Country Development Diagnostics Post-2015: Uganda

Country Development Diagnostics Post-2015: Uganda

Susanna Gable Hans Lofgren Israel Osorio-Rodarte DECPG, World Bank*

October 31, 2014

*

This paper was completed on October 31, 2014. It is a background paper for subsequent work that resulted in the volume Gable, Susanna; Lofgren, Hans; Osorio Rodarte, Israel. 2015. Trajectories for Sustainable Development Goals: Framework and Country Applications. World Bank, Washington, DC (https://openknowledge.worldbank.org/handle/10986/23122. We thank Mahmoud Mohieldin, Marilou Uy, and Jos Verbeek for overall guidance in this project, and Hans Timmer, Elena Ianchovichina, and Punam Chuhan for their valuable suggestions as peer reviewers. We are also grateful for comments from Lily Chu, Anton Dobronogov, Eric Feyen, Marcelo Giugale, Gloria Grandolini, Raj Nallari, Alberto Portugal, Sajjad Shah, Chris Thomas, and Debrework Zewdie. 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.

Table of Contents

TABLE OF CONTENTS

2

1

INTRODUCTION

3

2

BACKGROUND ON ECONOMIC PERFORMANCE AND STRUCTURE

4

3

STEP 1: SDGS RECENT DEVELOPMENTS AND CROSS-COUNTRY COMPARISONS

9

3.1 SDG TARGET INDICATORS: POVERTY AND SHARED PROSPERITY 3.2 SDG TARGET INDICATORS: INFRASTRUCTURE AND PRODUCTIVE CAPACITY 3.3 SDG TARGET INDICATORS: EDUCATION PRE-PRIMARY EDUCATION PRIMARY EDUCATION SECONDARY EDUCATION GENDER IN EDUCATION 3.4 SDG TARGET INDICATORS: HEALTH 3.5 SDG TARGET INDICATORS: CO2 EMISSIONS

9 12 19 20 21 25 27 29 32

4

STEP 2: SDG BUSINESS-AS-USUAL PROJECTIONS

34

AGGREGATE GROWTH PROJECTIONS PROJECTED BUSINESS-AS-USUAL 2030 SDG VALUES

34 36

4.1 4.2 5

STEP 3: BENCHMARKING DETERMINANTS AND IDENTIFYING SPENDING PRIORITIES 39

5.1 ECONOMIC GROWTH DETERMINANTS INVESTMENT AND SAVINGS HUMAN CAPITAL AND LABOR FOREIGN TRADE BUSINESS CLIMATE SUMMARY OF INSIGHTS FROM GROWTH DIAGNOSTIC 5.2 EDUCATION DETERMINANTS 5.3 HEALTH DETERMINANTS 5.4 IDENTIFYING SPENDING PRIORITIES FOR MORE AMBITIOUS SDG TARGETS

40 41 44 46 49 53 54 61 66

6

STEP 4: IDENTIFYING FISCAL SPACE

68

7

CONCLUSIONS

75

REFERENCES

77

APPENDIX

80

2

1

Introduction1

With the 2015 deadline for the current Millennium Development Goals (MDGs) drawing near, the global community is shaping a new set of international development goals for the longer term. The process involved consultations led by the UN Open Working Group guided by the 2013 report, “A New Global Partnership” of the UN High-level Panel (HLP). The work so far indicates that the post-2015 development agenda will encompass goals for social, economic, and environmental sustainability with broader coverage than the current MDGs.2 This paper presents a Uganda county diagnostic of the implications of the emerging global post-2015 agenda, here referred to as the Sustainable Development Goal (SDG) agenda. In setting the post-2015 SDGs, the global community will need to take cognizance of various challenges to implementation and financing at the country level. Against this background, the World Bank Group is developing a framework, with Uganda as the pilot study, to provide an initial understanding of the challenges policymakers will face in implementing key parts of the global SDG agenda in their countries (Gable et al., 2015). The framework tries to address questions such as: Given a country’s current conditions and assuming that it is striving to realize the emerging global agenda, what would be a set of ambitious yet feasible numerical development targets for it to achieve by 2030? What policy changes might the country’s government consider in order to accelerate progress? How could it create the fiscal space needed to pursue these policies? Applications of the framework provide an input to (or starting point for) more in-depth and policy-oriented country-specific analyses (which may be broad, or focused on a specific area of the agenda) with a critical role for country experts. While such analyses would benefit from the cross-country perspective that our framework offers, it needs to draw on additional countryspecific data and insights in order to offer a more detailed policy analysis. At the core of the diagnostic framework is benchmarking through simple cross-country regressions that draw on an extensive database of variables relevant to the post-2015 agenda; the database covers all lowand middle-income countries (subject to data availability). The framework strives to incorporate what has been learned from a large body of research on issues that underpin the post-2015 agenda. One lesson is that GNI per capita is related to many of the relevant indicators, both SDG target indicators and their “determinants”, i.e., indicators that may explain outcomes for SDG target indicators. For both indicator categories, the analysis is guided by cross-country regression of each variable on GNI per capita, permitting assessments of whether a country is under- or over-performing relative to its GNI per capita. When the relationship is tight, GNI per capita is treated as a summary measure of country capacity to achieve SDG outcomes and provide the inputs needed (given typical government prioritizations in contexts where governments invariably pursue multiple objectives).3 This does not translate into an assumption of GNI being a direct determinant of outcomes – it is merely a benchmark and starting point for discussions about how a country performs compared to a typical country at its income level. Hence, country performance is discussed with reference to “determinants”, representing policies and other 1

This paper is part of collaborative work on the Post-2015 Global Agenda involving DECPG and the Office of the World Bank Group President’s Special Envoy on MDGs and Financial Development, led by Mahmoud Mohieldin. 2 According to the HLP, the overall goals are to: end poverty; empower girls and women and achieve gender equality; provide quality education and lifelong learning; ensure healthy lives; ensure food security and good nutrition; achieve universal access to water and sanitation; secure sustainable energy; create jobs, sustainable livelihoods and equitable growth; manage natural resource assets sustainably; ensure good governance and effective institutions; ensure stable and peaceful societies; and create a global enabling environment and catalyze long-term finance. The Open Working Group is currently discussing a set of 17 development goals. 3 Technically, the underlying equation is Y = aXe where a = the constant, Y = an SDG indicator or determinant, X = GNI per capita, and e = elasticity. To generate linearity, the regressions are done in (natural) logarithmic form—i.e., ln Y = ln a + e ln X.

3

determining factors.4 The framework is designed to be flexible enough for replication across countries that vary strongly in initial conditions and policy priorities. The framework consists of four steps, further explained within their respective sections:  Step One benchmarks the current level of progress for each SDG for the country being analyzed relative to other countries, given GNI per capita, a variable that is highly correlated with most development indicators, including SDGs and their determinants. Accordingly, in this analysis, GNI per capita is treated as a summary indicator of the capacity of a country to achieve outcomes, for both SDGs and their determinants.  Step Two projects the country’s business-as-usual (BAU) GNI per capita and values for SDGs by 2030.  Step Three turns to the determinants of SDG outcomes—many of these are related to policies, including those that affect the efficiency and levels of public spending—pointing to ways of achieving outcomes that are more ambitious than those of the BAU projections. Policies may influence an SDG directly—health services may promote better outcomes for health SDGs—or indirectly, such as when measures that promote growth in household incomes per capita or increased access to sanitation have an indirect positive influence on health SDGs. In this step, therefore, we benchmark Uganda’s current levels of SDG determinants in relation to its GNI per capita and discuss potential changes in policy and spending in priority areas.  Step Four discusses ways to expand fiscal space for priority spending, through additional domestic or foreign financing (including taxes and foreign aid) and efficiency gains (achieved by reallocating spending from areas of lower priority and/or reducing spending in areas with technical efficiency gains without any service reduction). This analysis is applied to the specific case of Uganda: how and to what extent may it be able to create room for increased public spending in priority areas? Would such adjustments be advisable? What trade-offs may be involved? The paper is structured as follows: Section 2 consists of a brief review of Uganda’s economic development Section 3, 4, 5 and 6 applies Step One, Two, Three and Four respectively, for the case of Uganda; and Section 7 concludes.

2

Background on Economic Performance and Structure

As context for the SDG analysis in later sections, this section provides a brief background on Uganda’s economy, covering recent performance and basic data on the country’s current economic structure. Uganda has enjoyed high and sustained rates of growth in recent years, with GDP rising at 6.8 percent on average and per capita growth averaging 3.3 percent between 1990 and 2012, with the gap reflecting relatively rapid population growth. This is likely due to a combination of pro-market policies, successful institution building, and recovery from a low base (a “catch up effect”) after a period of political instability and violence.5 As noted by Hausmann et al. (2014), these reforms not only accelerated but also sustained Uganda’s growth. However, despite strong and sustained growth, Uganda still has very low levels of GDP and GNI per capita: US$547 and US$440 respectively in 2012 (Figure 2.1). In the context of the

4

In other words, the framework views policies and other, often country-specific, factors as “proximate drivers” that offer more direct explanations for country performance. Differences in levels for proximate drivers permit an understanding of why countries deviate from levels expected in relation to GNI per capita. 5 Most post-conflict rebounds run their course within six years (Collier and Hoeffler, 2001). Also, most growth episodes tend to expire within eight years or so, because they run-up against new binding constraints (Hausmann et al., 2005).

4

international financial crisis and commodity price volatility, GDP growth has slowed down since 2009 and in 2012 it was only 3.4 percent, which implied no increase in income per capita.

Figure 2.1: GNI per capita, 1982-2012 (constant 2005 US$)

Source: WDI, World Bank

Changes in the value-added shares for aggregate sectors have followed the standard pattern associated with growth: a declining role for agriculture, which was as high as 70 percent in the early 1980s, and stronger roles for industry and services (Figure 2.2, Panel 1). In 2011, Uganda’s shares—23.4 percent for agriculture, 25.4 for industry, and 51.2 for services—were roughly as expected relative to Uganda’s GNI per capita (for agriculture and services within the 95 percent confidence interval, but somewhat higher for industry).6 Figure 2.2, Panels 2-4 show expected values for the GDP shares for agriculture, industry, and services, derived from cross-country regressions covering all low- and middle-income countries with data for the most recent year (2011 for Uganda), with the shaded area indicating the 95 percent confidence interval for the prediction.7

6

Note that, in these and later graphs, income per capita (GNI per capita) is presented in logarithmic scale In accordance with the criteria described later in this paper, only the relationship between the share of agriculture and GNI per capita is considered here as “tight”. These and all other regressions in this paper are based on the most recent data available; countries are excluded if data is not available for 2005 or a more recent year. 7

5

Figure 2.2: Sectoral Value Added (% of GDP) Panel 1 (upper left): Uganda Agriculture, Industry, and Services values added (% of GDP), 1960-2011 Panel 2 (upper right): Agriculture value added (% of GDP) vs. GNI per capita Panel 3 (lower left): Industry value added (% of GDP) vs. GNI per capita Panel 4 (lower right): Services value added (% of GDP) vs. GNI per capita

.


In Uganda in 2009, 65.6 percent of those employed worked in agriculture, 6.0 percent in industry and 28.4 percent in services (Figure 2.3, Panel 1). Hence, in spite of having the smallest GDP share, agriculture in Uganda has the largest employment share, indicating that the sector also has the lowest value-added per worker. However, the employment shares by aggregate sector are—like the GDP shares—as expected for both agriculture and services (Figure 2.3: Panels 2-4). The fact that the share of employment in industry is lower than expected despite the somewhat higher-than-expected GDP share of the sector suggests that

6

labor in this sector has a relatively high value added per worker in comparison to similar countries; this has to be due to relatively high capital intensity and/or relatively high labor productivity. Figure 2.3: Employment Profiles by Sector Panel 1 (upper left): Uganda Shares of Total Employment by Sector, 2002-2009 Panel 2 (upper right): Employment in Agriculture (% of total) vs. GNI per capita Panel 3 (lower left): Employment in Industry (% of total) vs. GNI per capita Panel 4 (lower right): Employment in Services (% of total) vs. GNI per capita


Hence, the changing sectoral distribution of GDP has not been matched by a similar change in the distribution of the labor force, which is not atypical for a country at Uganda’s income level. Van Waegenberge and Bargawi (2011) report that self-employment and unpaid family members are increasing their shares of employment in agriculture and relate this to slow growth in job opportunities in other

7

sectors. Van Waegenberge and Bargawi (2011) further note that although the industrial sector (including mining, construction, and manufacturing) has seen an increase in its share of GDP, this has occurred mainly through growth in construction in contrast to a rather stagnant share for manufacturing. Uganda’s National Development Plan for 2010-2015 reports that, for the economy as a whole, value added per worker in Uganda is 68 percent, 96 percent, and 28 percent below the levels for India, China, and Tanzania respectively (Republic of Uganda, 2010). As the agricultural sector suffers from particularly low and worsening productivity levels, the National Development Plan emphasizes that “the agriculture sector therefore requires a strong stimulus if it is to absorb the increasingly large number of the labor force. Alternatively, other sectors of the economy (industry and services) will have to expand significantly in order to create opportunities for labor migration from the agricultural sector” (Republic of Uganda, 2010). Figure 2.4 shows the evolution of GDP shares for domestic final demands, exports, and imports. During the 1990s, the main changes were that the dominating GDP share of private consumption decreased while the shares of exports, imports, and private investment increased. The shares of public investment and consumptions have been relatively constant. Private consumption as a share of GDP was as expected in Uganda at 74.9 percent in 2011, while public consumption was significantly lower than expected—11.3 percent in 2011 rather than the expected 13.3 percent (Figure 2.5, Panels 1 and 2). The investment and trade aggregates are analyzed further in Section 3.1. Figure 2.4: Shares of Aggregated Demand, 1990-2011


8

Figure 2.5: Public and Private Consumption Panel 1 (left): Private Consumption (% of GDP) vs. GNI per capita Panel 2 (right): Public Consumption (% of GDP) vs. GNI per capita


3

Step 1: SDGs Recent Developments and Cross-Country Comparisons

In this section we trace the evolution of selected SDG indicators for Uganda since 1990, comparing the most recent data to expected values. Cross-country constant-elasticity regressions are used to assess whether a country is over- or under-performing for an SDG relative to its GNI per capita.8 Hence, for individual countries, deviations from predicted SDG values may be viewed as an indication of how well a country does relative to its capacity to achieve outcomes and provide inputs (determinants). We begin with indicators related to inclusive growth—poverty, the income share of the bottom 40 percent of population, the Gini coefficient, and malnutrition (Section 3.1). We then move on to infrastructure, specifically access to water, sanitation, electricity, roads, and information and communications technology (ICT), involving the Internet and mobile phone technology (Section 3.2). Education (preprimary, primary, and lower?? secondary, including gender issues) is covered in Section 3.3. Health (specifically rates of under-five mortality, maternal mortality, and indicators related to HIV, tuberculosis, and malaria) takes up Section 3.4, while Section 5 looks at CO2 emissions.

3.1

SDG Target Indicators: Poverty and Shared Prosperity

Elimination of extreme poverty (with US$1.25 per day as the poverty line) by 2030 is one of the twin goals of the World Bank Group. The second of these goals, shared prosperity, may be interpreted as requiring positive and above-average per-capita income growth for the bottom 40 percent. The indicators in crosscountry databases that are most directly related to the shared prosperity goal are the income share of the bottom 40 percent and the Gini coefficient. In a setting with positive per-capita private consumption 8

These simplified regressions are useful for current purposes (benchmarking and projections). However, they do not claim to sort out interactions between different indicators, a difficult task given high degrees of correlation, lagged effects, complex timeand space-specific relationships, and data limitations.

9

growth, a growing income share of the bottom 40 percent would almost invariably mean that the shared prosperity objective is satisfied. Moreover, if the income share of the bottom 40 percent is increasing, the Gini coefficient typically would be decreasing. During Uganda’s long period of sustained growth, the extreme poverty rate declined substantially, from 70 percent in 1992 to 38 percent in 2009 (Figure 3.1, Panel 1). Uganda’s current extreme poverty is high but within the expected range (Figure 3.1, Panel 2). Poverty computed on the basis of the national poverty line also declined drastically, with the strongest declines in urban areas. The share of population living below the national poverty line declined from 56 percent in 1992 to 24 percent in 2009. Poverty headcount in urban areas was reduced by two-thirds — from 29 to 9 percent — while the decline in rural areas, from 60 percent to 27 percent, was larger in percentage points. With this measure the poverty rate is significantly lower in both urban and rural areas (See Figure A.1 in Appendix). Figure 3.1: Poverty Panel 1 (left): Poverty Headcount rate, 1989-2009 Panel 2 (right): Poverty Headcount rate vs. GNI per capita


The strong poverty decline has coincided with a virtually unchanged Gini coefficient, at around 44 (Figure 3.2, Panel 1), suggesting that poverty reduction has been driven by growth in the average per capita levels and leaving Uganda with a level of inequality that is well above the average for low- and middle-income countries. The relationship between the Gini coefficient and GNI per capita is very loose and slightly increasing, suggesting that inequality is a reflection of country-specific factors and that it should not be expected to respond strongly or systematically to changes in income per capita. Similarly, only minimal changes were recorded between 1989 and 2009 for the income shares of each of the bottom two quintiles and their sum, the bottom 40 percent (Figure 3.3, Panel 1), leaving the latter share below the expected level (15.5 percent rather than 17.2 percent; Figure 3.3, Panel 2), a finding that is consistent with the higher than average Gini coefficient. The relationship between GNI per capita and the share for the bottom 40 percent is also very loose but inverse (as expected, given the inverse Gini-GNI per capita relationship).

10

Figure 3.2: Inequality Panel 1 (left): Gini, 1989-2009 Panel 2 (right): Gini vs. GNI per capita


Figure 3.3: Income Shares of Lower Quintiles Panel 1 (left): Income Shares of Lower Quintiles, 1989-2009 Panel 2 (right): Income Share of Bottom 40% vs. GNI per capita


The malnutrition indicator adopted here is the under-five underweight rate (measured by data on weight for age). Reduced malnutrition is not only an SDG in its own right but is also important for determining

11

positive effects on health and education, including the ability to learn. As with the poverty rate, the rate of malnutrition declined strongly during the recent decades of strong growth (Figure 3.4, Panel 1). The fact that poverty and malnutrition have declined is not surprising given a high correlation between the two, according to cross-country data for low- and middle-income countries.9 Currently, Uganda’s malnutrition rate is lower than expected (Figure 3.4, Panel 2): 14.1 percent of children under five years of age are underweight, while the expected number is 19.6 percent for a country with Uganda’s income per capita.10 Figure 3.4: Child Malnutrition Panel 1 (left): Malnutrition Prevalence, weight, (% of children under five years), 1988-2011. Panel 2 (right): Malnutrition Prevalence, weight, (% of children under five years) versus Income per capita.


3.2

SDG Target Indicators: Infrastructure and Productive Capacity

The overriding ambition for SDGs in infrastructure is to achieve universal access to water, sanitation, electricity, roads, and ICT. Success in this respect may contribute significantly to growth and help bend the above-mentioned curve that links higher per-capita incomes with a higher degree of inequality. In general, progress in these infrastructure areas is important because they provide inputs to production across the economy and may help also in achieving other SDGs. Water and sanitation access is important in its own right, but also because it contributes to improved health outcomes. Since 1990, the share of Uganda’s population with access to an improved source of drinking water has increased strongly—from 40 percent to 75 percent in 2011—while the improvement

9

The correlation coefficient between the two variables is 0.60 in non-log form and 0.72 in log form. The under-five underweight rate is defined as the percentage of children under the age of five years whose weight for age is more than two standard deviations below the median for the international reference population ages 0-59 months (WDI). 10

12

in sanitation access was moderate, from 27 percent to 35 percent in 2011 (Figure 3.5, Panel 1).11 In 2011, Uganda’s access shares were both slightly above expected levels (Figure 3.5, Panels 2 and 3). Challenges remains though in Uganda, not least for sanitation facilities in urban areas where Uganda is under performing compared to other countries (see Figure A.2 in Appendix) Figure 3.5: Improved Access to Water and Sanitation Panel 1 (upper left): Improved Water and Sanitation Sources, 1990 – 2011 Panel 2 (upper right): Improved Water Source vs. GNI per capita Panel 3 (lower left): Improved Sanitation Source vs. GNI per capita


11

An improved drinking water source is defined as one that is protected from outside contamination, in particular from contamination with fecal matter. An improved sanitation facility is one that hygienically separates human excreta from human contact (WDI).

13

Access to electricity is important in its own right, but is crucial also to many forms of production and SDG outcomes in areas such as health and education. The share of Uganda’s population with electricity access is low, just 14.5 percent in 2010 which, although an improvement over previous years, is still far below what is expected (Figure 3.6, Panel 1).12 As in most other countries, Uganda’s access to electricity is higher in urban than in rural areas; however, Uganda’s electrification is lagging behind other countries, especially in rural areas (Figure 3.6, Panels 2 and 3). Firms rely on generators to self-supply as much as 30 percent of their power needs, but still lose 10 percent of their sales due to inadequate power supply. While power outages are common, they are at the expected level (Figure 3.6, Panels 4). The high price of diesel (see below) makes the cost of self-generation, at US$0.46-US$1.44 per kilowatt-hour (kWh), two to six times more expensive than grid-based electricity (Ranganathan and Foster, 2012).

12

Data on electricity access are provided by the IEA (International Energy Association). The access indicator refers to the population share with access to electricity in their homes. While this definition leaves out access to production sectors, an indicator based on a broader definition would paint a similar picture.

14

Figure 3.6: Electrification Panel 1 (upper left): Access to Electricity Panel 2 (upper right): Rural Electrification Panel 3 (lower left): Urban Electrification Panel 4 (lower right): Power Outages in Firms

Source: EIA data, WDI, World Bank.

Uganda’s electricity production of 70.8 kWh per 1,000,000 people in 2010 covered power consumption of 64.5 kWh (Figure 3.7, Panels 1 and 2). However, these levels are both lower than expected (104 kWh for consumption and 103 kWh for production) compared to other countries at the same income per capita level. Since a drought-induced power supply crisis in the mid-2000s, Uganda has made considerable efforts to increase supply, in the process becoming less reliant on hydropower. Major power sector reforms took place, shifting the sector to a more market-based system. Generation capacity increased

15

from 280 megawatts (MW) to 568 MW between 2005 and 2009 (Ranganathan and Foster, 2012). However, this does not take account of the Bujagali hydropower project, completed in 2012. Although Uganda had moved to a more balanced hydro-thermal system in 2009 (EIA 2009), the addition of the Bujagali power station once again skewed the portfolio toward hydropower. As discussed in Ranganathan and Foster (2012), historic total production power costs in Uganda, at US$0.24 per kWh, are among the highest in Africa. In 2010, domestic tariffs covered only about two-thirds of the costs of power production. System losses at 35-40 percent in the late 2000s are also well above the African low-income benchmark of 24 percent.13 Hence, despite the reforms and very low electrification rates, the power sector continues to struggle with underpricing, high distribution losses, and high hydropower dependency. Figure 3.7: Electric Power Panel 1 (left): Electric Power Consumption (kWh per capita) Panel 2 (right): Electricity Production (kWh per capita)

Source: EIA data

Roads are important because they reduce the costs of goods and services (including SDG-related services), and offer producers more profitable access to markets beyond their immediate vicinities. Roads are similar to electricity in that, if financial resource constraints are overcome, improvements in relevant indicators should be relatively uncomplicated compared to, for example, improvements in health outcomes. Uganda’s national road network, which faces high traffic volume, has an adequate density and is in a fairly good condition.14 More specifically, the classified road density is four times the density of an average low-income African country (Ranganathan and Foster, 2012). Uganda’s road density, of 32.2 km per square km, is also substantially higher than the expected 8.9 km (Figure 3.8: Panel 1) while the number of motor vehicles per capita in Uganda is as expected (Figure 3.8: Panel 2). Despite the high road density, the paved roads carry twice as much traffic as those of an average low-income African country, and as much as Africa’s middle-income countries (Ranganathan and Foster 2012). This is not surprising since the 13

Revenue collection is the one area in which financial performance has clearly improved. In 2005, collection of power bills was around 84 percent; by 2009, this figure had increased significantly, to 94 percent, slightly above the average for low-income countries in Africa. 14 Cross-country data on roads, including updated data for Uganda is limited, but this section can instead base its conclusions on an infrastructure country diagnostic (Ranganathan and Foster, 2012).

16

population density is higher than expected (see Figure 4.11) with the number of motor vehicles as expected. Moreover, around 87 percent of Ugandan roads are in good or fair condition compared to only 72 percent in a typical African low-income country. However, in the latter category, Uganda’s share of roads in good condition is relatively low. Figure 3.8: The capacity of the current road network Panel 1 (left): Road Density vs. GNI per capita in Uganda Panel 2 (right): Motor Vehicles per capita vs. GNI per capita in Uganda


In order to diminish the overall pressure on the road transport system, it is important to consider alternative ways of transportation, especially for land-locked countries like Uganda. Figure 3.9 shows that within both air transport (carrier departures per capita) and rail transport (density of rail), Uganda is below what is expected. Air carrier departures from Uganda are 0.22 per 1,000 people, rather than the expected 0.33, and rail density is 0.0013 km per square km rather than the expected 0.0031. Figure 3.9: Air and rail transport Panel 1 (left): Air transport carrier departures per capita vs. GNI per capita Panel 2 (right): Length of rail lines per square km vs. GNI per capita

17

Source: WDI, World Bank.

Moreover, from a more disaggregated perspective, Uganda’s road network may not be geared toward facilitating shared prosperity. World Bank (2009) found that 54 percent of the district roads were in poor condition in 2008. District roads connect rural areas to markets, health centers, educational institutions, administrative centers, and other services. Moreover, despite the relatively high average road density in Uganda, the share of the population living within 2 km of an all-season road is not higher than in other low-income African countries (Ranganathan and Foster, 2012). Furthermore, while over half of the rural roads in other low-income African countries are in good or fair condition, barely 40 percent of the rural roads in Uganda qualify for this category. Finally, the road fatality rate has increased to 81 per 100,000, double the African average (World Bank, 2009). In sum, providing adequate resources for road maintenance remains a challenge, and further investment is needed to increase rural connectivity and improve road safety. In recent years, ICT has emerged as a force that, in multiple ways, promises to contribute to accelerated growth and SDG progress. Available data shows that use of the Internet and mobile phone subscriptions have proliferated greatly in Uganda in recent years, although fixed broadband subscriptions are not so common (Figure 3.10, Panel 1). From a cross-country perspective, Internet use and fixed broadband subscriptions are both higher than expected while mobile phone subscriptions are roughly as expected (Figure 3.10, Panels 2-4). Internet use in Uganda was 14.7 percent compared to the expected 5.3 percent, fixed broadband 0.106 compared to 0.06 percent, and mobile phone use 45.0 compared to 50.1 percent.

18

Figure 3.10: Information and Communications Technology Use and Access, 1995-2012 Panel 1 (upper left): ICT Users (per 100 people) Panel 2 (upper right): Broadband and Internet Subscriptions (per 100 people) vs. GNI per capita Panel 3 (lower left): Internet Use (per 100 people) vs. GNI per capita Panel 4 (lower right): Mobile Telephone Subscriptions (per 100 people) vs Income per capita


3.3

SDG Target Indicators: Education

Gross educational enrollment rates have risen in Uganda since 1970 but remain generally lower for higher levels of education (Figure 3.11), with the exception of pre-primary education, which started to expand

19

only recently.15 The main event in the educational system is extremely rapid expansion in primary education in the mid-1990s, bringing about gross enrollment rates well above 100 percent, showing that a substantial share of the students were not age-appropriate (in most cases, too old). Such high gross enrollment rates are not uncommon when primary access expands rapidly. The fact that Uganda’s primary gross enrollment rates now are declining indicates that older out-of-age children have completed the primary level; if in addition repetition rates decline (as they have since 2008—see Figure 3.13, Panel 4), students stay in school for fewer years, reinforcing this enrollment decline. The increases in secondary enrollment are also substantial, due to the links to primary education occurring with a lag. The expansion in tertiary education is much more modest, especially in terms of absolute numbers. In sum, it is only at the primary level that Uganda is now approaching universal access; currently, it is also the only level that is compulsory. The education-related SDGs on which we focus are related mainly to the pre-primary, primary, and secondary levels. Figure 3.11: Gross Enrollments in Schools, 1970-2013

Source: EdStats, World Bank

Pre-Primary Education Uganda’s pre-primary education is currently a three-year program for 3-5 year-olds with stated targets for increasing access and completion rates. Pre-primary education is only privately provided, with data available for gross and net enrollment rates. With the recent increase in pre-primary gross enrollment, Uganda is now at an expected level, even in terms of net enrollment (Figure 3.12, Panels 1 and 2).

15

The rate of gross enrollment at a level is defined as total enrollment for the level, irrespective of age, expressed as percentage of the total population in the official age group for the level.

20

Figure 3.12: Pre-Primary School Enrollment Panel 1 – Gross enrollment in pre-Primary School vs. GNI per capita Panel 2 – Net enrollment in pre-Primary School vs. GNI per capita


Primary Education At the primary level—seven years of schooling for 6-12 year olds—the SDG indicators should measure the extent to which the system is able to ensure that every child completes the level while meeting minimum learning standards for reading, writing, and counting. Unfortunately, the developing country coverage of the Program for International Student Assessment (PISA) has no data available for Uganda. Given these data constraints, therefore, we have based our analysis largely on indicators related to enrollment and the ability of the system to ensure that enrolled students complete primary education in a reasonably timely manner—i.e., without a large proportion of the students failing to enter primary school at the targeted age and/or repeating grades). The education system appears to be well on its way to providing universal and timely access to primary schooling (Figure 3.13: Panels 1 and 2). Cross-country data indicate that the relationship between gross enrolment and GNI is very loose but for net enrollment it is moderately tight. In Uganda, gross enrollment is close to average, but net enrollment is higher than expected; 90.8 percent of the primary age group was enrolled (i.e., only 9 percent were out of school) in 2011 compared to 78.3 percent that would be typically expected for a country at that income level. This is partly due to higher-than-average female enrollment (see further below). Similarly, Uganda’s net intake rate for the first grade of primary schooling was higher than expected, although the numbers have been declining since 2008 (Figure 3.13: Panels 3 and 4). This partly explains the declining primary gross enrollment rates in Uganda (in Figure 3.11), as well as the declining rate of repeaters (see below).

21

Figure 3.13: Primary School Enrolment and Intake Rates Panel 1 (upper left): Gross Enrollment (%) vs. GNI per capita Panel 2 (upper right): Net Enrollment (%) vs. GNI per capita Panel 3 (lower left): Net Intake for Grade 1 (% of school-age children) vs. GNI per capita Panel 4 (lower right): Net Intake Rate for Grade 1, 2002-2011


Enrollment is a prerequisite for the more ambitious SDG aim of ensuring completion. Uganda’s gross primary completion rate in 2011, 53.1 percent, is significantly lower than the expected 68.8 percent (Figure 3.14, Panel 1).16 This is mainly due to a very high primary school drop-out rate of 75.2 percent that

16

The gross primary completion rate measures the number of students completing the last grade of primary school as a share the population at the official completion age. More specifically, it is defined as the total number of students in the last grade of

22

is substantially higher than the expected 28.4 percent (Figure 3.14, Panel 2).17 Furthermore, the primary completion rate has been falling since the beginning of the 2000s (except for a temporary rise in the mid2000s) while primary dropout rates increased substantially. Although Uganda managed to reduce the dropout rate in the mid-2000s, it started increasing again from year 2008 (Figure 3.14, Panel 3). Completion should preferably be timely—i.e., with most of the population in the targeted age group (12 years old), completing the last grade at the age of 12 years. This would require a net intake rate for Grade 1 of primary school at near 100 percent as well as near-zero rates for both repetition and dropping out. A recent decline in Uganda’s repetition rate suggests that the country is moving in the right direction, notwithstanding an increase in the mid-2000s (Figure 3.15). However, the current dropout level (10.2 percent in 2011) is still too high to ensure an acceptable proportion of timely primary school completion. Nevertheless, from a cross-country perspective, Uganda is at the expected level relative to its income per capita. Uganda, therefore, is somewhat successful relative to comparator countries in enrolling primary students but relatively unsuccessful in bringing them to full and timely primary completion.

primary school, minus the number of repeaters in that grade, divided by the total number of children of official graduation age (WDI). 17 The dropout rate in primary education is the share of those who enroll in the first grade who, given current dropout rates by grade, would fail to reach the final (7th) grade of the primary cycle. The dropout rate equals 100 minus the survival rate to the last grade of primary education—the higher-than-expected dropout rate is consistent with findings from cross-country regression for the survival rate to the last grade. For Uganda, this rate is lower than expected, confirming the observation that the system is relatively unsuccessful in keeping primary students in school.

23

Figure 3.14: Primary School Completion Panel 1 (upper left): Primary Completion Rate vs. GNI per capita Panel 2 (upper right): Primary Dropout Rates vs. GNI per capita Panel 3 (lower left): Primary Completion and Dropout Rates, 2000-2011


24

Figure 3.15: Primary School Repetition Panel 1 (left): Repeaters vs. GNI per capita Panel 2 (right): Repetition rates, 2001-2011 (% total enrollment)


Secondary Education As Figure 3.11 showed, secondary gross enrollment has increased but is far below the corresponding numbers for the primary level. Given that Uganda and many other low-income countries at this time are approaching universal access at the primary level, looking at the next educational level is a logical next step. Figure 3.16, Panel 1 shows that Uganda’s gross enrollment in secondary of 27.6 percent is still substantially lower than the expected 47.5 percent. One reason this relatively modest rate is the low completion rate for primary education, leaving relatively few students eligible for secondary education. Another reason is a low progression rate from primary to secondary school: the 2011 rate of 58 percent is much lower than expected (Figure 3.16, Panel 2).18 Nevertheless, this rate has increased—in 1995 it was only 34 percent.19 There is no data on net enrollment for lower secondary school in Uganda, but the out-of-school rate (i.e., the inverse of net enrollment) is as expected, at 23 percent—suggesting that net enrollment is performing better than gross enrollment (Figure 3.16, Panel 3).20 This is suggested also by the relatively low level of repeaters compared to other countries (see below).

18

Progression to secondary school refers to the number of new entrants to the first grade of secondary school in a given year expressed as a percentage of students enrolled in the final grade of primary school the previous year (WDI). 19 Students dropping out from lower secondary itself would also add to the high lower secondary out-of-school rate; however, dropout rates are not available for lower secondary. 20 The out-of-school rate for children of lower secondary school age is defined as the number of children of official lower secondary school age who are not enrolled in lower secondary school expressed as a percentage of the population of official lower secondary school age (EdStats).

25

Figure 3.16: Secondary Education Panel 1 (upper left): Gross Enrollment in Secondary vs. GNI per capita Panel 2 (upper right): Progression to Secondary school vs. GNI per capita Panel 3 (lower left): Out-of-School Rate for Lower Secondary vs. GNI per capita


Despite low gross enrollment rates, the completion rate for secondary school is as expected (Figure 3.17, Panel 1).21 Note however that the proportion of repeaters in secondary education –2.0 percent – is lower than the expected 5.7 percent; i.e., compared to other countries, Uganda’s secondary enrollment rate is less inflated by this inefficiency. This is consistent with a relatively high net (age-appropriate) enrollment rate (Figure 3.17, Panel 2).

21

Uganda’s secondary completion rate is highly uncertain. Drawing on population, enrollment, and repetition data in EdStats, a rate of 9.4 percent was calculated for 2011.

26

To sum up, due to a high dropout rate during primary and a low progression rate to secondary (among those who graduate from primary), Uganda’s secondary gross enrollment rate is relatively low. Among those enrolled, a relatively high proportion is age-appropriate, and the completion rate is as expected, with a low repetition rate. In coming years, one major challenge will be to maintain or improve the functioning of this level (including low repetition and dropout rates) while absorbing a growing number of students. Figure 3.17: Secondary Completion and Repetition Rates Panel 1 (left): Secondary Completion Rate vs. GNI per capita Panel 2 (right): Repeaters (%) in Secondary vs. GNI per capita


Gender in Education Gender equality is a key feature of the post-2015 agenda, especially prominent in discussions led by the UN (2013) which is our main source for this section of the paper. Stronger equality in education may have a strong, long-term influence on equality in the labor market and society at large. To understand the remaining country-specific constraints within education, a gender perspective can be revealing. According to gender parity indices for gross enrollment (defined as the ratio between female and male gross enrollment rates at each level of education, Uganda’s educational system has become considerably more equal in recent decades (Figure 3.18). A gender parity of 0.83, for example, would mean that for every 100 males 83 females are enrolled. It is evident that the ratio has increased dramatically in Uganda for all educational levels with time series data since 1970, gradually and plausibly moving toward unity (i.e., gender equality in enrollment numbers), which can be expected wherever enrollment at a specific level is close to becoming universal. However, the higher the educational level, the lower the ratio. And since more educated citizens are more likely, on balance, to occupy leading positions, this phenomenon suggests that Uganda, in an important respect, is still far from attaining gender equality in the labor market and society at large. Our standard cross-country regressions indicate that, the higher the level of GNI per capita, the higher the gender parity index for enrollment—education enrollment becomes more equal as per-capita incomes rise—and that, the higher the level of education, the tighter the relationship between GNI per capita and this index. The latter observation reflects two patterns: differences in enrollment rates across countries

27

are more pronounced with higher levels of education, and lower overall enrollment rates for secondary and even more so, for tertiary education, are disproportionately related to low female enrollment rates at these higher levels. With regard to Uganda’s GNI per capita, the country has a gender parity index value higher than expected at primary level, as expected at secondary level, but lower than expected at tertiary level (Figure 3.18, Panels 2-4). Figure 3.18: Gender Ratios in Education Panel 1 (upper left): Gender Parity at Primary, Secondary, and Tertiary Levels, 1970-2011 Panel 2 (upper right): Female-Male Ratios in Primary vs. GNI per capita Panel 3 (lower left): Female-Male Ratios in Secondary vs. GNI per capita Panel 4 (lower right): Female-Male Ratios in Tertiary vs. GNI per capita


28

Gender equality in education may also be analyzed with indicators related to completion. In Uganda male and female primary completion rates have converged, mainly due to male completion declining (Figure 3.19, Panel 1). The high dropout rates in Uganda’s primary schools are also gender neutral, with a 75.5 percent rate for girls and a 47.8 percent rate for boys. Gender parity in survival rates also differs between males and females who are enrolled—i.e., independent of initial entry gaps (Figure 3.19, Panel 2 and 3). Compared to GNI per capita, the figures suggest that gender parity in survival rates to the final grade are as expected at the primary level, but lower than expected at the lower secondary level. (Data for the full secondary level is not available). As discussed earlier, though, a low survival rate may indicate simply a low level of repeaters, in which case data for other countries may be inflated by high female repetition rates. On the whole, in terms of total years of schooling and youth literacy rates (see Figure A.3 in appendix) males and females are at similar levels, and Uganda is performing as expected for years of schooling and better than expected for literacy for both gender. Figure 3.19: Gender Parity in Survival through Schooling Levels Panel 1 (left): Gender Parity Index for Survival to Final Grade of Primary vs. GNI per capita Panel 2 (right) Gender Parity Index for Survival to Final Grade of Lower Secondary vs. GNI per capita


3.4

SDG Target Indicators: Health

The SDG indicators we consider in the health area are the under-five and maternal mortality rates (U5MR and MMR) and those for HIV, malaria, and tuberculosis. Since 1990, Uganda’s MMR and U5MR have both declined considerably (Figure 3.20, Panel 1). MMR went from 600 to 440 deaths per 100,000 live births between 1990 and 2011, and U5MR fell from 178.2 to 68.9 deaths per 1,000 live births between 1990 and 2012. In comparison with these rates for countries with comparable GNI per capita (Figure 3.20, Panels 2 and 3), Uganda appears to be at the expected level for both maternal and under-five mortality rates.

29

Figure 3.20: Maternal and Under Five Mortality Rates Panel 1 (upper left): Maternal and Under-5 mortality rates, 1990-2012 Panel 2 (upper right): Maternal mortality rate vs. GNI per capita Panel 3 (lower left): Under-5 mortality rate vs. GNI per capita


Uganda’s HIV prevalence of 7.2 percent is substantially higher than the expected level, which is closer to 1 percent (Figure 3.21, Panel 1). However, given regional specificities—HIV disproportionately afflicts subSaharan Africa—the relationship with income per capita is weak. The regional average in sub-Saharan Africa is 5.2 percent while the median is 2.7 percent. Note, however, that even within this group Uganda is ranked 36th out of 45 countries with data. More limited cross-country HIV data for disaggregated groups within countries indicate a similar pattern: Uganda is an outlier also for children, adults, males, and females. The HIV prevalence dropped significantly from the early 1990s but has started to increase somewhat from the mid-2000s (Figure 3.21, Panel 2).

30

Figure 3.21: HIV Prevalence Panel 1 (left): HIV Prevalence (% of populations aged 15-49 years) vs. GNI per capita Panel 2 (right): Prevalence of HIV (% of population aged 15-49 years), 1990-2012

Source: HNP, World Bank

Malaria cases reported in Uganda have risen dramatically since the 1990s (Figure 3.22, Panel 1). They peaked at 45 percent in 2003 but typically stayed below 5 percent during late 2000s. In 2012 the level was 7.3 percent, higher than expected for Uganda’s income per capita level (Figure 3.22, Panel 2). Figure 3.22: Malaria Panel 1 (left): Malaria Cases Reported (% of population), 1992-2012 Panel 2(right): Malaria Cases Reported vs. GNI per capita


31

Among other diseases, tuberculosis has decreased in Uganda from almost 500 per 100,000 people in 1990 to less than 200 per 100,000 people in 2012; its prevalence is now below the expected value (Figure 3.23, Panels 1 and 2). Figure 3.23: Tuberculosis Panel 1 (upper left): Prevalence of Tuberculosis (% of population), 1990-2012 Panel 2 (upper right): Prevalence of Tuberculosis (% of population) vs. GNI per capita Panel 3 (lower left): Incidence of Tuberculosis (% of population) vs. GNI per capita


3.5

SDG Target Indicators: CO2 Emissions

From the perspective of global climate change, the most important issue may be to significantly reduce total CO2 emissions. In recent decades Uganda emissions have risen, in both per unit of GDP and per capita terms (Figure 3.24, Panel 1). Across countries, the emissions intensity of GDP is positively (but weakly)

32

related to GNI per capita while, as a corollary, there is a very strong relationship (R2 > 0.7) between CO 2 emissions and GNI per capita (Figure 3.24, Panels 2 and 3). Despite Uganda having followed a fairly typical pattern of economic transition vis-à-vis sector shares of GDP, the country has lower-than-expected levels for both indicators relative to its comparator GNI per capita. Compared with other regions of the world, though, there is a huge gap between low-income and high-income countries in CO2 emissions per person; the OECD, for instance, produces 100 times more CO2 per person than Uganda (Figure 3.24, Panel 4). Figure 3.24: Carbon Emissions Panel 1 (upper left): CO2 Emissions by Uganda in kg per GDP and metric tons per capita, 1982-2010 Panel 2 (upper right): CO2 Emissions (kg per US$ of GDP) vs. GNI per capita Panel 3 (lower left): CO2 Emissions (metric tons per capita) vs. GNI per capita Panel 4 (lower right): CO2 Emissions, 2010, Uganda and Selected Regions

Source: EIA and WDI, World Bank

Uganda’s relatively low levels of emission may be explained by the low level of electrification and its mix of energy sources. The country’s reliance on fossil fuels and hydropower in 2010 was as expected for fossil

33

fuels and on the higher end for hydropower (Figure 3.25, Panels 1 and 2). However, as noted in Section 3.2, by 2012 there were major additions in terms of hydropower, shifting the power balance further. The benefits of relying on relatively clean hydropower should, from Uganda’s perspective, be weighed against the increased vulnerability to drought that this entails. Figure 3.25: Sources of Electricity Generation Panel 1 (left): Electricity from Fossil Fuels (% of total) vs. GNI per capita Panel 2 (right): Electricity from Hydropower (% of total) vs. GNI per capita

Source: EIA data

4

Step 2: SDG business-as-usual projections

If the relationship between GNI per capita and an SDG is considered tight enough (which tends to be the case), then the GNI data for the country in question is used, not only to benchmark the initial SDG outcome but also to project business-as-usual SDG outcomes for 2030. A tight or moderately tight relationship refers to a significant GNI per capita variable and a good enough explanatory power of the regression (“tight” R2 > 0.3, “moderately tight” 0.1 < R2 < 0.3). Hence, to start with, we need projections of GNI per capita.

4.1

Aggregate growth projections

Aggregate growth projections covering most countries are produced by various international organizations, including the World Bank, IMF, CEPII, OECD, and IIASA, but also by most governments and other sources such as Hausmann et al. (2011). From the projections, it is difficult to determine which source is most reliable. Given the fact that available sources only project GDP while this paper focus on GNI data, we have to assume, for most countries quite reasonably, that projected GNI growth will not deviate substantially from projected GDP growth (both expressed in constant 2005 US dollars).22 In any 22

As indicated by the names of the terms, GDP is primarily a measure of production while GNI is an income measure, more specifically GNI = GDP plus net receipts from abroad of primary income (compensation of employees and property income). For most countries, the two measures are highly correlated; among low- and middle-income countries, they tend to diverge most

34

country case study, it is good practice to compare different projections and, if necessary, generate refined projections. Figure 4.1 uses three of these sources to show Uganda’s projected (indexed) levels of GDP per capita up to 2030 (and the historical development since 1990), while Table 4.1 presents growth rates. We opted for the CEPII projection, which has a growth rate for GNI per capita of 4.0 percent per year (at constant 2005 US dollars), translating to an increase from US$378 in 2011 to US$817 in 2030 (both at constant 2005 prices), a level similar to the current levels of countries such as Vietnam, India, and Senegal.23 Judging by the other data, an annual per capita growth rate of 4 percent seems realistic, if perhaps erring on the moderately optimistic side. Figure 4.1: Historical Data and Projections for Real GDP per capita (2011=100)

Sources: WDI, IIASA, OECD, and CEPII

strongly in countries where (net) FDI over time has represented a substantial share of total private investment, often in natural resource sectors, generating substantial profit remittances to the foreign investors. If additional information is available on how future GNI and GDP growth may differ for a country, then such information should be reflected in the GNI projections. 23 We chose the projections of CEPII due to a combination of factors, including a transparent model structure, clear documentation, and comprehensive country coverage.

35

Table 4.1: Historical and Projected Growth from Various Sources Source

Time period

Indicator (real values)

Comment

WDI WDI

Average annual growth (%) 3.3 3.2

1990-2012 1990-2011

GDP per capita GNI per capita

Data used in Figure 3 for period up to 2012 GDP per capita growth for 1990-2011 was 3.5 percent

CEPII OECD IIASA IMF (2013)

4.0 3.8 2.5 3.7

2013-2030 2013-2030 2013-2030 2013-2030

GDP per capita GDP per capita GDP per capita GDP per capita

Hausmann et al. (2014) Republic of Uganda 2014, pp. 27, 30, 53

3.3

2009-2020

GDP per capita

5.6

2014-2040

GDP per capita

4.2

Including oil revenues, adjusted for population growth Based on the Economic Complexity Index Calculation based on data for GDP growth and population in Uganda’s Vision 2040

Projected Business-as-Usual 2030 SDG Values

This section presents business as usual (BAU) projections for the SDG subset that is judged to have a tight enough relationship with income per capita. These projections may be seen as representing feasible targets for 2030; hence we refer to them as BAU targets. If Uganda’s current values were the expected values, the projected target for 2030 would simply be based on the SDG-specific income elasticity produced from the regressions and the projected income level for Uganda in 2030, i.e., Uganda would be assumed to follow the fitted line all the way to 2030, reaching the expected SDG level for a country at Uganda’s income per capita level at that time. However, for all SDGs Uganda is, like any other country, under- or over-performing (significantly or not) compared to the expected value. However, as countries make progress, approaching the extreme value for the various indicators (for example a net enrollment rate of 100), the extent to which they deviate from the expected values would tend to decline. To capture this phenomenon, our projections are adjusted to reflect a gradual convergence towards the expected value. It is important to note that these projections are based on typical rates of progress according to crosscountry patterns. Accelerated implementation of policies that, directly or indirectly (for the latter, most importantly by accelerating growth) have a positive impact on SDG indicators would make more ambitious targets feasible. In this process, access to external resources could have a significant positive impact. Issues related to such policies (and other determinants), as well as fiscal space, are discussed in Sections 3 and 4. Table 4.2 presents recent values for each SDG and the 2030 BAU projections when the relationships are considered tight or moderate. Given unpredictable technological changes, no projections are made for Internet use and mobile phone subscriptions. In terms of general interpretations, a weak relationship signals that income does not tend to bring countries to narrow outcome ranges or limit what is possible. However, this does not necessarily mean that it is easy to change the outcome, since it may depend on other determinants that are less prone to change than income – inherited inequalities may be one example. On the other hand, indicators with a tight relationship to GNI per capita may also differ strongly in terms of the extent to which GNI per capita (and the wide range of indicators that are associated with GNI per capita) limits what is possible. For infrastructure indicators that are determined by investments

36

by governments and their partners, if investment financing is available, rapid progress may still be possible without the income growth that is implied by the cross-country relationship. However, the challenges of speeding up progress are likely to be more severe for SDGs that are strongly dependent on interlocking aspects of incomes, education, health, and behavioral patterns involving millions of households. Table 4.2: Current and Projected Values for Selected SDG Indicators SDG GNI per capita (constant 2005 USD) Poverty headcount ratio at $1.25 a day (PPP) (% of population Malnutrition (weight for age: % of children under 5) Income share, bottom 40% (% of total income) GINI index Access to improved sanitation (% of population) Access to improved water (% of population) Access to electricity (% of population) Road density (km road per 100 sq. km of land area) Air transport (carrier departures per 1,000 capita) Rail road network (m per square km of land) Internet use (% of population) Mobile cellular subscriptions (% of population) Net enrollment, preprimary (%) Net enrollment, primary (%) Primary completion rate (%) Gross enrollment, secondary (%) Secondary completion rate (%) Maternal mortality (modeled estimate, per 100,000 live births) Under 5 mortality (per 1,000 live births) Prevalence of HIV total (% of population ages 15-49) Malaria reported Prevalence of tuberculosis CO2 emissions per capita

Recent value 378.1 38.0 14.1 15.5 44.3 33.9 74.8 14.6 32.2 0.22 1.29 14.7 45.0 13.6 90.9 53.1 27.6 9.4 310.0 68.9 7.2 7.3 175 0.11

BAU projection for 2030 817.6 11.5 8.8 44.8 80.7 31.0 35.8 0.77 20.4 66.1 41.6 146.3 42.7 1.3 109 0.39

Note: Green = Currently significantly over-performing; Red = Currently significantly under-performing; Black = Performing as expected; No projection = Too loose relationship with GNI per capita. Whether a specific deviation (positive or negative) reflects a stronger or weaker performance varies across indicators. For example, a positive deviation reflects weaker performance for poverty but stronger performance for water access. The terms over- and under-performance are used normatively; for example, for the maternal mortality rate, a lower rate than expected is reflected as over-performance.

Given a fairly tight relationship with GNI per capita, it is meaningful to project the extreme poverty rate for 2030. It declines to 11.5 percent, a BAU target that represents a substantial decrease from 38 percent in 2009 but is still far above the 3 percent global rate that the WBG has defined as the maximum level consistent with ending extreme poverty.24 The implication is that the government of Uganda, in order to

24 The 3

percent target (as opposed to zero percent) recognizes the fact that some amount of “frictional” poverty, stemming from unexpected fluctuations (often due to conflict or natural disasters), most likely will prevail also in 2030 (cf. World Bank, 2014).

37

eliminate extreme poverty, needs think of a package of measures that accelerates income growth, especially for those who live in the zone of poverty, below or only marginally above the poverty line. With regard to the two indicators related to shared prosperity, the Gini coefficient and the income share of the bottom 40 percent, we noted that, for each of these indicators, the relationship with GNI per capita is very weak, preventing us from projecting BAU values for 2030. Moreover, the estimated relationships reveal a tendency toward less success in prosperity sharing (a higher Gini coefficient and a lower income share for the bottom 40 percent) as GNI per capita increases. This means that, in order for a stronger element of shared prosperity to contribute to the SDG agenda, Uganda and other countries need to consider additional measures in order to “bend the curve” that associates higher per capita incomes with higher inequality.25 The projected access shares for 2030 are 80.7 percent for water and 44.8 percent for sanitation. Given a moderately tight relationship between GNI per capita and each of these two access indicators, this shows that Uganda is unlikely to reach anything near universal access in the absence of targeted efforts that go beyond what has been typical at Uganda’s income level. Note that this is the outcome even though Uganda is currently over-performing in both fields. Uganda’s current level of access to electricity is half of what is typical for a country at its income level. With a tight relationship with income per capita, the projected 2030 electricity access rate is only 31 percent. Given this, approaching universal access would require large efforts. The level of electric production per capita falls short of the expected level by more than 30 percent. The production shortcoming is aggravated by large system losses. Given this, improved access will require both higher production and reduced distribution losses. Fortunately, compared to SDG indicators in areas such as health and education, changes in electricity access depend less on changes in behavior and consumption by millions of households and more on major investments by a small number of public and/or private entities. Given this, rapid progress is a less-insurmountable challenge in this area. Road density in Uganda is several times higher than the expected level and the projected increase is marginal. (However, higher than expected density is not surprising given high population density; challenges for the road sector are discussed in Section 3.) However, both air and train transport infrastructure is lagging. For ICT indicators, given rapid and unforeseeable changes in both technology and costs, it is not meaningful to make projections up to 2030. However, Uganda appears to be positioned to benefit from ICT at least as well as comparator countries with a current level as expected for mobile phone use and an Internet use level that is almost three times larger than expected. If growth proceeds and emissions per unit of GDP both evolve as projected, CO2 emissions in Uganda would increase from 0.11 tons per capita to 0.39 tons per capita. This is still just half of the typical per-

25

It is possible to define changes in income distribution that reduce inequality according to only one of the two indicators (Gini coefficient and the income share of the bottom 40 percent). However, in practice, (changes in) the two indicators are likely to be highly correlated; for the universe of low- and middle-income countries, the correlation between Gini and the bottom 40 percent income share is -0.98. Moreover, if inequality decreases, poverty is also most likely reduced. This means that policies aimed at reducing inequality according to one measure are likely to do so also for other inequality measures. This inequality-poverty link could be reversed if inequality-reducing measures would prejudice growth. However, recent research suggests that, in so far as there is any relationship, it is more likely that less inequality is associated with stronger growth (Ostry et al., 2014).

38

capita emissions for a country in Sub-Saharan Africa today. Nevertheless, as part of the global partnership, Uganda and other countries should strive to keep such increases in check. Uganda is performing as expected in terms of net enrollment in pre-primary school, at 13.6 percent with the projected rate for 2030 at 20.4 percent. In primary education, Uganda’s net enrollment exceeds the expected value by 21.2 percent, with the projected value for 2030 at 92.8 percent (from 90.1 in 2011). For secondary education, only gross enrollment rates are available. Uganda’s current rate, 27.6 percent, falls short of the expected value by as much as 17.6 percentage points. The projected BAU 2030 value is 41.6 percent. In terms of keeping the enrolled students in school and completing each level, Uganda is lagging. The primary school completion rate is 22.8 percent lower than the expected level, and projections suggest a BAU 2030 target of 66.1 percent (compared to the current 53.1 percent). For secondary schooling, completion is as expected. The relationship is too loose though to make relevant projections for 2030. Hence, primary enrolment rates are over-performing while Uganda is facing a major challenge in terms of increasing completion rates faster than other countries in order to catch up with expected completion levels. For secondary school, its challenges are to accelerate increases in enrolment while maintaining completion rates. Uganda is performing as expected in terms of both MMR and U5MR. Given the project increase in GNI per capita, the projected values for MMR and the U5MR are 146.3 and 42.7, respectively. HIV prevalence is high in Uganda (as it is for many countries in sub-Saharan Africa); in order to tackle this scourge, targeted policies are needed. In terms of child malnutrition, Uganda is over-performing a typical country with 28 percent. Given a tight relationship with income per capita, the projected 2030 BAU level is a malnutrition rate of 8.8 percent among children.

5

Step 3: Benchmarking Determinants and Identifying Spending Priorities

In Step 3, we regress SDG determinants on GNI per capita. (In Step 1, we did this for SDG indicators). The identification of determinants is guided by the findings of country and cross-country research, limited to indicators that are available in cross-country databases. We emphasize those determinants that may be influenced by policy in the short to medium terms. The purpose is to assess the feasibility of policy changes that accelerate SDG progress and make more ambitious targets possible. Policies may influence SDGs in two ways: (i) by raising the level of GNI per capita, which in turn, through various channels, affects SDGs, and (ii) by improving country SDG outcomes relative to what is expected given the GNI per capita. The determinants—in our cross-country database represented by over 200 indicators—may be classified according to which of the following four areas they impact: economic growth, education, health, and climate change. In the fifth area that our approach covers—SDGs related to access to infrastructure—the basic approach is simpler: deviations are mainly viewed as indicating insufficient levels of efficient investments. It is important to note that some determinants influence several SDGs, and that SDGs may be determinants of other SDGs.26 Of course, the fact that cross-country analysis has shown that a certain determinant matters for an outcome does not necessarily mean that it is important in a specific country setting; conversely, a lack of evidence on the cross-country level does not necessarily mean a determinant

26

For example, access to electricity is an SDG in its own right and it is also likely to influence both education and health SDGs.

39

is unimportant for a specific country. In order to arrive at more definitive conclusions for a given country, it is necessary to assess and enrich the findings of our analysis, drawing on additional country information. A final note relates to shared prosperity. As we have seen, the strong poverty decline in Uganda has coincided with a virtually unchanged Gini coefficient, suggesting that poverty reduction has been driven by growth in the average per capita levels and leaving Uganda with a level of inequality that is well above the average for low- and middle-income countries. Moreover, the relationship between the Gini coefficient and GNI per capita suggests that inequality is a reflection of country-specific factors and that it should not be expected to respond strongly or systematically to changes in income per capita. Hence, for Uganda to address inequality it needs to make targeted efforts. Hence, whenever data allows, the results of the sample of the bottom 40% is presented, with special attention paid to indicators such as those related to education and health, access to finance, and secondary road infrastructure. In many ways, progress on the SDG agenda helps in making prosperity more shared.

5.1

Economic Growth Determinants

Acceleration of economic growth beyond the projected annual rate of 4 percent per capita growth up to 2030 (Section 4.1) would make it possible for Uganda to accelerate progress on the SDG agenda.27 The key reason is that the resulting increases in private and public incomes would permit higher real spending in SDG-related areas as well as more broadly in areas that reinforce the growth acceleration. The factors that make it possible to accelerate growth and reach higher income levels vary across countries and time periods.28 Nevertheless, research suggests that a set of factors play a positive role often enough to qualify for consideration when growth is analyzed across most country types. In fact, such research suggests that the SDG agenda also is a growth agenda: as noted by the Commission on Growth and Development (World Bank, 2008), successful countries have typically invested heavily in infrastructure, health and education, including the education of women, and made sure that the growth process is inclusive, keeping inequities under control. Other common ingredients for successful growth strategies include high levels of total (including private) investment and national savings, openness to trade, structural changes away from agriculture, urbanization, strong institutions (including an effective government), and reliance on the private sector (World Bank, 2008). In this section, we review Uganda’s performance with respect to the growth determinants that have not been discussed so far—in Section 2 we covered sectoral issues and in Section 3 several SDGs affecting economic growth, including human development and infrastructure.29 This remaining set of potential growth determinants is divided into the following groups: investment and savings30; business climate and

27

Selected list of references on the importance of income on SDGs: Poverty and employment – Dollar et al. (2013), Ravallion (2001), Kapos (2005); Infrastructure – Foster and Briçeno-Garmendia (2010), Anand (2008); Education – Clemens (2004), Baldacci et al. (2003, 2004); Health – Wagstaff and Claeson (2004), Bloom and Canning (2004), Rajkumar and Swaroop (2008). 28 Any discussion at cross-country level is, by necessity, generalized and insensitive to country specifics. For example, while FDI finances a larger private capital stock across the board, the extent to which it contributes to growth and development may more significantly depend on its contribution to technological development and tax revenues—the latter being especially important if profitability is high (such as FDI in natural resource sectors tends to be). Such issues are often sector- and country-specific, with the developmental contribution of FDI often depending largely on government capacity. The implication is that, while analysis of the type undertaken in this paper provides a cross-country perspective that may be indispensable, it should be regarded as only a starting point that needs to be complemented by country-specific analysis. 29 See for example Calderón et al. (2011), Agénor and Moreno-Dodson (2006), World Bank (2008), Estache et al. (2005), Rothman et al. (2010). 30 See for example World Bank (2008)

40

governance;31 human capital and labor;32 and foreign trade (openness and export structure).33 In each area, we take stock of Uganda’s performance and how it compares to that of other countries. Our aim is to identify areas in which policy changes potentially could bring forth growth acceleration and a permanent increase in GDP and GNI per capita. We conclude with a brief summary and discussion of how our findings relate to the current development strategy of Uganda’s government.34

Investment and Savings GDP growth may be explained by growth in factor productivity and factor employment. Among the factors, the stocks of available private and public physical capital are created by investment (gross fixed capital formation). Over the last decades, the GDP share of private investment in Uganda has increased substantially, from a mere 5 percent in the late 1980s to around 17-18 percent in recent years, whereas public investment has stayed quite constant, at around 5 percent of GDP, together bringing about a sizable increase in total investment (Figure 5.1, Panel 1). The bulk of private investment is carried out by the domestic private sector; however, the GDP share of FDI has also increased, especially in the last few years. Across countries, the GDP shares of all investment categories tend to vary considerably without any strong relationship to GNI per capita. Where weak relationships exist, private investment is positively related and public investment negatively related to GNI per capita (Figure 5.1, Panels 2-4). Compared to other countries and given its GNI per capita, Uganda’s private investment (17.5 percent of GDP) is somewhat higher and public investment (5 percent of GDP) somewhat lower than expected, leaving total investment at the expected level. While most private investment is domestically financed, foreign direct investment (FDI) may account for a significant part; its importance is often due not only to financial aspects—a foreign exchange inflow when investment takes place and an outflow when profits are repatriated, and tax revenues—but also to the technology that it embodies.35 In recent years, Uganda’s FDI has been highly volatile and increased dramatically (Figure 5.1, Panel 1). Across countries, FDI’s share of GDP share is increasing slightly in GNI per capita (Figure 5.1, Panel 5) while Uganda’s current, exceptionally high, level of 8.7 percent of GDP— directed mainly at the oil sector—is well above the expected level (3.1 percent of GDP).

31

See for example Knack and Keefer (1995), Acemoglu et al. (2001) See for example Barro (2001), Young (1995), Lin (2003), Sala-i-Martin (2004), World Bank (2008), Psacharopoulos and Patrinos (2004), Bloom and Canning (2005). 33 See for example Galor and Mountford (2006), Hausmann et al. (2007), Hausmann et al. (2011) 34 As we shall see, Uganda is in many respects a typical country for its income level. So, in order to grow faster than its peers it needs to develop a strategy that, with a realistic view of government capacity, identifies ways of a promoting a continuous process of structural transformation, with policies affecting human capital, business climate, and infrastructure, playing key roles. As long as this is well-grounded in domestic realities, it may well involve deviating from expected values. However, this analysis focuses primarily on areas that are likely to be binding constraints for further growth in Uganda, and where policy actions are most feasible (which does not automatically translate into desirable)—i.e., areas in which Uganda is falling behind other countries despite having the same resources and capacity (in terms of income per capita). Our analysis is not intended to generate policy prescriptions, but rather to direct the thinking of policymakers and analysts to areas highlighted by cross-country comparisons, and thereby to raise questions about potential policy changes. 35 The extent to which FDI contributes to sustainable growth and development is context-specific, depending often on the ability of governments to maximize technological spillovers, infrastructure investments available for broader use, and/or tax revenues. In Uganda’s context of FDI in oil sectors, tax and infrastructure aspects are particularly important. 32

41

Figure 5.1: Investment – Public, Private and Foreign Direct Panel 1 (upper left): Investment—Public, Private, and Foreign Direct (% of GDP), 1990-2012 Panel 2 (upper right): Investment—Total (% of GDP) vs. GNI per capita Panel 3 (middle left): Investment—Public (% of GDP) vs. GNI per capita Panel 4 (middle right): Investment—Private (% of GDP) vs. GNI per capita Panel 5 (bottom left): Investment—Foreign Direct (% of GDP) vs. GNI per capita

42


For most countries, the bulk not only of investment (as indicated above) but also of savings stems from domestic sources. Among other things, as with domestic investment savings’ volatility tends to be much lower, all things being equal, making it more desirable as a source of sustainable financing. Since 1990, Uganda’s gross (national) savings first increased, covering most of investment spending up until 2005, but then decreased substantially (Figure 5.2, Panels 1 and 2). In 2011, it was at 15.4 percent of GDP, roughly at the expected level and close to the average of all low- and middle-income countries. Figure 5.2: Gross National Savings Panel 1 (left): Gross National Savings and the Current Account Deficit (% of GDP), 1990-2011 Panel 2 (right): Gross National Savings (% of GDP)


43

Human Capital and Labor Improvements in human capital and growth in labor force participation may be major contributors to GDP growth. The level of human capital is often measured by average years of schooling among adults, ages 15 and above (even though such an attainment measure does not measure learning achievements). Uganda’s average, at 5.5 years, is not high, but it is still above the expected level (Figure 5.3). Looking at years of schooling among younger people, Uganda is performing as expected, also when only including those from the bottom 40% households (see Figure A.4 in Appendix). With a lag, the SDG agenda of universal primary schooling with emphasis on learning outcomes, and improved access to lower secondary education, should both boost the contribution of human capital to growth. Figure 5.3: Average Years of Total Schooling vs. GNI per capita


High labor force participation channeled into higher employment should raise the level of GDP, assuming no countervailing forces are in play. This consideration may be particularly important for groups that tend to be on the margin of the labor market: women and youth not in school. Figure 5.4, Panel 1 shows a cross-country relationship in which the male labor force participation rate for population aged 15 and above is inversely related to GNI per capita (reflecting, in part, that countries with higher incomes tend to have a higher proportion of age-related retirees). For Uganda, the male share is as expected for Uganda’s income per capita. For many developing countries the share of women participating in the labor force is small, usually for cultural reasons, and a gender GDP dividend would be likely if those women were to participate. However, in the case of Uganda, the female labor force participation rate is higher than expected, 75.9 percent of the female population rather than the expected 57.9 percent (Figure 5.4, Panel 2). In sum, an increase in the general labor force participation ratio is not likely to be a source of future GDP growth acceleration.36 36

It is important to note that, according to the System of National Accounts (SNA), GDP and labor force participation refer to production of marketable goods (including home consumption) and marketed services (not including home consumption). In virtually all countries, especially those with a low female labor force participation rate, women contribute the bulk of the labor that goes into the production of non-market services (including raising children, food preparation, and cleaning). Given this, the welfare effects of increased female labor force participation may be quite different from the GDP effects—female leisure (or sleep) time declines and/or the supply of non-marketed services, essential to household welfare, may also decline.

44

Figure 5.4: Labor Force Participation, Male and Female Panel 1 (left): Labor Force Participation Rate, Male vs. GNI per capita Panel 2 (right): Labor Force participation Rate, Female vs. GNI per capita


The age dependency ratio—i.e., the ratio between the population younger than 14 or older than 64 and the number of people of working age—gives a measure of how many dependents an average worker needs to care for (directly or through a social security system). Therefore, a lower dependency ratio means that a larger share of the population contributes to productive, income-raising work (unless the unemployment rate is higher), which tends to translate into a higher GDP per capita. Given that the population aged 15-64 years tends to have higher savings rates, this production and income increase may disproportionately add to savings and investments, contributing to accelerated GDP growth.37 After a strong increase, Uganda’s age dependency rate has declined since the early 2000s (Figure 5.5, Panel 2) due to a decline in fertility (discussed in Section 5.2). However, Uganda’s current ratio of 104 is still much higher than the expected 81, suggesting that it will fall further, generating a continuing growth dividend.

37 Young (1994, 1995)

argues that the accumulation of labor explained most of the economic miracle of the so-called Asian Tigers. Bloom, Cunning, and Malaney (2000) tie about one-third of the growth to the changing age structure.

45

Figure 5.5: Age Dependency Panel 1 (left): Age Dependency Ratio vs. GNI per capita Panel 2 (right): Age Dependency, 1960-2011


Foreign Trade Foreign trade, both exports and imports, may be important for growth, not only by permitting access to larger markets, but also by facilitating access to technology and offering domestic producers the opportunities and the challenges of a competitive environment. From a different perspective, trade may matter by influencing the productive structure, facilitating structural transformation in favor of exportoriented production in sectors where the country’s capabilities can be improved and in which it may manage to catch attractive prices in world markets. After steady increases since the early 1990s, exports and imports (as percentages of GDP) have remained stable since the onset of the global economic crisis in 2008-2009 (Figure 5.6: Panel 1). In 2012, Uganda’s exports at 23.7 percent of GDP were roughly as expected while its imports, at 34.5 percent of GDP, were below the expected level of 45.4 percent, i.e., adding up to a lower level of openness than expected (Figure 5.6: Panels 2-3). This may be partly due to Uganda being landlocked with relatively high transportation costs that reduce the attractiveness of foreign trade. Given this, the more or less constant trade deficit, at 10.8 percent of GDP in 2011, is lower than expected. The relatively small trade deficit indicates that the non-trade component of Uganda’s balance of payments has less of a surplus than expected – the sum of foreign savings, net current transfers, and net factor incomes is by definition smaller than expected – an outcome not necessarily driven by trade-related issues.

46

Figure 5.6: Balance of Trade Panel 1 (upper left): Exports, Imports and Trade Balance, 1990- 2012 Panel 2 (upper right) Exports (% of GDP) vs. GNI per capita Panel 3 (lower left): Imports (% of GDP) vs. GNI per capita


Uganda’s concentration of exports, measured by the Herfindahl index, is relatively low compared to other countries (Figure 5.7), suggesting that it does relatively well in diversifying its exports. A disaggregation of exports further reveal that Uganda, relative to other countries at the same income level, has a level of goods exports as expected but higher than expected service exports (see Figure A.5 in Appendix). Natural resource exports show a mixed picture with below expected values for ores and minerals and above expectations for fuel. Within agriculture both raw material and food is exported as expected. Manufacturing exports are higher than expected; among these both ICT and high-technology goods are substantially higher than expected. While cross-country data on service exports are more limited, the

47

figure also shows that export of insurance and financial services, as well as international tourism receipts, are higher than expected for Uganda. Hence, this suggests that Uganda is on track to a higher value export. Figure 5.7: Trade diversification vs. GNI per capita

An emerging approach in the analysis of the growth prospects for different countries, developed by Hausmann and coauthors, is based on the Economic Complexity Index (ECI) and the related Product Complexity Index (PCI) (Hausmann et al., 2011). The ECI is constructed on the basis of the characteristics of the basket of goods in which the country has a revealed comparative advantage: To what extent is the country’s basket diversified and focused on goods exported by relatively few countries? For a product, the questions are similar, related to whether or not the product in question is exported by a small number of countries with relatively diversified export baskets. Empirically, controlling for per capita income, countries with a higher ECI tend to grow faster in subsequent periods. In an ECI-based cross-country analysis of 128 countries, Uganda emerges as one of the countries that has made most progress in recent decades—between 1998 and 2008, the country ranked 10th in terms of its improvement in ECI ranking— and that has the best prospects for growth up to 2020; Uganda’s projected 3.3 percent annual growth in GDP per capita puts it in the 24th place. These issues, which are further developed in a recent analysis of Uganda (Hausmann et al. 2014), are worthy of further exploration. As with other countries, Uganda may be able to boost future growth by developing exports in products that have relatively high PCI values without at the same time requiring capabilities that, given Uganda’s current production pattern, are beyond its current reach.38 In this context, oil exports and their likely effect on the real exchange rate may be detrimental by inhibiting production of other tradables, and this could have a negative impact on Uganda’s long-term prospects for developing a complex, diversified, and technologically sophisticated economy. Oil revenues will have to be managed very carefully to minimize such effects and derive benefits from them in other ways, so that they have a net positive impact. Among other factors, the prospects for export development depend on the strength of a country’s trade infrastructure; the Logistics Performance Index tries to measure this strength on the basis of data on the 38

Exploiting the new-found natural resources may change this dynamic.

48

trade-related institutions and physical infrastructure.39 In Uganda, this index has a moderately tight positive association with GNI per capita (Figure 5.8). Encouragingly, according to this index, Uganda does have a trade infrastructure that is better than expected for its income level (2.82 rather than the expected 2.34 (1=low, 5=high)). Figure 5.8: Logistics Performance Index vs. GNI per capita


Business Climate The ability of the private sector to translate savings and investment into growth depends on their institutional environment. In order to assess the quality of this environment for purposes of this paper, we will review indicators related to the business climate: the World Bank Doing Business Index as well as indicators related to credit, the rule of law, and corruption. The Doing Business Index of the World Bank ranks countries according to the extent to which their regulatory environments enable the starting and operation of local firms. Uganda was ranked 132nd out of 189 countries in 2013—somewhat better than expected (145th) relative to GNI per capita (Figure 5.9).40

39

More specifically, the trade Logistics Performance Index reflects the combined influence of six indicators: (i) the efficiency of customs and border management clearance; (ii) the quality of trade and transport infrastructure; (iii) the ease of arranging competitively priced shipments; (iv) the competence and quality of logistics services; (v) the ability to track and trace consignments; and (vi) the frequency with which shipments reach consignees within scheduled or expected delivery times. 40 The Doing Business Index is defined on the basis of the average of separate percentile rankings for 10 indicators, related to starting a business, dealing with construction permits, getting electricity, registering property, getting credit, protecting investors, paying taxes, trading across borders, enforcing contracts, and resolving insolvency. (See www.doingbusiness.org/rankings)

49

Figure 5.9: Ease of Doing Business vs. GNI per capita


Access to credit, which is an important dimension of the investment climate, may be measured in many ways, six of which are displayed in Figure 5.10, Panels 1-6. Cross-country regressions indicate that Uganda is roughly at the expected levels relative to GNI per capita for credit to the private sector as percent of GDP, and interest rate spread, which is a measure of financial sector inefficiency in intermediation (Panels 1 and 2);41 the real lending rate (nominal lending rate less inflation) of 26 percent was well above the expected level of 18 (Panel 3), even though the value was misleadingly high (and the average of about 22 for the previous five years was also above the expected level). Moreover, (Panel 4), Uganda scores considerably above the expected level for the credit depth of information index (a scale of 0-6, with a higher value indicating stronger availability of information). Finally, the percentages with an account at a formal financial institution and with a loan from a formal financial institution are higher than expected in Uganda, both when the whole sample is used (Panels 5-6) and the sample with the bottom 40% (Figure A.6 in Appendix). In sum, also in terms of more detailed credit indicators, Uganda’s performance seems close to or even better than expected. Figure 5.10: Credit, Interest Rates and Lending Panel 1 (upper left): Credit to the Private Sector vs. GNI per capita Panel 2 (upper right): Interest Rate Spread vs. GNI per capita Panel 3 (middle left): Lending Interest Rate vs. GNI per capita Panel 4 (middle right): Credit Depth Information vs. GNI per capita Panel 5 (lower left): Loan from formal financial institution (% age 15+) vs. GNI per capita Panel 6 (lower right): Account at formal financial institution (% age 15+) vs. GNI per capita

41

The interest rate spread is the margin between the cost of mobilizing liabilities and the earnings on assets; a narrow spread means low transaction costs, and the consequent lower cost of investment funding may generate more rapid growth (WDI). The credit depth of information index, which is on a scale from 0-6, measures rules affecting the scope, accessibility, and quality of credit information available through public or private credit registries.

50

51


There are two additional aspects of the business climate and, more broadly, the institutional environment for all agents, that warrant consideration here: the rule of law index (a percentile ranking indicating the level of confidence in the rule of law), and the control of corruption index (a percentile ranking indicating how well the state does in limiting corruption). Uganda ranks considerably higher than expected for rule of law (45.5 compared to 14.3) and is at the expected rank for control of corruption (Figure 5.11). Figure 5.11: Rule of Law and Control of Corruption Indices Panel 1 (left): Rule of Law vs. GNI per capita Panel 2 (right): Control of Corruption vs. GNI per capita

Source: WB Governance Indicators, World Bank

52

A large proportion of rural population might be regarded as having a negative impact on long-term growth prospects, impeding access to international markets and the generation and application of new knowledge in production. In addition, the unit cost of services (both public and private) that are important for the post-2015 agenda and the difficulty of accessing them tend to be higher if the population density is low and concentrated in rural areas. By these measures, Uganda may be handicapped by its low urban population share—merely 16 percent in 2012—much lower than expected (Figure 5.12, Panel 1), although the hindrance may be mitigated by the fact that Uganda’s average population density is relatively high (Figure 5.12, Panel 2). Figure 5.12: Population Density and Distribution Panel 1 (left): Share of Urban Population vs. GNI per capita Panel 2 (right): Population Density vs. GNI per capita


Summary of Insights from Growth Diagnostic The growth-related indicators reviewed in this section place Uganda generally quite close to the levels expected for a country at its income level—doing relatively well for some indicators, such as private investment, average years of schooling, rule of law, trade logistics, and complexity of export structure. Uganda has the advantage also of a declining age dependency ratio that may facilitate savings and growth in the near future. At the same time the analysis suggests that its prospects for more rapid future growth may be enhanced by higher levels of public investment (especially the findings, below, related to government efficiency) and the development of export-oriented production that promises to better contribute to growth.42 Such a process of structural transformation may be facilitated by stronger urbanization. Other parts of this diagnostic address factors that may contribute to accelerated growth. Among these, Uganda seems to already be well-positioned in its overall road network and ICT infrastructure. In terms of future priorities, investment in electricity generation may be particularly important and easy to bring about within a relatively short period, as would be expansion of district and rural roads for better 42

As far as tradable industries are concerned, Hausmann et al. (2014) provide a detailed analysis of sectoral issues, suggesting that Uganda’s policymakers explore measures that include the provision of infrastructure to promote growth in agrochemicals and food processing.

53

connectivity and inclusive growth. In the long term, education may be of particular importance both for growth and the extent to which its fruits are shared. In further analysis, it would be important to view these findings, which emanate from a cross-country perspective, in the context of Uganda Vision 2040, the document that expresses the government’s current long-term development strategy (Government of Uganda 2013). Moreover, in addition to benchmarking Uganda to a typical country by its income, it would be informative also to expand this cross-country analysis by comparing Uganda to high-growth countries (where 7 percent or higher may be defined as high) that have sustained growth over a substantial period of time—what patterns have they followed in terms of investment, schooling, structural transformation in trade and production, and other indicators.43

5.2

Education Determinants

The review of education performance in Section 3.3 suggested that Uganda has done relatively well in getting students to enroll in primary school but less well in getting them to complete primary and, if they complete, to proceed to secondary, which has undercut lower secondary enrollment. Among other indicators and levels, rates are roughly as expected for pre-primary enrollment and lower secondary completion, while tertiary enrollment is below expectations. The determinants of educational outcomes that are specific to the educational system and available in cross-country databases (primarily EdStats) include government spending and indicators of physical inputs such as pupil-teacher ratios at different levels. 44 Available broader indicators that may influence educational performance in coming years include urbanization and infrastructure,45 health (including nutrition)46, and per-capita household incomes (or consumption) as previously discussed. Issues related to fertility, including among adolescents, may also be important, especially from a gender perspective. The efficiency of public education spending is discussed in Section 6.47 As in other areas, the impact of changes is context-specific. It may be helpful to think of the different determinants as inputs to a “production function” that brings about desired outcomes; the impact of improvements in one input may be restricted unless accompanied by changes in other inputs. More concretely, the impacts of spending changes in education and other areas depend on the government’s ability to maintain or improve efficiency in each area. In a world of diminishing marginal returns (other things being equal), the impacts depend also on initial real inputs and spending per student or per capita. Contextual indicators, such as the educational level of parents, matter but are difficult to influence in the short or medium term.

43

World Bank (2008) may be helpful in this regard. It is here important to view indicators in their specific context. For example, high-growth countries have typically had investment and savings rates of 25 percent of GDP or above. Given that both investment and savings are only weakly related to GNI per capita, such a change is not impossible for Uganda. However, such a change would involve a difficult trade-off between, on the one hand, current consumption (with tangible current and sometimes future welfare benefits) and, on the other hand, savings and investments (for which welfare benefits are deferred). It is important to note that high-performing countries have realized such high savings and investment rates in the context of rapid growth, suggesting that even high investment volumes have been efficient and consistent with respectable consumption growth. 44 With respect to the role of government spending, see for example ADB (2006), Dickson et al. (2010), Gupta et al. (2002), Baldacci et al. (2003), Glewwe and Kremer (2006), Majgaard and Mingat (2012). Findings supporting the importance of physical inputs are found in Glewwe and Kremer (2006), Lay (2010), Orazem and King (2008), Glewwe and Kremer (2006). 45 See for example Agénor and Moreno-Dodson (2006), Gupta et al. (2002), Lay (2010), Glewwe and Kremer (2006), 46 See for example Gupta et al. (2002) and Baldacci et al. (2004). 47 See for example Baldacci et al. (2004), Rajkumar and Swaroop (2008), Grigoli (2014), Herrera and Pang (2005).

54

With regard to Uganda and the broader indicators that may influence education, discussed in other parts of this report, one would expect that, given initial conditions, improvements in infrastructure (perhaps most importantly electricity access, in which Uganda is lagging) could have a significant positive impact on education by facilitating students’ ability to study and prepare. Accelerated growth in income per capita would enable households to spend on complementary inputs (such as housing, electricity, transportation, and nutritious food) and to better afford to avoid child labor. Turning to education-specific determinants, the government’s expenditures on each of the main levels of education (primary, secondary, and tertiary) have stayed relatively constant in recent years (Figure 5.13, Panel 1).48 Across countries, government spending per student, measured as percent of GDP per capita, tends to be positively related to GNI per capita for the primary level but negatively related for the secondary level. By this measure, Uganda’s government spending is less than expected on the primary level but as expected on the secondary level (Figure 5.13, Panels 2-3).49 For primary education, one may hypothesize that Uganda’s modest success in getting students to complete primary school (reflected in high dropout and repetition rates) may be due to insufficient funding of primary schools so that they cannot provide quality education.

48

In the absence of data on private education spending by level, this analysis is focused on government expenditures. Publicprivate enrollment shares suggest that the government plays the leading role on the primary level, whereas the private and government split is quite even on the secondary and tertiary levels. On the pre-primary level, all enrollment is private (WDI). 49 In order to provide a more complete picture, the discussion should be expanded to consider the role of the non-government sector in education.

55

Figure 5.13: Government Education Expenditures Panel 1 (upper left): Government Expenditures per Student—Primary, Secondary, and Tertiary, 2004- 2012 Panel 2 (upper right): Government Expenditures per Student—Primary vs. GNI per capita Panel 3 (lower left): Government Expenditures per student—Secondary vs. GNI per capita


It may be more illuminating to measure education spending at each level per person in the population in the relevant age group (for example, spending on secondary education divided by the population in the secondary age group). This measure would be more relevant for countries close to or at universal enrollment. By such measure, Uganda’s spending is lower than expected for primary, roughly as expected for secondary, but not relevant for tertiary before 2030 because of extremely low enrollment (see Figure A.7 in Appendix).

56

Pupil-teacher ratios provide an alternative, physical measure of education inputs, partly representing output of government spending. For Uganda, the ratio at the primary level is higher than expected (47.8 vs. an expected value of 40.1; Figure 5.14, Panel 1). This is consistent with below-expected spending per student, and suggests that the low completion rate may be due to lack of resources. However, Uganda’s primary pupil-trained teacher ratio is as expected, implying that additional teachers in comparable countries are largely untrained (Figure 5.14, Panel 3). By contrast, at the secondary level, the pupil-teacher ratio is considerably lower than expected (18.5 compared to an expected value of 26.7; Figure 5.14, Panel 2), despite the expenditures per student being as expected. Given this, one possible explanation for low enrollment and completion rates is a relatively small number of entrants, proceeding from primary school.

57

Figure 5.14: Pupil-Teacher Ratios, Primary and Secondary Panel 1 (upper left): Pupil-Teacher Ratio, Primary vs. GNI per capita Panel 2 (upper right): Pupil-Teacher Ratio, Secondary vs. GNI per capita Panel 3 (lower left): Pupil-Trained Teacher Ratio, Primary vs. GNI per capita


Beyond the confines of education policy, issues related to fertility may have a strong bearing on educational outcomes (rates of enrollment, repetition and dropouts, especially among females but also more generally). Teenage pregnancies prejudice the ability of girls to pursue secondary and tertiary education. High fertility rates in general have a negative impact on the resources (time and income) that households can spend on children due to a high number of dependents per household (see Section 4.1 for discussion of aggregated effects on economic growth). More concretely, the household has a harder

58

time covering financial costs of schooling while contending with higher alternative costs to support the family. This may be seen as a micro-level manifestation of the consequences of a macro-level indicator, a high dependency ratio. In Uganda, fertility rates among adolescents (aged 15-19) and females irrespective of age, have both declined since 1990 (Figure 5.15, Panel 1). However, current rates are still well above the expected level for Uganda’s income per capita (Figure 5.15, Panel 2). For every 1,000 adolescent girls in Uganda, 131 have given birth compared to the expected 81.4 given Uganda’s per capita income. The general fertility rate for Uganda is also higher than expected—6.1 compared to an expected 4.6 (Figure 5.15, Panel 3).

59

Figure 5.15: Fertility Rates Panel 1 (upper left): Adolescent Fertility and Overall Fertility Rates, 1960-2012 Panel 2 (upper right): Adolescent Fertility Rate vs. GNI per capita Panel 3 (lower left): Overall Fertility Rate vs. GNI per capita


Cross-country research indicates that, in the context of growing incomes per capita and gradual urbanization, the direction of change is as expected. Given that decisions about fertility are embedded in the social and cultural fabric of a country, the effectiveness of direct policy levers is not clear-cut. The most obvious levers are related to contraceptives (influencing the ability of women and, more broadly couples, to manage pregnancy rates). However, in Uganda the use of contraceptives is above the expected level—prevalence is at 30 percent while the expected value is 23.9 percent (Figure 5.16).

60

Figure 5.16: Prevalence of Contraception vs. GNI per capita


5.3

Health Determinants

The review of health indicators in Section 3.4 shows that, relative to its GNI per capita, Uganda is at the expected levels for the under-five and maternal mortality rates, suffers from a relatively high HIV incidence, but does better than expected for tuberculosis and child malnutrition. In this section, we discuss what policy may do to address the urgent task of further reducing these rates. There are several likely determinants specific to the health sector. The cross-country datasets consulted for this paper provide a limited set of determinants: public and private health spending50 and its efficiency51 (discussed in Section 6), the availability of real resources such as skilled personnel (physicians, health staff for birth attendance),52 and immunization rates. But, as with education, health outcomes are influenced also by determinants that reach far beyond the health sector itself. Research suggests that the following factors, all of which are part of the SDG agenda, may have a positive impact on health outcomes: education levels (especially among women),53 better nutrition, reduced fertility rates (a smaller number of pregnancies correlates with extended life expectancy among women),54 and access to infrastructure (including water and sanitation, electricity and roads) and urbanization.55 The latter may have an ambiguous impact, since while improved incomes and service access are beneficial, more severe air pollution and other environmental problems are detrimental.

50

See for example Wagstaff (2009), Gupta et al. (2002), Anand and Ravallion (1993), McGuire (2006), Hughes et al. (2010) See for example Lewis (2006), Grigoli and Kapsoli (2013), Herrera and Pang (2005), Gupta et al. (2002), Baldacci et al. (2004), Rajkumar and Swaroop (2008) 52 See for example Lay and Robillard (2009), Anand and Bärnighausen (2004), Fay et al. (2005) 53 See for example Baldacci, et al. (2004), Anand and Bärnighausen (2004), McGuire (2006), Lay and Robillard (2009), Fay, et al. (2005) 54 See for example Rajkumar and Swaroop (2008), Baldacci, et al. (2004) 55 See for example Wagstaff and Claeson (2004), Günter and Fink (2010), Fay, et al. (2005), ADB (2006). 51

61

Turning to the health-specific determinants, Uganda’s total health expenditure as a share of GDP has increased significantly since the end of the 1990s (Figure 5.17, Panel 1). The increase has been especially sharp in private expenditures although it has now stabilized. Comparator countries appear to differ quite widely in terms of the GDP shares spent on health across all income levels (Figure 5.17, Panels 2-4). Nevertheless, while total health expenditures typically stay unchanged in relation to GDP, there is a slight tendency for a larger share to be channeled through the public health system. In Uganda, current total health expenditures are very high, at 9.5 percent of GDP, compared to the expected level of 5.9 percent (which, as noted, is invariant to changes in GNI per capita, i.e., the 5.9 percent is close to the average for all low- and middle-income countries with recent data). Public expenditures are as expected, at 2.5 percent of GDP, while private expenditures are significantly higher than expected, at 7.0 percent of GDP (compared to an expected 3.0 percent). Cross-country data for total, public, and private health expenditures per capita show increases with income per capita and where relationships are tight (see Figure A.8 in Appendix). In line with the result when looking at expenditures per GDP, Uganda is spending less than expected on public health per capita, but more on private health per capita—leaving total health spending per capita slightly above what is expected. It is important to note, though, that public spending presumably has a stronger impact on the type of health issues included in the SDG, with a possible exception of non-communicable diseases, where private spending may be as important.

62

Figure 5.17: Health Expenditures Panel 1 (upper left): Health Expenditures, Private and Public, 1995-2011 Panel 2 (upper right): Health Expenditures, Total vs. GNI per capita Panel 3 (lower left): Health Expenditures, Public vs. GNI per capita Panel 4 (lower right): Health Expenditures, Private vs. GNI per capita


Health inputs may alternatively be measured by physical input indicators. For two such indicators—skilled physicians and the proportion of births attended by skilled health staff (Figure 5.18)—Uganda is roughly at, or slightly above, the expected levels. In 2011, the number of physicians for every 1,000 persons were 0.117 (with 0.083 as the expected level) while the proportion of births attended by skilled health staff was 57.4 percent (with an expected value of 52.0 percent). Both input levels seem consistent with outcomes for major health SDG indicators.

63

Figure 5.18: Skilled Physicians and Staff for Births Panel 1 (left): Availability of Physicians vs. GNI per capita Panel 2 (right): Birth Attended by Skilled Health Staff vs. GNI per capita


Immunizations against DPT (diphtheria, pertussis and tetanus) and measles are effective means of reducing child mortality. Uganda has made considerable progress in terms of both immunization indicators in recent decades (Figure 5.19, Panel 1). In 2012, the DPT and measles immunization rates among children aged 12-23 months were 78 and 82 percent, respectively. Compared to other countries, Uganda is within the expected range (Figure 5.19, Panel 2).

64

Figure 5.19: Immunization Panel 1 (upper left): Immunizations, DPT and Measles, 1981-2012 Panel 2 (upper right): Immunization, Measles vs. GNI per capita Panel 3 (lower left): Immunization, DPT vs. GNI per capita


In sum, in the area of health, Uganda’s results and aggregate inputs are quite close to what is expected given its GNI per capita, with the exception of private spending, which is well above the expected level. Given that private spending is driven by the needs of the households in a setting with relatively high inequality, it is quite possible that a private spending level above the expected does not translate into a significant addition to the resources that address the SDG agenda.

65

5.4

Identifying Spending Priorities for More Ambitious SDG Targets

This section concludes the analysis of determinants by highlighting findings for those that are most likely to be influenced by policy (Table 4.1), giving a flavor of the type of determinants that may be analyzed in a more detailed study.56 Note that the table only includes policy relevant indicators and only those where cross-country data is available. In addition, household incomes per capita (highly correlated with GNI per capita) and some of the actual SDGs (including those related to infrastructure—for example, access to safe water affecting health indicators) may also matter although already presented in Table 5.1. For those in red text, performance is significantly weaker than expected, given Uganda’s GNI per capita, suggesting that improvements in policies and outcomes in these areas may be most feasible. Table 5.1: Policy-Relevant SDG Determinants SDG Government consumption (% of GDP) Public investment (% of GDP) Logistic Performance Index Ease of doing business rank Public expenditure per student, primary (% of GDP per capita) Public expenditure per student, secondary (% of GDP per capita) Public expenditure per student, tertiary (% of GDP per capita) Public expenditure, primary (% of GDP) Public expenditure, secondary (% of GDP) Public expenditure, tertiary (% of GDP) Pupil-teacher ratio, primary Pupil-teacher ratio, secondary Public health expenditures (% of GDP) Contraceptive use (% of population) Physicians (per 1,000 people) Skilled staff at birth (% of births) Adolescent fertility rate (per 1,000 girls 15-19) Fertility rate (births per woman, 15+ years of age)

Recent value 11.3 6.7 2.8 132.0 7.6 20.7 45.6 1.8 0.8 0.4 47.8 18.5 2.5 30.0 0.12 57.4 131.0 6.1

Note: Green = Currently significantly over-performing; Red = Currently significantly under-performing; Black = Performing as expected; No projection = Too loose relationship with GNI per capita. Whether a certain deviation (positive or negative) reflects a stronger or weaker performance varies across indicators. For example, a positive deviation reflects weaker performance for poverty but stronger performance for water access. The terms over- and under-performance are used normatively; for example, for the maternal mortality rate, a lower rate than expected, is referred to as over-performance.

Decisions about spending priorities depend on policymaker priorities and assessments of available information. Such decisions are especially difficult when made in a situation such as Uganda’s, where large unmet needs coexist with a constrained capacity to scale up spending with retained efficiency. A crosscountry perspective can shed useful light on such decisions.

56

To exemplify, policy has a strong influence on health spending but little or no influence on parental education.

66

At the aggregate level, Uganda’s public spending-to-GDP ratio is low relative to its GNI per capita, for aggregate public consumption (at 11.3 percent of GDP in 2011, falling short by 2 percentage points) and, to a lesser extent, for aggregate public investment, suggesting that some expansion would not put excessive pressures on financing or institutional capacity. In this context, it is important to note that many successful high-growth countries have had exceptionally high levels of public investment, over-performing during a long period of time, after which they may reduce their levels (as a share of GDP). Education is not only important in itself and for economic growth, but for shared prosperity. At the primary education level, Uganda’s government spent around 7.6 percent of GDP per capita per student in 2011 (Table 5.1), which is less than the expected 11.0 percent. However, while spending per student as percent of GDP is less than expected, its spending on primary education as percent of GDP is as expected. The reason for this seeming contradiction is that enrollment is relatively high, largely due to high rates of repetition and enrollment of students who are older than the expected age for their grade. If repetition rates can be reduced and completion rates increased—something that would most likely require more spending per student—the GDP share for primary spending required to offer services similar to those of other countries will eventually decline as students graduate from the primary level. All things considered, an initial jump in the GDP spending share to 2.5 percent of GDP (compared to the current 1.8 percent of GDP) would raise spending to the expected level. However, even though such increased spending would raise per-student resources to what is typical for countries at Uganda’s GNI per capita, it still remains far below what may be needed to offer a quality primary education.57 At the secondary level, expenditures per student are as expected but the pupil-teacher ratio is lower than expected. Given the fact that the completion rate is as expected while the enrollment rate is below expectations (both rates are computed relative to the total population in relevant age groups), this suggests that the system, considering its level of spending, performs relatively well in terms of bringing enrolled students to completion. A more detailed investigation is needed to assess the room for efficiency improvements. As Uganda in the future meets the challenge of increasing the number of entrants that proceed from primary, the demands for public spending on secondary education will increase. As a result of expansion at lower levels, the demand for tertiary education will also increase, albeit with a lag. In 2011, public spending on tertiary education was 0.4 percent of GDP, less than expected. Like primary education, keeping spending per student as percent of GDP at expected levels may not be sufficient to offer a quality education.58 Moreover, fertility rates are significantly above the expected level. Studies suggest that households with many children end up giving the average child fewer resources and time for studies. Uganda’s higher-than-expected adolescent fertility rate renders many teenage girls unable to study or to fully benefit from their studies. Like a good education, good health is valuable in its own right apart from its contribution to economic development and shared prosperity. In health, under-five and maternal mortality rates are at expected levels while total health spending is higher than expected (9.5 percent compared to an expected 5.9 percent of GDP), suggesting that the scope for increasing total health spending is relatively limited. At a more disaggregated level, public spending is roughly as expected (2.5 percent of GDP) and private 57

In 2011, at PPP in constant 2010 US dollars, average public spending per primary student in low-income, middle-income, and high-income countries was $94, $554, and $6,353, respectively (UNESCO 2014a, p. 383; UNESCO 2014b, Table 11). 58 For Uganda and many other low-income countries, the education quality gap and challenge is particularly strong at the primary level. This is because at this level enrollment is higher and spending per student tends to grow faster than GDP per capita (raising the value for spending per student as percent of GDP per capita), reflecting initial over-enrollment relative to resources. At higher levels of education it is easier to manage the challenge: enrollment is smaller while growth in spending per student tends to be slower than growth in GDP per capita.

67

spending higher (7.0 percent of GDP compared to an expected level of 3.0 percent). In the short to medium runs, the ability of the public health sector to absorb additional spending while maintaining efficiency is severely constrained by a lack of qualified manpower, while waste is substantial, estimated at 13 percent of spending for 2005/2006 (Okwero at al. 2010, pp. 47, pp. 65-68). Meanwhile, the level of spending on current health MDGs is well below the recommended minimum—US$54 per capita at 2005 prices (Task Force on Innovative International Financing for Health Systems 2009, p. 11; WHO 2010, pp. 36-37); if projected growth rates are achieved, Uganda’s total health spending would not reach this level until about 2020. In other words, further financing for increased health services should still be a high priority, especially if the government managed to overcome the manpower and other constraints to increased absorptive capacity in the health sector. Regarding infrastructural development, investments and spending on operations and maintenance (in such sectors as water, sanitation, roads, electricity, and internet and communications technology, or ICT) are crucial for Uganda’s SDG agenda. But, despite having spent heavily on infrastructure during 20012009— at slightly above 10 percent of GDP, or US$1 billion per year (Ranganathan and Foster 2012, p. 43)—Uganda still lags behind comparator countries in electricity supply and access, air and rail transport and secondary roads. And although performing somewhat better than other countries, Uganda is severely challenged in achieving universal access to sanitation and considerably lacking in provision of running water and other services. According to Ranganathan and Foster (2012, p. 42), a program for accelerated (but still not unreasonable) progress may require annual spending of an additional US$400 million per year (in 2011 US dollars) through 2015, corresponding to around 2.4 percent of GDP. Given the importance of infrastructure access within the SDG agenda, and its key role in raising growth and contributing to a wide range of development goals – not least shared prosperity, it would be crucial to continue to improve services in this area up to 2030.

6

Step 4: Identifying Fiscal Space

The level and efficiency of public spending are typically among the determinants of the development of SDGs and their determinants. It is important to keep in mind that any given level of spending may take place within a wide range of policy frameworks, among other things with varying roles for public and private service delivery. In order to raise spending in priority areas, additional fiscal space is needed. It also matters to outcomes how resources are mobilized – the effects of additional aid are different from the effects of additional taxes. Here we primarily address fiscal space from a budgetary perspective since, by definition, budget resources are most directly controlled by policymakers. However, as will be noted, financing from NGOs and private investors may play an important complementary role. Our framework is comprehensive, analyzing the scope for creating additional fiscal space from taxes, fossil fuel subsidy cuts, overseas development assistance (ODA—i.e., grants and concessional loans), and other borrowing (domestic or foreign). It is also important to bring government efficiency into the analysis: if it is low initially, then improvements (which of course may be difficult) may release substantial resources for additional high priority spending without additional financing. If efficiency initially is high, then this source of fiscal space is less important. However, if so, the government is in a better position to use additional financing to scale up services and investments in priority areas while maintaining acceptable efficiency.59 The fiscal space analysis is based on cross59

The challenges of raising government efficiency in service delivery, in general and for services benefitting poor people in particular, is addressed in the seminal World Development Report of 2004, “Making Services Work for Poor People” (World Bank

68

country benchmarking of the current situation (when this is doable), domestic and foreign debt sustainability analysis, and anticipated major future developments, in addition to cross-country data, drawing on Uganda country information when needed. Table 6.1 summarizes the findings from the regressions, further discussed below. Table 6.1: Government fiscal space: Recent indicators and future directions of change. Impact on Comment Income and Efficiency Indicators Recent value future fiscal space Taxes (% of GDP) 13.0 + Likely increase (mainly due to revenues from oil sector) Fuel subsidies (% of GDP) 1.3 + Potential (and desirable) decrease. ODA (% of GNI) 10.1 Likely decrease. External Debt Stocks (% of GNI) 22.5 + Potential room to increase borrowing. Government efficiency + Potential (and desirable) increase. Non-oil taxes. Tax revenues are the main source of government financing in Uganda. Figure 6.1 shows how they have evolved since 1990, and benchmarks their current GDP share against those of other countries.60 As shown, Uganda’s tax revenue, at 13 percent of GDP in 2011, is as expected. The relationship with GNI per capita is not tight enough to project future changes on the basis of projected income growth. If non-oil tax policy were to change, then it would be important to consider the detailed design and likely effects on the SDG agenda of such changes, comparing the benefits from additional spending to the costs related to a reduction of the resources controlled by households and enterprises.61

2003). According to the report, the key to improved service delivery is institutional changes that strengthen relationships of accountability between policymakers, providers, and citizens. A large body of research stimulated by this report suggests that such institutional changes are possible but not easily implemented, largely because politicians in many settings may be able to resist accountability to citizens (Devarajan 2014; see also ODI 2014). 60 Figure 6 suggests, interestingly, that ODA per capita is unrelated to GNI per capita—i.e., there is no significant tendency to give higher aid per capita to the countries where needs are highest. 61 IMF (2013) suggests that, by 2018, an increase of 1.5 percentage points of GDP for non-oil would be feasible; Uganda would still remain within its expected range.

69

Figure 6.1: Tax revenues Panel 1 (left): Uganda—Tax Revenues 1990-2011 (% of GDP) Panel 2 (right): Tax Revenues (% of GDP) vs. GNI per capita

Sources: WDI, World Bank

Oil taxes. While considerable uncertainty is related to the oil sector – currently, 2018 is the expected starting year for production – it is likely that the sector will generate a substantial increase in tax revenues. According to one set of projections, the tax revenues from oil will reach 8 percent of GDP by 2023, after which they will decline gradually until 2045, when production ends and reserves are depleted; for the period 2016-2030, oil revenues may amount to an average of roughly 4.9 percent of GDP per year (IMF 2013, p. 57). Fossil fuel subsidies. Currently Uganda’s subsidy level is at around 1.3 percent of GDP, the equivalent of around 10 percent of government revenues, well within the typical range (Figure 6.2). A subsidy cut may be difficult politically. If implemented, it would add to fiscal space and contribute positively to the climate change agenda, a high priority given the global threat posed by climate change.

70

Figure 6.2: Fuel Subsidy (% of GDP) vs. GNI per capita

Source: EIA

ODA. The level of official development assistance (ODA), both as a share of GNI or in per-capita terms, has varied considerably over the years (Figure 6.3, Panel 1), and by both measures Uganda’s current ODA within the expected range (Figure 6.3, Panels 2 and 3). Uganda’s net ODA is at around 10.1 percent of GNI (9.4 percent of GDP), while the expected level is 11.1 percent of GNI. The cross-country relationship between GNI per capita and ODA (as % of GNI or GDP) suggests that Uganda’s ODA will decline relative to GNI and GDP while remaining constant in per capita terms. The likely advent of large oil revenues may lead to further cuts as donors turn to countries with more severe fiscal constraints. The projected 2030 level of ODA for Uganda – only taking the increased GNI per capita into account – is as low as 4.2 percent of GDP or, in an average year 2016-2030, at around 6.1 percent of GDP, i.e. a loss of 3.4 percentage points. To limit this loss, it may be possible to tap into global initiatives such as the Global Fund to Fight AIDS, Tuberculosis and Malaria.

71

Figure 6.3 Official Development Assistance (ODA) Panel 1 (upper left): ODA as % of GNI and ODA per capita, 1990-2011 Panel 2 (upper right): ODA as % of GNI vs. GNI per capita Panel 3 (lower left): ODA per capita vs. GNI per capita


Borrowing. Uganda’s external debt stocks have decreased substantially, not least following the Heavily Indebted Poor Country (HIPC). The current stock at 22.5 percent of GNI is lower than expected (Figure 6.4: Panel 1 and 2). Again the relationship to GNI per capita is not tight enough to make projections based on cross-country results. With an eye to the future, increased borrowing may create fiscal space (albeit at the cost of higher government indebtedness and interest payments). A recent IMF-World Bank Debt Sustainability Analysis (DSA) considers as sustainable an increase in Uganda's external public or publiclyguaranteed debt from 16 percent of GDP in 2012 to 22 percent in 2033, i.e. only moderately exceeds the rate of GDP growth, permitting additional annual borrowing of roughly 0.3 percent of GDP. In the DSA, it

72

was assumed that other debt stocks—public domestic and external private non-guaranteed—would not change from their current GDP shares of 13 percent and 10 percent, respectively (IMF 2013). Figure 6.4: Debt stock and debt service Panel 1 (left): Debt Stock and Debt Service, 1980-2012 Panel 2 (right): Debt Stock (% of GDP) versus GNI per capita


Government efficiency. A number of government efficiency measures are available (Box 4). On the basis of relationships between inputs and outputs, Grigoli and Kapsoli (2013) and Grigoli (2014) constructed indices for government efficiency in health and education spending, respectively; Dabla-Norris et al. (2011) developed a Public Investment Management Index (PIMI) that reflects actual practices in four areas (appraisal, selection, implementation, and evaluation). According to both the health and education indices, Uganda’s performance is below the expected levels; the education index is tightly related to GNI per capita, but the health index is largely uncorrelated (Figure 6.5, Panels 1 and 2). The PIMI is an intermediate case in terms of the tightness of its relationship to GNI per capita, and Uganda’s value is here roughly as expected. However, in terms of rankings for the 67 countries covered by the index, Uganda ranks 58th by income and 45th by PIMI, i.e., a case of over-performance. Meanwhile, the World Bank Governance Indicator measuring Government Effectiveness, which ranks countries by percentile and is designed to reflect perceptions of the quality of public services and policies, finds Uganda’s government much more effective than expected: rating the country at 14.8 percent and ranking it in 33rd place (high numbers reflect high effectiveness; Figure 6.5, Panel 3). For Regulatory quality Uganda is again performing better than expected (Figure 6.5, Panel 4), as for Rule of Law, already presented in Figure 5.11). Since the various indices measure different aspects of government performance, these findings may not be inconsistent. Among country-specific sources, scattered survey evidence also points to inefficiencies. For example, on any given day, roughly 15-20 percent of the teachers (including head teachers with supervisory responsibilities) are absent, with illness accounting for an almost negligible share of absences (UNESCO 2014a, pp. 31 and 267-268). Similarly, an analysis of local governments suggests, if all districts could be brought up to the health and education outcome-to-spending ratios of the best performing districts, then about one third of their budgets could be saved (World Bank 2013b, p. xiii). In sum, even though they are unpredictable, efficiency gains could potentially add considerable fiscal space.

73

Figure 5.5: Government Efficiency and Effectiveness Panel 1 (upper left): Health Expenditure Efficiency vs. GNI per capita Panel 2 (upper right): Education Expenditure Efficiency vs. GNI per capita Panel 3 (lower left): Government Effectiveness vs. GNI per capita Panel 3 (lower right): Regulatory Quality vs. GNI per capita

Sources: WB Governance Indicators, World Bank; Grigoli and Kapsoli, 2013; Grigoli, 2014

74

Box 6.1: Measures of Government Effectiveness On the basis of relationships between inputs and outputs, Grigoli and Kapsoli (2013) and Grigoli (2014) constructed indices for government efficiency in health and education spending; Dabla-Norris et al. (2011) developed a Public Investment Management Index (PIMI) that reflects actual practices in four areas (appraisal, selection, implementation, and evaluation). In addition, the World Bank Governance Indicators provide cross-country data on rule of law, government effectiveness, control of corruption, political stability and absence of violence, quality of regulations, and voice and accountability.

On balance, this information suggests the fiscal space for SDG priority spending could increase by as much as 4-5 percent of GDP.62 However, the extent of the increase is highly uncertain, not the least due to uncertainty regarding the future of the oil sector. In addition to the sources included in the table, it may be possible to attract additional FDI, especially for infrastructure investments, leveraged by additional government spending in this area. According to the IMF (2014), FDI is projected to continue at an average of 6 percent of GDP for the period up to 2033, i.e., reverting from the 2012 peak of nearly 9 percent, but remaining above the average for 2000-2012, which was 4.6 percent of GDP. It is important to note that, to varying degrees, trade-offs are involved when fiscal space is freed up and spending is increased according to priorities: policy makers need to think through scenarios for Uganda with and without major policy changes and the implications for the SDG agenda. The trade-offs may be least severe for success in raising government efficiency and ODA. For alternatives with different tax and subsidy policies, the net short- and long-run impacts of on different population groups should be considered. Additional borrowing increases the risk of unsustainable future debt levels. The advisability of scaling up priority programs would also depend on the government’s ability to maintain and if possible improve efficiency. To provide context, according to recent figures, total government spending amounts to around 20 percent of GDP (IMF 2013, p. 28). It would be a severe challenge to raise spending by 4-5 percent of GDP while maintaining acceptable efficiency. If spending is increased, it may be wise to make gradual adjustments and get guidance from frequent impact assessments. However, as we have seen, unmet needs for services and investments in key areas such as health, education, and infrastructure are considerable, suggesting that the case for scaled-up spending is strong, subject to availability of financing and the institutional capacity to make good use of the resources. If the projected increase in government oil revenues materializes, then the availability of financing does not appear to be the main constraint.

7

Conclusions

This paper has tested a diagnostic framework based on cross-country data to analyze the implications of the post-2015 SDG agenda at the level of individual countries, featuring Uganda as a pilot case. Subject to data availability and guided by the relevant literature, the framework compares current performance to expected values relative to GNI per capita, both for SDG indicators (i.e., indicators for which numerical

62

Using figures from the preceding discussion, a high estimate of the fiscal space increase may be as follows (all percent of GDP for an average year 2016-2030): 4.9 (oil taxes) + 1.5 (non-oil taxes) + 1.3 (fuel subsidy cuts) – 3.4 (ODA) + 0.3 (foreign borrowing) = 4.6. In addition, the government may be able to raise efficiency. However, as noted, the changes for individual items are uncertain, difficult to bring about, and/or subject to drawbacks (especially if increased spending is not efficient).

75

targets may be defined as part of the post-2015 agenda) and variables that influence the performance of the SDGs (determinants or indicators related to fiscal space). For the SDG indicators, projections are made to 2030 on the basis of projections for GNI per capita and the estimated cross-country relationship if the fit is considered sufficiently tight. Conclusions may be drawn regarding the findings for Uganda and the strengths and limitations of the framework. The findings for Uganda indicate that, for variables with a tight relationship to GNI per capita, the country’s performance was for the most part as expected relative to its GNI per capita; confirming that the development process for Uganda and most other countries follows regular patterns for many variables. The variables where Uganda deviated most strongly from a typical value tended to display a very loose association with GNI per capita (such as inequality). The fact that the country underperformed in various indicators may help set off alarms and define an agenda for more detailed, country-specific discussions, with the initial hypothesis that improvements are relatively feasible (for example, in the areas of primary school completion and progression to secondary school. In some areas, it was possible to hypothesize that causal linkages between variables were at work (e.g., between relatively weak primary outcomes and the allocation of relatively meager resources per student at this level, or in terms of links between fertility and female school outcomes at the secondary level). Regarding the ambitions of the SDG agenda, the results suggest that substantial yet only moderate progress should realistically be expected by 2030 for an economy like Uganda’s, that grows at a relatively rapid pace, unless substantial additional resources become available (i.e., in Uganda’s case, oil revenues). However, if such additional resources do become available, attention needs to be directed to the challenge of ensuring that the government is able to effectively scale up and manage spending. The framework used in this paper and the accompanying database offer analysts in developing countries and the broader international community useful starting points for assessing SDG targets and related policy and financing priorities to achieve them for virtually any low- or middle-income country. Such diagnostics can be conducted at fairly moderate cost, given that the multi-country database is readily accessible and can be used for cross-country analysis and benchmarking. However, it is important to note that, in order to permit more specific policy conclusions, the cross-country diagnostics that the framework offers should be linked to more detailed country-specific studies at country and sector levels.

76

References Acemoglu, D., S. Johnson, and J. A. Robinson, 2001. “The Colonial Origins of Comparative Development: An Empirical Investigation”, American Economic Review 91, pp. 1369-1401. ADB (Asian Development Bank). 2006. Key Indicators 2006: Measuring Policy Effectiveness in Health and Education. Manila, Philippines. Agénor, Pierre Richard and Blanca Moreno-Dodson. 2006. “Public Infrastructure and Growth: New Channels and Policy Implications”. World Bank Policy Research Working Paper 4064. World Bank: Washington, D.C. Anand, P. B. 2008. “Achieving the Water and Sanitation Millennium Development Goal,” pp. 90-122 in Ed. Mark McGillivray. Achieving the Millennium Development Goals. Palgrave MacMillan in association with the United Nations University – World Institute for Development Economics Research. Anand, Sudhir and Martin Ravallion. 1993. “Human Development in Poor Countries: On the Role of Private Incomes and Public Services.” Journal of Economic Perspectives, Vol. 7, No. 1, pp. 133–150. Anand, Sudhir and Till Bärnighausen. 2004. “Human resources and health outcomes: cross-country econometric study,” The Lancet, Vol. 364, pp. 1603-1609 Baldacci, Emanuele, Benedict Clements, Sanjeev Gupta, and Qiang Cui. 2004. “Social Spending, Human Capital, and Growth in Developing Countries: Implications for Achieving the MDGs”, IMF Working Paper wp/04/217, Washington DC. 25 Baldacci, Emanuele, Maria Teresa Guin-Sui, and Luiz De Mello. 2003. “More on the Effectiveness of Public Spending on Health Care and Education: A Covariance Structure Model.” Journal of International Development, 15: 709–725. Barro, R. J. 2001. "Human Capital and Growth." American Economic Review, 91(2): 12-17 Bloom and Canning, 2005, “Health and Economic Growth: Reconciling the Micro and Macro Evidence”, CDDRL Working Paper 42, Stanford Institute of International Studies. Bloom, D., Canning, D. and Sevilla, J. 2004. “The Effect of Health on Economic Growth: A Production Function Approach”. World Development, 32, 1-13. Bloom, D., D. Canning and P. Malaney, 2000. “Demographic Change and Economic Growth in Asia.” Population and Development Review, Vol. 26, pp. 257-90. Calderón, C., E. Moral-Benito, and L. Servén, 2011. “Is Infrastructure Capital Productive? A Dynamic Heterogeneous Approach”, World Bank Policy Research Working Paper 5682, Development Research Group, Macroeconomics and Growth Team. World Bank, Washington, DC. Clemens, Michael A. 2004. “The Long Walk to School: International education goals in historical perspective.” CGD Working Paper 37. March. Center for Global Development. Devarajan, S. 2014. “What the 2004 WDR Got Wrong”. Blog.

(http://blogs.worldbank.org/futuredevelopment/what-2004-wdr-got-wrong) Dickson, J., B. Hughes and M. Irfan, 2010. Advancing Global Education: Patterns of Potential Human Progress, Vol. 2, Ed. B. Hughes. Pardee Center for International Futures. Dollar, D., T. Kleineberg and A. Kraay, 2013. “Growth Still Is Good for the Poor”, World Bank Policy Research Working Paper 6568, Development Research Group, Macroeconomics and Growth Team. World Bank, Washington, DC. EIA, 2009. International Energy Statistics, United States Energy Information Administration, Washington, D.C. Estache, A., B. Speciale and D. Veredas (2005): “How much does infrastructure matter to growth in Sub-Saharan Africa?” unpublished manuscript.

77

Fay, M., D. Leipziger, Q. Wodon and T. Yepes. 2005. “Achieving Child-Health-Related Millennium Development Goals: The Role of Infrastructure.” World Development, Vol. 33, No. 8, pp. 1267-1284. Foster, V., and C. Briçeno-Garmendia. 2010. Africa’s Infrastructure. World Bank. Washington, D.C. Fouré, J., A. Bénassy-Quéré and L. Fontagné, 2010. “The Great Shift: Macroeconomic Projections for the World Economy at the 2050 Horizon”, CEPII Working Papers, No. 2012-03. Gable, S., H. Lofgren, and I. Osario-Rodarte. 2015. “Country Development Diagnostics Post-2015.” World Bank, Washington, DC. Galor, O. and A. Mountford, 2006. "Trade and the Great Divergence: The Family Connection," American Economic Review, American Economic Association, vol. 96(2), pages 299-303, May. Glewwe, Paul, and Michael Kremer. 2006. “Schools, teachers, and education outcomes in developing countries,” pp. 945-1017 in eds. Eric A. Hanushek and Finis Welch, Handbook of the Economics of Education, Vol. 2. Elsevier B.V. Grigoli, F. and J. Kapsoli, 2013. “Waste Not, Want Not: The Efficiency of Health Expenditure in Emerging and Developing Economies.” IMF Working Paper 13/187. International Monetary Fund, Washington, D.C. Grigoli, F., 2014. “A Hybrid Approach to Estimating the Efficiency of Public Spending on Education in Emerging and Developing Economies.” IMF Working Paper 14/19. International Monetary Fund, Washington, D.C. Günther, I. and G. Fink, 2010. “Water, Sanitation and Children’s Health: Evidence from 172 DHS Surveys”. World Bank Policy Research Working Paper 5275, Development Economics Prospects Group. World Bank, Washington, DC. Gupta, S., Verhoeven, M. and Tiongson, E. R. 2002. “The effectiveness of government spending on education and health care in developing and transition economies”. European Journal of Political Economy, 18, 717-737. Hausmann R. and B. Klinger, 2006. "Structural Transformation and Patterns of Comparative Advantage in the Product Space" Working Paper No. 128, Kennedy School of Government, Harvard University, Cambridge, MA. Hausmann, R. Hidalgo, C. Bustos, S., Coscia, M., Chung, S., Jimenez, J., Simoes, A. and Yildirim, M., 2011. The Atlas of Economic Complexity. Center for International Development at Harvard University. Hausmann, R. J. Matovu, R. Osire and K. Wyett, 2014. “How Should Uganda Grow?” ESID Working Paper No. 30, Effective States and Inclusive Development Research Centre (ESID), Manchester, U.K. Hausmann, Ricardo, Jason Hwang, and Dani Rodrik. 2007. “What You Export Matters.” Journal of Economic Growth 12 (1): 1–25. Herrera, S. and G. Pang, 2005. "Efficiency of public spending in developing countries: an efficiency frontier approach," World Bank Policy Research Working Paper Series, 3645, The World Bank. Hughes, B. R. Kuhn, C. Peterson, D. Rothman and J. Solórzano, 2010. Advancing Global Infrastructure: Patterns of Potential Human Progress, Vol. 3, Ed. B. Hughes. Pardee Center for International Futures. IMF, 2013. Uganda: 2013 Article IV Consultation and Sixth Review Under the Policy Support Instrument, Request for a Three-Year Policy Support Instrument and Cancellation of Current Policy Support Instrument—Staff Report; Public Information Notice and Press Release on the Executive Board Discussion; and Statement by the Executive Director for Uganda. IMF, Washington, D.C. Kapsos, S., 2005, Employment Intensity of Growth: trends and macro-determinants, ILO, Geneva, Knack S., and Keefer P., (1995), “Institutions and economic performance: cross-country tests using alternative institutional measures”, Economics and Politics, 7: 207-227. Lay, J. (2010). “MDG Achievements, Determinants and Resource Needs: What Has Been Learnt?” World Bank Policy Research Working Paper No 5320. World Bank, Washington, DC.

78

Lay, J., and A.-S. Robillard (2009). “The Complementarity of MDG Achievements: The Case of Child Mortality in Sub-Saharan Africa.” World Bank Policy Research Working Paper No 5062. World Bank, Washington, DC. Majgaard, K. and A. Mingat, 2012. Education in Sub-Saharan Africa: A Comparative Analysis, World Bank, Washington, DC. McGuire, James W. 2006. “Basic Health Care Provision and Under-5 Mortality: A Cross-National Study of Developing Countries,” World Development, Vol. 34, No. 3, pp. 405-425. 27 Orazem, P. and E. King, 2008. “Schooling in Developing Countries: The Roles of Supply, Demand and Government Policy.” Chapter 55 in T. P. Schultz and John Strauss, eds. Handbook of Development Economics, Volume 4. 2008. Amsterdam: North Holland. Ostry, J. D., A. Berg, and C. G. Tsangarides, 2014. “Redistribution, Inequality, and Growth.” IMF Staff Discussion Note SDN 14/12, Research Department. International Monetary Fund, Washington, D.C. Psacharopoulos, George, and Harry Anthony Patrinos. 2004. “Returns to investment in education: A further update.” Education Economics, Vol. 12, pp. 111–134. Ravallion, Martin, 2001. "Growth, Inequality and Poverty: Looking Beyond Averages," World Development, Elsevier, vol. 29(11), pages 1803-1815, November. Rajkumar, Andrew S. and Vinaya Swaroop. 2008. “Public spending and outcomes: Does governance matter?” Journal of Development Economics, Vol. 86, pp. 96-111. Ranganathan, R. and V. Foster, 2012. “Uganda’s Infrastructure: A Continental Perspective.” World Bank Policy Research Working Paper 5963, The World Bank Africa Region and Sustainable Development Department, February 2012 Republic of Uganda, 2010. National Development Plan (2010/11 - 2014/15). Poverty Reduction Strategy Paper. Kampala. Rothman, D., M. Irfan, E. Margolese-Malin, B. Hughes and J. Moyet, 2010. Advancing Global Infrastructure: Patterns of Potential Human Progress, Vol. 4, Ed. B. Hughes. Pardee Center for International Futures. Sala-i-Martin, X. 1997. “I just ran two million regressions”, American Economic Review, 87, 178-183. Taskforce on Innovative International Financing for Health Systems, 2008. More Money for Health, and More Health for the Money. World Health Organization, United Nations, New York. UN, 2013. “A New Global Partnership: Eradicate Poverty and Transform Economies through Sustainable Development.” The Report of the High-Level Panel of Eminent Persons on the Post-2015 Development Agenda, United Nations, New York. Van Waegenberge, E. and H. Bargawi, 2011. “Macroeconomic policy for “full and productive employment and decent work for all”: Uganda country study.” ILO Employment Working Paper No. 91. International Labor Organization, Geneva. Wagstaff, Adam and Mariam Claeson. 2004. The Millennium Development Goals for Health: Rising to the Challenges. World Bank. World Bank, 2003. “Making Services Work for Poor People”. World Development Report 2004. Washington, D.C. World Bank, 2008. “The Growth Report: Strategies for Sustained Growth and Inclusive Development.” The Commission on Growth and Development. The World Bank. Washington, D.C. World Bank, 2009. “Uganda Transport Sector Development Project.” Project Appraisal Document. November 12, 2009. Washington D.C. World Bank, 2013a. “Financing for Development Post-2015”. World Bank, October. Washington, D.C. World Bank, 2013b. “Republic of Uganda: Service Delivery with More Districts in Uganda Fiscal Challenges and Opportunities for Reforms”. Public Expenditure Review. Report No. ACS4421. June. Washington, D.C.

79

Young, A., 1995. “The Tyranny of Numbers: Confronting the Statistical Realities of the East Asian Growth Experience.” The Quarterly Journal of Economics 110 (3): 641-680.

Appendix Figure A.1: Poverty at national poverty line Panel 1 (left): Poverty headcount ratio at urban poverty line (% of urban population) vs. GNI per capita Panel 2 (right): Poverty headcount ratio at rural poverty line (% of rural population) vs. GNI per capita


Figure A.2: Water and Sanitation Access, Rural and Urban Panel 1 (upper left): Access to Improved Sanitation Facilities, Urban vs. GNI per capita Panel 2 (upper right): Access to Improved Sanitation, Rural vs. GNI per capita Panel 3 (lower left): Access to Improved Water Sources, Urban vs. GNI per capita Panel 4 (lower right): Access to Improved Water Sources, Rural vs. GNI per capita

80


81

Figure A.3: Gender in Education Panel 1: Years of schooling, 15-19 years, male vs. GDP per capita Panel 2: Years of schooling, 15-19 years, female vs. GDP per capita Panel 3: Literacy rate, 15-24 years, male vs. GDP per capita Panel 4: Literacy rate, 15-24 years, female vs. GDP per capita

Source: Gender Stats, World Bank

82

Figure A.4: Years of schooling Panel 1: Years of schooling, 15-19 years vs. GDP per capita Panel 2: Years of schooling, 15-19 years, bottom 40% vs. GDP per capita


83

Figure A.5: Disaggregated Sector Exports vs. GDP per capita

84

85


86

Figure A.6: Access to credit Panel 1 (left): Loan from formal financial institution (% age 15+), bottom 40%) vs. GDP per capita Panel 2 (right): Account at formal financial institution (% age 15+), bottom 40%) vs. GDP per capita

Source: FinDex, World Bank

Figure A.7: Education Expenditure by level per person in age group* Panel 1 (left): Primary Spending vs. GNI per capita Panel 2 (right): Lower Secondary Spending vs. GNI per capita

Source: EdStats, World Bank *Note: No data available for pre-primary or upper secondary. Tertiary enrollment too low for reliable measurement

87

Figure A.8: Health Expenditures per capita Panel 1 (upper left): Total Health Expenditures vs. GNI per capita Panel 2 (upper right): Public Health Expenditures vs. GNI per capita Panel 3 (lower left): Private Health Expenditures vs. GNI per capita


88