Assessing Frontline Service Delivery in Education Ritva Reinikka World Bank - DECRG HD week 2002
Motivation and Context Public spending traditionally analyzed from efficiency and equity viewpoint using budget allocations data Government performance becoming more of an issue Relationship between public spending and outcomes ambiguous at best –Large literature on growth and education and health outcomes
Identification Problem Budget allocations poor predictors of services beneficiaries receive when institutions are weak Problem of identification Household survey evidence shows that when measured by actual output education or credit, services important for poverty reduction The “ black box ” of service delivery
how to get this: The question is Real outcomes Improved learning Better health status Welfare –distribution –risk From this: ? Government expenditure
services Public providers Government expenditure Real outcomes Learning Health status Welfare –distribution –risk Private providers Household behavior Things are not that easy
Provision of Services Financing and provision are two key aspects of service delivery –Until now financing been given most attention –Provision becoming more of an issue (WDR 2003/04) Different types of providers –Government –Non-profits –Private for profit
Provider surveys Public expenditure tracking surveys (PETS) –Information on actual spending is scarce –Follow money through different tiers of government down to frontline facilities using sample survey techniques –Interviews and data collection from records Quantitative service delivery surveys (QSDS) –Frontline service providers/facilities in basic services –Inspired by micro-level household and firm surveys –Cover different type of providers (by ownership) –Triangulation of data
Uses of provider surveys PETS/QSDS can be combined –With each other –Downstream with household surveys –Upstream with public officials surveys Diagnostic tool –Stylized facts of service provision –Quantify “ moral hazard ” (asymmetric information) –Capacity building when a collaborative effort
Policy research Primary data for empirical research –Applies know microeconomic approaches to evaluation of public spending –Design of incentives –Decentralization –Voice, asymmetric information –Accountability and oversight –Participation of users and partnership with others
Uganda PETS Aggregate spending kept in check since 1992 (cash budgeting) and allocations improved in the late-1990s Little information on actual spending & service delivery –Only a perception survey of households –Diagnostic objective initially but panel also allowed research Hypothesis: primary enrollment stagnant despite a large increase in public spending No accounting information available on frontline service delivery units
Uganda PETS: school survey Survey of 250 schools to collect detailed quantitative data from school records Not possible to do on all spending items, sectors, or tiers of government –Aggregated salary data available at the center –Little data at the local government level and not forthcoming from officials Schools kept relatively good records
Uganda school survey Only 13% of non-wage expenditure reached the schools in on average Variation between schools –Statistical analysis shows that leakage depends on school characteristics (size, income, share of qualified teachers) PTA primary source of school funding Enrollment increased much more than indicated by national statistics (60% in 5 years) Health facility survey also carried out but didn ’ t work
Uganda PETS findings Responsibility for primary education delegated to districts –Decentralization initially worsened leakage Poor oversight by central government Allocation of resources based on relative bargaining power rather than efficiency and equity considerations
Impact Power of systematic information vs. anecdotes Information on transfers of funds to local governments regularly published in national media since 1996 Posters at schools to inform citizens about school-level funding from central government Publicity also signals central government oversight In 2000 and 2001 PETS carried out locally –80-90% of non-wage spending reaches schools but delays Tracking surveys expanded to other basic services
Tanzania PETS Accounting firm (1999), and NGO and research institute have implemented (2001) as part of PER Track pro-poor expenditures in priority sectors at all levels Combinations of documents, records, facility visits, interviews In 2001, for example, survey of 5 districts, 4 primary schools and 4 clinics in each districts (small sample)
Tanzania PETS Findings Non-wage leakage 57% in education 41% in health (1999) Larger delays in rural areas and non-wage recurrent (rather than salaries) Priority to council departments rather than service facilities Cash budgeting and aggregated records undermine transparency Information asymmetry, e.g. local administration versus parents
Ghana PETS Primary and secondary school and health facility pilot surveys –Recall method instead of records (not preferable) Central government and district level 80% of salary and 50% of non-salary expenditures reached primary schools Only 20% of public health spending reached districts
Honduras PETS Moral hazard and frontline health education workers –Ghost workers 2% in health;3-5% in education –Absenteeism 27% in health; 14% in education –Job migration 5% in health
Follow-up Cannot be overemphasized Uganda followed up immediately producing a huge improvement Tanzania initiated information dissemination and is beginning a public awareness campaign In Ghana little follow-up In Honduras little follow-up
Pilot Round of QSDS Education Laos: household survey link Papua New Guinea Uganda: information and voice Zambia: household survey link
Lessons from provider surveys Can be used to study different problems in service delivery Diagnostics of leakage in Africa Staff behavior in Honduran social services –Ghost, absenteeism, job capture by employees –Rigorous sampling, pre-testing required Qualitative approach yields hypotheses, quantitative surveys diagnosis and analysis