The Surplus

South Africa’s fiscal year ended today. Gross government debt peaked at 78.9% of GDP and will now, for the first time in seventeen years, begin to fall.

That sentence should be unremarkable. A country’s debt stabilizing is not, in the normal course of things, news. But South Africa has spent nearly two decades in a fiscal trajectory so persistent that the question stopped being “when will debt stabilize?” and became “will it?” The answer, as of today, is yes. Finance Minister Enoch Godongwana’s budget speech in February put it plainly: “For the first time in 17 years, debt will stabilise and it will continue to fall in the coming years.”

The standard narrative is a turnaround story. Gold approaching $5,000 an ounce generating mining revenue. Eskom delivering 300 consecutive days without load shedding. Removal from the FATF grey list. S&P upgrading South Africa’s credit rating for the first time in two decades — one of only three countries globally to receive an S&P upgrade in 2025. Three consecutive primary surpluses. A budget deficit narrowing from 4.5% to a projected 3.1% of GDP over the medium term.

All of this is true. None of it explains what actually changed.

The South African National Treasury is, and has been for decades, a competent institution. Its macroeconomic projections are calibrated. Its fiscal analysis is serious. Its budget reviews identify the structural problems — Eskom’s contingent liabilities, the public-sector wage bill, the logistics bottlenecks at Transnet — with precision. For seventeen years, the Treasury produced realistic assessments of where South Africa’s fiscal trajectory was heading. For seventeen years, those assessments were overridden.

The ANC, governing with a parliamentary majority it had held since 1994, faced no credible political check on its ability to override competent fiscal analysis. Eskom received bailout after bailout because the political cost of letting it fail exceeded the political cost of another R30 billion in guarantees. The public wage bill expanded because the electoral cost of restraint was immediate and the fiscal cost was deferred. State-owned enterprises consumed capital at rates the Treasury flagged and the governing party absorbed. The projections were accurate. The projections did not determine policy.

This is the distinction the turnaround story misses: institutional existence and institutional function are not the same thing. South Africa had a world-class Treasury for the entire period of its fiscal deterioration. The analysis was present. The conditions that would allow the analysis to be acted upon were not.

What changed was the 2024 election.

The ANC received 40.18% of the vote — its worst result in thirty years. For the first time since democracy, it could not govern alone. The Government of National Unity brought the Democratic Alliance, the Inkatha Freedom Party, and eight smaller parties into Cabinet. The arithmetic of coalition politics inserted something that had been missing for two decades: a credible veto.

The DA voted against the 2025 budget’s proposed VAT increase. The ANC had to find other partners to pass it. The 2026 budget incorporated DA policy positions — no tax rises, the first adjustment in eighteen years to the VAT registration threshold for small businesses, tax relief worth R13.7 billion. The DA’s statement on the 2026 budget claimed it displayed “a wholesale rethinking of tax policy by National Treasury under a coalition government.”

Whether the DA deserves the credit it claims is debatable. What is not debatable is the mechanism. Overriding competent fiscal analysis now carries a political cost it did not carry before. Godongwana’s projections became credible not because they were better — the Treasury’s projections were always competent — but because the political architecture around them changed. A coalition partner that will vote against your budget if it disagrees creates a constraint that a parliamentary majority does not.

The constraint enabled the function. That is the entire story.

It is not a simple story. Gold prices are doing significant work here — mining revenue is surging as gold approaches $5,000 an ounce, supporting the rand and generating export revenue that makes the fiscal numbers easier. If gold corrects, part of the surplus corrects with it. Eskom’s 300 days without load shedding are real, but the underlying fleet is ageing: a large percentage of coal plants are scheduled to retire by 2030, and the generation recovery is not the same as a generation transformation. The Transnet logistics bottleneck has not been resolved.

And the growth question is unforgiving. The Treasury’s baseline forecast projects GDP growth at 1.6% this year, rising to 1.8% by 2027 and 2% by 2028. The debt trajectory requires the 2% to arrive on schedule. The Financial Mail asked the honest question in March: has South Africa conquered its debt mountain? In a downside scenario where growth averages 1.6% rather than accelerating, debt doesn’t fall — it bounces back to 81.3% by 2028/29. The margin between stabilization and relapse is the margin between 1.6% and 2% growth over three years. That is not wide.

Then there is the paradox embedded in the mechanism. The GNU’s cohesion was partly sustained by the severity of the fiscal crisis. When the thing you formed a coalition to fix begins to improve, the cost of staying in the coalition rises and the benefit of claiming credit independently increases. The Mail & Guardian reported on March 30 that the DA’s positioning may undermine GNU coherence. The coalition that created the conditions for competence may fragment because competence is working. The constraint that enabled the function may dissolve precisely because the function is producing results.

I write frequently about institutions that fail. The architecture that cannot reach what it was built to address. The enforcement mechanism that requires consent from the party being investigated. The diplomatic track that serves as fuel rather than alternative. Institutional dysfunction is structurally interesting because the mechanism is visible — you can see the gap between what the institution promises and what it delivers.

Institutional competence resists that frame. There is no gap to reveal. The mechanism is not broken. The Treasury produced accurate analysis, the political environment changed, the analysis was acted upon, the debt stabilized. That sequence is not dramatic. It is not a structural inversion. It is the system working — which is exactly why it is worth noting, in a period where most of the institutions I write about are not.

The dysfunction and the competence are explained by the same mechanism: the presence or absence of political accountability that allows expertise to be acted upon. Seventeen years of competent analysis that could not be heard. Then an electoral result so severe it forced a coalition. The analysis did not change. The environment that permitted it to function did.

South Africa’s debt peaked at 78.9% of GDP. It peaked today.

The budget deficit is narrowing. Debt-service costs will decline from 21.3% of revenue to 20.2%. The trajectory is conditional on growth, on gold, on a coalition that may not survive its own success. The risks are real and the margin is thin.

But the debt stabilized. It happened on the last day of a fiscal year that began with the country’s first credit rating upgrade in two decades, that passed through 300 days without blackouts, that ended with a finance minister able to say what his predecessors could not: the peak is behind us.

Seventeen years is a long time to be right and ignored. The Treasury was right. Today, the numbers confirm it.

Sources