WPS8116 Policy Research Working Paper 8116 Education and Health Services in Uganda Quality of Inputs, User Satisfaction, and Community Welfare Levels Clarence Tsimpo Alvin Etang Quentin Wodon Poverty and Equity Global Practice Group & Education Global Practice Group June 2017 Policy Research Working Paper 8116 Abstract Good health and quality education are essential for eco- service. The quality of service is lowest for those living nomic growth and poverty reduction. Unfortunately, the in poor areas. This has implications for pupils’ learning quality of the education and health services provided in outcomes. Pupils in poor areas perform poorly on a stan- low-income countries is often low. Improving access and dardized test covering English, numeracy, and nonverbal quality of education and health are key policy goals for reasoning. Increased access to education was not accompa- Uganda. This paper builds on the Service Delivery Indicator nied by improvement in learning outcomes. Results from study by further exploring issues related to the quality of econometric analysis suggest that improvements in school service delivery in Uganda. The paper analyzes the quality facilities, improvements in the quality of teaching, and the of service from a poverty perspective, to contribute to the knowledge base of teachers could bring substantial gains ongoing policy debate on the quality of service delivery in student performance, particularly in poor communities. in Uganda, especially in the education and health sectors. Despite the low quality they face, if the poor are more sat- Combining data from the Service Delivery Indicator and isfied with the service, this has implications for demand for the Uganda National Household Survey surveys, the paper social accountability, as the poor often are not exposed to shows a strong correlation between welfare and quality of or ignore the standard of service to which they should refer. This paper is a joint product of the Poverty and Equity Global Practice Group and the Education Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank. org. The authors may be contacted at ctsimponkengne@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Education and Health Services in Uganda: Quality of Inputs, User Satisfaction, and Community Welfare Levels Clarence Tsimpo, Alvin Etang, and Quentin Wodon1 Keywords: Service delivery, Education, Health, Quality, Welfare and poverty, Uganda JEL codes: L88, I20, I10, I30, O55 1 Clarence Tsimpo is a Senior Economist in the Poverty and Equity Global Practice of the World Bank; Alvin Etang is an Economist in the Poverty and Equity Global Practice of the World Bank; Quentin Wodon is a Lead Economist in the Education Global Practice of the World Bank. The authors benefited from discussions with and/or comments from, among others Ruth Hill, Obert Pimhidzai, Waly Wane, Peter Okwero, Elizabeth Ninan Dulvy, and Innocent Mulindwa, and participants of the Uganda Poverty Assessment writers’ workshop. Peer reviewer comments from Andrew Dabalen and Christophe Rockmore are also gratefully acknowledged. We also thank Pablo Fajnzylber for providing comments on earlier drafts. Any errors or omissions are those of the authors only. The opinions expressed in the paper are those of the authors only and need not represent the views of the World Bank, its Executive Directors, or the countries they represent. The authors may be contacted at ctsimponkengne@worldbank.org. 1. Introduction There is broad agreement in the literature that the stock of human capital is a key driver of economic growth and prosperity, as well as poverty reduction. The stock of human capital is a concept that involves not only the overall available labor force, but also skills education and health. Thus, for the government of a country to ensure growth and economic development, it must invest in education and health. Good health and well-being, together with quality education, are part of the major development policy agenda as set in the Sustainable Development Goals (SDGs). Improving access to and demand for education and health are key policy goals for Uganda, because the lack of a qualified and healthy labor force is a constraint to growth, inclusive growth and poverty reduction. As pointed-out by Fox and Pimhidzai (2011), educational attainment is the most important predictor of income and welfare in Uganda. The Government of Uganda (GoU) is aware of this fact, and as evidence for this, education, health as well as physical infrastructure are among the investment priorities of the National Development Plan (NDP). In line with the NDP and in order to provide the environment for economic growth and prosperity, the GoU and its development partners have agreed that education, health and four other sectors (namely Agriculture; Energy; Works and Transport; Water and Sanitation) should get priority in the budget. Since 1997, the GoU has implemented a series of policies as well as investment in order to improve both the supply and the demand for education and health services. On the supply side, a set of policies has been implemented during the past decades in order to improve access. These policies include among others building and renovation of schools and health centers, purchase of adequate instructional materials, training, hiring and retaining teachers and health workers, drugs policy under the national medical store, fight against teachers’ and health workers’ absenteeism, hard to reach hard to say strategy, curriculum change, and decentralization. On the demand side, a series of policies and actions have been implemented as well. The most notable ones include the Universal Primary Education (UPE), the Universal Secondary Education (USE), School Feeding programs, Tuition programs, mama kits programs, and National immunization days, among others. Although there has been a big improvement over the past decade, all these inventions have yet to materialize into fully satisfactory performance of these keys sectors. The reforms that were implemented by the government were very successful in increasing access. Nevertheless, the supply systems were not ready to accommodate the increasing demand. Moreover, financial and human resources have not kept pace with the increase in demand which led to overcrowding and overstretched of physical infrastructure. As a consequence, it is believed that the increase in access was achieved at the expenses of good quality. In recent years, quality of service delivery has become a serious concern for all stakeholders including government and development partners. Yet, evidence suggest that beyond simple access/attainment, the quality of service delivery is very important for economic growth (Hanushek and Woessman, 2012). There is little evidence to support the policy debate on quality of service delivery in Uganda, especially for the education and health sectors. The few available data cover the education sectors only and are from Uwezo, the Uganda National Examination Board (UNEB) through the National Assessment of Progress in Education (NAPE). These data already provide evidence of 2 the quality deterioration. But they all lack the robustness and the specificity that is needed to properly study quality. In line with this concern, and in order to fill the data gap, the World Bank jointly with the African Economic Development Bank and the African Economic Research Consortium, developed and conducted in 2013 an innovative survey on the quality of the education and the health sector in Uganda: The Service Delivery Indicator (SDI) survey. The SDI survey applies a rigorous methodology to collect robust evidence on the quality of primary education and basic health services in a country. The availability of such survey made it possible to conduct an in- depth study of service delivery quality in Uganda. The main report of this survey was published by the World Bank and disseminate in Uganda widely (World Bank, 2013). This paper seeks to build on and extend the SDI study by further exploring issues related to the quality of service delivery in Uganda. In particular, this paper aims at analyzing quality of service from a poverty perspective. The purpose of this paper is to contribute to the ongoing policy debate on quality of service delivery in Uganda, especially in the education and health sectors. The paper aims to build on the richness of the data that where collected to provide a profile of quality of service along the welfare distribution. In addition, the paper shows the link between quality and the user’s satisfaction profile, which allows to draw policy recommendations in terms of community demand for results and accountability. This paper estimates the extent to which quality of service is positively correlated with welfare, and satisfaction is negatively correlated with welfare in Uganda. Section 2 presents the data sources used and the methods considered in the paper with a major focus on specific and critical indicators (absenteeism, caseload, etc.). Section 3 looks at the quality of inputs at the school and health center levels. Section 4 explores quality in terms of knowledge and behavior of the workers (teachers and health workers). Section 5 looks at outcomes at the school level and how this is correlated with teachers and school quality. Section 6 explores correlation between user’s satisfaction, quality of services and poverty. A conclusion follows. 2. Data This study draws heavily from two data sets: The Service Delivery Indicators (SDI) survey of 2013, and the Uganda National Household Survey (UNHS) of 2012/13. The SDI is used to compute the indicators on the supply and quality of services. The UNHS is used to rank facilities by welfare and to derive users’ satisfaction. The SDI is a survey that applies a rigorous methodology to collect robust evidence to gauge the quality of service delivery in primary education and basic health services in a country. Focusing on quality of service delivery, SDI surveys are standardized, allowing for data comparison across countries as well as between regions of a country. The indicators that can be computed using this survey can be ranked into three categories: (i) Provider competence and knowledge; (ii) Proxies for effort; and (iii) Availability of key infrastructure and inputs. In addition, for education, the survey captures information on service delivery outcomes, namely pupils’ knowledge. The SDI collects facility-based data from primary schools and health facilities. The sample frame is the list of facilities in the country. The measurement of these indicators is based on survey 3 instruments underpinned by rigorous research and embraces the latest innovations in measuring provider competence and effort (World Bank, 2013). The sample design for the Service Delivery Indicators is national and disaggregated by rural/urban locations and provider (education and health) type. Tables 1 and 2 provide details on the sample size for education and health respectively. Close to 400 schools and 400 health facilities across the country were selected and visited. The enumerators noted the presence/absence of 3,783 teachers in the school (and also in the classroom) during an unannounced visit to the facility; and 2,214 teachers were assessed on knowledge of English, mathematics and pedagogy. In the health sector, in addition to noting the presence/absence of 1,507 health workers, 736 providers were administered seven vignettes (also called ‘patient case simulations’). The Uganda National Household Survey (2012/13), is a nationally representative survey collected by the Uganda Bureau of Statistics (UBOS, 2013). The UNHS employs a two-stage sampling framework and it covers approximately 7,000 households. The socioeconomic module provides information on household consumption, which is used to derive household welfare indicator. Each district of the country is split into two parts: urban and rural. The average household welfare is then attributed to the facilities in the SDI survey. Subsequently, this allows to rank the facilities by quintile of welfare. The ultimate goal of the paper is to look at the quality of service along the welfare distribution. The community module of the UNHS has a section on the availability of services in the Local Council 1 (LC1).2 The questions in the module are as follows. First households are asked: “Is service available to members of the LC1 (even if they must travel to use it?)” Four answer options to choose from included: Yes, within LC1; Yes, outside LC1; No; and Not Applicable. Next, if the service is available in the LC1 or outside the LC1 (the first two possible answers to the previous question), households are asked six other questions: (1) “What is the name of the provider to which community members most often go to use the service?” (this question is not used in the analysis); (2) “What is the distance from village centre (i.e. geographical middle) to the service?”; (3) “What is the distance from village centre to [SERVICE] in kms?” ; (4) “What is the most common means of transport to the service?” (ten different modes of transportation are listed); (5) “What is the time taken to get to [SERVICE] from village centre using common means of transport? (in minutes)”; and finally (6) “How do you rate the quality of service offered by the service?” (three grades are available: Good, Average, and Poor). In the current paper, those saying that the quality is good are considered as satisfied with the service. 2 There are six levels of Local Councils in Uganda. The lowest level is the Local Council I (LC 1 or LC I), and is responsible for a village or, in the case of towns or cities, a neighborhood. 4 Table 1: Sample size and distribution for education facilities in the SDI survey Provider Location Region Welfare National Public Priv. Urban Rural Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Number Schools 308 76 71 313 95 91 33 60 105 71 77 70 64 102 384 Teacher (for absenteeism) 2,925 722 703 2,944 889 873 328 581 976 681 732 656 594 984 3,647 Teacher (for knowledge) 1,659 382 418 1,623 462 487 174 379 539 413 408 332 334 554 2,041 Pupils (for knowledge) 3,063 732 707 3,088 940 908 328 599 1,020 710 766 672 636 1,011 3,795 Share Schools 80.2 19.8 18.5 81.5 24.7 23.7 8.6 15.6 27.3 18.5 20.1 18.2 16.7 26.6 100.0 Teacher (for absenteeism) 80.2 19.8 19.3 80.7 24.4 23.9 9.0 15.9 26.8 18.7 20.1 18.0 16.3 27.0 100.0 Teacher (for knowledge) 81.3 18.7 20.5 79.5 22.6 23.9 8.5 18.6 26.4 20.2 20.0 16.3 16.4 27.1 100.0 Pupils (for knowledge) 80.7 19.3 18.6 81.4 24.8 23.9 8.6 15.8 26.9 18.7 20.2 17.7 16.8 26.6 100.0 Source: authors using the 2013 SDI survey. Table 2: Sample size and distribution for health facilities in the SDI survey Provider Location Region Welfare National Public Priv. Urban Rural Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Number Facilities 233 148 110 271 51 93 81 60 96 109 84 66 112 10 381 Health workers (for absenteeism) 1,464 762 763 1,463 319 500 521 352 534 575 455 397 724 75 2,226 Health workers (for knowledge) 490 223 160 553 96 194 107 144 172 231 184 113 167 18 713 Share Facilities 61.2 38.8 28.9 71.1 13.4 24.4 21.3 15.7 25.2 28.6 22.0 17.3 29.4 2.6 100.0 Health workers (for absenteeism) 65.8 34.2 34.3 65.7 14.3 22.5 23.4 15.8 24.0 25.8 20.4 17.8 32.5 3.4 100.0 Health workers (for knowledge) 68.7 31.3 22.4 77.6 13.5 27.2 15.0 20.2 24.1 32.4 25.8 15.8 23.4 2.5 100.0 Source: authors using the 2013 SDI survey. 5 3. The quality of inputs is lower in poor areas Primary schools located in better-off areas have higher quality inputs The SDI survey collected information on quality of inputs in the education sector. Pupil per classroom and the pupil to teacher ratios, teacher’s absenteeism, classroom environment, functioning parent-teacher association (PTA), functioning School Management Committee (SCM), and school inspection, are some of the indicators that can be derived from the survey. On various dimensions, quality of inputs seems to be of higher quality in better-off locations. First, consider the pupil per classroom and the pupil to teacher ratios (figure 1 and table 3). The pupil per classroom ratio and the pupil to teacher ratio are much higher for the poorest quintile compared to the richest. A typical classroom in the poorest quintile has 116 pupils, while the corresponding figure for the richest quintile is 58 pupils. A teacher in the poorest quintile has to attend to 58 pupils, while the teacher in richest quintile attends to 31 pupils. The overcrowding of pupils in poorer areas will certainly have negative consequences on learning outcomes. Another interesting finding is that the pupil to teacher and the pupil per classroom ratios are much higher in public schools compared to private schools. In public schools, a classroom has 86 pupils. This number is 35 pupils for private schools. The northern region, which also happens to be the poorest region in Uganda, has the worst performance in terms of pupil per classroom and the pupil to teacher ratios. Teacher absenteeism rates3 from the school or from the classroom (when teachers are in the school) are negatively correlated with welfare. Teachers are more likely to be absent in poor areas. For those in the poorest quintile, about four out of ten teachers are absent from school. The corresponding figure for the richest quintile is two out of ten teachers. The two poorest regions (Northern and Eastern) are also the regions with higher teacher’s absenteeism. Perhaps this absenteeism may be driven the fact that most locations in these regions are hard to reach and hard to stay.4 As illustrated in Table 3, field trip, salary retrieve, strike, and illness are the main reasons for teacher’s absence. Teacher’s absenteeism is much lower in private schools compared to public schools. In Uganda’s public schools, close to 60 percent of teachers were not in the classroom teaching. The corresponding figure for private schools is 39.8 percent. Teacher’s absenteeism induce inefficiency on public spending as they are paid, but not working for the equivalent amount of time. In addition, teacher’s absenteeism can hurt pupils’ educational achievement. When that teacher is repeatedly absent, pupils’ performance can be significantly impacted in a negative way. The more days a teacher is out of the classroom, the more likely their pupils will achieve lower scores in standardized tests (Mary Finlayson, 2009). 3 SDI uses a standardized, internationally benchmarked methodology to measure absenteeism, namely unannounced visits. SDI consists of two visits to each facility; the first is announced in advance so as to increase the likelihood of being able to collect the data underlying most of the indicators. The second visit, which happens during the 7 days following the first visit, is unannounced and its sole purpose is to ascertain the whereabouts of the providers. Providers who are not in the facility because it is not their shift are not considered absent. Health workers who are not in the facility because they are carrying out outreach activities are likewise not considered absent. 4 In a sense that it is difficult to realize infrastructural development, such as roads, electricity, buildings, safe water, etc. These areas are at times cut off from the rest of the district especially in rainy seasons. Because of these problems, hard to reach and hard to stay areas often failed to attract better human resource. 6 The SDI gathered information of the classroom environment for a primary 4 classroom, and infrastructure availability and the school level. These include availability of library, blackboard, electricity, etc. (see Table 3 for a detailed list).5 The learning environment in school is much better in richer areas. For instance, schools in richer areas are more likely to have a library, electricity, and work displayed on the walls. The classroom environment is of low standard in the Northern region, with no library available, no displayed material, and only 3 percent of schools have access to electricity. The classroom environment is much better in private schools compared to public schools. Overall, serious challenges remain when it comes to classroom environment, especially in line with the country’s ambition to become a middle-income country in the near future. Indeed, availability of a library, electricity or displayed material is still very low. For instance, only 8.8 percent of schools have a library. The corresponding figure for electricity is only 10.8 percent. Yet, connectivity to electricity will be important in operationalizing the “skilling Uganda” agenda, towards ICT and more vocational training. Availability of electricity will be crucial to achieve these goals that aim at leaning towards more operational and ready to be employed students at all levels. At the national level, three out of five schools have a functioning Parent-Teacher Association (PTA). Schools in poor areas are less likely to have a PTA. 46.6 percent of schools in the poorest quintile have a functioning PTA. The corresponding figure for the richest quintile is 55.2 percent. In the poorest region, the Northern, only 46.3 percent of schools have a functioning PTA. This finding is in line with previous studies that found that perception of the importance of school is low among the poor (Tsimpo and Wodon, 2015). Seven out of ten schools have a functioning School Management Committee (SMC). There is no apparent relation between welfare and availability of SMC in a school. Both PTA and SMC are more present in public schools. The probability of a school receiving the visit of an inspector during the school year is close to one for most schools. This is true across regions, regardless of wealth. The only exception is the Western region where up to 11 percent of schools did not receive an inspector visit during the year. The number of inspections carried at the school was found to be correlated to welfare. Inspectors tend to often visit schools that are located in better-off areas. Here again, issues related to the fact that poor areas are more likely to be hard to reach hard to stay areas, and transport issues are clear factors limiting the work of inspectors (Office of the prime Minister, 2012). 5 The factor on classroom environment is derived by running a factorial analysis on this list, all items are positively correlated to the first factor. This factor is used as factor for classroom environment (Figure 1). This is weighted average where the weight on each variable in the list is derived from the data. 7 Table 3: Primary schools – resources and management Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Basic resources Pupil per classroom 86.4 35.0 72.4 71.4 47.0 90.5 44.2 136.0 53.4 116.5 88.3 51.0 53.1 57.9 71.7 Pupil to teacher ratio 47.6 22.2 35.5 42.4 28.8 49.6 21.8 61.0 34.4 58.0 51.5 33.4 32.4 31.0 41.0 Pupil to textbook ratio 24.2 92.4 36.7 26.4 14.9 34.2 43.2 91.3 22.7 48.6 63.6 54.0 14.5 13.4 27.8 Teacher absenteeism School absence rate 27.4 13.2 14.2 26.3 21.8 26.0 10.2 34.7 18.5 36.9 25.1 17.2 22.6 18.0 23.8 Classroom absence rate 57.0 39.8 46.7 54.1 46.7 59.7 42.9 69.5 42.1 71.1 52.8 42.5 47.5 49.0 52.6 Reasons for absence Not his/her shift 1.8 5.7 0.0 2.7 0.9 0.0 0.0 4.8 4.8 3.8 2.3 4.0 0.0 1.2 2.4 Sick 17.0 21.0 21.7 16.9 18.6 18.3 15.2 14.9 18.3 11.8 17.2 28.7 14.6 22.8 17.5 Maternity leave 3.9 5.6 4.1 4.1 8.0 3.0 2.1 3.0 2.5 3.9 2.0 3.6 4.1 7.3 4.1 In training 2.3 1.5 0.3 2.4 0.8 2.8 2.9 1.9 3.4 0.7 5.3 5.0 1.1 0.5 2.2 Field trip 29.7 30.3 45.0 27.7 31.3 21.4 34.3 30.2 39.1 27.6 27.8 37.4 25.4 35.0 29.8 Funeral 2.9 0.0 0.0 2.8 4.8 0.6 0.0 3.6 1.2 2.2 1.8 1.8 6.4 0.0 2.5 Other approved absence 1.1 0.0 0.0 1.1 3.6 0.0 0.0 0.0 0.2 0.0 0.0 0.3 0.8 4.1 0.9 Gone to retrieve salary 24.6 25.0 21.2 25.1 23.2 46.3 43.4 11.5 9.6 26.8 32.1 4.6 29.9 21.2 24.7 On strike 10.1 7.6 2.4 10.8 4.7 7.1 2.1 21.6 6.4 16.4 10.9 2.3 7.2 5.2 9.8 Other 6.7 3.2 5.2 6.3 4.2 0.4 0.0 8.5 14.5 6.8 0.5 12.3 10.5 2.6 6.2 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Classroom environment Library corner/books for pupils 9.4 6.9 31.2 5.4 11.1 11.2 36.6 0.0 7.4 1.1 1.0 11.6 4.2 28.2 8.8 Blackboard/whiteboard in class 99.7 100 100 99.7 100 100 100 100 99.3 100 100 98.9 100 100 99.8 Chalk/marker to write 97.8 99.0 98.8 98.0 100 99.7 100 98.8 94.6 99.7 97.2 94.7 99.2 100 98.1 Working electricity connection 6.0 24.8 32.3 7.6 17.7 11.9 55.6 3.1 4.6 5.4 2.2 6.0 15.7 26.9 10.8 Children's work on walls 8.5 19.6 15.3 10.7 24.3 3.4 33.5 0.0 10.6 2.6 1.4 17.1 13.6 24.0 11.3 Other materials on walls 38.5 44.2 86.8 33.2 56.5 34.4 93.5 33.1 29.3 30.1 16.6 33.5 53.6 70.3 39.9 Hygiene in the classroom 77.9 70.8 91.1 73.9 70.2 61.8 100 85.9 86.3 74.9 69.4 84.6 71.2 80.8 76.1 Blackboard good for reading 86.3 90.4 89.8 87.0 87.6 72.3 100 97.3 93.4 87.5 84.1 92.2 85.1 87.9 87.4 Enough light for reading 95.8 95.6 98.0 95.4 92.7 93.1 90.3 98.1 99.8 94.7 95.7 95.1 95.1 98.3 95.7 Enough light for reading at back 90.7 90.2 88.1 90.9 85.5 82.0 79.0 97.3 99.3 90.5 91.3 94.4 90.1 86.1 90.6 Classroom environment index Mean 0.336 0.388 0.526 0.324 0.405 0.311 0.625 0.300 0.337 0.289 0.259 0.357 0.367 0.493 0.349 Median 0.268 0.268 0.409 0.268 0.337 0.268 0.599 0.268 0.268 0.268 0.268 0.268 0.332 0.441 0.268 Inspection, SMC, PTA Share with functioning PTA 69.1 38.9 69.0 60.2 58.2 60.4 46.2 46.3 73.8 46.6 54.6 72.6 78.4 55.2 61.3 Share with functioning SMC 76.2 50.8 78.8 68.3 54.8 66.3 66.4 77.9 81.5 69.9 69.9 76.3 65.4 65.9 69.6 At least one inspection (%) 98.1 82.5 100 93.2 92.1 99.2 100 98.5 89.0 99.1 85.4 99.3 89.5 97.0 94.1 Number of inspections, mean 5.3 3.6 7.7 4.4 4.0 6.0 5.5 3.9 5.1 4.1 4.6 6.4 3.0 6.2 4.8 Number of inspections, median 4 4 6 3 3 4 3 3 5 3 3 5 3 4 4 Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 8 Figure 1: Inputs in primary schools by welfare and sub region Pupil to classroom and Pupil to teacher Teachers Absenteeism ratio 160.0 80.0% 140.0 70.0% 120.0 60.0% 100.0 50.0% 80.0 40.0% 60.0 30.0% 40.0 20.0% 20.0 10.0% 0.0 0.0% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Western Quintile 5 Urban Eastern Urban Eastern Western Private Rural Central Kampala Private Rural Central Kampala Public Public Northern Northern Provider Location Region Welfare Provider Location Region Welfare Pupil per classroom Pupil to teacher ratio School absence rate Classroom absence rate Pupil to classroom Pupil to teacher ratio 80.0 80.0 70.0 North East 70.0 North East 60.0 60.0 50.0 50.0 Poverty headcount Poverty headcount 40.0 West Nile 40.0 We Mid-North Mid-North 30.0 30.0 East Central Eastern East Central Easte 20.0 20.0 Uganda Uganda 10.0 10.0 South-Western Mid-West South-Western Mid-West Central2 Kampala Central2 0.0 Kampala Central1 0.0 Central1 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 Pupil to classroom ratio Pupil to teacher ratio Classroom absence (teachers) Classroom environment 80.0 80.0 70.0 North East 70.0 North East 60.0 60.0 50.0 50.0 Poverty headcount Poverty headcount 40.0 West Nile 40.0 West Nile Mid-North Mid-North 30.0 30.0 East Central Eastern Eastern East Central 20.0 20.0 Uganda Uganda 10.0 10.0 Mid-West South-Western Mid-West South-Western Central2 Central2 Kampala 0.0 Central1 0.0 Central1 Kampala 40% 45% 50% 55% 60% 65% 70% 75% 80% 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 Classroom absence Factor classroom environment Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 9 Unlike education, there is little correlation between welfare and the quality of inputs in the health sector The SDI survey gathered information on the quality of inputs in the health sector. Caseload,6 absenteeism7 of health workers, availability of management structures, drugs availability, availability of infrastructure and equipment are some of the indicators that can be derived from the survey. There is a clear correlation between patient caseload and welfare, especially if one focuses on the median (Table 4 and Figure 2). Here we focus on the median to avoid the possibility of outliers driving the mean results. The poorer the area, the higher the patient caseload. Looking at the median, a health provider in the poorest quintile consulted 6 outpatients per day, against only 3 outpatients for the richest quintile. As a consequence, sick people in poor areas are more likely to face overcrowding and long queue while visiting their health centers. There is a lot of disparity across provider types (public-private) and regions of health workers’ level of effort. Public health providers’ caseload (6 outpatients per provider per day) was three times that of private providers (2 outpatients). Rural health workers’ caseload was more than twice that of urban health workers’. Health workers in the Northern region, were the busiest and received 6 outpatients on a daily basis. The Eastern and Western regions also had high patient caseload with 5 health provider consulting 6 outpatients each on a daily basis. Unlike the education sector, there is no apparent correlation between health workers’ absenteeism and welfare. At the national level, it is estimated that close to half of health workers were absent, including off duty workers. Excluding off duty, absenteeism rate remains high at 42 percent. The incidence of health workers’ absenteeism is quite similar across welfare quintile. There are important disparities across regions. Health workers are more likely to be absent in the Central region: close to six health workers out of ten where absent when including off-duty. Even if those who were off-duty are excluded, the Central remains the region with the highest absenteeism rate: 51 percent. Surprisingly, the absenteeism rate in the private health facilities is almost as high as in the public sector. For private providers, four out of five health workers were absent. On average about half of health facilities reported the presence of a functioning Health Facility Management Committee (HFMC) with important variations across provider types and regions. Notably, private facilities, especially private for profit tend not to have a HFMC. Only 29 percent of private health facilities have a HFMC, against 73.9 percent for public facilities. This is the factor driven the lowest incidence of HFMC observed in Kampala and in the fourth quintile. Most of the private for profit facilities are in Kampala and in the fourth quintile of welfare. Very few health facilities have a procurement committee or an audit committee (only 5.9 and 6.2 percent respectively). As a consequence, issues related to proper financial and resources management can 6 Patient caseload is and indicator that measures health provider’s level of effort. It is defined as the average number of outpatient visits a health provider attends to per working day. It is computed as the number of outpatient visits recorded in outpatient records in the three months prior to the survey, divided by the number of days the facility was open during the three-month period and the number of health workers who conduct patient consultations (i.e. excluding any staff who do not see patients). 7 See footnote 1 on how absenteeism is computed. 10 be an issue. Disciplinary Committees are available only in one out of five health facilities. The share of health facilities with a quality assurance committee is also low (12.6 percent). Disciplinary and quality assurance committees are more likely to be present in poor areas. Perhaps this is because quality issues are more prevalent in poor areas. For example, as shown above, rural health workers have twice the caseloads of urban health workers, and rural areas are more likely to be poor. In Uganda, most of the public facilities (90 percent) are push facilities.8 While most of the private ones are pull facilities. For public facilities, the drugs system is centralized under the National Medical Store (NMS). Under this organization, the NMS purchase drugs in bulk and handles the logistics of distribution across the country, as well as retrieving the expired drugs for proper disposal. This dichotomy between public and private providers is driving the story behind drugs. Facilities that are located in poor areas are mainly public facilities, and therefore fall under the push system. Private facilities, especially the private for profit, are located in better-off areas, as a consequence, facilities under the pull system are those located in better-off areas. The pull system seems to be effective. Availability of essential drugs is much higher in public facilities that are under the push system from the NMS. The six tracer drugs9 were available in 46 percent of public facilities. Availability of the tracer drugs falls to 15 percent for private facilities. The business model is likely to be determinant on the behavior of private for profit firms. They are there to make a profit, and will focus on drugs that patients will feel comfortable buying, and not on the tracers as recommended by the Ministry of Health. The presence of private pharmacies, and their comparative advantage in the drug business may be another factor explaining the behavior of private for profit providers. Surprisingly, the Northern region is the region with the highest availability of the tracer drugs. Efficiency of the NMS coupled with interventions of NGO are probable reasons for this. Amount the six tracer drugs, measles vaccine registered the highest out of stock. Thus acting on measles vaccine will increase the availability of tracers by a big margin. The SDI collected information on availability of basic infrastructure and equipment expected at a health facility. These include availability of electricity, piped water, toilets, ambulance, microscope, weighing scale, blood pressure machine, thermometer, malaria test kit, HIV test kit, etc. (see Table 4 for a detailed list10). In each case the equipment needed to be observed by the enumerator and assessed as functioning. The availability of these inputs is highly 8 Distribution from the main storage point to a lower level store or health facility may follow the push or pull system or a combination of both. In the pull system medicines requests are sent from the lower level. Adequate human resources are required to calculate medicines needs to last for a certain defined period of distribution. The push system is used mainly in situation where there is no adequate storage space or personnel to manage a range of products. In this case, a list of products is pushed from the higher level of warehouse to the health facilities during a defined time frame. A combination approach is used in situations where the regional or district stores are subsidiary of the Central Medical Store. In this case, the push system may be used to supply these subsidiary stores and the pull systems is used to distribute stock to the health facilities (see WHO website for more information). 9 The six major tracer drugs on Uganda’s essential medicines and health supplies (EMHS) list are: Artemisinin combination therapy (ACT); Cotrimoxazole; Measles vaccine; Oral Rehydration Salts (ORS sachets); Medroxyprogesterone acetate (Depo-Provera); and Fansidar. 10 The index on infrastructure/equipment availability is derived by running a factorial analysis on this list, all items are positively correlated to the first factor. This factor is used as index on infrastructure/equipment availability (figure 2). This is weighted average where the weight on each variable in the list is derived from the data. 11 and positively correlated with welfare. For example, health facilities in richer areas are more likely to have electricity, and piped water. This is not surprising as electricity and piped water networks are non-existent in poor areas, and alternatives are costly.11 Only one out of ten health facilities has a functioning ambulance. Ambulances are more likely to be available in richer areas. Surprisingly, availability of telephone (land line and cell phone) remains low in health facilities. All health facilities in the richest quintile have an adult weighing scale, while the corresponding figure for the poorest quintile is only 58 percent. Maternity waiting centers (antenatal rooms) are available in only 23.9 percent of health centers. This probably explains the fact that a low proportion of women delivered in formal health facilities under the attendance of a specialized health worker, despite the high rate of attendance for antenatal care and the mama kit program (Tsimpo and Wodon, 2015). The Northern and the Eastern regions, which happened to be the poorest, are also those with very limited availability of infrastructure and equipment in health facilities. 11 Solar system for example for electricity. 12 Figure 2: Inputs for health facilities by welfare and sub-region Health workers absenteeism Push or a pull facility and drugs availability 70% 100% 90% 60% 80% 50% 70% 40% 60% 30% 50% 40% 20% 30% 10% 20% 0% 10% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Eastern Western Private Rural Central Kampala Public Northern 0% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Provider Location Region Welfare Natio Welfare Absent including off-duty Absent excluding off duty Push Pull Both Drugs availability - 6 tracers (%) Caseload (median) Health workers absenteeism 45.0 45.0 40.0 West Nile 40.0 West Nile 35.0 35.0 Mid-North Mid-North 30.0 30.0 Poverty headcount Poverty headcount 25.0 25.0 East Central Eastern East Central Eastern 20.0 Uganda 20.0 Uganda 15.0 15.0 10.0 Mid-West 10.0 Mid-West South-Western South-Western Central2 Central2 5.0 5.0 Central1 Kampala Central1 0.0 Kampala 0.0 0 2 4 6 8 10 12 14 35% 40% 45% 50% 55% Caseload: average number of outpatient visits a health provider attends to per day Absent including off-duty (%) Availability of drugs(6 tracers) Infrastrcutre availability 45.0 45.0 40.0 West Nile 40.0 West Nile 35.0 35.0 Mid-North Mid-North 30.0 30.0 Poverty headcount Poverty headcount 25.0 25.0 Eastern East Central Eastern East Central 20.0 Uganda 20.0 Uganda 15.0 15.0 10.0 Mid-West 10.0 Mid-West South-Western 5.0 Central2 5.0 South-Western Central2 Kampala Central1 Central1 0.0 0.0 Kampala 0 10 20 30 40 50 60 70 80 0.100 0.150 0.200 0.250 0.300 0.350 0.40 Drugs availability - 6 tracers (%) Index of availabity of infrastructure Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 13 Table 4: Health facilities – caseload, workers, management, and drug availability Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Caseload Mean 10 2 2 8 7 8 1 6 10 8 5 11 3 3 6 Median 6 2 1 5 4 5 1 6 5 6 4 5 2 3 3 Absenteeism rate () Absent including off-duty 51.2 52.2 50.7 52.4 59.3 52.2 50.4 54.2 48.0 54.8 55.1 46.7 51.2 64.2 51.6 Absent excluding off duty 44.0 38.6 36.9 45.8 51.0 43.0 36.3 48.3 42.3 49.1 50.4 37.4 37.5 48.3 42.0 Reason for being absent Sick/maternity 8.2 4.1 4.2 8.4 5.9 8.7 4.7 11.0 5.6 11.3 7.7 3.5 5.4 0.0 6.5 In training/seminar 11.7 5.7 5.4 12.3 23.6 4.4 5.0 8.5 13.0 8.4 7.6 17.1 6.7 28.9 9.3 Official mission 7.4 4.6 4.9 7.3 6.9 14.8 1.9 8.0 4.9 11.9 12.0 3.6 2.5 13.0 6.2 Approved absence 13.8 18.7 17.7 14.3 21.4 13.4 17.8 11.7 13.4 11.8 17.9 15.6 17.1 5.4 15.8 Not his/her shift 26.1 44.8 45.5 24.4 30.6 30.6 46.2 21.1 23.4 19.7 20.6 32.3 45.4 48.0 33.8 Doing fieldwork 0.3 0.3 0.0 0.5 0.0 1.1 0.0 0.0 0.4 0.8 0.2 0.4 0.0 0.0 0.3 Not approved Absence 20.7 6.8 8.8 20.0 6.5 21.9 9.5 18.8 21.8 21.8 17.0 19.4 9.6 3.5 15.0 Gone to retrieve salary 0.7 0.0 0.0 0.7 0.0 0.7 0.0 0.7 1.1 0.7 0.4 1.2 0.0 0.0 0.4 Outreach 2.1 2.9 0.9 3.6 1.0 1.9 0.0 6.2 6.0 4.0 4.9 2.6 0.7 0.0 2.4 Other 8.9 12.2 12.6 8.4 4.0 2.7 14.7 14.1 10.5 9.6 11.6 4.4 12.6 1.2 10.2 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Management Health Facility Mgt Committee 73.9 29.2 19.3 72.3 84.8 67.7 11.7 67.0 70.9 75.6 74.9 67.7 20.8 85.8 52.3 Finance Committee 13.9 23.3 12.3 22.2 37.7 15.6 9.2 34.4 14.0 21.4 18.7 24.1 13.6 39.5 18.4 Procurement committee 2.1 9.9 4.7 6.6 20.0 3.0 5.4 6.6 0.6 6.4 4.1 9.6 4.7 0.0 5.9 Audit committee 2.6 9.9 5.4 6.6 14.1 5.5 5.5 9.0 1.5 6.7 1.0 12.4 5.0 12.6 6.2 Disciplinary Committee 22.8 25.2 13.4 30.3 27.8 31.9 11.8 52.3 16.1 37.8 25.0 26.0 14.7 17.1 24.0 Quality Assurance Committee 13.6 11.6 11.9 13.0 4.9 18.6 9.7 22.1 10.6 17.0 13.4 13.6 9.3 12.6 12.6 Push or pull facility Push 90.1 6.9 12.2 74.7 44.0 73.9 5.7 89.5 77.0 86.6 74.5 65.1 14.2 43.5 54.0 Pull 3.2 83.7 82.7 16.0 37.4 19.7 90.9 3.8 14.4 8.8 11.2 24.8 80.4 56.5 38.1 Both 6.7 9.4 5.1 9.3 18.6 6.4 3.5 6.7 8.6 4.6 14.4 10.1 5.4 0.0 7.9 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Drugs availability (6) 46 15 19 38 23 24 17 69 40 41 32 44 17 39 31 Availability by drug ACT 90.9 89.1 91.4 89.1 89.3 81.1 94.7 89.1 91.6 82.6 89.8 97.6 90.7 74.7 90.0 Cotrimoxazole 80.2 89.1 87.2 82.8 74.2 72.5 87.6 89.9 92.7 80.5 89.9 84.0 84.5 82.9 84.5 Measles vaccine 76.1 33.5 29.6 71.3 73.9 61.7 24.3 87.5 66.5 70.9 65.5 79.6 30.3 65.8 55.6 ORS sachets 78.8 81.5 81.9 79.0 74.3 67.4 81.7 90.7 85.4 77.0 82.4 77.4 82.0 87.4 80.1 Depo-Provera 98.5 66.0 81.6 83.6 61.1 85.2 82.2 94.2 87.5 90.3 78.0 85.1 79.8 73.1 82.8 Fansidar 84.4 70.8 70.5 82.3 88.7 80.4 68.2 94.4 73.9 86.8 81.0 78.0 71.2 87.4 77.9 Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. Note: ACT=Artemisinin combination therapy. 14 Table 4 (Continued): Health facilities - infrastructure and equipment Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Infrastructure/equipment Electricity 55.6 88.6 90.0 60.4 80.6 47.7 92.7 58.4 63.5 49.7 64.5 68.5 88.4 85.8 71.5 Piped water 12.9 50.3 60.2 13.3 19.8 8.3 65.1 4.1 23.0 5.6 22.1 16.6 56.3 57.4 31.0 Flush toilet for outpatients 4.1 34.5 48.9 0.5 1.1 0.0 55.8 1.3 1.4 0.0 0.0 1.6 46.4 22.4 18.7 Flush toilet for staffs 2.6 16.8 20.4 2.8 3.5 2.8 21.9 1.3 5.2 1.0 4.9 3.3 19.2 22.4 9.4 Functioning land phone 4.8 15.8 18.6 5.0 14.3 4.4 19.1 1.6 4.7 3.1 3.0 11.8 16.5 17.1 10.1 Functioning cellular 6.8 25.2 26.8 8.9 8.5 23.3 26.3 0.5 6.7 5.2 7.1 16.2 25.0 25.2 15.6 Functioning shortwave radio 2.0 0.9 0.0 2.4 0.0 3.1 0.0 4.0 1.6 3.4 1.9 1.9 0.0 0.0 1.5 Functioning computer 5.5 22.9 26.5 6.3 6.9 4.0 28.2 9.3 8.4 5.2 10.4 5.7 24.2 39.5 13.9 Access to email or internet 2.6 16.7 20.7 2.6 6.0 3.2 21.8 0.7 3.7 1.1 5.0 3.5 18.7 35.0 9.4 Functional ambulance 2.8 17.3 18.7 4.4 9.9 2.4 18.9 0.0 8.7 0.6 5.7 8.2 17.5 26.8 9.8 Maternity antenatal room 24.5 23.3 14.0 29.9 32.0 38.8 11.5 30.9 20.6 24.1 28.3 40.2 13.6 29.7 23.9 Adult weighing scale 70.4 80.6 78.5 73.4 86.4 64.5 77.0 68.2 79.8 58.3 79.1 88.5 76.3 100 75.3 Thermometer 74.0 92.5 94.6 75.8 94.2 60.0 96.0 79.4 78.8 67.2 84.4 77.3 93.6 100 82.9 Child weighing scale 78.1 48.3 36.4 80.3 73.4 85.1 28.8 76.1 83.4 77.2 90.5 84.0 33.4 95.5 63.8 Stethoscope 79.9 93.2 93.1 82.2 82.5 73.2 95.0 84.1 88.3 76.6 93.1 80.8 91.3 100 86.3 Infant weighing scale 48.6 33.5 28.3 49.3 79.1 48.2 22.2 32.5 46.9 38.1 51.3 54.6 31.8 59.4 41.3 Microscope 32.2 63.9 64.6 37.1 53.8 32.9 65.4 29.1 41.0 24.5 45.2 44.0 63.2 56.1 47.5 Glucometer 18.3 50.5 56.9 19.9 39.8 11.6 57.0 27.0 19.9 15.6 22.6 25.7 53.0 60.6 33.8 Malaria Test kit 80.8 62.0 54.6 82.1 76.1 66.5 53.6 87.4 90.5 75.9 84.7 83.0 58.1 57.7 71.7 Urine Dip kit 28.7 47.3 50.5 29.9 46.4 24.4 50.3 34.1 27.8 24.9 40.9 28.4 47.9 48.0 37.6 HIV Test kit 50.4 67.3 73.4 49.6 71.2 44.6 74.8 47.3 46.2 40.7 57.7 49.1 73.5 82.9 58.6 Tuberculosis test kit 20.7 11.3 17.2 15.6 15.2 22.8 14.1 20.2 11.9 16.5 16.4 17.4 15.2 30.8 16.2 Autoclave 15.7 22.9 30.7 12.3 11.4 11.1 30.0 20.0 14.5 14.3 20.1 6.6 27.8 25.3 19.2 Electric boiler/steamer 1.5 10.0 12.7 1.3 2.8 4.1 13.3 0.0 0.5 1.7 0.5 1.9 11.9 0.0 5.6 Electric dry heat sterilizer 1.6 12.6 16.7 1.0 4.2 0.0 17.8 0.7 2.3 0.4 0.5 2.2 15.8 14.2 6.9 Non-electric pot 51.6 59.1 57.1 54.1 58.8 52.2 57.0 44.1 59.6 43.0 64.0 60.6 55.4 50.8 55.2 Incinerator 9.9 12.6 7.9 13.2 14.0 2.4 6.7 26.8 14.4 14.8 12.4 11.1 8.6 9.7 11.2 Facilities index Mean 0.187 0.347 0.381 0.193 0.262 0.171 0.389 0.187 0.207 0.156 0.228 0.217 0.363 0.417 0.264 Median 0.133 0.275 0.287 0.159 0.213 0.123 0.284 0.157 0.163 0.118 0.215 0.170 0.281 0.329 0.208 Source: Authors using the 2013 SDI and the 2012/13 UNHS surveys. 15 4. Providers’ knowledge and attitudes Teachers had significant knowledge gaps. Teachers’ knowledge of the subjects they teach as well as pedagogical knowledge are very low, especially in poor areas The share of teachers with minimum content knowledge was observed based on the results of a customized teacher test administered to Primary 4 mathematics/numeracy and English teachers. The English test results were for teachers teaching English, and the mathematics test results were for teachers teaching mathematics. The tests were based on items from the curricula being taught in Uganda (World Bank, 2013). The objective of the teacher test is to examine whether teachers have the basic reading, writing and arithmetic skills that lower primary students need to have in order to progress further with their education. This is interpreted as the minimum knowledge required for the teacher to be effective and is the basis for the “Share of Teachers with minimum knowledge indicator”. The test results revealed that teachers’ knowledge of the subjects they teach was very low, and the pedagogical skills to transform their knowledge into meaningful teaching were even lower. On average, teachers scored 59 percent and 64 percent in the English and numeracy/mathematics tests, respectively (Figure 3 and Table 5). There is little difference on teachers’ knowledge by provider type. Performance of teachers in private schools is quite close to those of teachers in public schools. There is a clear positive relationship between teachers’ knowledge and welfare. Teachers’ knowledge increases with welfare of the location. For instance, teachers in the poorest quintile scored 56 percent and 59 percent, in the English and numeracy/mathematics tests, respectively. The corresponding figures for the richer quintile are 62 and 68 percent. As a consequence of the positive correlation between teachers’ knowledge and welfare, the Northern region which is also the poorest is the region where teachers’ scores are lowest for both the English and mathematics tests. The SDI survey also assessed teaching ability through a test of pedagogical knowledge. Survey estimation results suggest that pedagogy skills are disappointingly low, as reflected in the average score of 25 percent on the pedagogy test, and only 7 percent of teachers scored above 50 percent. Pedagogical knowledge appears to be slightly lower in poor areas, in rural areas, and in the Northern and Western regions, highlighting the positive correlation between educational quality and welfare. 16 Figure 3: Primary school teacher’s knowledge by welfare and sub region Teacher assessment (score) Factor analysis on teacher behavior 80% 80.0 70% North East 70.0 60% 60.0 50% Poverty headcount 50.0 40% 30% 40.0 West Nile Mid-North 20% 30.0 10% Eastern 20.0 East Central Uganda 0% Northern Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Private Kampala Urban Western Rural Central Eastern Public 10.0 Mid-West South-Western Central2 0.0 Central1 Kampala Provider Location Region Welfare Natio 20% 22% 24% 26% 28% 30% English Numeracy Pedagogical knowledge Teacher assessment: English Teacher assessment: Mathematics 80.0 80.0 70.0 North East 70.0 North East 60.0 60.0 Poverty headcount Poverty headcount 50.0 50.0 40.0 West Nile 40.0 West Nile Mid-North Mid-North 30.0 30.0 20.0 Eastern East Central 20.0 Eastern East Central Uganda Uganda 10.0 10.0 South-Western Mid-West South-Western Mid-West Central2 Central2 0.0 Central1 Kampala 0.0 Kampala Central1 54% 56% 58% 60% 62% 64% 50% 55% 60% 65% 70% 75% English score (teachers) Numeracy score (teachers) Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 17 Table 5: Assessment of teacher knowledge Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Teacher knowledge English 59 58 63 58 62 60 64 56 56 56 57 59 60 62 59 Numeracy 64 66 70 63 70 66 69 58 61 59 62 64 67 68 64 Pedagogy 26 24 28 25 28 24 29 26 23 26 23 23 26 28 25 Overall 45 45 48 45 49 45 49 43 43 44 43 45 47 48 45 Source: Authors using the 2013 SDI and the 2012/13 UNHS surveys. Table 6: Assessment of teaching quality Provider Location Region Welfare Nationa Centra Easter Kampal Norther Wester Pub. Priv. Urb. Rur. Q1 Q2 Q3 Q4 Q5 l l n a n n The teacher used local information to make learning relevant. 33.4 44.5 35.1 36.4 20.7 20.7 44.7 14.0 75.1 13.3 46.3 61.8 31.7 29.0 36.2 100. 100. 100. Teacher sitting or standing in front of the class? 95.8 91.3 98.1 94.1 97.1 100.0 76.5 95.7 100.0 100.0 84.6 0 0 0 94.6 Teacher going to individual children? 74.7 67.5 78.8 72.0 68.7 80.1 90.6 60.7 76.2 62.4 70.9 90.5 63.7 77.9 72.9 Teacher calling children by name while teaching? 91.5 86.3 98.2 89.0 92.1 84.0 100.0 93.7 91.0 87.3 87.3 96.2 88.2 92.3 90.2 Teacher smiling, laughing or joking with children? 66.0 70.9 91.4 63.7 54.4 71.7 89.8 71.3 71.8 68.5 76.7 79.9 41.6 68.8 67.3 Teacher hitting, pinching or slapping a child? 4.4 4.6 5.5 4.3 3.4 10.3 12.5 1.8 1.6 5.6 5.8 1.7 0.0 9.5 4.5 The teacher asked questions that required learners to recall information. 88.6 90.1 95.3 88.1 93.4 81.3 86.3 84.8 93.8 79.0 87.8 98.7 89.0 91.6 89.0 The teacher asked learners to carry out a task which allowed them to 86.2 89.9 90.0 86.8 77.9 74.8 91.3 79.9 92.9 93.0 78.2 demonstrate 94.4 97.6 100.0 87.2 The teacher asked questions that required learners to apply 68.5 75.0 69.7 70.2 75.2 62.2 77.6 60.1 77.1 52.9 81.0 77.6 68.7 72.0 70.2 Did the teacher ask questions which required learners to use their creativity an 60.2 75.4 77.0 62.1 70.6 75.6 73.3 12.2 75.8 30.3 76.5 76.3 62.2 78.5 64.1 Teacher gave feedback of praise, moral strengthening and/or encouragement 85.2 80.0 87.6 83.3 79.7 76.4 98.0 91.1 89.0 80.6 90.3 92.9 79.0 75.8 83.9 Teacher gave feedback that was correcting a mistake 91.9 83.5 93.5 89.2 85.2 83.0 83.6 97.3 96.0 89.0 90.4 96.6 89.1 83.2 89.8 Teacher gave feedback that was scolding at a mistake 15.2 25.4 15.9 18.1 12.9 34.4 14.8 2.3 17.2 14.5 38.8 6.0 14.9 14.3 17.8 Did the teacher introduce the lesson at the start of the class? 94.3 96.8 95.9 94.8 95.1 89.2 94.6 95.9 99.0 97.5 86.4 99.5 97.8 93.2 94.9 Did the teacher summarize the lesson at the end of the class? 26.3 23.5 32.5 24.5 24.6 25.3 24.8 38.6 19.7 27.7 29.2 26.9 13.1 30.8 25.6 Did the teacher assign homework to the class? 17.2 8.5 17.9 14.6 9.4 2.8 18.2 67.5 1.8 44.2 5.1 11.5 0.9 10.5 15.0 Did the teacher review or collect homework from the class? 20.9 15.7 29.9 18.1 14.0 16.6 29.5 54.4 7.8 37.3 12.8 18.8 6.1 21.5 19.6 Did the teacher use the local language of instruction? (language other than 54.8 58.2 22.3 60.6 69.0 48.8 22.4 24.8 67.9 30.8 67.0 58.9 71.4 51.9 English) 55.6 Factor analysis on teacher behavior Mean 0.67 0.66 0.74 0.65 0.64 0.69 0.74 0.59 0.66 0.642 0.634 0.754 0.705 0.699 0.670 1 5 6 8 6 0 6 7 7 Median 0.70 0.73 0.75 0.70 0.66 0.74 0.78 0.59 0.72 0.686 0.687 0.785 0.702 0.784 0.707 4 4 7 0 7 6 5 2 2 Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 18 Only a small proportion of health workers were able to diagnose all five tracer conditions. The accuracy of diagnoses is lower in poor areas Quality in the health sector was assessed using two indicators of process (adherence to clinical guidelines in five tracer conditions and management of maternal and newborn complications—as measured in the vignette interviews); and one indicator of outcomes (diagnostic accuracy in the five tracer conditions at the end of the vignette interviews). Three of the tracer conditions were childhood conditions (malaria with anemia, acute diarrhea with severe dehydration, and pneumonia), and two were adult conditions (pulmonary tuberculosis and diabetes mellitus). Two other conditions were included: post-partum hemorrhage, the most common cause of maternal death during birth; and neonatal asphyxia, the most common cause of neonatal death during birth (World Bank, 2013). Only a small proportion of health workers were able to diagnose all five tracer conditions. Only one in four health workers were able to do this. Malaria with anemia is the disease for which health workers exhibit a higher and very good knowledge of. Almost all of them (96.3 percent) were able to properly diagnose this condition. Notably, all health workers (99.9 percent) in the Central region were able to diagnose malaria and anemia. Providers in private facilities are slightly more competent in the diagnostic of malaria and anemia compared to their counterparts in public health facilities. The accuracy of diagnostics is lower in poor area, especially for acute diarrhea, pneumonia, diabetes mellitus, and pulmonary tuberculosis (PTB). The diagnostic assessment shows that health workers perform very poorly on acute diarrhea. Less than half (47 percent) were able to properly diagnose acute diarrhea. Performance on pneumonia and diabetes mellitus is also very low, with only 60 percent able to accurately diagnose each of these diseases. For all the diseases, health workers’ knowledge increases with welfare (figure 4, table 7). For those in the poorest quintile, only 16 percent were able to accurately diagnose the 5 tracer conditions. The corresponding figure for the richest quintile is 39.6 percent. The biggest knowledge gap across welfare quintiles is revealed through diagnosis of pneumonia. 85.3 percent of health workers in the richest quintile were able to properly diagnose pneumonia, against only 44.5 for those in the poorest quintile. The knowledge gap across quintile is also big (double digit) for acute diarrhea, diabetes mellitus, and PTB. Providers in private facilities are more competent in the diagnostic of pneumonia and diabetes Mellitus. Diagnostic accuracy was significantly higher in Kampala and lower in the Northern region. For example, in Kampala, 41 percent of providers were able to accurately diagnose all the 5 tracer conditions. In the Northern region, only 11 percent of providers were able to do so. Only half (54.3 percent) of providers were able to properly manage maternal and newborn complications (post-partum hemorrhage and neonatal asphyxia). There is no clear correlation between welfare and management of neonatal asphyxia. Proper management of post-partum hemorrhage increases with welfare. For example, for the richest quintile, 84.6 percent of providers were able to properly manage post-partum hemorrhage. The corresponding figure for the poorest quintile is 67.6 percent. Regionally, the worst performance is registered in the Eastern and Western regions where only 48.6 and 52 percent (respectively) of providers are able to properly manage neonatal asphyxia. The knowledge gap between these two regions and other regions regarding 19 neonatal asphyxia is very big. In other regions, at least 74 percent of providers were able to properly deal with neonatal asphyxia. Health workers’ knowledge gaps can have significant repercussions on the overall health sector outcomes, and ultimately economic development (Quote XXX). Moreover, because of medical errors due to limited knowledge, the society might be paying for health service, but in return is not getting value for money. As a result, the overall health status is not what it should be. Expenditure while sick does not guarantee that one will recover the stock of physical heath. Figure 4: Health workers knowledge by welfare and sub region Share giving the correct diagnostic (5 tracer conditions) 100% 45.0 90% 40.0 West Nile 80% 70% 35.0 Mid-North 60% 30.0 Poverty headcount 50% 25.0 40% Eastern East Central 30% 20.0 Uganda 20% 15.0 10% 10.0 Mid-West 0% South-Western Western Private Rural Central Public Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Eastern Northern Kampala 5.0 Central2 Central1 Kampala Provider Location Region Welfare Natio 0.0 0.0 10.0 20.0 30.0 40.0 50. All 5 Cases Exactly 4 Cases Exactly 3 Cases Exactly 2 Cases Only 1 Case No Case Share correct in all the 5 cases Post-partum hemorrhage and neonatal asphyxia 100% 45.0 90% 40.0 West Nile 80% 70% 35.0 Mid-North 60% 30.0 Poverty headcount 50% 25.0 40% Eastern East Central 30% 20.0 Uganda 20% 15.0 10% 10.0 Mid-West 0% South-Western Western Rural Central Public Private Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Eastern Northern Kampala 5.0 Central2 Central1 Kampala Provider Location Region Welfare Natio 0.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70. All 2 Conditions Only 1 Condition None of the conditions Share correct in all the 2 conditions Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 20 Table 7: Assessment of health worker knowledge Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 Diagnostics Acute diarrhea 45.3 50.6 58.2 39.2 45.3 41.0 61.0 28.8 41.5 36.1 40.7 38.3 59.2 50.6 47.5 Pneumonia 56.3 65.9 75.4 48.4 54.3 58.3 74.8 42.5 50.1 44.5 57.2 47.7 73.7 85.3 60.2 Diabetes Mellitus 57.8 66.6 76.6 49.7 51.2 51.9 77.2 57.7 50.4 49.1 55.2 50.0 74.5 62.3 61.4 PTB 86.9 89.5 91.9 84.9 91.2 91.7 90.7 77.4 84.0 81.6 87.5 87.4 91.0 94.8 87.9 Malaria/Anemia 95.3 97.9 96.6 96.1 99.9 98.7 96.0 95.6 93.2 96.4 94.5 97.1 96.6 97.8 96.3 Share correc7 All 5 Cases 25.9 28.7 39.4 17.5 17.1 21.7 41.3 11.3 21.8 16.4 14.4 20.0 39.7 39.6 27.1 Exactly 4 Cases 27.2 34.9 38.1 24.3 26.8 35.5 36.4 27.6 18.8 24.2 37.6 20.7 34.9 26.0 30.3 Exactly 3 Cases 21.4 21.5 12.0 28.8 42.2 14.4 11.6 26.6 29.6 23.7 26.6 30.1 14.3 24.3 21.5 Exactly 2 Cases 16.1 9.0 5.8 18.9 8.9 20.3 5.4 21.2 18.4 23.3 11.9 20.1 6.1 7.8 13.2 Only 1 Case 6.8 4.7 1.7 9.3 5.0 7.2 1.7 12.5 9.2 11.4 8.8 7.2 2.1 0.0 6.0 No Case 2.6 1.1 3.0 1.2 0.1 0.9 3.5 0.6 2.1 1.0 0.6 1.9 3.0 2.2 2.0 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Clinical knowledge PPH 68.6 76.0 71.7 71.5 69.7 74.7 71.2 74.2 69.3 67.6 72.3 71.5 72.9 84.6 71.6 Respiratory 62.8 73.6 75.5 60.7 74.0 48.6 80.5 74.5 52.3 65.3 50.6 57.0 79.4 58.3 67.2 Share correct All 2 Conditions 48.1 63.3 58.0 51.4 58.2 40.0 61.6 64.0 46.3 53.6 41.9 47.8 62.4 55.3 54.3 Only 1 Condition 35.3 23.0 31.2 29.5 27.4 43.4 28.4 20.8 29.0 25.8 39.0 32.9 27.5 32.3 30.3 None of the conditions 16.7 13.7 10.8 19.1 14.5 16.7 9.9 15.3 24.7 20.6 19.1 19.3 10.1 12.4 15.5 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Source: Authors using the 2013 SDI and the 2012/13 UNHS surveys. Note: PTB = Pulmonary Tuberculosis/Pneumonia/Chronic Bronchitis. PPH = Post-partum Hemorrhage. Respiratory = Respiratory distress syndrome/Birth asphyxia. 21 5. Outcomes Learning outcomes are highly correlated with poverty. Thanks to the UPE reform, children in poor areas are more likely to attend school, but they are failing to learn As part of the collection of the SDI survey, learning outcomes were measured for grade 4 pupils. The objective of the pupil’s assessment was to assess basic reading, writing, and arithmetic skills. The test was designed by experts in international pedagogy and based on a review of primary curriculum materials from 13 African countries, including Uganda (see Cunningham and Dowling, 2012). The pupil’s assessment also measured non-verbal reasoning skills on the basis of Raven’s matrices, a standard IQ measure that is designed to be valid across different cultures. This measure complements the pupil test scores in English and numeracy and can be used as a rough measure to control for innate student ability when comparing outcomes across different schools. Thus, the pupil assessment consisted of three parts: English, numeracy and non‐ verbal reasoning (NVR). Overall, pupils answered 47 percent of questions on the test correctly. The average score in English was 46 percent and the average score in numeracy was 43 percent. The average score on the non‐ verbal reasoning part was 57 percent. There is a lot of variation in learning outcomes across area of residence, regions as well as public-private schools. Performance of pupils in private schools was significantly higher, especially for English and numeracy. In private schools, pupils answered 65 percent of questions on the test correctly, against 43 percent for those in public schools. Performance in urban were higher in urban areas. Pupils in Kampala are by far the best performers. In Kampala, pupils answered 80 percent of the questions correctly. Conversely, the two poorest regions (Northern and Eastern regions) are the ones with the worst performance. Finding that pupils in private schools performed significantly better than those in public schools is consistent with the reasoning that the quality of inputs have implications on children learning outcomes. It was shown earlier that private schools had better quality inputs than public schools. For example, section 3 highlighted that private schools had better classroom environment, much lower teacher’s absenteeism, and smaller pupil to teacher and the pupil per classroom ratios - all of which can negatively affect pupils learning outcomes. Indeed, regression results below confirm this. Learning outcomes are strongly and positively correlated to welfare. Pupils who are in the richest quintile scored 66 percent overall while those in the poorest quintile scored only 34 percent (table 8). The biggest pupil knowledge gap is observed in the English test. Pupils in the richest quintile answered 69 percent of English questions correctly. For those in the poorest quintile, only 31 percent correctly answered English questions. The knowledge gap across the welfare distribution is also important for numeracy. But there is no apparent correlation between welfare and nonverbal reasoning. A wide range of factors can affect the ability of children to learn in school. Previous work for Uganda suggest that children from disadvantaged backgrounds are less likely to fare well. But school-level factors also play a role (Mulindwa and Marshall, 2013). Given that the SDI data combined information on both inputs and outcomes at school level, we can check whether the quality of input are in some ways related to outcomes. This is done using econometric approach. 22 This econometric analysis looks at the performance of grade 4 primary school students on tests for English, numeracy, nonverbal reasoning, and the student’s overall score. The explanatory factors include school level variables (whether the school is public or private; whether it has a parent- teacher association (PTA), whether it has a school management committee (SMC), the number of inspections carried at the school, an index for the school facilities and the environment in the classroom, and the pupil-teacher ratio), teacher-level variables (an index for teacher quality and measures of teacher knowledge for English and numeracy), child-level variables (age, gender, and whether the child had breakfast) and community-level variables (urban versus rural areas, regions, and welfare quintiles for the community). Results from the regression analysis of the correlates of student achievement are provided in Table 9. Performance in public schools is on average lower than in private schools. Schools with a lower pupil-teacher ratio tend to do better than schools with a higher pupil-teacher ratio. In most cases PTAs and SMC do not make much of a difference, but this could be because how, for example, a SMC operates may be more important than its mere existence. More inspections are associated with higher performance. Finally, better school/classroom facilities and to some extent management are also associated with better student performance. Teacher absenteeism reduces student performance, regardless of whether the teacher was absent from the school, or in school but not in the classroom. Better teaching in the classroom as measured though an index of teacher behavior leads to better student performance, as does a better score of the teacher on English and numeracy tests. Older children tend to perform worse (this may be capturing children who repeated previous grades). Girls also perform worse than boys. Children who had breakfast at home perform better (this may denote a higher socio-economic status of the household, and also point out to the importance of school feeding in disadvantaged areas). Children in urban areas perform better than in rural areas for some subjects. Performance is higher for children in Kampala and the Central region than in the Western, Northern, and Eastern regions. Children in poorer communities do no perform as well as children in better off areas as measured by the community’s welfare index. Generally speaking, these results are consistent with expectations, and have important policy implications. For example, the results suggest that improvements in school facilities as well as improvements in the quality of teaching and the knowledge base of teachers could bring substantial gains in student performance, especially in poor areas. A reduction in pupil-teacher ratio would also bring gains, although these are likely to be smaller, and may be costlier to achieve in terms of budgetary resources. Although one should be careful not to infer causality, it could be that strengthening the inspection regime would also bring gains, while by contrast PTAs and SMCs seem to have less of a beneficial impact. As illustrated by Mulindwa and Marshall (2013), members of PTAs and SMCs often lack the basics skills to properly play their role. Outcomes for the health sector are more complex to measure, thus the SDI did not collect such information. It is therefore not possible to make a similar assessment for the health sector here. 23 Figure 5: Pupil assessment (score) By provider, location and region By welfare 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Public Private Urban Rural Central Eastern Kampala Northern Wester Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Provider Location Region Welfare National English Numeracy Non verbal reasoning Overall score English Numeracy Non verbal reasoning Overall score English Numeracy 80.0 80.0 70.0 North East 70.0 North East 60.0 60.0 Poverty headcount Poverty headcount 50.0 50.0 40.0 West Nile 40.0 West Nile Mid-North Mid-North 30.0 30.0 20.0 Eastern East Central 20.0 Eastern East Central Uganda Uganda 10.0 10.0 Mid-West South-Western South-Western Mid-West Central2 Central2 0.0 Central1 Kampala 0.0 Kampala Central1 20% 30% 40% 50% 60% 70% 80% 90% 50% 55% 60% 65% 70% 75% English schore (pupils) Numeracy score (teachers) Nonverbal reasoning Overall score 80.0 80.0 70.0North East 70.0 North East 60.0 60.0 Poverty headcount Poverty headcount 50.0 50.0 40.0 West Nile 40.0 West Nile Mid-North Mid-North 30.0 30.0 Eastern Eastern 20.0 East Central 20.0 East Central Uganda Uganda 10.0 10.0 South-Western Mid-West Mid-West South-Western Central2 Central2 0.0 Central1 Kampala 0.0 Central1 Kampala 45% 50% 55% 60% 65% 70% 25% 35% 45% 55% 65% 75% 85% Non verbal observation (pupils) Overall score (pupils) Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 24 Table 8: Assessment of student performance Provider Location Region Welfare National Public Priv. Urb. Rur. Central Eastern Kampala Northern Western Q1 Q2 Q3 Q4 Q5 English 43 67 63 42 68 35 85 35 54 31 29 55 62 69 46 Numeracy 42 53 48 42 52 40 56 37 47 38 37 47 50 50 43 Nonverbal reasoning 56 62 60 56 62 56 67 51 59 52 56 60 59 60 57 Overall score 43 65 61 43 66 37 80 36 54 34 31 55 61 66 47 Source: Authors using the 2013 SDI and the 2012/13 UNHS surveys. Table 9 Correlates of pupil achievement (Probit model) Model 2: Model 3: nonverbal Model 4: overall Model 6: Model 7: nonverbal Model 8: overall Model 1: English Model 5: English numeracy reasoning score numeracy reasoning score coef t coef t coef t coef t coef t coef t coef t coef t School absence rate -0.264* 0.154 -0.219*** 0.071 -0.000 0.099 -0.231* 0.125 -0.317** 0.155 -0.265*** 0.079 -0.030 0.123 -0.392** 0.173 Classroom absence rate -0.303** 0.123 -0.063 0.056 -0.065 0.076 -0.245** 0.098 -0.263** 0.123 -0.056 0.062 -0.062 0.095 -0.286** 0.134 There is a PTA -0.009 0.063 0.018 0.028 0.120*** 0.040 -0.004 0.050 0.001 0.063 0.019 0.028 0.120*** 0.040 0.000 0.051 There is a SMC 0.070 0.064 0.010 0.029 0.044 0.042 0.060 0.051 0.074 0.064 0.013 0.029 0.047 0.041 0.067 0.050 Number of inspections 0.009* 0.005 0.010*** 0.002 0.007** 0.003 0.009** 0.004 0.009* 0.005 0.010*** 0.002 0.007** 0.003 0.009** 0.004 Index classroom 1.072*** 0.148 0.278*** 0.061 0.081 0.088 0.828*** 0.116 1.062*** 0.148 0.277*** 0.061 0.084 0.088 0.826*** 0.116 environment Index teacher behavior 0.371*** 0.137 -0.031 0.065 0.044 0.089 0.292*** 0.111 0.366*** 0.137 -0.029 0.065 0.054 0.089 0.293*** 0.111 Teacher score English 0.992*** 0.249 0.156 0.161 0.804*** 0.207 0.917*** 0.249 0.119 0.161 0.743*** 0.208 Teacher score numeracy 0.426*** 0.079 0.306*** 0.108 0.061 0.138 0.408*** 0.080 0.273** 0.108 0.018 0.139 Age -0.190*** 0.021 -0.014 0.009 -0.027** 0.012 -0.146*** 0.017 -0.189*** 0.021 -0.013 0.009 -0.025** 0.012 -0.145*** 0.017 Age squared 0.002*** 0.000 0.000 0.000 0.000* 0.000 0.001*** 0.000 0.002*** 0.000 0.000 0.000 0.000 0.000 0.001*** 0.000 Girl -0.054 0.047 -0.123*** 0.021 -0.066** 0.029 -0.067* 0.037 -0.054 0.047 -0.121*** 0.021 -0.066** 0.029 -0.068* 0.037 Has breakfast 0.157*** 0.050 -0.013 0.023 0.001 0.031 0.111*** 0.041 0.147*** 0.050 -0.016 0.023 -0.004 0.031 0.102** 0.041 Public school -0.626*** 0.072 -0.328*** 0.032 -0.171*** 0.044 -0.545*** 0.057 -0.531*** 0.081 -0.293*** 0.035 -0.107** 0.047 -0.458*** 0.064 Urban area 0.146 0.095 0.063 0.041 0.131** 0.058 0.133* 0.077 0.137 0.095 0.063 0.041 0.132** 0.058 0.131* 0.076 Region Central ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. Eastern -0.784*** 0.132 -0.319*** 0.061 -0.370*** 0.086 -0.699*** 0.111 -0.705*** 0.133 -0.291*** 0.061 -0.322*** 0.087 -0.635*** 0.111 Kampala 0.267* 0.137 -0.037 0.050 0.061 0.078 0.144 0.099 0.291** 0.138 -0.035 0.050 0.063 0.078 0.155 0.100 Northern -0.333** 0.150 -0.259*** 0.068 -0.452*** 0.095 -0.354*** 0.125 -0.249 0.152 -0.231*** 0.069 -0.399*** 0.096 -0.288** 0.126 Western -0.145** 0.068 0.001 0.032 -0.159*** 0.043 -0.119** 0.056 -0.122* 0.069 0.009 0.032 -0.144*** 0.043 -0.100* 0.057 Welfare quintile Q1 ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. ref. Q2 -0.157** 0.079 -0.156*** 0.036 0.008 0.048 -0.145** 0.063 -0.141* 0.079 -0.150*** 0.036 0.013 0.049 -0.129** 0.063 Q3 0.233* 0.142 -0.036 0.066 -0.112 0.090 0.159 0.120 0.208 0.142 -0.054 0.066 -0.145 0.090 0.124 0.120 Q4 0.631*** 0.143 0.049 0.065 -0.288*** 0.093 0.488*** 0.120 0.612*** 0.143 0.033 0.065 -0.315*** 0.093 0.460*** 0.120 Q5 0.590*** 0.144 0.033 0.066 -0.206** 0.091 0.447*** 0.121 0.555*** 0.145 0.012 0.066 -0.244*** 0.091 0.405*** 0.121 Pupil to teacher ratio (PTR) -0.005*** 0.002 -0.002*** 0.001 -0.004*** 0.001 -0.004*** 0.001 Constant 1.310*** 0.308 0.062 0.135 0.586*** 0.197 1.053*** 0.261 1.445*** 0.308 0.120 0.136 0.709*** 0.199 1.215*** 0.262 Number of observations 3,565 3,576 3,546 3,546 3,555 3,566 3,536 3,536 note: .01 - ***; .05 - **; .1 - *; Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. 25 6. Implications for social accountability The quality of services delivery, as well as social outcomes are very low in poor communities, yet these communities are more satisfied with the services The community module of the 2012/13 UNHS ask questions on perceptions of the quality of basics social services. Community members were asked to rank services according to their appreciation of the quality. The possible answers were the quality of service is “good”, “average” and “poor”. The top four rated services are private hospitals (by comparison government hospitals are ranked 13th), NGOs, pharmacies, and banks and financial institutions. Apart from three types of roads that are at the bottom of the ranking (community, feeder/district, and trunk roads), government primary schools and government health centers are poorly rated (see Tsimpo and Wodon, 2015b for more details). The perceived quality of service is negatively correlated with welfare. The poor are more likely to be satisfied with the service that they are receiving, although objective measures from the SDI survey suggest that it should be the opposite (Figures 6 and 7). In particular, for both education and health, the poor are more likely to be satisfied with the public provider compared to the non- poor. A phenomenon close to Inada conditions must be at play here. This means that the poor are so deprived that their expectations are low, and they quickly tend to be happy with the little service that they can get. On the contrary, the non-poor tend to have higher expectations, and therefore will be more demanding in terms of quality, and less satisfied even if objectively they are getting the best service in the country. Poverty can be a barrier for citizens to properly engage in social accountability Lower expectations from the poor can be a problem for social accountability. Social accountability is an approach towards building accountability that relies on civic engagement, in which citizens participate directly or indirectly in demanding accountability from service providers and public officials. Social accountability generally combines information on rights and service delivery with collective action for change. In Uganda, Social accountability has emerged as an important weapon in the fight for better governance and service delivery (Rosetti, 2002). Examples include U-report, Barazas, and Uganda Participatory Poverty Assessment Process. Besides low expectations, there are several other hypotheses for this observation. First, it could just be lack of information by the poor of what their options or choices are. For example, the supply of private facilities may not be available for the poor. Second, poor people just cannot hold providers accountable either because they cannot observe provider quality, or they do not have the power. Third, there also exists the possibility that the poor could be threatened if they engage in organizing themselves. Fourth and finally, the opportunity costs of organizing for a sustained period of time could be really high for the poor. The contrast between the objective measure of quality from the SDI and the perceived quality from the UNHS raises the question of the effectiveness/efficiency of community based monitoring or demand for accountability. If the population has low expectations, or is not exposed enough to know what quality to expect, how can it efficiently act or demand for quality service? Thus, the implication is that for social accountability to be effective and efficient, mitigations 26 actions must be taken to influence expectations of the citizens, especially the most disadvantaged and the poor. There are other examples in the literature that questioned effectiveness of social accountability. Empirical evidence of tangible impacts of social accountability initiatives is mixed (Jonathan A. Fox, 2015). In Benin for example, Keefer and Stuti (2014) found that government inputs into public education are no different in villages with greater access to community radio/information. Moreover, it is claimed that social accountability can fail if the political economy is not catered for (Anuradha Joshia et al., 2012). In a context where poverty and expectations are a problem, more should be done for social accountability to be effective. Figure 6: Quality of inputs, outcomes and satisfaction by welfare quintiles in education sector Absenteeism, pupil per classroom per teacher Teacher and pupil knowledge 50.0 100 50.0 80 45.0 90. 45.0 70 40.0 80. 40.0 60 35.0 70. 35.0 50 30.0 60. 30.0 25.0 50. 25.0 40 20.0 40. 20.0 30 15.0 30. 15.0 20 10.0 20. 10.0 10 5.0 10. 5.0 0.0 0.0 0.0 0% Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Satisfaction Government Satisfaction Private School absence rate Satisfaction Government Satisfaction Private Teacher score: English Classroom absence rate Pupil per classroom Pupil to teacher ratio Teacher score: Numeracy Pupil overall score Source: authors using the 2013 SDI and the 2012/13 UNHS surveys. Figure 7: Quality of inputs, outcomes and satisfaction by welfare quintiles in health sector Satisfaction and health workers absenteeism Satisfaction, drugs and knowledge 100.0 7 100.0 9 90.0 90.0 8 6 80.0 80.0 7 70.0 5 70.0 6 60.0 60.0 5 4 50.0 50.0 4 3 40.0 40.0 3 30.0 30.0 2 2 20.0 20.0 10.0 1 1 10.0 0.0 0 0.0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile Welfare quintile Satisfaction Govt HC Satisfaction Govt Hospital Satisfaction Govt HC Satisfaction Govt Hospital Satisfaction Private Clinic Satisfaction Private Clinic Satisfaction Private Hospital Satisfaction Private Hospital Absent including off-duty Absent excluding off duty Drugs availability - 6 tracers (%) Share correct diagnostic in the 5 cases 27 Satisfaction and child mortality Satisfaction and maternal health 100.0 100.0 1 90.0 90.0 9 80.0 80.0 8 70.0 70.0 7 60.0 6 60.0 50.0 5 50.0 40.0 4 40.0 30.0 3 30.0 20.0 2 20.0 10.0 1 10.0 0.0 0 0.0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Welfare quintile Welfare quintile Satisfaction Govt HC Satisfaction Govt Hospital Satisfaction Govt HC Satisfaction Govt Hospital Satisfaction Private Clinic Satisfaction Private Clinic Satisfaction Private Hospital Satisfaction Private Hospital Infant mortality (1q0) Child mortality (4q1) Median duration of amenorrhoea Percentage delivered in a health facility Source: authors using the 2013 SDI, the 2012/13 UNHS and the 2011 UDHS surveys. 7. Conclusion Building on the originality of the SDI survey, this paper has provided a descriptive analysis of service delivery in Uganda, focusing on inputs (providers’ behavior and knowledge, infrastructure) and learners’ outcomes. Moreover, combining the SDI and the UNHS surveys, the paper shows a strong correlation between welfare and quality of service. Those living in poor areas face the poorest quality of service. This also has implications for outcomes. Pupils in poor areas perform poorly on a standardized test covering English, numeracy and nonverbal reasoning. Increased access to education was not accompanied by improvement in learning outcomes. Econometric analysis suggests that improvements in school facilities as well as improvements in the quality of teaching and the knowledge base of teachers could bring substantial gains in students’ performance, especially in poor areas. By contrast PTAs and SMCs seem to have less of a beneficial impact. The perceived quality of service is negatively correlated with welfare. Poorer communities tend to have services of lower quality but are more satisfied with the services that they are receiving. Low quality of inputs in poor communities negatively affects outcomes such as student learning. The poor are more likely to be satisfied with the service that there are getting, although objective measures from the SDI survey suggest that it should be the opposite. This implies that the poor are so deprived that their expectations are low, and they tend to be happy with the little service that they can get. Conversely, the non-poor tend to have higher expectations, and therefore will be more demanding in terms of quality, and less satisfied even if objectively they are getting the best service in the country. The contrast between the objective measure of quality and the perceived quality has implications for social accountability mechanisms. If populations in poor areas have low expectations, their ability to monitor quality is weakened. Apart from the demonstration of the need to improve inputs for education and health facilities in Uganda, one of the implications of the 28 analysis is that for social accountability mechanisms to be effective, additional measures must probably be taken in order to enable populations in poor communities to ask for and obtain higher quality services. 29 References Etang Alvin and Clarence Tsimpo. 2016. Beyond Income Poverty: Non-Monetary Dimensions of Poverty in Uganda. Background paper for the Uganda Poverty Assessment. Fox Jonathan A. 2015. Social Accountability: What Does the Evidence Really Say?. World Development Vol. 72, pp. 346–361, 2015. Hanushek, Eric and Ludger Woessman. 2012. Do better schools lead to more growth? Cognitive skills, economic outcomes and causation” Journal of Economic Growth. Johnson, D., A. J. C. Cunningham, R. Dowling. 2012. Professional Teaching Competence across 17 African Countries – A Meta Analysis of Published Governmental Expectations. New York: World Bank. Joshia Anuradha and Peter P. Houtzagerb, 2012, “Widgets or Watchdogs?”, Public Management Review, Volume 14, Issue 2, 2012. Mulindwa. I. N., and Jeffery H. Marshall. 2013. Improving Learning in Uganda, Volume 2: Problematic Curriculum Areas and Teacher Effectiveness: Insights from National Assessments. World Bank Studies. Washington, DC: World Bank. Ssewanyana, N. S., A. J. Okidi, D. Angemi, and V. Barung. 2004. Understanding the Determinants of Income Inequality in Uganda, Working Paper 29, Oxford: CSAE, University of Oxford. Keefer Philip and Stuti Khemani. 2014. Radio’s Impact on Preferences for Patronage Benefits. WB-WPS No: 6932 Tsimpo C. and Q. Wodon. 2015. Coverage of Early Childhood Development Interventions in Uganda. Service Delivery and Poverty Studies, Volume 5, World Bank Processing. Tsimpo C. and Q. Wodon. 2015b. Availability of Services in Uganda: Education, Healthcare, and Infrastructure. Service Delivery and Poverty Studies, Volume 4, World Bank Processing. World Bank. 2013. Education and Health Services in Uganda: Data for Results and Accountability. 30