PRS Monitoring & Synthesis Project
1Briefing Note 4
Costing Poverty Reduction Strategies – Early Experience
A review of approaches to costing in 10 full PRSPs and 4 Progress Reports reveals that:
； All the PRSs provide some estimate of the cost of implementing PRS objectives and the
size of the subsequent financing gap. In most cases costings cover priority actions and
core sectors for a 3 to 5 year period. Costing exercises focus mainly on the cost of
achieving intermediate indicators rather than final outcome targets.
； A few PRSs attempt a comprehensive costing of priority actions (Malawi, Rwanda) but a
number also only cost the „additional‟ activities to be financed by HIPC funds and include
only HIPC resources in funding estimates (Burkina Faso). Some provide a breakdown of
recurrent and capital costs (Uganda, Malawi, Rwanda); others give basic unit cost
information (Uganda. Mozambique).
； Most costing is carried out „bottom-up‟ using activity-based unit cost estimates. Problems in
coming up with credible estimates stem from generally weak prioritisation, limited unit cost
data, inadequate links between programmes and indicators/targets (including in sector
programmes) and over-ambitious medium term targets.
； Credible costing depends on clear links between individual programmes and intermediate
indicators/targets, realistic medium term targets based on past performance rather than
politically determined aspirations and a closer integration of PRS objectives with an MTEF
(or similar forward budgeting framework).
； Some confusion exists over whether the role of costing is to work within financial
constraints imposed by the available resource envelope (as in the MTEF) or to estimate
the full extent of the fiscal gap between a country‟s poverty reduction needs set out in the
PRS and available resources.
Ten full PRSs (Burkina Faso, Mauritania, Mozambique, Malawi, Uganda, Tanzania, Rwanda, Honduras, Nicaragua, Bolivia) and 4 Annual Progress Reports (Burkina Faso, Uganda, Tanzania & Mauritania) were reviewed. Main Questions: How many PRSs contain costings? How complete are they? Any good examples of method? Any examples of best practice? What is useful/what is the required level of detail? What would be useful for Governments to know in order to cost PRSs?
The level of detail on costing found in most PRS documents is modest. Much relevant important information remains outside of the document and is hard to get hold of within the confines of a short review exercise. Limited information is provided on method. Cost estimates are only briefly commented on by JSAs. Costing issues are also hard to treat
1 Other Briefing Notes in 2001: Briefing Note 1 - PRSC - Vietnam and Uganda ;
Briefing Note 2 - Experience with the PRS process in South and South East Asia;
Briefing Note 3 – Board Responses to PRSPs.
separately from how targets are set, how indicators are chosen and the current state of PEM/MTEF reforms.
A crucial element of the PRS approach is assessing the fiscal implications of reaching medium and long-term poverty reduction targets. This is important in at least three ways:
(i) At the macro level, evaluating proposed poverty reduction measures in the light of
the expected resource envelope, including increments provided by HIPC;
(ii) Evaluating the fiscal viability of long run PRS targets and hence the fiscal
sustainability of the overall strategy;
(iii) At sector level, assessing the „relative‟ cost effectiveness of alternative
interventions in contributing towards a specific policy outcome (or target), and aiding
prioritisation between them (although not in isolation from an analysis of demand and
supply conditions/market failures etc), including making the case for extra funds
Three alternative approaches to costing PRS goals are:
(i) Average Costs: The cost of achieving a subset of goals using estimated average
costs of service and estimated population lacking access to priority services. Data
sources include: activity based budgeting and census/DHS/LSMS data for unserved
(ii) Multivariate Analysis/Determinants Analysis: Determinants of key poverty
outcomes (e.g. infant mortality) are used for sensitivity analysis with respect to
different policies. Alternative policies can be translated into unit costs, with total costs
calculated for attaining certain goals.
(iii) Cost Effectiveness: Ranking of existing projects and programmes according to the
relevance of their objectives to the PRS. Priority services and target groups identified.
Information obtained from sectors and donors about the most cost effective designs
for priority social services, including the costing of main and complementary inputs
Most of the PRSs reviewed rely on a simplified variant of the cost effectiveness approach. None have attempted the more complex determinants analysis, presumably because of the 2considerable data requirements and technical complexity.
Developing realistic and comprehensive cost estimates presents a number of methodological challenges for national authorities. The information requirements are substantial and often complex (especially in the social sectors where inputs and outputs are hard to measure), made more so by the fact that few of the parameters determining the cost of public actions are constant over time i.e. input prices change as do delivery systems making estimating the costs of reaching medium to long run targets complicated. Wage costs also tend to loom large in recurrent costs especially in priority sectors such as education and health, hence the need to be explicit about assumptions made regarding public sector wages and pay reform (see Uganda). Other reform processes, such as decentralisation, can also affect the cost of reaching a target, often during the life of the PRS itself.
2 It is not clear how much more value-added such an approach would provide to policymakers in the short run, although it is an approach that can offer more in terms of the identification of trade-offs and key inter-linkages within the overall strategy i.e. that the cost of reaching a target in health may lie equally with interventions in water as with those directly in health.
These challenges are not insurmountable. As this review finds a number of countries already have reasonable cost estimates for priority programmes because of existing SWAp arrangements and/or sectoral PER/MTEF processes. However there is still some way to go before all PRSs contain a complete estimate of the fiscal cost of medium and long-term development targets.
Table 1 (Annex 1) summarises the information available on costing from 9 full PRSs, 1 draft PRS (Rwanda) and 4 PRS Progress Reports (PPRs) (Burkina Faso, Uganda, Mauritania and Tanzania). This section summarises some of the main points.
How many PRSs contain costings? All of the 10 PRSs include some cost estimates. Of the 4 PPRs, two (Uganda and Tanzania) include quite detailed costing exercises that add important information over and above what was initially available in the full PRS.
How complete and detailed are they? Coverage varies.
； Most give broad cost estimates for selected poverty reduction measures or
programmes in priority sectors that are to be financed, at least in part, from HIPC
resources (Mauritania, Burkina Faso, Bolivia).
； Some attempt an overall estimate of the cost of the PRS covering existing and new
programmes (Uganda PPR, Honduras, Malawi, Rwanda), or priority/selected sectors
(Tanzania PPR, Burkina Faso, Nicaragua, Bolivia). Annex 1 Table 2.
； In a number of cases, cost estimates are disaggregated into recurrent and capital
costs for key sectors (Mauritania, Bolivia) or individual programmes/activities
(Uganda PRS & PPR, Malawi, Rwanda).
； Basic unit cost information is provided mainly for sectors with SWAps or SWAp-like
mechanisms (Mozambique, Uganda). In Tanzania the cost of health and education
programmes is based on an estimate of the ‟basic unit of service‟.
； None of the current round of PRSs makes a direct link between the cost of
programmes and the cost of directly achieving PRS targets. Most specify programme
costs relating to broad PRS themes or objectives such as “opportunities” or
“protection and security” (Bolivia, Malawi). Some PRSs limit the presentation of
public expenditure to „incremental activities‟ to be financed by HIPC debt relief and
external financing (Burkina Faso, Honduras). This „new project‟ approach precludes
consideration of how the existing expenditure programme might be reshaped, even at
the margin, to improve its cost effectiveness and impact on poverty reduction.
What information is provided on method? Only a few PRSs provide a description of the
method used to arrive at cost estimates:
； In Uganda, costings are based on ongoing work by the MFPED and line Ministries to
develop medium term poverty reduction targets, costings and priorities based on
sector plans. Final prioritisation is by Cabinet through the annual budgeting process.
Making fiscal costs fit within the available resource envelope calls for prioritisation
and revision of targets on an iterative basis. In the PPR the total cost of the PEAP is
considered unrealistic (resulting in a 37% financing gap!”, largely because of the cost
of the proposed pay reform. The PEAP programme is therefore under further revision
and reprioritisation to bring the estimated fiscal cost more in line with budget
； In Tanzania the costing of priority interventions is based partly on the analysis of
recurrent cost implications of sector programmes, and inputs from the PER and
MTEF. Technical studies defined „basic units of service‟ in health, education, water
and transport and their estimated costs. From these financing requirements were
derived for the medium term, subject to the available resource envelope. Special
attention was given to non-wage current outlays and development expenditure for key
interventions needed to attain PRS objectives.
； In Mozambique the JSA notes that cost data are not centralised, not uniform and not
easy to obtain outside of multi-donor funded sector programs (SWAps). In this
context the proposals for financing the PARPA have mainly relied on the
establishment of budget targets for the priority areas, based on a projected budget
envelope for 2001-5. Unit costs are provided for those sectors that have them
(education, health and infrastructure) as an illustration of future work to be
undertaken rather than as underpinnings of the expenditure programme.
； In Bolivia estimates of the financing required by the BPRS were performed on the
basis of the current structure of public investment programmes and projects,
projected revenues and expenditures and disbursements and grants. For the base
year 2001, projects in the PIP were classified according to the components and sub-
components of the BPRS and priority projects were identified for each component.
Financing requirements were then estimated for each component. The JSA notes that
the cost estimates are very general and relate only to the four pillars of the strategy
with no clear link to targets. Costing of the BPRS is also complicated by the fact that
municipalities will have considerable discretion to finance activities that are of high
priority at the local level, financed by both revenue-sharing funds and HIPC resources.
Unit cost data at the municipal level is currently only partial.
； In Rwanda the priority programmes for the PRS are a subset of the programmes
used in the MTEF. For each programme, the relevant agency identified and quantified
sub-programme activities, outputs and inputs. The specification and unit costing of
these was expected to be the same as for the MTEF. Unit costs were defined for
each input. The costing was to be guided by what was physically feasible, not by the
financial constraint (as in the MTEF). Hence the costing for the PRS was expected to 3be much larger than the existing resource envelope. The final costing also included
information on existing donors projects within priority programmes.
Are PRS costings linked with/to MTEFs? It is difficult to tell to what extent cost estimates presented in PRS documents are closely linked to those prepared as part of MTEF exercises. Not all of the countries included in the sample have MTEFs (Mauritania, Honduras, Nicaragua, Bolivia don‟t) and of those that do, few are fully functioning (Uganda is the exception). Few
PRS documents are clear about the linkages with the MTEF, although JSAs frequently mention links with the MTEF where they are relevant.
； Cost estimates are linked to an existing MTEF or preparations for an MTEF in 6
cases (Uganda, Tanzania, Burkina Faso, Mozambique, Rwanda and Malawi).
； In Tanzania the PER/MTEF process has been used to identify sector unit costs,
which are the same as those presented in the Progress Report. In Rwanda the
specification and unit costing of each activity was expected to be the same as for the
； In Malawi the PRSP costing was linked to the budget classification so that PRSP
expenditures could be easily tracked through the budget.
3 This approach raised a number of difficulties with the IFIs. The final version of the PRS now contains two „costed scenarios‟ – one subject to financial constraint the other unconstrained by the gross resource envelope. However the Rwandan approach invites the question of the role of the costing exercise. Is the role to match the policy priorities with current resources or to identify the overall fiscal gap between current resources and what is required to support a fully comprehensive poverty reduction strategy?
4A recent review of MTEFs in Africa by the Africa Region of the World Bank found that only a
subset of countries produce costings with data on programmes or activities and, of these, many are at an aggregate level (for example, the Mozambique health sector costs only three general programs „improving health service provision quality, improving health institutions and developing human resources‟) and some only include recurrent costs. In practice in most countries only a subset of sectors have produced costings to date, many of these are covered by SWAp type arrangements, although progress is being made (gradually). Examples from the review include:
； In Uganda sectoral objectives are presented in the PRS but sectoral expenditure
frameworks (SEFs) vary considerably in quality, and only some SEFs include
； In Tanzania, SEFs include strategies, objectives and priorities but again vary
considerable in quality. Some don‟t present detailed programme costings others don‟t 5present any costings. There is no standardised format.
； In Rwanda three sectors present some costings of different quality and there is
currently no standardisation.
； In Mozambique, five priority sectors present SEF costings based either on activities
or programs but there is no standardised format and most non-priority sectors present
costings according to their internal organisational structures.
Any examples of good practice? It is too early to comment on examples of best practice. However, in most cases countries are making the best of the information they have available and there are some examples of what might be called „improving practice‟.
； The Tanzania PPR shows how much can be done between the preparation of a full
PRS and the first annual progress report, particularly where there is a fairly
comprehensive PER/MTEF process in place. Much of the work for the cost estimates
in the PPR was done in the context of studies prepared for the MTEF. However, as
the JSA on the PPR notes, links between budget inputs, outputs and final outcomes
need to be further investigated, and further advances in the costing of achieving
proposed targets will require the finalisation of sectoral strategies as well as new
information not currently available.
； The Uganda PEAP and PPR present between them probably the most
comprehensive set of costings for a PRS, with disaggregation by recurrent and
capital costs for a number of key sectors. The quality of the estimates is variable,
however, with some program costings giving significant detail and others providing
only rudimentary estimates. The estimated shortfall between the total cost of the
PEAP and available financing (37%) illustrates the importance of a strong
prioritisation process and setting realistic sector goals and targets. A key implication
is the importance of basing medium term targets on existing information and past
performance rather than on politically determined „aspirations‟.
； The Malawi PRS gives a comprehensive costing of priority activities (and resources),
with targets and costs adjusted in the light of current implementation capacity.
Costing covers all Government and development partner activities except for a
number of large-scale infrastructure development projects that are treated separately
because they are generally funded by development partners and have high lumpy
costs. Inadequate information about some areas of donor activity is a problem.
4 Medium Term Expenditure Frameworks in the Africa Region: Preliminary Lessons from Experience. Draft August 2001. Phillipe Le Houerou and Robert Taliercio World Bank. 5 This has shifted somewhat with the work carried out for the PPR in November 2001. All priority sectors now include costings, except for agriculture which is still waiting for the finalisation of its sector strategy.
Costings are provided within each pillar of the MPRS, at goal and sub-goal level by
recurrent and development expenditure. A major weakness is that there is little
indication of how the gap between the gross resource envelope and the total cost of
the MPRS will be financed (as much as 10% of the total cost of the MPRS in 2002/3
While national governments have generally made the best of limited data and experience in costing PRSs, there is clearly still some way to go before cost estimates can be considered a reliable assessment of the fiscal impact of achieving PRS targets.
Among the issues affecting the quality of cost estimates include:
； Unclear rationale for prioritisation of interventions within or across key sectors
； Absence of sector strategies/plans or sectoral PERs providing credible unit cost
； Non-standardised formats for calculating costs within sector and donor programmes
； Limitations of unit cost data, particularly where complex services are being provided
； Over ambitious medium-term PRS targets
In addition there is some confusion in PRS documents as to whether the role of costing is to „constrain‟ priority actions to the gross resource envelope or to estimate the full scale of the
fiscal gap between the country‟s poverty reduction needs and available resources (a much larger figure). The Rwanda PRS is the clearest example of adopting the latter approach but one that subsequently ran into difficulties with the IFIs because of conflicts with crucial macroeconomic targets. The final document now contains two costing scenarios, one constrained by assumptions about future donor flows and domestic revenue growth, the other unconstrained by any financial limit.
Even where cost estimates are of reasonable quality, as in the case of Rwanda and Uganda, coming up with an overall estimate of the fiscal cost of reaching medium term targets is a major challenge requiring reprioritisation and reformulation of activities and programmes over time. It is therefore important not to view the process of costing in isolation. As Uganda has found, even if programmes are fully costed and embedded within the MTEF, there are still weaknesses in government capacity that prevent the implementation of programmes necessary to attain the targets. As the Uganda progress report explains in relation to a number of performance targets missed at the end of the first year “In several cases they represented costed input or output targets under the MTEF, where it is normal to assume that fiscal resources will be converted efficiently into inputs and outputs… In retrospect it is not surprising that the fast expansion of social services has caused implementation problems…”
The issue is two fold: first, although it may not be appropriate to impose too heavy a financial constraint on the initial costing exercise (so that the full extent of the fiscal gap can be fully revealed), constraints on physical implementation need to be taken full into account. Second, PRS targets are often highly ambitious and rarely based on evidence of past performance. This places huge stress on implementation and renders the costing process less than useful to serious policymakers.
Unlike the development plans of the 1950s and 1960s, the value of PRSs is that they aim to link intentions with budgets. Without adequate attention being given to cost estimates for PRSs and without the preparation of a realistic medium to long term public expenditure programme, the chances of the new PRS approach succeeding are significantly reduced.
In addition, in order for donors to be more persuaded to increase programmatic support for PRSs, governments need to move rapidly towards providing credible fiscal cost estimates.
Implications that might be of use to Government‟s embarking on a costing exercise are:
； In the initial phase it is likely to be difficult to come up with the unit costs of attaining
strategy targets, therefore in the short run use estimates of the overall financing
requirement for existing and new programmes (preferably disaggregated by
recurrent/capital costs) and aggregate these according to PRS objectives and goals.
Be clear that these are estimates of aggregate financing requirements not based on
detailed unit cost information.
； Commission technical studies on costing, preferably as part of an ongoing PER
and/or MTEF review. Focus on priority sectors first and gather as much information
as possible on service/activity unit costs, delivery system and sector demographic
variables. This will eventually allow calculation of the unit cost of reaching key
programme output/outcome targets. These can then be used to assess the cost of
attaining the more ambitious sectoral targets contained in the PRS. Link costings to
the budget classification in order to track future PRSP expenditures.
； As the full cost of attaining targets becomes known, it is vital to reconsider the cost
effectiveness of programmes and activities proposed (drawing on line Ministries and
external partners), as well as evaluating the viability of the targets themselves given
current and projected budget constraints.
； In the long run, effective costing requires better links between programmes and
intermediate indicators/targets, close integration of PRS objectives with an MTEF (or
similar forward budgeting framework), a fully integrated PIP, and a process of target
setting that is evidence based and transparent.
The limited evidence available regarding costings suggest that partner countries need assistance to improve the analytical basis of cost estimates. Given capacity constraints within partner countries and the technical difficulty associated with estimating costs and expenditures, it would be useful if the World Bank and donor community were to increase their support in this area. But considerable care is needed. External technical support should be relevant, appropriate and should not stifle or replace local efforts. Current guidance available in the PRS Sourcebook is not particularly helpful to capacity constrained national authorities. More intuitive guidance setting out key stages in the costing process, the types and sources of essential information, which stakeholders to involve and what stage and key linkages with MTEF/PER processes would be much more useful guide, particularly at the preliminary stages. Ultimately national authorities have to assume responsibility for producing robust and politically defensible estimates.
PRS Any How Complete? How Detailed? What Method? Any link with Good Comments
Costing? MTEF? Practice?
Burkina Faso Yes (i) Table showing No breakdown of Unit cost analysis Costs based in part JSA states „ The costing
additional costs of poverty recurrent/capital prepared for on preliminary work of the PRS action plan
reduction measures in costs; no discussion existing sector done for the 2001-3 is considered reliable
priority sector progs. to be of wage vs. other programmes. No MTEF. because it draws upon
financed largely by HIPC; recurrent costs; discussion of the long-term strategies
(ii) Annexes giving overall sector programme method or under finalisation in
cost of existing priority costs include physical assumptions re: priority sectors…..and
sector progs. and targets & unit costs technical efficiency. on preliminary work
projected amounts for but financial done, with Bank staff,
2001-3; projections seem to on the 2001-3 MTEF”
(iii) No clear link assume stable unit PRS Progress Report
established between total costs except for June2000-2001 notes
programme costs and annual 3% inflation continued efforts to
PRS targets. adjustment. improve budgeting
no further mention of
Mauritania Yes (i)Table showing program Estimates of Nothing on method No MTEF at time of JSA states „The costs of
of priority actions and recurrent/ capital Not clear what PRS preparation. the priority programmes
additional costs 2001-4. costs in social sectors assumptions are appear realistic but will
Linked to broad objectives are provided but not being made re: need to be reviewed as
but not targets per se. as part of costs of technical efficiency, PRS implementation
(ii) Exposition of additional financing priority input prices etc. proceeds and more
financing requirements to actions. detailed sectoral
meet costs of „additional‟ Note: PIP separate. knowledge become
priority actions No MTEF. available! After HIPC
(presumably based on and PIP financing
above unit costs) already committed, GoM
estimates financing gap
of close to $300million.
Targets ambitious. Mozambique Yes (i) Additional expenditure Rudimentary unit No discussion of Public expenditure JSA states: cost data
requirements for priority costs mainly for method. Cost data requirements are not centralised and
areas 2000-2005. sectors with SWAPs are not centralised anchored in MTEF uniform, and not easy to
(ii) Some average unit or SWAP like and uniform, and 2002-5, although obtain outside of multi-cost data for education, mechanisms not easy to obtain MTEF not fully funded sector programs health and infrastructure – (education, health, outside of multi-linked up with (SWAPs). These issue indicative of future work water/roads). funded sector annual budget are now being rather than as No disaggregation programs process. addressed in the context underpinning expenditure between (SWAPs). of the participatory PER program capital/recurrent or and new Public (ii) No direct link made clear idea of share of Financial Management between cost information wage costs in Law. In this context, and achievement of average unit costs. proposals for financing targets. of the PARPA have
mainly relied on the
establishment of budget
targets for the priority
Malawi Yes (i) Table showing total More detailed MPRS activities Linked to MTEF. Comprehensiv
PRS costings by pillar estimates of costs by defined in large e treatment of 2002-3 to 2004-5 objective categorised part through a Not clear whether resources and
(ii) Table showing more by recurrent & bottom-up the aggregate cost is a step detailed costings by pillar development approach. The resource constraint forward.
(e.g. sustainable pro-poor provided as annex. needs of the poor is the same as the Disaggregation growth), goal (e.g. were defined first MTEF (although by activity and sources of pro-poor MPRS costing is and strategies the implication is by recurrent/
growth) and sub-goal (e.g. comprehensive, designed to help that it is). development
increasing agricultural covering all them reduce their provides a
incomes). Government and poverty. These MPRS matrix clear picture of (iii) Resource gap based development partner activities were then outlines the overall
on estimate of gross activities (except for a costed, where responsible financing gap
resource envelope (based number of large possible using unit institution(s) for (although lack on total domestic tax, non-infrastructure costs applied to each activity so that of sufficient
tax revenue and development projects relevant targets eg: costing can be information
conservative estimates of that are expected to in education the easily translated about donor
donor inflows) minus be financed by total teacher wage into institutional flows is a
statutory & statehood donors/private sector bill was derived by allocations for problem).
expenditures and total and exist outside of multiplying the comparison with
MPRS costings. the budget). required number of the Budget.
teachers in a year
Resource envelope by the cost of
includes all sources paying each
of funding – domestic, teacher in that grants, loans, HIPC year.
etc. Note: not all
donor funds are Targets and costs covered because not were adjusted to all are included in the ensure realism, budget. GoM plans especially regards an suirvey of ongoing institutional projects and future capacity. Once all commitments to activities were ascertain the extent costed, the total to which MPRS cost of the MPRS activities are already was derived.
Next activities were
rescaled so that
total costings were
in line with total
Not clear how unit