Abstract
Selected Issues
What Led to the Doubling of Public Debt in the Last Decade? was Debt Good for Growth?1
A permanent increase of 4 percentage points of GDP in national government expenditure underlies the doubling of public debt in the last decade. The wage bill accounted for most of the expenditure increase (64 percent), followed by the interest bill (23 percent). The debt expansion, thus, financed a countercyclical fiscal policy centered on current spending, which likely shielded the impact of subdued economic activity, but had limited permanent effects on growth. Had resources devoted to wage increases and debt service payments been invested in more productive outlays, such as highly productive capital expenditure and reforms in key network industries, the growth gains would have been higher.
A. Introduction
1. South Africa’s fiscal policy, which has traditionally been sound, is now facing challenges. From FY94/95 to FY08/09 the national government debt declined from 48 percent of GDP to 26 percent in the context of strong economic growth. Since the global financial crisis, however, public debt has roughly doubled, reaching 53 percent of GDP by FY17/18. In the last two fiscal years, the fiscal position has faced significant revenue shortfalls in the context of slowing economic growth and governance weaknesses. Meanwhile, spending pressures have increased, and contingent liabilities from state owned enterprises (SOEs) have materialized. These developments triggered a significant increase in financing needs. Staff projections suggest that on current policies the growth outlook would remain lackluster and debt would continue to rise.
2. Favorable global financing conditions have mitigated financing risks so far, but conditions are becoming less benign. During recent years, South Africa, like other emerging markets, has benefited from foreign exchange inflows that supported the country’s borrowing needs. Nevertheless, borrowing costs have increased, especially after credit rating agencies cited low growth and contingent liabilities of SOEs as key vulnerabilities that led to sovereign debt downgrades. Moreover, the favorable global financing backdrop is changing, especially considering monetary policy expectations in the US, and recent changes in appetite toward emerging markets.
Local Currency Sovereign Bond Yield Spread to U.S.
(Basis Point)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: Bloomberg, L.P. and IMF staff estimates.Local Currency Sovereign Bond Yield Spread to U.S.
(Basis Point)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: Bloomberg, L.P. and IMF staff estimates.Local Currency Sovereign Bond Yield Spread to U.S.
(Basis Point)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: Bloomberg, L.P. and IMF staff estimates.3. Stabilizing debt dynamics at comfortable levels as soon as possible to strengthen economic resilience is advisable. This paper first takes stock of the factors behind the doubling of the national government debt and contingent liabilities between 2007 and 2017, and then extends the analysis at the level of the general government and SOEs. Given the critical role that expenditure played in the accumulation of debt, the paper identifies areas with room for gains in efficiency or rationalization, by using the IMFs Expenditure Assessment Tool (EAT), which benchmarks spending and its composition against comparators.2 The paper concludes with a discussion on the impact of debt accumulation on growth.
B. What Drove the Accumulation of Debt and Contingent Liabilities?
National Government Debt
4. The fiscal deficit was the main factor behind national debt accumulation between 2007 and 2017. Changes in the actual debt stock as a share of GDP are due to changes in the numerator (i.e., fiscal deficit, accumulation of cash balances, and a stock-flow adjustment term), or the denominator (i.e., nominal GDP). The stock flow adjustment term includes CPI adjustment of inflation-indexed debt, valuation changes of the foreign currency denominated debt, and other stock flow adjustment movements including from debt management operations. 3 Results show that the fiscal deficit accounted for 76 percent of the change in the numerator of the debt ratio, the stock flow adjustment for 18 percent, and the accumulation of the cash balances for the remaining 6 percent.
Breakdown of Debt Numerator Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and Fund staff estimates.Breakdown of Debt Numerator Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and Fund staff estimates.Breakdown of Debt Numerator Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and Fund staff estimates.5. An increase in expenditure as a share of GDP accounted for 90 percent of the increase in the deficit, while revenue shortfalls contributed only marginally. A decomposition analysis shows the contribution of each numerator component to the debt increase by breaking the fiscal deficit component into changes in expenditure and revenue as a share of GDP.
Expenditure increases undertaken at the time of the global financial crisis (2008 and 2009 compared to 2007) primarily on the wage bill (1.5 percent of GDP), purchases of goods and services (1 percent of GDP), and social benefits (0.5 percent of GDP) were not clawed back thereafter. Moreover, nominal expenditure ceilings (set based on more favorable growth assumptions) were only marginally adjusted down, under a de-facto presumption that weaker growth was a temporary phenomenon. Compounding these developments, the materialization of contingent liabilities (e.g., support to Eskom in 2015, and support to South African Airlines and the post office in 2017) also contributed in raising the public expenditure-to-GDP ratio.
Revenue, on the other hand, largely recovered. Revenue declined by about 3 percentage points of GDP to 24.1 percent of GDP in 2010, due to automatic stabilizer effects and the provision of tax relief. Since then, revenue recovered to 26.4 percent of GDP in 2017, buoyed primarily by the impact of the growth recovery on personal income, VAT, and international trade tax collections, and by tax policy measures, including bracket creep and changes in income tax rates, and adjustment of excise rates.
National Government Deficit, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: SARB Quarterly Bulletin (March 2018).National Government Deficit, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: SARB Quarterly Bulletin (March 2018).National Government Deficit, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: SARB Quarterly Bulletin (March 2018).6. The wage bill accounts for the bulk of the increase in total expenditure. A decomposition analysis looks at both the economic and the functional classifications of expenditure to identify areas of major changes:
The economic classification shows that the wage bill accounts for 64 percent of the increase in expenditure, followed by the interest bill (23 percent) reflecting the debt increase and higher borrowing costs, and social benefits (13 percent) reflecting an expansion in the coverage of social grants that occurred during the period. Reduced capital expenditure and grants (i.e., international transfers) broadly offsets increased expenditure on subsidies and to a lesser extent goods and services (not shown in chart).
The functional classification reflects consistently the economic classification results.4 The largest area of growth is “general public services”, which captures both the increasing wage and interest bills. Education and health, two of the largest employment sectors, follow — likely reflecting the contribution of the large wage bill increase. Social protection spending increase reflects rising social benefits.
Economic Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Economic Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Economic Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.
Functional Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018), 2018 Budget Review, and IMF staff estimates.Functional Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018), 2018 Budget Review, and IMF staff estimates.Functional Classification: Expenditure and Change, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018), 2018 Budget Review, and IMF staff estimates.General Government Debt
7. The results derived in the previous section remain largely unchanged when expanding the analysis to the general government.5 General government debt is estimated to have increased from 29.3 percent of GDP in 2007 to 55.7 percent in 2017, following a similar debt trajectory as the national government, and suggesting that the debt of the rest of the general government is between 2 and 3 percent of GDP. The debt accumulation arising from local governments is likely limited reflecting the presence of small surpluses or deficits. Provincial governments and extra-budgetary funds are not allowed to borrow. Social security funds have maintained a surplus during the projection period. Local government revenue increases contributed to the general government deficit reduction since 2007. Expenditure therefore played a bigger role in the deficit increase than at the national government level. The wage bill continued to be the main contributor to total expenditure increase (55 percent), followed by purchases of goods and services (19 percent), the interest bill (17 percent), and social benefits (14 percent). There is a net decline in the remaining expenditure categories of about 5 percent of total expenditure.6
SOE Debt
8. SOE debt has grown rapidly since 2007 driven primarily by non-financial SOEs.7 Total SOE loan and bond debt increased from 8 percent of GDP in 2007 to 15.6 percent in September 2017. Non-financial SOEs represented about 86 percent of SOE debt (13.5 percent of GDP) and 86 percent of the overall increase in SOE debt during the period. This reflects the difference in size between the non-financial SOE sector (37 percent of GDP in assets) and the financial SOE sector (5 percent of GDP in assets).
9. Growth in non-financial SOE debt has been driven by insufficient operational results to sustain investment. The non-financial SOEs, as a group, had cash flow deficits in 13 out the 14 most recent years, and the average cash flow deficit was about 1.6 percent of GDP in the last 10 years. Operational revenue has not been sufficient to sustain the investment program, which has averaged 3 percent of GDP per annum. While the average cash flow deficits remained broadly unchanged during the period due to some improvement in revenue collection, expenditure grew by 0.6 percent of GDP on the back of increased spending on goods and services and interest. Issues at Eskom, the largest non-financial SOE, are illustrative of difficulties SOEs have been facing, including collection of arrears, procurement of key inputs at high costs, overstaffing, operational inefficiencies, delays and cost overruns in the implementation of investment projects, and unfunded mandates.
10. SOE borrowing has increasingly taken the form of loans given diminishing interest in market bond issuances. In 2007, about 72 percent of non-financial SOEs obligations were bonds and 28 percent loans. Most recently, in 2017 the mix has been 51 percent for bonds and 49 percent for loans, with increasing loan placements with domestic financial institutions and reliance on multilateral financing. With deteriorating finances and limited market appetite for non-financial SOE bonds, a significant share of non-financial SOE debt is guaranteed by the government (see below).
Non-financial Public Enterprise Debt and Deficit, 2004–17
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Non-financial Public Enterprise Debt and Deficit, 2004–17
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Non-financial Public Enterprise Debt and Deficit, 2004–17
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.
Composition of Non-Financial SOE Debt, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Composition of Non-Financial SOE Debt, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Composition of Non-Financial SOE Debt, 2007–17
(Percent of GDP, calendar year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: SARB Quarterly Bulletin (March 2018) and IMF staff estimates.Contingent Liabilities
11. The SOE sector has been a key driver of the expansion in contingent liabilities for the budget. Guarantees on SOE loans are the most frequent types of contingent liabilities and the ones that grew the most. They amounted to 2.9 percent of GDP in FY07/08 and increased to about 9 percent in FY17/18. The largest loan guarantees were to Eskom and the independent power producers, which combined amounted to 7.3 percent of GDP in FY17/18. The independent power producers’ initiative was introduced in FY12/13 to alleviate energy shortages at the time.
Composition of Contingent Liabilities, 2007–17
(Percent of GDP, fiscal year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: National Treasury and IMF staff estimates.Composition of Contingent Liabilities, 2007–17
(Percent of GDP, fiscal year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: National Treasury and IMF staff estimates.Composition of Contingent Liabilities, 2007–17
(Percent of GDP, fiscal year)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: National Treasury and IMF staff estimates.12. Other sources of contingent liabilities have also been significant. The Road Accident Fund, which pays claims to victims of road accidents, has seen increased claims from 1.4 percent of GDP in FY07/08 to 4 percent in FY17/18, and has been the second most important source of growth of contingent liabilities outside SOEs. Contingent liabilities related to the medical assistance program for civil servants, and claims related to disputed bills together amounted to 2.2 percent of GDP in FY17/18. However, the amount has been declining over time.
C. Benchmarking of Spending Levels and Composition
Level of Expenditure and Composition by Economic Classification8
13. Total spending as a percent of GDP increased significantly more than in comparator country groups in 2007–16. Total spending grew by about 5.5 percentage points of GDP, a rate considerably higher than the one observed in SADC, EM, and OECD countries. Total spending is now broadly at par with the level of spending in other EMs and significantly above the SADC average.
Change in Total Spending, 2007–16
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Change in Total Spending, 2007–16
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Change in Total Spending, 2007–16
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.
General Government Spending, 2007 and 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.General Government Spending, 2007 and 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.General Government Spending, 2007 and 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.
Current and Capital Spending, 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Current and Capital Spending, 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Current and Capital Spending, 2016
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.14. The share of budgetary capital expenditures is lower than in the SADC and EM countries. This is because budget capital expenditure has been compressed to accommodate increasing wage and interest bill expenditure. However, this result must be interpreted with caution since in South Africa, SOEs carried out about 3 percentage points of GDP of capital spending on average that is not captured in these figures, given the coverage. Compared to the SADC, South Africa spends more on interest and other current expenditure because of its higher debt levels and more extensive social assistance expenditure.
South Africa: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.South Africa: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.South Africa: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.
SADC: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.SADC: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.SADC: Expenditure by Economic Classification, 2016
(Percent of total)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.15. The wage bill is on the high side of the EM distribution, and is driven by high compensation. Public employment of working age population is lower in South Africa than the EM average, while both the wage bill as a share of GDP and of public expenditure are above the EM average. This points to high compensation as the main reason for the higher wage bill. These findings are consistent with those in a special annex on public employment and compensation in the 2017 Medium Term Budget Policy Statement.
Benchmarking for Wage Bill and Employment
(Percent)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Assessment Tool.Benchmarking for Wage Bill and Employment
(Percent)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Assessment Tool.Benchmarking for Wage Bill and Employment
(Percent)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Assessment Tool.16. The quality of infrastructure compares favorably with respect to that of other EMs and SADC countries, but tends to lag that of OECD countries. Air transport and roads appear to be the areas where South Africa does better even compared to OECD standards. However, in general, the quality of infrastructure lags that of the OECD, in particular on ports and railroads.
Composition of Expenditure by Functional Classification9
Education
Infrastructure Quality
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Infrastructure Quality
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Infrastructure Quality
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.17. There are signs of inefficiency in education spending. Several SADC countries spend less at the primary education level and get better outcomes measured in terms of net enrollment. Also, EMs get considerably better outcomes with a similar level of expenditure. For example, with a similar spending level than South Africa, some SADC countries get a 95 percent enrollment rate. For a more detailed discussion of inefficiencies in education see Mlachila et al. (2018), and several auditor general reports for a discussion on wasteful expenditure in the sector.
Government Education Spending and Outcome, Primary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Government Education Spending and Outcome, Primary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Government Education Spending and Outcome, Primary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.
Government Education Spending and Outcome, Secondary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Government Education Spending and Outcome, Secondary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Government Education Spending and Outcome, Secondary 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.18. The ongoing shift of education spending toward providing free tertiary education spending is unlikely to benefit the most vulnerable. Since the 2016 budget review, spending reallocations were carried out to fund additional tertiary education subsidies. International evidence from many EMs (see Fiscal Policy and Income Inequality (2014), Fiscal Monitor (2007)) suggests that tertiary education spending tends to benefit disproportionately the better off—as is also the case in South Africa—since it is regressive in absolute terms. This is because the poor tend to have weaker educational backgrounds than the non-poor and are less likely to seek university studies. Moreover, students who complete tertiary education typically earn higher wages. As a result, those who can afford tertiary education would be willing to pay for it without any subsidization. By subsidizing students that would have paid in any case, less funds are available to help those in need.
Indicators of Health and Health System
(Per 1000 people, unless specified)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Indicators of Health and Health System
(Per 1000 people, unless specified)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.Indicators of Health and Health System
(Per 1000 people, unless specified)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.
Health Efficiency Frontier 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are average of SADC.Health Efficiency Frontier 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are average of SADC.Health Efficiency Frontier 1/
(Latest Data)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are average of SADC.19. Similar evidence of inefficiencies is present in healthcare. While health expenditure is slightly above EM and SADC countries, healthy life expectancy indicators (HALE) are considerably below what EM and several SADC countries achieve.10 Looking more broadly at other indicators, this is reinforced by relatively high infant mortality, lower life expectancy at birth, and lower number of physicians per 1000 inhabitants, compared to other EMs.
Social Protection
20. Spending on social assistance is considerably higher than in the OECD, EM, and SADC countries.11 This is mainly because of the 83 percent coverage of the poorest 20 percent of the income distribution, which is considerably broader than in most EMs (59 percent) and even OECD countries (79 percent). In terms of benefit incidence, the poorest 20 percent of the income distribution gets 20 percent of the benefits in South Africa compared to 28 percent in EMs on average and 34 percent in OECD countries according to this metric. These metrics likely do not take into account issues in the distribution of social assistance, where intermediaries force beneficiaries to get contracts for other services reducing the effective grant amount that gets to beneficiaries.
Social Assistance Spending
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: FAD and IMF staff estimatesSocial Assistance Spending
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: FAD and IMF staff estimatesSocial Assistance Spending
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Sources: FAD and IMF staff estimates
Social Assistance Coverage and Benefit Share of Poorest 20 Percent 1/
(Percent, latest value available)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Social Assistance Coverage and Benefit Share of Poorest 20 Percent 1/
(Percent, latest value available)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Social Assistance Coverage and Benefit Share of Poorest 20 Percent 1/
(Percent, latest value available)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the average of SADC.Economic Affairs
21. Energy subsidies are relatively large compared to SADC, EMs, and OECD countries. IMF (2015) estimates the subsidies at 13 percent of GDP. They derive primarily from coal and petroleum (8 percent of GDP) and electricity (1.2 percent of GDP). Subsidies stem primarily from the exemption of fuel from VAT (foregoing revenue estimated at 1.5 percent of GDP) and from the gap between excise collections on energy products and the valuation of the externalities they generate (including global warming, pollution, congestion).
Energy Subsidies by Product, 2015 1/
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the median for countries in the region.Energy Subsidies by Product, 2015 1/
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the median for countries in the region.Energy Subsidies by Product, 2015 1/
(Percent of GDP)
Citation: IMF Staff Country Reports 2018, 247; 10.5089/9781484371473.002.A001
Source: FAD Expenditure Assessment Tool.1/ Dashlines are the median for countries in the region.D. Was Debt Good for Growth?
22. The spending increase that drove the large debt accumulation helped smooth the impact of the global financial crisis, but likely did not have a material impact on growth. National government spending increased by 4 percentage points of GDP in the last decade, of which 3 points occurred during the global financial crisis years. With multipliers estimated between 0.4 and 0.9, the fiscal stimulus likely prevented a sharper decline in growth ranging between 1.2 and 2.7 percentage points during the crisis. However, the expenditure increase over the period 2007–17 was mainly explained by wage and interest payments outlays, and to a lesser extent social benefits. Such composition and the relative inefficiency of spending in health and education suggest that the increase in spending that drove up debt did not have material permanent effects on growth. These findings are broadly consistent with several studies on the impact of fiscal policy on economic activity in South Africa (e.g. Jooste et. al (2013) and those surveyed in Makrelov et. al. (2018)). Overall, although multiplier estimates vary, most studies suggest an impact of the government spending expansion on economic activity during the global financial crisis in the order of magnitude estimated by staff, but limited effects in the longer term12.
23. The expansion in the wage bill during the last decade has exacerbated budget rigidities, without raising productivity. Increases in the wage bill—a substantive part of the budget—have remained largely untouched during unfavorable economic times, forcing major compression in necessary goods and services and investment grants. The increase in the interest bill associated to the rising debt further compounded the rigidities created by a large wage bill. A meaningful reduction in the wage bill would improve budget flexibility and spread more fairly the burden of adjustment across all segments of the population.
24. SOE finances pose large fiscal risks to the budget, and their large investments have not been optimal for growth. The protracted deficits that SOEs have been running as a group during 13 of the last 14 years, and the need of most SOEs to get government guarantees to borrow, suggest that they have difficulties to sustain investment plans. Turnaround plans for SOEs are urgently needed to ensure their contribution to growth, and mitigate the fiscal risks they pose. Any new government guarantees should be made contingent on efficient performance in line with well-defined turnaround targets.
25. Spending on education and health has absorbed a considerable part of the debt increase, but weak quality service has not resulted in workforce productivity gains. The relatively weak outcomes South Africa achieves compared to other SACU countries and EMs point to the need to control personnel costs, which represent a sizable portion of total expenditure in these sectors, and implement the auditor general recommendations to reduce wasteful expenditure.
26. Social assistance benefits are a valuable policy option to compensate the poor for possible adverse effects from fiscal adjustment. The broad coverage and good targeting make social assistance benefits an effective choice to address adverse consequences of fiscal measures on the poor. Continued efficient spending in this area is likely to have an important payoff, particularly if issues related to the distribution of grants are addressed.
27. The significant debt accumulation observed in the last 10 years could have been used more productively. While the countercyclical fiscal policy at the time of the global financial crisis and the expansion of the safety net that the debt helped finance were appropriate, the significant resources that were spent on wages and associated borrowing costs could have been invested in more productive options. These include additional productive capital expenditure to support private sector activity at the national and subnational levels; structural reforms to increase the productivity of the economy, especially in network industries where large SOEs operate; and improvements in the efficiency of basic education and healthcare. All of these outlays would not only have contributed to increase the productivity of workforce, but also to reduce inequality.
References
Coady, D., I. Parry, L. Sears, and B. Shang. 2017. “How Large Are Global Fossil Fuel Subsidies?” World Development, Vol. 91, pp 11–27.
Garcia Escribano, M. and C. Liu. 2017. “Expenditure Assessment Tool (EAT)”, Technical Notes and Manuals, Fiscal Affairs Department, International Monetary Fund, Washington, DC.
International Monetary Fund. 2014. “Fiscal Policy and Income Inequality.” IMF Policy Paper, Washington, DC.
International Monetary Fund 2017. “Tacking Inequality” Fiscal Monitor, October.
Jooste, C., G. Liu, and R. Naraidoo. 2013. “Analyzing the Effects of Fiscal Policy Shocks in the South African Economy” Economic Research Southern Africa, May 2013.
Makrelov, K., C. Arndt, R. Davis, and L. Harris. 2018. “Fiscal Multipliers in South Africa. The importance of financial sector dynamics”, SA-TIED Working Paper 1, February 2018.
Mlachila, M. and T. Moeletsi. 2018. “Failing to Make the Grade: Causes and Consequences of South Africa’s Poor Educational Performance”, (forthcoming AFR Departmental Paper)
National Treasury of South Africa. 2017. Medium Term Budget Policy Statement, Annex B. “Compensation Data”. October. http://www.treasury.gov.za/documents/mtbps/2017/default.aspx
South Africa Auditor General PFMA and MFMA Reports, https://www.agsa.co.za
South African Reserve Bank. 2018. Quarterly Bulletin, March. https://www.resbank.co.za/publications/quarterlybulletins/pages/quarterly-bulletin.aspx
Prepared by Alejandro Simone; reviewed by Ana Lucía Coronel.
See Escribano and Liu (2017) for a discussion of the methodology and data sources. South Africa’s spending level and composition are benchmarked against the Southern African Development Community (SADC) countries, Emerging and Middle-Income Countries (EM), and, in some cases, against the Organization of Economic Cooperation and Development (OECD).
The analysis in this section is based on SARB’s national government public finance data, which is available already on a calendar year basis through 2017 by level of government. This allows to estimate what type of spending the grants to other levels of governments in the national government accounts are financing. It is assumed that, if local governments spend X percent of their total spending on wages, X percent of the grants the national government sends to the local government are spent on wages assuming money is fungible.
Given that functional classification data is not directly published for the national government, consolidated government level information was used as a proxy after applying a simple transformation to estimate calendar year data.
While general government revenue, expenditure, and financing data is available from SARB, consolidated general government debt data is not available. General government debt was proxied by adding the bond and loan debt information available to be consistent with the national government debt definition. Data is from the balance sheets of extra-budgetary funds, provincial governments, and local governments. Social security funds do not have loan and bond debt. Given that there is no information on intra-government debt this risks double counting.
The deficit accounts for a lower share of the change in the numerator (56 percent), which is picked up by the stock flow adjustment that includes residual errors. This change is unlikely to be meaningful, as it mostly represents the fact that general government debt is estimated with error.
Non-financial SOEs covered by the SARB data include Eskom (electricity), Transnet (transportation and logistics), Telkom (telecom), SANRAL (road construction), the water boards, and most of the largest non-financial enterprises and corporations. Similarly, financial SOEs include the Development Bank of South Africa, the Land Bank, and the Industrial Development corporation.
The fiscal coverage in FAD’s expenditure assessment tool is general government for most countries in the sample. For South Africa, the coverage is the consolidated government, which includes the central government, provincial governments, social security funds, transfers to local governments, and some public entities. The latest year available option in the tool is used in some cases to maximize the amount of data on comparators.
See footnote 8 for the fiscal coverage of the data.
Healthy life expectancy (HALE) is a measure of health expectancy that applies disability weights to health states to compute the equivalent number of years of life expected to be lived in full health.
Includes free services subsidy provided by the Treasury to local governments assuming it is targeted to the poor. Coverage is the number of individuals in the quintile who live in a household where at least one member receives the transfer divided by the number of individuals in that quintile. Benefit incidence is equal to the sum of all transfers received by all individuals in the quintile divided by the sum of all transfers received by all individuals in the population.
Multiplier estimates tend to be below one in the short term and negligible in the longer term in most studies. However, Makrelov et. al. (2018) argue that these estimates are low because they ignore important financial accelerator channels that were applicable to South Africa’s context during the global financial crisis. When these channels are incorporated, peak multipliers of 2.5 and 3.5 can be obtained depending on the response assumed for foreign savings and a more persistent response on output.
