On Wednesday afternoon, the Minister of Finance read out a long series of numbers, and Parliament applauded.

I have presented results to boards for most of my career. Nobody has ever applauded. You are asked what you said you would do, what you actually did, and why the two are different. You do not reach next year’s plan until you have answered for last year’s.

Let me be fair before I am critical, because the easy version of this argument is wrong.

South Africa does not hide its numbers. On the Open Budget Survey, we score 82 out of 100 for budget transparency, against a global average of 45. Treasury publishes what the country earned and spent every month within thirty days of month-end, faster than many listed companies.

So the documents exist. The problem is what they measure.

Government reports its spending against a budget it has already changed. The February budget is adjusted in October, money is moved between programmes, and that later version becomes the yardstick. Measured that way, departments spent about 99% of their money last year.

Measured against what was actually announced in February, nobody publishes the answer.

That is the difference between a variance report and a comparison of the second draft to the third. A chief financial officer reports against the budget the board approved. You do not get to redraw the budget in October and then report 99% success against the version you redrew.

And 99% of what, exactly? It tells you the money left the account. It says nothing about what arrived.

Take one paragraph of that speech. R3.8 billion to hire more than 3,500 social workers. Sexual offences courts expanded from 55 to 101. R186 million for forensic and investigative capacity. R157.4 million to improve “national coordination and oversight”.

The first two are budgets. They carry a unit, so you can divide, compare, and come back in a year and check. The second two are announcements. R157.4 million for coordination answers no question, because no question was asked. Nobody said what coordination costs, how much we need, or how much we already have.

A number with no requirement attached is not a budget. It is a figure.

Then there is what happens to money that is not spent, and this part should trouble anybody who has run a cost centre.

Under Treasury’s own rules, a department that underspends mostly loses it. Savings on salaries cannot be carried forward at all. A maximum of five per cent of a goods and services budget may be rolled over. Money a department declares it will not spend is shown as a straight reduction to its allocation.

So a manager who finds a more affordable way to do something keeps five cents in the rand and a smaller baseline next year. The rational response is to spend it before March. That is not a culture problem, and it is not a character problem. It is the incentive working exactly as designed.

And we now have the receipt. The Minister announced the completion of a review of infrastructure grants:

“Between 2017/18 and 2023/24, national government transferred R887 billion through infrastructure conditional grants. The review assessed whether these grants are delivering infrastructure and improving service outcomes. The findings were clear. While most funds are spent, projects are often delayed, incomplete or fail to deliver the intended benefits.”

Read it twice. R887 billion moved to provinces and municipalities. Confirmed spent. And seven years later we are told the projects were often late, unfinished, or did not do the thing they were for.

Nobody stole it. That is what makes it harder to fix. It is a system that counts the money leaving and never checks what arrived, and it will report that as 99%.

The International Monetary Fund made the technical version of this point to Treasury two years ago. It found that no reconciliation between the budget, the statistics, and the financial accounts is published anywhere, and it asked for one. Two years later, there still is not one.

None of which requires more money, which is why I find it hopeful.

It requires one table, twice a year, built against the original: what we said in February, what we spent, what it bought, where the gap came from, and who decided. Departments already produce most of it, programme by programme, in annual reports that arrive five to six months late and that almost nobody reads. Nobody has ever consolidated them.

And it requires one rule change. Let a department keep a genuine saving. Right now, we punish the only behaviour we claim to want.

A budget tells you what a government intends. A variance report tells you what it is capable of. We publish the first with great ceremony every year, and the second arrives quietly, late, and measured against a yardstick we moved.

The Auditor-General’s office has made the same point in harder language. Reporting that clean audits now cover only 12% of what the state spends, it noted that an unqualified outcome “is not one we should celebrate” because South Africans still want to know why they do not have services.

In February, the Minister will stand up and read out a new set of numbers.

Before anybody claps, somebody should ask what happened to the last set.

 

Yusuf Bodiat is a chartered accountant, certified director and former chief financial officer. He is the founder of Pragmatic Thinking, an independent strategic advisory firm, and serves as a non-executive director on several boards. He is the author of The Bottom Line: A CFO’s Blueprint for South Africa’s Turnaround, published in February 2026.