Bolivia is a lower-middle-income and predominantly urban developing country. In 2022, Bolivia had a GDP per capita of approximately $3 600 and was ranked 118th on the United Nations Human Development Index (HDI = 0.692), with a literacy rate of 94 percent. About 70% of the population lives in urban areas. In 2020, public expenditure on education represented 9.8% of GDP, significantly higher than the average for Latin America at 4.6% and the global average of 4.3 percent.¹

Despite a steady increase in public expenditure on education since 2010, the quality of the infrastructure in schools at all levels -i. e., at pre-primary through tertiary levels- still lags behind what is observed in many countries with similar development levels. This is particularly evident in rural communities, where, as ^{Regalsky and Laurie (2007)} note, school facilities often consist of simple adobe dwellings.

• ^{Regalsky and Laurie (2007)}
  The School, Whose Place Is This’? The Deep Structures of the Hidden Curriculum in Indigenous Education in Bolivia

  Comparative Education, 2007
  The School, Whose Place Is This’? The Deep Structures of the Hidden Curriculum in Indigenous Education in Bolivia

Educational outcomes are likely the result of many factors, such as parental support and availability, the student’s socioeconomic background, and their natural disposition toward learning. In addition to the previously mentioned factors, however, access to infrastructure facilities and services has also been shown to play an important role, especially in developing countries like Bolivia, where the quality of basic school infrastructure is often suboptimal.

The significance of school infrastructure cannot be overstated. It plays a crucial role in shaping a country’s educational landscape and fostering a conducive learning environment for students. Proper school infrastructure ensures that students have access to safe and well-equipped facilities, including classrooms, libraries, laboratories, and recreational areas, which are essential for their overall development. Additionally, well-maintained facilities enhance motivation and engagement for both students and teachers, enhancing the learning experience. By investing in modern and inclusive infrastructure, Bolivia can bridge the educational gap between urban and rural areas, promoting equal opportunities for all children to receive a quality education. Hence the timeliness of this study that uses the largest dataset of school institutions in the country to determine whether school infrastructure generates the aforementioned effects on key educational outcomes.

The objective of this study is to analyze the impact of school infrastructure on educational outcomes in the pre-primary, primary, and secondary levels during the 2015-2020 period. More specifically, how enrollment, promotion, abandonment, and failure levels are affected by access to basic infrastructure services like electricity, running water, and restroom facilities; and by access to basic infrastructure facilities like classrooms, sports fields, laboratories, libraries, and computer rooms. We also test whether school outcomes are affected by attending schools in urban centers; attending public schools; and offering a humanistic degree. The analysis is done at the aggregate level; by geographical region; and by school grade levels.

Even though the objective described above is clear, it is important to point out that education at all levels in Bolivia is a complex phenomenon likely to be affected by more than just available physical infrastructure. Among the many factors not explicitly accounted for in this study, the following are a few that ought to be considered when analyzing education in Bolivia: i) Urban vs. Rural Disparities: there is a significant gap between urban and rural areas in terms of teacher availability and quality. Urban schools tend to attract more qualified and experienced teachers due to better living conditions, professional opportunities, and resources. Rural areas, however, often face teacher shortages, and the teachers who do work there may have less training and fewer resources. This disparity directly affects the quality of education, as students in rural areas often experience less consistent, lower-quality instruction; ii) Cultural and Linguistic Diversity: Bolivia’s student population is culturally and linguistically diverse, with significant indigenous communities. The education system’s ability to cater to different languages and cultural contexts is crucial for equitable education. However, the current system often fails to fully accommodate this diversity, leading to lower educational outcomes for indigenous students; iii) Poverty and Access to Education: Poverty is a major barrier to education. Students from low-income families, particularly in rural areas, face difficulties accessing education due to costs associated with schooling (e. g., transportation, uniforms, and supplies) and the need to contribute to household income; iv) Income Inequality: Income inequality in Bolivia is stark and directly impacts educational opportunities. Children from wealthier families have access to better schools, private tutoring, and extracurricular activities, which enhances their educational outcomes. In contrast, children from poorer families often attend underfunded public schools, leading to a cycle of educational inequality; and v) Urban vs Rural Outlook: In urban areas, students may have higher expectations for future employment, particularly in professional or skilled jobs, which can motivate them to pursue higher education. In contrast, in rural areas, limited job opportunities and the need to work in agriculture or low-skilled jobs can reduce the perceived value of education, leading to higher dropout rates and lower educational attainment. Future lines of research would benefit from including these factors, emphasizing the impact of teacher training on sociocultural diversity, and analyzing the impact of introducing technology in rural schools, such as e-learning platforms or mobile education units, as a way to overcome infrastructure limitations and improve educational quality and coverage in remote areas.

In summary, our findings indicate that infrastructure services and facilities can have a significant positive impact on enrollment and promotion, though in most cases certain types of services and facilities, like basic utilities, computer rooms, and sports fields, have a comparatively larger impact. In terms of abandonment and failure, principal deterrents are access to sports facilities and the possibility of attending private schools in urban settings. A principal takeaway of the empirical findings is that improving basic infrastructure services and facilities can produce positive school outcomes. Our findings also indicate that other factors are important in explaining school outcomes, including type of school, geographical location, and whether the school offers a humanistic education.

Additionally, cultural and social factors influence Bolivian students’ decisions to continue or abandon school. In rural and indigenous communities, traditional customs and the need for children to contribute to household labor, particularly in agriculture, can pressure students to leave school early. Gender roles also influence educational decisions, with girls often expected to prioritize domestic responsibilities over education. Additionally, the lack of culturally relevant curricula and language barriers, especially for indigenous students, can lead to disengagement and dropout. Social expectations and the perception that education may not lead to better job opportunities, particularly in rural areas, further contribute to school abandonment. An analysis of these factors is beyond the scope of this paper, but surely need to be kept in one’s mind as the results presented here are analyzed and dissected.

The rest of the paper is organized as follows. Section 2 presents the literature review; the data and methodology are described in Section 3. Section 4 reports the results, and section 5 concludes.

The impact of infrastructure on school outcomes has been analyzed in detail both at the individual country level and for sets of countries. A sample of works includes ^{Barrett et al. (2019)} who provide a comprehensive review of current research studies that focus on how school infrastructure affects children’s learning outcomes; ^{Cuesta et al. (2016)} examine the economics and education literature from 1990 to 2012 to assess the extent to which specific types of school infrastructure have a causal impact on student learning and enrollment. They find some, but not conclusive evidence that access to toilets, laboratories, and drinking water facilities increases enrollment; and ^{Evans and Mendez (2021)}, focusing on Africa and finding that, among many other factors, new school constructions have a positive impact on enrollment in Burkina Faso, Zambia, and Niger.

• ^{Barrett et al. (2019)}
  The Impact of School Infrastructure on Learning: A Synthesis of the Evidence. International Development in Focus, 2019
  The Impact of School Infrastructure on Learning: A Synthesis of the Evidence. International Development in Focus
• ^{Cuesta et al. (2016)}
  School Infrastructure and Educational Outcomes: A Literature Review, with Special Reference to Latin America

  Economía, 2016
  School Infrastructure and Educational Outcomes: A Literature Review, with Special Reference to Latin America
• ^{Evans and Mendez (2021)}
  Education in Africa: What Are We Learning?

  Journal of African Economies, 2021
  Education in Africa: What Are We Learning?

Studies that focus on specific countries include ^{Cohen and Bhatt (2012)}, who analyze the United States and conclude that a lack of educational infrastructure is one factor that stymies efforts to improve literacy instruction; ^{Belmonte et al. (2020)}, focusing on Italy and finding that after the 2012 Northern earthquake, spending on school infrastructure increased standardized test scores in Mathematics and Italian language, with the effect being stronger for lower-achieving students and in Mathematics; and ^{Choudhuri and Desai (2021)} examining the relationship between piped water and access to liquefied petroleum gas (LPG) and children’s educational outcomes in rural India and finding that children aged 6-14 years, living in households that rely on the free collection of water and cooking fuel, have lower mathematics scores and benefit from lower educational expenditures than children living in households that do not collect water and fuel.

• ^{Cohen and Bhatt (2012)}
  The Importance of Infrastructure Development to High-Quality Literacy Instruction

  The Future of Children, 2012
  The Importance of Infrastructure Development to High-Quality Literacy Instruction
• ^{Belmonte et al. (2020)}
  School Infrastructure Spending and Educational Outcomes: Evidence from the 2012 Earthquake in Northern Italy

  Economics of Education Review, 2020
  School Infrastructure Spending and Educational Outcomes: Evidence from the 2012 Earthquake in Northern Italy
• ^{Choudhuri and Desai (2021)}
  Lack of Access to Clean Fuel and Piped Water and Children’s Educational Outcomes in Rural India

  World Development, 2021
  Lack of Access to Clean Fuel and Piped Water and Children’s Educational Outcomes in Rural India

Several authors have also analyzed different aspects of education in Bolivia, including ^{Faguet and Sánchez (2008)}, ^{Arrueta and Avery (2012)}, ^{Liberato et al. (2006)}, ^{Neiva de Figueiredo and Marca Barrientos (2013)}, ^{Moreno Cely et al. (2021)} and ^{Canelas and Niño-Zarazúa (2019)}. Specifically in Bolivia, some studies address infrastructure’s impact. For example, ^{Newman et al. (2002)} focus on how school infrastructure projects in rural Bolivia affect education outcomes and finding little evidence of improvement; a reason for the lack of improvement, they argue, may lie in the fact that these projects only address infrastructure improvement without addressing other concerns like teaching quality or access to nutritional meals; ^{Popova and Fabre (2017)} offer a mostly qualitative description of a project aimed at digital inclusion in the department of La Paz that improves teachers’ abilities to generate more effective interactive teaching methods and better learning environments; and ^{Farfán et al. (2015)}, also with a descriptive, qualitative analysis of digital inclusion in the Department of Tarija and its mostly positive impact on both teacher’s ability to integrate information and communications technologies into the educational process and students’ skills to access educational resources.

• ^{Faguet and Sánchez (2008)}
  Decentralization’s Effects on Educational Outcomes in Bolivia and Colombia

  World Development, 2008
  Decentralization’s Effects on Educational Outcomes in Bolivia and Colombia
• ^{Arrueta and Avery (2012)}
  Education Reform in Bolivia: Transitions Towards Which Future?

  Research in Comparative and International Education, 2012
  Education Reform in Bolivia: Transitions Towards Which Future?
• ^{Liberato et al. (2006)}
  Well-Being Outcomes in Bolivia: Accounting for the Effects of Ethnicity and Regional Location

  Social Indicators Research, 2006
  Well-Being Outcomes in Bolivia: Accounting for the Effects of Ethnicity and Regional Location
• ^{Neiva de Figueiredo and Marca Barrientos (2013)}
  A Decision Support Methodology for Increasing School Efficiency in Bolivia’s Low-Income Communities

  International Transactions in Operational Research, 2013
  A Decision Support Methodology for Increasing School Efficiency in Bolivia’s Low-Income Communities
• ^{Moreno Cely et al. (2021)}
  Transforming Higher Education in Bolivia by Linking Two ContraHegemonic Movements: Agroecology and the Decolonial Turn

  International Journal of Qualitative Studies in Education, 2021
  Transforming Higher Education in Bolivia by Linking Two ContraHegemonic Movements: Agroecology and the Decolonial Turn
• ^{Canelas and Niño-Zarazúa (2019)}
  Schooling and Labor Market Impacts of Bolivia’s Bono Juancito Pinto Program

  Population and Development Review, 2019
  Schooling and Labor Market Impacts of Bolivia’s Bono Juancito Pinto Program
• ^{Newman et al. (2002)}
  An Impact Evaluation of Education, Health, & Water Supply Investments by the Bolivian Social Investment Fund

  The World Bank Economic Review, 2002
  An Impact Evaluation of Education, Health, & Water Supply Investments by the Bolivian Social Investment Fund
• ^{Popova and Fabre (2017)}
  Digital Inclusion of Secondary Schools’ Subject Teachers in Bolivia

  International Journal of Education and Development Using Information and Communication Technology, 2017
  Digital Inclusion of Secondary Schools’ Subject Teachers in Bolivia
• ^{Farfán et al. (2015)}
  Digital Inclusion in Education in Tarija

  Cepal Review, 2015
  Digital Inclusion in Education in Tarija

It is evident that much remains to be learned about how infrastructure impacts school outcomes, particularly in developing countries. From the perspective of Bolivia, there is a lack of systematic, empirical analysis on this topic in Bolivia, hence the timeliness of this study that utilizes the most comprehensive database on all preprimary, primary, and secondary schools in the country to empirically test the impact of various infrastructure variables on specific school outcomes.

Data on infrastructure and school outcomes were sourced from the Bolivia’s General Planning Directorate at the Ministry of Education.² The data cover the period 2015-2020 for all pre-primary, primary, and secondary schools in the nine departments of the country. The average number of schools across all levels during this period is 15 955 and the maximum number of observations for the entire period is 95 766. The four outcome variables are yearly numbers of enrollment, promotion, abandonment, and failure.³ Infrastructure data is consists of facilities (number of classrooms, sports fields, laboratories, computer rooms, and libraries) and services (access to electricity, potable water, and functioning bathrooms; these three are captured in the ‘basic utilities’ variable.⁴

Additional variables tested include school location (urban or rural), school type (public or private), and whether the school offers a humanistic degree. A humanistic degree refers to the number of specialized seminars aimed at promoting technical skills; these seminars are available in schools that offer a ‘humanistic education’ in line with the Education Law passed by the Plurinational Legislative Assembly in 2010.⁵ The dataset is also disaggregated by sex (men and women).Tables 1a and b provides descriptive statistics for all variables and Appendix 1a, b, expands on the description of the dataset by presenting a sample of yearly statistics by region and grade level.

An initial assessment of the descriptive data over the six-year period (2015-2020) is that average enrollment is highest in Cochabamba (285.5 students), followed by Santa Cruz (265.3) and Oruro (196.7); the lowest enrollment occurs in Pando (90.9); the trend for average number of students promoted follows a similar pattern. For students who drop out, the highest average occurs in Santa Cruz (8.5 students), followed by Beni (7.6) and Cochabamba (6.6); abandonment is lowest in Potosí (3.4). The highest average number of students who fail to progress to the following grade occurs in Santa Cruz (8.5 students), followed Cochabamba (7.6) and Chuquisaca (4.5) thereafter; the lowest is recorded in La Paz (2.3). Access to basic utilities , such as electricity, water, and sanitation is highest in Tarija (92.0 percent) and lowest in Pando (67.1 percent). With infrastructure facilities, Pando has the lowest access to most percent of these facilities and Cochabamba has the highest access to classrooms (7.2). Oruro is where the highest number of specialized seminars were imparted (0.2), and Pando records the lowest number (0.0). Though Pando seems to fall short on most indicators, he data does not show a consistent pattern between infrastructure access and educational outcomes across departments in the country - Santa Cruz, La Paz, Cochabamba, and Tarija - fare better than the rest.⁶

A dated panel data is used throughout the study.⁷ The following equation is estimated using Generalized Least Squares (GLS):⁸

where Ai⁹ measures levels of enrollment, promotion, abandonment or failure in school i; Basic Utilities and Facility are the causing variables of interest; Other is an additional variable of interest (i. e., urban school, public school or humanistic degree); and ε _i is a standard error term.

When A _i is levels of enrollment or promotion, we assume β ₁ will be positive, as greater access to electricity, water, and working bathrooms -captured in the variable basic utilities- should exert a positive effect on attendance and performance. A similar argument can be made for β ₂, access to classrooms, sports fields, computer rooms, laboratories, and libraries should positively influence students’ desire to attend school and do well. With β ₃, the correlation depends on the regressor being considered. The expected sign of β ₃ depends on the specific regressor. For example, if the variable represents an urban school, we expect a positive coefficient, indicating better facilities and resources in urban schools (i. e., with more and better teachers, greater access to basic equipment and services, and greater access to financial resources) than their rural counterparts; and it is expected to be negative when the variable is public school due to the systematic limitations of public educational institutions, including insufficient governmental funding. The impact of humanistic degree is uncertain given that Bolivia’s efforts to expand technical education are still in early stages.

When A _i represents level of abandonment or failure, we generally expect is that greater access to services and infrastructure facilities should deter students’ desire to quit school and/or perform poorly, but it is recognized that other factors may influence these variables. School type and location may also impact student performance and abandonment rates; however, no a priori assumptions are made about their effects on these variables.

Table 2 presents the results of specification (1) using the full dataset, with enrollment levels as the dependent variable¹⁰ Columns 1-4 show results for the full sample, 5-8 for men, and 9-12 for women.

As expected, access to basic utilities, classrooms, sports fields, and computer rooms are positively correlated with levels of enrollment. All coefficients are positive and statistically significant at standard levels, reflecting that these infrastructure services and facilities play an important role in school enrollment. Across all samples, men, and woman access to basic utilities has the largest impact on enrollment. For instance, in column (1), all else equal, a coefficient of 327.495 means that for every unit increase in basic utilities, levels of enrollment increase by approximately 327.5 students. Classrooms, on the other hand, have the lowest impact, with an increase of only 12.2 students (column 1) per additional classroom.

Laboratories have a positive and significant impact for the entire sample and for men, but insignificant for women, highlighting that this facility does not seem to motivate women to enroll in school. A similar argument can be made for libraries: it has the expected positive and significant correlation for the full sample and women, but for men its coefficient is consistently insignificant. While these results do not elucidate the reasons why laboratories and libraries are conducive for enrollment to one group but not the other, they clearly point to the need for government or for the private institutions running the schools to address the disparities to ensure that these vital school facilities are equally supportive of all students, regardless of sex.

The coefficients for urban school and public school are the expected ones, reflecting that attending an urban school is positively correlated with enrollment and attending a public school is not. Specifically, and concentrating on the results depicted in column (4), all else equal, attending a school in an urban center would increase total enrollment by approximately 287 students, and attending a government-funded school would decrease it by roughly 112 students. The impact of humanistic degree is unclear; it is only statistically significant for men (at the 0.10 level).

Table 3 presents the results with promotion as the dependent variable, defined as students meeting the requirements to advance to the next grade. Results are reported for the full sample and disaggregated by sex.

The results are similar to the ones reported for enrollment and this is not entirely surprising, as one would expect most students to steadily move through the school system over time.¹¹ Infrastructure services and facilities play the expected, positive role, and the variable basic utilities has the consistently larger impact. For instance, and concentrating on column (1), all else equal, an additional unit of basic utilities would increase the level of students promoted by approximately 333. As was the case with enrollment, access to laboratories only seems to have the expected positive and significant impact on promotion in the full sample and for men; for women, the impact of this variable is statistically insignificant. Similarly, access to libraries is conducive for greater number of students promoted in the full sample and for women. For men, libraries do not seem conducive to greater promotion levels.

Similar to enrollment, laboratory access has a positive, significant impact on promotion for the full sample and men, but is statistically insignificant for women. The impact of schools offering a humanistic degree does not have a statistically significant impact on the dependent variable.

Table 4 presents results when the dependent variable is levels of abandonment. Abandonment refers to students who drop out of school before completing a school term.

The decision to drop out may involve factors beyond infrastructure or school type, so the results reported in Table 4 greatest impact on the dependent variable. For example, the coefficient for sports fields is consistently negative and significant, suggesting that access to sports fields reduces student dropout rates. In column (1), other things equal, an additional sports field would deter 1.129 students from abandoning school. The size of the coefficient diminishes when the regression is carried out for men and women but remains consistently negative and significant. Library access shows a positive correlation with drop out, and for women, it is statistically significant in all specifications, implying that access to this facility increases the number of people -especially women- dropping out of school. This result is a clear indication that libraries are not fulfilling their mandate of providing free access to books, magazines, newspapers, journals, and other resources that promote literacy, lifelong learning, and personal growth.¹² While the coefficient for classrooms is positive and significant, its small size suggests a negligible real-world impact on dropout rates.

The coefficients for urban school and public school are also positive and statistically significant, implying that attending schools in urban centers and funded by the government increase levels of abandonment. The coefficient for humanistic degree is negative and significant in all cases, implying that offering this specialized degree is an important deterrent for abandonment.

Table 5 presents results when the dependent variable is levels of failure. Failure refers to the number of students who fail to advance to the next grade.

Student failure to advance to the next level is also likely the result of many conflicting factors, including lack of effort, such as financial stress, and individual family dynamics, hence the need to take the results shown in Table 5 with caution. As was the case with abandonment, particular emphasis is placed on variables that are consistently significant and have the greatest impact on levels of failure. Access to sports fields appears to be the strongest deterrent against student failure. In all specifications, the coefficient for this variable is negative and statistically significant, implying that, all else equal and focusing on column (1), a unit increase in sports fields decreases the number of students failing to advance to the next level by 1.411. Key drivers of higher failure rates include attendance at public urban schools, availability of humanistic degrees, and access to libraries (especially for women and girls).¹³

Thus far, results have been analyzed at an aggregate level. However, in a country like Bolivia, where regional differences are significant, it may be the case that the results are contingent on the sample being analyzed o examine regional differences, we applied specification (1) to four subsets: Cochabamba, La Paz, Santa Cruz, and all other departments combined.¹⁴ Table 6 presents the results when the dependent variable is levels of enrollment.

Results reveal clear regional differences in how various factors - such as access to basic utilities and infrastructure- affect enrollment levels. Classrooms are the only variable with a consistent positive impact on enrollment across all sub-samples. Access to basic utilities driver higher enrollment only in Cochabamba, where its coefficient is positive and statistically significant in all cases; everywhere else, the impact of this variable is insignificant.¹⁵ Access to sports fields is positively and significantly correlated with enrollment in all departments except Santa Cruz, perhaps reflecting that in this Department -located in the lowlands of the country- there is much greater availability of public recreational spaces than are available in the valleys (Cochabamba, Chuquisaca, Tarija), the highlands (La Paz, Oruro, Potosí), and relatively poorer lowland regions of the country (Beni and Pando). In La Paz, access to computer rooms significantly boosts enrollment, while in the ‘Remaining Departments’ sub-sample, laboratory access is a key driver. Library access has mixed impact on the dependent variable: in Cochabamba, it is a significant driver for greater enrollment but in the ‘Remaining Departments’ subset it is a deterrent for it. This disparity may stem from poorer departments having less equipped libraries than wealthier areas like Cochabamba. The results with this variable for La Paz and Santa Cruz are less clear. Access to schools in urban centers is an important driver for greater enrollment everywhere except Santa Cruz, where there is seemingly little difference between rural and urban schools. Public schools are a drive higher enrollment in Cochabamba but a deterrent in La Paz and Santa Cruz, perhaps reflecting a relatively higher quality of public education in Cochabamba. Offering a humanistic degree deters enrollment in Cochabamba and La Paz, but an important driver in the ‘Remaining Departments’ sub-sample, a reflection of the importance of learning technical skills in the relatively less-developed regions of the country. To emphasize the importance of this variable in those departments, all else equal, a unit increase in humanistic degree would increase the level of enrollment between 44 (women) and 92 (total) students.

Table 7 shows the determinants of promotion, defined as students advancing to the next grade, across different regions.

The results are similar to those found when with enrollment as the dependent variable. Access to classrooms is important in all cases.¹⁶ Sports fields are significant drivers for promotion everywhere except in Santa Cruz; computer rooms are only important in La Paz; Libraries foment promotion in Cochabamba but deter it in the ‘Remaining Departments’; attending school in urban centers is conducive for greater enrolment everywhere except Santa Cruz; public schools are positively and significantly correlated with the dependent variable in Cochabamba; everywhere else, public education is a deterrent for greater enrollment; finally, a humanistic degree is only conducive for greater promotion in the ‘Remaining Departments’ subset.

Table 8 presents the regional determinants of abandonment, defined as the number of students who drop out before the school term ends.

As mentioned previously, the interpretation of results with levels of abandonment as the outcome variable is nuanced. The only consistent regressor found to increase abandonment rates across models is public school attendance. In all cases, the positive coefficient suggests that public schools may create environments more conducive to student dropout. Specifically, all else equal, a unit increase in public schools increases abandonment levels between 1.338 (La Paz, women) and 8.826 (Cochabamba, total) students. In La Paz, laboratory access is another significant driver of abandonment, with positive and significant coefficients suggesting that laboratories may discourage academic engagement. Urban school attendance is a catalyst for higher dropout rates in La Paz and in the sub-group ‘Remaining Departments’, highlighting issues in public education that should be addressed by regional and national authorities. In Cochabamba, the number of classrooms is also a driver for greater abandonment levels, but the absolute size of the coefficients for the three subsets (total, men, women) is negligible.

Factors that reduce dropout rates include access to sports fields and computer rooms in the remaining departments; access to laboratories and attending schools in urban centers in Cochabamba; and the offering of a humanistic degree in Cochabamba, La Paz, and Santa Cruz.¹⁷ The consistently negative and significant coefficients, reflecting that specific regional factors discourage students from abandoning school.

Table 9 reports findings when the dependent variable is levels of failure.

As was the case with abandonment, public school attendance is a driver for increasing numbers of students failing to advance to the next level, in all subsets, its coefficient is positive and statistically significant, reflecting that regardless of geography, attending publicly-funded schools increases the number of students with deficient academic performance. Attending to schools in urban centers -specifically in Cochabamba, La Paz, and the remaining departments- is an additional driver for greater failures. Lastly, schools offering a humanistic degree in La Paz and the remaining departments also seem to induce a greater number of students failing the grade. In La Paz, the number of classrooms is a driver for failure, but the absolute size of the coefficient in the three subsets (total, men, women) renders its real impact trivial.

Deterrents for failure include access to basic utilities in La Paz and Santa Cruz; access to computer rooms in Cochabamba and Santa Cruz;¹⁸ access to libraries in the remaining departments;¹⁹ and schools offering a humanistic degree in Santa Cruz.²⁰

Finally, Table 10 presents the determinants of enrollment based on the grade-level offerings of different schools. Computer rooms in Cochabamba and Santa Cruz; libraries in the remaining departments; and humanistic degree offerings in Santa Cruz. The hypothesis is that enrollment (along with promotion, abandonment, and failure rates) may vary not only by geographical location of the school, but also based on their different grade-level offerings. Appendices 2, 3, and 4 provides determinants of promotion, abandonment, and failure based on grade-level offerings.²¹

The results indicates that classrooms and urban schools are the only regressors showing a positive correlation with enrollment levels. Across all sub-groups, the coefficients for these variables are positive and statistically significant, indicate that classroom access and education in urban centers are important drivers for greater enrollment, regardless of the grades offered by a school.²² Access to basic services induce greater enrollment in schools offering pre-primary and primary levels and in schools offering all levels; access to sports fields is only consistently correlated with greater number of students enrolling in school in those institutions offering all levels;²³ access to computer rooms is consistently significant and positively correlated with the dependent variable in the pre-primary, primary and primary, secondary sub-groups; likewise with laboratories in the secondary and primary, secondary sub-groups;²⁴ access to libraries is important for greater enrollment in the pre-primary, primary sub-group and in schools offering all grade levels; public school is also important in promoting enrollment in schools offering pre-primary and primary levels, as well as in those offering all levels; and finally, humanistic degree is only important in those schools offering secondary, and primary and secondary education.

Regarding deterrents to enrollment, laboratories appears to reduce enrollment in schools offering all grade levels. This findings highlights both a problem and an opportunity for an impactful intervention, be it from the government or the private institutions administering a school. Libraries access also does not support higher enrolment in the secondary and primary, secondary sub-groups; lastly, the coefficient for public schools is consistently negative and significant for the primary, secondary sub-group, highlighting another potential area for improving public education in schools offering both primary and secondary education.

Overall, the empirical findings seem to confirm the expectations outlined in section 3.2 although notable differences emerge when analyzing by geographical region and grade-level offerings.

As noted in section 1, socioeconomics inequities in Bolivia can significantly the impact of infrastructure affects educational outcomes. In poverty-stricken areas-particularly rural regions in the highlandsstudents may face additional barriers such as poverty-stricken, limited access to school materials, or family obligations that detract from their ability to benefit fully from improved infrastructure. Wealthier areas with better infrastructure -primarily urban centers in the lowlands- may see more immediate educational gains. At the same time, disadvantaged communities might experience slower progress due to the compounded effects of inequality. These disparities can produce uneven outcomes, as the impact of infrastructure may be diminished or delayed in areas with greater socioeconomic challenges. With these caveats in mind, we present a summary of our principal findings below.

The study used Bolivian data to empirically analyze of the impact of school infrastructure on education outcomes was carried out for nationwide at the pre-primary, primary, and secondary levels during the 2015-2020 period. Specifically, the examined how enrollment, promotion, abandonment, and failure levels are influenced by access to essential infrastructure services like electricity, running water, and working bathrooms; and by access to basic infrastructure facilities like classrooms, sports fields, laboratories, libraries, and computer rooms. We also tested the impact of urban centers, attending public schools, and offering a humanistic degree impact school outcomes. A dated panel dataset was employed with results reported at the aggregate level; by geographical region; and by school grade levels.

The findings consistently demonstrate that infrastructure services and facilities significantly boost enrollment and promotion. However, certain elements-such as basic utilities, computer rooms, and sports fields-tend to have a stronger impact. At the aggregate level, urban school attendance significantly boosts enrollment and promotion. However, regional analysis, the results show that in Santa Cruz education in urban schools does not have a statistically significant effect on enrollment or promotion. At the aggregate level, public schools are shown to be a deterrent for enrollment and promotion, but when the analysis is carried by geographical area, the findings demonstrate that in Cochabamba attending government-funded schools increases levels of enrollment and promotion. Similarly, access to basic utilities is a primary driver for greater enrollment and promotion across all groups (all, men and women) at the aggregate level. Regionally, however, it only has a positive and significant impact in Cochabamba; its impact on promotion levels is only positive and significant (at the 10% level) in the women subset in Cochabamba. While offering a humanistic degree does not influence enrollment at the aggregate level but it is an important driver of it in the sub-group ‘remaining departments’ (Beni, Chuquisaca, Oruro, Pando, and Potosí); it is also an important driver for promotion in Santa Cruz (total and women) and in the remaining departments.

At the aggregate level, in terms of abandonment, principal deterrents are the existence of sports fields and the offering of a humanistic degree. Regionally, sports fields deter abandonment only in the remaining departments. In the remaining departments, access to computer rooms is a restraint on abandonment as is access to laboratories and attending urban schools in Cochabamba. Offering a humanistic degree significantly reduces abandonment in Cochabamba, La Paz, and Santa Cruz.

At the aggregate level, the main deterrent to failure is access to sports fields. Regionally, access to basic utilities is an important obstacle to failure in La Paz and Santa Cruz. In both Cochabamba and Santa Cruz (total and men groups), access to computer rooms reduces failure rates. Libraries are also important in preventing failure in the remaining departments (total, men). Lastly, offering a humanistic degree reduces failure rates in Santa Cruz (total and men groups). A similar analysis can be done when the determinants of school outcomes are studied based on school grade levels.

The main conclusion of this study is that both infrastructure services and facilities serve as significant positive determinant of school outcomes. The challenge for both government and private institutions is to ensure that new or renewed infrastructure serves the entire school-aged community and not only privileged sectors -i. e., urban areas vs rural areas; men vs women; wealthier departments vs poorer departments -within Bolivian society.

Future research could build on these findings. Specifically, given that physical infrastructure is important in determining school outcomes -a key finding of this study- subsequent research could examine how targeted teacher training programs that focus on cultural competence and bilingual education can improve educational outcomes for indigenous and rural students. Longitudinal studies that track students from different socioeconomic backgrounds could also provide insights into how income inequality affects educational and professional outcomes; these types of studies would inform, and guide policies aimed at reducing educational disparities and promoting social mobility. Lastly, as discussed in Section 1, future studies could focus on integrating technology in rural schools to address infrastructure limitations.