Education finance equalization, spending, teacher quality and student outcomes: The case of Brazil ’ s FUNDEF Nora GordonEmiliana Vegas UC San Diego The World Bank January 14, 2005
Structure of presentation Motivation Background on Brazil Key features of FUNDEF Related literature & this paper Data and descriptive statistics Findings Conclusions & policy implications
Motivation FUNDEF: an education finance reform implemented in 1998 Can provide useful evidence on the impact of education finance equalization strategies on access, quality, and equity of education
Background on Brazil mid-1990s: Brazil was characterized by enormous inequality across and within states in terms of education finance, access, and quality highly decentralized structure, with state and municipal education systems (26 states + DF, about 5,000 municipalities) basic education (Ensino Fundamental) is comprised of 2 levels: EF1 = grades 1-4 EF2= grades 5-8
Education finance in Brazil before 1998 By law, 25 percent of all state- and municipal-level taxes & transfers were mandated to be spent on education States & municipalities were (anecdotally) quite creative in their definition of education spending This led to enormous inequity in resources available for education within and across states (Soares 1998)
Regional disparities in education finance and access in the 1990s RegionMean PPE 1996 (R$) GER 1994 NER 1994 North Northeast South1, Southeast1, Center- West Sources: INEP and STN
FUNDEF: Key features Main feature is creation of a state fund to which state & municipal governments contribute 15 percent of specific taxes & transfers These contributions are then redistributed to the state & municipal governments on the basis of enrollment at least 60 % of FUNDEF revenues must be allocated to teacher salaries The federal government supplements the per student allocation in states where FUNDEF revenues per student are below a yearly established spending floor The law requires state & municipal governments to allocate 10% of FUNDEF-tapped and 25% of non- FUNDEF taxes & transfers to education
Previous research on FUNDEF Found that the reform: led to substantial increases in enrollment in municipal basic education systems, especially in the poorest regions (World Bank 2002) associated with positive effects on repetition, dropout and age-by-grade distortion (World Bank 2002, Abrahão de Castro 1998)
Previous research on education finance equalization reforms In the U.S., found: mixed evidence about the merits on reducing inequality in student achievement (Card & Payne 2002, Clark 2003) important to assess the extent to which previously allocated revenues for education are redirected to other areas (Hoxby 2001, Gordon 2004)
This paper 1.explores further how FUNDEF affected education expenditures by municipal & state governments, including the extent of crowd-out 2.examines the effect of the reform on state-level enrollment 3.analyzes how state & municipal governments allocated additional resources on inputs - teacher credentials and class size - and how these translate into student outcomes 4.evaluates the extent to which the reduction in spending inequality among states led to a decrease in inequality in student achievement
Data Education indicators from INEP ’ s annual school census for : student enrollment, number of teachers, teachers ’ educational attainment, age-by-grade distortion Annual financial data from STN (Treasury) for : State & municipal taxes & transfers, used to calculate FUNDEF (after 1998) and non-FUNDEF resources for education Expenditure data, used to calculate education expenditures Student achievement data (SAEB): Math and language standardized tests administered to 4 th graders in 2 years prior and 2 years post FUNDEF: 1995, 1997, 1999, and 2001 stratified sample of students, representative at the state level for state, municipal & private schools
Evolution of Enrollment in EF1, by region,
Evolution of Enrollment in EF2, by region,
Enrollment changes in % change EF1EF2 North Northeast Southeast South Central-West Total
Primary Net Enrollment Rates by Region,
Mean pupil:teacher ratios EF1EF (%) (%) N NE S SE CW Total
Share of teachers with more than primary education EF (%) N NE S SE CW Total
Mean age-by-grade distortion EF1EF (%) (%) N NE S SE CW Total
Regional averages of state per pupil spending change (%) N NE S9781,2261, SE8901,3541, CW8441,0051, Total7511,0101, (constant R$)
Regional averages of annual state FUNDEF per pupil allocations N NE S SE CW Legal minimum 315 EF1:333 EF2: EF1:363 EF2: EF1:418 EF2:438.9 (current R$)
Means and standard deviations of SAEB 4 th grade math scores MeanS.D.MeanS.D. N NE S SE CW Total
A short note on methodology We use regular OLS, instrumental variable and reduced-form approaches to identify the effect of FUNDEF on the variables of interest Our instrumental variable is the amount of education spending mandated by the reform, calculated using FUNDEF ’ s formula
Findings 1.To what extent did FUNDEF translate into increased education expenditures by municipal & state governments, including the extent of crowd-out? 2.To what extent did FUNDEF lead to increases in state-level enrollment? 3.How did state & municipal governments allocate additional FUNDEF resources on inputs - teacher credentials and class size - and how did these translate into student outcomes? 4.To what extent did the reduction in spending inequality among states led to a decrease in inequality in student achievement?
1. To what extent did FUNDEF resources translate into education spending? (1) Educ. exp. (2) Educ. Exp. (3) Educ. exp. per pupil (4) Educ. exp. per pupil Mandated spending (0.004)(0.005) Mandated spending PP (0.009) Enrollment (0.272)(0.000) Number of observations 22,879 17,957 R-squared
2. Effect of spending on state- level enrollment EF1 (grades 1-4)EF2 (grades 5- 8) OLSIVOLSIV Education spending (0.001) (0.001) (0.001) (0.001) Education spending*State bound by FUNDEF floor (0.004) (0.004) (0.002) (0.003) State-year observations162 Number of observations27
3. Effect of spending on class size EF1 (grades 1-4)EF2 (grades 5-8) OLSIVReduced form OLSIVReduced form Spending PP (0.004)(0.007)(0.031)(0.044) Mandated exp PP (0.007)(0.047) Obs.23,46017,95418,00216,44512,87312,903 R-squared
3. Effect of spending on teacher qualifications EF1 (grades 1-4)EF2 (grades 5-8) OLSIVReduced form OLSIVReduced form Spending PP (0.0002)(0.0003)(0.0001)(0.0003) Mandated exp PP (0.0003) Observations 23,51617,95418,00216,48212,87312,903 R-squared
3. Effect of spending on age-by- grade distortion EF1 (grades 1-4)EF2 (grades 5-8) OLSIVReduced form OLSIVReduced form Spending PP (0.001) (0.002) Mandated exp PP (0.001)(0.002) Observations 23,46317,95418,00214,40911,35911,388 R-squared
3. Effect of inputs on age-by-grade distortion EF1 (grades 1-4)EF2 (grades 5-8) OLSIVOLSIV Class size (0.001)(0.005)(0.000)(0.013) Share teachers with at least primary education (0.025)(0.031)(0.064)(0.079) Number of observations23,51618,00214,38811,343 R-squared
4. Effect of state-level mean per pupil spending on math achievement q(.25)q(.50)q(.75) Actual mean per pupil spending (0.000) Pseudo R-squared Number of observations78 (quantile regression results)
4. Effect of state-level inequality in per pupil spending on math achievement q(.25)q(.50)q(.75) Actual per pupil expenditure inequality 75/ (0.000) Pseudo R-squared Number of observations78 (quantile regression results)
Conclusions Revenue flows from FUNDEF fully translated into education spending FUNDEF led to increases in enrollment in those states most affected by the reform Additional resources from FUNDEF were used to reduce class size Legislation mandating that teachers have at least secondary education was successful
Conclusions (cont.) Reductions in class size and in the share of untrained teachers are associated with slight decreases in age-by-grade distortion Although changes in mean spending are not associated with higher student achievement, reductions in spending inequality may raise the achievement of students in the lower tail of the distribution