Report No. 33102-ECA Operationalizing the Health and Education Millennium Development Goals in Central Asia (In Two Volumes) Volume II: Kyrgyz Republic Health and Education Case Studies June 2005 Human Development Sector Unit Europe and Central Asia Region Document of the World Bank CONTENTS Acknowedgements ........................................................................................................................... iv OPERATIONALIZING THEMILLENNIUM DEVELOPMENT GOALSFOREDUCATION: A CASESTUDY OFTHEKYRGYZ Executive Summary ................................................................................................... REPUBLIC 1 1 Status ofMDGAttainment .............................................................................................................. . 9 9 Education Coverage: HighEnrollment, L o w Attendance......................................................... Objectives and Indicators.......................................................................................................... 9 Incomplete School Attendance ............................................................................................... 12 13 2 The Sources o f Unsatisfactory Education Performance.............................................................. . Deficiencies inEducation Quality .......................................................................................... 18 Incomplete School Attendance ............................................................................................... 18 3 Improving Education Performance............................................................................................... 26 . L o w Learning Achievement ............................................................................ 20 The Current Approach ...................................................................................... 26 Achieving Fuller Attainment of the Expanded Education MDG............................................ 26 OPERATIONALIZINGTHEMILLENNIUM DEVELOPMENT GOALS FORHEALTH: A CASESTUDY OFTHEKYRGYZ REPUBLIC Executive Summary ............................................................................................................................... 32 Introduction............................................................................................................................................ 36 1 MonitoringMDGHealthOutcomes toIdentifyGroups that are notAchieveingthe . M D G s ............................................................................................................................................... 37 Maternal and ChildHealth...................................................................................................... 37 Communicable Diseases: TB, HIV/AIDS and Brucellosis inthe Kyrgyz Republic................42 44 2 Identifying Interventions that are Effective for Raising MDGPerformance............................ 51 3 Targeting Interventions to Groups that are not Meeting the HealthM D G s............................. .. Non-communicable Diseases ................................................................................................... 54 Levels and Optimal Allocation o f Public Financing for the Health Sector ............................ 54 Addressing Service Delivery Bottlenecks inReaching Disadvantaged Groups with the Right Interventions............................................................................................................ 60 4: Strengthening Incentives for Providers and Users to Raise MDGOutcomes........................... 61 References ...................................................................................................................................... Conclusion ...................................................................................................................................... 62 67 B O X E S Education Boxes B o x 1: Factors Affecting Learning Achievement: The International Evidence .............................. 22 Health Boxes Box 2: Cost-Effective Interventions and Service Modalities for Achieving the Health MDGs...............51 Box 3: Community Action for Health inthe Kyrgyz Republic . Scaling Up the "Jumgal Model" to Achieve the HealthM D G s............................................................................................................. 53 i FIGURES EducationFigures Figure 1: LearningAchievement andPoverty IncidencebyUrbadRural Location ........................... 15 Figure 2: Learning Achievement and Percent Ruralby Oblast ................................................... 16 Figure 3: MeanUniversityAdmission Scoresand RuralProportion o fUniversityApplicants Figure 4: Low LearningAchievement and Poverty Incidence by Oblast ........................................ by Oblast, 2004 ......................................................................................................... 16 17 Figure 5: Average Household Expenditureson Education and Percent Poor by Oblast. 2001 ..................25 Figure 6: IMR 1997. (per 1.000 live births)................................................................................... HealthFigures 2004 38 Figure 7: Rayons and Towns with IMRAbove Oblast Average 1997-2004 ............................................ 39 Figure 8: U5MRby Oblast (selected years) .............................................................................................. 39 Figure 9: MMR per 100,000 Live Birthsby Oblast (selected years) ........................................................ 41 Figure 10: Notification Rate...................................................................................................................... 43 Figure 11: TB Mortality Rate inthe Kyrgyz Republic, With and Without Prisons.................................. 43 Figure 12: Distribution o f Visits (Primary Care + Outpatient Specialist) ................................................. 52 Figure 13: The Health Sector HasBeen Getting a Smaller and Smaller Share of the Government Budget Since 1996 (Health Expenditures as Percent o f Total Public Expenditures).......................................... 56 Figure 14: Local Governments Reduce Allocations for the Health Sector inOblasts Implementing HealthReforms....................................................................................................................................... 56 Figure 15: Public Expenditures on Healthare Not Pro-poor (Per Capita HealthExpenditures at 1995 Prices) ..................................................................................................................................................... 58 TABLES EducationTables Table 1: Achieving Fuller Attainment o f the ExpandedEducationMDG ........................................ 6 Table 2: Indicators o f Education Coverage inGrades 1-9 (inpercent) ......................................... 10 Table 3: Enrollment Ratesby Age and Selected Household Characteristics. Based on the 2001Household Budget Survey ..................................................................................... 11 Table 4: Number o f Children and Adolescents Aged 7-17 Who Never Attended School ........................ 13 Table 5: Incidence o f L o w LearningAchievement by Oblast. Eighth-Grade Assessment Results. October 2002 ................................................................................................ 14 Table 6: L o w Learning Achievement and Percent o f Rural Enrollments by Oblast. 2002 ...................15 Table 7: Average Scores in 8" Grade Science and Mathby Language o f Instruction. 2002 .................... 18 Table 8: Probit Estimates o f School Dropouts (1997198 Kyrgyz Poverty Monitoring Survey) ................20 Table 9: The Evolution of Actual and Planned Public Expenditures for Education as a Share of Total Public Expenditures and GDP. 1990-2007 ............................................................. 23 Table 10: Primary and Secondary Education Expenditures as a Share o f HouseholdIncome. 1996-1998 ............................................................................................................. 25 HealthTables Table 11: Achieving Fuller Attainment o f the ExpandedHealthMDG................................................... -33 Table 12: Comparison Across Oblasts o f Caloric Intake. Poverty and Anemia Rates.............................. 42 Table 13: Prevalence o f Brucellosis per 100,000 population, by oblast, 1990-2004 ................................ 44 Table 14: Cost-Effective HealthInterventions for Reaching HealthMDGsand Coverage inthe KyrgyzRepublic..................................................................................................................................... 46 11 Table 15: The Program o f State Benefits and Other HealthCare Benefits for Vulnerable Groups inthe Kyrgyz Republic Fundedthrough Republican and Local Government Contnbutions .......................... 52 Table 16: Budget Executionfor HealthInsurance Programs for Vulnerable Groups as Compared with Other Institutions, HealthPrograms ....................................................................................................... 57 Table 17: Budget Executionfor Children, Pensioners and Persons with Social Benefits Health Insurance ................................................................................................................................................ 58 Annex Table 1: Kyrgyz Republic: Health, Nutrition and Population, and Poverty, Total Population (Findingsfiom the DHS, 1997) .............................................................................................................. 63 Annex Table 2: List o f Organizations whose Representatives were Interviewed to Identify Bottlenecks and Interventions: ............................................................................................................... 66 111 . . I ACKNOWLEDGEMENTS This report was prepared by a team led by Sarbani Chakraborty (ECSHD), Team Leader and principal author (Kyrgyz Health MDG Case Study and Chapter 2). The team included: Michael Mertaugh (Kyrgyz Republic education case study), Elina Manjieva, Alan Wright, Shweta Jain and Julie Wagshal (all ECSHD) and Adilet Meimanaliev (Consultant). The peer reviewers for the study are: Sebnem Akkaya (Country Economist, ECSPE), Eduard Bos (Lead Population Specialist, HDNHE) and Rosemary Bellew (Lead Education Specialist, HDNEDU). The team benefited greatly from comments provided by the Quality Enhancement Review (QER) Panel consisting o f Daniel Ritchie, Akiko Maeda, Agnes Soucat, Robert Prouty, Rama Laksminarayan and Mercy Miyan Tambon. Comments were also provided by Chris Lovelace, Arup Banerji, ArminFilder, MaureenMcLaughlin and Peyvand Khaleghian. iv OPERATIONALIZINGTHE MILLENNIUMDEVELOPMENTGOALS FOREDUCATION: A CASE STUDY OF THE KYRGYZREPUBLIC EXECUTIVE SUMMARY Interpretingthe EducationMDGfor the Kyrgyz Republic i. The audience of this casestudyisGovernment anditspartnersinthe educationsector inthe Kyrgyz Republic. Its primary objective i s to make policymakers and other partners aware o f the unmet needs for attaining the education MDG, and to offer practical advice about options for meetingthem. A secondary objective i s to illustrate how the Ministry o f Education's new Education Budgeting and Strategic Planning Unit might approach the design o f interventions to meet specific educational objectives. ii. TheMDGsshouldbeintevpreted inthecontextof theCentralAsiansetting. Twooftheeight Millennium 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 o f primary schooling, and b) to eliminate gender disparity in primary and secondary education preferably by 2005, and at all levels by 2015. This study endorses the Kyrgyz Government's recommendation that the MDG goals and indicators should be interpreted in light of their broader development objectives. The recent Country Report on MDGimplementation' recommends that the education MDG be interpreted to consist of universal completion of an eleven-year cycle o f primary and secondary education. The MDG Country Progress Report also proposes that future monitoring o f education MDG attainment should cover education quality as well as enrollment. This case study reflects these recommendations o f the Government and partner MDGWorking Group. It interprets the education MDG for the Kyrgyz Republic to include universal completion o f an eleven-year cycle o f primary and secondary education with acceptable learning outcomes. Because school attendance i s a requirement for effective learning, it also includes full school attendance as a subsidiary goal.' This case study examines the needs for attaining this expanded education MDG objective, and proposes two sets of policy options to attain this goal (summarized inTable 1, below). Although the specific recommendations o f the case study pertain to the Kyrgyz Republic, it i s hoped that the framework of the case study will be applicable to other countries o f Central Asia, and to other sectors as well. iii. TheunfinishedMDGeducation agenda in theKyrgyz Republic involves raising attendance in basic education, raising education quality in the lowest performing schools, and raising secondary enrollments and attendance. As documentedbelow, enrollment inthe 9-year compulsory education cycle i s nearly complete, but falls well short o f the goal o f full coverage in 11-year primary and secondary education. Less visible but more threatening to the expanded MDG education goal are the high level o f non-attendance at the compulsory cycle, and the marked disparities in learning achievement across regions, which involves learning achievement below national standards in the lowest-performing schools. TheKyrgyz Republic: Millennium Development Goals Progress Report, Bishkek,ZO03. Since the educational dimensions of the gender MDGhave been attainedinthe Kyrgyz Republic, the Government andpartners have agreed to interpret this goal interms of equal employment opportunities for menand women. The goal of gender equality i s thus not addressed inthis education case study. 1 Attainment of the expanded education MDG would require very deliberate interventions to close these gaps. The Requirements for FullAttainment ofthe EducationMDG iv. The Government is doing much to promote the expanded education MDG, but could do more, 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 o f 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. Inthe case of the Kyrgyz Republic, public expenditure levels and patterns and the design of education programs indicate that the expanded education MDG is a serious priority o f the Government. A number o f activities are already under implementation by the Government which should contribute to attainment of the expanded education MDG. These include the support which is being provided under the Rural Education Project for introduction o f teacher performance incentives, for provision o f textbooks and educational materials, for strengthened student assessment, and for school grants. Inorder to fully attain the education MDG, these actions will needto be fully implemented, and brought together into a coherent results-oriented strategy, with additional actions specifically targeted to the groups with incomplete school attendance and low learning achievement. v. For reasons explained below,$ve generic actions are neededfor full attainment of the education MDG: e monitoringperformance, e targeting interventions to lowestperforming schools, e providing incentivesfor improvedperformance, e evaluatingpilot actions to identzJL what works, and e using budget resources selectively. vi. The first four generic actions are necessary in order to identify schools and students with sub- standard performance, to motivate and implement targeted actions to improve their performance, and to identify the educationally most cost-effective interventions for improving performance. The fifth generic action i s necessary in order to mobilize the necessary budget resources to improve sub-standard performance. As the case study documents, the schools in the poorest localities and regions have the lowest educational performance in terms o f 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. (Indeed, it i s their lack o f capacity to provide supplemental financing for their schools that explains a large part o f their poor performance in the first place). For that reason, targeted interventions to raise the learning achievement o f all students and schools to national standards would need to be financed with central budget resources. In an environment of scarce resources, mobilizing resources for targeted interventions to improve school attendance and learning achievement would require selectivity inuse o f central budget resources. vii. These actions to improve attainment of the education MDG can be applied in different levels of intensity. Table 1 summarizes what the Government i s already doing to promote attainment of the expanded education MDG, and presents options for fuller attainment o f the goal. It presents these actions under each of the five generic actions described above. Column 2, the base case scenario, summarizes actions which are currently under implementation which will promote attainment o f the expanded education MDG. Column 3, the intermediate option, presents an intermediate range o f actions which would lead to accelerated attainment o f the expanded education MDG. Column 4, the high-case option, 2 presents a high-commitment approach to attain the expanded education MDG. Actions are also distinguished by time frame (short-term and medium-term actions). A number o f these are admittedly "stretch" goals that would require a well conceived and consistently implemented advocacy program by the Government in order to ensure compliance by all parties - including the high-performance schools and regions whose education budgets could be reduced under these actions. The high-case approach includes adoption o f management incentives and a financing formula that give preference to schools in low-income areas (because o f their greater needs and lesser capacity to contribute from other sources). All recommended approaches involve greater use o f data on education outcomes and gradual improvement o fthese data over time. viii. Adopting this perjiormance-based approach to the education MDG would require a paradigm shiftfor the Government and thepublic -a change inwhich education performance i s no longer measured interms o fregistered enrollments, but interms of school attendance, learning achievement, and eventual labor-market outcomes, and in which performance o f the education system i s no longer judged by the success o f its best students, teachers, and schools, but its success in raising the performance o f its lowest- performing students, teachers, and schools to national learning standards. To the extent that public attention currently focuses on learning outcomes, it does so mainly for the right-side tail o f 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 o f its strongest parts - the Olympiad winners. Further progress on the MDGs and success in the broader objectives o f education would require a focus on left-side tail o f the distribution -the lowest-performing schools, teachers, and students. The Rationale for the ProposedApproach ix. Much of the shortfall in attainment of the educational MDG is attributable to resource and income constraints. As budget transfers from Moscow ended with the collapse o f the Soviet Union, the newly independent Government of the Kyrgyz Republic reduced the duration o f compulsory education from eleven years to nine years, shifted much o f the responsibility for financing public primary and secondary education to local go~ernrnents,~and authorized schools to raise money from parental contributions and from the rental o f facilities and other income-generating activities. Students were required to purchase textbooks that formerly had been provided free. Central budget financing was restricted to teacher salaries (which nonetheless fell sharply real terms) and school heat and electricity. These changes in financing had the virtue o f keeping schools in operation during a period o f unprecedented budgetary collapse, but they also led to a serious degradation o f the school teaching and learning environment in communities which were too poor to supplement central budget financing with contributions from parents and local governments. Average learning achievement in these schools often does not meet national standards, and some students have either dropped out o f school or do not attend school for a significant part o f the school year. Deficiencies inthe learning environment at school include the lack of essential teaching and learning materials inthe classroom, under-qualified and inexperienced teachers, and unfilled teaching positions in certain subject areas. These school-based impediments to learning are reinforced by income constraints at the household level which make it difficult for low- income students to afford textbooks and school materials, and sometimes lead to the necessity of child labor (and, consequently, to lower school attendance). Poorer communities also typically provide lower levels o f support to learning outside school due to lower levels o f parents' education and lower density o f educational stimuliand resources o f all kinds. Inspite ofthe intention of decentralizedfinance andmanagement ofprimary and secondary education, the scarcity of local government revenues has necessitated continued central budget financing o f core education expenditures - including, crucially, teachers' salaries. 3 x. The Government's approach to addressing the shortfall in attainment of the education MDG focuses largely on increasing resources for primary and general secondary schools by raising central budget financing for education as a share of GDP. The 2003 MDG Country Progress Report for the Kyrgyz Republic cites grade 1-9 enrollment ratios of 85.3% for 1993/94 and 95.0% for 2000/01. On the basis of that increase, it projects 100% coverage in grades 1 through 9 by 2015. But it also raises a cautionary note on the problems o f incomplete attendance, early dropouts, and quality disparities in education. It reports that these problems are especially widespread inrural areas, and attributes them to low household income. To address these problems, the MDGCountry Progress Report advocates a return to the level o f budget financing that the sector received in the latest years o f the Soviet period, which would require average GDP growth rates o f 6% per year through 2015, and an increase inthe Republican budget share for education from 3.9 % in 2001 to 6.5% in 2015.4 The Government i s making a serious effort to meet this goal. The Medium-TermBudget Framework (MTBF) for 2005 to 2007 calls for an increase inthe consolidated budget for education from 4.2% o f GDP in2004 to 5.3% o f GDP in2007.5 xi. More budget resources for education are a necessary but insuflcient conditionfor full attainment of the educationMDG;full attainment would also require a different approach to allocation of education budgets to schools -- based on equal performance rather than equal inputs. The current approach to education budget allocation treats all schools equally - as through they had equal needs, equal performance, and equal access to other resources. This approach essentially promotes a level playing field in terms of publicly provided inputs. But since some communities supplement central budget resources with very significant contributions from other sources, this equal-input approach leads to average levels o f learning achievement that are distinctly unequal, as the Monitoring Learning Achievement surveys document. It also results in allocations to schools that are, by themselves, only marginally keeping schools alive. For schools that rely exclusively on central budget financing, these allocations lead to substandard learning conditions and substandard learning outcomes. A strategy that gives first priority to attainment of the education MDGwould needto allocate education budgetresources very differently, giving priority to schools with sub-standard learning achievement. This would mean introducing a deliberate slope to the playing field in terms o f inputs in order to ensure that the lowest- performing schools meet national standards for learning achievement. Inthe current situation o f severely constrained budgets, it would also mean reducing - in some cases, possibly even eliminating - central budgetallocations for higher-performing schools with accessto other sources o f financing. xii. Full attainment of the education MDG would also require that central budgets for education finance a broader range of interventions. If the Republican budget for education were to continue to finance only teacher salaries and school utilities, the increase inbudget allocations which the Government has proposed would serve largely to raise the level o f teacher salaries. Higher teacher salaries are certainly needed, since average official teacher salaries are currently less than half the average salaries in public administration, and below the $1.OO per day poverty threshold. These extremely low salary levels have eroded teachers' morale and motivation, induced many teachers to take additional jobs, and exacerbated problems o f corruption. But budget resources are also need to cover other essential inputs to the learning process. For schools and communities which are too poor to be able to finance significant education inputs from their own resources, central budget resources would need to cover all essential requirements for effective teaching and learning. This means central budget financing not only of teachers' salaries and school utilities, but also o f textbooks, teachers' manuals, supplementary reading materials, and teaching and learning materials for classrooms. In addition to getting these classroom learning resources to all schools with sub-standard performance, full attainment o f the educational MDG would require that the central budget support whatever other interventions (such as extracurricular TheKyrgyz Republic: Millennium Development Goals ProgressReport, Bishkek, 2003, page 18. Table 11, KyrgyzRepublic Ministry o fFinance,Medium-termBudget Framework, 2005-2007, Bishkek,2004. 4 tutoring, student transport, or revision of the school calendar) may be necessary to ensure that all students attend school and meet national learning standards. These actions would need to be tailored to the particular needs o f each school. xiii. Full attainment of the education MDG may also require reallocating resources from other levels of education. If full attainment o f the education MDG required additional resources beyond those available from reallocation within the budget for primary and general secondary education, additional budget resources could be mobilized by reallocation from other education sub-sectors - in particular, form vocational secondary and higher education. Cost savings could be obtained by converting some vocational secondary schools to secondary general schools (which have significantly lower unit costs) and relyingmore heavily onprivate and local financing for highereducation. xiv. Better monitoring and evaluation of school Performance - particularly, in theform of external assessment of learning achievement - will help to identifi the target groupsfor MDG interventions and to design appropriate interventions to improve performance, but existing information provides a sufJicient basis to start theprocess. External assessmento f school performance i s currently limited to two sources: the Monitoring Learning Achievement Surveys which were carried out for 4' grade students in 2001and for gthgrade students in 2003, and the National University Entrance Examination which has been carried out annually since 2003. Although it will be important to improve assessment instruments over time, these sources provide a sufficient basis for starting the MDG targeting. At the start o f the process, the performance focus could be implemented using existinginformation on regional differentials in learning achievement and school attendance. Progressively, the targeting and design o f interventions should be refined on the basis o f better information on educational outcomes at the school level. Since income differences play such a large role in explaining differences in the learning environment, poverty data provide a good proxy for learning achievement and could be a valuable supplementary source of information for targeting. They are also available for the Kyrgyz Republic on a highly disaggregated basis. xv. Sustainability of theproposed results-oriented approach would require a more integrated M D G strategy. Currently, most of the targeted actions that are designed to improve performance o f low- performing schools are developed and implemented on a limited scale under projects financed by external partners. Often, the lessons of the pilot projects do not translate into system-wide improvements because the experiences o fthe pilot actions are not evaluated and usedto refine the design of the interventions and to replicate them on a wider scale. In order to be cost-effective and fully sustainable, these actions need to be brought together within an integrated MDGstrategy with clear performance goals and strong results monitoring and evaluation of interventions, and a process for wider replication o f pilot initiatives which are found to be cost-effective in improving school attendance and learning achievement. Ideally, these actions should be financed under the permanent budget process itself, rather than through ad-hoc, partner- financed projects. xvi. An integrated and effective M D G strategy would help mobilize additional donorfinancing. The proposed integrated approach does not mean that partner support i s not welcome or should be phased out. On the contrary, it would help mobilize more donor support ifthe Government i s able to put in place an effective mechanism for monitoring school performance, for targeting budgets and interventions to schools which are not meeting national learning standards, and for carrying out evaluations o f pilot actions to identify the most educationally cost-effective interventions to improve sub-standard school performance. Programmatic donor assistance could support this process, as in the health-sector SWAP6 which i s currently inpreparation. Sector-Wide Approach 5 Table 1: Achieving Fuller Attainment ofthe ExpandedEducationMDG 11. BASE CASE I 2.OPTlON INTERMEDIATE 3. HIGH-CASE OPTION MonitorPerformance I Short Term Conduct annual MLA surveys o f student achievement. Develop and pilot school report cards which are to include community monitoring o f non- attendance by students and teachers. Longer- Term Strengthen school-based assessment. Target Interventions Short Term Exempt 10% o f poorest students inall schools from textbook rental charges. Provide student social support (for clothing, transport, etc) for poorest students. Provide salary supplements for teachers inremoterural schools. Provide assignment grants for new rural school teachers in subjects with unfilled vacancies. 6 Longer-Term Provide Incentives for Improved Performance I Short Term Develop a new performance evaluation scheme for teachers which reflects teachers' effectiveness in improving learning achievement. Develop a new scheme o f salary incentives for highest performing teachers under the new performance evaluation scheme. Longer-Term Pilot the new performance evaluation scheme for teachers which reflects teachers' effectiveness in improving learning achievement. Pilot the new scheme o f salary incentives for teachers under the new performance evaluation scheme. Use BudgetResources Selectively Short Term Longer- Term 8 1. STATUSOFMDGATTAINMENT Objectivesand Indicators 1. Two o f the eight MillenniumDevelopment 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."7 The most recent MDGProgress Report for the Kyrgyz Republic,' interprets primary education to be the first four years of nine-year compulsory education, and on that basis reports that the education MDGhas been attained. It proposes that the MDG goals and indicators be interpreted inlight o f their broader development objectives. For countries which have met the literal MDG objectives, it advocates reinterpreting the MDGs to focus attention on unresolved needs related to the original goals. Inthe case o f the education MDG, it proposes that the goal be restated to consist of universal completion of a nine-year cycle of primary and secondary education. For the gender MDG, it proposes that the MDG focus on gender equality in employment rather than gender equality in education (which has been attained in the Kyrgyz Republic). The MDG Country Progress Report also proposes that future monitoring of education MDG attainment should cover education quality as well as enrollment. 2. The case study adopts this broader interpretation of the education MDG- to include the goals o f secondary education completion and universal access to education o f acceptable quality. This broader interpretation has been agreed through an extensive process o f consultation with Government and other partners, and i s more relevant to the country's actual situation and its aspirations. The case study does not directly address the gender MDG, which, under the proposed interpretation, largely transcends the reach o f education policy. As Table 3 records, female enrollments are generally as highas male enrollments at the compulsory level, but tend to fall behind male enrollments at higher levels. Female labor-force performance i s consistently below that for males in terms o f participation rates, earnings, and representationinmany professional occupations. (Teachers, however, are predominately female.) EducationCoverage: HighEnrollment,Low Attendance 3. Coverage at the National Level. Table 2 summarizes the available data on the extent o f coverage in basic education in the Kyrgyz Republic. The second column shows the gross enrollment ratiogestimates derived from registered enrollments and estimated population inthe 7-15-year age group. This series, which is also the basis o f the estimates provided inthe 2003 MDG Country Progress Report and the National Poverty Reduction Strategy, shows a sharp drop incoverage of the compulsory cycle of education early inthe transition, followed by a steady improvement through the end of the decade, and a slight drop in overall coverage since. On the basis o f the increase in coverage through the end o f the decade, the 2003 MDG Country Progress Report projects 100% coverage in grades 1through 9 by 2015. As a measure of completeness of enrollments, these gross enrollment ratio figures are subject to three sources o f error: First, they are biased upward because teacher positions and budgets depend upon registered enrollments. Second, they tend to overstate actual coverage because they include overage * http://ddp-ext.worldbank.org/ext/MDG/homePages.do.Republic: UnitedNationsDevelopmentProgramme, TheKyrgyz Millennium Development Goals Progress Report, Bishkek, 2003. The gross enrollmentratio expressesthe number of enrollments ina given stage of educationdivided by the estimatednumber of childreninthe normal age group for that stage. The net enrollmentratio i s the same ratio, except that the numerator includesonly enrollments of children within the normal age range for the education stage inquestion. children in the numerator but not inthe denominator. Third, they are subject to errors inthe denominator estimates o f school-age population. The calculatednet enrollment ratio shown in column 2 o f table 2 for 1998/99 i s restricted to enrollments o f students o f age 7-15 years in the numerator, and uses actual enumerated population for that age group for the denominator. lo This figure corrects for the second and the third sources of error, and is therefore a more reliable indicator of coverage of education for that year. Itimplies that there were 94,000 young people nationwide who were not enrolled inthe compulsory cycle o f schooling (grades 1-9) for the 1998/99 school year. Gross EnrollmentRatio Calculated Net Survey-Based derived from Registered Enrollment Ratiob Estimate of Grade Enrollmentsa 1-9 Net Enrollment Sources: a) MDGProgressReport, 2003, 2003 NationalPovertyReductionStrategy, andNational Statistics Committee, b) calculatedfrom registeredenrollments and 1999 enumeratedpopulation aged7-15 years, data from National StatisticsCommittee, c) HouseholdBudget Survey, 2001. 4. Household survey data provide another perspective on education system coverage. Although they are subject to sampling error, they are not subject to the three sources o f error in registered enrollments described above. Moreover, they have the additional advantage that, depending on how the questionnaire i s designed, they can provide an indication o f actual school attendance (which i s closer to the objective of learning achievement than is school enrollment). The findings of the Kyrgyz Household Budget Surveys lead to net enrollment ratio estimates o f 96.1% for the 2000/01 school year, and 96.4% for the 2001/02 school year. These figures are much higher than the calculated net enrollment ratio for 1998/99, implying a major improvement in school coverage in that two-year period. Part o f the difference inthe two figures could also result from a sampling error. lo numeratorofthisfigureisrestrictedtoregisteredenrollmentsbetweentheagesof7and15atthestartofthe The school year (fromTable 7.7, National StatisticsCommittee ofthe Kyrgyz Republic, Education and Science in the Kyrgvz Republic: StatisticalBulletin, Bishkek,2003); the denominator is the enumeratedpopulation aged7 through 15 years fromthe populationcensus which was conductedinMarch, 1999. 10 5. Enrollment differentials. As shown in Table 3, the data from the 2001 Household Budget Survey generally show very little variation in grade 1-9 enrollments by gender, parental education, and household income, with virtually identical average enrollment rates for poor and non-poor, for boys and girls, and for different levels of completed education o f parents. The single variable for which there are substantial enrollment deficits at the compulsory level (corresponding to the 7-15 year age group) i s residential location. Overall enrollment in basic education i s lower in rural areas than in urban areas. Children inrural areas tend to start school at an earlier age than children inrural areas, but they also tend to leave school at an earlier age: there i s a sharp fall-off in rural enrollments at age 17, whereas urban enrollments taper off much more gradually. Apart from these urbadrural enrollment differentials, there are also important differences across oblasts. The low enrolment rate for Bishkek (91.7%) results entirely from late age o f starting school: Only 21.2% o f 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 o f educational outcomes. This i s partly but not fully a consequence o f poverty, Although it i s true that Talas oblast has the second highest headcount incidence and depth o f poverty, Naryn oblast, with the highest headcount incidence and depth o f poverty, has a reasonably high rate o f school enrollment in the basic education age cohort (95.3%), and most o f its enrollment deficit results from late starts (with only 70.2% o f 7 year olds inNaryn oblast attending school). l1 6. In contrast to the situation for compulsory education, there are significant differences in enrollment rates by gender, parental education, and household income after 9-year basic education, when schooling i s no longer compulsory and opportunity costs o f school attendance are higher. Continuation from the compulsory cycle to upper secondary education (grades 10 and 11) i s much lower for rural population, for lower income quintiles, and for children o f less educated parents. Continuationrates are particularly low for Jalalabad, Naryn, and Talas oblasts - the oblasts with some o f the highest incidences o f poverty. IncompleteSchoolAttendance 7. Non-attendance takes several forms, including children who never start school, children who start but drop out before completing primary and secondary education, and childrenwho are enrolledinschool but miss a significant part of the school year due to illness or other causes. All are harmful to learning objectives. Raising levels o f school attendance i s thus an important part o f the challenge o f meeting the education MDG. In most countries, survey data show rates o f actual school attendance that are significantly below enrollment ratios which are derived from official registration data. InUzbelustan, for example, household survey responses imply rates o f school attendance that are about 15 percentage points below the enrollment ratios that are calculated from official enrollment data.12 Unfortunately, the questionnaires for the Kyrgyz Republic household budget surveys report end-of-year enrollments rather than actual school attendance. But it i s likely that the difficult conditions prevailing in most schools in the Kyrgyz Republic result in attendance rates which are lower than enrollment ratios to a significant degree. 8. Official data report that the phenomenon o f children who never attended school i s very limited, and has declined to a negligible level in recent years (Table 4). The other manifestations o f non- attendance - children who start but drop out before completing primary and secondary education, and children who are enrolled in school but miss a significant part o f the school year due to illness or other 'IKyrgyzRepublic, Enhancing Pro-Poor Growth, World Bank ReportNumber 24638-KG, August 23,2002. Uzbekistan: LivingStandards Assessment: Policies to Improve Living Standards, World Bank report number 25923-UZ; May, 2003. 12 causes - are more o f a threat to attainment of the education MDG. Several recent surveys report rates o f non-attendance in compulsory schooling that are 7 to 10 times higher than official estimate^.'^ The most comprehensive o f these estimates that from 2 to 3 % o f children o f compulsory-school age (aged 7 to 15) have dropped out o f school, that the average daily attendance rate i s 85 to 90% (implying 150,000 children absent from school on any given day), that 25% o f children (about 250,000 children) miss 20 or more days o f school per year, and that 5 to 10% o f children (50,000 to 100,000 children) miss 40 or more days of school per year.I4 These figures suggest that non-attendance for prolonged periods during the school year affects a large share o f enrolled students, and i s a serious threat to attainment of the MDG. As discussed in the following sections, interventions to address the causes of non-attendance need to be tailored to the prevailing constraints to attendance, which differ across schools. Table 4: Numberof ChildrenandAdolescentsAged 7-17 Who Never Attended School Source, Table 7.18, National StatisticsCommittee of the Kyrgyz Republic,Education and Science in the Kyrgyz Republic: Statistical Bulletin, Bishkek, 2003 Deficiencies in Education Quality 9. All o f the country sources on the educational MDGs -the 2003 Country Progress Report on the MDGs, the Country Report on EFA, and the National Poverty Reduction Strategy Paper - express a concern with deficiencies in education quality, especially in the form o f sub-standard learning conditions and learning outcomes in rural schools. External assessment o f learning achievement i s the preferred instrument for monitoring what students are learning and for comparing school performance across schools and over time. Because school examinations in the Kyrgyz Republic are school-based, they do not provide an objective basis for comparing student learning achievement across schools or regions. External assessmento f school performance at the compulsory level i s currently limited to the Monitoring Learning Achievement (MLA) surveys which were carried out with assistance from UNICEF and UNESCO for 5" grade students in November, 2000 and for sth grade students in October, 2002. The findings from these surveys reveal systematic variations in learning achievement by average income, by l3Eric Eversmann,SchoolAttendance in the Kyrgyz Republic: Reportfor UNICEFBishkek, 1999;Center for Public Opinion Studies andForecast, Children of 7-1I Years Who do Not Attend School (Naryn Town), Bishkek,2001, a reportprepared for Save the Children,U.K.; Julia Droeber andNargiza kyzy Abdyrahman, Educational Problems in Three Villages of Nookat Rayon, August, 2003, preparedunder the USAIDPEAKS Project. l4Eric Eversmann, School Attendance in the Kyrgyz Republic: Reportfor UNICEF Bishkek, 1999 13 urbadrural location, and by language o f instruction. Further information on differences in learning achievement i s available in the form o f applicant scores on the external examination for university entrance, which was introduced in the Kyrgyz Republic with USAID support in 2002. Because it i s limited to university applicants, this examination covers a self-selected sample of secondary education completers. Most, presumably, are o f above-average achievement. But the results o f the university entrance examinations nonetheless provide useful verification o f the patterns in differential learning achievement revealed by the Monitoring LearningAchievement Surveys. 10. Regional differences. The 2002 Monitoring Learning Achievement survey documents the sizeable differences in basic education learning achievement across oblasts. As shown in Table 5, Batken, Talas, Osh, and Naryn oblasts have the highest incidence o f l o w learning achievement; Bishkek has the lowest. Table 5: IncidenceofLow LearningAchievement by Oblast, Eighth-GradeAssessment Results, October, 2002 % receiving lowest category % receivinglowest category scores in grade 8 mathematics scores in grade 8 natural sciences Bishkek 18.8 39.8 Chui 45.2 74.9 Naryn 64.0 95.2 Issykul 45.5 70.5 Talas 75.0 93.8 Osh 68.8 83.1 Jalalabad 57.0 61.8 Batken 65.8 81.3 11. Urban-rural differences. Much o f the difference inlearning achievement by oblast reflects the lower performance of students in rural areas (Table 6 and Figures 1 and 2), where poverty is more widespread, where cash income i s scarce for many households, and where parental educational attainment i s lower than in urban areas. Average scores in grade 8 arithmetic and biology inrural schools were less than half as high as in urban areas, implying levels o f learning achievement well below national standards. The same inverse correlation between educational performance and % rural i s apparent in the pattern o f university admission scores (Figure 3). 14 Table 6: Low LearningAchievement andPercent ofRuralEnrollmentsby Oblast, 2002 Source: Center for Public Opinion Studies andForecast,Monitoring of Learning Achievement (8thGrade): National Suwey of Quality of Learning in the 8'hGrade, Bishkek,2003, andNational Statistics Committee o f the Kyrgyz Republic, Education and Science in the Kyrgyz Republic: Statistical Bulletin, Bishkek,2003). Figure1:LearningAchievementandPovertyIncidence byUrbanmuralLocation 80.0 70.0 60 0 50.0 40 0 30.0 Rural 20.0 10.0 0.0 Average Score in Average Score in 2003 Poverty Grade 8 Arithmetic Grade 8 Biology Incidence Headcount Index (Oh) Source: Center for Public Opinion Studies andForecast,Monitoring of Learning Achievement (Sth Grade): National Survey of Quality of Learning in the 81hGrade,Bishkek, 2003, and data from the 2003 HouseholdBudget Survey. 15 Figure 2: LearningAchievement andPercentRuralby Oblast I-- YOlowest math scores % lowestscience scores 1 YOrural Source: Center for Public Opinion Studies and Forecast,Monitoring of Learning Achievement (Grade 8), Bishkek, 2002-2003, and data from the 2003 HouseholdBudget Survey. Figure 3: Mean UniversityAdmission Scores andRuralProportion of UniversityApplicantsby Oblast, 2004 160 140 120 100 -+Mean Entrance Exam Score 80 60 I% Rural 40 20 0 12. Linksto poverty. As shown inFigure 4, low learning achievement is positively correlated with poverty incidence at the oblast level. In the oblasts with the highest incidence of poverty in the 2003 Household Budget Survey, the majority of Sth grade students in the MLA survey failed to obtain a satisfactory score o n the MLA tests for gthgrade math and natural sciences. 16 Figure4: Low LearningAchievementandPoverty Incidenceby Oblast 100.0 90.0 +YO of Grade 8 80.0 Students with Lowest 70.0 Score in Natural I 60.0 Sciences 50.0 --a--YOof Grade8 40.0 Students with Lowest 1 30.0 Score in Math I 20.0 10.0 0.0 2003 Po\~rty Incidence Headcount Index (YO) I I ~ Source: Source: Center for Public Opinion Studies and Forecast, Monitoringof LearningAchievement(??IhGrade): National Survey of Quality ofLearning in the81hGrade, Bishkek, 2003, and World Bank, Kyrgvz Republic: Enhancing Pro-Poor Growth,ReportNo. 24638-KG, August 23,2002. 13. Language of instruction. Another parameter on which there i s sizeable variation in learning achievement i s language o f instruction in schools, which i s closely related to the majority ethnicity in communities. In the 2002 Monitoring Learning Achievement Survey, the average scores o f students in Russian-language schools were consistently higher than those o f students in Uzbek-language schools, or Kyrgyz-language schools (Table 7). The average performance o f students in Uzbek-language schools was above that of students in Kyrgyz-language schools in math and physics, but below it in chemistry, biology, and physics. The 2002 Monitoring Learning Achievement survey found that lessons are cancelled most frequently in Kyrgyz-language schools, and almost never in Russian-language ~chools.'~ The same study found that Kyrgyz-language schools are more likely to be staffed by younger and less experienced teachers, and to suffer from unsatisfactory condition and availability o f teaching equipment. 'jIbid,page 57. 17 Table 7: Average Scores in SthGrade Science and Math by Language of Instruction, 2002 so a1 Survey of Quality of Learning in the 8IhGrade, Bishkek, 2003 2. THESOURCES OFUNSATISFACTORY EDUCATION PERFORMANCE 14. This section of the case study examines the sources o f unsatisfactory education performance in terms o f the two dimensions o f performance described inthe preceding section: deficient quality interms o f learning outcomes, and incomplete education coverage. Incomplete education coverage i s discussed here in terms o f reasons for non-attendance, including non-attendance by children who are ostensibly enrolled in school as well as non-attendance by children who are not enrolled (i.e., who never started school, or who were enrolled but dropped out before completing nine grades of schooling). 15. The preferred approach to understanding the sources o f unsatisfactory education performance is analysis which supports attribution o f causality - either through experimental design o f interventions and evaluation of outcomes, or statistical controls to isolate the partial contribution o f individual factors to observed outcomes. The latter approach i s possible inthe Kyrgyz Republic for looking at the reasons for incomplete school attendance or school dropouts, and i s the basis o f the analysis described in paragraph 19. School attendance is amenable to this approach because it is an objective and easily observable behavior, and because relatively rich data are available inthe form o f the findings o f the 2001 Household Budget Survey. Learningachievement i s more elusive. Although the Government is working to develop more objective indicators o f learning achievement as the basis for routine monitoring, the only currently available data set for comparing levels o f learning achievement i s too limited in scope and scale to support refinedstatistical analysis. For that reason, the analysis of sources o f low learning achievement in this case study is limited to the examination in the preceding section of differences in learning achievement for individual parameters on which there i s sizeable variation in learning achievement inthe 2002 Monitoring Learning Achievement survey and the university entrance examination, and a presentationin this section o f supporting evidence on the Kyrgyz situation from other sources, including findings from other studies. This i s supplemented by a summary of the literature on determinants of learning outcomes inthe United States and other data-rich countries (Box 1). Section C o f the case study describes the implications o f these findings interms o f the policy options for improving MDG attainment interms ofschool attendanceandlearning achievement intheKyrgyz Republic. Incomplete School Attendance 16. Negative sanctions played an important role in ensuring parent's compliance with compulsory education during the Soviet Union. Non-attendance was reported, and action was taken promptly with parents to ensure that their children returned to school. Negative sanctions continue to play a role in enforcement o f compulsory school attendance, but to a much smaller degree. The Government recognizes that there are legitimate reasons - many o f them poverty related- that keep some children out 18 o f school. Several recent policy initiatives have focused on addressing these income-related constraints. These include the introduction o f textbook rental rather than textbook purchase, and the program of student social support to help provide children from the poorest households with clothing and other necessities for attending school. 17. Some o f the income-related constraints to school attendance resulted from the actions taken to diversify education financing early inthe transition.16 As budget transfers from Moscow ended with the collapse o f the Soviet Union, the Government reducedthe duration o f compulsory education from eleven years to nine years, shifted much of the responsibility for financing public primary and secondary education to local governments," and authorized schools to raise money from parental contributions and from the rental o f facilities and other income-generating activities. Students were requiredto purchase textbooks and other learning materials that schools had formerly provided free. Schools were permitted to receive parental contributions, and many schools in more prosperous communities received very sizeable contributions form parents and form the community. In addition, many schools solicited contributions from parents, which amounted to virtual school fees. In some cases, these informal school fees were set at such a high level that they made it difficult for children from poorer families to attend school. To deal with this problem, the Government set ceilings for voluntary contributions from parents at 160 soms per student per year inurban areas (about US$4.00) and 100 soms inrural areas (about US$ 2.50). This has helped address the problem o f non-attendance, but it has not eliminated it. There are widespread reports that most schools treat the ceilings as non-voluntary school fees, that most parents accept them as such, and that some schools continue to ask for muchlarger amounts from parents. 18. Several recent studies investigate the causes o f non-attendance in primary and secondary school. All o f them report that income-related causes are an important factor in explaining non-attendance. A 2003 survey o f households inNaryn oblast found that 43% o f children aged 7-11who had dropped out o f school did so because they lackedclothing; 34% because they lackedtextbooks and educational materials, 9% because their parents "could not afford the cost o f schoo1ing."'* A 2004 study o f households in Jalalabad oblast reported a higher percentage o f non-attendance (82%) that was attributed by parents to lack o f clothing and shoes. Inthe same survey, 61% o f non-attendance was reportedly due to children's need to help inharvesting and agricultural work, and 52% to lack o f textbook^.'^ Some o f the problem o f non-attendance in rural areas i s seasonal, and results from the necessity o f children's participation in harvesting, planting, or other seasonal agricultural tasks. In one recent study in Jalalabad oblast, 75% of parents whose children were not attending school reported the reason as needing to help with the l6 These are described more fully inthe backgroundpaper that was prepared for the 2003 Public Expenditure Review: MichaelMertaugh, Kyrgyz Republic Public Expenditure Review: Budgeting and Expenditures in the Education Sector, February 24, 2003. Inspite ofthe intention of decentralized finance and management ofprimary and secondary education, the scarcity o f local government revenues has necessitated continued central budget financing o f core education expenditures -including, crucially, teachers' salaries. 18 Center for Public Opinion Studies and Forecasting, ChildrenAged 7 to 11 that do not Attend School, study commissioned by Save the ChildrenUK,Bishkek, 2003104. l 9Julia Droeber and Nargiza kyzy Abdyrahman, EMIS (Community-Based Education Management Information System) Report: Survey Conducted in Beshik Jon Village, Bazar-KorgonRayon, Djalalabad Oblast, Kyrgyzstan, sudy performed for USADIPEAKS Project, 26 May- 3 June, 2004. harvest.20 Focus-group discussions often include reports that household outlays for textbooks, parental "contributions", and other school-related costs deter school attendance by the poor." 19. The 2001 Poverty Assessment included a statistical analysis o f school dropouts, using data on household characteristics, school variables, and regional income levels from the 1997-98 Kyrgyz Poverty Monitoring Survey.22 This study defined as a dropout any child who started school but did not complete the eleven-year cycle o f primary and secondary education. The variables which were most strongly associated with dropping out o f school and significant at the 5% level are presented inTable 8. Ethnicity was found to have the strongest effect on school dropouts, with ethnic Uzbek and Tajik students having the highestprobability o f not completing secondary education. Ethnic Russian students also had a higher probability than majority ethnic Kyrgyz students o f not completing secondary education, although not as high as Uzbek and Tajik ethnicities. Adequate numbers of teachers in the school, higher than average household assets, and presence o f the father in the household were associated with lower probability o f dropping out o f school. Employedpersons, and males also hadhigher probabilities o f dropping out. Table 8: Probit Estimatesof School Dropouts (1997/98 Kyrgyz Poverty Monitoring Survey) Marginal Effect on Dropout Prnhahilitv + 0.1006 ~Uzbek or Tajik Ethnicity School has enough teachers Householdassets -- 0.0541 0.05 13 Russianethnicity +0.0506 Father is present +- 0.0384 0.0488 Individual is employed Male +0.0319 Source: Kyrgyz Republic: Poverty in the 1990sin the Kyrgvz Republic, World Bank reportnumber21721-KG, June, 2001, table 24. Low LearningAchievement 20. The role of resources. It is more difficult to document and infer causality for low learning achievement than it i s for incomplete school attendance. Inall settings, the debate about what leads to betterlearning outcomes inschools inevitably involves questions about the role o fresources. Inresource- rich environments, there i s evidence that more resources do not lead to better educational outcomes.23 In resource-poor environments, however, more resources are often needed in order to bring about improved learning outcomes (Box 1). This i s not to say that more resources necessarily lead to better outcomes: In reviewing the effectiveness o f development projects in reducing poverty, the 2004 OED Review of Development Effectiveness concludes that increased spending must be combined with measures to ensure 2o Julia Droeber andNargizakyzy Abdyrahman,Educational Problems in Three Villages of Nookat Rayon, August, ''2003, preparedunder the USAID PEAKSProject. Kyrgyz Republic: Consultations with the Poor, Reportpreparedfor the GlobalSynthesisWorkshop, September 22-23, 1999,World Bank. 22 KyrgyzRepublic:Poverty inthe 1990sinthe Kyrgyz Republic,World Bank reportnumber21721-KG, June, 2001. 23 See, for example,Eric A. Hanushek, "The Failure o f Input-BasedSchooling Policies," TheEconomic Journal, February,2003. 20 that expenditures on education reach poor population^.^^ Inthe Kyrgyz Republic, budget deprivationhas clearly led to a situation in which many schools in poor areas - particularly in rural areas - lack the essential resources for effective teaching and learning. The 2002 Monitoring Learning Achievement Survey, found, for example, that all urban schools surveyed were adequately equipped with blackboards for the teacher and chairs and desks for students, whereas 15% to 25% o f surveyed rural schools lack these essentials. Almost twice as many urban schools (85%) as rural schools (49%) were equipped with bookcases for teaching aids and supplies; teachers in urban schools were better provided with teaching documentation than teachers in rural schools. More than twice as many rural schools as urban schools were inadequately heated and provided with water. 25 In this resource-deprivedsetting, more adequate resource provision to poorly endowed schools i s a key instrument for education MDG attainment. We examine below the options for gettingmore resources to deprived schools. But first, it i s usefulto put the current resource situation for educationinto perspective. 21. Deficiencies in the teaching and learning environment in schools directly reflect the changing budgetary situation of schools. This, in turn, reflects the economic and fiscal shocks which the country faced early inthe transition, and still faces. Enrollments inprimary and secondary education increasedby 18% and inhigher education more than doubled between 1990 and 2000.26 Real education expenditures fell by 69% over the same period (Table 9), implying an implosion o f education resources with few historical parallels throughout the world. Although there has been some progress toward recovery since then and further increases planned, central budget resources per student remain well below half o f what they were in 1990. The sharp decIine in budget financing for education occurred in spite o f a roughly constant budget share for education - averaging about 20% o f total government expenditures, and reflected both the fall in GDP and the much smaller share o f the public sector under a market economy. Expressed inreal terms, public expenditures on education fell even more sharply by the end o f the 1990s- to less than one-third o f their 1990 level, reflecting the combined effect o f the smaller public sector and the sharp decline in GDP. Although there has been progress in raising the budget share since 2000, real education expenditures remain at less than half their 1990 level in absolute terms, and considerably less terms of outlays per student. 242004 Annual Review of Development Effectiveness: TheBank's Contribution to Poverty Reduction, Operations EvaluationDepartment, The World bank, 2005. 25 Center forPublic Opinion Studies and Forecasting, Monitoring o f Learning Achievement (8" Grade): National Survey of Quality o f Learning inthe 8'h Grade, Bishkek, 2003, page 66. 26 Michael Mertaugh, Kyrgvz Republic Public Expenditure Review: Budgeting and Expenditures in the Education Sector, February 24,2003. 21 Box 1:FactorsAffectingLearningAchievement: The InternationalEvidence There i s widespread debate about the factors affecting learning achievement, and there have many statistical attempts to identify the critical factors that lead to successful learning achievement. Few o f these have found robust links, which suggests that education is largely context-specific. Among the most consistent findings inthe research literature are that learning outcomes are better when students have textbooks, when teachers and students spend more classroom time on substantive teaching and learning tasks, and when this i s supplemented with more time spent by students on homework. A recent, comprehensive review of the research literature o n factors influencing student achievement reported the following results for developing countries and for the United States: EstimatedParameter Coefficients from 96 EducationalProductionFunctionEstimatesfor DevelopingCountries Statistically significant (%) Input Numberof Positive Negative Statistically estimates insignificant (%) Teacheripupil ratio 30 27 27 46 Teacher education 63 56 3 41 Teacher experience 46 3.5 4 61 Teacher salary 13 31 15 54 Expenditureipupil 12 50 0 50 Facilities 34 65 9 26 EstimatedEffectofKeyResourceson StudentPerformance, Basedon376 ProductionFunctionEstimatesfor the UnitedStates Statistically significant (`YO) Resources Number of Positive Negative Statistically estimates insignificant (%) Real classroom resources Teacher-pupil ratio 276 14 14 72 Teacher education 170 9 5 86 Teacher experience 206 29 5 66 Final aggregates Teacher salary 118 20 7 73 Expenditure per 163 27 7 66 pupil Other Facilities 91 9 5 86 Administration 75 12 5 83 Teacher test scores 41 37 10 53 Source: Eric A. Hanushek, "The Failure of Input-Based Schooling Policies," The Economic Journal, February, 2003. 22 Table 9: The Evolutionof Actualand PlannedPublicExpendituresfor Education as a Share of TotalPublicExpendituresand GDP, 1990-2007 Education Education Total Public Educationas GDP Education Expenditures Expenditures Expenditures YOof Total (in as % of (inmillions of (inmillionsof (inmillions of Public millions GDP current soms) 1995 soms) current soms) Expenditure o f current S soms) 1990 3.2 2,381.0 15.9 20.1 Yo 43 7.4 Yo 1991 5.6 1,775.5 24.4 23.0 Yo 93 6.0 % 1992 37.3 1,271.4 231.1 16.1 Yo 741 5.0 % 1993 227.2 906.3 1,225.8 18.5 % 5,355 4.2 Yo 1994 730.8 1,038.0 2,812.8 26.0 % 12,019 6.1 Yo 1995 1,064.9 1,064.9 4,6 10.5 23.1 % 16,145 6.6 Yo 1996 1,222.8 903.5 5,202.4 23.5 % 23,399 5.2 Yo 1997 1 314.0 937.6 6,695.7 22.6 % 30,686 4.9 % 1998 1,681.6 954.7 7,298.3 23.0 Yo 34,181 4.9 Yo 1999 1,892.3 780.9 9,042.2 20.9 % 48,744 3.9 % 2000 2,289.9 740.2 11,284.5 20.3 Yo 65,358 3.5 Yo 2001 2,847.6 854.3 12,257.0 23.2 % 73,883 3.9 Yo 2002 3,350.2 992.9 17,665.0 19.0 Yo 75,367 4.5 Yo 2003 3,753.6 1,072.0 16,890.6 22.2 Yo 83,872 4.5 % 2004 4,361.3 18,841.5 23.0 % 94,078 4.6 Yo 2005 4,03 1.7 19,536.7 20.6 % 100,784 4.0 % 2006a 5,067.1 22,630.9 22.4 Yo 111,504 4.5 Yo 2007b 6,055.0 26,879.2 22.5 % 124,729 4.9 % Figures on education shown inthis table do not include PIP but they include special means. a.draft budget. b.projected. Source: National Statistics Committee; World Bank database; National Statistics Committee o f the Kyrgyz Republic, Education and Science in the Kyrgyz Republic: Statistical Bulletin, Bishkek, 2003, and Kyrgyz Ministry of Finance, Drafl Budget Law as o f September 2005, Medium-Term Budget Framework 2006-2008,Bishkek, 2005. 22. What effect has this budgetary contraction had on education programs? Among the most serious effects are: 0 Widespread deprivation o f the basic educational materials needed for effective teaching and learning and for modernization o f education programs. The recent National Survey o f Primary Education Quality found, for example, that 80 % o f primary schools lacked a complete supply o f textbooks for students, 70 % lacked teachers' guides, 20 % lacked desks and chairs for students, 70 % needed repairs to school furniture, 23 % of schools lacked water supply, and 39 % lacked telephone^.^^ 0 A serious decline o f teachers' salaries, which has eroded teachers' morale and motivation, induced many teachers to take additionaljobs, and exacerbated problems o f 27Monitoring Learning Achievement: National Survey of Primary Education Quality, Center for Public Opinion Studies and Forecast, Bishkek, 2001. 28 Teacher salaries are low inboth absolute and relative terms. Inabsolute terms, they are only halfthe minimum consumption level for individuals, not to mention households. As a result, teachers lack motivation, and are compelled to work at other jobs inorder to support themselves and their families. Inhigher education, low faculty salaries have contributedto a serious problem of corruption, with students oftenpaying for admission and grades. 23 0 Increasing inequality in quality o f education and access to education as the shrinkage o f central budget support has led to increasing reliance on unevenly available local financing, parental contributions, tutoring, supplementary financing through rental o f premises and non-educational activities. On average, 35% o f the expenditures o f primary schools in urban areas and 25 % o f expenditures of rural schools come from non-budget sources. The increasing reliance on these non-budgetary sources i s not only a source o f increased inequality, but also threatens to displace the core teaching function of primary and secondary schools. 0 A cessation of new school construction which has led to excessively intensive use o f existing school facilities. Only 10 percent o f primary schools operate on a single shift. Fully 81 % o f rural schools and 71 % o f rural schools operate on double shifts. Nine percent o f rural schools and 19 % o f urban schools operate on triple shifts.29 0 The development o f parental contributions as an important source o f financing for school maintenance, fuel, and other necessities inurban schools. Parental contributions to urban schools often exceed $100 per year per student. Parents also contribute to schools in rural areas, but widespread poverty means that income from this source i s very limited. The reliance on parental contributions i s a major source o f inequity inquality o f education. 23. The effects of budget deprivation on learning achievement are reinforced by the effects o fpoverty at the household level and the policy o f shifting education costs from the school to students and parents. As shown in Figure 5, average household expenditures on education by oblast vary inversely with the incidence of poverty ineach oblast. Household outlays for educationinclude 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). Outlays for education constitute a larger burden for the poor - and particularly for the extreme poor - as shown in Table 10. The 2001 Review of Social Policy and Expenditures3' reported similarly high shares o f household expenditures for education - again, with higher expenditure shares for the poor than for the non-poor. This has ledto a serious deteriorationinquality o feducation. More significantly, it has ledto cynicismon the part of employers and the public about the significance o f higher education diplomas, except for the few institutions which have beenable to prevent or control corruption. Teacher salaries are also low ina relative sense. Average monthly earnings inthe education sector arejust 857 soms, versus the considerably higher average earnings inall other sectors except forestry and health. It i s notable, inparticular, that average teachers' salaries are less than 40 % of average earnings inpublic administration. 29ibid 30Kyrgyz Republic, Review of Social Policy and Expenditures, World Bank Report No. 22354 KZ, June, 2001. 24 Figure 5: Average HouseholdExpenditures on Education and Percent Poor by Oblast, 2001 120 ' 100 Household Education 80 Expenditures as YO 60 of Bishkek -E--Percent Poor 40 I 20 0 Source: Data from Kyrgyz HouseholdBudget Survey, 2001 Table 10: Primary and SecondaryEducation Expenditures as a Share of HouseholdIncome, 1996- 1998 Year ExtremePoor Other Poor Non-Poor 1996 21.8 % 19.5 % 16.7 % 1997 17.7 % 11.3 % 8.9 Yo 1998 20.1 % 13.9 % 9.6 % 24. The role of incentives. Incentives also play an important role in motivating better education performance, as the 2004 World Development Report3' emphasizes. The Government i s developing a performance-based evaluation system for teachers and an associated salary incentive scheme under the Rural Education Project. The project will also support pilot implementation of the new performance evaluation scheme and salary incentives intwo oblasts. The new teacher evaluation criteria will include indicators of teachers' contributionto improved learning achievement by their students. Full attainment of the education MDG would require that this new system o f performance evaluation include a deliberate focus on improving the Performance o f students who are not meeting national learning objectives, that a similar system o f performance evaluation and incentives be institutedfor education managers to motivate greater attention to improving the performance o f the lowest-performing schools and students. 3'WorldDevelopment Report 2004: Making Services Workfor Poor People, The Worldbank, 2003. 25 3. IMPROVING EDUCATIONPERFORMANCE The Current Approach 25. The Government's current approach to raising education performance focuses on two generic activities: a) raising the share o f budget financing for education, and b) implementingtargeted actions to improve the performance o f groups with incomplete attendance and low learning achievement: 0 Raisingthe share of budgetfinancingfor education. The 2003 MDG Country Progress Report for the Kyrgyz Republic cites grade 1-9 enrollment ratios o f 85.3% for 1993/94 and 95.0% for 2000/01. On the basis o f that increase, it projects 100% coverage in grades 1through 9 by 2015. Butit also raises a cautionary note on the problems of incomplete attendance, early dropouts, and quality disparities in education. It reports that these problems are especially widespread in rural areas, and attributes them to low household income. To address these problems, the MDG Country Progress Report advocates a return to the level o f budget financing that the sector received in the latest years o f the Soviet period, which would require average GDP growth rates o f 6% per year through 2015, and an increase in the Republican budget share for education from 3.9 YOin 2001 to 6.5% in 2015. The Government i s making a serious effort to meet this goal. The Medium-Term Budget Framework (MTBF) for 2006 to 2008 calls for an increase in the consolidated budget for education from 4.0% o f GDP in 2005 to 6.0% o f GDP in 2008. However, the feasibility o f achieving these targets i s under question - the education sector ceiling provided by the 2006 draft budget law i s lower than the one projected under MTBF. Moreover, historically annual budgets deviate significantly from MTBF projection^.^^ 0 Implementing targeted actions to improve education performance. The Government is implementing a number o f actions that are designed to address the constraints which contibute to incomplete school attendance and unsatisfactory learning achievement. Most of these interventions are being supported by external partners, including the Rural Education Project which includes support for development and piloting o f a newperformance incentive program for teachers. Ongoing targeted actions to improve education performance include: replacement of textbook purchase with textbook rental (at significantly lower annual costs for households), exemption o f textbook rental charges for the poorest 10% of students, provision o f clothing and other social support to the poorest students, and rural teacher salary supplements to attract qualified teachers to teach inremote, rural schools. These actions are either targeted to poor and low-performing students, or are explicitly intended to benefit poor and low-performing students. Other, untargeted actions such as developing and implementing learning improvement plans for schools and introducing teacher performance incentive^;^ should lead to better educational outcomes for all students, including those at particular risk o f incomplete attendance and low learningachievement. Achieving FullerAttainment of the ExpandedEducationMDG 26. As documented in the preceding section, resource constraints definitely contribute to low education performance. But simply raising the share o f education budgets in GDP to pre-transition levels j2Attachment 3, Kyrgyz Republic Ministry of Finance, Medium-term Budget Framework 2006-2008,Bishkek, 2005. j3These actions are being supported under the Rural Education Project, and were motivated by a concern to raise the performance o f substandard students and schools. under existing expenditure patterns and allocation rules would not solve the problems o f low education performance for two reasons: 0 First, because improved school attendance and improved learning achievement require interventions which are not currently financed by the budget, such as tutoring for students with learning difficulties and provision o f textbooks and other educational materials for students who cannot afford to buy them, and 0 Second, because the current system provides neither the means to identify schools and students with sub-standard performance nor the incentives for teachers, school principals, and education managers to concentrate their efforts on bringingthose low-performing students and schools up to national learning standards. 25. Full attainment of the expanded education MDG would require moving to a fundamentally different use o f budgetresources. Inparticular, it would require targeting allocations to schools with low performance, and financing a broader range o f interventions - including actions which are specifically intended to improve school attendance and to raise sub-standard learning achievement. It would also require the introduction o f incentives for teachers, school principals, and education managers to concentrate their attentions and efforts on raisingthe performance o f their lowest-performing schools and students. 26. Adopting this performance-based approach to MDG attainment would require a paradigm shift for the Government and the public - a change in which education performance i s no longer measured in terms o f registered enrollments, but in terms o f school attendance, learning achievement, and eventual labor-market outcomes, and in which performance o f the education system i s no longer judged by the success o f its best students, teachers, and schools, but its success inraising the performance o f its lowest- performing students, teachers, and schools to national learning standards. To the extent that public attention currently focuses on learning outcomes, it does so mainly for the right-side tail o f the distribution - the highest performing schools, teachers, and students. Government and the public tend to judge the strength o f the education system on the basis o f its strongest parts - the Olympiad winners. Further progress on the MDGs and success inthe broader objectives of education would require a focus on left-side tail o f the distribution -the lowest-performing schools, teachers, and students. 27. Resource allocation would needto reflect this new results focus. At a minimum, it would need to provide for targeted interventions to raise the performance o f the lowest-performing schools. More aggressive pursuit o f the performance-based approach would require a fundamentally different process for 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. But in spite o f these efforts, the basic resource allocation formula for schools remains input-based and norm-guided. A consequence of this approach i s that all schools receive central budget resources based on the number of teachers they have. This approach essentially promotes a level playing field in terms o f publicly provided inputs. But since many communities supplementthose inputs with very significant contributions from other sources, the results in terms o f learning achievement are distinctly unequal, as the Monitoring Learning Achievement surveys document. Focusing use o f central budget education resources on attainment o f the educational MDG as the first priority in the sector would require that the algorithm for allocating central budget resources in the education sector be reconfigured to introduce a deliberate slope in the playing field in terms o f inputs in order to ensure that the lowest-performing schools meet national standards for learning achievement, and requiring other schools to cover more o f their costs from other sources. If attainment o f the education MDGrequiredadditional resources, this selectivity inuse o f resources within 27 the existing budget allocations for primary and secondary education could be supplemented with selectivity across education sub-sectors by reallocating some o f the current higher education allocations - inthe central budget to primary and secondary education, and relyingmore heavily on private and local financing for higher education. 28. ProposedOptions. Column 1 of Table 1 summarizes what the Government is already doing to promote attainment of the expanded education MDG, and presents options for fuller attainment o f the goal. It does so under five generic actions which are key elements o f the options for fuller MDG attainment. Column 2, the base case scenario, summarizes actions which are currently under implementation which will promote attainment o f the expanded education MDG. Column 3, the intermediate option, presents an intermediaterange o f actions which would lead to accelerated attainment of the expanded education MDG. Column 4, the high-case option,presents a high-commitment approach to attain the expanded education MDG. Actions are also distinguished by time frame (short-term and medium-term actions). A number of these are admittedly "stretch" goals that would require a well conceived and consistently implemented advocacy program by the Government in order to ensure compliance by all parties - including the high-performance schools and regions whose educationbudgets could be reduced under these actions. The high-case approach includes adoption o f management incentives and a financing formula that give preference to schools in low-income areas (because o f their greater needs and lesser capacity to contribute from other sources). All recommended approaches involve greater use o f data on education outcomes and gradual improvement o f these data over time. 29. The following five sets o f actions are proposed for fuller attainment o f the expanded education MDGinthe Kyrgyz Republic: 0 Monitor Performance. Monitoring and evaluation of school performance - particularly, in the form o f external assessment o f learning achievement - play a key role in identifying the target groups for MDG interventions and in designing appropriate interventions to improve performance. External assessment o f school performance i s currently limited to two sources: the Monitoring Learning Achievement Surveys which were carried out for 4thgrade students in 2001 and for 8th grade students in 2003, and the National University Entrance Examination which has been carried out annually since 2003. Although it will be important to improve assessment instruments over time (as provided for under the Rural Education Project), these existing sources provide a sufficient basis for starting the MDG targeting. At the start o f the process, the performance focus could be implemented using existing information on regional differentials in learning achievement and school attendance. Progressively, the targeting and design of interventions should be refined on the basis o f better information on educational outcomes at the school level. School report cards (which are being developed for two pilot oblasts under the Rural Education Project) are a simple and practical instrument for involving parents and other community stakeholders inthe evaluation o f school performance. Since income differences play such a large role in explaining differences in the learning environment, poverty data provide a good proxy for learning achievement and could be a valuable supplementary source of information for targeting. They are also available for the Kyrgyz Republic on a highly disaggregated basis. The monitoring process should reward accurate feedback on system performance; there should be no negative sanctions for reportinglow performance. 0 Target Interventions to the Lowest-Performing Schools. Fuller attainment of the expanded education MDG in the Kyrgyz Republic would require not only that budget allocations for education continue to rise, but that budget allocations be targeted to students who are not attending school or not meeting national learning objectives, and include further deliberate actions to raise the performance o f those students. In part, this needs to involve equipping schools and students in the lowest performing schools better - if not to the standards o f the best 28 equipped, urban schools, at least to minimally acceptable standards - and providing incentives for teachers and education managers to focus more attention on the educational outcomes o f school attendance and learning achievement - particularly, for the lowest-performing students and schools. Where school attendance i s incomplete, it will also require tailored actions to identify and address the specific causes of non-attendance, including demand-side constraints. These could include, for example, expanding the textbook rental charge exemption to cover, where appropriate, more than the 10% o f students who are exempted under current policy, provision o f support to needy students for textbooks and educational materials, clothing, and transport, the provision o f school meals, or the adaptation o f the school calendar to accommodate seasonal needs for child labor. Provide Incentives for Improved Performance. As noted above, adopting a performance- based approach to MDG attainment would require a paradigm shift for the Government and the public. In order for that to happen, there would need to be a clear and consistent message from senior levels o f Government that such a change i s desired and needed. In order to motivate these changes at the school level and in local government, there would also need to be incentives for teachers, school principals, and education managers to focus on raising the performance o f substandard schools and students. The Rural Education Project i s supporting the development of an improved performance management systemfor teachers and principals, and payment of related performance incentives for teachers and principals for two academic years in two pilot ~ b l a s t s . ~ ~ It is expected that the newperformance incentives will be incorporated into the nationwide salary scale for teachers and principals if the results o f the pilot evaluation are positive. The purpose o f the new performance incentives is to improve student learning achievement by motivating teachers to pay more attention to what their students are learning and to raise the performance o f low-achieving students. To help attract and retain qualified teachers inrural schools in subjects with unfilledteacher positions, the Rural Education Project i s also financing a program o f rural teacher fellowships for about 300 new teacher graduates in scarce subject areas who agree to teach in rural schools throughout the country, starting in September, 2005. Additional rural teacher fellowships will be financed under the Asian Development Bank's Second Education Development Project. Implementing these performance incentives and rural location incentives on a nationwide basis would help motivate the focus on low-performers which i s needed for full attainment o f the expanded education MDG. Evaluate Pilots to Identify What Works. A number of small-scale interventions to improve school attendance and learning achievement are under implementation or planned -many o f them under partner-assisted projects. Correctly designed, pilot projects can be very useful for identifying cost-effective interventions to improve educational performance. But most pilot projects do not yield useful lessons for potential larger-scale implementation either because o f design deficiencies or because outcome evaluation i s not carried out. The preferred approach for evaluating the outcome of pilot projects i s an experimental design with random assignment o f schools into treatment and control groups (possibly, with alternative treatment mixes), or a quasi- experimental design with matched pairs o f treatment and non-treatment schools. The recommended options in Table 1 include a serious effort to design pilot activities to permit outcome evaluation, with actual evaluation o f outcomes when pilot projects are completed. In 34 The performance criteria inthe new system will rewardteachers for the quality of their classroomteaching, their capacityto diagnoseand supporttheir students' learningprogress, andthe progress(not just the level of achievement) of all their students (notjust their highestperforming students) against agreed standards. For school principals, the criteria will also cover leadership andmanagementperformance as well as student progress and performance. 29 order to improve the attribution o f causality to observed changes in learning achievement, changes in other variables which also affect learning achievement, such as classroom time spent on specific tasks and time spent on homework, should also be monitored and used in the evaluation o f outcomes. 0 Use Budget Resources Selectively. Budget allocation i s a crucial instrument for fuller attainment o f the expanded education MDG. At a minimum, budget allocation would need to provide for targeted interventions to raise the performance o f the lowest-performing schools. More aggressive pursuit o f the performance-based approach would require a fundamentally different process for 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. But in spite o f these efforts, the basic resource allocation formula for schools remains input-based and norm-guided. A consequence o f this approach i s that all schools receive central budget resources based on the number o f teachers they have. This approach essentially promotes a level playing field in terms o f publicly provided inputs. But since many communities supplement those inputs with very significant contributions from other sources, the results in terms o f learning achievement are distinctly unequal, as the Monitoring Learning Achievement surveys document. Focusing use o f central budget education resources on attainment o f the educational MDG as the first priority in the sector would require that the algorithm for allocating central budget resources in the education sector be reconfigured to introduce a deliberate slope in the playing field in terms o f inputs in order to ensure that the lowest-performing schools meet national standards for learning achievement, and requiring other schools to cover more o f their costs from other sources. If attainment o f the education MDG required additional resources, this selectivity in use o f resources within the existing budget allocations for primary and secondary general education could be supplemented with selectivity across education sub-sectors - for example, by reallocating some budget resources from secondary vocational education to secondary general education (for which the unit costs are considerably lower), or redirecting some o f the allocations for higher education inthe Republican budget to primary and secondary general education, and relying more heavily on private and local financing for highereducation. 30. Education Management. Finance and management o f primary and secondary education i s ostensibly the responsibility o f the 470 ail okmutus, or self-governing units o f civic administration. In principle, local governments manage primary and secondary schools and finance them from their own resources. But because most local governments lack the revenues to meet the costs o f operating their schools, core costs o f primary and secondary schools - consisting o f teacher salaries and utilities - are still financed from the Republican budget, while more prosperous communities supplement these funds from government and parental contributions o f various kinds.35This financing i s intended to be provided on a per-student basis, but inreality i s based on centrally set salary levels and centrally set norms on class size and minimumteaching loads. 31. The current approach to management and finance o fprimary and secondary education developed as an expedient solution to keep schools in operation when the country adopted decentralization early in the transition, and found that most local governments had neither the resources nor the capacity to manage their schools. This approach has the strong virtue o f having kept schools in operation through a period o f severe budget contraction. But it also has the important disadvantage o f providing neither the 35Thisprocessis describedmore fully inthe backgroundpaper that was preparedfor the 2003 Public Expenditure Review: MichaelMertaugh,Kyrgvz Republic Public Expenditure Review: Budgetingand Expenditures in the Education Sector, February 24: 2003. 30 means nor the incentive for local governments to improve the performance o f their schools or to manage themmore cost-effectively. More cost-effective use of existing resources couldbe animportant source of financing for targeted interventions to improve school attendance and raise performance o f low-achieving students. 32. Successful implementation o f the recommended performance-based approach to education management could be carried out in the current decentralized system o f education management, but it would be much simpler to do so in a centrally managed system. Central education management and financing would ease the task o f achieving full attainment o f the expanded education MDGby: 0 Internalizingthe external benefits o f education, and thus reduce the risk o f under-provision which characterizes all decentralized systems o f education finance,36 0 facilitating identification o f students and schools that are not meeting national learning standards, 0 making it easier for budget allocations to target the needs o f low-performing regions and schools which cannot afford to improve their performance with their own resources, and 0 facilitating learning about which interventions are most effective in raising sub-standard learningachievement. 33. Inaddition to getting the necessary resources to the schools with sub-standard performance, the intermediate and high-case options would also provide more flexibility in use o f budget resources - to include, for example, support for extracurricular tutoring or preschool education if these are found to be effective interventions for improving learning achievement o f the lowest-performing schools and students. Budget funds would cover whatever interventions are found through evaluation of pilot actions to be cost-effective inraising learning achievement to national standards. Inorder for this approach to be sustainable, targeted interventions to raise the performance o f students and schools which are not meeting national learning standards would need to be financed under the budgetprocess itself, rather than through ad-hoc partner-financed projects. This does not mean that partner support should be phased out. On the contrary, it would help to mobilize more donor support if the Government i s able to put in place an effective mechanism for monitoring school performance, for targeting budgets and interventions to schools which are not meeting national learning standards, and for carrying out evaluations o f pilot actions to identify the most educationally cost-effective interventions to improve sub-standard performance. 36 Because education provides benefits to the entire society as well as to the individual and the local community, reliance on local provision and local financing generally leads to provisiono f less than the economically efficient amount or quality of education. 31 OPERATIONALIZINGTHE MILLENNIUMDEVELOPMENT GOALS FORHEALTH: A CASE STUDY OF THE KYRGYZREPUBLIC EXECUTIVESUMMARY 1. The Kyrgyz Republic has committed to achieving the health MDGs. The objective o f this case study is to use the generic framework for analyzing the MDGs in Central Asia (focus on monitoring outcomes, identifying effective interventions, targeting the interventions, strengthening incentives for providers and users) for the analysis o f the healthMDGs inthe Kyrgyz Republic. The case study i s based on a review o f relevant literature on the topic and interviews with key stakeholders in the Kyrgyz Republic. The key findings o f the study are: 2. MonitoringOutcomes: There are significant disparities inMDGhealth outcomes inthe Kyrgyz Republic. Although there are concerns with the reliability and validity o f the MDG health outcome measures, using Ministry o f Health and other sources o f data, the case study finds that there are significant disparities in MDG health outcomes by region (oblast), urban and rural areas and among the poor and the non-poor. For example, while in some oblasts, health outcomes are steadily improving and these oblasts may even be on track for achieving the health MDGs, in other oblasts the situation i s worsening. To operationalize the health MDGs, it i s important to go beyond simply monitoring outcomes to also trachng progress on proxy indicators such as the coverage o f key programs such as immunizations, Directly Observed Therapy (DOTS)for TB. Disaggregatedanalysis o f these indicators i s also needed. The availability o f these data in the Kyrgyz Republic i s fragmented across the different vertical programs and disaggregated analysis i s not easily available. In situations where disaggregated analysis o f outcomes and intermediate variables i s available, it i s not integrated into a results oriented monitoring framework that promotes action and tracking o f the health MDGs. 3. IdentifyingInterventionsEffective in RaisingMDG Performance: Globally, there is a good evidence basedon service delivery interventions effective for achieving the health MDGs. Ninetypercent o f these interventions can be provided through primary care and public health services. The Kyrgyz Republic has maintained and improved upon the high primary care coverage levels o f the pre-transition period through a program o f targeted interventions aimed at strengthening primary care. Coverage o f immunization programs and skilled attendants during delivery i s also very high. In the absence o f disaggregated data, it i s not possible to evaluate coverage levels by urban and rural areas and oblasts. Data from qualitative studies indicate that access to emergency obstetric care (EmOc) i s a concern inrural areas. The quality of Safe Motherhood Programs also varies a lot and rural areas are particularly vulnerable to low levels o f quality. Coverage o f other key programs such as newborn health, Integrated Management o f Childhood Illnesses (IMCI) nutrition and harm reduction programs for prevention o f HIV/AIDS is low. Inthe majority of cases, the MDGinterventions are initiated and financed by donors and there are concerns with institutional and financial sustainability. 4. Targeting Interventionsto Groupsthat are not reachingthe MDGs: Targeting interventions to groups that are not reaching the MDGs requires that budget financing i s appropriately allocated to the right interventions and the right groups, and attention i s paid to service delivery bottlenecks in reaching these groups. The case study finds that there are concerns with targeting o f budget financing. Not only are the levels o f budget financing for health limited, but the allocated amounts have been steadily reducing inthe last 8 years. Through the healthreforms, progress has been made on improving allocation 32 for primary care. Nonetheless, the allocation pattern still favors hospital care. Regional allocation patterns are also skewed and some o f the poorest oblasts inthe country receive the least budget financing. The input-based budgeting system in the context of a resource constrained environment means that salaries are prioritized over non-salary items. This orientation has not served the health sector well in a context where the health sector has moved away from input to output-based budget methods (per case in the hospital sector and a capitation rate for primary care). In cases where the budget has been entirely defined by outputs (contribution per beneficiary for health services), these programs have been more vulnerable to cuts by the Government since they were not definedper traditional budget classifications. 5. There are various bottlenecks to reaching groups with worse health MDG outcomes. These bottlenecks need to be identified and addressed. For example, providing quality services in rural areas will require attracting and retaining skilled staff and improving access to medicines and emergency services. The Ministry o f Health recognizes these problems and has identified a strategy for strengthening rural health services by making greater use o f community-health workers (feldshers) and i s already training and providing medical equipment to feldshers. The next steps are to evaluate quality o f care among feldshers including uptake o f the training and strengthening the linkages between feldshers and the rest o f the health system. The rural pharmacy pilot i s testing how the problem o f pharmaceutical supply can be addressed. The M O H also has plans for strengthening ambulance services inrural areas. For addressing health problems such as micronutrient deficiencies, brucellosis and other lifestyle problems, a community-based health promotion program i s most effective. There are successful pilots that have tested community-basedhealth initiatives ("Jumgal Model" inNaryn and Talas). 6. Strengthening Incentives for patients and providers. The Kyrgyz Republic has achieved major improvement in the use o f primary health care services by developing appropriate incentives for providers and consumers. To accelerate action on the health MDGs, these incentives could be further fine tuned. For example, primary care doctors could be provided with performance bonuses to achieve expanded coverage o f preventive health services or healthpromotion actions. Table 11: AchievingFullerAttainment of the ExpandedHealthMDG Monitoringthe healthMDGs I Short Term MonitoringofMDGs (within 1-3 within wider monitoring years) and evaluationframework for the healthsector. 33 r Long-Term (in 3-6years) I Target Effective Interventions to Regions and Groups with worse Health MDG outcomes3' 1 Short Term Strengthenaccess to rural Scale-up Jumgal (within 1-3 healthcare by training community healthmodel years) feldshers, providing andruralpharmacy equipment and 'scheme to at least 7 out of rehabilitingfeldsher ambulatory points (FAPs) I , 9ob1asts. Strengthenambulance servicesfor ruralareas. i r Longer-Term Strengthenservice Evaluate gaps incoverage (in 3-5 years) delivery network linking inkey interventionsfor feldshersto the health achievingthe MDGs and promotion and primary develop actionplan for healthcare s! stem. further targeting of interventions Short Term Introducerisk-adjusted Introduceincentives Country-wide capitation for primary programsfor addressing implementation care doctors highpriority MDGs such as TB (improve coverage ofDOTS), andharm reductionprograms in selected pilot oblasts Longer-Term Introduceperformance Introduceperformance Scale-up performance bonuses for primary care bonuses for primarycare bonus system doctors linkedto monitoring outcomes for MDGs (such as quality and 37The proposedactions are basedon available information. 34 Short Term allocation formula for categorical grants (calculated on a per capita basis with adjusters) for the health sector in2006 Increase allocation for primary health care (defined as family doctors +feldshers) by at least 3 percent F Longer-Term 35 INTRODUCTION 7. The Kyrgyz Republic has committed to achieving the international health MDG targets. Specifically, by 2015, it plans to reduce by two-thirds the infant and under-five mortality rate, increase immunization coverage to 100 percent, reduce by three-quarters the maternal mortality rate, halt and begin to reverse the prevalence o f HIV, and other communicable diseases. The Kyrgyz Republic's communicable disease targets for the MDGs include brucellosis?' According to a World Bank report on MDG+, improving adult mortality from various causes including non-communicable diseases and trauma i s likely to improve life expectancy by 5.67 years in the Kyrgyz Republic. In comparison, reaching the maternal and child mortality MDG goals would buy 2.41 years in life expectancy (see Chapter 1o f main report for detail^).^' However, so far an an MDG+ agenda has not been defined for the Kyrgyz Republic. 8. The objective of this case study i s to apply a generic framework for analyzing and identifying policy options for full attainment of the health MDGs in the Kyrgyz Republic. The framework emphasizes the following: (i) monitoring MDGhealth outcomes to identify groups that are not achieving the MDGs, (ii) identifying interventions that are effective in raising MDG performance, (iii) targeting interventions to groups that are not achieving the MDGs, and (iv) strengthening incentives for providers and users to raise MDG outcomes. Where budget resources are a constraint - as is usually the case - it also involves selectivity inthe use ofbudget resources. 9. The case study discusses each element o f the framework and then identifies policy and programmatic options for the Kyrgyz Republic. The case study i s based on a review o f relevant literature o n the topic and interviews with key stakeholders in the Kyrgyz Republic. The list o f institutions/organizationsinterviewed for the case study i s includedinAnnex 1. 38UnitedNations. The Kyrgyz Republic: MillenniumDevelopment Goals: Progress Report, The UnitedNations, Bishkek, The Kyrgyz Republic. 39 World Bank. 2004. MillenniumDevelopment Goals for Health inEurope and Central Asia, The International Bank for Reconstructionand DevelopmentiThe World Bank, Washington DC. 36 1. MONITORINGMDGHEALTHOUTCOMESTO IDENTIFYGROUPSTHATARE NOT ACHIEVEING THE MDGS 10. Before discussing the status o f the health MDGs in the Kyrgyz Republic, it i s important to highlight data constraints in reliably tracking the health MDGs in the Kyrgyz Republic. Former Soviet Union (FSU) countries have not, as yet, adopted the internationally-accepted definition o f a live birth. This means that infant mortality rates are a significant underestimation in most FSU countries.40 The child mortality data in the Kyrgyz Republic suffers from similar problems, although there are opportunities for cross-validation o f the data using the 1997 Demographic and Health Survey (DHS), World Development Indicators (WDI) and TransMonee Data (TMD). Inaddition, the Kyrgyz Republic is one o f the pilot countries for adoption o f the international (WHO) definition o f infant and under-five mortality which led to an increase in recorded IMR and USMR in 2004. The increase i s expected to continue this year also as health facilities will implement more accurately the new standards. The measurement of maternal mortality suffers from many o f the same problems experienced with measuring maternal mortality in other parts o f the ~ o r l d . ~The measurement o f TB and HN/AIDS and other ' diseases also suffers from problems related to weak disease surveillance systems and disincentives for providers and patients to report communicable diseases. 42 Therefore, it i s clear that to accelerate action on the MDGs it i s important to improve tracking o f the healthMDGs. Maternal and Child Health 11. Disparities in Infant and Under-five Mortality Rate: At an aggregate level, childmortality indices inthe Kyrgyz Republic have continued to improve since the 1990s and in2004, the IMR for all of the Kyrgyz Republic was 25.6. Under-five mortality duringthe same time period has improved fi-om 41.3 to 29.5 per 1000 live births.43 This i s largely due to drop in the birth rate and improved spacing between births. However, aggregate data mask disparities by urban and rural areas, and by oblast. For example, as Figure 6 shows, some oblasts such as Issyk-kuland Chui are experiencing continuous improvements in IMR and USMR, while inother oblasts and cities, IMR is worsening (e.g. in Osh city, the IMR is almost 34 per 1000 live births and the under-five mortality in Batken i s 38, 30 percent higher than the national rate) Analysis of IMR from acute-respiratory infections (ARI),which i s the leading cause o f infant and under-five deaths, reveals an interesting picture o f disparities across oblasts. Although the DHS data i s from 1997, it i s worth mentioning the gaps in MDG health outcomes by income quintiles based on this data (see Annex Table 1). For example, according to the 1997 DHS, the IMR among the poorest quintile 40There are two factors that contribute to discrepancy inIMRand U5MR calculations: (i) use o f the Soviet the '' definition o f a live birthis still inuse, reducing IMRby 22-25 percent; (ii) misreporting o fbirths and infant deaths by medical personnel; (iii) two mainproblems is that live births are reported as still births or miscarriages and The ost-neonatal deaths as child deaths. Measurement o fmaternalmortality suffers from severalproblems: (i)i s often underreported since women die it outside the health system. Incountries where a large percentage o f women give birth at home, the problemo f underreporting i s magnified; (ii) maternal mortality may be misclassified. Health workers may not know why a woman died, or whether she was or had recently been pregnant. Evenifthe health worker does know, the information i s not always recorded. Deaths are sometimes intentionally misclassified, especially ifthey are associated with clandestine abortions, (iii) Methods usedto calculate maternal death rates are often complex and costly to use. The actual number o f maternal deaths ina specific place at a specific time is relatively small. Therefore, very large populations mustbe surveyed inorder to get accurate estimates. 42For example, surveillance systems for drug use, HIVIAIDS and STIs continue to depend on compulsory notificationiregistration systems and outdated screening programs which remain from the Soviet era. The Centers for Disease Control (CDC) has pilot sites for quality control inBishkek and Osh. There are no sentinel and second- generation surveillance system for HIViAIDS and STIs. 43These improvements are attributable to several social and economic changes inthe country including a drop inthe birthrate andimprovement inspacing of children. 37 was 83 as compared with 46 in the richest group per 1000 live births yielding a poorhich ratio o f 1.81. The pooririchratio for USMRwas 1.95.44 Figure 6: IMR 1997-2004 (per 1,000 live births) 1997 1998 1999 2000 2001 2002 2003 1 --cKyrgyzRepublic-.c- Batkenobi h Jalai-Abad ob1 --c-ksyk-Kulobi +NarynOb1 Osh ob1 Talas ob1 I Chui obi I Bishkek +Osh City 1 1 I ~~ 44Gwatkin D et al. 2000. Socio-economic Differentials inHealth, Nutrition and Populationinthe Kyrgyz Republic, The World Bank, Washington DC. 38 Figure 7: Rayons and Towns with IMR Above Oblast Average 1997-2004 ' 70 65 60 55 50 ~ 45 40 35 30 25 20 15 ~ 10 I 1997 1998 1999 2000 2001 2002 2003 I Batkensky rayon Jalal-Abadcrty Karakol crty --eNarynCity I I '--tKyrgyzRepublic +Chatkalsky Rayon -cNookatskyRayon -A- Talas Clty +--Alarmduskry rayon I Figure 8: USMRby Oblast (selected years) I 50 45 40 35 30 25 20 15 10 5 t o Kyrgyz Chui ksyk-Kul Talas Osh Jalal- Batken Naryn Bishkek Osh clty Republic Abad city I1997 o2000 I2003 2004 1 12. Underlying Determinants of Child Health: Immunization coverage i s a key contributor to reducing child mortality. Available data shows that immunization coverage in the Kyrgyz Republic i s high (almost 98 percent). The latest available data fi-om UNICEF corroborates a high immunization coverage rate (99% for measles and 98 percent for DPT). There are no significant disparities in immunization coverage either. One o f the concerns i s the fiscal sustainability o f immunizations since immunization are entirely funded by donors (see Table 4). The largest contributor to infant and under-five 39 deaths i s deaths during the perinatal and neonatal period.4s Infant deaths during this period are related to undiagnosed problems during pregnancy and poor quality o f management o f deliveries and newborns. The coverage of the neonatal integrated package in the Kyrgyz Republic i s still limited and there are various problems related to training o f staff and the availability o f medical equipment for managing perinatal emergencies (Table 4). 13. The second most important cause o f infant and under-five deaths i s the incidence of acute respiratory infections (ARI). The prevalence o f ARI has decreased significantly between 1996 and 2004 inall oblasts. However, the intra-country differences are still large with the ARI rate inBatkenoblast in 2003 exceeding the rate in Bishkek by about four times. Risk o f ARI i s linked to general nutritional status of children, exposure to cold as well as indoor air pollution.46 In the Kyrgyz Republic, rural households use indoor stoves for warmth in the winter and poor ventilation leads to high exposure to smoke. According to the 1997 DHS, the percentage of children moderately to severely underweight was higher among the poor as compared to the rich. Mortality from ARI i s a key indicator o f the quality of health services delivery, since ARI mortality can be reduced through immunizations, timely diagnosis o f danger signs by mother and access to quality health services (trained providers, access to antibiotics). Children diagnosed with ARI in three southern oblasts (Batken, Osh and Jalal-Abad) are much more likely to die from ARI compared to children living in Northern oblasts. Bishkek, Chui and Issyk-Kul have the lowest infant deaths due to ART. Integrated Management o f Childhood Illnesses (IMCI) has been piloted in several oblasts but there are still considerable financial and service access barriers to timely use o fhealth services (Table 4). There i s no data on mothers' recognitiono f danger signs for ARI. 14. Disparities in Maternal mortality: As in the case of child mortality, maternal mortality in the Kyrgyz Republic has improved from 62.9 in 1990 to 46.4 in 2004. Improvement in birth spacing has contributed to these positive changes. Nonetheless, pregnant women inthe Kyrgyz Republic, especially in some oblasts and rural areas, face a highprobability o f complications and possibly death from pregnancy- related causes. As inthe case of child mortality, large disparities inMMR are noted across oblasts, rayons and towns. Thus, in2003, MMR inTalas oblast was four times that o f Osh City and almost twice as high as the country a~erage.~'Talas oblast shows the most alarming trend. Maternal mortality ratio has risen sharply between 1997 and 2004 and i s significantly above the national average. Although starting 2001 it has been falling, the rate o f decline i s extremely small. Chui and Naryn oblasts also display worrying trends. Issyk-Kul has seen a dramatic decline but although this trend has leveled off in the past three years, it i s still above the national average. Osh oblast has experienced a steady downward trend and has stayed below the national average. MMR has been also declining inBishkek, although between 1998 and 2000 itwent up significantly. 45The perinatalperiod starts as the beginning of foetal viability (28 weeks gestationor 1OOOg) and ends at the end of the 7thday after delivery.Perinataldeaths are the sumof stillbirths plus early neonataldeaths. Neonataldeaths are during the first 27 days ofbirth. 46McCrackenJohn P and SmithKirk.1997.An AnnotatedBibliography on Preventionof Acute Respiratory InfectionsandIndoor Air Pollution (with special reference to children inDeveloping Countries).EnvironmentalHealthProject,Arlington, VA, USA 47However, as it has beennotedearlier the maternaldeaths even incountrieswith complete coverage of deaths in vital registrationmaybe misclassifiedinas muchas 50 percent o f cases. Thus, the trend data andhighly disaggregateddata have high standardvariations and measurementerrors. See for further discussionWagstaff et al., 2003; UNICEFIUNFPAIWHO, 2004. 40 Figure 9: MMRper 100,000 Live Births by Oblast (selected years) 180 I 160 140 I 120 100 80 ~ 60 40 20 0 Kyrgyz Batken Jalal- Issyk-Kul Naryn Osh* Talas Chui Bishkek Osh City Republic Abad I I 1I1997 2000 I2003 0 2004 1 I 15. Underlying determinants of maternal health: Coverage o f skilled attendance i s high (98%) but there are concerns with quality o f care (Table 14). According to the 1997 DHS, 9.4 percent o f women from poor households were likely to deliver at home, compared to 0.7 percent in the richest quintile. While the percentage of antenatal visits among women from poor andrichhouseholds was very high, rich women were twice as likely to have been seen by a medical doctor as compared with a medically trained person. The three leading causes o f maternal deaths are pregnancy disorders (gestosis), sepsis (infection) and bleeding. The highprevalence o f pregnancy disorders i s causedby lack o f timely antenatal care. For example, women from urban migrant communities and rural areas do not register early enough for regular check ups. Sepsis i s mainly due to poor hygiene and quality o f care during delivery. Also, mothers are increasingly delivering at home, especially in rural areas where transportation i s difficult and hospitalization can be These factors are further compounded by the weak skills o f health workers to deal with obstetric emergencies and lack o f essential drugs such as clonidin or oxytocin (Schuth, 2003). The prevalence o f anemia in pregnant women greatly increases risk o f maternal death. Between 2001 and 2003, the prevalence o f anemia inwomen o f reproductive age was lowest inIssyk-Kul oblast (26.2 percent in 2001 and 22.8 percent in 2003), closely followed by Chui (from 49.5 percent to 40.2 percent). Jalal-Abad on the contrary, shows consistently high and increasing rates o f anemia (from 61.8 percent in2001 to 76.5 percent in 2003). Talas oblast has the second highest rate o f anemia among pregnant women (62 percent in 2003). Although there are several strong socio-economic factors driving this problem, to a significant degree it is due to lack of knowledge and negligence on the part of health workers as well as women themselves. According to the Rapid Appraisal Study conducted in Chui and Issyk-Kul oblasts, 72 percent of all 261 women participating in the Study never took any iron supplements duringpregnan~y.~' 16. Micronutrient deficiency i s a growing problem in the country and one o f the leading causes o f morbidity and mortality among children. According to the 1997 DHS, 50 percent of children 6-36 months o f age suffer from anemia (1.4 percent from severe, 24 percent moderate, and 24.4 percent mild). Also, Vitamin A deficiency among children is becominga problem. In2003 a serological study detected 48Interview with Former Minister o f Healthof the Kyrgyz Republic, Dr.Tilek Memanaliev 49Scuth, 2003 41 that 33 percent o f children under 5 had Vitamin A deficiency, i.e. less than 20ug/dL5' Under a donor- funded program, Vitamin A supplementation has been increased to 90 percent but financial sustainability i s a concern. The national program for IDD was introduced in 2003 and i s entirely financed by donors. Only 40 percent o f the salt available for human consumption i s iodized(see Table 14 for details). Calories per day per person Anemia among pregnant women Extreme poverty rate (%) (%) 2001 2002 2003 2001 2002 2003 2001 2002 2003 Kyrgyz Republic 2136 2152 2180 56.2 44.5 53.8 13.5 13.8 9.4 Batken ob1 2517 2394 2301 65.7 47.9 48.2 10.9 14.3 5.7 Jalal-Abadob1 2168 2217 2356 61.8 54.6 76.5 8.3 11.7 5.4 ISSyk-Kul obi 2065 2055 2143 26.2 17.8 22.8 21.3 18.7 21.o 7.4 Naryn ob1 1928 1970 2081 54.7 48.3 38.1 36.3 34.7 Osh ob1 2092 2110 2008 58.2 42.4 55.6 19.9 17.4 17.0 Talas ob1 2139 2186 2229 66.1 44.8 62 25.4 21.8 23.1 Chui ob1 2331 2410 2492 49.5 39.4 40.2 5.9 6.0 3.8 Bishkek 1885 1872 1926 59.4 52.1 54.5 3.1 5.6 2.6 Osh City 2092 2110 2008 48.3 47.3 57.7 19.9 17.4 17.0 COMMUNICABLEDISEASES: HIV/AIDSAND BRUCELLOSIS THE KYRGYZ TB, IN REPUBLIC 17. HIVMIDS: The Kyrgyz Republic is experiencing a rapid growth o f HIV/AIDS. This rapid growth i s caused by the explosion o f HIV-infected injection drug users (IDU),which make up 85 percent o f the registered cases.51 Eighty-one percent of those HIV-infected are males, and 64 percent o f this group i s 29 years or younger. As i s indicated by the experience o f other ECA countries (Russia and Ukraine), there i s a high probability o f rapid spread of HIV infection among the IDU population and subsequently into the general population. Inaddition to IDUs, HIV-prevalence i s also high among some other high-riskgroups (sex workers). There i s a window o f opportunity to interrupt transmission of the virus in the Kyrgyz Republic, but only with a rapid scaling up o f "harm reduction" activities (needle exchange, drug treatment, efforts to reduce risky sexual behavior) to about 60% o f intravenous drug users from the current level o f about 5%. 18. Tuberculosis (TB): The Kyrgyz Republic has one o f the highest prevalence o f TB in the ECA Region. A growing concern i s the rise in Multi-drugresistant (MDR) TB. Despite the introduction o f the modern method o f TB control inthe general health system (Directly Observed Therapy, Short-course, or "DOTS"), TB continues to pose a public health threat to the KyrgyzRepublic. Apart from the problem in the general population, TB in prisons poses a particular challenge, because o f the high rates o f infection there, the policy o f granting amnesty to TB-infected persons that was inplace untilrecently, and because the "prison health system" i s not coordinated with the general health system. MDR TB inprisons i s high and increasing because DOTS i s not properly functioning in the prisons. MDR TB i s spread to the general population because MDR TB i s not treated in the prisons. In addition, there i s no system o f exchange o f information between the prison and civilian health systems and prisoners infected with TB are "lost" to the health system. The situation with TB i s most serious inJalal-abad and Osh oblasts and in 50 (Schuth, 2003). 5 1InformationonHIViAIDS inthe Kyrgyz Republic is derived from: Ministry o fHealth.World Bank.2005: Reversingthe Tide:The HIViAIDSEpidemic inCentralAsia, The World Bank, WashingtonDC. 42 Bishkek city. The overall incidence rates in cities are higher than in rural areas. TB i s also a major problem among HIV-infected individuals. There is an immediate needto upgrade the ability o f the prison health system to detect and treat TB and MDR TB usingmodern methods. Related to this is the need to coordinate TB control efforts with the general health system o f the country to ensure continuity o f care when prisoners are released and to share skilled staff and expensive diagnostic equipment. New investments are needed to establish the capacity to detect and treat MDR TB. The positive impact o f the recent change in the legislation that used to grant amnesty to prisoners on the grounds of moderate or severe TB is already clear. Figure 10: Notification Rate (1)Notification rate, all new cases GUlN and civilian system; (2) Notification rate, all new cases civilian system 180.0 - 160.0 - iE I4O.O -' 1 1 1::: I2O.O -1 - 11 -Notification rate,all new cases GUN and civilian system 40.0 - --(c Notification rate,all new 20.0 1 cases civilian system Source: ICRC Report, 2005 Figure 11: TB MortalityRate in the Kyrgyz Republic, With and Without Prisons 30.0 - 25.0 - i 20.0 1 16 9 E I5.O - * 1-TB mortality rate - with prisons I 7 6 5.0 +TB mortality rate without prisons 0.o ~~ ...~ 43 19. Brucellosis: Brucellosis i s an infectious disease which i s spread from animal to human beings and i s characterized by highfever andjoint pains. Ifuncontrolled, it can cause enlargement o f the spleen. As Table 13 shows, the prevalence of brucellosis has been steadily increasing in the Kyrgyz Republic (from only 19.1 in 1996to 43.9 per 100,000 population in2004). As inthe case o f the other MDGs, there are disparities in prevalence. For example, prevalence in Batken and Talas are almost three times the national rate. Prevalence i s the lowest inBishkek (only 6). Brucellosislargely affects the rural population that i s involved inagriculture and cattle rearing and affects adults and children alike. Table 13: Prevalenceof Brucellosis per 100,000 population, by oblast, 1990-2004 % change 1990 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 97-04 ISSVk-KUl 16.9 30.7 44 47.6 66.7 64 55.9 102.4 72.7 70.2 67.1 297 Jalal-Abad 10.9 12.8 14.1 18.3 22 24.2 25.2 39.5 32.8 68.4 37.6 245 Naryn 15.3 15.2 16.5 18.1 16.1 18 80.1 84.1 89.4 112.8 98.2 542 Batken 11.2 14.3 11.2 67.8 145 119.9 Osh 11.8 9.7 15.3 6 6 7.8 7.4 17.3 15.8 15.2 16.9 43 Talas 29.9 11.8 13 21.9 22.7 17.5 33.1 78.5 49.4 76.7 115.9 288 Chui 17.4 25.5 35.2 42.6 26.6 28.4 32.3 39.7 38.8 45.2 44.5 156 Osh City 5.3 11.2 Bishkek 3.4 5.3 8.8 5.9 7.1 8.3 13.2 12.3 6.9 6.8 6.20 82 Kyrgyz Republic 12.8 14.5 19.1 19.1 19.3 20.1 24.9 36.7 35.2 50.3 43.9 243 NON-COMMUNICABLE DISEASES 20. The Kyrgyz Republic i s facing the epidemiological transition and non-communicable chronic diseases such as diabetes and hypertension are becoming more and more common contributing to morbidity and mortality from strokes and heart attacks. The underlying factors for NCDs in the Kyrgyz Republic are similar to other ECA countries (diet, consumption o f tobacco and alcohol). Generally adults rather than children are more likely to suffer from NCDs and pensioners are most at risk. The percentage o f adults smoking (defined as +15 age group) has declined in the Kyrgyz Republic from 19.8 percent to 16 percent (between 1998-2001) with a comparable decline for the poor and non-poor, although the decline i s slightly higher among the non-poor (3.5% among the non-poor as compared with 1.4% among the extremely poor). Incomparison to smoking rates, alcohol consumption rates inthe Kyrgyz Republic have increased from 22 to 37 percent between 1998 - 2001. The rates o f alcohol consumption among the extreme poor have doubled as compared to only a 16 percent increase among the non-poor. The percentage drinking alcohol more than once a week has doubled with the sharpest increase among the poor.52 21. Despite the importance of disaggregated data, the case study finds that this information is not systematically integrated into key policy documents of the Government, such as the National Health Policy, the Comprehensive Development Framework (CDF) and the National Poverty Reduction Strategy (NPRS). This prevents operationalizing the health MDGs into key policy priorities o f the Government as well as the medium-termbudget framework. The key bottlenecks to regular disaggregated monitoring o f health MDGs and integration into policy are largely institutional, political and technical, i.e., : (i) historical orientation, since there was limiteduse o f disaggregated data under the pre-transition social and political system, (ii)lack o f incentives for line ministries to focus on disaggregated analysis and 52 World Bank. 2003.Enhancing Pro-poor Growth : Kyrgyz Republic, The International Bank for Reconstruction and Development, The World Bank. 44 monitoring, since these policy objectives are not promoted through macro policy instruments such as the PRSP and the medium-termbudget framework, (iii) technical capacity inline ministries and other limited institutions (such as the National Statistical Office that i s responsible for poverty monitoring) and (iv) gaps inthe quality and quantity o f available data. For example, there are reliability and validity problems with measuring many of the health outcomes and disincentives for providers and patients to report communicable diseases. Under the health reforms, the Kyrgyz Republic has piloted an impressive health policy analysis programthat combines regular monitoring and evaluation with use o f this information for evidence-based policy-making (the Dfid-Financed WHO Health Policy Analysis Unit). The next steps are to ensure financial and institutional sustainability o f the unit within an overall plan for streamlining and strengtheningthe institutional structure and functions o fthe MOH. Y z 4 N 92uE 9 eeme ij I a, ic, a w 2. IDENTIFYINGINTERVENTIONSTHAT ARE EFFECTIVE FORRAISING MDGPERFORMANCE 22. There i s a good global evidence base on health interventions effective for raising MDG performance (see Table 14). Although this table does not include interventions for brucellosis and non- communicable diseases, these interventions are also known and for some, cost-effectiveness data i s also available. For example, a highly cost-effective intervention against brucellosis i s expanding an animal vaccination program. This requires intersectoral action with the Ministryo f Agriculture and Livestock.54 Cost-effective interventions for addressing cardiovascular diseases i s a way to promote healthy lifestyles (diet, smoking and alcohol consumption). It is important to note that the effective interventions for addressing MDGs can be effectively provided through three service delivery models (see Box 2). Therefore, for fuller attainment o f the MDGs, it is important to analyze coverage o f the various services delivery models as well as specific programs under the various service-delivery modalities. Box 2: Cost-Effective Interventions and Service Modalities for Achieving the HealthMDGs There is a substantial global evidence base on the most cost-effective interventions needed to achieve the MDGs. These interventions canbe organized and deliveredthrough three service delivery modalities: COMMUNITY-BASED SERVICES: This includes promotion o f health behaviors for core MDGs as well as MDG+ (e.g. breastfeeding, adoption of health lifestyle), environmental hygiene, health education relying on community health workers supervised by relevant higher level institutions (e.g. Ministry o f Health institutions on healthpromotion). POPULATION-BASED PREVENTIVE CARE: such as immunization, antenatal care, Vitamin A supplementation etc. These services are delivered through health centers or health posts with community health workers playing a critical role. For remote areas, mobile health teams provide a critical link between community health workers and other parts o f the service delivery network. CLINICAL-BASED INDIVIDUAL CARE: Includes providing first-contact clinical services with referral. Includes primary health care and secondary health care as the service delivery points. 23. The case study finds mixed results in the coverage o f these interventions. Under the Kyrgyz health reforms, the implementation o f which began in 1996, a key objective was to strengthen primary care. Coverage o f primary care during the pre-transition was very high, although the service delivery model was fragmented. In rural areas, under the Soviet system, there were health posts (feldsher ambulatory points or FAPs) staffed by feldsher (community health workers) with limited medical training and midwives. Rural polyclinics (SVA) were staffed by medical personnel. In urban areas, there were polyclinics that served adults, children and women. This fragmented system was expensive to manage and was also cumbersome for patients. For example, a mother had to go different doctors for her own medical care and that o f her children. Under the health reforms, a system o f family medicine has been introduced and rural polyclinics have disappeared and replaced by family doctors. The family doctors are linked to a family medicine center that oversees a network of facilities. The FAPs remain as before. Currently, 95 percent o f the Kyrgyz population i s enrolled with a family doctor. Primary health care i s free and covered under the constitutionally-guaranteed Benefits Package (see Table 15). As Figure 12 shows, utilization o fprimary care and outpatient care by the poor shows a generally positive trend. 54F Rothand J Zinsstag. 2001. Improvements inHuman Healththrough interventions inthe veterinary sector: The case o fBrucellosis inMongolia, The Swiss Tropical Institute, Basle, Switzerland. 51 Table 15: The Programof State Benefits and other HealthCare Benefitsfor VulnerableGroupsin the KyrgyzRepublicFundedthroughRepublicanand LocalGovernment Contributions Beneficiary Category Services Sources of Revenue General Population Free primary care services Local budget consisting o f local (consultations and emergency government contributions and drugs) for all enrolled with a categorical grants. Categorical family doctors; referredhospital grants almost finance 50 percent care with co-payments o f total local government spending on health care and are therefore very important Exemptedpopulation suchas war Free primary care for all enrolled Local government funds for veterans, cancer and TB patients, with a family doctors, referred health supplement the additional low-income pensioners hospital care without co-payment costs for covering co-payments for these groups Pensioners, children, persons Reduced co-payments for Republicanbudgetcontributions with social benefits inpatient care, reduced co- to MHIF payment for outpatient drugs Figure12: Distributionof Visits (primary care outpatientspecialist) + 16 14 12 10 I +.Visit rate 2001 8 , --.-Visit ---rate2003 6 -~ 0 1 Source: Jakab et al. 2005. (Jakab et a1.2005 "Who benefits from the Kyrgyz Single-Payer System: Analysis o f the incidence o fPublic Expenditures" Forthcoming) 24. The Kyrgyz healthreformshave also focused on the reform o f the public healthsystem, although the reforms are still inthe early stages. One key weakness inthe public health system o f the Soviet times was the Sanitary Epidemiological Services (SES). This system did not promote efficient and effective diseases surveillance and i s completely out o f date with current epidemiological practices for disease surveillance. So, one key reform i s reforming the SES system. The other key reform i s implement a health promotion program which i s in line with international good practice on health promotion and focuses on enabling people to understand the determinants o f health and to increase control over these determinants. Health promotion i s intrinsically linked to empowering communities and individuals within community to get involved inhealthaction and is critical for implementingintersectoral actions on health. Key institutional mechanisms for promoting health promotion have been established (Republican Center for Health Promotion, linkages with family doctors). Nonetheless, the coverage o f health promotion programs i s still very low and there i s still a need to link the macro actions on health promotion with community-based actions. The Jumgal Project in Naryn (Box 3) i s a good example o f this and provides an approach to scaling-up the coverage o f community-based health promotion interventions. Adequate financing i s another key bottleneck inscaling-up healthpromotion interventions. 52 Box 3: CommunityActionfor Healthinthe Kyrgyz Republic - ScalingUpthe "JumgalModel" to Achieve the HealthMDGs Community Action for Health (CAH) i s defined as "collective efforts by communities which are directed towards increasing community control over the determinants o f health and thereby improving health." Community Action for Health focuses on health priorities identified by communities and therefore are often intersectoral in natures, addressing problems such as water and sanitation, micronutrient deficiencies, social problems. Intersectoral community-based actions combined with health system interventions on health MDGs i s needed to accelerate progress inreaching the health MDGs. In2002, the Ministryo f Healthand the Kyrgyz-Swiss HealthReformSupport Project beganthe implementation of C A H in one rayon (Jumgal) o fNaryn oblast. Subsequently, the program was extended to all villages in all rayons in Naryn and from 2004 to all o f Talas oblast. There are now around 200 health committees inboth oblasts combined. The project helped form Village Health Committees (VHC) that would work with the community on identifying and addressing priority health problem. The community prioritized anemia (linked to maternal mortality), hypertension (non-communicable disease linked to MDG+), brucellosis (core health MDG), women's health (linked to STD and maternal and child health). Underlying problems identified include: access to medicines in the village, clean drinking water, improvednutrition, access to medical treatment. The VHCs in collaboration with Ministry o f Health institutions (namely Health Promotion Units) are addressing these problems. As a resulto f this work: The number o f households using iodized salt increased from 71% to 90% inone year Households were convinced o fthe importance o f consuming more vegetables to improve nutrition (especially iron and Vitamin deficiency); among others about ten times more households planted beans to address anemia, especially among women o f reproductive age. As a result, 30,000 households planted bean seeds 0 Improved community awareness o f ways to prevent brucellosis Communities started to work on changing traditions around alcohol consumption Yearly running costs o f this programme are estimated to require around 1.6% o f the health budget o f Naryn oblast. Additional costs were: for establishing committees around 100 USD per health committee, for all four health actions mentioned together around 60,000 USD. Source: Schuth T. 2004. Community Action for Health in Kyrgyzstan: Approach and first Results o f the Pilot Project inNarynOblast, Kyrgyztan, Swiss Red Cross, Bishkek, The Kyrgyz Republic 25. Coverage of key interventions for specifically attainting the MDGs shows a mixed picture. Coverage of key programs such as newborn health, tacking micronutrient deficiencies and integrated management o f childhood illnesses i s quite low. Most o f these programs are donor financed and have not been fully institutionalized within the Ministry o f Health. Each of these donor financed programs are linked to vertical programs within the Ministry o f Health which impedes better collaboration across the vertical programs and addressing intersectoral issues. In addition to coverage, access and quality o f care i s a concern with programs such as safe Motherhood, especially Emergency Obstetric Care (EmOC). Access to EmOC in rural areas is difficult. Transportation i s difficult as many emergency stations do not have gasoline. Also, there are problems with late referrals and rural health workers and midwives do not have the skills to deal with complications. 53 3. TARGETINGINTERVENTIONSTO GROUPS THAT ARE NOTMEETING HEALTH THE MDGS 26. There are two dimensions to targeting groups that are not reaching the MDGs: (i) ensuring that budget financing i s adequately targeted at the right interventions and groups and (ii) addressing the service delivery bottlenecks inreaching disadvantaged groups with the right interventions. Each o f these topics i s further discussedbelow. Levels and OptimalAllocation of PublicFinancingfor the HealthSector 27. As the 2004 WDR argues, "the budget is a critical link on the long route to accountability connecting citizens to providers through politicians and policymakers. The biggest payoffs to service delivery are likely to come for a few key actions o f governments, i.e., spending wisely and predictably in line with priorities and coordinated across sectors." 55 j6 28. One o f the key bottlenecks for maintaining the health reforms and accelerating action on the health MDGs inthe Kyrgyz Republic i s public financing for health. This i s largely linked to problems in budget formulation and execution. Inthe last few years, the Kyrgyz Republic has achieved progress on improving aggregate fiscal disciple through fiscal con~olidation.~'However, the goal o f using the budget for allocative efficiency and equity and strengthening operational impact has yet to be achieved. Key weaknesses in PEM include: weak linkages between public resources and strategic expenditure priorities o f the Government, wide divergence between initially approved budgets and the allocations for non- protected items with large cuts in the provision o f complementary inputs in social and economic development, poor internal and external controls and weaknesses inmonitoring and evaluating the impact o f public spending. The present M&E system i s very simple and i s limited to monitoring inputs.j* 29. There are two key elements to the budget probleminthe health sector: 0 The levels o fpublic sector allocations to the health sector. 0 The continuing problems of inappropriate targeting of public financing on health (geographical and program targeting). 30. Since the health reforms were introduced, the health sector has been getting a smaller and smaller share of the government budget (Figure 13): The Kyrgyz Republic continues to spend very low amounts on health. This i s partly explained by macroeconomic constraints. However, what i s less well explained i s why allocations for health, as a percentage o f total Republican and local public spending has seen a steady decline, i.e., from 13.5 percent in 1996 (when the reforms began) to 7.2 percent in2004 (Republican) and from 25 percent to 22 percent for local. Since public financing covers the Program o f State Guarantees (PSG) and the MHIF program for vulnerable groups, if the allocated levels o f public financing are less than what i s required to provide the PSG, the gaps have to be met through other means, j5World Bank, World DevelopmentReport:Making Services Work for Poor People. International Bank for ReconstructionandDevelopment.The World Bank, WashingtonDC. 56 There is a substantialliteratureonpublic spendingandhealthoutcomes. For details see: WDR 2004: Making Services Work for PoorPeople, The World Bank. BidaniB, andRavillion M:Decomposingsocial indicators using distributional data. Journalof Econometrics,1997, 77:125-139; WagstaffA. Childhealth on a dollar a day: some tentative cross-countrycomparisons. SocialScience and Medicine, 2003; 57: 1529-38; Filmer D et al: Weak Links ina Chain:A diagnosis ofhealthpolicy inpoor countries,World Bank ResearchObserver 2000; World Bank. 2004. Rising to the Challenge:Achieving the HealthMDGs. The World Bank, WashingtonDC. 51 Generalgovernmentfiscal deficit has declinedfrom 6 percent of GDP in2001to 4 percentin2004. 58 World Bank.2005. Kyrgyz Republic: ProgrammaticPublic ExpenditureReview: ConceptNote, The World Bank, WashingtonDC. 54 i.e., greater dependence on out-of-pocket payments and/or cutting services through implicit rationing. Analysis o f household data shows that, in all income groups, households spent a greater share on health care in 2003 than in 2000. The growth in health expenditures does not appear to burden any particular socio-economic group, but nor are the poorest particularlyprotected.59 31. Why didthis happen given that the health reforms andMDGswere endorsed by the President and fully integrated into the Comprehensive Development Framework (CDF) and the National Poverty Reduction Strategy (NPRS). Linkages between the healthreforms and the MDGswere explicit with there i s an extensive legal-normative base, including laws, state policies, state and national programs, intersectoralprograms and concepts that aim at improving the health systemand achievinghealthMDGs. 32. The input-based budget formulation process penalized the health sector for rightsizing health infrastructure and personnel: One o f the key objectives o f the reform was to right size the excessive health infrastructure and health personnel. The savings for efficiency gains were supposed to be reinvested into improving quality and equity o f health care. Under the Soviet system, the health budget for a facility was formulated on the basis of the number of beds, which in turn determined the number of staff. This meant that facilities in the capital city and hospitals were well funded while primary care, especially in rural areas, was severely under funded. Utility costs were prohibitive. The health reforms introduced incentive-based payment systems. As a result o f these reforms, the number o f health personnel in the health sector has reduced by over 30 percent, rural hospitals have merged, the number ofbedshas beenreduced and buildingshave been closed. This meant that under the input-based budget system, reductions in line items resulted in a reduction in allocations. Since health was the only sector that had moved to output-based payment systems, it was difficult for the MOF to determine how to treat health differently from all other sectors. The budget formulation, execution and reporting requirementswere set-up to track inputs and not outputs. 33. The Formalization of Co-payments has Negatively Affected Local Health Budgets: Local governments finance between 30-60 percent o f costs of delivering the PSG. Since the implementation o f the health reforms, allocations for the health sector have declined. As Figure 14 shows, inthe oblasts that implemented health reforms (Issyk-kul and Chui, local government allocations have declined). One o f the main reasons for this i s that local governments believe that since the health sector i s collecting co- payments, they do not need additional funds. Attempts by the health sector to explain that co-payments are not a new or additional source o f revenue but rater formalization of informal payments has had no impact. Inthe absence o f a fiscal accountability framework with local governments, the health has had no power to influence local government allocations for the health sector. ~ 59 WHO/Dfid Health Policy Analysis Unit). 55 Figure 13: The HealthSector hasbeenGettingaSmaller andSmaller Shareof the Government Budget Since 1996(healthexpendituresas YOof totalpublicexpenditures) 13.5% I 11.6% 11 7% A lo.*% 99% 10.1% 0% ~ I 1995 1996 1997 1998 1999 2000 2001 2002 2003 1 +State Budget -I--RepublicanBudget b Local Budget 1 Source: WHO HPAUAnalysis basedonTreasury and MHIF data Figure 14: LocalGovernmentsReduceAllocationsfor the HealthSector inOblastsImplementii HealthReforms 33% 32 % 3 i% 30% 29 I 28% 2 7% 2 6 X 25% 2 4 % 23% 1995 1996 1997 1998 1999 2000 2001 .+lssyk-Kul --I-- Chui Single Payer mother oblasts Source: WHO-HPAU, 2003 34. Operationalizationof the Medium-termBudgetFramework(MTBF) has been Slow: Given the problems with implementing output-based budget formulation in the health sector, reforms in the health sector have welcomed PEM reforms aimed at strengthening the MTBF. As the 2004 WDR mentions: "properly implemented a medium-term expenditure framework can reduce incentives for 56 bureaucratic gaming and reveal the true costs o f the political choices being made in the budget. It can usefully address information asymmetry between the M O F and line ministries." However, attempts to implement an MTBF in the Kyrgyz Republic began in the late 1990s and a series o f adjustment operations have included conditions on the MTBF and donors have provided technical assistance. Despite these interventions, MTBF implementation has been very slow. Slow progress i s due to limited institutional and technical capacity within the MOF, lack of reforms within MOF that would link the annual budget formulation process to the MTBF, as well as lack o f incentives in line ministries to implement the MTBF. Ina context where budget execution i s a key problem and line ministries grapple with cash problems, it was hardfor line ministries to take the MTBF seriously. This reinforces that inthe absence o f a solid foundation o fbudget execution and reporting, it i s difficult to implement and MTBF. 35. Budget Execution favors Inputs: Unpredictability inallocations for the health sector is another major problem, puttingthe health sector under stress to manage the delivery o f health services. Here too, some aspects o f the health budget, which are strongly linked to policy priorities o f the Government suffers. The MHIF Program for Vulnerable Groups covers a contribution from the national budget to the health insurance fund for a reducedhospital co-payment for pensioners, children and persons with social benefits and coverage under the outpatient drug package. Poor budget execution on this program simply means that pensioners and children have to pay out-of-pocket for hospitalization and drugs. In the budget, this allocation shows up as a single program and is not broken down by line item (salaries, medicines, etc). Budgetexecution for this programhas been consistently lower than budget execution for sectors. Since sectoral allocations largely cover salaries, and salaries and pension contributions are protected items, programs that are calculated on the basis o f outputs (programs) suffer. Table 16provides a snapshot o f the first halfof the 2004 fiscal year. Trend data for the last six years shows a similar trend. Budget execution for first half of Fiscal Year (2004) Category % execlref budget Ministryo fHealth (Rep.institutions, salaries, other) 89.80 ,MHIF 28.60 Children`s health insurance 27.61 Social beneficiareis 29.13 I bensioners 130.43 I TDMHIF (Local andCG) 85.60 bishkek 85.50 Osh 78.40 Jalal-Abad 89.20 Batken 69.00 Chui 96.40 Issyk-Kul 85.20 I kalas 192.70 I I baryn 188.30 I otal for the sector (local and Republican) 71.10 57 Table 17: BudgetExecutionfor Children, Pensioners and Personswith Social BenefitsHealth Insurance Source: HPAU calculations based on Treasury and MHIF data for 2000; TCUiPIU o f HeatlhI1Project, Quarterly Report, May 2005. 36. The second key dimension of public financing for health is targeting resources for allocative efficiency and equity. Although the health reforms aimed at this, the changes in favor o f worse-off regions and from specialized care to primary care has been slow. 37. Input-based Budget Formulation Impeded Improving Equity in Resource Allocation: Inaddition to reducing the overall levels o f financing available for the health sector, input-based budget formulation also impeded improvements in equity inresource allocation. As Figure 15 shows, there i s considerable disparity inpublic financing across oblasts. Bishkekcity (with the largest number o f health facilities and personnel) receives the most public financing, while Naryn and Talas (some o f the poorest oblasts in the country with poor MDGhealth outcomes) get the least (Figure 15). Figure 15: Public Expenditures on Health are Not Pro-poor (per capita healthexpendituresat 1995 prices) 3500 3000 j I 313,6 - - 2500 I __ -- -__ 2000 -_ 1 ~- - - - 150 0 100 0 500 I 0 0 local republican MHIF Source: WHO H P A UAnalysis basedon Treasury, MHIF data 58 38. Improving equity in resource allocation i s linked to the implementation o f different methods o f budget formulation, i.e., on a per capita basis and adjusted for indicators such as demographics (children and elderly consumer more health services), sex (women also are heavy uses o f health services), rural and high mountain areas (it i s more expensive to get health inputs to rural areas and to attract medical staff). Under a World Bank adjustment operation, the MOF is implementing changes in the calculation of categorical grants. Categorical grants finance between 60-80 percent o f local financing for health and education and were so far calculated on the basis o f inputs (salaries). In2006, the categorical grants for health will be calculated on the basis on a per capita formula adjusted for a demographic and rural coefficient. The operationalization o f a different mechanism for calculating categorical grants has not been easy in a context where the MOF does not know how it will monitor an output-based budget to oblasts. So far, the M O F could claim that it was paying salaries. Under an output-based system, it will be a more complicated. Given that the Republican-local government fiscal accountability framework i s weak, it i s not sure what will happen if oblasts violate the rules, and the M O F then faces a problem o f unpaid salaries. Nonpayment o f salaries (especially o f teachers and doctors) could generate huge political problems for the national government and they simplywould rather avoid these problems. 39. Reallocations across levels o f care has been politically difficult: Given that Bishkek city i s a major consumer o f public financing for health and there i s substantial duplication in service delivery arrangement and large health facilities which are inefficient, one o f the objectives o f the reforms was to right size Bishkek city, which would mean that Ministry o f Health allocations to institutions inBishkek could be used for other purposes (e.g. financing primary care inrural areas). Almost 50 percent o f the M O Hbudget i s spent on these institutions. Some o f the most powerful people inthe medical community are directors o f these institutions and therefore, right from the start, there was huge resistance to rightsizing in Bishkek city. Arguments were made that Bishkek city was providing a substantial amount o f services to poor communities and the rural population and downsizing in Bishkek would negatively impact the poor. Despite the fact that the data showed that 98 percent o f patients in these institutions were from Bishkek, there was limited progress on restructuring. 40. The Bishkek city government also made various excuses since it was aware that once Bishkek city and Republican budgets were pooled and case based payments introduced, Bishkek city hospitals would be under pressure to downsize and may even have to merge with Republican hospitals since the latter were more popular with patients. The fact that Ministry o f Finance didnot support the reforms and did not promote operational efficiency in health and other sectors meant that Ministry o f Health did not feel any pressure or accountability to downsize in Bishkek city. Under strong pressure from the donors, some rightsizing has taken place, but the impact on resource allocation i s not felt. The savings and reallocations from hospital to primary care in the Kyrgyz Republic so far are largely the result o f restructuring rural hospitals. 41. In the context of the sectoral Medium-termBudget Framework, the M O H has not provided leadership in prioritizing across all health sector programs in the context o f the health ceilings. For example, in the 2005 - 07 MTBF, the MOH argued for a 24 percent increase for the Ministry o f Health budget, which includes Republican institutions in Bishkek while the Health Insurance Program for Vulnerable Groups was severely under-funded. For example, in year 1 o f the 05-07 MTBF, the gap between needs and availability for the MHIF Program was 76%. This increased to 113 percent in 2007. In contrast the gap between needs and availability for the MOH portion of the budget was steadily reducingover 05-07. Inthe absence o f the MOF asking specific questions on why priority programs were under-funded, the MOH felt no accountability to target public financing for vulnerable groups. 59 Addressing ServiceDelivery Bottlenecksin ReachingDisadvantaged Groupswith the Right Interventions 42. As Section 1 and 2 have already discussed, certain groups are at the disadvantage in terms o f coverage and quality o f key effective interventions known to impact MDG health outcomes. The underlying factors for the gaps incoverage and quality are complex and there i s some information on the underlying factors for the disparities, although there i s a need for more systematic tracking o f this information. 43. Lack of Skilled Health Personnel, Drugs and Medical Equipment in Rural Areas: One of the key supply side barriers inproviding quality health care inrural areas in the difficulties o f attracting and retaininghealthpersonnel. The majority o f staff inrural areas are above fifty and when they retire, it i s difficult to replace them. There are monetary disincentives since staff i s rural areas are paid very poorly. The monetary issues will be resolves with the introductiono f per capita based categorical grants (in 2006). However, there are also none-monetary factors (intellectual and social isolation, the lack of amenities such as electncity and transport, absence of professional support). These factors are harder to address. Some countries have addressed this problem by attracting young medical graduates for a fixed term appointment in rural areas with substantial financial incentives (allowances, waiver o f student loans). Another option i s training local recruits in medicine and using community health workers 60(feldshers). Feldshers in the Kyrgyz Republic are already frontline providers for 25 percent o f the population. This population lives insome o f the most remote areas o f the country that are completely cut- off duringthe winter months. Therefore, addressing the skills levels o f feldshers, providing basic medical equipment and drugs, providing professional support and development to feldshers by strengthening linkages with family doctors and providing ambulances i s key. Feldshers also play a critical role in integrating community, population and clinical care. This linkage i s key to accelerating progress on MDGs. 44. The SpecialProblemof MigrantsinUrbanAreas: Access to and utilizationo f health services among the internal migrantpopulations inthe capital i s becoming increasingly important inimproving the general health status in the country. Migrants from rural areas are unaware o f the rules and regulations governing access to health care inurban areas. For example, the majority o f migrants do not know that their health insurance benefits are portable and they have free access to a family doctor as they would have in their place o f residence. Untimely access to health services i s major cause o f child and maternal mortality among migrants. Some o f the same factors that contribute to worse health outcomes in rural areas are also present among migrants (poor nutrition, cultural factors and norms). It i s also likely that stress plays a key role among migrants associated with living in a new environment but under difficult conditions. However, there are no targeted interventions so far for these groups. 45. Social Determinants of Health: Recent research on health inequalities show that different groups may respond very differently to the same health intervention. The response o f groups i s mediated by a range of social and economic factors (social determinants o f health). Little is known about the social determinants o f health. However, generally, poor households have less access to information as compared with non-poor households.61 This can affect the adoption o f health lifestyles or result in delays inseeking treatment. Poor household may also face material barriers to improvinghealth such as lack of access to green vegetables or other nutritious food. Cultural factors affecting poor health outcomes are also more entrenched in rural as compared with urban areas. These complex intersectoral issues were 6oWorld Bank.2005.Global Monitoring Report.The World Bank, WashingtonDC. 61Rockfeller Foundation.2003.UnderstandingHealthinequalities,The Rockfeller Foundation,New York. New York. 60 best addressed through strengtheningcommunity-based health promotion and strengthening the interface between public health, health promotion and primary care. Interventions tackling the social determinants of health are so far in a pilot stage in the Kyrgyz Republic and need to be evaluated and scaled-up (see Box 3 on Jumgal Model). 4: STRENGTHENINGINCENTIVESFOR PROVIDERSAND USERSTO RAISEMDGOUTCOMES 46. The 2004 World Development Report emphasizes the importance o f incentives in reaching the poor with better health services. The Kyrgyz health reforms have already contributed a great deal to improving incentives for providers and users. The implementation of legally enforceable contracts between services providers and policy makers has enhanced accountability. The achievements in output- based payment systems and contracting have been the most successful in the hospital sector where it has contributed to downsizing o f the infrastructure and personnel while improving the delivery of services. In the primary care sector, there i s still considerable work that neededto be done to improve incentives. For example, primary care providers are currently paid on the basis on an unadjusted capitation rate. This means that primary care doctors have little incentive to adequately treat children, women and the elderly or the poor who may have worse health problems. There i s a need to risk-adjust this capitation rate. In addition, as the experience of other countries has shown, primary care doctors can be encouraged to achieve specific preventive and health promotion objectives by linkingpayments to the achievement o f these goals. For example, providers could be encouraged to improve the quality and effectiveness o f DOTS coverage or of antenatal care. The issues o f risk-adjusted capitation and bonus payments are recognized by the Government and there are plans to address this inthe next phase o f healthreforms. 47. The health reforms have also addressed consumer incentives for timely utilization of care by reducing financial access barriers and making sure that the majority o f the population has incentive to use primary health care as the first point of contact in the health system. Despite this, consumers in rural areas still face access barriers since they have to travel to reach a family doctor. Although the co- payment rates are low, as the available information shows, the co-payments are not low enough for certain sections o f the population. Transport costs are also a concern. For addressing specific diseases problems such as TB and STIs, there i s also a need for consumer incentives to encourage reporting and completion o f treatment. 61 CONCLUSION 48. Table 11describes in detail policy and program actions that can be taken by the Kyrgyz Republic for full attainment of the MDGs. Many o f these actions are already includedunder the new health reform program of the Government (Manas 11). For rapid attainment o f the health MDGs, the Kyrgyz Republic has to accelerate action on rural health services delivery addressing the bottlenecks o f skilled staff, medicines and ambulances as well as gaps inquality o f care inmaternal and child health, TB, HIV/AIDS, brucellosis and management of chronic diseases. Many of these interventions are already under implementation. Another high-impact intervention for fuller attainment o f the health MDGs in the Jumgal Model. Rapid scaling-up o f the Jumgal Model i s therefore strongly recommended. Currently, internal migrants are a vulnerable group and yet there i s no program for addressing the needs o f this group. This needs to be urgently identified, implemented on a pilot basis, evaluated and scaled-up, Finally, as this case study indicates, there i s a need for continued investigation o f the underlyingfactors for disparities incoverage o f services among various groups. 49. All interventions must be ensconced within a result-based monitoring framework for the MDGs and appropriate levels and optimal allocations of budget financing for health. If the interventions are not embedded within these two areas, there i s a danger o f continued problems with fragmentation, lack o f prioritization and an unstable fiscal and institutionalcontext for full attainment and sustainability o f the healthMDGs. e M m - z m r - 9 9 9 0 0 0 0 0 0 0 -3 W 4 4 4 3 3 3 -4 o m m V N W ; N 0 0 0 0 0 0 2 z vi q o w t- m 0 09 mo! -z t- v! 0 3 3 3 3 N b . c? 2 2 -9 m c! 9 r- 0 m 3 00 sE L u z 3 CCI N 3 5 -h 3 cd 0 .n D m 3 " z0 m d 5 0 z P 0 $. ? o ' : g o o 2 v) Annex Table2: Listof Organizationswhose Representativeswere Interviewedto Identify BottlenecksandInterventions: 1. Bishkek Territorial Department o fHealthInsurance Fund 2. Mandatory Health Insurance Fund 3. Medical Information Center, Ministry o f Health, 4. Primary Health Care Department, Ministry o f Health 5. Migration Services Department, Ministryo f Foreign Affairs 6. Republican AIDS Center 7. Project HOPE 8. ICRC 9. Swiss Red Cross, Kyrgyz-Swiss HealthReform Support Project 10. UNAIDS 11. UNICEF, Health Project 12. UNICEF, Nutrition Project 13. U S A I D Regional Mission for Central Asia, Health Specialist 14. Department of Obstetrics and Gynecology, Kyrgyz Russian Slavic University. 66 REFERENCES Asymbekova, Gulnara. Unpublished Report. Analysis o f the Maternal Mortality for 1999inthe Kyrgyz Republic.Ministry of Health of the Kyrgyz Republic, Bishkek. Evans, T. et al. 2002. Challenging Inequalities inHealth: From ethics to action. Rockefeller Foundation, NewYork, NY. Godinho, Joana et al. 2005. Reversing the Tide: Prioritiesfor HIV/AIDSPrevention in Central Asia. The World Bank, Washington DC. Godinho, Joana et al. 2005. Stopping Tuberculosis in Central Asia: Priorities for Action The World Bank, Washington DC. International Organization for Migration and ProfessionalManager Consulting. 1999.Research on International Migration inthe Kyrgyz Republic. Bishkek.September. MinistryofHealthofthe KyrgyzRepublic, Department ofHealthCare ReformsProject Implementation Unit.Quaterly Report, May 2005. Roth, F. and J. Zinsstag. 2001. Improvements inHuman Healththrough interventions inthe veterinary sector: The case o fBrucellosis inMongolia, The Swiss Tropical Institute, Basle, Switzerland. Schuth, Tobias. 2003. Access and Quality o f Primary HealthCare with Focus on Mother and ChildCare: Rapid Appraisal Study inChui and Issyk-KulOblasts, Kyrgyzstan. UNICEF, inthe Frame o f CARK M C HForum, August. Swiss RedCross HealthReforms Support Project. Unpublished.Distribution o f Sprinkles to Children Aged 6-36 Months, Pregnant Women and Lactating Mothersto Reduce andPrevent Anemia inN a y Oblast, Kyrgyzstan. Project Proposal. UNFPA. 2002. People, Poverty and Possibilities: Making Development Work for the Poor. State o f the World Population. UNICEF, Ministryo f Healtho f the Kyrgyz Republic and Center for Study o f Public Opinion and Forecast. UnpublishedReport. Study o f Anemia Among Women of Kara-SuuRayon (District), Osh Oblast. UNICEF, Ministry o f Healthof the Kyrgyz Republic and Center for Study o f Public Opinion and Forecast. 2001. Hb Level Research inWomen o f Reproductive Age and Children Under Three Years o f Age inNaryn Oblast. Bishkek. UNICEF Office inCAR and Kazakhstan, Kazakh Academy o f Nutrition, Kyrgyz Research Institute of Obstetrics and Pediatrics. 2003. EvaluatiodAssessment of Vitamin A Deficiency Prevalenceinthe Kyrgyz Republic.Almaty. UNICEF. Unpublished.Inputsinto a Mid-TernReview of National Efforts to Eliminate Micronutrient Malnutrition inCARK. Endo f Trip Report, June 15 -July 8,2002. UnitedNationsDevelopment Program. 2003. The Kyrgyz Republic:Millennium Development Goals Progress Report, The UnitedNations, Bishkek. World Bank. 2004. MillenniumDevelopment Goals for HealthinEurope and Central Asia: Relevance and Policy Implications.The World Bank. Washington DC. Wuhib, Tadesse et al. 2003. "Underestimation o f Infant Mortality Rates inOne Republic ofthe Former Soviet Union." Pediatrics Vol. 111No. 5 May 2003. Accessed on March 21, 2005 at http://pediatrics.aappuplications.orgc/cll/l1/5/e596 Key StatisticalData Sources for Health Case Study Kyrgyz Republic Demographic and Health Survey, 1997. Medical InformationCenter o f the Ministryo f Health. Health o f the Population and Activities of the HealthFacilities o f the Kyrgyz Republic. 2004, 2003. Bishkek. Medical InformationCenter of the Ministry of HealthInternalDatabase. Mandatory HealthInsurance Fundof the Kyrgyz Republic Database. National Statistical Committee. 2004. Food Security and Poverty Information Bulletin. Bishkek. National Statistical Committee Household Budget Survey Data. UNICEF TransMONEE Database. 68