r and g revisited
Is the fiscal position sustainable? Yes, but only just – and with primary balance forecasts that are somewhat optimistic – leaving the framework open to deterioration on shocks.
We approach this question in two ways:
The classic Blanchard r versus g approach, where we show that that r > g for many years to come, meaning that the Treasury will need to run a (rather worryingly large) primary surplus indefinitely to even have a chance of stabilising debt.
Bottom-up. We forecast forward a 10-year baseline estimate for the fiscus from now until 2033 using plausible but bullish assumptions on growth and spending restraint.
While we share the market’s view that the National Treasury is overoptimistic on debt stabilisation, we do not think that debt will blow up (helped by its structure and low share of FX debt). Our baseline is for debt stabilisation to occur in FY2027, as the energy constraint lifts and years and years of budget restraint finally stabilises expenditure. The problem of course, is if slumps in growth occur, debt will shift higher and higher still.
The downside risk is that the 2024 Election delivers a fiscal expansionist government (or even just one that only runs a primary balance or small deficit), and debt levels deteriorate faster. The major upside risk is that state-owned enterprises are restructured and no longer bleed the fiscus dry in the long run.
This is the first of a series of papers on pressing fiscal issues.
Where are we now?
South Africa has a debt problem. Over the very long sweep of its history, the country has seen up and down cycles of debt. But the post-2008 period has been the most significant period of debt accumulation since the 1991-1998 period, and government debt (as a share of GDP) is now at post World War II levels.
The accumulation of debt is due to rapid expenditure growth. This has been primarily on wages which grew quickly post 2008.
Tax increases have not helped. Attempt to raise taxes have had negative effects, with estimates of tax multipliers suggesting that these have not supported increased revenue but may have reduced it.
There is no evidence that the debt expansion has provided any fiscal stimulus. Fiscal multipliers suggest that expansionary fiscal policy has been an overall drag on growth. The negative effects (higher interest rates on borrowing and higher taxes) have outweighed any positive economic stimulus.
With this in mind, we ask the question: is the fiscal position sustainable? We approach this in two ways: the theoretical r and g approach and a bottom-up forecast of the fiscal position over the next ten years. We compare our estimates to those of the Treasury and the IMF.
Approach 1: r and g
Debt service costs and growth give us a quick way of establishing sustainability. The simplest and most intuitive approach to considering fiscal sustainability is by comparing r, the average interest rate paid on net government debt,1 with g, the nominal growth rate.2 The mathematics tell us that when r > g, then the government needs to run a primary fiscal surplus to stabilise debt; while if r < g, then government can run a small primary deficit but still stabilise net debt. 3
Intuitively, this is because revenues grow at g and (with some simplifying assumptions) debt grows at r. So, if r > g, then debt growth will outpace revenue growth. Houston, we have a problem. Taking the data back to 1980, we see that generally periods of r > g are associated with increases in debt to GDP.
Our estimate of nominal r is around 6.8%, based on an estimate of the effective nominal interest rate on government gross debt. This may seem rather low – surely with bond yields closer to 11 per cent, average borrowing costs should be significantly higher? The answer is three-fold. First, issuance is along the curve; second, the Treasury has a long-dated debt portfolio, much of it issued when yields were lower; and third, new offshore issuance is increasingly in cheap international financial institution (IFI) money, e.g., from the World Bank or IMF. Of course, the risk is that the cost of debt will start to spiral as redemptions start to kick in. The second risk, which is exactly what has happened to Kenya and Ghana, is cheap IFI money supports an unsustainable fiscal path and only delays the reckoning. (See box).
We can estimate this by looking at the Treasury’s redemption profile over the next few years, and the cost of issuance coming to derive a shadow interest rate on debt.
Second, why the large difference depending on gross versus net debt? The government holds reasonably large balances for an emerging market – approximately 4.8% of GDP at the end of 2022/23.
We see r rising slowly toward 8% as debt is refinanced at higher rates. What is the trajectory going forward? r is low because of the historical issuance strategy. As debt continues to accumulate, however, and needs to be re-issued at ever higher rates, the risk is that r rises quickly. This is the debt-spiral problem that most concerns the market, unsurprisingly.
Over the next three years, for example, approximately 10% of all debt will be refinanced. While over the medium term this is manageable, if interest rates remain high for several years, then r could rise over time and create significant problems.
The entire debt stock will need to be refinanced within 11 years. If the fiscal consolidation takes too long, and interest rates remain elevated, then this will be a new squeeze on the fiscal position.
Our baseline estimate is that due to refinancing, r will rise over time from the current upper 6s to closer to 7.5%. We base this on our estimate of the refinancing projections.
Our view is that long-run growth will remain relatively flat while the structural reform agenda remains stuck. Inflation is firmly under control, suggesting that long-run g of around 6.5%. (The bounce up in g in 2024/25 is due to our view of higher inflation).
Our baseline is thus that the Treasury will need to run a primary fiscal surplus for some time to counteract the rising rate on government debt. We can derive the required primary surplus. We estimate that under reasonably moderate assumptions (set out below), the Treasury will be able to achieve a flat primary balance that keeps debt growth reasonably flat. We do not, however, at this stage see that the Treasury will be able to reduce debt. The main budget deficit slowly improves toward the 4% of GDP level. This highlights how weak the long-term fiscal position remains – because interest payments are projected to slowly rise to 5% of GDP level, the Treasury will have to “keep running to keep still” for quite some time – i.e., run a primary surplus just to maintain sustainability. (The r – g approach above shows the same).
This is a narrow achievable window with a number of risks, which we set out in detail in the next section. Moreover, historically, before 2008, the Treasury was able to manage an average primary deficit of 1.8%.
This leads us to our forecast for the debt trajectory based on these assumptions. We also do a full bottom-up process to derive the forecast in the next section.
Our issuance framework highlights some of the issues with debt levels here and debt sustainability.
We can see the significant Eskom debt relief additions to the gross funding requirement – i.e. even though the headline or above the line budget numbers are “ok”, they rise by another 20% on the cash needed for debt relief.
Switch auctions are not being used to dampen short run redemptions but will be required medium run as the burden of redemptions rise.
Significant cash drawdown is required this year in order to prevent SAGB issuance moving higher. We pencil in an additional ZAR12.3bn cash usage this year, but really more than this is impossible given NT’s risk views on cash stocks required.
To contain SAGB issuance increases, we have to up t-bill issuance – push FRN issuance much harder (the run rate is more like ZAR78bn for the year currently), undertake a benchmark S’ukuk every year and add in an additional USD SOAF issuance in addition to the budget-pencilled in IFI financing.
All of this still requires a step up in SAGB issuance next year from ZAR4.5bn/week from ZAR3.9bn/week this year (/currently). However, if the FRN, S’ukuk and wider push is not possible or if NT are risk averse and don’t want to use extra cash, then a further increase in SAGB issuance this year to about ZAR4.1bn/week would be required.
The practical implications here are that there are very limited options given the state of the fiscus, no additional backstops given where cash levels are and a wall of Eskom debt relief that cannot be hidden from the funding requirement and redemptions and so debt levels will have to rise faster.
Approach 2: Bottom-up
The second approach is a full ten-year forecast, Our baseline bottom-up forecast projects the fiscal position ten years into the future. This allows us to estimate how debt will evolve and whether the long-term fiscal position of the country will remain sustainable.
We project forward the main spending and revenue items for the next decade, together with the impact of rising interest rates on the overall debt stock to derive and estimate of debt-to-GDP.
Spending
It is our view that the National Treasury’s budget forecast for spending over the next three years (and by extension the next ten years) is overly optimistic. There are four main areas where we differ from their forecast: compensation, social grants, employment programmes and state-owned enterprises. Each of these is subject to political forces and the risks posed by the election, which we discuss in detail below.
Compensation
The wage agreement has come in higher than National Treasury’s projection. We add the full estimated cost of R37.4 billion4 in 2023/24 to the Treasury’s projection of R701 billion for wages. But we deduct it from elsewhere as the Treasury has indicated that it will try its best to absorb the cost. In addition, higher than expected inflation implies necessary higher than expected wage settlements in each of the years ahead.
We assume that compensation growth is rises at CPI + 1% over the next ten years. A significant risk is the need to increase headcount to deal with service delivery risks. However, as noted below, our population growth estimates are very moderate, suggesting that there is no coming population boom that will require additional staff. This is, however, not true spatially. Population growth is fastest in Gauteng and the Western Cape.
The historic experience has been that these two provinces add headcount, but that the other provinces do not reduce headcount. Moreover, the fast-growing provinces see a deterioration in key metrics (e.g., policeman per 1000, doctors per 1000). The slow-growing provinces see an improvement in aggregate service delivery measures, but at a microspatial level (e.g., at a district). Rural hospitals and schools run empty while urban hospitals get fuller.
Another way of looking at this is real spending per capita for certain services. Michael Sachs and his team at Wits have highlighted the spatial dynamics of spending. A glaring example is education (see Error! Reference source not found.). This creates a long-term spending risk as provinces such as KZN who are under resourced may demand more resources. Already there is a debate about adding “rurality” into the spending formula.
State-owned enterprises
The inability to reform state-owned enterprises poses an ongoing and significant fiscal risk. Treasury always underestimates the ability of the Department of Public Enterprises to break the budget framework (sometimes twice a year). The large state-owned enterprises, particularly Eskom and Transnet, have Cabinet-approved reform strategies but implementation has been delayed (e.g., the Eskom unbundling was first announced in 1998, and then again in 2019).
State-owned enterprise bailouts are crowding out both social and infrastructure spending, causing significant long-term economic damage (over and above the damage of unreformed enterprises). Over the last ten years, unbudgeted for state-owned enterprise bailouts have been approximately R331 billion, equivalent to the entire budget for the old age grant. If we assume no change in political party in the 2024 Election, it is likely that political deadlock will continue and “unexpected” “emergency” bailouts for usual suspects (Eskom, SAA, Denel, Post Office) will be another R100 billion over the next ten years.
The explicit costing of the Eskom debt takeover provides some certainty and reduces risk. That said, the National Treasury framework does not include provision for non-Eskom SOE bailouts, and there is an almost infinite list. The SAA equity injection has been disastrous to say the least, Denel continues to have difficulties, Post Office has indicated it may be applying for business rescue, etc.
Our debt trajectory excludes the consolidation of Eskom into gross debt. Many analysts (not least Moody’s) consolidate Eskom into the overall debt-to-GDP ratio. This is intuitively correct – as a guaranteed SOE that is in distress, the sovereign is implicitly and basically explicitly on the hook. But the IMF gross debt rules allow it to be disclosed separately. To make our figures comparable to National Treasury’s we, reluctantly, do the same.
Grants
Our baseline is for a basic income grant is that the existing social relief of distress grant will simply be converted and remain R350 per month, rising by inflation. (See our research note on the topic). The converse is true for the social relief of distress grant. The take-up remains reasonably low, and the eligible population is growing at around 1% a year. We assume that SRD/BIG numbers rise by population growth (1% a year).
The market has shown significant concern about the noise around the introduction of a more expensive basic income grant. Our view is that this is unlikely. That said, the indefinite extension of the social relief of distress grant (R350 a month for around 8 million people costing R30 billion a year is inevitable. In our alternative scenarios section below, we assume a basic income grant at the Food Poverty Line.
One risk that is not often discussed is the ageing of the South African population. 5 This will lead to increased take up of the old age grant, which is currently the largest grant costing R99.1 billion a year. Those over 60 are growing at 2.31 % a year (post-COVID), compared to overall population growth rate post COVID of 1.06 % a year. We assume pre-COVID growth rates of 3% for our forward estimates of take up of the old age grant.6 It is thus the largest and fastest growing grant component over the next ten years.
The other large grant by numbers is the Child Support Grant. 14 million beneficiaries receive the child support grant of R480 per month. We expect that the Child Support Grant will grow at the same rate as the under 19 population. This growth rate has been trending upwards over time but is still below overall population growth. (How can the overall population grow faster than the birth rate? The answer is a significant life expectancy gains, particularly as HIV/AIDS has been brought almost entirely under control. In short – there are fewer people dying than are being born, so the population is growing but ageing.)
Employment programmes
The Presidency has articulated strong support for employment programmes. This is a fiscal risk (albeit reasonably small). As there is a risk that it will be both employment programmes and a basic income grant.
Revenue
Revenue has consistently tracked GDP over the past twenty years, with the notable exception of. Absent any major shift in the mix of taxes (which we think is unlikely), The only significant risk is significant reductions to the VAT rate, either through a reduced rate or through additional exemptions, e.g., for chicken.
Overall debt to GDP projection
This leads to our overall debt projection, which is for a reasonably flat profile over the next decade, bearing in mind the significant risks.
Notably, we do not see debt beginning to decline as a share of GDP. The reasons for this flow from our analysis above – the Treasury has been reasonably successful at getting spending growth under control, but we don’t see that there will be much scope for them to reduce expenditure growth to less than 2% a year in real terms. The pressures are simply too many – the state-owned enterprises will continue to inhale cash without economic benefit, the rural provinces (all ANC controlled) will continue to demand a larger share of the budget, and the need for increased spending on core services (such as education and health) will remain. There will also need to be a slow structural shift in spending as the population ages: In grants, there will automatically be greater share of spending on the relatively more expensive old age pension, while a smaller share on child support grants; and the shift from education spending to health will be slow and complex as population ages at different rates in different provinces. Finally, this could obviously all be solved with a much more aggressive structural reform programme – energy and logistics reform will unlock higher economic growth rates, which will create a virtuous cycle of an improved fiscal position and more capacity to spend on health and education.
Footnotes
[1] The market focusses on the consolidated gross debt to GDP, which we think is the wrong one to look at. While consolidated gross debt to GDP headline number is widely quoted (not least by the Treasury), is not a particularly helpful ratio for a debt sustainability analysis. This is because cash balances are not netted out.
[2] Blanchard, Olivier. (2019). Public debt and low interest rates. American Economic Review 109(4), pp. 1197-1229.
[3] Havemann, Roy, and Hylton Hollander. (2022) Fiscal policy in times of fiscal stress: Or what to do when r> g. WIDER Working Paper No. 2022/52.
[5] Daviaud et al (2019).
[6]https://www.statssa.gov.za/publications/P0302/MidYear2022.pdf
References
Blanchard, Olivier. (2019). Public debt and low interest rates. American Economic Review 109(4), pp. 1197-1229.
Daviaud, E., Kelly, G., Cornell, J., Geffen, L., & Solanki, G. (2019). Population ageing in South Africa: trends, impact, and challenges for the health sector. South African Health Review, 2019(1), 175-182.
Havemann, R., & Kerby, E. (2021). Reigniting Economic Growth: Lessons from three centuries of data (No. 854). Economic Research Southern Africa. https://econrsa.org/publications/reigniting-economic-growth-lessons-from-three-centuries-of-data/
Havemann, R., and Hollander, H.. (2022). Fiscal policy in times of fiscal stress: Or what to do when r> g. No. 2022/52. WIDER Working Paper, 2022. https://sa-tied-archive.wider.unu.edu/article/fiscal-policy-times-fiscal-stress-or-what-do-when-r-g
Sachs, M., Ewinyu, A.K. and Shedi, O. 2022. Public Services, Government Employment and the Budget. Public Economy Project. SCIS Working paper | Number 39. https://www.wits.ac.za/media/wits-university/faculties-andschools/commerce-law-and-management/research-entities/scis/images/PEP%20Public%20services%20and%20employment%20rep ort_WP39.pdf
Disclaimer
This research report was issued by Intellidex UK Ltd.
Intellidex aims to deliver impartial and objective assessments of securities, companies or other subjects. This document is issued for information purposes only and is not an offer to purchase or sell investments or related financial instruments. Individuals should undertake their own analysis and/or seek professional advice based on their specific needs before purchasing or selling investments.
The information contained in this report is based on sources that Intellidex believes to be reliable, but Intellidex makes no representations or warranties regarding the completeness, accuracy or reliability of any information, facts, estimates, forecasts or opinions contained in this document. The information and opinions could change at any time without prior notice. Intellidex is under no obligation to inform any recipient of this document of any such changes.
Intellidex, its directors, officers, staff, agents or associates shall have no liability for any loss or damage of any nature arising from the use of this document.
Disclosure
The opinions or recommendations contained in this report represent the true views of the analyst(s) responsible for preparing the report. The analyst’s remuneration is not affected by the opinions or recommendations contained in this report, although his/her remuneration may be affected by the overall quality of their research, feedback from clients and the financial performance of Intellidex group entities.
Intellidex staff may hold positions in financial instruments or derivatives thereof which are discussed in this document. Trades by staff are subject to Intellidex’s code of conduct which can be obtained by emailing mail@intellidex.co.za.
Intellidex may have, or be seeking to have, a consulting or other professional relationship with the companies, sovereigns or individuals mentioned in this report. A copy of Intellidex’s conflicts of interest policy is available on request by emailing mail@intellidex.co.za. Relevant specific conflicts of interest will be listed here if they exist.
Intellidex is providing independent advice and independent research to a wide range of investors and financial institutions on Eskom. Intellidex’s interactions with all clients on Eskom may include business confidential information but does not include MNPI and so does not provide a conflict. Intellidex does not ‘act’ or ‘advocate’ for or ‘represent’ any of these clients. Intellidex has regular interactions with government, Eskom and other related entities connected with the Eskom situation but does not provide paid consulting services or paid advice to any of these entities. These interactions are governed by Intellidex’s own conflicts of interest policy as well as secrecy rules of the respective institutions or state-owned companies.
Copyright
© 2023. All rights reserved. This document is copyrighted to Intellidex UK Ltd.
This report is only intended for the direct recipient of this report from an Intellidex group company employee and may not be distributed in any form without prior permission. Prior written permission must be obtained before using the content of this report in other forms including for media, commercial or non-commercial benefit.
Please respect our business - this report is intended for subscribers to Intellidex Capital Markets Research, selected media and policy makers only.
Do not forward to non-subscribers without our permission.
If you are not a subscriber please contact the author to discuss the benefits of a subscription.
I hope you are right, Roy. What worries me is the the passive approach the NT takes to budgeting. Instead of trying to prioritise according to policy and efficacy considerations, it simply top-slices every department and leaves them to prioritise internally. The effect this inevitably has (after the wage increase in particular) is that the labour intensive functions of government, which are very often the public facing departments, will suffer much more from loss of capacity. Public services will deteriorate (further) and damaged organisations wont be fixed. The top-slice approach is lazy and cowardly, and means that difficult choices are not made with national priorities in mind.