Report No. 10564-BR Secondary Education and Training in Brazil: Adapting to New Economic Realities (In Two Volumes) Volume Il: Annexes June 8, 1992 Human Resources Operations Division Country Department I Latin America and the Caribbean Regional Office FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY ALN1 Currency Unit = New Cruzado (January 1989-March 1990) = Cruzado (from February 1986 to Janurary 1989) = Cruzeiro (prior to February 1986) AVERAGE EXCHANGE RATES NCz$ 1.00 = US$ 0.16390 (November 16, 1989) US$ 1.00 = NCz$ 6.10000 (November 16, 1989) 1988 US$ 1.00 = Cz$ 262.02 1987 US$ 1.00 = Cz$ 39.23 1986 US$ 1.00 = Cz$ 13.66 1985 US$ 1.00 = Cr$ 6.20 FISCAL YEAR January 1 - December 31 FOR OMCLAL USE ONLY ISSUES IN BRAZILIAr SECONDARY EDUCATION VOLUME TWO: ANNEXES Table of Contents ANNEX I: Math and Portuguese Achievement in Secondary Schools in Brazil ANNEX II: The Finance and Costs of Secondary Education in Brazil ANNEX III: Demand For and Supply of Private Secondary Education in Brazil ANNEX IV: Modelo Para Projeglo da Matricula No Segundo Grau - Brasil (Projections Model for Forecasting Secondary School Enrollments in Brazil) This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. ANNEX I Page 1 of 9 MATH AND PORTUGUESE ACHIEVEMENT IN SECONDARY SCHlOOLS IN BRAZIL 1/ Introduction 1. As background to this report, the Ministry of Education and the World Bank commissioned the Carlos Chagas Foundation to design and administer a standardized test of mathematics and Portuguese to a sample of secondary studonts in four cicies (F taleza, Salvador, Sao Paulo and Curitiba). The achievement test wa. designed for students at the end of their third (and final) year of secondary school, and comprised test items that had appeared previously on vestibular examinations for Brazilian universities. The test was accompanied by a background questionnaire for each student and a questionnaire about the school that was completed by the school director. Both tests and questionnaires were administered in November, 1988. The overall purpose of the study was to explore differences in the relative effectiveness of various types of school in enhancing student achievement and in reducing performance differences between students from different social class backgrounds. Sample 2. Fo- each city, a stratified random sample of schools was identified, with replacements; stratification was based on school type (Federal Technical, SENAI schools, teacher training [magisterio] Z_hools, and general secondary schools), time of shift (day or night) and ownership (public or private); it was designed to be representative of the actual distribution of schools and students in each state. Within each school, all students present on the day of test administration were requested to complete the test and accompanying questionnaire, although participation was voluntary. 3. The voluntary nature of participation may have affected the overall estimates of achievement for individuals, schools, and cities. Private school and day shift students were slightly oversampled in Fortaleza, biasing average scores upwar in this city. In Sao Paulo, private schools and day shift students *ere underrepresented, driving average scores downward. In Salvador, public schools were oversampled. Strong collaboration from the State Secondary of Education in Parana 1/ This Annex was written by Marlaine E. Lockheed and Barbara Bruns of The World Bank. Qinghua Zhau provided expert computational assistance. Support for this research was provided by a World Bank research grant, RPO 674-84. For further analysis of the secondary school achievement tests, see "Avilacao do Rendimento Escolar de Alunos da 3o serie do Segundo Grau - Subsidios Para Uma Discussao Preliminar," by Dr. Heraldo Vianna of the Carlos Chagas Foundation (1989). -2- ANNEX I Page 2 of 9 resulted in the most representative sample for this city. Because thesample of schools was fairly small, it was difficult to capture the diversity of private secondary schools; in particulaL, low-tuition schools were not sampled in sufficient quantity to analyze separately. This is unfortunate, because so little is known about the effectiveness of low- tuition private schools and the characteristics of their students. Instruments 4. Fnur instruments were developed: a wathematics test, a Portuguese test, a student background questionnaire, and a school questionnaire. The mathematics and Portuguese tests were designed to measure understanding of the basic secondary school curriculum in these areas, and included one or more item for each area of content. (For example, verb tenses and reading comprehension in Portuguese and linear functions and trigonometry in mathematics.) Items wex. selucted on the basis of "item facility," a statistical measure of item discrimination. Reliabilities fo- the total tests were high, with Cronhach's alpha = .84 for mathematics and .75 for Portuguese. However, reliabilities calculated separately for students in teacher training programs ("magisterio" students) were substantially lower, .30 for mathematics and .58 for Portuguese. Analytic sample 5. Data were obtained from 2648 students and 66 schools; after cleaning the data (range and logic checks) and .naking necessary corrections, usabic data were obtained from 2611 students and 62 schools. Schools 46, 50 and 51 were deleted for insufficient student-level data (9, 3 and 10 cases respectively) and School 39 was deleted for insufficient school-level data (all fields were blank). Differences in achievement by school types 6. On average, students enrolled in different types of school performed differently on the tests; the mean scores of students in all eight types of school are presented in Table 1 (Appendix Table 24 shows results by region). The resulting ranking of school by student performance is: (i) Federal Technical Schools (day shift) (ii) Private General Secondary Schools (day shift and night shift) (iii) Federal Technical Schools (night shift) (iv) Private Teacher Training Schools (day shift) (v) Public General Secondary Schools (day shift) (vi) SENAI Schools (vii) Private Teacher Training Schools (night shift) (viii) Public Teacher Training Schools (ix) Public General Secondary Schools (night shift) -3- ANNEX I Page 3 of 9 7. In each region, there was a reasonably wide range in scoreB from the top schools to the bottom schools, and -irtually the same ranking in each region. The average scores for each type of school across regions were remarkably similar. Table 1s Mean achievement scores for students in different types of secondary schools, Brazil, 1988 Mathematics Portuguese School Type Mean S.D. Mean S.D. Federal technical 22.60 2.18 21.11 0.88 SENAI 12.75 0-53 16.79 0.82 Teacher training* 11.24 1. 36 16.79 2.04 General secondary* 13.70 5.79 17.00 3.28 Note: Maximum score in mathematics is 45; maximum scorp in Portuguese is 35. *Includes public and private schools of this type. 8. On average, students in federal te nnical schools scored significantly higher on both mathematics (about 10 points) and Portuguese (about 5 points) than students in any of the o'her types of schools. Students in teacher training schools scored lowest in mathematics, with students in SENAI schools scoring only about 1.5 points higher than students in teacher :raining and students in general secondary schools about one more point higher. In Portuguese, performance at all types of schools other than the federal technical schools was equivalent. 9. Average differences between schools, however, do not mean that the schools are solely responsible for the differences. Differences in student selection practices can also account for differences in achievement. Since some types of schools, such as the federal technical schools, SENAI schools and the best private schools are selective, it is likely that the average ability of their students will be higher. Ideally, achievement tests should be carried out on the same students at two or more points in their education, for example as entering first year students and just prior to graduation. This "value added" approach allows a more direct assessment of school effectiveness, irresoective of student ability (although non-school factors such as changes in students' motivation or family circumstances will always play some role in determining achievement). To date, such types of longitudinal studies have not been undertaken at the secondary level in Brazil. Such studies would be a valuable next step. -4- ANNEX I Page 4 of 9 Analytic Model and Results 10. The main objective of this analysis was to determine the relative impact of school type and student background on student performance. To achieve this, a multi-level modelling package, called a hierarchical linear model (HLM) 2/ was used. The advantage of the HLM procedure over ordinary least squares (OLS) regression analysis is that dLM can provide more accurate estimates of the importance of school-level factors (i.e., it correctly estimates the standard errors for school-level coeft:cients) A second advantage of the HLM procedure is that it can analyze the factors, what influence differences in student achieve.nent within schools at the same time as it evaluates the factors that influence student achievement across scl'ools. 11. The first result of the analysis was that school differences are more important in explaining differences in student math test scores than in explaining Portuguese test scores. This result is consistont with research from other countries which shows that, as one might expect, the more abstract the subject matter, the more important formal schooling is for developing student achievement. As can be seen in Table 2, almost two- thirds of the variation in student math scores can be attributed to the schools they attend, whereas on the Portuguese test, most of the variation is attributable to individual differences. Table 2: Results of variance component analysis: Brazil secondary school achievement, 1988 (% of Total Variance) Score School Individual Mathematics 62.38 37.62 Portuguese 36.36 63.64 12. The second result was that within any given school, socio- economic differences among students had only very small effects on their test scores. Table 3 shows this through the very low coefficients on "social class differentiation" (which refers to differences within schools), .29 for mathematics and .31 for Port%.-uese. The implication of 2/ A. Bryk, S. Raudenbush, M. Seltzer and R. Congdon, Jr. (1988) An Introduction to HLM: Computer Program and User's Guide, Chicago: University of Chicago. For further information and references, see M.E. Lockheed and Nicholas Longford "A Multi-Level Model of School Effectiveness in a Developing Country." World Bank, WPS 242 (1989). Page 5 of 9 this is that a student from a low income or lcw social class background who attends a good scL.ol will perform close to the average for that school. Table 3: The HLM unconditional model Estimated Effects Coefficients error t-stat p-value School mean Achievement Mean mathematics 14.29 .75 19.04 .000 Mean Portuguese 17.36 .41 42.39 .000 Social Class Differentiation Mean mathematics .29 .15 1.99 .046 Mean Portuguese .31 .12 2.54 .011 13. However, it is also very clear from the data that different types of secondary schools in Brazil generally attract different types of students. Based on information from the student questionnaires which accompanied the achievement tests, all students in the sample were given a socioeconomic status (SES) index score ranging from -1 (low status) to +1 (high status). The SES score was a composite index comprised of father's occupation, father's education, mother's education, and family income. 14. As can be seen from Table 4, students in federal technical schools and SENAI schools are disproportionately male and come from higher social class backgrounds, while students in teacher training schools are disproportionately female and come from lower social class backgrounds. The average socioeconomic level for students in general secondary schools in this sample was in the mid-range, but it should be recalled that this average is composed of private (generally high SES) and public (generally lcder SES) school students. Other differences are that students in general secondary schools spend more hours per week working, and students in federal technical schools spend fewer hours working that any of the others. -6- ANNEX I Page 6 of 9 Table 4: Average characteristics of students attending different types of secondary schools, Brazil 1988 School Ty - Federal Teacher General Student characteristic technical SENAI train,ng secondary Sex (% female) 26.0 6.8 96.1 58.7 Age in years 18.5 19.4 18.9 18.7 SES (factor score) .18 .19 -.29 .02 Family size 4.7 4.2 4.4 4.5 Hours working weekly 5.8 8.6 8.0 14.5 Sample size 192 118 309 1992 15. In order to try to separate all of these various influences on student achievement, the analysis used three different groups of variables. First, to take individual background characteristics into account, four variables were used: sex, age, family size, and the number of hours per week that a student worked. These are presented in the first part of Table 5 as "fixed" variables. 16. Second, to analyze the effects of school type on average student performance, five different variables were used: the average SES of students in the school, and four dummy variables corresponding to the type of school. The first three are federal technical, teacher training, and SENAI schools. General secondary schools were the omitted variable, which means that the results for the other schools are in relation to the average results for general secondary schools. Similarly, because both general secondary schools and teacher training schools are found in both the public and private sector, a fourth dummy variable was used to test for public/private sector effect on achievement. These are presented in the middle section of Table 5 as "Mean Achievement" variables. 17. Finally, the extent to which different types of schools reduced the effect of social class background on achievement was studied. This was done by looking at the variations in student scores within a school in light of the variations in social class background within that school. This is presented in the bottom section of Table 5 as "SES Achievement Gap." -7- ANNEX I Page 7 of 9 Table 5: Effect of school type on achievement and social class differentiation Mathemaatics Portuguese Independent variable Coeff. t-stat Coeff. t-stat Fixed Female -1.48*** -7.06 0.58** 3.02 Age -0.37*** -5.84 -0.48** -8.24 Family size 0.05 0.27 0.05 1.11 Working hours -0.03*** -4.65 -0.02** -3.62 Mean achievement Intercept 13.92*** 22.29 16.81*** 43.71 Average SES 4.71*** 4.74 3.82*** 6.42 F?deral technical 9.28*** 5.20 3.68** 3.44 SENAI -1.52 -0.63 -0.84 -0.58 Teacher training -1.37 -1.06 -0.09 -0.12 Private 3.20* 2.12 -0.20 -0.22 SES achievement gaa Base 0.11 0.54 0.21 1.26 Federal technical 0.68 1.31 -0.25 -0.57 SENAI -0.69 -1.07 -0.37 -0.70 Teacher training -0.35 -0.80 -0.28 -0.74 Private 0.02 0.06 0.22 0.72 * p < .05, ** p < .01, *** p < .001 18. The results are presented in Table 5. Regarding the effect of individual characteristics, the analysis showed that older students performed less well on both the mathematics and Portuguese tests than did younger students, and girls outperformed boys in Portuguese, but boys scored significantly higher than girls on mathematics. Family size was unralated to achievement in either mathematics or Portuguese. The number of working hours had small but highly significant negative effect on students' scores in ,-jth math and Portuguese. 19. With respect to the effects of school type, the results show that, in comparison with students in general secondary schools, students in federal technical schools score sig, 'icantly higher in mathematics (by about 9 points out of a maximum score of 45) and in Portuguese (by about 3 points out of a maximum of 35). For both teacher training and SENAI students, performance differences vis-a-vis students in general secondary schools appear to be mainly due to differences in average student background. For example, although SENAI students' raw achievement scores were slightly higher, once the effect of the higher socioeconomic status of SENAI students is taken into account, there is no significant difference between these students' performance and that of general secondary students. -8- ANNEX I Page 8 of 9 Conversely, students in teacher training schools had low raw scores, but when the fact that these students come from lo :er SES backgrounds is taken into account, the gap virtually disappears. 20. The results also show a "private school effect": c'ntrolling for school type and student background, students in private schools outperformed students in public schools in mathematics (by about 3 points out of 45). There was no significant diff-renc, in Portuguese performance, however. Finally, the results showed tha. garuiess of school type, student test acores are significantly affectt by t-.e average SES level of students the school. If average SES is high, all students (even those which may be individually from a lower class background) do better (by almost 5 points in math and 4 points in Portuguese). 21. The "SES achievement gap" presented in Table 5 measures the effects of social class variations within schools on student achievement. Although as discussed earlier (Table 2), these effects ;-re very small, there was slight evidence that some schools do a better )ob than others in "equalizing" the achievement of students from different backgrounds. This "' 10 minimum salaries I 16.2 23.1 Total 100.0% 100.0 130.0 Source: IBGE, Anuaric Estatistico do Brasil, 1983; PNAD, 1982. F. Conclusions and Policy Issues The above analysis leads to the following conclusions: * Brazil's investment, especially its public investment in secondary education is small relative to other Latin American countries; * secondary education spending has decreased in real terms since 1980 but has recovered somewhat from the low spending levels of 1985; * the private sector plays an especially important role in Brazilian secondary education, both in terms of providing and financing education; * unlike primary education, federal transfers play a relatively unimportant role in financing secondary education provided by other levels of government; * federal expenditures on secondary education heavily emphasize technical/vocational education; - 30 - ANNEX II Page 20 of 31 * technical/vocational education in Brazil is very expensive relative to general secondary education; * earmarking provisions in the new Constitution may lead to increased educational spending in general, including increased spending at the secondary level; * opportunities for secondary education are less equally distributed than those for primary education and are very unequally distributed across income groups within Brazil. These conclusions and the supporting analysis raise several policy questions, most of which require further study prior to offering firm policy recommendations. Among the important policy questions regarding finance of secondary education are the following: 1. Should Brazil increase its investment in secondary education, perhaps by increasing secondary education's share of total public education spending? 2. How should investment in secondary education be increased--through increased outlays on public education or increased subventions to expand the supply of private education; by expanding general secondary or technical/vocational education? To date the federal government has acted to expand technical/ vocational education, 26/ especially in the public sector. The fact that per pupil expenditures in SENAI and the federal technical schools are eight times per pupil expenditures in state secondary schools, which primarily offer general education, reveals the high budgetary cost of expanding technical/vocational education. 3. Should federal transfers be increased in size to stimulate a larger supply of secondary education by the states and private sector; should those transfers, also, be reoriented from technical/vocational to general secondary education? Direct federal expenditures in 1985 were Cz$ 7.5 billion and resulted in the instruction of less than 100,000 students. If transferred to the states, this same expenditure would have educated almost 750,000 students. Increased transfers to both 26/ Recently introduced changes in the federal income tax have eliminated the educational deduction, thereby increasing the after-tax price of private education relative to public education. - 31 - ANNEX II Page 21 of 31 the states and private schools accompanied by incentives (e.g., matching grants) might increase the supply of secondary education still more. 4. Should enforcement of Constitutional earmarking provisions for education be encouraged and strengthened as a means of increasing secondary education investments, or should earmarking and its enforcement be discouraged as poor public finance policy in general? 5. Should equality of opportunity for secondary education be increased across regions and income groups, perhaps by changing funding priorities from expensive technical to less costly general secondary education? Had federal funds been used to fund state-provided secondary education rather than federal technical education, the overall secondary education enrollment rate could have increased from 19.7 percent to 22.8 percent. In other words, the enrollment rate in all regions could have been increased to that (23%) found in the Southeast region. Similarly, disparities in opportunities between income groups could be significantly reduced through reallocation of federal secondary education expenditures. - 32 - ANNEX II Page 22 of 31 Table A Total Post-Primary Education and Training in Brazil, 1983 Annual Hours Student-Program Enrollment Graduates Per Course Hour$ Academic/Technical Federal Agricultural (EATF) 13,554 1,187 16,088,598 Industrial (ETF) 55,313 1,187 65,656,531 State 1,574,752 900 1,417,276,800 Municipal 137,716 900 123,944,400 Private Secondary Schools 1,129,845 900 1,016,860,500 Vestibular CursinhoB n.a. SENAI Habilitacao Program 1,413* SUBTOTAL 2,911,180 2,639,827,000 (46.1% of total) (188.1% of total) Vocational Training !FNAC 946,975 746,787 62 53,272,653 SENAI 496,561 400,825 249 103,698,231 SENAR 152,544 152,544 10 1,432,784 FUNCEP 3,603 3,204 107 328,270 MPAS 301,000 283,006 200 41,255,610 Ministry of 189,174 171,704 95 6,070,734 Agriculture (EMBRATER, CEPLAC) Ministry of 636 590 45 28,833 Industry and Commerce (PLANALSUCAR) - 33 - ANNEX I I Page 23 of 31 Table A (cont'd) Total Post-Primary Education and Training in Brazil, 1983 Annual Hours Student-Program Enrollment Graduates Per Course Hours Academic/Technical Ministry of 9,528 9,514 100 546,424 Communications (TELEBRAS, ECT) Private Schools n.a. n.a. n.a. n.a. Curso Livre Private Enterprises 211,602 202,833 108 21,760,981 Acordos de Insencao Lei 1,193,392 1,193,392 108 128,896,340 Double-Counting 101,111 90,670 546,425 SUBTOTAL 3,403,994 3,073,729 95 356,729,445 TOTAL 6,315,174 2,996,556,000 (100%) (100%) * SENAI/Sao Paulo. Source: Paul Ammann, As Teorias o a Pratica da Formacao Profissional, Ministerio de Trabalho, Brasilia, 1987; MEC/SEEC; SENAI. - 34 - ANNEX II Page 24 of 31 Table B Education Expenditures by Region, 1986 (in Cruzados of 1987) Category of ExDenditure BRAZIL North Northeast Southeast South CenterWest Total State Government 1,234.8 55.4 227.5 643.7 203.6 104.7 Expenditures (billions) State Government Education 244.7 11.8 42.9 130.6 40.9 18.4 Expenditures (billions) State Government Secondary 21.8 1.8 6.6 6.7 4.1 2.6 Education Expenditures (billions) Federal PTA Transfers to 326.0 25.1 158.2 101.1 22.1 19.4 States for Regular Secondary Education (millions) Federal PTA Transfers to 56.9 25.2 18.1 10.1 6.2 7.1 States for Adult Secondary Education (millions) Federal PROTEC Transfers 1,174.8 195.1 255.4 287.5 391.6 45.3 to States for Secondary Education in 1987 (millions) Federal FAE Transfers to 141.2 Individuals (millions) Federal FAE Work 73.7 Scholarships (millions) Federal Agricultural 1,901.3 Schools (millions) Federal Industrial 3,770.0 Schools (millions) Total Secondary School 3,142.6 131.8 745.2 1,552.2 504.3 209.2 Enrollments, 1986 (thousands) Population 15,918.3 908.9 4,640.0 6,748.5 2,508.6 1,112.3 Aged 15-19 _______________________ Source: SEEC/MEC, Balancos Gerais da Uniao; IPEA, Rslatori@ Anual de Acompanhamento--1987. - 35 - ANNEX II Page 25 of 31 Table C Sao Paulo Secondary Education Expenditures, 1986 (millions of Cruzados of 1987) Percent of Total ExDenditures TOTAL EXPENDITURES ON SECONDARY EDUCATION 3,863 Recurrent Expenditures 3,610 93.5% Salaries 3,311 85.7 Teachers 2,733 Administrators 413 Others 166 Books & Supplies .0 Food Program 59 Health Program 3 Building Maintenance 237 Capital Expenditures 253 6.5 Construction 205 Repairs 32 Misc. 16 rcent of ^ tal Revenues TOTAL REVENUES FOR 55,581 ALL EDUCATION* State General Revenues 46,707 84.0% Education Salary Tax 7,492 13.5 Own-Source Revenues 1,095 2.0 Federal Transfers 287 0.5 * excluding borrowing Note: total expenditures on primary education in 1986 was Cz$ 49,644 million. Sourcet Secretaria da Educacao, Sao Paulo. - 36 - ANNEX II Page 26 of 31 Table D Sao Paulo Secondary Education Expenditures and Enrollments (state schools only; in Cruzados of 1987) Year Expenditures Enrollments Expenditures (billions) (millions) Per Pupil 1980 2.66 416,216 6391 1981 2.62 436,920 5997 1982 2.53 462,916 5465 1983 3.15 486,489 6475 1984 1.94 515,239 3766 1985 2.38 545,372 4364 1986 3.68 552,277 6664 1987 3.48 539,397 6452 1988 539,809* Source: Secretaria da Educacao, Sao Paulo. - 37 - Page 27 of 31 Table B Unit Costs in Secondary Education in Sao Paulo (in Cruzados of 1987) PulcPiae SiES^;* Direct Recurrent Costs SEE 1 & 2 5,030 7,193 Paro 1 & 2 9,443 2 only 13,551 Braga 1 & 2 16,740 2 only 14,353 SENAI total 61,737 textiles 52,728 plastics 145,226 graphic arts 105,617 ceramics 96,960 mechanics 38,719 precision mechanics 50,284 metallurgy 39,847 ital Costs Paro 1 & 2 1,729 2 only 6,191 Braga 1 & 2 10,703 2 only 12,070 sonnel Costs 85.7% % of Total urrent Costs SEE 89.9% Paro 96.S5 *Annual cost per student enrolled in Habilitacao Profissional in 1985. Sources: Paro (1982); Secretaria de Estado da Educacao (Sao Paulo); Braga and Cyrillo (1988); SENAI/Sao Paulo, 1987. - 38 - ANNEX II Page 28 of 31 Notes to Table E There have been several studies of the costs of secondary education in Sao Paulo state; unfortunately, they vary in terms of methodology, sample, and time of data collection, thereby making it difficult to compare the unit costs of different types of institutions. The Secretaria de Estado da Educacao (SEE) carried out a study of the direct and indirect costs of education as its contribution to a national study of unit costs conducted under the auspices of a World Bank loan to the Ministry of Education. While the national study focused on primary schools, the Sao Paulo also included secondary schools in its sample. Data were collected at three different dates: April 19e5, September 1986, and July 1987; high inflation combined with sharp changes in real teacher salaries result in very different nominal and real unit costs over time. The data reported in Annex 7 come from the private and public (i.e., state) secondary schools included in the SEE sample; these schools may not be representative of Sao Paulo. The only cost data included are personnel and school materials; indirect administrative costs (of SEE) and capital costs are ignored. While these data are, thus, incomplete and not necessarily representative, they do have one advantage over the other studies reported here: they include both public and private secondary nchools in the same sample thereby permitting a comparison. The study by Paro (1982), also, included a sample of primary and secondary schools but of the public sector only. The sample included twenty-eight schools offering secondary education; data on both recurrent and capital costs were collected in December 1980 or January 1981. Only equipment and building values were included in capital costs due to the lack of information on actual land values. Building values were set at the cost of new construction, and the annual rent was imputed as 1/35th of the building value. Equipment values were not directly measured but, rather, were estimated based on educational standards for primary schools; equipment was assumed to have a useful life of 6.6 years. The strong assumptions employed in calculating capital costs in this study and the unknown biases in the resulting numbers suggest caution in interpretation. The study by Braga and Cyrillo (1988) surveyed a sample of private primary and secondary schools during the 1985 school year. The survey relied on the proprietors of the private schools to report the value of land, buildings, and equipment. These capital costs were found to be high, primarily due to high land values for schools located in urban areas. The Braga and Cyrillo study is the only one of those reported here to report relatively complete and accurate estimates of capital costs. Finally, the Sao Paulo SENAI district has calculated the recurrent costs of secondary school instruction for different technical fields for all schools in the SENAI system. Since SENAI personnel as well as facilities are frequently used for a variety of training and instructional - 39 - ANNEX II Page 29 of 31 activities, costs were allocated according to the proportion of faculty time spent on secondary education; the determination of how faculty spend their time is carefully done, and the estimates of recurrent unit costs can be regarded as quite accurate. The omission of capital costs from these calculations, however, is serious. The high quality of SENAI facilities and the high capital costs usually easociated with technical education suggest unit capital costs are likely to be much higher than those found for general secondary education in the Braga and Cyrillo study. Each of the studies reviewed provides useful information, while no one study answers all relevant questions. The SEE study does not yield representative results but is the only study to allow a comparison of recurrent costs in public and private schools; unit costs in private schools are found to be about forty percent higher than in public schools. The Paro study provides a more representative estimate of unit costs in public secondary schoo' , and they are considerably higher than that found in the SEE study. The Braga and Cyrillo study provides a representative estimate of unit costs in private secondary schools; they are higher than those Paro found for public secondary schools. The Braga and Cyrillo study is the only one to provide a believable estimate of capital costs, and these are found to be as much as forty-six percent of total unit costs in secondary education 27/ and considerably higher than the capital costs estimated by Paro (which excluded land values). Finally, the SENAI results demonstrate the recurrent unit costs of technical education are between three and ten times those of general education. 27/ The respondents to the survey, however, had an incentive to overstate capital costs if they thought the results might be used in influencing public policy regarding private school tuition. Maximum tuition charges are set by allowing a "fair" return on capital investment. - 40 - ANNEX II Page 30 of 31 Table F Secondary Education Expenditures (Current + Capital) Per Pupil, by Providor and Program, 1985 (in constant Cruzados of 1987) Government Total Secondary School Total Secondary School Cost/ and Program Expenditures Enrollments Student (billions) Federal, Total 7.5 99,422 75,436 * Agriculture Schools (1.5) 13,664 109,778 * Industrial Schools (3.2) 54,089 59,162 State, Total 18.0 1,780,155 10,111 Municipal, Total 0.7 132,333 5,290 SENAI 0.7 7,543 86,823 Private Schools* 14.2 1,004,228 14,155 TOTAL 45.8 3,023,681 * Private school expenditures per student are calculated as 1.4 times the expenditure per pupil in state secondary schools, based on cost analyses for the state of Sao Paulo; see Table E. - 41 - ANNEX II Page 31 of 31 Table G Secondary Education Expenditures by Level of Government, 1975 - 1985 (billions of 1987 Cruzados) 1975 1980 1985 Federal 1/ 5.3 8.7 7.5 State 20.2 21.0 18.0 Municipal 0.2 0.8 0.7 2/ Total 25.8 30.5 6.2 Total Number of Secondary School 1.9 2.8 3.0 Students (millions) Number of Secondary n.a. 1.5 2.0 School Students in Public Schools (millions) Expenditure Per n.a. 20,222 13,022 Public Secondary School Student (Cruzados) 1/ excluding transfers 2/ estimated - 43 - ANNEX III Page 1 of 19 DEMIAND FOR AND SUPPLY OF PRIVATE SECONDARY EDUCATION IN BRAZIL 1/ A. Role of the Private Sector Private ee'cation plays an especially important role in Brazil. Twelve percent of primary school pupils, 34% of secondary school pupilo, and 59% of university pupils are enrolled in private institutions. In addition, large numbers of students are enrolled in private pre-schools, private "cursos livres" offering instruction ranging from hair cutting to computer programming, and "cursinhos" which prepare secondary school graduates to take university entrance ("vestibular") examinations. 2.1 overall, the private education sector can be accurately described as dynamic, innovative, and heterogenous in terms of the quality of instructional services. Private education can be either non-profit or for- profit, and entrepreneurs have in the past found the returns sufficiently high to undertake sizeable investments in the sector. Private Market Share The percentage of students enrolled in private secondary schools has declined considerably during the 1980s, from 46.5% in 1980 to 33.3% in 1985. This decline is attributable to two phenomena: (i) the economic crisis that reduced personal incomes and, thus, decreased demand for private secondary education and (ii) an increase in the supply of public secondary education. Table 1 shows that after enjoying a relatively stable or growing share of secondary school enrollments in the 1970s, the private sector has suffered a serious decline in every region of the country since 1980. 1/ This Annex was prepared by Donald Winkler. 2v In Sao Paulo alone, about 8900 private educational establishments are registered with the Private School Syndicate. Approximately half offer "cursos livres," one-quarter are pre-schools, and the remaining one-quarter offer either primary or secondary education. - 44 - ANNEX III Page 2 of 19 Table 1 Private Share of Enrollments in Secondary Education by Region, 1970 - 1985* 1270 1975 1980 1985 BRAZIL 45.1% 45.3% 46.5% 33.3% North 38.3 31.7 31.8 17.5 Northeast 43.2 42.0 45.3 34.1 Southeast 49.1 50.0 53.5 36.7 South 38.5 40.4 33.2 30.3 Center-West 30.6 1.2 36.8 23.1 * Enrollments at the beginning of the school year; the private share of enrollments computed on the basis of year-end enrollments shows a similar trend over time but the reduction in private share is not as large, decreasing nationally from 45.4% in 1970 to 37.7% in 1985. Source: MEC/SEEC data. The private sector shAre is, however, still very important. At current unit costs (expressed in 1987 purchasing power), the public sector would have to increase its spending by US$ 275 million were it to decide to completely finance all private secondary school instruction. 3/ Another way of looking at the importance of private secondary education is given in Table 2. The secondary education gross enrollment rate was 21% of the 15-19 age group in 1985. Were the supply of private secondary education to abruptly disappear, public secondary schools could only educate 14.1% of the age group, approximately equal to the level of educational opportunity available in 1975. 3/ Private secondary schools enrolled almost 1.1 million students in 1986, and the per pupil expenditure in state-operated schools was US$ 257.4 (in 1987 prices). - 45 - ANNEX III Page 3 of 19 Table 2 Gross Enrollment Rates Including and Excluding Private Secondary Education Enrollments by Region, 1970 - 1985 1970 1975 1980 1985 BRAZIL Public and Private 9.8% 16.6 20.8 21.0 Public Only 5.4 9.1 11.1 14.1 North Public and Private 6.7 12.3 16.1 15.1 Public Only 4.1 8.4 11.0 12.5 Northeast Public and Private 6.0 10.2 13.9 15.8 Public Only 3.4 5.9 7.6 10.4 Southeast Public and Private 13.3 21.3 26.0 25.4 Public Only 6.8 10.7 12.1 16.1 South Public and Private 9.4 16.1 21.9 22.4 Public Only 5.8 9.6 14.6 15.6 Center-West Public and Private 8.1 14.2 18.6 20.2 Public Only 5.6 9.8 11.8 15.5 Source: MEC/SEEC data. As shown in Table 3, the private sector share does increase with family income level, but a high proportion of students from even the lowest income quartile attend private schools. The heterogeneous quality of secondary education results in the provision of both low-cost instruction to low income families and high-cost instruction to high income families. As a ratio of per capita family income, the cost of private education does not vary greatly by income quartile. - 46 - ANNEX III Page 4 of 19 Table 3 Private School Shares, Tuition Charges, and Family Income by Income Quartile, 1982 Income Quartile First Second Third Fourth Total Percent of Students Enrolled 28.8 30.7 34.4 51.0 43.0 in Private Secondary Schools Average Monthly Private 7.36 12.13 19.07 41.96 33.58 Secondary School Tuition Paid in 1982 (US$) Average Per Capita Family 30.50 46.94 83.52 255.67 193.79 Income of Children in Private Secondary Schools in 1982 (in US$) Ratio of Tuition to Per 0.241 0.258 0.228 0.164 0.173 Capita Family Income Source: Annex A. B. Demand for Private Education Private education is typically supplied in one of two situations: (i) excess demand for education exists given the public supply or (ii) public supply fails to satisfy the demand for heterogenous quality levels. 4/ Private secondary education in Brazil thrived initially due to excess demand; the public sector failed to provide sufficient student- places. Private secondary education still exists in part because the public sector fails to provide secondary schools and teache-s; this is especially true in the rapidly-growing periphery of urban areas. Increasingly, however, private education exists because large numbers of families are dissatisfied with the perceived quality of public education. Several factors, of course, simultaneously influence the demand for private education: the availability of free public education, the relative quality of private education, family income, and private school tuition levels. Data do not exist to estimate a formal model of private scbool demand and supply, but the basic model can be used to verbally analyze enrollmen' Lrends in recent years. 4/ See James (1987). - 47 - ANNEX III Page 5 of 19 Availability of Free Public Education The supply of public secondary education has increased considerably in recent years. As long as demand continued to increase apace this presented little threat to private school enrollments. But as overall demand stagnated in the early 1980's, increased public supply was reflected in decreased demand for private secondary education. As shown in Table 2, the overall secondary education participation rate barely changed between 1980 and 1985, yet the private sector share diminished significantly. Table 4 demonstrates this for each -.egion in Brazil. Overall, public secondary education enrollments increased by 420,000 between 1980 and 1985, while private school enrollments decreased by 306,700 in the same time period. Over the 1980-84 time period, the annual enrollment growth rate was -6.1% for private schools and +6.2% for public schools. Table 4 Private Market Share in Secondary Education, 1985 Change in Change in Private Public Percent of Percent of Enrollments Enrollments Region Establishments Enrollments 1980-1985 1980-1985 North 27.1 17.5 -11,100 +33,900 Northeast 49.8 34.1 -17,100 +151,400 --Ceara 66.8 47.4 Southeast 46.7 36.7 -249,0C0 +253,800 --Sao Paulo 35.4 30.6 South 42.3 30.3 -16,300 +10,800 Center-West 21.7 23.1 -13,200 +53,800 BRAZIL 1/ 43.9 33.3 -306,700 +420,000 1/ Annual enrollment growth for the period 1980-1984 was -6.1% for private schools and +6.2% for public schools. Source: MEC/SEEC, Sinopses Estatisticas de Ensino do 2 Grau. - 48 - ANNEX III Page 6 of 19 The public sector can expand the supply of secondary education relatively easily and at low unit costs by offering night session secondary instruction in buildings which are used for primary school instruction during the day. Sao Paulo adopted this policy in the early 1970s, and the results can be seen in Table 5. The private share of day session enrollments declined by 5.1 points between 1980 and 1985, while its share of night session enrollments decreased by 12.2 points. Furthermore, the private share of night session enrollments has not grown since 1985, in contrast to day session enrollments. While this may reflect several factors, the most plausible hypothesis is that the supply of public secondary school instruction nas continued to increase much more rapidly in night than day sessions. Table 5 Private Share of Enrollments in Secondary Education in The State of Sao Paulo, 1960-1987 Private Share Private Share Private Share of of of Year Total Enrollments 1%) Total Day Session Total Night Session Enrollments Jj) Enrollments 1%) 1960 1970 37.3 1975 39.0 1980 40.6 42.3 39.4 1981 38.7 1982 35.3 39.0 33.0 1983 33.3 37.3 30.9 1984 31.1 35.9 28.3 1985 30.8 37.2 27.2 1986 32.8 41.1 28.2 1987 35.1 45.1 29.7 1988* 32.8 46.9 26.3 *preliminary data Sources Annex Table C. - 49 - ANNEX III Page 7 of 19 Excess demand for public education is unlikely to be a primary cause of enrollment growth in private education over the next decade for several reasons. First, there is apparently some excess supply of public school places at present, in part because enrollment capacity and population growth are not located in the same place. Second, total secondary education enrollments are projected to grow at a moderate annual rate over the next decade, and the percent of the 15-19 year old age group attending secondary school is expected to grow only slightly. The modest growth projections reflect, in part, the fact that a high proportion (72-77%) of primary school graduates already enroll in secondary education. Relative Oualitv of Private Education Increasingly private education is demanded because of perceived differences in the quality of instruction. Educational observers in Brazil note the shift in enrollments by upper income families away from public to private secondary schools over the past two decades as reflecting perceived changes in the relative quality of private schools. This shift in demand is not unique to Brazil; as participation rates in public education increase, the average level of knowledge of entering students decreases, and public education is perceived to decline in quality. This perception of higher quality private education appears to exist among all income groups in Brazil, which explains in part why surprisingly large numbers of lower income parents send their children to low-cost, low-tuition private schools over no-tuition public schools. Perceptions of quality differences do have some basis in fact. Pass rates on university entrance examinations, for example, tend to be higher for graduates of private secondary schools, and graduates of public secondary schools are more likely to need to take additional coursework ("cursinho") to pass the vestibular exams. 5/ Private school quality is also perceived as being higher due to the lower frequency of strikes and resulting longer effective school-year than is often found in the public sector. Student-teacher ratios are not, however, necessarily smaller in S/ FUVESP, the organization which administers university entrance examinations for the University of Sao Paulo, reports a pass rate (in 1988) of 9.1% for private school graduates versus 4.4% for municipal secondary school graduates, 5.4% for state secondary school graduates, and 15% for the small number of graduates of the federal technical schools. FUVESP also reports that taking a preparatory "cursinho" makes almost difference in the performance of private school graduates but a large difference in the performance of public school graduates. In 1987, for example, only 3.2% of public school graduates not having a "cursinhou passed the university entrance examination compared with an 8.1% pass rate for those having taken a "cursinho". - 50 - ANNEX III Page 8 of 19 private schools. 6/ Finally, quality may be perceived as higher due to the lower dropout rate during the school year: in 1984 23.7% of students entering public secondary schools dropped out by year's end; the corresponding number for private secondary schools was 14.4%. 7/ Law 5692 to Law 7044 According to Brazilian scholars, the catalyst in altering popular perceptions about the relative qualities of private and public education was Law 5692 passed in 1971. 8/ Law 5692 required the vocationalization of secondary education in Brazil; all pupils were to receive practical, vocationally-oriented training in addition to more academic, general education. The law was widely implemented in public secondary schools, but was not rigorously enforced in private schools. 9/ The result was public schools having a stronger vocational orientation than the private schools. The move to vocationalize secondary education was egalitarian in the sense of preparing everyone for practical work, but it was also unrealistic in that most parents send their children to secondary school in order to gain admission to college. 10/ The result was that parents began to view private secondary education as doing a better job at providing the knowledge required to gain admission to good quality universities. Access to high quality, free public universities became worth the investment in private secondary education. Law 5692 served to increase the demand for private secondary education but, more importantly, changed the reputation and perceived quality of private education. Hence, when Law 7044 was passed in 1982 6/ See Annex Table B, which shows the student-teacher ratio as being only slightly smaller in private than public secondary schools in Brazil in 1984. 7/ Of course, none of these perceived differences in output are necessarily attributable to more school resources or higher effectiveness in private schools. Higher output in private schools may simply reflect the socio-economic and motivational characteristics of the children enrolled. 8/ See, for example, Luscher and Mafra (1986). 9/ Ambiguity in the definition of vocational education and variation in the ways in which such education can be provided made enforcement difficult in any case. 10/ This was even more true in the early 1970s, when the secondary education enrollment rate was lower. - 51 - ANNEX III Page 9 of19 rescinding universal, obligatory vocational education no immediate shift occurred either in terms of instructional content of the public schools or in demand for public secondary education. Law 5692 did not in and of itself cause the shift in demand by higher income groups for private education, but it helped to speed the process. rAcomes In addition to the availability of free public secondary education and the perceived relative quality of private education, income and prices play an important role in affecting demand for private secondary schools. Part of the decline in private enrollments between 1980 and 1985 can be attributed to the economic crisis which reduced real family income. As seen in Table 5, recent enrollment data for Sao Paulo indicate an increase in the private school share of total secondary enrollments since 1985; all of the increase, however, is due to growth in daytime enrollments. Day sessions and night sessions appear to be two distinct secondary education markets. More than half (56%) of all secondary school students attend night sessions. Higher income families, however, are more likely than lower income families to send children to private schools, and a majority of children from higher income families attend day sessions. Most children from lower income families on the other hand attend night sessions; AB shown in Table 6, one explanation for the large number of students attending night sessions is the high proportion of students who work. Table 6 Income Distribution, Enrollment Shares, and Work Experience in Secondary Education, 1982 Percent of Percent of Students in Students in Private Secondary Income Education Education Classification Who Work < 1 minimum salary 29.0% 25.3% 1-2 minimum salaries 34.9 41.0 2-5 minimum salaries 36.2 43.8 5-10 minimum salaries 41.9 45.5 > 10 minimum salaries 60.2 31.3 Source: IBGE, Anuario Estatistico do Brasil, 1983; PNAD, 1982. - 52 - ANNEX III Page 10 of 19 Private School Tuition Another factor which ma- explain resumed growth in private secondary education enrollments since 1985 is lower real prices. The Plano Cruzado froze private school tuitions at relatively low levels, and subsequent price controls kept real tuition levels low. Table 7 Private School Monthly Tuition Rates in Sao Paulo, Ceara and Parana by Type of Secondary School, 1987 and 1988 (in current cruzados) Minimum Maximum Mean TYPE OF SCHOOL Sao Paulo 1/ . General Education 1 43,041 3,233 . Teacher Training 1,937 8,980 3,924 . Technical Accounting 1,408 6,480 3,234 Data Processing 3,043 8,600 5,886 * Adult Education (supletivo) 1,807 6,760 3,852 Ceara (March 1988) 2/ . Fortaleza Only 1st and 2nd years 926 7,630 4,267 3rd year 926 9,219 5,078 . Interior Only 1st and 2nd years 350 2,807 1,470 3rd year 708 2,807 1,595 . Total State 1st and 2nd years 350 7,630 3,347 3rd year 708 9,219 4,015 Parana (November 1988) 3/ . General Education 871 38,276 13,840 _____________________________ - 53 - ANNEX III Page 11 of 19 Sources: Note: The same data converted into US dollars appears in Statistical Appendix Table 42 of the main report. 1/ Data from commissao de Encargos Sociais of the Conselho Estadual da Educacao for September 1987. 2/ Data from the Conselho Estadual de Educacao of Ceara, March 1988. 3,' Data from the Conselho Estadual de Educacao of Parana, November 1988. Tnere is no reliable information on how private school tuition levels have changed over time. There is evidence, however, of wide variation in tuition levels. Table 3 demonstrated how tuition paid varies with family income levels. Table 7 provides further evidence that tuition levels vary widely by location of school, type of education offered, and year of education. Tuition charges tend to be lower outside than inside the capital city, lower for the first two years than the final year of instruction, lower for teacher education and vocationally-oriented instruction than for general instruction, and lower for adult than regular instruction. There is, however, very substantial variation within each of these categories. The price actually paid by parents deviates from published tuition rates for several reasons. First, the schools themselves typically offer discounts for families enrolling more than one child. Second, the schools offer some need-based partial or full scholarships. In Sao Paulo, 6-8% of all students receive scholarships, usually for the full amount of tuition. 11/ In Brasilia, on the other hand, approximately 20% of all private students are on scholarship. 12/ Third, until 1989 families who chose to itemize taxes could deduct a maximum of USS 641.5 (in 1987) in educational expenses per child from their taxable income. The -alue of this deduction depended on the family's tax rate, which varied from 5% to 60%; for families in the highest income category the after-tax price of private education was reduced by more than half. The elimination of this deduction has increased the relative price of private secondary education to current itemizers and can be expected to decrease the demand for private schooling. How large this effect might be, however, has not been studied. Finally, the state and federal governments provide some schol&rships for students to attend private secondary schools. The states (e.g., Ceara) scmetimes provide scholarships for private schools where public schools do not exist {e.g., in fast growing areas of the periphery of Fortaleza). 11/ Braga and Cyrillo (1988). I2/ These are data self-reported by the private school syndicate-- Sindicato dos Estabelecimentos de Ensino do Distrito Federal, A Ouestao Salarial e a Realidade da Escola Particular no Distrito Federal, 1986. - 54 - ANNEX III Page 12 of 19 These scholarships are usually large enough to pay only for low-cost private secondary education, and the schools may require students to pay an additional amount. 13/ The federal government provides private scholarships through the Educational Assistance Foundation (FAE--Fundacao de Assistencia ao Estudante); these scholarships can only be used in non-profit schools. In 1987 it funded 62,234 scholarships with a total value of US$ 1.38 million; the value of each scholarship was thus less than one month's tuition charge at a low-cost private school. 14/ C. Regulation and Supply of Private Secondary Education In 1969 the responsibility for regulating both public and private secondary schools was assigned to the state education councils (CEE-- Conselho Estadual de Educacao). A year prior to opening, a new private school must register with the council, stating its curriculum, faculty qualifications, and tuition rate, which it is initially free to set. In principle, the council monitors all schools and assures they comply with all educational requirements and regulations. In practice, they only enforce regulations when confronted with consumer complaints. The council is composed of twenty-four individuals appointed by the governor; their positions are not full-time, and the council's staff is too small to consistently monitor and enforce regulations. The Council assigns responsibility for regulation of tuition charges to a special price commission (Comissao de Encargos Sociais) which includes representatives from parents, students, public schools, and private schools. The rules adopted for regulating tuition charges vary from state to state. Parana simply allows tuition rates to rise with the consumer price index. Sao Paulo allows a 10% markup, and includes an estimate of depreciated capital in costs. Ceara allows a 10% markup; actual building rent but not imputed rent can be included as a cost. Brasilia allows a rental/depreciation charge equal to 8% of capital value and a 10% markup on total costs. Other rules exist with respect to what proportion of teacher salary increases can be passed on in the form of tuition increases. Tuition price controls can reduce the quantity and quality of secondary education supplied privately if they are set lower than equilibrium market prices. The evidence is not clear for Brazil. The only careful study of the issue is provided by Braga and Cyrillo for Sao Paulo Li 1985, who conclude that actual tuition charges are set by the market 13/ Relative to the total public secondary education budget, private school scholarship assistance is negligible; Ceara allocates less than one percent of its secondary education budget to scholarships. 14/ FAE, Relatorio 1987. - 55 - ANNEX 'I1 Page 13 of19 below the maximum prices set by the council. Other evidence suggests that tuition price controls do effectively constrain tuition charges. The Plano Cruzado, for example, requirid reductions in real tuition charges. In February 1988 the federal government removed all tuition price controls only to quickly reinstitute them when a large number of private schools responded by raising tuition levels. The number of private secondary schools contracted slightly between 1980 and 1985, but not as sharply as the decline in enrollments, and there is some evidence of excess capacity in the sector. 1S/ Braga and Cyrillo conclude that while most private schools more than cover their recurrent costs, few cover their capital costs fully, which discourages new private investments in schools. In conclusion, tuition price controls have clearly been constraining at times in the past, but in general tuition levels appear to be more strongly influenced by market forces than binding price controls. The greatest harm done by price controls is to introduce an additional element of uncertainty into schooling investment decisions by entrepreneurs. The new constitution will influence the supply of private education by increasing the price of labor (due to mandated worker benefits), possibly liberating price controls, and possibly increasing the supply of public education. 16/ D. Conclusions and Policy Issues The private sector share of secondary education enrollments declined between 1980 and 1985, partly in response to increased public supply. At the same time, private schools have maintained a reputation for offering higher quality instruction than public schools. Neither of these trends is likely to be reversed in the long run. Public policy affects private secondary education through (i) regulations, (ii) subsidies, and (iii) government competition. Aside from tuition charges, private education is not stringently regulated in Brazil, 15/ However, as noted earlier, there is also some evidence (see Table 5 for Sao Paulo) that private secondary school enrollments have begun to increase again since 1985. 16/ What the new constitution says about price controls is not yet clear; the private schools believe it eliminates them, while the state education councils believe not. The supply of public secondary education may increase as a result of increases in revenues earmarked for education, especially at the federal level. - 56 - ANNEX III Page 14 of 19 perhaps because it receives so little funding. 17/ While tuition price controls are nnt always constraining, strong arguments can be made for their elimination. At a minimum, the existence of price controls makes it difficult for established schools to follow a strategy of simultaneously raising instructional quality, costs, and tuition charges. Also, by increasing uncertainty regarding future pricing policies, tuition price controls bring about some reduction in the supply of private secondary education. The only important subsidy to private secondary education in Brazil has been the federal tax subsidy, which was eliminated in the recent federal income tax reform. This was a positive step as tax subsidies for education in a country such as Brazil, where only a small share of the population files personal income taxes, tend to be quite regressive. On the other hand, both efficiency and equity arguments can be made for targeted subsidies to private education. An efficiency argument is that a given amount of public funds expended in the form of subsidies can stimulate a larger supply of private education than can be directly provided by government. An equity argument is that subsidies can be targeted to lower income families, something which is difficult to accomplish through direct public provision. Increased government supply of secondary education does not necessarily raise secondary education participation rates. The evidence for 1980-1985 in Brazil suggests that increased public enrollments were almost precisely offset by reduced private enrollments. To the extent this remains true, only ideological grounds provide a justification for increased government supply of secondary education. Several policy recommendations follow from the above analysis: 1. Tuition price controls on private secondary education should be phased out to further stimulate private supply. 2. Government should introduce new programs to subsidize the private provision of secondary education; subsidies could be (i) institutional in the form of matching grants to cover capital construction costs and (ii) individual in the form of need-based scholarships (or vouchers) to students to attend private schools. 3. Given phased-out price controls and private sector subsidies, government should expand the provision of public secondary education only in special circumstances where the private sector response is for some reason inadequate. 17/ James (1987) notes the correlation across countries between the amount of public funding received by private schools and the amount of public regulation to which they are effectively subjected. - 57 - ANNEX IIx Page 15 of 19 These recommendations are consistent with the fiscal and educational environment currently facing Brazil--relatively low levels of national educational investment, constrained public expenditures for education, and a dynamic private education sector. These recommendations are not, however, consistent with the experience of many other countries, where the private sector share of secondary education enrollments declines as per capita income increases and public sector regulations increase with the level of subsidies to private schools. The experience of other countries as communicated by James is either (i) low public regulation, low public subsidies, and v )w private market shares or (ii) high public regulation, high public subbidies, and higher private market shares. With only 37% of the age group currently enrolled in secondary education, Brazil's secondary education system will undergo dramatic changes in the future. It has the opportunity to develop its own unique model. - 58 - Page 16 of 19 Table A Secondary Education Statistics by Family Income Quartiles, 1982 Income guartile First Second Third Fourth TOTAL Percent of Students Entering 66.1 63.3 69.8 79.0 72.3 Secondary School in 1982 Who Graduated From Public Primary School in 1981 Percent of Students Entering 60.5 62.4 74.9 81.8 77.4 Secondary School in 1982 Who Graduated From Private Primary School in 1981 Percent of Students Enrolled 28.8 30.7 34.4 51.0 43.0 in Private Schools Percent of Students Enrolled 64.5 70.2 67.3 46.6 56.1 in Night Sessions Percent of Students i? 60.6 66.9 63.7 49.3 57.2 Public Schools Enrolled in Night Sessions Percent of Students in 73.9 77.7 74.0 44.0 54.7 Private Schools Enrolled in Night Sessions Average Age of Students 18.7 17.9 17.3 16.2 16.9 Entering Secondary School Average Monthly Private $7.36 12.13 19.07 41.96 33.58 School Tuition Paid in 1982 (US$) Average Per Capita Family $30.50 46.94 83.52 255.67 193.79 Income of Children in Private Schools in 1982 (in US$) Source: Pesquisa Nacional por Amostra de Domicilios (PNAD), 1982. - 59 - ANNEX III Page 17 of 19 Table B Ratio of Students to Schools and Teachers in Secondary Education Students/ Students/ Schools Teachers Brazil (1984) Federal Schools 695.9 12.9 State Schools 416.9 15.5 Municipal Schools 195.8 12.6 Private Schools 244.2 11.9 Mias Gerais (1979) Public Schools 307.8 16.6 Private Schools 323.6 17.5 imas Gerais (1973) Public Schools 221.0 10.8 Private Schools 174.7 10.6 Sore________sce_nd_fa_18) Source: SEEC/MEC; Luscher and Mafra (1986). - 60 - ANNEX II I Page 18 of 19 Table C Sao Paulo State Secondary School Enrollments (in thousands) State Municipal Private Year Day Night Day right Day Night 1975 181.4 132.4 1980 174.8 241.4 3.0 14.4 130.5 166.5 1981 1982 173.1 289.8 1983 1984 185.5 329.7 2.6 14.0 105.2 135.6 1985 184.3 361.1 1.8 14.2 109.8 140.2 1986 173.7 378.5 2.6 14.0 123.3 154.5 1987 162.7 376.7 2.3 13.9 135.7 165.4 1988 164.2 411.6 2.4 14.3 137.5 151.9 Source: SEE, Sao Paulo. - 61 - ANNEX III Page 19 of 19 Table D Tuition Rates and Family Income by Region, 1982 (in US$) Monthly Tuition Charge Average Income Per Private Secondary Capita of Children Schools in Private Secondary Schools BRAZIL $33.58 $193.79 * North 30.98 189.58 . Northeast 23.29 124.66 . Southeast 37.95 214.36 . South 29.55 191.61 . Center-West 38.23 236.23 Sr:_uai___orae ii_PA 19 Source: Pesquisa Nacional por Amostra de Domicilios (PNAD), 1982. - 63 - ANNEX IV Page 1 of 3 MODELO PARA PROJEgAO DA MATRfCULA NO SEGUNDO GRAU - BRASIL-' A metodologia 6 constituida de duas fases. A primeira diz respeito ao ajuste do modelo, onde os parametros do modelo sAo estimados. A segunda 6 a fase de projeqao, na qual o modelo estimado 6 usado para projetar as matriculas. Inicialmente a primeira fase sera descrita, para depois entao, explicar-se as projeg6es de matriculas. Considere os seguintes valores e notagoes: Mtp : matricula no Primeiro Grau no ano t; Mt, : matricula no Segundo Grau no ano t; Ntp : alunos novos no Primeiro Grau no ano t; Nt0 : alunos novos no Segundo Grau no ano t; atj: proporgao de alunos do Grau i no ano t que no ano t+1 estar& no Grau _j (taxa de transig&o). E' claro que o nuimero de alunos matriculados num nivel no ano t+1 6 igual ao numero de alunos novos no ano t+1 adicionado do nuimero de alunos matriculados neste nivel no ano t que continuam no mesmo nivel no ano t+l. Assim pode-se escrever Mt+Ip = at,,Mt, + Nt+p; Mt+ _= ata4t, + Nt+l' atj4t. + at .Mt . Conhecidos as matriculas no Primeiro e Segundo Grau nos anos t e t+1 e o nurmero de alunos novos no ano t+1 6, ent&o, possivel estimar as taxas de transig6es (at,j). Com isto tem-se o modelo ajustado que podera ser usado para projeg8es de matriculas. As taxas de transic8es atj sao calculadas como: at., = (Mt+'P + Nt+' )/Mt; at., = Nt+l /Mtp e at,, = (Mt+I, + Nt+* )/MtHL if This annex was prepared by Joao Batista Gomes-Neto. - 64 - ANNEX IV Page 2 of 3 Por outro lado se conhecemos a matriz de transi,&o para o Primeiro Grau (no caso brasileiro e de 8 series) podemos entao estimarmos 7 Mti Mt9 aLpp= (rt1 + p1L) + rt, i=1 Mtp Mtp onde MtI : matricula na serie i do Primeiro Grau no ano t rti : taxa de repet&ncia na s6rie i do Primeiro Grau no ano t e pti : taxa de promogao da serie i para a serie i+l do Primeiro Grau no ano t. Esta ultima f6rmula foi usada para estimarse at, onde a matriz de transigoes para o Primeiro Grau foi estimada atraves do Modelo PROFLUXO, de Fletcher e Ribeiro. Neste trabalho as matriculas no Primeiro e Segundo Grau foram obtidas das PNADs (Pesquisa Nacional por Amostra de Domicilios) do IBGE. J& os alunos novos no Primeiro e Segundo Grau foram obtidos usando-se as seguintes f6rmulas: Ntp = Tt -pt e Nt+ ,,= Pt8Mt8 onde, Ttp 6 a taxa de participago (de uma geragao) na primeira serie do Primeiro Grau e foi estimada em 90% atraves do PROFLUXO (Fletcher e Ribeiro), usando dados da PNAD de 1982; pt7 6 a populac&o de 7 anos de idade no ano t; estes valores foram estimado atraves de uma regressAo exponencial usando-se dados das PNADs de 1981 a 1987. pt8 6 a taxa de promog&o da oitava serie do Primeiro Grau que se matricular& no ano seguinte na primeira serie do Segundo Grau; esta taxa tambLm foi estimada atrav6s do PROFLUXO usando a mesma PNAD de 1982; os valores obtidos foram de 60% (rela,co entre a matricula inicial da primeira s6rie do Segundo Grau e a matricula inicial da oitava serie do Primeiro Grau) ou 81% (relaq&o entre a matricula inicial da primeira se6rie do Segundo Grau e o numero do concluintes do Primeiro Grau). - 65 - ANNEX IV Page 3 of 3 Mts 6 o numero de alunos matriculados na oitava s6rie do Primeiro Grau no ano t. Seus valores foram obtidos das PNADs de 1981 a 1987. atpp e atP. foram estimados para os anos de 1982 e 1983. No modelo utilizam- se a m6dia dos dois valores estimados at,, foi estimado para os anos 1982 a 1986. Usam-se a media aritmgtica destes cinco valores estimados. Os valores obtidos foram: at = 88,8%, atps = 3,4% e at.. = 75,1% De posse do modelo ja estimado e com a hip6tese de que os alunos somente entram no sistema na primeira serie do Primeira Grau, podemos projetar as matriculas do Primeiro e do Segundo Grau, desde que se saiba o numero de alunos novos no Primeiro Grau. Utilizando-se o valor da taxa de participagao na primeira serie do Primeiro Grau estimada em 90% por Fletcher e Ribeiio atraves do PROFLUXO, e supondo que esta participa,co nao mude ao longo do tempo, pode-se estimar os alunos novos no Primeiro Grau desde que se projete a populaq&o de 7 anos de idade. A proje,co da populaq&o de 7 anos foi obtida atrav6s de uma regressao exponencial usando-se dados das PNADs de 1981 a 1987. Como as PNADs excluem a Zona Rural da Regiao Norte, e importante frisar que as projegoes aqui consideradas tarnb6m excluem esta Zona. Isto, porem, n&o invalida os resultados, tendo em vista que, de acordo com o Censo de 1980, esta Zona corresponde apenas a 3.5% da popula,&o total do pais. Usando o modelo estimado (i.e., a'.. = 88,8%, at., = 3,4%, at8, = 75,1% e TtL = 90%) tem-se na tabela I a projec&o de matriculas no Segundo Grau para o Brasil no periodo Qtc 1990 a 2010. Tabela I - Matricula no 2o Grau Brasil 1990 - 2010 Ano Matriculas (em mil) 1990 3477 1995 3848 2000 4371 2005 5060 2010 5917