Grants and dispersion of local education spending in Brazil

ABSTRACT

Grants and dispersion of local education spending in Brazil. Regional Studies. Since several federations rely on a decentralized system of public education, understanding local responses to intergovernmental grants is a relevant issue in regional economics. This paper investigates whether different types of grants (block, categorical and matching) increase the dispersion of education spending and impact spending inequalities across Brazilian municipalities from 2004 to 2010. The results suggest that in a decentralized fiscal system public funding based on block grants, unlike funding from categorical grants, favours differences in tastes and leads to a higher dispersion of public provision across jurisdictions. Matching grants do not produce a consistent effect on dispersion.

摘要

巴西地方教育支出的补助与散布。Regional Studies. 由于若干联邦政府依赖去中心化的公共教育系统，理解地方对于政府间的补助之回应，便是区域经济中的相关议题。本文探讨不同的补助类别（固定补助，分类补助与配对补助）是否增加巴西各个市政府自2004年至2010年的教育支出散布，并影响支出的不均等。研究结果显示，在去中心化的财政系统中，与分类补助资金不同的是，根据固定补助的公共资助偏好风格的差异，并导致各管辖区的公共供给散布更广。配对补助对于散布并未产生一致的效应。

RÉSUMÉ

Subventions et dispersion des dépenses locales pour l’éducation au Brésil. Regional Studies. Étant donné que plusieurs fédérations sont tributaires d’un système décentralisé d’éducation publique, comprendre les réactions locales à des bourses intergouvernementales est une question pertinente dans l’économie régionale. La présente communication tente d’établir si différents types de subventions (globales, par catégories ou de contrepartie) augmentent la dispersion des dépenses pour l’éducation et influent sur les inégalités des dépenses entre différentes municipalités du Brésil, de 2004 à 2010. Les résultats semblent indiquer que, contrairement aux subventions par catégorie, dans un système fiscal décentralisé, les financements publics basés sur des subventions globales favorisent les différences dans les goûts, et portent à une dispersion supérieure des systèmes publics entre les juridictions. Les subventions de contrepartie ne produisent pas un effet homogène sur la dispersion.

ZUSAMMENFASSUNG

Staatliche Transfers und Streuung der lokalen Bildungsausgaben in Brasilien. Regional Studies. Da sich mehrere Bundesstaaten auf ein dezentralisiertes System der öffentlichen Bildung verlassen, ist ein Verständnis der lokalen Antworten auf staatliche Transfers ein relevantes Thema für die regionale Ökonomie. In diesem Beitrag wird untersucht, ob verschiedene Arten von Transfers (Block, kategorisch und angepasst) die Streuung der Bildungsausgaben verstärken und sich auf die Ungleichheit bei den Ausgaben verschiedener brasilianischer Gemeinden im Zeitraum von 2004 bis 2010 auswirken. Aus den Ergebnissen geht hervor, dass eine dezentralisierte staatliche Finanzierung über das Fiskalsystem auf der Grundlage von Blocktransfers im Gegensatz zu einer Finanzierung durch kategorische Transfers Unterschiede beim Geschmack fördert und zu einer stärkeren Streuung von öffentlichen Zuwendungen in den verschiedenen Hoheitsgebieten führt. Angepasste Transfers erzeugen keine gleichmäßige Auswirkung auf die Streuung.

RESUMEN

Subvenciones y dispersión del gasto educativo local en Brasil. Regional Studies. Dado que varias federaciones confían en un sistema descentralizado de la educación pública, entender las respuestas de ámbito local para las subvenciones intergubernamentales es una cuestión relevante en la economía regional. En este artículo analizamos si los diferentes tipos de subvenciones (bloque, categóricas y compartidas) hacen aumentar la dispersión del gasto en educación y repercuten en la desigualdad de los gastos en los municipios de Brasil en el periodo entre 2004 y 2010. Los resultados indican que en un sistema fiscal descentralizado, los fondos públicos basados en subvenciones en bloque, a diferencia de los fondos de subvenciones categóricas, favorecen las diferencias en las preferencias y conducen a una mayor dispersión de la prestación pública en las jurisdicciones. Las subvenciones compartidas no producen un efecto coherente en la dispersión.

KEYWORDS:

• dispersion on education spending
• local public goods
• grants
• fiscal decentralization

关键词:

• 教育支出的散布
• 地方公共财
• 补助
• 财政去中心化

MOTS-CLÉS:

• dispersion des dépenses pour l’éducation
• Produits publics locaux
• subventions
• fiscal décentralisation fiscale

SCHLÜSSELWÖRTER:

• Streuung der Bildungsausgaben
• lokale öffentliche Güter
• Finanzausgleich
• fiskale Dezentralisierung

PALABRAS CLAVES:

• dispersión en el gasto educativo
• bienes públicos locales
• subvenciones
• descentralización fiscal

JEL:

• H52
• H77
• R10
• R50

ACKNOWLEDGEMENTS

Two referees and an associate editor provided excellent comments.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

SUPPLEMENTAL DATA

Supplemental data for this article can be accessed at http://10.1080/00343404.2017.1287891

ORCiD

Enlinson Mattos http://orcid.org/0000-0001-7351-1280

Notes

1. See Albouy (2012) for a comprehensive discussion of grant efficiency.

2. Fisher and Papke (2000) analyse the effect of different types of educational grants on local government spending in the United States. In a study more similar to this paper, Ahlin and Mork (2008) analyse the impact of local preferences and grant types on school resources. Unlike this paper, they do not consider dispersion of spending; and Cascio et al. (2013) focus on the impact of Title I reform on the access of minority groups to education opportunities. Overall, their results coincide to show that following school financing reforms, equalization in spending increases and that disadvantaged groups obtain additional resources.

3. States provide the majority of FUNDEB resources. Although the central government regulates the FUNDEB constitution, each state forms its own fund. Therefore, there are 27 funds (26 state funds and the capital district fund).

4. In 2010, the minimum annual spending per pupil was fixed at R$1414.85, or approximately US$790. In the sample, FUNDEB resources represented a population-weighted average of approximately 46% of total municipality educational expenses; additionally, in 2010, just 4% of municipalities spent less on education than the amount received from FUNDEB. See Table 1 for more details.

5. As MPF (block) grants are a piecewise function of population, many authors have explored the exogenous nature of the MPF criteria using IV approaches (Brollo & Nannicini, 2012; Litschig & Morrison, 2013). Arvate et al. (2015) use the same IV strategy to approach the flypaper effect in Brazil. For the population cut-offs and their respective factors, see also Table A4 and Figure A1 in Appendix A in the supplemental data online.

6. The standard deviation (SD_t) and the average (X_t) refer to the entire sample for each year. The idea is to capture essential variation across localities (between), which is widely used to analyse expenditure volatility. The i-municipality is not excluded from the SD_t and X_t measures because the sample is composed of quite a large number of observations (around 4200 municipalities in each year), so it should not affect the sample average.

7. This approach assumes zero correlation across groups (i and j do not have correlated errors), but it allows for within-group correlation (within i). See Nichols and Schaffer (2007) for a discussion of clustered errors and fixed effects.

8. The former is available from the Secretaria do Tesouro Nacional (Brazilian Secretary of Treasury Service), the later from the Instituto Nacional de Estudos e Pesquisas Educacionais.

9. Panel data for the share of the poor population and personal income are not available from the IBGE for municipalities in Brazil during the period under analysis.

10. See http://www.tse.jus.br/eleicoes/estatisticas/estatisticas-eleitorais-2014/.

11. Table A1 in Appendix A in the supplemental data online presents some indicators (mean and SD) of Brazilian municipalities split into three different groups according to population size.

12. Results are similar if one excludes only the 27 state capitals subject to different criteria for unconditional transfers.

13. Most of the inequality analysis in this section follows Murray et al. (1998).

14. The main assumption behind the regression discontinuity method in the current context is that in the case of a lack of grants, municipalities close to the cut-off point should exhibit similar results in the dependent variable (education spending). For details, see Van der Klaauw (2002).

15. A robust IV approach considers that the instruments present a high correlation with the instrumented variable and that instruments are not correlated with the error term [corr(..

16. In this last case, the estimation of H follows:(6.1) where:(6.2) (6.3)

17. For results with grants in level specification, see Tables A5 and A6 in Appendix A in the supplemental data online. Results are qualitatively similar to logarithm specification but instruments are weaker.

18. Some first-stage results are presented in Table A3 in Appendix A in the supplemental data online. Results of the J-Sargan test also reinforce the idea that the model does not suffer from endogeneity problems.

19. Figure A2 in Appendix A in the supplemental data online reports the variation of education spending and grants by year for both the top (25% of the localities with the highest average spending per pupil) and bottom quartiles. Interestingly, the variation in growth on average spending per pupil is more associated with block than categorical grants in both quartiles.