenditurestrategies in place, such as an initial sector review, setting out the sector's policy objectives and classifying sectoral activities into broad programme areas to meet the objectives. Sector ceilings are basedon overall policy priorities and limited analyses of the main economic categories of expenditures. 261, In the next phase of development, an intermediate MTEF process, countries undertake more advanced analyses for the macro/fiscal framework and sector expenditurestrategies. The setting of overall expenditureceilings is based on some cross-sectoral analyses (overall government priorities e.g., poverty reduction, investment/recurrentratio etc.). Onthe sector side, sector ministries (perhaps initially at the pilot stage) have begun to analyse their existingactivities by broad programme and activity in order to prioritise existing activities and identify those which are no longer deemed to be priorities. The budget allocation mechanism has been changed to enable the budget to reflect more clearly government policy priorities by program/activity, as reflectedin sector expenditure strategies, in additionto the items of expenditure. Initial changes to the budget classificationto reflect sectoral and ministry programmes are beingput in place. 43 In Tajikistan this is the Law on State Finances. 44 While it is understood that the same agencies should be involved in both the strategic and detailed estimate preparation phases of an integratedMTEFhudget process (Le. that the MTEF process should preciselynot be the responsibility of separate groups or bodies), it is often useful to initiate the process of developing an MTEF by entrusting a steering body with the task of ensuring that this new way of doing business is harmoniously mainstreamed with existing processes and that those are adjusted as and when needbe. 87 262. When a country has a more advanced MTEF process, the macro/fiscal framework includes all resources, including external project finance, extra-budgetary finance and off-budget resources. Macro/fiscal projections are based on a more sophisticated forecasting model and on detailed policy analyses o f proposed revenue policy changes and macro feedbacks. Sector ceilings are based on detailed analyses o f cross-sectoral issues. With respect to sector expenditure strategies, initial or detailed costing o f existing activities (both recurrent and capital) have been carried out, areas o f cost savings and improvements to programme efficiency and operations have been identified, and broad expenditure implications o f the new or restructured programme activities have been calculated. In this level, new prioritized activities to achieve policy objectives have been identified, a clear mechanism has been established for prioritising against policy objectives, and a mechanism is in place for allocating resources based on cutting or phasing out non-priority activities and funding resources to the highest priority and most cost-efficient activities. 263. At the sector level, the output is a detailed, fully-costed and prioritized medium-term sector expenditure strategy, with medium-term expenditure plans which contain detailed expenditure (recurrent and investment) implications for programs/activities and which are consistent with sector ceilings. In the most advanced cases, performance measurement indicators and analyses o f sector performance are carried and sector budgets have been restructured to realize cost and efficiency savings. Finally, a mechanism to cost out and prioritize new proposals and integrate them into the MTEF has been established. At this level, changes to the budget classification and Chart o f Accounts are fully in place, and budgets are executed in accordance with programmes and activities linked to policy priorities. 264. Tajikistan is effectively still at the preparation stage although it has implemented one o f the main components o f the primary stage (the macro/fiscal projections). The main impediment to further progress in the primary stage and to progression onto the intermediate stage is the fragmentation o f the budge process discussed in Section 4.4, which prevents meaningful sector ceilings being used as budgetplanning tools. Developmentof MTEFswithin BroaderMeasuresto Strengthen PFM 265. As part o f the process for introducing a more strategic phase to the budget, Governments should pay attention to the wider budget systems on which budget plans depend. The MTEF is an integral part o f the budget/PFM cycle; weaknesses in any o f the links in the chain potentially undermine the ability o f the MTEF to be used to achieve Government policy objectives. 4sThere is a continuum with respect to including performance elements in the budget process, from merely introducing performance indicators alongside budgets, for sectors to monitor and report on in their sector expenditure strategies, to a process of linking budget allocations not to broad programmesprograms of activities but rather to specific outputs or performance targets. Countries at IevelLevel 3 in the development of an MTEF process are likely to start at the simpler end of the continuum. The more advanced along this continuum a country moves the greater are the analytical capacity requirements. 88 Firstly, ifthe broad strategic allocations in the MTEF are not translated into approved allocations in the appropriated annual budget, then the resources employed preparing a strategic budgetingphase are wasted. Secondly, well-made budget plans which are ignored or changed significantly during budget implementation undermine the credibility of the budget process and any strategic decisions that are made. More sophisticated planning of expenditures is futile if there are not also mechanisms to ensure that budget implementationis in line with plans, and if good information is not availableon budget performance. Specifically, ifthe supportingbudget systems (e.g., the budget classificationsystem) are unableto allocate resourcesto the priority policy areas in line with the strategic budget plans, then there could be a mismatchbetween the use of budgetary resources and the achievement of budget policies. If accounting and reportingsystems do not allow the recordingand monitoringof budget programs, then central agencies, Governments and Parliaments will be unable to monitor if budgetobjectivesare beingmet. Finally, interms of scrutinizingthe MTEFhudget, if the executive, the legislature, and civil society do not understand the budget in its strategic form, then it is more difficult to achieve an appropriate match between budget policiesand resources. I Table4.2: Broad phasesof the development of an MTEFprocess Pre-MTEF Primary Intermediate Advanced MTEF MTEFprocess MTEFDrocess Drocess Macro-jscal Based on Simple Inclusiveof all pamework Establishment simple (e.g., domestic resources; more (top-down) and/or financial model; stronger sophisticated model; strengthening programming) macro/fiscal alternativescenarios of basic legal framework; analysis; may informdecision- and focus on inclusionof making institutional domestic externalfinance parameters, revenues Setting of including Overall policy Some analysis Deepedmore sector ensuring that priorities, of inter-sectoral comprehensiveanalysis ceilings (top- a modern littleho analysis prioritiesincl. of inter-sectoral down) Organic ofthese basic policy priorities Budget Law is prioritiesor of prioritiesand in place. cross-sectoral cross-cutting issues issues Sector Limited sector Intra-sectoral Costingof activities expenditure expenditure broadly and stricter strategies strategies prioritised prioritization (bottom-up) strategic/ mechanisms inform programme intra-sectoralstrategic framework in expenditure I place framework. 266. Governments frequently introducestrategic elements to the budget as part of a wider PFM reform program. The decision of when best to introduce a strategic budgetingprocess within such a reform program is an important one. In practice, as 89 discussed above, when best to undertakethe development of an MTEF depends onthe basic state of public financial management. Ifa PEFA46or other diagnosticanalysis of PFM indicatesa large number of low scores, then the basic requirementsof a PFM system are unlikely to be inplace. G. STRATEGYFORPHASEDIMPLEMENTATION OFTHE MTEF 267. Restructuringbudget relations along the lines discussed in Section 3 will be difficult and involve risks. Although the existing budget process is far from optimal as a process for allocating resources in line with policy objectives, it does actually allow a budget to be prepared and implemented: at least as a bureaucratic exercise it is effective. Even this could be threatened by an ill planned restructuring of the budget process, which could be highly disruptive to the functioning of Government. To minimize the risks, it will be necessary to move gradually, with the MTEF introduced in phases over many years. The strategy recommended here involves a twin track approach. The existing budget process will remain in place but reforms will gradually be introduced in the way in which the MOF handles the centralized functions of the budget process, while simultaneously sector budgeting will be introduced in a pilot sector. The next step will then be to bring more sectors of the budget intothe MTEF, so that the existingbudget processes are gradually replacedby new ones. 268. The full introduction of the MTEF will take many years, and will need to be accompanied by complementary reforms to public financial management, such as strengthening of treasury systems, auditing, monitoring and evaluation and payroll management. One of the important findings o f the World Bank study on the implications o f weak PEM capacity for the PRSP approach, is that PEM reform should be undertaken in a holistic manner: improving one link in the PEM system will not generate better budget outcomes if other links in the system are unreformed and remainweak (Andrews and Moon, 2003). The Government will need assistance to implement the MTEF, and complementary PFM reforms, from donors; in particular, it will need technical assistance for the MOF and the line ministries in the pilot sectors. Because the budget systems in CIS countries are markedly different from those in most other developing countries, the former face unique challenges in implementing PEM reforms. It will, therefore, be important that TA provided by donors can draw on the expertise gained in implementingPEM reforms elsewhere in the CIS region. 269. In terms o f the sequencing discussed in Section 4.6, the three phases set out below are intended to take Tajikistan from the preparatory stage (with some components ofthe primary stage) through into the intermediarystage. Preparatory Phase 270. The earliest that implementationof the MTEF can start will be the 2007 budget process, which will produce budget estimates for 2008 and budget projections for the following two years. Before the 2007 budget process begins, several preparatory 46See PEFA Secretariat, Public Financial Management: Performance Measurement Framework, June 2005 90 measures will have to be taken. These are: (i)obtaining political approval for the introduction of the MTEF through a Government Re~olution;~'(ii)reviewing all relevant legislation (e.g., the Law on State Finances) to determine if any revisions to legislationare needed; (iii) select one (or more) pilot sectors; and (iv) draw up a new budget timetable with an earlier start date to allow for a longer budget process. 271. The best candidate for a pilot sector is the education sector for several reasons. It is already benefiting from TA designed to strengthen budgeting allocations in the sector and, under the Education for All Initiative, the sector will receive grants, a condition of which is the preparation of a sector strategy. T he preparatory phase should prepare the Ministry of Education(MOE) for the expanded role it will play in the budget process. A study shouldbe carried out to assess the best way to restructure the budget process within the education sector so that the MOE is given the responsibility of coordinating all budget requests from the KBOs in its sector and preparinga single consolidatedsubmissionfor the MOF. A chart shouldbe prepared by the MOE showing all of the KBOs in the education sector and the expenditures which they control. The MOE should draw up budget instructions to be sent to all KBOs in the education sector, which request details of budget requests and give guidelines on the form and magnitudethat these should take. Internalrestructuringof the MOE will also be necessary so that more staff is allocated to working on the budget process. Phase 1 272. Phase 1 should start with the beginning of the 2007 budget process, which shouldbe no later than March 2007, and last for two years. 273. The key tasks for the MOF will be to prepare a budget resource envelope at the start of the budget process which can provide the basis for allocating sector ceilings. The European Union (EU) is planning to provide technical assistance in the area of macroeconomic and tax revenue forecasting which will assist the MOF to prepare the budget resource envelope. Other tasks which should be implementedby the MOF in phase 1 are the setting up o f a process for allocating the aggregate budget resource envelope into sector ceilings, involving drawing up criteria for making these allocations (which in principle should be linked to policy objectives inthe PRSP) and an intra-governmentalprocess for reachinga political consensus on the allocations, so that they are not merely a technicalexercise carriedout by the MOF. 274. In the education sector, the minimum requirement of phase 1 will be to establishfunctioning relationshipsbetweenthe MOE and all other KBOs in the sector to allow the education sector budget submission to be prepared on schedule. The MOE should also begin to prepare a medium-term expenditure plan for the sector, which can guide the sector's budget allocations. This will requirecapacity building in the MOE and probably long term TA. n education sector working group should also be established, comprisingkey stakeholders amongthe KBOs inthe sector. 275. Appendix 2 sets out the main actions requiredto implementthe MTEF in both the MOF and the pilot sectors in the preparatoryphaseand phase 1. 47The Government Resolutionendorsingthe MTEF was passed in October 2006. 91 Phase2 276. Phase 2 should begin inthe first quarter of 2009. The key tasks of phase 2 will be fourfold. 277. First, the MOF should consolidate the processfor preparingits budget resource envelope and the allocation of this into sector ceilings. In particular,a formal process for agreeing the sector ceilings within government should be established: One option would be to usethe NationalBudget Commission. 278. Secondly, the education sector, havingrestructured the bureaucratic links in the budget process to channel budget submissions through the MOE, should then focus on improvingthe quality of budget allocations by applying technicalexpertise in the MOE to evaluate expenditure requests, cost new proposals and assess the links between expenditures, inputs, outputs and outcomes in the sector. Obviously these qualitative improvements in the budget process will take time to be realized. The SWG should also be strengthened so that it can play an effective role in the budget process. 279. Thirdly, responsibility for determining all education related capital expenditures in the CSIP should be transferred to the MOE from the MOET. The education sector expenditure ceiling will, therefore, cover both current and capital expenditures. 280. Fourthly, two other sectors should begin implementing sector budgeting in phase 2, undertaking similar reforms to those carried out in the education sector in phase 1. Healthwould be a good candidate for a secondpilot sector, not least because it is so critical for achieving the MDGs. The social protection sector can also be includedinthis phase. The restructuringof the budgetprocess inthe social protection sector should be more straightforwardthan in other social sectors because the local governments play only a minor role inthis sector. Phase3 281. The third phase will probably begin in 2011 and is likely to last for more than one year, as it will involve bringingthe rest o f the budget into the MTEF. The MOF should prepare new budget instructions for all LLMs and the KBOs which remain outside of sectors. These instructions should include the three year expenditure ceilings. All of the remaining sectors should be brought into the MTEF, drawing on the experiences and lessons learned by the pilot sectors to reorganize the budget process within the sectors. All sectors will take over responsibilityfor the relevant capitalprojects under the CSIP. 282. Tax collections should be centralizedin phase 3, with all collectionsother than those taxes specifically earmarked for local authority funding being submitted to the central treasury. Local authorities will, therefore, receive their funds in the form of transfers from the centraltreasury, instead o f retainingtaxes when they are collected. 283. The final step inthe implementationof the MTEF will be the integrationof the PIP projects intothe sector ceilings of each sector. 92 H. CONCLUSION 284. The existing budget system, characterized by a very high level of fragmentation among over one hundred KBOs and a very limited role played by line ministries is a major constraint to the introduction of a sector-based MTEF in Tajikistan. Reforming the budget system is, therefore, a prerequisite for implementationof a meaningful MTEF. The key reforms involve restructuringthe budget process into sectors, so that a sector expenditure ceiling can be allocated in a coherent manner among competingexpenditure demands in the sector, accordingto the sector's expenditure priorities. Ineach sector, a lead line ministry should be put in charge of coordinatingthe budget process inthat sector. 285. It is necessaryto be realisticabout the pace at which the reformsrecommended here can be implemented. The institutional reforms to the budget system will be difficult to implement and take considerable time; hence they should be introduced cautiously, in a phased manner, to minimize the risks of severely disrupting the budget process. The creation of functioning sectors, with sector wide expenditure ceilings and budget allocations, should be implemented first on a pilot basis beginning in the education sector. The pilot phase is likely to last two years before other sectors can be brought into the MTEF. The MTEF will place much greater responsibilities on line ministries to coordinate and lead budget planning in their respective sectors, but capacities for budget planningin line ministriesare very weak, and so strengtheningthese capacitiesthroughtraining and technicalassistancewill be imperative if the MTEF is to succeed. Consequently, substantial financial and technicalassistance from donors will be essentialto support the implementationof the MTEF. 93 APPENDIX 1 MEDIUM-TERMBUDGETFRAMEWORK Table 1 Budgetfor 2006 and Forecast for 2007 and 2008; thousands of Somoni tevenue and Grants 2006 ievenues and Grants to state budget 150000C If which VAT repaid -1oooc ievenues and Grants 151000C tax revenues 1383736 income individuals 9027E 110279 123512 profit tax from corporations 425% 45410 50859 minimum tax on enterprise profits 22847 23192 25975 social taxes 17200C 197000 225000 tax on real estate 14972 17217 19283 land tax 8312 8338 9340 single tax from agric producers 66203 66865 71546 sales tax inc 58837 131125 146860 cotton fibre 58112 74125 83020 aluminum 725 57000 63840 tax on reta.il sales 30134 33748 37798 VAT 505532 552134 618390 domestic 124990 145210 162635 external 380542 406924 455755 road user tax 47173 47941 53694 simplified tax 5835 8158 9137 excise tax 81901 91105 104281 domestic 11331 13886 15840 external 70570 77219 88441 customs duties 108121 132768 147364 patent fee 12072 15202 17026 vehicle owner tax 8700 10596 11867 bonus and royalty 6262 7189 8051 state duty 16468 17621 19736 processed product tax 85531 ion tax revenues 88264 95606 114311 grants 38000 40000 40000 ) w e : Ministry of Finance 94 Table 1(continued) Medium Term Budget Framework: Budget for 2006 and Forecast for 2007 and 2008; thousands of Somoni ixpenditures 2006 200; 2001 ,tal expenditures 154200C 1696991 190453( Iublic admin bodies & foreign econ relations 184656 20321; 22806: -aw enforcement structures 23871E 262706 29483 Social Spheretotal 76311E 84043: 94453( education 33607C 37035( 41608' health 106482 11730: 13251; social protection 19165C 210911 236701 compensations 3724C 4098: 4599! culture and sports 57411 6318: 7090s other social services 34264 37701 4232( interest payments 5108E 5510; 5770: externaI 3508E 3749! 3794; domestic 1600C 17601 1976; rota1economic services 23382 25782: 29219( agriculture and agro-indistrial complex 49691 5493! 6237~ transport and communications 74842 8236f 9282: mining and construction 1153E 1269f 14241 fuel and energy 25575 28296 32581 utilities 67246 741O! 8405f other economic services 4935 543' 609! ither expenses 7060C 7769f 87191 `IP spending from external sources 33600C 36400( 40400( `otal expenditures including external PIP 187800C 2060991 230853( ludget Deficit and Financing ludget Deficit excluding PIP -4200C -4550' -5050' nancing 4200C lomestic 150151 privatisation receipts changes in govt deposits proceeds from securities domestic debt repayment ixternal 22784s Foreign debt repayment -108151 -10160f -11501( External financing to PIP 33600C 36400( 40400( leficit including PIP -37800C -40950' -45450` iurce: Ministry of Finance 95 APPENDIX2: ACTION PLANOF STEP TO IMPLEMENT MTEFINPILOT SECTOR 2006-2007 Time Period Ministry of Finance Lead Line Ministry for Pilot Sector (MOE) October 2006 Designateeducation sector as pilot sector and Ministry of Education(MOE) as Lead Line Ministry for that sector October 2006 Send out budget instructionsfor implementingMTEF in education sector November- Draw up new budget timetable December for 2007, with earlier start date 2006 for budgetprocess November- Meet with MOEto discuss Meetwith MOFto discuss plans December plans for implementingMTEF - for implementingMTEF in 2007 2006 in2007 November- Meet with MOE to discuss TA Meet with MOFto discuss TA December requirements from donors requirements from donors: make 2006 requeststo donors for TA December Assist MOEwith budgetdata Draw up list of all KBOs in 2006 on KBOs educationsector and obtain from MOF budgetestimates for 2005 and 2006 for eachof these KBOs December Organise trainingworkshops for Attendtraining workshops 2006 -January senior officials and budget 2007 officials in MOE and selected KBOs in education sectors January 2007 Strengthen budget departments by re-allocatingstaffto them January 2007 Write to all KBOs in education sector informingthem of new budget proceduresfor implementingMTEF January - Work with MOEto draft new Work with MOFto draft budget February 2007 budget instructionsto be sent by instructionsto be sent by MOEto MOEto all KBOs ineducation all KBOs in education sector sector January - Makeprojectionsof budget March2007 resource envelope for State Budgetusingforecasts of revenues, grants, debt service, borrowing,privatisation receipts January-March Consultwith NationalBudget 2007 Committeeon criteria for allocating budget resource 96 envelope among sectors February- Meet withjoint PublicFinance March2007 Managementdonor missionto discuss and agree mediumterm assistance for MTEF and PFM reforms March2007 Determinesector ceilings for education sector: send sector ceilingsto MOE April 2007 Advise MOE on dividing Divide sector ceilingfrom MOF consolidated sector ceilings into into ceilings for each KBO, based ceilingfor eachKBO on projectionsfor protected items, sector policies,etc. April 2007 Sendbudget instructions, includingceilingfor each KBO, to each KBO in sector May-June Meetwith eachKBO insector to 2007 discuss budget allocation July 2007 Receivebudget requests from KBOs in sector; review each requestand revise ifnecessary August 2007 Submit consolidatedsector budget submissionto MOF August 2007 Meetwith MOEto finalise Meetwith MOFto finalise sector sector budget submission budget submission September Amalgamate consolidated 2007 budget estimates from education sector with budget estimates for all non pilot sector KBOs October - Begindraftingmediumterm December sector expenditure plans 2007 November Review implementationof 2007 Review implementationof 2007 2007 budget process in pilot sector budget process in pilot sectors with MOE; identifyproblems with MOF; identifyproblemsand and necessary adjustments for necessary adjustments for 2008 2008 budget process; identify budget process, identifywhere where more training, TA is more training, TA is required required November - Trainingof senior officials and Attend trainingorganized by December budget officers in MOF, MOE, MOF 2007 LLMsfor future MTEF sectors and selected KBOs 97 APPENDIX 3: REGIONALLESSONSFROM THE DEVELOPMENTOF MTEFS Experience of developingMTEFs inother CIS countries indicatesthat, whilst there are a number of common lessons, there is no standardized methodologyor blueprint for introducinggreater strategic elements into the budget process. The development of an MTEFprocess needsto be adaptedto suit the particularcircumstances of each country, includingthe extent to which fiscal discipline hasbeenestablished, the relative strength of the role of the MOF inthe budget process, relativecapacities in the MOFcomparedwith sector ministries, andthe degree ofpolitical commitmentby the Governmentandby MOF. The introductionof MTEFs in the CIS region has led to a number of achievements, including providing a tool for the Government to maintain fiscal stability and discipline through publicizingand committingitself to the overall budget framework. In addition, MTEFs have provided GovernmentParliament, sector ministries, and sub-national governments with a more comprehensive source of analytical informationon the budget (i-e., trends and key issues); a common problem in many countries is the lack of basic informationand analysis on where funds are beingspent. This is particularly important when centralagencies do not have influence, or little or indirect, over the budgets of sub-nationalentities. MTEFs have provideda forum for communicatingthese budget parametersto budget stakeholders. In some MTEFs, budgetary allocationshave begun to enable resources to be moved towards higher priority sectors and to achieve a better balance between economic items (e.g., reform of pay structures and greater allocations for operations and maintenance). Inmore advanced MTEFs, expenditure analyses at programlevel have focussed on general strategic directions for a relatively small number of basic programmes, in some cases accompanied by the introduction of a programme-level classification. Some MTEF experiences have had success in generating a more active dialogue with the MOF over funding levels and justification for activities to be funded. In general, the broad lessons from the MTEFs can be linked to issues of timing and sequencing and institutionalcapacities. In particular, more established MTEFs have tended to develop more slowly and organically, illustrating sequencing and timing issues as discussed in the section below. n other words, these processes have not attempted to go for a "big-bang" approach to MTEF introduction but instead are working through the different stages of development (see separate section on sequencingMTEFs). Interms of specific lessons, a number of common themes have emerged amongst the countries inthe regionwhich have introducedMTEF processes. Begin with the macro/fiscalframework (top-downprocess) In terms o f the appropriate timetable and specification for MTEF development, experience indicates that relatively greater progress has tended to be made in establishingand strengtheningthe macro/fiscalframework, followed by the rest of the top-down process. Slower progress has been made in incorporating sector strategy 98 exercises into the process and getting concrete changes in budget practices leadingto resources being spent (not just allocated) according to strategic priorities. This reflectstwo aspects of transitional countriescompared to other regions: (i)the weaker institutional role of MOF; and (ii)the relatively stronger capacities of staff and staff with more relevant experience in MOF than in sector ministries. First, a strong, directional role for the MOF in the budget process, common in other regions, is less prevalent in transitional countries. This reflects the often unchanged management role and capacities of MOFs. In most cases, the structures and responsibilitiesof boththe MOF and finance departments in the sector ministrieshave not changed significantly since independence, although their roles have grown in importance as constraints on resources have increased. In particular, the role of the MOF in facilitating policy co-ordination though the budgetary process is not yet widely understood. This can underminethe necessary fiscal management role of the MOF and its overall co-ordination and management of the budget process as a whole (in particular, articulating and defending budgetary trade-offs), which is central to a well-functioning MTEF process. Thus, the first step in developing an MTEF in the region is to build up the macro/fiscal framework in order to set the overall resource constraintsfor sectoralexpendituredecisions. Second, progress in developing sector expenditure strategies and sub-sectoralpolicy priorities within constrained resources in the region has been hampered by relatively weaker analytical capacities found in sector ministries. Many budgedfinance departments in sector ministries continue to be small, and budgetary analytical skills tend to be quite limited; these departments tend to be staffed mainly by accountants, reflecting their previous role under central planning. In many MOFs, by contrast, there is evidence of the employment of new staff with relatively greater analytical skills. Importance of ensuring that the basics are in placejirst An MTEF process is not a panacea for PFM weaknesses. Lessons from experience indicatethat a numberof PFM basics needto be in place to undertake successfullythe development of an MTEF. First, achieving and sustaining aggregate fiscal discipline i s critical. As experience in the region suggests, the MTEF can assist both with the initial achievement of aggregate fiscal disciplineand with sustainingit. However, it is easier to begin such a process if the Government has already become accustomed to the discipline imposedby ensuringa realistic macro-fiscalframework. Secondly, the introduction of MTEFs can be undermined by the lack of a credible budget process at the outset. Above all, a basic budget system needs to be in place. If there is not a credible annual budget process, including basic budget discipline (e.g., in immediate post-conflict countries), then there will be little gained from trying to introducea strategic phase to such a weak or nonexistent budget process, and it could well prove to be a distraction from building such a process. Other PFM reforms which assist in the implementation of the MTEF include: (i) having in place a modern budget classification system, such as the GFS, so that expenditure data can be analysed by functiodsector and by economic item; (ii)the promulgationof a revised Organic Budget Law and extended budget calendar to give sufficient time for the analysis, preparationand considerationo f MTEF and budgetary 99 parameters; and (iii)strengthening expenditure control systems so that budgets may be more easily executedas planned. Implementingan MTEF can provide the impetus to the MOF to implement a wider package of PFM reforms, including the necessary institutional capacity building, in order to makethe MTEFmore successful. Extent of Political Engagement It is often assumed, particularlywith respect to other regions, that sufficient political engagement is a necessary pre-conditionfor introducinga successful MTEF process since the MTEF is about ensuring that resource allocations reflect a government's priorities. In the region, there have been positive MTEF experiences where explicit political commitmenthas been strong at the beginningof the process, as well as good exampleswhere the processhas helpedto build political support. On the one hand, what is often lacking in the process of introducinga more strategic budget process is strong political guidance and leadership on relative policy priorities. In the absence of such guidance, it is difficult for the MOF and other central ministries to make inter-sectoralresource allocation decisions. A t the same time, without strong Government engagement in the process, it is not clear that there is commitment to the medium-termbudgetary allocations, thus potentiallyundermining the credibility of the MTEF process (e.g., Cabinet decisions on budget allocations may not be inline with MTEF ceilings). On the other hand, in the CIS, governments' motivation has often been generated subsequently by demand-pull factors by senior technicians, whereby senior budget personnel, usually in the MOF, find that the MTEF facilitates their ability to determine the overall resource framework (in terms o f setting a hard budget constraint) to guide sector ministries in budget preparation, evaluate sector ministry budget proposals against this framework, and provide analyses and justification to Government officials on budget parameters, all of which in turn can create the demand for more analyticalbudgetary information. In addition, at least in theory, the MTEF should enable government officials and MOF staff to have a more informed dialogue with the International Financial Institutions (IFIs), although it is not clear how muchthis has improvedin practicein the region. Some MTEFs in the region have achieved considerable success in involving governments in the key decision-makingpoints, such as approvalof the initial macro- fiscal framework and the overall framework and Budget Framework Paper/MTEF document. This review is often facilitated by the establishment of MTEF Steering Committees to oversee the process. One also sees evidence of a growing interest by Parliaments in some countries to review the MTEF; strengthened Government and Parliamentary scrutiny of the overall MTEF framework and resource movements (ceilings) in relation to governmentPRSP policies will be important. Encouraging wider public interestin the strategic budget process is a longer-termpriority. External pressure has led to the introduction of some MTEFs in the region. The ability of donors to facilitate internal demand for an MTEF depends on the Government's initial willingness to reform, which in turn often depends on there beingprior experience of a productiveGovernment-donordialogue. 100 Extentof Analytical and Other InstitutionalCapacities As indicated above, institutional weaknesses faced by countries introducing more strategic elements in the budget process include weak analytical and policy-making capacities, exacerbated by existing management structures which do not reflect the greater need for these capacities. In many cases, the structures and responsibilitiesof both the MOF and finance departments in the sector ministries have not changed significantlysince the advent of strategic budgeting, althoughthese departments' roles have grown in importance. In particular, the greater policy analysis role of MOFs is often not evident inthe departmental structure, staffing, andjob specifications. Similarly, the analysis of investmentand recurrent expenditures often continues to be the responsibility of separate departments within a ministry (or even separate ministries); sometimes due to institutional weaknesses, the MTEF does not always succeed in integrating the processes for allocating recurrent and development expenditures. Effective strategic budgets which aim to link policies to an integrated budget require some co-ordination arrangements (e.g., an Inter-Ministerial Budget Guideline Committee). The lack, or poor specification, of such arrangements can be problematic. Similar issues arise in line agencies with regard to policy development/analysis, planningand strategic budgeting. Significant difficulties have occurred in processeswhich have been overly demanding of analyses before sufficient analytical capacities are in place to support these requirements, e.g., for policy development or expenditure analysis. This can be wasteful of staff time and can distract time and attention away from more feasible strategic prioritisationprocesses which would be informedby simpleranalyses. Finally, there have been difficulties in aligning programmes with budget classifications due to existing classifications being based on organizational (institutional) lines. This has prevented resources from being allocated by programmes during implementation and monitoring information on expenditures programmesto be collected. Existenceof SupportingPFMReforms The lack of accompanying and supporting reforms in the wider budget cycle, particularly expenditure execution, accounting, reporting and control processes, can undermine improvements in budget planning. Thus, planned budget allocations which reflectsectoral priorities may not actually be spent accordingto those priorities. Without accompanying changes in budget systems and procedures, particularly the use of normative techniques in sectoral resource allocations, it is very difficult to restructure sectoral budgets. The use of norms tends to increase inefficiencies in sectoral allocations through: (i) removing from programme managers the responsibilityfor introducinginnovationsthat allow for improvements in productivity and in the quality of public service provision; (ii) providing incentivesto managersto increase input use as a means of maximising their budget allocations; and (iii) reducingbudget analysis in sector ministriesto a process of checking whether norms have been appliedcorrectly rather than challengingthe assumptions on which budgets are basedand consideringalternatives. 101 LegaVOrganizationalFramework for MTEF In CIS countries, greater emphasis than in other areas is placed on the legal framework for budgeting. Revising the Organic Budget Law to incorporate the explicit requirement to prepare an MTEF and incorporating it in the Budget calendar has beena critical factor in supportingthe MTEF in countriesin the region. However, this is not strictly necessary, as the modem budget systems laws in many CIS countriesare sufficient to cover the introductionof an MTEF. Timing for MTEFPreparation The lessons from the MTEFs which have been carried out in the regionunderlinethe importance of beginning the process early and devoting sufficient time to it. In practice, in CIS countries, this is likely to mean beginningin January or February, for submission of the BFP/MTEF document to Government in June, prior to the beginningof the Annual Budget process, with the issuingof budget guidelines.Thus, the first half of the year would be devoted by the MOF and sector ministries, if involved, to preparinga Budget Framework Paper/MTEFdocument. This would give sufficient time for an initial macro-fiscalframework to be submitted to Government, or a governmental Budget Committee, for initial approval before the framework is finalised and included in the Annual Budget Guidelines, which are often issued in May or June. This would facilitate a logical process from strategic budget phase to detailed budget allocations, including involvement of Governments at key decision- making points (Le,, the macro/fiscal framework in MarcWApril and the detailed resourceframework includingsector/ministry ceilings). Designof an MTEF Inthe designof an MTEF, it is importantthat expectations be realistic and achievable. A key indicator o f initial success is ifthe Government begins to request explicitly the analyses included in the MTEF. Generating this interest can be facilitated by the requirement for approvalby the Government of MTEF parameters. At the same time, in order for the MTEFs to be sustainable, the issue of institutionalcapacities must be addressed, and inthe meantime, governments and MOFs should adopt an appropriate sequence for MTEF preparation. It is also importantthat the ceilings are determined at an appropriate level of detail. In MTEFs in other regions, budget and forward estimates are shown by programme. With some exceptions, most MTEFs in the region show their estimates by functiodsectorand economic classificationor by functiodsector,ministry/agencyand economic classification. In the sector expenditure strategies undertaken by some countriesin the region, an attempt has beenmade to analyse the sectors' expenditures by major programme. Changes in the classification system will be required to introduceprogrammaticcategories, althoughin some cases these may follow the sub- functional classification. The programbudgets introducedin some countries in the regionprovide a framework for bottom-upanalysis of sector expenditures. These exercises ask sector ministriesto set out their mission statement or "vision", and request narrative justification for expenditure requests, which tend to be presentedby sub-function. Sector expenditure analyses, prepared as part o f the MTEF, link this type of bottom-upanalysis with the 102 top-down policy and strategic framework. The main conceptual difference between programmebudgetsand sector expenditure strategies is in the levelof analytical detail required. In general, analysis of expenditures by sub-functions (as in the program budgets) can be very detailed, making policy and expenditure trade-offs between programmes difficult. Whilst the MTEF is part of the budget process, the initial focus of many MTEFs has been on producing an MTEF document, setting out the medium-term macro/fiscal strategy and the resource framework, including the sectoraland/or ministry ceilings. This document can help establish the MTEF as a key part of the budget process and providestransparency of future budgetary parameters and thereby can promote fiscal discipline. As the MTEF process evolves, experience shows that the importance of reviewing these documents to ensure that the document does not become overly complicatedand the preparationof the document become an end in itself, rather than part of the underlyingbudget planningprocess. 103 Exampleof Annual Budget Calendar in Countrywith a January-DecemberFiscalYear Date Task JanuaryIFebruary Preparationof MTEF methodologicalguidelines Februaryhlarch Preparationof initial macro/fiscalframework for MTEF periodto Government April Presentationof initial macro/fiscalframework for MTEF periodto Government Preparationof MTEF budget paper by MOFand presentationto Government Determinationof ministerialbudgetceilings by Government 1 June Issuanceof budget preparationinstructions,budget forms and indicative budgetceilings June/July Ministriesdetermine spending prioritiesand prepare budget requests. Submissionto MOF by mid-July. August MOF to scrutinisesubmissions and to resolve minor discrepancieswith spending ministries August/September Meetingswith ministriesat MOF. Early September Preparationof economic reportand revisionof MTEF resource envelope. Collationand summarisationof budget data. 15 September Submission of draft Budget to Government 15 October Submission of draft Budget to BudgetaryCommitteeof Parliament October Preparationof Budget Address and updateof Economic Reportin line with the country's EconomicDevelopment Policy document 1 November Submission of Budget to Parliament 1 December Approval of Annual Budget Law by Parliament 1 January Issuanceof authoritiesto spend 1 January Distributionof Budget March, June, Quarterlyreports by lineministriesto MoF on budget September, execution December March/April Annual report by Governmentto Parliamenton execution of previousyear's State Budget 104 5. PUBLIC EXPENDITUREAND FINANCIAL ACCOUNTABILITY (PEFA) ASSESSMENT A. INTRODUCTION 286. This chapter provides a summary of the PEFA Assessment of Tajikistan's PerformanceReportundertakenbetweenOctober 2006 and April 2007.48The purpose of the PEFA assessment is to provide the GOTwith an objective assessment of the country's PFM systems. It aims to support a better understanding of the overall fiduciary environment of the budget and assist in identifyingthose parts of the PFM systems most in need of reform. As such it should contribute to a common understanding among the GOT and the donors wishing to support further PFM reforms. 287. The PEFA study team was composed o f three national and two international consultant^.^^ The main GOTpartner for the PEFA Assessment was the MOF. The MOF appointed a Working Group which coordinatedthe Government's participation and input into the PFM Assessment. Meetings were held with a majority of government institutions involved in the PEFA assessment (MOF, MOET, SFCC, MOR, the parliamentary budget committee, Aid Coordination Unit), and with key PEFA partners(IMF, DFID, EU). Subsequently,the MOF prepareda self-assessment report, which was a key component ofthe assessment process. Methodology 288. The PEFA methodology is set out in the Public Financial Management Performance Measurement Framework (available at www.~efa.orq).~~ It is based on 28 indicators covering a country's PFM system, and 3 indicators addressing the interaction of donors with a country's budget process and PFM system. The PEFA methodology has been in use since 2005 and has been applied to more than 60 c~untries.~' 289. PEFA Assessments provide cross-country comparable indications of the effectiveness of PFM systems, and of their improvements over time. They do not provide, however, an analysis of the causes of existingweaknesses, nor an indication o f the PFM system's ability to deliver development objectives, e.g. poverty reduction outcomes. 48 The Main Reportwill be publishedseparately. 49 The Swiss State Secretariat for EconomicAffairs, SECO, was the main sponsoringpartner of the assessment. 50 The PEFA assessment framework is also available in Russianthrough this website. 5 1Further information, including the methodology in Russian, can be found at: www.pefa.org. 105 Each indicator is scored on a scale from A to D. The basis for these ratings are the minimumrequirements set out inthe methodology. 290. The main sources of information used for this PEFA assessment were: (a) official GOTreports and data; (b) external evaluations and reports (WB, IMF, and others); and (c) numerous semi-structuredinterviews with key users and providers of PFM information and other stakeholders (Government representatives; donor representatives;membersof parliament; representativesof selected non-governmental organizations).The self-assessment report playedan importantrole interms of taking stock of current institutionalfeatures, practices, and laws concerningpublic financial management in Tajikistan; and provided a key source on which the Performance Reportdraws. B. INTEGRATEDASSESSMENT OF PFMPERFORMANCE 291. The Public Expenditureand FinancialAccountability PerformanceAssessment Resultsare summarized in Table 1, Credibility of the Budget (PI-1 to PI-4) 292. Inthe period2003 to 2006, the GOTbased its budget planningon conservative revenue forecasts. A s reflectedin indicator PI-3, this is generally positive, but it has led to a situation in which the budget outturn has differed considerably from budget plans; i.e., actual revenues have exceeded budgeted revenues and this has enabled actual expenditures to exceed budgeted expenditures. Although this discrepancy has declined, it is still significant. Furthermore, the actual composition of expenditure differs from the composition in the budget plan. Given the tendency for actual revenues to exceed budgeted revenues it is important to increase the transparency of the processes for allocating additional expenditures; including a more consistent involvement of the national parliament. This may also require more precise legal rules. Expendituresshould be planned and discussed during the budget formulation process, restricting the use of supplementary budgets within-year to unexpected emergency requirements. Payment arrears have been declining and are currently at low levels;, but the informationthat is recorded does not provide a clear distinction between paymentsdue and arrears. Comprehensiveness and Transparency(PI-5 to PI-10) 293. The Government has undertaken initial steps to improve the comprehensiveness and transparency of the budget; but further improvements are needed. The GOThas introducedfunctional and economic budget classificationsthat are generally in line with the 1986 GFS standard. However, an administrative classification is still missing; and until 2006, PIP resources were not included in the functional breakdown. The Social ProtectionFund was integrated more closely with the budget in late 2006. Sub-nationalgovernments are covered by the treasury system and central government monitors their fiscal situation. Reporting, monitoring and oversight of the fiscal risks from State Owned Enterprises (SOEs) is weak; this is a major concern with regardto large enterprises and in the context of still sizable quasi- fiscal activities. The set of information included in the annual budget submitted for parliamentaryscrutiny is relativelycomprehensive, but lacks analyticalmaterialon 106 Table 1: Overview of Assessment Results Indicator IIIssue covered Rating A. PFMOut-Turns:Budget Credibility II I project and program aid D3 IIProportion of aid that is managedby use of national procedures ] D 107 new policy initiatives and their budgetary implications. Parliament is also not informed about tax exemptions which reduce the overall amount of budgetary funds. There is some public access to fiscal information, but the information providedtends to be partialand incomplete. Accessibility of informationshould be improved further. Policy-based Budgeting (PI-11 and -12) 294. Some steps towards establishingmore policy-basedbudgetinghave been made and a renewed effort at establishing an MTEF was begun in 2006. The annual budget process is orderly, in terms of following key dates set out in the budget calendar. However, the institutionalstructures and systems for linking policies and budgets are still weak. Cabinet-leveldiscussions of the budget and of key priorities and trade-offs to be reflected in fiscal planning are not well developed. A cabinet-level Budget Commissionwas established in 2004, but its impact has remained marginal.An initial set of sector strategies has been elaborated; but thus far, only the education sector strategy is at a more advanced stage. Linking sector policy goals to realistic costing estimates will requirefurther work for a number of years. 295. Capital budgeting is fragmented, and needs to be further integrated with recurrentbudgets. The high share of capital spending in total expendituresmeansthat discretionary resources are a substantial share of the budget, which underlines the need for policy-basedbudgeting. Since a large part of capital spending is financed by development partners, it also underlinesthe need for a more effective approach to aid management on the side of the GOT,and for the harmonization and alignment with national priorities on the side of donors. Some steps in this direction have been undertaken in 2007 in terms of budget presentation. However, institutionally, capital budget planning remains separate and sub-divided into domestically financed and externally financed `streams', and aid policy and aid management structures continue to require greater attention. Predictabilityand Controlin Budget Execution (PI-13 to PI-21) 296. A basic degree of predictability and control in aggregate budget executionhas been achieved, but there are still systemic weaknesses. A core Treasury system has been established; and the Government plans to extend autoimmunization to local levels. Cash management has worked in a situation of continuously higher than budgeted revenues. Debt data recording and reporting has been improved; but debt sustainabilityanalyses still needto includedomestic debt. 297. Important weaknesses in budget execution are the following: (a) effectiveness in tax collection is a concern, with significant levels of tax arrears especially from large enterprises, (b) public payroll controls are currently weak, (c) procurement reform has begun, but while legal reform has been undertaken, the institutionalization of the new system is still missing, and (d) the internal audit function is very weak. Furthermore, as indicatedabove, the process of allocating additionalresources during budget execution is not sufficiently clear and transparent. This means that that there is a situation of frequent ad hoc bidding for additional funds from various sources (presidential fund, excess revenue) and of generating own resources by budget organizations which have not been sufficiently taken into account at the budget planning stage. Expenditures budgeted and organized at short notice during the 108 process of budget execution are unlikely to be employed most effectively. Furthermore, further improvements in cash-flow forecastingand management will be neededas the discrepancy betweenplanned and actual revenue is reduced. 298. According to the Public Administration Reform Strategy endorsed by the Government and the President in 2006, payroll controls and internal audit are to be addressed in the short to mediumterm. It is currently not clear what further reforms are envisaged with regard to the revenue system and the cash-flow forecasting and managementfunction. Accounting,Recording,and Reporting(PI-22 to PI-25) 299. Basic systems for accounting, recording, and reportingare in place. Accounts reconciliation is carried out in a timely fashion. In-year reports are prepared on a monthly and quarterly basis by the Treasury. The information appears to be reliable. However, a modern reliable "double entry" Financial Management Information System is a vital priority for the future if quality accounting and reporting is to be achieved. A key weakness is the fact that the budget execution report is not being submitted for external audit on an annual basis. Furthermore, accounting standards do not comply with international standards, and the standards used are not systematicallydisclosed in fiscal documents. External Scrutiny and Audit (PI-26to PI-28) 300. Externalscrutiny and audit are weak. Legislative scrutiny of the annual budget law is satisfactory; but the absenceof an administrativeclassificationlimits the ability of parliament to hold members of the executive to account. Tajikistan's External Audit Institution continues to be accountable to the president and currently lacks independence as required by international (INTOSAI) standards.52 The audit is largely focused on compliance with rules (regularity audits) which should be regarded as an adequate interim stage, given that the State Financial Control Committee (SFCC) was only established in 2002. Some training on other forms of audit (performance/value-for-money audits) has been received. Parliament is currently not reviewing audit findings. As noted above, the SFCC does not receive the Government'sfinancial statement for annual review (contrary to provisionsstipulated by law), but rather has undertaken bi-annualaudits of the budget (by conductingon- site audits of the MOF - in addition to audits of KBOs). Audit findings are primarily submittedto the president, and to parliament only in summary format. In late 2006, the SFCC was transformedinto an Agency of State FinancialControl and Combating Corruption. This is a problematic change; as the roles o f external audit and of combatingcorruptionare distinctand should be kept separate accordingto established internationalpractice. Donor Assistance (D1 to D3) 301, Donor assistance to Tajikistan still largely follows a traditional, project-based approach. Use of country systems has been very limited thus far as these systems are 52.Tajikistan's external audit institution currently is not a member of INTOSAI. It holds membership in ECOSAI, the regional associationfor the countriesof the Economic CooperationOrganization. This regional group is not a member of INTOSAI. 109 perceivedto be weak. An intensifieddialogue on aid effectiveness and the actions that the GOTand donors can take to improve aid practices would be desirable. Donor coordination should be a higher priority for the Government, given Tajikistan's substantialaid receipts (around 10% of GDP) and the importance of these relative to domestically generated revenues. Donors need to make greater progress towards harmonizationand alignmentof their aid. Increasingthe share of budget support may be one way o f encouraging a more comprehensive approach to Tajikistan's development needs; but it will require further progress with regard to the quality of the public financialmanagementsystems (and in particular internaland externalaudit functions). c. ASSESSMENT OF THE IMPACT OF PFMWEAKNESSES 302. PFM systems have three objectives: macro-fiscal discipline, strategic allocation of resources, and operational or technical efficiency. These three objectives are linked. Fiscal discipline is the basis without which neither a strategic allocationof resources nor operationalefficiency is possible. 303. The findings of the strengths and weaknesses of the current PFM system have the following implications in terms of these three objectives. Tajikistan has made important steps towards establishingmacro-fiscaldiscipline. However, the situation remains fragile due to systemic risks and weaknesses, as shown by the substantially increased fiscal deficit projected in 2007 (which will be funded by external borrowing). It will be essential not to lose sight of the fundamental issue of macro- fiscal discipline and the need to further strengthen relevant structures - including better revenue collection, sustainable debt management, and better oversight of fiscal risks arisingfrom SOEs and quasi-fiscalactivities. These issues should be addressed as a priority. 304. Furthermore, significant progress still needs to be made with regard to the strategic allocation of resources and the technical efficiency in the utilization of public funds. The current situation reflects initial reform efforts and priorities since the re-buildingof the state and of the PFM system begun in 1997 (with an emphasis on re-establishingfiscal discipline). In the early 2000s, efforts were undertaken to improve the strategic allocation of resources (introduction of an MTBF), but they failed to have a systemic impact due to a combination of over-ambitious goals, the absence of sufficient political will in the context of wider political challenges, and capacity constraint^.^^ 305. A new round of `second-generation' reforms is currently under preparation. To the degree that these reforms will be executed, they are likely to bring improvements with respect to the strategic allocation of resources (especially the implementationof an MTEF) and to technical efficiency. The earlier experience implies that comprehensiveness and sequencing of these further reform efforts will require careful attention, especially with regard to: (i)avoiding overburdening the MOF and other key institutions during the reform process; (ii) ensuring that refdrms cover implementationand do not stop at the adoptionof new legislation;(iii)ensuring that initiatives are complementary and well linked(especially with regardto financial 53See World Bank, ImplementationCompletionReport on IBTA2 (2006). 110 and sectoral management information systems; and complimentarity between information systems and underlying processes); (iv) ensuring that reforms are well adaptedto the contextof Tajikistan. Macro-fiscal Discipline 306. The use of conservative budget estimates and the ability to limit demands on the budget have been important assets in the context of a very resource-constrained and uncertaineconomic environment in Tajikistan. However, there is some concern that the commitment to fiscal discipline is weakening with the contraction of large external loans to fund infrastructure projects. A core Treasury system has been established and equipped with the requisite tools for managing budget executionand reporting. While the situationhas been stable in recent years, important risks remain. Tax revenue is still rather low, in part due to tax exemptions as well as to weaknesses in tax administration (and resulting high tax arrears). Also, there are substantial quasi-fiscalactivities by state owned enterprises (notably in the electricity industry), and the practice of issuing state guarantees for debts incurred by SOEs may be resumed (see PI-17). Monitoring of fiscal risks arising from SOEs is weak. Furthermore, after a period o f debt-write offs that helped to reduce the debt burden, Tajikistan is currently acquiring substantial new external debt. The fiscal deficit and debt situation will require careful monitoring in order to ensure that the debt burden does not again becomeunsustainable. 307. Finally, the current lack of a multi-year expenditure framework poses risks; particularly regarding the insufficient estimation of recurrent costs associated with large investmentsthat are beingundertakenor planned. Strategic Allocation of Resources 308. Open and policy-basedcompetition for resources is not yet well developed in Tajikistan's PFM system. The fragmentationof the Government at the central level and the evolvingsystem of intergovernmentalrelationspose challenges for the policy based allocation of resources. Transparency and public discussion of spending priorities in the legislature, civil society, and the media are still in their infancy. The GOT needs to decide whether to enhance the role of the cabinet-level Budget Commissionor to consider a different mechanism for the discussion of the budget at this level. Giventhe high share of investments, the discretionary part of the budget is substantial, and hence requires strategic decision-making. In this regard, there is also an urgent need for greater government leadership and policy debate on the management and allocation of aid resource^.'^ Furthermore, weaknesses in revenue collectionthat result in an overall reduction of resources should be part of the policy debate. Renewed efforts at establishing an MTEF have been initiated. The full implementation of an MTEF will be a medium-longterm effort, in particular with regard to developing information on the cost of programs and realistically costed sector strategies. 309. Inthe last three years, spending increasedon average by more than 8% during budget implementation.The uneven allocation of additional resources has resulted in 54PIP funding has accountedfor between 20 and 30 per cent%of total budgetaryresources per year. 111 substantial shifts in the sectoral allocation of funds, thus making allocative decisions taken duringbudget planningless binding. The main assets of the current system are the relative orderliness of the budget planningand executionprocesses, which are an important prerequisite for the capacity to translate policy choices into actual fiscal allocations. Operationalefficiency 310. Operational efficiency is important given that public services have to be delivered with very limited resources. Some measure of predictability in resource flows to core activities has beenachieved (while there still may be greater difficulties at local levels), and the internal control and reporting environment is in line with current technical capacities (including still limited automating of the Treasury system). 311. Key areas of concern with regard to operational efficiency are the following: first, procurement has yet to move from the stage of legal reform to full implementationin order to reap benefits in terms of efficiency. Secondly, institutions for internaland external audit are weak, do not cooperate, and are almost exclusively focusing on regularity in individualtransactions, rather than on systemic issues and on the effectiveness with which public funds are used. A strategy for improving internal control and audit is under discussion, and discussions have recently intensifiedon how to improvethe externalaudit function. 312. Furthermore, the expectation of additional funds becoming available during budget execution, distorts some of the incentive for budget organizations and their management to plan within a clear financial envelope. Currently, there is little transparency to citizens (as service users) regarding public funding allocations to front-line service providers such as primary schools or health facilities; and thus the potential for direct accountability between service users and service providers is not used. D. PROSPECTSFORREFORM PLANNING AND IMPLEMENTATION 3 13. A new round of PFM reforms and capacity buildingcovering a range of areas i s under discussion between the GOTand donors. This follows previous reform and capacity building efforts which have been o f mixed success (see World Bank, April 2006); including the successful establishment of a core Treasury and budget management system since the late 1990s, and a range o f legal reforms in the 2000s. Donor-supportedefforts to reform the system of intergovernmentalfinances and to introducean MTBF were not successful inthe early 2000s. 314. The overall environment is evolving (greater availability of resources, gradually improving capacity, step-by-step familiarization with reform ideas and options) and thereforeconditions for further reformare improving. 315. The key reforms and capacity building efforts currently under discussion includethe following: 112 budget planning: establishment of an MTEF (including a strengthening of financial planning and management capacities in line ministries); revenue forecasting (already under way); introduction of an administrative budget classification; budget execution: further development of the Treasury system, reform and expansionof internalcontroland audit; reform of intergovernmentalfiscal relations): per capita allocation of funds in social sectors; introduction of greater control of sector ministriesover sector- relatedspendingat local levels; revenue administration; reformof government structuresand organizationand of the public service; further donor support for the development of an independent external audit function Further donor support to develop the role of the parliament in scrutinizingall aspects ofthe budget process. As this list reflects, the reform agenda is broad and ambitious and will require good prioritization and sequencing. As a signatory to the Paris Declaration, Tajikistan has committed itself to undertake improvements of its PFM system. Exercisinggovernment ownership and active government management of this reform agenda will be crucial in maximizing the benefits from the financial and technical support that is beingofferedby donors. 3 17. It will be importantalso to pay attention to links and interactions between the various components of the reform agenda that is under discussion. For example, the further development of the treasury system and changes in the role and capacities for public financial management in line ministries should be well coordinated; similarly, there are important linkages between reforms in sector planningand management and changesto the system of intergovernmental(fiscal) relations. 3 18. Furthermore, while reforms are rightly focused on achieving improvements with regard to strategic allocation of resources, there are concerns about the apparent ambitiousness of these reforms, given the starting position. How exactly the introductionof the MTEF will be sequenced is a vital consideration.For establishing a meaningful medium-term planning system, basic building blocks of the fiscal system need to be strengthened further, including improvements in revenue administration,budget classification, payroll controls, procurement, and internal and externalaudit. 319. The present tendency of substantial budget deviations at the sector level implies considerable levels of unpredictabilityeven over a 12 month period. Until this unpredictabilityis reduced, attempts to program resources over the medium term will be futile. Furthermore, an importantobjectivewill be to find ways of integrating 113 capital and recurrent budgeting (with implications also for aid policy and management). 320. Evenif not the entire reformagenda as outlinedcan be put on track in the short to mediumterm (2-3 years), it i s likely that PFM systems in Tajikistanwill see further improvements over the coming period. This should lead to improvements in repeat rounds of the PEFA assessment at least inthose areas where substantive reform and capacity buildingefforts canbe set in motion. 114 REFERENCES Andrews, Matthew and Alister Moon (2003), PRSP and PEM Capacities in ECA PRSPCountries(mimeo), ECSP World Bank. Aschauer, DA. (1989), "Is public expenditure productive?", Journal of Monetary Economics 23, 177-200. Barro, R. J. (1990), "Government spendingin a simple modelof endogenous growth," Journal ofPolitical Economy 98, S103-Sl25. Barro, R. J. (1991), "Economic growth in a cross section of countries", Quarterly Journal of Economics 106,407-43. Barro, R.J. and Lee, J-W. (200l),"International data on educational attainment: updates and implications", Oxford Economic Papers 3, 541-563. Bearing Point. Information about the execution of the project "Reform of Economic Policy of CentralAsian Republics (Tajikistan) since January 2003, no date. Bose, N., Haque, M.E. and Osborn, D. (2003), Public expenditure and economic growth: A disaggregated analysis for developing countries', Working Paper, University of Manchester. De Broeck, M. and Koen, V. (2000), "The Great Contractions in Russia, the Baltics and the other countries of the Former Soviet Union: A View from the Supply Side", Working Paper WPlOOl32, IMF. Devarajan, S., Swaroop, V. and Zou, H. (1996), "The composition o f public expenditureand economic growth," Journal of Monetary Economics 37, 313-344. Easterly, W. and Fischer, S. (1994), "The Soviet Economic Decline: Historical and RepublicanData", Policy ResearchWorking Paper 1284, World Bank. Easterly, W. and Rebelo, S. (1993), "Fiscal policy and economic growth: an empirical investigation," Journal of Monetary Economics 32, 417-458. Ebinger, Jane (2006), "Measuring Financial Performance in Infrastructure: An Application to Europe and CentralAsia", WPS3992, World Bank. European UniordGovernrnent of Tajikistan (2006), Strategy for Developing Public Financial Control in the Republic of Tajikistan. Governmentof Tajikistan (2005), Public Administration Reform Strategy. Government of Tajikistan (2006), Self-assessment on the effectiveness of public expenditure management. 115 Green, C.J. and Murinde, V. (1998), "Modelling the macroeconomic policy framework for an emerging market economy", The Manchester School, 66, 3, 302- 330. Harrison, Graham, Esil Martinsen and Sundar Somodi (2005), "A Program for Tax AdministrationReform", TechnicalReport FiscalAffairs Department, IMF. Herrera, Santiago and Gaobo Pang (2005), "Efficiency of Public Spending in DevelopingCountries:An EfficiencyFrontierApproach", mimeo, World Bank. Le Houerou, Philippe and Robert Taliercio (2002), "Medium Term Expenditure Frameworks: From Concept to Practise. Preliminary Lessons from Africa", Africa RegionWorking Paper Series No 28, World Bank. Ilolov, Mamdasho, and Khudoiyev Mirodasen (200l),"Local Government in Tajikistan," in: Muteanu, Igor and Victor Popa (eds.): Local Governments in Eastern Europe, in the Caucasusand Central Asia, Budapest:Open Society Institute. InternationalMonetaryFund (2006A), Republic of Tajikistan: Sixth Review Under the Poverty Reduction and Growth Facility. Country Reportno. 06/62. IMF (2006B), "Republic o f Tajikistan - Ex Post Assessment of Longer Term Program Engagement", IMF Country Report 06/63. IMF and World Bank (2007), "Tajikistan: Joint Fund-World Bank External Debt SustainabilityAnalysis", mimeo. International Monetary Fund (2007). Republic of Tajikistan: Report on The Observanceof Standardsand Codes (ROSC), Fiscal TransparencyModule. Kronenberg, Tobias (2004), "The curse of natural resources in the transition economies", Eco.nomicsof Transition, Vol 12 (3), 399-426 Landau, D. (1986), "Government and economic growth in the less developed countries: and empirical study for 1960-1980", Economic Development and Cultural Change, 35(1), 35-75. Lensink, R., Hermes, N.and Murinde, V. (2000), "Capital Flight and Political Risk", Journal of International Money and Finance, Vol. 19, pp. 73-92. Levine, R. and Renelt, D. (1992), A sensitivity analysis of cross-country growth regressions,American Economic Review, 82, 942-63. Lorie, Henri (2003) `Priorities for Further Fiscal Reforms in the Commonwealthof IndependentStates' Working Paper WP/03/209, IMF. Loukoianova E. and Unigovskaya, A. (2004), "Analysis of Recent Growth in Low- IncomeCIS Countries", WorkingPaper WP/04/151, IMF. Matovu, J. (2005), "Sources of Economic Growth" in Tajikistan: Selected Issues and Statistical Appendix, Washington, DC, IMF. 116 Neel, Gerard (2006), Technical Report (EU: TAJIKISTAN: Streamlining Policy PlanningProcesses:PRSP and PIP). Pro-gramma(2006) Horizontal Functional Analysis of the State Executive Bodies of the Republic of Tajikistan, PreliminaryResults (presentation). Radev, Dimitar, John Zohrab, John Matovu and Sailendra Pattanayak (2005), "Public Financial Management and Treasury Reforms" Technical Report, Fiscal Affairs Department, IMF. Segura-Ubiergo, Alex, Alejandro Simone and Sanjeev Gupta (2006), "New Evidence on Fiscal Adjustment and Growth in Transition Economies" Working Paper WP/06/244, InternationalMonetary Fund. Summers, Victoria and Katherine Baer (2003), "Revenue Policy and Administration in the CIS-7: Recent Trends and Future Challenges", paper prepared for the Lucerne Conference of the CIS-7 Initiative. Umarov, K. and Repkine, A. (2004), "Tajikistan's growth performance: the first decade of transition" in G.Ofer and R. Pomfret(eds), The Economic Prospects of the C.Z.S.: Sourcesof Long Term Growth, Edward Elgar, Cheltenham. Varoudakis, Aristomene (2006), "Public Finance, Growth and Governance in Europe and Central Asia", Preliminary Findings for the 2007 ECA Regional Fiscal Study, World Bank. Wood, A. and Mayer, J. (2001), "African Development in a Comparative Perspective", Cambridge Journal of Economics 25, 369-394. World Bank (2003), Country Procurement Assessment Report. World Bank (2004A), Country Financial Accountability Assessment. World Bank (2004B), Tradeand TransportFacilitation in Central Asia. World Bank (2004C), Central Asia: Regional Electricity Export Potential Study, Europe and CentralAsia Region, World Bank. World Bank (2005A), Republic of Tajikistan Poverty Assessment Update,Report NO:30853-TJ. World Bank (2005B), Tajikistan: TradeDiagnostic Study, ReportNo 32603-TJ. World Bank (2005C), Republic of Tajikistan: Health Sector Note, Report No 29858- TJ. World Bank (2005D), Tajikistan: Public Expenditure and Institutional Review. World Bank (2005E), Tajikistan Energy Utility Reform Review. 117 World Bank (2006A), "Energy Sector Development of Kyrgyz Republic", presentation, to the Government and multilateral and bilateral financial institutions, Biskek, June. World Bank, (2006B), Social Protection in the Electricity and Gas Sectors in Tajikistan: Policy Note. World Bank (2006C), Europe and CentralAsia Inpastructure Study. World Bank (2006D), Implementation CompletionReport: IBTAZ. World Bank (2006E) Aid EffectivenessReview: Tajikistan. 118 Annex I.The Implicationsof Public Expenditure for Economic Growth in Tajikistan55 Victor Murinde 1. Introduction The motivation of this study is to examine the implicationsof various forms o f public expenditure for supporting the priority of the Tajik government to sustain high rates of economic growth. Average growth rates of 8-10% per annum recorded during 1998-2004 were largely driven by recovery of the economy from civil war (1992- 1997), with unused capacity being brought back into production once peace and security had been restored. Although output of the economy is still around 30% below its level in 1990, much of the capital stock of the productive sectors dates from the Soviet era and is illsuited for the demands of a market economy. There are already signs that the recovery is beginning to falter, with growth slowing to 7% in 2005. Consequently, sustaining high rates of GDP growth in future will require new investments or a combination of new investmentand more productive use of existing factors of production. It is anticipated that future growth will be influenced by the nature of public expenditure, through its impact on the factors that drive long term sustainable economic growth, such as human capital formation through education and health services, the macroeconomic environment, as well as constructionand maintenance of public infrastructure. The Government is undertaking budgetary reforms, one of the aims of which is to improve the allocationof public expenditureso that it can support sustainable economic growth. Hence, the objective of the study is to modelthe growth effects of government expenditure, to identify the main constraints to economic growth andto assess their implicationsfor public expenditure. In spite of the growing research attention to the relationship between government spending and economic growth, a consensus has not emerged regarding what type of government expenditures are growth promoting for developing and transition economies. For example, some studies have found a significant and positive relationship between government size, as measured either by the level of total public expenditure or by the level of public consumptionexpenditure, and economic growth. However, some other studies have found a significant but negative relationship (see, in particular, Landau 1986; Grier and Tullock 1989; Barro 1990, 1991). Between these conflicting results, some other studies have found the relationship to be insignificant (for example, Levine and Renelt 1992). In general, the conflicting results have been attributed to differences in empirical models, data sets, econometric techniques, and sample economies. The choice of sample economies is particularly important because most of the existing findings on the growth effects of public spending are drawn either from the experiences o f developed countries or large samples pooled from developed, developing and transitioneconomies, with the exceptionof the work by Landau(1986) and Devarajan et al. (1996)). What would be more helpful, in the context of this study, is to obtain 55 This backgroundstudy for the PPER was preparedwith the financial support of DFID. 119 more findings that are based on serious empirical work on the institutional structure and the process by which public expenditure variables influenceeconomic growth in transition economies only, in contrast to broad cross-country samples dominated by developedcountries. Inaddition, it is argued that the differences arise from the use of different sets of conditioning variables and initial conditions which are used by various studies on the impact of fiscal variables on endogenous growth (see, Levine and Renelt 1992). Further, it is argued that the conflicting results partly arise from failure by researchers to acknowledge the role of the Government budget constraint in growth models that include fiscal variables. The argument is that in order to undertake a meaningfulevaluationofthe effectsof taxes or expenditures on economic growth, it is importantto recognisethe flow-of-fundconcept about the sources as well as the uses of funds in the growthprocess. For example, while the studies by Landau (1986) and Devarajanet al. (1996) are importantto this study because they focus on developing countries, they suffer from the possibility of omission bias, in the sense that they only focus on the expenditure side of the budget constraint and ignore the revenue side. Hence, the main objective o f this study is to empirically identify the specific components of government expenditure which significantly impact on economic growth, and which could be used as part of a public expenditure review in order to target sustainable economic growth. We invokeendogenous growth theory and apply it in two steps. The first step is to use the model to study the growth effects of government expenditure in eight CIS countries (Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyz Republic, Moldova, Mongolia and Tajikistan) for the period 1996-2004 and identify the patterns of government expenditure that are associated with economic growth regionally. The second step is to use Seemingly Unrelated Regression Equations (SURE) techniques to isolate the coefficients specifically and hence identify the types of public expenditure which are most likely to support sustainedeconomic growth in Tajikistan. Clearly, the emphasis of the study is on the government expenditure side; the government revenue side and the financing variables, such 'as government budget surplus/deficitand tax revenue, are included in the model in order to incorporatethe government budget constraint and to avoid the coefficient biases that would result from their omission. We test the importantof each government expenditurevariable on economic growth; we also test for the variables as a group. We explicitly recognize that Tajikistan is a resource constrained transition economy, hence it is important to investigate the impact of each sector-specific government expenditure, for example, government expenditure on education or health. The information from sector-specific government expenditure is useful for informing policy, especiallywhere the allocationof limitedpublic resourcesbetweenthe sectors i s a critical issue Le., shall the Government spend more on health or on education, giventhat resourceconstraints meanthat it cannot spendmore on both. The remainder of this paper is structuredinto four sections. Section 2 highlights the institutional issues, including geography and historical legacy, which may constrain the possible impact of Government expenditure on economic growth in Tajikistan. The body of existing knowledge on the growth effects of public expenditure is surveyed in Section 3, to provide a point of departure for our specific study on 120 Tajikistan and to provide contrasting experiences in other countries. Section 4 presents the model and reports the estimationand testing results. The main findings, conclusions and policy recommendations are presentedin Section 5. 2. Institutional Issues: Economic Geographyand Historical Base Tajikistan shares a common Soviet economic heritagewith other CIS countries, and this institutionalbackground influencesthe sample selectioncriteria for the empirical work in this report. The argument is that the economic geography and historical developments are important institutionalfactors in constructinga regional sample for studying the drivers of economic growth in the area. A sample that recognizes both factors includes the CIS-7, namely Tajikistan, Azerbaijan, Uzbekistan, the Kyrgyz Republic, Kazakhstan, Armenia, Georgia and Moldova, as well as Mongolia, which although not a CIS member, is a landlocked country which had a state controlled economy. We exclude some regional neighbors of Tajikistan which have different institutional features, including Afghanistan, Pakistan, India and China. Afghanistan is a failed state which has suffered civil conflict for more than two decades . Indiaand Pakistanare much more established as market-drivencountriesrather than transitional economies. China is a transition economy but its size makes it very different from Tajikistan. An important comparative experience of these neighboring countries relates to the changes in economic growth rates following independencefrom the Soviet Union and formation of CIS in the early 1990s. Like most other CIS countries, Tajikistan suffereda sharp decline in economic growth in the six years following independence in 1991. The economy recovery began inthe late 1990s. Broadly, there are at least two interesting lines o f enquiry which are important for empiricalwork and policy design. The first issue is to disentangle the factors which explain growth, especially in the periodsince 1996. In view of the existing literature, surveyed briefly in this study, this line of enquiry suggests an endogenous growth model. The second line of enquiry, which is strongly related to the first one, is to identify the most promising public expenditure interventions which will accelerate economic growth. Overall, on the basis of the institutional and historical factors of Tajikistan, and given the existing knowledge of other transition economies, it is plausible to suggest the following key candidate avenues and critical factors for . economic growth in Tajikistanduringthe next 10years. .. Human Capital and Education: Includingeducation at the primary school and secondary school levels. Life Expectancy and Health:These includehealthservices and expansion of the healthand sanitationsectors. Capital expenditure and private investment: The main type of investment . required is mainly private, such as in agro-exports. Labor intensive exports increase wages and value added, and hence growth. Miwant workers and remittances: Remittances form Tajiks working abroad may be expanded to supplant Foreign Direct Investment (FDI) and exogenous Dinflows. The type of interventionrequired includesnew investment in health, education, and other social service sectors. 121 Physical and human capital in Taiikistan: Current indications are that physical and human capital remains well above the level that is needed to produce the present GDP per capita of US$310; it is also far above the level that can be sustained with the present GDP per capita. It is not clear, though, whether physicaland human capital is used efficiently. Indicationsare that it is not, but we need to invoke a reliable measure of the efficiency o f physical and human capitalandthen apply it to obtainserial indicatorsover time. 3. Public Expenditureand Economic Growth: What Do We Know? From the start, it is useful to note that the strand of literature on public expenditure and economic growth is slightly different from the strand of literature on isolatingthe factors that explaineconomic growth in a neo-classicalgrowth model. Inthe case of the latter, researchers tend to use a growth accounting framework to analyse the sources of economic growth; for example, De Broek and Koen (2000) and Loukoianova and Unigovskaya (2004) for CIS countries and Matovu (2005) for Tajikistan. In general, the growth accountingframework disaggregates GDP growth into three components: the growth in the capital stock, the growth in the labour force (and labor productivity) and the growth in total factor productivity (TFP). TFP is a residual which picks up all of the influences on the productivity of factors of production, as well as measurement errors. The fundamental assumptions underpinningthe growth accounting exercises arise from a constant return to scale productionfunctionwith elasticitiesof outputwith respect to laborand capital. The strand of literatureon public expenditure which is relevantto this report mainly derives from endogenous growth models. As noted by Le and Suruga (2006), there have been quite a number of empirical studies analysing the impact of public expenditures on economic growth. However, the findings are not unanimous. On the one hand, some findings support the idea that public investment has a positive impact while public consumptionhas a negative impact on economic growth. For example, the findings by Aschauer (1989), Barro (1991) and Easterly and Rebelo (1993) support the argument that government expenditures spent on main physical infrastructure have a significant positive impact on private-sector productivity. In addition, using a sample of 39 developing countries during the period 1990-2000, Gupta et al. (2002) find a positive relationship between growth and fiscal policy (government budgetary balance, the financing of budgetary deficits, and expenditure composition), and conclude that in order to achieve a sustained fiscal adjustment, governments should reallocate expenditures on wages and salaries to more productivity uses such as capitalexpenditures. On the other hand, some researchers have challenged the above findings by specifically arguing that the relationship between government expenditure and economic growth is not positiveper se, but tends to be sector-specificand ultimately depends on the type of Government expenditure. For example, Landau (1986) categorises government expenditure into five groups (consumption, education, defence, transfers and capital expenditure) in order to conduct a comprehensive study of the impact of government expenditure on economic ,growth, using cross section data of 96 countries over various time periods between 1961 and 1976. It is found that each type of government expenditure has either significant negative or insignificant positive effect on economic growth. In addition, Grier and Tullock 122 (1989) use a panel data analysis on 24 OECD countries in the period 1951-1980and 89 other countries in the period 1961-1980 and find similar inconsistentrelationships betweenpublic expenditure and economic growth as those found by Landau (1986). Moreover, in a sample of Asian countries over the period 1970-1990, Devarajanet al. (1996) find that an increase in the share of current expenditure has positive and statistically significant growth effects, while Government capital expenditures (transport and communication) have either a negative or insignificant positive impact on economic growth, in sharp contrast to findings by previous researchers, such as Aschauer (1989) and Barro (1990) who regardthese expendituresas productive. A main problem with the existing literature stems from the variations in the models used, estimation and testing techniques and the heterogeneity of the countries included in the sample countries. Another problem arises from the differences in the motivationsof the studies, rangingfrom attempts to model growth and convergenceto motivation to tease the main variables that drive growth which could be used as drivers for public expenditure review. The latter is consistent with the objectives of this study. A relevant study along these lines is the recentwork by Bose, Haque and Osborn (2003), who examine the growth effects of government expenditure for a panel of thirty developing countries over the decades of the 1970s and 1980~~with a particular focus on sectoral expenditures. It is found that the share of government capital expenditure in GDP is positively and significantly correlated with economic growth, but current expenditure is insignificant. It is also found that, at the sectoral level, government investment and total expenditures in education are the only outlays that are significantly associated with growth once the budget constraint and omitted variables are taken into consideration. Hence, on the basis of the foregoing, what we know is that a substantial volume of empirical research has been directed towards identifying the elements of public expenditure which influence economic growth. We also note that the literature varies in terms of data sets, econometric techniques, and often tends to yield conflicting results and hence policy recommendations. Following Bose, Haque and Osborn (2003), we aim to address the shortcomings in the existing literature and also draw from our knowledge of the institutional structure of the Tajik economy. For example, we explicitly recognise the implications of the government budget constraint in our empirical work, specifically by considering both the sources and the uses of funds simultaneously for a meaningful evaluation of the effects of taxes or expenditures on economic growth. 4. Econometric Analysis and Results 4.1An Endogenous Growth Model with Government Expenditures We modelthe growth effects of government expenditure for a panel of eight countries (seven CIS countries plus Mongolia) during the period 1996-2004, when these countries were experiencing economic recovery. We invoke a methodology which differs from previous research in two ways. First, we explicitly recognize the role of the government budget constraint. Second, we control for possible biases arising from omitted variables. Specifically, we aim to isolate the specific components of governmentexpenditurewhich significantly impacton economic growth. 123 We invoke an endogenous growth model following Bose, Haque and Osborn (2003). We specify a baseline model which captures endogenous growth variables as well as the government expenditurevariables, as follows: J K GROWTH,,= a,,, xa,CONDS,,, -k xPkGOVTSPENk,, + + xynOTHERn, N + E , , (1) /=I k=l n=l where the superscripts denote country i at time t for each vector of variables; GROWTH represents growth of real GDP per capita; the vector CONDS captures the main explanatory variables that commonly appear as conditioning variables in endogenous growth models; the vector GOVTSPEN represents the explanatory variables which capture government expenditure, hence these are the variables that are of particular interest for the study, namely government expenditures and their major components at aggregate and sectoral levels, expressedas percentages of GDP; the vector OTHER includes all other explanatory variables which are added to the model in order to test the sensitivity of the model, thus we include variables that often have been included in previous studies as indicators for monetary polices, trade polices, and market distortion; and E is a stochastic term. We follow Bose, Haque and Osborn (2003) to incorporate the government budget constraint in the model. It is argued that empirical investigation of the impact of government spending on growth, which does not incorporate a government budget constraint, is vulnerable to bias that arises from omitted variables. We consider a simple government budget constraint specified in Green and Murinde (1998), as follows: GS- TX = AMON + B O N D + AFM(2) Where GS = total government spending, TX = total revenue (including tax as well as well as non-tax revenue), MON = narrow money; BOND = government domestic borrowing, and FRN = foreign borrowing. The Governmentbudget identity specifies that Government expenditure must equal Government revenue or otherwise the balance would be funded by the issue of new money, borrowing domestically or borrowing abroad. For this matter, we cannot include all elements of government spending as well as elements of government revenue; we need to drop at least one element. We therefore include in the model at least one government expenditure variable together with the tax revenue variable. Altogether, we include only the following variables: Government capital expenditure, Government recurrent expenditure, Government expenditure on education, Government expenditure on health, Government expenditure on education and health, domestic revenue and Government budget deficit or surplus; all the variables are given as percentage of GDP. 4.2 Data and Measurement The transition economies we take as the sample are members of the CIS who share not only geographicalproximity with Tajikistan but also trong regionaleconomic ties: these are Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan and Moldova. We also include Mongolia in the sample. Hence, we focus on these 8 transition 124 economies. Our dataset is retrieved from World Bank online data supplemented by World Bank Country Economic Reports and Public Expenditure Reviews, for the period 1996-2004. Hence, overall, we constructed a panelof 72 observations. The measurement of the variables in the model is as follows. The endogenous variable of the model, GROWTH, is measured as growth rate of realGDP per capita. The explanatory variables for including in the vector COND are consistent with those used in recent work by, for example, Levine and Renelt (1992), Barro (1991) and Bose, Haque and Osborn (2003). Thus, the vector CONDS includes the main variables that commonly appear as conditioning variables in endogenous growth models: LFEX=Log of life expectancy; PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; HCGEN = initial human capital (PS) variable; NINV = Non government investment (a proxy for private investment); ILFEX = Log of life expectancy at initial period as at 1996; IGDPL = Log of initial GDP per capita as at 1996; we include this variable in order to control for possible effects of convergence on output growth. The variable HCGEN is constructed in the context of Landau (1986) as the weighted sum of the initial enrolment ratios in primary and secondary schools. The weights are Ifor primary school enrolment ratio and 2 for secondary school enrollment ratio, hence, these weights are approximations to the relative values of two types of education. We measure PRIM and SECS using school enrolment ratios rather than the average schooling years because the latter were not available; moreover, previous studies, such as Easterly and Rebelo (1993) and Bose, Haque and Osborn (2003) use school enrolment ratios in the sense that the ratios are probably better available measures of investment in education than the average schooling years. In the analysis, we use the HCGEN variable rather than its subcomponents because of the high milticollinearity between the separate enrolment rates of PRIM and SECS. Overall, therefore, the variables included in vector COND capture the key argument o f the endogenous growth literature, namely that human capital and institutional factors are important determinants of economic growth in Tajikistan. The vector GOVTSPEN represents the variables which capture government expenditures and their major components at the aggregate and sectoral levels, expressed as percentages of GDP. We include the following: CEXP = Government capital expenditure; REXP = Government recurrent expenditure; EEXP = Government expenditure on education; HEEP = Government expenditure on health. We also tried combinations of EEXP and HEEP to capture government expenditure on services; we also constructed total government expenditure as the sum of CEXP and REXP. The vector OTHER includesall other variables which are added to the model in order to test the sensitivity of the model. We follow Easterly and Rebelo (1993) and define the explanatory variables in the vector OTHER to include variables that often have been included in previous studies as indicators for monetary polices, trade polices, and macroeconomic stability; we included the following: BMON = Broad money (M2) (% of GDP); TTDE = Trade ratio (export plus import as % of GDP); ALIT = Adult literacy (as % of total population); INFL = Inflation rate, computedas the rate of change o f the consumer price index. The variable BMON is included in order to control for the effects of monetarypolicy. The variable TTDE is included in order to 125 control for the degree of openness. It has been shown by previous studies, notably Levine and Renelt (1992) that the two variables are significantly correlated with economic growth. In addition, as discussed above, we include variables to capture the government budget constraint, namely: DREV = Domestic revenue; FDEF = Government surplus / deficit. Becausethe government budget constraint is an identity (see equation 2), we need to exclude an element of the identity and thus avoid perfect collinearity. Another context for the inclusion of DREV is in order to make a proper assessment about the growth effects of public spending, because while the provision of public goods is growth-enhancing, the distortionary taxes that need to be raised to fund the provisionof the public goods may have growth-diminishingeffects. As a summary of the above, Appendix Table A1 presents the variables used in the empiricalwork, the measurementofthe variablesand the sources of data. Table 1: Descriptivestatisticsof the mainvariablesin a panelof commonwealth HCGEN 129.46762 109.93558 -0.10328 -1.76059 CEXP 0.059748 0.050000 0.94008 0.047908 REXP 0.20706 0.079493 0.46280 2.22195 ALIT 9.61250 29.49781 2.77749 5.87745 PRIM 49.54023 43.65270 -0.24769 -1.94013 SECS 39.96369 38.54898 -0.039259 -1.95863 * EEXP 0.035046 0.024183 0.61222 0.096532 LHEEP 0.018339 0.014560 1.06593 1.00471 FDEF -0.043 103 0.039203 -0.27623 -0.30656 1BMON 10.14413 10.077471 I 1.94156 14.95123 I TTDE 0.81803 0.40459 -0.4 1098 0-50306 INFL 0.12638 0.15527 1.93515 7.18600 Note: GROW = Average growth rate in GDP per capita; LFEX = Log of life expectancy; ILFEX= Life expectancy at initial period; IGDPL = initial GDP per capita, as at 1996; NINV = Non government (Private investment) (% of GDP); DREV = Tax revenue (% of GDP); HCGEN = initial human capital (PS) variable; CEXP = Government capital expenditure (% of GDP); REXP = Government recurrent expenditure (% of GDP); ALIT = Adult literacy (as % of total population); PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; EEXP = Government expenditure on education (% of GDP); HEEP = Government expenditureon health(% of GDP); FDEF = Government surplus / deficit (% of GDP); BMON = Broad money (M2) (% of GDP); TTDE = Trade ratio (export plus import as % of GDP); INFL = Inflation rate, computed as the rate of change of the consumer price index. 126 HCGEN, the initial human capital variable is constructed in the context of Landau (1983) as the weighted sum of the initial enrolment ratios (%) in primary and secondary schools. The weights are 1 for primary school enrolment ratio and 2 for secondary school enrolment ratio, hence, these weights are approximationsto the relative values of two types of education. We use the HCGEN variable because of the high multicollinearity betweenthe separateenrolment rates of PRIMand SECS. 127 4.3 Patterns of GovernmentExpenditure and Economic Growth Variables in CIS, including Tajikistan We first check the data for normal distribution and other univariate properties, using the descriptive statistics reported in Table 1. We also check the bivariate relationshipsof the endogenous variable and the explanatory variables, using a correlationmatrix reported in Table 2. Table 1 shows that the endogenous variable has a reasonable mean economic growth rate of 5% but the high standard deviation o f 4.5% suggests dramatic swings inthe growth rates of the sample economies during 1996-2004.In general, however, the main government expenditure variables show a normal distribution in the sense that they are generally symmetric about the mean values. Overall, there are very limited problems of skewness and kurtosis in the data for most variables, accordingto the statisticsreported in Table 1, except perhapsfor the inflation rate variable. \ The correlation matrix in Table 2 reports the bivariate relationshipsof all variables in the model, and thus providesthe first insight into how each explanatory variable is correlated with the endogenous variable as well as each of the other explanatory variables in the model. These correlations should give us a first glimpse into patterns of relationships in trying to pin down the impact of each element of government expenditure on economic growth. Hence, Table 2 should be read against Appendix A. T he latter presents 11 charts of scatter diagrams which show the patterns of government expenditure variables and other variables of the model against the economic growth variable for the eight sample countries for the period 1996-2004. Some interesting patterns emerge from these charts, regarding the relationship between government expenditure variables and economic growth in the transition economies. Three mainpatterns have importantpolicy implications, as highlightedbelow. First, EEXP and GROWTH show a patternthat supports the argument of higher economic growth even at low levels of government expenditure on education, hence consistentwith the human capital argument. In addition, we find a pattern of high primary school enrolment rate (PRIM) and high economic growth, which is consistent with humancapital arguments. Also, there is a pattern of high secondary school enrolment rate (SEC) and high economic growth, which is consistent with human capital arguments. Hence, primary school education and secondary school education are important for economic growth. Again, this is the human capital argument and suggests government intervention to build human capital through education.The correlationcoefficients reported in Table 2 support this pattern. The correlation between economic growth rate per capita and the composite human capital variable (HCGEN) is 0.4699; even the components o f human capital are highly correlated with growth, primary school enrolment rate at 0.3361 and secondary school enrolment rate at 0.4798. In addition, government expenditure is also correlated with economic growth by 0.2931. The important policy implication, subject to further investigationin the empirical part of this study, is that government expenditure in educationas a proportionof GDP (EEXP) is importantfor economic growth. Second, CEXP and GROWTH have a pattern showing positive associated between economic growth and government capital expenditure (CEXP), even at lower levels of capital expenditure, there is positive growth. Table 2 shows a correlation coefficient of 0.1513 between CEXP and GROWTH. A higher coefficient, though, is reported for non government investment (NINV) and economic growth at 0.2630, which suggests that capital expenditure is positively correlated with economic growth, but that it is private 129 investment that has a higher impact. Third, there is a high correlation coefficient of 0.604 between the budget surplus and economic growth, which emphasizes not only budgetary discipline but also the importance of an `enablingmacroeconomic environment'. Overall, the patterns in the charts in Appendix A are generally consistent with the results from preliminary data analysis presented in Tables 1 and 2. The patterns will be more rigorously investigatedusingeconometric methods inthe remainder of this section. 4.4Initial Estimation and TestingResults As the starting point in the analysis, we are interested in investigating whether the government expenditure variables in vector GOVTSPENare significantly correlatedwith economic growthafter controllingfor the variables inthe COND vector. Hence, we run a restricted version of the model in equation (I), the gamma coefficients are where constrained to be zero. The regressions are run in such a way that we add one GOVTSPENvariable at a time. These initial regressions are intended to capture a central idea of the endogenous growth literature namely that human capital is an important determinant of economic growth. In addition by including initial GDP in these initial regressions,we are tryingto explorethe effectsof convergenceon growth. Table 3: Estimation and Testing Resultsfrom an EndogenousGrowth Modelfor a Panel of Commonwealth Independent States (CIS) includingTajikistan: The Impact of Initial Variable Model Model Model Variant Model Model Variant 1 Variant 2 3 Variant 4 Variant 5 C 0.038 -61.977 -68.624 -52.416 -123.171 (2.830) (-2.817) (-3.194) (-2.492) (-2.3 15) [0.006] [0.007] [0.002] [0.15] [0.024] CEXP 0.509 0.470 (4.639) (3.4 19) [O.OOO] [0.001] EEXP 0.510 0.583 (1.437) (1.653) I I [0.156] [0.103] ILFEX I1 I 14.601 16.167 12.354 I 28.983 (2.820) l(3.199) II (2.496) I (2.316) I HCGEN I I 0.628 I 0.676 I (1.274) I I (1.407) 1 1 [0.208] I [0.164] FEDEF (7.230) (3.871) (6.207) (3.742) (6.201) [O.OOO] [O.OOO] [O.OOO] [O.OOO] [O.OOO] AZ 0.028 -0.03 1 0.106 (1.983) (-1.996) (1.967) [0.052] [0.294] [0.054] INFL 0.025 (0.888) [0.3781 GE 0.055 -0.039 -0.04 1 -0.041 130 (3.583) (-1.996) (-2.100) (-2.1 15) [0.001] [0.050] [0.0401 [0.038] KA 0.035 III 1.287 1.427 1.087 2.660 (2.310) (2.780) (3.157) (2.456) (2.306) [0.024] [0.007] [0.002] [0.017] [0.024] KY IIII IIIIII0.026 0.748 0.830 0.642 1.581 (1.992) (2.754) (3.134) (2.451) (2.302) [0.051] [0.008] [0.003] [O.O 171 [0.025] ML 0.326 0.678 0.759 0.587 1.530 (0.228) IIIIIII(2.572) (2.950) (2.290) (2.254) [0.004] [0.025] [0.028] MO I II 1.141 1.272 0.980 2.389 (2.716) (3.106) (2.414) (2.2 84) J0.0031 [0.019] [0.026] TJ I 1.449 1.609 1.225 2.997 (3.121) (2.425) (2.290) ro.0031 ro.0I81 r0.0251 SSR 0.060 0.059 0.060 0.061 0.057 SER 0.031 0.031 0.031 0.031 II ARS III 0.539 1II 0.516 0.5 10 I0.511 ,I 0.030 0.549 Note: Below the estimatedcoefficients are reportedtl t-values in (.) bracketsand the p-value in [.] brackets. In the above regressions, the endogenous variable is GROW = Average growth rate in GDP per capita. The rest of the variables are defined as follows, as in Appendix Table A 1: CEXP = Government capital expenditure (% of GDP); REXP = Governmentrecurrent expenditure (% of GDP); EEXP = Governmentexpenditure on education (% of GDP); HEEP = Governmentexpenditureon health(% of GDP); DREV = Tax revenue (% of GDP); FDEF = Government surplus / deficit (% of GDP); NINV = Non Government(% of GDP); BMON = Broad money (M2) (% of GDP); PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; HCGEN = initial humancapital (PS) variable; LFEX = Log of life expectancy; TTDE = Trade ratio (export plus import as % of GDP); ALIT = Adult literacy (as % of total population); INFL =Inflation rate, computedas the rate of change of the consumerprice index; IGDPL = Log of GDP per capita at initial period. Table 3 presents the estimation and testing results from the endogenous growth model with government spending variables Le., with only the explanatory variables in vectors COND and GOVTSPEN. Five model variants are presentedto allow for inclusion of all government spending variables. It is shown that of the variables in the COND vector, only initial life expectancy variable shows a statistically significant relationship with economic growth in model variants 2-4. This result is consistent with the endogenous growth literature in which life expectancy is a crucial determinant of economic growth; the sign in each of the five model variants is positive, suggesting a positive relationship between life expectancy and economic growth. The variable for initial human capital is found to have a positive effect on growth in model variants 2 and 4; although the coefficientsare not statisticallysignificant. However, we do not find any evidence to show that there is convergence among this group oftransition economies. This is not surprising,perhaps, giventhat these economies are still in the process oftransitionto a market economy. In terms of the explanatory variables in the GOVTSPEN vector, the most significant results are obtained for Government capital expenditure (CEXP) in model variants 1 and 5, and Government expenditurein the education sector (EEXP) in model variants 2 and 3. The results for Government expenditure on health (HEEP) and recurrent expenditure (REXP) were not statistically significant and were dropped from the modelvariants 1-5. 131 The significant association between the share of Government capital expenditure in GDP and economic growth is not entirely surprising in the light of the conclusions drawn by previousstudies (see, for example, Easterlyand Rebelo 1993; Cashin 1995; Fuente 1997). Our results are also consistent with those of Bose, Haque and Osborn (2003) in a panel study that includedcapitalexpenditure in the regressionfor developingcountries. Our results which show a positive relationship between Government expenditure on education and economic growth differ from conclusions drawn by previous studies, irrespectiveo f whether these are based on data for a large pool of countries or developing countries (e.g., Landau 1986; Devarajan 1996). These earlier results indicate that the association of this variable with growth is either insignificant or non-robust. We find expenditure on education to be significant(albeit weakly); it is shown that one percentage point increase in government expenditure on education in relation to GDP is associated with an increase in the average growth rate of real GDP per capita by 0.5 percentage points. This effect may be explained in terms of the fact that increased expenditure on education amounts to an investment in human capital, which raises the potential for economic growth, accordingto the endogenousgrowthliterature. The model reported, whose estimation and testing results are reported in Table 3, is augmented with variables from vector OTHER and from the government budget constraint. The inclusion of these additional variables also amounts to a check for the robustness o fthe explanatory variables in the GOVTSPENvector. 4.5 A Checkfor the Robustness of the Results We now conduct checks for the robustness of our initial findings by augmenting the model with explanatoryvariables from the vector OTHER in the model (equation 1). The variables includedinthis vector are suggestedby the previous literature. For example, we include the ratio of broad money (M2) to GDP (BMON) and the trade share of GDP (TTDE), following Easterly and Rebelo (1993) and Levine and Renelt (1992). In our study, the ratio of broad money to GDP is included in order to control for the effects of monetary policy, especially as these countriesare in transitionto a market economy. The trade share of GDP is includedin order to control for the degree of openness, given that economic growth in Tajikistan seems to largely depend on its trading links with the neighbouringcountries. We also includethe growth rate of the terms of trade (TTDE2) as an additional variable. This variable captures the adverse effect of trade shocks that Tajikistan and a number o f transition economies in the region experienced during the sample period. Previous research has shown that the growth rate o f the terms of trade is significantly and positively relatedto economic growth in developing countries (see, for example, Deverajanet al. 1996). 132 Table 4: RobustnessCheck on the Results from an EndogenousGrowth Model for a Panelof CommonwealthIndependentStates (CIS) includingTajikistan Variable Model I Model I Model Model Variant 1 Variant 2 Variant 3 Variant 4 CONSTANT 0.114 II 0.103 II 0.094 III 0.864 (3.572) (3.194) (2.800) 10.154 [0.001] [0.002] [0.007] [0.878] LFEX 0.810 III0.234 III0.328 II 0.847 (0.043) (0.121) (0.0 17) (0.441) [0.966] [0.904] [0.9861 [0.661] NINV -0.896 0.011 -0.025 0.029 (-0.102) (0.125) (-0.287) (0.337) [0.919] [0.901] I I [0.775] [0.738] DREV -0.205 -0.083 -0.089 -0.035 (-1.076) II1 (-0.437) I (1.267) III(-1.666) [0.286] [0.664] [0.211] [0.102] HCGEN 0.678 II 0.776 II 0.616 II 0.371 (1.237) (1.390) (1.100) (0.654) [0.221] [O. 1701 [0.276] [0.516] CEXP 0.740 I (2.401) HEEP I [0.020] 0.233 (0.338) [0.737] EEXP 0.772 0.90 1 (1.559) (1.774) [0.1251 [0.82] FDEF 0.642 0.561 0.615 0.546 (0.201) (2.789) (3.098) (2.885) [3.196] [0.007] [0.003] [0.006] BMON 0.019 0.054 0.156 0.213 (0,139) (0.385) (0.977) (1.391) [0.8901 [0.702] [0.333] [O. 1711 BMON(-1) -0.146 -0.087 (-1.396) (-0.804) [O. 1691 [0.425] TTDE 0.969 -0.199 0.026 0.02 1 (0.244) (-0.050) (0.642) (0.537) [0.808] [0.960] [OS231 [0.594] INFL 0.025 0.0 17 0.675 -0.010 (0.805) (0.566) (0.201) (-0.3 10) [0.424] [0.574] [0.842] [0.758] INFL(-1) 0.045 0.027 (1.463) (0-030) [O. 1501 [0.372] AZ -0.029 -0.026 -0.030 0.055 (-1.142) (0.942) (1.168) (1.345) 10.2581 [0.3501 [0.2481 [0.185] GE -0.926 -0.020 0.0 12 0.061 -0.471 (-1.027) (0.563) (1.544) [0.639] [0.3091 [0.5761 [O. 1291 KA -0.032 -0.034 -0.028 0.055 133 (-1.747) (-1.790) (-1.464) (1.702) [0.086] [0.079] [O. 1491 [0.095] KY -0.044 -0.029 -0.043 0.0322 (-2.00 1) (-1.407) (-1.895) (1.303) [0.O501 [O. 1651 [0.064] [O. 1991 ML -0.065 0.067 -0.045 0.069 (-1.356) (-1.354) (-0.908) (1.215) [0.181] [O. 1811 [0.3681 [0.230] MO -0.060 -0.048 -0.066 (-1.980) (-1.604) (-2.055) [0.053] [0.114] [0.045] TJ -0.068 -0.05 18 -0.070 0.017 (-2.004) (-1.569) (-1.988) (0.455) [0-0501 [O. 1221 [0.052] [0.65 11 SSR 0.060 0.062 0.052 0.049 SER 0.033 0.034 0.320 0.03 1 ARS 0.463 0.441 0.05 10 0.521 The test for robustness, which is applied in this study, follows Levine and Renelt (1992) and Lensink, Hermes and Murinde (2000), among others. According to this test, the variables in the GOVTSPEN vector are robust if they can neither lose their statistical significance nor experience a sign reversal when other variables have been added to the model i.e. the results of the variables in the vector are not sensitive to the addition of new explanatory variables. In this context, the results reported in Table 4, specifically the results for CEXP in model variant 4 and EEXP in model variants Iand 3, confirm that there is a robust relationshipbetweenthe explanatory variables inthe GOVTSPENvector and economic growth and the rest of the transition economies in the sample. Moreover, we find that the statistical significanceof EEXP increases, further emphasizing the role o f government expenditure on education in economic growth. The initial results regarding the role of government capital expenditure in economic growth are also confirmed. Specifically these results seem to hold for model variant 3 in which the country dummy variable is statistically significant. The evidence therefore further underscoresthe role of government expenditure on education in economic growth, with specific reference to Tajikistan. We extendthe test for robustness by includingexplanatoryvariables from the government budget constraint. As we earlier argued, previous studies on the relationship between government spending and economic growth, which omit the government budget constraint, are potential biased. The results incorporating the variables from the Governmentbudget constraint,are reported inTable 5. It is shown in Table 5 that the effects of includingthe budget constraint, and also jointly considering significant expenditure components, improves dramatically the performance of the initial variables in vector COND and the explanatory variables in vector 134 GOVTSPEN. Inparticular,the new results identify two GOVTSPENvariables, which are important for economic growth in the sample transition economies, namely, government expenditure on education (model variant 1) and government capital expenditure (model variant 2). These results imply that the earlier findings reportedin Tables 4 and 5 for the variables in the vector GOVTSPEN are not distorted by the addition or omission of the growthof the terms oftrade (TTDE), the ratio of broadmoney (M2) to GDP (BMON) and the variables for the government budget constraint, specifically, budget deficit or surplus (FDEF) and domestic revenue (DREV). Of particular interest to this study, is that the results identify Government expenditure on education as the key sectoral allocation for public expendituresupport in order to promote economic growth, as can be notedfrom the significanceof the country dummy in modelvariant 1. 135 Table 5: Further Robustness Check on the Evidence from an Endogenous Growth Model for a Panelof CommonwealthIndependentStates (CIS) includingTajikistan Variable ModelVariant 1 ModelVariant 2 Model Variant 3 C 0.103 0.016 0.073 (3.649) 111(0.407) 111(0.015) J0.0011 [0.658] [O.OOO] NINV(-1) 0.036 0.065 (0.657) (1.431) Josi3j io.1571 DREV -0.236 -0.214 io.1341' (-1.520) I (-1.705) [0.093] HCGEN 0.866 (1.746) J0.0861 EEXP 1.033 0.321 (2.435) (1.386) FDEF 0.644 I 0.699 (3.946) (6.546) JO.OOO] [O.OOO] INFL(-1) 0.495 0.022 (1.786) CEXP FDEF (5.481) BMON (1.332) [O. 1881 BMON(-I) -0.103 (-1.714) [0.091] A2 -0.026 0.077 (-1.682) (2.235) 10.0981 [0.029] GE 0.088 (2.771) io.oo7j KA -0.021 II 0.068 (-1.400) (2.348) LO. 1671 [0.022] KY -0.037 0.038 (-2.037) (2.031) 10.0461 [0.047] AR 0.039 (3.046) [0.003] ML -0.715 0.055 -0.036 (-2.771) (1.771) (-2.672) 10.0071 [0.082] [O.0lo] M O -0.054 (-2.233) [0.029] TJ -0.058 0.033 (-3.644) (1.204) [0.001] [0.233] SSR 0.056 0.065 0.065 SER 0.03 1 0.032 0.032 ARS 0.529 0.494 0.494 136 Note: In the above regressions, the endogenous variable is GROW = Average growth rate in GDP per capita. The rest of the variables are defined as follows, as in Appendix Table A l : CEXP = Government capital expenditure (% of GDP); REXP = Government recurrent expenditure (% of GDP); EEXP = Government expenditure on education (% of GDP); HEEP = Government expenditure on health (% of GDP); DREV = Tax revenue (% of GDP); FDEF = Government surplus / deficit (% of GDP); N M V = Non Government investment (% of GDP); BMON = Broad money (M2) (% of GDP); PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; HCGEN = initial human capital (PS) variable; LFEX = Log of life expectancy; TTDE = Trade ratio (export plus import as % of GDP); ALIT = Adult literacy (as % of total population); M F L = Inflation rate, computed as the rate of change of the consumer price index; IGDPL = Log of GDP per capita at initial period. It is interestingto note from Table 5 that our results suggest that the government budget deficit or surplus as a proportion of GDP has a significant and positive impact on economic growth (model variant 2). This result is consistent with the earlier results reported in Table 4 (model variants 1-5). The evidence in both Tables 4 and 5 suggests that, on average, a one percentage point increase in the budget surplus is associated with an increase in real per capita growth rate of about 0.6 percentage points. Interestingly, although the finding is at variance with some earlier research (e.g., Bose, Haque and Osborn, 2003), it is generallyconsistent with other earlier findings which suggest that for developing countries, a one percentage point increase in the Government surplus (as a percentage of GDP) is associated with an increase in growth rate of real GDP per capita by an average of 0.15 percentage points.Our interpretationof this variable is that budget outlaysare critical intransition economies, and muchof their impact depends on how they are financedand how they are utilized. Furthermore, we test for endogeneity. It is usefulto bear inmindthe fact that government spending variables and the other economic variables in the model evolve jointly, in the sense that government expenditure variables affect economic growth, but it is also possible to experience reverse causality. To ascertain whether reverse causality is indeed the case in our model, we estimated the growth regression in equation (1) using three stage least squares (3SLS). We chose instruments by usingsome of the originalvariables and lags of the other variables. The initial variables in the vector COND enter as their own instruments because these variables are exogenous to the sample. Our results show that even after controlling for possible endogeneity or reverse causality, the signs and levelsof significanceofthe coefficientsfor the significant expenditurevariables(Le., total capitalexpenditure, expenditure in the education sector) still hold. 4.6 Country-spec$c Evidencefor Tajikistan@om Seemingly UnrelatedRegression Equations (SURE) So far, the econometric results we reported above were based on panel data, because we could not estimate the model on Tajikistan alone using annual data for 1996-2004; degrees of freedom problems would have violated the asymptotic accuracy of the estimationresults. We now proposeto tease out the coefficientestimates for each country so that we can isolate the coefficientsfor Tajikistan. We estimate and test the modelon a regional sample using the Zellner procedure of seemingly unrelatedregression equations (SURE)which allow for individualcountriesto feature inthe panelestimates. The results from SURE estimation and testing are reported for all the eight sample countries, including Tajikistan, in Table 6 for 7 countries, namely Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyz Republic, Mongolia and Tajikistan. We report further results inTable 7, this time for all the 8 countriesin the sample. 137 Table 6: SURE (ZELLNER)EstimationResults from an EndogenousGrowth Model for Tajikistan (TJ): Country-SpecificEvidence AR AZ GE KA KY MN TJ Constant 0.678 -0.118 0.315 0.246 0.164 -0.116 -0.104 (0.039) (0.464) (0.014) (-0.105) (-0.1 11) 0.932 10.9113 0.106 -0.228 0.458 0.880 0.880 (9.591) (-11.258) (4.317) (1.060) (1.060) [0.0101~ [o,i ioj [o.oooj [0.0001 [0.0001 [ o m ] 10.2891 NINV 0.794 0.063 1.170 0.145 0.319 -0.050 0.039 (7.896) (1.793) (7.369) (3.509) (-2.540) (-0.312) (0.383) [O.OOO] [0.073] [O.OOO] [O.OOO] [0.011] E0.7551 [0.702] DREV 0.924 -0.132 -2.509 1.137 -0.986 0.067 -0.278 (5.015) (-2.292) [0.0001 (-0.763) (-7.393) (12.255) (-3.595) (0.875) [0.445] JO.OOO] [O.OOO] [O.OOO] L0.3821 10.0221 CEXP -1.022 1.269 11.267 -1.379 -1.171 0.146 1.716 (-7.909) (4.157) (10.952) (-8.232) (-3.255) (0.794) (7.552) [O.OOO] [o.oooj _[O.OOO] [O.OOO] [0.001] [0.427] ~0.000] SSR 0.282 0.466 0.294 0.128 0.588 0.559 0.397 SER I 0.626 I 0.805 0.639 0.421 0.904 0.881 0.743 0.940 0.978 0.680 0.568 0.872 AR AZ GE KA KY MN TJ LFEX -0,158 -0.870 0.784 0.250 (-1.742) (-4.726) (6.227) (2.482) [0.082] [O.OOO] [O.OOO] [O.O 131 NINV 0.602 -0.064 0.175 -0.272 0.534 -0.493 (11.423) (2.534) (3.976) (-2.385) (5.568) (-5.693) [O.OOO] [0.011] [O.OOO] [0.017] [O.OOO] _[O.OOO] DREV 0.269 0.063 0.093 -1.718 0.657 (3.690) (8.395) (4.146) (-3.968) (8.441) [O.OOO] [O.OOO] [O.OOO] [O.OOO] [O.OOO] HCGEN 0.168 -0.239 0.190 0.708 0.185 (5.958) (-6.078) (6.724) (3.442) (4.699) rn nnni rn nnni rn Ann1 rn A n i i [O.OOO] 2.073 (-2.250) (9.055) (-2.030) (4.524) (-8.437) (4.4 17) [0.024] [O.OOO] [0.042] [O.OOO] [O.OOO] [O.OOO] SSR 0.423 0.270 0.578 0.221 0.551 0.284 0.569 SER 10.767 0.613 0.896 0.554 0.875 0.628 0.889 RS 0.941 0.945 0.882 0.962 0.701 0.781 0.816 Adult literacy (as % of total population); INFL = Inflation rate, computed as the rate of change of the consumer price index; IGDPL = Log of GDP per capitaat initial period. The sample countries are: AR=Armenia; AZ=Azerbaijan; GE=Georgia; KA=Kazakhstan; KY=Kyrgyz Republic; ML=Moldova; MN=Mongolia; and TJ=Tajikistan. Table 7: Further SURE (ZELLNER)EstimationResults from an Endogenous I Growth MI le1for Tr i$;tan (TJ)~: Country-! iecific Evil AR AZ KA KY TJ C -0.230 0.1116 -0.018 -0.943 0.523 0.263 (-4-641) (5.317) (39.196) (12.603) [O.OOO] [O.OOO] [O.OOO] [O.OOO] ILFEX 1.949 0.171 -0.23 1 -0.841 (9.5 14) [0.0001 (10.128) (-6.987) (-22.781) (-99.486) [O.OOO] [0I0001 [0*0001 NINV 1.460 -3.220 0.155 L 9.601 (-26.564) (-21.768) (7.245) [O.OOO] [O.OOO] [O.OOO] DREV 0.385 -0.237 -0.147 (9.926) (-4.93 1) (-40.050) (-27.319) [O.OOO] [O.OOO] [O.OOO [O.OOO] HCGEN -4.405 -2.189 -0.320 -2.961 2.824 (-18.524) (-4.375) (-8.842) (-15.728) (18.443) [O.OOO] [O.OOO] [O.OOO] [O.OOO] [O.OOO] EEXP 1.251 1.460 -0.459 1.152 0.1332 4.113 (23.482) 9.601 (-3.554) (40.659) (6.890) (45.495) FDEF 1 i0.0001 [O.OOO] y.go] 1 [O.OOO] [O.OOO] [O.OOO] -0.908 2.413 -0.274 (-4.920) (71.863) (24.067) (21.891) (-5.643) ~0.000] [0.000] 0.985 -0.479 0.163 0.136 (7.163) (163.688) (-11.717) (13.210) (29.862) [O.OOO] [O.OOO] 0.512 0.227 (13.242) (15.310) I [O.OOO] I [0.000] ROW = Average growth rate in GDP per capita. The rest of the variables aredefined as follows, as in Appendix TableA I : CEXP = Government capital expenditure (% of GDP); REXP = Government recurrent expenditure (% of GDP); EEXP = Governmentexpenditure on education(% o f GDP); HEEP = Government expenditure on health (% of GDP); DREV = Tax revenue (% of GDP); FDEF = Government surplus / deficit (% of GDP); NINV = Private investment (% of GDP); BMON = Broad money (M2) (% of GDP); PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; HCGEN = initial human capital (PS) variable; LFEX = Log of life expectancy; TTDE = Trade ratio (export plus import as % of GDP); ALIT = Adult literacy (as % of total population); INFL = Inflation rate, computed as the rate of change of the consumer price index; IGDPL = Log of GDP per capitaat initial period. Table 6 reports the ZELLNER results for explanatory variables from vectors COND and GOVTSPEN. In general, the main results for Tajikistan are clear. First, the evidence suggests that government expenditure in education is positively and significantly correlatedwith economic growth in Tajikistan(Table 6, modelvariant B, column 8). The 139 evidence which supports the positive role of education in economic growth is also found for the human capital composite variable in Tajikistan, which by construction encompasses primary and secondary schooling (Table 6, model variant B, column 8). Second, it is shown that the share of government capitalexpenditure in GDP is positively and significantly correlated with economic growth (column 8 of Table 6, model variant A). The ZELLNER estimation results in Tables 6 are augmented with explanatory variables from the vector OTHER and from the government budget constraint and the results are reportedinTable 7. Ingeneral, the results for Tajikistanreportedin Table 7 show that the two main government expenditure variables in the GOVTSPEN vector, Government spending on education and Government capital expenditure, are able to meet the robustness test Le., they are robust even if other explanatory variables from the vector OTHERare includedinthe model. It is shown in Table 7 that all of the explanatory variables in the COND vector are statistically significant for Tajikistan. Importantly, the non government investment variable shows a significant relationshipwith economic growth in Tajikistan. This result is consistent with a number of previous empirical studies (e.g., Levine and Renelt 1992). The results are also consistent with the basic predictionof the neoclassical growth theory. The evidence also shows that the trade share o f GDP has a significant positiverelationship with economic growth. Further, it is found that the ratio of broad money to GDP is positively and significantly related to economic growth in Tajikistan. The domestic revenue and fiscal deficit variables have mixed results, but in most cases conform to the role of the budgetconstraint. 5. Conclusionand Recommendationsfor Tajikistan PublicExpenditureReview 5.I Summary and Conclusions This study has focused on identifying the elements of public expenditure that have a significant relationship with economic growth, with respect to the CIS-7 transition economies, and with special reference to Tajikistan. Our study is innovativeinbuildingon the theoreticalliteratureon government expenditure and endogenous growth by proposing and implementing a novel way of estimating an endogenous growth model on Tajikistan and 7 other transition economies. We depart from most of previousresearchindevelopingcountries by consideringthe implicationsof the government budget constraint. We also depart from previous work in this area by focusingon the transitioneconomies. Interms of econometric methodology, first, we estimatedthe modelon a panel dataset of transition economies, namely Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyz Republic, Moldova, Mongolia and Tajikistan for the period 1996 - 2004, hence 72 observations. We run the base regressions in order to establish whether the government spending variables have significant impact on growth after controlling for the CONDS variables. Basically, we run a restricted version of the model in equation (l), where the gamma coefficientsare constrained to be zero. The regressions are run in such a way that we add one GOVTSPEN variable at a time. These initial regressions are intended to capture a central idea of the endogenous growth literaturenamely that human capital and institutional factors are important determinants of economic growth. In addition by includinginitial GDP in these initial regressions, we are trying to explore the argument o f 140 possible effects of convergence on output growth in the sample economies. We then conducted checks for the robustness o f our initial findings by addingto the model vector OTHER, to obtain the augmented model. Further,since we could not estimate the model on Tajikistanalone usingannual data for 1996-2005because degreesof freedom problems would violate the asymptotic accuracy of the estimation results, we tried to tease out the coefficientsfor each country so that we can isolate the coefficientsfor Tajikistan. Hence, we estimated and tested the model on a regional sample using the Zellner procedure (SURE,or seemingly unrelatedregressionequations) which allow for individualcountries to feature in the panel estimates. We also checked for robustness within the Zellner procedure. The evidence from the study is unequivocal in supportingthe prevalent view in modem growth theory that education is an important key to economic growth. Hence, the most importantfinding of this study, from an empirical as well as a policy perspective, is that education is the key sector to which public expenditure should be directed in order to promote economic growth in Tajikistan. From an econometric perspective, this finding is important and reinforcesearlier findings by Bose, Haque and Osborn (2003) whose study focused on developing economies. In addition, the study is important in providing evidence which counters the earlier results by Landau(1986) and Devarajanet al. (1996), who find negative or insignificant positive effects of educationexpenditureon growth for a sample of developing economies. We argue that the results of our study and those obtained by Bose, Haque and Osborn (2003) are more plausible because they incorporate the government budget constraint. Moreover, our study is more plausible because it reflects the economic structure of Tajikistan and the neighboring transition economies which form our sample. We also find that the share of Government capital expenditure in GDP is positively and significantly correlated with economic growth. This result provides the basis for Ggovernment intervention in supporting public investments in key sectors such as transport and communication in order to achieve economic growth. However, this variable seems to be conditionalon the non-government investment variable (which is a proxy for private investment). In fact, the non-Government investment share of GDP is associated with economic growth in a significant and positive manner, especially in the Zellner estimates for Tajikistan. Also, we find strong evidence to support humancapital variables in the 8 countriesand in Tajikistan. In particular, the initial human capital variable, expressed as enrolment in primary and secondary schools was consistently significant. This suggests the need for donors and gvernments to support the fundingof primary and secondary schooleducation. Also, life expectancy is consistentlysignificantin explainingeconomic growth in all the 8 countries, and in Tajikisatn, in particular. This finding lends support to increased expendituresto support healthand sanitationin Tajikistan. Finally, we must acknowledge the limitations of our small sample, although the data we used are the most comprehensive database available at the moment for Tajikistan and her regionalneighbors. 5.2Recommendationsfor Programmatic Public Expenditure Review for Tajikistan In terms of recommendations for the PPER, the evidence from this study gives rise to information that is particularly useful for the Tajik government, which is resource 141 constrained and where the allocationof limitedpublic resources betweenthe sectors is an issue of paramount importance. First, the empirical finding which unequivocally supports the association between government expenditure on education and economic growth is importantfor Tajikistan in the sense that the finding underscoresthe argument that the priority should be to allocate scarce government resources towards the education sector. The results are also important in that they provide evidence to support the argument that expenditure on education should be increased. Second, the evidence supports the human capital argument in the endogenous growth story, especially in terms of the combination of primary school enrolment rates and secondary school enrollment rates in the 8 transition economies and in Tajikistan. This evidence lends support to the case made above for public expenditureallocation in favor of education and services in the education sector. But, the evidence on human capital variables goes a bit further than that. Specifically, the evidence which supports strong association between life expectancy and economic growth in the sample economics in general and in Tajikistan in particular, lends support to increased expenditures to support healthand sanitationinTajikistan. Third, the finding that the Government budget surplus as a proportion of GDP has a significant and positive impact on economic growth is important, given the magnitude of the impact: on average, a one percentage point increase in the budget surplus is associated with an increase in real per capita growth rate of about 0.6 percentage points. Our recommendation is based on our interpretationof this variable, which is that the budget outlays are critical intransitioneconomies, and much of their impact depends on how they are financed and how they are utilised. In this context, development assistance can be an important part of budget support, specifically targeted to expenditures in the social services sector such as education and health. Fourth, we recommend the use of private capital, in addition to the current capital expenditures, perhaps in a public-private partnership. Broadly, our evidence that the share of government capital expenditure in GDP is positively and significantly correlated with economic growth, and the rest of the sampletransitioneconomies, providesthe basis for Government intervention in supporting public investments and expenditures in other key sectors such as transport and communication in order to achieve economic growth. However, this variable seems to be conditional on the non government investment variable in Tajikistan. Infact, the non government investment share of GDP is associated with economic growth in a significant and positive manner, especially for Tajikistan. Hence, this finding may be interpreted to suggest that while government capital expenditure is associatedwith economic growth, the financingthe constructionof roads, bridges and other capitalinfrastructureprojectsshould at least involve the private sector, perhaps in a public-private partnership, rather than being exclusively the reserve of government spending, especially in: strategically targeted new investment in infrastructureprojectsand rehabilitationof existing infrastructurenetwork; Fifth, we recommend budget support for the Tajikistan government by international donors in order to deliver an enabling macroeconomic environment that fosters long term sustainable growth. This recommendation is based on the evidence from this study which underlinesthe role of macroeconomic factors, including broad money, domestic revenue and budget discipline. 142 Finally, the targeting o f government expenditure towards capital expenditure in say infrastructure and transport, education and health should bear in mind the institutional, regionaland historicalfactors of Tajikistan. 143 ources, 1996-2004 data source I World Bank and IMF Country Reports World Bank website CEXP Government capital expenditure (% of GDP) World Bank and IMF Country Reports WorldBank website EEXP Government expenditure on education(% of GDP) World Bank and IMF Country Reports World,Bank website HEEP Government expenditure on health(% of GDP) World Bank and IMF Country Reports World Bank website DREV Tax revenue (% of GDP) World Bank and IMF Country Reports World Bank website FDEF Government surplus/ deficit (% of GDP) World Bank and IMF Country Reports World Bank website NmV Private investment (% of GDP) World Bank and IMF Country Reports World Bank website BMON Broad money(M2) (% of GDP) World Bank and IMF Country Reports World Bank website PRIM Primary schoolenrolmentratio World Bank and IMF Country Reports World Bank website SECS Secondary school enrolmentratio World Bank and IMF Country Reports World Bank website HCGEN Initial human capital (PS) variable Generated from PRIM and SECS- see notes below LFEX Log of life expectancy World Bank and IMF Country Reports World Bank website TTDE Trade ratio (export plus import as % of GDP) World Bank and IMF Country Reports World Bank website ALIT I Adult literacy (as % of total population) World Bank and IMF Country Reports Notes:GROW Averagegrowth rate inGDP per capita. The rest o the variablesare definedas follows, as = in Appendix Table Al: CEXP = Government capital expenditure (% of GDP); REXP = Government recurrent expenditure (% of GDP); EEXP = Government expenditure on education (% of GDP); HEEP = Government expenditure on health (% of GDP); DREV = Tax revenue (% of GDP); FDEF = Government surplus / deficit (% of GDP); NINV = Private investment(% of GDP); BMON = Broad money (M2) (% of GDP); PRIM = Primary school enrolment ratio; SECS = Secondary school enrolment ratio; HCGEN = initial human capital (PS) variable; LFEX = Log of life expectancy; TTDE = Trade ratio (export plus import as % of GDP); ALIT = Adult literacy (as % of total population); INFL = Inflation rate, computedas the rate of change of the consumer price index; IGDPL Log of GDP per capita at initial period. = HCGEN, the initial human capital variable is constructed in the context of Landau (1983) as the weighted sum of the initial enrolment ratios (%) in primary and secondary schools and in higher education. The weights are 1 for primary school enrolment ratio and 2 for secondary school enrolment ratio, hence, these weights are approximations to the relative values of two types of education. We use the HCGEN variable because o f the high milticollinearity between the separate enrolment rates of PRIM and SECS. 144 APPENDIX A: Patternsof Government Expenditureand Economic GrowthVariables in CIS, includingTaiikistan Figure 1:Patternsof GovernmentRecurrentExpenditure(REXP)and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 Growth Vx REXP 0.500 0.350 0.300 0.250 0.204 + a \.I50 + 0.100 0.050 - I - - * * I I -0.100 0.000 0.100 0.200 Growth 145 Figure2: Patternsof GovernmentCapital Expenditure(CEXP) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 Growth Vs CEXP 0.450 - 0.400 - 0.350 - 0.300 - 0.250 - IccExp] -0.10 -0.05 0.000 0.050 0.100 0.150 0.200 0 0 Growth 146 Figure3: Patternsof DomesticRevenue @REV) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 Growth Vs DRE V 0.500 0.450 + + + $ +$ + + + + 0.050 - I n nnn I 1 -0.100 0.000 0.100 0.200 Growth 147 Figure4:Patterns of Net Investment(NINV) and Economic Growth (GROWTH) in CommonwealthIndependent States (CIS), 1996-2005 iGrowth VsNINV 0.500 4 0.450 4 0.400 0.350 4 4 0.300 0.250 4% 'p4 4 r n n -** 4 I I -0.100 0.000 0.100 0.200 Growth 148 Figure5:Patterns of Broad Money (BMON) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 0.500 1 + 0.400 0.300 -/ 0.250 + + [*BMON] ++ +++ + 0.050 I nnnn I I I -0.100 0.000 0.100 0.200 Growth 149 Figure6:Patterns of Terms of Trade (TTDE)and Economic Growth (GROWTH) in CommonwealthIndependent States (CIS), 1996-2005 Growth Vs TTDE 5.000 4.500 4.000 3.500 3.000 2.500 2.000 I - - Y A . Y I I -0.100 0.000 0.100 0.200 Growth 150 Figure7:Patterns of Inflation (INFL)and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 5.000 3.000 2.000 I +INFLI 1.000 -0.100 0.c00 0.100 0.200 -1.000 - Growth 151 Figure8:Patterns of GovernmentExpenditureon Health (HEEP) and Economic Growth (GROWTH) in CommonwealthIndependent States (CIS), 1996-2005 Growth Vs HEEP 0.5 - 0.45 - 0.4 - 0.35 - El8 0.3 - 0.25 - 0.2 - -0.100 0.000 0.100 0.200 Growth 152 Figure9:Patterns of Life GovernmentExpenditureon Education(EEXP) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996- 2005 Growth Vs EEXP Bw 0.3 - 0.25 - 0.2 - 0.15 - + n 1 - + -0.100 0.000 0.100 0.200 Growth 153 Figure 10:Patterns of Primary SchoolEducation(PRIM) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 120 100 sd E %; 60 PRIM 40 20 I -0.100 0.000 0.100 0.200 G r O W t h 154 Figure 1l:Patterns of Secondary School Education (LFEX) and Economic Growth (GROWTH) in CommonwealthIndependentStates (CIS), 1996-2005 Growth VsSECS 100 I 2 O 1 -0.100 0.000 0.100 0.200 Growth 155 Annex 11. Expendituresof the State budget of the Republicof Tajikistan 2005-2010 BudgetItems Total Revenuesand Granb 1,256.0 1,414.5 1,510.1 1,823.5 1,847.7 2goo.o 2,730.0 3,319.2 3,992.0 in percent to GDP 17.7 19.6 18.0 19.7 19.3 21.2 21.3 22.I 22.5 TotalGeneralGovernmentExDenditures 1.729.9 1.712.7 1.936.7 2.126.8 3.290.9 3.562.9 4.095.5 4,352.0 4.633.1 in percent to GDP 24.4 23.8 23.I 22.9 34.5 32.8 32.0 29.0 26.2 including: -state budget: 1,290.7 1,3 13.4 1,541.7 1,643.4 1,943.2 2,215 1 2,857.9 3,469.6 4,169 1 - wages - Centralized and salaries 274.6 269.I 378.0 381.3 474.0 483.I 693.0 0.0 0.0 State Investment Programme (CSIP) 137.2 238.4 192.2 2611.8 245.3 349.7 298.0 342.7 394.I --- extra-budgetary Public Investment Program(PIP) 350.3 310.2 329.5 348.0 489.7 489 7 547.0 344.0 304.0 State Loan(Chinese) 0.0 0.0 0.0 0.0 778.5 778.5 570.4 398.4 (special) funds 88.9 89.1 65.6 135.4 79.5 79.5 120.2 140.0 160.0 1. Governanceand Public Administration 198.8 243.2 208.6 292.9 260.0 293.3 319.4 353.0 406.3 in percent to G D P 2.8 3.4 2.5 3.2 2.7 2.7 2.5 2.3 2.3 -including: state budget: 160.0 204.4 180.7 225.7 223.4 256.7 275.0 330.1 396.1 -- wages and salaries 33.9 32.6 50.2 46.2 64.3 64.3 91.2 Centralized State Investment Programme (CSIP) 32.2 75.6 40.I 66.9 38.5 99.0 38.5 42.0 48.5 -- extra-budgetary Public Investment Program(PIP) 1.6 1.7 2.0 0.7 24.3 24.3 23.2 22.9 10.2 (special) funds 37.2 37.2 25.9 66.5 12.3 12.3 21.2 2. Defense 102.4 96.3 124.9 105.0 149.2 157.4 194.0 232.9 279.4 in percent to GDP 1.4 1.3 1.5 1.I 1.6 1.4 1.5 1.5 1.6 including - state budget 102 4 96 3 124.9 103.8 149.2 157.4 194.0 232.9 279.4 -- wagesand salaries 18 I 16.2 26.9 27.6 33.0 33.0 41.0 CentralizedState Investment Programme (CSIP) 4 0 15.8 4.5 11.9 4.9 9.5 5.9 7.5 8.7 -- extra-budgetary(special) Public InvestmentProgram(PIP) 0 0 0 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 funds 0 0 0.0 0.0 I .2 0.0 0.0 0.0- 3. Law-EnforcementAgencies 111.0 115.1 131.0 138.3 161.0 168.6 216.7 216.1 259.2 in percent to GDP 1.6 1.6 1.6 1.5 1.7 1.6 1.7 1.4 1.5 including - state budget 91 9 96 1 113.8 117.9 138 2 145.9 180.0 216.1 259.2 -- wages and salaries 32 0 32.2 49.9 53.6 58.6 58.6 73.0 Centralized State Investment Programme (CSIP) 4.7 7.9 7.3 6.4 7 5 13.0 9.2 12.5 15.0 -Public Investment Program(PIP) 0 0 0 0 0 0 0.0 0 0 0.0 0.0 0.0 0.0 - extra-budgetary (special) funds 19 I 19 1 17.2 20.4 22 8 22.8 36.7 SOCIAL BLOC 665.4 676.1 853.9 907.2 1,103.0 1,219.1 1,477.6-1,588.9 1,777.8 in percent to G D P 9 4 9.4 10.2 9.8 11.6 11.2 11.6 10.6 10.0 156 including: - state budget: 600.2 609.8 764.1 801.8 961.4 1,077.5 1,297.0 1,501.6 1,756.4 - wages - Centralized and salaries 179.5 174.8 237.1 231.2 299.I 308.I 448.3 0.0 0.0 State Investment Programme (CSIP) 20.2 31.0 52.6 39.9 69.3 97.7 97.6 120.5 137.1 -- extra-budgetary Public Investment Program (PIP) 41.5 42.6 71.4 73.4 108.9 108.9 132.4 87.3 21.4 (special) funds 23.6 23.7 18.4 32.0 32.8 32.8 48.1 0.0 0 0 4. Education 281.6 290.2 381.0 372.7 506.4 552.9 689.7 705.7 820.7 inpercent to GDP 4.0 4.0 4.5 4.0 5.4 4.7 4.6 including: - state -- wages budget: 244.8 253.1 332.5 317.7 428.5 475.1 562.0 674.6 809.4 andsalaries 139.2 137.9 189.3 183.7 237.1 246.I 308.8 Centralized State Investmenr Programme (CSIP) 10.9 20.0 26.4 16.9 39.3 47.7 55.0 60.0 60.0 - Public Investment Program(PIP) 18.3 18.5 33.1 30.9 49.3 49.3 85.7 31.1 11.3 - extra-budgetary (special) funds 18.5 18.6 15.4 24.1 28.6 28.6 42.0 5. Health 97.2 90.4 117.9 125.3 178.3 189.7 254.8 311.9 316.9 in percent to GDP 1.4 1.3 1.4 1.4 2.0 2.I 1.8 including: - state budget: 89.3 82.4 105.7 105.4 131.0 142.4 213.0 255.7 306.8 -- wages and salaries 29.9 29.0 39.5 38.9 51.2 51.2 122.6 Cennalized State Investment Programme (CSIP) 6.8 8.2 11.8 9.I 13.0 18.8 16.0 23.0 30.5 -- extra-budgetary Public Investment Program(PP) 5.7 5.8 11.1 17.7 47.2 47.2 41.4 56.2 10.1 (special) funds 2.2 2.2 1.1 2.2 0.027 0.0 0.4 6. Social Insurance and Social Protection 245.8 250.8 295.7 342.1 344.1 382.1 435.8 461.5 508.4 in percent to GDP 3.5 3.5 3.5 3.7 3.4 3.I 2.9 including - state budget 228.2 232.4 268.5 3170 331 7 369.7 430.5 461.5 508.4 -- wages and salaries 5.3 2.8 1.9 1.8 3.0 3.0 4.8 Centralized State Investment Programme (CSIP) 0.7 0.7 0.9 1.9 2.9 8.6 9.7 16.0 21.0 -- extra-budgetary Public Investment Program(PIP) . 17.5 18.3 27.2 24.8 12.4 12.4 5.3 (special) funds 0.1 0.1 0.0 0.3 0.01 0.0 8. Cultureand Sport 40.7 44.7 59.3 67.0 97.2 109.8 131.8 inpercent IO GDP 0.6 0.6 0.7 0.7 0.8 0.7 0.7 including: - state budget: 37.9 41.9 57.4 61.7 70.2 90.3 91.5 109.8 131.8 - wages and salaries - Centralized 5.I 5.I 6.4 6.8 7.8 7.8 12.2 State Investment Programme (CSlP) 1.8 2.I 13.5 12.0 14.1 22.6 16.9 21.5 25.6 -- extra-budgetary Public Investment Program (PIP) 0.0 0.0 0.0 0.0 0.0 0.0 (special) funds 2.9 2.9 I.9 5.3 4.I 4.I 5.7 ECONOMIC BLOC 522.9 517.8 493.2 605.0 1,452.1 3,196.4 1,363.8 -1,098.0 836.2 in percent to GDP 7.4 7.2 5.9 6.5 10.7 7.3 4.7 including - state budget 210.8 246.7 233.7 319.8 391 5 469.9 563.8 -- wages and salaries 7.2 13.4 13.9 22.7 39.5 Centralized State Investment Programme (CSIP) 75.2 106.9 84.7 133.2 125.I 130 6 138.3 141.3 161.I - Public Investment Program (PIP) 306.1 265.0 255.4 273.4 350 0 350 0 387 6 229.7 272.3 157 -- extra-budgetary State Loan(Chinese) 0.0 0.0 0.0 778.5 778.5 570.4 398.4 0.0 (special) funds 6.0 6.0 :::1 11.8 11.7 11.7 14.2 0.0 0.0 7. Housing and Communal Services 97.6 120.9 125.2 139.7 196.3 148.7 141.2 155.0 in percent to GDP 1.4 I.7 1.5 1.8 1.2 0.9 0.9 including: -state budget: 55.9 80.2 65.0 101.6 82.8 139.4 IO7 6 129.2 155.0 - wages and salaries 1.6 4.2 8.6 9.4 11.2 11.2 16.6 29.8 19.2 40.6 23.7 24.9 28.5 ~ Centralized State Investment Programme (CSIP) 34.0 40.0 -- extra-budgetary(special) Public InvestmentProgram(PIP) 39.1 38.1 19.7 19.6 53.7 53.7 41 1 12 I funds 2.6 2.6 0.0 4.0 3.2 3.2 9. Fuel and Energy Complex 97.5 103.2 164.2I 175.6 571.8 571.8 350.7 408.6 115.7 inpercent to GDP 1.4 1.4 2.0 1.9 6.0 5.3 2.7 2.7 0.7 including: - state budget: 15.1 17.7 25.2 38.2 45.2 45.2 60.0 72.0 86.4 -- wages andsalaries 0.0 0.0 0.0 0.0 0.4 0.4 Centralized State Investment Programme (CSIP) 14.2 16.7 24.2 20.8 43.2 39.I 55.3 42.0 45.0 --- extra-budgetary(special) Public Investment Program(PIP) 82.4 85.5 139.0 137.4 104 5 104.5 83.9 27.2 29.3 State Loan (Chinese) 0.0 0.0 0.0 0.0 422.1 422.1 206.9 309.4 funds 0.0 0.0 0.0 0.0 0 0 0.0 IO. Agriculture, Fishery, and Hunting 89.3 84.3 111.0I 102.6 176.3 184.3 232.9 206.9 254.3 inpercent to G D P 1.3 1.2 1.3 1.1 1.8 1.7 1.8 1.4 1.4 including: - state budget: 41.4 37.9 49.7 44.4 57.4 65.4 73.5 88.2 105.9 -- wages andsalaries 4.3 4.3 3.2 6.6 4.7 4.7 CentralizedState Investment Programme (CSIP) 8.I 11.1 8.8 10.7 10.5 18.1 12.6 15.4 17.8 -Public InvestmentProgram(PIP) 46.4 44.8 60.8 56.4 116.4 116.4 151.6 118.6 148.4 - extra-budgetary (special) funds 1.5 1.5 0.5 1.8 2.5 2.5 7.7 11. lndustrv and Construction 18.2 23.7 15.7I 45.2 20.8 21.5 27.8 31.7 38.0 in percent to GDP 0.3 0.3 0 2 0 5 0.2 0.2 0.2 0.2 0.2 including - state budget 17.7 23 2 15 3 44 6 20.3 21 0 26.4 3 1 7 38 0 -- wages and salaries 0.3 0.3 0 5 0 5 0.7 0.7 Centralized State Investment Programme (CSIP) 6.3 19.7 8 4 41.1 15.4 17.7 18.7 22.6 26.0 -- extra-budgetary(special) Public InvestmentProgram(PIP) 0.0 0.0 0 0 0 0 0.0 0.0 0.0 0.0 0.0 funds 0.5 0.5 0 4 0 6 0.5 0.5 1.4- 12. Transoort 215.1 176.1 112.9I 151.9 536.1 539.2 594.7 298.9 260.2 inpercent IO GDP 3.0 2.4 1.3 1.6 5.6 5.0 4.6 2.0 1.5 including: - state budget: 76.1 78.8 74.8 87.2 99.8 102.9 115.0 138 0 165.6 - wages and salaries 0.3 4.0 0.9 5.3 1.2 1.2 30.0 29.6 24.2 19.9 30.2 ~ Centralized State Investment programme (CSIP) 31.3 22.0 25.3 29.5 --- extra-budgetary(special) Public Investment Program (PIP) 138.3 96.6 35.9 60.0 75.4 75.4 111.0 71 9 94.6 State Loan (Chinese) 0.0 0 0 0 0 0.0 356.4 356.4 363.6 89 0 funds 0.6 0.6 2.2 4.6 4.6 4.6 5.1 13. Other Economic Services 5.4 9.7 4.7 I 4.5 7.4 8.0 9.0 10.8 13.0 158 in Dercent to GDP 0.1 I 0.I 0.I 0.0 0.I 0.I 0.I 0.I 0.I including: - state budget: 3.7 3.8 6.4 7.0 9.0 10.8 13.0 -- wages and salaries 0.8 0.8 0.9 0.9 Centralized State Investment Programme (CSIP) 0.8 0.0 0.6 1.2 2.8 ~ 0.0 1.0 2.0 -- extra-budgetary Public Investment Program (PIP) 0.0 0.0 0.0 0.0 (special) funds 1.o 0.7 1.o 1.o 14. Other Expenditures 129.5 125.1 78.4 165.6 203.3 524.2 723.1 914.2 in percent to GDP 1.5 0.8 1.7 1.9 4.I 4.8 5.2 including - state budget 124.5 74.4 159.I 196.8 520.4 719.0 914.2 -- wages and salaries 0.0 0.4 0.0 0.0 Centralized Slate Investment Programme (CSIP) 2.9 2.5 0.0 0.0 8.5 18.9 23.7 -- extra-budgetary Public Investment Program (PIP) 0.6 0.5 6.5 6.5 3.8 4.1 (special) funds 3 0 3 1 0.0 3.5 0.0 0.0 - - Budget DeficiUSurplus -34.7 101.1 -31.6 180.0 -95.5 84.9 -127.9 -150.3 -177.1 inpercent to GDP -0.4 1.9 -1.0 0.8 -1.0 -1.0 -1.0 GDP 8,400.0 9,272.1 9,550.0 10,865.0 12,790.0 15.030. 0 17,710.0 1 percent of GDP 71.0 I 72.0 84.0 92.7 95.5 108.7 127.9 150.3 177.1 159