Resource allocation and budgetary mechanisms for decentralized health systems: experiences from Balochistan, Pakistan.

Resource allocation and budgetary mechanisms for decentralized health systems: experiences from Balochistan, Pakistan Andrew Green,1 B. Ali,2 A. Naeem,3 & D. Ross4 This paper identifies key political and technical issues involved in the development of an appropriate resource allocation and budgetary system for the public health sector, using experience gained in the Province of Balochistan, Pakistan. The resource allocation and budgetary system is a critical, yet often neglected, component of any decentralization policy. Current systems are often based on historical incrementalism that is neither efficient nor equitable. This article describes technical work carried out in Balochistan to develop a system of resource allocation and budgeting that is needs-based, in line with policies of decentralization, and implementable within existing technical constraints. However, the development of technical systems, while necessary, is not a sufficient condition for the implementation of a resource allocation and decentralized budgeting system. This is illustrated by analysing the constraints that have been encountered in the development of such a system in Balochistan. Keywords: Pakistan; health care rationing; budgets; financial management; public sector; health services accessibility; community health services, organization and administration; models, economic. Voir page 1033 le re´sume´ en franc¸ais. En la pa´gina 1034 figura un resumen en espan˜ol. Background Health sector reform is now amain focus of attention for the ministries of health in many developing countries. Policy discussions often focus on the development of a more efficient service through initiatives such as distinguishing between the func- tions of ‘‘purchasing’’ and ‘‘provision’’, the develop- ment of the public/privatemix, greater autonomy for hospitals, and the development of district-based systems. Critical to the last of these is the develop- ment of appropriate systems for allocating resources from central to lower administrative levels. For many countries, the existing system of allocating resources, particularly financial resources, to lower levels in the health service is inconsistent with decentralization policies and the pursuit of equity. However, less attention has been paid to the development of resource allocation processes. This article analyses issues involved in the development of such resource allocation and budget- ary systems, drawing on the experience of work conducted under the Balochistan Health Systems Strengthening Component (BHSSC) of the Second Family Health Project (FH2P). The BHSSC seeks to develop institutional capacity to support a more decentralized and effectively functioning district health system. These objectives are directly linked into the wider Government of Balochistan’s Social Action Programme (SAP), which places priority on primary care services and decentralization from the provincial level towards district-based management. The decentralization strategy being developed and implemented within the project is based primarily around the strengthening of district management and planning capacity brought about by increasing management skills, improving manage- ment systems and developing more decentralized organizational structures. This is combined with a strengthening of provincial planning systems to provide strategic policy guidance to districts and builds upon earlier work carried out to establish a provincial health planning system (1). Within such a decentralization process, one necessary precondition for achieving equity is the development of systems for allocating resources to districts in line with health needs. The objective is to develop an approach that allows for central resource planning and local health care programming (2). The present article is limited to a consideration of resource allocation within the government health sector and does not consider the overall levels of sectoral funding or resource flows to 1 Senior Lecturer in Health Planning and Economics, Nuffield Institute for Health, University of Leeds, 71–75 Clarendon Road, Leeds L52 9PL, England (e-mail: a.t.green@leeds.ac.uk); and Consultant to the Second Family Health Project, Balochistan, Pakistan. Correspondence should be addressed to Mr Green at the former address. 2 Associate Professor and Chairperson, Department of Administrative Sciences, University of Balochistan, Balochistan, Pakistan; and Consultant to the Second Family Health Project, Balochistan, Pakistan. 3 Assistant Professor, Department of Administrative Sciences, University of Balochistan, Balochistan, Pakistan; and Consultant to the Second Family Health Project, Balochistan, Pakistan. 4 Assistant Director of Primary Care, East Riding Health Authority, Yorkshire, England; and former Health Planning and Management Adviser, Health System Strengthening Component, Second Family Health Project, Balochistan, Pakistan. Ref. No. 99-0174 Policy and Practice 1024 # World Health Organization 2000 Bulletin of the World Health Organization, 2000, 78 (8) nongovernmental elements. The resource allocation and budgeting system in Balochistan is primarily incrementalism. There are various weaknesses, including, most importantly, mismatches between the population health needs, the requirements of existing facilities, and the budgets set. The strategy adopted by the project to initiate a more appropriate resource allocation system has involved the follow- ing: analysis of previous resource allocation patterns; development of proposals for modifying resource allocation systems; and development of necessary support systems and organizational structures for linking the resource allocation process into the provincial and district planning and budgeting systems. Despite these technical developments, the project has been unable to produce significant changes in the allocation processes. This is largely the result of failing to gain sufficient support in key areas of government, at both the political and bureaucratic level. The objectives of this article are as shown below. . First, we explore the more technical issues involved in resource allocation in developing health systems, an area in which little has been published. . Second, through an analysis of the Balochistan public health sector, we illustrate that technical solutions alone, in the absence of wider political ownership, will not lead to change. It is argued that any redesign needs to take place within a specific health system context, and this is illustrated with an assessment of the Balochistan system. . Finally we examine the strategy adopted in Balochistan to move from the current inappropri- ate system to a more appropriate one through a phased approach, and draw lessons from the experiences and difficulties encountered. Approaches to resource allocation and budgeting Resource allocation is taken to be the overall allocation of financial resources to decentralized management areas within the government health service. It is closely related to budgeting, which is concerned with statements of specific expenditure plans within these broad allocative ceilings. An effective resource allocation mechanism is a key factor in supporting decentralized health systems, which are deconcentrated, i.e. responsibility and authority are decentralized within the public health system. This contrasts with a devolved system, where local government takes responsibility for health care (3). Mechanisms for devolved systems require different decision-making and allocative processes (4) and are not discussed here. Decentralization (5) provides an opportunity to respond to local needs within a national equity-focused policy for allocating resources, though there are potential dangers in inappropriate decentralization (6). Misallocation of financial resources is widely accepted as an important cause of poor health service performance and inequity (7). Frequently, previous budget allocations (incre- mentalism), current service or facility patterns, capital developments or political factors heavily influence resource allocation. Such approaches fail to address efficiency and equity objectives. Research has been carried out on resource allocation between different activities (8, 9) and the cost-effectiveness of particular interventions. How- ever, less attention has been paid to allocations between different geographical administrative areas, reflecting perhaps the relatively recent interest in decentralization. While the first of these possibilities focuses on efficiency, the second is more concerned with issues of equity. Despite the development of complex systems of resource allocation in industria- lized countries, there are relatively few documented examples of needs-based resource allocation systems in developing countries. Examples include Papua New Guinea, which developed a complex goal programming model (10), Zambia (11) and South Africa (12). An equity-focused policy would require a shift in resource allocation, away from a mechanism based on existing facilities, to one based on an assessment of the needs of particular areas and their population groups. Fig. 1 sets out a conceptual model for such a process. One of the best documented examples of this approach is the United Kingdom National Health Service (NHS), which in the 1970s developed a resource allocation formula (13) based on a model with the following components: – the health needs of a specified population; – the relative costs of different services; – the relative costs associated with different areas; – the costs associated with non-service delivery, such as teaching costs; – the use by patients in one area of services in another (cross-boundary flows). Within a public sector resource allocation system, a further criterion may be the level of private sector health care, as this may reduce the potential need for public sector resources. The most contentious of these components is the assessment of needs. Potential measures include a combination of the following: – the size of the population; – age and sex ratios; – direct measures of morbidity; – mortality ratios as an overall proxy for different levels of health need; – specific indicators of deprivation (to reflect potentially higher levels of relationship between morbidity and mortality, and higher health care costs). In the NHS resource allocation model, need was split into different components including the following: acute care, maternity, chronic and psychiatric, each with different proxy measures. Issues over measure- Resource allocation and budgetary mechanisms for decentralized health systems 1025Bulletin of the World Health Organization, 2000, 78 (8) ment (e.g. 14, 15) have centred around whether mortality is a good proxy for morbidity; whether including wider causal deprivation indices, alongside direct health measures, leads to double counting of need; whether morbidity measures are influenced by accessibility to services; the weighting given to different aspects; whether the actual funding based on components of need are translated into equivalent services at the local level; and which services should be centrally provided. Such formulae have tended to concentrate on allocation of recurrent funding, with capital allocations being related to existing capital stock. There are further issues around the implementa- tion of such formulae. These include those discussed below. . The optimal speed of implementation. This is related, in part, to the absorptive capacity of under-funded areas; the ability to reduce funding of ‘‘over-resourced’’ areas; the overall growth of funding; and the political strength of different areas and their ability to resist a relative reduction in allocation. . The need to take account of local revenue generation. Some areas, because of their economic conditions, may be able to raise revenue locally (through user charges, for example). Under an equity-focused system, such areas might be expected to receive fewer funds from a central source. Compensatory mechanisms, however, may discourage local revenue generation. . The existence of adequate information systems. A complex system that is not backed up by appropriate and credible information, such as data on population size, can become easily discredited. . The dangers of perverse incentives, which send the wrong ‘‘signals’’ to managers. The most obvious of these is where high measures of health are seen to be ‘‘rewarded’’ by high resource flows, with no compensating mechanism to encourage managers to improve the health situation. . The difficulties of incorporating cross-boundary flows, without encouraging cost-shunting. Where patients may cross health care ‘‘boundaries’’ to seek care from neighbouring health services, some mechanism is required to recompense the receiv- ing health authority for the additional workload. Without this there is a danger of patients being encouraged to seek health care from outside their catchment areas, particularly when the cost is high. As a result of such constraints, and in particular the paucity of good information, countries considering the adoption of such formulae may need to start with very basic allocative formulae, based primarily or wholly on population distribution, before developing more sophisticated formulae such as those used by the NHS. Over time, it would be possible to incorporate an element of, or surrogate for, ‘‘need’’ without getting involved in formulae that are too complex. Such indicators may be used to weight basic population data, to reflect differing levels of need in similar-sized populations. In Balochistan, an example might be female literacy rates, which in many parts of the province are extremely low and, given the relationship between female education and health status, may provide a potential added measure of need. Diagnosis of the situation in Balochistan Any redesign of a resource allocation and budgetary systemhas to take into consideration the health system to which it applies. Universal systems of reform are unlikely to be successful (16). The following sets out key features of the Balochistan health system which have a bearing on the form of allocative system. The health care system Balochistan Province is situated in the aridmountain- ous south-western quadrant of Pakistan. The 26 dis- tricts of the province are themain administrative unit. Balochistan’s total population is estimated to be around 6.6 million, but densities are low (19 per km2) and predominantly rural in distribution. Health status is poor, with an estimated infant mortality of 180 per 1000 (17). Policy and Practice 1026 Bulletin of the World Health Organization, 2000, 78 (8) The health care system is based around government provision of services, although there is a significant and growing private sector that accounts for around half of all service delivery (18). State services are organized into a variety of primary level service delivery outlets feeding into small district level hospitals. Larger hospitals and training institu- tions are located mainly in Quetta, the provincial capital. There are various poorly integrated vertical programmes. The health care system suffers from a number of deficiencies attributable, in part, to problems in the planning and management of services. Most notable among these are high rates of staff absenteeism, limited availability of drugs and supplies, and a critical shortage of female health workers. The resultant poor quality of service leads to low levels of utilization and limited impact upon health problems. In such a context a resource allocation system needs to pay particular attention to its impact on quality and utilization. Existing resource allocation and budgeting system The existing resource allocation and budgeting system is based upon budgetary demand covering both the development and revenue sides of the budget. In the Pakistan public sector, the terms revenue and development budgets are used to denote what elsewhere may be known as the recurrent and capital budgets. The system was designed to be ‘‘bottom-up’’ and as such is consistent with a decentralized approach. In theory, budget submis- sions are proposed by budget holders, typically district health officers, responsible for district primary care services, and hospital medical super- intendents. However, the failure of the majority of budget holders to participate in the process has effectively led to a centralization of the processwithin the provincial Health Department. Any submissions are aggregated, reviewed and subjected to central modification and political influences. No clear policy guidelines or estimates of future financial availability are applied in drawing up the budgets. While some norms are deployed, it is widely recognized that these are outdated and bear little relation to real service costs. Although adjustment is made centrally in recognition of new facilities, the main allocative driver is the historical budget rather than health or health service needs. This leads to the practice of budget holders using the virement process (transfer of items from one financial account to another) to try to adjust budgets after they have been set, rather than formulating robust initial budgets, even though virement is itself a cumbersome process. Specific aspects of the system are discussed further below. Overall budget structures One of the issues surrounding the development of resource allocation processes relates to the complexity of the relationships between the existing budgets. In Balochistaneachhealthcare systemmayhaveanumber of budgets, the minimum usually being a recurrent and a capital budget. In addition there may be budget streams associated with project activity. For example, there is a permanent (on-going) budget, which is routinely approved with some inflator, as well as two forms (new and continuing activities) of a temporary budget, which are set out each year in the schedule of new expenditure. The relationship between these is shown diagrammatically in Fig. 2. In addition, a multi- donor supported Social Action Programme (SAP) has been set up, which though in theory is fully integrated into the government budget system, continues to be viewed in some quarters as an additional budget process. As personnel costs are contained within the permanent budget, they are largely protected from budgetary cuts. This has led to a 50% increase in allocation to personnel costs since 1981, and a concomitant relative decrease in non-salary items. Alongside the revenue budgets (the ultimate responsibility of the provincial FinanceDepartment), there are development budgets in the form of the Annual Development Programme (ADP) that are controlled by a separate Planning and Development Department. In theory an ADP comprises a number of planning proposals which set out the development and revenue implications of a project for overall approval. In practice the revenue implications are considered less critically at the time of the ADP. The number and type of budgets tend to cause confusion, particularly in terms of the relationship between development items and on-going revenue items. First, as a result of the stringent controls on the use of the permanent revenue budget by the Finance Department, a habit has arisen of using the develop- ment budget to finance what effectively are ongoing revenue items. Second, the parallel nature of the revenue budgets, which now include an allocation for SAP activities, leads to complexities in budgeting procedures and a danger of overemphasis on process rather than strategy. In addition, the lack of linkage between development and revenue budgeting has led to an over-extension of future revenue requirements. Resource allocation and budgetary mechanisms for decentralized health systems 1027Bulletin of the World Health Organization, 2000, 78 (8) The potential scale of this problem is exemplified in Table 1, which sets out the gross long-term revenue implications of the 1995–96 ADP. At constant prices, a 22% real increase would be required to accommo- date the projected revenue implications in 1998–99. This problem is a product of both a failure in the past to recognize the critical strategic importance of viewing the two forms of budget together and, partly as a result of this, the lack of coordination between the two budget arms of the Department. In addition to the aspects discussed above, the internal budget structures are complex and involve a large number of budget heads and line items (known as objects of classification). This is the result of a growth in activities without a concomitant rationalization of budget structures, and makes allocations difficult to analyse, monitor and change. As we have seen, different budgets (e.g. revenue, development and SAP) have different decision-making and accounting systems, leading to inconsistencies. In Balochistan, powers for virement are extremely centralized at a variety of levels (Directorate, Secretary of Health, or Finance Department) and are dependent on the form and level of re-appropriation/virement that is re- quired. While more robust budget formulation would reduce the need for virement during the year, it is inevitable that some re-appropriation is likely to be necessary. This complexity, together with the cen- tralized control on virement or powers, leads to an over-rigidity in the system and to difficulties in budget formulation and management at the district level. Financial guidelines The development of decentralized district budgets needs to occur within a framework of overall policy guidelines. In particular, guidance is needed on overall health strategy, on the criteria by which resources are to be allocated, and on likely levels of future budget growth. Without such a framework, decentralization can easily become fragmented and decentralized plans become unsustainable. This is well illustrated by the situation in Balochistan, where there are a lack of useful explicit central financial guidelines. There are three levels at which the issuance of guidelines would help. Currently the provincial Finance Department does not provide global sectoral revenue budget estimates for the following year. This causes difficulties for the Department of Health, which is forced to develop its budget in the absence of guidance as to the likely overall budget levels. Similarly the Planning and Development Department does not provide any formal estimate of the likely capital allocation for the following year. In the absence of any central guidance, the Department of Health, in turn, does not provide any guidance for service managers concerning their likely allocation. Information base Any budget system needs to be based on robust information that includes health needs, service patterns, and costs. Information systems may not exist for providing the appropriate information for allocating resources to decentralized levels. For Balochistan there is a significant lack of information about either utilization or health care costs, both of which are critical for the development of budgets. Although a health management information system is being developed and should eventually assist in providing health utilization data, it will be some time before it is in a position to provide such information routinely. One particular difficulty relates to popula- tion data, where there are concerns over the reliability of estimates based on outdated and disputed 1981 census data. A census was carried out in 1998 which may therefore overcome this constraint to a significant degree. This is further discussed below. Professional expertise Up till nowwe have focused on the budget structures and processes. In addition, it is essential that any allocative process be handled by staff with appro- priate professional expertise and who recognize the critical importance of budgets for achieving policy. The current allocative process in Balochistan is administered predominantly by clerics, and though formally there are various points at which there is administrative or technical scrutiny and approval, in practice it is minimal. Linked to much of the above, either as a cause or effect, is a general lack of budgeting expertise in theDepartment ofHealth. For example, there is no professional accountant — a major constraint on the system and a reflection of the medicotechnical dominance in the Department of Health. This, linked to the lack of budgeting skills at the service manager level, is an important factor in budgeting failures. There has been little training in budgeting and this, combined with the complexity of the process, is a disincentive to active involvement by professional or policy-level staff. The situation is exacerbated by the frequent transfer of professional staff, which has caused problems in maintaining continuity. Table 1. Additional revenues required to meet projected health budgetsa Sector Additional revenues required % of (Rs x 106) Total 1996–97 1997–98 1998–99 Long-term recurrent requirements Primary 35.8 51.4 122.2 467.0 79 Secondary 0.0 0.0 6.9 14.1 2 Tertiary 0.0 37.5 59.4 101.5 17 Support 0.0 5.7 11.0 12.3 2 Total 35.8 94.6 199.6 594.9 100 a The projected data were based on the implications of the Balochistan 1995–96 Annual Development Programme, Balochistan Department of Health Development Budget. See ref. 19. Policy and Practice 1028 Bulletin of the World Health Organization, 2000, 78 (8) Political context The political context, within which decentralization takes place, is critical to the type of decentralization and its success. In Balochistan a number of issues are important in this respect, including the political relationship between the national and provincial governments, the frequent changes in government, the role of the military in politics, and the strength of the medical profession and its role within health care management. The fragile and transitory nature of government tends to result in a short-term policy perspective and a desire to maintain central political control that may run counter to the development of more rational and decentralized systems of budget- ing. Finally, in a system where public sector salaries are very low, the potential for themisuse of power for individual financial gain cannot be ignored. The products of all this are budgets and patterns of resource allocation with high degrees of inequity and inefficiency. The budgets that result are as described below. . Inefficient between-line items, particularly be- tween salary and non-salary items. For example, salary costs may be as high as 80% of total costs in primary level facilities (19). . Inequitable between similar service units. For example, allocations to hospitals with apparently similar capacity or demands (as indicated by the number of beds or utilization levels) can vary by up to a factor of two, and to primary facilities by as much as five (20). . Inequitable between populations with apparently similar health needs. For example, allocations per capita to primary level services by district vary from 18 to 159 Rs (US$ 1.00 = ca. Rs 32 at the time of the 1995–96 ADP), with an overall provincial average of Rs 37 per capita (20). The above-mentioned constraints were analysed at an early stage in theBHSCCproject, fromwhich it became clear that the allocative systemwasnotworking in away that would either induce efficient practice or promote equity between districts, regardless of how the latter was defined and measured. This situation has led to a deep sense of frustration at various levels of the health service, but in particular at the level of the districts. Overall, thebudgetingandresourceallocationsystemis viewed as an obstacle rather than a support to the development of decentralized health care. It was clear that the project would need to put emphasis on improving the resource allocation system as part of its decentralizationobjectives.We turnnowto theoptions that were considered. Modifying the resource allocation and budgeting system Technical and organizational design issues Described below is the process of modifying the resource allocation system operating within Balochi- stan. The following refers to intraprovincial resource allocation, rather than interprovincial, and only to revenue budgets. It was accepted early on that it would be impossible to include development budgets in the system redesign and that they would have to be tied into the budgeting system through the planning process for the medium term. Within Balochistan, it was agreed that the general criteria for choosing an allocative system to districts were as follows: – impact on equity; – impact on efficiency ; – transparency; – feasibility including data availability, technical capacity to operate, ability to reduce over-capacity where appropriate, and capacity to absorb growth where appropriate; – consistency with other government systems; – flexibility to allow medium- to long-term refine- ment. Alternative models Following diagnosis of the existing situation, criteria were applied to four potential allocative models, including the current one. All the models require a mechanism for allocating a portion of the overall Department of Health budget to cover central costs and those programmes that cannot be managed by districts. These include central and divisional admin- istration, tertiary hospitals and other institutions and training. While it is possible to envisage a situation whereby the bulk of these are also decentralized with districts purchasing them back, this is unlikely to be feasible for some time given current managerial and information constraints. Model A. Incrementalist (current model). District budgets are based on the previous year’s allocation (or expenditure), increased pro rata, though the possibility exists for new budgets to be added through the annual schedule of new expendi- ture. Though this approach is administratively simple and non-threatening it neither promotes efficiency nor equity. Model B. Health facility requirement. Bud- gets would be set to ensure that the major existing primary and secondary level facilities are provided with adequate resources to allow them to operate effectively and more efficiently. This would not lead to an improvement in the distribution of resources between districts. However, it would improve the quality of care at existing facilities, and indeed provides rewards to facilities with high utilization rates. This is an important feature in the Pakistani system where there is no reward, and indeed some disincentives, for improving utilization at facilities that are currently under- or inappropriately re- sourced. Budgets would take into account the real costs and utilization rates of existing facilities. However, to avoid major swings in allocation to different districts, whichmight run counter to equity, ceilings on increases would need to be set. Such ceilings could initially be based on a crude per capita basis and would take into account absorptive Resource allocation and budgetary mechanisms for decentralized health systems 1029Bulletin of the World Health Organization, 2000, 78 (8) capacity. Explicit prioritization would also need to be made between facilities within districts, as this approach on its own will not lead necessarily to adequate overall resource levels in districts. Districts would have to make their own prioritization decisions between facilities, based on their assess- ment of differential need. Some training would be required for district managers and central staff, but the close links to existing facilities would mean that it was politically acceptable and intuitively understand- able. A systemwould need to be put in place to ensure the above elements were handled at both provincial and district levels. Model C: Health service requirements. The third approach is an enhancement of the second. Through strong links into district planning, budgets are tied more closely to the overall health service requirements of the district population, rather than the health facility requirements. Districts which, through the planning process, are considered to be underserved would receive particular planning atten- tion and above average allocation growth rates. Development expenditure that is related to plans, and subsequent revenue implications, would be a major mechanism for this. Judgements as to which districts were underserved, and to what degree, would be formed on the basis of similar information to that in Model D (e.g. population and morbidity), but would not be tied formally into a formula. There would be similar implementation implications as to Model B, but an additional need for strengthened central planning capacity. Model D: Population-based resource alloca- tion. The final model involves the allocation of resources on the basis of population, possibly weighted by factors such as age, sex, specific health needs, density, cross-boundary flows and different costs of health care delivery. Under this, district managers set specific budgets within an overall allocation provided by the formula. In a system of purdah, the ability to access services anonymously may be important and may lead to additional cross- boundary flows (21). The main strength of this model lies in the potential to promote equity. Similar training require- ments to those of the other models would be needed. However, the weaknesses in the current information system constitute a major constraint to this model. Furthermore the political opposition from ‘‘losing’’ districts could be major obstacle. Each of the resource allocation models has advantages and disadvantages. While there was general support for Model D in principle, it was initially felt that it would be impossible to introduce in the short-term, largely as a result of the lack of confidence in the population data. Instead a progressive movement through the models towards Model D was chosen. The current incrementalist approach (Model A) is widely regarded as inap- propriate. Development of better systems to meet the operating requirements of current facilities (Model B) would improve their technical efficiency and engender support in the system for the budgeting process. However, it was recognized that this should only be an intermediate step since this, by itself, will not overcome the inefficient and inequitable dis- tribution of current primary and secondary level resources. Model C (through its strong emphasis on links between planning and budgeting) would allow a greater development of health needs sensitivity and responsiveness. The last stage in the progression towards full budgetary decentralization would in- volve the allocation of resources on the basis of population, weighted to incorporate other aspects of need (Model D). Supporting components There are various components required in order to develop the chosen allocative system, including those discussed below. Improved information As we have seen, information systems appropriate to the development of a decentralized resource allocation system may not be available and will need to be developed through both routine data collection and research. Within Balochistan, development of a health management information system is proceeding. Full provincial coverage has been achieved in primary level facilities, and a substantial amount of health service activity data is now flowing through the system. This will eventually provide more detailed information on non-financial inputs and service outputs. Studies have been undertaken to develop a range of unit costs of routine activities in primary level facilities and district hospitals (22, 23). The first of these studies provided important information on the actual costs and ‘‘standard’’ levels of resources for primary facilities, taking account of utilization, case mix and other district level costs such as supervision and monitoring (Table 2). The difference between actual costs and standard requirements gives an indication of the ‘‘funding gap’’ to be filled if services are to run at an improved level of efficiency under Model B. Differ- ential funding needs can be calculated on a district basis and under-funding of existing facilities occurring in the primary health sector can now be estimated. Furthermore, up until 1998–99 it had been expected that SAP would provide real significant increases in health sector spending. This would permit the use of a policy of differential growth (based on the relative size of the gap) to under-resourced areas, rather than achieving resource shifts by cutting absolute alloca- tions to ‘‘over’’-resourced areas. This, it had been hoped, would minimize political opposition from ‘‘losing’’ districts. Unfortunately, the 1998–99 budget suffered a significant cross-sector cut in response to government concerns over increasing regional poli- tical tension following nuclear testing. One of the major constraints against developing a population-based allocative formula was the lack of general confidence in the official population figures Policy and Practice 1030 Bulletin of the World Health Organization, 2000, 78 (8) and the political ramifications of these for different population groups. However, in 1998 a census was carried out, the results of which, formally at least, are expected to be accepted. One result of the census has been an indication that the Department of Health would now be prepared to consider a faster move towards incorporation of population in any formula. Development of central planning and budget- ing systems.Decentralization is often interpreted as a weakening of the central position. This, we would suggest, is misguided. Instead, a change in role is needed, with concomitant energy put into a redevelopment of this role. Within Balochistan, annual guidelines (24) have been developed which provide specific guidance on service policy, and broad service delivery targets. All these elements are combined to form the district planning and budget- ing guidelines, designed to be issued annually to budget holders by the central health planning unit. In Balochistan, the Planning Cell was supported by the project to produce financial guidelines for district officers. In developing these it was proposed, inter alia, that: – no district would be allowed to suffer a real net loss in resource levels; – targets would be set to guide managers in the line item allocation of funds within their overall budget: in particular, these would focus on increasing the size of the non-salary budget; – hospitals and central programmes would receive increases in line with inflation. Development of district planning and budget- ing systems. One of the key rationales for the development of a decentralized system is the ability to develop plans for districts based on their specific needs, within an overall policy envelope. In Balochi- stan, a district planning cycle has been developed (25) that enables district officers to relate their identified health problems to implementable solutions. Service managers are encouraged to define objectives linked to interventions and activities for the coming three years. Using the resource guidelines and financial informa- tion, the intention is that they are then able to develop more rational and appropriate development and revenue budgets. These budgets can then be sub- mitted to the centre for checking and consolidation prior to submission to the Department of Finance. Implementation experiences The above section has outlined the technical work carried out to develop an improved allocative system, consistentwith the stated decentralization aims of the Department of Health. This section describes the implementation history of this aspect of the project. Following early preparatory work, agreement was reached on the way forward as described earlier. Costing studies were then carried out; and training in the development of district plans and budgets provided. Initially the BHSSC project was designed to operate province-wide. The financial year 1996–97 was used as a trial year to introduce revised planning and budgetary processes. District plans and devel- opment budgets were developed by 12 of the 26 districts in the province. Unfortunately, inade- quate time was set aside to prepare revenue budgets and no district was able independently to submit a worked proposal. Due to the limited success of the district-level budgeting process, central allocation guidelines based on unit cost information were applied to existing district budgets. This produced more rational, but still centrally generated, draft district budgets for 1997–98. Insufficient ‘‘owner- ship’’ within the Budget Section of the Department ofHealth, however, led to even these drafts not being accepted and introduction of the system being delayed a further year. Following a mid-term review of the overall project, it was decided to focus more intensively on six trial districts. Though this had little effect on the design issues for a resource allocation system, it did have implications for the human resource aspects of the project. By concentrating resources on a limited number of districts it was, in theory, possible to provide greater support for the development of their budgeting and planning capacity. However, the general problem of frequent staff transfers, referred to earlier, lessened the actual impact of this. Furthermore, the concentration on trial districts gave rise to (unfounded) suspicions that the allocation system might favour such districts in terms of the overall level of resources. Inevitably also the shift in focus to trial districts lessened the inputs of the project at the provincial level. The trial districts, with technical support from the project, produced draft budgets for 1998–99, but these were not accepted by the Department of Health, ostensibly on the grounds that they were late. The process was started earlier in the following year and trial districts again produced draft budgets for 1999– 2000 based on their facilities’ needs and using the information provided from the costing studies. Again, however, they were not incorporated in the con- solidated budget, with concerns about the favoured status of the trial districts again being voiced. Following the 1998 population census the Depart- ment of Health is now considering a shift to centrally determined budgets on the basis of a complex combination of population and facilities (in effect a Table 2. Revenue costs per month for types of primary care facilitya Civil dispensaries Basic health units Rural health centres Budget Personnel Budget Personnel Budget Personnel allocation cost allocation cost allocation cost (Rs) (%) (Rs) (%) (Rs) (%) Actual costs 21 062 81 21 062 81 46 043 80 Standard 42 455 52 54 555 49 210 213 46 requirements a Actual personnel costs and standard requirement personnel costs are expressed as a percentage of their respective budget allocations. See ref. 22. Resource allocation and budgetary mechanisms for decentralized health systems 1031Bulletin of the World Health Organization, 2000, 78 (8) combination of Model B andModel D). However, the project is drawing to an end and no changes are likely to be seenwithin the next year, with any future changes likely to be implemented through SAP. Conclusions and lessons to be drawn In this article we have described one aspect of a broader decentralization project. A number of other aspects are ongoing (including the development of decentralized planning capacity and monitoring processes). However, it is clear that the technical work carried out to support the development of the resource allocation and budgeting system in line with decentralization has not yet led to change. Although it is expected that when this project ends the technical aspects of the work will be taken over by the continuing SAP processes, it is important to analyse the implementation failure and see what lessons can be drawn for other decentralization processes. The causes of failure identified in the article are summarized in Table 3. One useful framework for analysis of the sort of policy changes discussed here is that proposed by Walt &Gilson (26), which suggests the need to assess not only the policy content but also the context, processes, and actors. The focus of the project’s work in this area was on the content of the reform to the allocative and budget system, with considerable attention being paid to alternative models and their technical robustness. As part of this, the context in which the reforms were being proposed was analysed, paying particular attention to the wider government plan- ning and budgeting systems, information availability, and skill levels. Possibly insufficient attention was paid to the decision-making culture. The culture of centralized decision-making and an attendant proce- durally driven bureaucracy, coupledwith the frequent transfer of staff, means that decentralization both challenges the organizational and management culture and is in fact high risk, albeit with high potential returns in terms of health impact. The process of change to a new system requires support at all levels. The project had always recognized the sources of resistance to decentralization broadly, and budgeting and resource allocation in particular. Unfortunately, the level and sources of resistance to new systemswas underestimated. Various actors can be identified as perceiving the process as threatening. These include the following: the senior professionals in the Health Directorate concerned at a loss of role, status and power; clerical staff currently responsible for centralized budgeting concerned about the devel- opment of a system which may sideline them; and other central departments, most importantly the Department of Finance, concerned about the poten- tial dangers of decentralization. In addition, politicians able to use the current allocative system as a means of maintaining a political base may be resistant. Under- pinning all of this is the critical issue as to whether there is shared understanding of and genuine support for equity, the main driving force behind a needs- based resource allocation system. Lastly, there are questions as to the process adopted by the project to introduce the changes. The changes were driven externally by a project operating close to, but largely external to, the government systems. As such it may have underestimated the need for greater internal ownership or championing of the changes by either a senior bureaucrat or politician. The shift by the project halfway through from a province-wide intervention to one using trial districts may also have led to a reduced opportunity for change. Under a province-wide approach greater support from potential winners may have been feasible, and accusations of favouring trial districts would not have been possible. What lessons can be drawn? . First, it is easy to underestimate the various sources and depth of resistance. Greater attention needs to be given to the political dimensions of such projects, to seek ownership of the process of change. While resource allocation may appear as a ‘‘technical’’ issue, it clearly is muchmore than that. Furthermore, the type of changes involved in a new resource allocation system may be viewed as technically challenging by staff with little manage- ment, let alone economic, training and as such resisted. Under such circumstances, projects need to recognize that the processes of change may need to be slower to develop a critical mass for change. This unfortunately, does not accord easily with the time horizons of most projects. Further- more, a clear and strong champion is needed from within the system. However, one of the structural difficulties with donor projects is that they tend to operate in parallel with the mainstream govern- ment system. The SAP, which is designed as an integral part of government, does in theory provide greater opportunity for genuine owner- ship to be developed. It is possible that the degree of genuine commitment to equity objectives could have been explored further with an analysis of the Table 3. Causes of intervention failure Lack of appropriate credible information on health needs and service costs Resistance from politicians Resistance from bureaucrats and health service mangers Lack of clear and shared understanding and commitment to equity Complex budget structures Centralized decision-making processes and culture Interventions from project external to government Difficulty of reforming one public sector in isolation Difficulty of using trial districts Lack of appropriate budgeting skills Lack of central guidelines Policy and Practice 1032 Bulletin of the World Health Organization, 2000, 78 (8) variations in health outcomes between districts (as opposed to inputs). . Second, it needs to be recognized that, in decentralization projects, as much attention needs to be given to supporting the necessary changes at the centre as at the periphery. While the district staff embraced the technical proposals, the major resistance was encountered at the centre. . Third, it is worth recognizing the difficulty of reforming one element of the public sector in isolation. Many of the problems faced by the health sector were shared by, or arose from, other sectors. It is certainly arguable that single-sector reform may not be feasible, though it can also be argued that a single-sector ‘‘lead’’ may be necessary to pilot new government-wide ap- proaches. However, in these cases, central government support is assumed. Within the health sector itself, it may be that in areas such as resource allocation reform, changes need to be introduced simultaneously across the system rather than through trial districts. . Fourth, while the attempt to improve the resource allocation system may appear as a failed interven- tion, it is important to recognize three positive outputs of the process. First, as a result of the project, technical capacity has improved in a number of areas. Health economics capacity, with practical costing skills, has been developed within the University of Balochistan. This has led to positive collaboration with benefits to all parties and has provided the Planning Cell with access to external technical resources. Second, the skills and interest of district managers has also shown notable qualitative improvement and there is a broader recognition amongst this group of the potentially important role of a district manager under an appropriate decentralized system. Third, the tech- nical work that has been carried out will form the basis for change when there is greater political readiness to adopt genuine decentralization. n Acknowledgements The work reported in this article was carried out as part of a project funded by the United Kingdom Department for International Development (DFID). The views expressed are, however, not necessarily endorsed by DFID. The overall FH2P is additionally cofunded by the World Bank, Kreditanstalt fu¨r Wiederaufban (KfW) and the Government of Balochistan, and has the objectives of improving the health status of the population of Balochistan; increasing the effectiveness of the existing health care network; and building institutional capacity to realize the above objectives and set the stage for future interventions. We acknowledge the helpful com- ments made by reviewers of the article. Re´sume´ Allocation de ressources et me´canismes budge´taires pour des syste`mes de sante´ de´centralise´s : expe´rience du Be´loutchistan (Pakistan) Le pre´sent article de´crit le travail accompli au Be´lout- chistan (Pakistan) dans le cadre du deuxie`me projet de sante´ familiale (composante Renforcement des syste`mes de sante´ du Be´loutchistan). Il s’agissait d’e´laborer un syste`me de sante´ publique permettant d’allouer des ressources et d’e´tablir un budget, qui soit fonde´ sur les besoins et re´alisable malgre´ les difficulte´s techniques locales, comme les lacunes du syste`me d’information. Le syste`me envisage´ devait par ailleurs eˆtre conforme a` la politique du Be´loutchistan consistant a` de´centraliser le secteur de la sante´ publique. Le syste`me actuel est base´ sur une augmentation progressive des allocations budge´taires fixe´es pre´ce´demment et il n’est ni efficace ni e´quitable. Pour reme´dier a` la situation, nous avons cerne´ les principales questions politiques et techniques qui entrent en ligne de compte dans l’e´laboration d’un syste`me plus approprie´ d’allocation de ressources et de budge´tisation. Nous avons commence´ par exposer diffe´rentes me´thodes d’allocation de ressources du niveau central a` la pe´riphe´rie et e´tudie´ les e´le´ments techniques utilise´s pour de´terminer une formule applicable a` un tel transfert. Nous avons pre´sente´ un mode`le conceptuel d’allocation de ressources en fonction des besoins et examine´ les conditions ne´cessaires a` la mise en œuvre de ce mode`le. Nous avons ensuite de´crit le syste`me de sante´ actuel du Be´loutchistan, en mettant l’accent sur les me´canismes budge´taires. Ceux-ci sont complexes, car il y a plusieurs budgets qui sont tous controˆle´s de manie`re diffe´rente. Les de´cisions sont souvent prises pour des raisons administratives plutoˆt que strate´giques et elles peuvent eˆtre influence´es par des facteurs politiques. Il s’ensuit que les budgets sont inefficaces et ine´quitables. Pour e´laborer un syste`me d’allocation plus rationnel, nous avons envisage´ quatre options. Celles-ci sont de´crites, ainsi que les crite`res retenus pour de´gager un accord sur le syste`me qui a finalement e´te´ choisi. Malgre´ cet accord, le syste`me n’a pas e´te´ pleinement mis en œuvre et l’article tente d’analyser les raisons de cet e´chec partiel. On peut tirer les lec¸ons suivantes de l’expe´rience du Be´loutchistan. . Premie`rement, il est facile de sous-estimer les sources et l’ampleur de la re´sistance. C’est la` un obstacle majeur a` l’application d’une nouvelle politique, car le processus de changement qui doit aboutir a` son adoption exige un soutien a` tous les niveaux. Cependant, plusieurs groupes voient dans ce processus une menace, et il est primordial de savoir si l’e´quite´, principal e´le´ment moteur d’un syste`me d’allocation de ressources fonde´ sur les besoins, be´ne´ficie d’un re´el et large soutien. En outre, l’habitude de la centralisation de la prise de de´cision et de la bureaucratie proce´durie`re, combine´e avec la mutation fre´quente de personnel, fait que la Resource allocation and budgetary mechanisms for decentralized health systems 1033Bulletin of the World Health Organization, 2000, 78 (8) de´centralisation met en danger la culture de l’organisation et de la gestion et qu’elle est conside´re´e comme un grand risque. Il faut donc accorder une plus grande attention aux dimensions politiques des projets de de´centralisation et recon- naıˆtre que les re´formes doivent parfois eˆtre plus lentes afin de de´gager une masse critique favorable au changement. De plus, on a besoin d’un chef de file incontestable et puissant qui soit issu du syste`me. Toutefois, l’une des difficulte´s structurelles inhe´rentes aux projets des donateurs est qu’ils tendent a` fonctionner paralle`lement au syste`me gouverne- mental traditionnel. . Deuxie`mement, dans les projets de de´centralisation, il faut appuyer les changements ne´cessaires avec autant de vigueur au niveau central qu’a` la pe´riphe´rie. . Troisie`mement, il convient de reconnaıˆtre la difficulte´ de re´former un seul e´le´ment du secteur public. Bon nombre des proble`mes qui se posent dans le secteur de la sante´ sont communs a` d’autres secteurs, ou proviennent de ceux-ci, et l’on peut soutenir que la re´forme d’un seul secteur n’est pas re´alisable. . Quatrie`mement, si la tentative d’ame´liorer le syste`me d’allocation de ressources peut paraıˆtre un e´chec, il est important de relever les re´sultats positifs du processus. Graˆce au projet, les capacite´s techniques ont e´te´ renforce´es dans plusieurs domaines. On a constate´ une nette ame´lioration qualitative des compe´tences et de l’inte´reˆt des administrateurs de district, lesquels sont de plus en plus nombreux a` reconnaıˆtre le roˆle potentiel- lement important d’un administrateur de district dans un syste`me convenablement de´centralise´. Le travail technique qui a e´te´ accompli permettra d’ope´rer le changement lorsque la volonte´ politique de proce´der a` une ve´ritable de´centralisation sera plus grande. Resumen Asignacio´n de recursos y mecanismos presupuestarios para sistemas de salud descentralizados: la experiencia del Baluchista´n (Pakista´n) En este artı´culo se describe el trabajo llevado a cabo en el Baluchista´n (Pakista´n) como parte del Segundo Proyecto de Salud Familiar (Componente de Fortalecimiento de los Sistemas de Salud del Baluchista´n). El objetivo consistı´a en desarrollar en el marco de la salud pu´blica un sistema de asignacio´n de recursos y elaboracio´n presupuestos que estuviera basado en las necesidades y fuese realizable a pesar de las limitaciones te´cnicas existentes a nivel local, relacionadas, por ejemplo, con el sistema de informacio´n. El sistema tenı´a que ser tambie´n coherente con la polı´tica baluchistanı´ de descentralizacio´n del sector de la salud pu´blica. Actualmente el sistema vigente en el Baluchista´n se basa en un gradualismo histo´rico y no es ni eficiente ni equitativo. Para abordar esta cuestio´n hemos identificado diversas cuestiones polı´ticas y te´cnicas fundamentales para el desarrollo de un sistema ma´s apropiado de asignacio´n de recursos y preparacio´n de presupuestos. Empezamos exponiendo a grandes rasgos distin- tos enfoques para asignar recursos de zonas centrales a zonas perife´ricas, examinando diversas cuestiones te´cnicas relacionadas con la eleccio´n de la fo´rmula ido´nea para determinar esas asignaciones. Se presenta un modelo conceptual para establecer una asignacio´n de recursos basada en las necesidades, examina´ndose paralelamente los requisitos de aplicacio´n de un sistema de esa naturaleza. A continuacio´n se describe el actual sistema de salud del Baluchista´n, prestando especial atencio´n al sistema presupuestario. Es e´ste un sistema complejo, con varios presupuestos, controlados todos ellos de diferente manera. Las decisiones se adoptan a menudo con criterios administrativos antes que estrate´- gicos, y pueden verse influidas por factores polı´ticos. El resultado son unos presupuestos ineficientes y no equitativos. A fin de desarrollar un sistema de asignacio´n ma´s racional, consideramos cuatro opciones. Se describen dichas opciones, ası´ como los criterios utilizados para llegar a un acuerdo respecto al sistema finalmente elegido. Pese al acuerdo logrado, el sistema no se ha llevado a la pra´ctica en su totalidad, por razones que se intenta analizar en el artı´culo. Nuestra experiencia en el Baluchista´n nos ha ensen˜adovarias lecciones, segu´n se resumeacontinuacio´n. . En primer lugar, es fa´cil subestimar las causas y la magnitud de la resistencia a las medidas. Este factor dificulta sobremanera la aplicacio´n de una nueva polı´tica, pues un cambio tal requiere apoyo a todos los niveles. Sin embargo, diversos grupos ven en el proceso de cambio una amenaza, y una cuestio´n decisiva es si existe o no un aute´ntico apoyo generalizado en favor de la equidad, concepto que constituye la principal fuerza impulsora de un sistema de asignacio´n de recursos basado en las necesidades. Adema´s, la existencia de una cultura de centraliza- cio´n de la adopcio´n de decisiones y de una burocracia dependiente de procedimientos, unida a los frecuen- tes traslados de personal, hacen de la descentraliza- cio´n tanto un desafı´o para la cultura de la organizacio´n y la gestio´n como un proceso de alto riesgo. Ası´ pues, es necesario prestar ma´s atencio´n a las dimensiones polı´ticas de esos proyectos, y admitir que a veces hay que frenar el ritmo de las reformas para poder lograr una masa crı´tica favorable al cambio. Adema´s, hay que disponer de un aliado inequı´voco y firme dentro del sistema. Sin embargo, una de las dificultades estructurales que plantean los proyectos de los donantes es que tienden a funcionar paralelamente al sector principal de la Administra- cio´n. . Segundo, en los proyectos de descentralizacio´n hay que procurar apoyar tanto los cambios necesarios en el centro como los requeridos en la periferia. . Tercero, conviene reconocer las dificultades que supone intentar reformar por separado un elemento del sector pu´blico. Muchos de los problemas afrontados por el sector de la salud afectaban Policy and Practice 1034 Bulletin of the World Health Organization, 2000, 78 (8) tambie´n a otros sectores, cuando no procedı´an de ellos, lo que respaldarı´a la idea de que la reforma de sectores aislados quiza´ no sea viable. . Cuarto, si bien puede parecer que el intento de mejorar el sistema de asignacio´n de recursos ha sido una intervencio´n fallida, es importante reconocer los resultados positivos del proceso. Ası´, como conse- cuencia del proyecto, la capacidad te´cnica ha mejorado en varias a´reas. Las aptitudes y los intereses de los administradores de distrito han experimentado mejoras cualitativas, y entre esas personas hay una mayor conciencia del importante papel que puede llegar a desempen˜ar un administrador de distrito en un sistema descentralizado. El trabajo te´cnico que se ha llevado a cabo sentara´ las bases para aplicar los cambios cuando exista una mayor voluntad polı´tica de proceder a una verdadera descentralizacio´n. References 1. Green A et al. Health planning in Pakistan: a case study. International Journal of Health Planning and Management, 1997, 12: 187–205. 2. Segall M. Planning and politics of resource allocation for primary health care: promotion of meaningful national policy. Social Science and Medicine, 1983, 17: 1947–1960. 3. Collins C. Management and organization in developing countries. Oxford, Oxford University Press, 1994. 4. Cassells A, Janovsky K. A time of change: health policy, planning and organization in Ghana. Health Policy and Planning, 1992, 7: 144–154. 5. Mills A et al. Health system decentralization: concepts, issues and country experience. Geneva, World Health Organization, 1990. 6. Collins CD, Green AT. Decentralization and primary health care: some negative implications in developing countries. International Journal of Health Services, 1994, 24: 459–476. 7. World Bank. World development report 1993. Investing in health. New York, Oxford University Press, 1993. 8. Mahapatra P, Berman P. Resource allocation for public hospitals in Andhra Pradesh, India. Health Policy and Planning, 1995, 10: 29–39. 9. Murray CJL, Kreuser J, Whang W. Cost-effectiveness analysis and policy choices: investing in health systems. Bulletin of the World Health Organization, 1994, 72: 663–674. 10. Chae YM et al. Application of goal programming to improve resource allocation for health services in Papua New Guinea. International Journal of Health Planning and Management, 1989, 4: 81–95. 11. Kalumba K. Towards an equity-oriented policy of decentraliza- tion in health systems under conditions of turbulence: the case of Zambia. Geneva, World Health Organization, 1997 (unpublished document WHO/ARA/97.2). 12. Doherty J, van der Heever A. A resource allocation formula in support of equity and primary health care. Johannesburg, University of Witwatersand, Centre for Health Policy, 1997, Paper No. 47 (248–250). 13. Resource Allocation Working Party. Sharing resources for health in England. DHSS, London, 1976. 14. Mays N. Measuring need in the National Health Service Resource Allocation Formula: standardized mortality ratios or social deprivation? Public Administration, 1987, 65: 45–60. 15. Carr-Hill R, Sheldon R. Rationality and the use of formulae in the allocation of resources to health care. Journal of Public Health Medicine, 1992, 14: 117–126. 16. Collins CD et al. International transfers of National Health Service reforms: problems and issues. Lancet, 1994, 344: 248–250. 17. Government of Baluchistan. Ready reckoner of health statistics. Quetta, Statistics Cell, Health Directorate, 1996. 18. Government of Pakistan. Pakistan Integrated Household Survey: final results. Islamabad, PIHS Project, Federal Bureau of Statistics, 1991. 19. Government of Baluchistan. Situational analysis of health in Baluchistan. Quetta, Planning Cell, Health Department,1995. 20. Government of Baluchistan. Situational analysis of health in Baluchistan. Quetta, Planning Cell, Health Department, 1993. 21. Tarin E, Thunhurst C. Community participation — with provider collaboration, World Health Forum, 1998, 19: 72–75. 22. Barkat A et al. Research study: unit costs of primary level health care facilities. Quetta, Second Family Health Project, British Council, 1996. 23. Barkat A, Naeem A. Research Study: Unit costs of Divisional and District Hospitals. Quetta, Second Family Health Project, British Council, 1997. 24. Government of Baluchistan. District planning and budgeting guidelines. Quetta, Planning Cell, Health Department, 1998. 25. Government of Baluchistan. District planning and budgeting manual. Quetta, Planning Cell, Health Department, 1998. 26. Walt G, Gilson L. Reforming the health sector in developing countries: the central role of policy analysis. Health Policy and Planning, 1994, 9: 353–370. Resource allocation and budgetary mechanisms for decentralized health systems 1035Bulletin of the World Health Organization, 2000, 78 (8)