Equality of Opportunities from a Fiscal Perspective: Education in Liberia January 26, 2010 Jose Cuesta and Ana Abras PRM PR April 25, 2011
We measure EqOpp, so what? Diagnostics serve several purposes: Understand distribution of opportunities Monitoring over time, across regions, internationally Understand key obstacles to universal access BUT how to go beyond diagnostics? Fiscal side of policy interventions Pilot: education spending in Liberia
Education in Liberia: Spending Liberia Education CSR2010 & PEMFAR’09: – At 3.2% of GDP public education spending in Liberia is low compared to SSA averages – But has doubled from 2004 to 2008 – International aid (2007): US$ 38 million – Household spending: US$ 27 million – Public spending on education: US$ 12.2 million – In 2008: US$ 15 million (aid) and US$ 23.6 million
Education in Liberia: Policies WB CSR 2010 Policy Recommendations: Increase enrolment at age 6: abolish entry exams Narrow regional disparities: Prioritize spending into poor areas; school lunches Reduce gender disparities: provide scholarships to needy girls in targeted areas Increase budget in 2ry as enrolment increases Improve quality of education: training & certification Improve management: HHRR database for teachers linked to payroll
Public Education in Liberia: Interventions Sectoral interventions (FTI-Catalytic Fund 2010): – School construction – Text books for grades 4-9 – School grants – School health (de-worming) – Management: community involvement and payroll management – Revise teachers’ salaries and incentives to rural areas
How can HOI fiscal analysis help? Supplementing the distributive dimension of sector diagnostics – From BIA to Opp BIA, focusing on opportunities rather incomes/expenditures/wealth Linking (the fiscal side of) policy proposals with improvements in opportunities – Simulate the effect of budgetary changes and or composition on opportunities
What can and cannot do? – The analysis focuses on fiscal dimension: Amount and Distribution of resources (by level, region, gender) – Most useful to analyze increases in spending (increasing teachers’ salaries), targeting to poor areas or poor girls, elimination of fees, free text books, changes in international aid – Cannot say much on issues such as de-worming, quality training, management reforms
MODELLING PROBABILITY – Estimate a logit model, the dependent variable is the opportunity (attending school age 6-15) and independent variables are the circumstances (child gender, hh head gender, education and age, region, u/r, number of children in hh, single parent, mother alive, father alive. Use CWIQ 2007
Linking HOI with Fiscal Policies Fiscal Simulations: Cost & efficiency implications of improving Educational Opportunities TRADITIONAL BIA: – What is the distribution of public spending across income/asset levels? “OPP” BIA – What is the distribution of public spending across opportunity groups? SIMULATION: How much would an additional dollar spent on education affect the distribution of educational opportunities among children? How much would it cost to close access gaps across children? Benefit Incidence Analysis and Opportunities
Benefit Incidence Analysis STEPS FISCAL / SECTOR DATA Public spending on education by level (not possible by region) Total beneficiaries Unitary gross benefit (per beneficiary) HH DATA Private out of pocket contributions per beneficiary FINALLY Unitary Net benefit per beneficiary PEMFAR 2009 Q1 (PoorestQ2Q3Q4 Q5 (Rich est) Unit cost household spending on education Unit cost government spending on education Primary education Secondary education Higher education Total Per Student Expenditure on education, by Quintile and Level of Education, 2007 (US$)
Distribution of unitary cost of primary school Quintiles of wealthQuintiles of opportunities Quintiles of wealth: Q1: ; Q ; Q ; Q ; Q Quintiles of probability: Q1: ; Q2: , Q3: ; Q4: ; Q5: Public spending in education is progressive but for the wrong reasons
Who is getting how much? Share of public spending on education captured by group and probability of access Progressive but clearly not pro-poor: Opp BIA shows it more clearly
FISCAL POLICY SIMULATION Sim- STEP 1: MODELLING PROBABILITY – Estimate a logit model, the dependent variable is the opportunity (attending school age 6-15) and independent variables are the circumstances (child gender, hh head gender, education and age, region, u/r, number of children in hh, single parent, mother alive, father alive Sim – STEP 2: SPENDING AS CIRCUMSTANCE – Logit model expanded to include gross unitary benefit as circumstance – A new probability is estimated for each household
FISCAL POLICY SIMULATION Sim – STEP 3: POLICY SHOCK – Assuming each parameter constant in Step 2, new distributions of gross unitary benefits are allocated to each household (i.e., higher $, new beneficiaries) and its probability re-estimated Sim – STEP 4: ATTRIBUTION – The difference in probabilities between Step 3 and Step 2 is the attributed effect on the policy simulation
A few considerations Technical Issues ENDOGENEITY: Expected at aggregated level but not at hh level POLICY INSTRUMENT AS CIRCUMSTANCE Level of public spending exogenous to individual Interpretation WHAT DOES THIS SIMULATION TELL US? From a hh level, the extent to which personal circumstances act as obstacles of an opportunity From a fiscal point of view, how spending (S) may counteract personal circumstances (D)
4 simulations SIMULATION 1: MORE RESOURCES FOR ALL An additional 70% of public spending on education–thought perhaps as increasing teachers’ salaries- is transferred as it is right now SIMULATION 2: NO FEES PAID BY HOUSEHOLDS An additional subsidy to households that report paying fees SIMULATION 3: NO NON-FEE COSTS An additional subsidy to households that report paying for non-fee items SIMULATION 4: REDISTRIBUTION public spending on primary education from children in household in urban areas is redistributed among all children in household in rural areas
Simulation Results
SIMULATION 2: NO FEES PAID BY HOUSEHOLDS An additional subsidy to households that report paying fees Smallest win: Urban female child with educated head Largest win: Rural male child with educated head and Rural male child with non educated head
SIMULATION 3: NO NON-FEE COSTS An additional subsidy to households that report paying for non-fee items Smallest win: Urban male child with educated head Largest win: Rural male child with non educated head
Wrapping up 1.Opp BIA provides a supplemental darker picture than the traditional BIA 2.Circumstances act as serious obstacles to improve opportunities 3.“Policy matters”: simply increasing resources for all or taking away from some to give to others may not do the job 4.How policy is done will have determine the pattern of winners and losers, although simulations show that effects are not large.
Thank you!
HOI for Education Opportunities CWIQ 2007 HOI significantly different from coverage for 3 of 4 opportunities Much better picture for enrollment than starting or finishing primary on time late entry into schools
HH wealth, location and HH head characteristics (education. Gender) or child characteristics (age, gender) are most important for access to education opportunities In 2010, preliminary results suggest ethnicity and religion seem to matter as well
Comparison with a selection of African countries -I (late 2000’s) Liberia Liberia lags behind most countries, but exhibits common patterns Attendance improves with age late entry into school in all countries Significant gap between HOI and coverage, esp. for lower age group
Comparison with a selection of African countries -II (late 2000’s) Liberia HOI and coverage low for all countries, and Liberia does better on both than some of the other countries Evidence of late entry into school for all countries