South Africa's failing infrastructure acts as an invisible tax, raising costs, constraining investment and deepening the country's fiscal crisis.
South Africa is paying an infrastructure tax that does not appear in any budget, is imposed by no law and is collected by no government department. It is paid every time freight fails to reach a port, water disappears through leaking municipal pipes or electricity bills climb faster than household incomes. National Treasury's own diagnosis, set out in the Medium Term Budget Policy Statement (MTBPS) tabled on October 21, 2026, identifies three principal risks to growth: "infrastructure bottlenecks, weak confidence and limited fiscal space." Yet these are not three disconnected problems. Infrastructure failures deepen the other two, imposing an unlegislated levy on households and businesses while draining the productive capacity of the economy. Unlike a conventional tax, this one raises no revenue, finances no public service and leaves its proceeds nowhere. It simply makes South Africa more expensive to live in, operate in and invest in.
The fiscal consequences of this infrastructure burden become clearer when placed alongside the government's growth outlook and competing expenditure commitments. The same statement cut the 2026 forecast for real gross domestic product (GDP) growth to 1.2 percent, from 1.6 percent at the February Budget. On the day, Finance Minister Enoch Godongwana told Parliament that the adjustments budget carried R823 million (about US$50 million) for the deployment of the South African National Defence Force (SANDF) in support of the police. Those two figures sit side by side in a budget that has little room for either. This analysis argues that South Africa's weak infrastructure now works as an unlegislated levy, collected through tariffs that outrun inflation and through freight and water that never reach the customer. A further share is collected indirectly through debt-service costs, as borrowing partly reflects the repeated financial support extended to state-owned infrastructure companies. Although the fiscal consolidation measures outlined in the MTBPS are credible on their own terms, they do not yet raise public investment sufficiently to shrink the levy. All U.S. dollar figures in this analysis use the Federal Reserve's H.10 noon buying rate of R16.5868 per dollar for October 23, 2026, the latest published at the time of writing, and are rounded to the precision of the rand figure.
The metaphor needs a test before it carries an argument, and the cleanest place to run one is the household electricity bill. The National Energy Regulator of South Africa (NERSA) said in its draft market inquiry report of June 2026 that Eskom tariffs rose 937 percent between 2007 and 2024, against a 155 percent rise in consumer prices over the same period. The chart below compares the two.
Adjusting the cumulative tariff increase for consumer price inflation reveals how much of the increase represents a rise in the real cost of electricity. The real tariff multiple is the ratio of the two price indexes, each set to 1 in 2007.
Here is the real electricity price in 2024 relative to 2007, is the cumulative tariff increase (937 percent, written as 9.37) and is cumulative consumer price inflation (155 percent, or 1.55). The result is an IndraStra calculation from NERSA's figures. In real terms, the price of a unit of Eskom electricity in 2024 was approximately 4.1 times its 2007 level. Part of that rise reflected the long-overdue move toward cost-reflective tariffs, since Eskom priced below cost for years. Even so, much of the cost being recovered reflects the expense of building and operating a generation fleet that consistently delivered below its intended capacity.
The rises have not stopped, though they have slowed. NERSA approved increases of 8.76 percent for Eskom's direct customers from April 1, 2026 and 9.01 percent for municipal customers from July 1, 2026. Treasury projects consumer inflation of 4.4 percent for 2026, so both increases run at roughly twice the general rate. Municipalities add their own markups before electricity reaches most homes, introducing another layer of cost between generation and consumption.
Electricity tariffs illustrate the price dimension of the infrastructure levy. Water reveals its physical dimension more directly: a substantial share of what municipalities treat and pump never reaches the billing system. In a written reply to Parliament published on July 1, 2026, the Minister of Water and Sanitation put national non-revenue water at 47.3 percent in 2023/24, against 47.4 percent a year earlier. The figure covers physical leaks, metering failures and unbilled use together. The provincial spread, charted below, runs from 28.5 percent in the Western Cape to 60.6 percent in Mpumalanga.
Chart 2: Nearly half of municipal water is lost or unbilled, and the share rose in six of nine provinces (non-revenue water as a percentage of system input, 2023/24; official statistics). Source: Department of Water and Sanitation, reply to National Assembly question NW3669, July 1, 2026. Data cut-off: October 30, 2026.
The water losses fall on paying customers in a way that resembles a tax more than a market price. A utility that cannot bill almost half its output must either raise tariffs on the half it can bill, run down maintenance or ask for transfers. The ministerial reply showed losses rising between the two years in KwaZulu-Natal, Mpumalanga, Gauteng, the Free State, North West and the Western Cape. They fell in the Eastern Cape, Limpopo and the Northern Cape. Year-on-year moves in provincial figures can reflect better measurement as well as worse networks, so the national level matters more than any single change.
Beyond electricity and water, freight transport represents a third infrastructure channel through which network inefficiencies impose costs on the wider economy. Unlike the deterioration visible in municipal water systems, Transnet's latest results point to measurable, although incomplete, operational recovery. Transnet's annual results, released on September 10, 2026, reported revenue of R88.6 billion (US$5.3 billion) and a profit of R4.6 billion (US$277 million) for the year ended March 31, 2026. According to the Freight Rail report, rail volumes rose 4.9 percent to 167.87 million tonnes, from 160.07 million tonnes a year earlier and 151.78 million tonnes in 2023/24. The chart below sets those volumes against the company's targets.
The size of that shortfall can be measured directly against the company's own plan. The target gap is the missing tonnage as a share of the target.
Here is the shortfall, is the 2025/26 target of 180.6 million tonnes and is actual volume of 167.87 million tonnes. The calculation is IndraStra's, from Transnet's figures. About 12.7 million tonnes of planned freight did not move by rail in 2025/26. Some of it was diverted to road transport, increasing logistics costs and placing additional pressure on the national road network, while some did not move at all. Container handling at the ports showed the same pattern, with 4.048 million twenty-foot equivalent units (TEU) against a target of 4.38 million, the results release said. Transnet also cut capital investment to R23.3 billion (US$1.4 billion), and the company's targets for 2026/27 stay close to the old ones, at 180 million tonnes on rail.
These operational losses would be less damaging if the state had sufficient investment capacity to modernize its networks and restore their productive efficiency. The fiscal data, however, point in the opposite direction. South Africa's total fixed investment has fallen as a share of output for a decade. The Government Technical Advisory Centre (GTAC), an agency of National Treasury, reported in March 2026, citing Statistics South Africa, that gross fixed capital formation (GFCF) fell from 17.5 percent of GDP in 2016 to 13.7 percent in 2025. The National Development Plan (NDP) target is 30 percent. The chart below splits the decline between the public and private sectors.
The public decline is the one that bears on the infrastructure levy, because the state and its companies own the networks in question. Public investment fell from 6.3 percent to 4.1 percent of GDP, a drop of about 35 percent in its share, by IndraStra calculation. Real fixed investment fell 3.9 percent in 2024 and 2.2 percent in 2025, according to the Statistics South Africa fact sheet for the fourth quarter of 2025. The GTAC deck also put unspent infrastructure budgets at about R23.3 billion (US$1.4 billion) over the three fiscal years from 2022/23 to 2024/25. Money was allocated and not used, which points to delivery capacity as well as funding.
Funding is still the binding constraint for the medium term, and the MTBPS shows why through its interest bill. Debt-service costs reach R441.5 billion (US$26.6 billion) in 2026/27 and R509.3 billion (US$30.7 billion) in 2029/30, according to Table 4.1 of the MTBPS. Payments for capital assets, the budget line closest to direct public investment, are R148.4 billion (US$8.9 billion) and R173.5 billion (US$10.5 billion) in the same years. The chart below compares the two lines year by year.
That ratio is simple to compute, and it barely moves across the forecast. The interest-to-capital-expenditure ratio divides interest by capital payments.
Here is the interest-to-capital-expenditure ratio in fiscal year , is consolidated debt-service costs and is payments for capital assets, both in R billion. Both ratios are IndraStra calculations from Table 4.1. By the end of the framework the state still spends almost R3 on interest for every R1 it spends directly on capital assets. Treasury frames the same pressure in revenue terms. Debt service takes "21 cents of every R1 of revenue," the MTBPS said. An IndraStra check gives 441.5 divided by main budget revenue of 2,136.1, or 20.7 percent, which matches. Capital payments, for their part, equal about 5.5 percent of the R2,716.9 billion that consolidated government spends in 2026/27, also by IndraStra calculation.
Comparing interest with the main service functions puts the same pressure in plainer terms. Debt-service costs in 2026/27 exceed spending on basic education, at R363.4 billion (US$21.9 billion), and on health, at R311.3 billion (US$18.8 billion), according to Table 4.2 of the MTBPS. They are more than double the R181.5 billion (US$10.9 billion) allotted to the Treasury function group for economic regulation and infrastructure. The chart below ranks the main lines.
The link between that interest bill and infrastructure runs through the state-owned companies, whose repeated rescues and debt-relief arrangements contributed to the government's broader fiscal exposure. Gross loan debt stands at R6.35 trillion (US$383 billion), or 78.2 percent of GDP, in 2026/27 and reaches R7.24 trillion (US$436 billion) in 2029/30, according to Table 3.8 of the MTBPS. The chart below traces the ratio.
Chart 7: The debt ratio peaked in 2025/26 but falls by less than two percentage points over four years (percent of GDP, end of fiscal year; 2025/26 outcome, later years estimates; official statistics). Source: National Treasury, 2026 MTBPS, Table 3.8. Data cut-off: October 30, 2026.
Annexure A of the MTBPS shows how much infrastructure risk still sits behind that debt. The state's guarantee portfolio fell from R982.4 billion (US$59.2 billion) in March 2026 to R960.6 billion (US$57.9 billion) in June 2026. Within the portfolio of guarantees drawn by state-owned companies, Eskom accounted for most of the reduction. Its exposure declined by R43.4 billion (US$2.6 billion) following debt-relief transfers and the settlement of a R38 billion (US$2.3 billion) bond. Treasury assessed the overall risk as medium, with the main residual risk in Transnet and Eskom. It said Transnet's finances remain constrained by "ageing infrastructure, maintenance backlogs, high debt and rising debt-service costs." Eskom's risk is easing, the annexure said, though municipal arrears to the utility remain a liquidity risk. That last point closes a loop. The same weak municipal finances that leave water unbilled also leave electricity bills unpaid.
The strongest counterargument to the infrastructure levy thesis is that the government has already begun changing course, and the MTBPS provides substantive evidence of this shift. The primary balance, revenue minus non-interest spending, improves from R80.8 billion in 2025/26 to R275.7 billion in 2029/30 (US$4.9 billion to US$16.6 billion), and the debt ratio has turned. Payments for capital assets grow 5.3 percent a year over the medium term, faster than interest at 5.1 percent. Spending on buildings and other capital assets grows 8.8 percent a year. Transfers to public corporations fall from R68.3 billion (US$4.1 billion) to R46.6 billion (US$2.8 billion), potentially signalling a gradual reduction in reliance on direct fiscal support to state-owned enterprises. The table below lists the main financing tools the MTBPS describes.
Instrument | Status in the 2026 MTBPS | Amount (rand) | US$ (approx.) |
|---|---|---|---|
Budget Facility for Infrastructure (BFI) | Approved across 41 projects to date | More than R100 billion | More than $6 billion |
BFI earmark in medium-term framework | Set aside for approved projects | R20 billion | $1.2 billion |
Infrastructure bonds, first issue | 10- and 15-year bonds, December 2025 | R11.8 billion | $0.7 billion |
Infrastructure bonds, planned stock by 2029/30 | Quarterly auctions since July 2026 | R76.2 billion | $4.6 billion |
Credit Guarantee Vehicle (CGV) | Incorporated August 2026; starts with the transmission grid | N/A | $553 million in conditional commitments; $1.5–2 billion in guarantees supported |
Metropolitan trading services reform | Grants over the medium term, including R8.6 billion in incentives | R27.7 billion | $1.67 billion |
The scale of those tools is where the counterargument loses force, because each is modest next to the costs it is meant to offset. More than R100 billion of BFI approvals spread over several budget cycles compares with R441.5 billion in a single year of interest. Although these figures cover different periods and types of expenditure, their relative scale illustrates the magnitude of the government's financing challenge. Treasury's own appraisal concedes the efficiency problem. The MTBPS cites an International Monetary Fund (IMF) Public Investment Management Assessment (PIMA) that found an efficiency gap of 40 percent in South Africa's public investment, against an emerging-market average of 36 percent. On that measure, a rand of public capital buys less infrastructure in South Africa than elsewhere, so more money alone does not shrink the levy. The MTBPS also lists two Transnet corridor projects under BFI consideration. The Cape Corridor project totals R18.1 billion (US$1.09 billion), of which R8.7 billion (US$522 million) is requested, and the Central Corridor project totals R13.6 billion (US$822 million), with R6.0 billion (US$362 million) requested. Both remain proposals as of the October 2026 MTBPS.
Treasury's own growth scenarios point the same way and give the counterargument its best footing. If the second phase of Operation Vulindlela, the joint Presidency and Treasury reform program, is carried out in full alongside higher investment, the MTBPS projects growth of 2.3 percent by the end of the medium term, against 2 percent in the baseline, and GDP 3.1 percent above the baseline by 2034. A scenario built on better public investment lifts average growth to 1.8 percent over the medium term, against 1.6 percent, and leaves GDP 1.6 percent higher by 2034. Those are Treasury's modeled outcomes, not forecasts. The gap between the two scenarios suggests, as this analysis reads them, that reform of how the networks are run carries much of the gain. Electricity market reform is a case in point. The MTBPS dates the launch of a competitive wholesale market to April 2027, after delays.
The defense expenditure shows how that tight budget handles pressures from outside the infrastructure agenda. The R823 million for the SANDF's support to the police came through the adjustments budget, not through new revenue. Vote 23, the Department of Defence's budget vote, rose from a main appropriation of R57.6 billion (US$3.47 billion) to an adjusted R61.7 billion (US$3.72 billion), according to Table C.11 of the MTBPS. The table below sets out the vote at budget level.
Department of Defence (Vote 23), R million | 2025/26 | 2026/27 |
|---|---|---|
Starting appropriation | 59,072 (adjusted) | 57,606 (main budget) |
Special appropriation | N/A | 280 |
In-year adjustments | N/A | 3,852 |
Adjusted appropriation | 59,072 | 61,738 (US$3.72 billion) |
Audited outcome or spending to date | 62,331 (outcome) | 30,968 (April to September) |
The 2025/26 outcome exceeded the adjusted appropriation by R3.26 billion (US$196 million), by IndraStra calculation from Table C.11. Consolidated spending on defense and state security grows only 1.6 percent a year, from R62.4 billion (US$3.76 billion) in 2026/27 to R65.5 billion (US$3.95 billion) in 2029/30, below projected inflation, according to Table 4.2. How that money divides between personnel and equipment is a separate question for a later IndraStra analysis. For the infrastructure argument, the relevant point is narrower. The contingency reserve rises to R10 billion (US$603 million) for El Niño and other risks, the minister said, and temporary fuel levy relief from November 4, 2026 is funded by returning R10 billion (US$603 million) from the Central Energy Fund to that reserve. Each in-year call on the reserve, including the defense deployment, draws on the same narrow pool from which unplanned infrastructure needs would be met.
The implications of an unlegislated levy differ according to who bears its cost, and the evidence allows them to be stated with some precision. For households and firms, the 9.01 percent municipal electricity increase now in force, combined with non-revenue water levels near 47 percent, means that utility costs will remain under pressure and could continue to rise faster than the 3.8 percent inflation Treasury projects for 2027. For Transnet, the test over the next year is whether rail volumes reach the 180 million tonne target for 2026/27, and whether the Cape and Central Corridor projects advance from BFI consideration to approval. For National Treasury, the February 2027 Budget is the next marker. It will show whether payments for capital assets remain on the 5.3 percent growth path set out in October, and whether the infrastructure bond programme expands on schedule towards R76.2 billion. For investors, the CGV's first transmission guarantees will indicate whether private capital can assume some of the network burden that the state can no longer finance alone.
Apple News, Google News, Feedly, Flipboard, and WhatsApp Channel
DISCLAIMER: This article is an independent research analysis based on publicly available government publications, official budget documents, regulatory disclosures and institutional data available as of October 30, 2026. All calculations, ratios and currency conversions identified as IndraStra calculations are derived from the cited sources and are presented for analytical purposes. Fiscal projections, infrastructure targets and economic growth scenarios reflect official estimates or stated policy objectives and should not be interpreted as guaranteed outcomes. The term "infrastructure tax" is used metaphorically to describe the indirect economic costs arising from infrastructure inefficiencies, service-delivery failures and constrained public investment. It does not refer to an actual tax, statutory levy or government revenue instrument. The analysis and interpretations presented are those of the author and do not necessarily represent the views of the institutions cited.
COVER IMAGE: The cover image is AI-generated and is used for illustrative purposes only. It does not depict actual people, places, or events.
