Educational development and costing in Mexico, 1977-1990 - a cross-state time-series analysis

A View from LATHR No. 25 EDUCATIONAL DEVELOPMENT AND COSTING IN MEXICO, 1977-1990 A CROSS-STATE TIME-SERIES ANALYSIS by Juan Prawda and George Psacharopoulos Human Resources Division Technical Department Latin America and the Caribbean Region The World Bank November 1991 "A View from LATHR' is a series of occasional flyers produced by the Human Resources Division of Latin America and the Caribbean Technical Department of the World Bank for the purpose of stimulating discussion among staff on key issues facing the sector. The views expressed here are those of the authors and should not be attributed to the World Bank. Abstract The paper uses time series (1977-1990) and cross-sectional (States) data on a number of educational development and cost indicators in Mexico to answer the question on how schooling has been affected during the economic downturt of the 1980's. The results show that, in spite of severe cuts in budgetary allocations to education (mainly in terms of lower teachers' salaries), educational provision continued to improve throughout the period. This apparent puzzle is explained in terms of the use of a more efficient management of the school system. Preschool coverage has increased dramatically between 1977 and 1990, now covering nearly three quarters of the eligible population. The expansion of preschool education contributed to a higher primary school completion rate. This Mexico natural experiment provides a lesson that a country does not have to wait for an economic crisis in order to introduce cost-saving measures in education. The authors wish to express their gratitude to Luis David Figueroa, Director of Information Systems and Analysis, within the General Directorate of Planning, Programming and Budgeting of the Mexican Ministry of Education for his help in providing the relevaft statistical and unit cost information used in this paper. Laurence Wolff, Eduardo Velez and Sam Carlson commented on earlier drafts of the paper. I. Introduction Mexico is a country with an adult illiteracy rate of about 8 percent and 6.2 grades of average schooling. The current net primary enrollment ratio for children 6 to 1 1-years-old is 98 percent' and the primary completion rate of a six year cohort of 57 percent. Although far from desirable standards, Mexico fares above the average;registered for Latin America as a region.2 However, these averages hide enormous variation among the 32 states (including the capital city). For example, in 1989 the adult illiteracy rate in Chiapas (16 percent) was over five times the one in Mexico City (3 percent), and the adult average number of years of schooling in Oaxaca (4 grades) was half of the one corresponding to Nuevo Leon. Also Mexico is a country that was hit hard by the economic crisis of the last decade, where real per capita income fell 21 percent in the 1981-1988 period. How has education been affected by the economic crisis, especially in the poorest states? Considerable evidence exists showing that text-books, writing materials and teacher quality consistently influence student achievement.' Acknowledging the above, this paper combines time-series and state data for Mexico to answer a major question that has been posed World-wide since the early years of the debt crisis: How does an economic downturn affect educational development? In the process of doing so we also explore other non- quantitative determinants of educational development, such as management style and the effect of class size on unit costs. II. Data Description The following data have been compiled for a fourteen year period (1977 to 1990) for each of the country's 31 states and its capital city. 'Prontuario Estadistico, Cultural y Presupuestario. Direcci6n General de Planeaci6n, Programaci6n y Presupuesto (DGPPP), Secretarfa de Educaci6n PNblica (SEP), M6xico, 1990. The following averages are reported by OREALC-UNESCO for Latin America and the Caribbean: 86 percent net primary enrollment; 14.7 percent preschool coverage for the 0 to 5 year cohort (21.3 percent for Mexico), and 11 percent adult illiteracy. Situaci6n Educativa de Amdrica Latina y el Caribe, 1980-1987, UNESCO-OREALC. See also Rockwell (1990). * Compendio Estadistico por Entidad Federativa. DGPP, SEP, M6xico, 1990. * From a high US$2,840 in 1981 to US$2,250 in 1988. World Tables, 1989-1990 Edition. The World Bank and The Johns Hopkins University Press, 1990. * Fuller, B., Raising School Quality -in-Developing Countries. What Investment Boast Learning. The World Bank Discussion Papers. No. 2, Washington, D.C., 1986. 2 Educational Development Indicators (in percent) . preschool coverage -- the proportion of 5-years old in a given state enrolled attending any modality of preschool education. . incomplete schools -- the percentage of primary schools in a given state offering less than six grades, the legal duration of the primary school cycle. . primary efficiency or completion rate -- the percentage of students exiting sixth grade in year t out of students entering first grade in year t-6. . secondary absorption -- the percentage of primary school leavers in a given state entering secondary education in the same state. . student-teacher ratio -- the number of students per teacher in preschool, primary and secondary education. . management -- the quality of a state's educational management determined by expert opinion (coded -1 "below average", 0 "average", 1 "above average"), using six objective criteria (see footnote to Table A-3). Cost indicators (gross to the Government in 1980 constant MX pesos) . preschool unit cost -- the average cost of preschool education in a given state per student year . primary school unit cost-- the average cost of primary education in a given state per student year. secondary school unit cost -- the average cost per secondary education in a given state per student year . administration cost -- school administration expenditure in a given state divided by the state's population (reminder that about 90 percent of the disbursed budget goes to pay salaries). . teachers' salaries -- gross monthly salary of preschool and primary teachers, and hourly salary for secondary teachers; it excludes monetary supplements for didactic material, seniority, vacation premiums, and yearly bonus. 3 Economic Indicators per capita income - the state's per capita income (in 1980 constant pesos) delta income -- the average annual rate of change of per capita income in constant prices between 1980 and 1985 (in percent) Population Indicators . population -- the state's population (in millions) . delta population -- the average annual rate of change in the state's population between 1980 and 1988 (in percent). M. Descriptive Statistics Table 1 presents the means and standard deviations of the variables. The Annex tables show the value of the mean variables for each state and year. Figure 1 portrays the 14 year trend of the educational development indicators. There has been improvement in all indicators, although for some the improvement has been more dramatic than others. The sharp increase in preschool enrollment and decrease of incomplete primary schools at the national level stem from the sustained implementation, since 1978, of the following objectives in the Mexican Six Year Educational Plan and the Educational Plan of the 1982-1988 Administrationg: (a) provide 70 percent preschool coverage to 5 year-old children; (b) provide 100 percent primary education coverage to all children in the 6 to 14 age cohort; and (c) increase the primary school completion rate to 75 percent. As a consequence of the above, and the implementation of other specific measures', the national primary completion rate experienced a slight increase of about one point of a percent per year in the period of analysis. - All mean statistics and regressions reported here are weighted by the state's population in 1988. * Secretarta de Educaci6n Ptblica, Memorias 1976-1982, Vol 1, (Politicas Educativas), Secretarfa de Educaci6n Piblica, M6xico, 1982. * Programa Educativo 1983-1988. Secretarfa de Educaci6n Ptiblica, M6xico, 1984. Implementation of compensatory and-preventive special programs for children at risk of failure and repeaters. 4 Table 1 Means and Standard Deviations of the Variables Symbol Variable Mean Std.Dev. N PRES Preschool coverage 44.92 24.77 448 INCOMPS Incomplete primary Schools 28.11 17.51 448 PRIMEFF Primary school efficiency 53.61 14.08 448 SECABS Secondary school absorption 81.94 13.37 448 STPRE Preschool Student-teacher ratio 32.60 5.04 448 STPRIM Primary Student-teacher ratio 36.95 5.09 448 STSEC Secondary Student-teacher ratio 18.56 1.85 448 CPRE Preschool unit cost 6654 3651 416' CPRIM Primary unit cost 4831 1900 416 CSEC Secondary unit cost 10800 5034 416 CAPMPC Adm. cost per capita 148 67 416 MANAGE Management style -.03 .53 448 POP80 State population in 1980 4181 2906 448 POP88 State Population in 1988 5151 3835 448 Y80 State P/C income in 1980 63864 33041 448 Y85 State P/C income in 1985 62454 32096 448 DELTAP State population change, 80-88 2.52 1.58 448 DELTAY State inc. change, 80-85 -.24 2.47 448 Source: Based on information supplied by the Direcci6n General de Planeaci6n, Programaci6n y Presupuesto Secretarfa de Educaci6n Ptiblica, M6xico 1990. 'The cost time-series are available since 1977. The national average secondary absorption did not increase at the same rate as the primary completion rate and oscillated around the 80 percent plateau. As a consequence, further education opportunities were hindered for many primary graduates, especially in the rural areas. Figure 2 shows that the unit cost of all levels of education followed a downward trend during the period under investigation, the most dramatic one being that for secondary education, which decreased by about 65 percent in the 1977-1990 period. The preschool unit cost went down by half between 1977 and 1989. However, this is not an indication of efficiency as the 5 year-old enrollment grew 6.2 percent per year, but the preschool teacher labor force increased 14 percent per year. As a consequence, the preschool student teacher- ratio decreased from 39 in 1977 to 26 in 1990 (figure 3). Given that about 90 percent of 5 Fioure 1 Educational Development Indicators Trends 100% Sec. school absorption so%- 60% Primary education efficiency 40%- Incomplete primary schools 20% Pre-school coverage 0% - I I II I I I I I I t Year 1977 1980 1983 1988 1990 Fiqure 2 Educational Unit Cost and Per Capita Income Trends Unit Cost (1980 Mex.Pes.) GNP per Capita (US$ 1980) V3000 21000- -2500 Per capita income - 2000 16000 - - 1500 Primary 11000- teachers' salaries Secondary cost ------ -1000 000 Primary coo -re-school cos- 500 1000 1 0 1978 1980 1982 1984 1986 1988 1990 6 preschool expenditures are earmarked for teachers' salaries, the above unit cost reduction is explained by a real decline in teachers' wages", as explained below (table 2). The same reasoning explains the decline in the primary unit cost, which was reduced by 42 percent during the period under investigation. Whereas the primary education enrollment increased 1.4 percent per year", the primary teacher labor force has steadily increased at 4.1 percent per year12. The student teacher rAtio decreased 31 percent in the period of analysis (figure 3). The most plausible explanation of the decline in primary education unit costs seems to be again a decline in primary teachers' wages,14 which account for about 90 percent of the total federal budget expenditure in primary education. The secondary unit cost decrease is partially explained by enrollment growth efficiency (the student-teacher ratio increased from 17 to 19 in the period of analysis)", and by a very sharp decline in real teacher wages. In Figure 2 we have superimposed to the cost data the country's real per capita income and the primary teachers' real gross monthly salary. The decline in education unit costs since 1981 mirrors nearly exactly the drop in per capita income and teachers real gross monthly salaries. The decline in preschool, primary, and secondary unit costs is also explained by a reduction in real terms of the federal expenditure in such levels. Whereas in 1983 the disbursed federal budget for the above educational levels was 198.5 billion pesos (53.8 percent of the total disbursed budget), in 1987 it decreased to 121 billion constant pesos (50.4 percent of the total share)", for a total decline of 39 percent. 10 Real wages for preschool and primary education teachers declined by about 41 percent in the 1976-1991 period. "Primary education enrollment went from 12 million students in 1976 to a high of 15.3 million in 1983, and has gradually, but consistently, decreased ever since to its current level of 14.6 million in 1990. 12 From 272,000 in 1976 to 426,000 in 1983, and then to 478,000 in 1990. ' From 44.1 in 1976 to 30.5 in 1990. 14 It is also interesting to note that, after 1983, while primary education enrollment has shown a steady annual decrease of 0.7 percent, the hiring of primary school teachers has increased 1.7 percent per year. * Secondary enrollment went from 2.1 million students in 1976 to 4.4 in 1989 (5.9 percent increase per year). The secondary teaching force grew from 125,000 to 245,000 in the same period (5.3 percent increase per year). 1 Compendio Estadistico del Gasto Educativo. DGPPP, SEP, Mexico, 1990. 7 Table 2 Gross Teachers' Salaries and Minimum Wages (Constant Mexican Pesos of 1980) Average Teachers' Salaries Year Preschool, Secondary Minimum Primary Wage (Monthly) (Hourly) (Daily) 1976 10,323 616 158.5 1977 9,054 535 164.9 1978 9,027 531 160.2 1979 8,920 520 154.2 1980 10,070 581 140.7 1981 10,095 540 145.3 1982 8,828 422 120.7 1983 5,828 255 103.1 1984 5,649 223 97.4 1985 6,374 275 97.5 1986 5,147 217 87.9 1987 5,802 280 68.7 1988 3,451 183 86.8 1989 3,888 182 76.0 1990 4,091 185 78.9 1991 6,135 213 81.0 Annual Decrease (percent) 3.4 6.8 4.4 Source: Direcci6n General de Planeaci6n, Programaci6n, y Presupuesto, Secretarfa de Educaci6n Piblica, M6xico, 1991. Comisi6n Nacional de Salarios Mfnimos*, Serie 1976-1991, M6xico 1991. Table 2 presents the preschool, primary, and secondary teachers' wages by year in the period 1976-1991, as well as the minimum daily wages of unskilled workers. The gross monthly preschool and primary teachers' salary decreased almost 41 percent in real terms in the above mentioned period, that is 3.4 percent per year. On the other hand, the gross hourly secondary teachers' salary decreased 65 percent in the same period, that is, 6.8 percent per year (twice the decrease in primary teachers' salaries). The teachers' union negotiated with the Government for better primary teachers' wages than for any other educational level, as primary teachers control the union and represent the largest share of the 8 total teaching labor force. Further evidence of the teachers' union bargaining power is that the decrease of primary teachers' salary in the 1976-91 period was one third less than the minimum daily wage decrease for unskilled workers (4.4 percent per year). Figure 3 shows the behavior of the student-teacher ratio in the three levels of education over time. The reduction of this ratio (i.e. less students per teacher) has been constant throughout the period for preschool and primary education, unaffected by the economic crisis. The student-teacher ratio for secondary education has slightly increased. The reduction of the preschool and primary student-teacher ratios is partially explained by: (a) the teachers union pressures targeted at the Ministry of Education to hire all the graduates from official primary and preschool Teachers Training Institutions, in spite of existing teacher wastage" in the public primary schools in almost all the large urban areas of the country; and (b) the rapid expansion of preschool services which required the hiring of all available preschool teachers. The enrollment increase in secondary education had to be met with fewer teachers due to their scarcity and higher cost. Figures 4, 5, and 6 show the behavior of the educational development indicators by a state's management style. Management style has been defined on the basis of the following objective criteria: (a) turnover rate of the SEP's state Directors in the period 1978-1991; (b) number of training courses undertaken by the SEP's state technical staff in a 3-year period; (c) use of microplanning and programming techniques, by SEP's state technical staff to diagnose current situation (process monitoring) and forecast short term (not more than 3 years) education demand and to request and disburse resources in an efficient way; (d) quality control on the educational statistical information gathered annually by SEP's state agencies; (e) number of days the public schools operate in a given year as percent of the legal school year duration; and (f) number of teachers' strikes in a given year. The weights, scale used for each variable, and the ranking procedure appear in Table A-3 in the Annex. As expected, above-average-managed Ministry of Education (SEP) state agencies, which not necessarily had the "easier task", achieved larger: (i) increases of preschool coverage of 5 year-olds; (ii) reduction of primary incomplete schools; and (iii) increases of primary completion and secondary absorption rates than the below-average-managed SEP state agencies (figure 4 ). They are also more efficient in preschool as their student-teacher ratios are higher (figure 5). The student-teacher ratio in secondary education is almost the same in the three categories of management in SEP's state agencies (figure 5). " In 1987, there were about 6,000 public primary teachers in Mexico City, not counting the Metropolitan area (14 percent of the public teaching labor force) collecting a bi-weekly salary and not working. The demographic changes in the Mexican large urban areas--rural immigration to the slums, and the emigration of the better off to the suburbs-and the rigidity to transfer teachers from one school district to another imposed by the Union, has made the above situation worse with time. Direcci6n General de Planeacidn, Programaci6n y Presupuesto, Secretarfa de Educacidn Piblica, 1989. 9 Fiqure 3 Student-Teacher Ratio and Per Capita Income Trends Student-Teacher Ratio GNP per Capita (US$ 1980) 50 3000 2500 4o0- Per Capita Income Pre-school Primary 2000 30 - 1500 20 - Secondary 1000 10 - 500 1978 1978 1980 1982 1984 1988 1988 1990 Fioure 4 Educational Development Indicators by Management Style 100- 80- 60- 40- 20- 0- Pres. Cov. % 1com. Prim. Prim. Com. Rate Trans. to Sec.Ed. Below Average E Average Good Management Style 10 Figure 5 Student-Teacher Ratio by Management Style 40- 30- 20- 10- 0 Preachool Primary Secondary Slow Average M Average = Good Management Style Fioure 6 Unit Cost by Management Style 14000- 12000- 10000- 8000- 6000- 4000- 2000- 0000 Preschool Primary Secondary Below Average M Average 0 Good Management Style 11 Figure 6 shows that well managed SEP state agencies have higher unit costs than the average and below average managed agencies. Since the former states have a better preschool coverage, higher primary completion and secondary absorption rates than the latter states, the marginal cost of serving and/or retaining one extra student appears to be an increasing function of enrollment. This explanation seems to suggest that low enrollment states do not yet experience diminishing returns as it is more costly to serve students living in small, isolated rural communities than in the urban or tural-urban areas, which are more densely populated. It also shows that cost-effective models have not been generally implemented to serve a more costly clientele". On the other hand, it could also suggest that well managed SEP state agencies are crowding pupils in, with associated quality loss. Unfortunately, there is no cognitive performance data to test this hypothesis. The pattern shown in figure 6 may also be due to the fact that non-salary education expenditures" as a share of the total disbursed budget might be larger in the better managed SEP's state agencies. Better managed states are more efficient in deploying available resources and seem to have less waste. IV. Explaining Educational Development Table 3 presents the results of a model attempting to explain differences in educational development among states over the 14 year period of our investigation. The rationale for inclusion of the independent variables is as follows. "Year" has been introduced in order to capture the time-trend element associated with our pooling of the state and time-series data together. We expect this variable to capture any unmeasured characteristics associated with the change of educational development indicators over time, over and above the ones referring to income and population changes depicted by our "delta" variables. It also gives an idea of policy sustainability, as in the period of analysis three different administrations (and four Secretaries of Education) were in charge of implementing sectoral policies. "Delta Population" has been included in the model on the assumption that rapid population growth in a given state, other things being equal, would have retarded educational development because of the inability of the federal/state to finance it. * The exception being some non-conventional preschool models serving dispersed rural and ethnic 4 and 5 year olds, and the telesecundaria, developed during the seventies, currently serving about 10 percent of total secondary enrollment in Mexico. The telesecundaria was designed and implemented to serve primary graduates from small rural and isolated communities. See Arena, E. "Mexican Telesecundaria System: A Cost Analysis", Paper presented at the World Bank Seminar on Education Technology, Washington DC, 1988. '9 These are resources allocated to: produce local didactic and teaching materials; repair school furniture; provide school maintenance; and support PTA committees. 12 "Delta Income" has been included in order to test the hypothesis that during the economic crisis educational development has suffered. "Management "has been included in order to test on whether better-managed states have exhibited a higher level of educational development. The unit cost of education at the different levels has been included on the assumption that the higher the unit cost of a given educational level, e.g. preschool, the more difficult it would be for the particular state to develop education at that level. "Preschool coverage" enters in a recursive manner in the function explaining primary education completion, on the assumption that a better-developed preschool system would facilitate the role of primary education. Similarly, "Primary completion" enters as determinant of secondary school absorption, on the assumption that states where primary school completion rates are higher, offer a better raw input to secondary education. It does not necessarily mean however, automatic increases in secondary absorption rates. As shown by the regression results in Table 3, preschool coverage follows an upward trend, regardless of the economic crisis. This must be due to the strong demand for education at that level and sustained policy implementation since 1978 by the different administrations20. Good management is positively associated with increased preschool coverage, which means efficient deployment and targeting of available preschool resources as shown in Figure 421. As expected, rapid population growth and higher unit cost are - negatively associated with preschool coverage, as the incremental preschool supply has to be earmarked to more expensive clientele, usually living in remote and disperse rural villages, with fewer children in preschool age, where teachers seldom want to work. However, the use of less expensive non-conventional preschool models, such as Cursos Comunitarios and Prescolar Indigena, in conjunction with lower real preschool teachers' salaries, as explained below, has contained the preschool unit cost increase. The number of incomplete primary schools shows a reduction over the years, as a consequence of sustained and effective implementation of stated policies by the different 2 This is an important point given the "pendulum" effect 'characteristic of consecutive Mexican government administrations. 1 Preschool growth has been mainly targeted to poor urban and rural 4 and 5-year old children since 1979. As a consequence, whereas in 1976 only a meaningless percent of total preschool enrollment was rural (communities with 2,500 or -less. inhabitants), by 1990 almost 23 percent of total preschool enrollment (about 621 thousand 4 and 5 year olds) is rural. 13 Table 3 Explaining Educational Development Indicators Incomplete Independent Preschool Primary Primary Secondary Variable Coverage Schools Completion Absorption Year 4.607** -2.946** 1.844** -.845** Delta Income -.264 2.034** -2.518** -1.106** Delta Population -4.290** 1.128* -1.358** -3.240** Management 7.455** -10.266** 13.410** 3.874** Preschool Cost -.001** - - - Primary School Cost - -.002** .002** Secondary School Cost - - - -.0005** Preschool Coverage - - -.066 .101* Primary School Completion - - - .425** Constant -9071 5875 -3604 2187 R' .839 .554 .639 .518 N 416 416 416 416 Note: The stars next to the regression coefficients indicate: * Statistical significance at 10% ** Statistical significance at 5% authorities, and concrete interventions22. This variable is, paradoxically, positively associated with the change of income, implying that states experiencing decreases in per capita income (usually the poorest like Oaxaca, Michoacan, and Chiapas), are also the ones Opening of double shifts, allocation of-an additional teacher, construction and/or refurbishing of one extra classroom, allotment of teaching resources. 14 completing more primary schools, as a consequence of targeting specified policies23. The percent of incomplete primary schools is negatively associated with the unit cost of primary education, indicating that the marginal cost of completing primary schools, by adding extra teachers and classrooms, and/or establishing double shifts, even for few students, is an increasing function of the enrollment; this negative association is also explained by a drastic reduction of the student-teacher ratio associated with the incremental grades that are added to complete a primary school. The positive correlation of incomplete primary schools with population indicates that the states that experienced large population increases, mainly through internal migration and/or higher than average birth rates, also displayed the slowest reduction of incomplete primary schools. The negative correlation of incomplete primary schools with management shows that most of the below average managed states, usually poor rural states, were also the ones with the highest share of primary incomplete schools, and thus were targeted by specific sectoral policy actions to reduce the number of primary incomplete schools. The primary school completion rate follows an upward trend, indicating sustained policy interventions since 197624 to achieve stated objectives. Primary completion is: (a) negatively related to the change in income indicating that states with larger decreases in per capita income, usually the poorest, are also the ones with the smaller completion primary rates25; (b) negatively related with population, indicating that states with limited population growth, usually the most developed, achieved the highest primary completion rates; and (c) positively affected by good management, displaying effective deployment and targeting of resources and specific interventionS26. ' The largest proportion of primary incomplete schools during the late seventies and early eighties were located in states that experienced real decreases in per capita income. ' Such as increasing the supply of preschool education targeted to poor urban and rural children which are more at risk during their primary education; completing incomplete primary schools; and implementing preventive and compensatory interventions specially designed and targeted to primary students at risk. Chiapas primary completion rate for the 1985-1990 cohort of students (27.5 percent) is almost one third the one for Baja, California, (76.2 percent). Chiapas had a 1985 per capita income of 40,195 MxPs, whereas Baja, California, had 79,198 MxPs. Although poor in an absolute sense, Chiapas and Oaxaca experienced income gains between 1980 and 1985 (see Annex Table 1). - I Microplanning techniques were utilized to identify clusters of primary schools registering below state average primary completion rates, to be targeted with specific pedagogical interventions (such as reinforced reading and math teaching programs) and additional education inputs (reading resource centers, and teaching materials). 15 Under the assumption that preschool coverage helps to boost the primary education completion rate, we fitted a lagged model in which the primary school coverage in a given year affects primary education completion seven years later. (In doing so we have of course lost seven years of observations for each state). The regression results shown in Table 4 underline the importance of preschooling in determining primary school completion. Other things equal, every 10 percentage point increase in preschool coverage translates to a nearly 4 percentage point increase in primary education completibn. This evidence reinforces other worldwide available research that children provided with a stimulating environment consistently learn more quickly and increase their likelihood of completing primary school faster than children from more deprived backgrounds." Table 4 Primary Education Completion as a Function of Lagged Preschool Coverage Variable Coefficient Preschool Coverage 7 Years Earlier .378 *** Year -.552 Primary Unit Cost .003 Delta Y -.614 Delta P 2.123 Constant -2001 R 2 .323 N 224 ** Statistical significance at 1 percent or better. 'Brown, B., Long Term Gains from Early Intervention. Boulder Co. West View Press, 1978. Halpern, R., "Effects of Early Childhood Intervention on Primary Progress in Latin America." Comparative Education Review, 30,2, 1986. Schiefelbein, E., and Simmons, J., "Determinants ofSchool Achievement: A Review of Research for Developing Countries." International Development Research Center, Ottawa, 1981. 16 Primary completion rates are positively correlated with costs, given that the incremental primary graduates are likely to be more expensive, insofar as they emerge from isolated, less densely populated, and inaccessible rural primary schools. Secondary school absorption shows a negative time trend indicating that during the last 14 years the number of primary graduates increased more than 9 times faster than the secondary first grade supply capacity28. This situation is a reflection of: (a) lack of adequate cost-effective alternatives (beyond telesecundaria) to serve primary graduates from very small and isolated primary schools; and (b) lack of economic resources hindering the continuation of the formal learning process of poor primary graduates, proven by the negative association with delta-income. The secondary absorption rate shows a negative association with the change in population, indicating that fast population growth implies more primary graduates (especially from small rural and isolated primary schools), which in turn are marginally more costly and difficult to serve. It is positively associated to management, indicating that better managed states also registered higher secondary absorption rates than the average and below average managed states. As expected, the secondary absorption rate is positively correlated with preschool coverage (the preschool gains are felt beyond primary education), and with the primary school completion rate. Also as anticipated, secondary absorption rates are negatively associated with secondary unit cost indicating that states with larger secondary freshman enrollment as a proportion of primary graduates, mostly in urban areas, had to be served with lesser number of secondary teachers, mostly less available in the market. An attempt has been made in Table 5 to explain the student-teacher ratio of the different educational levels. Beyond the already documented year trend, population increase is strongly associated with higher student-teacher ratios for all levels of education. Increases in income, characteristic of better off states (most of which have a "good" management style) are also associated with an increase in the student-teacher ratio for preschool and primary education, as they are more efficient in deploying their scarce resources (figure 5). The positive association between preschool and management could stem: (i) from the fact that in well managed states, the preschool enrollment increments were served more efficiently (with less teachers) than in not-so-well managed states, where wastage is more obvious; or (ii) * Whereas primary graduates increased at an annual rate of 4.7 percent (from 1.1 million in 1977 to 1.9 million in 1989), the secondary absorption rate increased at 0.5 percent in the same period (from 78.1 to 83.4 percent). ' The states that experienced the largest real changes in per capita income, usually the poorest states in the country, were also the states experiencing the smallest marginal changes in secondary absorption rates. There are cases were the secondary absorption rate in some of these states even decreased ( Chiapas has a 76.3 secondary absorption rate in 1977 and 75.7 in 1990). 17 better managed states experienced inward migration, especially during economic decline, and therefore were able to replace drop-outs and/or raise class sizes by following the line of least administrative resistance. There is a negative association of the primary student-teacher ratio with management, indicating that below average managed states, usually the poorest in the country, have also higher primary student-teacher ratios. One explanation seems to be the continued existence of a large number of overcrowded rural multigrade schools, characteristic of many poor, mostly rural and below-average-managed, states in Mexico. Another explanation, is that rural primary schools in poor states are, by large, understaffed. Table 5 Explaining Student-teacher Ratios Preschool Primary Secondary Independent student-teacher student-teacher student-teacher Variable ratio ratio ratio Year -.979** -1.074** .114** Delta Population .524** .734** .352** Delta Income .212** .297** .010 Management .447* -.700** -.825 Constant 1974 2164 -210 R' .640 .759 .203 N 448 448 448 ** Statistical significance at 5% V. Explaining Educational Costs Table 6 presents the results of a model attem'pting to explain the relationship between educational unit costs with managerial style, the yearly trend, and per-capita income and population changes. For all levels, the unit cost is decreasing over time, as explained before and illustrated in figure 2. The other variables, however, affect educational costs in a different way for 18 each level. Thus "good" management is associated with increased primary unit costs, yet lower secondary school costs. As explained above, the secondary education enrollment increase has been served more efficiently (Figure 3). With a national coverage of 82 percent there is still room to grow with efficiency. Thus, well managed states, characterized by higher secondary absorption and completion rates, have reduced unit costs along the growth process. Secondary unit costs are a decreasing function of enrollment and completion rates". Table 6 Explaining Educational Costs Independent Preschool Primary Secondary Variable unit cost unit cost unit cost Management -29 923** -686* Delta Income 209** -203** 469** Delta Population 118 -142** 580** Year -306** -387** -963** Student-teacher ratio 235** -83** -431** Constant 605,030 755,364 1,928,063 R 2 .355 .451 .607 N 416 416 416 * Statistical significance at 10% ** Statistical significance at 5% States that experienced an increase in income in the eighties, also have a higher pre- school and secondary school unit costs31. One explanation could be that in these states good management was able to reduce the dropout and repetition rates and thus fill " This situation is bound to change when secondary education coverage surpasses the 90 percent mark, and diminishing returns are present due to more expensive (rural) students. 3 Preschool and secondary enrollments- are still on the increase; primary enrollments has been on the decrease since the middle of the eighties. 19 in school classes with successive age group students. Another explanation is that these states experienced inward migration and classes were filled up by immigrants, a sign of good luck. But change in income has been also associated with a fall in the primary education unit cost. This situation seems to be partially explained by the inward, mostly rural, migration to the cities, predominantly characteristic in more urban and industrialized states, where: (a) there are less rural schools, and thus, less expensive clientele to serve and more students per teacher; and (b) the upper primary grades have more students, as the primary completion rates are higher. Population increase (not necessarily of the primary age cohort), partially generated by rural migration into the urban areas, has contributed to a lower primary unit cost, as one might expect, especially when the primary enrollment has been consistently decreasing in the last six school years." On the other hand, population increase is associated with a higher secondary school cost as the extra students are marginally more expensive to serve. As expected, the primary and secondary unit costs are inversely associated with the student-teacher ratio. However, there is a positive association with preschool unit costs, that is, if the student-teacher ratio decreases so does the preschool unit cost. This situation can be explained by the portfolio of conventional and non-conventional models to serve the preschoolers, especially those living in the rural and ethnic communities, where the .Government has been aiming the supply growth in the last decade. Preschool in the rural and ethnic areas of Mexico is being served primarily by non-conventional models (Cursos comunitarios and Prescolar Indigena) which are on the average 60 to 65 percent less expensive than the conventional models (implemented in urban and urban-rural areas). Total enrollment in both non-conventional modalities represents over 12 percent of the total federally sponsored preschool enrollment; the national average student teacher-ratio for the conventional model is almost 29; for the ethnic non-conventional model is 23; and for Cursos Comunitarios is 9. In short, because the growth in preschool is mainly targeted to rural and ethnic preschoolers using less expensive non-conventional models, as the student-teacher ratio decreases, so does the preschool unit cost. VI. Conclusion - There are a few conclusions stemming from this analysis. First, the economic decline of the 1980's has not affected, as one might expect, the expansion of education provision. Educational provision seems to have improved steadily despite the economic crisis. The historical in-built inefficiencies of the educational system inherited from 1921-1978 period were consistently reduced (although not eliminated) from 1978 on, with the use of a more 3 See footnote No. 11. 20 rational allocation and targeting of scarce resources based on the use of macro and microplanning techniques, programming, budgeting, evaluation, and information systems. Second, the administrative decentralization process implemented in the education sector since 1978 transferred the management and accountability to state agencies representing the Ministry of Education, making them more responsive to regional educational needs. Those agencies that benefitted from all the interventions imbedded in the so called "good management practices", explained above, performed much better than those that did not. Third, the variable most affected by the economic crisis is teachers' salaries. These have been cut severely, as illustrated in table 2, leading to lower unit costs of education.. It is noteworthy that in spite of such sharp reductions in teachers' salaries, teachers kept working in the classroom, although no measure of their work quality is available. Double shifts and moon-lighting during out-of-class schedules seem to have been popular means of compensating for the loss of real income. Furthermore, in spite of the salary decrease during the economic crisis, primary school teachers fared better than secondary teachers and minimum wage earners. Fourth, the use of cost-effective models, such as telesecundaria and non-conventional preschool models targeted to rural and ethnic student populations, seem to have been successful in serving a more expensive clientele in an efficient way. Fifth, investments made in early stimulation, especially targeted to poor urban and rural children under six years old seem to pay off, as it reduces the likelihood of primary dropout and repetition rates. Has educational quality suffered in this process? We do not know with the data in hand. We need measurements of student learning. But as judged from the more traditional physical and educational development indicators (completion and dropout rates), efficiency has improved during this period. Another lesson from the preceding analysis is that, perhaps, a country does not have to wait for an economic crisis in order to introduce cost-saving measures in education. The natural experiment in Mexico has shown that there are circumstances this can be done without loss in efficiency. 21 REFERENCES Arena, E. "Mexican Telesecundaria System: A Cost Analysis", Paper presented at the World Bank Seminar on Education Technology, Washington DC, 1988. Brown, B., Long Term Gainsfrom Early Intervention. Boulder Co. West View Press, 1978. Comisión de Salarios Mínimos, Serie 1976-1991, México, 1991. Compendio Estadístico del Gasto Educativo. DGPPP, SEP, Mexico, 1990. Compendio Estadístico por Entidad Federativa. DGPP, SEP, México, 1990. Cronología de la Secretaría de Educación Pública. Dirección General de Planeación , Programación y Presupuesto, Secretaría de Educación Pública, 1991. Estadísticas Educativas del Sistema Educativo Mexicano, Series 1976-1990. Dirección General de Planeación, Programación y Presupuesto, Secretaría de Educación Pública, México, 1990. Halpern, R., "Effects of Early Childhood Intervention on Primary Progress in Latin America." Comparative Education Review, 30,2, 1986. Prawda J., Teoría y Praxis de la Planeación Educativa en Mexico. Editorial Grijalbo, Mexico, 1985. Prawda J., Logros, Inequidades y Retos del Futuro del Sistema Educativo Mexicano, Editorial Grijalbo, Mexico, 1989. Prawda J., Análisis Sectorial del Sistema Educativo Mexicano, International Institute of Educational Planning, UNESCO, Research Report 87, Paris, 1990. Programa Educativo 1983-1988. Secretaría de Educación Pública, México, 1984. Prontuario Estadístico, Cultural y Presupuestario. Dirección General de Planeación, Programación y Presupuesto (DGPPP), Secretaría de Educación Pública (SEP), México, 1990. Rockwell, E., "Calidad de la Educación Primaria: Contexto y Proceso en el Programa de Modernización Educativa," Avance y Perspectiva, 9, Abril, 1990: 104-111. Secretaría de Educación Pública, Memorias 1976-1982, Vol I, (Políticas Educativas), Secretaría de Educación Pública, México, 1982. Schiefelbein, E., and Simmons, J., "Determinants of School Achievement: A Review of Research for Developing Countries. " International Development Research Center, Ottawa, 1981. Segundo Informe de Gobierno. Secretaría de Programación y Presupuesto, Mexico, 1990. World Tables, 1989-1990 Edition. The World Bank and the Johns Hopkins University Press, 1990. 22 Table A-1. Variable Means by State STATE ID PRES INCOMPS PRIMEFF - SECABS MANAGEMENT Aguascal 51.6 24.0 60.8 81.6 1 BajaCal 58.4 8.6 67.7 94.9 1 BajaCalS 64.7 23.0 61.4 97.41 1 Campeche 46.1 42.1 40.3 85.3 0 Coahuil 57.3 21.0 64.9 86.4 1 Colima 61.8 31.8 51.9 91.2 0 Chiapas 40.6 59.2 25.8 78.3 -1 Chihuahu 53.8 32.5 52.4 75.1 0 DistFede 50.4 3.9 73.2 107.0 0 Durango 42.0 36.8 49.1 74.1 0 Guanaju 40.0 38.4 48.1 68.0 0 Guerrero 52.3 41.2 36.6 74.7 -1 Hildalgo 51.1 39.9 44.9 77.4 -1 Jalisco 43.3 32.2 51.1 71.6 0 Mexico 31.0 18.7 64.2 74.1 0 Michoaca 38.8 37.2 37.2 73.6 -1 Morelos 58.7 17.8 67.4 94.3 1 Nayari 64.0 35.4 54.0 87.1 0 NuevLeo 53.3 19.3 70.1 91.7 1 Oaxac 56.5 38.4 34.9 66.5 -1 Puebla 39.5 34.2 45.5 75.8 -1 Queret 55.9 30.1 54.4 70.8 0 QuinRo 58.3 46.6 45.7 98.9 1 SanLuPo 48.7 35.1 47.0 87.5 0 Sinalo 36.5 34.8 53.4 88.5 0 Sonora 54.4 26.1 59.7 97.5 1 Tabasc 46.8 33.3 39.0 80.4 1 Tamaul 49.6 27.0 60.4 . 84.3 -1 Tlaxcal 58.2 20.0 61.6 87.1 1 Veracru 35.0 39.0 40.5 75.6 -1 Yucatan 53.6 40.0 35.3 89.1 0 Zacate 45.2 35.7 44.8 67.0 0 -continued 23 Table A-1 - continued STATE ID CPRE CPIM CSEC CADMPC DELTAY DELTAP Aguascal 5308 4730 12186 199 1.3 3.3 BajaCal 8992 5340 12771- 121 .5 1.6 BajaCalS 5985 6545 16438 195 -3.1 5.1 Campech 5220 4241 14015 211 -3.0 4.4 Coahuil 6061 5626 10879 154 -.5 2.2 Colima 5999 4970 11869 205 1.9 2.0 Chiapas 7200 3476 14744 182 1.0 2.0 Chihuahu 8122 4880 12117- 137 1.5 .9 DistFede 4478 6178 6402 139 -.9 1.4 Durango 9516 5261 11702 161 2.2 1.5 Guanaju 8971 3556 10168 154 2.3 1.6 Guerrero 7797 4826 11005 207 -2.0 2.1 Hildalgo 4499 4189 10108 218 -2.8 1.6 Jalisco 6779 4305 14402 121 1.0 2.1 Mexico 6493 4823 9870 91 -3.4 6.0 Michoaca 6258 3586 9977 197 .6 2.1 Morelos 4737 5559 8713 179 -2.8 1.6 Nayari 6200 4547 11418 193 -1.2 3.4 NuevLeo 7600 5341 13483 97 -.4 2.5 Oaxac 4699 3345 11793 214 3.2 .9 Puebla 5321 3857 9812 155 -.8 2.1 Queret 5399 4048 11963 212 .9 2.9 QuinRo 6900 5548 18322 242 -2.7 8.4 SanLuPo 9566 3799 10931 180 1.5 2.0 Sinalo 705 5318 11392 140 -.1 2.8 Sonora 8392 5671 11295 142 -1.0 1.8 Tabasc 8970 3408 14572 209 11.4 2.2 Tamaul 5790 5145 10907 152 -.7 1.6 Tlaxcal 648 6369 9837 176 9.6 1.8 Veracru 7675 5993 11148 136 -.5 2.3 Yucatan 4818 3924 13054 . 151 -.6 2.2 Zacate 395 3454 10492 159 1.5 .7 24 Table A-2 Variable Means by Year YEAR PRES INCOMP PRIMEF SECABS STPRE STPRIM STSEC 1977 13 43 46 76 40 44 17 1978 14 41 48 74 39 43 18 1979 15 41 49 76 38 43 18 1980 21 39. 50 80 35 41 18 1981 30 36 52 81 34 39 18 1982 38 32 53 86 33 38 18 1983 46 30 55 86 32 37 19 1984 52 25 53 85 32 36 19 1985 58 21 55 83 30 35 19 1986 64 18 55 85 30 34 19 1987 66 19 56 84 29 33 19 1988 68 16 59 84 28 32 19 1989 71 17 59 84 28 31 19 1990 72 16 60 83 27 31 19 YEAR CPRE CPRIM CSEC CADMPC 1978 10835 6172 18876 71 1979 11210 5886 16213 120 1980 7983 6173 14446 120 1981 7065 5246 12654 106 1982 7729 6871 14041 203 1983 7098 -6576 12336 177 1984 5199 4002 7361 149 1985 5540 4380 9065 267 1986 5352 4279 8568 199 1987 4282 3192 6097 133 1988 4684 3312 6692 98 1989 4667 3255 6675 101 1990 4854 3462 7381 185 25 Table A-3 SEP's State Agencies Management Quality Criterion Weighted State A B C D E F Sum Rating Aguascalientes 2 1 2 1 1 1 1.30 1 Baja California 2 1 1 1 1 1 1.15 1 Baja California Sur 2 1 1 1 1 1 1.15 1 Campeche 3 1 2 2 2 2 2.00 0 Coahuila 1 1 1 1 1 1 1.00 1 Colima 2 1 2 2 1 1 1.45 0 Chiapas 3 1 3 3 3 3 2.40 -1 Chihuahua 1 1 1 2 2 1 1.45 0 Durango 1 1 . 2 2 3 2 2.00 0 Guanajuato 2 1 2 2 2 1 1.75 0 Guerrero 2 1 2 3 3 3 2.40 -1 Hidalgo 2 1 2 3 3 3 2.40 -1 Jalisco 2 1 1 2 2 2 1.70 0 M6xico 2 1 1 1 2 2 1.55 0 Michoacdn 3 1 3 3 3 3 2.40 -1 Morelos 2 1 1 1 1 1 1.15 1 Nayarit 2 1 2 2 2 1 1.85 0 Nuevo Le6n 1 1 1 1 1 1 1.00 1 Oaxaca 3 1 3 3 3 3 2.40 -1 Puebla 3 1 2 2 3 2 2.15 -1 Quer6taro 2 1 2 2 2 2 1.35 0 Quintana Roo 3 1 1 1 1 1 1.30 1 San Luis Potosf 2 1 2 2 2 1 1.75 0 Sinaloa 3 1 2 2 2 1 1.75 0 Sonora 1 1 1 1 1 1 1.00 1 Tabasco 3 1 1 1 1 1 1.30 1 Tamaulipas 3 1 2 3 3 2 2.45 -1 Tlaxcala 3 1 1 1 1 1 1.30 1 Veracruz 3 1 2 3 3 2 2.45 -1 YucatAn 3 1 2 2 2 2 2.00 0 Zacatecas 3 1 2 2. 2 2 2.00 0 IU 1.75 a 00.49 Code: A: Management turnover: 1, if three or less managers in 14 years; 2, if between 4 and 6 mangers; 3, if 7 or more. Weight is 15 percent. B: Courses taken: 1, if at least one per year; 2, if at least one every 2 years; 3, if at least ofie every 3 years. Weight is 15 percent. C: Use of scientific planning and programming methodologv: 1, if always used; 2, if sometimes used; 3, if seldom used. Weight is 15 percent. 26 D: Quality control of education statistics: 1, if there is 2 or less percent error in the sampling when quality control is performed in the state; 2, if error is above 2 percent but below 5 percent; 3, if error is above 5 percent. Weight is 15 percent. E: Effective days of classes in the school calendar: 1, if public schools operate an average of 95 percent or more of the legal school calendar; 2, if public schools operate an average of between 90 and 95 percent of the school calendar; 3 if public schools operate an average of 89 or less percent of the legal school calendar. Weight is 30 percent, as this is considered a very important variable for student learning. F: Teachers' strikes: 1, if one or less strikes in one school year; 2, if two strikes in one school year; 3, if 3 or more strikes in a school year. Weight is 10 percent, as strikes are also dependent on variables outside the control of the SEP's state agency management. Ranking=Good (+1), if weighted average is greater or equal to 1 and less than 1.37. (1 X < A-0.75 a) Average (0), if weighted average is greater or equal to 1.37 but smaller or equal to 2.11 (j-0.75a9 X JL+0.75o) Below average (-1), if weighted average is greater than 2.11 (I+0.75c< x) Source: Elaborated by the General Directorate of Planning, Programming and Budgeting of the Mexican Ministry of Education based on Cronologia Anual de la SEP. - Continued from back Page No. 16 "What do we think about Health Care Finance in Latin America and the Caribbean?" by Philip Musgrove, September 1991 No. 17 "Population Growth, Externalities and Poverty" by Nancy Birdsall and Charles Griffin, September 1991 No. 18 "Wage Trends in Latin America" by Alejandra Cox Edwards, September 1991 No. 19 "Investment in Science Research and Training: The Case of Brazil and Implications for Other Countries" by Laurence Wolff, with contributions from George Psacharopoulos, Aron Kuppermann, Charles Blitzer, Geoffrey Shepherd, Carlos Primo Braga and Alcyone Saliba, September 1991 No. 20 "Prenatal and Perinatal Health Care: A Diagnostic Instrument" by Francisco Mardones, September 1991 No. 21 "Maternal Anthropometry in Prenatal Care: A New Maternal Weight Gain Chart" by Pedro Rosso, September 1991 No. 22 "Poverty and Inequality in Latin America and the Caribbean During the 70s and 80s: An Overview of the Evidence" by Dominique van de Walle, September 1991 No. 23 "Social Indicators in Latin America and the Caribbean: A Compilation of Statistics from 1970 to the Present" by George Psacharopoulos and Bill Wood, October 1991 No. 24 "ICETEX - A Student Loan Success Story in Colombia" by Samuel Carlson, October 1991 No. 25 "Educational Development and Costing in Mexico, 1977-1990: A Cross-State Time- Series Analysis" by Juan Prawda and George Psacharopoulos, November 1991 Views from LATHR No. 0 "The Magnitude of Poverty in Latin America in the 1980s" September, 1990 No. 1 "An Ounce of Prevention is Worth How Much Cure? Thinking about the Allocation of Health Care Spending" by Philip Musgrove, September 1990. No. 2 "Decentralization and Educational Bureaucracies" by Juan Prawda, November, 1990 No. 3 "What Should Social Funds Finance?: Portfolio Mix, Targeting, and Efficiency Criteria" by Margaret E. Grosh, December 1990 No. 4 "Financial Balance in Chile: The ISAPRES (Instituciones de Salud Previsional) Health Care System and the Public Sector" by'Philip Musgrove, January, 1991 No. 5 "Population, Health and Nutrition Issues in the Latin American and Caribbean Region and the Agenda for the 90's" by Oscar Echeverri, January, 1991 No. 6 "Population and Family Planning in the 1990's: Reconciling Macro and Micro Issues" by Bruce D. Carlson, February, 1991 No. 7 "The Feasibility of Student Loans in Latin America: A Simulation" by Samuel Carlson and Guozhong Xie, March, 1991 No. 8 "Transforming the Vicious Circle - The Costs and Savings of School Inefficiency in Mexico" by Samuel Carlson, April 1991 No. 9 "Colombia's "Escuela Nueva": An Education Innovation" by Eduardo Velez, May 1991 No. 10 "Health Technology Development and Assessment: Do LAC Countries Have a Choice?" by Oscar Echeverri, June 1991 No. 11 "The Recurrent Cost Factor in the PHR Sector" by Jacob van Lutsenburg Maas, July 1991 No. 12 "The Burden of Death at Different Ages: Assumptions, Parameters and Values" by Phillip Musgrove, August 1991 No. 13 "Government Expenditure on Social Sectors in Latin America and the Caribbean: Statistical Trends" by Hongyu Yang, August 1991 No. 14 "From Manpower Planning to Labor Market Analysis" by George Psacharopoulos, September 1991 No. 15 "An Update on Cholera in the Americas" by Francisco Mardones, August 1991 - Continued on inside Page