Why IMF debt sums for Kenya did not add up

Macroeconomic—and therefore political—developments in Kenya in 2024 were dominated by IMF projections for government debt.

When, in mid-2024, IMF staff projected that public debt was on a rising trajectory, they swooped on Nairobi and demanded immediate policy corrective action—culminating in short order in the 2024 Finance Bill, protests, and major political realignments.

But, it turns out as confirmed by the IMF World Economic Outlook (WEO) issued this week, those IMF debt projections made mid-year for 2024 were plain wrong—they did not add up—on a very big scale.

Accounting Identity.

To see this, note that public debt projections are based on a simple accounting identity. 

According to that identity, and absent below-the-line transactions like privatisation and exchange rate revaluation effects:  debt stock tomorrow = debt stock today, plus borrowing today, plus discrepancy.

So simple is that identity that it even largely holds in Kenyan data, with discrepancies seldom exceeding one percent of GDP, after exchange rate revaluation effects are accounted for.

Wayward IMF Debt Projections

So in the first quarter of 2024, prior to that infamous IMF mission and applying this identity, the IMF publication was projecting public debt at end-2024 to be 73 percent of GDP, unchanged from end-2023. 

That number seemed very high (to us) at the time because it implied a lot of debt appearing without passing through the budget—given the IMF’s projection at the time of government borrowing for 2024. And the bigger the revaluation effect, the bigger that implied estimate.

But we held our tongue then, thinking there may be some large below-the-line transactions in prospect of which we were then unaware, but which the authorities might have shared with IMF staff. We nevertheless stayed alert.

The plot thickened in the October 2024 WEO when, remarkably, after the additional borrowing approved by the IMF in response to the protests, the IMF projection for debt at end 2024 actually went down, a lot, from the original projection of 73 percent of GDP, to 69.9 percent.

Some of that was understandable revaluation effects given the shilling's strength after January 2024. That still left a big part out because that new number still implied that some 4 percent of GDP of debt would appear without passing through the budget. That size of discrepancy was far out of line with Kenyan norms. So now we were really suspicious.

Read: 2025: A sneak peek into economic joys and pains

Nevertheless, we were stunned to learn from the April IMF WEO of 2025 that the IMF now projects that public debt at end-2024 was just 65.6 percent of GDP, over four percentage points of GDP lower than it projected in October. None of that is revaluation effects because the Shilling exchange rate has been unchanged since then. 

So, end-2024 debt is some seven percentage points of GDP below the IMF’s original projection of just a year ago. And after allowing for exchange rate revaluation effects the identity holds in these latest IMF numbers with a discrepancy within the usual range of -1.2 percent of GDP. 

But the big news is that the 4 percent of GDP discrepancy has disappeared in the last months of 2024, and the IMF projection for debt has plummeted accordingly.

What happened?

Clearly, some of the fall in overall projected debt ratios over the whole of 2024 was due to exchange rate revaluation effects. But those are evidently negligible from mid-2024 to now.

So the only explanation apparent to us is that IMF staff projections during 2024 simply did not add up. They contained a simple but very large “add factor”—of the order of four percent of GDP—which has only now been expunged from IMF staff spreadsheets.

Had the IMF staff been aware of their error, and so had they been aware that debt was heading down sharply towards 65 percent and not to 73 percent of GDP, it is highly unlikely that their demands for policy correction in mid-2024 would have been anything like as virulent as they were. And indeed, with the lower debt number now, the IMF have lowered their medium-term primary surplus target for Kenya by ½ a percentage point of GDP.

So, absent the four percentage point of GDP error through 2024, the whole macroeconomic and political history of Kenya for 2024 could have been very different—including protesters’ lives saved.

Other Guardians

A further troubling question is why the Treasury Mandarins and economic advisors in the Presidency did not spot and correct these large IMF algebraic errors on debt in real time. Those officials, after all, were completely aware of all below-the-line transactions pending and other relevant data. And though the IMF staff may be overwhelmed and error-prone, they are not monsters; had the mathematical error been drawn to their attention, they would have corrected it. Yet this evidently did not happen until now—far too late.

Read: Address fears as MPs seek views on tax Bills

So either these officials also did not spot this critical and sustained IMF algebraic error or they had some reason—perhaps instruction from bosses—to let it slide.

And we at the IEA have lessons to learn too. In April 2024, we were too quick to give the IMF the benefit of the doubt in regard to its algebra. We promise not to do that again.

Transparency

The IMF often—and rightly—recommends accountability as a cure for many macroeconomic ills.  In that light, and in light of other IMF errors on public debt elsewhere in Africa, including recently in Mozambique and Senegal, and ongoing negotiations for a successor program for Kenya, it is incumbent on the IMF at the very least to account fully to the public for these major lapses in its projections for 2024 regarding the pivotal indicator for the program, and to explain what steps it will take to prevent their recurrence.

And the same burden falls on the Kenyan authorities to explain why they did not correct this absolutely critical IMF algebra at the time.

Kwame Owino is the CEO of the Institute of Economic Affairs and Maureen Barasa is a programme officer at IEA.