Kenya sold assets for infrastructure. Its new fund plans to lend the money back to Treasury.

On 9 September, National Infrastructure Fund chief executive James Mworia described a machine that appeared capable of financing itself.

The fund would place its seed capital in long-term government securities yielding about 12 to 14 percent. A projected return of 12.5 percent would produce roughly KSh42 billion a year, he said. The principal would remain intact. The income, combined with private capital and project debt, would finance airports, roads, power systems, water works and other commercial infrastructure.

Six days later, the High Court removed the largest reported component of that machine.

A three-judge bench declared the Government's sale of a 15 percent Safaricom stake to Vodacom illegal, null and void. The judges quashed the approvals and agreements behind the transaction and ordered the shares restored to the State. The sale had closed on 30 June and supplied KSh204.3 billion to the infrastructure fund.

Recent disclosures put the fund's seed capital at KSh310.3 billion. On that figure, the Safaricom transaction accounts for 65.8 percent. Unless a higher court stays or overturns the orders, the fund cannot continue presenting that money as settled capital.

The judgment arrived one day after Parliament's Finance and National Planning Committee called for an extensive rewrite of the fund's investment policy. MPs found insufficient rules for project selection, risk, leverage, liquidity, returns, conflicts of interest and public reporting. The court separately found that depositing the Safaricom proceeds in the fund did not ring-fence them to identified projects with costs and locations.

Kenya has created an infrastructure investor before establishing a stable capital base, a complete investment rulebook or a public register showing which projects will receive the money. The fund's first clear strategy is to buy debt from the government that created it.

The capital base does not reconcile.

The published figures no longer support a single opening balance.

The Kenya Pipeline Company public offer raised KSh106.3 billion. The Safaricom sale was valued at KSh204.3 billion. Together they produce KSh310.6 billion, close to the KSh310.3 billion cited in recent disclosures. Parliament also approved a separate KSh40.2 billion upfront payment for future dividends on the Government's remaining Safaricom shares. Recent parliamentary reporting has referred to a KSh340 billion fund.

None of those combinations produces the same number. No opening financial statement has publicly reconciled the assets, cash received, transaction costs, dividend arrangement and current legal status of each contribution.

A 12.5 percent return on KSh310.3 billion produces KSh38.8 billion. The same return on KSh340 billion produces KSh42.5 billion. The KSh42 billion projection therefore appears to use a capital base close to KSh340 billion, while public explanations of its sources use KSh310.3 billion.

That discrepancy should be resolved in an audited opening balance, not left for the public to reverse-engineer from speeches.

The High Court order makes the reconciliation urgent. If KSh204.3 billion must be returned as the Safaricom transaction is unwound, the fund may be left principally with the Kenya Pipeline proceeds and any other lawfully transferred assets. At 12.5 percent, KSh106.3 billion would generate about KSh13.3 billion before tax, costs and any valuation changes. The distance between KSh13.3 billion and KSh42 billion changes the number, size and timing of projects the fund can support.

The safe return competes with the stated purpose.

Mworia's proposal is commercially intelligible. Kenya has a limited stock of assets it can sell. Spending every privatisation shilling once would leave the fund depleted. Preserving the principal and using recurring income can turn finite proceeds into permanent investment capacity.

The National Infrastructure Fund Act supports that structure. Section 41 limits annual expenditure and commitments to annual income plus surplus income carried forward. Section 43 permits the board, with the Cabinet Secretary's approval, to invest surplus funds in government securities.

Treasury bills auctioned on 27 August yielded 8.769 per cent for 91 days, 8.940 per cent for 182 days and 9.032 per cent for 364 days. The returns targeted by the fund sit further along the yield curve. August bond auctions cleared at rates ranging from 11.24 to 14.44 percent. The September offer included securities carrying coupons of 12.873 per cent and 12.5 per cent, with about 12.6 and 29.6 years left to maturity.

A 12.5 percent target therefore requires substantial exposure to long-dated debt rather than temporary parking in short bills. That introduces duration risk. If market rates rise and the fund needs cash before a bond matures, it may have to sell below its purchase price. If it holds to maturity, part of the capital intended for infrastructure remains committed to financing the Treasury for years.

The draft investment policy reportedly set a minimum expected equity return of 7 percent for infrastructure projects. A fund offered about 12.5 percent by the sovereign borrower will need a compelling reason to accept construction, demand, operating and political risks for a project that clears a 7 percent floor. The policy must state whether the two returns are nominal or real, gross or net, and how public benefits that do not appear in project cashflows will be valued.

Without that discipline, the easiest way for the fund to report a healthy return will be to postpone the difficult work for which it exists.

The borrower pays the fund with public money.

Government entities, semi-autonomous agencies and public funds held KSh513.31 billion in Treasury bills and bonds by 28 August, up from KSh467.78 billion in June 2025, according to Central Bank data. The Treasury plans to raise KSh987.4 billion in the domestic market during the financial year ending June 2027.

The policy behind the increase is deliberate. In May 2025, the Treasury told Parliament that selected public entities with balances at the Central Bank would invest money not immediately required in government securities on non-competitive terms. It also said a circular would require entities banking commercially to invest surplus balances directly, bypassing intermediaries.

A non-competitive investor does not name the yield. It receives the weighted average rate established by competitive bids and accepts the allocation. Direct investment can remove a bank's intermediation margin and keep interest income within the public sector. Properly matched to spending dates, it is better cash management than leaving billions dormant in accounts.

On the consolidated public balance sheet, the fund gains an asset and the Treasury gains an equal liability. Interest received by the public entity is paid by the Exchequer from taxes, new borrowing or other revenue. The country has not earned new income merely because one State institution owes another.

The transfer can still improve cash management. It cannot substitute for the economic return of a completed road, airport, power line or water system.

Public entities bidding non-competitively do not set the market rate, but their money reduces the volume the Treasury must raise from investors able to refuse, demand a higher return or move their capital elsewhere. As the captive share expands, the marginal pressure that reveals the true price of sovereign risk becomes weaker.

Lower borrowing costs are useful. A price produced with less contestable demand deserves less confidence as a measure of risk.

Short-term parking is not a permanent investment policy.

By June 2025, the Affordable Housing Fund held KSh45.48 billion in Treasury bills and had reported KSh4.2 billion in earnings. Housing Principal Secretary Charles Hinga said the Treasury-bill position had fallen to zero by August 2026 as procurement accelerated and projects absorbed the cash.

Short bills allowed the money to earn a return while remaining close to the date on which it was needed. The investment ended when the program was ready to spend.

The National Infrastructure Fund's return target points to long bonds, while its chief executive has described preserving the seed capital as a continuing principle. The bonds are not simply a waiting room for projects. They are intended to become the income engine that finances the fund's equity contributions.

That model can make the Treasury the fund's preferred customer. Government paper arrives already structured, carries no construction timetable and pays without the fund having to resolve land acquisition, tariffs, procurement, demand forecasts or political interference. Infrastructure must pass every one of those tests.

The National Infrastructure Fund was presented as a route away from debt-led development. Investing privatisation proceeds in Treasury bonds moves the debt one room down the public balance sheet. The Exchequer still pays the coupon. The fund then uses part of that coupon as equity and asks private investors to finance the project alongside it.

Kenya would have sold a productive public asset, lent the proceeds back to itself and called the interest a new source of infrastructure capital.

Parliament has found the missing guardrails.

The Finance and National Planning Committee wants project appraisals to cover construction and completion risk, operations, demand, law and regulation, environmental and social exposure, foreign exchange, interest rates and political conditions. It wants intended users' ability and willingness to pay tested rather than assumed. It has called for clearer rules on eligible sectors, project ranking, leverage, liquidity, exposure limits, monitoring, reporting and conflicts of interest.

The existing policy caps a single sector at 40 per cent of the fund and a single project at 20 per cent. Those limits control concentration after a project is admitted. They do not establish that the project is viable, that its tariff is affordable, that its procurement is defensible or that the public support behind it has been disclosed.

A requirement that projects support substantial non-recourse debt can provide another market test. Lenders whose repayment depends on a project's own cashflows will examine demand, contracts, completion and operating risks. Any government guarantee, revenue floor, availability payment or other support must then appear as a contingent public liability. Private finance should not be described as risk transfer when the taxpayer remains behind the repayment.

The court's Safaricom decision adds a prior test. Every shilling entering the fund must have a lawful source and an identified public purpose. A broad list of sectors is not a project pipeline. A deposit into the fund is not ring-fencing.

Publish the fund before marketing the return.

The fund should publish an audited opening balance showing every capital source, date of receipt, legal authority, transaction cost and restriction. The Safaricom proceeds and the KSh40.2 billion dividend arrangement should appear separately, together with the steps being taken to comply with the High Court orders and any stay or appeal.

Its investment register should show Treasury holdings by instrument, purchase date, maturity, coupon, acquisition yield and unrealized gain or loss. A liquidity schedule should match those maturities against expected project commitments. The public should be able to see whether government securities are temporary cash management or a permanent allocation.

Each proposed project should then have a named vehicle, total cost, expected equity return, user-charge model, demand study, completion date, private investors, debt terms, government support and conflict disclosures. The fund can protect commercially sensitive negotiations without concealing the public exposure.

Kenya did not create the National Infrastructure Fund to become a permanent customer for Treasury debt. Public cash awaiting a signed, viable project can earn interest. Capital cannot remain indefinitely in bonds because the coupon is comfortable and construction is hard.

An infrastructure fund earns its name when money enters a ring-fenced project with a contract, a revenue model, a completion date and accounts the public can inspect. Until then, it is financing the Government that was supposed to need less financing.

The State can lend to itself. A coupon paid from one public account to another is not a road.