School reorganization reforms: The case of school networks in Colombia

Elacqua, Gregory; Sánchez Torres, Fabio; Santos, Humberto Working Paper School reorganization reforms: The case of school networks in Colombia IDB Working Paper Series, No. IDB-WP-1044 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Elacqua, Gregory; Sánchez Torres, Fabio; Santos, Humberto (2019) : School reorganization reforms: The case of school networks in Colombia, IDB Working Paper Series, No. IDB-WP-1044, Inter-American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0001844 This Version is available at: https://hdl.handle.net/10419/208194 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode School Reorganization Reforms: The Case of School Networks in Colombia Gregory Elacqua Fabio Sanchez Humberto Santos IDB WORKING PAPER SERIES Nº IDB-WP-01044 A ugust 2019 Education Division Inter-American Development Bank A ugust 2019 School Reorganization Reforms: The Case of School Networks in Colombia Gregory Elacqua Fabio Sanchez Humberto Santos Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Elacqua, Gregory M., 1972School reorganization reforms: the case of school networks in Colombia / Gregory Elacqua, Fabio Sánchez, Humberto Santos. p. cm. — (IDB Working Paper Series ; 1044) Includes bibliographic references. 1. School management and organization-Colombia. 2. Educational change-Colombia. 3. Academic achievement-Colombia. I. Sánchez, Fabio. II. Santos, Humberto. III. Inter-American Development Bank. Education Division. IV. Title. V. Series. IDB-WP-1044 Copyright © Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 AttributionNonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/ legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore, the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license and these statements, the latter shall prevail. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. http://www.iadb.org 2019 1 School reorganization reforms: The case of school networks in Colombia Gregory Elacqua Education Division, Social Sector Department Inter-American Development Bank Fabio Sánchez Department of Economics, Universidad de Los Andes Humberto Santos Education Division, Social Sector Department Inter-American Development Bank Abstract To contribute to the debate about the reorganization of the public-school supply, in this paper we analyze a reform introduced in Colombia in 2001 that merged several independent small schools into a single educational institution, with the same name, administration, educational project and school principal. Specifically, we estimate the differences in student achievement and measures of teacher characteristics and technological infrastructure between school sites belonging to a multi-site institution with single-site schools. Our results suggest that there are no differences in standardized test scores, but at the same time, we found that school sites belonging to multi-site schools, especially those in larger networks (more than 6 sites) and located far away from the main site of the school, tend to have younger teachers that earn lower salaries and are more likely to have a temporary contract. We also find evidence that more isolated sites from large-sized schools have less of a probability to have access to the Internet and a computer room than single-site schools. From the point of view of public policy, these results provide suggestions of potential reforms that should be implemented to increase cooperation between sites within the same school. Keywords: school reorganization, Colombia, educational achievement JEL Classification: I210, I240, I280 Introduction Due to demographic, economic and political trends, small schools have become a pressing problem in many countries and regions. Because these generally face higher costs per student, school networks with a large number of small schools tend to be expensive to operate (Ares, 2014). In many education systems, the solution has been to close small schools and replace them with larger schools (Abu-Ghaida, Alonso, and Sánchez, 2011; Ares, 2013). However, in some situations, school consolidation or restructuring can be difficult to implement, especially when families do not have options available or when transportation 2 costs are too high (Elacqua et al., 2012). For example, in many rural and remote areas with low population density, schools are facing declining enrollments. Given the geographic location of these schools, most cannot attract more students, and consolidation may be problematic for different reasons such as travel distances and lack of adequate facilities (Ares, 2014). Another policy that some systems have adopted is to foster cooperation between schools, in order to reduce costs and increase efficiency. Specifically, school systems have created clusters of sites, where several independent schools are managed by a single administration. Advocates argue that this policy could improve the efficiency of school networks, generating economies of scale, maintaining geographical coverage and avoiding the closure of small rural schools. On the other hand, skeptics argue that, in this model, headquarter schools could utilize most of the network’s resources, increasing the inequity in the allocation of inputs and resources, and that coordinating the functioning of the cluster is more difficult, and introduces higher costs, when the cluster is composed of many institutions and when the distance between schools in the network is significant. In 2001, the Colombian government instituted a reorganization process that resulted in the merging of several small schools into a single institution, with the same name, administration, educational project and budget (Corpoeducación, 2004; 715 Law of 2001). The reform had four main objectives: (i) guarantee school attendance at all levels of basic education; (ii) promote school retention, by avoiding movements of students from one site to another; (iii) improve student learning and reduce repetition, by following the same institutional educational project, and (iv) increase efficiency through a more intensive use of physical infrastructure, materials, and equipment. In this paper we examine the school reorganization reform that Colombia instituted in 2001. Specifically, we estimate differences in educational outcomes measured by results on the SABER test and measures of schooling inputs, comparing school sites belonging to a multi-site institution with schools functioning independently (single-site schools). Although there are several educational systems that have introduced school cluster and network reforms, there is scant evidence on the impact of these policies on student outcomes. Colombia’s school reorganization reform provides an interesting context to explore this topic. The national scope of the reform and the availability of data at the school site level provide a unique opportunity to examine this reform. Moreover, the debate about school reorganization is a hot topic in Latin America, because several systems in the region are considering policies to optimize the school supply in rural and isolated zones. For example, in Peru, the Ministry of Education recently approved guidelines for the organization and operation of the Rural Educational Networks (RER) to strengthen the management of basic education institutions located in rural areas. In Chile, a recent study commissioned by the Ministry of Education recommends significant changes to the per pupil voucher system to reduce the pressure on the financial viability of small and rural schools, especially during a period of marked demographic decline. In Ecuador, the government recently announced that it would reverse a rural school closure policy. This paper is organized as follows. In section 1 we review the literature about the reorganization of school supply and school networks. Section 2 describes the reform in Colombia. In section 3 we introduce the 3 econometric strategy and the data. Section 4 presents descriptive statistics and section 5 describes the results. Finally, section 6 discusses the findings and presents some policy implications. 1. Literature Over the past decades, the set of policies school systems have adopted to face the challenge of small and isolated schools has expanded beyond merging or closing schools. There is growing interest in developing informal and formal co-operation between schools, a focus on the expansion or reduction of specific educational services within schools, and the reorganization of specific grade levels. School systems have developed different models of cooperation. For instance, the joint provision of specialized educational services or curricula, sharing human resources, facilities and infrastructure, jointly purchasing materials or services, coordinating student transportation, or making professional development training available to teachers from multiple schools. There are two main models, which differ in the level of autonomy of the cooperating units and the intensity of collaboration (OECD, 2017). In the first model, a group of independent schools cooperate and share resources. For example, in Belgium (Flemish Community), the government provided incentives in the form of additional staff resources that can be shared between schools that belong to an association. These associations are free to adjust the scope of their collaboration based on their respective needs, ranging from communities with low-intensity co-operation on selected issues to those that share a wide range of services and resources. Another example of this model are the centers for specialized services and teacher collaboration. For example, in Estonia, the regional counselling centers have centralized the provision of specialized services related to the diagnosis and accommodation of special educational needs of multiple schools in the region, thus allowing them to be provided at an efficient scale. Another example is in Chile, where “rural micro-centers” (microcentros rurales) provide teachers in rural areas with a space to meet, to collaborate, and to share best practices to address their common challenges1 (OECD, 2017). In the second model, a group of schools are organized as a cluster, under a consolidated administration (OECD, 2017). Giordano (2008) introduced a typology of different models, based on the relations between the schools in the network and the way in which the clusters are created, distinguishing between bottomup and top-down, voluntary vs. mandatory, selective vs. universal, highvs. low-intensity cooperation approaches, financially autonomous vs. financially supported, and between clusters that serve primarily as tools for external control vs. those that promote internal development. Advocates of these policies argue that the joint administration of multiple schools could improve the efficiency of school networks and the capacity to provide education, generating economies of scale, maintaining geographical coverage and avoiding the closure of small rural schools. For example, a group of schools can share their equipment and infrastructure, or motivate high-quality teachers to remain in rural areas, offering them better opportunities for development and a wider professional community (OECD, 2017). At the same time, economic theory suggests several possible reasons why costs (measured 1 For more information about microcentros, see https://rural.mineduc.cl/el-microcentro/ 4 per pupil) might vary with school size. Some authors have maintained that there may be economies or diseconomies of scale in schooling (Andrews, Duncombe and Yinger, 2002). For instance, costs may decline with size as the fixed costs of schools are spread over a larger student body. Fixed costs might include the physical infrastructure of the school as well as personnel, such as principals, or other inputs, such as library books. The key is that these inputs are, in some measure, indivisible. Alternatively, economies of scale may reflect the gains from specialization and division of labor, for example, if teachers are able to specialize in offering courses in their areas of expertise (Stiefel et al, 2009). Despite the expected positive effects, educational integration has presented several challenges (Corpoeducación, 2004; Giordano, 2008). First, skeptics argue that, when several schools form a network, headquarter schools could utilize most of the network’s resources, increasing the inequity in the allocation of inputs and resources. This is a risk, especially in those networks where there is a school that concentrates resources and decision making. Second, critics are also be concerned that the newly formed networks become a sum of the weaknesses that exponentially increase the problems that each of the integrated institutions face. Third, integration may have negative impacts on social capital and cohesion in the surrounding communities, due to the loss of autonomy of the local school (Ares, 2014). For example, in a study on school consolidation in the state of Arkansas, Nitta et al. (2010) found that, teachers and parents report a reduction in the interaction with each other. After consolidation, the school authority is perceived by parents as being more distant, centralized, and concentrated. The larger distance to the school organization and to decision-makers makes interaction costlier and reduces the possibility that parents will participate in school activities (Strang, 1987). Finally, coordinating cluster activities is more difficult when the cluster is composed of many institutions and when the distance between schools in the network is significant. Giordiano (2008) shows that large clusters can have coordination problems which reduce the cooperation between schools and make the supervision and support of the network difficult. Coordination costs (e.g. transportation and communication) can also create diseconomies of scale because they limit the ability of principals (or districts) to manage a large school network or generate congestion in the utilization of fixed resources. Portugal and England have adopted school cluster policies. In 2005/06, Portugal, implemented a reform to consolidate schools into networks to address inefficiency and drastic regional inequalities among small and remote schools. Prior to the reform, rural areas had predominately small schools with poor infrastructure and low performance while schools in urban areas were often overcrowded and relied on double shifts. To address this problem, the Ministry of Education co-operated with local governments and school executive boards to close small, underperforming schools. The government also introduced school clusters to create incentives for cooperation between schools (OECD, 2017; Ares, 2014). A small school cluster typically consists of one headquarter school that offers the second and third cycle of education and 4-5 schools (from 20 to 400 pupils) that offer the first cycle education. Although there is no causal evidence of the impact of the policy on student outcomes, an evaluation based on interviews with a wide range of stakeholders suggests that the change had a positive impact on student retention and learning, reduced the isolation of teachers, optimized the supply of different grade levels, and increased the retention of teachers and the availability of resources for students and teachers (Matthews et al, 2009). 5 The United Kingdom created school federations to improve results in low-quality small schools. The federation policy merged two or more schools to form a single ‘split - site’ school with one head teacher and governing body. Studies on the impact of federations and collaboration in rural areas suggest that the policy created incentives for schools to pool resources, develop a broader curriculum, and increase professional development opportunities (Chapman, 2015). Chapman and Mujis (2014) used a quantitative methodology involving a matched sample of federated and non-federated schools and multilevel modelling techniques, to explore the impact of federations on student outcomes. They found evidence of a positive impact on overall student attainment, and that this federation effect is greater in federations where higher performing schools partner with lower performing ones. However, there is also evidence that in some cases, small schools in the federation can feel dominated by more centrally located or larger ones. When this occurs, schools have fewer incentives to collaborate because of mistrust (Mujis, 2008). Another potential concern in the reorganization process is the unequal allocation of inputs between rural and urban sites (e.g. teachers). 2. Reorganization of the school supply in Colombia In 2001, the Colombian government implemented a reorganization process that resulted in the merging of several small schools into a single institution. Unlike past efforts, this was the first time that the reorganization became mandatory for schools in the public sector (escuelas oficiales)2. First, an educational institution, according to Law 715, enacted in 2001, must offer all grades (K-11), grades 1-9 or grades 10-113 and belong to a network of two or more school sites that previously functioned as independent schools before the reform (sedes educativas). One of the schools in the network operates as headquarters (sede principal) and receives the resources from the central government and from territorial entities4 and employs the administrative staff. The main site’s school principal can assign or reallocate teachers to the rest of the sites (sedes adscritas), especially those with temporary contracts, manages the network’s budget and defines the educational project of the school (Econometría, 2013). The reform had four main objectives: (i) guarantee school attendance at all levels of basic education; (ii) increase grade retention, reducing dropout rates that often occur when students transfer to other schools; (iii) improve student learning and decrease repetition, following the same institutional educational project, and (iv) 2 For example, Ley 115 de 1994 (General Education Law) states that schools must have the administrative infrastructure and supports of the pedagogical activity to offer at least preschool and basic education. It also indicates that if they do not offer the complete educational cycle, they should establish agreements with other schools, to guarantee the continuity of the educational process of their students. For details about former attempts to reorganize the public supply, see Corpeducación (2014). 3 In Colombia, the school system is divided into three levels: básica primaria, (grades 1-6), básica secundaria (grades 6-9) and educación media (grades 10-11). The equivalent in the ISCED classification is primary, lower secondary, and upper secondary respectively. 4 In Colombia, education is managed by Certified Territorial Entities (Entidades Territoriales Certificadas or ETC) that can be Departaments (Departamentos or Sub-national level) or Municipalities (Municipios or local level) (Ley 715 de 2001). Departments and districts were certified automatically. The education in municipalities that are not certified is managed by Departments. In the first stage after the reform that introduced the ETCs, certification was granted to those municipalities that exceeded 100 thousand inhabitants (Brutti, 2014). Currently, municipalities with less than 100 thousand inhabitants can apply to be certified after an evaluation of the technical, administrative, and financial capacity to manage the school network. 12 medium-sized schools (8.4%). In the case of 9th grade, the dropout rate is higher in sites from large schools (10.2%), followed by sites from medium-sized schools (8.0%) and independent sites (7.5%). The data also indicate that sites in larger networks are more likely to have younger teachers with lower monthly salaries16, and less likely to have a permanent contract than colleagues that work at independent sites. The difference between single-site and medium-sized institutions are much smaller. These differences can also be observed in access to information and communication technology infrastructure. Sites belonging to large institutions are less likely to have access to the Internet and computer rooms. Independent sites are less likely to have access to the Internet, compared to those belonging to mediumsized institutions. 5. Results 5.1. Learning outcomes and dropout rates Tables 5-8 present the results of estimations for the percentage of students in the lower level (insuficiente) in SABER 5th and 9th grade in Language and Mathematics. When controls are not included at the school site or Territorial Entity level (Specification 1), the results show that the sites that belong to multi-site schools, both medium (2-5 sites) and large (6 sites or more), have a higher percentage of students in the insufficient level than independent sites (between 3 and 5 percentage points more, depending on the area and grade). This gap is even higher when considering those sites located farthest from the main site of the school (Specification 9). However, when we control for the characteristics of the sites and Territorial Entities (Specifications 2 and 3), the results are no longer significant, as well as when Territorial Entity fixed effects are included (Specification 4). According to these results, the sites that belong to larger networks have lower educational results, which is explained mainly by their location, socioeconomic composition, and other observable characteristics included in the regression. Thus, our results suggest 16 In Colombia, there are two different pay scales for teachers in the public sector. Teachers that entered in the teacher career before 2002 are governed by Decree 2277 of 1979 and those hired after that year by Decree 1278 of 2002. The current pay scale is divided in three different grades, defined by the level of education of teachers: i) Normal-school or technical teacher, ii) University graduate in education or non-graduate professional, and iii) University graduate in education or non-graduate professional with master or doctorate degree. In each grade there are four levels (A, B, C and D) through which teachers increase as they gain experience and pass a competency assessment. The evaluation is voluntary and can be presented by teachers and directors aspiring to be promoted or relocated in the scale. There is a base monthly salary in each combination of grade and level. The old pay scale consists of 14 levels, each one having a higher monthly base salary. The education level of the teacher determines at what level she enters. For example, a teacher with a Normalist degree entered Level 4, while a teacher with a Bachelor's degree entered Level 7. Increases on the scale are determined by experience; every three years teachers go to the next level of the scale, which brings a salary increase of approximately 12 percent (Ome, 2013). If a teacher acquires additional education, (which does not necessarily lead to a formal education qualification), he can substitute this additional education for experience, implying that he will ascend faster. 13 that there is not a statistically significant relationship between belonging to a multi-site institution and the academic performance of the institutions17. Regarding the fifth-grade dropout (last grade of primary education) (Table 9), the results show that, without controlling for other factors (Specification 1), the sites belonging to medium-sized institutions have a dropout rate lower than the independent sites (-1.1 pp). But at the same time, the sites that belong to large institutions have higher dropout rates (+1.4 pp) than independent sites. However, this relation is no longer significant once we control for the characteristics of the sites (Specification 2). We also find that, with respect to the independent sites that do not have lower secondary (reference category), all the other categories of sites show lower dropout rates (Specifications 5-8), which implies that having lower secondary school in the same site or having that level available at another site within the school, reduces the likelihood of dropping out for a fifth-grade student. For example, the dropout rate is 6.3 and 4.9 percentage points lower (for medium and large networks respectively) in sites in which the integration make available lower secondary (i.e. do not have lower secondary, but other sites in the same school offer this level), with respect to single-site schools that do not have lower secondary (Specification 8). The latter was precisely one of the objectives of the reorganization of the educational supply. This is relevant, because a recent paper shows that the highest dropout rate is in the transition between 5th and 6th grade (Sanchez et al, 2018). In the ninth grade (last grade of lower secondary) the results are different (Table 10), since in this case, the sites belonging to medium and large institutions have a higher dropout rate than independent sites, especially those located at a greater distance from the main site. This result holds even after controlling for the characteristics of the sites and territorial entities they belong to. The difference with respect to 5th grade is that 86% of the independent sites that have 9th grade also have the upper secondary level at the same site, which reduces the probability to drop out (See Table 4 for descriptive statistics). This shows that the impact of the policy is greater in those schools that only have the primary level and can now access the next level in the same institution. 5.2. School inputs and technological infrastructure Tables 11, 12, 13 and 14 show evidence that, other things held constant, sites belonging to multi-site schools, especially those in the largest networks, have teachers with lower average monthly salaries (Table 11), more likely to have temporary contracts (Table 12) and younger (Table 13 and 14), than teachers that work at single-site institutions. The difference between site categories is greater when the site is farthest 17 The results do not change if the model is estimated only with the non-main sites of the multi-site schools and when we restrict the sample just to sites located outside of the capitals of the departments (32 municipalities) and Bogotá. We also used propensity score matching to estimate the average treatment effect (ATT) of belonging to a multi-site school, using the same control variables in Tables 3-6 to estimate the propensity score and imposing a common support by dropping treatment observations whose pscore is higher than the maximum or less than the minimum pscore of the controls. We also estimated the Lee Selection Model for SABER 5th (Lee, 1983), in which selectivity is modelled as a multinomial logit (as opposed, for instance, to a univariate Probit as in the Heckman model) using the availability of lower secondary at the school site as the exclusion variable. In both cases, the results show that there are no significant differences in the SABER results between sites belonging to multi-site schools (medium or large) and single-site schools. These results are available upon request. 14 away from the main site (Specifications 9-11), which supports the hypothesis that there are inequities within schools with several sites, affecting those more isolated from the main site. For example, a teacher in a school site that belongs to a large institution but is more than 5 km away from the main site, earns a monthly base salary of almost 220,000 pesos (≈72 USD) less than a teacher in a single-site school (Specification 11). On the contrary, there are no significant differences between the latter and the teachers in medium-sized schools that are within a radius of 1 km from the main site. However, when we add the ETC fixed effect to the regression, the results change, and school sites belonging to medium-sized institutions have teachers with higher average salaries, with a higher probability of being older and having a permanent contract than the rest of categories. On the other hand, results show that the sites of large institutions (more than 6 sites) and especially those farthest away from the main site, are less likely to have access to the Internet and a computer room (Tables 15 and 16) than single sites. In this case, the results are significant in all specifications, even when ETC fixed effects are included. For example, a site that belongs to a large school but is located more than 5 km away from the main site, has a probability of having internet 11.1 percentage points less than in a single-site school (Specification 12). On the contrary, there are no significant differences between the latter and school sites located within a radius of 1 km from the main site. However, it is interesting that belonging to a medium-sized network, especially when it is close to the headquarter school, increases the likelihood of having an Internet connection. In this case it may be that belonging to the network allows access to the sites with an Internet connection. These results are consistent with qualitative assessments indicating the existence of deficiencies in the infrastructure in school sites belonging to multi-site schools and problems to recruit quality teachers, which is exacerbated by the distance to the main site of the school, restricting the learning opportunities offered to their students (Econometría, 2013). 6. Discussion The objective of this paper is to contribute to the debate about the reorganization of the public-school supply. We analyze a reform introduced in Colombia in 2001 that merged several independent small schools into a single educational institution, with the same name, administration, educational project and school principal. To inform that debate, we estimate the differences in student achievement and measures of teacher characteristics and technological infrastructure between school sites belonging to a multi-site institution with single-site schools. We find that there are no significant differences in SABER test scores between sites belonging to multisite schools and single-site schools, after controlling for characteristics of the school site and the territorial entity to which the school belongs. However, we found suggestive evidence that school sites, especially the larger networks (more than 6 sites) and located far away from the main site of the school, tend to have younger teachers that earn lower salaries and are more likely to have a temporary contract. However, it is not possible to disentangle how much of this inequity is explained by teacher preferences for schools during the hiring process (e.g. because applicants with high scores in the teacher contest may 15 be less willing to choose vacancies in locations far from the main headquarters due to lower autonomy and isolation) and how much is explained by decisions of the school principal or education secretary about the allocation of teachers within schools (e.g. focus on the results of the main site of the school). This is a topic to be explored in future research. We also find evidence that more isolated sites from large-sized schools have less of a probability to have access to the Internet and a computer room than single-site schools. Considering the long distances that exist between sites in large-sized schools, the lack of connectivity may mean high coordination costs, which reduce the possibility of generating cooperation and exchange of inputs and personnel within school networks. One way to address this problem is to focus public infrastructure investments from the central government and territorial entities could focus on school sites in remotes areas that belong to multi-site schools. Another option is to modify the funding formula in Colombia to consider special transfers to institutions with networks of school sites. However, preliminary results presented in this draft must be taken with caution, because, as was shown in the descriptive statistics, there are very large differences in observed variables between school sites belonging to multi-site institutions and those functioning as a single-site school. Even after controlling for a set of characteristics of the sites, there is little common support between treated and control sites, since sites belonging to networks are much smaller, are much less likely to have the next level available (lower secondary or high-school), are more likely to have non-traditional educational models, are more likely to be located in rural areas, and are more likely to be a single-teacher school than single-site schools. This is more pronounced in larger sized school networks with more than 6 sites. This lack of common support can affect the results and bias the estimations. So, it is difficult to conclude that the statistical relationships found are causally attributable to belonging to a multi-site school. However, from the point of view of public policy, what is relevant is that these sites, especially those far from the main site, lack the basic infrastructure to work in a network and have difficulties in recruiting quality teachers. This conclusion implies that interventions should focus on these schools. Another difficulty is the lack of evidence about the implementation of the network formation process and the moment in which each institution merged, given that this process was implemented individually by each entity and for the time that has elapsed since the implementation, which makes it difficult to determine potential explanatory variables of the treatment. Another limitation of this study is that, due to a lack of information, we do not analyze potential cooperation mechanisms between different sites of the same school that can compensate for the lack of infrastructure and quality teachers. The existence of this cooperation may explain why we do not find differences in test scores after considering observable differences between school sites. Some qualitative evidence suggest that these cooperation mechanisms exist in practice (Econometría, 2013). For example, this research suggests that school principals promote visits to support teachers from other schools or to discuss pedagogical issues several times a year, the most frequent is to advise teachers, whose frequency varies between 11 times a year in Bolívar up to 19 times in Norte de Santander. Joint meetings are also held for cultural or pedagogical activities that fluctuate between 4 per year in Cundinamarca and 8 in Bolívar. In general, there is access, from the library, the computer room and the sports facilities of the main site, to the students of other sites. This coincides with the response of the number of days that 16 teachers and students have used the resources of the main office, where Bolívar has 15 days a year, Cundinamarca 3, Guaviare 7 and Norte de Santander 9. At the same time there is evidence of teachers from the main site teaching in other sites and teachers working in several sites in the same school network. 17 Figures and tables Table 1. Distribution of school sites according to number of sites of the school schools sites 1 site N 3,292 3,292 % 33.3 7.6 2-5 sites N 3,831 12,391 % 38.7 28.5 6 or more sites N 2,769 27,797 % 28.0 63.9 Total N 9,892 43,480 % 100 100 Table 2. Distribution of schools, by number of sites and grades offered 1 site 2-5 sites 6 or more sites N % N % N % Preschool (PS) 8 0.2% 0 0.0% 0 0.0% Primary (P) 437 13.3% 9 0.2% 3 0.1% Lower secondary (LS) 1 0.0% 0 0.0% 0 0.0% Upper secondary (US) 3 0.1% 0 0.0% 0 0.0% PS-P 1,297 39.4% 320 8.4% 219 7.9% P-LS 15 0.5% 0 0.0% 0 0.0% LS-US 23 0.7% 3 0.1% 2 0.1% PS-P-LS 216 6.6% 311 8.1% 456 16.5% P-LS-US 8 0.2% 5 0.1% 4 0.1% PS-P-LS-US 1,275 38.7% 3,183 83.1% 2,085 75.3% Only adults 9 0.3% 0 0.0% 0 0.0% Total 3,292 100.0% 3,831 100.0% 2,769 100.0% Table 3. Distribution of schools, by number and location of sites 1 site 2-5 sites 6 or more sites N % N % N % Only rural sites 2,277 69.2% 1,792 46.8% 1,971 71.2% Only urban sites 1,015 30.8% 1,641 42.8% 75 2.7% Urban and Rural sites 0 0.0% 398 10.4% 723 26.1% Total 3,292 100.0% 3,831 100.0% 2,769 100.0% 18 Figure 1. Total enrollment range within multi-site schools Figure 2. Distance to the main site range within multi-site schools 0 .05 .1 .15 .2 .25 .3 0 .05 .1 .15 .2 .25 .3 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 2-5 sites 6 or more sites Fraction total enrollment range (max-min) 0.2 .4 .6 .8 0.2 .4 .6 .8 050 100 150 200 2-5 sites 6 or more sites Fraction distance to main site in km. (max) 19 Figure 3. Distribution of school sites according to SES of student population Primary Lower Secondary 29.1% 18.4% 28.3% 20.4% 19.6% 29.4% 16.6% 24.3% 26.1% 20.1% 29.0% 15.1% 13.9% 8.7% 1.2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1 site 2-5 sites 6 or more sites Q1 Q2 Q3 Q4 Q5 16.2% 22.3% 41.2% 18.2% 25.1% 34.0% 24.4% 28.9% 21.0% 33.8% 22.1% 3.8% 7.4% 1.6% 0.0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1 site 2-5 sites 6 or more sites Q1 Q2 Q3 Q4 Q5 20 Table 4. Descriptive statistics of the database mean Nmean Nmean N Outcome variables p_insuficientelenguaje_5_grado percentage of students in the lower level (insuficiente) of SABER 5th grade Language test 15.7% 778 18.9% 4899 19.3% 3257 p_insuficientematematicas_5_grado percentage of students in the lower level (insuficiente) of SABER 5th grade Mathematics test 37.3% 778 41.7% 4891 41.7% 3245 p_insuficientelenguaje_9_grado percentage of students in the lower level (insuficiente) of SABER 9th grade Language test 16.2% 762 19.9% 2936 20.2% 1985 p_insuficientematematicas_9_grado percentage of students in the lower level (insuficiente) of SABER 9th grade Mathematics test 21.1% 759 24.5% 2918 25.6% 1935 asignacion_basica_m_m base salary of all teachers in the site (MM Colombian pesos 2016) 2.3 2784 2.3 11169 2.0 22950 p_profesores22_edad percentage of teachers (20-25 years) in primary education 2.1% 3014 2.2% 10313 5.2% 25134 p_profesores32_edad percentage of teachers (20-25 years) in lower secondary education 0.9% 1414 1.1% 3880 2.2% 3837 p_profesores26_edad percentage of teachers (>40 years) in primary education 65.3% 3014 63.0% 10313 48.2% 25134 p_profesores36_edad percentage of teachers (>40 years) in lower secondary education 58.9% 1414 59.7% 3880 44.6% 3837 p_profesores1_con percentage of teachers with permanent contract in the teacher career path 81.6% 3092 88.4% 11853 76.6% 26474 tieneint_n percentage with access to the internet 49.2% 3092 61.9% 11853 27.8% 26474 tieneaul_n percentage with access to computer room 73.8% 3092 74.0% 11853 44.2% 26474 desercion5 dropout rate 5th grade (%) 9.8% 2033 8.4% 7559 11.0% 7909 desercion9 dropout rate 9th grade (%) 7.5% 1373 8.0% 3535 10.2% 2425 School site characteristics matr_tot total enrollment 467 3292 345 12391 78 27797 unidocente_2 percentage single-teacher in primary 40.9% 3014 29.7% 10313 73.8% 25134 unidocente_3 percentage single-teacher in lower secondary 3.5% 1414 3.2% 3880 20.4% 3837 zona2 percentage rural 69.2% 3292 56.3% 12391 91.6% 27797 quintil5_1 percentage in SES quintil 1 5th grade 29.1% 2508 18.4% 7162 28.3% 12047 quintil5_2 percentage in SES quintil 2 5th grade 20.4% 2508 19.6% 7162 29.4% 12047 quintil5_3 percentage in SES quintil 3 5th grade 16.6% 2508 24.3% 7162 26.1% 12047 quintil5_4 percentage in SES quintil 4 5th grade 20.1% 2508 29.0% 7162 15.1% 12047 quintil5_5 percentage in SES quintil 5 5th grade 13.9% 2508 8.7% 7162 1.2% 12047 quintil9_1 percentage in SES quintil 1 9th grade 16.2% 1226 22.3% 3267 41.2% 2770 quintil9_2 percentage in SES quintil 2 9th grade 18.2% 1226 25.1% 3267 34.0% 2770 quintil9_3 percentage in SES quintil 3 9th grade 24.4% 1226 28.9% 3267 21.0% 2770 quintil9_4 percentage in SES quintil 4 9th grade 33.8% 1226 22.1% 3267 3.8% 2770 quintil9_5 percentage in SES quintil 5 9th grade 7.4% 1226 1.6% 3267 0.0% 2770 antiguedad years of operation 30 3085 29 11824 24 26420 p_matricula_modelo1 percentage of students in a traditional model of teaching 51.3% 3248 59.6% 10951 17.9% 26779 tipo_jornada1 percentage full-time 2.4% 3292 10.4% 12391 30.3% 27797 tipo_jornada2 percentage double-shift 26.6% 3292 29.5% 12391 3.7% 27797 tipo_jornada3 percentage only morning shift 61.6% 3292 46.0% 12391 56.6% 27797 tipo_jornada4 percentage other type of school day 9.4% 3292 14.1% 12391 9.4% 27797 etnias2 percentage of students belonging to an ethnic group in primary 24.5% 3248 16.3% 10951 12.4% 26779 repitente22 percentage of students who have repeated a course in primary 2.6% 3248 3.1% 10951 2.9% 26779 cont_primaria1 percentage only primary site (it does not have lower secondary level neither at the site nor at the school) 53.4% 3248 10.1% 10951 8.3% 26779 cont_primaria2 percentage with lower secondary at the site 46.6% 3248 26.4% 10951 13.0% 26779 cont_primaria3 percentage only primary site but with lower secondary in the school -3248 63.5% 10951 78.8% 26779 etnias3 percentage of students belonging to an ethnic group in lower secondary 16.0% 1538 14.2% 4148 12.2% 4346 repitente23 percentage of students who have repeated a course in lower secondary 3.4% 1538 3.7% 4148 2.8% 4346 cont_secundaria1 percentage only lower secondary site (it does not have upper secondary level neither at the site nor at the school) 15.1% 1538 8.7% 4148 17.1% 4346 cont_secundaria2 percentage with upper secondary at the site 84.9% 1538 81.5% 4148 57.7% 4346 cont_secundaria3 percentage only lower secondary site but with upper secondary in the school -1538 9.9% 4148 25.2% 4346 distancia_sede_alcaldia distance between the school site and the main building of the municipality (km) 10.5 3131 11.4 11651 15.8 25572 distancia_sede_principal distance between the school site and the main site of the school (km) -3131 5.7 7905 11.0 22853 Notes: SABER scores and dropout rates includes only schools with more than 6 students in 5th and 9th grade Distance to the main site includes only sites different from the main site of the school 2-5 sites 6 or more sites 1 site Variable name Description 21 Table 5. Main regression for percentage of students in the lower level (insuficiente) of SABER 5th grade Language test (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites 0.032*** -0.001 -0.000 0.004 (5.70) (-0.21) (-0.06) (0.69) 6 or more sites 0.036*** -0.008 -0.002 0.010 (5.48) (-1.02) (-0.21) (1.32) total enrollment 0.000 0.000 0.000* 0.000 0.000 0.000* 0.000 0.000* 0.000** (1.24) (1.78) (2.16) (1.19) (1.81) (2.48) (1.60) (2.00) (2.69) single-teacher 0.017 0.019 0.023 0.017 0.019 0.022 0.014 0.017 0.020 (1.19) (1.43) (1.67) (1.18) (1.41) (1.63) (0.99) (1.27) (1.43) rural -0.020** -0.029*** -0.025*** -0.020** -0.028*** -0.023*** -0.021** -0.029*** -0.025*** (-2.98) (-4.04) (-3.60) (-2.83) (-3.83) (-3.30) (-3.03) (-4.07) (-3.68) quintile 2 SES (5th grade) -0.053*** -0.048*** -0.049*** -0.053*** -0.048*** -0.048*** -0.052*** -0.048*** -0.047*** (-4.28) (-3.93) (-4.08) (-4.27) (-3.92) (-4.06) (-4.18) (-3.86) (-3.96) quintilE 3 SES (5th grade) -0.086*** -0.076*** -0.070*** -0.086*** -0.077*** -0.070*** -0.085*** -0.076*** -0.069*** (-7.09) (-6.35) (-5.94) (-7.09) (-6.36) (-5.97) (-7.02) (-6.29) (-5.85) quintilE 4 SES (5th grade) -0.155*** -0.145*** -0.110*** -0.155*** -0.145*** -0.110*** -0.155*** -0.144*** -0.109*** (-12.49) (-11.57) (-8.85) (-12.47) (-11.57) (-8.88) (-12.42) (-11.51) (-8.77) quintilE 5 SES (5th grade) -0.223*** -0.212*** -0.160*** -0.223*** -0.212*** -0.161*** -0.223*** -0.211*** -0.159*** (-17.16) (-15.70) (-12.03) (-17.11) (-15.69) (-12.07) (-17.11) (-15.65) (-11.96) % of students belonging to an ethnic group in primary 0.124*** 0.112*** 0.109*** 0.124*** 0.112*** 0.109*** 0.125*** 0.112*** 0.109*** (9.54) (8.54) (7.33) (9.52) (8.53) (7.30) (9.64) (8.61) (7.36) % of students who have repeated a course in primary 0.060 0.027 0.043 0.061 0.027 0.044 0.060 0.027 0.044 (1.38) (0.67) (1.16) (1.39) (0.68) (1.20) (1.39) (0.68) (1.18) years of operation -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** (-4.35) (-4.63) (-4.93) (-4.34) (-4.61) (-4.91) (-4.24) (-4.55) (-4.83) % of students in a traditional model of teaching (1) 0.006 -0.005 -0.011 0.006 -0.005 -0.011 0.006 -0.004 -0.010 (0.74) (-0.66) (-1.45) (0.74) (-0.67) (-1.49) (0.81) (-0.58) (-1.27) double-shift 0.066*** 0.054*** 0.030*** 0.066*** 0.053*** 0.031*** 0.067*** 0.054*** 0.032*** (8.02) (6.35) (3.45) (7.93) (6.26) (3.43) (8.09) (6.42) (3.59) only morning shift 0.034*** 0.033*** 0.015 0.034*** 0.033*** 0.014 0.034*** 0.033*** 0.014 (4.31) (4.23) (1.65) (4.30) (4.19) (1.61) (4.32) (4.25) (1.65) other type of school day 0.026** 0.024** 0.011 0.026** 0.024** 0.011 0.027*** 0.025** 0.012 (3.21) (2.88) (1.22) (3.13) (2.82) (1.22) (3.32) (2.99) (1.41) distance school site-municipality (1-5 km) -0.017** -0.022*** -0.003 -0.017** -0.022*** -0.003 -0.018*** -0.023*** -0.005 (-3.12) (-3.74) (-0.57) (-3.11) (-3.72) (-0.54) (-3.40) (-3.89) (-0.80) distance school site-municipality (5-20 km) -0.005 -0.008 0.008 -0.005 -0.008 0.008 -0.006 -0.009 0.007 (-0.81) (-1.17) (1.11) (-0.81) (-1.16) (1.15) (-0.95) (-1.24) (0.99) distance school site-municipality (20-100 km) 0.003 0.004 0.024** 0.003 0.004 0.024** 0.003 0.004 0.024** (0.40) (0.44) (2.68) (0.39) (0.44) (2.69) (0.34) (0.41) (2.63) distance school site-municipality (>100 km) 0.001 0.004 0.022 0.001 0.004 0.022 0.003 0.005 0.024 (0.02) (0.12) (0.74) (0.03) (0.12) (0.75) (0.08) (0.16) (0.81) Total per student spending in the ETC -0.000*** -0.000*** -0.000*** (-10.09) (-10.13) (-10.06) % low-SES students in the ETC (5th grade) (2) -0.268*** -0.268*** -0.264*** (-6.97) (-6.91) (-6.84) total public enrollment ETC 0.000 0.000 0.000 (1.06) (1.04) (1.07) Percentage of rural students in the ETC 0.163*** 0.164*** 0.162*** (6.00) (5.99) (5.95) Heterogeneous effects by availability of the next level 1 site: with lower secondary -0.009 -0.011 -0.022 -0.031 (-0.22) (-0.29) (-0.57) (-0.77) 2-5 sites: no changes with integration (3) 0.034 -0.013 -0.023 -0.028 (0.84) (-0.33) (-0.59) (-0.69) 2-5 sites: integration make available lower secondary 0.015 -0.012 -0.020 -0.023 (0.39) (-0.29) (-0.53) (-0.56) 6 or more sites: no changes with integration 0.038 -0.018 -0.022 -0.020 (0.96) (-0.46) (-0.59) (-0.50) 6 or more sites: integration make available lower secondary 0.020 -0.019 -0.023 -0.019 (0.51) (-0.48) (-0.59) (-0.46) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.027*** -0.001 -0.000 0.004 (4.68) (-0.24) (-0.06) (0.65) 2-5 sites & 2-5 km to principal site 0.048*** 0.009 0.005 0.015 (5.71) (0.92) (0.58) (1.68) 2-5 sites & >5km to principal site 0.053*** -0.004 -0.003 0.010 (4.48) (-0.32) (-0.19) (0.79) 6 or more sites & <1km to principal site 0.021** -0.011 -0.004 0.008 (3.13) (-1.35) (-0.54) (0.98) 6 or more sites & 2-5 km to principal site 0.051*** -0.000 0.004 0.019 (4.78) (-0.03) (0.32) (1.73) 6 or more sites & >5km to principal site 0.063*** 0.006 0.010 0.028* (5.73) (0.48) (0.78) (2.32) constant 0.155*** 0.263*** 0.495*** 0.266*** 0.164*** 0.274*** 0.515*** 0.295*** 0.155*** 0.260*** 0.490*** 0.261*** (31.94) (16.08) (17.57) (15.21) (4.15) (6.39) (10.91) (6.68) (31.93) (15.75) (17.22) (14.84) Fixed effects ETC level No No No Yes No No No Yes No No No Yes N8502 6327 6327 6327 8495 6327 6327 6327 8502 6327 6327 6327 r2 0.003 0.175 0.194 0.272 0.006 0.175 0.194 0.272 0.009 0.175 0.195 0.273 aic -5501.813 -5725.594 -5871.239 -6334.189 -5506.437 -5719.726 -5865.579 -6329.404 -5541.780 -5723.447 -5866.590 -6334.663 bic -5480.669 -5583.790 -5702.425 -5557.642 -5464.153 -5557.664 -5676.507 -5532.599 -5492.443 -5554.633 -5670.765 -5531.106 t sta ti s ti cs in parenthes es *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) (3) It includes sites with lower secondary and those not having lower secondary in the school The categories omitted are: 1 site; quintile 1 SES (5th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no lower secondary 28 Table 12. Main regression for percentage of teachers with permanent contracts (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites 0.058*** -0.006 -0.000 0.050*** (7.36) (-0.74) (-0.03) (5.93) 6 or more sites -0.057*** -0.048*** -0.044*** 0.017 (-5.87) (-4.59) (-4.13) (1.45) total enrollment -0.000*** -0.000*** -0.000*** -0.000** -0.000** -0.000 -0.000*** -0.000*** -0.000*** (-9.10) (-8.83) (-5.66) (-2.79) (-2.90) (-1.30) (-8.78) (-8.43) (-4.96) single-teacher -0.010 -0.014 -0.012 -0.027*** -0.028*** -0.023** -0.009 -0.014 -0.013 (-1.38) (-1.94) (-1.72) (-3.47) (-3.64) (-3.16) (-1.14) (-1.83) (-1.89) rural 0.014 -0.000 0.004 0.025** 0.010 0.016 0.014 -0.001 0.003 (1.66) (-0.03) (0.45) (2.78) (0.99) (1.74) (1.61) (-0.09) (0.38) quintile 2 SES (5th grade) 0.077*** 0.077*** 0.073*** 0.080*** 0.080*** 0.075*** 0.077*** 0.076*** 0.073*** (8.86) (8.97) (9.21) (9.18) (9.29) (9.47) (8.83) (8.95) (9.22) quintilE 3 SES (5th grade) 0.102*** 0.102*** 0.098*** 0.104*** 0.104*** 0.099*** 0.102*** 0.102*** 0.098*** (10.60) (10.81) (11.10) (10.75) (10.99) (11.18) (10.60) (10.82) (11.08) quintilE 4 SES (5th grade) 0.100*** 0.102*** 0.106*** 0.099*** 0.101*** 0.104*** 0.099*** 0.101*** 0.105*** (9.59) (9.52) (10.04) (9.48) (9.48) (9.87) (9.59) (9.53) (10.00) quintilE 5 SES (5th grade) 0.063*** 0.065*** 0.080*** 0.066*** 0.069*** 0.077*** 0.064*** 0.065*** 0.080*** (5.24) (4.76) (5.98) (5.51) (5.07) (5.82) (5.24) (4.77) (5.98) % of students belonging to an ethnic group in primary -0.203*** -0.214*** -0.228*** -0.205*** -0.217*** -0.231*** -0.204*** -0.214*** -0.228*** (-9.65) (-10.10) (-9.97) (-9.72) (-10.24) (-10.13) (-9.66) (-10.09) (-9.99) % of students who have repeated a course in primary -0.055 -0.048 -0.074 -0.061 -0.058 -0.074 -0.054 -0.047 -0.075 (-0.79) (-0.65) (-1.10) (-0.88) (-0.78) (-1.09) (-0.78) (-0.64) (-1.11) years of operation 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (5.13) (5.06) (6.19) (5.17) (5.09) (6.20) (5.05) (4.99) (6.19) % of students in a traditional model of teaching (1) 0.076*** 0.071*** 0.058*** 0.073*** 0.068*** 0.058*** 0.076*** 0.072*** 0.059*** (9.15) (8.59) (6.80) (8.70) (8.05) (6.79) (9.13) (8.62) (6.86) double-shift 0.015 0.028* 0.032* 0.019 0.029* 0.035* 0.016 0.031* 0.035* (1.20) (2.22) (2.30) (1.55) (2.28) (2.57) (1.34) (2.40) (2.50) only morning shift -0.019 -0.026* -0.004 -0.023 -0.030* -0.004 -0.018 -0.025* -0.003 (-1.64) (-2.20) (-0.27) (-1.93) (-2.45) (-0.26) (-1.52) (-2.06) (-0.21) other type of school day -0.019 -0.021 -0.003 -0.010 -0.014 0.005 -0.018 -0.019 -0.001 (-1.59) (-1.73) (-0.22) (-0.87) (-1.16) (0.41) (-1.52) (-1.57) (-0.06) distance school site-municipality (1-5 km) 0.020* 0.033** 0.021* 0.020* 0.032** 0.022** 0.017 0.029** 0.017 (2.11) (3.23) (2.50) (2.07) (3.09) (2.66) (1.80) (2.88) (1.93) distance school site-municipality (5-20 km) -0.023* -0.007 -0.019 -0.020 -0.006 -0.017 -0.021* -0.006 -0.020* (-2.04) (-0.59) (-1.93) (-1.86) (-0.56) (-1.73) (-1.97) (-0.57) (-2.02) distance school site-municipality (20-100 km) -0.107*** -0.088*** -0.096*** -0.104*** -0.087*** -0.095*** -0.103*** -0.085*** -0.095*** (-6.98) (-5.50) (-6.89) (-6.77) (-5.41) (-6.79) (-6.74) (-5.32) (-6.84) distance school site-municipality (>100 km) -0.077 -0.078 -0.123* -0.069 -0.071 -0.118* -0.074 -0.075 -0.122* (-1.54) (-1.55) (-2.45) (-1.40) (-1.43) (-2.37) (-1.48) (-1.48) (-2.41) Total per student spending in the ETC 0.000*** 0.000*** 0.000*** (7.71) (6.66) (7.72) % low-SES students in the ETC (5th grade) (2) -0.176* -0.212** -0.169* (-2.32) (-2.68) (-2.21) total public enrollment ETC -0.000 -0.000 -0.000 (-0.34) (-0.48) (-0.31) Percentage of rural students in the ETC 0.194*** 0.213*** 0.190*** (3.47) (3.74) (3.40) Heterogeneous effects by availability of the next level 1 site: with lower secondary 0.044*** -0.136*** -0.128*** -0.078*** (3.50) (-9.45) (-8.24) (-4.11) 2-5 sites: no changes with integration (3) 0.065*** -0.088*** -0.083*** -0.024 (5.56) (-6.69) (-5.56) (-1.30) 2-5 sites: integration make available lower secondary 0.088*** -0.060*** -0.053*** 0.011 (7.87) (-4.79) (-3.74) (0.59) 6 or more sites: no changes with integration -0.048** -0.149*** -0.139*** -0.068*** (-3.28) (-9.52) (-8.42) (-3.32) 6 or more sites: integration make available lower secondary -0.033** -0.086*** -0.085*** -0.020 (-2.58) (-6.11) (-5.52) (-1.01) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.084*** -0.009 -0.005 0.046*** (11.13) (-1.18) (-0.54) (5.45) 2-5 sites & 2-5 km to principal site 0.066*** 0.010 0.016 0.074*** (6.87) (0.96) (1.44) (6.50) 2-5 sites & >5km to principal site -0.032* -0.019 -0.010 0.049** (-2.29) (-1.23) (-0.62) (3.03) 6 or more sites & <1km to principal site -0.023 -0.045*** -0.044*** 0.018 (-1.91) (-4.24) (-4.10) (1.55) 6 or more sites & 2-5 km to principal site -0.003 -0.031* -0.026* 0.036** (-0.28) (-2.47) (-2.02) (2.68) 6 or more sites & >5km to principal site -0.110*** -0.063*** -0.057*** 0.013 (-9.21) (-4.67) (-4.09) (0.88) constant 0.824*** 0.824*** 0.655*** 0.740*** 0.803*** 0.877*** 0.748*** 0.786*** 0.824*** 0.823*** 0.651*** 0.737*** (128.35) (48.16) (14.63) (33.30) (81.67) (46.33) (15.00) (28.88) (128.34) (48.06) (14.47) (33.09) Fixed effects ETC level No No No Si No No No Si No No No Si N37280 19629 19629 19629 37280 19629 19629 19629 37280 19629 19629 19629 r2 0.018 0.112 0.122 0.170 0.019 0.120 0.128 0.173 0.033 0.114 0.123 0.170 aic 32341.570 8904.795 8705.381 7783.882 32320.957 8744.851 8576.956 7716.400 31782.305 8885.781 8689.259 7770.541 bic 32367.149 9070.375 8902.500 8690.630 32372.115 8934.085 8797.730 8646.802 31841.988 9082.900 8917.917 8708.828 t stati s ti cs in parentheses *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) The categories omitted are: (3) It includes sites with lower secondary and those not having lower secondary in the school The categories omitted are: 1 site; quintile 1 SES (5th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no lower secondary 29 Table 13. Main regression for percentage of teachers over 40 years in primary (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites -0.027** -0.063*** -0.034** 0.038** (-2.85) (-6.25) (-3.19) (3.20) 6 or more sites -0.177*** -0.134*** -0.111*** -0.003 (-18.81) (-11.68) (-9.56) (-0.23) total enrollment -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** (-10.57) (-9.17) (-5.80) (-6.48) (-5.65) (-3.74) (-10.73) (-9.19) (-5.68) single-teacher -0.021* -0.026** -0.017* -0.038*** -0.040*** -0.027** -0.014 -0.023* -0.016 (-2.38) (-3.05) (-2.00) (-4.15) (-4.41) (-3.12) (-1.55) (-2.56) (-1.86) rural -0.071*** -0.090*** -0.090*** -0.066*** -0.084*** -0.082*** -0.070*** -0.089*** -0.090*** (-5.35) (-6.70) (-6.93) (-4.81) (-6.08) (-6.12) (-5.31) (-6.68) (-6.97) quintile 2 SES (5th grade) 0.027** 0.029** 0.035*** 0.030** 0.032** 0.037*** 0.025** 0.028** 0.035*** (2.75) (2.97) (3.74) (3.11) (3.28) (3.93) (2.61) (2.88) (3.71) quintilE 3 SES (5th grade) 0.038*** 0.045*** 0.064*** 0.042*** 0.065*** 0.038*** 0.045*** 0.064*** (3.64) (4.33) (6.33) (4.00) (4.60) (6.39) (3.58) (4.28) (6.30) quintilE 4 SES (5th grade) 0.033** 0.056*** 0.101*** 0.035** 0.057*** 0.100*** 0.033** 0.056*** 0.101*** (2.62) (4.39) (7.87) (2.76) (4.44) (7.76) (2.59) (4.35) (7.85) quintilE 5 SES (5th grade) 0.001 0.059** 0.121*** 0.011 0.064*** 0.119*** 0.002 0.058** 0.121*** (0.07) (3.09) (6.26) (0.61) (3.41) (6.18) (0.13) (3.09) (6.26) % of students belonging to an ethnic group in primary -0.051** -0.077*** -0.083*** -0.055** -0.081*** -0.086*** -0.054** -0.079*** -0.083*** (-3.05) (-4.43) (-4.38) (-3.26) (-4.66) (-4.57) (-3.22) (-4.52) (-4.40) % of students who have repeated a course in primary 0.023 0.058 0.066 0.015 0.047 0.066 0.026 0.059 0.066 (0.45) (1.17) (1.27) (0.29) (0.95) (1.27) (0.51) (1.19) (1.28) years of operation 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (7.07) (7.12) (6.54) (6.96) (7.07) (6.52) (6.98) (7.05) (6.52) % of students in a traditional model of teaching (1) 0.057*** 0.068*** 0.048*** 0.054*** 0.064*** 0.048*** 0.056*** 0.068*** 0.048*** (5.22) (6.21) (4.28) (4.98) (5.82) (4.27) (5.14) (6.13) (4.25) double-shift 0.055*** 0.091*** 0.050** 0.058*** 0.090*** 0.053** 0.052*** 0.089*** 0.050** (3.66) (5.96) (3.03) (3.86) (5.87) (3.23) (3.43) (5.79) (3.04) only morning shift 0.099*** 0.085*** 0.052*** 0.093*** 0.080*** 0.053*** 0.099*** 0.085*** 0.053*** (8.27) (6.92) (3.60) (7.75) (6.48) (3.63) (8.24) (6.93) (3.63) other type of school day 0.039** 0.041** 0.017 0.046** 0.046** 0.025 0.035* 0.039** 0.017 (2.68) (2.81) (1.13) (3.16) (3.14) (1.65) (2.39) (2.67) (1.13) distance school site-municipality (1-5 km) -0.013 0.026* 0.027* -0.014 0.023* 0.028* -0.009 0.026* 0.026* (-1.11) (2.17) (2.25) (-1.21) (1.98) (2.33) (-0.79) (2.15) (2.15) distance school site-municipality (5-20 km) -0.084*** -0.034** -0.034** -0.082*** -0.034** -0.033* -0.077*** -0.031* -0.033* (-6.76) (-2.62) (-2.60) (-6.59) (-2.65) (-2.48) (-6.17) (-2.39) (-2.52) distance school site-municipality (20-100 km) -0.181*** -0.127*** -0.110*** -0.177*** -0.126*** -0.109*** -0.174*** -0.123*** -0.109*** (-11.85) (-8.22) (-6.87) (-11.62) (-8.18) (-6.82) (-11.31) (-7.91) (-6.74) distance school site-municipality (>100 km) -0.121* -0.105 -0.054 -0.111 -0.098 -0.051 -0.122* -0.105 -0.054 (-2.09) (-1.80) (-0.93) (-1.95) (-1.71) (-0.89) (-2.11) (-1.80) (-0.93) Total per student spending in the ETC 0.000*** 0.000*** 0.000*** (8.88) (7.94) (8.77) % low-SES students in the ETC (5th grade) (2) 0.290*** 0.241** 0.283*** (3.72) (3.01) (3.60) total public enrollment ETC -0.000*** -0.000*** -0.000*** (-4.63) (-4.80) (-4.60) Percentage of rural students in the ETC 0.099 0.121* 0.101 (1.90) (2.29) (1.93) Heterogeneous effects by availability of the next level 1 site: with lower secondary -0.081*** -0.190*** -0.146*** -0.070** (-5.43) (-10.48) (-7.70) (-3.12) 2-5 sites: no changes with integration (3) -0.090*** -0.163*** -0.118*** -0.022 (-6.54) (-10.27) (-6.81) (-0.99) 2-5 sites: integration make available lower secondary -0.050*** -0.155*** -0.105*** -0.007 (-3.82) (-10.01) (-6.23) (-0.30) 6 or more sites: no changes with integration -0.239*** -0.259*** -0.214*** -0.086*** (-16.03) (-15.11) (-12.00) (-3.53) 6 or more sites: integration make available lower secondary -0.209*** -0.200*** -0.162*** -0.036 (-16.15) (-13.23) (-10.07) (-1.55) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.004 -0.052*** -0.029** 0.037** (0.38) (-4.99) (-2.64) (3.12) 2-5 sites & 2-5 km to principal site -0.034** -0.071*** -0.036* 0.041* (-2.81) (-5.05) (-2.46) (2.56) 2-5 sites & >5km to principal site -0.108*** -0.089*** -0.052** 0.034 (-7.28) (-4.98) (-2.86) (1.73) 6 or more sites & <1km to principal site -0.124*** -0.121*** -0.106*** -0.002 (-10.53) (-9.55) (-8.43) (-0.15) 6 or more sites & 2-5 km to principal site -0.143*** -0.134*** -0.106*** 0.002 (-13.09) (-9.29) (-7.18) (0.14) 6 or more sites & >5km to principal site -0.226*** -0.158*** -0.126*** -0.010 (-21.72) (-11.09) (-8.79) (-0.57) constant 0.661*** 0.696*** 0.177*** 0.419*** 0.699*** 0.781*** 0.294*** 0.466*** 0.661*** 0.693*** 0.182*** 0.420*** (87.61) (30.05) (3.35) (16.57) (61.56) (30.72) (5.02) (15.02) (87.60) (30.10) (3.43) (16.45) Fixed effects ETC level No No No Si No No No Si No No No Si N36721 19629 19629 19629 36721 19629 19629 19629 36721 19629 19629 19629 r2 0.027 0.074 0.089 0.123 0.028 0.080 0.093 0.124 0.035 0.075 0.090 0.123 aic 46268.532 22254.275 21932.743 21378.599 46215.125 22139.330 21862.444 21352.674 45949.831 22244.553 21933.430 21384.922 bic 46294.065 22419.855 22129.862 22285.346 46266.192 22328.565 22083.218 22283.076 46009.409 22441.672 22162.088 22323.208 t stati s ti cs in parentheses *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) The categories omitted are: (3) It includes sites with lower secondary and those not having lower secondary in the school The categories omitted are: 1 site; quintile 1 SES (5th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no lower secondary 30 Table 14. Main regression for percentage of teachers over 40 years in lower secondary (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites 0.004 0.012 0.011 0.024 (0.38) (0.96) (0.90) (1.95) 6 or more sites -0.146*** -0.042** -0.047** -0.000 (-12.08) (-2.97) (-3.25) (-0.02) total enrollment -0.000** -0.000* -0.000 -0.000** -0.000* -0.000 -0.000** -0.000* -0.000 (-3.01) (-2.28) (-0.64) (-2.92) (-2.15) (-0.54) (-3.05) (-2.28) (-0.60) single-teacher 0.029 0.043 0.031 0.024 0.034 0.022 0.017 0.026 0.016 (1.01) (1.46) (1.07) (0.77) (1.09) (0.72) (0.55) (0.84) (0.51) rural -0.064*** -0.070*** -0.055*** -0.064*** -0.071*** -0.055*** -0.063*** -0.069*** -0.054*** (-4.44) (-4.81) (-3.71) (-4.49) (-4.85) (-3.68) (-4.37) (-4.76) (-3.63) quintile 2 SES (9th grade) 0.049*** 0.049*** 0.070*** 0.050*** 0.050*** 0.071*** 0.050*** 0.050*** 0.072*** (3.79) (3.67) (5.31) (3.85) (3.76) (5.38) (3.84) (3.75) (5.41) quintilE 3 SES (9th grade) 0.046** 0.049** 0.100*** 0.046** 0.050** 0.100*** 0.047** 0.051** 0.102*** (2.89) (3.00) (5.70) (2.92) (3.04) (5.74) (2.96) (3.08) (5.82) quintilE 4 SES (9th grade) 0.026 0.046* 0.138*** 0.026 0.047* 0.139*** 0.026 0.047* 0.140*** (1.41) (2.27) (6.29) (1.44) (2.33) (6.31) (1.41) (2.32) (6.36) quintilE 5 SES (9th grade) -0.015 0.008 0.116** -0.012 0.011 0.118** -0.015 0.009 0.118** (-0.43) (0.21) (3.11) (-0.36) (0.29) (3.15) (-0.42) (0.25) (3.17) % of students belonging to an ethnic group in lower secondary -0.015 -0.028 -0.034 -0.016 -0.028 -0.033 -0.014 -0.027 -0.032 (-0.79) (-1.43) (-1.43) (-0.84) (-1.46) (-1.41) (-0.76) (-1.38) (-1.38) % of students who have repeated a course in lower secondary -0.065 -0.050 0.056 -0.061 -0.047 0.059 -0.059 -0.044 0.063 (-0.97) (-0.76) (0.80) (-0.90) (-0.70) (0.83) (-0.86) (-0.66) (0.88) years of operation 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (4.35) (4.20) (3.47) (4.28) (4.13) (3.43) (4.34) (4.20) (3.50) % of students in a traditional model of teaching (1) 0.108*** 0.113*** 0.079*** 0.109*** 0.115*** 0.082*** 0.109*** 0.115*** 0.081*** (7.96) (8.23) (5.53) (7.98) (8.30) (5.66) (7.89) (8.24) (5.57) double-shift 0.099*** 0.101*** 0.065*** 0.099*** 0.102*** 0.066*** 0.100*** 0.103*** 0.067*** (5.51) (5.49) (3.36) (5.53) (5.52) (3.41) (5.54) (5.58) (3.44) only morning shift 0.104*** 0.086*** 0.047* 0.104*** 0.086*** 0.048** 0.105*** 0.087*** 0.048** (6.23) (5.12) (2.55) (6.24) (5.15) (2.60) (6.28) (5.21) (2.63) other type of school day 0.045** 0.033* 0.004 0.046** 0.034* 0.006 0.046** 0.035* 0.007 (2.71) (1.99) (0.25) (2.75) (2.07) (0.35) (2.77) (2.13) (0.39) distance school site-municipality (1-5 km) 0.002 0.011 0.003 0.002 0.011 0.003 0.002 0.012 0.003 (0.16) (0.82) (0.23) (0.19) (0.84) (0.24) (0.19) (0.86) (0.24) distance school site-municipality (5-20 km) -0.027* -0.004 -0.014 -0.027* -0.003 -0.014 -0.028* -0.004 -0.015 (-2.01) (-0.25) (-0.91) (-2.01) (-0.24) (-0.92) (-2.08) (-0.29) (-0.95) distance school site-municipality (20-100 km) -0.144*** -0.121*** -0.101*** -0.143*** -0.120*** -0.101*** -0.144*** -0.121*** -0.101*** (-8.64) (-6.86) (-5.42) (-8.62) (-6.81) (-5.39) (-8.68) (-6.85) (-5.41) distance school site-municipality (>100 km) -0.133** -0.125** -0.095* -0.132** -0.124** -0.094* -0.134** -0.126** -0.096* (-3.13) (-2.94) (-2.11) (-3.11) (-2.91) (-2.08) (-3.16) (-2.95) (-2.11) Total per student spending in the ETC 0.000*** 0.000*** 0.000*** (3.74) (3.68) (3.77) % low-SES students in the ETC (5th grade) (2) 0.047 0.048 0.054 (0.61) (0.62) (0.70) total public enrollment ETC -0.000*** -0.000*** -0.000*** (-5.78) (-5.88) (-5.79) Percentage of rural students in the ETC 0.039 0.039 0.038 (0.75) (0.75) (0.73) Heterogeneous effects by availability of the next level 1 site: with upper secondary 0.114** -0.065 -0.064 -0.036 (3.23) (-1.38) (-1.35) (-0.81) 2-5 sites: no changes with integration (3) 0.104** -0.048 -0.047 -0.009 (3.03) (-1.04) (-1.02) (-0.21) 2-5 sites: integration make available upper secondary 0.114** -0.080 -0.070 -0.018 (2.85) (-1.40) (-1.22) (-0.33) 6 or more sites: no changes with integration -0.039 -0.104* -0.107* -0.036 (-1.13) (-2.24) (-2.32) (-0.80) 6 or more sites: integration make available upper secondary -0.070 -0.084 -0.077 -0.005 (-1.83) (-1.59) (-1.47) (-0.09) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.012 0.013 0.012 0.024* (1.08) (1.08) (0.98) (1.99) 2-5 sites & 2-5 km to principal site -0.016 0.010 0.018 0.054 (-0.53) (0.26) (0.47) (1.45) 2-5 sites & >5km to principal site -0.158*** -0.053 -0.046 -0.042 (-4.14) (-1.11) (-0.97) (-0.88) 6 or more sites & <1km to principal site -0.119*** -0.046** -0.052*** -0.005 (-9.77) (-3.25) (-3.60) (-0.32) 6 or more sites & 2-5 km to principal site -0.187*** -0.068 -0.062 -0.013 (-7.84) (-1.88) (-1.69) (-0.37) 6 or more sites & >5km to principal site -0.202*** 0.004 0.009 0.054 (-10.54) (0.14) (0.27) (1.67) constant 0.592*** 0.440*** 0.302*** 0.319*** 0.492*** 0.499*** 0.359*** 0.346*** 0.592*** 0.439*** 0.294*** 0.313*** (61.42) (16.82) (5.36) (10.45) (14.56) (9.70) (4.92) (6.68) (61.40) (16.64) (5.17) (10.06) Fixed effects ETC level No No No Si No No No Si No No No Si N8970 6782 6782 6782 8970 6782 6782 6782 8970 6782 6782 6782 r2 0.038 0.109 0.115 0.171 0.040 0.109 0.115 0.171 0.045 0.110 0.116 0.172 aic 7726.632 4777.013 4740.011 4478.556 7714.851 4778.890 4741.510 4481.832 7666.628 4775.935 4738.220 4475.258 bic 7747.937 4920.275 4910.562 5263.090 7757.460 4942.619 4932.527 5286.831 7716.339 4946.486 4936.059 5287.080 t stati s ti cs in parenthe ses *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) The categories omitted are: (3) It includes sites with upper secondary and those not having upper secondary in the school The categories omitted are: 1 site; quintile 1 SES (9th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no upper secondary 31 Table 15. Main regression for existence of access to the internet in the site (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites 0.097*** 0.059*** 0.022* 0.004 (8.38) (5.59) (2.04) (0.27) 6 or more sites -0.233*** -0.034** -0.060*** -0.054*** (-21.54) (-2.87) (-5.07) (-3.42) total enrollment 0.000*** 0.000*** 0.000*** 0.000*** 0.000** 0.000*** 0.000*** 0.000*** 0.000*** (11.28) (8.97) (9.01) (4.17) (2.65) (3.97) (10.82) (8.12) (7.63) single-teacher -0.230*** -0.218*** -0.201*** -0.190*** -0.184*** -0.173*** -0.217*** -0.199*** -0.180*** (-22.71) (-21.48) (-20.05) (-17.67) (-17.41) (-16.63) (-20.60) (-18.88) (-17.29) rural -0.083*** -0.065*** -0.060*** -0.116*** -0.097*** -0.090*** -0.082*** -0.064*** -0.060*** (-5.29) (-4.15) (-3.88) (-7.25) (-6.07) (-5.69) (-5.25) (-4.05) (-3.87) quintile 2 SES (5th grade) 0.066*** 0.067*** 0.071*** 0.062*** 0.062*** 0.066*** 0.065*** 0.065*** 0.068*** (6.83) (6.94) (7.47) (6.37) (6.46) (7.02) (6.70) (6.73) (7.19) quintilE 3 SES (5th grade) 0.097*** 0.094*** 0.100*** 0.097*** 0.094*** 0.100*** 0.096*** 0.093*** 0.099*** (8.73) (8.69) (9.36) (8.82) (8.70) (9.39) (8.72) (8.62) (9.27) quintilE 4 SES (5th grade) 0.145*** 0.130*** 0.128*** 0.154*** 0.135*** 0.134*** 0.145*** 0.129*** 0.128*** (10.10) (9.29) (9.47) (10.76) (9.67) (9.89) (10.14) (9.22) (9.45) quintilE 5 SES (5th grade) 0.187*** 0.136*** 0.122*** 0.197*** 0.139*** 0.130*** 0.190*** 0.135*** 0.123*** (10.94) (7.66) (6.96) (11.50) (7.85) (7.47) (11.08) (7.63) (7.00) % of students belonging to an ethnic group in primary -0.052** -0.023 -0.062** -0.050** -0.020 -0.053** -0.057*** -0.028 -0.063** (-3.10) (-1.31) (-3.13) (-2.95) (-1.14) (-2.62) (-3.34) (-1.62) (-3.15) % of students who have repeated a course in primary 0.086 0.033 0.120* 0.098 0.045 0.119* 0.091 0.036 0.124* (1.37) (0.54) (2.25) (1.60) (0.75) (2.24) (1.46) (0.60) (2.34) years of operation 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (5.90) (5.81) (3.45) (5.70) (5.60) (3.36) (5.86) (5.77) (3.36) % of students in a traditional model of teaching (1) 0.019 0.006 0.054*** 0.029* 0.014 0.056*** 0.017 0.002 0.048*** (1.46) (0.48) (4.36) (2.27) (1.13) (4.54) (1.34) (0.18) (3.89) double-shift 0.081*** 0.060*** 0.048** 0.066*** 0.043* 0.034 0.073*** 0.047** 0.035 (4.44) (3.39) (2.58) (3.69) (2.42) (1.83) (3.98) (2.60) (1.90) only morning shift -0.067*** -0.027* -0.037* -0.066*** -0.032* -0.039** -0.070*** -0.031* -0.038* (-4.69) (-2.08) (-2.44) (-4.67) (-2.45) (-2.58) (-4.89) (-2.37) (-2.51) other type of school day 0.052** 0.076*** 0.059*** 0.026 0.047** 0.035* 0.042* 0.061*** 0.044* (2.85) (4.44) (3.39) (1.45) (2.76) (1.98) (2.29) (3.57) (2.55) distance school site-municipality (1-5 km) 0.036** -0.006 0.008 0.036** -0.008 0.004 0.050*** 0.010 0.028* (2.96) (-0.49) (0.62) (3.00) (-0.67) (0.35) (4.01) (0.80) (2.16) distance school site-municipality (5-20 km) 0.036** -0.027* -0.011 0.034** -0.029* -0.015 0.050*** -0.011 0.007 (2.83) (-2.05) (-0.79) (2.68) (-2.28) (-1.12) (3.79) (-0.80) (0.54) distance school site-municipality (20-100 km) -0.022 -0.091*** -0.058*** -0.025 -0.093*** -0.060*** -0.012 -0.079*** -0.046** (-1.40) (-5.68) (-3.53) (-1.63) (-5.86) (-3.66) (-0.73) (-4.85) (-2.71) distance school site-municipality (>100 km) -0.070 -0.087 -0.041 -0.079 -0.096* -0.049 -0.075 -0.096 -0.049 (-1.47) (-1.78) (-0.86) (-1.76) (-2.06) (-1.07) (-1.55) (-1.89) (-1.01) Total per student spending in the ETC -0.000*** -0.000*** -0.000*** (-12.87) (-11.45) (-13.21) % low-SES students in the ETC (5th grade) (2) -0.233** -0.286** -0.282** (-2.71) (-3.21) (-3.25) total public enrollment ETC 0.000*** 0.000*** 0.000*** (9.38) (8.96) (9.54) Percentage of rural students in the ETC -0.101 -0.094 -0.087 (-1.76) (-1.61) (-1.51) Heterogeneous effects by availability of the next level 1 site: with lower secondary 0.620*** 0.158*** 0.103*** 0.070* (43.27) (8.42) (5.19) (2.54) 2-5 sites: no changes with integration (3) 0.479*** 0.176*** 0.113*** 0.079** (34.10) (10.11) (5.98) (2.76) 2-5 sites: integration make available lower secondary 0.339*** 0.098*** 0.033 0.012 (25.37) (5.83) (1.85) (0.42) 6 or more sites: no changes with integration 0.231*** 0.147*** 0.085*** 0.073* (14.64) (7.98) (4.52) (2.39) 6 or more sites: integration make available lower secondary 0.016 -0.027 -0.072*** -0.067* (1.32) (-1.72) (-4.41) (-2.28) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.237*** 0.077*** 0.045*** 0.017 (20.25) (7.04) (4.01) (1.28) 2-5 sites & 2-5 km to principal site -0.030* 0.038* -0.011 -0.053** (-1.97) (2.39) (-0.67) (-2.91) 2-5 sites & >5km to principal site -0.141*** 0.033 -0.019 -0.053* (-8.60) (1.70) (-0.95) (-2.47) 6 or more sites & <1km to principal site -0.092*** 0.001 -0.015 -0.021 (-6.73) (0.08) (-1.15) (-1.28) 6 or more sites & 2-5 km to principal site -0.239*** -0.064*** -0.100*** -0.120*** (-18.75) (-4.09) (-6.61) (-6.33) 6 or more sites & >5km to principal site -0.298*** -0.060*** -0.101*** -0.111*** (-26.16) (-4.13) (-6.96) (-5.96) constant 0.499*** 0.490*** 1.103*** 0.727*** 0.204*** 0.449*** 1.077*** 0.706*** 0.499*** 0.483*** 1.135*** 0.742*** (54.56) (19.05) (18.51) (23.61) (20.05) (16.08) (16.33) (17.89) (54.55) (18.73) (18.89) (23.95) Fixed effects ETC level No No No Si No No No Si No No No Si N37280 19629 19629 19629 37280 19629 19629 19629 37280 19629 19629 19629 r2 0.095 0.276 0.303 0.342 0.154 0.290 0.314 0.350 0.142 0.278 0.306 0.346 aic 47966.594 22071.371 21343.718 20385.985 45423.766 21690.808 21039.727 20155.700 45982.264 22029.096 21254.405 20271.094 bic 47992.173 22236.951 21540.837 21292.732 45474.924 21880.042 21260.500 21086.102 46041.947 22226.215 21483.063 21209.381 t stati s ti cs in parentheses *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) The categories omitted are: (3) It includes sites with lower secondary and those not having lower secondary in the school The categories omitted are: 1 site; quintile 1 SES (5th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no lower secondary 32 Table 16. Main regression for existence of access to computer room in the site (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 2-5 sites -0.030** -0.081*** -0.063*** -0.012 (-2.94) (-8.17) (-6.00) (-1.27) 6 or more sites -0.325*** -0.162*** -0.148*** -0.079*** (-30.84) (-13.39) (-12.05) (-5.98) total enrollment 0.000*** 0.000*** 0.000*** 0.000* 0.000 0.000** 0.000*** 0.000*** 0.000*** (6.34) (6.12) (9.24) (2.29) (1.71) (2.87) (4.96) (4.68) (7.41) single-teacher -0.204*** -0.211*** -0.210*** -0.180*** -0.185*** -0.180*** -0.173*** -0.180*** -0.179*** (-19.73) (-20.66) (-20.90) (-16.59) (-17.29) (-17.16) (-16.07) (-16.95) (-17.08) rural -0.013 -0.023 -0.025 -0.048*** -0.059*** -0.063*** -0.013 -0.022 -0.027 (-1.01) (-1.66) (-1.82) (-3.57) (-4.16) (-4.44) (-0.98) (-1.59) (-1.93) quintile 2 SES (5th grade) 0.100*** 0.100*** 0.102*** 0.100*** 0.098*** 0.098*** 0.096*** 0.096*** 0.097*** (9.66) (9.63) (10.04) (9.73) (9.54) (9.72) (9.35) (9.29) (9.60) quintilE 3 SES (5th grade) 0.133*** 0.132*** 0.135*** 0.140*** 0.137*** 0.138*** 0.132*** 0.131*** 0.133*** (11.65) (11.70) (12.29) (12.37) (12.17) (12.55) (11.61) (11.56) (12.05) quintilE 4 SES (5th grade) 0.143*** 0.145*** 0.146*** 0.159*** 0.154*** 0.153*** 0.144*** 0.143*** 0.144*** (10.37) (10.32) (10.22) (11.55) (11.01) (10.83) (10.47) (10.19) (10.06) quintilE 5 SES (5th grade) 0.098*** 0.098*** 0.114*** 0.128*** 0.113*** 0.127*** 0.104*** 0.098*** 0.115*** (5.99) (5.38) (6.19) (7.90) (6.29) (6.96) (6.41) (5.35) (6.26) % of students belonging to an ethnic group in primary -0.120*** -0.123*** -0.125*** -0.124*** -0.125*** -0.113*** -0.131*** -0.132*** -0.126*** (-6.79) (-6.77) (-5.82) (-6.96) (-6.86) (-5.27) (-7.30) (-7.17) (-5.87) % of students who have repeated a course in primary -0.013 0.005 0.022 -0.012 0.004 0.020 -0.002 0.011 0.029 (-0.23) (0.08) (0.38) (-0.20) (0.07) (0.34) (-0.03) (0.19) (0.50) years of operation 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (7.06) (6.90) (5.52) (6.49) (6.52) (5.29) (6.93) (6.80) (5.40) % of students in a traditional model of teaching (1) 0.101*** 0.104*** 0.093*** 0.110*** 0.109*** 0.098*** 0.098*** 0.098*** 0.086*** (7.68) (7.96) (7.05) (8.42) (8.39) (7.44) (7.53) (7.63) (6.51) double-shift 0.042** 0.069*** 0.075*** 0.028 0.042* 0.055** 0.027 0.050** 0.060** (2.59) (4.00) (3.90) (1.70) (2.43) (2.83) (1.66) (2.90) (3.14) only morning shift 0.014 0.015 0.017 0.004 -0.001 0.013 0.009 0.010 0.017 (0.91) (0.91) (0.90) (0.25) (-0.08) (0.67) (0.57) (0.62) (0.88) other type of school day 0.080*** 0.087*** 0.090*** 0.058** 0.054** 0.061** 0.058*** 0.065*** 0.070*** (4.52) (4.80) (4.64) (3.27) (2.96) (3.12) (3.30) (3.57) (3.61) distance school site-municipality (1-5 km) 0.019 0.033* 0.014 0.016 0.026 0.010 0.046*** 0.055*** 0.039** (1.44) (2.37) (1.02) (1.27) (1.85) (0.70) (3.48) (3.95) (2.68) distance school site-municipality (5-20 km) -0.016 0.002 -0.011 -0.015 -0.003 -0.015 0.017 0.028 0.017 (-1.10) (0.10) (-0.69) (-1.02) (-0.17) (-0.98) (1.19) (1.81) (1.04) distance school site-municipality (20-100 km) -0.112*** -0.092*** -0.078*** -0.108*** -0.093*** -0.079*** -0.083*** -0.070*** -0.054** (-6.53) (-5.12) (-4.13) (-6.34) (-5.19) (-4.21) (-4.80) (-3.83) (-2.85) distance school site-municipality (>100 km) -0.050 -0.048 0.047 -0.045 -0.048 0.040 -0.058 -0.058 0.038 (-0.93) (-0.89) (0.89) (-0.83) (-0.87) (0.75) (-1.12) (-1.11) (0.73) Total per student spending in the ETC 0.000*** 0.000*** 0.000*** (6.25) (6.76) (5.70) % low-SES students in the ETC (5th grade) (2) 0.051 -0.115 -0.017 (0.58) (-1.27) (-0.20) total public enrollment ETC 0.000 0.000 0.000 (1.64) (0.72) (1.94) Percentage of rural students in the ETC 0.062 0.111 0.077 (1.02) (1.80) (1.29) Heterogeneous effects by availability of the next level 1 site: with lower secondary 0.300*** -0.126*** -0.106*** -0.067** (20.77) (-7.13) (-5.59) (-2.97) 2-5 sites: no changes with integration (3) 0.205*** -0.102*** -0.079*** -0.038 (13.54) (-5.80) (-4.10) (-1.69) 2-5 sites: integration make available lower secondary 0.058*** -0.194*** -0.172*** -0.130*** (3.95) (-11.17) (-9.08) (-5.86) 6 or more sites: no changes with integration -0.046** -0.142*** -0.117*** -0.055* (-2.63) (-7.31) (-5.76) (-2.21) 6 or more sites: integration make available lower secondary -0.219*** -0.270*** -0.257*** -0.213*** (-14.94) (-15.40) (-13.74) (-8.96) Heterogeneous effects by distance to the main site 2-5 sites & < 1km to principal site 0.098*** -0.047*** -0.035*** -0.002 (10.08) (-4.96) (-3.51) (-0.17) 2-5 sites & 2-5 km to principal site -0.123*** -0.107*** -0.090*** -0.059*** (-8.83) (-6.92) (-5.65) (-3.75) 2-5 sites & >5km to principal site -0.279*** -0.151*** -0.133*** -0.098*** (-16.88) (-7.80) (-6.78) (-5.06) 6 or more sites & <1km to principal site -0.139*** -0.079*** -0.071*** -0.021 (-10.01) (-5.96) (-5.32) (-1.54) 6 or more sites & 2-5 km to principal site -0.314*** -0.206*** -0.194*** -0.148*** (-24.89) (-12.82) (-11.87) (-8.62) 6 or more sites & >5km to principal site -0.422*** -0.242*** -0.228*** -0.180*** (-35.89) (-15.57) (-14.40) (-10.66) constant 0.753*** 0.697*** 0.379*** 0.592*** 0.610*** 0.794*** 0.549*** 0.692*** 0.753*** 0.679*** 0.424*** 0.610*** (95.40) (27.19) (6.55) (21.15) (49.51) (27.52) (8.41) (20.41) (95.40) (26.77) (7.35) (21.79) Fixed effects ETC level No No No Si No No No Si No No No Si N37280 19629 19629 19629 37280 19629 19629 19629 37280 19629 19629 19629 r2 0.084 0.235 0.240 0.259 0.109 0.247 0.252 0.271 0.140 0.246 0.250 0.269 aic 50689508.0 21611349.0 21490828.0 21164512.0 49631728.0 21320789.0 21190725.0 20849694.0 48340910.0 21337066.0 21240089.0 20913091.0 bic 50715086.0 21776929.0 21687947.0 22071260.0 49682885.0 21510024.0 21411498.0 21780096.0 48400593.0 21534185.0 21468747.0 21851378.0 t stati s ti cs in parenthe ses *p<0.05; ** p<0.01; *** p<0.001 Notes: ETC: Certified Territorial Entity (1) The opposite of the traditional model is a flexible education model, which has specific educational materials and defined training processes. These models are designed to serve diverse populations, in situations of vulnerability and who have difficulty accessing the regular model (e.g. Escuela Nueva, Post-Primaria, Telesecundaria). (2) low-SES student: student whose mother didn’t complete secondary education (5th grade) The categories omitted are: (3) It includes sites with lower secondary and those not having lower secondary in the school The categories omitted are: 1 site; quintile 1 SES (5th grade); full-time school day; distance school site-municipality (< 1 km); 1 site: no lower secondary 33 Appendix Table A1. Comparison of the sample of sites used for the estimation in Table 3 (SABER 5 Language) (≥ 6 students) with the unrestricted sample Variable Restricted Unrestricted Without controling SES (>=6 students) Mean Mean Mean p_insuficientelenguaje_5_grado 18.0% 16.8% 18.6% 2-5 sites 56.1% 38.4% 54.8% 6 or more sites 33.6% 56.8% 36.4% total enrollment 525 262 486 single-teacher 3.3% 42.2% 3.8% rural 47.4% 74.8% 46.7% quintile 2 SES (5th grade) 16.1% 24.4% - quintile 3 SES (5th grade) 24.9% 25.4% - quintile 4 SES (5th grade) 38.3% 23.5% - quintile 5 SES (5th grade) 11.9% 6.0% - % of students belonging to an ethnic group in primary 10.1% 8.7% 10.6% % of students who have repeated a course in primary 2.9% 2.8% 2.9% years of operation 31 28 30 % of students in a traditional model of teaching 77.2% 43.1% 76.5% double-shift 41.7% 19.7% 40.2% only morning shift 31.1% 43.2% 33.3% other type of school day 16.3% 11.5% 15.7% distance school site-municipality (1-5 km) 29.8% 27.7% 29.6% distance school site-municipality (5-20 km) 37.5% 45.8% 36.9% distance school site-municipality (20-100 km) 10.9% 13.4% 11.1% distance school site-municipality (>100 km) 0.8% 0.6% 0.8% Total per student spending in the ETC 2,926,843 2,999,635 2,926,128 % low-SES students in the ETC (5th grade) (2) 52.6% 56.8% 52.6% total public enrollment ETC 197,875 197,716 198,769 Percentage of rural students in the ETC 34.0% 39.3% 34.1% Number of sites 6,237 13,581 8,303 34 References Abu-Ghaida, D.; Alonso, J.D., and Sánchez, A. (2011). Reforming Education Financing Systems in Transition Countries: An Introduction to Six Case Studies, in J.D. Alonso and A. Sánchez (eds.), Reforming Education Finance in Transition Countries: Six Case Studies in Per capita Financing Schemes. World Bank, Washington, DC. doi/abs/10.1596/978-0-8213-8783-2 Ares Abalde, M. (2014). School Size Policies: A Literature Review. OECD Education Working Papers, No. 106, OECD Publishing, Paris. https://doi.org/10.1787/5jxt472ddkjl-en. Ayala, M.C. (2017). Efecto de los docentes provisionales sobre desempeño académico: Evidencia para la educación secundaria oficial en Colombia. Universidad de los Andes. Retrieved from: http://2017.economicsofeducation.com/user/pdfsesiones/159.pdf?PHPSESSID=so7ujho24h9n1v4iv2pn3 1fvo2 Bertoni, E.; Elacqua, G.; Jaimovich, A.; Rodriguez, J.; and Santos, H. (2018). Teacher Policies, Incentives, and Labor Markets in Chile, Colombia, and Peru: Implications for Equality. IDB Working Paper Series Nº IDB-WP-00945. Available at https://publications.iadb.org/en/teacher-policies-incentives-and-labormarkets-chile-colombia-and-peru-implications-equality Brutti, Z.; and Sanchez, F. (2016). New Teachers for Colombia: Is Quality Control Working? Draft. Universidad de Los Andes. Retrieved from: https://lacer.lacea.org/bitstream/handle/123456789/64418/lacea2016_new_teachers_colombia.pdf?se quence=1 Chapman, C.; and Muijs, D. (2014). Does school-to-school collaboration promote school improvement? A study of the impact of school federations on student outcomes. School Effectiveness and School Improvement, 25(3), 351-393. DOI: 10.1080/09243453.2013.840319 Chapman, C. (2015). From one school to many: Reflections on the impact and nature of school federations and chains in England. Educational Management Administration and Leadership, 43(1), 46-60. https://doi.org/10.1177/1741143213494883 Corpoeducación. (2004). La Nueva Institución Educativa: Orientaciones para su integración. Econometría. (2013). Estudio de los efectos de la fusión de establecimientos educativos en contextos rurales en cuatro entidades territoriales. Retrieved from: http://aprende.colombiaaprende.edu.co/sites/default/files/naspublic/per/documentos/5.%20Estudio_F usion_Sedes_final.pdf Elacqua, G.; Martínez, M.; Santos, H.; and Urbina, D. (2012). School closures in Chile. Access to quality alternatives in a school choice system. Estudios de Economía, 39(2), 179-202. https://estudiosdeeconomia.uchile.cl/index.php/EDE/article/view/31042/43315 35 García, S.; Maldonado, D.; Perry, G. E.; Rodríguez, C.; and Saavedra, J. E. (2014). Tras la excelencia docente: ¿Cómo mejorar la calidad de la educación para todos los colombianos? Bogotá: Fundación Compartir. Retrieved from: http://www.fundacioncompartir.org/pdf/Traslaexcelenciadocente18.02.2014.pdf Giordano, E. (2008). School clusters and teacher resource centres. UNESCO International Institute for Educational Planning. Retrieved from: http://unesdoc.unesco.org/images/0015/001597/159776e.pdf Heckman J.J. (1978). Dummy endogenous variables in a simultaneous equations system. Econometrica, 46 (4), 931–960. DOI: 10.2307/1909757 Instituto Colombiano para la Evaluación de la Educación. (2016). Guía de Interpretación y Uso de Resultados de las pruebas Saber 3°, 5° y 9°. Stiefel, L.; Schwartz, A.E.; Iatarola, P.; and Chellman, C.C. (2009). Mission matters: The cost of small high schools revisited. Economics of Education Review, 28 (5), 585-599. https://doi.org/10.1016/j.econedurev.2009.01.005 Lee L.F. (1983). Generalized Econometric Models with Selectivity. Econometrica, 51, 507-512. DOI: 10.2307/1912003 Ministerio de Educación Nacional (2005). Guía No. 9 Guía Certificación de Municipios Menores de Cien Mil Habitantes. Retrieved from: https://www.mineducacion.gov.co/1759/w3-article-81012.html Andrews, M.; Duncombe, W.; and Yinger, J. (2002). Revisiting economies of size in American education: are we any closer to a consensus? Economics of Education Review, 21 (3), 245-262. https://doi.org/10.1016/S0272-7757(01)00006-1 Matthews, P.; Klaver, E.; Lannert, J.; Conluain, G.; and Ventura, A. (2009). Policy Measures Implemented in the First Cycle of Compulsory Education in Portugal: International Evaluation for the Ministry of Education. Office for Education Statistics and Planning, Ministry of Education, Portugal. Retrieved from: http://www.oecd.org/education/school/42065538.pdf Muijs, D. (2008). Widening opportunities? A case study of school to school collaboration in a rural district. Improving Schools, 11(1): 61–73. DOI: https://doi.org/10.1007/978-94-007-0283-7_9 Nitta, K.; Holley, M.; and Wrobel, S. (2010). A phenomenological study of rural school consolidation. Journal of Research in Rural Education, 25 (2), pp. 1-19. Retrieved from http://sites.psu.edu/jrre/wpcontent/uploads/sites/6347/2014/02/25-2.pdf OECD. (2017). Organisation of the school offer. OECD Review of Policies to Improve the Effectiveness of Resource Use in Schools. Ome, A. (2013). El Estatuto de profesionalización docente: Una primera evaluación. Cuadernos de Fedesarrollo, 43. Retreived from: https://www.repository.fedesarrollo.org.co/handle/11445/156 36 Sánchez, F.; Velasco, T.; Ayala, M.C.; and Pulido, X. (2016). Trayectorias de Permanencia, Deserción y Repitencia en la Educación Secundaria Colombiana y Sus Factores Asociados. Documentos de Trabajo No. 36. Available at http://dx.doi.org/10.2139/ssrn.2927172