THE WORLD BANK GROUP AFRICA REGION POVERTY REDUCTION & ECONOMIC MANAGEMENT J U Ly 2 0 1 2 IssUE 3 South Africa 71553 Focus on Inequality of Opportunity Economic Update South Africa Economic Update Focus on Inequality of Opportunity © 2012 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street NW Washington, DC 20433 USA All rights reserved This report was prepared by the staff of the Africa Region Poverty Reduction and Economic Man- agement. The findings, interpretations, and conclusions expressed herein are those of the authors and do not necessarily reflect the views of the World Bank’s Board of Executive Directors or the countries they represent. Photos by John Hogg/World Bank. The report was designed, edited, and typeset by Communications Development Incorporated, Washington, DC. Contents Acknowledgments v Foreword vi Executive summary vii Section 1 Recent economic developments 1 Global economic prospects 1 Recent trends in South Africa 5 Economic outlook for South Africa 11 Risks to the outlook 12 Section 2 The state of human opportunity in South Africa 15 Opportunities among children 18 Inequality in labor market opportunities 30 Conclusions 38 Annex A Illustrating the Human Opportunity Index with a simple example 41 Annex B Three key properties of the Human Opportunity Index 42 Annex C Estimating the Human Opportunity Index from household survey data 43 Annex D Opportunities and circumstances for children 44 Annex E Inequality in employment opportunities 46 Notes 47 References 53 Boxes 2.1 The Human Opportunity Index 18 2.2 Measuring inequality of opportunity 23 2.3 Contrasting opportunities of three children 30 Figures 1.1 Daily sovereign credit default swap rates since January 2010 1 1.2 South Africa has been an exception in the global equity market sell-off 2 1.3 Gross capital flows to developing regions have fallen 2 1.4 Global currency movements in May 3 1.5 Declining Purchasing Managers’ Indices indicate a manufacturing contraction 3 1.6 Commodity prices have declined as global economic uncertainty increases 5 1.7 Change in GDP growth, fiscal balance, and general government gross debt between 2007 and 2011 9 iii South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity 1.8 Core and headline Consumer Price Index inflation and Producer Price Index inflation, 2009–12 10 1.9 Current account, trade balance, and terms of trade, 2007–12 11 1.10 Bond and equity purchases by nonresidents 11 1.11 South Africa’s industrial production growth is highly correlated to developments in Europe 13 1.12 Percentage deviation in GDP growth in 2012 from baseline in the event of a further escalation in the Eurozone crisis 14 iv 1.13 Commodity exporters to be hardest hit by a 20 percent decline in metal and mineral prices 14 2.1 GDP per capita and the Gini index for consumption 15 2.2 Income sources by deciles 16 2.3 Growth incidence curves for South Africa, 1995–2005 16 2.4 School attendance (ages 10–14) 20 2.5 Finished primary school on time 21 2.6 Access to safe water on site 21 2.7 Access to improved sanitation 22 2.8 Access to electricity 22 2.9 D-Indices and HOIs for some key opportunities for South African children, 2010 23 2.10 Change in the HOI and decomposition of changes, 2002–10 24 2.11 Average annual change in the HOI for finishing primary school on time 25 2.12 Average annual change in the HOI for access to safe water on site 25 2.13 Average annual change in the HOI for access to improved sanitation 26 2.14 Average annual change in the HOI for access to electricity 26 2.15 Contribution of circumstances to overall inequality, 2010 27 2.16 Snapshot of the vulnerability profile 28 2.17 Inequality-adjusted coverage and inequality between groups differentiated by circumstances, education, and age 32 2.18 D-Index and inequality-adjusted coverage by employment, circa 2008 33 2.19 Decomposition of sources of inequality 33 2.20 How circumstances contribute to inequality 34 2.21 Profiles of top quintile and bottom quintile by probability, 2012 34 2.22 D-Index and inequality-adjusted coverage by location, 2008 and 2012 35 2.23 Decomposition of sources of inequality, 2008 and 2012 35 2.24 Inequality-adjusted coverage and D-Index for workers of different age groups, 2008 and 2012 36 2.25 Decomposition of sources of inequality by age group 37 Tables 1.1 The global outlook in summary 4 1.2 GDP growth by main sectors (gross value added), 2008–2012q1 6 1.3 Gross domestic expenditure growth by component, 2008–2012q1 7 1.4 Labor market transition matrices 8 1.5 Consolidated government fiscal framework, 2010/11–2014/15 9 1.6 Macroeconomic outlook, 2008–14 12 D1 What constitutes opportunities for South African children? 44 D2 What constitutes circumstances in the analysis of children’s opportunities? 45 D3 HOI, D-Index, and coverage for 2002, 2005, and 2010 45 E1 Definition of variables to measure “opportunities� in labor market 46 E2 Definition of circumstances and characteristics in labor market 46 Acknowledgments This edition was prepared by a core team com- Country Directors for South Africa) and John prising Fernando Im and Sandeep Mahajan Panzer (Sector Manager, AFTP1). (Co-Task Team Leaders), Allen Dennis, Sailesh The team is grateful for comments from Tiwari, Alejandro Hoyos Suarez, Shabana Claus Astrup, McDonald Benjamin, Haroon Mitra, and Phindile Ngwenya. Ambar Narayan Bhorat (University of Cape Town), Andrew (Lead Economist, PRMPR) was a special guest Dabalen, Shantayanan Devarajan, Quy-Toan co-author, leading the work on inequality of Do, Patrick Kabuya, Jose Antonio Leiva, opportunity, the focus topic for this issue. Peer Gladys Lopez-Acevedo, Konstantin Makrelov reviewers are John Newman (Lead Poverty Spe- (National Treasury), Jaime Saavedra, Philip cialist, SASEP) and Jose Molinas Vega (Senior Schuler, Marco Scuriatti, and Simi Siwisa. Economist, LCSPP). The report was prepared A team at Communications Development under the overall guidance of Ruth Kagia and Incorporated, led by Bruce Ross-Larson, Asad Alam (respectively, former and current edited, proofread, and laid out the report. v Foreword The World Bank is pleased to present the third of outcomes that reflects differences in innate South Africa Economic Update, with a focus human capabilities, effort, education, experi- on inequality of opportunity. ence, and skills. But a recognized goal for pub- In keeping with the earlier format, the lic policy is to ensure at least the equality of report has two parts. Section 1 provides an eco- opportunity for every individual in a country. nomic update and assesses the challenges and Many countries have used this new approach near-term prospects facing the South African to develop targeted policies to promote such economy. In particular, it looks at the impli- equality of opportunity and to monitor and cations for South Africa of the resurgence of evaluate the success of public programs. uncertainty in global financial markets, the We hope that the evidence-based analysis surge in capital flows to safe-haven assets, the presented here will promote debate and spur continuing Eurozone crisis, and signs of slow- new thinking on how the challenge of high down in some of the large emerging market inequality can be addressed in South Africa, economies. and how the lives of all of its people can be Section 2 focuses on inequality of opportu- improved. nity in South Africa. For the first time, using innovative techniques, this section presents an analysis of the interlinked inequality of oppor- Asad Alam tunities for children and for access to employ- Country Director for South Africa ment. Every society has a degree of inequality The World Bank vi Executive summary World GDP growth is set to fall from 2.7 percent precrisis 2008 levels, indicating only partial in 2011 to 2.5 percent in 2012, with a sharper recovery in these sectors. decline projected in developing countries. Against the prospects of a noticeably weaker The turmoil of recent months is estimated to global economy than envisaged at the time have knocked 0.2 percentage points from the of the November 2011 Economic Update, annual growth in the Eurozone, now projected medium-term growth projections for South to contract 0.3 percent this year. Growth is Africa have been revised downward. GDP expected to slow by close to 2 percentage points growth is projected at 2.5 percent in 2012, lower in both China and India and hold at just under than the 3.1  percent forecast in the Novem- 3 percent in Brazil. These drags on growth are ber 2011 update and the 2.7 percent forecast expected to ease somewhat, and global growth in the 2012 Budget. The main reason for the should strengthen during 2013 and 2014. Even lower projection is the further slowdown in the so, both developing country and high-income European Union and in China, the two main country GDP will grow less quickly than during export destinations for South Africa. An eas- this century’s precrisis years. ing of global financial market tensions would South Africa’s GDP growth, which showed contribute to a pickup in global demand, which nascent signs of strengthening in the latter should support gradual acceleration of growth part of 2011, could not maintain the momen- in South Africa toward its potential of 3.5 tum. The slowdown in 2012q1 was under- percent. An important constraint on a faster pinned by a sharp decline in mining output pickup in growth is likely to be bottlenecks and an almost across-the-board growth slow- in electricity supply, which is already rubbing down in services (finance being the excep- against peak demand. tion). These more than offset the marked The major downside risks remain tied to pickup in manufacturing and a turnaround how the situation in Europe unfolds. Europe in agriculture, which posted its first growth is South Africa’s largest export market, so in five quarters. The manufacturing pickup— industrial production growth in the two econ- striking given the weak economic outlook omies is strongly correlated. The form that an of the Eurozone, South Africa’s largest des- escalation of the crisis might take in the cur- tination for manufacturing exports—likely rent economic context, if one occurs, is very reflected the recovery that began in 2011q4 uncertain—partly because it is impossible following a sharp decline in the preceding to predict what exactly might trigger it, and two quarters because of industrial action in partly because the powerful forces unleashed mid-2011: 2012q1 output was only 0.4 percent could easily take a route very different from higher than in 2011q1. Manufacturing, min- the one foreseen by standard economic ing, and agriculture production remains below reasoning. vii South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Lower commodity prices present another legacy of the apartheid system, which denied downside risk. With much of the recent the non-whites (especially Africans) the chance increase in commodity prices driven by strong to accumulate capital in any form—land, demand from Asia (particularly China), a finance, skills, education, or social networks. cooling of the Chinese economy beyond pro- At the heart of high inequality lies the inabil- jections, coupled with weak demand from ity to create employment opportunities on a Europe, would dampen commodity prices, hit- large enough scale. Unemployment stands at ting South Africa’s growth prospects. Of devel- 25.2 percent (33.0 percent, including “discour- viii oping countries, South Africa would be among aged� workers), among the world’s highest. No the 10 countries to be hit most by the drop in surprise then that despite an almost 30 percent commodity prices. Indeed, simulation results increase in per capita GDP since the late 1990s, show GDP growth to fall by some 1.7 percent- reductions in poverty have been modest at best. age points. This would have been untenable without the An upside risk for South Africa and other growing social assistance grants. Noncontribu- developing countries is the possibility of a tory and means-tested (except for foster care) Growth has also stronger recovery in global demand than financial transfers from the budget account for embedded in the current baseline. While less more than 70 percent of the income of the bot- been highly uneven likely, this could come from an improvement tom quintile (up from 15 percent in 1993 and in its distribution, in market sentiment, perhaps due to additional 29 percent in 2000). With the social grants, progress on the reform agenda or to better- the entire spectrum of population ranked by perpetuating than-anticipated outturns in high-income income percentiles saw income growth between inequality and countries. For developing countries such 1995 and 2005. But without the grants as part as South Africa where some postcrisis slack of income, those below the 40th percentile saw exclusion remains, a stronger than expected recovery a significant decline in their income. In other in demand could be absorbed fairly easily and words, without the grants, two-fifths of the converted into improved living conditions and population would have seen its income decline lower unemployment. in the first decade after apartheid. Even after accounting for the equalizing The state of human opportunity role of social assistance, income inequality in South Africa remains extraordinarily high. To reduce it to While GDP growth—if modest by global com- more reasonable levels over the long run, social parisons—has averaged a credible 3.2 percent assistance is clearly not enough and needs to a year since 1995 (1.6 percent per capita), it has be complemented by other initiatives. These proven insufficient to absorb the wave of new would include a special focus on human capital entrants to the labor market from dismantling development, particularly among children and apartheid’s barriers. The potential for growth youths. has been held back by industrial concentration, Social consensus and policy action can skill shortages, labor market rigidities, and become more likely when the sources of chronically low savings and investment rates— inequality are decomposed and the issue is the latter, despite the fairly high and improving framed around the notion of equity rather than rates of return to capital. equality. A girl (let’s call her Thandiwe) born in Growth has also been highly uneven in the township of Tembisa outside Johannesburg its distribution, perpetuating inequality and to a single, uneducated mother earning R2,000 exclusion. With an income Gini of around 0.70 a month and with four other siblings should in 2008 and consumption Gini of 0.63 in 2009, have an equal shot at becoming a doctor or an South Africa stands as one of the most unequal engineer as a boy (let’s call him Andries) with countries in the world. The top decile of the one sibling, born in a two-parent household in population accounts for 58 percent of the coun- Sandton, Johannesburg. And so should a girl try’s income, while the bottom decile accounts (let’s call her Nothando), born in a family with for 0.5 percent and the bottom half less than similar characteristics as Thandiwe’s, but living 8 percent. In large part, this is an enduring in the rural area of Eshowe in Kwa-Zulu Natal. What would it take to equalize Nothando’s, any other personal circumstance beyond their Thandiwe’s, and Andries’s chances of success control and considered by society to be an in life? Ample research shows that access to a unjust source of exclusion. basic set of goods and services during child- For South Africa, the circumstances are per- hood can be an important (if far from perfect) sonal and family-related: gender of the child; predictor of future outcomes, including edu- ethnicity; household composition—presence of cation achievements and earnings. Access to the spouse of the household head in the house- quality basic services such as education, health hold, total number of children ages 0–16 in care, essential infrastructure (like water, sani- the household, and whether both parents live ix tation, and electricity), and early childhood in the household; education of the household development provides an individual, irrespec- head; other household head characteristics— tive of background, the opportunity to advance gender and age of the head; orphan status— and reach his or her human potential. whether both parents are alive; and location Analyzing such opportunities for children of the household—urban townships, informal in South Africa can help better understand the settlements, other urban areas, or rural areas. nature and causes of inequality of outcomes Note that, in covering the townships and infor- Opportunities among observed among adults. Opportunities among mal settlements, a broad brush overlooks their children in South children can also be reliable predictors of eco- remarkable variety in size, population density, nomic mobility across generations and over length of time since establishment, rural or Africa vary widely time. For instance, if access to economic oppor- urban location, and distance from major urban tunities, in the form of jobs (and earnings), centers. across different types credit, and ownership of land and financial The opportunities considered are those of goods and services assets is correlated with the circumstances of that enable South African children to realize an individual (such as race and location of resi- their productive potential directly, by enhanc- dence), it reinforces the link between children’s ing their human capital and creating a safe circumstances and their opportunities in life. phsyical environment, and indirectly, by pro- viding access to infrastructure amenities that Opportunities among children help ensure a decent quality of life and facili- Basic opportunities are defined here as a sub- tate the accumulation of human capital. The set of goods and services for children, such as human development component of opportu- access to education, safe water on site, or elec- nities comprises exposure to early childhood tricity, that are critical in determining opportu- development programs (children ages 0–4 nities for economic advancement in life. These years), school enrollment (at ages 6–11 years either are already affordable by society at large, and 12–15 years), timely completion of primary or could be in the near future, given the avail- school (ages 13–15), adequacy of school infra- able technology. Universal provision of basic structure and teachers (as reported by parents), opportunities is a valid and realistic social goal. and having health insurance. Opportunities among children are mea- sured in this report by the Human Opportu- The main findings. Opportunities among chil- nity Index (HOI), which is the coverage rate of dren in South Africa vary widely across dif- a particular basic service adjusted by how equi- ferent types of goods and services. Some tably the service is distributed among groups opportunities, such as school attendance for differentiated by circumstances. In discount- children under age 16 and access to telecom- ing inequitable access, the HOI reflects how munications, are nearly universal with an HOI personal circumstances—for which children above 90 percent. Others are well below uni- cannot be held accountable—affect their basic versal (an HOI of around 60 percent or below) opportunities. This means that two societies and distributed with high inequality among with the same coverage rate for any service can children of different circumstances, as for hav- have different HOIs if citizen access to that ser- ing health insurance, access to safe water on vice in one society is determined to a greater site and improved sanitation, and adequate extent by gender, race, family background, or space without overcrowding—and finishing South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity primary school (13–15-year-olds). Still, other an HOI of around 30 in 2002 to more than opportunities—such as access to ECD pro- 90 in 2010—and a big increase in the HOI for grams, safety in the neighborhood, and access electricity are other milestones in improving to electricity—are well below universal, but opportunities for children in South Africa. have low to moderate inequality attributable to Major challenges are the limited and circumstances. unequal access to safe water on site and In international comparisons, South Africa improved sanitation and the opportunity to fares well on school attendance but ranks finish primary school on time or be exposed below most comparators on the HOI for com- to ECD programs, along with the general lack x pleting primary school on time and access to of physical safety—all of which create the con- safe water on site, improved sanitation, and ditions for children to develop their human even electricity. Trends suggest improvements, potential. The research literature is rich with but the gaps with other countries are generally evidence linking childhood opportunities to not closing. Except for electricity, where South future success. Preschoolers with low cognitive Africa’s average annual progress has been development have lower school achievement Close to 70 percent exceptional, the progress on the other four and earn lower wages in adulthood. And early dimensions puts it in the bottom half of inter- childhood education has substantial long- of the bottom national comparators. term impacts, ranging from adult earnings to income quintile was Inequality of opportunity among children retirement savings. Access to safe water and in South Africa is shaped to a varying degree improved sanitation are particularly critical unemployed in 2008 by different types of circumstances. Whether a inputs for child health, a determinant of nutri- child lives in a township/informal settlement tion status. or a rural area as opposed to other urban area, and education of the household head con- Inequality in labor market opportunities tribute the most to inequality of opportunity The ability of individuals to find jobs com- in most cases where inequality is moderate or mensurate with their qualifications—irrespec- high. Location is particularly important for tive of circumstances—is crucial for economic opportunities related to infrastructure; and mobility and reductions in inequality. This is education of the household head contribute especially true in South Africa, with its chroni- the most to inequality in finishing primary cally high unemployment. Two key facts bring school on time and having health insurance, out the impact of exclusion from labor markets underscoring the lock of the family’s socio- on income inequality in South Africa. First, economic background on children’s future. close to 70 percent of the bottom income quin- Gender of the child contributes appreciably to tile was unemployed in 2008. Second, even inequality only in finishing primary school on those employed face sharp disparities in wage time. Ethnicity contributes to inequality in all earnings based on race, gender, location, and opportunities, but does not rank among the union membership. top two contributors for any. This implies not In analyzing inequality in labor market sta- that race- and gender-based disparities are not tus and its links to individual circumstances, important, but that race and gender are not this section uses a modified version of the the most important factors by themselves when approach in the previous section. The analysis the role of other circumstances (some corre- here provides a rough but nonetheless intui- lated to race and gender in the first place) are tive measure of the inequality between groups taken into account. of working-age adults differentiated by various The overall picture is therefore mixed. On attributes. It allows an assessment of the contri- the positive side, South Africa has achieved bution of circumstances (as opposed to that of near-universal access to primary education, a acquired or innate characteristics such as edu- necessary first step for equalizing opportunities cation and age) to inequality between groups among children and an important success for in the labor market. the education system to build on. The meteoric Three definitions of a desirable employment rise in access to telecommunications—from status are used for the analysis. Working-age adults (ages 15–64) are considered to have few jobs, but also by the inequality that persists desirable employment if they: between different groups in their access to • Have a job (any job), as opposed to being these few jobs. Reflecting the adverse effects unemployed or discouraged from seeking of the global financial crisis, the inequality- employment. adjusted coverage fell for finding employment • Are employed full-time, as opposed to between the first quarters of 2008 and 2012. being unemployed, underemployed, or In part, this reflected fewer supply of jobs on discouraged. account of the global crisis. But this was also • Are employed in the formal nonagricultural because of a rise in inequality between groups, xi sector, as opposed to being employed in the which cannot be attributed to the global crisis. informal sector outside agriculture. The causes of inequality in labor markets The first two definitions are closely aligned have changed in the past four years. The con- with the common notions of unemployment tribution of education has increased, while that and underemployment. The third definition of circumstances of gender and ethnicity has is restricted to a sample of only those who are fallen slightly. Where a person seeking employ- employed in the nonagricultural sector. ment lives, however, matters more now than it The employment Two types of individual attributes are con- did four years ago. situation appears sidered: circumstances (gender, ethnicity, and The rising importance of education carries location—urban township and informal settle- its own challenges: the disadvantages conferred to be particularly ments, other urban, or rural) and characteris- by unequal opportunities in education earlier tics (education and age of the worker). Unlike in life are an increasingly consequential stum- challenging for gender and ethnicity, location is somewhat bling block to the social and economic mobility young workers and debatable as a circumstance because working- of individuals, for whom having a job is crucial. age individuals have some control over where For employment in the formal sector outside residents of townships, they live. In practice, however, spatial mobility agriculture, a measure of quality jobs with informal settlements, is restricted by cost considerations, (lack of) some stability, education accounts for more social networks, and cultural and family ties. In than half the inequality in employment, the and rural areas South Africa particularly, mobility out of rural contribution having risen 8 percentage points areas and townships and informal settlements in four years. The particularly strong impact is fairly restricted because of historical reasons, of education on the likelihood of having a for- most affecting the poorest. mal sector job points to a high and rising skills The list of circumstances is shorter than premium in the labor market, confirming find- that for children because of data constraints, ings in the literature. The wage inequality from such as the lack of information about paren- the skills premium is a key driver of income tal characteristics for working-age adults. The inequality in South Africa. characteristics of education and age of the indi- The employment situation appears to be vidual are included as (imperfect) measures particularly challenging for young workers of acquired qualifications and experience. and residents of townships, informal settle- The contribution of circumstances to inequal- ments, and rural areas. An individual’s age is ity would indicate the extent of inequality of an unusually large contributor to inequality opportunity in the labor market—the part of in employment in South Africa, more than for inequality, produced by conditions in the labor many other middle-income countries, with the market, which is not explained by qualifica- odds increasingly stacked against the youngest tions acquired by the individual. workers. Inequality of opportunity, the part of inequality attributable to circumstances that The main findings. Given its high unemployment an individual has little or no control over, is rates, it is not surprising that South Africa does also higher among young workers than among much worse on labor market opportunities older workers. Thus not only do young workers than other middle-income countries. It is still face a disadvantage in the labor market due interesting and telling, however, that its rela- to their age, they also compete for jobs in a tive performance is explained not only by too market seemingly more “unfair� in allocating South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity opportunities among the young: a young per- policy. Making use of it, a number of coun- son’s chances of having a job, a full-time job or tries in Latin America, including Brazil and formal sector job, seem to depend more on cir- Peru, have begun confronting their inherent cumstances beyond their control than the skills inequalities with proactive and increasingly they have acquired. well-targeted social policies, with positive ini- In addition to being young and living in cer- tial results. South Africa, with its entrenched tain locations, being a woman and non-white inequality inextricably linked to its people’s still matters, increasing the likelihood of being varied circumstances, has a chance to forge a unemployed or underemployed significantly similar path of policy correction. xii (over and above any impact of these attributes Extraneous circumstances that a child is on education). born into (ethnicity, location, gender, and In some ways, the contributions of specific family background) were found in this report circumstances to inequality in the labor mar- to variably affect the child’s access to basic ket (like race and location) could be overstated opportunities in South Africa. Moreover, here, since circumstances related to parental some of the circumstances (location and eth- The presence of socioeconomic background, which are likely to nicity in particular) are also important for be correlated with these attributes, are missing inequality in employment opportunities later multiple deprivations from the analysis due to a lack of data. With- in the child’s life. This raises the prospect of points to the need out clear evidence, the contributions of race a vicious cycle of adverse circumstances that and location are best interpreted as reflect- compounds inequalities over multiple stages for policy programs ing socioeconomic factors, including (but not in life, and over the lives of multiple genera- in different sectors exclusively) race and location, in explaining tions. A rural black African girl, for example, inequality in the labor market. growing up with far fewer opportunities to (health and education, develop to her full potential is also less likely for example) to Conclusions to find employment opportunities as a young An equitable society would not allow circum- adult. Put differently, the child with disadvan- coordinate closely stances over which the individual has no con- tageous circumstances not only has to work trol to influence her or his basic opportunities harder to overcome the disadvantages, but after birth. Whether a person is born a boy or having done so, finds that these reemerge a girl, black or white, in a township or leafy when seeking employment as an adult. More- suburb, to an educated and well-off parent or over, the disadvantages do not stop with otherwise should not be relevant to reaching one person—they get transmitted across his or her full potential: ideally, only the per- generations. son’s effort, innate talent, choices in life, and, How to break through this vicious, self-per- to an extent, sheer luck, would be the influ- petuating cycle of inequality in South Africa? encing forces. This is at the core of the equal- To be sure, that would involve leveling the play- ity of opportunity principle, which provides a ing field in the quality of education children powerful platform for the formulation of social get and the employment opportunities they and economic policy—one of the rare policy face as young adults, irrespective of location, goals on which a political consensus is easier gender, or ethnicity. Paying special attention to to achieve. the water, sanitation, and health care needs of As with any other policy, an intuitive and rural areas and townships, and overcrowding objective measure of progress is crucial. The in townships would also be important. development of the human opportunity index, Policy design needs to recognize that chil- extensively presented in this report for various dren of certain circumstances are vulnerable childhood and employment related opportu- to deprivations in multiple dimensions simul- nities, has helped fulfill this requirement for taneously. For example, black South Africans the equality of opportunity objective. Together living in rural areas, and with household heads with a robust data gathering and monitor- who did not complete primary schooling, are ing and evaluation system, the HOI can help much more likely not to complete primary improve the targeting and efficacy of social school, be exposed to an ECD program, or have access to health insurance. The presence that a dynamic system involving policy experi- of multiple deprivations points to the need for mentation (from incentives for training and policy programs in different sectors (health hiring of young workers to monitorable and and education, for example) to coordinate incentive-based delivery of public services), closely in order to achieve better efficiency and backed by rigorous impact evaluation and the best results. greater participation of communities in the Of course, there are no simple, elegant pol- actual delivery of basic public service delivery icy solutions in the quest for equity. One impor- and in the feedback loops for policymakers, is tant lesson from international experience is crucial. xiii SECTION 1 Recent economic developments Global economic prospects Increased Eurozone jitters have reversed The resurgence of uncertainty in global finan- earlier improvements in market sentiment cial markets that began in early May 2012 has Markets have been in heightened turmoil since dampened the nascent global economic recovery early May. First, tensions rose sharply, sparked shaping up in the first four months of the year.1 by fiscal slippages, banking downgrades, and The renewed market nervousness, which has political uncertainties in the Eurozone, where since subsided, caused the price of risk to spike measures of financial market tension, such as upward globally. Signs of slowdown in the large credit default swap rates, rose close to their emerging market economies, particularly China, peaks in the fall of 2011 (figure  1.1). Some also contributed to the global economic distress. calm was then restored with coordinated Large-scale deleveraging of western banks and action by EU governments to provide relief ongoing fiscal consolidation in high-income to undercapitalized banks and debt-laden countries further add to the global headwinds member states. Credit default swap rates for that will moderate growth. Significantly, the most developing countries followed a similar large emerging market economies have greatly pattern. compressed macroeconomic policy space to buf- Global equities suffered a sharp sell-off in fer the latest round of external shocks. May and June (South Africa being a notable Figure Daily sovereign credit default swap rates since January 2010 1.1 East Asia & Paci c Europe & Central Asia Latin America & Caribbean Middle East & North Africa 500 400 300 Basis points 200 100 0 Jan. July Jan. July Jan. July 2010 2010 2011 2011 2012 2012 Source: World Bank DEC Prospects Group; Datastream. 1 South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity exception), wiping out $5.9 trillion in valua- flows were down 36 percent in May, as the tion, giving up almost all of the gains over the number of IPO deals in the first five months preceding four months (figure 1.2). With a of the year fell to 43 (raising $5.4 billion) from flight to safety amid heightened risk aversion, 79 ($17.4 billion) in the previous year. Overall yields on high-spread economies have been bank lending remains subdued, but it has been driven up, while those of safe-haven assets surprisingly resilient in recent months despite declined. rising external market volatility and deleverag- Capital flows to developing countries picked ing. Capital flows over the first six months of 2 up slightly in June after falling sharply in May the year were 23 percent below the same period due to renewed Eurozone tensions. Gross capi- in 2011. tal flows to developing countries fell 45 per- A surge in capital flows to safe-haven assets cent in May from the previous month. Bond caused the U.S. dollar to appreciate sharply issuance by developing countries declined against most other currencies in May (figure most, largely due to a 62 percent (m/m) drop 1.4). But much of the change reflects a strong in corporate borrowing (figure 1.3).2 Equity dollar, not weak local currencies. For example, Capital flows to placements fell to $5.3 billion in May on lower in May the euro fell 6.6 percent against the dol- activity in China and Brazil. Year-to-date equity lar, but on a trade-weighted basis it fell only developing countries picked up slightly Figure South Africa has been an exception in the global equity market sell-off in June after falling 1.2 Year to date Q1 Q2 20 sharply in May due to renewed 10 Eurozone tensions Percent 0 –10 Emerging BRICs Mature United Germany Japan South markets markets States Africa Source: Bloomberg data. Figure Gross capital flows to developing regions have fallen 1.3 Equities (new issues) Syndicated bank lending Bond issuance Total ows 60 40 Percent 20 0 Jan. June Jan. June 2011 2011 2012 2012 Note: Bars represent two-month moving averages of reported flows. Line shows raw unsmoothed data for the total of those flows. Source: Dealogic; World Bank DEC Prospects Group. Figure Global currency movements in May 1.4 Local currency unit/US$ Nominal effective exchange rate Italy Switzerland Sweden Spain Swaziland Bulgaria Brazil Botswana Chile Croatia Germany India 3 Mexico Hungary South Africa New Zealand Russian Federation Namibia Poland 0 3 6 9 12 15 Source: World Bank DEC Prospects Group. Industrial production 3 percent. Overall, developing country curren- Renewed tensions will add to headwinds has weakened and cies fell about 2.9 percent in nominal effective and keep global GDP gains modest terms (about 40 percent of their depreciation Industrial production has weakened and busi- business confidence against the dollar), with commodity exporters ness confidence has taken a beating since the down about 3.6 percent, against 2.5 percent start of the second quarter. Global industrial has taken a beating for countries not relying heavily on commodity production decelerated to 2.1 percent annual- since the start of exports. ized growth in May compared with 7.8 percent While financial market turmoil has receded in the previous month; industrial production in the second quarter somewhat in recent weeks, progress has been the Eurozone fell 7 percent in May. The global slow in addressing some of the underlying Purchasing Managers’ Index (PMI) descended problems of the Eurozone. The lack (or slow into sub-50 territory, suggesting that global man- pace) of fiscal convergence, the competitive- ufacturing output shrank in June, with all eco- ness differentials across countries of the com- nomic regions weakening (except China, whose mon currency area, and the need to contain official PMI improved slightly; figure 1.5). The sovereign country borrowing costs to guaran- decline in sentiment is consistent with a scenario tee debt sustainability in the medium and long where firms and consumers are holding back on runs remain the main unresolved challenges. spending because of greater uncertainty. Figure Declining Purchasing Managers’ Indices indicate a manufacturing contraction 1.5 China (of cial) South Africa Eurozone Other developing countries Other high-income countries 60 Purchasing Managers’ Index 55 50 45 40 Jan. Feb. Mar. Apr. May June Jul. Aug. Sep. Oct. Nov. Dec. Jan. Feb. Mar. Apr. May June 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2011 2012 2012 2012 2012 2012 2012 Note: A Purchasing Managers’ Index greater than 50 implies expansion. Source: Haver Analytics; Markit; World Bank DEC Prospects Group. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity The fresh worries, emanating mostly from 2.5 percent in 2012, with a sharper decline the unresolved sovereign debt and banking projected in developing countries (table 1.1). crises in Europe, are exacerbated by signs of The turmoil of recent months is estimated slowdown in the large emerging market econ- to have knocked 0.2 percentage points from omies, particularly China, that have run up the annual growth in the Eurozone, now pro- against capacity constraints and have much jected to contract 0.3 percent this year. Growth more limited macroeconomic policy space to is expected to slow by close to 2 percentage counter external shocks than at the start of the points in both China and India and hold at just 4 global financial crisis. 3 Large-scale deleverag- under 3 percent in Brazil (compared with 7.5 ing of western banks and ongoing fiscal con- percent in 2010). These drags on growth are solidation in high-income countries further expected to ease somewhat, and global growth add to the global headwinds that will moder- should strengthen during 2013 and 2014. Even ate growth. so, both developing country and high-income World GDP growth is set to fall from 2.7 country GDP will grow less quickly than during percent in 2011 (and 4.0 percent in 2010) to this century’s precrisis years. Both developing country and high- Table The global outlook in summary 1.1 (percent change from previous year, unless otherwise indicated) income country Indicator 2010 2011 2012f 2013f 2014f GDP will grow less Global conditions quickly than during World trade volume (GNFS) 13.0 6.1 5.3 7.0 7.7 Consumer prices this century’s G-7 countriesa,b 1.2 2.4 1.9 1.8 2.0 Commodity prices (US$ terms) precrisis years Nonoil commodities 22.5 20.7 –8.5 –2.2 –3.1 Oil price (US$ per barrel)c 79.0 104.0 106.6 103.0 102.4 Oil price (percent change) 28.0 31.6 2.5 –3.4 –0.6 Manufactures unit export valued 3.3 8.9 0.9 1.2 1.5 Interest rates US$, 6-month (percent) 0.5 0.5 0.7 0.8 1.1 €, 6-month (percent) 1.0 1.6 1.0 1.1 1.4 Real GDP growthe World 4.1 2.7 2.5 3.0 3.3 Memo item: worldf 5.1 3.7 3.3 3.9 4.2 High income 3.0 1.6 1.4 1.9 2.3 OECD countries 2.9 1.4 1.3 1.8 2.2 Eurozone 1.8 1.6 –0.3 0.7 1.4 Japan 4.5 –0.7 2.4 1.5 1.5 United States 3.0 1.7 2.1 2.4 2.8 Non-OECD countries 7.4 4.8 3.6 4.3 4.1 Developing countries 7.4 6.1 5.3 5.9 6.0 Sub-Saharan Africa 5.0 4.7 5.0 5.3 5.2 South Africa 2.9 3.1 2.5 3.2 3.5 Memorandum items Developing countries Excluding transition countries 7.8 6.4 5.5 6.1 6.2 Excluding China and India 5.6 4.4 3.6 4.3 4.5 PPP = power purchasing parity; f = forecast. a. Canada, France, Germany, Italy, Japan, the United Kingdom, and the United States. b. In local currency, aggregated using 2005 GDP weights. c. Simple average of Dubai, Brent, and West Texas Intermediate. d. Unit value index of manufactured exports from major economies, expressed in U.S. dollars. e. Aggregate growth rates calculated using constant 2005 dollars GDP weights. f. Calculated using 2005 PPP weights. Source: World Bank. If global conditions deteriorate, developing international crude oil prices were $25 lower countries will be hit hard and are unlikely to than their first quarter highs, and copper and bounce back as quickly as they did from the aluminum prices were down 10 percent and 16 2008/09 crisis. Although a resolution of the percent, respectively. Food prices fell less due tensions implicit in the baseline is still the most to tightness in edible oils and to weather con- likely outcome, a sharp deterioration of condi- cerns. Part of the recent decline in commodity tions cannot be ruled out. While the precise prices also reflects the appreciation of the U.S. nature of such a scenario is difficult to predict, dollar. developing countries could be expected to take 5 a large hit. Simulations at the World Bank sug- Recent trends in South Africa gest that developing country GDP could decline relative to baseline by more than 4 percentage Fragile growth momentum points in some regions, with commodity prices, GDP growth, which showed nascent signs remittances, tourism, trade, finance, and inter- of strengthening in the latter part of 2011, national business confidence all mechanisms could not maintain the momentum. The for transmitting the tribulations of the high- slowdown in 2012q1 was underpinned by a GDP growth, which income world to developing countries. (The sharp decline in mining output and an almost showed nascent signs possible impact on South Africa is discussed in across-the-board growth slowdown in ser- the risks to the outlook section.) vices (finance being the exception; table 1.2). of strengthening in the These more than offset the marked pickup Commodity prices have softened as in manufacturing and a turnaround in agri- latter part of 2011, global economic uncertainty increased culture, which posted its first growth in five could not maintain Eurozone tensions and global growth con- quarters. The manufacturing pickup—strik- cerns have added to the downward pressure on ing given the weak economic outlook of the the momentum commodity prices. After strengthening in the Eurozone, South Africa’s largest destination first quarter of 2012, most commodity prices for manufacturing exports—likely reflected have since retreated below their end-2011 lev- the recovery that began in 2011q4 following els (figure 1.6). The fall of prices was particu- a sharp decline in the preceding two quar- larly sharp in May as the European debt crisis ters because of industrial action in mid-2011: intensified and China’s growth slowed. The 2012q1 output was only 0.4 percent higher initial easing in prices occurred even as global than in 2011q1. Mining activity was affected by economic activity was firming, but it has been industrial action, safety stoppages, weak exter- accentuated with financial market tensions nal demand, and electricity demand manage- and weak growth expectations. As of July 3, ment drive by Eskom.4 A doubling of growth Figure Commodity prices have declined as global economic uncertainty increases 1.6 Agriculture Energy Metals and minerals 250 200 US$ prices (2005 = 100) 150 100 50 0 Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. 2005 2006 2007 2008 2009 2010 2011 2012 Source: World Bank. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Table GDP growth by main sectors (gross value added), 2008 –2012q1 1.2 (percent, seasonally adjusted annual rate) 2008 2009 2010 2011 2011q1 2011q2 2011q3 2011q4 2012q1 Primary sector –0.1 –4.2 4.0 0.0 –4.7 –6.0 –14.7 –1.0 –11.2 Agriculture 16.1 –1.5 0.5 –0.4 –4.8 –9.5 –6.9 –5.0 3.4 Mining –5.6 –5.4 5.5 0.2 –4.6 –4.5 –17.8 0.7 –16.8 Secondary sector 2.9 –6.9 4.3 2.1 10.1 –6.5 –0.5 3.5 6.4 Manufacturing 2.6 –10.1 5.4 2.4 12.8 –8.8 –0.7 4.2 7.7 6 Electricity –3.1 –1.4 1.7 1.3 3.1 1.0 –2.6 1.2 –0.1 Construction 8.5 7.8 0.9 0.8 1.2 0.8 1.8 1.9 3.8 Tertiary sector 4.6 1.0 2.3 3.6 3.8 4.0 4.2 3.5 3.0 Wholesale and retail 1.0 –1.2 3.5 4.4 2.4 5.2 6.1 5.2 3.0 Transport 4.0 0.9 2.0 3.3 4.1 4.3 2.3 2.9 2.5 Finance 7.4 1.0 2.0 3.5 5.3 2.7 4.5 2.3 4.1 Manufacturing, the Government services 4.5 3.8 2.7 3.9 3.1 5.1 4.2 4.4 2.3 Personal services 3.9 –0.9 –0.1 2.4 2.6 2.8 2.5 3.0 1.7 mainstay of growth GDP growth 3.6 –1.5 2.9 3.1 4.6 1.0 1.7 3.2 2.7 in the first quarter, Source: South African Reserve Bank. appears to be faltering in construction (3.8 percent q/q seasonally to disposable income (to 74.7 percent). Stag- adjusted annual rate, or saar) also contributed nant employment prospects (see next section) to the robust economic performance posted and the negative effects of the global situation by the secondary sector. Manufacturing, on consumer confidence, purchasing power mining, and agriculture production remains (potentially through the exchange rate and below precrisis 2008 levels, indicating only headline inf lation), and household wealth partial recovery in these sectors. (through asset and housing prices) are likely to More recent output trends exhibit growing weigh on consumer purchasing decisions mov- vulnerabilities of the real economy. Manu- ing forward. This can be already observed in facturing, the mainstay of growth in the first the moderation in retail sales and the sharp quarter, appears to be faltering, as reflected in plunge in business confidence in the retail sec- June’s 5.4 point decline in the PMI to 48.2, its tor (from 61 in 2011q1 to 39 in 2012q2, accord- lowest value since August 2011. A below-50 rat- ing to the RMB/BER Business Confidence ing usually reflects contraction in the sector. Index). In addition to the weaker external demand, Gross fixed capital formation remained the decline in the PMI also reflects softening strong despite losing some momentum in domestic demand, highlighted by a sharp 10 2012q1. There were wide variations among the percent decline in the new sales orders compo- different components, as the increase in the nent of the PMI. The global growth concerns growth of capital outlays by the government noted earlier do not bode well for commodity and public corporations (mainly by Eskom and prices or mining prospects (see section on risks Transnet) partly buffered the steep slowdown to the outlook). of capital formation by private enterprises. Gross domestic expenditure growth mod- Moreover, the bleak global economic out- erated to 4.3 percent q/q saar in 2012q1, as look, the recent downward trend in commod- final consumption and investment expendi- ity prices, the fragile nature of the domestic ture posted slower rates of increase (table 1.3). economic recovery, and the drop in business Household consumption expenditure growth confidence signal little hope for a surge in fell fractionally short of the growth of real investment, especially since capacity utilization household disposable income, resulting in a in the manufacturing sector is still below pre- marginal fall in the ratio of household debt crisis levels. Table Gross domestic expenditure growth by component, 2008 –2012q1 1.3 (percent, seasonally adjusted annual rate, unless otherwise indicated) 2008 2009 2010 2011 2011q1 2011q2 2011q3 2011q4 2012q1 Total final consumption 2.7 –0.1 4.0 4.9 7.0 2.4 4.1 5.2 2.9 Household consumption 2.2 –1.6 3.7 5.0 6.2 3.4 3.8 4.6 3.1 Durables –8.3 –12.6 18.1 15.7 19.6 13.5 17.5 16.6 8.2 Semidurables 4.7 –1.9 2.0 7.0 12.5 7.8 6.7 7.3 2.9 Nondurables 0.2 –1.0 1.3 2.9 5.0 2.1 2.0 2.2 3.5 7 Services 6.1 0.5 3.4 3.8 2.9 1.0 1.4 2.9 1.6 Government consumption 4.5 4.7 4.9 4.5 9.4 –0.4 4.8 7.3 2.2 Gross fixed capital formation 13.3 –3.2 –1.6 4.4 4.4 5.0 5.9 7.2 5.3 Private 9.2 –9.6 –0.8 5.3 4.9 5.3 5.4 6.2 1.8 Government 10.9 –3.4 –9.5 0.8 –0.5 1.7 3.4 7.8 9.3 Public corporations 36.2 22.2 1.8 4.2 6.2 6.0 9.0 9.6 13.1 Change in inventories (R billions) –9.0 –28.0 –1.8 4.8 7.1 3.1 4.0 4.9 5.9 An estimated 650,000 Gross domestic expenditure 3.5 –1.6 4.2 4.3 4.6 1.4 4.8 5.1 4.3 additional workers Source: South African Reserve Bank. have joined the ranks Labor markets To develop a more nuanced assessment of of the unemployed High structural unemployment has been exac- the churning within the South African labor since the end of 2008 erbated by the negative cyclical effects of the markets, we compute a six-month transition global financial crisis. An estimated 650,000 matrix for the period that preceded the reces- additional workers have joined the ranks of the sion of 2009 (2008q1–2008q3) and compare it unemployed since the end of 2008, a number with the transition matrix for 2011q3–2012q1 that rises to more than 1.8 million with those (table 1.4).5 A similar comparison was also made who were discouraged and stopped looking using a transition matrix for 2011q1–2011q3 to for a job. As a result, unemployment increased isolate the effects of seasonal factors, but the from 21.9 percent in 2008q4 to 25.2 in 2012q1, results did not change materially. The diagonal even though real GDP had recovered to its pre- elements in the tables indicate the probability crisis level by 2010q2. of staying in the same employment status dur- The narrow unemployment rate increased ing the six months, while the nondiagonal ones to 25.2 percent (broad, 33.8 percent) in indicate the probability of moving from one 2012q1, compared with 23.9 percent (32.7 status to another.6 percent) in 2011q4 and 25.0 percent (33.4 per- A comparison of the two panels suggests the cent) in 2011q1. Manufacturing and construc- following: tion saw the most significant job losses in the • The most visible differences in labor market last quarter, despite being among the fastest transitions, before and after the crisis, per- growing sectors in the quarter. The quarterly tain to those unemployed or discouraged. climb in the unemployment rate can be partly The persistence in these vulnerable catego- explained by seasonal factors. On an annual ries increased significantly following the basis, the economy added 304,000 jobs. But downbeat and uncertain conditions brought the increase in the number of discouraged about by the global crisis. workers, the strong growth in the working-age • Almost 70 percent of the unemployed kept population, and the significant number of non- looking but did not find a job within six economically active persons that transitioned months in the recent period, up from 53 into the labor force more than offset these percent before the crisis. Similarly, only 15 gains, leading to a 0.2 (0.4) percentage point percent of the unemployed are likely to find increase in the narrow (broad) unemployment a job within six months from now, down rate. from 26 percent in 2008. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Table Labor market transition matrices 1.4 (percent) 2011q3–2012q1 Employed Unemployed Discouraged Other inactive Employed 91.4 4.0 1.3 3.2 Unemployed 15.1 69.2 4.6 11.1 Discouraged 9.0 13.0 57.1 20.9 Other inactive 3.1 7.0 4.8 85.0 8 2008q1–2008q3 Employed Unemployed Discouraged Other inactive Employed 88.7 5.1 1.1 5.1 Unemployed 25.9 52.8 3.9 18.0 Discouraged 15.4 21.1 33.7 29.7 Other inactive 5.5 7.1 2.5 85.0 Source: Quarterly Labour Force Surveys (2008q1 and 2012q1). Deficit and public sector borrowing • The probability of a discouraged worker public expenditures, but the situation makes it targets were lowered finding a job fell by more than 6 percentage politically hard to cut back social spending. On points from an already low 15 percent in the the other hand, bond markets—the mainstay somewhat to fully previous period. Persistence in this category of budget financing for South Africa—are jit- restore market (that is, the chances of a discouraged worker tery worldwide, and credit rating agencies are remaining in that category) increased 70 taking an increasingly dim view of countries confidence percent. The discouraged worker’s chances with records of persistent budget deficits and of moving to unemployment also fell almost debt accumulation, especially if combined with 40 percent, reflecting less motivation to a level of policy and political uncertainty. This begin looking for a job. unpleasant reality was brought home to South • The transitions from being employed to Africa by recent outlook downgrades—from other statuses (unemployed, discouraged, stable to negative—by Fitch, Moody’s, and S&P. and other inactive) remained similar before Predictably shaped by these forces, the and after the crisis, reflecting relative stabil- 2012 Budget carried few elements of surprise ity in the position of those already holding (table  1.5). Deficit and public sector borrow- a job. The probability of staying employed ing targets were lowered somewhat to fully after six months was slightly lower in the pre- restore market confidence. Social expenditures crisis period relative to the recent period. remained the dominant component of spend- ing. And the increase in infrastructure spend- Fiscal policy ing over the 2012/13–2014/15 MTEF period Fiscal policy in South Africa (as elsewhere) is was kept at under 5 percent compared with the required to walk a very fine line. On the one MTEF cycle in the previous budget. Recogniz- hand, the fragile domestic economic recovery, ing the shortfalls in the municipalities’ capacity closely tied to the uncertain global situation, to carry out infrastructure spending, the bud- takes away the possibility of a rapid rollback get highlighted a proposed Cities Support Pro- of the countercyclical fiscal stance in recent- gramme, currently under Cabinet review, that year budgets. Major infrastructural shortfalls would put metropolitan authorities in charge persist, requiring stepped-up budgetary inter- of spatial planning, public transport, and vention together with greater private partici- managing the provision of utilities, backed by pation enabled by necessary institutional and robust capacity-building support. A careful and regulatory reforms. South Africa’s high unem- phased implementation (over 14 years) of the ployment and an unfinished agenda for public much-debated national health insurance frame- service delivery (see the focus on inequality of work was announced, subject to careful evalua- opportunity) cannot be resolved through higher tion of its financial viability. The budget carried Table Consolidated government fiscal framework, 2010/11–2014/15 1.5 (percent of GDP, unless otherwise indicated) Outcome MTBPS Budget Budget forecast 2010/11 2011/12 2012/13 2012/13 2013/14 2014/15 Revenue 27.5 27.7 27.0 27.4 27.8 28.0 Expenditure 31.7 32.5 32.2 32.1 31.7 31.0 Budget balance –4.2 –4.8 –5.2 –4.6 –4.0 –3.0 Interest cost 2.4 2.6 2.7 2.7 2.8 2.7 Public sector borrowing requirement 6.5 7.1 7.8 7.1 6.2 5.0 9 Total net government debt 29.7 33.3 36.7 36.0 37.8 38.5 Southern African Customs Union transfers (R millions) 17,906 21,763 38,983 42,151 37,245 41,416 Source: National Treasury of the Republic of South Africa, 2012 Budget. a welcome thrust on measures to strengthen the between 2007 and 2011 (figure 1.7). Although government’s financial management systems enabled by the relatively favorable position, Inflationary pressures (especially those related to procurement, con- this was a much more pronounced increase have receded in recent tract management, and supply chain manage- than the other BRICS and several other upper ment) for better value for money. middle-income countries. The fiscal stance months, helped by a For 2011/12, preliminary estimates of rev- remains sustainable overall, but lack of prog- enue collection were R4 billion higher than the ress in addressing structural problems that con- moderation in global revenue estimates in the 2012 Budget (R743 tribute to high unemployment and inequality, food and petrol billion), while preliminary estimates of govern- low growth, and festering elements of political ment spending were R4 billion lower than the uncertainty have led to a more cautious assess- price increases estimates set in the 2012 Budget (R968 billion), ment of the fiscal position by market analysts. resulting in a revised budget deficit of 4.5 per- cent of GDP, 0.3 percentage points lower than Inflation previously envisaged. Inflationary pressures have receded in recent As in other emerging markets, South Africa’s months, helped by a moderation in global food fiscal space has been curtailed after adopting and petrol price increases. After hitting the a countercyclical stance in recent-year bud- upper threshold (6 percent) of the inflation tar- gets. General government gross debt increased get band in October 2011—and remaining at or by more than 10 percentage points of GDP above that for six months—headline Consumer Change in GDP growth, fiscal balance, and general government gross debt between Figure 2007 and 2011 1.7 GDP growth (percent) General government General government budget balance (percent of GDP) gross debt (percent of GDP) 20 10 Percent 0 –10 South China Upper BRICs Russian Brazil India Africa middle-income Federation countries Source: IMF World Economic Outlook, April 2012; staff estimates. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Price Index inflation fell to 5.5  percent (y/y) heightened uncertainty and weaker global in June (figure  1.8). Food and petrol infla- growth prospects go hand-in-hand with lower tion, which account for roughly 18 percent of international oil and commodity prices. On the consumption basket, declined from 11 and the other hand, greater uncertainty emanating 30 percent (y/y) in October to 6.0 and 14.2 per- from the Eurozone would lead to spikes of risk cent in June, respectively. Producer price infla- aversion, higher volatility of the exchange rate, tion also subsided over the same period, falling and potentially a further depreciation of the to single-digit annual increases. A similar down- rand. Depending on several factors, this may ward trend can be observed in the Kagiso PMI lead to increases in imported good prices and 10 price index. After remaining in the low 70s second-round effects (see box 1.1 on exchange in the previous four months, 10 index points rate pass-through in South Africa in the South below the monthly average recorded in the last African Economic Update of November 2011). quarter of 2011, the Kagiso price index fell to On balance, the risks to the inflation outlook 65.1 in June, also suggesting a moderation of have abated in recent months. Together with input costs in the manufacturing sector. the weakened growth outlook, this justifies the Greater uncertainty South Africa’s inflation outlook remains South African Reserve Bank’s July 19 decision subject to domestic and global risks. On the one to lower the repurchase rate from 6.0 percent emanating from hand, inflationary pressures have thus far orig- to a record low 5.5 percent, its first interest rate the Eurozone would inated on the supply side, mostly from global cut in 20 months. commodity prices. Demand-driven inflation- lead to spikes ary pressures have remained subdued, as the External sector of risk aversion, domestic economy is yet to fully recover from The current account deficit rose to 4.9 percent the 2009 recession and capacity utilization in of GDP in 2012q1 from 3.6 percent in 2011q4 higher volatility of the manufacturing sector remains below the as the deficits in the trade account and in the the exchange rate, levels before the global financial crisis (79.9 services, income, and current transfers account percent in 2012q1, compared with 85.3 percent widened (figure 1.9). The deficit in the trade and potentially a in 2007q1). A deteriorating global economic account edged higher in 2012q1, rising to 1.4 further depreciation outlook is also likely to restrain domestic con- percent of GDP and reinforcing the downward sumer and business confidence, further damp- trend that started in 2011q1. The terms of trade of the rand ening prices through lower external demand maintained in 2012q1 the downward trend (with the Eurozone still being South Africa’s that started in early 2011 as increases in inter- major trading partner and the main recipi- national oil prices outpaced the increases in ent of its manufactured goods). Moreover, nonoil commodity prices. The services, income, Core and headline Consumer Price Index inflation and Producer Price Index inflation, Figure 2009–12 1.8 Food Excluding food, nonalcoholic beverages, petrol, and energy PPI in ation for domestic output Headline CPI in ation Petrol 20 40 15 20 Percent Percent 10 0 5 –20 0 –40 Jan. 2009 Jan. 2010 Jan. 2011 Jan. 2012 May 2012 Source: Statistics South Africa. Figure Current account, trade balance, and terms of trade, 2007–12 1.9 Trade balance Savings to investment rate Current account balance Terms of trade, including gold 5 130 0 120 Percent Percent 11 –5 110 –10 100 January January January January January January 2007 2008 2009 2010 2011 2012 Source: South African Reserve Bank 2012; staff calculations. GDP growth is projected at Figure Bond and equity purchases by nonresidents 2.5 percent in 2012, 1.10 Bonds Equities Rand/US$ lower than the Bonds and equities exchange rate 200 10 3.1 percent forecast in the November 100 9 Rand/US$ exchange rate 2011 update and the US$ millions 0 8 2.7 percent forecast in the 2012 Budget -100 7 -200 6 July 19, Sept. Nov. Jan. Mar. May June 25, 2011 2011 2011 2012 2012 2012 2012 Source: Citigroup; Johannesburg Stock Exchange; staff calculations. and current transfers account also deteriorated Economic outlook for South Africa due to higher net dividend payments to nonres- Against the prospects of a noticeably weaker idents, reinforcing the net effect on the current global economy than envisaged at the time account and, thus, the country’s dependence of the previous Economic Update in Novem- on foreign savings. ber 2011, medium-term growth projections As noted earlier, risk aversion has spiked in for South Africa have been revised downward recent months, leading to a general flight to (table 1.6). GDP growth is projected at 2.5 per- safety, with lower capital flows in general to cent in 2012, lower than the 3.1 percent fore- developing countries and widespread deprecia- cast in the November 2011 update and the tion of emerging market exchange rates against 2.7 percent forecast in the 2012 Budget. The the dollar. Against this backdrop, the South main reason for the lower projection is the African rand has also weakened against the dol- further slowdown in the European Union and lar, reaching levels not seen since 2011q4. Bond in China, the two main export destinations and equity flows have recovered somewhat for South Africa. Under the baseline assump- from their plunge in the latter part of May, but tions applied here, the current global finan- remain below the levels in April 2012. cial market tensions would ease somewhat South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Table Macroeconomic outlook, 2008 –14 1.6 (percent change, unless otherwise indicated) 2008 2009 2010 2011 2012 2013 2014 Real GDP growth 3.6 –1.5 2.9 3.1 2.5 3.2 3.5 Household consumption 2.2 –1.6 3.7 4.9 3.4 3.5 3.7 Government consumption 4.5 4.7 4.9 4.6 4.1 4.1 4.1 Gross fixed investment 13.3 –3.2 –1.6 4.3 3.8 4.2 5.0 Exports, GNFS 1.8 –19.5 4.5 6.0 2.4 5.0 5.5 12 Imports, GNFS 1.5 –17.4 9.6 9.4 6.8 6.8 7.9 Headline CPI 9.9 7.1 4.3 5.0 6.0 5.3 5.1 Current account balance (percent of GDP) –7.2 –4.0 –2.8 –3.3 –4.3 –4.7 –4.8 Source: Staff calculations; National Treasury of the Republic of South Africa estimates and projections. in 2013 and 2014, contributing to a pickup in capital spending by the government and pub- Domestic demand, global demand, which should support gradual lic corporations. acceleration of growth in South Africa toward Consistent with the public sector’s ambi- particularly consumer its potential GDP growth of 3.5 percent. An tious plans to boost spending on infrastruc- spending, will continue important constraint on a faster pickup in ture projects, and with persistent pressures to growth is likely to be bottlenecks in electricity increase social spending, government spend- to be an important supply, which is already rubbing against peak ing will remain critical over the first two years driver of growth demand and will continue to do so until fresh of the forecast period. But the planned fiscal large-scale generation capacity comes onboard consolidation would tend to reduce govern- and Eskom’s demand-side management mea- ment spending’s contribution to growth in sures for greater energy efficiency take firmer the medium term. Due to weak demand from root. Europe, slowing demand from Asia, and strong Domestic demand, particularly consumer import growth, net exports will place a drag spending, will continue to be an important on growth in 2012. However, a pickup in global driver of growth—albeit at a less forceful pace demand in the outer years of the forecast than previously envisaged. Consumer spend- should support an increase in export growth in ing will continue to be supported by robust those years. wage increases, historically low interest rates, and expanded access to unsecured credit. Risks to the outlook Retail sales, an indicator of the strength of The major downside risks remain tied to how consumer spending, were up 5.9 percent (y/y) the situation in Europe unfolds. Europe is in 2012q1. But there are signs of a moderation South Africa’s largest export market, so indus- in 2012q2, as picked up by a sharp decline in trial production growth in the two economies the RMB/BER Business Confidence Index for is strongly correlated (0.78; figure 1.11). the retail sector. Persistent high unemploy- The form that an escalation of the crisis ment, elevated consumer debt, and stable or might take in the current economic context, if declining housing prices will put a check on one occurs, is very uncertain—partly because it the expansion in consumer spending. After is impossible to predict what exactly might trig- two years of negative growth, private invest- ger it, and partly because the powerful forces ment expanded 4.4 percent in 2011. The unleashed could easily take a route very dif- fragile domestic economic outlook, the deteri- ferent from the one foreseen by standard eco- orating business conditions, the excess capac- nomic reasoning. ity still prevailing in the manufacturing sector, With that caveat, for purely illustration pur- and the weak external demand do not provide poses, two hypothetical downside scenarios are a supportive environment for private invest- considered here. ment. So, private investment is projected to • Scenario 1. The first scenario assumes a moderate in 2012, offset only partly by higher serious credit squeeze faced by one or two South Africa’s industrial production growth is highly correlated to developments in Figure Europe 1.11 (industrial production volumes, 3m/3m seasonally adjusted annual rate) South Africa Eurozone 30 20 10 0 13 Percent –10 –20 –30 –40 Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. 2005 2006 2007 2008 2009 2010 2011 2012 South Africa’s large Source: World Bank; Datastream. current account deficit continues to be a small Eurozone economies, but an other- the deceleration in its GDP growth much wise fairly orderly resolution to the crisis sharper than in other large developing coun- downward risk to that limits contagion from spilling beyond tries (figure 1.12). Under Scenario 1, 1.5 per- the growth outlook these economies. Although the scenario centage points would be shaved off South assumes that borrowing costs increase in Africa’s GDP growth in 2012 relative to the other European countries and that banks baseline growth of 2.5 percent. Under Sce- tighten lending conditions due to losses nario 2, some 2.2 percentage points would be in the directly affected economies and to taken off GDP growth in 2012. heightened uncertainty and risk aversion, Lower commodity prices present another the banking stress is contained and does downside risk. With much of the recent not spread to the rest of the high-income increase in commodity prices driven by strong world. The main direct transmission chan- demand from Asia (particularly China), a nel to the rest of the world (including cooling of the Chinese economy beyond pro- South Africa) would be through weaker jections, coupled with weak demand from demand for its exports. However, the indi- Europe, would dampen commodity prices, rect channels of transmission arising from hitting South Africa’s growth prospects. To the confidence effects are likely to be evaluate this, the impact of a 20 percent fall in even stronger—with precautionary savings nonoil commodity prices is simulated (figure among firms increasing and heightened 1.13). Once again, among developing coun- risk aversion delaying spending by firms. tries, South Africa is among the 10 countries As a result, both consumer and investment to be hit most by the drop in commodity prices. demand would decline. Indeed, the results show GDP growth to fall by • Scenario 2. This scenario assumes a disorderly some 1.7 percentage points. resolution to the crisis. A freezing of credit South Africa’s large current account defi- is assumed to spread to two larger Eurozone cit (reflecting low domestic savings), financed economies, generating similar declines in mostly through portfolio flows, continues to be GDP and imports of these economies. Euro- a downward risk to the growth outlook. A dis- zone GDP is assumed to fall 8.5 percent rela- orderly resolution of the Eurozone crisis may tive to the baseline in 2013, pushing the rest heighten volatility in financial markets and dry of the world into a deep recession. up capital inflows, with portfolio flows mostly Simulations of the two scenarios show a affected by swings in market sentiment. Greater major potential impact on South Africa, with volatility in the rand exchange rate—and a South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Percentage deviation in GDP growth in 2012 from baseline in the event of a further Figure escalation in the Eurozone crisis 1.12 Small crisis Big crisis 0.0 –0.5 –1.0 Percent –1.5 14 –2.0 –2.5 –3.0 Kenya Brazil India Malaysia Nigeria Turkey China South Russian Africa Federation Source: World Bank. An upside risk for South Africa and other Commodity exporters to be hardest hit by a 20 percent decline in metal and mineral Figure prices developing countries 1.13 Paraguay is the possibility of Uruguay Argentina a stronger recovery Kyrgyz Republic Belize Chile in global demand Uzbekistan Bolivia South Africa Uganda Colombia Benin Swaziland Namibia –8 –6 –4 –2 0 Percent deviation from GDP forecast in 2013 Source: World Bank. sharp correction in domestic absorption and embedded in the current baseline. While less the current account—would likely drag down likely, this could come from an improvement the fragile domestic economic recovery. More in market sentiment, perhaps due to additional than 20 percent of total gross fixed capital progress on the reform agenda or to better- formation was financed by nonresidents in than-anticipated outturns in high-income 2012q1, up from 14 percent in 2011q4, showing countries. For developing countries such the domestic economy’s increased vulnerability as South Africa where some postcrisis slack to capital flow reversals. remains, a stronger than expected recovery An upside risk for South Africa and other in demand could be absorbed fairly easily and developing countries is the possibility of a converted into improved living conditions and stronger recovery in global demand than lower unemployment. SECTION 2 The state of human opportunity in South Africa South Africa, the continent’s largest economy year since 1995 (1.6 percent per capita), it has by far and its only G-20 member, displays proven insufficient to absorb the wave of new strikingly high and persistent inequality and entrants to the labor market from dismantling marginalization for an upper middle-income apartheid’s barriers. The potential for growth country. The contradictions, a legacy of the has been held back by industrial concentra- country’s apartheid past, are on display mostly tion, skill shortages, labor market rigidities, and along racial lines and spatially demarcated chronically low savings and investment rates— boundaries. Peering past the first-world living the latter, despite the fairly high and improving conditions of urban South Africa, it is not too rates of return to capital.8 hard to see the downcast situation of town- Growth has also been highly uneven in ships, informal settlements, and former home- its distribution, perpetuating inequality and lands, a large minority of whose residents lack a exclusion. With an income Gini of around 0.70 job or the means to even look for one, being so in 2008 and consumption Gini of 0.63 in 2009, spatially disconnected from market access and South Africa stands as one of the most unequal employment opportunities.7 countries in the world (figure 2.1).9 The top While GDP growth—if modest by global com- decile of the population accounts for 58 per- parisons—has averaged a credible 3.2 percent a cent of the country’s income, while the bottom Figure GDP per capita and the Gini index for consumption 2.1 All countries African countries South Africa 80 South Africa, 2006 South Africa, 1993 South Africa, 2009 60 South Africa, 2000 South Africa, 1995 Gini index 40 20 0 5 6 7 8 9 10 11 12 Log of GDP per capita, PPP (2005 international $) Source: World Bank 2012. 15 South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity decile accounts for 0.5 percent and the bottom Noncontributory and means-tested (except half less than 8 percent.10 In large part, this is for foster care) financial transfers from the an enduring legacy of the apartheid system, budget account for more than 70  percent of which denied the non-whites (especially Afri- the income of the bottom quintile (up from cans) the chance to accumulate capital in any 15 percent in 1993 and 29 percent in 2000).14 form—land, finance, skills, education, or social With the social grants, the entire spectrum of networks. population ranked by income percentiles saw At the heart of high inequality lies the inabil- income growth between 1995 and 2005 (figure 16 ity to create employment opportunities on a 2.3). But without the grants as part of income, large enough scale.11 Unemployment stands at those below the 40th percentile saw a signifi- 25.2 percent (33.0 percent, including “discour- cant decline in their income. In other words, aged� workers), among the world’s highest. No without the grants, two-fifths of the population surprise then that despite an almost 30 percent would have seen its income decline in the first increase in per capita GDP since the late 1990s, decade after apartheid.15 reductions in poverty have been modest at best.12 Even after accounting for the equalizing At the heart of This would have been untenable without the role of social assistance, income inequality growing social assistance grants (figure 2.2).13 remains extraordinarily high.16 To reduce it to high inequality lies the inability to Figure Income sources by deciles create employment 2.2 Labor market Remittances Capital Government Other 100 opportunities on a 80 large enough scale 60 Percent 40 20 0 Bottom Ninth Eighth Seventh Sixth Fifth Fourth Third Second Top Quintile Source: Leibbrandt and others 2010. Figure Growth incidence curves for South Africa, 1995–2005 2.3 Including social grants Excluding social grants Annual growth in per capita income (percent) 20 10 0 –10 –20 0 20 40 60 80 100 Poorest percentile by per capita income (percent) Source: Bhorat and van der Westhuizen 2011. more reasonable levels over the long run, social basic set of goods and services during child- assistance is clearly not enough and needs to hood can be an important (if far from perfect) be complemented by other initiatives. These predictor of future outcomes, including edu- would include a special focus on human capital cation achievements and earnings. Access to development, particularly among children and quality basic services such as education, health youths. care, essential infrastructure (like water, sani- The high and possibly rising patterns of tation, and electricity), and early childhood inequality predictably polarize the political development provides an individual, irrespec- and economic debate in the country. While tive of background, the opportunity to advance 17 the perils of an unequal society are univer- and reach his or her human potential. sally acknowledged, the notion of unfairness Analyzing such opportunities for children requiring policy intervention rarely finds con- in South Africa can help better understand the sensus. Some argue, with justification, that nature and causes of inequality of outcomes individuals cannot and should not be rewarded observed among adults. Opportunities among equally irrespective of their effort, choices in children can also be reliable predictors of eco- life (whether to pursue higher education, for nomic mobility across generations and over Social consensus and example), or innate talent. If a person works time. For instance, if access to economic oppor- policy action can harder than another, it is only fair that that tunities, in the form of jobs (and earnings), person be suitably compensated; the same for credit, and ownership of land and financial become more likely someone who happens to be more talented assets is correlated with the circumstances of than another. an individual (such as race and location of resi- when the issue is There is much broader agreement on the dence), it reinforces the link between children’s framed around the notion that predetermined circumstances such circumstances and their opportunities in life. as gender, ethnicity, place of birth, or family Until recently, framing the debate in such notion of equity origins should not play a role in determining terms had not been possible in developing rather than equality people’s economic, social, and political success. countries because of lack of an intuitive mea- In other words, a person should not have fewer sure of equality of opportunity among chil- opportunities in life just because she is a girl dren (akin to the Gini coefficient) that could or just because she is born of a certain color. be readily applied to the data typically avail- This is the core principle behind the concept able in a developing country. New techniques of equality of opportunity—the framework in such as the Human Opportunity Index (HOI) this report for South Africa.17 at the World Bank in partnership with others, Social consensus and policy action can and their application in the Latin America become more likely when the sources of and Caribbean region in the last decade, have inequality are decomposed and the issue is helped fill some of that gap (box 2.1). This framed around the notion of equity rather report applies some of these techniques to bet- than equality. A girl (let’s call her Thandiwe) ter understand the underpinnings of inequal- born in the township of Tembisa outside ity in South Africa. Johannesburg to a single, uneducated mother The first part of this focus section limits the earning R2,000 a month and with four other list of opportunities to those provided to an siblings should have an equal shot at becoming individual in childhood. For a child, opportu- a doctor or an engineer as a boy (let’s call him nities are synonymous with access to (and use Andries) with one sibling, born in a two-parent of) basic goods or services, such as basic educa- household in Sandton, Johannesburg (see box tion, health, safe water on site, and improved 2.3 below). And so should a girl (let’s call her sanitation, while “individual effort� is mostly Nothando), born in a family with similar char- irrelevant, because the family, society, or the acteristics as Thandiwe’s, but living in the rural government (and not the child herself) are area of Eshowe in Kwa-Zulu Natal. responsible for ensuring whether she will have What would it take to equalize Nothando’s, access to them. The focus on children also Thandiwe’s, and Andries’s chances of success has implications for public policy. Academic in life? Ample research shows that access to a research has found interventions that equalize South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Box The Human Opportunity Index 2.1 The academic literature offers a number of ways to measure inequality of opportunity. Among these, our purpose is best served by a scalar measure easily computable from the typical data available in developing countries. Developed by Bank staff in collabora- tion with external researchers, the Human Opportunity Index (HOI) is an intuitive measure of a society’s progress toward the equi- table provision of opportunities for all children.1 It has been used in two regional reports in Latin America, a study (in draft) for 20 Sub-Saharan countries, and numerous studies conducted (or in progress) for countries around the world. The focus of most stud- ies has been children, with some applications to labor markets in recent studies. The HOI measures in a single indicator the coverage rate of a particular service, adjusted by how equitably the available services 18 are distributed among groups differentiated by circumstance. Its construction involves aggregating circumstance-specific coverage rates in a scalar measure that increases with overall coverage and decreases with the differences in coverage among groups with different sets of circumstances. The index runs from 0 to 100. A society that has universal coverage in a particular opportunity (say, primary school enrollment) would score 100. By contrast, a society that has an average primary enrollment of 50 percent that is unequally distributed in favor of children of certain circumstances (say, urban children) will have an HOI below 50, with the exact value depend- ing on how unequal enrollment is among children of different circumstances (see annex A). The measure also has a number of desirable and intuitive properties (see annex B). Interventions that All results are subject to the caveat that the HOI is estimated for a specified list of the circumstances, which could change if this list were to change. But the HOI for any opportunity cannot be higher if more circumstances are added to the existing list. Thus equalize opportunities if a society wants to measure equality of opportunity for a larger number of circumstances (and groups) than considered here, the HOI here will serve as an upper bound to the “true� HOI that would take all circumstances into account. To compute the HOI for a earlier in life are much certain opportunity for the children of a country, household survey data are essential, with information at the individual (child) and more cost-effective household level. Note: and successful than 1. Barros and others 2009. those later in life opportunities earlier in life to be much more across generations. Of these, only employ- cost-effective and successful than those later in ment is covered in the second part, because of life.18 data limitations for the other dimensions and The report poses three key questions for the the centrality of high unemployment in South equality of opportunities among South African Africa. The results on opportunities in employ- children: ment are less definitive than those for children • What is the status of South African chil- and need to be treated with caution, due to the dren’s access to basic opportunities (educa- conceptual difficulties in measuring inequal- tion, health, water, sanitation, and so on) in ity of opportunity among adults with the data coverage and distribution among children limitations in South Africa. of different circumstances? The main questions posed in the second • How has South African children’s access part are: to opportunities changed over time, and • What is the status of opportunities in the to what extent is the change attributable South Africa labor market in coverage and to changes in the scale and inequality of inequality, and how have these changed over coverage? time? • What circumstances shape inequality of • What are the “contributions� of circum- opportunity in South Africa and, conse- stances relative to acquired attributes to quently, what are the profiles of the most inequality in employment status, such as vulnerable groups among South African education and age (or experience)? children? • How do labor market opportunities The second part of the focus section exam- vary among different subgroups of the ines inequality of opportunities in employment, population? which encompasses a number of dimensions (in addition to the basic opportunities in child- Opportunities among children hood)—including employment, earnings, Basic opportunities here are defined as a sub- assets, and credit—that perpetuate inequality set of goods and services for children, such as access to education, safe water on site, or elec- status—whether both parents are alive; and tricity, that are critical in determining opportu- location of the household—urban townships, nities for economic advancement in life. These informal settlements, other urban areas, or either are already affordable by society at large, rural areas.22 Note that, in covering the town- or could be in the near future, given the avail- ships and informal settlements, a broad brush able technology. Universal provision of basic overlooks their remarkable variety in size, pop- opportunities is a valid and realistic social goal. ulation density, length of time since establish- Opportunities among children are mea- ment, rural or urban location, and distance sured in this report by the Human Opportu- from major urban centers. 19 nity Index (HOI), which is the coverage rate of The opportunities considered are those that a particular basic service adjusted by how equi- enable South African children to realize their tably the service is distributed among groups productive potential directly, by enhancing differentiated by circumstances. In discount- their human capital and creating a safe physi- ing inequitable access, the HOI reflects how cal environment, and indirectly, by providing personal circumstances—for which children access to infrastructure amenities that help cannot be held accountable—affect their basic ensure a decent quality of life and facilitate the Access to education, opportunities. This means that two societies accumulation of human capital (see annex D safe water on site, with the same coverage rate for any service can for details).23 The human development compo- have different HOIs if citizen access to that ser- nent of opportunities comprises exposure to and electricity are vice in one society is determined to a greater early childhood development programs (chil- extent by gender, race, family background, or dren ages 0–4 years), school enrollment (at critical in determining any other personal circumstance beyond their ages 6–11 and 12–15), timely completion of pri- opportunities control and considered by society to be an mary school (ages 13–15), adequacy of school unjust source of exclusion.19 infrastructure and teachers (as reported by for economic The HOI therefore is a unique representa- parents), and having health insurance. Objec- advancement in life tion of progress toward universal coverage of tive measures of student achievement or quality basic opportunities and the fairness of allocat- of learning were not available. ing those opportunities, in a single indicator.20 Exposure to early childhood development It brings equity to the forefront of policy- (ECD) is a valuable (if partial) measure of making with an operational measure to track cognitive inputs early in a child’s development progress. process, which research has shown to have What determines the choice of the circum- long-term impacts on learning.24 stances to be taken into account? There are no Having health insurance, an indirect mea- universal selection criteria. The determination sure of access to quality health services, has is best made by societal consensus based on recently received considerable policy attention what the society considers, for ethical reasons, in South Africa with the envisaged National as characteristics that should not interfere with Health Insurance program. Ideally, this mea- access to basic goods and services. From a prac- sure would be complemented with more mean- tical point of view, the choice inevitably is also ingful data on child health in South Africa guided by the kinds of data available, particu- such as nutritional outcomes, access to preven- larly nationally representative household sur- tive care, and access to pre- and postnatal care vey data that is essential for the analysis.21 for mothers, which are absent in the dataset For South Africa, the circumstances are for this analysis. A child’s physical well-being personal and family-related: gender of the also depends on access to safe water on site and child; ethnicity; household composition—pres- improved sanitation. ence of the spouse of the household head in Infrastructure services include access to the household, total number of children ages electricity, safe water on site, and improved 0–16 in the household, and whether both par- sanitation and living in households that are ents live in the household; education of the not overcrowded—all critical for a safe, stable, household head; other household head charac- and a stimulating childhood. Access to safe teristics—gender and age of the head; orphan water on site, for example, is known to reduce South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity the risk of water-borne diseases, the leading developing countries, including many with cause of illness and undernourishment in similar levels of development. Recognizing the children, which in turn affect their education limited coverage of similar analysis for other outcomes and earning potential.25 Electricity countries, the comparisons are restricted to enables nighttime reading and is a healthier Latin America and a few large African coun- source of energy than fuel, and it opens up tries and to five measures of opportunity, access to other opportunities, such as sources with five HOIs and the associated inequality of information (radio, television, and Inter- of access measured by the dissimilarity index 20 net). Improved sanitation reduces the risk of (D-Index; figures 2.4–2.8 and box 2.2). illness. Overcrowding has detrimental effects South Africa does extremely well in terms on mental health, social relationships at home, of school attendance for children ages 10–14, and child care.26 Access to telecommunications where its HOI exceeds that of almost all the enhances exposure to information, the ability comparator countries and the D-Index is rela- to communicate, and even access to such ser- tively small (see figure 2.4). However, in a com- vices as health care and education. parison of school performance (completion of South Africa does The opportunity to grow up in a safe neigh- primary school on time), South Africa is sur- borhood is important for a child’s cognitive passed by a comfortable margin by most of its extremely well in and psychological development. The salience Latin American peers: in fact, it is more in line terms of school of safety in the socioeconomic fabric of the with the other African countries and the poor- country is perhaps best illustrated by the fact est Latin American countries, in terms of a high attendance for that crime and theft were regarded as the top D-Index and relatively low HOI (see figure 2.5). children ages 10–14 five problematic factors of doing business in The HOIs for access to safe water on site and South Africa—more important than access to improved sanitation show that South Africa, finance, infrastructure, and tax regulations.27 though superior to the other African countries, Unsafe areas also tend not to have quality is well behind most Latin American countries schools, hospitals, and other such amenities. except for the poorest ones. For example, on access to safe water and improved sanitation, What is the status of South African South Africa’s HOIs are on par with those children’s access to basic opportunities? for El Salvador and Honduras, respectively, It is useful to begin by comparing South Afri- with comparable D-Indices. In other words, can children’s status with that of peers in other children’s access to safe water and improved Figure School attendance (ages 10 –14) 2.4 D-Index HOI 20 100 15 75 Percent Percent 10 50 5 25 0 0 Jamaica (2002) Guatemala (2006) Honduras (2006) Nicaragua (2005) Ecuador (2006) El Salvador (2007) Panama (2003) Paraguay (2008) Mexico (2008) Colombia (2008) Venezuela, RB (2005) Uruguay (2008) Peru (2008) Costa Rica (2009) Dominican Republic (2008) Argentina (2008) Brazil (2008) Chile (2006) Nigeria (2008) Ghana (2008) Zambia (2007) Zimbabwe (2010–11) Namibia (2006–07) Kenya (2008–09) South Africa (2010) Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. Figure Finished primary school on time 2.5 D-Index HOI 40 100 30 75 Percent Percent 20 50 21 10 25 0 0 Jamaica (2002) Guatemala (2006) Brazil (2008) Nicaragua (2005) El Salvador (2007) Honduras (2006) Dominican Republic (2008) Paraguay (2008) Costa Rica (2009) Colombia (2008) Panama (2003) Venezuela, RB (2005) Peru (2008) Uruguay (2008) Ecuador (2006) Chile (2006) Argentina (2008) Mexico (2008) Zambia (2007) Ghana (2008) Nigeria (2008) Kenya (2008–09) Namibia (2006–07) Zimbabwe (2010–11) South Africa (2010) The HOIs for access to safe water on site and Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. improved sanitation show that South Figure Access to safe water on site Africa is well behind 2.6 D-Index HOI most Latin American 80 100 countries except for 60 75 the poorest ones Percent Percent 40 50 20 25 0 0 Jamaica (2002) Nicaragua (2005) Honduras (2006) El Salvador (2007) Peru (2008) Panama (2003) Colombia (2008) Guatemala (2006) Paraguay (2008) Ecuador (2006) Dominican Republic (2008) Mexico (2008) Brazil (2008) Venezuela, RB (2005) Uruguay (2008) Chile (2006) Costa Rica (2009) Argentina (2008) Nigeria (2008) Ghana (2008) Zambia (2007) Kenya (2008–09) Zimbabwe (2010–11) Namibia (2006–07) South Africa (2010) Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. sanitation is lower and much more influenced In the international comparisons, only a by circumstances in South Africa than most limited set of circumstances—gender, family Latin American countries, including Brazil. structure, socioeconomic status of parents, and South Africa does relatively better on access location—and opportunities was used due to to electricity, but still ranks below a majority of data limitations for the other countries and the these countries on HOI. While coverage is not need to ensure comparability. These are now universal in South Africa, a low D-Index indi- expanded to cover the full range of informa- cates that existing coverage is equitably distrib- tion available for South Africa for a deeper uted across children of different circumstances. look at the country. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Figure Access to improved sanitation 2.7 D-Index HOI 80 100 60 75 Percent Percent 40 50 22 20 25 0 0 Jamaica (2002) El Salvador (2007) Guatemala (2006) Honduras (2006) Panama (2003) Nicaragua (2005) Dominican Republic (2008) Paraguay (2008) Ecuador (2006) Peru (2008) Argentina (2008) Mexico (2008) Colombia (2008) Brazil (2008) Venezuela, RB (2005) Chile (2006) Costa Rica (2009) Uruguay (2008) Kenya (2008–09) Ghana (2008) Nigeria (2008) Zambia (2007) Zimbabwe (2010–11) Namibia (2006–07) South Africa (2010) The high coverage rates conceal variable Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean quality of education, countries; and Demographic and Health Surveys for African countries. widely recognized to be a major issue Figure Access to electricity in South Africa 2.8 D-Index HOI 80 100 60 75 Percent Percent 40 50 20 25 0 0 Jamaica (2002) Nicaragua (2005) Honduras (2006) Panama (2003) Guatemala (2006) Peru (2008) El Salvador (2007) Ecuador (2006) Paraguay (2008) Dominican Republic (2008) Brazil (2008) Mexico (2008) Uruguay (2008) Venezuela, RB (2005) Costa Rica (2009) Chile (2006) Argentina (2008) Colombia (2008) Kenya (2008–09) Zambia (2007) Zimbabwe (2010–11) Namibia (2006–07) Nigeria (2008) Ghana (2008) South Africa (2010) Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. Several features of South African children’s The high coverage rates nevertheless con- access to opportunity stand out (figure 2.9). ceal variable quality of education, widely recog- On the positive side, as noted, the country is nized to be a major issue in South Africa. The very close to universal coverage in primary issues of quality are brought out by standard- education, with almost uniformly high school ized international test scores that consistently attendance and healthy measures of opportuni- rank South Africa low relative to global com- ties for school infrastructure and numbers of parators.28,29 The survey data for this analysis teachers. For each of these opportunities, there do not cover test scores for children.30 When an is little inequality in accessing them, evident indirect (and imperfect) measure of school per- from the extremely low D-Indices. formance is used—whether a child completed Box Measuring inequality of opportunity 2.2 The dissimilarity index (D-Index) is the measure of inequality of opportunity used in the analysis. It is the ratio of the penalty due to inequality to the coverage rate of a particular basic good or service. Intuitively, the D-Index measures the share of available opportunities that needs to be reallocated across circumstance groups in order to achieve equality of opportunity. A D-Index above 5 percent indicates non-negligible inequality of opportunity, and one above 10 can be considered problematic for policymaking. primary school on time—the result is similarly For infrastructure services, except telecom- discouraging, with an HOI of only 51 percent. munications, inequalities are generally high, 23 The D-Index for finishing primary school on evidenced by a D-Index that varies between 6 time (at 10 percent) indicates some disadvan- percent (access to electricity) and 23 percent tage among certain groups because of their (access to safe water on site). The opportunity circumstances. Delaying completion of (or fail- to live and grow up in a household with access ing to complete) primary school has important to safe water on site, for example, is limited to bearing on lifetime educational attainment of 63 percent of children ages 0–15, with close to a a child and thus represents significant disad- quarter of those opportunities being inequitably Opportunities are vantages for opportunities in later life. allocated among circumstance groups (D-Index limited in exposure Opportunities are also limited in the two of 23 percent), which produces an HOI of 48 per- other human development–related areas: expo- cent. Access to improved sanitation is similarly of children to any sure of children to any kind of ECD program limited, with only 67 percent of children living between ages 0 and 4 and the chances for a in households that have either a flush toilet (con- kind of ECD program child to be covered by health insurance. The nected to public sewerage or a septic tank) or a between ages 0 and D-Index is extremely high at 47 percent for chemical toilet or pit latrine. A high D-Index of the latter, reflecting the major role of extra- 18 percent again suggests that the access to this 4 and the chances for neous circumstances in children’s access to opportunity is influenced by the children’s cir- a child to be covered health care. cumstances, resulting in a low HOI of 55 percent. by health insurance Figure HOIs and D-Indices for some key opportunities for South African children, 2010 2.9 D-Index HOI 50 100 40 80 30 60 Percent Percent 20 40 10 20 0 0 Have health insurance Access to safe water on site No overcrowding Finish primary school on time (ages 13–15) Good environmental Access to improved sanitation ECD exposure Safe location Access to electricity Adequate information in school Access to telecommunications Adequate teachers in school School attendance (ages 12–15) School attendance (ages 6–11) Source: Authors’ calculations based on General Household Survey (2010). South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity A significant proportion of children ages in school, and growing up in households that 0–15 years (36 percent) live in areas not con- are not overcrowded showed much more mod- sidered safe. A D-Index of 4 percent produces est gains. an HOI (61 percent), not very different from Changes in the HOI are decomposed into the coverage rate. So, an overall lack of per- a composition effect, scale effect, and equal- ceived safety is the problem, as opposed to high ization effect (see figure 2.10). The first refers inequality, in how these conditions affect chil- to the change in the distribution of circum- dren of different circumstances. stances that can be the result of broad demo- 24 graphic changes, economic growth, or social How has South African children’s access progress. The second refers to proportional to opportunities changed over time? or parallel changes in coverage rates for all The HOIs for South Africa are well short of groups, perhaps as a result of a broad-based universal coverage in many cases and generally public policy. The third signifies a change in are either at par with or behind comparator the coverage rate for the vulnerable groups for countries in Latin America, a region marked by a given overall coverage, indicating the equity The most noticeable high inequality. But have the HOIs improved trend in society, perhaps as a result of targeted over time? They appear to have measurably policies. progress appears to improved for each of the opportunities since For the opportunities with the largest have been in access to the early 2000s, if with sharp variations in their improvements, most of the gains were driven progress (figure 2.10).31 by the scale effect—a general expansion of the improved sanitation, The most noticeable progress appears to coverage rate of services across children of all electricity, and have been in access to improved sanitation, circumstances. But for the opportunities of electricity, and telecommunications—for which improved sanitation and telecommunications, telecommunications the HOI increased by more than 50 percent- an appreciable gain was also driven by the age points—and adequate infrastructure in equalization effect with opportunities improv- school (as reported by parents). Opportunities ing more than proportionately for the weaker corresponding to access to safe water on site, circumstance groups. The composition effect access to health insurance, school attendance was small in most cases, highlighting the slow- for 12–15-year-olds, having adequate teachers moving changes in the relative circumstances Figure Change in the HOI and decomposition of changes, 2002–10 2.10 Equalization Scale Composition Total change (2002–10) 70 60 Percentage point change 50 40 30 20 10 0 Have health insurance School attendance (ages 12–15) Adequate teachers in school Access to safe water on site Finish primary school on time School attendance (ages 6–11) Adequate infrastructure in school Access to improved sanitation Access to telecomunications No overcrowding Access to electricity Source: Authors’ calculations based on General Household Surveys (2002 and 2010). of children. But it did account for 66 percent (which precluded rapid improvement), that of the (small) improvement in access to health comparison was not included. insurance and 60 percent of the improvement Except for electrification, where it has an in access to safe water on site. outstanding record on improving access, the How does South Africa’s progress compare rates at which opportunities have expanded with that of other countries? Figures 2.11–2.14 in South Africa are distinctly lower than the show the average annual percentage point majority of Latin American countries and com- change in the HOI for 23 of the 25 countries parable with other Sub-Saharan countries. On shown in figures 2.4–2.8, for four opportuni- the annual average rate of HOI improvement 25 ties: finishing primary school on time and in finishing primary school on time and in access to safe water on site, improved sanita- access to safe water and improved sanitation, tion, and electricity.32 Given the already high South Africa ranks 16th, 14th, and 11th, respec- HOI for school attendance in South Africa tively, of 23 countries. Figure Average annual change in the HOI for finishing primary school on time For the opportunities 2.11 Namibia Peru Colombia with the largest Dominican Republic Honduras Mexico El Salvador improvements, most of Zimbabwe Brazil Kenya Uruguay the gains were driven Guatemala Paraguay Venezuela, RB by a general expansion Zambia South Africa Chile Costa Rica of the coverage Jamaica Panama Ghana rate of services Argentina Nigeria –1 0 1 2 3 4 across children of Percentage point change all circumstances Note: The surveys are for different years across countries. The average survey years are 1997 and 2007, and the average period between two survey years for a country is nine years. These are broadly comparable with South Africa: surveys from 2002 and 2010 with an elapsed time period of eight years. Average annual changes in the HOI (in percentage points) are considered to make the changes comparable across countries and independent of the base period HOI. Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. Figure Average annual change in the HOI for access to safe water on site 2.12 Mexico Paraguay Uruguay Panama Brazil Namibia Guatemala Chile Honduras Dominican Republic Peru Costa Rica Argentina South Africa Zimbabwe Venezuela, RB El Salvador Ghana Kenya Colombia Zambia Nigeria Jamaica –1 0 1 2 3 4 Percentage point change Note: The surveys are for different years across countries. The average survey years are 1997 and 2007, and the average period between two survey years for a country is nine years. These are broadly comparable with South Africa: surveys from 2002 and 2010 with an elapsed time period of eight years. Average annual changes in the HOI (in percentage points) are considered to make the changes comparable across countries and independent of the base period HOI. Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Figure Average annual change in the HOI for access to improved sanitation 2.13 Brazil Peru Mexico Chile Costa Rica Guatemala Dominican Republic Paraguay Colombia Argentina South Africa Namibia Venezuela, RB Jamaica Panama Ghana 26 El Salvador Zimbabwe Uruguay Nigeria Kenya Zambia Honduras –1 0 1 2 3 4 Percentage point change Note: The surveys are for different years across countries. The average survey years are 1997 and 2007, and the average period between two survey years for a country is nine years. These are broadly comparable with South Africa: surveys from 2002 and 2010 with an elapsed time period of eight years. Average annual changes in the The importance of HOI (in percentage points) are considered to make the changes comparable across countries and independent of the base period HOI. Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean location is particularly countries; and Demographic and Health Surveys for African countries. pronounced for Figure Average annual change in the HOI for access to electricity opportunities related 2.14 South Africa to infrastructure El Salvador Jamaica Ghana variables Guatemala Peru Dominican Republic Colombia Paraguay Brazil Panama Chile Honduras Namibia Mexico Costa Rica Zimbabwe Uruguay Kenya Nigeria Argentina Venezuela, RB Zambia –1 0 1 2 3 4 Percentage point change Note: The surveys are for different years across countries. The average survey years are 1997 and 2007, and the average period between two survey years for a country is nine years. These are broadly comparable with South Africa: surveys from 2002 and 2010 with an elapsed time period of eight years. Average annual changes in the HOI (in percentage points) are considered to make the changes comparable across countries and independent of the base period HOI. Source: Authors’ calculations based on General Household Surveys (2002 and 2010) for South Africa; national household surveys for Latin America and the Caribbean countries; and Demographic and Health Surveys for African countries. What circumstances shape inequality or a rural area as opposed to other urban of opportunity in South Africa? area) and education of the household head (a For the opportunity of finishing primary broad proxy for the socioeconomic status of schooling (13–15-year-olds), the D-Index of the household) contribute most to inequality overall inequality of opportunity is 10 percent, of opportunity. The importance of location more than half explained by the education of is particularly pronounced for opportunities the household head and the gender of the child related to infrastructure variables such as (figure 2.15). 33,34 Note that decompositions access to safe water on site, improved sanita- are shown only for the seven opportunities for tion, and electricity, ref lecting the disad- which the D-Index is 5 percent or higher, the vantaged situations of rural areas and the level at which the measure generally becomes townships and informal settlements. Education of policy concern. of the household head accounts for a sizable Location of the household (whether a child share of inequality in finishing primary school lives in a township and informal settlement on time, having health insurance, exposure to Figure Contribution of circumstances to overall inequality, 2010 2.15 (decompositions of D-Index greater than 5 percent) Gender Ethnicity Household composition Orphan status Education of the household head Other household characteristics Location D-Index 100 50 Marginal contributions (percent) 80 40 27 60 30 Percent 40 20 20 10 0 0 The children most School attendance (ages 6–11) School attendance (ages 12–15) Adequate teachers in school Adequate infrastructure in school Access to telecomunications Safe location ECD exposure Access to improved sanitation Access to safe water on site Have health insurance No overcrowding Access to electricity Finish primary vulnerable to not completing primary school, getting ECD exposure, and not having access to Source: Authors’ calculations based on General Household Survey (2010). safe water on site ECD programs, and access to infrastructure are less important by themselves when other (particularly electricity and adequate space in circumstances (which may be shaped by race are overwhelmingly the house), underscoring the lock of the fam- and gender in the first place) are taken into black Africans ily’s socioeconomic background on children’s account. This also implies that many of the future. apparent racial and gender gaps in opportuni- Household composition—whether both ties for children in South Africa today could be parents of the child live at home and the total narrowed if opportunities could be equalized number of children up to the age of 16 who across groups differentiated by socioeconomic live in the same household—intuitively enough status and, above all, location. features most prominently for the opportunity to grow up in a house with no overcrowding. Profiles of some vulnerable groups of South African Ethnicity contributes to inequality in all oppor- children. Figure 2.16 profiles some of the vul- tunities but does not rank among the top two nerable groups of children for a select group contributors to inequality to any. Gender of the of opportunities. For each opportunity, com- child contributes to inequality only for finish- parisons are made between children in the low- ing primary school on time. That gender has est quintile (bottom 20 percent) and those in such a limited role in explaining inequality in the highest quintile (top 20 percent) for their accessing infrastructure facilities is not surpris- chances of accessing a service. For ease of pre- ing, given that access is measured at the house- sentation, only three of the circumstances are hold level (and not the individual child’s). highlighted: location, ethnicity, and education Orphanhood—tragically common in South of the household head. Africa—appears to matter little in its contribu- The children most vulnerable to not com- tion to inequality of opportunity. pleting primary school, getting ECD expo- These decompositions should not be taken sure, and not having access to safe water on to imply that race- and gender-based dispari- site are overwhelmingly black Africans, under ties are not important. Rather, race and gender household heads who have either no education South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Figure Snapshot of the vulnerability profile 2.16 Finish primary school on time Bottom quintile Top quintile 100 Ethnicity Education Location 75 Percent 50 28 25 0 n ed n e e te te te te y T) en / l ra em hip iar hit on ca sia ple ple ple ple N ur Ru t fri ttl ns rt /A W N IS/ lo om om om om kA Te Children living in se w an Co (N al To nc yc nc yc di ac n In yi yi Bl ar ar ba ar ar im nd Ur im nd m Pr co townships and Pr co or Se inf Se informal settlements No overcrowding Bottom quintile Top quintile or rural areas are also 100 Ethnicity Education Location much more likely to 75 live in overcrowded households Percent 50 25 0 n ed n e e te te te te y T) en / l ra em hip iar hit on ca sia ple ple ple ple N ur Ru t fri ttl ns rt /A W N IS/ lo om om om om kA Te se w an Co (N al To nc yc nc yc di ac n In yi yi Bl ar ar ba ar ar im nd Ur im nd m Pr co Pr co or Se inf Se (continued) or some primary education, and live in rural goods and services (box 2.3). Some opportu- areas. Black Africans make up 80 percent of nities, such as school attendance for children the country’s population but only 61 percent under age 16 and access to telecommunica- of the children most likely to finish primary tions, are nearly universal with an HOI above school on time. Children living in townships 90 percent. Others are well below universal and informal settlements or rural areas are (an HOI of around 60 percent or below) and also much more likely to live in overcrowded distributed with high inequality among chil- households than those in nontownship urban dren of different circumstances, as for hav- areas. ing health insurance, access to safe water on site and improved sanitation, and adequate Inequality of opportunities for space without overcrowding—and finishing children: main findings primary school (13–15-year-olds). Still, other Opportunities among children in South opportunities—such as access to ECD pro- Africa vary widely across different types of grams, safety in the neighborhood, and access Figure Snapshot of the vulnerability profile (continued) 2.16 Access to safe water on site Bottom quintile Top quintile 100 Ethnicity Education Location 75 Percent 50 29 25 0 n ed n e e te te te te y T) en / l ra em hip iar hit on ca sia ple ple ple ple N ur Ru t fri ttl ns rt /A W N IS/ lo om om om om kA Te South Africa se w an Co (N al To nc yc nc yc di ac n In yi yi Bl ar ar ba ar ar im nd Ur im nd m Pr co has achieved Pr co or Se inf Se ECD exposure near-universal access Bottom quintile Top quintile 100 to primary education, Ethnicity Education Location a necessary first 75 step for equalizing opportunities Percent 50 among children 25 0 n ed n e e te te te te y T) en / l ra em hip iar hit on ca sia ple ple ple ple N ur Ru t fri ttl ns rt /A W N IS/ lo om om om om kA Te se w an Co (N al To nc yc nc yc di ac n In yi yi Bl ar ar ba ar ar im nd Ur im nd m Pr co Pr co or Se inf Se NIS/NT = noninformal settlement and nontownship. Source: Authors’ calculations based on General Household Survey (2010). to electricity—are well below universal but time puts it in the bottom half of international have low to moderate inequality attributable comparators. to circumstances. The overall picture is therefore mixed. On In international comparisons, South Africa the positive side, South Africa has achieved fares well on school attendance but ranks below near-universal access to primary education, most comparators on the HOIs for finishing a necessary first step for equalizing oppor- primary school on time and access to safe water tunities among children and an important on site, improved sanitation, and even electric- success for the education system to build on. ity. Trends suggest improvements, but the gaps The meteoric rise in access to telecommunica- with other countries are generally not closing. tions—from an HOI of around 30 in 2002 to Except for electricity, where South Africa’s aver- more than 90 in 2010—and a big increase in age annual progress has been exceptional, the the HOI for electricity are other milestones in progress on access to safe water and improved improving opportunities for children in South sanitation and on finishing primary school on Africa. South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Box Contrasting opportunities of three children 2.3 Thandiwe and Nothando are black African girls, growing up in households headed by individuals with no education. Thandiwe lives in an urban township and Nothando in a village. Andries is a white boy, living in a household headed by someone who completed secondary education and in an urban area that is not a township or informal settlement. How does their access to basic opportu- nities in life differ? It seems that both Thandiwe’s and Nothando’s chance of attending school up to the age of 15 are pretty high and just as good as Andries’s, a major achievement of post-apartheid South Africa (see box figure). But compared with Andries, they have lower chance of finishing primary school by age 15, having exposure to an early childhood development (ECD) program by age 4, and living in a 30 safe environment with adequate space in the household. Their chance of having health insurance is virtually zero, compared with more than 60 percent for Andries, and their access to most measures of physical infrastructure, such as telephones, electricity, safe water on site, and improved sanitation, falls short. Thandiwe and Nothando’s opportunities also vary greatly relative to each other, even though they are similar in all respects other than where they reside. For access to safe water on site, improved sanitation, and electricity, Thandiwe has a large advantage over Nothando, reflecting the vast gap in these services between urban townships/informal settlements and rural areas. For the opportuni- ties of primary school completion and access to ECD and telecommunications, Thandiwe has a small advantage over Nothando. Living Access to safe water in a township, however, puts Thandiwe at a disadvantage compared with Nothando in terms of a safe environment and adequate living space. and improved Nothando Thandiwe Andries sanitation are School attendance (ages 12–15) School attendance (ages 6–11) particularly critical Access to telecomunications Safe location inputs for child health Access to electricity ECD exposure No overcrowding Finish primary school on time Access to improved sanitation Access to safe water on site Have health insurance 0.0 0.2 0.4 0.6 0.8 1.0 Probability of access to the opportunity Source: Authors’ calculations based on General Household Survey (2010). Major challenges are the limited and Inequality in labor market opportunities unequal access to safe water on site and The ability of individuals to find jobs com- improved sanitation and the opportunity to mensurate with their qualifications—irrespec- finish primary school on time or be exposed tive of circumstances—is crucial for economic to ECD programs, along with the general lack mobility and reductions in inequality. This is of physical safety—all of which create the con- especially true in South Africa, with its chroni- ditions for children to develop their human cally high unemployment. Two key facts bring potential. The research literature is rich with out the impact of exclusion from labor markets evidence linking childhood opportunities to on income inequality in South Africa. First, future success. Preschoolers with low cognitive close to 70 percent of the bottom income quin- development have lower school achievement tile was unemployed in 2008. 37 Second, even and earn lower wages in adulthood. 35 And those employed face sharp disparities in wage early childhood education has substantial long- earnings based on race, gender, location, and term impacts, ranging from adult earnings to union membership. retirement savings.36 Access to safe water and In analyzing inequality in labor market sta- improved sanitation are particularly critical tus and its links to individual circumstances, inputs for child health, a determinant of nutri- this section uses a modified version of the tion status. approach in the previous section. 38,39 The analysis provides a rough but nonetheless intui- The list of circumstances is shorter than tive measure of the inequality between groups that for children because of data constraints, of working-age adults differentiated by various such as the lack of information about paren- attributes. It allows an assessment of the contri- tal characteristics for working-age adults. In bution of circumstances (as opposed to that of addition to the circumstances, the characteris- acquired or innate characteristics such as edu- tics of education and age of the individual are cation and age) to inequality of opportunity in included as (imperfect) measures of acquired the labor market. qualifications and experience. Three definitions of a desirable employ- 31 ment status are used for the analysis. Working- What is the status of opportunities in age adults (ages 15–64) are considered to have the South African labor market? desirable employment if they (annex E): The relationship between circumstances and • Have a job (any job), as opposed to being having a desirable employment status (under unemployed or discouraged from seeking the three definitions here) can work through employment. two channels: a direct effect (belonging to • Are employed full-time, as opposed to a particular race or gender group can affect The ability of being unemployed, underemployed, or the likelihood of getting a job just by virtue of individuals to find jobs discouraged. these circumstances), and an indirect effect • Are employed in the formal nonagricultural (circumstances can also influence the educa- commensurate with sector, as opposed to being employed in the tion, choices, and efforts of a person, which in informal sector outside agriculture. turn influences the likelihood of getting a job). their qualifications— The first two definitions are closely aligned The analysis below will focus on the first chan- irrespective of with the common notions of unemployment nel to assess inequality of opportunity in the and underemployment. The third definition labor market, due to conditions in that market. circumstances— is restricted to a sample of only those who Figure 2.17 summarizes the levels and is crucial for are employed in the nonagricultural sector. changes in employment status and inequal- The first two definitions are thus driven by ity between 2008 and 2012, using comparable economic mobility what an individual considers desirable. To be Quarterly Labour Force Survey data for both and reductions unemployed (or underemployed), for example, periods. The bars show an inequality-adjusted is a less desirable outcome only for those who coverage (IAC) rate, and the lines show in inequality would like to be employed (or employed full- inequality between groups with varied char- time). For the third definition, the assumption acteristics. The IAC is analogous to the HOIs is that having a formal sector job is the pre- computed for children, with the crucial dif- ferred status for those employed outside agri- ference being that the IAC and the inequality culture, which seems reasonable. measure (or D-Index) are based on not just cir- Two types of individual attributes are con- cumstances—gender, ethnicity, and location— sidered: circumstances (gender, ethnicity, and but also education and age.41 location—urban townships and informal settle- Thus the D-Index here reflects inequality ments, other urban, or rural) and characteristics between groups attributable to the direct effect (education and age of the worker). The defini- of circumstances and differences in education tion of circumstances, as in the previous section, and age. And it can be interpreted as the per- is clearly satisfied by gender and ethnicity. Loca- centage of total jobs that need to be reallocated tion is more debatable as a circumstance because across groups for all groups to have identical working-age individuals have some control over coverage rates. Differences in education, in where they live. In practice, however, spatial turn, can arise from two types of effects: that of mobility is restricted by cost considerations, (lack innate talent and effort, and the indirect effect of) social networks, and cultural and family ties. of circumstances on education—likely to have In South Africa particularly, mobility out of rural occurred at an earlier stage of an individual’s areas and townships and informal settlements is life and not related to conditions in the labor fairly restricted because of historical reasons, market. The direct contribution of circum- most affecting the poorest.40 stances to inequality, which can be attributed South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Inequality-adjusted coverage and inequality between groups differentiated by Figure circumstances, education, and age 2.17 2008 2012 2008 2012 100 16 Inequality-adjusted coverage (percent) 75 12 D-Index (percent) 50 8 32 25 4 0 0 Employed Employed full-time Employed in formal sector (nonagriculture) Note: The inequality-adjusted coverage for formal sector employment is higher than for the other two categories because it reflects the situation of only those who are Among the employed whereas the other two categories encompass the entire labor force (with or without employment). Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). circumstances, location appears to contribute to distortions in the labor market, would be finding a job were not only the lowest in South an estimate of inequality of opportunity in the Africa, but they were also the most strongly asso- most to inequality labor market. ciated with education, age, and circumstances.43 in employment The key facts that emerge: • The IAC (or opportunity) for being What are the contributions of circumstances status, followed by employed fell between 2008 and 2012, while relative to acquired attributes to ethnicity and gender IACs for full-time employment and formal inequality in employment status? employment in nonagriculture improved To unpack the nature of inequality in the labor slightly. Even as the employment situation market, overall inequality is decomposed into worsened, those employed are now less likely the contributions of education, age, and the to be underemployed and those employed direct effect of circumstances (which, as noted outside agriculture are more likely to have earlier, is distinct from the indirect effect formal employment. through education).44 In most cases, more than • The fall in IAC for employment between half the between-group inequality is from the 2008 and 2012 is due to a fall in the coverage education and age attributes: this is the more rate and a rise in inequality. While the cov- acceptable part of the inequality because it is erage rate declined from 72 to 66 percent based on attributes (experience and skills) during this time, between-group inequality that should matter for rewards in job markets increased—where the groups are differenti- (figure 2.19). The remaining (unacceptable ated by education, age, and circumstances. part of) inequality is attributable directly to • Inequality between groups in full-time circumstances—an estimate of inequality of employment increased from 2008 to 2012. opportunity in the labor market. The direct contribution of circumstances to inequality has South Africa compared with other middle-income fallen or remained unchanged in the past four countries. For an international comparison, simi- years, while that of education has increased, lar IACs for employment are shown for 17 other most significantly for employment in the for- middle-income countries in Latin America and mal sector outside agriculture. the Caribbean and Europe and Central Asia. Among the circumstances, location appears South Africa (13th among the 18 countries in to contribute most to inequality in employment per capita GDP) was ranked the lowest by IAC of status, followed by ethnicity and gender (figure employment and had the worst between-group 2.20). Between 2008 and 2012, the relative con- inequality linked to finding a job (figure 2.18).42 tribution of location to inequality in employment In other words, an individual’s chances of and full-time employment increased, while there Figure D-Index and inequality-adjusted coverage by employment, circa 2008 2.18 D-Index Inequality-adjusted coverage 15 100 12 80 9 60 Percent Percent 6 40 33 3 20 0 0 South Africa Turkey Romania Colombia Serbia Kazakhstan Bulgaria Ukraine Dominican Republic Brazil Argentina Venezuela, RB Peru Chile Russian Federation Mexico Belarus Ecuador Being a non-white young female, Note: Computed using similar but not identical definitions of circumstances across countries. Results for Latin America and the Caribbean and Europe and Central Asia may not match those from Quarterly Labour Force Surveys from the same countries. educated below Source: Authors’ calculations based on Latino Barometro (2008) for Latin America and the Caribbean countries; Life in Transition Surveys (2006) for Europe and Central Asian countries; and Quarterly Labour Force Survey (2008q1) for South Africa. secondary level, and living in rural areas or Figure Decomposition of sources of inequality in urban townships or 2.19 Circumstances Age Education informal settlements Employed 2008 2012 appears to define the vulnerability of being Employed 2008 full-time employed or being 2012 employed full-time (nonagriculture) formal sector Employed in 2008 2012 0 20 40 60 80 100 Percent Note: The contribution of each circumstance is expressed as percent of total between-group inequality measured by the D-Index. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). were small declines in that of ethnicity and gen- secondary level, and living in rural areas or der. Employment in the formal nonagricultural in urban townships or informal settlements sector stands out as the case where gender has appears to define the vulnerability of being little or no contribution to inequality between employed or being employed full-time. Employ- groups, while ethnicity and location contribute ment in the formal sector shows no pattern by more than for the other two cases. gender or age, but sharp differences appear by education, ethnicity, and location.45 Differ- Profiles of inequality in employment opportunity. entiating by location, rural residents have the Figure 2.21 profiles the most and least vulnera- highest vulnerability and are even less likely to ble groups for their chances of being employed be employed or employed full-time than town- under the three employment categories. Being ship residents. Among those already employed, a non-white young female, educated below rural workers have almost zero likelihood of South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Figure How circumstances contribute to inequality 2.20 Gender Ethnicity Location D-Index 30 15 Inequality-adjusted coverage (percent) 20 10 Percent 34 10 5 0 0 2008 2012 2008 2012 2008 2012 Employed Employed full-time Employed in formal sector (nonagriculture) The chances of finding Note: The contribution of each circumstance is expressed as percent of total between-group inequality measured by the D-Index. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). employment for a township and informal Figure Profiles of top quintile and bottom quintile by probability, 2012 settlement resident 2.21 Bottom quintile Top quintile Employment in formal sector and especially a rural Employment Full-time employment (nonagriculture) 100 100 100 resident do not stack 75 75 75 up favorably against Percent 50 50 50 a compatriot living in 25 25 25 other urban areas 0 0 0 Female Non-white TS or IS Rural Below secondary education Age (average) Female Non-white TS or IS Rural Below secondary education Age (average) Female Non-white TS or IS Rural Below secondary education Age (average) TS = township; IS = informal settlement Note: “Probability� refers to probability of having an opportunity, defined as “employment,� “full-time employment,� and “employment� in formal sector (for those employed outside agriculture). Source: Authors’ calculations based on Quarterly Labour Force Surveys (2012q1). working in the formal sector outside agricul- urban areas (figure 2.22).46 This suggests that ture, while township workers fare far better. unemployment and underemployment are much more serious in townships and informal How do labor market opportunities vary settlements and particularly rural areas. among different subgroups of the population? Inequality between groups in employment (measured by the D-Index) and full-time Location. How does the place of residence employment is high across the urban land- affect employment prospects in South Africa? scape, with the bias again stronger against resi- Clearly, the chances of finding employment for dents of townships and informal settlements. a township and informal settlement resident Inequality between groups is extremely high in and especially a rural resident do not stack up rural areas (much higher than in urban areas), favorably against a compatriot living in other and it got significantly worse for employment Figure D-Index and inequality-adjusted coverage by location, 2008 and 2012 2.22 D-Index Inequality-adjusted coverage 20 100 15 75 Percent Percent 10 50 35 5 25 0 0 Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural 2008 2012 2008 2012 2008 2012 For residents of urban Employed Employed full-time Employed in formal sector (nonagriculture) townships, informal TS = township; IS = informal settlement Note: Circumstances/characteristics are gender, ethnicity, education, and age. The D-Index refers to between-group inequality (groups differentiated by gender, ethnicity, settlements, and rural education, and age) among TS/IS or non-TS/IS workers. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). areas, age appears to be the biggest and full-time employment between 2008 and to employment and full-time employment, an 2012. especially difficult situation for the township factor in explaining For residents of urban townships, informal and rural youth (figure 2.23). Education and the inequality in settlements, and rural areas, age appears to be gender are a distant second. For having a for- the biggest factor in explaining the inequal- mal sector nonagricultural job (among those their access to ity (D-Index) between groups in their access already employed), the role of education has employment and full-time employment Figure Decomposition of sources of inequality, 2008 and 2012 2.23 Education Age Ethnicity Gender 100 80 60 Percent 40 20 0 Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural Urban non-TS/IS Urban TS/IS Rural 2008 2012 2008 2012 2008 2012 Employed Employed full-time Employed in formal sector (nonagriculture) TS = township; IS = informal settlement Note: Circumstances/characteristics are gender, ethnicity, education, and age. Columns show the percentage contribution of every circumstance/characteristic to the D-Index (between-group inequality among groups differentiated by gender, ethnicity, education, and age) for each type of worker. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity grown significantly in recent years and almost the least level playing field when looking for single-handedly determines the inequality— employment or full-time employment. They and much more so in townships and rural also experienced the steepest declines in IAC areas than for residents of other urban areas, for employment between 2008 and 2012: 25 for whom ethnicity also matters. percent for 15–24-year-olds and 15 percent for 25–29-year-olds, while their inequality of Age group of workers. The age group of work- opportunity increased 42 percent and 16 per- ers, as already indicated, contributes much to cent, respectively. For underemployment, all 36 between-group inequality for employment and age groups other than the youngest improved full-time employment in South Africa. This their situation over the period. The IACs and is unusual. For the other 17 middle-income D-Indices in formal sector employment out- countries, the average contribution of age to side agriculture were similar across all five age between-group inequality in employment was groups. 19 percent in 2008, less than half the 42 per- Inequality of opportunity in the labor mar- cent for South Africa. Even though there is ket, indicated by the contribution of circum- Inequality of wide variation across countries, age-induced stances (gender, ethnicity, and location) to inequality in employment was the highest in between-group inequality, is the highest for the opportunity in the South Africa in magnitude, while only Belarus youngest age group (15–24- and 25–29-year- labor market is and the Dominican Republic had a higher per- olds). Around 70 percent of inequality of centage contribution of age to inequality than opportunity in finding employment or full- the highest for the South Africa. time employment within these groups is youngest age group Breaking down the working-age popula- attributable to circumstances over which tion into five age groups draws out the differ- they have no control. Education matters for ences in opportunity and inequality between job opportunities of most age groups (figure age groups.47 For employment and full-time 2.25). But education matters little for inequal- employment, IAC increases and inequality ity in employment and full-time employment (D-Index) declines progressively for older among the youngest workers, and the extent workers. Improvements are especially large to which education matters increases substan- between the age groups 15–24 and 25–29 and tially as workers get older. For inequality in then between 25–29 and 30–39 (figure 2.24). formal employment (among those employed Youths (ages 15–24 and 25–29) experience outside agriculture), education contributes Inequality-adjusted coverage and D-Index for workers of different age groups, Figure 2008 and 2012 2.24 Inequality-adjusted coverage D-Index 100 20 75 15 Percent Percent 50 10 25 5 0 0 25 24 30 29 40 39 50 49 5 25 4 30 29 40 39 50 49 5 25 24 30 29 40 39 50 49 5 25 4 30 29 40 39 50 49 5 25 4 30 29 40 39 50 49 5 25 4 30 29 40 39 50 49 5 –6 –2 –6 –6 –2 –6 –2 –6 –2 –6 – – – – – – – – – – – – – – – – – – – – 15 15 15 15 15 15 2008 2012 2008 2012 2008 2012 Employed Employed full-time Employed in formal sector (nonagriculture) Note: Circumstances/characteristics are gender, ethnicity, education, location, and age. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). Figure Decomposition of sources of inequality by age group 2.25 Education Age Location Ethnicity Gender 100 80 60 Percent 40 37 20 0 15–24 25–29 30–39 40–49 50–65 15–24 25–29 30–39 40–49 50–65 15–24 25–29 30–39 40–49 50–65 Employed Employed full-time Employed in formal sector (nonagriculture) Note: Circumstances/characteristics are gender, ethnicity, education, and age. Columns show the percentage contribution of every circumstance/characteristic to D-Index, Among circumstances, for each age group. Source: Authors’ calculations based on Quarterly Labour Force Surveys (2008q1 and 2012q1). location matters the significantly to inequality for all age groups, reflected fewer supply of jobs on account of the most across the board, but again matters the least for the youngest global crisis. But this was also because of a rise age group. in inequality between groups, which cannot be followed by ethnicity Thus the opportunities of employment and attributed to the global crisis. and then gender full-time employment seem to be the least The causes of inequality in labor markets meritocratic for workers below age 30, with have changed in the past four years. The con- meritocracy improving progressively with age. tribution of education has increased, while that Even though inequality of opportunity is gener- of circumstances of gender and ethnicity has ally lower (and meritocracy higher) for formal fallen slightly. Where a person seeking employ- employment outside agriculture (among those ment lives, however, matters more now than it already employed), the age divide in inequal- did four years ago. ity of opportunity persists here as well: circum- The rising importance of education carries stances matter the most for the youngest age its own challenges: the disadvantages conferred group. Among circumstances, location matters by unequal opportunities in education earlier the most across the board, followed by ethnic- in life are an increasingly consequential stum- ity and then gender. Interestingly, even among bling block to the social and economic mobility the group of youngest workers, differences of individuals, for whom having a job is crucial. in age matters substantially for the chance of For employment in the formal sector outside being employed and employed full-time. agriculture, a measure of quality jobs with some stability, education accounts for more Inequality of employment opportunities: than half the inequality in employment, the main findings contribution having risen 8 percentage points Given its high unemployment rates, it is not in four years. The particularly strong impact of surprising that South Africa does much worse education on the likelihood of having a formal on labor market opportunities than other mid- sector job points to a high and rising skills pre- dle-income countries. It is still interesting and mium in the labor market, confirming findings telling, however, that its relative performance in the literature.48 The wage inequality from is explained not only by too few jobs, but also the skills premium is a key driver of income by the inequality that persists between differ- inequality in South Africa. ent groups in their access to these few jobs. The employment situation appears to be Reflecting the adverse effects of the global particularly challenging for young workers financial crisis, the inequality-adjusted cover- and residents of townships, informal settle- age fell for finding employment. In part, this ments, and rural areas. An individual’s age is South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity an unusually large contributor to inequality (but not exclusively) race and location, in in employment in South Africa, more than for explaining inequality in the labor market. many other middle-income countries, with the odds increasingly stacked against the youngest Conclusions workers. Inequality of opportunity, the part of An equitable society would not allow circum- inequality attributable to circumstances that stances over which the individual has no control an individual has little or no control over, is to influence her or his basic opportunities after also higher among young workers than among birth. Whether a person is born a boy or a girl, 38 older workers. Thus not only do young workers black or white, in a township or leafy suburb, face a disadvantage in the labor market due to an educated and well-off parent or otherwise to their age, they also compete for jobs in a should not be relevant to reaching his or her market seemingly more “unfair� in allocating full potential: ideally, only the person’s effort, opportunities among the young: a young per- innate talent, choices in life, and, to an extent, son’s chances of having a job, a full-time job, sheer luck, would be the influencing forces. or a formal sector job seem to depend more on This is at the core of the equality of opportunity Whether a person circumstances beyond their control than the principle, which provides a powerful platform skills they have acquired. for the formulation of social and economic is born a boy or a In addition to being young and living in cer- policy—one of the rare policy goals on which a girl, black or white, tain locations, being a woman and non-white political consensus is easier to achieve. still matters, increasing the likelihood of being As with any other policy, an intuitive and in a township or unemployed or underemployed significantly objective measure of progress is crucial. The leafy suburb, to an (over and above any impact of these attributes development of the human opportunity index, on education). extensively presented in this report for various educated and well-off It is sobering to note that inequality of childhood-  and employment-related oppor- parent or otherwise opportunity in employment—the part of tunities, has helped fulfill this requirement inequality attributable to circumstances— for the equality of opportunity objective. The should not be relevant would be higher than what is estimated here HOI is a powerful tool to measure and track to reaching his or if the indirect effect of circumstances through a society’s progress on equitable distribution education were also taken into account. The of basic opportunities. Together with a robust her full potential analysis here focused on inequality produced data gathering and monitoring and evalua- in the labor market and did not take into tion system, it can help improve the targeting account the unequal distribution of opportu- and efficacy of social policy. Making use of it, a nities earlier in life (such as timely completion number of countries in Latin America, includ- of schooling or ECD exposure) that matter for ing Brazil and Peru, have begun confronting human capital formation and are also affected their inherent inequalities with proactive and by these circumstances. increasingly well-targeted social policies, with In some ways, the contributions of specific positive initial results. South Africa, with its circumstances (like race and location) to entrenched inequality inextricably linked to its inequality in the labor market could also be people’s varied circumstances, has a chance to overstated here, since circumstances related forge a similar path of policy correction. to parental socioeconomic background, which Extraneous circumstances that a child is are likely to be correlated with these attributes, born into (ethnicity, location, gender, and are missing from the analysis due to a lack of family background) were found in this report data. Recall that gender and race had a smaller to variably affect the child’s access to basic contribution to inequality of opportunity for opportunities in South Africa. Moreover, children once the contributions of other fac- some of the circumstances (location and eth- tors, including socioeconomic attributes of nicity in particular) are also important for the child’s family and parents, were accounted inequality in employment opportunities later for. Without clear evidence, the contributions in the child’s life. This raises the prospect of of race and location are best interpreted as a vicious cycle of adverse circumstances that reflecting socioeconomic factors, including compounds inequalities over multiple stages in life, and over the lives of multiple generations. to deprivations in multiple dimensions simul- A rural black African girl, for example, grow- taneously. For example, black South Africans ing up with far fewer opportunities to develop living in rural areas, and with household heads to her full potential is also less likely to find who did not complete primary schooling, are employment opportunities as a young adult. much more likely not to complete primary Put differently, the child with disadvantageous school, be exposed to an ECD program, or circumstances not only has to work harder to have access to health insurance. The presence overcome the disadvantages, but having done of multiple deprivations points to the need for so, finds that these reemerge when seeking policy programs in different sectors (health 39 employment as an adult. Moreover, the disad- and education, for example) to coordinate vantages do not stop with one person—they get closely in order to achieve better efficiency and transmitted across generations. the best results. How to break through this vicious, self-per- Of course, there are no simple, elegant petuating cycle of inequality in South Africa? policy solutions in the quest for equity. One To be sure, that would involve leveling the play- important lesson from international experi- ing field in the quality of education children ence is that a dynamic system involving policy There are no simple, get and the employment opportunities they experimentation (from incentives for training elegant policy face as young adults, irrespective of location, and hiring of young workers to monitorable gender, or ethnicity. Paying special attention to and incentive-based delivery of public services), solutions in the the water, sanitation, and health care needs of backed by rigorous impact evaluation and rural areas and townships and to overcrowding greater participation of communities in the quest for equity in townships would also be important. actual delivery of basic public service delivery Policy design needs to recognize that chil- and in the feedback loops for policymakers, is dren of certain circumstances are vulnerable crucial. ANNEx A Illustrating the Human Opportunity Index with a simple example Consider two societies A and B in which half The imposed “penalty� can be interpreted as the population lives in rural areas and the the share of the total number of opportunities other half in urban areas. Now consider a basic that need to be redistributed to ensure equita- opportunity such as access to primary educa- ble access based on the equality of opportunity tion. Say that 50 percent of all children go to principle. In society A, this will constitute “real- school in both societies. Looking at the over- locating� 25 percent of total enrollments from all coverage, they both will appear similarly urban children to rural children. Therefore, placed. But suppose we also know that in soci- the penalty would be 25 percent and the HOI, ety A, no rural child attends a school; while in which is the coverage minus the penalty, would society B, 50 percent of both rural and urban equal 25 percent. For society B, there is no children attend school. The Human Opportu- inequality based on location, and the penalty is nity Index (HOI) discounts the coverage rate zero. This implies that the HOI is 50 percent, or of 50 percent by imposing a “penalty� when equal to the coverage. So, based on the equality access is more unequal based on circumstances of opportunity criteria, society B is more equal such as location. than society A, even though average enrollment rate is the same in both societies. 41 ANNEx B Three key properties of the Human Opportunity Index First, it is sensitive to scale—if access improves An important caveat is that this measure is for all groups by, say, a factor of k (additively sensitive to the set of circumstances chosen for or multiplicatively), then the Human Opportu- analysis. But this is mitigated by an additional nity Index (HOI) changes by the same factor k. property that is highly desirable given that it is Second, it rewards Pareto improvement—if the seldom possible to identify all relevant circum- coverage rate improves for one circumstance- stances for any population and opportunity: group without decreasing coverage rates for the HOI will not be higher if more circum- the remaining groups, the HOI will rise. Third, stances are added to the existing set of circum- the measure will always improve if access stances in the analysis. This implies that the changes in a way that the more vulnerable computed inequality serves as a lower bound to groups (groups with coverage rates lower than the “actual� inequality where all circumstances the overall coverage rate) have higher access. of interest could be included in the analysis. Source: Barros, Molinas Vega, and Saavedra 2010. 42 ANNEx C Estimating the Human Opportunity Index from household survey data To construct the HOI, we need to obtain the circumstances (such as between parental conditional probabilities of access to oppor- education and location). Given the number tunities for each child based on the circum- of circumstances (all are dummy variables), stances. To do so, we can estimate a logistic limited sample sizes, and the large number of model, linear in the parameters β, where event countries and opportunities for which these I corresponds to accessing the opportunity regressions have to be run, including inter- (such as safe water), and x the set of circum- actions would lead to intractable problems stances (such as gender of the child and educa- in at least some cases. The interaction terms tion and gender of the head of the household). are thus omitted, even though translating the An important caveat to the logistic estima- exact definition of the D-Index to the logis- tion model is that the list of regressors does tic regression model would require including not include any interaction terms between these terms. Source: Barros, Molinas Vega, and Saavedra 2010. 43 ANNEx D Opportunities and circumstances for children Table What constitutes opportunities for South African children? D1 Opportunities Description Human development School attendance (ages 6–11) Currently attending educational institution for children ages 6–11 School attendance (ages 12–15) Currently attending educational institution for children ages 12–15 Finish primary school on time Completed primary (seventh grade) education for children ages 13–15 (ages 13–15) Adequate infrastructure in school No lack of books and no facilities in bad conditions (parental responses) Adequate teachers in school Teacher quality not poor, no lack of teachers, and no classes too large/too many learners (parental responses) ECD exposure Exposed to an early childhood development program in any way for children ages 0–4 Have health insurance Covered by a medical aid or medical benefit scheme or other private health insurance for children ages 0–16 Basic infrastructure No overcrowding Equal or below 1.5 person per room for children ages 0–16 Access to safe water on site Source of drinking water is piped water (dwelling or yard), borehole, or rainwater for children ages 0–16 Access to improved sanitation Type of toilet facility is flush toilet (public sewerage or septic tank), chemical toilet, or pit latrine/toilet with ventilation pipe (in the yard or dwelling) for children ages 0–16 Access to electricity The household has a connection to the MAINS electricity supply for children ages 0–16 Access to telecommunications Access to a telephone or cell phone in the household for children ages 0–16 Security Safe location Members of the household feel very safe or rather safe living here 44 Table What constitutes circumstances in the analysis of children’s opportunities? D2 Dimension Circumstances Details Gender Gender Ethnicity Ethnicity African/Black, Coloured, Indian/Asian, and White Household composition Presence of the spouse of the household head in the household Total number of children ages 0–16 in the household Presence of the father in the household Presence of the mother in the household 45 Orphan status Is the father alive? Is the mother alive? Education of the household head Level of education of the household head None, primary incomplete, primary complete, secondary incomplete, secondary complete, and tertiary Other household head characteristics Gender of the household head Age of the household head Location Urban: not informal settlements/not townships Urban townships and/or informal settlements Rural areas (including some informal settlements) Table HOI, D-Index, and coverage for 2002, 2005, and 2010 D3 (percent) 2002 2005 2010 HOI D-Index Coverage HOI D-Index Coverage HOI D-Index Coverage School attendance (ages 6–11) 89.8 2.1 91.7 94.9 1.2 96.0 98.2 0.4 98.5 School attendance (ages 12–15) 95.4 1.2 96.6 96.9 0.8 97.7 97.6 0.5 98.1 Finish primary school on time (ages 13–15) 42.5 14.8 49.9 49.4 12.5 56.5 50.7 10.0 56.3 Adequate infrastructure in school 70.4 5.4 74.4 76.4 5.4 80.7 90.6 0.9 91.4 Adequate teachers in school 87.1 1.1 88.1 86.8 1.9 88.5 92.6 0.7 93.2 ECD exposure (ages 0–4) — — — — — — 59.9 5.7 63.5 Have health insurance 5.4 56.8 12.4 5.0 55.7 11.3 7.3 47.4 13.8 No overcrowding 42.2 19.6 52.5 41.8 19.6 52.0 48.9 16.5 58.5 Access to safe water on site 40.3 28.2 56.1 44.7 23.2 58.3 48.3 23.3 63.0 Access to improved sanitation 27.8 38.1 44.9 41.3 23.0 53.6 54.6 18.4 66.9 Access to electricity 60.7 12.9 69.7 68.3 10.0 75.9 77.0 6.1 82.0 Access to telecommunications 30.7 23.7 40.2 60.8 9.4 67.1 91.5 1.9 93.3 Safe location — — — — — — 61.3 4.0 63.8 — = not available. Note: 2005 estimates have been computed using province (and not urban/rural/townships) as the location circumstance. Source: Authors’ calculations based on General Household Surveys (2002, 2005, and 2010). ANNEx E Inequality in employment opportunities Table Definition of variables to measure “opportunities� in labor market E1 Status Universe of workers Desirable labor market status Less desirable labor market status Free from unemployment or Those in labor force (employed, Employed part-time or full-time Unemployed or discouraged from employed unemployed, or discouraged) participation in labor market Free from underemployment or Those in labor force (employed, Employed for 35 hours a week or more Underemployed, unemployed, or employed full-time unemployed, or discouraged) discouraged from participation in labor market Employment in the formal sector In the age group 15–64, Employees in firms with more than five Employed in the nonagricultural (nonagricultural workers) those who are employed in the employees or firms who deduct income tax sector, but not in a firm or as an nonagricultural sectors from salaries. Or employers, own account employer who meets the definition workers, and persons helping unpaid in of formality their household business, who are registered for either income tax or value-added tax Note: All variables refer to the working-age group of ages 15–64. “Underemployed� refers to those who are working less than 35 hours a week and looking for full-time employment. Table Definition of circumstances and characteristics in labor market E2 Circumstances Gender Male/female Ethnicity African/Black, Coloured, Indian/Asian, and White Location Urban: not informal settlements/not townships; urban townships and/or informal settlements; and rural areas (including some informal settlements) Characteristics Education of the worker None, primary incomplete, primary complete, secondary incomplete, secondary complete, and tertiary Age of the worker Number of years 46 Notes 1. This section on global economic prospects we also used the relevant variable for the draws on parts of the June 2012 Global person’s education attainment. Economic Prospects report prepared by 6. The employment categories are: employed, the Development Economics group of unemployed, discouraged worker, and the World Bank. The full report is acces- other inactive. We constrain the sample to sible at http://siteresources.worldbank. the working-age population (15–64) and org/INTPROSPECTS/Resources/334934 to those observations that we were able to -1322593305595/8287139-1339427993716/ match. GEPJune2012_Full_Report.pdf. 7. Townships and informal settlements, far 2. In part this reflects the fact that develop- removed as they are, are also ill-served by ing country borrowers (especially corpo- any form of public transportation, having rates) had issued international bonds at a thus to rely largely on private taxis that record pace up until April, taking advan- charge a hefty amount—on average close to tage of low funding costs and robust inves- 20–30 percent of the daily minimum wage tor appetite for emerging market debt. for a round trip to work by some accounts. 3. South Africa and the rest of the region do 8. World Bank 2011b. not face the capacity constraint problem, 9. The literature finds that roughly half with manufacturing capacity utilization the overall inequality is generated by running at 80 percent and severe unem- between-race inequality, with the within- ployment rates. race inequality contributing the other 4. South African Reserve Bank 2012. half. There is a weaker consensus on the 5. We use the fact that sample for each Quar- contribution of the two components to terly Labour Force Survey is divided into the increase in inequality since apartheid. four subgroups called rotation groups, Leibbrandt and others (2010) find that with one of these groups being rotated the contribution of inequality between each quarter. This implies that roughly racial groups fell over 1993–2008 while half of the households will be available that of the inequality within racial groups in the sample after six months, which will increased. Bhorat and van der Westhuizen allow us to examine the flows from a given (2011) find the opposite trends for the two employment status to another. The match- subcomponents for the 1995–2005 period. ing process to construct the transition 10. Leibbrandt and others (2010), based on matrices was done using the household’s 2008 National Income Dynamics Survey. unique identification number, the worker 11. Wage income contributes to more than 85 age, the race, and the gender—and for percent to the overall income inequality those observations that were inconclusive (Leibbrandt and others 2010). 47 South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity 12. Bhorat and van der Westhuizen (2011) over multiple years, and covers relevant report that over 1995–2005 poverty reduc- topics. It is an annual household survey tion was much more pronounced at a specifically designed to measure various lower poverty line, suggesting that those in aspects of living conditions of households deeper poverty experienced more signifi- and conducted regularly by Statistics cant income gains. Leibbrandt and others South Africa since 2002. And it covers six (2010) argue that while the pace of poverty broad areas: education, health and social reduction over 1993–2008 was modest, development, housing, household access the depth of poverty measure saw a much to services and facilities, food security, 48 more noticeable improvement, again sug- and agriculture. gesting that those at the lower end saw 22. For 2005, households could not be mapped faster improvement. to townships so the location variable refers 13. The grants—whose main components are simply to the nine regions of the country. the Child Support Grant, State Old Age While this makes the 2005 results not quite Pension Grant, the Disability Grant, and comparable for those of 2002 and 2010, the Foster Care Grant—have increased in comparisons over time are possible since coverage from 3 million recipients (7 per- differences in the definition of one cir- cent) in 1997 to more than 16 million in cumstance are unlikely to affect the HOI 2012 (32 percent), with the amount as a in a significant way. Results for all years are share of GDP increasing from 2.5 percent shown in annex table D.3, while the main in 1996/97 to 4.0 percent in 2011/12. report refers to results from 2002 and 2010 14. Leibbrandt and others 2010. only. 15. Bhorat and van der Westhuizen 2011. 23. For an example of how opportunities in 16. Bhorat and van der Westhuizen (2011) other countries are chosen, see the case of find that in 2005 income Gini would Côte d’Ivoire in Abras, Cuesta, and others increase from 0.72 (including social assis- (2012). tance grants) to 0.77 (without the grant 24. Barnett (1995) reviews 36 studies of proj- incomes). ects and large-scale public programs 17. Perhaps most important for the pro- (including preschool education, Head posed work are the contributions of John Start, child care, and home-visiting pro- Roemer, whose 1998 work Equality of Oppor- grams) to examine their long-term effects tunity was the first to formalize an equality on children of low-income families. of opportunity principle (Roemer 1998). According to him, ECD programs can 18. See, for example, Chetty and others (2010) produce large short-term benefits for chil- for evidence that early childhood educa- dren on IQ and sizable long-term effects tion has substantial long-term impacts, on school achievement, grade retention, ranging from adult earnings to retirement placement in special education, and social savings. Child malnutrition has also been adjustment. shown to generate life-long learning diffi- 25. Cabral, Lucas, and Gordon (2009) find culties, poor health, and lower productiv- that the main health threat posed by ity and earnings over a lifetime (Alderman drinking unsafe water is infectious diar- and others 2001; Hoddinott and others rhea—the leading cause of mortality for 2008). children under age 5 and estimated to 19. Barros and others 2009; Barros, Molinas cause 1.5 million deaths a year. Jalan and Vega, and Saavedra 2010. Ravallion (2003) find that the prevalence 20. Barros and others 2009; Barros, Molinas and duration of diarrhea among children Vega, and Saavedra 2010. under 5 in rural India are significantly 21. For South Africa, we rely on data from the lower on average for families with piped General Household Surveys (2002, 2005, water than for those without it. Alderman and 2010), as it is representative of the and others (2011) show that private behav- population, available for and comparable ioral choices and policies that affect the health and nutrition of children in rural 34. These decompositions are based on a Pakistan have important effects on school Shapley Value approach, which essen- enrollment and thus on eventual produc- tially estimates the contribution of each tivity. Improved nutrition increases enroll- circumstance to the D-Index as the mar- ments, and more so for girls, thus closing a ginal change (increase) in the value of the portion of the gender gap. D-Index when the circumstance is added 26. For example, Gove, Hughes, and Galle to all possible combinations of other cir- (1979) in their study on Chicago. cumstances. It involves taking the weighted 27. World Economic Forum 2011. average of all marginal contributions 49 28. World Bank 2011a. when the kth variable is added to all pos- 29. The Global Competitiveness Report 2011–2012 sible combinations of the other variables. ranks South Africa 127 of 142 on the Qual- See Hoyos and Narayan (2011) and Abras, ity of Primary Education index, 133 on the Cuesta, and others (2012) for theory and Quality of Educational System index, and applications. 138 on the Quality of Math and Science 35. Currie and Thomas 1999; Case and Pax- Education index (World Economic Forum son 2006. 2011). 36. Chetty and others 2010. 30. Information on parents’ own subjective 37. Leibbrandt and others 2010. appraisal of the adequacy of infrastructure 38. See Abras, Hoyos, and others (2012) for and teachers in the schools their wards a detailed discussion and application of attend, which the surveys do have, is far this approach, which is conceptually simi- from ideal for assessing the opportunity of lar to other approaches in the academic quality education. More than 90 percent of literature on inequality of opportunity. the parents whose children were enrolled They infer inequality of opportunity from in primary schools in 2010 appear to have a distribution of outcomes among individ- been satisfied with the level of teaching uals of different “types� (defined by their and physical infrastructure in schools with circumstances), where such inequality is the HOIs being fairly similar to the cover- typically measured as the between-group age rate (implying low inequality). inequality across these types, distinct from 31. Note we can also analyze changes between inequality within each type. For an exam- 2002 and 2005 and 2005 and 2010. But ple and a discussion of recent literature, given the problems with comparability see Ferreira, Gignoux, and Aran (2011) between results from 2005 and the other and Ferreira and Gignoux (2011). In our years due to differences in the location case, the outcome (job status) is different variable, we focus on changes from 2002 from the typical variables (income, con- to 2010 in the main text. Annex table D.3 sumption) used in the literature, in that it presents results for all three years. is a discrete variable. This necessitates the 32. The surveys are for different years across use of a modified HOI-type index, which countries. The ranges for survey years are incorporates the D-Index that is essentially 1994–2002 (average of 1997) and 2002–11 a measure of between-group inequality. (average of 2007). The average period 39. The analysis in this section is subject to a between two survey years for a country number of caveats (also see Abras, Hoyos, is nine years. All these are broadly com- and others 2012). The main challenges parable with South Africa (surveys from relate to the lack of information on paren- 2002 and 2010, and time period of eight tal characteristics of working-age individu- years). As before, we revert to the analysis als from labor force surveys; and that the of a smaller subset of opportunities using a observed outcomes in the labor market are smaller set of circumstances for the sake of an equilibrium phenomenon, which does comparability. not contain information about the indi- 33. The same charts for 2002 and 2005 are vidual’s preferences and reservation wage, presented in annex D. the effort that he or she puts into finding South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity a job, unobserved attributes like talent and 44. For the decomposition, we use the Shapley initiative, and even (in a reliable way) the method discussed earlier. quality of the job. Information on wages 45. The results from decompositions and pro- or earnings, which would help to some files should be interpreted with caution extent, is absent from the recent Quarterly due to the caveats mentioned earlier. The Labour Force Survey data available to us. problem created by the short list of cir- The approach we adopt addresses in part cumstances is especially important to high- some of these issues, but the results are light. The fact that potentially important best seen as indicative and suggesting cor- circumstances are missing—like parental 50 relations as opposed to direction or causal- education and income/wealth—implies ity of effects. that the “contribution� of ethnicity and 40. Gelderblom (2007), reviewing the literature location to inequality (likely to be corre- on migration in South Africa, concludes lated with parental education and wealth) that constraints associated with the costs will at least partly reflect that of the miss- and risks of migration are high for the poor ing circumstances. This would mean that and especially high among those in more the role of these circumstances would distant areas, far from transport and tele- overestimate their “true� contribution communication links, with a more recent to inequality. So, while the profiles and history of labor migration, and thus weakly contributions to inequality fairly reflect developed social networks. It also applies how each attribute contributes to inequal- more particularly to the poorest, whose net- ity between the groups defined by these works will be more weakly developed. attributes, there may well be other attri- 41. The results—particularly the changes butes that are important, if they could be over time and across opportunities—look included in the analysis. similar when IAC and the D-Index are 46. An important caveat to the results for the estimated based on circumstances (gen- subgroups in this section (by location or der, ethnicity, and location) only and not age) is that the IAC measure used here education and age. (Results are available (algebraically identical to HOI, which on request.) The levels of IAC (D-Index) is based on circumstances only) is not are generally higher (lower) in this case, subgroup-consistent. This implies that compared with those in figure 2.17, since the IAC for the working-age population, the number of groups is smaller. Unlike for any measure of employment, is not figure 2.17, these results do not distinguish an average of the IACs for the subgroups between the direct and indirect effects of (by location or age). An alternate mea- circumstances on inequality and reflect sure of opportunities, the Geometric both channels of effect: distortions in HOI (GHOI) has the property of sub- the labor market and the influence of cir- group consistency, but it is less intuitive cumstances on education that would have and harder to understand and represent occurred prior to an individual entering graphically (Barros and others 2010). the labor market. Rather than using a different measure 42. The average per capita GDP (purchasing altogether for doing the subgroup analy- power parity) in 2010 of the 17 comparator sis, we have opted to continue using the countries was around $12,280, compared original HOI-type measure. This makes with $10,565 for South Africa, making it a little difference to the conclusions, for the reasonable (if slightly richer) comparator IACs using the GHOI methodology are group on average. highly correlated with the original IACs. 43. While countries with higher employment The IACs using GHOI method are avail- tend to have lower inequality between able upon request. groups, the magnitude by which South 47. Four of these groups constitute 20–22 Africa’s inequality exceeds that of its com- percent of the working-age population, parators is still striking. whereas the oldest group constitutes about 15 percent—making the groups quite com- relative wages of skilled workers between parable in size. 1995 and 2005. In 1995 and 2005, the 48. Using data from 1995 to 2002, Bhorat share of wage income in total income was (2004) found unemployment to be higher much higher for those in the top four for low educational attainment groups. deciles than for the bottom deciles (Bho- Bhorat, van der Westhuizen, and Jacobs rat, van der Westhuizen, and Jacobs 2009). (2009) found evidence in favor of a ris- While highly skilled workers, concen- ing skills premium and argued that a shift trated among the better-off households, from the primary and secondary sectors to were rewarded with high wages, lower and 51 the tertiary sector has driven the increased unskilled workers were either poorly paid demand for skilled labor, boosting the or unable to find employment. References Abras, A., J. Cuesta, A. Hoyos, and A. Narayan. Bhorat, H., and C. van der Westhuizen. 2011. 2012. “Equality of Opportunities and Fiscal “Pro-poor Growth and Social Protection in Incidence in Côte d’Ivoire.� Policy Research South Africa: Exploring the Interactions.� Working Paper 6048, World Bank, Washing- Input paper prepared for the National Plan- ton, DC. ning Commission. Abras, A., A. Hoyos, A. Narayan, and S. Tiwari. Bhorat, H., C. van der Westhuizen, and T. Jacobs. 2012. “Inequality of Opportunities in the 2009. “Income and Non-Income Inequality Labor Market: Evidence from Life in Tran- in Post-Apartheid South Africa: What are the sition Surveys in Europe and Central Asia.� Drivers and Possible Policy Interventions?� Background Paper for the World Develop- DPRU Working Paper 09/138. University of ment Report 2013, World Bank, Washington, Cape Town, Development Policy Research DC. Unit, Cape Town, South Africa. Alderman, H., J. Behrman, V. Lavy, and R. Cabral, C., P. Lucas, and D. Gordon. 2009. “Esti- Menon. 2001. “Child Health and School mating the Health Impacts of Unsafe Drink- Enrollment: A Longitudinal Analysis.� The ing Water in Developing Country Contexts.� Journal of Human Resources 36 (1): 185–205. Aquatest Working Paper No. 01. Aquatest 2 Barnett, W. Steven. 1995. “Long-Term effects of Project, Water and Health Research Centre, early childhood programs on cognitive and University of Bristol, UK. school outcomes.� The Future of Children 5 Case, A., and C. Paxson. 2006. “Stature and (3): 25–50. Status: Height, Ability, and Labor Market Barros, R., J. R. Molinas Vega, and J. Saavedra. Outcomes.� NBER Working Papers 12466, 2010. “Measuring Progress Toward Basic National Bureau of Economic Research, Opportunities for All.� Brazilian Review of Cambridge, MA. Econometrics 30 (2). Chetty, R., J. Friedman, N. Hilger, E. Saez, Barros, R., F. Ferreira, J. Molinas Vega, and J. D. Schanzenbach, and D. Yagan. 2010. Saavedra. 2009. Measuring Inequality of Oppor- “How Does Your Kindergarten Classroom tunities in Latin American and the Caribbean. Affect Your Earnings? Evidence from Proj- Washington, DC: World Bank. ect STAR.� NBER Working Paper 16381, Bhorat, H. 2004. “Labour Market Challenges in National Bureau of Economic Research, the Post-Apartheid South Africa.� South Afri- Cambridge, MA. can Journal of Economics 72 (5): 940–77. Currie, J., and D. Thomas. 1999. “Does Head Bhorat, H., and M. Oosthuizen. 2009. “Deter- Start Help Hispanic Children?� Journal of minants of Grade 12 Pass Rates in the Post- Public Economics 74 (2): 235–62. Apartheid South African Schooling System.� Ferreira, F., and J. Gignoux. 2011. “The Mea- Journal of African Economies 18 (4): 634–66. surement of Inequality of Opportunity: 53 South AfricA Economic UpdatE—FocUs on inEqUality oF opportUnity Theory and an Application to Latin Amer- Jalan, J., and M. Ravallion. 2003. “Does Piped ica.� Review of Income and Wealth 57 (4): Water Reduce Diarrhea for Children in 622–57. Rural India?� Journal of Econometrics 112 (1): Ferreira, F., J. Gignoux, and M. Aran. 2011. 153–73. “Measuring Inequality of Opportunity with Leibbrandt, M., I. Woolard, H. McEwen, and Imperfect Data: The Case of Turkey.� Journal C. Koep. 2010. “Employment and Inequality of Economic Inequality 9 (4): 651–80. Outcomes in South Africa.� Southern Africa Gelderblom, D. 2007. “Does Poverty Constrain Labour and Development Research Unit 54 Migration in South Africa? Evidence, Expla- and School of Economics, University of Cape nations and Implications.� Development South- Town. ern Africa 24 (2): 241–55. Roemer, J. 1998. Equality of Opportunity. Cam- Gove, W., M. Hughes, and O. Galle. 1979. bridge, MA: Harvard University Press. “Overcrowding in the Home: an Empiri- South African Reserve Bank. 2012. Quarterly cal Investigation of its Possible Pathological Bulletin: June 2012. Pretoria. Consequences.� American Sociological Review World Bank. 2011a. Accountability in Public Ser- 44 (February): 59–80. vices in South Africa: Selected Issues. Washing- Hoddinott, J., J. Maluccio, J. Behrman, R. Flores, ton, DC. and R. Martorell. 2008. “The Impact of Nutri- World Bank. 2011b. South Africa Economic tion during Early Childhood on Income, Update: Focus on Savings, Investment, and Inclu- Hours Worked, and Wages of Guatemalan sive Growth. July 2011 Issue. Washington, DC. Adults.� The Lancet 371 (February): 411–16. World Bank. 2012. World Development Indicators Hoyos, A., and A. Narayan. 2011. “Inequality of 2012. Washington, DC. Opportunities among Children: How Much World Economic Forum. 2011. The Global Com- Does Gender Matter?� Background Paper petitiveness Report 2011–2012. Geneva. www3. for World Development Report 2012, World weforum.org/docs/WEF_GCR_Report_ Bank, Washington, DC. 2011-12.pdf.