Bonds, made plain

When you buy a Treasury bond you are lending money to the Government of Kenya. In return it pays you a fixed coupon twice a year and returns your face value on the maturity date. A Treasury bill works differently: there is no coupon at all. You buy it below face value and receive the full face value at maturity — the gap is your interest.

Three families matter in Kenya. FXD — fixed coupon bonds, 2 to 30 years. IFB — infrastructure bonds, which are exempt from withholding tax and therefore usually the highest net yield on the board. T-bills — 91, 182 and 364 days, auctioned weekly.

Three things sit between the headline number and your bank account.

Withholding tax. 15% on bonds shorter than ten years, 10% on ten years and longer, 0% on infrastructure bonds. Tax applies to coupon income only, never to the principal returned at maturity.

The price you pay. A coupon is a percentage of face value, not of what you paid. Buy a 16% bond at 108 and you are earning 16 shillings on 108 spent, not on 100. Yield to maturity accounts for this; the coupon alone does not.

Accrued interest. Buying between coupon dates means paying the seller for interest already earned. The quoted “clean” price excludes it; the “dirty” price you actually settle includes it.

T-bill rates are quoted as a discount rate, and it misleads in both directions at once. A 91-day bill quoted at 8.79% prices at about 97.86 per 100. You earn 2.14 on 97.86 spent over 91 days, and rolling that four times a year annualises to roughly 9.08% — higher than the quote. Then 15% withholding tax takes it down to about 7.68% — lower than the quote.

The gap narrows with tenor, and this page said the opposite until 30 July 2026. CBK's quote is already a simple annual yield on the price you pay, so the only thing the effective rate adds is compounding — and you can only compound what you can reinvest. A 91-day bill rolls four times a year and picks up about 29 basis points doing it; a 364-day bill is held once and lands within a hundredth of a point of its quote. Only the net figure is comparable to a bond's net yield.

Register at dhowcsd.centralbank.go.ke with your National ID, KRA PIN and bank details. Approval usually takes one to three business days. There is no broker and no commission — you deal with the Central Bank directly.

At the auction you choose non-competitive or competitive bidding. Non-competitive means you accept the weighted average rate the market clears at, and is available up to Ksh 50 million — this is what almost every retail investor should use. Competitive bidding means naming your own rate, and if you bid too aggressively you simply get nothing.

Bids close before the auction date — Thursday 2:00pm for bills. Once allotted, you must fund the settlement by the value date or forfeit.

Putting everything into one bond creates two problems: your entire holding is exposed to one day's rate, and the money is locked until one date that may not be when you need it. A ladder solves both by holding several bonds maturing in different years.

The real power is maturity matching. If school fees fall due in 2033, holding a bond that matures in 2033 means the principal arrives when the invoice does — no selling into a bad market, no guessing. The same logic builds a retirement income or an emergency reserve.

Government bonds carry minimal credit risk in shillings, but they are not risk-free. Rates rise, prices fall — sell early after a rate rise and you lose money. Inflation erodes fixed coupons: 12% net against 6.4% inflation is about 5.6% in real terms. Liquidity is thin for many issues, so selling quickly may mean selling cheaply.

The practical defence is simple: buy what you can hold to maturity, and match maturities to when you actually need the money.

In a competitive bid you name a rate, and the direction is easy to get backwards: a low rate is the aggressive bid — you are offering the government cheap money, and if the auction clears above you, you are simply filled; bid below where it clears and CBK may never reach you. A high rate is the generous one — likely rejected when demand is strong, because CBK fills the cheapest money first and stops when the offer is covered.

The evidence for a sensible number is public: every auction result CBK has published. Two traps in reading it. First, a re-opening keeps its original name, so a “15-year” bond may be sold with five years left — compare by time left to run, never by the label. Second, demand figures are per auction, not per bond — one auction often covers several bonds, and dividing one bond's bids by the whole auction's offer makes demand look far weaker than it is.

Or skip the question: non-competitive bidding (up to Ksh 50 million) takes the weighted average rate the auction clears at, and is the right choice for almost every retail investor. Our auction radar shows where comparable paper has been clearing — and keeps a public record of its own predictions, scored against what actually happened.

A broker's sale quote arrives as a wall of numbers: dirty price, clean price, accrued interest, consideration. The structure is simple. The clean price is the bond's market value per 100 of face value. Accrued interest is the share of the next coupon you have already earned by holding — the buyer owes it to you. Clean plus accrued is the dirty price, which is what the buyer actually pays. Then commission and levies come out of your side.

Two questions decide whether the deal is fair. First: does the quoted price actually match the quoted yield? It is arithmetic, not opinion — and one detail can move the answer by thousands of shillings: some infrastructure bonds repay part of their principal early (amortisation), and pricing one as if all principal arrives at maturity overstates its value. Second: what must a replacement earn? Charges make your realised yield slightly worse than the headline, and if you are leaving a tax-free bond for a taxable one, the replacement must gross up for withholding tax just to break even. Selling a tax-free bond to buy a taxable one at the same rate is a pay cut.

Our sale evaluator does all of this from the numbers on the quote itself — it reproduced a real broker's contract note to the cent before we shipped it.

Every figure this app has shown you until now is nominal — counted in shillings, without asking what those shillings will buy. A 13.6% coupon that leaves you 11.56% after withholding tax sounds like getting meaningfully richer. At the CPI print published beside it, you are getting richer by about half that.

Work it out by dividing, not subtracting. Real return is (1 + your return) ÷ (1 + inflation) − 1. Take a 12% net yield against 6% inflation: that is 5.66%, where subtracting gives 6.00%. Thirty-four basis points sounds like pedantry; it is not, because the error grows with both rates and it always flatters you. Round numbers here to keep the arithmetic legible — the calculator uses the current rate, whatever it is today.

Then there is the part a yield figure structurally cannot show. A bond returns a fixed nominal face value at maturity, however long it has been away. On a fifteen-year bond at 6% inflation, Ksh 100 of principal comes back worth about Ksh 42 in today's money. On a long bond that is where most of the real loss lives, and no amount of coupon arithmetic reveals it.

One consequence worth carrying: an infrastructure bond's tax exemption is worth proportionally more in real terms than in nominal ones, because the exempted amount is measured against a much smaller base once inflation has been taken off. Judging an IFB against an FXD on headline yield alone undervalues the IFB systematically.

The calculator now shows all of this for whichever bond you are looking at, with the inflation rate on a slider — because holding one month's reading constant for fifteen years is an assumption, not a forecast. Kenyan inflation was above 9% as recently as 2023. Move it and watch what survives.

A broker names a price. You have no idea whether it is generous, ordinary or poor, and Kenya publishes no secondary-market benchmark a retail buyer can check it against. This app holds no exchange prices at all — deliberately, because we hold no licence for them.

What we do hold is every auction CBK has settled. So the question can be turned around: solve the yield your quoted price implies, and set it beside what comparable paper has actually been clearing at in the primary market. That does not make a quote right or wrong — but “you are being offered 80 basis points less than the government has been paying for the same risk over the same horizon” is a fact you can act on.

Four things have to match or the comparison is noise. Gross against gross, because clearing rates are pre-tax. Tax status, because investors accept a lower gross yield on an IFB for the exemption — measured on our own archive, an IFB priced at par reads +509bps against a blended pool, which is an artefact and not an opportunity. Remaining term, never the tenor in the issue code, because re-openings keep their original code for life. And recency: the 7–12 year band medians 12.78% over the last year and 13.67% over two.

All four at once is expensive. On the shipped archive this can judge 29 of the 58 outstanding bonds; for the rest it says so rather than widening the net until it has something to say. A median of two prints dressed up as a market level would be worse than silence.