Decentralized Public Service and Financial Management Sector Development Program (RRP CAM 41392-013) PROGRAM IMPACT ASSESSMENT Decentralized Public Service and Financial Management Subprogram 1 I. Introduction 1. The ADB s Staff Operating Manual (OM) requires proposed Programs to properly justify the loan amount suggested. 1 The impact assessment should attempt to ex ante estimate the costs and benefits of the proposed Program as well as discuss the importance of the sector and adjustment costs as required. This Program Impact Assessment (PIA) presents a methodology for assessing the impact of the proposed first subprogram of the Decentralized Public Service and Financial Management (DPSFM) sector development program on stakeholders. 2. This impact assessment estimates the reform program adds approximately $300 million to GDP, equivalent to 2.5% using 2010 levels. The benefits arise from improved state allocative and operational efficiencies. The costs of the reform program in terms of government administrative costs and fiscal costs of implementing reforms amount to around $140 million, thus the Program providing a net benefit to the economy of some $160 million. 3. In line with best international practices and prior PIAs, this section applies, to the evaluation of costs and benefits of the proposed program, a four-step methodology including: a) the definition of the problem (Section II); b) the definition of the impact and outcomes of the proposed program (Section III); c) a range of options considered for achieving the defined impacts and outcomes (Section IV); d) the assessment of costs and benefits (Section V). Section VI summarizes the adequacy of the program financing proposed. II. Definition of the problem 4. Uneven distribution of economic growth and large disparities in local economic development. Despite robust growth performance over the decade leading into the global financial crisis, large disparities in living standards remain between urban and rural areas. GDP growth averaged over 10% between 2003 and 2007 and, after falling to below 1% in 2009. It has again reached healthy levels of nearly 6% in 2010. Cambodia has however anchored this impressive growth upon a narrow economic base, where four sectors (garments, tourism, construction and agriculture) make up for most of the national income. 2 The concentration of Cambodia s economic base in a limited number of sectors goes a long way in explaining geographical differences in development and living standards across the country. 5. Economic growth has ushered in a substantial reduction of poverty rates, from 47 per cent in 1993 to 30 per cent in 2007. However, in a pattern common to other Southeast Asian nations, the reduction on poverty rates has come accompanied of significant increases in income inequality. During the same period, the Gini coefficient shot from 0.35 to 0.43, high for the country s per capita income levels. Gains against poverty have been unevenly distributed geographically. The poverty headcount in urban areas dropped on average from 25% in 2004 to 22% in 2007 (the headcount was a mere 1% in Phnom Penh in 2007). In comparison, some 35% of the population was poor in rural areas in 2007, down from 40% in 2004. Disparities are 1 OM Section D4/BP issued in 7 April 2010. 2 ADB. 2012. Cambodia: Country Poverty Analysis. Manila.
2 greater if geographical regions are analyzed separately. For instance, both the Plateau/Mountain Region (52%) and the Tonle Sap Region (43%) largely exceeded national poverty headcount averages. 6. Even greater disparities are observed in access to basic social services. The percentage of families with access to safe drinking water varies from 92% in Phnom Penh, to less than 30% in Mondul Kiri Province. The average distance from a village to a health center in Stung Treng Province is 25 km, compared with a national average of less than 5 km. Largely as a result of limited access to services, human development indicators for rural areas lag behind those for urban areas. Infant mortality rates are about three times higher in rural areas (64 deaths per 1000 children) than urban areas (22 deaths). 3 7. Inefficiencies in the delivery of basic services partly contribute to the disparities across regions. Improved tax revenue collection from economic growth has allowed important increases in public expenditure on basic services such as education, health, or infrastructure. However, despite the improvement in available resources, service delivery systems remain weak. The inefficient structure of intergovernmental fiscal relations, characterized by unclear functional assignments between the national and the subnational levels of administrations (SNAs), and a largely underdeveloped financing system for SNAs, partly accounts for the observed low performance. In addition, there is an absence of accountability mechanisms between the different levels of administration and their constituencies. As a result, beyond the regular elections, there are few possibilities for the voters to convey their preferences and needs in terms of service delivery. 8. The government s National Program for Democratic Development at the Sub-national Level (NP) aims to address this challenge by developing the management systems of SNAs (with an emphasis at the District/Municipal level) based on the principles of democratic participation. Improvements to the transparency and accountability of the systems can promote local development and delivery of public services to meet the needs of citizens and contribute to poverty reduction within the respective territories. The proposed reforms implicitly require a redesign of the system of intergovernmental fiscal relations, from the reassignment of service delivery responsibilities across levels of administration, to the definition of adequate and efficient sources of revenues for SNAs. 9. Fledging subnational administrations and central government agencies will require large investments in capacity and institutional development. The comprehensive decentralization and deconcentrations reforms outlined by the government in the NP will demand the development of: (i) oversight and monitoring capacity at the central government agencies to ensure efficient implementation of reforms and the refinement of policy proposals; and (ii) the development of capacities at the sub-national level of administration for public financial management. 10. Decentralization can assist economic growth and poverty reduction through a variety of channels: (i) higher economic growth through allocative efficiency in the delivery of public services to the poor; (ii) productive efficiency in service delivery, and (iii) providing incentives for improved revenue collection and public expenditure. The first channel is through allocative efficiency. Efficiency gains in public spending from reassignment of expenditure functions to the local government may boost growth and reduce inequality. This 3 Government of Cambodia 2010. National Committee for Democratic Development at the Subnational Level. Commune Database Indicators. Phnom Penh.
3 occurs because local governments are often better placed to identify local spending priorities and more likely to procure products and services locally compared with central governments. Evidence from international experiences show that the share of education and health in total local government expenditures increases with decentralization. Local economic development can also be assisted if local governments increase the scope and improve the quality of public infrastructure spending. In Cambodia, the generalized use of the Commune/Sangkat fund on rural road rehabilitation signals such a preference. As participatory and accountability mechanisms between local administrations and their constituencies are developed, decentralization may improve targeting in the delivery of public services to the poor. The magnitude of the impact on poor will depend on the pro-poor elements of the local government budgets, the quality in the design of national anti-poverty programs and the development of efficient local public financial management systems. 11. The second channel is through the impact of potentially higher productive efficiency in the delivery of basic services at the local level. Fiscal autonomy can lead to higher output per unit of labor, thus leading to economic growth. The use of local contractors, and local materials, for the delivery of local services may offer important savings in public expenditure. Finally, the third channel refers to the incentives for revenue mobilization (and thus local public expenditure) from the assignment of own revenue sources to SNAs. The reason is that local populations may be more willing to contribute from tax and non-tax sources if they know those revenues will be used for local service delivery. III. Impact and Outcome of the Sector Development Program 12. The expected impact of the subprogram is improved access and coverage of basic services across the country. The expected outcome is elected subnational administrations increasingly taking on responsibilities for equitable basic service and infrastructure delivery. The proposed program includes a policy-based loan, and a project loan with two components: assistance to the construction of rural district offices, and institutional and capacity development for local public financial management. Subprogram 1 of the policy-based loan has XX reform actions, which the government completed between 2010 and 2012. IV. Options to address the sector s challenges 13. The policy-based loan is structured around seven major outputs: (i) the development of a gender sensitive policy and regulatory framework for decentralization; (ii) the establishment and strengthening of the NCDD-S for efficient coordination of decentralization and deconcentration policies, supporting adequate female participation; (iii) the implementation of the basic pillars of a system of intergovernmental fiscal relations; (iv) the development of an efficient system of decentralized public financial management; and (v) the establishment of an accountability mechanism in the SNAs that favors the participation of women. Each of the outputs includes triggers and milestones that are drawn from the 3 year Implementation Plan for Sub-national Democratic Development (IP3). 14. The project loan has two main outputs: (i) construction of up to 40 rural district offices, and (ii) institutional and capacity development for local public financial management. Construction of district offices in rural areas. The project will provide selected rural districts with basic premises and equipment. The districts receiving this assistance will be selected on the basis of need, using poverty, population, and geographical considerations as selection criteria. The project will not entail any land acquisition. Only public land, upon presentation of the land title, will be used for construction. Construction will still require the preparation, for the
4 selected locations, of a due diligence report, using ADB standard checklists, discussing the potential impact on land acquisition and resettlement. Institutional and capacity development for local public financial management. The outputs of this component are: (i) enhanced oversight, monitoring and support capacity at the Ministry of Economy and Finance and relevant provincial departments, and (ii) development and strengthening of an efficient system of intergovernmental fiscal relations. This project component will be implemented with intermittent inputs over a period of twenty-four (24) months starting in January 2013, and the estimated value of the component is $1.5 million. 15. In this program impact assessment, we focus on the analysis of the costs and benefits derived from the reforms incorporated on the policy-based component of the sector development program loan. V. Impact Analysis 16. This section presents estimates of the economy-wide benefits and costs of the DPSFM Sector Development Program, subprogram 1. The assessment of net benefits associated to the package of reforms and the project components contained in this program would ideally require the consideration of both the static and dynamic dimensions of the reform. That is the first and subsequent rounds effects of the activities implemented. In a static analysis, we concentrate in the impact of the variation of a single variable on our variable of interest, assuming all other relevant factors remain immutable. 17. A dynamic assessment would in turn require forecasting the levels of a wide range of variables over the medium to long term, and then estimating the expected costs and benefits of the set of reforms proposed, based on the evolution of the whole set of variables over time. For dynamic cost benefit analysis, computer general equilibrium models have been widely used as a preferred methodology. 18. At this early stage of reforms in Cambodia, scarcity of data limits both approaches to impact assessment importantly, but especially the dynamic evaluation of the impact of reforms. Accordingly, this analysis is limited to the partial assessment of static costs and benefits. Despite its obvious limitations, it is possible to draw substantial insights from the short term static analysis of the net benefits expected from the reforms. And in fact, our preliminary assessment below suggests that very sizeable gains can be obtained from the proposed program. Benefits Estimation 19. The inability of central governments to match service delivery to the different preferences and needs of local populations can result in important inefficiencies. It is commonly argued that, as a small country with a relatively ethnically homogeneous population, preferences for service delivery across Cambodia do not differ significantly. The reality is that differing preferences for local services are not solely due to ethnic differences but to other factors such as geography, population levels, or even the inertia of past service delivery that may have had a greater emphasis on some services than others. Moreover, local populations in Cambodia have not had the chance to express their preferences until recently, and neither has local administrations been able to deviate from the instructions received from line Ministries as to the use of the public funds they receive. Although arguably the first ever transfer to local governments, the Commune/Sangkat Fund (CSF), is mostly spent on rural roads, there is some variation on the mix of infrastructure projects implemented. In certain regions, irrigation projects
5 receive greater priority for instance. Rather than differences on the type of individual services communities prefer, differences concentrate in the sectoral priorities faced by the localities. 20. As decentralization and de-concentration reforms in Cambodia progress, it would be possible to ascertain if the mix of services provided by local administrations differs from that currently being offered, once the SNAs are endowed with some discretionary resources. Availability of resources will be a key determinant in any case. It is unlikely that the SNAs discretionary authority can be exercised in the absence of adequate level of funds. 21. In 2006, NCDD started a District Initiative Program (DIP), which remains the only source of available data to ascertain investment priorities from Districts and Municipalities derived from the use of unconditional transfers. Initially, the DIP covered 24 districts and municipalities throughout Cambodia, but was expanded in subsequent years to cover 106 districts in 2010. The DIP provided D/Ms with grants to finance inter-commune projects (i.e. projects that would benefit two or more communes in a D/M). Each D/Ms received, on average, approximately US$20,000 per year, which it could allocate to infrastructure and non-infrastructure projects on a positive list. In 2009 and 2010, D/M allocated over 90% of funds for infrastructure on water resources projects (such as dam renovation, water gate construction and canal repair) and rural roads. Agriculture and natural resources management accounted for about 85% of total funds spent on non-infrastructure projects (mainly training and demonstration projects). 4 22. The observed concentration of funds from the DIP on a small number of sectors would indicate a high and largely unfulfilled demand for infrastructure in those sectors, on the basis of central government led spending. As such, the returns from the re-allocation of funds to those projects would probably offer higher returns to investment and efficiency gains. There are no estimates however of the efficiency gains from the re-assignment of public funds to the discretionary use of sub-national administration in Cambodia. As implementation of reforms proceeds, it will be possible to evaluate the gains from the reforms. Despite the difficulties, a first estimate is provided here. 23. In Indonesia, a study of the impact of allocative efficiency gains in the growth of the regional sector rendered an estimated coefficient of 0.36. That is, every percentage point gained on allocative efficiency translated into a 0.36% increase in the size of the regional sector (measured as total expenditure at the sub-national level). The data was used from a panel data of regional governments between 1996 an 2005. 5 The efficiency gains were achieved via improvements to the local PFM systems through a mix of initiatives - legislative and regulatory measures to improve execution of expenditure functions and better align fiscal transfers with development objectives at the local level, investments in PFM systems and capacity development of staff in local governments and measures to reduce costs and wastage. 4 The relatively high share in natural resource management (NRM) may have been caused by the availability of Danish Cooperation (DANIDA) funds that could only be allocated to NRM-type projects. 5 Estimates of allocative efficiency are reported in D. Tirtosuharto (2010). The Impact of Fiscal Decentralization and State Allocative Efficiency on Regional Growth in Indonesia. Journal of International Commerce, Economics and Policy, Vol. 1 (2) pp 287-307.
6 24. In the early stages of decentralization, allocative efficiency gains are expected to the sizeable. But the process of decentralization in Cambodia differs significantly from that of Indonesia in a variety of ways. The most important one perhaps is the size of the local government sector, which will only grow slowly in Cambodia in contrast with the large devolution of functions witnessed almost overnight in Indonesia. Current level of SNAs expenditure in Cambodia (excluding de-concentrated expenditures) amounted to some $120 million or close to 7% of total public expenditure. This is a similar figure than at the beginning of the decentralization process in Indonesia, and is expected to increase as decentralization of functions proceeds. In light of the government comprehensive reforms, SNA expenditure could reach 20% of total government expenditure by 2015, or an estimated $350 million. Assuming a similar coefficient allocative efficiency over SNA expenditure (0.36), and a 20% increase in efficiency over the period up to 2015, the allocative efficiency gains could be estimated at around $20 million per year (well over $100 for the duration of the program). 25. The proposed creation and implementation of the Sub-national Investment Facility is expected to benefit SNAs in a variety of ways: (a) by generating additional resources for local investment from both government, private sector and development partners to the tune of $150 million; (b) by reducing transaction costs associated to the procurement and contract implementation of local investment projects. 26. Substantial gains are expected from improved efficiency in revenue mobilization. In particular from the assignment and development of the property tax for SNAs, but also from improved revenue collection from non-tax instruments. Being conservative, we estimate an increase of 50% in local tax revenue collection between 2010 and 2014 (including all levels of administration), equivalent to $40 million. Non-tax revenues, derived from fees and contributions collected from public service delivery, as expected to double as a proportion of general non tax revenues from 2.5% to 5%, or an estimated gain overall of $10million. Table 1. Potential Economic Gains from DPSFM Subprogram 1 1. Economic growth from allocative efficiency 2. Resource mobilization from SNIF 3. Improved efficiency in revenue mobilization Gains (million USD) Share of GDP (% 2010) 100 0.850% 150 1.275% 50 0.425% Total 300 2.550% Costs Estimation 27. Adequate, albeit static, estimation of the costs associated to the proposed program requires the consideration of costs borne not only by Government agencies, but by the financial institutions and relevant stakeholders as well. For the former, we classified costs among: 1) administrative costs (or directly derived from the implementation of the program by Government agencies); 2) enforcement costs (incurred during the enforcement or regulations and monitoring of compliance); and 3) fiscal costs (associated to required budgetary expenditures, foregone tax revenue, financial costs, etc).
7 28. On that basis, an initial estimate of costs would include: Table 2: Cambodia: Potential Costs to Government from DPSFM SP 1 Types of Adjustment Costs National and Regional Governments 1. Administrative costs (i) Costs incurred in legislative and regulatory initiatives estimated $1.0 million (new laws, subdecrees, circulars and guidelines). (ii) Costs incurred in developing various policy papers, reform strategies and action plans estimated at $1.0 million (including IP3 2014-2016 and annual reviews and action plans). (iii) Costs from the establishment of NCDD-S and new coordination mechanisms at the different tiers of government estimated at $1.5 million. (iv) Administrative and investment costs from the development and implementation of the financial management procedures for SNAs at $8 million. (v) Administrative costs from the design and implementation of the SNIF at $5 million. Total administrative costs at $16.5 million 2. Fiscal costs (i) The permanent budget operating costs of the NCDD-S, new coordination mechanisms, institutions such as SNIF and annual reporting requirements estimated at $5 million per year or a discounted present value lump sum of $90.0 million. (ii) The permanent costs of operating SNA financial management systems estimated at $1.5 million per year or discounted present value lump sum of $25.0 million. (iii) Capacity development for SNAs on financial management systems and other elements of the systems of intergovernmental fiscal relations estimated at $8 million. Total fiscal costs discounted present value lump sum of $123 million Total estimates Total discounted present value lump sum of $139.5 Source: ADB staff estimates. VI. Program Financing 29. The government has requested a loan of $37.46 million to finance the first subprogram of the DPSFM Sector Development Program from Asian Development Bank Special Resources. The size of the loan is based on a number of factors. The key considerations include: (i) (ii) (iii) The potential importance of the sector to the economy. The set of policy reforms and project initiatives proposed may conservatively add $300 million to GDP or approximately 2.5 percentage points over the duration of the program. The adjustment costs. The policy adjustment costs to the Government are conservatively estimated as a discounted present value lump sum of $139.5 million (Table 2). These include the administrative, investment and fiscal costs to the Government in implementing the reforms. As a result, the expected net benefit from the program is estimated at some $160 million. The reform program includes several politically sensitive and complex policy and institutional reforms. To support the reforms, the program amount loan provides leverage to the executing and implementing agencies to support implementation.
8 (iv) The program loan also reflects the government s development financing needs as well as the need to conform to the overall financing requirement from the country partnership strategy period.