Mexico's preschools : coverage, equity and impact

16 I 82'2 LCSHD Paper Series Department of Human Development Mexico's Preschools: Coverage, Equity and Impact John H. Y. Edwards Xiaoyan Liang May 1998 The World Bank Latin America and Caribbean Regional Office  LCSHD Paper Series No. I Van der Gaag and Winkler, Children of the Poor in Latin America and thle Caribbean No. 2 Schneidman, Targeting At-Risk Youth: Rationales, Approaclhes to Service Delivery and Monitoring and Evaluation Issues No. 3 Harrell, Evaluacion de los Programas para Nifiosy Jovenes Vulnerables No. 4 Potashnik, Computers in the Schtools: Clile's Learning Network No. 5 Barker and Fontes, Review and Analysis of International Experience witlh Programs Targeted on At-Risk Youtht No. 6 Lewis, Measuring Public Hospital Costs: Empirical Evidencefrom the Dominican Republic No. 7 Edwards, Bruce and Parandekar, Primary Education Efficiency in Honduras: Whiat Remains to be Done? No. 8 Winkler, Descentralizacion de la Educaci6n: Participacion en el Afanejo de las Escuelas al Nivel Local No. 9 Meza, Descentralizacion Educativa, Organizacion y Manejo de las Escuelas al Nivel Local: EDUCO, el Caso de El Salvador No. 10 Espinola, Descenrtralizacidn Educativa, Organizaci6n y Man ejo de las Escuelas al Nivel Local: El Caso de Chiile No. 11 Guedes, Lobo, Walker and Amaral, Gestion Descentralizada de la Educacion en el Estado de Minas Gerais, Brasil No. 12 Cominetti and Ruiz, Evolucion del Gasto Puiblico Social en America Latin a: 1980 - 1995 No. 13 Bedi and. Edwards, The Impact of School Quality on thle Level and Distribution of Earnings: Evidencefrom Honduras No. 14 Duthilleu1, Do Parents Matter? The Role of Parental Practices on Fourth Graders' Reading Comprehension Achievement in Montevideo Public Schools No. 15 Villegas-Reimers, The Preparation of Teachers in Latin America: Chlallenges and Trends No. 16 Edwards and Liang, Mevico's Preschools: Coverage, Equity and Impact No. 17 Soares, The Financing of Education in Brazil: Witli Special Reference to the North, A\ortlieast and Center- West Regions No. 18 Sahrni, Equity and Quality ini Private Education: The Haitian Paradox  Human Development Department LCSHD Paper Series No. 16 Mexico's Preschools: Coverage, Equity and Impact John H. Y. Edwards Xiaoyan Liang May 1998 Papers prepared in this series are not formal publications of the World Bank. They present preliminary and unpolished results of country analysis or research that is circulated to encourage discussion and comment; any citation and use of this paper should take account of its provisional character. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and should not be attributed in any manner to the World Bank, its affiliated organization members of its Board of Executive Directors or the countries they represent. The World Bank Latin America and the Caribbean Regional Office  ii  TABLE OF CONTENTS EXECUTIVE SUMMARY ....................v I. INTRODUCTION .1 II. THE ROLE OF PRESCHOOL EDUCATION IN CHILD DEVELOPMENT .4 III. DESCRIPTION OF THE MEXICAN PRESCHOOL SYSTEM . 6 IV. PRESCHOOL ENROLLMENT AND COVERAGE .13 V. IMPACT 1: PEDAGOGICAL EFFECTS .19 VI. IMPACT 2: EFFECT ON MATERNAL EMPLOYMENT .26 VII. IMPACT 3: EQUITY .28 VIII. THE COST OF MEXICAN PRESCHOOL PROGRAMS ................................ 31 IX. SUMMARY OF FINDINGS .................................................. 34 REFERENCES .................................................. 39 ANNEXES .................................................. 41  AX  EXECUTIVE SUMMARY Summary. The purpose of this document is to describe the administrative and pedagogical structure of the Mexican preschool system and to evaluate its effect on higher levels of schooling, female labor force participation and equity. After a search through the professional literature for the roles that preschool can play, the Mexican system is described. Existing data sources are then analyzed to depict it numerically and search for evidence of its impact. Findings indicate that coverage is high, but that the system is quite costly and appears to have very little impact on school readiness, on mothers' decision to work, or on social equity. The study ends on a cautionary note, pointing out shortcomings in the existing data sources. Recommendations include rnodifications in the preschool system and ways of developing better evaluation data. Setting. A recent report notes that in 1992, 25.4 percent of Mexico's 84 million people were living in poverty. The past two decades have seen tumultuous economic evolution. Change was accented --some would say accelerated-- by two major external developments: the end of the oil boom and the fonnation of the North American Free Trade Association. A precipitous fall in per-capita income during the 1980's was followed by growth in the early 90's and a second near-collapse in 1994. Poverty is omnipresent but unevenly distributed across regions and ethnicities. Rates tend to be higher in the south, especially in rural areas where there are concentrations of indigenous peoples. The current climate in Mexico favors fundamental changes that will lead to stable and sustained growth. Government Strategy. Educational development is one of the keystones of the government strategy for long term poverty-alleviating growth. Areas of low educational achievement and attainment coincide with the foci of poverty. The most pressing problems concern internal efficiency and educational attainment. In 1990 about 56 students were completing primary school for every 100 who had enrolled six years earlier. This "terminal efficiency rate" ranged from 27 percent in the southern state of Chiapas to over 80 percent in Mexico City. In 1994 about 98 percent of Mexican children age 9 were enrolled in primary schools, yet only 85 percent of 15 to 19-year-olds had completed grade 6 and only 58 percent of 20 to 24-year olds had completed high school. Problems of early childhood (ages 0 to 3) and primary school (mainly ages 6 to 12) have been fairly well scrutinized. They have received a considerable amount of investment, including World Bank funded interventions such as the Initial Education Project (PRODEI) and Basic Education Projects I and II (PARE and PAREB). Issues. The Mexican preschool sector has received much less focused attention. The major issues facing it are (i) the impact on primary school performance, (ii) the cost per year of preschool attendance relative to other methods of improving primary efficiency and grade attainment, (iii) the link between preschool attendance and maternal labor force participation and 1 Staff Appraisal Report: Mexico Program of Essential Social Services (PROSSE). World Bank & Inter-American Development Bank, May 1995.  vi -hence-the potential short term role of preschools in poverty alleviation, (iv) the equity implications of differential access to preschools. Findings. The findings in this document should be viewed with circumspection, even as preliminary, until Mexico develops the kind of precision data that are best suited for tailoring policies around these very relevant issues. Readers should also bear in mind that the evaluation methodology employed herein considers only the measurable aspects of preschools ignoring the many unquantifiable benefits that are sometimes attributed to center-based early education. Setting these caveats aside, the existing data do not paint an optimistic picture. The major findings are that (i) preschool coverage for 5-year olds is high, but that (ii) coverage is inequitable, (iii) unit costs are high, (iv) there is no convincing evidence that attending a Mexican preschool will enhance primary school performance and that (v) preschool attendance is not linked to maternal labor force participation. Finally, (vi) there is evidence that the preschool system is drifting towards provision of free childcare on demand. If it continues unit costs will increase rapidly in the near future, generating a large fiscal impact and placing in peril quality improvements that are much needed elsewhere in the sector. Coverage. Mexican preschools concentrate on children of age 5. For this age group, public preschool coverage averages about 63% nationally. Private preschools enroll another 10%. This places Mexico at about the same level as Austria and New Zealand. It compares favorably with most of the other countries in region and with coverage in some higher income OECD countries like Sweden, Spain and Portugal. Equity. There is gender equity in access: about the same proportion of girls and boys are enrolled in all 23 Mexican states. However, the data indicate that coverage is inequitable across income groups. It is not so much that enrollment constitutes privilege, as that non-enrollment is linked to under-privilege. Preschool attendance is the norm for the top two-thirds of the income distribution. Household income and parental education are substantially lower among the families of children that are not enrolled. Cost. Per-student cost in Mexico's preschool program is high in relation to the cost of primary schooling. In 1994 current expenditures for one student-year in the General Preschool program exceeded the cost of a year of primary school by 12 percent, yet a preschool day lasts less than half as long as a day of primary school. This means that in order to be economically justified solely on the basis impact on primary efficiency, each student-year of preschool would have to improve primary school efficiency by more than one year. Impacts of this magnitude have never been reported in the literature. Impact on Primary School Efficiency. Existing data show that attending a Mexican preschool has little or no impact on primary school efficiency. Impact on entry-level ability scores and on repetition rates is small and likely to vanish once better data can simultaneously control for the relatively privileged family background of preschool graduates.  vii Maternal Labor Force Participation. The predicted propensity to enroll a 5-year old child in preschool is found to have no statistically significant impact on the likelihood that its mother will seek employment. The reason for this is probably that most preschool programs last for only 3 hours. Unless academic preschoois provide pre and aftercare options, working mothers are likely to make alternative, full-day childcare arrangements. The uncertain fiscal impact of free childeare. The typical preschool is designed to operate two 3-hour shifts per day. In the face of growing maternal demand for full-day childcare, some preschools are opting to provide it in what are known as "mixed modality" programs. The mixed modality operates a full-day shift that subsumes the "regular" preschool curriculum during three morning hours. This means that only one shift per day is possible, effectively doubling costs. The constitution mandates that all publicly provided education should be free. Mixed modality schools appear to interpret this constitutional mandate as including extended childcare services. Parents are charged only the nominal cost of providing lunch. The mixed modality is relatively new. Official statistics indicate neither how many centers operate in this manner nor how many students they enroll. What is clear is that continued development in this direction could eventually double the per-student cost of preschool, placing a severe strain on the Mexican educational budget.  I. INTRODUCTION This paper examines the role of Mexican school-based pre-primary programs in a three- dimensional context formed by pedagogical function, family need for substitute childcare and social concern for equity. Issues Facing Preschool Education The major issues facing the Mexican preschool system are: * its impact on primary school performance; * the cost per year of preschool attendance relative to other methods of improving primary efficiency and grade attainment; * the link between preschool attendance and maternal labor force participation and -hence-the potential short term role of preschools in poverty alleviation; and * the equity implications of differential preschool coverage patterns. In understanding educational systems, the study of preschools is distinguished from the study of primary schools by the fact that preschooi enrollment is voluntary. Parents decide whether or not to send their children. The analytical framework adopted here allows for preschool attendance to be either investment or consumption behavior. When viewed as an investment, it can have both academic and non-academic payoffs. As a consumer good, preschool enrollment can benefit the parents as well as the children. In this document the methodology that is employed to understand the function of preschools consists of first describing the constraints on choice behavior imposed by the institutional and family economic context and then examining the impact of enrollment. Preferences revealed in observed enrollment patterns then divulge the implicit relative importance of each role within a given national context. The constitutive pedagogical issue for educational policy formulation is the impact of center-based preschool programs on the depth and duration of learning. Does the preschool experience lead to better grades and more years of school attainment? If so, is the magnitude of this impact greater than what would obtain if the same resources were spent directly on primary schools? Preschool attendance may have non-academic benefits as well. Some of these are easy to quantify and measure and others are not. An example at the first extreme is the child-care role played by preschool centers. Childcare is an immediate and easily measured benefit that is independent of any impact on future academic achievement. It can be of considerable importance too. If enrollment in a center enables the mother to work, this will boost household income. Her earnings can be measured and compared against the social cost of this alternative childcare. The issues of childcare, per se, and the future academic impact of childcare are linked only to the extent that this academic impact reduces maternal misgivings by enough to  2 significantly affect her labor force participation decision. At the opposite, hard-to-measure extreme are the dividends suggested by developmental psychologists and educators. They have argued that pre-school experiences build social skills and .a sense of self confidence along with other desirable personal attributes. These individual improvements may eventually translate into effects such as a lower propensity to commit criminal acts, but gains of this type are not easy to measure. It is conceivable that some parents view the enrollment of their child in preschool as an act of consumption. Preschools are fum. They provide toys, playmates and organized activities, which the children consume and enjoy. Preschools also let the mother consume a few hours of leisure each day. Whatever the reasons that enrollment is chosen, matters of coverage and access are rife with equity implications. For instance, if preschool attendance enhances learning and lifetime school attainment, theA .overage will bear on the future distribution of income through the education/earnings nexus. If attendance serves to facilitate paid maternal labor, then it will bear on the current distribution of income as well. Equity issues become especially relevant when -- as in the Mexican case studied here--the state is legally required to provide preschool services free of charge. Then, even if preschool school is pure consumption, who enrolls becomes an equity issue because costs are paid out of general tax revenues. Background In Mexico, primary school officially starts at age 6. However, actual entry age varies considerably and children enter primary school with very diverse exposures to organized education and learning activities. Before age 4, some will have been enrolled in an initial education program where mothers received coaching from social promoters on parenting skills. The increased awareness of early education promoted through the Initial Education Project (PRODEI) may have prompted their parents to enroll them into a preschool when they turned age 4 and keep them there until age 6. So, while some children enter grade 1 with as much as six years of exposure to the educational system, others may have never so much as joined a playgroup. The World Bank's involvement in Mexican education development starts at birth through its funding of PRODEI, the initial education program that covers the ages from 0 to 3. As a result of a study on initial education status (World Bank 1990), in 1992, the World Bank issued an IDA credit of US$80 million to help the government expand its initial education coverage. Under PRODEI, home visits are used as a channel to impart proper parenting knowledge and skills. It is estimated that by the time it is completed in 1999 PRODEI will have benefited nearly 1.8 million low-income children below the age of 4. In addition to this early childhood program, the World Bank has been heavily involved in the Mexican primary education sub-sector. Two World Bank projects have already targeted children in the primary school age range (a total of US$662 million) and a third basic education  3 project is currently being prepared. This new project will range all 'ihe way from preschooi to grade 9. Finally, a higher education student loan program is also now in place. Analysis of preschools will serve two purposes. First of all, the World Bank approach to the education sector has been to have analysis accompany each of the sub-sector projects. This approach has left an infornation gap: we know very little about children after initial education but before primary enrollment, i.e. children of ages of 3 to 5. Also, given that the third education project discussions include preschool, this study will serve to help focus attention on this relatively neglected age group. The Mexico Initial Education Strategy (World Bank 1992), advanced two arguments for an integrated early child development approach (age 0 to 3). First, it reasoned that a child's physical cognitive, social and emotional development is positively related to proper care and early stimulation. Second, it argued that investments in education, health and nutrition early in life can bring a high return by increasing a child's productivity in later years. We complement the initial education strategy paper by focusing specifically on the various dimensions of impact resulting from education of somewhat older children (ages 3 to 5). We step back to point out several messages from the child development literature that are often neglected by its strongest advocates. Summary and Outline The main issues therefore are coverage, impact, cost and equity. This paper examines each issue in turn in a Mexican context formed as much by the legal environment, administrative structure and curricului-n design of the school system as by the distribution of education and income across families. Wherever possible characteristics of the Mexican case are contrasted to preschool systems elsewhere in the Latin American region and OECD. After a brief review of the professional literature, the Mexican system is outlined and existing sources are analyzed to describe coverage and search for evidence of impact. Findings indicate that coverage is high, but that the system is quite costly and has very little impact on school readiness, on mothers' decision to work, or on improving social equity. The study ends on a cautionary note, pointing out shortcomings in the existing data sources. Recommendations include modifications in the preschool system and ways of developing better evaluation data. The paper has 9 sections. Section 2, discusses the role of preschool, using evidence from the professional literature. Section 3 describes the Mexican system, covering legal, curriculum and administrative aspects. Section 4 estimates enrollment and coverage. Sections 5, 6 and 7 examine impact on primary education efficiency, maternal labor force participation and equity, respectively. Section 8 reports unit costs and compares them to other school levels. The document ends with a summary and discussion of the findings in Section 9.  4 II. THE ROLE OF PRESCHOOL EDUCATION IN CHILD DEVELOPMENT Many terms have been used to describe the education or care provided to children before they officially enter primary school. They include "day care", "preschool", "initial education" and "early child development". These terms overlap and the programs they describe may also differ in their targeted beneficiary as well as in their approaches to child development as reflected by implicit philosophy or explicit curriculum. For example, Young (1995) is very careful to define early child development as services provided to children from 0-8 years old integrating nutrition, health, and education. At the opposite extreme, UNESCO's effort to develop international data on preschool and pre-primary programs does very little in the way of distinguishing between different types of program content. In this paper, the term "preschool" is used in a very restricted sense. It is defined to mean school-based education provided to children between the ages of 3 and 5. It is important to keep in mind the analysis that follows concerns only the costs and benefits of the formal type of early education that is nrovided in an organized setting with a pronounced curriculum or philosophy and which, for the most part, employs professionally trained teachers. It does not assess any other type of services that are provided to young children, such as integrated health and nutrition services. Most of the research on the impact of early education has been carried out within the developed nations of the North America and Europe. Early in the 1980s, The Consortium for Longitudinal Studies found that, on average: * Preschool programs increased individual scores on standard intelligence tests. These increases remained statistically significant for three to four years after the preschool experience. But the gains tend to fade after third grade. * Arithmetic and reading achievement scores of children who went to preschool were higher than those of children who had not participated in preschool programs. e Preschool graduates had higher self-esteem, valued achievement, and had higher occupational aspirations than the other children. * Children who went to preschool were less likely to be placed in special education or remedial classes than those who did not. They were also more likely to meet standard school requirements and to graduate from high school. (Lazar, 1983) The findings from The Consortium for Longitudinal Studies were later reinforced by the independent findings of Weikart and his associates, published in 1984, on the long-term effects of the Perry Preschool Project, which used the High/Scope curriculum model (Berrueta-Clement, Scheweinhard, Barnett, Epstein, & Weikart, 1984). However, both studies were implemented under ideal laboratory-like conditions. They had better-trained staff than are available to the typical preschool program (Currie and Thomas, 1995). And, in the case of the Perry Preschool Project the sample was very small: 58 treatments and 65 "controls".  5 Evaluations of more typical early intervention programns have, in fact, found less dramatic or even no difference between those who attended preschool and those who did not (Evans 1985). Head Start is a U.S. program is aimed at improving the school-readiness, health and social skills of children from disadvantaged backgrounds. It is a case in point. This one of the largest, best targeted, longest running and best-studied experiments in the use of preschool programs to affect outcomes later in life. Yet, in a critical review of early intervention programs by Haskins (1989) concludes that, although there is strong support for the immediate impact of typical Head Start programs on children's intellectual performance, there is "virtually no evidence" of long- term impact on specific measures of life success such as those found for participants in the Perry Preschool Project. Furthermore, in their review of the Head Start literature, Currie and Thomas (1995) write that "Policymakers and the general public appear to believe that the benefits of Head Start are well known and well documented. However. a careful reading of the literature reveals that credible studies that demonstrate lasting effects of Head Start are limited. The studies that do exist are typically restricted to small geographic regions and specific racial groups." (p. 341) If this can be said of the evidence on Head Start impact, it is certainly true of the preschool-impact literature generally. Most preschool impact evaluations are rife with methodological problems stemming from simultaneity and selection bias as well as from subject- attrition. After the most rigorous analysis of Head Start to date Currie and Thomas find that the program has no lasting impact on test score performance or grade repetition probabilities of African Americans. There does appear to be some evidence of longer-lasting impact on poor European American participants, but the authors also point out that selection into this supply- constrained program is playing an unknown role. Differences in program quality and in post Head Start reinforcement in the home and in school may be as important to impact durability as program participation per se. Scarcest is the evidence from developing countries. In addition to the problems just outlined, this specialized strand of the literature faces additional shortcomings that often include low data quality and poor research methods. Yet, existing evidence seems to point to the increase of school readiness, increased enrollment, reduced repetition and dropout rates and higher achievement in primary schools (see review in World Bank 1992). The evidence of impact seems to be particularly strong when interventions are targeted at disadvantaged segments of the population. The child development research boil down to a handful of messages. * There is some evidence of early education impact on children's performance in primary schools. Children from disadvantaged families, in particular, may benefit from such programs.  6 * However it is not meaningful to simply ask whether preschool education is good or bad. Early education is not inherently or inevitably beneficial or harmful. A great variety of environments for children are united under a general rubric and impact on children's development is linked to the variation in service quality. * The most important message is that school, home and out-of-home settings are confederated and mutually influential. Child development is multiply determined: by sources within the child, such as temperament, neurological integrity, and impairment; by factors in the child's immediate environment, such as quality of relationships and interactions with parents and quality of out-of-home care; and by factors in the child's larger social environment, including the immediate neighborhood and the broader culture. These factors do not operate separately but interact in a complex fashion to influence developmental outcomes. Preschool education therefore must be viewed as one of many sources of influence on children's development and one that interacts in complex ways with numerous others. It must be kept in mind that public lducation policy controls only a subset of the factors that influence child development and scholastic achievement. The next section describes the Mexican preschool system and subsequent sections evaluate its impact in a broader socio-economic context. III. DESCRIPTION OF MEXICAN PRESCHOOL SYSTEM 2 A 3-year program covering children of ages 3, 4 and 5 constitutes the Mexican preschool system. It is not a pre-requisite for primary school, nor does law require attendance. The official school day lasts for 3 hours and the school year coincides with the primary school year. Legal Framework The legal status of Mexican preschool education is somewhat equivocal in that the state is obligated to provide this service free of charge, but enrollment is not obligatory. Article 3 of the Mexican constitution states that, * "Every individual has the right to receive an education. The State --Federation, states and municipalities-- will impart preschool, primary, and secondary education. Primary and secondary education are obligatory" (emphasis added). Article 31 of the constitution holds parents legally responsible for obligatory schooling, again implying that preschool is not a legal obligation: 2An excellent summary of the Mexican preschool system can be found in B. Garcia and M. Gonzalez "Estructura Actual del Sistema Educativo y Modalidades de educacion preescolar". For a history of evolution of the system, see V. Patrowitsh and B. Garcia "Sintesis Hist6rica de la educaci6n Preescolar en Mexico". Both articles are in B. Garcia Cabrero (Ed.) La Educaci6n Preescolar en Mexico: Un Estudio Diagn6stico, CONAFE, 1996.  7 "... it is the obligation of the Mexicans to make their children or tutors attendpublic or private schools for the purpose of obtaining primary and secondary education." Chapter IV, article 37 of the General Law of Education makes it clear that preschool is not an academic obligation either: . "Preschool education is not a prerequisite for primary school." But the General Law of Education 6/12/93, (article 3) reiterates children's entitlement to the service, * "The state is obligated to provide educational services so that the entire population is able to pursue preschool, primary, and secondary education". Furthermore, if parents decide to enroll their children, the state is legally obligated to pay for the service. According to Article 3, paragraph 4 of the Constitution * "All education which the state imparts shall be free." The word "all" has been interpreted to include preschool. In summary, preschool is not obligatory and it is also not a prerequisite for admission to grade school. The state is required to provide free preschool services, but parents are not legally obligated to send their children. This implies that it is a one-way obligation. Preschools have to attract students in a way that is not necessary at the primary level. The Curriculum Curriculum Design. Nearly everyone in the official preschool establishment, from the director of national preschool education to individual teacher's assistants, cite the French psychologist Jean Piaget as the origin of the system's philosophy toward preschool education. This view of children's development contrasts with the behaviorist view of the child as a passive receptor of knowledge, which is acquired through imitation and practice and is internalized through processes of reward and punishment (Kamii, 1985). For Piaget, true learning is not something handed down by the teacher, but something that is constructed by the child. Constructivists view learning as a process of spontaneous invention and discovery. Learning is centered through play. Accordingly, the teacher should not try to impose knowledge on the child, but rather should find materials that will be interesting and challenging and then permit the child to solve problems on its own (Crain 1992). In public preschools the constructivist view of development translates into physically organizing classrooms into different activity corners or social settings, such as a kitchen, reading corner or "office" supplied with context-giving toys and reading materials. It is also reflected in  8 the official curriculum book that emphasizes leaming through playing and activities rather through formal instructions and drills. However, in practice, fairly large class sizes (around 20 children per classroom), compromise the ability of teachers to guide or scaffold the play of individual children. Officially Recognized Centers. All preschools registered with the SEP are bound by the same curriculum guidelines. Officially recognized centers must follow the standard curriculum. Information on enrollment, age and basic schooi characteristics are well maintained and documented. Textbooks are developed and distributed centrally by the Direcci6n de Educacion Preescolar. Four SEP offices are involved in curriculum development and program design. 1) Direcci6n de Educaci6n Inicial 2) Direcci6n de Educaci6n Preescolar 3) CONAFE 4) Direcci6n General de Educaci6n Indigena Private, Unofficial Centers. Unregistered centers that enroll children in this age group can introduce curriculum variations at will. Mexican administrators are very aware that they are responsible for a preschool system that exists in a legal limbo. An immediate outgrowth of this intermediate legal status is that many private preschools do not follow the official curriculum or satisfy licensing requirements. Official recognition of preschools --which presumably conveys information on quality to parents3-- is therefore used to entice private schools to follow official guidelines, rather than as a coercive mechanism. It is presumed that those schools that forgo official recognition do so because they are below SEP quality standards, blurring the distinction between daycare and preschool. Another problem in evaluating most private providers is that those centers that do not follow the official guidelines are not in SEP databases. There is no systematically gathered information on them. SEP officials believe that informal coverage is large and growing, but almost nothing is known about coverage, size, location or quality of most private centers. An informal survey recently conducted by the Direcci6n General Preescolar estimates' that unregistered preschools in the Mexican Capital outnumber official ones and account approximately 40% of total enrollment. Target Age Mexican law and SEP regulations officially recognize 3 school-based preschool grades (henceforth K1, K2 and K13, respectively). Since promotion is automatic, children are assigned to grades according to their age. The design correspondence is: Age 3 Grade KI Age 4 Grade K2 Age5 GradeK3 3See Edwards, Fuller and Liang (1996)  9 The last column of Table 1 reveals that despite this design, in practice the Mexican school-based preschool system mainly serves older children in grade K3. This topic is discussed further below. Administrative Structure Until recently the federal government ran nearly all the public schools in the country. Then the 1992 decentralization delegated most direct responsibility for school administration to the 31 states and the F.ederal District. While the scope of central government administration is much reduced, SEP continrues to run some schools. DIF and a few other organizations run a handful of preschools as part of compensatory family programs. In addition, each secretariat of the central government runs its own preschools under the CENDI program (Centro de Desarrollo Infantil). SEP recognizes 23 program types. Table I illustrates the administrative complexity of this system. The majority of students (52%) are enrolled in preschools that were under federal jurisdiction until 1992 but.are now run by state educational agencies ("Jardin de Ninos Federal, Transferido"). The second largest category, with 17% of registered enrollment, is composed of other "general" schools that were administered by the states even before the 1992 reforms. Indigenous schools are in third place with just over 8% of enrollments. Some programs operate in only one state, others in all 32 states. Some of the programs are extremely specialized, perhaps experimental, such as the CENDI "Reclusi6n Federal" which serves the children of female prisoners and has only one school with 25 students. Mexican law requires that any establishment --public or private-- which employs 50 or more women must establish a daycare center. Table 1 indicates that, in practice, very few private establishments comply with this regulation. Only 425 private CENDI are registered with SEP, enrolling 11,020 students. The government does appear to comply. Table 1 ("CENDI Otra Secretaria " and "CENDI Estatal") adds credence to the claim that each of the Federal and State Secretariats sets up preschool CENDI's in all of their major offices throughout the country. Classroom Crowding Some information on the quality of each type of Mexican program is also provided in Table 1. Classroom crowding measures vary substantially. The students per class measure ranges from a low of 5.8 to a high of 30.5 and the student/teacher ratio from about 7.2 to 29.8.  TABLE 1: Basic Statistics of the Mexican Preschool System PROGRAM NAME States w/ | Schools | Total | Teachers | Classrooms Students |Students per| Percent 1Percent Age I Program | Enrollment. I I per Class Teach. | Girls | 5 Jardin De Nifhos Federal 1 1,127 200,061 6,710 7,611 30.5 29.8 49.3% 50.0% Jardin De Nifhos Otra Sria. 4 306 6,051 317 278 6.8 19.1 49.6% 45.1% Preescolar Jardiii De Nifios Fed. (T) 31 24,849 1,616,356 62,424 68,618 20.3 25.9 49.7% 58.7% Preescolar Gen. Prom. Tec. Fed. (T) 18 3,272 52,387 3,503 2,867 6.4 15.0 49.8% 48.0% Preescolar Gein. Prom. Bach. Fed. (T) 15 1,517 29,140 1,698 1,214 7,5 17.2 50.6% 47.9% Jardin De Ninios Asist. Fed. (T) 1 30 4,322 161 161 27.0 26.8 48.9% 50.9% Preescolar General Estatal 23 6,580 525,613 18,825 20,316 23.2 27.9 49.8% 53.9% Preescolar General Particular 32 4,377 219,897 12,621 16,539 16.6 17.4 49.8% 44.4% Preescolar Indigenia Federal (T) 23 7,339 255,247 11,469 11,927 13.7 22.3 49.9% 53.6% Preescolar Indigena Estatal 2 400 6,724 453 422 5.8 14.8 49.6% 34.3% Preescolar Conafe Federal 32 7,401 80,976 7,401 5,070 10.9 10.9 41.5% 47.6% Preescolar Agrie. Migrante Fed. 5 97 1,325 97 90 13.7 13.7 48.1% 54.3% Preescolar Conafe Infantil 5 31 341 31 23 11.0 11.0 49.0% 63.0% Preescolar Conafe Indigena Fed. 5 250 1,808 250 148 7.2 7.2 49.9% 57.5% Preescolar Cendi Sep Federal 1 34 2,091 83 65 20.3 25.2 50.3% 40.3% Preescolar Cendi , Reclusion Fed. I 1 25 1 1 8.3 25.0 44.0% 32.0% Preescolar Cendi Otra Secretaria 27 416 18,913 912 1,066 15.4 20.7 48.8% 42.4% Preescolar Cendi Sep Federal (T) 30 117 7,981 355 609 24.5 22.5 48.1% 43.3% Preescolar Cendi Estatal 28 187 11,113 484 902 20.8 23.0 48.0% 39.9% Preescolar Cendi Autonomo 11 20 1,670 86 101 24.2 19.4 47.2% 36.8% Preescolar Cendi Particular 25 425 11,020 773 1,129 10.5 14.3 48.6% 41.3% Preescolar Dif Estatal 8 1,375 64,946 2,618 2,485 16.1 24.8 49.2% 59.6% Preescolar Dif Particular 1 153 6,076 277 293 21.8 21.9 49.3% 55.6% TOTAL .... N.A. 60,304 3,124,083 131,549 141,935 18.4 . 23.7 49.5% 55.0% (7): Indicates Federal program transferred to the States in 1992.  Most of the CENDI operate for a full day and not in shifts. Table 1 data therefore imply that the typical CENDI has a class size of 20. Most of the CONAFE classes are very small, averaging about_11 students. In contrast, the "Jardin de Ninos Federar' shows crowding measures that are about 50% higher. Though this program operates only in the Federal District, enrollment surpasses 200,000. Such schools also tend to be much larger, enrolling an average of 178 students, compared to the national average of 52 students per preschool. The much higher student per class and student per teacher measures suggest a moderately serious crowding problem in a large number of schools. It also suggests that such problem schools are of a particular type and can be located fairly easily with existing SEP statistics. The programs that appear to be most crowded fall under the "gereral preschool" rubric and account for the majority of enrollment. Unfortunately, SEP daca do not identify which schools operate two half-day shifts (the norm) and which ones care for children for a full day under the so-called "mixed- modality". This important omission makes it difficult to interpret the more extreme crowding measures. Pedagogical Structure On the surface, the Mexican preschool school system appears to be administratively complex, but the educational establishment views the preschool system from a pedagogical perspective. From this viewpoint it is a much more concise system 4 that consists of only 3 types * General Preschool; * Indigenous Preschool; and * "Community", "Rural" or "CONAFE" preschools. Furthermore, since the same curriculum and objectives bind all 3 programs they differ only in who and how these are delivered to students. General Preschools cater to children that live in small to large urban areas and speak Spanish as their native tongue. Most of them live in urban settings making it is easier for General Preschools to attract and retain college-educated teachers. Since 1989, General Preschool teachers are required to be normal school graduates with a specialty in preschool education. The General Preschool norm places students in school for 3 hours per day, allowing schools to operate in 2 shifts. The morning shift lasts from 9 A.M. until 12 noon and the afternoon shift from 1 to 4 P.M. This schedule is clearly not designed to substitute for parental childcare. With the disappearance of extended --and some would argue of even nuclear-- families and increased female employment, there appears to be a growing demand for full day services. 4For instance, see B. Garcia Cabrero (1996) op. Cit.  12 In some cases parents have access to a "mixed" modality option, where children are enrolled for an extended program that lasts from about 7:00 A.M. until 5 P.M. This modality receives the name "mixed" because 3 hours consist of the General Preschool program while the rest of the day is standard daycare. Parents are required to contribute the (nominal) amount that it costs to feed lunch, but otherwise they pay no tuition or fees for extended care. The mixed modality would seem to be especially attractive to working mothers. In fact, access to such schools could provide the incentive that mothers in low- income families need to enter the labor market (see below). The legal and administrative standing of mixed modality schools is less clear than it is for the regular General Preschools. The constitution obligates the government to provide 3 hours of free "academic" preschool to whomever should request it. General Preschools satisfy this requirement exactly. The constitution does not obligate the government to extend care beyond that time, much less to provide the care for free. It is clear that the mixed modality arose in response to family demand for it, but it is not clear how the government came to shoulder the significantly higher cost per student that it must imply. At present, the mixed modality does not even appear in official statistics. Demand for this variant is reportedly growing. By more than doubling the hours of care, continued full public funding could have a noticeable short term impact on cost per student. In the long term, full day care will also conflict with the current 2-shift system thereby affecting long-term capital costs, too. The economic (and not just pedagogical) basis and design of such programs needs further thought. Indigenous P'-eschools are run by the Direcci6n General de Educaci6n Indigena. The main difference in these schools is that the language of instruction is not Spanish, but one of the approximately 60 native languages of Mexico. Preference is given to Normal school graduates, but the main requirement is fluency in a native language. Teaching materials differ from those used in the General Preschools, as they emphasize local culture and are often written in a minority language. CONAFE is charged with providing educational opportunities to the more remote areas of the country. This is why their schools are most often referred to as "Community" or "Rural" schools. The CONAFE charter brings them into contact not only with poor, Spanish-speaking children but also with indigenous groups, and with the offspring of migrant farm workers. In order to reach these populations and provide them with instruction, CONAFE has relaxed teacher education requirements. In one program community "promoters" are in their iast year of high school. In others they may have even less formal education.  13 TABLE 2: Characteristics of 3 Program Types (1995) PROGRAM NAME Schools Tot. Enrollmt # Teachers Classrooms Stud/Class Stud/Teach General 44,786 2,777,662 111,848 124,255 22.4 24.8 Indigenous 7,739 261,971 11,922 12,349 21.2 22.0 CONAFE 7,779 84,450 7,779 5,331 15.8 10.9 Table 2 groups the data by the three basic types. About 90% of the students are enrolled in General preschools, 8.4% in Indigenous and the remaining 2.7% in CONAFE schools. The most notable difference across these broader types is that CONAFE classes and student teacher ratios are far smaller than in the General and Indigenous schools. IV. PRESCHOOI i-NROLLMENT AND COVERAGE This section of the paper examines how many students are enrolled and reports on their ages and gender. It then develops the best feasible estimates of coverage using SEP and ENIGH data. Data Quality The quality of educational data available in Mexico does not match the size and sophistication of the country's educational system. In fact, Mexico is surprisingly data- poor. There is no one source that reports what share of the population is in school, who they are, what it costs to educate them and how much they are paying for the education. The main micro-data resources are "raw" and designed for other purposes. These include: * "SIDEC` (SEP) school-level data for enrollment, ages and number of teachers * PARE (CONAFE) student level ability test data for test scores and repetition status, in the 4 poorest states * INEGI Census data for population * INEGI household survey data for crossing enrollment with socio-economic status * SEP and Secretaria de Hacienda reports for unit cost data The empirical portion of this study constitutes original analysis. It is formed by quilting together bits of relevant information and inference from all of these sources. Each one has its own limitations as none was designed for the specific purpose of evaluating the cost and effectiveness of an educational system.  14 SIDEC databases on enrollments are good, well managed and available in a format that makes them easily accessible. They contain information on number of teachers and classes, along with data on student enrollments by gender, age and grade. The national statistics bureau (INEGI) gathers demographic data through the census every 10 years. This makes it possible to estimate program coverage, but only at a very aggregated level. Enrollment Table 3 shows total preschool enrollments, by state. The preschool program nominally includes separate grades for each of ages 3, 4 and 5. If the three ages were represented proportionately, each one would account for about 33% of enrollment. Table 3 shows that in fact the share of 5-year old children ranges from 41.7% in Chiapas to 67.4% in Baja California. When 6-year olds are added, the share of older children rises by an another 5 percent in each state. Program design notwithstanding, coverage is strongly skewed towards the older children. Access to preschool does not depend on gender in Mexiko. At a national level, 49.5% of preschool students are girls. Table 3 shows that gender ratios are very close to one-half in all three grades in all States.5 Table 2 also showed gender equality in access across the major program types. CONAFE rural schools deviate significantly from this national pattern, but they account for only about 2.5% of total enrollment. Coverage In contrast to enrollment, which only counts the number of students, coverage measures this number in relation to some norm. The norm most often employed is the total number of people in the country who are of the design age group. For the Mexican preschool system this would be children of age 3 for grade KI, age 4 for K2 and age 5 for K3. Coverage Estimates from SEP data. Unfortunately, another data problem is encountered at this point. SEP only gathers information on the numerator for this ratio, not the denominator. In order to get around this difficulty, data from a second SEP series on primary school enrollment were used to calculate lower and upper bounds for net enrollment ratios by State.6 The magnitude of the estimates was then checked against a national average derived from INEGI household survey data.7 Evidence from the two sources converges to indicate that about 63% of 5 year olds are enrolled in public preschools in Mexico. 5Data for Aguascalientep have probably been mis-coded. 6Attempts to estimate coverage shortfalls at the municipal level, by subtracting SEP enrollment figures by age from INEGI population estimates for the same age, yielded many nonsensical negative numbers. 7Unfortunately, INEGI data only permit coverage estimates for the 5 year olds, not 3 or 4 year olds.  15 ;; 777 0 : X State Total |Age 4 Age$5: % Age 5 Girls_Kl Girls _K2 Girls_ K3 Aguascalientes 32,1051 10,435 19,487 60.7% 62.6% 87.4% 66.3% Baja California 62,344 20,238 42,008 67.4% 52.3% 49.7% 48.9% Baja California Sur 14,663 6,475 7,855 53.6% 51.3% 50.2% 49.0% Campeche 26,438 10,254 12,357 46.7% 49.4% 49.4% 50.1% Coahuila 75,474! 24,979 44,026 58.3% 51.7% 50.1% 49.0% Colima 18,834 6,280 10,896 57.9