Geographic inequality and learning in Peru: Evidence from the reform period of 2007-16.

Contributors:

University of Sussex, UK: Sonja Fagernas, Panu Pelkonen, Juan-Manuel del Pozo Segura, Diego de la Fuente Stevens

U. Antonio Ruiz de Montoya: Adriana Urrutia Pozzi-Escot (also Transparencia Perú),

Funding: British Academy

Date: May 2023

A set of significant educational reforms was implemented in Peru during the decade of 2010. The education budget increased substantially from around 2011 onwards (see Figure below). The period 2011-16 also coincided with a new government. The learning levels of primary schools pupils, measured as test performance, increased significantly during this reform period.

Figure 1 Developments in the education budget Figure 2 2^nd grade test scores in public schools

Our research project focuses on the changes in geographic inequality in relation to primary schooling, during the reform period of 2007-2016. We have sought to answer the following questions

1) Did inequality in learning between different regions/departementos fall as a result of the reforms? How about inequality between urban and rural areas? The answer turns out to be no, so we focus on explaining the persistent urban advantage in learning, including factors beyond the reforms.

2) Did the improvement in public sector learning levels allow public schools to catch up with private schools? How did this affect the position of private schools in urban Peru?

3) How was the regional education budget allocated over the reform period? To what extent did it relate to the educational needs of the regions? Did observable political factors affect how the educational budget was allocated across the regions?

To answer these questions, we use data from a range of sources, including school census data for all primary schools (Censo Escolar) and test score data for all 2^nd grade pupils (Evaluación Censal de Estudiantes), data on the education budget from the Ministerio de Economía y Finanzas, the national household surveys (ENAHO) and the national population census (Censo de Población y Vivienda). Our findings are summarized below.

1) Rural-urban inequality and learning [link to the study]

Several aspects of the educational reforms, and especially the budget, favored remote or rural regions. Public sector schools in rural areas experienced larger improvements in school inputs, such as infrastructure and teachers. Despite this, learning in rural public schools did not catch up with learning in urban schools on average. While Reading and Mathematics test scores improved in rural areas, they also improved in urban areas, in some cases even faster than in rural areas.

Figure 3 Key comparisons between urban and rural areas

Why did inequality in learning persist to this degree? Our statistical analysis provides the following explanations for the public sector urban learning premium. 1) Half of the urban learning premium can be explained by differences in school resources. 2) The remainder can be attributed to larger school size in urban areas, as well as factors beyond the school system, such as household socioeconomic status. The reform period was characterized by robust economic growth, which appears to have benefitted urban areas somewhat more. Therefore, the reforms and educational investment can only explain a part of the increases in learning.

From the perspective of an individual, learning could also be improved also by moving from a rural to an urban area. A parallel development to the educational reforms was rapid rural-urban migration in several parts of the country. We show that rural-urban migration potentially accounted for 20-30% of the improvement in learning in many regions. Moving from a rural primary school to an urban secondary school is associated with a small increase in learning, in both Reading and Mathematics.

Overall, the closing of learning gaps across areas can be challenging, because educational policy cannot influence all factors that affect learning, such as parental socioeconomic status, wealth of an area, or even school size. To reduce regional or rural-urban inequalities, educational policy should target school infrastructure and multigrade teaching, although the latter can be a challenge in rural areas.

2) Effect of the reform on private schools [link to the study]

Across the developing and emerging economies, low-fee private schooling is advancing rapidly due to low quality and capacity constraints in the public sector. The educational reforms carried out in Peru provide a possibly rare example of how public investment in education may reverse the balance between the public and private sector.

Figure 4 Key comparisons between public and private urban schools

Our analysis here focuses on urban areas and the period of 2007-2016. We show that as public school test scores for 2^nd grade pupils improved, the private sector learning premium in urban primary schools largely vanished. While private schools have been better resourced on average, learning levels in urban public sector schools improved significantly from 2011 onwards and largely caught up with private schools.

The growth of private primary school enrolment in turn started to level around 2014, soon after the rapid improvement in the public sector began (see figures below).

Figure 5 Private and public primary enrolment in urban Peru

Given the average improvement in the quality of public schools in the reform period, we can assume that competitive pressure on private schools increased. We construct a measure of competition faced by private schools from public schools, based on the number of public schools within a specific radius for each private schools. The data provides information of the location of each school. Our statistical analysis reveals that post 2011, competitive pressure on private schools increased more if there were more public schools nearby, the private schools were of lower quality in terms of learning and parental income in the area was lower, implying weaker purchasing power. In these cases, private schools were more likely to close down and their enrolment declined more in response to the improvement in public sector schooling in the post-2011 period.

The presence of low quality, low-fee private schools is potentially a significant problem for the development of skills in Peru. Our analysis is not a comprehensive study of this sector, and regulation evidently has an important role to play. Our results do suggest that significant investments in the public sector which improve its quality, have the potential to slow the growth of the private sector, and crowd out under-performing, private schools. A further analysis would be warranted to study whether the impact is long-lasting. The key message for policy would be to guarantee the existence of a good public sector school within a safe and convenient walking distance for all families in urban and peri-urban settings.

3) Allocation of the regional education budget: needs and politics [link to the study]

Our period of analysis coincides with two government periods in Peru (2006-2011 and 2011-2016). The regional education budget increased significantly since 2011. It largely consists of transfers from the central government to regions. Based on estimations, we divide the regional education budget into a share that can be explained by resource needs of the regional education systems, and a more discretionary component.

We find that over the period, the allocation of funding to regions became more “rational” in the sense that it more closely tracks the expected funding needs. In the latter period, funding was directed more towards regions that lagged behind in terms of school infrastructure, and that employed more teachers per pupil population, implying a larger wage bill.

However, there is also discretion in the allocation of funds. As might be expected, transfers of tax income from natural resource related activities (‘canon’ payment) are associated with higher regional funding, beyond resource needs. Finally, our analysis suggests that political lobbying may affect the discretionary component.

We find that if a larger share of districts in a region has a member of congress, the region receives more infrastructure funding, and has a larger number of new schools. Members of congress tend to come from larger cities and this tends to concentrate them to fewer districts. However, if a region has a number of successful candidates from more rural locations or smaller towns, the share of districts with a member of congress can increase. This in turn can influence budget negotiations.

Overall, we do not find that regions that obtained more funding than expected, managed to translate this funding into larger improvements in learning. Rather, learning outcomes improved across regions, irrespective of some discrepancies in funding.