DG DEVCO Staff Seminar on Social Protection - from strategies to concrete approaches -

26 - 30 September 2016, Brussels

SESSION 8 Fiscal Incidence in South Africa

Jon JELLEMA

Associate Director for Africa, Asia & Europe, CEQ Institute [email protected] www.commitmentoequity.org

Two Main Questions

• How do and spending in South Africa redistribute income between the rich and poor?

• What is the impact of taxes and spending on poverty and inequality in South Africa?

Based on: Inchauste et al., (2015) “The Distribuonal Impact of in South Africa.” CEQ Working Paper No. 29. Tulane University and the Center for Inter-American Policy and Research. Taxes What taxes were assessed?

Incidence 2010/11 analysis (% of GDP)

Total General Government Revenue 30.9 17.5 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 VAT 6.9 6.9 General fuel levy 1.3 1.3 Specific excise duties 0.9 0.8 International trade taxes 1 … Other indirect taxes 0.3 … Other taxes 2.5 … Non-tax revenue 3.8 …

Sources: Stats SA - Financial statistics of consolidated general government, 22 November 2012 for totals. Line items under direct and indirect taxes from 2013 Budget Review, National Treasury Direct taxes are absolutely progressive. 87% of direct taxes are collected from the top 10% of the distribution…

South Africa Concentration Curves of Direct Taxes (share paid by market income deciles) 1.0

0.9 Direct taxes 0.8 0.7 Market Inc 0.6 45 Degree Line 0.5 0.4 0.3 0.2 0.1 Cumulative Proportion of market income/tax income/tax of market Proportion Cumulative 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Cumulative Proportion of the Population

Source: Own estimates for South Africa based on IES 2010/11. …but less so than in other countries...

Progressivity of South Africa’s Direct Tax System: The Kakwani Coefficient

Peru (2009) 0.43

Mexico (2010) 0.30

Ethiopia (2011) 0.28

Brazil (2009) 0.27

Uruguay (2009) 0.25

Armenia (2011) 0.23

South Africa (2010) 0.13

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

Sources: Armenia (Younger et al, 2014), Bolivia (Paz et al, 2014), Brazil (Higgins and Pereira, 2014), Ethiopia (Hill et al, 2014), Indonesia (Jellema et al 2014), Mexico (Scott, 2014), Peru (Jaramillo, 2014), Uruguay (Bucheli et al, 2014), and own estimates for South Africa based on IES 2010/11. In contrast, indirect taxes are slightly regressive on account of excise taxes South Africa Concentration Curves of Indirect Taxes (share paid by disposable income deciles)

100% 90% 80% 70% 60% 50% 40% 30% disposable income/tax

Cumulative proportion ofCumulative 20% 10% 0%

Cumulative proportion of the population Disposable Income VAT Excise Tax Fuel Levy 45 Degree Line

Source: Own estimates for South Africa based on IES 2010/11. Overall, the tax system is globally progressive.

South Africa. Concentration Curves of All Taxes, 2010 1.0 (share of market income) 0.9 Market Inc 0.8 45 Degree Line 0.7 All taxes 0.6 0.5 0.4 0.3 0.2 Cumulative Proportion ofCumulative Tax 0.1 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Cumulative Proportion of the Population

Sources: Own estimates for South Africa based on IES 2010/11. How progressive is social spending in South Africa? Government Spending: 2010

Incidence 2010/11 analysis (% of GDP)

Total General Government Expenditure 34.8 14.9 Primary government spending 32.2 14.9 Social Spending 17.6 14.9 Total Cash Transfers 3.8 3.8 State old age pension 1.3 1.3 Child Support Grant 1.1 1.1 Disability grant 0.6 0.6 Other grants 0.6 0.6 Foster care grant 0.2 0.2 Other Transfers: Free Basic Services 0.5 0.5 In-kind transfers 12.6 11.1 Education 7 7 Health 4.1 4.1 Housing and urban 1.5 … Other social spending 1.1 … Non-Social Spending (incl. public sector pensions) 14.6 …

Sources: Stats SA - Financial statistics of consolidated general government, 22 November 2012 for total. Line items under direct and indirect taxes from 2013 Budget Review, National Treasury. For Free Basic Services, data represents 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. Primary spending is large relative to other MICs

Primary Government Spending (Share of GDP) 45% 41% 41% 40% 32% 33% 35% 28% 28% 26% 30% 24% 24% 25% 18% 18% 20% 16% 15% 10% 5% 0%

Total cash transfers Education Health Other social spending Non-social spending Total primary spending Direct Transfers and Pensions South Africa spends generously on direct transfers.

Public Social Assistance in the form of Cash Transfers (share of GDP) 4.5% 4.2% 4.0% 3.7% 3.5% 3.3% 3.0% 2.5% 2.3% 2.5% 2.0% 2.0% 2.0% 1.3% 1.5% 1.0% Share of GDP of GDP Share 0.7% 1.0% 0.4% 0.4% 0.5% 0.0%

Source: For Latin America see: Lustig and Pessino, 2014; Paz et al, 2014, Higgins and Pereira, 2014; Scott, 2014; Jaramillo, 2014, Bucheli et al, 2014; Lustig et al, 2014. Preliminary results for Armenia (Younger et al, 2014) Ethiopia (Hill et al, 2014), Jordan (Serajuddin et al, 2014), Sri Lanka (Arunatilake et al, 2014), and own estimates for South Africa based on IES 2010/11. Direct cash transfers as a whole are strongly progressive…

Direct Cash Transfers by Category Concentration Curves for Transfers and Lorenz Curve for

100% Market Income 90% Lorenz for Market Income Direct Transfers 80% Old -age pension 70% Child support grant Disability grant 60% Population Shares 50%

40%

30%

20%

10%

Cumulative Proportion of Income/Spending Proportion Cumulative 0% 0 1 2 3 4 5 6 7 8 9 10 Cumulative proportion of the population by market income deciles Sources: Own estimates for South Africa based on IES 2010/11. Cash transfers are huge relative to the market incomes of the poor…

1200% South Africa. Incidence of Direct Cash Transfers (share of market income by decile) 1000%

800%

600%

400% Share of income market 200%

0% Poorest 2 3 4 5 6 7 8 9 Richest decile decile

Old -age pension Child support grant Disability grant Child care dependency grant Child foster care Grant-in-aid Other grants

Source: Own estimates based on IES 2010/2011. …much more so than other MICs.

Incidence of Direct Cash Transfers (share of market income deciles) 1200% 4.5% 4.0% 1000% 3.5% 800% 3.0% 2.5% 600% 2.0% 400% 1.5% 1.0% 200% 0.5% 0% 0.0% Argentina Armenia Brazil Mexico South Uruguay (2009) (2011) (2009) (2010) Africa (2009) (2010) Poorest decile Richest decile Share of GDP (right axis)

Sources: Argentina (Lustig and Pessino, 2014), Armenia (Younger et al, 2014); Brazil (Higgins and Pereira, 2014), Mexico (Scott, 2014), Uruguay (Bucheli et al, 2014), and own estimates for South Africa based on IES 2010/11. Free Basic Services The incidence of FBS depends on its delivery method,… • Municipalities use different criteria for deciding on indigent households. Given that we can not directly identify the value of free basic services, we model two extremes:

• Benchmark scenario: – assumes all household connected to the national electricity grid equally benefitted from the inverted block tariffs and received an equal share of government subsidies for free basic services. – FBS are treated as an indirect subsidy.

• Sensitivity scenario: – assumes that the subsidy for free basic services is allocated equally among indigent households who are connected to the electricity grid. – household is indigent if market income< R18,000 per year – FBS are treated as direct transfers since municipalities that target FBS typically deliver these through rebates. If FBS were targeted nationwide, there would be clear advantages for the poor, as they would be more progressive

Concentration Curves for Free Basic Services 100%

90% Lorenz for Market Income

80% FBS as a transfer FBS as a subsidy 70% Population Shares 60%

50%

40%

30%

20%

10% Cumulative Proportion of Income/Spending Proportion Cumulative

0% 0 1 2 3 4 5 6 7 8 9 10

Cumulative proportion of the population by market income deciles Source: Own estimates based on IES (2010/11). In-kind transfers on health and education Public education and health spending are large compared to the market incomes of the poor,…

South Africa. Incidence of Education and Health Spending (share of market income deciles) 800% 700% 600% 500% 400% 300% 200% Share of market income income of market Share 100% 0% Poorest 2 3 4 5 6 7 8 9 Richest decile decile

Education Health

Source: Own estimates using IES, 2010/11 and NIDS 2008w1. …more so than in other MICs. This is true for education….

Incidence of Education Spending relative to its size (share of benefits by market income deciles; spending as a share of GDP) 800% 9.0%

700% 8.0% 7.0% 600% 6.0% 500% 5.0% 400% 4.0%

300% of GDP Share 3.0% 200% 2.0%

100% 1.0% Incidence as a share of Market Income Income of Market share as a Incidence 0% 0.0%

Poorest decile Richest decile Share of GDP (right axis)

Source: For Latin America see: Lustig and Pessino, 2014; Paz et al, 2014, Higgins and Pereira, 2014; Scott, 2014; Jaramillo, 2014, Bucheli et al, 2014; Lustig et al, 2014. Preliminary results for Armenia (Younger et al, 2014) Ethiopia (Hill et al, 2014), Jordan (Serajuddin et al, 2014), Sri Lanka (Arunatilake et al, 2014), and own estimates for South Africa based on IES 2010/11. …and health.

Incidence of Health Spending relative to its size (share of benefits by market income deciles; spending as a share of GDP) 500% 7.0% 450% 6.0% 400% 350% 5.0% 300% 4.0% 250% 200% 3.0% 150% 2.0% of GDP Share 100% 1.0% 50% 0% 0.0% Incidence as a share of Market Income Income of Market share as a Incidence

Poorest decile Richest decile Share of GDP (right axis)

Source: For Latin America see: Lustig and Pessino, 2014; Paz et al, 2014, Higgins and Pereira, 2014; Scott, 2014; Jaramillo, 2014, Bucheli et al, 2014; Lustig et al, 2014. Preliminary results for Armenia (Younger et al, 2014) Ethiopia (Hill et al, 2014), Jordan (Serajuddin et al, 2014), Sri Lanka (Arunatilake et al, 2014), and own estimates for South Africa based on IES 2010/11. Primary and secondary education spending disproportionally benefits those at the bottom of the distribution…

100%

90%

80% Lorenz for Market Income 70% Pre-school 60% Primary School 50% Secondary School 40% College education Adult education 30% University education 20% In-kind Education 10% Population Shares

Cumulative Proportion of Income/Spending Proportion Cumulative 0% 0 1 2 3 4 5 6 7 8 9 10 Cumulative proportion of the population by market income deciles

Source: Own estimates using IES, 2010/11. …but no so for post-secondary education

South Africa (% of target population through secondary school) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Preschool Primary Secondary Post Adult University secondary training training centres

y < 1.25 1.25 < y < 2.5 2.5 < y < 4 4 < y < 10 10 < y < 50 y > 50

Source: Own estimates using IES, 2010/11. In contrast, health spending is more progressively distributed than education…

South Africa. Concentration Shares of Health Spending (share of benefits concentrated in each Market Income Decile) 12%

10%

8%

6%

4%

2%

0% Poorest 2 3 4 5 6 7 8 9 Richest decile decile

Source: Own estimates using IES (2010/11) and NIDS (2008w1). Health spending is also progressive in absolute terms.. Concentration of Health Spending and Lorenz Curve 100% for Market Income 90% Lorenz for Market Income 80% In-kind Education 70% In-kind Health 60% Population Shares 50% 40% 30% 20% 10% 0% Cumulative Proportion of Income/Spending of Income/Spending Proportion Cumulative 0 1 2 3 4 5 6 7 8 9 10 Cumulative proportion of the population by market income deciles

Source: Own estimates using IES, 2010/11 and NIDS 2008w1. With health utilization more evenly distributed across socioeconomic groups than in other MICs

Health Share of Individuals in Beneficiary Households by Income Group 100% 90% 80% 70% 60% 50% 40% 30% Share of Target Population Population Target of Share 20% 10% 0% Armenia Bolivia Brazil Ethiopia Indonesia South Africa

y<1.25 1.25 50

Source: Armenia (Younger et al, 2014), Bolivia (Paz et al, 2014), Brazil (Higgins and Pereira, 2014), Ethiopia (Woldehanna et al, 2014); and Indonesia (Jellema et al, 2014). For South Africa, own esmates based on IES 2010/2011. Social spending as a whole is strongly progressive

South Africa. Concentration coefficients for Spending

Free basic services (as transfers) (0.39) Care dependency (0.37) Child support grant (0.34) Foster Care grant (0.32) Disability grant (0.25) Primary education (0.19) Pre and Primary education (0.19) Old age pension (0.17) Grant-in-aid (0.15) Secondary School Education (0.12) Pre-school education (0.11) Health Spending 0.00 Other grants 0.01 Free basic services (as indirect subsidies) 0.16 Tertiary Education Spending 0.50 Market income Gini 0.77 (0.60) (0.40) (0.20) - 0.20 0.40 0.60 0.80 1.00

Source: Own estimates using IES, 2010/11.

What is the net impact of taxes and government transfers on inequality and poverty?

30 Inequality falls substantially with Government interventions,…

The Gini Coefficient before and after Taxes, Transfers and Free Services, Education and Health Falls by over one-fifth 0.85 0.771 0.75 0.75 0.694 0.695 0.65 0.596 0.55

0.45

0.35

0.25 Market Income Net Market Disposable Post-Fiscal Final Income Income Income Income - direct + transfers - VAT, Fuel, + Educ, taxes & FBS excise +Health …more so than in other middle-income countries…

Change in Gini: Disposable vs Market Income (in Gini points)

0.00 -0.02 -0.04 -0.06 -0.08 -0.10 -0.12 -0.14 -0.16 -0.18 -0.20 Source: Armenia (Younger et al, 2014); Bolivia (Paz et al, 2014); Brazil (Higgins and Pereira, 2014); Ethiopia (Woldehanna et al, 2014); Indonesia (Jellema et al 2014); Mexico (Sco, 2014); Peru (Jaramillo, 2014); Uruguay (Bucheli et al, 2014); Lusg(2014) based on Costa Rica (Sauma et al, 2014), El Salvador (Beneke de Sanfeliu et al, 2014), and Guatemala (Cabrera et al, 2014); and own esmates for South Africa based on IES 2010/11. …but inequality is still higher after fiscal policy than inequality prior to fiscal policy in other countries

Gini Coefficient for Each Income Concept

Armenia 0.771 (2011) 0.75 Brazil (2009) 0.694 Indonesia 0.65 (2012) 0.579 0.596 Jordan (2010)

0.55 Mexico (2010)

Peru (2009) 0.45 0.439 South Africa (2010) 0.35 Sri Lanka (2009) Uruguay 0.25 (2009) Market Income Net Market Income Disposable Income Post-Fiscal Income Final Income

- direct - VAT, + Educ, taxes + transfers - Fuel +Health & FBS - Excise Poverty also declines substantially…

60% 52.3% 52.5% 50.1% 50% 46.5% 46.7% 45.1% 40.8% 41.0% 39.6% 40% 34.2% 32.2% 29.0% 30% 23.4%

20% National food poverty line1 20.2% Official consumption based (food poverty line) 10% National lower bound poverty line 2 Official consumption based (lower bound) 0% National upper bound poverty line3 Market Income Net Market Income Disposable Income Post-fiscal Income

- Indirect + direct taxes - direct transfers taxes + indirect subsidies 34 With the effect on poverty larger than other middle income countries. Poverty Headcount at $2.50 per day (PPP) 70 Armenia 60 Bolivia 50 46.2 46.4 Brazil 40 39 Indonesia Jordan

Percent 30 33.4 Mexico 20 Peru 10 South Africa Sri Lanka 0 Market Income Net Market Disposable Post-Fiscal Uruguay Income Income Income

- Indirect - direct + direct taxes taxes transfers + indirect 35 subsidies Direct cash transfers are highly effective at redistributing fiscal resources towards the poor...

Poverty Reducing Effectiveness -13% 4.0

3.5 -11%

3.0 -9% 2.5 -7% 2.0

-5% 1.5 Effectiveness Indicator Indicator Effectiveness Change in US$2.50-a-day Poverty Headcount Headcount US$2.50-a-day Poverty in Change -3% 1.0

0.5 -1% 0.0 Argentina Brazil (2009) Indonesia Mexico Peru (2009) South Africa Uruguay (2009) (2012) (2010) (2010) (2009)

Change in Poverty Rate Effectiveness Indicator

Source: Argenna (Lusg and Pessino, 2014); Brazil (Higgins and Pereira, 2014); Indonesia (Jellema et al), Mexico (Sco, 2014); Peru (Jaramillo, 2014), Uruguay (Bucheli et al, 2014); and Inchauste, Lusg, Maboshe, Purfield and Woolard (2014) for South Africa based on IES 2010/11. …and in reducing inequality.

-0.06 Inequality-reducing Effectiveness 1.6 1.4 -0.05 1.2 -0.04 1.0 -0.03 0.8 0.6 -0.02 Change in Gini Gini in Change 0.4 Effectiveness Indicator Indicator Effectiveness -0.01 0.2 0.00 0.0

Change in Gini from Direct Transfers Effectiveness Indicator

Source: Argenna (Lusg and Pessino, 2014); Brazil (Higgins and Pereira, 2014); Indonesia (Jellema et al), Mexico (Sco, 2014); Peru (Jaramillo, 2014), Uruguay (Bucheli et al, 2014); and Inchauste, Lusg, Maboshe, Purfield and Woolard (2014) for South Africa based on IES 2010/11. Conclusions

• South Africa uses its fiscal instruments to effectively reduce market income poverty and inequality through a slightly system and highly progressive social spending

• On the tax side, fiscal policy relies on a mix of progressive direct taxes such personal income taxes, slightly regressive indirect/consumption taxes that when combined generate a slightly progressive tax system.

• On the spending side of fiscal policy, social spending is not only progressive, but it also contributes to large reductions in poverty and inequality.

• In fact, South Africa performs very well when compared with other middle income countries: it achieves the most “redistribution” compared to the other middle income countries in the CEQ analysis. Conclusions (2)

• However, there are concerns about the quality of such spending which begs the question could more be done to improve the quality of such services so to allow education and health spending to maximize their potential in reducing poverty and inequality.

• Although fiscal policy is going a long way towards achieving redistribution, the level of inequality and poverty in South Africa after taxes and spending remains unacceptably high.

• Addressing the twin challenges of poverty and inequality going forward in a way that is consistent with fiscal sustainability requires higher and more inclusive economic growth. This would be particularly important in addressing the need for jobs and higher incomes, especially at the lower end of the income distribution, helping narrow the gap in incomes between the rich and the poor and reinforce the effectiveness of fiscal policy. Questions for additional analysis…

• Potential further targeting of free basic services • Distributional impacts of proposed health insurance, given sources of funding considered • Fiscal space for further redistributive policies • Potential labor disincentive effects of transfers • – efficiency tradeoffs • Inter-generational mobility and implications for transfers in future - Quality of education