92167 South Africa Economic Update Fiscal Policy and Redistribution in an World Bank Group  |  November 2014  |  Volume 6 Unequal Society South Africa Economic Update Fiscal Policy and Redistribution in an Unequal Society © 2014 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 Country Management Unit for Southern Africa, the Macroeconomics and Fiscal Management Global Practice, and the Poverty Global Practice. 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. Cover photos: top, World Bank staff; bottom, ©iStock.com/stevenallan The report was designed, edited, and typeset by Communications Development Incorporated, Washington, DC. Contents Foreword   v Acknowledgments   vi Executive Summary   1 Section 1 Recent Economic Developments   5 Global economic developments and prospects    5 Recent trends in South Africa    8 Notes   20 Section 2  Fiscal Policy and Redistribution in an Unequal Society: Two Questions Answered    21 The government’s fiscal tool kit to tackle poverty and inequality    22 What is fiscal incidence analysis?    25 Question 1: How do taxes and spending in South Africa redistribute income between the rich and the poor?   28 Question 2: What is the impact of taxes and spending on the rates of poverty and inequality in South Africa?     42 Conclusion: addressing the twin challenges of poverty and inequality in South Africa going forward   45 Notes   47 References   49 Boxes 2.1 Caveats and data limitations    26 2.2 Structure of consumption of the poorest and richest deciles    31 2.3 Comparing the bottom and top of South Africa’s income distribution    35 Figures 1.1 GDP outcomes disappoint   5 1.2 Industrial production growth decelerated across all regions in the third quarter    6 1.3 Capital flows remained robust in 2014    7 1.4 Metals prices are declining as China slows    7 1.5 The economy has failed to sustain the growth momentum    9 1.6 Growth in South Africa continues to trail its emerging market peers    9 1.7 Unemployment remains high   12 1.8 New entrants and the long-term unemployed find it difficult to find employment    13 iii S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y 1.9 The economy needs to generate more than 1.2 million jobs to return to the precrisis employment ratio and absorb new entrants    14 1.10 Headline inflation has moved back within the inflation target band    15 1.11 Petroleum and food price inflation has begun to moderate    15 1.12 The current account deficit widened to 6.2 percent of GDP in 2014q2    16 1.13 Foreigners were net sellers of bonds but net purchasers of equity    17 1.14 The outlook for growth has been successively revised downward    18 1.15 Policy rate expectations have been revised downward since the start of 2014    19 2.1 Composition of taxes   23 iv 2.2 Social spending and subsidies    24 2.3 Definitions of income underpinning the Commitment to Equity fiscal incidence analysis   25 2.4 The progressivity of taxes and transfers    27 2.5 Personal income taxes are progressive    29 2.6 Progressivity of South Africa’s direct tax system: the Kakwani index    30 2.7 Concentration curves of indirect taxes    32 2.8 Incidence of indirect taxes    33 2.9 Concentration curves of all taxes, 2010    34 2.10 Progressivity of direct cash transfers by category: concentration curves for transfers and Lorenz curve for market income    34 2.11 South Africa’s direct cash transfer programs in a national and international perspective   36 2.12 Incidence and concentration curves for free basic services    37 2.13 Progressivity of education spending by category: concentration curves for transfers and Lorenz curve for market income    38 2.14 Share of education benefits by income group    39 2.15 Progressivity of health spending: concentration curves and Lorenz curve for market income   39 2.16 Health spending: incidence and beneficiary households by income group    40 2.17 Concentration curves of benefits, 2010    41 2.18 Concentration coefficients for spending    41 2.19 Change in Gini coefficient: disposable versus market income    44 2.20 Poverty and inequality reducing effectiveness of direct transfers    46 Tables 1.1 GDP components   10 1.2 Aggregate demand components   11 1.3 Likelihood of finding a job for new entrants and those unemployed, 2014q1–2014q2   13 1.4 Economic outlook   17 2.1 General government revenue collections, 2010/11    23 2.2 General government expenditure, 2010/11    24 2.3 Progressivity of direct taxes    28 2.4 Progressivity of indirect taxes    32 2.5 Measuring the progressivity of indirect taxes    33 2.6 Share of benefits going to each income group    36 2.7 Poverty and inequality indicators at each income concept    43 2.8 Average per capita income in each market income decile    43 2.9 How the Gini coefficient for each income concept compares across CEQ countries    44 2.10 How the poverty headcount rate at $2.50 PPP a day for each income concept compares across CEQ countries   45 Foreword Addressing poverty and inequality is South what is the impact of taxes and spending on Africa’s greatest challenge. It is also at the the rates of poverty and inequality in South heart of the National Development Plan. Africa? The analysis puts the results in an Much progress has been made since the end international context that shows that South of apartheid in 1994, with South Africa using Africa is achieving a sizable reduction in pov- its tax and benefit system, as part of its devel- erty and inequality through its fiscal tools. opment program, to alleviate poverty and This Update also reviews some recent inequality. To this end, the government has economic developments and assesses South expanded its social assistance programs and Africa’s economic prospects: domestic factors devoted considerable resources to providing and a fragile global recovery pose significant education and health services and improving headwinds to South Africa’s growth perfor- access to other basic services like electricity mance, but high inequality restrains growth and water. and accentuates social stresses. This sixth edition of the South Africa We hope that the analysis in this Update Economic Update focuses on the role of fis- will help inform and deepen the ongoing cal policy in addressing the twin challenges debate on the broader policies needed to of poverty and inequality in South Africa. attack poverty and inequality as elaborated It provides an analysis based on the innova- in the National Development Plan. tive use of fiscal and household survey data to answer two main questions: How do taxes Asad Alam and spending in South Africa redistribute Country Director for South Africa income between the rich and poor? And World Bank v Acknowledgments This edition was prepared by a core team (Practice Manager, GMFDR). Cecilia Moyo comprising Catriona Mary Purfield (AFCS1), (AFCS1) helped process the report at various Fernando Im (GMFDR), and Gabriela stages. Inchauste (GPVDR), based on analytical The team is grateful for comments from inputs from Ingrid Woolard and Mashekwa South Africa’s National Treasury and the Maboshe (Consultants, UCT), Precious Zikh- South African Reserve Bank. In particular, ali (GPVDR), and Nora Lustig (Tulane Uni- we would like to thank Michael Sachs and Ian versity). Gerard Kambou contributed to the Stuart of the National Treasury who initiated global developments part of section 1 based and facilitated this edition’s special focus on the World Bank’s Global Economic Prospects. section. We are also grateful for the input Peer reviewers were Blanca Moreno-Dodson received from government counterparts, (Lead Economist, GMFDR), Samuel Freije- nongovernmental agencies, and researchers Rodriguez (Lead Economist and Sector who attended a workshop on the preliminary Leader, LCC1C), and Servaas van der Berg findings of the special focus section in May (Consultant, Stellenboch University). The 2014. We would also like to thank Danny Bra- report was prepared under the overall guid- dlow and Chris Loewald and their colleagues ance and supervision of Asad Alam (Country at the South African Reserve Bank for their Director, AFCS1) and Sudarshan Gooptu helpful suggestions and inputs. vi Executive Summary Economic developments Growth momentum in South Africa has and prospects faded progressively since 2011, reflecting The global economic recovery remains growing domestic constraints. Real GDP uneven, as growth in the United States is growth declined from a postcrisis peak of gaining momentum but appears to be at 3.6 percent in 2011 to just 1.9 percent in 2013 risk of stalling in the Euro Area and Japan. and to 1.3 percent y/y in the first half of 2014. U.S. growth is expected to gain pace over the Domestic factors—particularly indus- rest of the year and into 2015 as employment trial action, constraints related to electric- prospects boost real income growth and con- ity and transport infrastructure, and skills fidence. Following the Euro Area’s exit from shortages—have played a major part in the recession in 2013q1, GDP was flat in 2014q2, economy’s lackluster performance. Min- and preliminary data for the third quarter ing and manufacturing output contracted suggest slowing growth momentum amid sharply in the first half of 2014, reflect- weak domestic demand, ongoing balance ing the impact of a five-month strike in the sheet adjustments, a fragmented banking platinum sector and a subsequent strike by sector, and rising geopolitical risks. In Japan, metal workers. Monthly data from the third a sales tax hike in April caused a more signif- quarter suggest that manufacturing produc- icant contraction in activity than expected, tion has subsequently recovered but that while exports failed to pick up. mining output is still contracting as the sec- Prospects continue to be for a slow, tor struggles to regain prestrike production uneven, and fragile strengthening of the levels. One bright spot has been construc- global recovery. Global growth is likely to tion, where activity has been robust. Against remain at about 2.5 percent in 2014 and rise this backdrop, unemployment has remained to 3.2 percent in 2015–17, as the recovery in stubbornly high, and declining food and fuel the United States gains traction, activity in prices have helped ease inflation pressures. the Euro Area and Japan picks up modestly, Fiscal space has declined, and the slow- and growth in China slows. A deepening down in growth has placed public finances of geopolitical risks and renewed bouts of under pressure. The gross stock of public financial market turbulence pose downside debt stood at 45.9  percent of GDP at end- risks to the global forecasts. In Sub-­ Saharan 2013/14, up almost 10  percentage points Africa, growth is expected to rise from since 2010/11. Revenue collections are 4.6 percent in 2014 to 5.2 percent in 2015–17. expected to fall short of the budget target on The Ebola epidemic has reduced growth in account of shortfalls in the corporate income the three most affected countries—Guinea, tax, value-added tax, customs duties, and Liberia, and Sierra Leone—and poses down- fuel levy. side risks to the regional outlook should the Noting that fiscal consolidation could no crisis escalate. longer be postponed, the recent Medium 1 S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Term Budget Policy Statement advanced fis- relations and quickly address power and cal consolidation measures to underpin fis- infrastructure gaps risks further undermin- cal targets and stabilize the debt burden. ing business confidence and investment With the new measures, the Medium Term prospects. South Africa urgently needs to Budget Policy Statement safeguards the accelerate economic growth by addressing decline in the deficit from a revised 4.1 per- infrastructure constraints and broaden- cent of GDP in 2014/15 to 2.5 percent of GDP ing structural reforms if it is to reduce the in 2017/18, which is expected to stabilize the unacceptably high levels of joblessness and gross debt burden at around 49.8  percent inequality prevailing in the economy. 2 of GDP in 2017/18. The package contains a combination of spending cuts and yet to Fiscal policy and redistribution be announced tax increases that will raise in an unequal society 0.6 percent of GDP a year in the next two fis- South Africa has made progress toward estab- cal years. Even so, the fiscal targets remain lishing a more equitable society. Since the subject to downside risks from possible short- end of apartheid, the government has used falls in real GDP growth, rising borrowing its tax resources to fund the gradual expan- costs, and contingent liabilities related to the sion of social assistance programs and scale finances of state-owned enterprises. up spending on education and health ser- The current account remains high, leav- vices. It thus was able to reduce poverty con- ing South Africa susceptible to shifts in inves- siderably. But progress in achieving greater tor sentiment. The current account deficit income equality has proved elusive. Inequal- widened to 6.2  percent of GDP in 2014q2, ity of household consumption, measured by reflecting a deterioration in the trade deficit the Gini coefficient on disposable income, to levels not seen since 2006. Import growth, increased from about 0.67 in 1993 to around while declining, continues to outstrip growth 0.69 in 2011, among the world’s highest. in exports that suffered from industrial With fiscal space becoming more con- action and declining terms of trade. The cur- strained, this Update explores whether the rent account continued to be largely financed government is making the best possible use by capital inflows. of fiscal policy to reduce poverty and inequal- Our forecast for real GDP growth has ity. It provides an analysis based on the inno- been marked down to 1.4  percent for 2014 vative use of fiscal and household survey data and 2.5  percent for 2015, from 2.7  percent to answer two main questions: and 3.4  percent in the previous Update. 1. How do taxes and spending in South This revision largely reflects the impact of Africa redistribute income between the prolonged labor unrest and the constraints rich and the poor? in infrastructure—particularly in electric- 2. What is the impact of taxes and spend- ity—on domestic production and exports. ing on poverty and inequality? Our baseline scenario envisages a slow and This Update is the first study in South gradual return to modest economic growth Africa to use the Commitment to Equity over the medium term as public infrastruc- methodology developed by Tulane Univer- ture investment helps ease infrastructure sity, which allows the impact of fiscal policy constraints and investment and household on inequality and poverty in South Africa to consumption gradually regain pace with be measured and then compared with that in strengthening external demand and improv- 12 middle-income countries that have used ing business and consumer sentiment. the methodology. The outlook is subject to significant In answer to the first question, this domestic and external downside risks. South Update finds that the tax system is slightly Africa is vulnerable to potential bouts of progressive, and spending is highly progres- financial market volatility given its reli- sive. In other words, the rich in South Africa ance on portfolio flows to fund the current bear the brunt of taxes, and the government account deficit. A sharp slowdown in growth effectively redirects these tax resources to in China could adversely affect demand for the poorest in society to raise their incomes. South Africa’s commodity exports. On the On the tax side, fiscal policy relies on a mix domestic front, failing to stabilize labor of progressive direct taxes—such personal income taxes and slightly regressive indi- services net of taxes. Inequality goes from rect taxes—that when combined generate a a situation where the incomes of the richest slightly progressive tax system. Direct taxes decile are more than 1,000 times higher than (personal income and payroll taxes) are the poorest to one where they are about 66 progressive, since the richer deciles pay a times higher. As a result, the Gini coefficient proportionally higher share of total direct on income falls from 0.77, where it lies before tax collections than their share of market various taxes and social spending programs income. And because these taxes make up are applied, to 0.59 after these fiscal inter- a fairly high share of GDP, they help narrow ventions are incorporated. Still, the level of the gap in incomes between the rich and the inequality remaining is higher than what 3 poor. Indirect taxes are slightly regressive: all other countries in this sample start with the four poorest deciles contributed about before they apply fiscal policies. 5.0 percent of total indirect tax collections, In sum, fiscal policy already goes a long compared with their share of 4.8  percent way toward redistribution. Even so, the level in total disposable income. This regressiv- of inequality and poverty in South Africa ity at the lower end of the income distribu- after taxes and spending remains unaccept- tion largely reflects the impact of excises, as ably high. But South Africa’s fiscal deficit value-added and fuel taxes are progressive. and debt indicators show that the fiscal space South Africa uses its fiscal instruments to spend more to achieve even greater redis- very effectively, achieving the largest reduc- tribution is extremely limited. Addressing tions in poverty and inequality of the 12 mid- the twin challenges of poverty and inequal- dle-income countries. As a result of South ity going forward in a way consistent with Africa’s fiscal system, some 3.6 million people fiscal sustainability will require better qual- are lifted out of poverty, measured as those ity and more-efficient public services. It will living on less than $2.50 a day (in purchasing also require faster and more-inclusive eco- power parity dollars). The rate of extreme nomic growth to address the need for jobs poverty is cut by half. The share of the popu- and higher incomes at the lower end of the lation living on $1.25 a day or less falls from income distribution—to narrow the gap in 34.4  percent to 16.5  percent, ref lecting incomes between the rich and the poor and the impact of cash transfers and free basic to reinforce the effectiveness of fiscal policy. SECTION 1 Recent Economic Developments Global economic developments prospects boost real income growth and con- and prospects fidence. Investment is projected to rise in line with strong corporate profits and favor- The recovery in high-income countries able financing conditions. continues but remains very uneven Meanwhile, growth in the Euro Area and Growth was weaker than expected so far this Japan appears to have stalled. The Euro year, with disappointing economic activity in Area’s modest recovery appears to be stall- several major countries (figure 1.1). Growth ing amid weak domestic demand, ongoing in the United States, the Euro Area, and balance sheet adjustments, a fragmented Japan averaged 0.6 percent in the first half banking sector, and rising geopolitical risks. of 2014, but their recoveries have diverged Euro Area GDP was flat in 2014q2, following considerably. a small uptick in 2014q1. Output in Germany, Growth in the United States has been Italy, and France contracted in the second gathering momentum. U.S. growth recovered quarter. Weak growth and falling inflation strongly in 2014q2 from the weather-induced prompted the European Central Bank to fur- sharp contraction in 2014q1, aided by rising ther loosen monetary policy and announce employment and investment growth, a still- measures to support bank lending to house- accommodative monetary policy, and easing holds and firms. Preliminary data from the fiscal consolidation. The recovery is expected Euro Area for the third quarter also suggest to gain pace over the remainder of the year slowing growth momentum. In Japan, a sales and to continue well into 2015 as employment tax hike in April caused a more significant Figure GDP outcomes disappoint 1.1 2013, rst half 2013, second half 2014, rst half 5 4 GDP growth, annualized (%) 3 2 1 0 –1 United States Japan Euro Area Developing countries Source: World Bank and Haver Analytics. 5 S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y contraction in activity than expected, while Growing uncertainty begins to exports failed to pick up despite a weak yen. weigh on financial markets Although unemployment is low, labor force Notwithstanding a weak start to the year, participation remains below precrisis levels, equity markets had risen to all-time highs and wage growth has remained weak. and government bond yields had fallen to record lows by September only to witness Developing country growth was steady in the considerable volatility in October. Through first half, but there are signs of a slowdown September, U.K. and U.S. benchmark stock Across the major emerging markets, growth indexes, in particular, had risen to record 6 in the first six months of 2014 has been highs on the back of strengthening macro broadly steady. Following a subdued first data, still-accommodative U.S. monetary quarter, growth in developing countries policy, and credit easing by the European accelerated to an annualized rate of 4.2 per- Central Bank, but growing uncertainty cent in 2014q2. Developing country indus- about global growth prospects has started to trial production expanded at an annualized weigh on investor sentiment. The European rate of 4.8 percent in the first half of 2014. Central Bank’s announced policy measures Among high-income While still higher than that of high-income have led to a weakening of the euro against countries, industrial production growth the dollar over the past months. This has countries growth remained below the average growth rate of generated capital flows into U.S. long-term patterns have 7.6 percent achieved between 2000 and 2013. bond markets but also into risky assets such After a soft start to the year, growth acceler- as emerging market stock markets. The con- diverged. Growth in ated in China in 2014q2 to reach 7.7 percent, tinuing accommodative monetary stance the United States is reflecting the impact of the mini-stimulus of the European Central Bank could help package launched in March. In the wake of counteract somewhat the global impact of gathering momentum, national elections, improved business sen- eventual monetary tightening in the United but the Euro Area and timent boosted growth in India. However, States. escalating geopolitical tensions weighed on Capital f lows to developing countries, Japanese economies growth in Eastern Europe. which weakened in early 2014 in a market sell- appear to be stalling Incoming data for the third quarter show off, resumed strongly beginning in March that industrial production decelerated across 2014 and were up 13.3 percent through end- developing countries. Industrial production August from the year earlier (figure 1.3). grew at a seasonally adjusted annualized rate Much of this increase reflects bond issuance (saar) of 4.7  percent q/q in August, down by Chinese entities, which accounts for an from 5.2  percent in July, reflecting a slow- unprecedented quarter of all developing- down in the large emerging countries (figure country bond issuance. 1.2). Industrial production slowed in China More generally, year-to-date gross capital and Mexico and contracted 6.8 percent q/q flows have increased to developing countries (saar) in India and 6.2 percent in Brazil. in all regions, except Europe and Central Figure Industrial production growth decelerated across all regions in the third quarter 1.2 10 Industrial production growth (%) Developing countries 5 World High-income countries 0 -5 Jan. Mar. May July Sep. Nov. Jan. Mar. May July 2013 2013 2013 2013 2013 2013 2014 2014 2014 2014 Source: World Bank. Figure Capital flows remained robust in 2014 1.3 Bank Bond Equity 200 150 $ billions 100 7 50 0 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 East Asia Europe and Latin America Middle East and South Sub-Saharan and Paci c Central Asia and Caribbean North Africa Asia Africa Note: Data are for January–August. Source: World Bank. Capital flows to developing countries, Asia, where bank flows have dropped sharply, reflecting good crop prospects. Meanwhile, partly as a result of tensions in Ukraine and the decline in the price of metals was halted which weakened in sanctions on the Russian Federation. in 2014q3, with the World Bank metals price index rising 2.6  percent (q/q) (figure 1.4). early 2014 in a market Robust supply and weakening Base metals drove the increase in prices, sell-off, were up demand from China weigh on various rising 5.3  percent (q/q), while iron ore saw commodity and metal prices a steep drop in prices. The strengthening 13.3 percent through Oil prices have moved down from a range of in metal prices during 2014q3 was broad- end-August from $100 per barrel to $111 per barrel between based, with prices of nickel, copper, lead, June and September 2014 to a low of $83 aluminum, and zinc all increasing. However, the year earlier per barrel by mid-October. Robust supply with Chinese markets in surplus and capacity prospects due to increased output from continuing to rise, metal prices are expected Iraq, Libya, and the United States, along to decline more than 5 percent in 2014. The with weak economic data for China and World Bank precious metals price index, Europe, are placing downward pressure on which declined 0.5  percent in 2014q3 com- oil prices. pared with the previous quarter, is 4.5 per- Agricultural prices experienced broad- cent lower than a year ago. The index fell to based declines in 2014q3, with the overall a four-year low in September, with the prices price index down 5  percent for the quar- of platinum and gold down 1.3 percent and ter and 3  percent lower than a year ago, 3.6 percent (y/y), respectively. Figure Metals prices are declining as China slows 1.4 200 Nominal $ (index, 2010 = 100) 150 Energy Agriculture 100 Metals and minerals 50 Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. 2007 2008 2009 2010 2011 2012 2013 2014 Source: World Bank. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Prospects are for a modest pickup challenging postcrisis global environment in global growth, led by the United where external demand is weaker and there States, with developing country is a withdrawal of fiscal stimulus, especially growth picking up more slowly in major emerging markets. In addition, Prospects continue to be for a slow but structural bottlenecks—including poor uneven strengthening of the global recovery, business environments, inadequate public which is likely to be fragile amid rising risks. infrastructure, and weak global trade—have Global growth is likely to remain at about capped longer term growth and productivity 2.5 percent in 2014, similar to that in 2012– gains for a number of developing countries. 8 13, with global activity struggling to gain On the back of the strengthening recovery momentum. Global growth is expected to in high-income countries, growth in develop- reach 3.2 percent in 2015–17 as the recovery ing countries is expected to pick up to 5.2 per- in the United States gains traction, activity in cent in 2015–16. Although broadly in line with the Euro Area and Japan picks up modestly, potential, this is about 2  percentage points and growth in China slows. A deepening of lower than the 7.3 percent average of the pre- geopolitical risks and renewed bouts of finan- crisis boom years, highlighting the need for Growth in Sub‑Saharan cial market turbulence pose downside risks structural reforms to address capacity con- to these forecasts. In Sub-­ S aharan Africa, straints and boost medium-term growth. Africa is expected to the Ebola epidemic has reduced growth pros- In China, growth is expected to slow from rise from 4.6 percent pects in the three most affected countries— 7.7 percent in 2013 to 7.4 percent in 2014, and Guinea, Liberia, and Sierra Leone. World to average 7.1 percent in 2015–17 as the coun- in 2014 to 5.2 percent Bank estimates suggest that the forgone try transitions away from an investment-led in 2015–16 output for these three countries could reach growth strategy toward greater emphasis on $359 million in 2013 prices. domestic consumption. Developing countries In the United States, GDP growth for 2014 with significant trade exposure to the United is expected to expand by 2.2  percent, up States should gradually gain momentum, from 1.9 percent in 2013, and the improving while those reliant on Euro Area demand are job market and upturn in investment spend- expected to face headwinds that should grad- ing is expected to lift growth to 3 percent or ually ease over the course of 2015 and 2016 thereabouts in 2015–17. In the Euro Area, a as growth in the Euro Area recovers. Com- slow improvement in credit and labor mar- modity exporters, particularly metals pro- ket conditions should provide some momen- ducers in Sub-­ S aharan Africa, will remain tum, but investment prospects remain under pressure as growth in China moder- subdued and precautionary savings are still ates, and demand and prices of certain com- high. Exports should gradually pick up, sup- modities, including metals and coal, are ported by strengthening demand from the adversely affected. Reflecting robust growth United States and a weakening euro. Against in Nigeria, continuing infrastructure invest- this backdrop, growth is expected to aver- ment, and increased agricultural production age 0.8  percent in 2014 and gradually rise in the region, growth in Sub-­ Saharan Africa to 1.3 percent in 2015, although risks to the is expected to rise from 4.6 percent in 2014 projection are to the downside, particularly to 5.2 percent in 2015–16. The normalization in light of geopolitical tensions. In Japan, of U.S. monetary policy will gradually raise monetary policy accommodation and reform global borrowing costs in 2015, despite the commitments will provide ongoing support, expected loosening in the Euro Area, and but fiscal consolidation is expected to keep will sharpen investor concerns about eco- domestic demand subdued throughout 2015, nomic fundamentals. with exports only recovering slowly. Real GDP growth is expected to average 1 percent Recent trends in South Africa in 2014, down from 1.5 percent in 2013, and pick up moderately to 1.2 percent in 2015. Growth continues to disappoint in the face For developing countries, growth is of pressing social and development needs expected to edge up to 5.0 percent in 2014. Growth momentum in South Africa has This rate remains below long-run his- faded progressively since 2011. Real GDP torical average levels and reflects a more growth declined from a postcrisis peak of 3.6 percent in 2011 to just 1.9 percent in 2013. failed to achieve two consecutive quarters Following a –0.6  percent q/q (saar) output of rising economic growth rates. Figure 1.5 contraction in 2014q1, the economy managed shows that an acceleration in growth has to grow by a paltry 0.6 percent q/q in 2014q2. always been followed by a moderation in the As a result, real GDP rose by just 1.3 percent growth rate in the next quarter. This erratic y/y in the first half of 2014, the lowest head- pattern reflects the impact of growing labor line growth since the onset of the global unrest, increasingly binding infrastructure financial crisis. Since 2011, the economy has and skills constraints, and still-weak external Figure The economy has failed to sustain the growth momentum 9 1.5 Year over year 5 4 Real GDP growth (%) 3 Growth momentum 2 fades as labor strife, power and 1 other infrastructure 0 2011q1 2011q2 2011q3 2011q4 2012q1 2012q2 2012q3 2012q4 2013q1 2013q2 2013q3 2013q4 2014q1 2014q2 constraints, and Quarter over quarter moderating export 6 prices take their toll Real GDP growth, seasonally adjusted (%) 4 2 0 –2 2011q1 2011q2 2011q3 2011q4 2012q1 2012q2 2012q3 2012q4 2013q1 2013q2 2013q3 2013q4 2014q1 2014q2 Note: Blue bars indicate quarters in which growth picked up relative to the previous quarter, whereas red bars flag quarters in which economic activity moderated. Source: Statistics South Africa. Figure Growth in South Africa continues to trail its emerging market peers 1.6 175 India Real GDP, seasonally adjusted 150 (index, 2007q1 = 100) Indonesia Turkey 125 Brazil Russian Federation South Africa 100 75 2007q1 2008q1 2009q1 2010q1 2011q1 2012q1 2013q1 2014q1 Source: Federal Reserve Bank of St. Louis, Statistics South Africa, and World Bank staff calculations. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y demand. Growth in South Africa also contin- in 2014q2. Manufacturing production fell by ues to trail most of its peers (figure 1.6), and 2.1 percent q/q (saar) in 2014q2, somewhat if it fails to pick up substantially, progress in milder than the contraction in 2014q1. Even reducing South Africa’s high rates of unem- so, the manufacturing sector subtracted 0.8 ployment, poverty, and inequality will prove and 0.4 percentage point, respectively, from challenging. headline growth in the first and second quar- Domestic factors, particularly industrial ters. In August, however, manufacturing pro- action and supply-side constraints related duction recovered smartly from the strike, to infrastructure and skills shortages, have rising 2.2 percent m/m (saar). 10 played a major part in the economy’s lack- The one bright spot has been the construc- luster performance. Output in the primary tion sector, where activity remained robust, sector contracted in the first two quarters of with output expanding at close to 5 percent 2014, though at a slower rate from –17.2 per- q/q (saar) in 2014q2. Real value added in cent q/q (saar) in 2014q1 to –5.1 percent in the tertiary sector also rose by 1.8  percent 2014q2. This contraction was bought about q/q (saar), sustaining growth at 2014q1 rates, by a protracted five-month strike in the helped by robust performance of transport, Domestic factors, platinum sector that caused output in the storage, and communication and general mining and quarrying sectors to contract by government services (table 1.1). particularly –24.7 percent and –9.4 percent, respectively, Labor unrest, along with policy uncer- supply‑side in the first two quarters of the year. As a tainty and increasingly binding electricity result, the mining sector subtracted 1.6 per- supply constraints, has shaken already bat- constraints related centage points from headline GDP growth tered business confidence. Despite a 5-index to infrastructure and in 2014q1 and another 0.5 percentage point point rebound in the RMB/BER Business from headline growth in the second quarter. Confidence Index to 46 in 2014q3, most skills shortages, have Production data through August showed that respondents still rate current business condi- played a major part the mining sector continued to struggle to tions as less than satisfactory. The index has regain prestrike production levels, with out- been below the neutral threshold of 50 since in the economy’s put contracting 3.1 percent m/m (saar). 2013q2. Other business confidence indica- lackluster performance Reflecting the spillover effects from this tors such as the SACCI Business Confidence strike on the motor vehicle, parts, and acces- Index and PMI Expected Business Condi- sories and other transport equipment sec- tions also point to a weak business environ- tor, as well as the impact of a separate strike ment, despite a mild recovery prompted by by metal workers in this sector, real value the end of the five-month-long labor unrest added in the secondary sector also declined in the mining sector. Table GDP components 1.1 Percent, seasonally adjusted and annualized Sector 2010 2011 2012 2013q1 2013q2 2013q3 2013q4 2013 2014q1 2014q2 GDP at market prices 3.1 3.6 2.5 0.8 3.2 0.7 3.8 1.9 –0.6 0.6 Primary sector 4.0 0.2 –2.0 7.5 –4.7 8.9 12.8 2.9 –17.2 –5.1 Agriculture, forestry, and fishing 0.4 –0.1 2.0 –4.4 –3.0 3.6 6.4 2.3 2.5 4.9 Mining and quarrying 5.7 0.3 –3.6 13.4 -5.4 11.4 15.7 3.1 –24.7 –9.4 Secondary sector 4.5 2.7 1.8 –5.9 9.6 –4.5 9.2 1.0 –2.7 –0.9 Manufacturing 5.5 3.3 2.1 –7.9 11.7 –6.6 12.3 0.8 –4.4 –2.1 Electricity, gas, and water 2.5 1.5 –1.6 –2.8 5.1 3.8 –5.6 –0.4 0.1 –0.6 Construction 0.7 0.3 2.3 2.5 2.3 2.1 3.1 2.8 4.9 5.0 Tertiary sector 2.5 4.1 3.2 2.0 2.2 1.3 1.5 2.0 1.8 1.8 Wholesale and retail trade, catering, and accommodations 3.8 4.4 3.8 2.1 3.1 1.3 2.3 2.2 2.1 –0.2 Transport, storage, and communication 2.0 3.1 2.4 2.1 1.5 2.6 1.6 1.9 1.7 4.0 Finance, real estate, and business services 2.2 4.7 3.7 3.3 3.5 1.2 1.5 2.4 2.0 1.5 General government services 3.1 4.2 2.8 0.1 0.2 0.4 0.9 1.5 1.7 2.9 Personal services 0.4 2.3 2.1 1.2 1.6 1.6 1.3 1.8 1.0 1.2 Source: Statistics South Africa. From the demand side, growth in final 2014q2. Higher capital expenditure by the domestic expenditure moderated to 1.8 per- general government was more than offset by cent q/q (saar), ref lecting rapidly slowing contractions in capital outlays by both state- household consumption and depressed owned enterprises (particularly in electric- investment (table 1.2). Household spending, ity and transport) and private businesses. which represents about two-thirds of aggre- After gaining some pace in 2013, investment gate demand, continued its trend of declin- by the private sector worryingly contracted ing growth that started in 2013q3, rising by 1.1 percent in 2014q2. Increasing volatil- by a mere 1.5 percent q/q (saar) in 2014q2. ity and uncertainty emanating domestically The slowdown was particularly pronounced (industrial action, wage settlements, produc- 11 in the areas of durable and semidurable tion stoppages, policy uncertainty, and the goods expenditure. Widespread unemploy- deteriorating domestic economic outlook) ment, the high number of working days lost and growing external concerns (rising geo- due to industrial action (7.5  million work political tensions and uncertainties about days were lost due to strike action in the the growth outlook in China and the Euro first half of 2014, compared with 1.8  mil- Area, and the impact of monetary normal- lion in the first half of 20131), still high lev- ization in the United States) likely delayed Growth in final els of household indebtedness (debt stood or halted long-term investment decisions domestic expenditure at 73.5  percent of disposable income in by private businesses. Moreover, capacity 2014q2), and tightening credit conditions utilization by large manufacturing enter- moderated to that resulted in credit growth to the house- prises still remains well below precrisis lev- hold sector slowing through end-August— els at 80.4 percent in 2014q2—or about 1.0 1.8 percent quarter all have worked to constrain consumption. and 4.0  percentage points below the levels over quarter, Retail sales for July and August also point to recorded in 2013q2 and 2008q2, respec- continuing weak momentum in household tively. Finally, the pace of contraction in net reflecting rapidly consumption. exports eased, shaving 1.3 percentage points slowing household Gross fixed capital formation growth from headline growth in 2014q2, compared also edged sharply lower, from 2.6  percent with 3.5  percentage points it subtracted in consumption and q/q (saar) in 2014q1 to just 0.5  percent in 2014q1. depressed investment Table Aggregate demand components 1.2 Percent, seasonally adjusted and annualized, unless otherwise noted Component 2010 2011 2012 2013q1 2013q2 2013q3 2013q4 2013 2014q1 2014q2 Total final consumption 4.4 4.7 3.7 2.5 2.3 2.0 2.0 2.5 1.7 1.5 Final consumption expenditure by household (PCE) 4.4 4.9 3.5 2.4 2.5 2.1 2.0 2.6 1.8 1.5 Durable goods 18.8 16.1 11.1 5.9 12.6 9.4 6.9 7.9 2.8 1.4 Semidurable goods 3.6 5.9 6.2 7.6 8.5 7.1 3.1 6.7 6.1 2.0 Nondurable goods 1.8 3.1 2.7 2.4 2.7 0.5 0.2 2.2 –0.4 0.7 Services 4.0 3.6 1.7 0.1 –2.1 0.1 1.7 0.3 2.2 2.0 Final consumption expenditure by general government 4.4 4.3 4.0 2.8 1.7 1.5 2.0 2.4 1.4 1.6 Gross fixed capital formation (investment) –2.1 4.2 4.4 3.8 5.6 7.0 3.1 4.7 2.6 0.5 General government –9.2 9.7 6.2 1.5 2.6 9.0 9.2 3.5 4.6 8.9 Public corporations –1.5 –0.6 4.9 –1.6 0.1 0.4 0.8 3.1 6.0 –0.7 Private business enterprises –0.5 4.6 3.9 6.2 8.2 8.6 2.4 5.5 1.0 –1.1 Change in inventories (R millions) –1,988 7,865 9,850 7,868 16,388 3,260 –22,304 1,303 –14,404 –12,348 Residual item (R millions) 1,390 –8,956 –7,977 –4,398 –11,359 –18,092 –23,811 –14,415 –27,794 –27,155 Gross domestic expenditure 3.9 4.6 4.0 5.3 3.2 –0.8 –3.6 2.2 2.7 1.8 Exports of goods and services 9.0 6.8 0.4 5.3 3.2 –0.8 –3.6 4.2 5.4 –11.4 Imports of goods and services 11.0 10.0 6.0 21.5 7.3 7.0 –18.9 4.7 16.3 –5.2 Net exports (R millions) –85,212 –107,848 –140,640 –162,545 –163,567 –155,561 –117,155 –149,707 –134,388 –141,070 Gross domestic product 3.1 3.6 2.5 0.8 3.2 0.7 3.8 2.2 –0.6 0.6 Source: South African Reserve Bank. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Labor markets: employment growth is being job seekers in 2014q2. Of this number, fig- driven mainly by gains in the community and ure 1.8 shows that about 38.7  percent were social services sector, while the manufacturing new entrants trying to find a first job, and and mining sectors continue to shed jobs another 18.3  percent were individuals who Against the backdrop of disappointing last worked more than five years ago. Table growth, labor market outcomes have, unsur- 1.3 suggests that only about 9 percent of each prisingly, shown very little improvement of these groups of unemployed will be able to in the aftermath of the global financial find a job in the next three months. To put crisis. Both narrow and broad unemploy- it in context, a person who has been unem- 12 ment (which includes discouraged workers) ployed for less than a year is twice as likely remained elevated in 2014q2 at 25.5 percent to find a job in the next three months as a and 33.4  percent, respectively, just 0.1  per- person trying to find a first job, or a person centage point shy of their postcrisis peaks who has not held a job in the last five years. (figure 1.7). It is estimated that the economy will need Employment rose in the second quarter, to generate more than 1.2 million jobs if it almost entirely due to gains in the commu- is to close the “ jobs gap”—the difference The economy needs nity and social services and transport sectors between the current level of employment and and in private households. However, employ- the level of employment required to return to create 1.2 million ment in the traded-goods sectors remains to the precrisis absorption rate (the ratio of jobs to return depressed. Agriculture shed about 39,000 employed persons to working-age popula- jobs q/q. Employment in manufacturing tion) and absorb new entrants (figure 1.9).3 employment to contracted by 60,000 jobs q/q while employ- Even under the most optimistic of the sce- precrisis levels and to ment in mining2 declined by 1,000 jobs q/q. narios, our calculations suggest that it would Between 2008q4 and 2014q2, about 352,000 take at least three years to close this gap.4 absorb new entrants jobs were lost in the manufacturing sec- Against this backdrop, the take-up of the new tor. Agriculture shed another 137,000 jobs, youth employment tax incentive is encourag- whereas mining employment fell by 28,000 ing. As of August 2014, some 23,500 employ- jobs during the same time period. These job ers claimed the incentive for at least 209,000 losses have been offset by gains in transport young workers.5 But given the strained labor (117,000), finance (243,000), and community relations environment and policy uncertainty and social services (700,000). However, this (such as restrictions on labor brokers, imple- growth in employment was insufficient to mentation of a national minimum wage, and absorb new labor market entrants since the proposals to introduce a strike ballots for all global financial crisis began (see below). workers), together with weak investor confi- Unemployment is largely structural dence and the overall slowdown in growth, and long term. About 7.6  million people risks to the employment outlook are to the were either unemployed or discouraged downside. Figure Unemployment remains high 1.7 Broad unemployment Under average job creation (employment growth of 2.6 percent) rate Under fast job creation (employment growth of 4.5 percent) 40 30 Percent 20 10 0 2008q1 2009q1 2010q1 2011q1 2012q1 2013q1 2014q1 2015 2020 2025 Note: Shaded area represents forecast. Source: Quarterly Labour Force Surveys, Statistics South Africa. Figure New entrants and the long-term unemployed find it difficult to find employment 1.8 About 7.6 million persons were either unemployed or have left the labor force due to bleak job prospects in 2014q2 6 Narrow unemployment rate, 25.5% 4 Millions Broad unemployment rate, 33.4% 13 2 0 Unemployed Discouraged workers New entrants and persons who last worked more than five years ago represent the vast majority of those unemployed New entrants and long-term unemployed Last worked more than represent about ve years ago 18% 57 percent of those New entrants 39% unemployed Job losers, job leavers, and re-entrants 43% Source: Quarterly Labour Force Surveys, Statistics South Africa. Table Likelihood of finding a job for new entrants and those unemployed, 2014q1–2014q2 1.3 Percent Other not Employed Unemployed Discouraged economically active Unemployed 12.9 65.3 7.3 14.6 Short term (less than a year) 20.8 57.1 7.9 14.2 Long term (more than a year) 9.0 69.2 7.0 14.8 New entrants 8.8 66.0 6.3 18.9 Have not worked in more than five years 8.7 69.2 7.5 14.6 Source: Quarterly Labour Force Surveys, Statistics South Africa. Fiscal policy: weaker than expected economic budget had targeted a gradual reduction in growth challenges fiscal adjustment plans the budget deficit from 4.0 percent of GDP Fiscal space has declined, and the slow- in 2014/15 to 2.8 percent of GDP by 2016/17, down in growth has placed public finances to help stabilize the debt burden by the end under pressure. The stock of public debt of this period. But the budget had assumed has increased considerably since the global that growth would reach 2.7 percent in 2014 financial crisis. The gross debt burden of and rise above 3  percent by the end of the the general government stood at 45.9  per- forecast period. The marked slowdown in cent of GDP at end-2013/14, up about 10 per- economic growth in 2014 relative to the fore- centage points since end-2010/11. The 2014 cast has made these targets difficult to reach. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y The economy needs to generate more than 1.2 million jobs to return to the Figure precrisis employment ratio and absorb new entrants 1.9 Jobs gap Under average job creation (employment growth of 2.6 percent) Under fast job creation (employment growth of 4.5 percent) 2.0 1.5 1.0 0.5 14 Millions 0.0 –0.5 –1.0 –1.5 2009q1 2010q1 2011q1 2012q1 2013q1 2014q1 2015 2020 Note: The initial starting point is prerecession, when we assume that the job gap was 0 in 2008q4. The difference between the absorption rate in each quarter and the one in 2008q4 gives an estimate of the number of jobs needed. From 2009q1 to 2014q2 (bars), the working-age population is reported by the Quarterly The Medium Term Labour Force Survey. Data for 2015–20 (red and blue lines) are based on two assumptions (fast and average job creation, and average working-age population growth of 1.8 percent for 2011q1–2014q2). Source: Quarterly Labour Force Surveys, Statistics South Africa. Budget Policy Statement set out a Reflecting concerns about the implications The adjustment package comprises spend- of the growth slowdown for budget finances, ing and revenue measures of just more than package of measures in 2014q2 Standard and Poor’s downgraded 0.6  percent of GDP a year in the next two to show how it will South Africa’s long-term foreign currency– years. The key adjustment measures include: denominated debt by one notch to BBB–, • A reduction in the annual noninterest undertake fiscal one notch above subinvestment grade, while spending ceiling by 0.2 percent of GDP consolidation and Fitch revised its outlook on its BBB rating on (R 10 billion) in 2015/16 and 0.3 percent the same debt from stable to negative. of GDP (R 15 billion) in 2016/17. safeguard sustainable The October 22 Medium Term Budget • Tax policy and administrative reforms public finances Policy Statement (MTBPS) proposes a pack- to generate additional revenue of about age of spending and tax measures to safe- 0.3 percent of GDP a year in the next two guard the fiscal consolidation path set out years. in the 2014 budget. It envisages much weaker • Strengthening the budget preparation economic growth: real GDP is expected process, including a review of outer year to grow 1.4  percent in 2014 and to remain spending plans to ensure more-efficient subdued over the medium term, recovering resource allocation and the inclusion of to 3.0  percent only in 2017, well below the a contingency or fiscal buffer of 0.9 per- National Development Plan targets. Due cent of GDP (R 45 billion) in the 2017/18 to weaker growth, tax revenue collections spending ceiling. in 2014/15 are projected to fall short of the • A freeze on the government’s personnel budget target by some 0.3  percent of GDP headcount. (R 10 billion) because of the underperform- • Deficit neutral financing of state-owned ing corporate income tax, customs duties, enterprises through the sale of nonstra- value added tax, and fuel levy. Noting that tegic government assets. a turning point had been reached and that The exact details of the tax measures will fiscal consolidation could no longer be post- be informed by the findings of the ongoing poned, the MTBPS advanced fiscal consoli- Davis Tax Commission and will be imple- dation measures to safeguard the planned mented in the 2015/16 budget. adjustment path to stabilize the debt burden. The adjustment package is welcome, As a result of the new package, the MTBPS but the deficit and debt targets still remain safeguards the decline in the deficit from subject to considerable downside risks. a revised 4.1  percent of GDP in 2014/15 to Lower-than-projected growth continues to 2.5  percent of GDP in 2017/18 to help the represent a significant challenge to revenue gross debt burden stabilize at 49.8 percent of collections. Rising borrowing costs could GDP in 2017/18. put further pressure on fiscal consolidation efforts, particularly as monetary policy nor- within the South African Reserve Bank’s malizes in some advanced countries and inflation target band (5.9 percent y/y in Sep- global interest rates rise. The targets assume tember). After five consecutive months in that state-owned enterprises can repair their which headline CPI inflation remained above finances and government guarantees of their the upper threshold of the inflation target, borrowing can be contained within existing driven mainly by strong increases in food limits. Finally, the wage bill poses consider- and nonalcoholic beverages (figures 1.10 and able risk as the targets assume that wage 1.11), inflation moderated mainly on account growth is contained to inflation and person- of falling petroleum prices. Core inflation nel numbers stabilize. Clearly if the ongoing (excluding food and nonalcoholic beverages, 15 public sector wage negotiations result in an petroleum, and energy) also eased, falling to above-inflation pay increase for civil ser- 5.6 percent y/y in September compared with vants, further measures would be necessary 5.8 percent y/y in August. The rand depreci- to offset the impact and safeguard the adjust- ated by 33 percent against the dollar between ment targets. the end of 2012 and the end of September 2014, and further rand depreciation could Inflation and monetary policy: the Reserve pose upside risk to the inflation outlook. Headline CPI inflation Bank is gradually normalizing monetary policy With headline inflation still close to the appears to have Headline CPI inf lation appears to have upper limit of the inflation target band, we peaked at 6.6 percent in May and June and expect the Reserve Bank to continue gradual peaked at 6.6 percent has subsequently moderated, falling back tightening of its monetary stance to anchor in May and June 2014 Figure Headline inflation has moved back within the inflation target band 1.10 Headline in ation Core in ation In ation target band 8 6 Percent 4 2 0 Jan. July Jan. July Jan. July 2012 2012 2013 2013 2014 2014 Source: Statistics South Africa. Figure Petroleum and food price inflation has begun to moderate 1.11 25 Petroleum 20 Price increase (%) 15 Administered prices 10 5 Housing and utilities Food and nonalcoholic beverages 0 Jan. July Jan. July Jan. July 2012 2012 2013 2013 2014 2014 Source: Statistics South Africa. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y inflation expectations. The Monetary Pol- deficit in the services, income, and current icy Committee kept the repurchase rate transfer account reflected one-off gross divi- unchanged in September at 5.75  percent. dend receipts from abroad and a decline in The reference rate was previously increased gross dividend payments to nonresidents. by 25 basis points in July. This followed an But in 2014q2 the trade deficit deteriorated increase of 50 basis points in January. to 2.8 percent of GDP, a level not seen since Weakening oil prices and moderating 2006q4. Although the rand value of merchan- food prices are expected to contribute to a dise imports contracted in 2014q2 by 9.7 per- more benign backdrop for domestic infla- cent (saar)—mainly due to lower volumes of 16 tion in the second half of 2014. Depending oil imports, base metals, machinery, electri- on data developments, the Reserve Bank has cal equipment, and vehicles and transport signaled that it will continue on its path of equipment—it was insufficient to offset the gradual monetary policy normalization given sharper contraction in export growth. Indus- how close inflation remains to the upper tar- trial action in mining and manufacturing get band in the context of the upside risks resulted in the export earnings of nongold from the exchange rate and wage increases goods exports contracting by 23.4  percent The improvement that exceed productivity gains. In recent (saar) in 2014q2. Lower external demand— statements, the Reserve Bank has noted that particularly from Europe and Asia—and in the current despite the increase in policy rates in 2014, declining U.S. dollar prices for South Afri- account balance in real interest rates remain slightly negative ca’s exports that have caused South Africa’s and supportive of the weak domestic econ- terms of trade to deteriorate also contributed 2014q1 proved to omy, while acknowledging that the sources to poor export performance. be short-lived of subpar growth remain outside the realms Absent substantial dividend receipts from of monetary policy. It called for an improve- abroad, the services, income, and current ment in relationships between management transfers deficit widened from 2.4 percent of and labor “to get South Africa back to work,” GDP in 2014q1 to about 3.4 percent of GDP while stressing the urgent need to implement in 2014q2. A quarter-on-quarter improve- structural reforms to achieve higher and ment in the net services account was coun- more inclusive growth.6 tered by deterioration in the net income and net current transfers accounts. External sector: the current account deficit The widening current account was remains high, leaving South Africa susceptible largely financed by capital inflows. Foreign- to shifts in global financial market sentiment ers were net sellers of bonds (–$3.7 billion) The current account deficit widened to but net purchasers of equity ($2.9  billion), 6.2 percent of GDP in 2014q2 (figure 1.12). with cumulative nonresident portfolio out- A short-lived improvement to 4.2 percent of flows amounting to –$874 million (through GDP in 2014q1 on account of a sharply lower 2014q3), contrasting with the net inflows Figure The current account deficit widened to 6.2 percent of GDP in 2014q2 1.12 Trade balance Net services Current transfers Terms of trade Income Current account balance (excluding gold) 5 125 0 115 Index (2005 = 100) Percent of GDP –5 105 –10 95 2006q1 2007q1 2008q1 2009q1 2010q1 2011q1 2012q1 2013q1 2014q1 Source: South African Reserve Bank. of $4.9  billion and $9.2  billion recorded 2.5  percent for 2015, from 2.7  percent and in 2013 and 2012 during the same period 3.4 percent in the previous update (table 1.4 (figure 1.13). and figure 1.14). This revision largely reflects Foreign direct investment rose to reach the impact of prolonged labor unrest and the 1.3  percent of GDP in the second quarter, constraints of capacity—particularly in the helping bring total capital inflows to 3.4 per- electricity sector—on domestic production cent of GDP in 2014q2. Although the level of and exports. errors and omissions has fallen, they remain Our baseline scenario envisages a slow and large, at 1.3 percent of GDP, and account for gradual return to modest economic growth about a fifth of the overall current account over the medium term. The pickup in eco- 17 deficit. International reserves fell somewhat nomic activity in 2015 is likely to be lower than from $49.8 billion by end-August to $49.1 bil- previously expected. Although public infra- lion by end-September (some 19.8 weeks of structure investment is expected to ease sup- import coverage). ply constraints, especially as additional power comes onstream from the Medupi electricity Economic outlook: the growth forecast plant in 2015, constraints in power availabil- for South Africa has been revised down to ity are expected to persist as other parts of The forecast 1.4 percent for 2014 and 2.5 percent for 2015 the electricity grid undergo overdue mainte- anticipates modest The outlook for growth has continued to nance. We expect investment and household deteriorate since the previous South Afri- consumption, a major driver of the fragile growth in household can Economic Update in February 2014. economic recovery, to gradually regain pace Our forecast for real GDP growth has been as external demand strengthens and business spending and marked down to 1.4  percent for 2014 and and consumer sentiment improve. investment Figure Foreigners were net sellers of bonds but net purchasers of equity 1.13 Bonds Equities Bonds and equities Rand/$ exchange rate 150 12 100 10 $ millions (22-day moving average) 50 8 Exchange rate 0 6 –50 4 –100 2 –150 0 Jan. July Jan. July Jan. July 2012 2012 2013 2013 2014 2014 Source: Citigroup, Johannesburg Stock Exchange, South African Reserve Bank, and World Bank staff calculations. Table Economic outlook 1.4 Percent, unless otherwise noted Indicator 2012 2013 2014 2015 2016 Household consumption 3.5 2.6 2.2 2.3 2.7 Government consumption 4.0 2.4 1.8 1.6 1.7 Gross fixed capital formation 4.4 4.7 3.2 3.9 4.3 Exports 0.4 4.2 3.8 4.1 4.6 Imports 6.0 4.7 –0.8 3.5 4.2 GDP 2.5 1.9 1.4 2.5 2.8 Headline inflation 5.6 5.7 6.0 5.8 5.2 Current account deficit (percent of GDP) 5.2 5.8 5.7 5.6 5.5 Source: South Africa National Treasury, South African Reserve Bank, and World Bank staff calculations. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure The outlook for growth has been successively revised downward 1.14 4 3 Percent 2 18 1 0 2014 2015 2014 2015 2014 2015 2014 2015 World Bank, South Africa IMF, World National Treasury South African Economic Update Economic Outlook Reserve Bank Source: World Bank (South Africa Economic Update 4, 5, and 6), International Monetary Fund ( World Economic Outlook, October 2013, April 2014, and October 2014), National Treasury (2013 Budget, 2013 MTBPS, and 2014 Budget), and South African Reserve Bank (Quarterly Bulletin, January 2014, July 2014, and September 2014). Consumption spending is expected to Our forecasts assume that labor relations As a result, South Africa’s negative output recover only slowly, normalize, so that long-running labor dis- gap, estimated at about 1.2 percent of poten- putes are avoided. We expect South African tial GDP for 2014, is expected to narrow only constrained by high firms to continue to leverage the ongoing slowly over the forecasting horizon. levels of joblessness recovery in high-income countries as well as The moderation in domestic absorption robust economic growth in the rest of Sub-­ is expected to lead to a gradual adjustment and household Saharan Africa, now the most important in the current account balance. Due to the indebtedness destination for South Africa’s nonmineral slowdown in consumption and investment exports and a major investment destination demand and higher costs due to depreciation for many of South Africa’s companies and of the rand, imports are expected to remain banks. somewhat subdued. But external demand is Even so, South Africa is expected to set to strengthen over the forecast horizon, as underperform compared with many other economic activity picks up in both developed emerging market economies going forward and developing economies. Recoveries in key as domestic constraints hamper its ability export markets, including the United States, to take full advantage of the strengthening should help offset a further slowdown in global economic recovery. In the baseline China and should spur a recovery in exports forecast, consumption spending is expected that will help boost economic growth. Should to recover only slowly, constrained by high South Africa succeed in normalizing its levels of joblessness and declining purchas- labor relations, addressing its infrastructure ing power by households due to higher debt constraints, and containing wage increases servicing costs. In addition, moderation in in line with productivity gains, improving unsecured lending growth following the export performance should help gradually fallout from the unwinding of African Bank narrow the current account deficit. We nev- is likely to have an additional dampening ertheless expect the current account deficit impact on consumer spending. Private sec- to remain elevated, partly reflecting structur- tor investment is also expected to remain ally low savings of the South African econ- relatively subdued, driven by the anticipated omy (see South Africa Economic Update 1 in scaling back of investment in the gold and 2011) given the context of high rates of struc- platinum mining sectors as well as continu- tural unemployment. ing uncertainties about the labor environ- ment and reliability of the power supply that Risks to the outlook remain to the downside weigh on business sentiment. Meanwhile, the The outlook is subject to significant domestic fiscal consolidation needed to stabilize debt and external downside risks. South Africa’s burden and rebuild fiscal space will limit gov- recent weak economic performance can be ernment’s ability to further stimulate growth. attributed mainly to domestic developments, and domestic risks remain elevated due to boom, would adversely affect demand for increasing uncertainty over the direction of South Africa’s exports as well as its terms of policies, insufficient progress in addressing trade. Moderating growth in China is hurt- the security of electricity supply, and ongoing ing prices and volumes of key metal exports tensions in labor relations. South Africa, along from South Africa. Our estimates suggest with other developing countries and emerg- that in a scenario where prices of metals ing markets, also faces ongoing risks from the and agricultural commodities decline by inevitable normalization of monetary policy 15 percent from the baseline in 2014, South in the United States and a slowdown of growth Africa’s trade deficit could deteriorate by up in China on both commodity demand and to 1  percent of GDP. Compounding such a 19 prices. In addition, geopolitical tensions and shock could be the risk of a prolonged bout public health threats in Sub-­ Saharan Africa of weaker economic growth in the Euro could also weigh on investor sentiment. Area, one of South Africa’s major nonmin- South Africa is particularly vulnerable to eral goods export destinations. Should the potential bouts of financial market volatil- Ebola crisis in West Africa escalate, growth ity as global monetary conditions normal- in the Sub-­Saharan Africa, one of the recent ize, given its large current account deficit bright spots for South African exports, could South Africa faces and reliance on portfolio flows to fund it. slow. There is also a risk of indirect spillovers ongoing risks from Monetary policy in high-income countries is into the local tourism sector as international expected to diverge, with the Euro Area and and African leisure and business travelers the inevitable Japan expected to keep monetary conditions become more risk averse about travelling to loose amid risks of deflation and stagnation and within Africa. normalization of (figure 1.15). But the U.S. Federal Reserve is But domestic risks represent the most sig- monetary policy in projected to start raising policy rates soon. nificant downside to the economic outlook. The current emerging market context of Our baseline forecast relies on the assump- the United States still relatively buoyant financial markets and tion that labor relations return to normal. and a slowdown of exceptionally low yields carries the risk that Failure to stabilize domestic labor relations bouts of financial market volatility that could risks further undermining business confi- growth in China lead to a sharp reversal in capital flows, caus- dence and investment prospects, which would ing growth and investment to decline sharply make addressing South Africa’s fiscal and in South Africa. Deeper geopolitical tensions external deficits and high levels of unemploy- (for example, in Iraq or Russia) could also ment more difficult. Structural vulnerabilities trigger greater risk-based aversion to invest- and the more limited fiscal space to weather ment in emerging markets. shocks underscore the urgency to adopt bold A sharp slowdown in growth in China, or and far-reaching structural reforms to arrest the disorderly unwinding of its real estate what could risk becoming a secular decline Figure Policy rate expectations have been revised downward since the start of 2014 1.15 March 31, 2014 Current 4 United States 3 United Kingdom Percent 2 Euro Area 1 Japan 0 Mar. July Nov. Mar. July Nov. Mar. July Nov. Mar. July Nov. Mar. July Nov. 2014 2014 2014 2015 2015 2015 2016 2016 2016 2017 2017 2017 2018 2018 2018 Note: Policy rate expectations from overnight indexed swaps: Euro Overnight Index Average for Euro Area, Tokyo Overnight Index Average for Japan, Sterling Overnight Index Average for the United Kingdom, and federal fund rate for the United States. Source: World Bank and Bloomberg. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y in South African growth. South Africa needs number of jobs and the number of jobs urgently to accelerate the pace of economic that would preserve the absorption rate growth by addressing infrastructure con- at the same level as in 2008q4 (46.2 per- straints, policy uncertainty, and broaden- cent), taking into account the average ing structural reforms if it is to reduce the growth in the working-age population unacceptably high levels of joblessness and (1.8 percent). inequality prevailing in the economy. 4. Average growth is assumed to be the average y/y employment growth for Notes 2011q1–2014q2 (2.6  percent). Fast 20 1. South African Reserve Bank 2014c. growth is the maximum y/y employment 2. For mining, we use formal employment growth for 2011q1–2014q2 (4.5  percent reported in the Quarterly Employment in 2013q4). Statistics, given the fact that mining is a 5. National Treasury of the Republic of very clustered industry. South Africa 2014a. 3. More precisely, the jobs gap is calculated 6. Statement of the Monetary Policy Com- as the difference between the current mittee in July 2014 and September 2014. SECTION 2 Fiscal Policy and Redistribution in an Unequal Society: Two Questions Answered Since the end of apartheid, South Africa has has expanded social assistance programs made progress toward establishing a more in line with this mandate and spent siz- equitable society. In particular, advances in able resources, by the standards of middle- areas such as electrification and access to income countries, on health and education education and health services have increased services. By 2013/14, total government spend- equality of opportunities.1 There has also ing amounted to 33.2 percent of GDP, with been a sizable reduction in the levels of pov- more than half of it devoted to social spend- erty in recent years. Between 2006 and 2011, ing. Meanwhile the tax system generated the proportion of the population living in pov- considerable resources for redistribution, erty (using the national upper bound poverty with total general government revenue col- line) fell from 57.2 percent to 45.5 percent.2 lections amounting to 29.2  percent of GDP However, progress toward greater income in the same year. However, with the overall equality has proved elusive. Inequality of budget deficit now at about 4 percent of GDP household consumption, measured by the and debt burden close to 40 percent of GDP, Gini coefficient on disposable income, fiscal space to further expand social spend- increased from about 0.67 in 1993 to around ing has become more limited. In such an 0.69 in 2011,3 one of the highest levels in the environment, the question becomes whether world. The richest quintile of the popula- the government is making the best possible tion accounted for 61.3 percent of national use of fiscal policy to reduce poverty and consumption, while the poorest quintile inequality. accounted for 4.3 percent in 2011.4 In large The objective of this Update’s special part, this is an enduring legacy of the apart- focus section is to comprehensively assess the heid system. The National Development Plan distributional impact of government taxa- sets the ambitious goal of eliminating pov- tion and spending. It conducts a fiscal inci- erty and reducing inequality. It targets cut- dence analysis to assess how personal income ting the Gini coefficient to 0.60 by 2030 by and consumption taxes along with social raising employment and the share of income spending redistribute resources among the of the two poorest quintiles of the income different deciles of South Africa’s income distribution from 6 percent to 10 percent.5 distribution.6 The government has used the tax and The analysis seeks to provide answers to benefit system to alleviate inequality and pov- two main questions: erty in South Africa. The Constitution’s Bill 1. How do taxes and spending in South of Rights established citizens’ rights to health Africa redistribute income between the care, food, water, and social assistance. It rich and the poor? required the state to fulfill these rights 2. What is the impact of taxes and spend- progressively and to the best of its ability. ing on the rates of poverty and inequal- Since the end of apartheid, the government ity in South Africa? 21 S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y In providing answers to these questions efficiency of public services but also higher our analysis takes advantage of the most and more-inclusive economic growth to help recent Income and Expenditure Survey create jobs and lift incomes. (IES) from 2010/11.7 The survey contains Looking to the rest of this special focus data on household income, expenditures, section, we first provide an overview of the cash transfers, and utilization of educa- key fiscal tools used in South Africa to redis- tional services collected from some 25,328 tribute income between the rich and the households covering over 95,000 individu- poor, followed by an overview of the method- als. What makes this analysis unique relative ology and some caveats about its application. 22 to earlier studies 8 is that it uses the Com- We then proceed to examine the evidence to mitment to Equity (CEQ) methodology, 9 address the two key questions posed by this which allows comparison of the impact of special focus section. fiscal policy on inequality and poverty in South Africa to that in other middle-income The government’s fiscal tool kit countries. As described below, the CEQ uses to tackle poverty and inequality a consistent approach to assess how taxes Since the end of apartheid, the government South Africa is and spending work to benefit the poor and has progressively expanded its fiscal tool kit alleviate inequality in a set of comparable to help address poverty and inequality while currently grappling middle-income countries: Armenia, Brazil, maintaining sound fiscal policy. It broad- with slowing economic Bolivia, Costa Rica, El Salvador, Ethiopia, ened the tax base and built an efficient tax Guatemala, Indonesia, Mexico, Peru, and administration to generate the resources it growth, a high fiscal Uruguay. South Africa ranks as one of the needed to progressively expand the social deficit, and a rising most unequal countries of CEQ participant safety net for the poor. However, in recent countries, if not among all middle-income years fiscal space has become more limited. debt burden countries, given its Gini coefficient of 0.69. Reflecting the impact of the global finan- The proportion of the population living in cial crisis and the slowdown in economic poverty at 33.4 percent10­—measured by the growth in South Africa, the government pur- international benchmark of $2.50 a day sued a countercyclical policy that preserved (purchasing power parity, PPP, adjusted)— spending in the face of declining revenue is also higher than in many other middle- collections. As a result, the overall fiscal def- income countries with similar levels of GNI icit rose to a peak of 4.3 percent of GDP in per capita. For example, the poverty rate is 2012/13, up from a surplus of about 1.3 per- 11 percent in Brazil and 4 percent in Costa cent of GDP in 2008, before declining some- Rica.11 what to 4  percent of GDP in 2013/14. The Briefly, this Update has two main find- overall net debt burden rose from 22.9 per- ings. First, the burden of taxes falls on the cent of GDP in 2008/09 to 39.7  percent of richest in South Africa, and social spending GDP in 2013/14, and in an environment results in sizable increases in the incomes of slow economic growth it could rise even of the poor. In other words, the tax and higher. Given a more constrained fiscal envi- social spending system is overall progres- ronment going forward, the contribution of sive. Second, fiscal policy in South Africa fiscal policy to reducing market-determined achieves appreciable reductions in poverty levels of inequality and poverty has particu- and income inequality, and these reduc- lar relevance. tions are in fact the largest achieved in the On the revenue side, the tax system in emerging market countries that have so far South Africa generates, by middle-income been included in the CEQ. Yet despite fiscal country standards, considerable resources policy being both progressive and equaliz- for potential redistribution. Just over half of ing, the levels of poverty and inequality that South African’s tax collections of 27.1  per- remain are unacceptably high. South Africa cent of GDP in 2010/11 came from direct is currently grappling with slowing economic taxes: the personal income tax (PIT), cor- growth, a high fiscal deficit, and a rising debt porate income tax, and payroll taxes in the burden. In this context, addressing the twin form of unemployment insurance and the challenges of poverty and inequality will skill development levy (table 2.1). South require not only much-improved quality and Africa relies more on PIT and less on indirect Table General government revenue collections, 2010/11 2.1 Percent of GDP 2010/11 Incidence analysis Total general government revenue 30.9 17.5 Tax revenue 27.1 17.5 Direct taxes 14.3 8.5 Personal income tax 8.5 8.5 Corporate income tax 5.6 — Other direct taxes 0.1 — Indirect taxes 10.4 9.0 23 Value-added tax 6.9 6.9 General fuel levy 1.3 1.3 Specific excise duties 0.9 0.8 International trade taxes 1.0 — Other indirect taxes 0.3 — Other taxes 2.5 — Nontax revenue 3.8 — Government spending — is not included in the incidence analysis. Source: Statistics South Africa (2012b) for totals. Line items under direct and indirect taxes are from National Treasury (2013). in South Africa is also somewhat higher than the Figure Composition of taxes 2.1 average for middle- Direct taxesa Indirect taxes 30 15 income countries GNI per capita (2011 PPP $ thousands) at 32.2 percent of Share of GDP (%) 20 10 GDP in 2010/11 10 5 0 0 Ethiopia Bolivia Guatemala El Salvador Armenia Indonesia Peru South Africa Costa Rica Brazil Uruguay Mexico (2011) (2009) (2010) (2011) (2011) (2012) (2009) (2010) (2010) (2009) (2009) (2010) a. Direct taxes include corporate income tax collections in addition to the personal income tax. Source: Lustig forthcoming. or consumption taxes than other CEQ coun- GDP in 2010/11. This total compared with tries (figure 2.1). a middle-income country average of about Our analysis focuses on South Africa’s 27.6  percent of GDP for those two years. major tax items within the two groupings of South Africa’s social government spending direct and indirect taxes (see table 2.1): the (as a share of GDP) is among the highest PIT and payroll taxes within direct taxes, and in our CEQ sample (figure 2.2). Compared the value-added tax (VAT), specific excise with other big spenders, South Africa spends duties on alcohol and tobacco, and general somewhat more on education and less on fuel levy within indirect taxes. These items health and direct cash transfers than Bra- make up about two-thirds of South Africa’s zil, but more on direct cash transfers than total tax revenue.12 Bolivia. Total government spending in South Just more than half of South Africa’s total Africa is also somewhat higher than the expenditure was devoted to social spending average for middle-income countries. Total (table 2.2). Over the past decade, the num- government spending, excluding inter- ber of beneficiaries receiving social grants est payments, amounted to 32.2  percent of doubled from almost 8 million in 2003/04 to S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Social spending and subsidies 2.2 Direct transfers Education Health Subsidies 30 15 GNI per capita (2011 PPP $ thousands) Share of GDP (%) 20 10 24 10 5 0 0 Ethiopia Bolivia Guatemala El Salvador Armenia Indonesia Peru South Africa Costa Rica Brazil Uruguay Mexico (2011) (2009) (2010) (2011) (2011) (2012) (2009) (2010) (2010) (2009) (2009) (2010) Note: Excludes contributory old-age pensions. Subsidy outlays are not considered part of social spending. Source: Lustig forthcoming. Over the past decade, the number of Table General government expenditure, 2010/11 beneficiaries receiving 2.2 Percent of GDP social grants rose to 2010/11 Incidence analysis 15.8 million in 2013/14, Total general government expenditure 34.8 14.9 Primary government spending 32.2 14.9 mainly reflecting Social spending 17.6 14.9 Total cash transfers 3.8 3.8 the expansion of Old age pension (noncontributory) 1.3 1.3 Child support grant 1.1 1.1 direct cash transfers Disability grant 0.6 0.6 to children and Other grants 0.6 0.6 Foster care grant 0.2 0.2 the elderly Other transfers: free basic services 0.5 0.5 In-kind transfers 12.6 11.1 Education 7.0 7.0 Health 4.1 4.1 Housing and urban 1.5 — Other social spending 1.1 — Nonsocial spending (including public sector pensions) 14.6 — — is not included in the incidence analysis. Note: For free basic services, data represent the amount transferred under the equitable share formula for 2010/11 to municipalities to compensate them for providing basic services to poor households, and was provided by the Financial and Fiscal Commission of South Africa. Source: Statistics South Africa (2012b) for totals. Line items under direct and indirect taxes from National Treasury (2013). 15.8 million in 2013/14, mainly reflecting the Other items included in social spend- expansion of direct cash transfers to children ing include 0.5 percent of GDP dedicated to and the elderly. The child support grant, the provision of basic services such as power, introduced in 1998, was initially targeted at sanitation, water supply, and refuse removal children ages 0–7 years, with the age limit (free basic services) that are provided free to progressively raised to its current level of 18 low-income households who earned less than years. The age limit for the old-age grant for R 18,000 in 2010 ($2,466).14 A further 12.6 per- men was also lowered from 65 years to 60 cent of GDP was spent on in-kind transfers, years to equalize it with that of women. Total with 4.1 percent of GDP on health and 7 per- spending on all direct cash grants, at some cent of GDP on education. Finally, 1.5 percent 3.3 percent of GDP in 2010/11, is more than of GDP was devoted to housing and urban in- twice the median spending across developing kind transfers, including Reconstruction and countries.13 Development Programme housing. What is fiscal incidence analysis? contributions to the Unemployment Fiscal incidence analysis assesses how vari- Insurance Fund and Skills Development ous taxes and components of social spending Fund from the market income calcu- work to redistribute income among differ- lated in step 1. ent deciles of the population. It examines 3. Disposable income is constructed by add- the questions of who pays when the govern- ing direct cash transfers to net market ment collects taxes and who benefits when income from step 2. This measure is the government spends. The analysis consists closest to the household consumption of allocating taxes and social spending to on which the Gini coefficient in South households or individuals to compare their Africa is usually constructed. In South 25 incomes before and after taxes and transfers. Africa, direct cash transfers include, for The most common fiscal incidence analysis, example, the old age, child, disability, the accounting approach, which we use in and foster grants, and, in our study, free this special focus, examines what is paid and basic services. received without assessing the behavioral 4. Postfiscal income adds the impact of indi- responses that taxes and public spending may rect taxes and subsidies to the dispos- trigger among individuals or households. able income derived in step 3. In South Fiscal incidence We measure per capita income before and Africa, indirect taxes in this analysis analysis examines after each fiscal intervention using the steps include VAT, excises on alcohol and set out in the Commitment to Equity Hand- tobacco, and the fuel levy. who pays when the book15 as follows (see also figure 2.3): 5. Final income adds in-kind benefits (health 1. Market income comprises pretax wages, and education) to postfiscal income government collects salaries, and income such as rent, inter- from step 4. taxes and who est, and dividends. To assess the size of fiscal intervention at 2. Net market income subtracts direct each step, we use data on general govern- benefits when the taxes such as the PIT and employee ment revenues and spending from national government spends Definitions of income underpinning the Commitment to Equity fiscal incidence Figure analysis 2.3 BENEFITS TAXES Market income Wages and salaries; income from capital; private transfers; contributory pensions – Personal income and payroll taxes Net market income + Direct transfers Disposable income + Indirect subsidies – Indirect taxes Post scal income In-kind transfers + – Co-payments; (free government services in education and health) user fees Final income Source: Lustig and Higgins 2013. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y and fiscal accounts. These data are mapped beneficiaries in the survey aligns well with to the relevant measure of income taken administrative fiscal data.17 Since the IES from IES 2010/11. From the IES, we take has no information on health care use, the the reported net tax market income of each analysis imputed values of health spending household and divide it by the number of from the 2008 National Income Dynamics household members to arrive at per capita Study,18 a nationally representative house- income. The measure of income includes hold sample.19 From the fiscal administrative imputed rent on owner-occupied housing but data, we use the amount spent by the govern- excludes the value of own production since ment on a particular service divided by the 26 that is not separately identified in the sur- number of beneficiaries to approximate the vey. Using information about the tax system, level of benefit received. This nonbehavioral including statutory rates, thresholds, and approach amounts to asking by how much exemptions, we simulate the amount paid in would household income have to increase if taxes.16 the household had to pay for education or The IES also provides information on health services at the full cost incurred by the educational enrollment by level and type of government. institution (public or private), as well as on Box 2.1 discusses some limitations of the the receipt of cash transfers. The number of methodology and data used in the study. Box Caveats and data limitations 2.1 There are some important caveats about the scope of the fiscal incidence analysis conducted in this special focus section. First, by considering the poverty and redistributive effects of fiscal policy, we do not offer a full analysis of whether specific taxes or expenditures are desirable. When one tax or expenditure is found to be more redistributive to the poor than another group, the temptation is to conclude that the former is preferable. However, redistribution is only one of many criteria that matter when making public policy. Good tax policy will aim to be sufficient, efficient, and simple in addition to being equitable, and public spending will aim (among other goals) to provide the minimal functions of a state (such as security) and invest in public goods (such as infrastructure) that are necessary to ensure prosperity in addition to improving equity. By assessing the equity of taxes and spending, the results of this focus section are but one input to public policymaking, a factor that should be weighed with other evidence before deciding that a tax or expenditure is desirable. Second, the analysis does not take into account the quality of services delivered by the government. This limitation is particularly pertinent to the analysis of spending on health, education, and free basic services, as we discuss in more detail later. Second, the analysis excludes some important taxes and spending such as corporate income, international trade, and property taxes, and spending such as infrastructure investments due to the lack of an established methodology for assigning these outlays across households. Finally, it does not capture the growing debate on how asset accumulation and returns to capi- tal affect income inequality. Turing to the data used in the analysis, we also find limitations. The methodology used for data collection in the IES survey follows internationally accepted best practice, asking respondents to use a diary as well as recall methods to record their activi- ties over a two-week period. It is generally believed that the quality of the incomes and consumption data from the survey is good. However, there is some concern that the share of food consumption of the extreme poor in South Africa is much lower than one would expect, potentially pointing to some underreporting at the bottom of the income distribution. The IES does not separately identify own-produced goods, which could lead to some of this underreporting and lack of comparability with international findings. As in other countries, there are questions about the ability of a survey of this type to collect adequate information on households at the top of the distribution. Finally, the amounts reported in the IES as expenditure on alcohol and tobacco are 17 percent of that reported in South Africa’s National Accounts. This reflects a large number of households reporting zero consumption of these items. To try to control for these possible shortcomings and biases, we conduct various robustness tests. In addition to calculating the various measures of income in figure 2.3 from the reported level of net income in the IES, we also cross-check findings using the reported level of disposable income, which closely approximates consumption, to calculate the steps in figure 2.3. We also follow previous studies and use effective rather than statutory rates to correct for possible tax avoidance. In cases of gaps between levels of expenditure reported in the survey and other sources, we scale down the aggregate reported in the fiscal accounts to match that total reported in the IES.1 This effectively amounts to assuming that the survey provides the correct distribution of spending, for example, on alcohol and tobacco, but the wrong levels. Further details on the assumptions and vari- ous robustness tests are available in Inchauste and others (forthcoming). Note 1. For instance, following previous fiscal incidence analysis done for South Africa (Crosoer, Leibbrandt, and Woolard 2005), our analysis adjusts for the underreporting of alcohol and tobacco consumption in the IES relative to that reported in the national income accounts. Using the CEQ methodology, it is pos- share of the transfer received by the bottom sible to measure how the redistributive pro- X percent of the population is higher than its cess implemented through the fiscal system share in the population, a transfer is progres- impacts poverty and inequality. Broadly, the sive not just relative to the rich but also in impact on inequality of the fiscal system as absolute terms: that is, the per capita transfer a whole or any intervention in particular is higher for the poorest deciles and declines depends on two factors: the level of taxation as income rises.23 and spending and the progressivity of taxes Figure 2.4 presents an illustration of a and transfers.20 Lorenz curve where the population is ranked One common way to measure the pro- along the horizontal axis using market 27 gressivity of taxes is by comparing the share income, and the cumulative shares of taxes of a specific tax that is collected from each paid or transfers received is plotted along decile of the population relative to the share the vertical axis. The latter are concentration of total income each decile receives. The curves. population is ranked from the poorest to Going forward, we use the following richest decile using income per capita. The descriptions when referring to how a spend- shares are cumulated so that at the highest ing program or tax redistributes income: The progressivity of level of per capita income, the share of taxes • Progressive (regressive): a transfer (tax) taxes can be measured collected is equal to the total, or 100  per- whose concentration curve is above the cent. This is known as the tax redistribu- Lorenz curve for market income but by comparing the tion approach.21 A tax is progressive if the below the line of perfect equality (where cumulative share of a tax paid by the bottom each decile pays taxes or receives trans- share of a specific X percent of the population is lower than its fers of equal amounts). The transfer is tax collected from share in income. progressive only in relative terms. On the spending side, a transfer or spend- • Absolute progressive: When the concentra- each decile of the ing program is progressive if the cumulative tion curve for a spending program is population relative share of the total spending on the transfer above the line of perfect equality, the received by the bottom X percent of the transfer is also progressive in absolute to the share of population is higher than its share of mar- terms, in the sense that the monetary total income each ket income. When this happens, a transfer amount received falls as income rises. is equalizing in the sense that inequality • Neutral: A transfer (tax) whose concen- decile receives measured after receipt of the transfer will be tration curve coincides with the Lorenz lower than it was before receipt of the trans- curve of market income. fer.22 In the case of spending, it is also useful • Regressive (progressive): A transfer (tax) to compare the share of spending received whose concentration curve lies below the by decile of the total population. When the Lorenz curve of market income. Figure The progressivity of taxes and transfers 2.4 Transfer: progressive in absolute terms Line of perfect equality; transfer: neutral in absolute terms; bene ts, taxes, or income Cumulative proportion of tax: upper bound of regressive Transfer: progressive in relative terms; tax: regressive Transfer: regressive; tax: progressive Lorenz market income curve; transfer or tax: neutral in relative terms if on this line Cumulative proportion of population (by market income) Source: Lustig and Higgins 2013. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Fiscal policy poverty and inequality: two questions answered Question 1: How do taxes and spending in South Africa redistribute income between the rich and the poor? The tax and spending system is progressive: the burden of taxes falls on the richest and social spending results in sizable increases in the income of the poor. 28 Are taxes progressive? and the concentration shares of direct taxes What is the incidence of taxation in South and their two components in subsequent Africa? We assess the incidence of each tax columns. separately before presenting a summary • The PIT is quite progressive (at each assessment of how progressive PIT, pay- decile). The burden of the PIT is borne roll taxes, and consumption taxes are com- overwhelmingly by the richer deciles. Direct taxes are bined. Thus, the analysis evaluates the tax The two richest deciles (the richest system along only one dimension, its impact decile) of individuals generated over progressive and work on equity. It does not assess other impor- 97  percent (87  percent) of total PIT to reduce inequality tant features of a tax system, such as its collections while their share in mar- efficiency—which measures the amount col- ket income was equal to 81.4  percent in South Africa lected given the rate—simplicity, and ease of (63.7 percent). administering. • Payroll taxes in the form of the Skills Development Levy and contributions to Direct taxes: personal income the Unemployment Insurance Fund are and payroll taxes progressive up to the eighth decile. How- Direct taxes in our incidence analysis for ever, table 2.3 shows that these payroll South Africa are PIT and payroll taxes, which taxes are locally regressive for the 9th and comprise contributions to the Skills Develop- 10th deciles. The 10th decile pays a lower ment Levy and the Unemployment Insurance share of total contributions (58.4  per- Fund. We assess what share of market income cent) than its share in market income is paid in these taxes by each decile. (63.7 percent), which reflects the effects Direct taxes are (at each decile) progres- of the income cap on contributions to the sive and work to reduce inequality in South Unemployment Insurance Fund.24 Africa. Table 2.3 shows the cumulative distri- Relative to other countries in the CEQ bution of market income in the first column sample, figure 2.5 shows that the richest Table Progressivity of direct taxes 2.3 Cumulative distribution and cumulative concentration shares (%) Contributions to the Unemployment Decile Market income Direct taxes Personal income taxes Insurance Fund and Skills Levy Poorest 0.1 0.0 0.0 0.0 2 0.3 0.0 0.0 0.0 3 0.7 0.0 0.0 0.1 4 1.6 0.0 0.0 0.4 5 3.1 0.1 0.0 1.2 6 5.8 0.4 0.1 3.3 7 10.3 1.2 0.5 8.1 8 18.6 4.0 2.5 18.3 9 36.3 15.7 13.1 41.6 Richest 100 100 100 100 Note: These are the cumulative distributions of market income for the population ordered by market income—in other words, the Lorenz curves for market income by decile. Source: Inchauste and others (forthcoming) based on IES 2010/11. Figure Personal income taxes are progressive 2.5 A. Concentration shares of personal income taxes, South Africa 100 Share of taxes by income decile (%) 75 50 29 25 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Incidence of personal income taxes and contributions relative to revenue collections Poorest decile Richest decile Share of GDP Total The Kakwani index 20 20 of progressivity for Share of taxes by income decile (%) 15 15 South Africa is 0.13, Share of GDP (%) compared with 0.27 10 10 for Brazil and 0.30 for Mexico, showing that 5 5 South Africa’s direct 0 0 tax system is less Armenia Brazil Mexico Peru South Africa Uruguay (2011) (2009) (2010) (2009) (2010) (2009) progressive than those Source: Armenia (Younger and Khachatryan forthcoming), Brazil (Higgins and Pereira 2014), Mexico (Scott 2014), Peru (Jaramillo 2014), Uruguay (Bucheli and others 2014), and South Africa (Inchauste and others forthcoming) based on IES 2010/11. in other countries decile in South Africa pays about 18.5 per- Kakwani index—the tax concentration coef- cent of its market income in PIT.25 Those in ficient, as described above, minus the Gini the bottom half of the income distribution coefficient on income. If the Kakwani index do not pay PIT because their market income is greater than zero, the tax is progressive (in is below the PIT threshold. 26 By contrast, the tax redistribution sense). If it is equal to Brazil collects almost a similar amount to zero, the tax is neutral; and if it is less than South Africa in direct taxes as a share of zero, the tax is regressive and poorer deciles GDP, 27 and households in its richest decile pay proportionally more of their income in pay about 11  percent of market income in taxes than their share in income relative to direct taxes (or about 5  percent of market richer deciles. As a practical rule and in line income in PIT), while those in its poorest with international practice, we have defined decile pay about 1  percent. The difference as neutral those taxes for which the Kakwani in effective rates of market income collected index lies between –0.1 and 0.1. in direct taxes between the two countries The Kakwani index confirms that direct reflects both the steeper PIT tax rate struc- taxes in South Africa (which combine PIT ture in South Africa (which peaks at a top and payroll taxes) are progressive but less rate of 40 percent, compared with a top stat- so than in other countries (figure 2.6). The utory rate of 27.5 percent in Brazil) and its Kakwani index of progressivity for South exemption threshold, which helps exclude Africa is 0.13, compared with 0.27 for Bra- poorer households. zil and 0.30 for Mexico. This result, showing A conventional summary measure of pro- that South Africa’s direct tax system is less gressivity also used in the tax literature is the progressive than those in other countries, S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Progressivity of South Africa’s direct tax system: the Kakwani index 2.6 0.5 0.4 Kakwani index 0.3 0.2 30 0.1 0.0 South Africa Armenia Uruguay Brazil Ethiopia Mexico Peru (2010) (2011) (2009) (2009) (2011) (2010) (2009) Source: Armenia (Younger and Khachatryan forthcoming), Brazil (Higgins and Pereira 2014), Ethiopia (Hill, Tsehaye, and Woldehanna forthcoming), Mexico (Scott 2014), Peru (Jaramillo 2014), Uruguay (Bucheli and others 2014), and South Africa (Inchauste and others forthcoming) based on IES 2010/11. Value-added and fuel may seem surprising at first given South households make their consumption deci- taxes are progressive, Africa’s more progressive statutory PIT rate sions taking into account government cash but excise taxes structure, its higher effective tax burden at transfers as part of their income. As a result, the upper end of the income distribution, they consume much more than their labor, are regressive and its high share of direct taxes as a share or market, income would allow them to con- of GDP. 28 However, the underlying distri- sume. In the absence of such transfers, they bution of pretax market income in South would thus have paid much less in indirect Africa is much more unequal than in other taxes than they actually did. Box 2.2 com- countries: the Gini coefficient of market pares the consumption basket of the richest income of 0.771 in South Africa compared and poorest deciles in the income distribu- with 0.579 in Brazil and 0.511 in Mexico. tion in South Africa. Since the Kakwani index subtracts the Gini In terms of the progressivity of indirect coefficient of income from the tax concen- taxes, figure 2.7 and table 2.4 show that indi- tration coefficient, it is lower in South Africa rect taxes are only slightly regressive. 30 Up than in other countries. Although direct to the seventh decile, the share paid of total taxes in South Africa are working to redis- indirect taxes exceeds their cumulative share tribute, they therefore face against strong of disposable income by only a small margin. headwinds from the underlying inequality VAT and the fuel levy are progressive, with in earnings. all deciles paying a lower share in such taxes than their share of disposable income. By Indirect taxes contrast, excise taxes are outright regressive: We undertake an incidence analysis of VAT, the poorest deciles pay a substantially higher excises on alcohol and tobacco, and the fuel share of the total than their share of dispos- levy. These three groups of taxes comprise able income. This is the result of the fact that 9 percent of GDP, or about a third of South the poor consume proportionately more of Africa’s tax base. VAT accounts for roughly the “sin goods.” a quarter of tax revenue in any given year, 29 The slightly regressive nature of indi- excise duties contribute about 3.5  percent rect taxes is most clearly seen in the fact of tax revenue, and the general fuel levy that the four poorest deciles accounted for contributes about 5.2  percent of total tax 4.78 percent of total income distribution but revenue. paid 4.95  percent of total indirect tax col- We assess the incidence of indirect taxes lections, while the two richest deciles (the with respect to disposable income (which richest decile) of the income distribution is defined as market income minus direct paid 75.0 percent (56.9 percent) of the total taxes plus direct transfers, and is roughly revenue indirect tax collections when their equivalent to consumption) rather than with share in total disposable income was about respect to market income. We do this because 74.5 percent (56.7 percent).31 Box Structure of consumption of the poorest and richest deciles 2.2 As in most countries, the consumption baskets of the poorest and the richest deciles in South Africa are quite different. In 2010/11, food made up 36 percent of total consumption in South Africa but only 7 percent of total consumption for the richest decile (box figure 1). Although the poorest decile is more likely to consume goods that are zero-rated, such as basic foods, the poor consume other goods that are subject to indirect taxes, including clothing, household maintenance, and personal care items. Box figure 1. Structure of consumption A. Poorest decile 31 Miscellaneous Alcohol and tobacco 1% 19% Communication 2% Food Recreation 2% 36% Household maintenance 3% Insurance 5% The burden of indirect Rent, energy, Transport 8% and water taxes in South Africa is 15% quite even across the Clothing and textiles 9% income distribution B. Richest decile compared with Food other middle‑income 7% Rent, energy, countries and water 13% Miscellaneous Clothing and textiles 4% 39% Transport 8% Insurance 15% Alcohol and tobacco 1% Communication 3% Recreation 4% Household maintenance 6% Source: Inchauste and others (forthcoming) based on IES 2010/11. Figure 2.8A shows that the share of dispos- decile, 3.4  percent, compared with 2.4  per- able income paid in VAT increases from just cent for the poorest decile.32 When these two under 9.5 percent of the disposable income factors are combined, the burden of indirect of the poorest decile to almost 12 percent of taxes in percentage of disposable income is the disposable income of the richest decile. quite even across the income distribution However, excise taxes on alcohol and tobacco compared with other middle-income coun- tend to make up a higher share of the dis- tries (figure 2.8A). For instance, in Brazil and posable income of the poorer deciles. The Mexico the overall burden of indirect taxa- poorest decile pays about 4.3 percent of dis- tion rises more progressively with income posable income in such excises, compared than it does in South Africa (figure 2.8B). with 0.6  percent for those in the richest The Kakwani index shows that indirect or decile. However, this is partly offset by the consumption-based taxes are in aggregate fuel levy, which makes up a slightly higher broadly neutral. Table 2.5 shows the over- share of the disposable income of the richest all index is –0.003. VAT and fuel levy are S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Concentration curves of indirect taxes 2.7 Cumulative share of disposable income (%) 100 75 Line of perfect equality 50 Excise tax 32 Lorenz curve for 25 disposable income Value-added tax Fuel levy 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. The progressivity of direct taxes works Table Progressivity of indirect taxes 2.4 Cumulative distribution and cumulative concentration shares (%) to outweigh the Decile Disposable income Value-added tax Excise tax Fuel levy Indirect taxes slight regressivity Poorest 0.5 0.5 3.4 0.4 0.7 of indirect taxes 2 1.5 1.4 7.0 1.1 1.7 3 2.9 2.6 10.7 2.2 3.1 4 4.8 4.2 15.4 3.6 5.0 5 7.3 6.6 21.2 5.5 7.4 6 10.9 9.8 29.0 8.5 11.0 7 16.4 14.8 39.2 13.6 16.4 8 25.5 23.2 52.5 22.8 25.4 9 43.2 40.4 70.8 42.0 43.1 Richest 100 100 100 100 100 Note: This is the cumulative distribution of disposable income and tax concentration shares ordered by disposable income decile. Source: Inchauste and others (forthcoming) based on IES 2010/11. both slightly progressive, each with a Kak- taxes (driven by the regressivity of excise wani index of about 0.02. In contrast, excise taxes). In other words, the progressivity of taxes are regressive, with a Kakwani index of direct taxes works to outweigh the slight –0.302. regressivity of indirect taxes, resulting in a tax system that is progressive globally. Overall impact of taxes in South Africa How progressive in the aggregate are direct Is social spending progressive? and consumption taxes in South Africa? How does social spending work to redis- To assess the progressivity of direct and tribute tax resources to benefit the poor in indirect taxes, we add direct and indirect South Africa? As we saw earlier, South Africa taxes and measure their incidence relative to has higher social spending than other mid- market income. The Kakwani index for both dle-income countries. But spending more taxes combined is equal to 0.028, reflect- does not mean that the poor always ben- ing that the taxes covered in this Update efit from such programs. Poorly targeted or are globally progressive. However, in figure designed social programs often result in the 2.9 we can see that the concentration curve benefits leaking to higher income groups. As and the Lorenz curve cross, indicating that with the tax side of the government’s budget, the system is not progressive everywhere. we therefore assess the question of who ben- The slight regressivity at the lower end of efits from social spending in South Africa by the income distribution largely reflects the examining each social program individually impact of the slight regressivity of indirect before combining them to assess the overall Figure Incidence of indirect taxes 2.8 A. South Africa Value-added tax Excise tax Fuel levy 20 Share of disposable income (%) 15 10 33 5 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Middle-income countries Share of disposable income by income decile (%) Poorest decile Richest decile Share of GDP 20 20 15 15 Share of GDP (%) 10 10 5 5 0 0 Bolivia Brazil Indonesia Mexico South Africa Note: Deciles are ranked by market income. Source: Bolivia (Paz Arauco and others 2014), Brazil (Higgins and Pereira 2014), Indonesia (Jellema, Wai-Poi, and Afkar forthcoming), Mexico (Scott 2014), and South Africa (Inchauste and others forthcoming) based on IES 2010/11. Table Measuring the progressivity of indirect taxes 2.5 Value-added tax Fuel levy Excise tax Indirect tax Concentration curves Progressive everywhere Progressive everywhere Regressive everywhere Regressive except at the 9th and 10th deciles Kakwani index 0.020 0.025 –0.302 –0.003 Note: If the Kakwani index is greater than zero, the tax is progressive. If it is between –0.1 and 0.1 the tax is neutral, and if it is below –0.1, the tax is regressive. Source: Inchauste and others (forthcoming) based on IES 2010/11. benefits from social spending. We assess the our analysis also includes health and educa- incidence of direct cash transfers—the old- tion spending. Together, these items account age noncontributory pension, the child for 43 percent of total spending and 85 per- support grant, the disability grant, the foster- cent of social spending. care grant, and other grants such as the care dependency grant. We also examine free Direct cash transfers basic services (water, electricity, and sani- Direct cash transfers as a whole are progres- tation) provided by the government to the sive in absolute terms. The cash amount poor, under the assumption that they are a received declines as market income rises, as form of a direct transfer to the poor. Finally, shown by the red line in figure 2.10. Which of S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Concentration curves of all taxes, 2010 2.9 100 Cumulative share of tax paid (%) 75 Line of perfect equality 50 Lorenz curve for market income 34 25 All taxes 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. Among large cash Progressivity of direct cash transfers by category: concentration curves for transfer programs, the Figure transfers and Lorenz curve for market income child support grant 2.10 A. Large cash transfer programs 100 stands out as being Child support grant Cumulative share of transfers (%) the most progressive Direct transfers 75 Disability grant Old-age pension Line of perfect equality 50 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Small cash transfer programs 100 Child foster care Cumulative share of transfers (%) Direct transfers 75 Grant-in-aid 50 Other grants Line of perfect equality 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. the various cash transfers programs in South The cash transfer program that stands out Africa are the most progressive? We show as being the most progressive, more so than the concentration curves of the largest cash the average of all direct cash transfers com- transfer programs in figure 2.10A and those bined, is the child support grant. Among the for the smaller programs in figure 2.10B. larger programs, the next most progressive program is the old-age pension, which is also pensions and disability and child support progressive in absolute terms. The disability grants in boosting the incomes of the poor. grant is as about as progressive as the average The impact of these transfers in raising the cash grant for the poorest deciles. Among income of the poor in South Africa is far the smaller programs, the most progressive larger than in other middle-income coun- one is child foster care. tries in our CEQ sample, including Brazil. In It appears that cash transfers in South Brazil, outlays on direct transfers—at 4.2 per- Africa do reach the poor. The bulk of cash cent of GDP—are larger than in South Africa transfers go to the bottom of the income dis- and include expenditure on the well-known tribution: 69 percent of all cash transfers go Bolsa Família conditional cash transfer pro- 35 the four poorest deciles. This partly reflects gram. Yet these transfers raise the market the fact that the share of households with incomes of the poorest decile by only a factor school-age children and the elderly is higher of 2 (figure 2.11B). at the bottom of the distribution than at Even so, table 2.6 also shows that there are the top (box 2.3). The IES shows that about some cash transfers directed to individuals in 66 percent of the poorest decile households higher income groups, with nearly 18 percent have children under 18 years of age, com- of old-age pensions and 11 percent of disabil- Cash transfers in pared with 37  percent in the richest decile. ity grants going to households with incomes South Africa reach the Some 28 percent of households in the poor- above $10 a day (PPP). est decile have a pension-age adult in it, com- poor: 69 percent of all pared with 22  percent in the richest decile. Free basic services Moreover, by directing transfers to families Do the free basic services of water, electricity, cash transfers go the with children and the elderly, the transfer sanitation, and refuse removal provided by the four poorest deciles system is very effective at targeting the poor government benefit South Africa’s poorest? because some 40  percent of those who are Because municipalities do not report the exact classified as living below the national lower value of such services that they provide free to bound poverty line of R 433 ($59.31) a month households, we cannot directly identify their in local 2010/11 prices were under the age of exact rand value. Therefore, we examine two 15, while 23 percent were over the age of 60.33 scenarios that represent the extremes of what In monetary terms, direct cash transfers we understand is the general practice adopted received from the government boost the mar- by municipalities in providing such services: ket incomes of those in the poorest decile • First, we assume that the amount of more than 10-fold. As shown in figure 2.11A, money allocated by the central govern- this largely reflects the effect of the old-age ment for free basic services (0.5 percent Box Comparing the bottom and top of South Africa’s income distribution 2.3 In 2010/11, there were 1.1 million households, or about 5 million people, in the poorest decile. The average per capita income was R 200 a person that year, with many in this decile “de facto” reporting zero, or near-zero, market income before receiving government transfers (box table 1). In contrast, the income of those in the richest decile was more than 1,000 times larger at R 204,639 ($28,443) a person. Persons in households in the poorest decile of South Africa’s income distribution were more likely to live in rural areas, have larger households, and were on average younger with more dependents than those at the top of the income distribution. A higher share of households in the poorest decile had a pension-age adult. Box table 1. Features of households at bottom and top of the income distribution Feature of household Poorest decile Richest decile Percentage of households in urban areas 47 92 Average household size (people) 4.5 2.7 Average age (years) 24 35 Years of schooling 6 13 Percentage of households with children younger than 18 years 66 37 Percentage of households with pension-age adult 28 22 Per capita annual income before taxes and cash transfers (rand) 200 207,369 Per capita annual income after taxes and cash transfers, free basic services (rand) 2,131 141,075 Source: Inchauste and others (forthcoming) based on IES 2010/11. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y South Africa’s direct cash transfer programs in a national and international Figure perspective 2.11 A. Incidence of direct cash transfers, South Africa Old-age pension Child support grant Disability grant Child care dependency grant Child foster care Grant-in-aid Other grants 1,250 Share of market income (%) 1,000 750 36 500 250 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Incidence of direct cash transfers, middle-income countries Poorest decile Richest decile Share of GDP Share of market income by income decile (%) 1,500 5 1,200 4 Share of GDP (%) 900 3 600 2 300 1 0 0 Argentina Armenia Brazil Mexico South Africa Uruguay (2009) (2011) (2009) (2010) (2010) (2009) Source: Argentina (Lustig and Pessino 2014), Armenia (Younger and Khachatryan forthcoming), Brazil (Higgins and Pereira 2014), Mexico (Scott 2014), Uruguay (Bucheli and others 2014), and South Africa (Inchauste and others, forthcoming, based on IES 2010/11. Table Share of benefits going to each income group 2.6 Percent of total, by level of market income in purchasing power parity dollars a day More than Less than $1.25 $1.25–$2.50 $2.50–$4.00 $4.00–$10.00 $10.00–$50.00 $50.00 Total Old-age pension 46.7 13.4 8.0 14.3 14.7 2.9 100 Disability grant 50.4 14.6 8.0 16.0 9.9 1.1 100 Child support grant 55.1 16.3 9.4 14.3 4.7 0.2 100 Care dependency grant 64.5 13.3 5.3 14.2 2.2 0.6 100 Foster care grant 53.3 17.1 10.7 11.7 6.5 0.6 100 Grant-in-aid 41.8 13.9 8.4 21.0 12.7 2.2 100 Other grants 32.5 7.5 10.2 20.0 17.9 11.9 100 Free basic services 40.9 12.7 7.7 17.5 16.2 5.0 100 Population shares 40.3 16.2 9.0 16.2 14.4 4.0 100 Source: Inchauste and others (forthcoming) based on IES 2010/11. of GDP) to municipalities is distributed two old-age pensions, or about R 24,000 equally among households that are con- a year). This effectively assumes that free nected to the electricity grid and are basic services are closer in nature to a indigent (those with incomes of less than targeted cash transfer program, because there are municipalities that choose to households.34 In monetary terms, combining deliver these services as cash rebates tar- direct cash transfers with free basic services geted to the poor. helps boost the market income of the poorest • Second, we assume that all households decile 11-fold. connected to the national electricity grid equally benefit from inverted block tariffs In-kind transfers: education and health and receive an equal share of the central In assessing how education and health spend- government allocations for free basic ser- ing benefit the poor we have to caution that vices. This would make free basic services our analysis does not address the quality of closer to an untargeted indirect subsidy. such spending.35 We use government expen- 37 The results of our analysis show that if free diture data on the various forms of education basic services were targeted nationwide along and health services to estimate unit costs of the lines of our first scenario, there would these programs. The analysis thus assumes be clear advantages for the poor, since free that the actual benefit received by individu- basic services would be progressive in abso- als is equal to the amount spent per capita. lute terms (figure 2.12A). In contrast, if free This assumption reflects a clear limitation basic services were delivered as indirect sub- of the analysis because the quality of school Combining direct cash sidies, they would be progressive only in rela- infrastructure, teachers, and health clinics transfers with free tive terms, since a larger share of the benefit and hospitals varies across the country. would go to the richest deciles (figure 2.12B). A few words of caution are thus warranted basic services helps In sum, targeting in all municipalities would to explain how our findings on targeting may improve effectiveness in reaching the poorest not translate into a commensurate actual boost the market income of the poorest Figure Incidence and concentration curves for free basic services decile 11-fold 2.12 A. Concentration curves for free basic services Cumulative share of basic services received (%) 100 Free basic services as a transfer 75 Line of perfect equality 50 Free basic services as a subsidy 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Incidence of free basic services As a transfer As a subsidy 250 Share of market income (%) 200 150 100 50 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y impact on the poor. Despite good policy and on a student’s race has now been eliminated: relatively high spending levels in relation to while in the early 1990s the average white GDP for education and health, actual perfor- child received a spending subsidy for educa- mance and outcomes in these sectors have tion that was 4.5 times as much as that of a been disappointing.36 For example, in educa- black child, the disparity was eliminated by tion, South Africa achieves test scores in read- 2006.41 Any remaining gap in spending per ing and math at grade 6 that are below the pupil is caused by the fact that more highly Southern and Eastern Africa region average qualified teachers tend to be concentrated test scores, even though many of these com- in richer schools, implying a slight bias in 38 parator countries spend the same or less per salary expenditure per student to these capita on education. 37 The 2011 Trends in schools, but this is virtually balanced by the International Math and Science Study showed higher allocations of spending on norms and large improvements in scores for grade 9 standards in poorer schools. Although it is learners relative to 2002, but South African true that schools in more affluent neighbor- students were still ranked in the bottom five hoods are able to supplement state resources of 42 countries. Moreover, the Trends in with privately funded school fees, the public Despite good policy International Math and Science Study showed financing of schools is more or less equal. As that the average scores on math and science a result, public spending per student aver- and relatively high for the South Africa’s best-performing stu- aged R 11,000 in 2011, and about 78 percent spending levels in dents (those in the 95th percentile) were of learners (more than 8 million students) in below the average scores achieved by students 80 percent of public schools (close to 20,000 relation to GDP for in Singapore, Chinese Taipei, the Republic of schools) benefited from no-fee schools.42 education and health, Korea, Japan, Finland, Slovenia, and the Rus- With these important considerations in sian Federation.38 In health, despite steady mind, the results for education show that actual performance improvements, South Africa still has compar- when we “monetize” the value of education and outcomes in atively high levels of maternal and infant mor- spending, it disproportionally benefits those tality by middle-income country standards, at the bottom of the distribution (figure these sectors have while its level of health spending (public and 2.13). Public spending on education is pro- been disappointing private) of just more than 8 percent of GDP is gressive in absolute terms. Reflecting rela- comparatively high.39 tively high levels of spending as well as the Another important consideration is how nationally high rate of enrollment in the spending per student in education varies by education system (more than 97 percent par- race. One of the major features of apartheid ticipation for 7–15-year-olds and 83 percent social spending was the large gap in per capita for 16–18-year-olds),43 the poor benefit from spending per schoolchild: per capita funding primary and secondary education spending for white students was 10 times that of African to a greater extent than the rich. However, learners.40 The gap in public financing based they derive less benefit relative to higher Progressivity of education spending by category: concentration curves for Figure transfers and Lorenz curve for market income 2.13 Cumulative share of education spending (%) 100 Primary school Secondary school 75 Adult education In-kind education In-kind transfers, total 50 Line of perfect College University equality education education 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. income groups from college and university at the bottom of the distribution. In figure education spending because of lower rates of 2.15, we plotted the concentration curves for attendance by the poor at these institutions. direct and in-kind transfers as a whole, and While spending on postsecondary and uni- for education spending and health spending. versity education is still equalizing, these cat- Health spending is progressive in absolute egories are progressive only in relative terms, terms to a (roughly) similar extent as educa- with college education being more progres- tion spending. sive than spending on university education.44 The monetized value of health spend- While the share of benefits in postsecondary ing makes up a larger share of the market and university education is relatively small incomes of those at the bottom of the income 39 for the poor, about half of spending on adult distribution (figure 2.16A): health spending training centers goes to households with is nine times the size of the market incomes incomes of less than $4 a day (figure 2.14). of the poorest decile. Health care use in Our findings for South Africa are not unique, South Africa is, however, more evenly dis- since much of tertiary education spending in tributed across socioeconomic groups than Armenia, Bolivia, and Brazil benefits higher in other middle-income countries (figure income groups in as well. 2.16B). Public spending on health is relatively The poor benefit from We find that health spending makes up well targeted not because poorer people use primary and secondary a larger share of market incomes of those health facilities more, but rather because education spending Figure Share of education benefits by income group to a greater extent 2.14 Income group: Less than $1.25 $1.25–$2.50 $2.50–$4.00 $4.00–$10.00 $10.00–$50.00 More than $50.00 than the rich, but they 100 derive less benefit Share of education bene ts (%) 75 relative to higher income groups from 50 college and university education spending 25 0 Preschool Primary Secondary Postsecondary Adult training University training centers Source: Inchauste and others (forthcoming) based on IES 2010/2011. Progressivity of health spending: concentration curves and Lorenz curve for Figure market income 2.15 100 In-kind health Cumulative share of spending (%) In-kind transfers, total 75 In-kind Direct transfers education 50 Line of perfect equality 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Health spending: incidence and beneficiary households by income group 2.16 A. Incidence of health spending, South Africa 1,000 Share of market income (%) 750 500 40 250 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. Health spending share of individuals in beneficiary households by income group Health spending makes Income group: Less than $1.25 $1.25–$2.50 $2.50–$4.00 $4.00–$10.00 $10.00–$50.00 More than $50.00 up a larger share of 100 market incomes of Share of target population (%) 75 those at the bottom of the distribution 50 25 0 Armenia Bolivia Brazil Ethiopia Indonesia South Africa Source: Armenia (Younger and Khachatryan forthcoming), Bolivia (Paz Arauco and others 2014), Brazil (Higgins and Pereira 2014), Ethiopia (Hill, Tsehaye, and Woldehanna forthcoming), and Indonesia (Jellema, Wai-Poi, and Afkar forthcoming). For South Africa, Inchauste and others (forthcoming) based on IES 2010/11. they use public health services more than analysis received by each decile. Remem- richer people compared with those in other ber that spending concentration curves for countries, such as Bolivia and Brazil, where a transfer targeted to the poor are above most beneficiary households are nonpoor. the line of perfect equality if they are pro- The share of households that reported using gressive in absolute terms (when the share public clinics—the main pillar of the pub- of benefits going to the poorest deciles is lic health system in South Africa—has risen higher than the share of market income of steadily from 44.5 percent in 2004 to 59.6 per- those deciles). In figure 2.17A, the bottom cent in 2012.45 The public sector, which spent half of the income distribution receives just more than 4 percent of GDP on health in about 70 percent of spending on direct trans- 2010/11, serves roughly 83 percent (41.3 mil- fers and 54 percent of spending on in-kind lion) of the South African population.46 transfers but accounted for about 3 percent Some 17 percent (8.3 million) of people have of total market income across deciles. The medical aid and mostly use private facilities, results thus confirm that when combined, with total private sector health-related spend- direct transfers (cash transfers and free basic ing of about 4.3 percent of GDP.47 services) are progressive in absolute terms. In-kind transfers in the form of education Overall impact of social spending and health services are also progressive in How progressive is South Africa’s social absolute terms but not to the extent of direct spending? Again, we use concentration transfers (figure 2.17B). curves whose vertical axis measures the In addition to the concentration curves, proportion of the spending program under we calculate concentration coefficients for Figure Concentration curves of benefits, 2010 2.17 A. Direct transfers 100 Cumulative share of bene ts (%) 75 Direct transfers In-kind transfers 50 Line of perfect equality 41 25 Lorenz curve for market income 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile B. In-kind transfers 100 In-kind transfers in the form of education Cumulative share of bene ts (%) 75 In-kind and health services Direct transfers health In-kind education are progressive in 50 Line of perfect equality absolute terms but 25 not to the extent Lorenz curve for market income of direct transfers 0 Poorest 2 3 4 5 6 7 8 9 Richest decile decile Source: Inchauste and others (forthcoming) based on IES 2010/11. each type of spending and compare them in absolute terms. Spending is absolutely pro- to the market income Gini coefficient. Con- gressive if it is less than –0.1. As shown in fig- centration coefficients are calculated in the ure 2.18, this is true for preschool education, same manner as the Gini coefficient. If the grant-in-aid, old-age pensions, primary edu- concentration curve is above the diagonal, cation, the disability, foster care, and child the concentration coefficient is negative and support grants, and most notably for the care we can conclude that spending is progressive dependency grant. What is also very clear is Figure Concentration coefficients for spending 2.18 Free basic services (as transfers) Care dependency Child support grant Foster care grant Disability grant Primary education Preprimary and primary education Old-age pension Grant-in-aid Secondary school education Preschool education Health spending Other grants Free basic services (as indirect subsidies) Tertiary education spending Market income Gini coef cient –0.4 –0.2 0.0 0.2 0.4 0.6 0.8 Concentration coef cient Source: Inchauste and others (forthcoming) based on IES 2010/11. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y that when free basic services are targeted untargeted indirect subsidy, they are progres- as cash transfers, they are absolutely pro- sive only in relative terms when compared gressive. When they take the form of an with the Gini of market income. Question 2: What is the impact of taxes and spending on the rates of poverty and inequality in South Africa? Taxes and social spending achieve the largest reductions in income inequality and pov- 42 erty of the 12 CEQ countries. Thanks to progressive taxes and spending, of those in the fourth decile of the income fiscal policy makes a substantial contribution distribution (table 2.8), a more than 10-fold to reducing market income inequality and increase. Even if indirect taxes are taken into the level of poverty in South Africa (table account, the increase is still about 10-fold. 2.7). When the impact of the taxes included South Africa ranks at the top in terms of Direct taxes and in this study (PIT, VAT, excise taxes, and the scale of redistribution of the CEQ com- the fuel levy), cash transfers, free basic ser- parator countries (figure 2.19 and table cash transfers reduce vices, and spending on education and health 2.9).48 Fiscal policy is equalizing in the sense the level of extreme are taken into account, fiscal policy reduces that it works to reduce the gap in incomes inequality, measured by the Gini coefficient, across the distribution. The reduction in income poverty by from 0.771 to 0.596—a decline of 0.175 Gini the Gini coefficient of 0.175 Gini point (or more than two-thirds, point, or 22 percent. almost a quarter) is twice as large as achieved Reflecting the impact of taxes, cash trans- in Brazil, the next-best performer (see table from 34.4 percent fers, and free basic services, the proportion 2.9). Even so, the Gini for final income in to 11.7 percent of the population living in extreme poverty South Africa, which reflects the full impact of ($1.25 a day) falls from 34.4 percent in terms redistribution through fiscal policy, remains of market income to 16.5 percent. The share higher than Brazil, the second most unequal living on $2.50 a day falls from 46.2 percent country in our sample, starts with before it to 39.0 percent. But it is also useful to look begins to redistribute via its fiscal system. The only at transfers (thus excluding the impact Gini coefficient for market income in Brazil is of indirect taxes) by examining the rate of 0.579 and falls to 0.439 following the impact poverty with respect to disposable income, of fiscal policy. In South Africa, the Gini coef- which captures the amount of resources an ficient remains higher, at 0.596, after taking individual has available to spend or consume into account redistribution via fiscal policy. after receiving cash transfers from the gov- South Africa also leads the comparator ernment. Using this benchmark of dispos- countries in terms of the amount of poverty able income, the reduction in poverty is more reduction achieved through fiscal policy. impressive. Direct taxes and cash transfers Table 2.10 shows that after the impact of reduce the level of extreme income pov- taxes, cash transfers, and free basic services, erty, usually measured as an income of less South Africa generates the largest reduc- than $1.25 a day (PPP), by more than two- tion in poverty of the countries in the CEQ thirds, from 34.4 percent to 11.7 percent (see sample using the international benchmark table 2.7). of share of the population living on less than In monetary terms, the redistribution $2.50 a day (PPP). Most notably, consump- implemented through South Africa’s fis- tion taxes such as VAT and excise taxes do cal system through taxes and cash transfers not reverse the reduction in poverty associ- is sufficient to bring those with a per capita ated with direct transfers, so that postfiscal market income of a mere R 200 a year ($19) poverty (column 4) is still lower than net in the poorest decile to a per capita income market income (column 2). This contrasts of disposable income of R  2,363 ($223). In with the results in several other countries, other words, fiscal policy works to lift those in including Brazil (see table 2.10).49 the poorest decile of the income distribution The reduction in poverty ref lects the to almost reach the average market income impact of social spending, which works to Table Poverty and inequality indicators at each income concept 2.7 Market Net market Disposable Postfiscal Final income income income income income (1) (2) (3) (4) (5) Column 1 Column 2 Column 3 Column 4 – direct taxes + cash transfers – indirect taxes + in-kind Indicator transfers Inequality indicators Gini coefficient 0.771 0.750 0.694 0.695 0.596 Theil index 1.222 1.119 0.973 0.971 0.724 90/10 ratio 198.9 173.3 32.7 33.2 12.5 43 Headcount poverty indicators (%) National food poverty linea 40.8 41.0 23.4 29.0 — Official consumption based (food poverty line) — — 20.2 — — National lower bound poverty lineb 46.5 46.7 34.2 39.6 — Official consumption based (lower bound) — — 32.2 — — National upper bound poverty linec 52.3 52.5 45.1 50.1 — $1.25 a day (PPP) 34.4 34.4 11.7 16.5 — $2.50 a day (PPP) 46.2 46.4 33.4 39.0 — Fiscal policy works $4.00 a day (PPP) 54.3 54.6 48.5 53.1 — to lift those in the a. The food poverty line was set at R 210 a month in 2005/06 using March 2006 prices. Adjusted for inflation it was R 321 a month in 2010/11. b. The lower bound poverty line was set at R 300 a month in 2005/06 using March 2006 prices. Adjusted for inflation it was R 443 a month in 2010/11. poorest decile of the c. The upper bound poverty line was set at R 431 a month in 2005/06 using March 2006 prices. Adjusted for inflation it was R 620 a month in 2010/11. — is not available. Source: Inchauste and others (forthcoming) based on IES 2010/11. income distribution to almost reach Table Average per capita income in each market income decile the average market 2.8 Rand income of those in the Market income Net market income Disposable income Postfiscal income (1) (2) (3) (4) fourth decile of the Column 1 Column 2 Column 3 Market income decile – direct taxes + cash transfers – indirect taxes income distribution Poorest decile 200 200 2,363 2,131 2 736 735 2,997 2,669 3 1,497 1,493 3,691 3,264 4 2,761 2,748 4,679 4,106 5 4,925 4,887 6,609 5,755 6 8,653 8,535 9,970 8,627 7 14,793 14,397 15,662 13,481 8 27,119 25,762 26,658 22,828 9 57,711 51,994 52,661 44,822 Richest decile 207,639 166,52 166,803 141,075 Note: Column 3 excludes free basic services. Source: Inchauste and others (forthcoming) based on IES 2010/11. substantially boost the incomes of the poor basic services. The net cash position of the in South Africa. We find that households in household after taxes and transfers is posi- the poorest decile receive cash transfers and tive for the six poorest deciles. Once the free basic services that are worth 11 times monetized value of education and health ser- their market income (or 32 times their mar- vices is included, the poorest decile receives ket income if the monetized value of educa- transfers and services worth some R 6,900 (or tion and health services is also included). $945 in 2010/11) a year per capita from the This compares with the burden of PIT and government, compared with the R 724 ($99) consumption taxes, which amounted to two that they pay in taxes. Only the three richest times market income. Households in the deciles of the market income distribution pay bottom half of the income distribution thus more in taxes than they receive in all forms receive far more in direct transfers and free of cash and in-kind benefits. S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Change in Gini coefficient: disposable versus market income 2.19 0.00 Change in Gini coef cient –0.05 –0.10 44 –0.15 –0.20 Ethiopia Indonesia Guatemala El Salvador Peru Armenia Bolivia Mexico Uruguay Costa Rica Brazil South Africa (2011) (2012) (2010) (2011) (2009) (2011) (2009) (2010) (2009) (2010) (2009) (2010) Source: Armenia (Younger and Khachatryan forthcoming), Bolivia (Paz Arauco and others 2014), Brazil (Higgins and Pereira 2014), Costa Rica (Lustig forthcoming, based on Sauma and Trejos 2014), El Salvador (Beneke de Sanfeliu, Lustig, and Oliva 2014), Ethiopia (Hill, Tsehaye, and Woldehanna forthcoming), Guatemala (Cabrera, Lustig, and Morán forthcoming), Indonesia (Jellema, Wai-Poi, and Afkar 2014), Mexico (Scott 2014), Peru (Jaramillo 2014), and Uruguay (Bucheli and others 2014); Inchauste and others (forthcoming) for South Africa based on IES 2010/11. The reduction in poverty for the Table How the Gini coefficient for each income concept compares across CEQ countries amount spent 2.9 Market income Net market income Disposable income Postfiscal income Final income (effectiveness) is also (1) (2) (3) (4) (5) Column 1 Column 2 Column 3 Column 4 the highest among – direct taxes + cash transfers – indirect taxes + in-kind transfers Armenia (2011) 0.403 0.393 0.373 0.374 0.357 Commitment to Bolivia (2009) 0.503 0.503 0.493 0.503 0.446 Equity countries Brazil (2009) 0.579 0.565 0.544 0.546 0.439 Costa Rica (2010) 0.508 0.500 0.489 0.486 0.393 El Salvador (2011) 0.440 0.436 0.430 0.429 0.404 Ethiopia (2011) 0.322 0.315 0.305 0.302 0.299 Guatemala ( 2010) 0.551 0.550 0.546 0.551 0.523 Indonesia (2012) 0.394 0.394 0.390 0.391 0.369 Mexico (2010) 0.511 0.497 0.488 0.481 0.429 Peru (2009) 0.504 0.498 0.494 0.492 0.466 South Africa (2010) 0.771 0.750 0.694 0.695 0.596 Uruguay (2009) 0.492 0.478 0.457 0.459 0.393 Note: Figures for Bolivia and Indonesia include indirect taxes only. Source: Armenia (Younger and Khachatryan forthcoming), Bolivia (Paz Arauco and others 2014), Brazil (Higgins and Pereira 2014), Costa Rica (Lustig forthcoming, based on Sauma and Trejos 2014), El Salvador (Beneke de Sanfeliu, Lustig, and Oliva 2014), Ethiopia (Hill, Tsehaye, and Woldehanna forthcoming), Guatemala (Cabrera, Lustig, and Morán forthcoming), Indonesia (Jellema, Wai-Poi, and Afkar forthcoming), Mexico (Scott 2014), Peru (Jaramillo 2014), and Uruguay (Bucheli and others 2014); and Inchauste and others (forthcoming) for South Africa based on IES 2010/11. These findings could simply reflect the Africa by 13  percentage points in 2010/11 fact that South Africa spends more than before the impact of indirect taxes. 50 The other countries on its social programs. A reduction in poverty for the amount spent natural follow-up question is how effective (effectiveness) is also the highest in our CEQ are direct transfers in reducing poverty and sample of middle-income countries and inequality given the amount spent in terms reflects the combination of effective target- of GDP? This formulation allows us to com- ing and the relatively large amount spent pare across different programs within South on these absolutely progressive programs. Africa as well as to compare South African Similarly, direct transfers reduce the inequal- programs with similar programs in other ity coefficient by 0.055 of a Gini coefficient developing countries. Figure 2.20A shows (figure 2.20B). The change in inequality due that the 3.8 percent of GDP spent on direct to in-kind transfers, such health care and transfer programs (cash transfers and free education, works to amplify the already high basic services) reduced the poverty rate for benefits from social transfers. South Africa’s those living on less than $2.50 day in South social spending of 14.9  percent of GDP on How the poverty headcount rate at $2.50 PPP a day for each income concept Table compares across CEQ countries 2.10 Market income Net market income Disposable income Postfiscal income (1) (2) (3) (4) Column 1 Column 2 Column 3 – direct taxes + cash transfers – indirect taxes Armenia (2011) 31.3 32.0 28.9 34.9 Bolivia (2009) 19.6 19.6 17.6 20.2 Brazil (2009) 15.1 15.7 11.2 16.3 Costa Rica (2010) 5.4 5.7 3.9 4.2 El Salvador (2011) 14.7 15.1 12.9 14.4 45 Guatemala ( 2010) 35.9 36.2 34.6 36.5 Indonesia (2012) 56.4 56.4 55.9 54.9 Mexico (2010) 12.6 12.6 10.7 10.7 Peru (2009) 15.2 15.2 14.0 14.5 South Africa (2010) 46.2 46.4 33.4 39.0 Uruguay (2009) 5.1 5.1 1.5 2.3 Note: Figures for Bolivia and Indonesia include indirect taxes only. South Africa uses its Source: Armenia (Younger and Khachatryan forthcoming), Bolivia (Paz Arauco and others 2014), Brazil (Higgins and Pereira 2014), Costa Rica (Lustig forthcoming, based on Sauma and Trejos 2014), El Salvador (Beneke de Sanfeliu, Lustig, and Oliva 2014), Ethiopia (Hill, Tsehaye, and Woldehanna forthcoming), Guatemala (Cabrera, Lustig, and Morán forthcoming), Indonesia (Jellema, Wai-Poi, and Afkar forthcoming), Mexico (Scott 2014), Peru (Jaramillo 2014), and Uruguay (Bucheli fiscal instruments very and others 2014); and Inchauste and others (forthcoming) for South Africa based on IES 2010/11. effectively to reduce cash transfers, free basic services, and edu- and health spending, which together account poverty and inequality cation and health services together reduces for just over 40 percent of total spending. inequality by 0.14 of a Gini coefficient. The results show that South Africa uses its In sum, South Africa’s fiscal system lifted fiscal instruments very effectively to reduce some 3.6  million individuals out of poverty poverty and inequality. The tax system is when measured as those living on less than slightly progressive, and spending is highly $2.50 a day (PPP). When taxes and all social progressive. In other words, the rich in spending are combined, the gap in incomes South Africa bear the brunt of taxes, and the between the rich and poor goes from a situa- government effectively redirects these tax tion where the incomes of the richest decile resources to the poorest in society to raise are more than 1,000 times higher than the their incomes. As a result of the fiscal system, poorest to one where they are about 66 times some 3.6 million individuals are lifted out of higher. poverty when measured as those living on less than $2.50 a day (PPP). Inequality goes from Conclusion: addressing the twin a situation where the incomes of the richest challenges of poverty and inequality decile are more than 1,000 times higher than in South Africa going forward the poorest to one where they are about 66 This Update’s special focus section sought to times higher. The Gini coefficient falls from shed light on the important question of how 0.77 before taxes and spending to 0.59 after South Africa’s government is using fiscal pol- fiscal policy is applied. icy to address the twin challenges of poverty On the tax side, fiscal policy relies on a and inequality. It sought answers to two main mix of progressive direct taxes, such as the questions: How do taxes and spending work PIT and slightly regressive indirect/consump- to redistribute income? And what reductions tion taxes, that when combined generate a in poverty and inequality rate have been slightly progressive tax system. Direct taxes achieved by fiscal policy? To address these (PIT and payroll taxes) were progressive, questions we analyzed the incidence of PIT since the richer deciles pay a proportionally and payroll contributions, VAT, excise taxes higher share of total direct tax collections on alcohol and tobacco, and the general fuel than their share of market income. More- levy, which comprise just over two-thirds of over because they make up a relatively high total tax collections. On the expenditure share of GDP, they help narrow the gap in side, we analyzed the incidence of direct cash incomes between the rich and poor. Indirect transfers, free basic services, and education taxes are slightly regressive; the four poorest S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y Figure Poverty and inequality reducing effectiveness of direct transfers 2.20 A. Poverty-reduction effectiveness Change in poverty rate Effectiveness indicator –20 4 Percentage point change of those living on less than $2.50 a day –15 3 Effectiveness indicator –10 2 46 –5 1 0 0 Argentina Brazil Indonesia Mexico Peru South Africa Uruguay (2009) (2009) (2012) (2010) (2009) (2010) (2009) B. Inequality-reduction effectiveness Fiscal policy Change in Gini coef cient from direct transfers Effectiveness indicator goes a long way –0.06 1.8 toward achieving –0.05 1.5 Change in Gini coef cient Effectiveness indicator redistribution, but –0.04 1.2 higher growth and –0.03 0.9 job creation will be –0.02 0.6 required to reduce –0.01 0.3 poverty and inequality 0.00 0.0 Argentina Brazil Indonesia Mexico Peru South Africa Uruguay going forward (2009) (2009) (2012) (2010) (2009) (2010) (2009) Note: Changes in poverty and inequality reflect changes from net market to disposable income in all countries. Source: Argentina (Lustig and Pessino 2014), Brazil (Higgins and Pereira 2014), Indonesia (Jellema, Wai-Poi, and Afkar forthcoming), Mexico (Scott 2014), Peru (Jaramillo 2014), and Uruguay (Bucheli and others 2014); Inchauste and others (forthcoming) for South Africa based on IES 2010/11. deciles contributed about 5.0 percent of total that while there is some scope to improve indirect tax collections, compared with the targeting of certain social programs like their share of 4.8 percent in total disposable free basic services, cash transfer programs income. This regressivity at the lower end of are already well targeted and quite sizable the income distribution largely reflects the in terms of outlays. Education and health impact of excises because VAT and fuel tax spending also benefit the poorer parts of the are progressive. income distribution relatively more than the On the spending side of fiscal policy, social rich. However, there are concerns about the spending is not only progressive but also quality of such spending, which suggests that contributes to large reductions in poverty more could be done to improve the quality and inequality. Direct transfers are progres- of such services to ensure that education and sive in absolute terms, since they effectively health spending maximize their potential in target the poor (who are largely children reducing poverty and inequality. and old-age pensioners), and are sizable in In sum, fiscal policy already goes a long terms of GDP, all of which leads to important way toward achieving redistribution. Even so, reductions in poverty and inequality. In fact, the level of poverty and inequality in South South Africa performs very well when com- Africa after taxes and spending remains unac- pared with other middle-income countries: ceptably high. More can and needs to be it achieves the most redistribution compared done to improve the quality of service deliv- with the other middle-income countries ery. But South Africa’s fiscal deficit and debt in the CEQ analysis. Our analysis suggests indicators show that the fiscal space to spend more to achieve even greater redistribution is income, which equates roughly with extremely limited. Addressing the twin chal- consumption. lenges of poverty and inequality going forward 12. The largest omitted item is corporate in a way that is consistent with fiscal sustain- income tax, which accounts for about ability will require higher and more inclusive 21 percent of tax revenue. economic growth. This would be particularly 13. World Bank 2009. important in addressing the need for jobs and 14. All rand values reported for 2010/11 in higher incomes at the lower end of the income this section are calculated at the average distribution, to narrow the gap in incomes exchange rate in 2010/11 of R 7.3 per $. between the rich and the poor and reinforce 15. Lustig and Higgins 2013. 47 the effectiveness of fiscal policy. 16. For PIT, we assume the burden of the tax is borne by the recipient of the income. Notes For consumption taxes, like VAT, excise 1. World Bank 2012. taxes, and the fuel taxes, we assume 2. Statistics South Africa (2014a), tab. 3, that the tax is borne by the consumer. using the national upper bound poverty Detailed consumption data in the IES line of R 620 a month. allows us to estimate the burden on 3. Statistics South Africa 2014a, tab. 5. income of the VAT, the fuel levy, and 4. Statistics South Africa 2014a, tab. 5. specific excise duties on alcohol and 5. National Planning Commission 2012. tobacco. 6. Inchauste and others (forthcoming) 17. National Treasury 2014b. present in more detail the methodology, 18. Southern Africa Labour and Develop- assumptions, and analysis used in this ment Research Unit 2009. special focus section. 19. See Inchauste and others (forthcoming) 7. Statistics South Africa 2012a. app., for full details on the methodology 8. See, for example, van der Berg (2009), adopted. van der Berg and Moses (2012), and 20. For more on the definitions of progres- Leibbrandt and others (2010). Bosch sivity, see Lustig and Higgins (2013). and others (2010) calculate the impact of 21. See, for example, Duclos and Araar social grants, free basic services, housing (2006), p. 136. subsidies, and direct taxes on the Gini 22. See, for example, Lambert (2002). coefficient. Bhorat and Van Der West- 23. To establish progressivity as defined huizen (2012) examine how social grants in the literature, it is not necessary for impacted the growth in expenditure lev- transfers (taxes) to be progressive (either els by the bottom deciles of the income relatively or absolutely in the case of distribution after 1994. transfers) at every point (that is, for every 9. See Lustig and Higgins (2013). Led by individual) in the distribution. Transfers Nora Lustig since 2008, the CEQ is a (taxes) can be globally progressive even project of the Center for Inter-American if they are not everywhere progressive. Policy and Research and the Depart- See, for example, Lambert (2002) and ment of Economics at Tulane University, Duclos and Araar (2006). the Center for Global Development, and 24. Under the Unemployment Insurance the Inter-American Dialogue (www.­ Fund, employers and employees each commitmentoequity.org). The World contribute 1  percent of earnings up Bank has partnered with Tulane Uni- on earnings up to R  14,872 a month versity to apply the CEQ framework ($1,487). to Armenia, Ethiopia, Jordan, South 25. When using consumption as the welfare Africa, and Sri Lanka. measure instead of market income, the 10. Based on disposable income; see table results are nearly identical. The richest 2.7. The poverty line is measured in pur- decile pays 18  percent relative to their chasing power parity terms. total consumption. 11. The poverty line is measured in purchas- 26. In 2010/11, individuals with an income ing power parity terms. The headcount of less than R  120,000 a year ($16,438) ratios are measured using disposable (who made up more than half of all S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y taxpayers) were not required to file and Development Programme (RDP) tax returns. In 2010/11, the tax thresh- housing in our analysis. However, the old (that is, the taxable income below IES does have information on whether which no PIT was payable) was R 54,200 any member of the household received ($7,424) for individuals below age 65 an RDP house from the government. and R  84,200 ($11,534) for individuals Based on this we find that the distribu- over the age of 65. The top marginal tion of RDP beneficiaries and house- tax rate was 40 percent and kicked in at holds ranked by market income deciles is R 525,000 a year ($71,917). relatively flat, with relatively small shares 48 27. In Brazil, direct taxes include PIT, pay- in the poorest and richest deciles. See roll taxes, and property taxes. Inchauste and others (forthcoming) for 28. The top marginal PIT rate in Brazil further details. (27.5 percent) is lower than the top rate 36. National Planning Commission 2012. in South Africa (40 percent). In Mexico, 37. Presidency of the Republic of South the four poorest deciles are exempt from Africa 2014; Southern and Eastern paying PIT and benefit from an employ- Africa Consortium for Monitoring Edu- ment subsidy (or negative income tax). cational Quality 2011. 29. For 2010/11, zero-rated goods reduced 38. HSRC 2012. the VAT revenue intake by R 34 ­ billion, 39. Presidency of the Republic of South 1.2  percent of fiscal year GDP, and Africa 2013. exempt goods and services reduced rev- 40. Presidency of the Republic of South enue by another R 1 billion, or 0.03 per- Africa 2014. cent of fiscal year GDP (National 41. Van der Berg 2006, 2009. Treasury 2014). 42. Presidency of the South Africa 2014. 30. The results are nearly identical if we use 43. National Planning Commission 2012. market income or consumption as the 44. Note that students are captured in sur- welfare measure. veys at the places they find themselves 31. Disposable income includes free basic when studying, which in some cases may services as a form of transfer. When free not be the same as their households of basic services are treated as an indirect origin. As a result, it may appear that subsidy and therefore excluded from some students from very poor house- disposable income, the share of the two holds are not actually appearing in the richest deciles in disposable income is survey as poor. 75.1 percent. 45. Presidency of the Republic of South 32. Our estimates of the fuel levy include not Africa 2014. only the direct amount of tax paid, but 46. National Planning Commission 2012. also “second round effects” that include 47. National Planning Commission 2012. the impact of higher fuel prices on trans- 48. See Lustig (forthcoming) for full details. port, which in turn affects all other com- 49. Lustig forthcoming. modities that use transport as an input. 50. This corresponds to the difference in For more details on how this was done, the poverty rates for net market and dis- see Inchauste and others (forthcoming). posable income in table 2.10. In line with 33. Statistics South Africa 2014a. standard practice in incidence studies, 34. We obtain qualitatively the same we do not include the impact of educa- results when measuring with respect to tion and health spending on poverty consumption. because recipients may not be willing to 35. Due to lack of detailed administrative pay in cash terms an amount equivalent data on housing values, we have not to the outlays on these services in the included spending on Reconstruction budget. References Beneke de Sanfeliu, Margarita, Nora Lustig, Estimation with DAD. New York: Springer and José Andrés Oliva. 2014. “La inciden- and International Development Research cia de los impuestos y el gasto social sobre Centre. la pobreza y la desigualdad en El Salvador.” Higgins, Sean, and Claudiney Pereira. 2014. Bhorat, Harron, and Carlene Van Der Wes- “The Effects of Brazil’s Taxation and Social thuizen. 2012. “Poverty, Inequality and Spending on the Distribution of House- the Nature of Economic Growth in South hold Income.” In The Redistributive Impact Africa.” Development Policy Research Unit of Taxes and Social Spending in Latin America, Working Paper 12/151, University of Cape ed. Nora Lustig, Carola Pessino, and John Town. Scott, special issue, Public Finance Review 42 Bosch, Adel, Jannie Rossouw, Tian Claassens, (3). and Bertie du Plessis. 2010. “A Second Hill, Ruth, Eyasu Tsehaye, and Tassew Wolde- Look at Measuring Inequality in South hanna. Forthcoming. “The Distributional Africa: A Modified Gini Coefficient.” Impact of Fiscal Policy in Ethiopia.” Back- School of Development Studies Working ground paper for World Bank Poverty Paper No. 58, September 2010. Assessment. Bucheli, Marisa, Nora Lustig, Máximo Rossi, HSRC (Human Sciences and Research Coun- and Florencia Amábile. 2014. “Social cil of South Africa). 2012. Towards Equity Spending, Taxes and Income Redistri- and Excellence, Highlights from TIMSS 2011: bution in Uruguay.” In The Redistributive The South African Perspective. Pretoria. Impact of Taxes and Social Spending in Latin IMF (International Monetary Fund). 2013. America, ed. Nora Lustig, Carola Pessino, World Economic Outlook: Transitions and Ten- and John Scott, special issue, Public Finance sions, October 2013. Washington, DC. Review 42 (3). ———. 2014a. World Economic Outlook: Recov- Cabrera, Maynor, Nora Lustig, and Hilicías ery Strengthens, Remains Uneven, April 2014. E. Morán. Forthcoming. “Fiscal Policy and Washington, DC. the Ethnic Divide in Guatemala.” ———. 2014b. World Economic Outlook: Lega- Crosoer, David, Murray Leibbrandt, and cies, Clouds, Uncertainties, October 2014. Ingrid Woolard. 2005. “Asset-Based versus Washington, DC. Money Metric Poverty Indices in South Inchauste, Gabriela, Nora Lustig, Maskekwa Africa: An Assessment Using the Chronic Maboshe, Catriona Purfield, and Ingrid Poverty Research Centre RSA 2002 Sur- Woolard. Forthcoming. “The Distribu- vey.” SALDRU/CSSR Working Papers 109, tional Impact of Fiscal Policy in South Southern Africa Labour and Development Africa.” World Bank Policy Research Work- Research Unit, University of Cape Town. ing Paper, Washington, DC. Duclos, Jean-Yves, and Abdelkrim Araar. 2006. Jaramillo, Miguel. 2014. “The Incidence of Poverty and Equity: Measurement, Policy, and Social Spending and Taxes in Peru.” In 49 S OUT H A FRICA ECONOM IC U P D AT E — F IS C A L P O LI C Y A N D R ED I S T R I B UT I O N I N A N UN EQ UA L S O C I ET Y The Redistributive Impact of Taxes and Social America, ed. Nora Lustig, Carola Pessino, Spending in Latin America, ed. Nora Lustig, and John Scott, special issue, Public Finance Carola Pessino, and John Scott, special Review 42 (3). issue, Public Finance Review 42 (3). Presidency of the Republic of South Africa. Jellema, Jon, Matthew Wai-Poi, and Rythia 2013. Development Indicators 2012. Pretoria. Afkar. Forthcoming. “The Distributional ———. 2014. Twenty-Year Review. Pretoria. Impact of Fiscal Policy in Indonesia.” Sauma, Pablo, and Juan Diego Trejos. 2014. World Bank, Washington, DC. Gasto Público Social, Impuestos, Redistribu- Kakwani, Nanak C. 1977. “Measurement of ción del Ingreso y Pobreza en Costa Rica. CEQ 50 Tax Progressivity: An International Com- Working Paper No. 18. Center for Inter- parison.” Economic Journal 87 (345): 71–80. American Policy and Research, Depart- Lambert, Peter. 2002. The Distribution and ment of Economics at Tulane University, Redistribution of Income, 3rd ed. Manchester, and Inter-American Dialogue. UK: Manchester University Press. Scott, John. 2014. “Redistributive Impact and Leibbrandt, M., I. Woolard, A. Finn, and J. Efficiency of Mexico’s Fiscal System.” In Argent. 2010. “Trends in South African The Redistributive Impact of Taxes and Social Income Distribution and Poverty since the Spending in Latin America, ed. Nora Lustig, Fall of Apartheid.” OECD Social, Employ- Carola Pessino, and John Scott, special ment and Migration Working Papers, No. 101. issue, Public Finance Review 42 (3). Paris: OECD Publishing. South African Reserve Bank. 2014a. Quarterly Lustig, Nora. Forthcoming. “Taxes, Transfers, Bulletin, January 2014. Pretoria. Inequality and the Poor in the Develop- ———. 2014b. Quarterly Bulletin, July 2014. ing World. Round 1.” CEQ Working Paper Pretoria. 23, Center for Inter-American Policy and ———. 2014c. Quarterly Bulletin, September Research and Department of Economics 2014. Pretoria. at Tulane University and Inter-American Southern Africa Labour and Development Dialogue. Research Unit. 2009. National Income Lustig, Nora, and Sean Higgins. 2013. Com- Dynamics Study 2008, Wave 1 [dataset]. Ver- mitment to Equity Assessment (CEQ): Estimat- sion 5.2. Cape Town. ing the Incidence of Social Spending, Subsidies Southern and Eastern Africa Consortium for and Taxes. Handbook. CEQ Working Paper Monitoring Educational Quality. 2011. No. 1, Center for Inter-American Policy “Trends in Achievement Levels of Grade and Research, Department of Economics 6 Pupils in South Africa.” SACMEQ Policy at Tulane University, and Inter-American Brief No. 1. Paris. Dialogue. Statistics South Africa. 2012a. “Income and Lustig, Nora, and Carola Pessino. 2014. Expenditure Survey.” Pretoria. “Social Spending and Income Redistribu- ———. 2012b. “Financial Statistics of Con- tion in Argentina in the 2000s: The Ris- solidated General Government.” Pretoria. ing Role of Noncontributory Pensions.” ———. 2014a. “Poverty Trends in South Pubic Finance Review. Published online 20 Africa: An Examination of Absolute Pov- November 2013. erty Between 2006 and 2011.” Pretoria. National Planning Commission. 2012. Our ————. 2014b. “The South African MPI: Future—Make it Work: National Development Creating a Multidimensional Poverty Plan 2030. Pretoria. Index Using Census Data.” Pretoria. National Treasury. 2013. Budget 2013 . van der Berg, Servaas. 2006. “The Targeting Pretoria. of Public Spending on School Education, ———. 2014a. Medium Term Budget Policy State- 1995 and 2000.” Perspectives in Education 24 ment 2014. Pretoria (2): 49–64. ———. 2014b. Budget 2014. Pretoria. ———. 2009. “Fiscal Incidence of Social Paz Arauco, Verónica, George Gray Molina, Spending in South Africa, 2006”. A Report Wilson Jiménez Pozo, and Ernesto Yáñez to National Treasury. Aguilar. 2014. “Explaining Low Redistribu- van der Berg, Servass, and Eldrige Moses. tive Impact in Bolivia.” In The Redistributive 2012. “How Better Targeting of Social Impact of Taxes and Social Spending in Latin Spending Affects Social Delivery in South Africa, 2012.” Development South Africa 29 ———. 2014a. South Africa Economic Update 5: (1). Focus on Export Competitiveness. Washington, World Bank. 2009. “Levels and Patterns of DC. Safety Net Spending in Developing and ———. 2014b. South Africa Economic Update 6: Transition Countries.” Washington, DC. Fiscal Policy and Redistribution in an Unequal ———. 2012. “South Africa Economic Society: Two Questions Answered. Washing- Update: Focus on Inequality of Opportu- ton, DC. nity.” Working Paper No. 71553. Washing- Younger, Stephen D., and Artsvi Khachatryan. ton, DC. Forthcoming. “Fiscal Incidence in Arme- ———. 2013. South Africa Economic Update 4: nia.” Background Paper for World Bank 51 Focus on Financial Inclusion. Washington, Armenia Public Expenditure Review. Washing- DC. ton, DC: World Bank The World Bank 442 Rodericks Road, Lynnwood, Pretoria 0081, Republic of South Africa Tel: +27 (0) 12 742 3100 Fax: +27 (0) 12 742 3135 Twitter: #SouthAfricaEU www.worldbank.org/za www.facebook.com/WorldBankSouthAfrica