
7 minute read
Notes
municipalities that are doing slightly better in education outcomes may be less likely to engage in the technical assistance program. • A crucial element of Ceará’s RBF mechanism is the use of general transfers as an incentive to improve education. Ceará’s incentive mechanism links general transfers to education results, allowing mayors to use the transfers in any sector, even one not directly related to education. Such freedom provides incentives for all of municipal government, including the mayors and secretaries of finance. • Two necessary preconditions for establishing a results-based mechanism in education are a decentralized school system and a robust monitoring and evaluation system. A system of incentives requires that subnational governments have autonomy over how to manage their schools. The Ceará case shows that a highly decentralized system with well-designed incentives and support to the municipalities can be very successful in improving student learning. In addition, the state government of Ceará has a monitoring and evaluation system, which is critical for establishing incentives based on education results and identifying municipalities that need more support.
NOTES
1. This chapter benefited from comments and suggestions from Samer Al-Samarrai, Blane
Lewis, Lars Sondergaard, Halsey Rogers, Pedro Cerdán-Infantes, Emanuela di Gropello,
Pablo Acosta, Kjetil Hansen, and Fabiano Colbano. 2. The exceptions include federal schools that provide primary and upper secondary education. many states and a few municipal governments also have their own universities and vocational training institutes. 3. The Brazilian National Congress has recently approved a reform of FUNDEB under which the federal contribution will gradually increase to 23 percent by 2026, starting with 12 percent in 2021. 4. The IDEB is calculated at the school, municipal, state, and national levels and is based on two components: student performance on the nationwide Basic Education Evaluation
System (SAEB) and student passing rates (IDEB = N ∙ P, where N = normalized student performance at the end of each school cycle, 0 ≤ N ≤ 10 and P = harmonic mean of student progression of all grades in each school cycle, 0 ≤ P ≤ 1). The index is calculated every two years and is coupled with targets that make it possible to assess whether schools, municipalities, states, and the country are making progress toward improving education quality. 5. For more details on the Ceará education model, see Loureiro et al. (2020). 6. The establishment of the Fund for the Development of Primary and Lower Secondary
Education (FUNDEF) and, later, FUNDEB has minimized this issue. Since the establishment of these funds, there has been an ongoing process of devolving lower secondary education to municipalities. However, there is wide variation among states. According to the 2018 Education Census, the share of municipal enrollment compared to total public enrollment in lower secondary education ranges from as low as 2 percent in Paraná to 28.5 percent in São Paulo, 44 percent in mato Grosso do Sul, and 49 percent in Amazonas, and as high as 74 percent in Rio de Janeiro and 94 percent in Ceará. 7. Public universities have enrolled 24.3 percent of the country’s 8 million tertiary students.
The federal government is the main public provider of higher education, accounting for 62 percent of the enrollment in public institutions, followed by the states with 32 percent, and the municipalities with 6 percent. 8. For a broader discussion of the general transfers in Brazil, see World Bank (2020). 9. The amount transferred by federal government to the states is defined each year using the following algorithm: (1) the states are ranked by value per student (amount of FUNDEB resources in the state divided by total enrollment) considering both the state network and the municipalities in each state; (2) the federal government calculates the amount of resources that would need to be transferred to the state with the lowest value per student to reach the amount per student in the state with the second-lowest value per student;
(3) the equalization process is repeated comparing the value per student in each state with the value per student in the next-highest state; (4) the process stops when the top-up funds allocated by the federal government would exceed the total allocated amount for the federal contribution to FUNDEB. 10. Because FUNDEB is procyclical, that is, the total amount of funds available each year is a fraction of government revenues, the minimum spending per student is defined endogenously by a top-down approach, rather than by a bottom-up approach, under which the amount to be allocated for education is defined on the basis of an “ideal cost.” This choice is made relevant by Brazil’s strong fiscal constraints. See World Bank (2020). 11. According to education census data, there were 46.6 million students enrolled in public basic education in 2007 under FUNDEB, while 28.4 million were in primary and lower secondary education under FUNDEF. Therefore, FUNDEB covered 64 percent more students than those covered by FUNDEF. 12. Added were the Causa mortis and Donations Tax (ITCmD), the motor vehicles Property
Tax (IPvA) and a 50 percent quota of the Rural Lands Tax (IITR). 13. The amendment also increased the minimum percentage of spending on teacher salaries with FUNDEB funds from 60 percent to 70 percent. 14. As of 2019, states such as Pernambuco, Alagoas, Espírito Santo, and Amapá had changed their ICmS redistribution criteria, following Ceará’s experience. 15. TPAIC was organized under three main components: literacy support, pedagogical use of student assessment, and strengthened governance. Literacy support comprises a series of activities, such as designing and delivering textbooks, strengthening teacher training with a focus on classroom practice, fostering the reading culture, and supporting the expansion of ECE. The pedagogical use of student assessment involved financial and technical support for the implementation of local learning assessments and training municipal and school professionals to make a systematic use of assessment results. Strengthened governance included a cascade model to support and train municipal teams and incentives for exchanging best practices among schools. See Loureiro et al. (2020). 16. The devolution of primary and lower secondary education to municipalities is one of the main results of FUNDEB policy, but this process evolved differently in different Brazilian states. Ceará was one of the few states that already had high levels of decentralization in 2007. Since then, other states have devolved the responsibility for managing primary and lower secondary education to municipal governments, but none of them has established a framework of collaboration between the state and municipal governments with concrete technical support as Ceará did. Having seen Ceará’s outstanding education results, some other states have recently started to design similar frameworks. 17. For more details on the education reforms in Ceará, see Loureiro et al. (2020). 18. For a discussion of how information systems and accountability can reduce political bargaining, see Toral (2019). 19. Data envelopment analysis showed that if all municipal and state networks used best practices in resource allocation and management, the average IDEB score of primary and lower secondary students in Brazil would rise from 4.5 to 6.4. When disaggregating the results for lower (1st to 5th) and upper grades (6th to 9th), there is evidence that IDEB scores could increase from 5 to 7 and from 4 to 6, respectively. For details, see World Bank (2017). 20. Carneiro and Irffi (2017) use four different control groups and apply a difference-in-differences approach. The first control consisted of municipal schools from all Brazilian states with an ICmS rule different from Ceará’s. The second was restricted to states in the Northeast, which have regional similarities. The third removed from the previous groups states that had external learning evaluations on the assumption that this policy might have affected the schools’ scores. The fourth compared Ceará’s municipal schools with schools run by the state. In the first and the second control groups, the authors considered two scenarios, one with and another without propensity score matching. 21. Brandão (2014) used a difference-in-differences model and two control groups picked from Ceará’s neighboring states: Piaui, Pernambuco, Paraiba, and Rio Grande do Norte.
The first control consisted of municipalities right at the border that belong to the same microregion, according to the Brazilian Institute of Geography and Statistics (IBGE) classification. The second was formed through propensity score matching. As the latter was more balanced with treated municipalities, the author considered the control group formed through propensity score matching to conduct the winner-loser and richer-poorer subgroup analysis. The education variable consisted of SAEB scores in Portuguese and math.