Operationalizing the Millenium Development Goals for ...documents.worldbank.org/curated/en/193711468301146785/pdf/415… · This study examines the actions necessary to operationalize

Operationalizing the Millenium Development Goals for Education, A Case Study of the Kyrgyz Republic

Executive Summary

i. The primary audience of this case study is Government and its partners in the education sector. Its primary objective is to make policymakers and other partners aware of the unmet needs for the education MDGs, and to offer practical advice about options for meeting them. A secondary objective is to illustrate how the Ministry’s new strategic analysis unit might approach the design of interventions to meet specific educational objectives. ii. Two of the eight Millenium Development Goals (MDGs) which were adopted by the United Nations in 2000 address education objectives. They are: a) to ensure that all boys and girls complete a full course of primary schooling, and b) to eliminate gender disparity in primary and secondary education preferably by 2005, and at all levels by 2015. This study examines the actions necessary to operationalize these goals in the Central Asian context. It does so through a case study of the Kyrgyz Republic. It is hoped that the framework of the case study will be applicable to the other countries of Central Asia as well.1 iii. The case study comprises five sections:

• Status of MDG Attainment. The first describes the status of educational MDG attainment and documents the unmet needs involving school attendance and learning achievement. The situation of the Central Asian countries differs from that of most countries at similar levels of income. The interpretation of what it means to attain the education MDGs in Central Asia reflects this fact. Until now, Government attention to the education MDGs has focused on school enrollments. The high level of registered school enrollments has led to an impression that the educational MDGs are well on the way to being met. This case study argues that the spirit and intention of the MDGs calls for more attention to education outcomes in terms of whether students are attending school and what students are learning. Measured by those criteria, the education MDGs have not been attained and may not be met by the agreed target dates.

• The Causes of Low Educational Performance. The second assesses the causes of low school attendance and low learning achievement, focusing on deficiencies of the learning environment in schools and the supporting environment in low-income areas, and documenting the link between poverty and low educational performance.

• Equal Budgets, Unequal Outcomes. The third section examines the reasons why educational performance is lower in areas with lower income, in spite of government policy and a financing formula which give equal treatment to all schools. This section describes the role of contributions of households and local governments in more prosperous areas, and discusses the implications for attracting and retaining qualified teachers. It also describes the problem of the weaker supporting environment for educational performance in low-income areas, including lower parental education and lower saturation with educational stimuli of all kinds. This section prepares the way for the recommendation in the following section that the Government adopt a financing formula that leads to a “sloping playing field” that compensates for these educational deficiencies of low-income areas, rather than the level playing field under current policy.

1 Comprising Kazakhstan, Tajikistan, and Uzbekistan.

41512P

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

ed

2

• Recommendations. The fourth section describes the various actions which the Government is already taking to address the educational MDGs and recommends further actions to accelerate their attainment. These recommendations are presented in the form of a table which distinguishes short-term actions from medium-term actions, and sets out three progressive levels of intensity of actions to achieve the educational MDGs. The higher-intensity approaches would accelerate MDG attainment. The maximum-intensity approach includes adoption of management incentives and a financing formula that give preference to schools in low-income areas (because of their greater needs and lesser capacity to contribute from other sources). All approaches involve greater use of data on education outcomes and gradual improvement of these data over time.

• Source of Financing. The fifth section argues that actions to accelerate MDG attainment should be financed from the Republican budget, and discusses the possible sources of budget financing. These are: a) increased budget allocations for the sector, b) internally-generated savings in use of existing budgets, and c) reallocation of budgets for other activities within the sector. Better documentation of needs and results would strengthen the claim for increased budget resources for the sector, but an overall increase beyond the increase already planned under the Medium-Term Budget Framework cannot be assumed. There is significant potential for savings under the second source -- through more efficient use of budget resources within the sector. Developing the incentive framework for improved efficiency will be a long-term effort. In order to be effective, this will need to include redefined roles for local governments and central education authorities in order to remove the current role conflicts which prevent rationalization of the school network and more efficient deployment of teachers. It will also require the political will to shed unneeded teachers. The third source – reallocation from other activities within the sector – is a better prospect in the short term. Since resources for primary and secondary education consist almost exclusively of teacher salaries, the only potential budget area from which significant funds could be reallocated is higher education. The case study discusses several financing alternatives for higher education as a means of freeing resources for accelerated MDG attainment.

3

Current Program Accelerated Program Maximum Program Short term actions - Student social

support program - Rural teacher incentive program - Rural teacher fellowship program

Enlarge

Medium term actions - Pilot on capitation financing

Adopt performance-based financing formula

The case study will comprise four sections: first section will describe the ststus of 1.MDG Status In MDG status section, cite UNDP Country Report for KG Present MDG+ issues: school attendance and learning achievement 2.Causes of low school attendance and low learning achievement

a. Attendance: evidence from Kinnon Scott and annex chapter in poverty report, UNICEF and SOROS reports on school attendance, data from 2003 survey (new regression?) b. Learning achievement: Present evidence on poverty links, absence of textbooks, lower parental education, etc. MLA and ACCELS data. Any room for new analysis of ACELS data? Do MLA data include information on availability of educational materials in schools, whether children have textbooks, etc? Correlation of average education of parents by oblast and MLA scores? Cite general findings from production function literature. Who in the Ed Advisory? Ask XXX

3. Interventions to address the causes of low school attendance and low learning achievement. 4. Constraints resources incentives implementation 5. Recommendations ServiceandequipUNICEF attan Why is Government not providing more resources? Hypotheses about why Government is not providing more resources: Other political priorities (revealed preference) MOE doesn’t document needs and use of existing resources.

4

Needs are not apparent, because of focus on enrollments. A New Paradigm for Education iii. The case study proposes a number of specific interventions to improve performance on the education MDGs. But meeting the education MDGs will require more than these specific actions. It will require a paradigm shift for the Government – a change in which education performance is no longer measured in terms of registered enrollments, but in terms of school attendance, learning achievement, and labor-market success, and in which performance of the education system is no longer judged by the success of its best students, teachers, and schools, but its success in raising the performance of its lowest-performing students, teachers, and schools to national learning standards. Implementing this new approach would also require a fundamentally different approach to resource allocation in the sector. The inherited approach to resource allocation for education was resolutely input-based. There were efforts to replace this with an output-based capitation formula early in the transition, in the desire to introduce incentives for improved efficiency. which would provide in the expectation that In spite of efforts to introduce capitation financing early in the transition, the resource allocation formula remains input-based and norm-guided. A consequence of this approach is that all schools receive resources based on the number of teachers they have. e basic principle of the This approach formula To the extent that public attention focuses on learning outcomes, it does so mainly for the right-side tail of the distribution – the highest performing schools, teachers, and students. Government and the public tend to judge the strength of the education system on the basis of its strongest parts – the Olympiad winners. Further progress on the MDGs and success in the broader objectives of education will require a focus on left-side tail of the distribution -- the lowest-performing schools, teachers, and students. Resource allocation will need to reflect this focus. The Resource Constraint iv. Resources are inevitably an important part of the MDG story in the Kyrgyz Republic. At independence, Kyrgyzstan inherited a relatively well-developed education system, without the resources to maintain it. As the 2003 Public Expenditure Review documents, the Government has struggled to maintain the system. In general, this has involved progressive cutbacks in budget financing to barely life-support levels. Central budget financing for primary and secondary education has essentially been limited to teacher salaries,2 which have declined seriously in real terms. In relatively affluent – largely, urban – communities, additional financing from local 2 For a brief period, local communities were expected to pay teacher salaries, with the result that teachers in most communities were not paid.

5

governments and households has helped maintain essential teaching and learning materials and has supplemented teacher salaries enough to retain qualified teachers. But in poorer communities and most rural areas the minimal budget financing and the lack of supplementary financing have led to a depletion of teaching and learning materials, a lack of teachers in certain subject areas, and a general degradation of the teaching and learning environment in schools. These schools are a hollowed-out version of the schools in more affluent communities. The Proposed Approach v. More resources, stronger incentives. The case study argues that the risks to MDG attainment in the Kyrgyz Republic are predominately related to resources and incentives. It proposes that meeting the MDGs is largely a challenge of equipping schools and students in the lowest performing schools better – if not to the standards of the best equipped, urban schools, at least to minimally acceptable standards – and providing incentives for teachers and education managers to focus more attention on the MDG goals of school attendance and learning achievement – particularly, of the lowest-performing students and schools. At the start, this approach can be implemented using existing information on regional differentials in learning achievement and school attendance. Progressively, the targeting and the interventions themselves should be refined on the basis of better information on educational outcomes at the school level. vi. The importance of central financing. As the case study documents, the schools in the poorest localities and regions have the lowest educational performance in terms of school attendance and learning achievement -- often, below national learning standards. They are thus the least able to finance the interventions which are needed to improve education performance (and, indeed, it is their lack of capacity to provide supplemental financing for their schools that explains a large part of their poor performance in the first place). For this reason, interventions to improve their performance will need to be financed from the Republican budget. vii. MDGs – not the only goal. The approach to analysis in this case study recognizes that while MDGs are a stated goal of education programs and policy in the Kyrgyz Republic, they are not the only goal. Education programs and policy also promote a number of other legitimate objectives, including developing a sense of nationhood and civic responsibility, and providing the skills for productive employment. Public expenditures provide an indication of “revealed preferences” or implicit priorities for public-sector expenditures. In the case of the Kyrgyz Republic, public expenditure levels and patterns indicate that the education MDGs are a serious priority of the Government. But the case study argues that the needs of meeting the MDGs in basic education call for additional resources. viii. Mobilizing additional resources – the importance of selectivity. There are three potential sources of additional central budget resources for MDG targeted interventions: increased budget allocations for the sector, internally-generated savings in use of budgets, and reallocation from other activities within the sector. The case study discusses the options for mobilizing resources for MDG interventions from the second and third sources, since increased budgets for the sector cannot be assumed. There is significant potential for savings under the second source -- through more efficient use of budget resources within the sector -- but developing the incentive framework for improved efficiency will be a long-term effort. In order to be effective, this will need to include redefined roles for local governments and central education authorities in order to remove the current conflicts which prevent rationalization of the school network and more efficient deployment of teachers. It will also require the political will to shed unneeded teachers. The third source – reallocation from other activities within the sector – is a better prospect in the short term. Since resources for primary and secondary education

6

consist almost exclusively of teacher salaries, this implies reallocation from higher education. Reallocation of budget resources from higher education and increased reliance on private financing would be an appropriate source of financing for MDG-targeted education interventions for two reasons: first, because basic education is compulsory, and thus should be free, and second, because basic education generates significant external benefits whereas the benefits of higher education are largely private. Shifting resources from higher education to meet the needs of better performance on the MDGs in basic education would entail obvious tradeoffs among legitimate educational objectives. Deciding whether those tradeoffs are justified is a matter for resolution through the political process. ix. Actions under implementation. A number of activities are already under implementation by the Government which should contribute to better performance on the education MDGs by addressing the resource and incentive constraints described in the case study. These include the support which is being provided under the (FY 2005) Rural Education Project for teacher performance incentives, for textbooks and educational materials, for school grants to improve school attendance and learning achievement, and for improvements to the financing formula for education. The findings of this case study underline the importance of those actions for meeting the educational MDGs. They also suggest refinements in some of the planned actions which could improve their impact on MDG outcomes. x. Additional actions. Beyond these actions which are already in progress, the case study describes a range of options for pursuing better performance under the education MDGs, involving more or less intensive efforts on the part of Government. A minimalist approach would involve expanding existing programs of student social support (including exemption from textbook rental fees) for low-income students who are at greatest risk of non-attendance and low learning achievement. A more ambitious approach would involve a new resource allocation formula based upon school performance, providing both negative sanctions and positive incentives for low-performing schools to improve their performance. Between these extrema, the case study describes a range of intermediate options involving combinations of policies to promote the education MDGs.

7

• Focus on Results. The primary pillar is the need to manage education programs and

education resources with a focus on these educational outcomes (as opposed to the current focus of maintaining existing infrastructure and staffing). A corollary of this pillar is the need to: a) monitor educational outcomes (attendance and learning achievement, b) investigate the causes of incomplete attendance and low learning achievement, and c) implement targeted interventions to address the needs of students who are not attending school or not meeting stated learning goals. Those interventions will need to be maintained until the learning goals are met. The interventions will need to be tailored to the particular needs of individual risk groups, and will need to be flexible enough to address both supply constraints and demand constraints. Admittedly, the production function for learning is not well known. But there are some universal truths about the learning process that can guide the design of targeted interventions -- e.g., that effective learning usually requires a student, a teacher, and a textbook. Currently, a non-trivial number of students lack textbooks. Ensuring that all students have textbooks should be a key priority. In a situation of constrained resources, Pillar 1 implies very different allocation rules for use of budget resources. Pillars 2 and 3 describe the new allocation rules. (In that sense, Pillars 2 and 3 are derived from Pillar 1.)

• Pillar 2: Be More Selective in Use of Budget Resources. The Central Asian countries

cannot afford to maintain their inherited first-world education systems with third-world levels of per-capita income. The message here will be to concentrate use of budget resources on the base of the education pyramid, with increased reliance on private financing for higher education and job-specific training.

• Pillar 3: Use Existing Resources More Efficiently. Getting better results from limited

budgets will require not only selectivity, but also using existing resources more efficiently. This section will describe the constraints to using resources more efficiently, and will recommend measures to relieve those constraints.

8

2. As the case study documents, the education MDG is essentially met in letter, but not in spirit. Virtually all children, boys and girls, attend school through the current compulsory nine-year cycle of basic education. The educational MDG is currently not met in spirit because educational outcomes in terms of learning achievement for a significant proportion of basic education students are not satisfactory even by the Government’s own definition. Full attainment of the educational MDG would require improved performance on the part of low-achieving students. The Government is in the process of raising the bar for universal education, and is moving to a compulsory twelve-year education cycle. Full compliance with this goal will require special efforts. These will need to go beyond expansion of capacity, because enrollment rates in the upper grades of secondary schooling currently fall short of full coverage. Demand-side constraints as well as supply-side constraints contribute to that outcome. 3. The report examines the causes of low learning achievement, and concludes that two causes – both income related – are primarily responsible for this outcome. The first is the financing formula for education , which leads to major differences in education quality between poor (often rural) communities and more prosperous (often rural) communities. The second is household poverty, which contributes to low educational performance by limiting school attendance and undermining access to textbooks and other vital learning resources. The report recommends specific measures to address both causes of low learning achievement.

Structure of the education system

4. The Kyrgyz education system comprises preschool education for children between one and six years of age, primary education (grades one through four), lower secondary education (grades five through nine), upper secondary education (grades ten and eleven), and higher education. Education is compulsory in grades one through nine. In addition to the formal sector, there are numerous out-of-school institutions and organizations to meet the education and training needs of the population. Under the Soviet Union, the education system emphasized Russian language instruction, but also offered instruction in the Kyrgyz, Uzbek, and Tajik languages. Since independence, the education system continues to offer instruction in the four languages, but has moved to give greater emphasis to Russian language instruction. Textbooks in many subjects the upper grades are still available only in Russian.

Education Coverage 5. There are two basic approaches for examining the extent of education coverage: through registration data and through survey data. Registration data have the advantage of being more comprehensive and longitudinally comparable, but are subject to upward bias (among other reasons, because teaching positions and resources depend upon registered enrollments). Registration data, at least in published form, also are limited to formal enrollments, and thus do not shed light on how actual school attendance may differ from registered enrollments. Survey data offer the potential advantage of distinguishing between enrollments and actual patterns of school attendance. Because survey data are less likely than registration data to provide an incentive for biased responses, they may provide greater objectivity in results. And, by including

9

other variables, survey data may offer insights into causal factors affecting enrollments and school attendance.

6. Registration Data. As shown in Table 1, the largest changes in enrollments occurred in preschool education and higher education. Recorded enrollments in preschool education declined precipitously throughout the 1990s, with virtually all of the decline occurring in schools managed by ministries other than the MEC. Enrollments in higher education more than doubled. Enrollments in primary and secondary education increased at an average of slightly under 2% per year, with an initial decline in enrollments in grades 5 through 9 and a rapid recovery in the latter years of the decade. At the same time, university enrollments have increased significantly -- from 10.8 % of the age group in 1990 to 15.2 % 1998, largely due to growth in enrollments in private institutions.

Table 1 – Enrollments by Level, 1992/93-2000/01 Preschool Grades

1-4 Grades

5-9 Grades 10-11a

Total Grades 1-11

Higher Education

1991/92 190,100 1992/93 143,200 367,000 458,500 114,000 939,500 38,414 1993/94 92,200 378,900 454,400 101,000 934,300 38,384 1994/95 58,900 386,200 460,800 97,800 944,800 39,902 1995/96 46,100 471,900 402,900 95,900 970,700 47,416 1996/97 47,300 475,800 425,000 105,400 1,006,200 53,102 1997/98 46,100 472,100 446,700 121,100 1,039,900 63,498 1998/99 46,601 470,746 467,469 140,418 1,078,600 75,196 1999/00 45,004 466,250 479,479 154,384 1,100,113 88,924 2000/01 45,768 459,721 557,483 99,808 1,117,012 101,446 2001/02 45,052 454,692 555,297 107,836 1,117,825 109,830 2002/03 46,003 450,744 548,757 164,530 1,164,031 103,577

Source: National Statistics Commission. Higher education enrollments are for full-time, day students only. a. General education only

7. What do these enrollment figures imply about coverage of education programs? Clearly, coverage in preschool education as a percentage of the age group has declined sharply, whereas coverage in higher education has increased. For primary and secondary education, the evidence is mixed. The Government’s official country report for the Education for All program3 reports a gross enrollment ratio for primary education of 97.5% and a net enrollment ratio of 97.1% for 1998, with virtually identical rates for boys and girls.4 These figures are the result of official enrollment figures for primary education divided by the Government’s estimate of population for the normal primary education age group.

3 Country Report on Education for All, Bishkek 1999. 4 The gross enrollment ratio expresses the number of enrollments in a given stage of education divided by the estimated number of children in the normal age group for that stage. The net enrollment ratio is the same ratio, except that the numerator includes only enrollments of children within the normal age range for the education stage in question. Because the gross enrollment ratio includes overage children in the numerator but not in the denominator, it overstates the actual coverage of education. By excluding overage and underage children from the numerator, the net enrollment ratio attempts to correct for that error.

10

8. Firmer gross enrollment rate estimates are available for 1999, when the new population census provided fresh figures for the denominator. These are particularly valuable because they allow disaggregation by oblast. Table 2 presents gross enrollment ratios by oblast for primary and secondary education -- grades 1 through 11.5 The national gross enrollment ratio of 88.4% is surprisingly close to the 90.2% figure for grades 1 through 9, implying little attrition in the upper secondary cycle. What is particularly surprising is the low gross enrollment ratio of 77.1% for Bishkek municipality, and the high enrollment ratios for Batken and Naryn. These findings may reflect higher attrition to pursue employment opportunities in Bishkek, and the absence of employment opportunities in Batken and Naryn oblasts.

Table 3a – Grade 1-11 Gross Enrollment Ratios by Oblast, 1999/2000

1999/2000 enrollments, grades 1-11

1999 enumerated population,

aged 7-17 years

1999 gross enrollment ratio

Bishkek 104,278 135,246 77.1 % Chui 158,298 183,842 86.1 % Naryn 66,700 71,901 92.8 % Issykul 98,286 110,589 88.9 % Talas 51,116 56,778 90.0 % Osh 301,733 332,170 90.8 % Jalalabad 221,712 248,094 89.4 % Batken 101,494 109,808 92.4 % Total 1,103,617 1,248,428 88.4 % Source: National Statistics Commission.

5 Equivalent enrollment figures by oblast for grades 1 through 9 are not currently available.

11

Table 3b – Grade 1-4 Gross Enrollment Ratios by Oblast, 1999/2000

1999/2000 enrollments, grades 1-4

1999 enumerated population,

aged 7-10 years

1999/2000 gross enrollment ratio

Bishkek 42,000 49,482 84.9 % Chui 66,800 68,207 97.9 % Naryn 27,800 26,935 103.2 % Issykul 41,500 42,489 97.7 % Talas 21,400 21,672 98.7 % Osh 128,500 131,419 97.8 % Jalalabad 96,700 97,553 99.1 % Batken 41,700 42,128 99.0 % Total 466,400 479,885 97.2 % Source: National Statistics Commission.

Table 3c – Grade 5-9 Gross Enrollment Ratios by Oblast, 1999/2000

1999/2000 enrollments, grades 5-9

1999 enumerated population,

aged 11-15 years

1999/2000 gross enrollment ratio

Bishkek 46,600 58,124 80.2 % Chui 70,900 85,945 82.5 % Naryn 28,100 33,845 83.0 % Issykul 42,900 51,396 83.5 % Talas 21,700 26,167 82.9 % Osh 131,000 149,532 87.6 % Jalalabad 93,600 113,038 82.8 % Batken 44,500 50,814 87.6 % Total 479,300 568,861 84.3 % Source: National Statistics Commission.

12

Table 3d – Grade 1-9 Gross Enrollment Ratios by Oblast, 1999/2000

1999/2000 enrollments, grades 1-9

1999 enumerated population,

aged 7-15 years

1999/2000 gross enrollment ratio

Bishkek 88,600 107,606 82.3 % Chui 137,700 154,152 89.3 % Naryn 55,900 60,780 92.0 % Issykul 84,400 93,885 89.9 % Talas 43,100 47,839 90.1 % Osh 259,500 280,951 92.4 % Jalalabad 190,300 210,591 90.4 % Batken 86,200 92,942 92.7 % Total 945,700 1,048,746 90.2 % Source: National Statistics Commission.

10. Household survey results on school attendance usually provide a valuable reality check on official enrollment figures. In most countries, survey data show rates of actual school attendance that are significantly below enrollment ratios which are based on official registration. In Uzbekistan, for example, household survey responses imply rates of school attendance that are about 15 percentage points below the enrollment ratios that are calculated from official enrollment data.6 But in the Kyrgyz Republic, the results of the Kyrgyz Poverty Monitoring Surveys which were carried out in 1996, 1997, and 19987 lead to levels of school attendance for primary education that are very close to the net enrollment ratios that are reported in the Education for All report. All these estimates are sensitive to the estimates of population that are adopted for the denominator of the calculation of enrollment ratios. The Government’s reported enrollment ratios are derived from registration data on enrollments and estimated population in the school-year age groups. If one calculates denominators instead from the single-year enumerated population aged 7 through 15 from the 1999 population census, it leads to a much lower gross enrollment ratio for grades 1 through 9 of 90.2% for the 1999/2000 school year. This finding is much more consistent with findings on basic education coverage from other countries in the region. It implies a total of at least 200,000 children of compulsory school age that are not enrolled in school. 6 Uzbekistan: Living Standards Assessment, draft World Bank report, August, 2002. 7 Kyrgyz Republic: Poverty in the 1990s in the Kyrgyz Republic, World Bank report No. 21721-KG, June 2001.

13

, and proha, under which republican-budget financing provides minimal support for private financing for differential , which because . formally met goal and the consto status ofnote reports the findings of the 2001 Household Budget Survey (HBS) on education/poverty linkages, and compares them to the findings from the 1996, 1997 and 1998 Kyrgyz Poverty Monitoring Surveys (KPMS). In addition to other information on household characteristics, the 1996-98 KPMS surveys included several questions on education, including the reasons for school dropout. This information was the basis for the analysis of education/poverty linkages which was reported in the 2001 Kyrgyz Poverty Study8 and the 2002 Kyrgyz Vulnerability Study.9 The KPMS surveys were not continued after 1998, but a new household survey on living standards is being conducted in January, 2003 which uses an improved version of the KPMS questionnaire. A full analysis of the results of that survey is planned for FY 2004. In the meantime, the 2001 HBS provides the main basis for updating the findings of the 1996, 1997 and 1998 KPMS surveys on the linkages between education and poverty. Coverage of the 2001 questionnaire is more limited than for the KPMS surveys. Thus, the findings of the 2001 survey do not allow comparisons with all the findings reported in the earlier studies. The 1996-98 surveys used a common questionnaire, sampling frame, and sampling plan. All of these changed for the 2001 survey. Thus, the findings of the 2001 survey do not provide the same degree of comparability with the earlier surveys that the three KPMS provide among themselves. Nevertheless, until the results of the 2003 survey are available, the findings of the 2001 survey provide a valuable indication of recent trends in education and poverty in the Kyrgyz Republic. Education and Enrollments 2. Basic Education. Data from the 1996-1998 KPMS surveys showed that school enrollment is virtually universal for the compulsory cycle (grades 1 through 9). As shown in Table 1, enrollment rates in grades 1 through 9 varied little by gender, by household income, and by most other household characteristics in the KPMS surveys. Data from the 2001 HBS survey generally show even less variation, with virtually identical average enrollment rates for poor and non-poor, and for boys and girls. The single variable for which there are substantial enrollment deficits at the compulsory level is residential location. This is not so much a question of differences by urban/rural location as it is of differences among specific oblasts. The low enrolment rate for Bishkek (91.7%) results entirely from late age of starting school: As shown in Annex Table 1, only 21.2% of 7-year-olds surveyed were attending school in Bishkek, versus 100% for ages 8-15. In contrast, the low enrollment rate for Talas oblast (89.5%) reflects significant numbers of children not attending school at all ages between 7 and 15 – a much more serious problem in terms of poverty consequences. This is not simply a result of poverty. Although it is true that Talas oblast has the second highest headcount incidence and depth of poverty, 10 Naryn oblast, with the highest headcount incidence and depth of poverty, has a reasonably high rate of school enrollment in the basic education age cohort (95.3%), and most of its enrollment deficit results from late starts (with only 70.2% of 7 year olds in Naryn oblast attending school). Hopefully, the results of the 2003 survey will shed further light on the reasons for low enrollments in Talas oblast.

Table 1 – Enrollment-Rate Differentials in Basic Education (Grades 1-9)

8 Kyrgyz Republic: Poverty in the 1990s in the Kyrgyz Republic, World Bank Report Number 21721-KG, June, 2001. 9 Poverty and Vulnerability in the Kyrgyz Republic, 1996-1998, draft report by Jane Falkingham, Ceema Namazie, and Amani Siyam, 10 September, 2002. 10 Kyrgyz Republic, Enhancing Pro-Poor Growth, World Bank Report Number 24638-KG, August 23, 2002.

14

(in percent)

Enrollment Rate 1996 (KPMS)

1997 (KPMS)

1998 (KPMS)

2001 (HBS)

Gender Boys 96.9 95.8 97.3 96.2 Girls 96.2 94.9 97.5 96.1 Poverty status Poor 95.7 95.3 97.1 96.1 Non-poor 98.0 95.4 98.0 96.2 Educational attainment of household head

Below primary completion 93.7 93.4 95.5 96.4 Primary completion 94.3 93.7 94.6 95.8 Secondary vocational completion 97.4 96.4 96.9 93.0 Secondary general completion 96.6 95.4 98.5 97.0 Higher education completion 98.7 96.0 98.5 97.5 Residence Urban 97.3 94.6 97.5 95.1 Rural 96.3 95.7 97.1 96.6 Oblast Bishkek 98.0 93.9 98.3 91.7 Issykul 98.2 97.8 98.3 95.5 Jalalabad 92.6 96.0 95.0 98.7 Naryn 99.4 99.5 98.9 95.3 Batken 99.2 Osh 98.3 95.3 98.6 98.9 Talas 94.0 95.5 99.2 89.5 Chui 93.4 92.5 94.9 90.5 Source: Poverty and Vulnerability in the Kyrgyz Republic, 1996-1998, draft report by Jane Falkingham, Ceema Namazie, and Amani Siyam, 10 September, 2002, and data from the 2001 Kyrgyz Household Budget Survey. 3. Secondary and Higher Education. At post-compulsory levels of education, there has been more change over time and more variance in enrollments. There have been significant increases in enrollment rates for secondary and higher education since the 1998 KPMS. Enrollment rates for females remain above those for males. Higher education enrollment rates almost doubled in the aggregate, and increased almost four-fold for poor households. Uneducated parents have a high rate of compliance in educating their children through the compulsory cycle, but consistently withdraw their children from school at the completion of the compulsory cycle. The decline in enrollments is slowest for the most educated parents. As in most countries, higher education enrollment rates are highest among households whose head also completed higher education. But the 80.9% secondary education enrollment rate for households whose heads completed only primary schooling and the 38.6% higher education enrollment rate for households whose heads completed only primary schooling show that there is significant access to post-compulsory education for children of less educated parents. This development is particularly impressive because most of the growth in higher education enrollments has been in new, fee-paying places in public universities. Students from poor households either received high enough scores on the entry examination to entitle them to budget-financed places, or found the

15

means to pay for fee-paying places in public or private universities. In view of this evidence on improving access of the poor to secondary and higher education, the very low enrollment rates for secondary education (32.3%) and, especially, higher education (2.3%) for households whose heads completed secondary vocational education is an unexpected finding. It suggests a pattern of intergenerational stratification in which children of blue-collar workers rarely progress to secondary and higher education. If this pattern is confirmed in further analysis, it would call for particular efforts to raise enrollments in this target group.

Table 2 – Enrollment-Rate Differentials for Preschool and Secondary Education

(in percent)

Enrollment Rate 1996 1997 1998 2001

Secondary Education11 Gender Boys 51.0 60.4 Girls 56.3 64.8 Poverty Status Poor 44.3 57.5 Non-poor 65.0 68.8 Educational attainment of household head

Below primary completion 40.9 Primary completion 80.9 Secondary vocational completion 32.3 Secondary general completion 56.0 Higher education completion 72.5 Higher Education12 Gender Boys 18.3 34.0 Girls 19.7 37.0 Poverty status Poor 7.3 27.4 Non-poor 38.7 44.5 Educational attainment of household head

Below primary completion 0.0 Primary completion 38.6 Secondary vocational completion 2.4 Secondary general completion 19.2 Higher education completion 56.9 Source: Poverty and Vulnerability in the Kyrgyz Republic, 1996-1998, draft report by Jane Falkingham, Ceema Namazie, and Amani Siyam, 10 September, 2002, and data from the 2001 Kyrgyz Household Budget Survey. Household Expenditures on Education

11 Figures for 1996 are unweighted averages for ages 16 through 18. Figures for 2001 are weighted averages for ages 16 through 18. 12 Unweighted averages of enrollment rates for ages 19 through 21.

16

4. Household expenditures on education constitute an important part of education financing. As shown in Figure 1, average household expenditures on education by oblast vary inversely with the incidence of poverty in each oblast. Household outlays for education include textbook rental, purchase of school supplies, fees for private education, tutoring, and “contract” programs in public secondary and higher education, and contributions to schools and teachers either for specific purchases or implied services (including favorable treatment in examinations, grading, and admission).

Figure 1 – Average Household Expenditures on Education and % Poor by Oblast, 2001

0

20

40

60

80

100

120

Bishke

k

Issyk

ul

Jalal

abad

Naryn

Batken Osh

Talas

Chui

HouseholdEducationExpenditures as %of BishkekPercent Poor

Source: Data from Kyrgyz Household Budget Survey, 2001

5. The 2001 Kyrgyz Poverty Study reported that household costs of schooling are an important factor affecting school enrollments – particularly in secondary education: “In short, expenditures on schooling at the secondary level are high for all income groups, but the burden they represent to the extremely poor is almost twice that for the non-poor.”13 The regression analysis presented in that report, based on the 1998 KPMS data, found that the level of household expenditures on secondary education had a small but significant effect in increasing the likelihood of secondary school dropouts. There are also numerous indications that the costs of education deter school attendance by the poor, and that the different capacities of households to contribute to schools affect the learning environment in the school and the quality of education which it provides.14

Table 3 – Secondary Education Expenditures as a Share of Household Income, 1996-1998

Year Extreme Poor Other Poor Non-Poor 1996 21.8 % 19.5 % 16.7 % 1997 17.7 % 11.3 % 8.9 %

13 Kyrgyz Republic: Poverty in the 1990s in the Kyrgyz Republic, World Bank Report Number 21721-KG, June, 2001. 14 Kyrgyz Republic: Consultations with the Poor, Report prepared for the Global Synthesis Workshop, September 22-23, 1999, World Bank.

17

1998 20.1 % 13.9 % 9.6 % Source: Kyrgyz Republic: Poverty in the 1990s in the Kyrgyz Republic, World Bank Report Number 21721-KG, June, 2001, Table 6.7.

6. Interpretation of survey results on the levels of household expenditures on education is problematic. As shown in Table 3, the Kyrgyz Poverty Monitoring Surveys yielded very high shares of household expenditures for secondary education in 1996, 1997, and 1998 – over 20% for the extreme poor. And the 2001 Review of Social Policy and Expenditures15 reported similarly high shares of household expenditures for all levels of education – again, with higher expenditure shares for the poor than for the non-poor (Table 4). But the expenditure figures from the 2001 Household Budget Survey yield implausibly low expenditure figures – averaging just 1.4 % of household income for poor households with children attending school, and 2.0 % for non-poor households. It is difficult to reconcile the relatively modest size of reported average household expenditures for primary and secondary education in 2001 – expressed either in absolute terms or as a share of total household expenditures – with the findings of statistical analysis and beneficiary assessment that these expenditures are a significant deterrent to school attendance. It is also difficult to explain the ironic, opposite relationship observed for higher education, where enrollments have grown rapidly for all groups, including the poor, in spite of the very high levels of household expenditures reported for higher education by all income groups (Table 5). Resolution of these ambiguities and clarification of the underlying relationships will require further analysis, including analysis of the January, 2003 survey findings.

Table 4 – Total Education Expenditures as a Share of Household Income, 1997

Extremely Poor Other Poor Non-Poor All Households 16.9 % 14.1 % 13.1 % 13.6 %

Source: Kyrgyz Republic, Review of Social Policy and Expenditures, World Bank Report No. 22354 KZ, June, 2001, Volume 1, Table 17.

7. The findings of the 2001 Household Budget Survey, summarized in Table 5, show much smaller shares of household expenditures on education. They also indicate that poorer households devote a smaller share of their expenditures to education than more prosperous households – which is the reverse of the expected relationship and the relationship found in the earlier KPMS surveys. Household expenditures on higher education by all categories of households dwarf expenditures on primary and secondary education. Although poverty is much more widespread in rural areas than in urban areas,16 rural households on average spend almost three times as much as urban households on higher education (and virtually nothing on primary education). Bishkek (the most prosperous oblast) reports the lowest level of average household expenditures on higher education of any oblast. Together with the expenditure/income relationship described above, this finding suggests that it is non-poor households in rural areas that account for the bulk of this average. One explanation for the high cost of higher education for rural households (and the low cost for Bishkek, for example) is the necessity for rural households to finance housing and living expenses for their students at (largely urban) universities.

15 Kyrgyz Republic, Review of Social Policy and Expenditures, World Bank Report No. 22354 KZ, June, 2001. 16 Table 1.8, Kyrgyz Republic, Enhancing Pro-Poor Growth, World Bank Report Number 24638-KG, August 23, 2002.

18

Table 5 - Household Expenditures on Education by Level and Category of Expenditure, 2001 (households with enrolled members only, in soms per enrolled household member) Purchase and rentala Fees and Contributionsb Total ed. expend. Books Notebooks, Higher Secondary Primary in soms as % of paper, and education Education Schools per Total Stationery student h.h. expen Total 12 84 1715 45 15 395 1.6 Urban 22 104 918 51 54 497 2.0 Rural 7 75 2540 43 2 349 1.4 Oblast Bishkek 37 125 819 55 52 757 2.4 Issyk-Kul 3 70 1029 33 24 271 1.4 Jalal-Abad 7 85 1761 3 2 188 1.1 Naryn 5 67 1181 11 0 148 1.0 Batken 6 97 2995 36 7 440 1.6 Osh 11 64 2007 29 15 349 1.6 Talas 3 42 2579 9 2 264 1.5 Chui 9 104 2575 172 39 639 2.0 Economic welfare status Not poor 19 108 2303 93 21 743 2.4 Poor 7 68 851 13 12 155 1.0 Quintile 1 4 52 166 12 7 90 0.9 Quintile 2 6 74 1091 2 15 169 1.0 Quintile 3 10 78 1000 28 15 238 1.5 Quintile 4 13 99 1364 76 35 573 2.1 Quintile 5 27 124 3056 125 9 1009 2.7 a Includes all households with at least one enrolled member at any level of the three listed levels of education. b For each level of education, includes only households with at least one enrolled member in that level. c Includes only levels of education in table, I.e. excludes pre-school education. 8. The quintile average expenditures for education at all levels indicate that demand for education is highly income elastic. Since fees for private universities and for contract places in public universities are ostensibly uniform for all students, the much lower average expenditures for higher education by poor households imply either that students from poor households are more successful in obtaining the free (budget-financed) places in public universities than students from non-poor households, or that students from higher-income households consistently enroll in the more expensive programs in public and private universities. This income/expenditure relationship is obscured when one looks at data on place of residence.

Differences in Education Quality 9. Household expenditures on education have a strong effect on enrollments, but their effect on education quality is of even greater concern. Because budget resources for education are essentially limited to financing teacher salaries, parents are routinely asked to make contributions

19

to schools in order to finance school maintenance, school recurrent expenditures, and the purchase of crucial educational materials such as library books and computers. As a result, schools in more prosperous communities tend to provide a better teaching and learning environment than schools in poorer communities. One conspicuous example concerns fuel for heating: Classrooms in schools in the poorest communities often lack functioning stoves or fuel, and are too cold for effective teaching and learning for much of the year. This problem is especially acute at high altitudes, where the incidence of poverty is greater and where the winter weather conditions are most severe. Even within schools, differences in household income lead to different education quality because poor households often cannot afford to rent textbooks or purchase school supplies for their children,17 and are more likely to be undernourished and to miss school to help with income-earning tasks. 10. As a result of these factors, the differences between the best and the worst schools are very great – representing enormously different conditions for teaching and learning effectiveness. Student assessment results illustrate the effects of these differences on student learning achievement. In general, the incidence of poverty and secondary scores move inversely, with Bishkek and Chui oblasts having the lowest incidence of poverty and the highest secondary examination results (Figure 2). Test results for grade 5 do not correlate as closely with poverty incidence by oblast. In particular, Batken oblast has a much lower pass rate in the grade 5 test than its poverty incidence would suggest. This could be related to learning problems for non-Kyrgyz or Russian-speaking population, which is more numerous in Batken than in other oblasts. Speaking a language other than Russian or Kyrgyz was found in the 1998 KPMS to lead to higher probability of school dropout.18 This ethnicity effect on learning achievement may not be as apparent in the secondary examination results because self selection after completion of the compulsory cycle effectively eliminates most students whose native language is other than Kyrgyz or Russian.

Figure 2 – Percentage Poor, Grade 5 Test Pass Rate, and Secondary Examination Scores by Oblast

17 The author visited one rural primary school in Chui oblast in which only 40% of the students had textbooks. When asked why this was so, the principal replied that the other students could not afford to rent textbooks. She also explained that the school text-book rental library had enough books for another 40% of the students. When asked why the school did not make these books available to some of the students without textbooks, she replied that that would undermine the incentive for other students to rent their books. 18 Kyrgyz Republic: Poverty in the 1990s in the Kyrgyz Republic, World Bank Report Number 21721-KG, June, 2001.

20

0

20

40

60

80

100

120

140

Bishke

k

Issyk

ul

Jalal

abad

Naryn

Batken Osh

Talas

Chui

Percent Poor

Pass Rate in Grade5 Test2002 SecondaryScores

Source: Table 2.1, Kyrgyz Republic, Enhancing Pro-Poor Growth, World Bank Report Number 24638-KG, August 23, 2002; Ministry of Education and Culture data, and Monitoring Learning Achievement: National Survey of Primary Education Quality, Kyrgyz Ministry of Education and Culture, UNICEF, UNESCO, and the Center for Public Opinion Studies and Forecasts; Bishkek, 2001. Education and Earnings 11. Based on the findings of the 2001 Household Budget Survey, the 2002 Kyrgyz Poverty Report showed a generally inverse relation between duration of completed education and the incidence of poverty (Table 6).

Table 6 – Headcount Incidence of Poverty by Education of Household Head, 2001 Education level Absolute Poverty (%) Extreme Poverty (%)

Higher/incomplete higher 35.8 9.9 Secondary vocational 52.6 21.8 Secondary general 66.7 32.4 Incomplete secondary 56.3 27.2 Primary 68.1 36.3 Below primary 78.0 19.1 All households 56.4 24.7

Source: 2001 HBS. Weighted estimates. 12. Converting the income data from the 2001 HBS into same years of schooling and estimating the returns to education and work experience yields the following regression (in which all coefficient estimates are significant at the 1% confidence level):

ln monthly income = α + .070 years of schooling + .040 experience - .001 experience2 + ε Education and experience contribute significantly to earnings, with the effect of experience diminishing over time. Education is thus an effective instrument for poverty alleviation.

21

Benefit Incidence Analysis 13. Combining the expenditure and education unit cost information prepared for the recent Public Expenditure Review19 with the education and poverty findings of the 2001 HBS leads to the incidence analysis presented in Table 7. Public expenditures on primary education predominately benefit the poor and rural population. Public expenditures on secondary education also predominately benefit the poor and rural population, but less so than expenditures on primary education. Public expenditures on higher education primarily benefit the non-poor and the urban population. Thus, public expenditures on primary education are the most pro-poor form of public expenditures on education.

Table 7 – Benefit Incidence of Public Expenditures on Education by Level of Education and Poverty Status of Beneficiaries, 2001

Primary Secondary Higher By Income Quintiles 1st Quintile 28 % 21 % 7 % 2nd Quintile 22 % 20 % 9 % 3rd Quintile 21 % 22 % 24 % 4th Quintile 16 % 20 % 23 % 5th Quintile 13 % 17 % 37 % By Poverty Status Poor 67 % 58 % 35 % Non-Poor 33 % 42 % 65 % By Residence Urban 27 % 32 % 58 % Rural 73 % 68 % 42 %

19 Kyrgyz Expenditure Review: Fiscal Policies for Growth and Poverty Reduction, December 16, 2002 draft, The World Bank.

23

Annex Table 1 – Enrollment Rates by Age and Selected Household Characteristics, 2001

Age in years 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 7-15 16-18 19-22 Total 15.7 79.3 97.2 98.0 97.7 99.1 98.9 98.4 97.7 96.7 92.4 57.3 38.0 39.5 37.5 29.2 23.6 96.1 62.9 32.7 Urban 14.2 57.6 97.3 96.2 100.0 99.5 98.9 99.8 100.0 98.3 90.4 77.6 58.0 59.2 60.4 56.8 35.6 95.1 75.7 52.0 Rural 16.5 87.0 97.2 98.8 96.7 99.0 98.9 97.9 96.8 96.0 93.4 45.1 27.9 28.5 27.7 12.2 14.0 96.6 55.9 21.3 Oblast Bishkek 3.1 21.2 94.5 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 93.9 84.9 71.6 73.5 78.3 40.6 91.7 92.7 62.2 Issyk-Kul 3.3 62.7 95.9 100.0 100.0 100.0 100.0 100.0 97.9 95.5 90.8 71.9 47.2 51.9 60.0 31.8 25.8 95.5 72.6 41.2 Jalal-Abad 0 100.0 96.4 96.8 98.3 98.5 100.0 100.0 98.0 100.0 100.0 46.6 7.8 8.2 19.0 0 5.2 98.7 53.2 7.9 Naryn 10.9 70.1 100.0 100.0 100.0 100.0 93.1 100.0 97.0 94.3 97.6 30.5 17.4 10.1 22.2 5.2 12.9 95.3 55.2 12.6 Batken 0 100.0 94.8 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 14.7 26.9 23.3 15.8 14.2 0.4 99.2 43.3 14.6 Osh 43.4 100.0 100.0 100.0 100.0 100.0 98.8 100.0 98.1 94.7 84.9 42.4 38.3 38.7 49.4 11.5 13.2 98.9 56.8 30.5 Talas 3.1 78.2 91.6 94.5 93.1 87.5 93.4 95.3 82.0 88.4 87.9 56.2 7.4 23.2 9.7 8.8 8.2 89.5 51.5 12.6 Chui 11.3 45.4 100.0 90.2 88.0 100.0 100.0 89.5 97.3 96.4 85.5 83.4 35.6 57.3 27.8 51.9 38.5 90.5 66.6 43.3 Gender Female 12.7 75.1 96.5 99.2 96.5 99.5 99.2 99.7 97.1 97.0 93.8 62.1 38.6 41.8 41.5 27.7 23.7 96.1 65.2 33.9 Male 18.7 82.8 97.9 96.6 99.0 98.7 98.6 97.1 98.3 96.4 91.0 52.4 37.5 36.9 34.1 31.0 23.5 96.2 60.6 31.4 Economic welfare status Not poor 23.5 72.5 97.7 96.1 97.3 99.8 100.0 100.0 98.0 97.4 95.9 64.0 46.5 46.6 45.4 41.5 34.9 96.2 68.1 41.9 Poor 12.6 82.4 96.9 98.7 97.9 98.7 98.3 97.5 97.5 96.2 89.9 51.7 30.9 32.4 30.4 19.5 7.7 96.1 58.7 23.4 Quintile 1 11.7 83.3 97.4 100.0 99.5 99.5 97.3 98.5 96.8 96.6 86.1 48.4 26.6 21.1 15.7 9.9 1.9 96.6 55.4 13.1 Quintile 2 14.5 78.2 94.2 98.9 96.2 99.1 100.0 96.3 96.3 94.9 94.8 59.5 20.8 19.0 27.9 10.1 4.2 95.0 59.5 14.9 Quintile 3 12.3 82.1 100.0 95.9 97.8 97.5 98.2 97.8 100.0 97.8 87.7 47.6 48.4 56.7 41.1 53.8 26.5 96.4 61.1 44.6 Quintile 4 22.9 71.2 99.2 100.0 94.8 99.6 100.0 100.0 99.5 98.8 93.6 68.2 43.4 35.6 40.1 29.1 28.5 96.4 70.0 33.7 Quintile 5 25.2 76.6 95.7 93.2 100.0 100.0 100.0 100.0 95.8 95.3 100.0 62.1 46.2 58.3 57.2 52.5 37.3 96.2 67.7 50.3

Educational attainment of household head

24

Higher 21.0 75.6 100.0 100.0 100.0 100.0 100.0 100.0 99.4 98.8 96.3 66.9 50.0 66.5 72.0 60.5 34.7 97.5 72.5 56.9 Paraprofessional 25.0 68.7 98.3 98.1 95.8 98.1 98.7 94.8 95.8 94.2 94.2 69.8 41.3 45.5 45.7 27.7 27.8 94.3 68.0 37.6 Complete secondary 7.4 85.7 96.6 97.9 97.7 99.5 99.3 99.9 99.0 97.6 89.5 45.2 30.2 24.2 20.1 19.3 11.9 97.0 56.0 19.2 Incomplete secondary 18.9 74.3 100.0 91.6 100.0 100.0 100.0 100.0 86.8 100.0 64.0 56.7 0 4.5 0 0 4.7 93.0 32.3 2.4 Primary 8.7 90.8 86.5 100.0 100.0 100.0 92.0 100.0 100.0 100.0 100.0 79.1 63.7 0 0 0 60.8 95.8 80.9 38.6 Below primary 34.8 87.5 100.0 100.0 100.0 100.0 100.0 100.0 91.3 88.3 100.0 5.0 48.7 0 0 0 0 96.4 40.9 0

Source: 2001 Kyrgyz Household Budget Survey (weighted results) WB05718 C:\Documents and Settings\wb05718\My Documents\Kyrgyz Republic\Education and Poverty in the Kyrgyz Republic, revised draft.doc MMertaugh, January 31, 2003 9:59 AM