AI-generated Image: Failed Convergence

The Bank of Zambia cut the policy rate by 25 basis points to 13.25 per cent on 13 May. In February, it cut 75 basis points. The deceleration from 75 to 25 is the concession the data extracted.

Between those two decisions, the April bond auction undersubscribed. The fiscal deficit exceeded two months’ budgeted allowance. The Finance Minister tabled a supplementary budget adding domestic borrowing from the securities market that had just rejected the government’s offering. The MPC eased. The environment into which it eased did not. This note maps the fiscal and financing evidence first, then assesses the monetary policy decision against it.

The Supplementary Budget

On 24 April 2026, the Government of the Republic of Zambia offered K6.3bn in bonds to the domestic market. The market bid K2.7bn, a subscription rate of 43 per cent. The government allocated K1.3bn, rejecting higher-yielding bids. Six days later, on 30 April, the Finance Minister tabled Supplementary Estimates No. 1 of 2026 in the amount of K26.3bn. The financing plan includes K7.5bn in additional net domestic borrowing from the same securities market. This K7.5bn financing figure is separate from the K7.5bn expenditure allocation under Head 21 (Loans and Investments); the first is a source of funds, the second a use.

The supplementary invites attribution to the Iran war. The composition does not support it as a direct cause. The war is cited as context in the ministerial statement: “the war in the middle east which has led to Government foregoing about US$200.0 million in 3 months, thereby affecting several macro and fiscal parameters.” That USD200m in foregone fuel tax revenue is real and sits on the revenue side. It is not an expenditure item within the K26.3bn. The supplementary expenditure allocations are domestic in origin. The war may have accelerated the tabling by compressing fiscal space through the revenue loss, but the spending items themselves were committed before the Middle East conflict began. The war determined when the supplementary arrived, not what it contained.

The 2026 budget, presented on 26 September 2025 under the IMF’s Extended Credit Facility and approved by parliament in December, targeted a fiscal deficit of 2.1 per cent of GDP and capped net new domestic borrowing at K21.6bn (2.3 per cent of GDP). The sixth and final ECF review was completed on 27 January 2026. The IMF praised reform progress at programme close. The post-programme question was whether fiscal discipline would survive once the external anchor no longer bound the calendar. The supplementary, tabled three months later, provides the answer.

The budget was made under the anchor. The supplementary was made without it. The timeline is three months.

Parliament referred the estimates to the Expanded Planning and Budgeting Committee, which expressed support but cautioned against “continued heavy reliance on borrowing and expenditure realignments.” The House approved the estimates on 11 May. The committee report has not been published. The primary estimates document has not been released for independent verification.

The FRA Drag on Primary Balance

In September 2025, the President directed FRA to purchase all available maize. The planned purchase was 543,000 MT. FRA purchased 1,667,921 MT, valued at K11.3bn against an approved budget of K2.4bn. Treasury funded K3bn directly. A K5bn commercial loan covered the next tranche. The residual unfunded arrear stood at approximately K3.3bn. FRA’s secure storage capacity is approximately 1.5m MT. It purchased 1.7m MT. The excess sits in open-air storage.

The supplementary confirms the consequence. K7.4bn to Agriculture is the largest single allocation, covering FRA replenishment and FISP arrears. The Expanded Planning and Budgeting Committee expressed concern over “continued accumulation of arrears owed to agro-dealers and financing challenges facing the FRA despite increased allocations.”

The K5bn loan converted a social commitment into interest-bearing sovereign debt. New purchases from a record 2025/26 harvest (exceeding 3.87m tonnes) will compound the position. FRA has announced it will purchase a minimum of 500,000 MT for the 2026/27 season from May to October 2026. The 2026/27 purchase price has not been gazetted. Even at the 2025/26 price of K340/bag, the cost of buying only the gap to the 2.5m MT strategic reserve target is K5.4bn against a K2.1bn allocation. In an election year, the political incentive to buy more is identical to the pressure that produced the September 2025 directive.

FRA cannot offload current stock without loss.

The government has never run a fiscal exposure of this magnitude through FRA. The combined pressure of K5bn loan servicing, the K3.3bn unfunded arrear, and new purchases above the K2.1bn allocation represents a drag on the primary balance measured in the low billions annually.

The conditional escape depends on climate. If El Nino materialises alongside sustained disruption to global fertiliser and agrochemical supply, FRA’s stock shifts from liability to strategic asset (Tindale, 2026). China’s restriction on sulphuric acid exports from May 2026 appears politically durable (Duesterberg and Aibel, Hudson Institute). Without these conditions, the stock deteriorates or is sold below cost.

FRA does not publish audited financial statements. These numbers are reconstructed from ministerial statements, parliamentary records, and budget documents.

What These Numbers Measure

Fiscal Balance = Revenue and Grants − Expenditure (excluding amortisation)

When the answer is negative, the government must finance the gap.

Primary Balance = Revenue and Grants − Expenditure (excluding amortisation and interest payments)

The primary balance reveals whether the government can cover current operations from current revenue before the cost of past debt. The IMF tracked this number as its core measure of fiscal sustainability throughout the programme. A primary deficit means the government cannot fund even its operational spending from revenue.

The Secretary to the Treasury presented Q1 fiscal data to stakeholders in April. Revenue, expenditure, and financing appeared as three resource categories. He did not subtract one from the other. Neither the fiscal balance nor the primary balance appeared. The single most important number in the presentation was absent.

Fiscal Performance Through Q1

The fiscal deficit consumed 40 per cent of the annual target in two months. Q1 financing of K17bn implies approximately 88 per cent was consumed in one quarter. Both figures predate the war or were materially determined before it could explain them.

January ran a fiscal deficit of K8.5bn, consuming more than a quarter of the full-year allowance in a single month. Expenditure reached 21 per cent above target. FRA funding overshot its monthly target by 404 per cent. FISP disbursements ran 39 per cent above plan. The primary balance swung from a targeted surplus of approximately K3.9bn to a deficit of K1.6bn, a K5.5bn miss.

February reversed the expenditure pattern. Spending pulled back 23 per cent below target. Revenue deteriorated. VAT collected 42 per cent below plan. Domestic VAT reached K198m against a target of K897m, 78 per cent below. Grants came in at K30m against a full-year budget of K12.1bn. The full expenditure breakdown for February was not published.

Through two months, the cumulative fiscal deficit stood at K7.8bn against a full-year budget of K19.4bn. That is 40 per cent of the annual target consumed in 17 per cent of the year. January fully precedes the war. February’s fiscal outturn was materially determined before the 28 February shock could explain it.

The Q1 aggregates confirmed the trajectory. Tax revenue ran 4.8 per cent below target. Grants fell 67 per cent short (Secretary to the Treasury’s Q1 presentation). Transfers, where FRA and FISP sit, reached K15.0bn against K7.3bn projected, 104 per cent above. Total financing reached K17.0bn against K7.9bn planned. The government borrowed more than double what it had projected. The ZRA Commissioner separately confirmed that Q1 revenues fell short by K1.7bn. The Secretary to the Treasury’s own presentation identifies two Hormuz transmission channels for Zambia: fuel import dependency and imported fertiliser inputs.

The Q1 financing figure independently confirms the scale of the deficit. In standard fiscal accounting, the deficit approximately equals the financing requirement. Q1 financing of K17.0bn against a full-year target of K19.4bn implies that approximately 88 per cent of the annual deficit was consumed in one quarter. The Secretary to the Treasury’s expenditure figure (K59.77bn) may differ from the MEI definition, and the presentation did not separate interest from amortisation, so the fiscal balance cannot be computed precisely from Q1 data alone. But the financing line is unambiguous. The government financed K17bn in one quarter against a plan of K7.9bn.

Moody’s Ratings projected in April that the deficit could reach 4 per cent of GDP, nearly double the 2.1 per cent target. The Finance Minister said “that may be slightly on the high side.” The supplementary budget, tabled the same week, suggests otherwise.

The revenue base itself may be weaker than the budget assumed. Preliminary estimates show the economy grew 3.8 per cent in 2025, unchanged from 2024 and below the trajectory the 2026 budget was built on. Copper production fell 12.5 per cent from January to 63,253 tonnes in February (Ministry of Mines). The ZamStats trade data measures a separate metric, exports rather than mine output, and shows a parallel decline: refined copper export volumes fell 11.2 per cent from 77,200 tonnes in February to 68,600 tonnes in March, with export earnings down 10.5 per cent from K19.4bn to K17.3bn as LME copper prices dropped 3.6 per cent to USD12,499 per tonne. Export volumes exceed production because Zambian smelters process DRC copper concentrate alongside domestic ore. Cumulative Q1 2026 export volumes remain 6 per cent above Q1 2025, but the monthly direction is downward. Mineral royalties, which are levied on Zambian mine output, held above their February target on price strength, but declining production at any price level compresses the royalty base.

The Grants Hole

The grants line is close to zero. Through Q1, grants reached K789m against K2.4bn projected (Secretary to the Treasury’s Q1 presentation). Against the full-year budget of K12.1bn, the run-rate is concerning.

Part of this connects to the stalling of Zambia’s health aid arrangement with the United States. PEPFAR provided approximately USD367m annually, roughly 60 per cent of total US development support and funding over 60 per cent of the national HIV programme. The US shifted to bilateral memoranda of understanding requiring data-sharing provisions and, in some cases, minerals access. By December 2025, fourteen African countries had signed: Kenya, Rwanda, Uganda, Nigeria, Mozambique, Ethiopia, Botswana, and seven others. Several agreements have been challenged in court. Zambia has not signed. Zimbabwe rejected the terms on data sovereignty grounds. Ghana rejected similar terms. South Africa’s bilateral relationship collapsed over separate tensions. Tanzania and the DRC have not signed. The Foreign Minister called the proposed terms “unconscionable.” The deadline passed without resolution.

The 2026 health budget is K26.2bn. The budget speech acknowledged the risk of losing external health support: the K6.4bn allocation for drugs and medical supplies "represents an increase of 30.0 percent from 2025" and was explicitly intended "to cover the gap following the withdraw of some external support" (paragraph 168). Whether other health budget lines were also adjusted in anticipation is not disclosed. The budget projected K12.1bn in total grants from cooperating partners. Whether that figure assumed continuity of PEPFAR funding is also not disclosed. PEPFAR at approximately K6.9bn annually (USD367m at the current rate of ZMW18.86) represents more than half the grants target. A joint ZIPAR and United Nations analysis of the 2026 budget identified a K21bn financing gap to meet international health commitments (2 October 2025). Through Q1, grants reached K789m against K2.4bn projected. The gap between the budget's assumptions and the grants outturn is widening regardless of which assumptions the budget made.

This is one side of the structural constraint we mapped in “The Forced Choice” (February 2026). The United States conditions health aid on minerals access. China, Zambia’s largest bilateral creditor at over USD4bn, conditions engagement on political alignment. The RightsCon cancellation in May illustrated the other side. The IMF programme, which imposed conditions but also provided resources and diplomatic cover, is no longer in place. The grants line collapsing is the fiscal expression of a closing external support architecture. We covered the political economy of this constraint in “The Forced Choice” at canarycompass.com.

With the fiscal hole widening and external grants stalling, the burden of funding the state falls entirely on the domestic securities market.

Financing Performance Through April

The domestic borrowing programme originally targeted K106bn in gross securities issuance for 2026, including the refinancing of maturing debt. Of that total, K21.6bn represented net new domestic financing. The supplementary adds K7.5bn in additional net domestic financing, raising the net domestic borrowing requirement to K29.1bn and the effective gross issuance target to approximately K113.5bn.

The bond auction tells the story. In January, bids reached K10.1bn against K4.2bn offered. In February, K21.3bn. Offshore capital took 49 and 69 per cent of allocations. By April, bids fell to K2.7bn against K6.3bn. The 7-year drew K281m against K1.6bn. Rejected bids at the long end carried yields up to 100 per cent. Bond demand fell 87 per cent from the February peak.

T-bills are operating inside a corridor trap. Three of four tenors offer insufficient compensation relative to the overnight interbank deposit floor. The 30 April auction showed the pressure: the 182-day was rejected at 15.3 per cent (accepted at 11.5), and the 364-day at 14.0 (accepted up to 12.7). Investors demanded higher yields. The government, through its agent, the Bank of Zambia, rejected them.

April did not reduce the burden. The supplementary added more than April delivered. The required monthly run-rate has risen from K8.42bn to K9.70bn despite a month of issuance having passed. May has no bond auction. Even if the T-bill auctions in May and June each deliver K2.2bn in net financing (requiring subscription above 100 per cent because T-bills maturing within the year offset gross issuance), the remaining balance entering the June bond auction rises further. The monthly run-rate climbs above K10.7bn. The June bond auction, offered at K6.3bn, would need to deliver approximately K10.6bn in net issuance to break even on a single month's requirement.

Unless June produces a significant financing month, issuance sizes in the second half will need to increase. The current structure of K6.3bn per bond auction and K2.2bn fortnightly in T-bills cannot reasonably be expected to sustain the required monthly run-rate. Unless there is a significant change in the global environment, Zambia may have to let yields rise to attract the participation the programme requires.

June: The Corrected Maturity Wall

Our earlier estimate of K8.3bn in June bond maturities, sourced from Reuters, included credit-linked note wrappers. A credit-linked note gives offshore investors exposure to the same domestic bond through an international clearing structure; it is not additional debt. These CLNs carry XS ISINs and are classified as Eurobonds in the Reuters database, but they are derivative representations of domestic ZM-ISIN bonds structured for international clearing through Euroclear. The Reuters data flags them with a “DR” (duplicated) marker. The outstanding amounts on the domestic and international ISINs are identical. Stripping these wrappers corrects the maturity profile.

June carries the largest maturity concentration in the remaining calendar. Bond maturities of K4.7bn and bill maturities of K6.3bn will provide recycled liquidity, but the amount is smaller than the K14.6bn headline we cited. Whether investors roll at maturity or exit depends on global risk appetite, yield compensation, and confidence in the fiscal trajectory.

A weak June auction would not only widen the immediate financing gap but compromise Q3 and Q4 even if conditions subsequently improve, because the catch-up burden compounds forward. If neither robust bidding interest nor rollovers materialise through June, the required yield concession to clear the market widens.

Some regional data suggests the April deterioration was systemic. South Africa saw record bond outflows in March. Kenya’s switch auction collapsed at 87 per cent in April, with the CBK Governor attributing it directly to geopolitical uncertainty. The CBK ran three switch auctions in 2026. The January and March switches oversubscribed. By the third, on 13 April, six weeks into the war, demand collapsed to KSh2.56bn against a KSh20bn target.

But the systemic explanation has an expiry date. Even if the global environment normalises, concerns are shifting onto domestic fiscal performance. Moody’s projection of a 4 per cent deficit did not require the war. It required only the domestic fiscal performance that the MEIs have confirmed. The war amplified what the structure was already producing. Offshore demand may have been front-loaded into January and February. The maturity calendar between March and April offered limited recycled cash (no bond auction is scheduled for May, which also carries limited maturities). Those factors would have thinned demand regardless. If the global environment stabilises and offshore still does not return at scale, the fiscal trajectory becomes the larger explanation.

BoZ committed to introducing Liability Management Operations in Q2. No modalities have been published as of mid-May. May, which carries no bond auction, could be the window. Kenya’s experience suggests the current environment is not favourable for switch operations.

The Offshore Dependency

That positioning was earned. The copper story, the reform trajectory, and the debt restructuring completion attracted capital into a carry trade at yields of 16-17 per cent with a strengthening kwacha. Offshore liked the administration. Investors who had met the President were encouraged by his ambitions for the country. The macro direction was credible.

In three months, non-residents absorbed K12.5bn of the K16.5bn in net new securities issued. The domestic market contributed K4.0bn. The Bank of Zambia had raised the non-resident participation cap from 5 per cent to 23 per cent in January. Q1 2026 was the most offshore-dependent quarter of domestic financing in Zambia’s post-restructuring history.

The total 2026 cash flow obligation to non-residents on the existing portfolio at end-December stood at K24.6bn (Debt Statistical Bulletin Q4, Table 27): face value maturities of K10.5bn, discount payments of K4.5bn, and coupon payments of K9.6bn. Net additions in Q1 (K12.5bn in face value) do not retire the annual obligation. Total cash owed to non-residents on the portfolio exceeds what any single quarter's inflow contributed. When the April bond auction collapsed, servicing this obligation became a financing question yet again.

The kwacha’s Q1 appreciation rested on mining sector FX supply of USD915.7m, up from USD759.4m in Q4 2025. Net sales by mining companies reached USD626.0m. Gross international reserves reached a historic high of USD6.5bn in February before closing Q1 at USD6.2bn. The March decline included USD114.7m in government payments for fuel procurement. Outstanding foreign exchange demand orders built to USD33.1m at end-March from zero at end-December. At USD6.2bn, reserves provide a substantial buffer against near-term currency pressure. The kwacha risk is not immediate. It is conditional: if copper volumes continue to decline and fuel imports continue to draw on reserves, the FX supply arithmetic shifts over quarters, not weeks.

The January and February bond auctions proved the confidence was real. Bids of K10.1bn and K21.3bn, with offshore taking 49 and 69 per cent of allocations, showed that investors priced the reform trajectory, the debt restructuring completion, and the IMF-anchored fiscal path. That confidence carried Zambia through the immediate post-programme period. The question still remains whether it survives without the programme. An IMF anchor does not prevent fiscal shocks. It provides the institutional framework through which shocks are absorbed without breaking market confidence. When a country with an active programme faces an oil shock, investors expect the programme conditions to constrain the fiscal response. When the programme has ended and the supplementary arrives three months later, investors must assess fiscal discipline on the government's own record. The April auction was the first assessment. June will be the second.

Even in the best case where the global environment normalises and offshore sentiment recovers, the Zambia they return to is not the Zambia they bought into in February. They priced a 2.1 per cent deficit under an IMF anchor with domestic borrowing capped at K21.6bn. They now face a widening deficit, domestic borrowing at K29.1bn, a grants architecture that has not recovered, an FRA exposure measured in billions, and no programme. The same yield that was attractive in February may not compensate for the deteriorated fiscal position. Offshore may return, but at a higher yield threshold. The government may get the participation it needs, but only by conceding on price. A stable currency makes this possible. It removes FX risk and preserves the carry on exit. But it requires the yield leg to adjust upward. If the election resolves without disruption and yields adjust into a stable currency, offshore participation could recover in Q3. If the conflict de-escalates, oil falls, and June clears strongly, Zambia gets a window. The structural items (FRA, wages, arrears) would persist, but the systemic alibi would expire faster, shifting the test from shock management to domestic fiscal credibility.

At K113.5bn in effective annual issuance, the market depends structurally on offshore participation. Non-residents held 28.9 per cent of total securities and absorbed 75.8 per cent of Q1 net issuance. That participation is not marginal. It is the financing. Each incremental percentage point is harder to attract. If offshore saturates, two paths remain: fiscal consolidation reduces the borrowing requirement, or the domestic market steps up. Several 2025 bills aimed at improving credit access would redirect bank balance sheets toward private-sector lending. That is good policy in normal times. In a year when the government needs domestic banks to strongly absorb a portion of K113.5bn in securities issuance, those bills create tension. The domestic bid for government paper could thin at the same time the borrowing requirement has not shrunk.

The Finance Minister told Bloomberg in April that the local bond market “has been very useful” and that there is “scope for more” yield reduction, “not based on force, but based on what investors can see and judge.” The April auction subscribed at 43 per cent.

Election-year fiscal loosening is standard frontier-market behaviour. The issue is whether this loosening was priced in. The budget signalled discipline. The supplementary reversed it. Offshore will extend latitude in an election year. A deficit consuming nearly half the annual allowance in two months, before the war, is not latitude. If the systemic environment normalises and the fiscal data does not improve, the country-specific explanation takes over.

It is against this backdrop of fiscal expansion and evaporating market demand that the central bank’s rate decision must be measured.

The Monetary Policy Decision

The Committee cited a favourable inflation outlook, the expected maize harvest, exchange rate stability, and upside risks from the Middle East conflict. Inflation declined from 11.2 per cent in December 2025 to 6.8 per cent in April, inside the 6-8 per cent target band. The forecast projects it remains within the band through 2028, averaging 6.8 per cent in 2026 and moderating to 6.1 per cent in 2027. The drivers cited are exchange rate appreciation and anticipated lower maize prices.

Zambia’s inflation outturn is unusually benign relative to the scale of the fuel shock. The reason is not monetary policy. It is a fiscal subsidy. The MPC statement itself acknowledges this: fuel prices “could, actually, have been higher than they currently are had it not been for the tax relief that the Government has provided, notably the suspension of excise duty and zero-rating of value added tax on petroleum products for three months.” The Committee identified the mechanism, noted that it suppresses measured inflation, and cut anyway. Inflation protected by revenue foregone is not inflation conquered by structural adjustment. The maize harvest mechanically suppresses food prices, the largest weight in the basket, but the FRA is buying the surplus at K340/bag, converting the CPI benefit into a fiscal liability measured in billions.

The ERB price build-up makes the subsidy measurable. The counterfactual pump price follows a simple formula:

P_counterfactual = (P_actual + Excise_suspended) × (1 + VAT_rate)

The table shows three stories. Petrol carried excise of K2.34 per litre and VAT at 16 per cent through March. In April, both were suspended. Without the suspension, petrol would cost K34.21. It costs K27.15. The government is absorbing K7.06 per litre. Diesel carried excise of K0.75 and VAT at 16 per cent. Without the suspension, May diesel would cost K40.30. It costs K33.99. The subsidy is K6.31 per litre, up from K5.63 in April as international diesel prices rose 23 per cent in a single ERB review cycle. Kerosene, which has never carried excise or VAT within the current pricing structure, shows what full pass-through looks like: K21.06 in March, K32.26 in April, K35.05 in May. The government is purchasing the 6.8 per cent inflation reading at K6.31 per litre of diesel and K7.06 per litre of petrol.

The April CPI, published by ZamStats the week before the MPC meeting, already showed the strain. Monthly non-food inflation (month-on-month) jumped from 0.4 per cent in March to 1.3 per cent in April. The driver was fuel: diesel at the pump had risen 28 per cent between the March and April ERB reviews (K23.25 to K29.78), even with the suspension in place. Transport inflation (year-on-year) swung from minus 0.3 per cent in March to plus 1.3 per cent in April. The May ERB review adds another 14 per cent to diesel. The pipeline feeding into the May and June CPI readings is rising, not stable.

What the statement did not address is more revealing than what it did. The word “fiscal” does not appear. There is no mention of the supplementary budget tabled twelve days before the meeting. No mention of domestic borrowing rising 35 per cent. No mention of the April bond auction subscribing at 43 per cent. No mention of grants at 67 per cent below target. The entire fiscal and financing convergence documented in this note is absent from the central bank’s assessment.

The forecast improved while international fuel prices roughly doubled. The BoZ cites exchange rate appreciation and lower maize prices as the drivers. Both are real. The forecast improvement depends heavily on the assumption that food and exchange rate disinflation outweigh fuel price pass-through. If the tax suspension expires and oil prices remain elevated, that assumption faces its hardest test in the final quarter. The statement does not disclose how much of the improved path depends on the temporary fiscal tax relief it acknowledges elsewhere.

The three-month suspension window expires on 30 June. It is unlikely to. No government removes a fuel subsidy two months before a general election, although a partial restoration remains possible if oil prices fall. The suspension will run through August at minimum, and likely through September or beyond. The September MPC will therefore assess inflation produced entirely under fiscal suppression. The Committee will have no unsuppressed data to evaluate. If the suspension expires at end September, full pass-through hits the October pump price. October is the last month of the lower seasonal inflation cycle (May to October: harvest effects, lower food prices month-on-month). November begins the planting season, when food inflation rises structurally. The fuel shock and the seasonal turn would compound in the same quarter. The Committee meets 28-29 September with subsidised data, holds or adjusts the rate, and then faces the compounding shock without a scheduled decision point. The 25 basis point cut was not cautious. It was a forecast built on the assumption that someone else keeps writing the cheque.

The BoZ’s mandate is price stability, not fiscal financing. The Committee would argue inflation is within target and real rates remain positive. But the inflation reading the mandate is being measured against depends on a fiscal subsidy the central bank does not control. The statement identifies the subsidy but does not model what happens when it expires.

The May decision collided with the fiscal and financing pressures documented above. The stable exchange rate, at ZMW18.86, should be the foundation for attracting offshore participation back into the bond market. A stable currency removes FX risk and preserves the carry on exit. But the return requires both legs: currency and yield. The MPC moved the yield leg in the wrong direction.

The 30 April T-bill auction showed bidders demanding 15.3 per cent on the 182-day and 14.0 per cent on the 364-day. The Bank of Zambia rejected those bids. At 13.25 per cent, the policy rate pushes the overnight floor lower, attempting to close the corridor trap from above. The auction data suggests the market wants to move in the other direction. Bonds are a different market where yields are set by auction clearing rates, fiscal trajectory, and global risk appetite. But the signal matters. A central bank that eases while the fiscal position is deteriorating tells the bond market that it trusts its own inflation forecast more than the fiscal data warrants. If that forecast depends on a temporary subsidy, the market prices accordingly. In an environment where the government needs K9.70bn per month from the domestic market, the disconnect between the policy rate and the market's required yield carries a cost.

The next MPC meeting is 28-29 September. The Committee has locked itself into 13.25 per cent through the June bond auction, the fiscal data for March through June, the supplementary implementation, and the August general election. The most consequential fiscal and financing period of the year will unfold without a scheduled rate decision.

Close

Seven stars needed to align. That was the assessment in January, when the refinancing wall was mapped and the buffers were thin. The arithmetic left no margin for shocks on either the fiscal or the financing side simultaneously.

The fiscal deficit reached 40 per cent of the annual target in two months. Q1 financing of K17bn implies approximately 88 per cent was consumed in one quarter. The FRA liability of K11.3bn followed a September 2025 directive. The wage increment was negotiated before the war. The constituencies were created by constitutional amendment. All of this preceded the Middle East conflict. Of K16.3bn in net new supplementary expenditure, no major allocation is directly identifiable as war expenditure. The war’s direct fiscal cost appears on the revenue side through foregone fuel taxes. Its indirect cost appears through financing pressure and reserve drawdowns. When the suspension is extended through the August election, as it almost certainly will be, that revenue loss roughly doubles.

The government now plans to raise the supplementary addition from a securities market that subscribed at 43 per cent six days before the supplementary was tabled. The required monthly run-rate has risen to K9.70bn, 15.2 per cent higher than at end-Q1, despite a month of issuance having passed. The Secretary to the Treasury presented Q1 data without computing the fiscal balance. The grants line sits well below plan while the PEPFAR arrangement remains unresolved. The budget speech acknowledged the withdrawal of external health support nine months ago. The FRA carries billions in exposure outside headline fiscal reporting.

The Secretary to the Treasury’s own conclusion in April was that “Zambia has reached a decisive moment where recent stabilisation gains must be carefully preserved and strengthened.” The Q1 data he presented in the same briefing, and the supplementary budget tabled the same week, suggest those gains are not being preserved. They are being spent.

The data in this note was available to every analyst covering Zambia. The MEIs are published. The auction results are published. The budget speech is published. The supplementary statement is published. The local analytical ecosystem did not produce the stress-test before the supplementary arrived.

Zambia’s fiscal challenges are compounded by an analytical environment where commentary calibrated to access rather than accuracy disables the early warning function that independent analysis is supposed to perform. Each of these spending decisions looked costless in isolation. The maize purchase was a political win for farmers. The wage increment was a win for civil servants. The constituencies were democratic expansion. But they accumulate. They compound. They sit on the balance sheet as arrears and commercial loans until a supplementary arrives to regularise them. When the accumulated cost crystallises, it transmits to the taxpayer through debt monetisation that puts a floor under inflation. Or it transmits through fiscal adjustment that cuts the services the spending was supposed to fund. This is not a new cycle. It is the cycle that required IMF intervention in the first place.

The fiscal position is slipping while financing is tightening. Yields may have to move higher to compensate. That big February statement may yet turn into a concession.

For the full analytical framework, see The 2026 Refinancing Wall, Domestic Market Absorption, Copper Output and the 2026 Royalty Arithmetic, Growth Without Diffusion, A Little Here a Little There (fuel subsidy counterfactual methodology), The Acid Test, The Forecast Is Not the Evidence, and The Forced Choice at canarycompass.com.

Sources: Bank of Zambia MPC Statement (13 May 2026, meeting 11-12 May), Governor’s Media Presentation (13 May 2026), Monetary Policy Report (February 2026), and auction results (January, February, April 2026). Ministry of Finance Monthly Economic Indicators (January, February 2026). Ministry of Finance Quarterly Debt Statistical Bulletin (Q3 2025, Q4 2025). Secretary to the Treasury Q1 presentation (April 2026). Musokotwane ministerial statement to Parliament (20 February 2026). Supplementary Estimates No. 1 of 2026 (ministerial statement, 30 April 2026; approved 11 May 2026). 2026 Budget Speech (26 September 2025). ZRA Q1 revenue confirmation. ZamStats, The Monthly, Volume 277 (April 2026). Energy Regulation Board, Review of Petroleum Pump Prices (April and May 2026). Reuters maturity profile (corrected for CLN wrappers). Finance Minister Situmbeko Musokotwane (Bloomberg, April 2026; Parliament, 30 April 2026). Foreign Minister Mulambo Haimbe (May 2026). Moody’s Ratings (April 2026). Civil Society for Poverty Reduction. Expanded Planning and Budgeting Committee. Agriculture Minister Reuben Mtolo (January 2026). Tindale (2026). Duesterberg and Aibel, Hudson Institute (2026). CBK Governor Kamau Thugge (Business Daily, April 2026). UNDP/ZIPAR health financing analysis. Canary Compass, “A Little Here, a Little There” (1 April 2026).

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu or X @InfinitelyDean.

AI-generated Image: Same profile. Different verdict.

The first thing I remember about football is crying.

Not mine. Theirs. It was July 1990 and I was six years old, sitting in a living room in Nairobi, not really understanding what was happening on the screen. Italy were playing Argentina in the semi-final of the World Cup, at the Stadio San Paolo in Naples. I did not know then that Naples was Diego Maradona’s city. That the stadium belonged to his club. That the Neapolitan crowd had been torn all evening between their country and their god. I did not understand any of that. I just saw the Italian players weeping at the end, and Maradona smiling, and I knew something terrible had happened to the team I had chosen without knowing why.

My dad supported the Germans. So did every older person I knew. The German machine. I think it was because Bundesliga matches used to come on terrestrial television in Kenya back then, alongside Serie A, and that was the entertainment long before the Premier League launched and took over everything. But I did not want the Germans. I wanted the Italians. And there was a player with a ponytail who had come off the bench in that semi-final, came close to winning it, scored his penalty in the shootout, and still lost. His name was Roberto Baggio.

Four years later, the World Cup was in the United States for the first time. I was ten. Baggio was no longer the substitute. He was il Divin Codino, the Divine Ponytail, and he had carried Italy to the final almost single-handedly. The equaliser against Nigeria in the final minutes, when Italy were seconds from elimination, then the winning penalty in extra time. The winner against Spain. Both goals against Bulgaria in the semi-final. He was the player I wanted to be, in the way that only a ten-year-old can want to be someone.

The final was on a Sunday. Brazil against Italy, at the Rose Bowl in Pasadena. In Nairobi, kick-off was half past ten at night. My parents would not let me stay up on a school night. I went to bed knowing the match was happening without me.

I woke up on Monday morning and ran to my parents’ room. Who won? Brazil. I found out later that day how. The match had finished goalless after extra time. It went to penalties. Baresi missed for Italy. Brazil missed too. But by the time the count reached 3-2 to Brazil, with Massaro saved and Dunga converted, there was one kick left for each side. My favourite player, the man who had carried Italy to that final, stepped up to take the fifth penalty. If he scored, Brazil’s fifth taker still had the chance to win it. But Baggio never gave him the opportunity. He put it over the bar. It was over. Brazil were world champions and I was heartbroken before breakfast.

I tell you this because it explains what comes next. But not in the way you might expect.

I did not become a Manchester United fan because of defensive football. I became one in 1993, during the first Premier League season, because of a Frenchman with raised collars who walked with his chest puffed out like he was a king. Eric Cantona had nothing to do with catenaccio. He was swagger, audacity, and the understanding that football could be theatre. That is a story for another day. The point is that my two footballing allegiances, Italy and United, came from completely different places. Italy gave me the tears and the tradition. United gave me the collars and the arrogance. One was built on structure. The other was built on personality.

I did not know these two traditions would meet in a statistics table 30 years later.

The season that should trouble every United fan who mocks Arsenal was not the season they remember with warmth. Not Ronaldo’s 31-league-goal year. Not the Treble. It was 2008/09. The season Edwin van der Sar went 1,311 minutes without conceding a goal. Fourteen consecutive clean sheets. Sixty-eight goals scored in 38 games. Just 1.79 per game. Consider what that means. The year before, Ronaldo alone had scored 42 across all competitions. Then United added Berbatov to a front line that already had Ronaldo, Rooney, and Tevez. Four world-class attackers. And the team scored fewer league goals than the season before, not more. The lowest-scoring campaign in a seven-year window. And 90 points. Only the 1999/00 squad, scoring 97 goals, ever collected more in the 38-game era. United’s second-highest points total came from their lowest-scoring season. That is not a coincidence. It is a statement about what defensive solidity can produce, even when the attack has every reason to produce more.

My attacking club’s second-best ever points haul came from a season where we added a world-class striker and scored fewer goals. I just never looked at the numbers until this week.

Now look at Arsenal in 2025/26. Five points clear at the top of the Premier League with three games to play. A Champions League final in Budapest at the end of the month. And somehow, the most criticised side in English football.

Both campaigns produce under two goals per game. This is not because Arsenal sit deep and defend. They average between 55 and 58 per cent possession this season depending on the source, they dominate territory, they keep the ball. But when it comes to converting that possession into goals, the output is remarkably low for a title-leading side. For reference, Manchester City’s centurion season in 2017/18 produced 106 goals, or 2.79 per game. Liverpool’s title in 2019/20 hit 2.24. United’s own 1999/00 squad managed 2.55. By those standards, both the 08/09 United and this Arsenal side produce elite defensive profiles with attacking output that looks modest by champion standards. Arsenal can be attractive. They have the players for it. But not often enough, and not consistently enough, for the numbers to show it.

Ferguson’s side was more efficient at converting that defensive platform into wins. A 74 per cent win rate versus 66 per cent. Only six draws in 38 games versus Arsenal’s seven in 35. In a defence-first campaign, draws are where points leak. United’s ability to grind out 1-0 results rather than settling for goalless stalemates is what pushed them from an 82-point season into a 90-point one.

But the profile is the same. Two sides winning through solidity, not volume.

Where they diverge is how the goals arrive. United’s 68 goals in 2008/09 came through a front four who were individually among the most watchable attackers in the league’s history. Ronaldo’s free kicks. Berbatov’s first touch. Macheda’s injury-time curler on his debut against Villa. The goals were rare but often spectacular. Individual brilliance behind a wall of defensive structure. United do not appear in the all-time Premier League records for corner goals that season, suggesting fewer than 13 from set pieces. Open play and moments of genius were the primary engine.

Arsenal in 2025/26 are a different machine. They can play. Saka on the right, Eze cutting inside, Odegaard threading passes through lines. On their best days, they are as watchable as anyone in the league. But the goals tell a different story. They have scored 17 from corners, breaking a Premier League record first set in the league’s inaugural season. Declan Rice delivers. Gabriel and Saliba attack the ball. Timber arrives at the back post. Set pieces account for a striking share of their goals, and the corner record alone is enough to show how deliberately this attack has been engineered. When open play does not produce, the corners do. It is tactically impressive and objectively effective. It is closer to manufacturing than improvisation.

Both sides win ugly by modern standards. Neither is consistently entertaining over 38 games. United’s 2008/09 campaign has been mythologised around the moments that survived in memory: a Ronaldo free kick, a Berbatov touch, Macheda off the bench. But those were the exceptions. Most of the 68 goals were 1-0 grinds behind a defensive wall. Arsenal are the same: Saka can beat three men, Eze can curl one into the top corner, but most of the output comes from Rice’s delivery meeting Gabriel’s forehead. The difference between the two campaigns is not that one is beautiful and the other is ugly. It is that the memorable goals from 2008/09 came from open play, while Arsenal’s come from dead balls. An individually brilliant goal and a corner headed in by a centre-back produce the same point. They do not produce the same highlight reel. That is the gap people are reacting to, even if the goals-per-game numbers are almost identical.

I put this data in front of a group of friends this week. Fellow United fans, mostly. Intelligent people. The response was instructive.

One friend argued from the wrong season entirely, citing 2007/08 when United scored 80 goals, rather than 2008/09 when they scored 68. When corrected, he said “Bye” and left the conversation. Another questioned where I got the data, as though the source of the compilation could change the numbers. A third said I was spreading propaganda to suit my narrative. Nobody engaged with the actual comparison. Nobody said: that is interesting, I did not know that. The tribal instinct overrode the evidence. A United fan cannot see the defensive record that defined his club’s 2008/09 title reflected in Arsenal’s current campaign. It is like complimenting your rival’s wife. The data is right there. The loyalty says no.

But what struck me was not the tribal refusal. It was the pattern underneath it.

The same week, I published an essay about the global content economy. I had spent weeks convinced that controversy was the only product that worked in the podcast market. The data said otherwise. Fewer than twenty of the top-ranking programmes across the major US charts are built on political controversy. The loud ten per cent commands ninety per cent of the cultural oxygen. I had been measuring my work against a room that did not represent the market.

I watched the same thing in financial markets. The current bull market sits at all-time highs, yet the mood is universally miserable. People trained on smooth, controlled, upward trends do not know what to do with a market that grinds higher through volatility and narrow breadth. They call it broken. It is not broken. It is winning ugly.

In every case, one dominant style had trained people to expect a particular version of the thing. Possession football. Viral controversy. Smooth returns. When the thing showed up in a different form, they rejected it rather than recognised it. The style became the standard. The standard became invisible. And serious people started abandoning what worked because it did not look like what was winning.

A friend in my football group called Arsenal’s style a total disgrace to the football world. Another said he would rather watch the women’s league than accept defensive football as a benchmark. They have a word for it. They call it haram ball. Forbidden. Illegitimate. Not real football.

I do not find it hard to watch. I find it hard to watch consistently over a 38-game season, the way I find any single approach hard to sustain interest in over that stretch. But in a knockout tournament, where every game is squeaky bum time, where one goal decides everything? I am more than comfortable with it. I grew up watching it through Italy. Baggio’s Italy were capable of brilliance in flashes and suffocation in between. So were Maldini’s Milan. I did not expect to find the same profile in my own club’s second-best ever points haul. And I certainly did not expect to find it in Arsenal. The tradition is not about refusing to attack. It is about controlling the game and being ruthlessly efficient when the moments arrive, even if the moments arrive from a corner. I just never had to defend that tradition until the people calling it haram were my own United fans, looking at a mirror they refused to recognise.

This is why I think it matters that Arsenal win.

Football felt richest to me in the 1990s. I was a boy, and the game I was watching had room for everything. Capello’s defensive Milan coexisted with Cruyff’s total football Barcelona. Sacchi’s pressing sat alongside Hiddink’s counter-attacking Netherlands. Multiple systems could win. No single style was orthodoxy. The ecosystem was diverse. Then Guardiola flattened it. First at Barcelona, then at Bayern, then at City. Tiki-taka and its descendants became the only acceptable way to play. The rest of the sport followed. Possession coaches were hired across the continent. Playing out from the back became mandatory. The tactical diversity that defined the game I grew up watching compressed into a single model. I may be romanticising the era because I was young, but I am not romanticising the compression. You can see it in the prestige language of the sport: the coaching hires, the academy curricula, the obsession with playing out from the back, and the suspicion directed at anything that wins another way. Italy, the country that defined defensive football, abandoned its identity to chase what looked like evolution. Two consecutive World Cup absences followed. The nation that won four World Cups through defensive discipline tried to become something it was not, because the dominant style had redefined what good looked like.

I said this to a friend on Wednesday. I told him: you are mistaking a dominant style for evolution.

He told me to support Arsenal in peace and stop trying to justify it.

If Arsenal win the league and the Champions League holding over 55 per cent of the ball and scoring a record number of goals from corners, it breaks the Pep-ball monopoly. Not destroys it. Breaks the monopoly. It tells every young coach on every continent that there is more than one way to build a winning side. That tactical diversity is not regression. That what the last quarter-century trained us to call beautiful is not the only version of the game that wins. The rotation has always been there. Total football arrived in the 1970s with Cruyff’s Netherlands, then faded after two consecutive World Cup final defeats: to West Germany in 1974 and Argentina in 1978. The generation aged out and the style lost its international dominance. It returned when a new Dutch generation, Gullit, Van Basten, Rijkaard, won the 1988 European Championship, then faded again when Capello’s Milan dismantled Cruyff’s Barcelona 4-0 in the 1994 Champions League final, and came back under Guardiola for the longest run of dominance any single style has held. If Arsenal win well, I think it fades again. And something else takes its place. And the game is better for it.

There is a ten-year-old boy in Nairobi who woke up on a Monday morning in July 1994, ran to his parents’ room, and learned that his favourite player had missed. The World Cup returns to the United States this summer, 32 years later. He will be 42. It took him all that time to understand that Baggio’s Italy and Arteta’s Arsenal are the same tradition, wearing different colours, playing in different decades. Both judged by people who mistake the dominant style for the only style.

The numbers do not care what you call it. They sit there, waiting for someone to look.

Statistics sourced from Opta Analyst, StatMuse, the Premier League, and FBref. Arsenal data as of 35 games played (2 May 2026).

AI-generated image of the geographic penalty.

There is a product being sold to hundreds of millions of people every day, and most of them do not know they are buying it. It arrives on their phones disguised as information. It sounds like news. It carries enough verifiable fact to feel credible. But its purpose is not to inform. Its purpose is to hold attention long enough to monetise it.

The business model is straightforward. Take a real event. Thread a narrative through it that is partly true and partly constructed, selectively framed to maximise emotional response rather than to clarify. Deliver it with high production quality and emotional conviction. The consumer receives something that feels like analysis but functions like entertainment. The old version of this product was the Hollywood movie. It came with one or two trailers, a release date, and a screen you chose to sit in front of. The new version is the video podcast, cut into short-form clips and fed into every platform simultaneously. It is a daily trailer that never stops releasing, hitting the consumer from every angle, every scroll, every feed. The movie never pretended to be real. And it never followed you home.

The creators who command the most attention in this economy are not the most informed. They are the most watchable. Dysfunction turned into product, sold at scale, consumed as insight.

I have written before that discomfort is not violence, and that free speech requires harder ground than most societies are currently willing to maintain. That argument still holds. But it was incomplete. It addressed the listener’s fragility. It did not address the supplier’s incentive. When the business model of the content economy is optimised to blur the boundary between truth and narrative, the problem is no longer that people cannot tolerate hard speech. The problem is that they can no longer distinguish it from performance.

The Perception Distortion

The data tells a different story from the one most people carry in their heads. Across the major US podcast charts, fewer than twenty of the top-ranking programmes are built primarily on political controversy. On the right: Tucker Carlson, Candace Owens, Ben Shapiro, Patrick Bet-David, Megyn Kelly, Matt Walsh, Dan Bongino. On the left: Pod Save America, MeidasTouch, The Young Turks, Rachel Maddow, The Bulwark, Ezra Klein. That is roughly ten per cent of the most listened-to programmes. The categorisation is mine, and the boundary is not clean. Many of these programmes do substantive work alongside their partisan framing: policy analysis, long-form interviews, investigative reporting. The label “controversy-driven” does not mean every episode is noise. The phenomenon is not uniform, and it spans the political spectrum. A decade ago, the most commercially successful form of ideological rigidity in the content economy came from the cultural left: policing language, enforcing orthodoxies around identity and appropriation. Today, the most commercially dominant version comes from the right: institutional distrust, anti-establishment certainty, and alternative knowledge communities built around suspicion of official narratives. At their core, neither version rewards sustained engagement with counter-evidence, even when individual episodes do substantive work. Both monetise conviction. The mechanism is the same. The direction has rotated. The distinction I am drawing is about a business model, not an ideology: programmes whose commercial engine runs on partisan emotional activation, where the audience’s political convictions are the product being monetised. This is not to say that other genres avoid emotion. All media activates emotion. True crime manufactures suspense. News selects for resonance. The distinction is between content that uses emotion to illustrate reported events and content that uses emotion to advance political claims presented as factual analysis. The rest of the chart includes programmes that are sensational, trivial, or shallow. But they are not built to blur the line between fact and narrative for political effect.

What dominates the chart is comedy, true crime, news, storytelling, and long-form interview. Crime Junkie is audio-first. The Daily is audio-first. NPR is audio-first. The BBC is audio-first. Many of these shows now distribute clips or full episodes on YouTube, but their primary product is designed for audio consumption. These programmes chart consistently, attract loyal audiences, and generate revenue without manufacturing outrage. They are the quiet majority of a market that most people assume is dominated by provocation.

The analysis uses the US market deliberately, not because the content economy is exclusively American, but because the controversy-driven content that shapes global perception is predominantly American in origin. These are American products consumed globally through short-form clips on X, TikTok, and Instagram. The platforms through which African creators operate, Spotify, YouTube, Apple Podcasts, are US-headquartered with US-benchmarked CPM structures. The perception distortion this essay describes is driven by American content that travels through global feeds and shapes how creators in other markets, including Africa, assess the landscape. Chinese platforms operate under state censorship, which is a different dynamic entirely. Indian and other regional markets have their own ecosystems. This essay is not claiming to describe every content economy on earth. It is describing the content economy as experienced by African creators, which is predominantly mediated through US-origin platforms and US-produced controversy content.

The split between audio and video reinforces the point. In the top ten, the balance is roughly even. But further down the chart, the formats diverge. True crime remains overwhelmingly audio. News is almost entirely audio. The video shift is concentrated in personality-driven, controversy-adjacent content. The loudest room is not the biggest room. It is simply the one with the best acoustics.

This matters because perception shapes behaviour. A content creator entering the market today surveys the landscape and sees controversy rewarded, substance ignored, and volume valued over rigour. That perception is measurably wrong. But it is powerful enough to make serious builders question whether their work has a market. I know, because I questioned it myself.

The Geographic Penalty

If the perception distortion is the first problem, the payment architecture is the second. And for African creators, the second problem is far more damaging than the first.

The Africa Creator Economy Report, published in Lagos in January 2026, found that six in ten African creators earn less than USD100 per month from their digital work. More than half earn less than USD62. Even adjusted for purchasing power, the gap remains substantial, though the raw dollar comparison overstates the disparity in real consumption terms. Ad revenue accounts for just 5.8 per cent of creators’ income. The TikTok Creator Fund is not available in most African markets. Spotify’s revenue per stream in many African countries runs three to four times lower than in Europe or the United States, a function of cheaper subscriptions, ad-supported listening, and lower CPM rates. The Afrobeats hit “Coup du marteau,” produced by Tam Sir and the unofficial anthem of the 2023 Africa Cup of Nations, has surpassed 100 million YouTube views. The song earned substantially across global streaming, licensing, and performances. But the platform revenue from its African audience tells a different story. The roughly 10 million views routed through Senegal, the only francophone Sub-Saharan market where YouTube payouts are available, generated approximately 971 euros in ad revenue. The remaining views, overwhelmingly from Ivory Coast, Cameroon, Mali, and the broader francophone continent, generated zero platform revenue for the creators. Tam Sir earned despite the African payment layer, not because of it. Most African creators do not have a continental hit that crosses into global markets. For them, the platform payment layer is the primary income source, and it is the layer that pays the least.

An economist would argue that lower CPMs are not a penalty but a price signal: advertisers pay less to reach consumers with lower purchasing power, and the platform applies the same revenue-share algorithm to every market. That is technically correct. But when the platform’s pro rata distribution model locks that differential into every stream, every view, and every click, the price signal becomes a ceiling that individual creators have no mechanism to negotiate. The market is also growing: Sub-Saharan Africa’s recorded music revenue rose 22.6 per cent in 2024, roughly five times the global average. But the growth rate makes the infrastructure gap more urgent, not less. A rapidly expanding audience generating rapidly expanding revenue that does not yet flow to African creators at scale is a widening gap, not a closing one.

A minority of African creators have built viable businesses through direct brand relationships, sponsorship negotiations, and audience-funded models that bypass the platform payment layer entirely. These creators are not waiting for platforms to fix the plumbing. They are building around it, pioneering revenue models, from live podcast events to direct brand incubation, that global platforms have not imagined. Their success does not weaken the broader argument. It sharpens it. When the only path to sustainability requires exceptional individual effort and constant improvisation around underdeveloped infrastructure, the infrastructure is the gap.

The CEO of Afripods, a pan-African podcast hosting platform headquartered in Nairobi, stated the gap plainly: no major platform has built an Africa-specific creator payment programme at parity with Western markets. African and non-Western platforms are beginning to fill the space. Boomplay, headquartered in Beijing with major operations in Lagos, pays creators in local currencies and has tens of millions of active users. Audiomack has significant and growing African market penetration. But the scale remains disproportionate. A Zimbabwean podcaster with 30,000 subscribers described his audience as “huge for us and Zimbabwe” but, on a global scale, “a drop in the ocean.” Even BBC Africa has struggled to find regular advertising on its podcasts, according to industry participants interviewed by Jamlab. Meanwhile, Showmax, widely regarded as Africa’s largest homegrown streaming service and one of the continent’s most significant commissioners of original content, ceased operations on 30 April 2026, after losses surged to R4.9 billion. This came two years after Amazon pulled out of African originals entirely.

The monetisation infrastructure that sustains creators in New York, London, and Los Angeles does not exist at equivalent scale for creators in Nairobi, Lagos, or Lusaka. This is not a quality gap. It is a plumbing gap. The pipes that carry revenue from audience to creator were designed for markets with high CPM rates, card-based payment systems, and premium subscription bases. African markets operate on mobile money, prepaid data, and ad-supported tiers that generate a fraction of the revenue per listener. The result is a geographic penalty: African content travels globally, African payment does not.

The BBC asked the question this week: why are African creatives earning less on YouTube? The answer lies in the payment architecture. The platform’s royalty model distributes revenue based on total streams weighted by the economic value of the listener’s market. Within that model, an American listener generates more advertiser revenue than a Kenyan listener, because the advertiser’s expected return is higher in wealthier markets. The content is identical. The compensation is not.

The Regulatory Inversion

If the global market will not pay African creators fairly, the logical response would be for African governments to build the infrastructure that closes the gap: invest in payment systems, negotiate platform terms collectively, fund measurement tools that give African audience data commercial weight. Too little of that is happening at the scale required. Some governments and development institutions are investing in digital infrastructure. The Pan-African Payment and Settlement System (PAPSS) is already operational and processing cross-border transactions. Kenya has a Creative Economy Support Bill in its legislative pipeline. Rwanda’s digital economy push is among the most advanced on the continent. But the gap between policy ambition and creator monetisation remains wide. And in too many jurisdictions, what is growing faster than infrastructure investment is regulatory control over speech.

Cybercrime across African markets is real and damaging: SIM-swap fraud syndicates, mobile money theft, coordinated disinformation campaigns, AI-generated child exploitation material, and foreign-backed election interference. These are documented threats that cost African consumers and businesses billions. Legislation designed to address them serves a legitimate purpose, and some prosecutions under these laws, targeting financial fraud, identity theft, and exploitation, have delivered proportionate outcomes. The problem lies in the absence of safeguards that prevent legitimate instruments from being repurposed for political control.

In April 2025, Zambia enacted the Cyber Security Act and the Cyber Crimes Act. The laws were debated in parliament, amended during committee stage, and subject to civil society consultation, however imperfect that process was. Some provisions address genuine cybercrime. But international observers and civil society groups warned that other provisions, broadly worded, granted sweeping powers to criminalise dissent and enable intrusive surveillance. Those warnings have been borne out. Munir Zulu, a sitting member of parliament at the time of the post and a controversial, combative political figure with a documented history of inflammatory public statements, was convicted of seditious practices and sentenced to 18 months with hard labour. His offence was claiming, on Facebook, during a period of genuine political tension, that the president planned to dissolve parliament and hold early elections. The claim was wrong. Even some opposition figures criticised it. The punishment was eighteen months of hard labour. Three citizens were arrested for statements about the president’s health. The Zambian government did not respond to requests for comment from Human Rights Watch. And on 29 April 2026, the Zambian government withdrew venue access for RightsCon, a leading global summit on human rights and technology, prompting the organisers to cancel the event days before it was scheduled to open in Lusaka. The decision was reportedly driven by a combination of domestic political calculations, security concerns about certain participants, and Chinese government displeasure about invited delegates from Taiwan. The conference venue, the Mulungushi Centre, had been refurbished with USD60 million in Chinese funding, described at the time as “a gift with no strings attached.”

Zambia is not an outlier. It is a pattern, and the pattern extends well beyond any single administration. The essay uses Zambia as a detailed case precisely because the reform mandate that brought Hichilema to power makes the trajectory more instructive. Leaders who arrive on reform tickets and then build control architectures are not unique to Lusaka. The pattern recurs across the continent.

Tanzania imposed a full internet shutdown on election day in October 2025. The African Union’s own Election Observation Mission acknowledged the blackout’s impact on election monitoring, with observers and rights groups questioning the elections’ democratic credibility. Uganda imposed a near-total blackout during its 2026 elections. Togo restricted access to key social media and messaging platforms for 43 days in 2025 amid political unrest. Gabon’s social media restrictions cost nearly CFA30 billion per month, according to NetBlocks estimates. Kenya’s government ordered all television and radio stations to stop live coverage of the June 2025 demonstrations and physically cut broadcasting signals. Kenya’s High Court suspended the ban within hours on the same day and permanently struck it down five months later as unconstitutional, but the damage, a live blackout during active protests, had already been done. The government also passed legislation in 2025 expanding its powers over online content. Ethiopia has recorded at least 30 internet shutdowns since 2016. Sudan’s military government imposed a cybercrime law with severe penalties for social media users and reporters, with watchdogs warning that vaguely drafted provisions expose online speech to years of imprisonment. Nigeria’s proposed Digital Economy Bill would grant a single agency authority over virtually every pillar of the digital economy, from AI and cloud services to cybersecurity and platform governance, centralising control in a manner that its critics argue has less to do with modernisation than with capture. In 2024 alone, 21 shutdowns were recorded across 15 African countries. The estimated cost to sub-Saharan economies in 2025 was USD1.11 billion.

These interventions are usually defended through the language of consumer protection, national security, or cybersecurity. Some of those concerns are real. But in case after case, broad digital powers have also been used to restrict dissent, limit protest coverage, pressure civil society, or control information flows around elections. For young African creators, that produces a double constraint: global platforms monetise their audiences weakly, while domestic states increasingly regulate the terms on which they can operate.

The Missing Sunset Clause

There is a parallel in trade that clarifies the underlying problem.

Free markets work until a subsidised actor distorts them at scale. China’s state-backed industrial subsidies made open competition untenable for domestic manufacturers in dozens of countries. The response, tariffs and trade barriers, was not inherently wrong. But the difference between a defensible tariff and destructive protectionism is a single mechanism: the sunset clause. A time-bound intervention that protects domestic capacity while it builds, with a mandatory review and an expiry date. Without the sunset, protectionism becomes permanent rent-seeking. The state captures the benefit. The consumer bears the cost. The domestic industry never matures because it never has to.

The same logic applies to the speech and information-control provisions of digital regulation, not to the criminal provisions addressing fraud, identity theft, or exploitation, which should be permanent. If the problem is misinformation degrading the information commons, a time-bound, independently reviewed intervention directed at building the public’s evaluative capacity is defensible. Media literacy programmes with measurable outcomes. Transparency requirements for algorithmic recommendation systems. Disclosure rules for sponsored content. All of these can carry sunset clauses. All can be reviewed against evidence. All can expire.

I write this as someone who argued, less than a year ago, that no regulation can substitute for personal courage in public dialogue. That conviction has not changed. What has changed is my recognition that the information environment in which that courage operates is itself being engineered, from one side by the market and from the other by the state. The case for intervention is not abstract. In Kenya’s 2007 post-election crisis, unregulated vernacular radio, not the state, incited ethnic violence with a documented body count. That was not discomfort from hard debate. It was direct incitement to mass killing, the category of speech that even the most committed defender of open dialogue acknowledges as a legitimate limit. The question is whether the regulatory response to that real danger can be trusted not to become the greater danger: permanent state control over speech that is merely uncomfortable, inconvenient, or politically embarrassing. The sunset clause is the minimum condition under which any intervention remains distinguishable from control.

This essay identifies the design principle that any credible institution would need to embody. The principle is time-bound intervention with independent review. The institution does not yet exist. The legislative calendars of most African parliaments are already overwhelmed, and the review mechanism would need to sit outside the parliamentary cycle, which raises its own governance questions.

Among the major digital laws enacted across the continent in the past three years, none that I have reviewed contains a meaningful sunset clause on its speech-related provisions. Some include parliamentary reporting obligations or are subject to judicial review, as South Africa’s Cybercrimes Act and Kenya’s Computer Misuse Act have demonstrated. Kenya’s High Court nullification of the broadcast ban is the closest example on the continent of binding judicial review working in real time, but even there, the ban was imposed and enforced before the court intervened. These are real, if unevenly distributed, oversight structures. But they are not the same as a mandatory expiry that forces re-justification. The Zambian case is instructive precisely because Hichilema came to power on a reform mandate. Civil society groups, including CIPESA and the Paradigm Initiative, have documented how digital laws across the continent are increasingly drafted with broad provisions that serve suppression alongside their legitimate functions. The African Declaration on Internet Rights and Freedoms provides a normative framework, but it lacks enforcement teeth. The cyber laws were enacted under the banner of modernisation. In too many jurisdictions, their architecture serves control.

The enforcement gap is real and must be named honestly. The African Commission on Human and Peoples’ Rights condemned Tanzania’s election shutdown. Nothing followed. The ECOWAS Court found Senegal’s 2023 shutdown unlawful. The ruling took two years. Kenya’s permanent constitutional ruling against the broadcast ban is a genuine precedent, the kind of institutional development the continent needs more of. Whether the next government respects it is a different question. African governments have a documented tendency to abrogate constitutional provisions and ignore judicial orders when they become inconvenient. Many of the continent’s leaders aspire to the Singapore model of rapid state-led development, but Singapore’s authoritarianism was embedded in predictable, consistently applied rule of law. The authority came with discipline. The control came with predictability. That combination remains rare on the continent. The institutional architecture for digital rights exists on paper across Africa. It lacks binding authority in practice. Domestic judiciaries are slow, compromised, or both. Civil society organisations that might challenge these laws are themselves being criminalised.

The sovereignty objection will be raised against any proposal for external enforcement. It is always raised. And it is sometimes legitimate. But sovereignty is not a shield for suppression. A government that arrests citizens for social media posts about the president’s health is defending presidential fragility, not sovereignty. The distinction matters. And the leaders who would need to sign a binding digital rights treaty are, in too many cases, the same leaders currently building the infrastructure those treaties would constrain. The circularity is the problem.

Any enforcement mechanism that emerges must be African-designed and African-governed. External imposition would replicate the colonial structure this essay critiques. But African-led cannot mean government-led when the governments are the ones exercising the powers in question. The design challenge is institutional independence within a sovereignty framework. It has not been solved.

The Content Resource Curse

The pattern of peripheral value extraction that dependency scholars, from Samir Amin to contemporary analysts, have documented in commodity markets is reproducing itself in the content economy. The prescription here is not Amin’s delinking but a demand for equitable integration: the same platforms, with payment infrastructure that works for all markets. Africa exports its raw minerals for processing elsewhere, capturing a fraction of the value. The analogy is imperfect: content is not a finite resource, and creators retain ownership in ways that miners do not. But the geography of value capture is parallel in form. African creators generate attention, audiences, and cultural product. The value is captured disproportionately by platforms headquartered in San Francisco, Stockholm, and Beijing. The revenue flows through payment infrastructure built for other markets. The creator receives the equivalent of a royalty on unprocessed ore.

The objection that African governments face competing fiscal priorities, health, education, physical infrastructure, is real. But it misreads what the content economy requires. This is not a demand for state-built platforms or publicly funded ad exchanges. It is a recognition that the digital creative economy is one of the few sectors where a young person with a phone and a data connection can create a livelihood without waiting for formal employment to absorb them. On a continent where over 60 per cent of the population is under 25 and youth unemployment is structural, treating the conditions for digital participation as a luxury amounts to a misallocation of a different kind.

Part of the challenge is generational. Many of the continent’s economic planners still operate within an industrial development paradigm where value creation means manufacturing, mining, and formal employment. The content economy does not look like productive work to policymakers who came of age before smartphones. But it is a livelihood bridge that requires less capital, less time, and less state permission than any industrial programme, and it is available now.

If African governments will not build the monetisation infrastructure and will not permit the speech that makes content creation viable, the market will impose its own verdict. Talent will migrate, digitally if not physically, to platforms and jurisdictions that both pay and permit. The continent will continue to produce creative output whose commercial value is captured disproportionately outside its borders.

The loud ten per cent of the global content economy will continue to command disproportionate attention. But they are not the market. They are the distortion. The broader market is quieter, more diverse, and still significantly audio-led outside the most visible personality-driven segment. It rewards consistency, trust, and depth over provocation. African creators belong in that market. They do not yet have access to its payment layer at equivalent terms, and their own regulatory environment is tightening the space in which they can operate.

The question is not whether to regulate. It is whether the regulation has an expiry date, an independent reviewer, and a purpose beyond the next election cycle. Until African governments can answer that question credibly, the content economy on the continent will remain trapped between a global market that will not pay and a domestic state that will not permit. The loudest voices will keep commanding the room. The ones with something to say will keep being told to be quiet.

In September 2025, I argued in “Discomfort Is Not Violence” that free speech requires harder ground than most societies are willing to maintain. That argument addressed the listener. This essay addresses the supplier and the regulator. The two pieces are companions. The ground has not become harder. It has become contested from both sides.

Sources

Africa Creator Economy Report 2.0, Communique and TM Global, January 2026. African Declaration on Internet Rights and Freedoms. Broadcast Media Africa, “Monetising Podcasts and On-Demand Audio in Africa,” April 2025. Broadcast Media Africa, “Africa’s Media Industry Confronts Revenue Shift,” April 2026. CIPESA, “Africa’s Digital Dilemma: Platform Regulation vs Internet Freedom,” May 2025. Ecofin Agency, “African Artists Face a ‘Geographic Penalty’ in Streaming Payouts,” February 2026. Freedom House, “Zambia: Freedom on the Net 2025.” Human Rights Watch, “Zambia: Summit on Human Rights, Technology Effectively Canceled,” 1 May 2026. ISS Africa, “Internet Shutdowns Won’t Solve Central Africa’s Political Crises,” April 2026. Journal of Democracy, “How Zambia’s Cyber Laws Rebrand Repression,” August 2025. Netzpolitik, “Internet Shutdowns in Africa: A Human Rights and Democratic Crisis,” February 2026. Paradigm Initiative. Reuters Institute, “African Podcasters Are Now Recognised Globally. Can They Transform This Success into a Viable Business?” Spotify Charts; Podtrac Multi-Channel Podcast Rankings, March 2026. TechCabal, “Podcasting in Africa Is on the Rise. Why Is It Not Profitable Yet?” August 2022. TechCabal, “These African Countries Passed Major Tech Laws in 2025,” December 2025. TechPoint Africa, “Six in Ten African Creators Earn Less Than $100 Monthly,” January 2026. UN News, “UN Warns of Rising Internet Shutdowns,” January 2026. Variety, “After Canal+ Shutters Showmax, Is the Dream for Cutting-Edge African Content Over?” March 2026.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu or X @InfinitelyDean.

AI-generated Image: The data does not sort. By anything. In any direction.

For two weeks I have been watching South Africa from Nairobi.

On 19 April, Tabeth Chidziva, a Zimbabwean street vendor in Hillbrow, was shot dead on camera. Her family says she was defending her pregnant daughter. Viral posts reduced it to a dispute over a plate of food. The CCTV footage went viral within hours. In Durban, protests led many businesses to shut their doors from 22 April after the March and March movement escalated pressure on foreign-owned traders. Foreign-owned shops have been looted in KwaZulu-Natal and the Eastern Cape. Ghana and Nigeria both issued formal advisories to their nationals. The African Commission on Human and Peoples’ Rights released a statement of condemnation. And across all of it, videos circulate on X with the same algorithmic urgency, some current, some recycled from years ago with fresh captions. Tabeth Chidziva’s own family had to ask the public to stop spreading false versions of her death.

The feed is a room. It selects for content that confirms the category it has assigned you. If you are watching xenophobia footage, the feed will show you more xenophobia footage, until the footage becomes the country. South Africa is reduced to a single story. The algorithm does not do this with malice. It does it because confirmation drives engagement, and engagement is what it optimises for. You end up believing you have seen the whole picture when you have only seen the corner the feed chose for you.

I am a black foreigner who lived in South Africa. I should know what this looks like from the inside. And I do. But the data I carry does not match the story the feed is telling, and it does not match the story my category is expected to tell either.

In my first year at the University of the Witwatersrand, I experienced xenophobia. It came from black South Africans. I did not think much of it at the time. I filed it somewhere between irritation and culture shock and moved on. I never experienced racism from a white person in South Africa. That sentence will cost me something, and I am going to let it sit there anyway, because it is what happened.

But that sentence is not the whole picture either. Some of my closest friends are black South Africans. People I joined the graduate programme with. People I studied with. People I spent long nights cramming with before exams. One of the people I can call at any hour, for any kind of help, is a black South African. The xenophobia I experienced in my first week and the friendships I built over the years that followed exist in the same country, in the same city, on the same campus. Both are real. Neither cancels the other.

At university, some classmates who received poor marks would attribute them to racial bias. The exams were marked by student number. The marking was anonymous. I checked my own papers against the criteria and I could not see the evidence for the claim. Maybe the weaker English that some students carried from under-resourced schools showed through even without a name attached. That is a real problem with real origins. But it is a problem of upstream schooling, and calling it racism at the point of marking skips every link in the chain to land on the conclusion the category supplies. The frame arrived before the evidence. I watched it happen in real time, and I noticed I was the one checking the paper while others were reaching for the category.

The pattern repeated across every environment I entered. At my Opus Dei high school, where the priests were mostly Spanish and Kenyan with an American, I was an Adventist kid inside a Catholic structure that was not built for me. The maths teacher who liked me most was black. He died recently, and I carry that. The commerce teacher who backed me was also black. The Swahili teacher from Read These, She Said expected me to fail. He was also black. Three black teachers, three readings of the same student. At the Indian school that followed, a full scholarship placed me among a mostly Indian student body. At the A-level award ceremony for examinable subjects under British curriculum schools across East Africa, there were two black students in the room. I was one of them. In my career, white practitioners across five nationalities opened many significant doors. My mentor, a white South African, still guides me. The person who presented the most professional friction was a black Zambian, working under a white South African who was decent to me. And alongside all of them, others showed up at different points. Black, white, Indian, across every creed. People who shared ideas and gave assistance when it mattered.

I am listing these details because removing them would be dishonest. The data does not sort. By anything. In any direction.

These were specific rooms. They were unusual rooms. Most people who look like me were not in them. A pattern can be real and still fail to explain every room. I am not claiming my experience disproves anything structural. I am claiming it is mine, and I have not been willing to suppress it. But saying that out loud turns out to be harder than I expected.

Xenophobia is category enforcement by violence. Tabeth Chidziva was not shot because of who she was. She was shot because of what she was filed under. Foreign. The category was enough.

The feed is category enforcement by curation. It does not show you the black South African who would answer your call at two in the morning. That is not engaging content.

Education is category enforcement by credential. The categories arrive through coursework, through reading, through institutional authority. They feel like knowledge because they were acquired through the process we associate with knowledge. Contradictory data is treated as error, not evidence.

And the pressure to suppress your own contradictory evidence is category enforcement by silence. When you hold data that complicates the narrative, the expectation is that you will keep quiet. One side will extract a sentence and use it to say racism is overstated. The other will extract a different sentence and say you have been captured by proximity. Both will strip the context. When I have made observations like these before, it is black people who have called me a house negro. The enforcement comes from inside the category. That is what makes it effective. When you suppress what you actually experienced, you are deferring to your demographic’s expected testimony over what your own rooms showed you.

Over the weekend I sat with a group of educated professionals at a bar in Nairobi. The conversation turned to why populist movements and figures like Trump continue to garner support across parts of Africa and the West. They listened intently, but challenged vehemently. The default was immediate: his supporters are misinformed, under-educated, consuming the wrong media. The room could not sit with the possibility that education itself might have installed the categories being defended. Indoctrination is something that happens to the uneducated. It cannot happen to us. We did the reading.

Someone in different rooms collected different data. Some of that data may reflect real structural damage carried from upstream. Some of it may reflect a category that was reached for before the evidence was checked. I do not know which, because I was not in their rooms. What I know is what my rooms showed me, and I have checked it against the evidence I have.

The videos are still arriving. I watch the footage and I hold what anyone would hold. The violence is real. The suffering is real. People who look like me are being targeted by people who look like me, in a country where the people who invested in me look nothing like either group, and the people who became my closest friends look like both. I carry all of that at the same time. It does not resolve into a clean position. I do not think it is supposed to.

What are you carrying that does not resolve? And have you reported it honestly, or have you let the room you are in tell you what you saw?

AI-generated image of a Risk board mid-game.

The Canary Compass podcast is live on Spotify and YouTube. Tuesday episodes explore the week’s note from a different angle. Saturday episodes companion the Friday Reflection. Bonus episodes on Thursday when two notes publish in the same week.

I had a losing record at chess.

There was a kid in our neighbourhood who played regularly, and I played against him whenever the board came out. I lost more than I won. I never kept exact count, but the record was clear enough. He was better than me. That much I knew.

What I did not know, because it never occurred to me to ask, was whether he was actually any good. He was the only person I ever played. My entire assessment of my own strategic ability was built on a sample size of two. For all I knew, he was also terrible. I just happened to be slightly more terrible. I accepted the verdict and carried it quietly for years: I am not a strategy person.

The trouble with childhood verdicts is the instrument, not the cruelty. You receive them at eight or twelve or fifteen, and they harden into facts about yourself that you stop examining. The losing record hardened into a conclusion I did not revisit.

My sister had a different kind of honesty.

I sang at church the way most children sing at church. Loudly and with conviction, without any evidence that the sound leaving my mouth was the sound I intended. My sister sat next to me, and by the second hymn she had reached her limit. She told me to keep quiet because I was embarrassing her. There was no room for interpretation.

That verdict travelled with me through secondary school, where I tried out for the boys’ choir. I did not get in on the first attempt. Or the second. On the third attempt, the teacher let me in, and I suspect it was mercy rather than range. When we performed, I was placed next to the microphone at the front. My tone was off. The mic did not help.

My sister was right. The choir teacher confirmed it by letting me in out of pity, and the microphone made the verdict public. I could not sing, and an honest woman had told me so early enough that I only lost a school term of dignity rather than a career. Someone loved me enough to say it plainly.

So one verdict was wrong and built on nothing. The other was right and built on proximity and honest love. The same person received both, in the same childhood, and could not tell the difference at the time. I did not know that verdicts needed auditing. I thought they were just things that were true about you.

The games I actually loved told a different story.

Scrabble came first. I was top of the boys’ club in our neighbourhood, and the feeling of placing a seven-letter word on a triple score was the closest thing to flight I had at fifteen. Every player looks at the same tiles and the same dictionary. What separates them is who can see the combinations nobody else is staring at. I did not know the phrase “pattern recognition” then. I just knew I could see the words the other kids could not.

Then there was Risk.

Risk was the one that mattered, though I did not know why at the time. Multi-front, with alliances forming and dissolving at the table, and no player telling you their real intentions. You absorb losses on one front to win on another. Timing matters as much as position.

And then there are the dice. You can read the board and position the armies well, and the numbers still fall against you. A campaign planned across three turns collapses because you rolled badly. The game is designed to break your plans. The skill is building a position where one bad roll does not end you. Fortify before you attack. Absorb the loss and wait for the next roll.

I loved it because it felt like the right shape. Multiple moving parts and no certainty, with your nerve being tested every time the dice went against you. Nobody explained any of this to me. I gravitated toward it the way you gravitate toward a room where the temperature feels right.

It took twenty years and a career in financial markets to understand that Risk had been teaching me my profession before I entered it. The kid with the losing chess record turned out to be built for multi-front positions under imperfect information, where resilience was the only currency that mattered. The chess record was accurate. The board was wrong.

The audit question I wish someone had handed me at fifteen is simple. Whose board am I playing on? Before you accept a conclusion you have carried for years, ask who set the test and whether the board matched the gift. My sister was right because the instrument was the right one for what she was measuring. Most of the verdicts we carry do not have that pedigree.

This weekend, pick one small old label. The thing someone said about you that you never revisited, the one you still introduce yourself with. Ask whose board you were playing on. Sometimes the verdict holds, and you just needed someone honest enough to say it. But occasionally you will find that the game you lost was never the game you were built to play, and the losing record was a story about the board, not about you.

AI-generated Illustration: The two-gate filter and three-pool architecture. Gate One applies a locked fiscal metric uniformly across 55 AU members. Gate Two codifies five override triggers. Pool Three is the capital pool; Pool One and Pool Two are preconditions that move countries through the tiers.

The Canary Compass podcast is live on Spotify and YouTube. Tuesday episodes explore the week’s note from a different angle. Saturday episodes companion the Friday Reflection. Bonus episodes on Thursday when two notes publish in the same week.

0. Executive Summary

Sub-Saharan African sovereign issuers have paid 150 to 200 basis points more at issuance than their fundamentals justify in primary bond markets. The cost compounds across every issuance and represents at least USD5bn in additional interest on current outstanding bonds. The premium is not a grievance. It reflects the institutional capacity markets actually price: whether contracts hold through political cycles and whether fiscal systems absorb shocks without external supervision. African multilateral institutions that should lower the continent’s cost of capital, including Afreximbank, the Trade and Development Bank, and the Africa Finance Corporation, fund themselves in private markets at material spreads over comparable benchmarks, with the precise ranges varying by issuer, tenor, and rating class as earlier Canary Compass research documented (African Multilaterals, June 2025). Africa owns neither the outcomes-measurement layer that verifies capital deployment into productive enterprise nor the capital pool deep enough to set its margin.

This essay introduces the first layer of a capital architecture designed to close that gap. The filter built here is a locked fiscal metric applied uniformly across all 55 African Union members, with a two-gate system combining objective measurement with codified governance overrides. A tier structure determines which countries have earned the fiscal discipline that makes capital deployment safe. The filter is not the product. It is the gate. Part III builds what sits behind it: a four-level measurement engine. Level 1 grades government fiscal data through the confidence system set out below. Levels 2, 3, and 4 use real-time satellite, blockchain, and IoT verification as complementary layers, addressing the 12 to 24 month audit lag that makes public reporting too uneven to support live allocation on its own.

A note on the series. Part I was published in January 2026 and anticipated that Part II would test remittance cycle correlation and institutional research validation. The architecture that emerged demanded a different sequence: the filter must be built before it can be tested. The series is now five parts. Part I (Push, Pull, Friction) established the signal. This essay builds the filter. Part III (Screen, Score, Deploy) applies it across all 55 AU members and constructs the capital pools. Part IV (Hold, Absorb, Prove) addresses safeguards. Part V (Measure First) specifies the minimum viable product.

1. Three Gaps

Part I established the signal: three forces converging on a structural inflection. AI compressing the returns to labour that sustained Black Atlantic economic mobility. Social media collapsing the distance between diaspora and continent. And a structural absorption gap in the industries that could deploy returning capital into productive enterprise: agro-processing, energy, logistics, construction, water infrastructure, healthcare, digital infrastructure. The diaspora professional in London, whether fourth-generation African American or first-generation Nigerian immigrant, whose Vanguard index fund compounds on companies he has never visited. The entrepreneur considering Nairobi over Atlanta. The pension trustee allocating capital. They all face the same operational question Part I left unanswered. If diaspora capital were to flow into African enterprise, infrastructure, and sovereign instruments, what would price them, what would measure the underlying fiscal health, and which countries would earn access?

The existing frameworks do not yet provide a sufficiently Africa-specific, independently owned measurement layer. The information environment they produce is part of the problem.

Gbohoui, Ouedraogo, and Some, writing for the IMF in 2023, analysed 1,592 primary sovereign bond issuances from 89 countries between 2003 and 2021. After controlling for credit ratings, debt levels, GDP, reserves, and institutional quality, they found that Sub-Saharan African countries pay 150 to 200 basis points more at issuance than their fundamentals predict (IMF WP/23/130). Olabisi and Stein, extending their original study through 2022, find the premium persists at approximately 100 basis points. On current outstanding bonds, that represents at least USD5bn in additional interest unexplained by standard fundamentals (Journal of African Trade, 2024, Vol. 11).

Markets do not price fairness. They price the likelihood that contracts will be honoured through full political and economic cycles. Gbohoui et al. also find that the premium narrows substantially when structural factors, including budget transparency, informality, financial development, and institutional quality, are accounted for. That finding is not a refutation. It is a diagnosis. The premium reflects measurable structural conditions, not sentiment. Complaining about it changes nothing. Without the filter, capital allocators default to external assessments and the USD5bn on today’s bond stock compounds with every new issuance through the decade. Measuring those conditions, publishing the results, and creating incentives to improve them is the only path that moves the price. This is the argument I made in From Grievance to Design (Canary Compass, December 2025). This series builds the engine that article called for.

Measurement is where it begins. Africa’s interest-to-revenue burden, the share of government income consumed by the cost of past borrowing, produces two numbers depending on who is counting.

The near two-to-one gap is a statistical artefact with three causes: aggregation method (revenue-weighted average versus median), sample composition (all-Africa versus SSA-only), and the weight of high-ratio economies. Egypt reaches 84 per cent in FY2023/24 first ten months (IMF CR 24/274), but Kenya (~30 per cent), Nigeria at the federal level (~41 per cent), and Angola (~32 per cent) also pull the revenue-weighted average sharply higher. The IMF covers Egypt in its Middle East and Central Asia REO rather than its SSA sample, which accounts for part of the divergence between the two figures but does not eliminate it. Both methodologies are defensible. No single African institution currently produces a standardised, country-comparable, independently verified fiscal dataset that resolves this ambiguity.

The third gap is capital. Ratha and Mohapatra, writing for the World Bank in 2011, estimated African diaspora annual savings at approximately USD53bn, of which USD30.5bn from the Sub-Saharan African diaspora (Migration and Development Brief No. 14). No post-2020 academic update applying the same methodology exists. Remittances to Sub-Saharan Africa reached approximately USD54bn in 2023 (World Bank Migration and Development Brief No. 40) and are projected at roughly USD57 to 58bn in 2025 (KNOMAD); all-Africa flows are estimated to have crossed USD100bn in 2024 (World Bank/KNOMAD estimates). Remittances are flows to households rather than a stock of investable savings, so the 2011 figure is treated as an unchanged conservative floor rather than extrapolated forward. LemFi, a diaspora fintech founded in London, already processes USD1bn in monthly transactions across remittance corridors. The rails exist. The instruments that would convert diaspora savings into African productive assets at scale, with governance structures diaspora investors would trust, do not.

These three gaps are the architecture’s reason for existence. What follows is the engine: a locked fiscal definition, a two-gate system combining measurement with codified governance overrides, four tier bands that determine Pool Three access terms, and a migration mechanism that moves countries between tiers annually on published data.

2. The Locked Definition

Every figure in this architecture traces to a single metric applied uniformly across all 55 African Union members: gross interest payments divided by total government revenue excluding grants, on a central government basis. The architecture uses this ratio rather than the more familiar debt-to-GDP because debt stock is a balance sheet number while interest-to-revenue is a cash flow number. A government does not default because its debt is large. It defaults because it cannot service the debt from current income. Cash flow kills. Grants are excluded from the denominator to isolate structural tax-base capacity from external donor support outside the sovereign’s structural control. The ratio determines which tier a country occupies. The tier determines Pool Three access terms. Pool One and Pool Two are implemented by countries and jurisdictions independently of tier status; their implementation moves countries through the tiers by improving the fiscal indicators Gate One measures. Part III specifies all three layers; Part I set out the absorption industries into which deployment flows.

The choices embedded in that sentence matter.

Rwanda illustrates the calibration. Its ratio is 13.7 per cent excluding grants (11.6 per cent including grants). The excluding-grants figure governs tier assignment. Rwanda’s organic fiscal discipline is what secures Tier 1 Anchor status.

The central government basis is the default because that is the level at which most African sovereigns manage public debt and at which most IMF Article IV fiscal tables report. Where a material gap exists between central and consolidated government ratios, exceeding three percentage points, the architecture publishes both and specifies which governs pricing. The pricing input is always the level of government that holds the debt and would access the capital pools.

This matters most for Nigeria. IMF Country Report 25/157 publishes two fiscal tables. Table 4 covers the Federal Government alone, the entity that issues FGN bonds, Treasury bills, and Eurobonds. Table 5 covers consolidated general government (FGN plus 36 states, the FCT, and 774 LGAs). Federal interest-to-revenue is approximately 41 per cent (Table 4), placing Nigeria in Tier 4; consolidated interest-to-revenue is approximately 20.8 per cent (Table 5). The FGN holds approximately 92 per cent of total public debt; state and LGA revenue enters the consolidated denominator without proportional debt service. The architecture uses the federal basis because FGN is the entity that borrows, issues, and must service. The consolidated figure is published alongside as context.

The tier-determining ratio uses a three-year rolling average of the three most recent fiscal years available. Single-year figures are published as current indicators and inform quarterly monitoring but do not determine tier assignments. Single-year metrics destroy pricing stability: Angola’s ratio swings between 31.9 and 23.8 per cent on kwanza movements. The rolling average delivers predictable classifications for normal FX volatility. Acute shocks exceeding 25 per cent year-on-year trigger the FX liquidity circuit breaker under Gate Two, overriding the rolling average for interim reassessment.

Every figure carries a confidence grade (A, B, C, or X, from verified IMF fiscal table to no verifiable source) and a full source citation. Rwanda’s 13.7 per cent, for example, carries Grade A: extracted directly from IMF CR 25/319, page 8, fiscal table. A ratio derived algebraically from IMF aggregates carries Grade B. A ratio estimated from news sources where no current Article IV exists carries Grade C. Grade X applies where no verifiable fiscal data exists, including South Sudan in reporting-gap years, Sahrawi Arab Democratic Republic, and any country whose statistical system collapses in conflict. Grade X countries sit Below Floor until data resumes. During the data gap, the Canary Codex Lab provides technical assistance targeted at statistical capacity rebuilding, with a published pathway back to Tier 4 when verifiable data is restored. Territories outside the African Union, including Somaliland, fall outside scope by design. Source hierarchy: IMF Article IV fiscal tables, programme reviews, national budget documents, then World Bank and AfDB datasets. Part III publishes the complete verification chain.

3. Two Gates

A single locked metric applied without exception would be clean but dangerous. A country with a low interest ratio and a collapsing economy would receive anchor pricing. A country recovering from restructuring would be treated identically to a country that never defaulted. The architecture avoids both errors by operating as a two-gate system.

Gate One is the locked metric. It determines the baseline tier assignment for every country. It is objective, formula-driven, and universally applied. Four tiers, calibrated against the IMF SSA median of approximately 12 to 14 per cent, as set out in Table 2. Countries in fiscal distress or data opacity sit Below Floor, with no Pool Three access until their metrics improve to Tier 4 or better.

The Tier 1 ceiling at 15 per cent positions anchor pricing for countries whose interest burden sits near or below the continental median. These thresholds are institutional operating bands calibrated against the IMF SSA median, not econometric optima derived from a regression. The Phase One pilot will back-test the 15 per cent threshold against observed SSA default and stress events from 2000 to 2025, with results published transparently.

Each tier carries a governance discount, the reduction in borrowing cost that the architecture’s independent measurement and multi-country diversification are designed to produce. The discount is framed as a testable hypothesis. The literature on patriotic discounts (Ketkar and Ratha, 2010) supports 50 to 100 basis points under conditions of strong governance. Israel Bonds provide the strongest empirical anchor at 100 to 200 basis points below comparable sovereigns over multiple decades; India’s 1998 Resurgent India Bonds showed a more modest and less consistent discount, with the wider patriotic discount literature cautioning against cross-case generalisation. The Phase One pilot will test three scenarios: 100, 150, and 200 basis points. If the achievable discount falls below 100 basis points, the architecture’s value proposition narrows to measurement and transparency rather than pricing compression. That outcome is published transparently.

Consider what even the conservative scenario means. A pension fund manager in Nairobi allocating to Kenyan government securities sees instruments priced by the same external methodologies that produced the Africa premium. The Africa Credit Rating Agency, AU-sponsored and Mauritius-headquartered, is scheduled to issue its first sovereign ratings in June 2026, replicating the sovereign rating function under African ownership. The Canary Codex engine addresses a different layer. It is not a credit rating agency. It is a tiered fiscal filter bound to Pool Three access terms, measuring which African sovereigns have earned specific deployment conditions rather than pricing probability of default. AfCRA rates. The Canary Codex engine assigns access. The two are complementary. The measurement is the public good.

Gate Two is override and eligibility governance. Five codified triggers can modify a Gate One assignment.

The first is the fiscal fragility override. A country with interest-to-revenue below 15 per cent but exhibiting severe fragility receives a Tier 1 Conditional designation with restricted pool access until fragility indicators improve. The triggers are specific and any one is sufficient. Fiscal deficit exceeding 10 per cent of GDP for two consecutive years. IMF sovereign risk rating of High or In Distress. Reserve cover below three months of imports (IMF reserve adequacy benchmark). Tax-to-GDP ratio below 13 per cent, the IMF tax capacity tipping point above which sustained growth and state capacity durably improve (Gaspar, Jaramillo and Wingender, IMF WP/16/234). Data opacity preventing independent verification.

Algeria illustrates why the primary metric alone is insufficient. Algeria qualifies for Tier 1 with a ratio of approximately 1.8 to 3.0 per cent in FY2024/25 (IMF CR 25/270), one of the lowest on the continent. But Algeria’s fiscal deficit stands at 13.9 per cent of GDP, its sovereign risk rating is High, and hydrocarbon revenues are declining structurally. Anchor status on a low interest ratio while the fiscal position deteriorates would fail every investor looking past the headline. Algeria receives Tier 1 Conditional.

Ethiopia presents a different problem. Its interest-to-revenue ratio of approximately 9.5 per cent (IMF CR 25/188) also qualifies for Tier 1. The low ratio is real. But it appears to reflect concessional borrowing at sub-2 per cent average interest rates rather than fiscal discipline. Ethiopia’s tax-to-GDP ratio of 7 to 8 per cent is among the lowest globally, meaning even modest interest payments would consume a larger share if the debt portfolio shifted toward market rates. The fragility is in the revenue base, not the interest cost. Ethiopia receives Tier 1 Conditional until tax-to-GDP exceeds 13 per cent, aligned with the IMF tax capacity tipping point. The path out is specific: whichever trigger placed a country in Conditional governs exit, and meeting that trigger threshold for two consecutive annual assessments lifts the designation. The architecture tells countries exactly what they must fix and exactly when the restriction ends.

The second is the fiscal distress override, activated when total debt service, interest plus principal, exceeds 100 per cent of total government revenue regardless of the interest-only ratio. This trigger catches refinancing walls and bullet-maturity events that the primary metric alone misses. It is one mechanism for the architecture to say no to a country whose headline ratio looks workable but whose cash flow reality does not.

The third is a Watch Negative designation, activated when a country’s Gate One ratio is within Tier 1 but the underlying fiscal trajectory places the ratio above 15 per cent within five years. Botswana is Tier 1 on current data with an interest-to-revenue ratio of approximately 3 to 5 per cent, among the lowest on the continent. But diamond revenues are declining structurally, the fiscal deficit is projected at 9.3 per cent of GDP in FY2025/26 and 8.9 per cent in FY2026/27 (S&P, March 2026), and S&P downgraded the sovereign to BBB negative on 13 March 2026. Net debt is projected to reach 37.4 per cent of GDP by 2029 from a recent net asset position. The interest-to-revenue trajectory points toward the Tier 1/2 boundary by the end of the decade. Botswana carries Tier 1 Watch Negative, signalling that the classification is not permanent.

The fourth is emergency reclassification, activated upon sovereign default or restructuring announcement, IMF programme suspension for breach of quantitative performance criteria, or discovery of material fiscal misrepresentation. It requires a published determination by the measurement authority with full documentation. Senegal is the test case. Senegal’s interest-to-revenue ratio of approximately 19 to 22 per cent would place it in Tier 2 on the primary metric alone. The Court of Auditors revealed hidden borrowing initially estimated at USD7bn, subsequently expanded toward USD13bn. The IMF now shows total public sector debt at 132 per cent of GDP at end-2024 (IMF PR 25/360, 6 November 2025). The IMF Mission Chief for Senegal, Edward Gemayel, described the hidden debt as of a magnitude he had “never seen” in Africa. The USD1.8bn Extended Credit Facility was suspended in June 2024; S&P downgraded Senegal to CCC+ in November 2025. Senegal sits Below Floor under emergency reclassification despite being a WAEMU anchor. The arithmetic and the triggers admit no exception.

The fifth is the FX liquidity circuit breaker, activated when a sovereign experiences year-on-year cumulative currency depreciation exceeding 25 per cent against the weighted average of its external debt denominations. Acute FX collapses destroy hard-currency debt service capacity instantly, long before the three-year rolling average reflects the damage. The circuit breaker overrides the rolling average for interim reassessment within 90 days of the trigger event. Angola’s 2023 kwanza devaluation of approximately 39 per cent against the US dollar between May and June 2023 is the case reference. A shock of that magnitude would have triggered interim reassessment under this rule, separate from the normal year-to-year volatility the rolling average is designed to handle.

Gate Two overrides operate downward only. A country with a favourable Gate One assignment that exhibits fragility, trajectory risk, misreporting, or acute FX pressure receives a stricter designation. A country whose Gate One assignment already reflects a difficult fiscal position does not receive an upgrade. The architecture prices the arithmetic; it does not soften arithmetic outcomes through narrative context. Kenya’s Tier 3 position reflects the interest burden accumulated through sustained market-rate borrowing that Kenya has continued to service. Nigeria’s Tier 4 on the federal basis reflects FGN debt service reality on the entity that issues and must repay. Both countries move through tiers by improving the underlying arithmetic. Neither receives an upward override for structural context alone.

4. How the Engine Moves

Tiers are reassessed annually, using the most recent audited fiscal data, published simultaneously for all participating countries on a fixed date. African fiscal calendars are not synchronised. The annual reassessment uses the most recent completed fiscal year for each country. The architecture publishes what is available, grades its confidence, and updates as new data arrives. Where a country’s most recent audited fiscal data is older than 24 months, the confidence grade drops one band and the tier assignment carries a Watch designation until current data is published. This staleness penalty incentivises rapid financial reporting and prevents prolonged audit lag from masking deterioration. Quarterly monitoring indicators signal trajectory but do not trigger tier migration between annual reassessments.

Countries exiting debt restructuring present a specific problem. During grace periods, a country may exhibit an artificially low interest burden because restructured instruments have not yet stepped up to market-rate coupons. Kenya’s interest-to-revenue ratio sits in the Tier 3 range, servicing a heavy burden at significant fiscal cost. Zambia, post-restructuring under the G20 Common Framework, shows a materially lower near-term service burden (IMF CR 25/225), with the interest-only component lower still. On the primary metric, post-restructuring Zambia sits closer to Tier 1 while Kenya sits in Tier 3. Investors may reasonably prefer Zambia’s clean slate. But the clean slate reflects concessional grace period terms that will escalate when coupon step-ups activate through 2028-2031. Without a fairness mechanism, the architecture would assign Zambia anchor pricing on temporary terms and then violently reclassify it when those terms reset.

The architecture applies a three-year smoothing adjustment. In the first year following restructuring, the tier assignment blends the current ratio at 40 per cent with the pre-restructuring ratio at 60 per cent. Year two shifts to 70/30. Year three returns to 100 per cent current data. The three-year window tracks typical coupon step-up schedules in Common Framework and London Club restructurings, where concessional terms generally expire within that horizon. The weights approximate the net present value of step-up coupons over a typical Common Framework restructuring profile; this recognises that headline ratios during grace periods misstate the true service burden. Weights are fixed ex ante and published before the first annual reassessment; if the Phase One pilot demonstrates recalibration is needed, any revision is published before implementation. The smoothing ensures that countries which stretched to honour their obligations are not penalised relative to countries that did not. Ghana receives identical treatment.

5. What the Engine Is

A question this invites: why fiscal as the gate? Enterprises and projects vary enormously in performance within any sovereign. The answer is that the filter governs the deployment environment, not project selection. A high-performing enterprise in a sovereign facing fiscal distress sees returns extracted through capital controls, emergency taxation, FX restrictions, or payment freezes long before operating performance matters. The sovereign filter defines the envelope within which project-level selection becomes meaningful. Levels 2 and 3, which track deployment and sectoral absorption, operate inside that envelope, not as a substitute for it.

The filter does not guarantee absorption. A country can qualify for Tier 1 on the fiscal metric while its financial sector remains shallow, dollarised, or concentrated in sovereign exposure. Pool Three eligibility is preserved at the tier level, but realised allocation depends on the absorption capacity that Pool One domestic financial sector reforms build over time. A Tier 1 country with deep private credit markets, extended tenor structure, and working collateral infrastructure can absorb Pool Three flows at scale. A Tier 1 country without these conditions cannot. Fiscal discipline is the gate. Absorption capacity is what determines how much the gate admits.

Tiered access mechanisms exist. IDA graduation, the IMF’s PRGT and GRA windows, AfDB pricing, and EU Structural Funds all use differentiated access on development metrics. What is new in this architecture is not the tiering but what the tiers unlock.

The tier system built in this essay is Level 1 of a four-level measurement engine. Level 1 measures sovereign creditworthiness: the fiscal filter. Part III builds the remaining three levels. Level 2 measures what borrowed capital actually builds, verified through satellite imagery and blockchain-tracked deployment alongside government self-reporting. Level 3 measures whether capital reaches the absorption industries Part I identified. Level 4 measures whether deployed capital strengthens household balance sheets: youth employment, land title registration, financial inclusion, income diversification, enterprise formation. Mainstream sovereign risk frameworks, including S&P, Moody’s, and Fitch methodologies and the IMF DSA and AfDB sovereign assessment framework, rate instruments at issuance on fiscal and institutional variables. Part II’s Level 1 fiscal filter operates in that space and contributes confidence-graded African-owned verification, the Productive Asset Discount, and codified Gate Two overrides. Levels 2, 3, and 4, built in Part III, extend the architecture beyond rating at issuance into outcomes-conditional pricing across staged Pool Three tranches. Subsequent tranche terms reflect prior-tranche verification: Level 2 confirms what capital built, Level 3 confirms whether it reached the absorption industries Part I identified, Level 4 confirms household impact including youth employment, financial inclusion, land title, and enterprise formation. Pricing moves with verified outcomes rather than with promises at issuance. This is not a rating function. It is a capital deployment architecture that prices sovereigns differently depending on what their prior borrowing actually achieved. That mechanism is the architecture’s distinctive contribution relative to both mainstream frameworks and AfCRA, and it is what distinguishes this work from a repackaging of IMF fiscal tables under African ownership.

Two uniform refinements extend the locked Gate One definition. They are not price adjustments. They are shaping mechanisms. The architecture uses pricing to change how sovereigns borrow, not only to assess how they have borrowed.

The first is a Productive Asset Discount. Mainstream sovereign risk frameworks address the productive-versus-consumptive distinction qualitatively; the PAD operationalises it in pricing. Interest tied to verified self-liquidating revenue-generating assets is discounted from the Gate One ratio. A sovereign borrowing to build a revenue-generating port or energy asset is not pricing-equivalent to one borrowing to fund recurrent expenditure. Sovereigns seeking Anchor terms have a direct structural incentive to borrow productively. The pricing mechanism shapes the borrowing.

Phase One operates in beta form. The beta does not accept the blunt IMF Article IV capital expenditure classification, which groups a non-revenue-generating parliament building with a self-liquidating toll road under the same CapEx line. It requires project-level verification through national audit office classifications or equivalent African development finance institution project assessments. The discount applies only where the underlying debt has ring-fenced revenue accounts, audited cash flow segregation, and covenant-protected servicing tied to the specific asset’s revenue stream. The empirical record on ring-fencing in African sovereigns is uneven: Chad’s 2000s oil revenue framework and Ghana’s Petroleum Revenue Management Act both show how political override can erode integrity in practice. Level 2 satellite verification and Level 3 blockchain tracking in Phase Two upgrade the detection layer.

The second refinement is a denomination weighting. Hard-currency interest carries heavier weight than local-currency interest, because a central bank can theoretically monetise domestic obligations in ways it cannot with external hard-currency debt. Denomination weighting follows the Phase One pilot.

Five elements combine that do not elsewhere: a locked sovereign fiscal gate applied across 55 AU members, confidence-graded African-owned verification, real-time deployment and household-outcome tracking across Levels 2 through 4, a productive-versus-consumptive debt distinction via the Productive Asset Discount with ring-fencing conditions, and a measurement-plus-incentive architecture binding domestic reform to Pool Three access under mission-oriented governance. Each element has prior art. The integration does not.

Part I’s central claim demands this filter. The return pathway must be enterprise-led. A continent adding over 150 million to its working-age population by 2030 needs diaspora members as enterprise builders in countries whose fiscal discipline creates space for productive investment. The filter identifies which countries have earned that space. The strategic question, as Part I argued, is no longer which jobs to seek but which productive assets to own.

When applied continent-wide, as Part III does in full, the engine produces uncomfortable results. Fewer than a third of the 55 members qualify for Tier 1 Anchor on the uniform excluding-grants metric, and several of those carry Conditional or Watch designations. Nigeria, the continent’s largest economy, sits in Tier 4 on the federal government basis that governs its pricing. Egypt, at 84 per cent in FY2023/24, is the continental outlier by a factor of six. Senegal, a WAEMU anchor, is Below Floor. The engine says what the locked definition produces when applied uniformly.

The measurement is a public good. This requires institutional form. The measurement authority is structured as an African-sponsored independent entity. Political mandate sits with the AU Specialized Technical Committee on Finance, Monetary Affairs, Economic Planning and Integration; the AfCFTA Secretariat provides convening support on trade-adjacent questions but does not co-own technical governance. The board is nominated by African development finance institutions, African research institutions, and private-sector representatives, with technical governance separate from direct AU supervision. Funding is African-sourced by design: anchor commitments from African DFIs, secondary contributions from African institutional investors, and subscription revenue from African financial institutions using the framework. External non-African donor funding is excluded to preserve independence from external policy conditionality. Methodology, source citations, and confidence grades are published openly under a permissive public licence. Sovereigns hold the right of appeal with reasoned response, methodology changes require notice and a consultation window, and the authority is externally audited annually. These provisions make the architecture durable across political cycles.

This is not the development model. It is the fiscal and governance gate inside a wider mission-oriented African capital platform. Part III specifies the pools that deploy into the absorption industries Part I identified. Part IV addresses safeguards. Part V specifies the minimum viable product. Together these parts define the missions, reciprocal conditions on deployed capital, public value capture mechanisms, and state capability building. The filter clears fiscal space. The mission work fills it.

The architecture does not argue that Africa is mispriced and deserves better. It builds the measurement system, publishes the results, and lets the pricing follow. Countries that fail the metric are excluded by arithmetic, not by editorial choice. Structure before sentiment.

6. What Comes Next

Part III applies the filter across the 55, constructs the three capital pools, and takes up the deployment question. The absorption universe Part I opened is being refined through parallel workstreams. The Mineral Trilogy maps Africa’s position in the critical minerals value chain and the non-extractive manufacturing path that sits alongside it, beginning with Forced Choice on coalition logic. The Misaligned Transition disaggregates energy into firm and variable components and takes up the financing architecture behind each, previewed in EVs Are the Last Mile, Not the First. Part III inherits the taxonomy these workstreams develop.

Part III also constructs the three capital layers. The structure is asymmetric. Pool Three is the architecture’s capital pool in the conventional sense. Diaspora capital sits outside African sovereign regulatory reach, and a vehicle that channels it into productive enterprise brings new capital into African deployment. Pool Three operates through a two-strand framework and a coalition of the eligible governing diaspora capital access. It starts with a Phase One vehicle and scales to the African Diaspora Reserve over Phase Two and Three. Pool One and Pool Two are not capital pools in this sense. They are preconditions. Pool One is a programme of domestic financial sector reforms drawing on the Structure Before Sentiment series, combined with a measurement layer that publishes country-comparable indicators of capital market structure. Its policy toolkit overlaps with established IMF, World Bank, and AfDB recommendations; its distinctive contribution is measurement, incentive architecture, and African institutional ownership. Pool Two is a regulatory reclassification mechanism. It shifts pan-African institutional capital, a pension and insurance base of roughly USD775bn (OECD/AFC 2025), from ‘foreign’ to ‘pan-African’ classification in allocation caps. Bicameral governance and sub-regional corridor phasing structure the reclassification. Phase One mobilisation potential, at regulatory ceilings around 5 per cent, sits in the order of USD30 to 40bn across eligible jurisdictions. Parts IV and V address safeguards and specify the minimum viable product.

The filter is built. Part III puts it to work.

The 2026 Outlook continues in Part III: Screen, Score, Deploy

Sources

Sources: IMF WP/23/130 (Gbohoui, Ouedraogo, and Some 2023), Journal of African Trade Vol. 11 (Olabisi and Stein 2024), AfDB African Economic Outlook 2025, IMF Sub-Saharan Africa Regional Economic Outlook October 2025, World Bank Migration and Development Brief No. 14 (Ratha and Mohapatra 2011), World Bank Migration and Development Brief No. 40, KNOMAD 2025 remittance projections, DMO Nigeria Annual Report 2024, IMF Country Reports (CR 24/316 Kenya, CR 25/157 Nigeria, CR 25/188 Ethiopia, CR 25/189 Ethiopia Selected Issues, CR 25/225 Zambia, CR 25/270 Algeria, CR 25/319 Rwanda, CR 24/274 Egypt), IMF Selected Issues Paper 2025/108 (Hegab, Ethiopia’s Tax System), IMF Press Release 25/360 (Senegal, 6 November 2025) and RFI interview with IMF Mission Chief Edward Gemayel, 6 November 2025, OECD and Africa Finance Corporation, Institutional Investment in Africa (2025), Ketkar and Ratha (2010) on diaspora bond discounts, diaspora bond empirical precedents (Israel Bonds; Resurgent India Bonds 1998), S&P Global Ratings (March 2026 Botswana sovereign action; November 2025 Senegal sovereign action), Banco Nacional de Angola 2023 exchange rate data, Canary Compass architecture v7.0, From Grievance to Design (Canary Compass, December 2025), Forced Choice (Canary Compass, Mineral Trilogy Part I), EVs Are the Last Mile, Not the First (Canary Compass, December 2025), Structure Before Sentiment series Parts 1-4 (Canary Compass, November-December 2025).

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu or X @InfinitelyDean.

AI-generated image: Read These, She Said

The school sent my son’s report last week. His Swahili mark came in at around half of what the school expects at his level. Below where any parent wants to see their child sitting.

Then I read the teacher’s comments.

Focused. Disciplined. A cooperative learner. Strives for excellence. Demonstrates admirable leadership by guiding peers and participating actively in group tasks.

I sat with both pages longer than I should have needed to.

I was once that boy. In primary school I was good at Swahili. My mother went to the PTA meetings and the teacher would tell her Dean is going to do well.

Then high school happened and something in my head flipped. I stopped understanding the language. I stopped enjoying it. I kept studying, but the studying was not landing, and at sixteen that is a powerful argument for giving up on yourself. For four years my mother sat across from a Swahili teacher who had a steady, polished line that gained a little more pity each time. Dean is not quite getting it. Dean struggles. Dean is not a Swahili student.

At the last PTA of the final year, the teacher gave her the definitive version. I don’t think Dean is going to do well for Swahili.

My mother did not argue with him. She went out and bought me practice papers. Not just Swahili. All the subjects. In those days, hawkers sold exam practice papers for school leaving students outside bookshops and at various spots around town. The Swahili paper covered all three components: language, essays, literature. They arrived in my room without ceremony. Read these, she said.

The night before the exam, I was not reading my mother’s paper.

I was with two of my friends and schoolmates at a dining room table. The country was in one of its power rationing seasons. This was long before solar solutions were affordable, so we worked under one of those lamps you charged during the day and ran at night. Somebody had given us a paper and sworn, on whatever is swearable at sixteen, that it was leakage. The real exam, smuggled out.

We had each put in our own work. The paper was supposed to be extra. But when someone hands you what they swear is the real exam, you take it seriously. We went through it, cross-referencing, arguing over answers, nodding wisely. Then we all went home.

Nothing from that paper came up. Not a question, not a proverb, not even a comma. Kenya has a long tradition of leakage. Ours was not leakage. Ours was stationery.

On the morning of the exam I walked into the school library. Students were quizzing each other on methalis. One boy would call out a proverb and three others would fire back the synonym and the meaning. Like a prayer they had been saying for years. Like the catechism we had crammed for two years straight in Catholic school.

I stood at the back and listened.

I did not know a single one of them.

My heart sank, flat, like a stone through still water. I was going to fail this exam. Everyone in this library was going to beat me. The teacher was going to be right.

I left the library. I went to the toilet. I locked the door. I sat down and cried.

I have not told that story to many people. My mother knows. A few of my mentees know. It was the boy version of every Idols contestant who walks in because his mother, his whole life, could not bear to tell him he could not sing. I understand that mother. My eldest son still thinks he can dance.

I prayed. I was spiritual then. I still am now, but back then it was on steroids. I did not pray for knowledge. I prayed because I had run out of anything else.

And something said, quite quietly, go and read the Swahili paper your mother bought you.

The paper was in my bag. So was the Kamusi ya Methali. I had studied the paper before, along with the other subjects, but there are over two thousand proverbs in the kamusi and a sixteen-year-old cannot hold them all. I pulled both out in the toilet stall, still crying, and went back to the paper with the kamusi open beside it. I crammed what I could in the time I had left and walked into the exam hall.

The proverb on the exam was the synonym to the one from my mother’s paper.

I had the synonym and the original. The meaning was fresh. The story wrote itself.

The literature paper came in the afternoon and that one was mine anyway, because the theme was neo-colonial and I have always had strong opinions about people putting on makeup to look like something they are not.

I came out of those final school leaving exams with an A for my worst subject. Two months later, when the results were announced, my best friend told me to my face that they had to be fake. There is no way, he said, you could have beaten me in Swahili. He had a point, statistically. He was also wrong. Because somewhere in his calculation, he forgot my mother.

I have been thinking about my mother this week because of a man on X.

He told me I only ever condemn the previous administration. That my writing on that administration is uniformly negative. He did not call it a verdict. He called it an observation. He was hoping I would hear the difference.

He is right about one thing. When I have written about the 2011 to 2021 period, the lost decade, and especially the 2015 to 2021 years, my conclusions have been almost without exception negative. Money supply, inflation, external debt, the default, the pivot to local debt, private issuances coming due as refinancing walls in someone else’s lap. That work is recent. I began publishing publicly in 2023, looking backwards at a period already closed.

I wrote back that the mirror has no favourites. It only reflects. You can dislike what it shows.

On labour policy, which is not my wheelhouse, I would credit that same administration openly for worker protections among the strongest Zambia has had. If it lands in their favour, it lands in their favour.

He was not reading me. He was reading the red marks on the page and calling them the whole page.

That exchange pointed to a wider habit. A journalist publishes a story sourced from people he will not name. Another outlet cites the journalist. A third cites the second. Each retelling hides the anonymity under one more layer of citation. By the fourth round, the claim reads as consensus, and no one thinks to ask the question the first piece never answered.

I know the habit because I have done the same thing with a public figure I had formed a view on. I had a settled position on him, and I had repeated it enough times that it felt like knowledge.

The evidence did not arrive all at once. It accumulated. For a long time I could fit it inside the position I already held. At some point the position could no longer hold it. Structure before sentiment cuts both ways. If the evidence keeps pointing, I have to let it move me, even when the direction is inconvenient.

I had to credit him on one register. I could not credit him on the rest, and the rest still stands where it stood. Holding both at once is harder than either verdict alone.

Grievance can be very addictive.

Three or four years ago, someone could send me a sharp argument and I would match grievance with grievance for two hours and walk away thinking I had won something. I had not. I had just done a longer version of what the other person was doing.

I do not know when the framework shifted. I had to sit still long enough to notice I was doing it.

Sometimes the reader is not a person.

Sometimes it is the register you have been using to grade the world. The one that circles what is wrong in red and takes the rest for granted.

The habit of only seeing what is wrong does not stay small.

My son will not remember his Swahili mark in twenty years. He will remember whether the adults in his life learned how to read him.

I am writing this because my mother learned how to read me.

I am trying to learn how to read him.

Prefer to listen? Canary Compass is now a podcast.

YouTube: Canary Compass on YouTube

Spotify: Canary Compass on Spotify

New episodes every Tuesday and Saturday, with bonus episodes on Thursday.

Sulphuric acid: the compound beneath the food system and the industrial metals complex.

Price and supply data as of 12 April 2026. Strait of Hormuz developments current to 12 April. This note traces the sulphuric acid deficit through five distinct African exposures: refining architecture in Nigeria, sovereign subsidy exposure in Kenya, agricultural forward-risk in Zambia, hydrometallurgical mining in the DRC, and phosphate production in Morocco.

Three developments since 19 March have broken the arithmetic of the Hormuz closure. The architecture was already visible in my Quick Take of that date, which mapped the downstream cascade from crude oil through sulphur, fertiliser, and fuel to specific African country exposures. What has changed is that three sources of supply for a single industrial chemical have been impaired simultaneously. The implications run from phosphate processing in Morocco to copper leaching pads in the DRC to planting schedules across East and Southern Africa.

The chemical is sulphuric acid. The three doors are now closed.

Day 44

A ceasefire between the United States and Iran was announced on 8 April. It did not hold in substance. The Strait of Hormuz has not reopened to normal commercial shipping. On 12 April, after 21 hours of negotiations in Islamabad failed to produce an agreement, President Trump announced that the US Navy will blockade the Strait, preventing access to Iranian ports while permitting transit to non-Iranian destinations (CENTCOM, 12 April). Separately, Trump directed the Navy to seek and interdict any vessel in international waters that has paid Iran’s toll. The US has ordered mine clearance to begin. Two destroyers (USS Frank E. Peterson and USS Michael Murphy) transited the Strait on 11 April as part of a broader CENTCOM operation to establish a safe passage corridor.

The closure has escalated, not resolved. For 43 days, Iran controlled access. Now both the US and Iran claim authority over the same waterway. Crude oil has been partially rerouted through Saudi Arabia’s East-West pipeline (restored to 7m bpd capacity, Al Jazeera, 12 April) and the UAE’s ADCOP corridor to Fujairah. No equivalent bypass exists for elemental sulphur, LNG, ammonia, urea, or finished fertiliser. These commodities transit the Strait by ship or they do not transit at all.

Even under a best-case scenario in which commercial shipping resumes within weeks, the physical pipeline has been drained. The last cargoes loaded before the war are arriving at their destinations now. After those clear, four to six weeks will pass before the first post-reopening cargoes reach end users. For sulphur, the lag is longer: it must be shipped, converted to acid, transported to processing plants, and applied. Reopening the Strait is the beginning of a recovery, not its conclusion. Hapag-Lloyd estimates normalisation at six to eight weeks minimum. Other shipping analysts place it at months.

In my 19 March assessment, I described the IRGC toll mechanism as a uniform feature of Iran’s selective Hormuz doctrine. That framing was too simple, and the situation has since moved past it. Ed Finley-Richardson challenged the toll claim on structural grounds. The IRGC is OFAC-designated. Compliant owners cannot pay a sanctioned entity. The USD2m figure traced to two ships early in the conflict with no credible subsequent confirmation. The New York Times, citing US officials, reported on 10 April that Iran mined the Strait and left a single cleared corridor open to vessels willing to pay for passage in yuan or cryptocurrency. Trump’s 12 April blockade order now closes the other side of the trap: any vessel that paid Iran’s toll faces US interdiction.

Four categories of vessel are now operating:

The grey middle, which produced a thin trickle of transits well below normal levels, is now squeezed from both directions. Iran charges for entry. The US interdicts for having paid. Commercial shipping that needs certainty from both sides has neither.

The Three Doors

Sulphuric acid sits inside fertilisers, copper leaching, nickel and cobalt processing, oil refining, lead-acid batteries, and a wide range of chemical chains. Craig Tindale describes it as one of the base layers of civilisation. Remove flow at the margin and the effects propagate through industrial systems we never see until they seize up. Acid prices had already surged roughly 500 per cent before the conflict began, according to Tindale’s market analysis, reflecting pre-existing tightness in the global sulphur market.

Three sources of supply have now failed simultaneously.

The first door is Hormuz. Industry estimates place Persian Gulf seaborne sulphur exports at approximately 44 per cent of the global total (S&P Global Trade Atlas). The Gulf states are among the largest exporters of elemental sulphur, a byproduct of oil refining and gas processing. When Gulf refineries shut and gas plants close, sulphur supply does not become expensive. It stops. The standard conversion ratio is one tonne of elemental sulphur to approximately three tonnes of sulphuric acid. The 15 to 20 million tonnes of sulphur that normally transit the Strait therefore represent 45 to 60 million tonnes of acid equivalent. Tindale places the range at 40 to 70 million tonnes when including recovered acid and direct acid shipments. That supply has been effectively unavailable since 28 February.

The second door is China. Bloomberg reported on 10 April that Chinese authorities will halt sulphuric acid exports from May. The ban covers smelter acid from copper and zinc processing and sulphur-based acid. Only electronic-grade acid is exempted. Ivanhoe Mines founder Robert Friedland notes China exported 4.6m tonnes of sulphuric acid in 2025 according to preliminary Chinese customs data. Major destinations included Chile (32 per cent), Indonesia (15 per cent), Morocco (12 per cent), Saudi Arabia (12 per cent), and India (9 per cent). Chinese exports were already falling sharply in early 2026, with January-February volumes at roughly half the prior-year level. The door was narrowing before Beijing shut it.

The third door is Russia. Ukrainian strikes on Russian refining and export infrastructure have reduced Russian sulphur output by an estimated one million tonnes, according to Tindale’s assessment. Russia is among the world’s largest sulphur producers. The strikes were designed to impose costs on Moscow, not to tighten the global sulphur balance. The global sulphur balance does not distinguish between intended and unintended causes of disruption.

Three sources. All three impaired in the same window. This convergence did not exist on 19 March.

The Math

Tindale walks the stoichiometry (the calculation of reactants and products in a chemical process) in public. The numbers are large enough to matter at the level of food security for entire regions.

In the wet-process route for phosphate fertiliser production, sulphuric acid is the acidulating agent used to convert phosphate rock into phosphoric acid. On straight stoichiometry, 3m tonnes of H₂SO₄ yield approximately 1.8m tonnes of phosphoric acid, equivalent to roughly 1.30m tonnes of P₂O₅. That supports approximately 2.8m tonnes of DAP-equivalent fertiliser at 46 per cent P₂O₅. At a common field application rate of 60 kilograms of P₂O₅ per hectare, that covers roughly 22 million hectares. At global average cereal yields and caloric conversion, that is approximately the area required to feed 330 million people for a year. The conversion from hectares to people fed is illustrative, but the order of magnitude holds across reasonable yield and calorie assumptions.

I verified the chemistry independently. The industrial reaction uses fluorapatite, not pure tricalcium phosphate, and the stoichiometry holds. Tindale’s rounding is minor and consistently in the same direction.

Now scale it to Hormuz. If the Strait locks in 40 to 70 million tonnes of acid equivalent per year, the phosphate fertiliser production at risk is measured in billions of people’s food supply.

The gas and nitrogen fertiliser channel, which I mapped in detail on 19 March, has not improved. Urea spot prices remain at approximately USD700/t. Chinese export quotas on nitrogen-potassium blends are unlikely to be lifted before August. With the Gulf physically blocked and China halting acid exports, the market has simultaneously lost its two largest sources of replacement supply. The sulphur chain breaks phosphate production. The gas chain breaks nitrogen production. Trade policy restricts substitution between them.

Morocco: The Phosphate Exposure

This is where China’s second door lands hardest on Africa. Morocco is the world’s largest phosphate producer, and OCP Group’s processing operations at Jorf Lasfar and Safi are among the largest consumers of sulphuric acid on the continent. China shipped 12 per cent of its 4.6m tonnes of acid exports to Morocco in 2025. That supply is now halted from May. Morocco also sources sulphur from the Gulf, which is blocked. OCP has captive acid capacity from its own sulphur-burning plants, but the margin of insulation depends on how much imported feedstock those plants require. A sustained disruption to both Gulf sulphur and Chinese acid tightens the input chain for the world’s most important phosphate exporter, compounding the global fertiliser deficit at its source. Morocco is also a PAPSS member.

The Net Exporter Fallacy

There is a persistent assumption that net oil exporters are shielded from global price shocks. Oil is a globally priced commodity. No country sells oil to itself at a domestic discount.

The United States produces and refines. Its exposure is price, partially buffered by SPR releases and domestic refining capacity. Nigeria produces but cannot adequately refine. Its exposure is price and structure simultaneously. The mechanics of Nigeria’s refining gap, including the Dangote throughput shortfall and the reimport paradox, are documented in my 19 March assessment.

The question that has sharpened is settlement. Why has Dangote not moved to settle intra-African fuel trade through PAPSS?

The Pan-African Payment and Settlement System is now live in 19 countries with over 160 commercial banks and 15 national switches connected, according to Afreximbank’s February 2026 disclosures and TechCabal reporting. It has launched three products: the Instant Payment System, the PAPSS African Currency Marketplace (PACM, launched July 2025 to address the USD5bn currency bottleneck in intra-African trade), and the PAPSSCARD. It settles cross-border transactions in local currencies within 120 seconds. Nigeria, Kenya, Zambia, Morocco, Algeria, Egypt, and Tunisia are all connected.

Fuel is among the largest intra-African trade flows denominated in foreign currency. Dangote has refining capacity and PAPSS has settlement infrastructure, neither of which was operational at current scale during the last supply shock. Both are operational. Neither is connected to the other.

Connecting Dangote to PAPSS for fuel settlement requires more than a technical switch. It requires overcoming bilateral FX liquidity constraints, Afreximbank net settlement limits, and Central Bank of Nigeria regulatory approvals for large-scale commodity transactions. These are real obstacles. But PAPSS has been live since 2022 and the failure to pilot even a single cargo suggests a lack of political will to test the architecture, not a technical impossibility.

Tchakarova’s systemic analysis documents China’s yuan energy corridor accelerating during the crisis. Twenty-three institutional participants are now processing yuan-denominated energy transactions, up from twelve on 22 March. Beijing has been building this architecture since at least 2018, when it launched yuan-denominated oil futures on the Shanghai International Energy Exchange. The Hormuz closure is accelerating a programme that predates it by years. Africa has the equivalent infrastructure and is not using it.

The cost of inaction is quantifiable. A single 30,000-tonne Dangote petrol cargo settled via PAPSS PACM would remove approximately USD35-40m in FX demand from the correspondent banking system and settle in under 120 seconds. Afreximbank confirmed in February 2026 briefings that the remaining gate is a one-time CBN commodity-settlement approval and bilateral FX netting agreements with the importer’s central bank. China required 23 institutional participants to move energy trades into yuan in the first six weeks of the crisis. Africa already possesses the rails. The delay is a revealed preference for USD intermediation even under blockade conditions.

The minimum step is a pilot settlement of a single Dangote fuel cargo through PAPSS between Nigeria and a neighbouring importer. The commercial parties exist. The rails exist. What is required is a decision.

African Agricultural Exposure: The Next Season, Not This One

The main maize crop across most of East and Southern Africa was planted between October and December 2025. That crop is in the ground. Kenya’s long rains season is underway, and fertiliser for the current planting was largely procured before the conflict began. Zambia’s primary maize harvest is being collected now.

The disruption window does not hit the current crop. It hits the next one.

Zambia. The winter wheat crop planted from May to August falls directly inside the disruption window. More consequentially, procurement for the 2026/27 FISP season will be tendered at prices shaped by the current supply environment.

Zambia’s maize story is more complex than the fertiliser transmission alone. The country sits at an unusual junction: maximum current abundance meeting maximum forward risk.

The ZamStats Crop Forecasting Survey for the 2024/25 season (completed) provides the verified baseline:

The 2025/26 season (being harvested now) has no government crop forecast yet. Supply chain practitioners with direct visibility into production volumes provide a forward projection:

The consumption figures reflect a genuine disagreement between government data and practitioners in the value chain. ZamStats reports national requirements of 3.54m tonnes including human consumption, industrial use, seed retention, and reserve replenishment. Practitioners with direct visibility into the supply chain place total consumption, including industrial and feed, at 2.5m tonnes. Both claim to be measuring the same thing. The gap of over one million tonnes is itself a signal: if the baseline consumption figure is contested by that margin, the reliability of any surplus calculation derived from it is limited.

The FRA carryover traces to a specific event. In September 2025, President Hichilema directed FRA to “buy all the maize.” FRA purchased between 1.6 and 1.7m tonnes from the 2024/25 harvest (Presidential Delivery Unit, March 2026; Zambian Observer, November 2025), well above both its financing envelope and its storage capacity.

The storage constraint compounds the problem. Zambia’s secure national storage capacity is approximately 1.5m MT. FRA is holding 1.6 to 1.7m MT, with hundreds of thousands of tonnes under tarpaulins (Zambian Observer, November 2025). The system was already straining before the new harvest arrives.

The government has allowed maize exports, itself a signal of the surplus magnitude. Zambia has historically restricted or banned exports to protect domestic food security. But the regional picture makes exporting difficult. All three major Southern African producers are sitting on surpluses simultaneously.

The reserve-building imperative extends beyond Zambia. All three major Southern African producers are sitting on surpluses at precisely the moment when two distinct threats are converging on the 2026/27 planting season. The first is the Hormuz fertiliser cascade traced throughout this essay: sulphuric acid, phosphate, and nitrogen supply chains all impaired simultaneously, with procurement for the next season’s inputs already being priced in this environment. The second is the El Nino probability. NOAA’s Climate Prediction Center gives a 62 per cent chance that El Nino conditions emerge by June-August 2026 and persist through at least the end of the year. The International Research Institute for Climate and Society places the probability higher, at 72 to 80 per cent through late 2026, with a one-in-three chance of a strong event exceeding +1.5°C (NOAA CPC, March 2026; IRI, March 2026). Several leading forecast models are projecting intensity levels consistent with a potential super El Nino by winter 2026/27. The 2023/24 El Nino produced severe drought across Zambia, Zimbabwe, and Malawi. A comparable or stronger event during the 2026/27 growing season would hit a region whose fertiliser inputs are already constrained by the Hormuz closure. Either threat alone would justify building reserves from the current surplus. Both arriving in the same window removes the question of whether to build them.

South Africa has already taken a significant share of the Katanga deficit market in the DRC and is the dominant regional exporter, with Zimbabwe absorbing 30 to 35 per cent of SA’s 2.4m tonne export forecast (Wandile Sihlobo, Agbiz). Tanzania is moving in the opposite direction from reserve-building. NFRA has contracted approximately 1.2m tonnes of grain for sale: 650,000 tonnes to Zambia (half delivered), 500,000 to the DRC, and 35,000 to the World Food Programme (Milling MEA, February 2025). NFRA plans to sell a further 1m tonnes in 2025/26 and launched a 534,000-tonne public tender in October 2025 at a floor price of TSh850,000/t for maize (The Citizen, October 2025). The storage constraint explains the urgency. NFRA’s total capacity is 776,000 tonnes, expanding to 1.1m under construction, against a government target of 3m by 2030 (TanzaniaInvest, President Hassan). Tanzania cannot simultaneously hold strategic reserves and store a surplus it cannot physically accommodate. It is exporting by necessity. If the El Nino and fertiliser shocks materialise in the 2026/27 season, Tanzania will have sold its buffer into a region that was also sitting on surpluses at the time. Zimbabwe will absorb an estimated 720,000 to 840,000 tonnes.

The pricing structure explains why the surplus is hard to move, and why the compression is already underway:

Two distinct markets are operating simultaneously. Old season stock from the 2024/25 harvest, already dry and stored in FRA sheds, is being sold through ZAMACE at approximately USD330/t and trading at factory gate in Lusaka at around USD345/t. This is FRA offloading inventory, not the floor price for the new harvest. New season maize from the 2025/26 harvest, still at high moisture content and in low volumes as collection begins, is already trading at approximately USD280/t and falling as harvest volumes increase and moisture drops. The gap between old season and new season will close as the harvest progresses.

South African white maize was trading at R3,506/t on SAFEX in January 2026, down roughly 50 per cent year on year (Grain SA). At current exchange rates, that converts to approximately USD214/t. Add transport to Katanga and the landed price reaches approximately USD250/t. Zambia’s new season maize at USD280 is already within striking distance of SA’s competitive price. As harvest volumes increase, prices compress further and that gap narrows.

The FRA buying floor tells a different story. The last confirmed price is K340 per 50kg bag, announced in June 2025 for the 2025 crop marketing season. The 2026 price has not been announced. FRA typically opens depots in late May or June. At the current kwacha rate of 19.05, K340 converts to approximately USD357/t, which is 28 per cent above where new season maize is already trading and 43 per cent above SA landed Katanga. Meanwhile, the Hormuz oil price spike has fed directly into domestic logistics: local freight costs have risen approximately 15 per cent, driven by elevated Brent crude benchmarks and end-of-March fuel price adjustments by the Energy Regulation Board (Nathan Kakhoma, FX and Commodities Risk Specialist). The surplus is harder to move physically at the same time it is harder to sell competitively.

The natural question is whether FRA will lower the floor to enable exports. History says it will not. No recorded reduction exists since at least 2017. The floor moves in one direction.

In surplus years, the government maintains the price and buys more volume. In deficit years, the price rises. Maize pricing in Zambia is a deeply political issue, and the August 2026 general election makes a reduction virtually impossible. No government drops the maize floor three months before an election.

The exchange rate has compounded the trap. The kwacha appreciated roughly 33 per cent against the dollar over the past twelve months, from approximately 28 to 19. The FRA floor barely moved in nominal kwacha terms (K330 to K340, a 3 per cent increase). But at the extremes of the kwacha's twelve-month range, the same nominal floor converts from approximately USD236/t to USD357/t. A year ago, at the weaker end of that range, Zambia's FRA floor sat below SA's export parity of approximately USD250. Today it sits 43 per cent above it. The kwacha appreciation, not FRA policy, is what priced Zambian maize out of regional competition.

For prices to compress toward the USD250 export parity, one of three things must happen: the kwacha depreciates back toward 25 to 28, FRA reduces the floor (no historical precedent, election year), or private market prices fall below the FRA floor as the surplus overwhelms domestic absorption capacity. The structural logic of the third path runs in two stages. If the surplus depresses prices in the near term, the compression creates a window to build reserves at lower cost. If fertiliser-driven yield reduction and the El Nino conditions described above then compress next season’s production, that window closes and prices rebound. The question is whether the reserve-building happens during the window or after it shuts.

The storage constraint is the binding problem:

FRA is already operating beyond secure capacity. The strategic reserve target of 2.5m tonnes exceeds the secure infrastructure by a million tonnes. At current construction timelines, the gap between the reserve target and the physical capacity to hold it is unlikely to close before the next planting season.

The strategic response is to rotate the stock. FRA is already selling old holdings through ZAMACE at approximately USD330/t (~K6,300/t per ZAMACE), below the buying price of USD357/t (K340/50kg = K6,800/t, converted at ZMW19.05), absorbing the loss. In 2024, the subsidy on FRA sales to millers amounted to K80 per 50kg bag, roughly 24 per cent of the purchase price (Zambian Observer). The rotation needs to accelerate: sell the deteriorating stock, accept the fiscal cost, refresh with new harvest grain at lower prices as the compression works through, and build toward the 2.5m tonne reserve target using current season revenues. The alternative is grain rotting in open-air storage while the next season's input costs spike.

Reports from the Zambian agricultural supply chain indicate that farmers are already weighing a significant shift from maize to soya for the 2026/27 season. The driver is input costs: fertiliser requirements for soya are a fraction of those for maize, roughly one-tenth by some estimates. That is the acid shortage and the fertiliser price spike transmitting directly into planting decisions on the ground. If that shift materialises at scale, Zambia’s 2026/27 maize production contracts, and the surplus that looked unmanageable in April 2026 becomes a strategic reserve by April 2027.

Southern Africa’s record regional production is masking a forward risk: the fertiliser cascade and the El Nino probability are converging on the 2026/27 season. The question is whether Zambia reads the current season or the next two.

Kenya. The short rains season from October to December is the planting cycle exposed to procurement conducted under current supply conditions. Kenya imports virtually all of its fertiliser, approximately 760,000 to 836,000 tonnes per year (IEA Kenya). The government’s National Fertiliser Subsidy Programme budgeted KES20bn for 2024/25, targeting 12 million bags (600,000 tonnes) at a subsidised price of KES2,500 per 50kg bag against a market price above KES6,000. The subsidy is denominated in shillings and priced against dollar-denominated imports. When the input price rises in dollar terms and the shilling weakens simultaneously, the fiscal cost of maintaining the subsidy expands from both sides. The programme’s centralised distribution model, which routes through NCPB depots averaging 18km from farmers rather than the 6km to private agro-dealers under the previous system, adds logistical friction that compounds during supply disruptions.

Tchakarova’s cascade analysis, drawing on the Tindale framework, identifies 28 April as the date on which the northern hemisphere spring nitrogen application window closes, as noted in my 19 March assessment. That date is now two weeks away. No ceasefire reached after it can restore 2026 crop yields in the US Corn Belt or South Asian cereal production. The food price transmission lag is approximately 90 to 120 days, placing the initial global food inflation impact in late July to September.

For Africa, the calendar is different but the logic is the same. The procurement decisions being made in the next eight to twelve weeks for the 2026/27 planting season will embed current supply conditions into African food prices well into 2027 regardless of when the Strait reopens. The food price impact of this disruption is not a near-term event. It is a six to nine month transmission from supply shock to retail price, running through procurement, planting, harvest, and distribution. The prices that African households pay for food in Q4 2026 and into 2027 are being set by the input costs being locked in this month.

The Copper Signal

The sulphuric acid shortage runs a second downstream channel through African commodity exports.

The DRC is the world’s second-largest copper producer, with output forecast at approximately 3,400 kilotonnes in 2026 (Fastmarkets, analyst estimates). What distinguishes the DRC is the share that depends on acid-intensive processing. Solvent extraction and electrowinning (SX-EW), the hydrometallurgical process used to recover copper from oxide ores, produced roughly 1.5m tonnes of cathode in 2025, approximately 45 per cent of national output against a global average of 15 per cent. The DRC also produces roughly 70 per cent of the world’s mined cobalt, which uses the same leaching process. The battery supply chain sits downstream of the same chokepoint.

Friedland warned on 2 March that over 90 per cent of sulphur imported into Africa comes from the Gulf and that oxide operations could face closure within three weeks. We are six weeks past that warning. Kamoa-Kakula, his own operation, is partially insulated. Its copper smelter, Africa’s largest at 500,000 tonnes per annum of copper concentrate processing capacity, produces sulphuric acid as a byproduct at 1,200 tonnes per day (Ivanhoe Mines, January 2026 production disclosure), ramping toward 700,000 tonnes of acid per annum at steady state. That acid is sold locally to DRC Copperbelt operations. Zambia banned acid exports in September 2025.

A distinction matters here. Zambia's Copperbelt is structurally different from the DRC's. The Zambian deposits are predominantly sulphide at depth, and smelters like Kansanshi produce acid as a byproduct of processing sulphur-rich ore rather than importing elemental sulphur. Smelter output data suggests total Zambian acid production of approximately 1.5m tonnes per year. Domestic prices sit at roughly USD250/t, a fraction of what the DRC pays. The system is fragile, with KCM and Mopani operating below historical capacity, Mopani's smelter underfed even after improved production following UAE investment. But the dependency runs through different geology.

The DRC’s exposure is a direct import dependency on Gulf sulphur, and the deficit is severe. DRC acid demand runs at approximately 6m tonnes per year (CRU Group, BC Insight, 2024 data). Domestic production is limited outside Kamoa-Kakula, whose disclosed output of 1,200 tonnes per day equates to an annualised production rate of approximately 400,000 tonnes (Ivanhoe Mines, January 2026). By April 2026, Friedland reported throughput of approximately 1,600 tonnes per day as the smelter ramps toward its 700,000-tonne steady state target. The gap between 6m tonnes of demand and well under a million tonnes of domestic production is what Gulf sulphur was filling.

Bloomberg Intelligence estimates 50 to 60 per cent of DRC copper output depends on sulphuric acid. Goldman Sachs reports DRC producers hold approximately three months of acid inventory, meaning a short disruption is manageable but a prolonged one erodes the projected refined copper surplus. The IEA notes the annual copper TC/RC benchmark settled at USD0/t in January 2026 (Reuters), with sulphide smelters surviving on byproduct revenues including acid and gold. Rising acid prices strengthen smelter margins. The pressure falls on the other side: the DRC's oxide operations, which consume acid rather than produce it, face input costs that have quadrupled in two years. If acid supply does not return, it is the SX-EW plants that cut production, and refined copper supply falls.

The bull case for copper later this year is acid-supply-driven, not demand-driven. It runs through the same chokepoint and the same chemical threatening phosphate fertiliser and food security. The competition between the mining sector and the fertiliser sector for a shrinking pool of sulphuric acid has received almost no attention outside specialist circles. Karim Fawaz observes that the scarier disconnect is between oil markets and broader equity, capital, and FX markets, which remain unfazed. The repricing has not arrived. The acid chain will be part of its transmission.

The Investment Implication

Sulphuric acid sits beneath both the food system and the industrial metals complex. Phosphate fertiliser and copper cathode production depend on it. Both are physically constrained by the same chokepoint and the same supply failures. The geopolitical contests producing these fractures will not end with this ceasefire. The Hormuz closure is a preview of how resource corridors behave under great power competition.

Practitioners in the Zambian fuel supply chain observe that the Russia-Ukraine disruption, which was smaller in magnitude, took roughly twelve months to normalise across oil and fertiliser markets. This crisis involves the physical closure of the world’s most critical commodity corridor for 44 days and counting, with a ceasefire that collapsed into a US naval blockade. We should be planning for at least twelve months. The evidence suggests longer.

The cascade multiplies; it does not add. Energy prices drive fertiliser costs, which drive planting decisions, which drive harvest volumes and food prices. Food prices drive political instability in import-dependent states. A ceasefire removes the trigger. It does not reverse the trajectory. The procurement timeline described above is the investment timeline.

This essay has traced a single chemical compound from a blocked chokepoint through phosphate fields, copper leaching pads, and planting decisions across the continent. The fiscal cost of Zambia’s next FISP tender, the price Kenya pays for its next fertiliser shipment, the margin on every tonne of DRC copper cathode, and the viability of Morocco’s phosphate processing all sit downstream of the same three supply failures.

The Hormuz closure is not an isolated event. It sits inside a pattern of supply corridor disruptions that has been accelerating since the tariff escalation of April 2025, running through the Venezuela-Guyana corridor and now into the Gulf. The confrontations driving these fractures are structural features of great power competition. The volatility they produce is likely to recur over the next five years at least. That changes the investment horizon from waiting for the crisis to pass to positioning for a world in which supply corridors are periodically contested.

Institutional portfolios are already reflecting this. Private real assets from wealthy clients are growing three times faster than traditional financial channels (Oliver Wyman, 2026). BlackRock’s 2026 outlook describes the shift as private markets moving from peripheral to essential against a backdrop of structural inflation. The 60/40 model, built for open supply chains and predictable monetary cycles, is giving way to frameworks anchored in productive physical systems. The capital suited to that world is connected to what those contests are repricing: the agricultural land where the maize-to-soya signal tells you fertiliser costs are rewriting planting economics, the water systems whose criticality the Gulf desalination crisis has measured in litres per household per day, the energy infrastructure that does not depend on a 34-kilometre channel, and the strategic minerals the battery supply chain requires. The capital that assumed the pre-April 2025 supply architecture would hold is the capital most exposed to discovering it will not.

The Architecture Question

Forty-four days after the Strait closed, the ceasefire has collapsed into a US naval blockade. Three supply failures that did not exist on 19 March have converged on a single chemical beneath both the food system and the industrial metals complex.

China’s yuan energy corridor is accelerating during the crisis. Africa has settlement architecture and is not using it. The fertiliser to be procured for the next African planting season will be priced in this supply environment. The copper that the DRC produces, and that the global energy transition requires, depends on an acid chain running through the same chokepoint. The phosphate that Morocco processes, and that global agriculture requires, depends on the same chemical from the same blocked sources.

The headline is still oil. The downstream is already being priced into fertiliser tenders, acid spot markets, and procurement calendars across the continent.

The infrastructure exists. The crisis exists. The arbitrage exists. What is missing is the decision to connect them before the next procurement cycle prices the acid shock into African food budgets and copper cathode margins for the next twelve to eighteen months.

Structure precedes sentiment.

The full downstream cascade analysis, including detailed country-level fuel and fertiliser mechanics for Nigeria, Kenya, and Zambia, is available in “Quick Take: The Shock That Is Not One,” published 19 March 2026 on Canary Compass.

Sources

Craig Tindale, raw materials and industrial cascade analysis, public thread and Simultaneity Metrics Framework contributions, April 2026. Robert Friedland, Ivanhoe Mines founder and executive co-chairman, public disclosures on Chinese sulphuric acid exports (preliminary Chinese customs data), Kamoa-Kakula operational data, and DRC Copperbelt acid supply, March-April 2026; Ivanhoe Mines corporate disclosures, January and March 2026. Ed Finley-Richardson, maritime analyst, public thread on Hormuz toll mechanism and OFAC compliance architecture, 8-9 April 2026. New York Times, reporting on Iranian mine-laying and cleared corridor toll mechanism, sourced to US officials, 10 April 2026. US Central Command (CENTCOM), mine clearance operations announcement, 11 April 2026; blockade scope clarification (non-Iranian port transit permitted), 12 April 2026. Al Jazeera, Saudi East-West pipeline restoration to full capacity, 12 April 2026. Hapag-Lloyd, shipping normalisation timeline estimate, April 2026. President Donald Trump, Truth Social statement on US naval blockade of Strait of Hormuz, 12 April 2026. Velina Tchakarova, geopolitical and systemic risk analyst, “The Global System Rupture” series, March-April 2026. Karim Fawaz, observation on physical oil and financial market disconnect, April 2026. Bloomberg, confirmation of China sulphuric acid export halt, 10 April 2026. Bloomberg Intelligence, DRC copper-sulphuric acid dependency estimates, March 2026. S&P Global Trade Atlas, Persian Gulf seaborne sulphur export share. International Energy Agency, copper smelter TC/RC benchmark and byproduct revenue analysis, March 2026. Goldman Sachs, DRC acid inventory estimates and refined copper surplus projections, April 2026. Reuters, Antofagasta/CSPT TC/RC settlement reporting, January 2026. CRU Group and BC Insight, DRC acid demand estimates (2024 data, November 2025). USGS, DRC copper production data (2025 estimates). Fastmarkets, DRC copper production forecasts and TC/RC reporting. Zambia Statistics Agency, 2024/25 Crop Forecasting Survey. Food Reserve Agency, 2025 crop marketing season purchase prices and volumes. Open Zambia, FRA floor price announcement June 2025. Zambian Business Times, FRA historical pricing and SAFADA commentary, 2020-2023. Milling MEA, FRA 2023 pricing announcement. Presidential Delivery Unit, FRA record 1.6m MT purchase, March 2026. Zambian Observer, “FRA’s Deepening Fiscal Trap,” November 2025, storage capacity, FRA purchase data, and miller subsidy mechanism. Nathan Kakhoma, FX and Commodities Risk Management Specialist, LinkedIn, April 2026, Lusaka maize market pricing and freight costs. Zambia Agricultural Commodity Exchange (ZAMACE), FRA selling programme data. AGRA Food Security Monitor, Tanzania maize production and surplus, August 2025. USDA Foreign Agricultural Service GAIN Report, Tanzania grain and feed annual, April 2025. Milling MEA, NFRA grain sales and contracts, February 2025. The Citizen (Tanzania), NFRA grain tender, October 2025. TanzaniaInvest, NFRA storage expansion and food security bond, July 2025. FAO, Tanzania total cereal production estimate. NOAA Climate Prediction Center, ENSO Diagnostic Discussion and El Nino probability forecast, March 2026. International Research Institute for Climate and Society (IRI), ENSO probability forecast, March 2026. South African NAMC SASDE, maize production and export data, October 2025. Wandile Sihlobo (Agbiz), South Africa maize export analysis. Grain SA and Business Day, SAFEX white maize spot pricing, January 2026. Trading Economics, USD/ZMW exchange rate data. Nation Africa, Kenya fertiliser subsidy budget reporting. IEA Kenya, fertiliser import data. IFPRI, Kenya National Fertiliser Subsidy Programme assessment. Afreximbank and TechCabal, PAPSS operational data including country count, bank connectivity, and PACM launch, February-March 2026. Oliver Wyman, private wealth asset allocation trends, 2026. BlackRock, Private Markets Outlook for US Wealth, 2026.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu or X @InfinitelyDean.

AI-generated image of The Tower of Babel: Everyone is still building. Nobody is coordinating. The tower stands unfinished not because it was destroyed, but because the builders stopped understanding each other.

This week I was shocked at some of the things I heard from highly capable people. It was a peak, but the bewilderment has been building for some time. I said publicly that I was 50 per cent convinced it was time to grab a holy book, because what I was encountering had stopped making sense and I was no longer sure the explanation was entirely rational.

So I reached for one. That is the trader in me. When the model breaks, you go back to first principles. And first principles for a four-thousand-year-old problem might actually be four thousand years old.

The first story that matched what I was observing was not from Genesis. It was older. Around 2100 BCE, a Sumerian text called Enmerkar and the Lord of Aratta recorded a king building a ziggurat while the god Enki confused the languages of the inhabited world. The Assyriologist Samuel Noah Kramer called it “The Babel of Tongues: A Sumerian Version.” It was written in southern Mesopotamia, in the land we now call Iraq, beside the conflict that prompted this essay. The Genesis account tells a version of the same story, composed later in the tradition. God left the tower standing, the builders still holding their skills, their tools, their ambition. He confused only the languages. The project failed not from external force but from internal incoherence.

Four thousand years later, the builders are still standing. The languages are still confused.

I have written before about information overload. This is different. Overload is a volume problem. What I observed this week is a coherence problem. People using identical words inside completely different realities. They had information. They lacked a common frame to process it through. Babel is the absence of a shared language, not the absence of language itself. Each dialect works perfectly within its own community. The collapse is between them, not within them.

There are at least two such dialects operating simultaneously. The first has detached entirely from verifiable evidence. In this dialect, public figures have been secretly replaced. Global events are scripted performances. Timelines contradict what can be verified but cohere within their own internal logic. The people speaking this language are not stupid. They are pattern-matching inside a closed mythology where coherence has replaced evidence as the standard of truth. This dialect persists because it serves a function the institutional language will not perform: it provides a moral architecture for people who feel the world has stopped making sense. When institutions fail people repeatedly and the institutional dialect insists the models are fine, some people abandon the institutions entirely and build a replacement cosmology. The replacement is often absurd. The departure was rational.

The second dialect is institutional. It references real data and operates within recognisable analytical frameworks. But the frameworks have stopped updating. A political scientist whose core body of published work sits in air power, terrorism, and political violence publishes a newspaper commentary. The commentary argues that a country with a GDP between USD400bn and USD475bn now qualifies as a global superpower alongside nations whose economies are tens of times larger. The claim fails on the evidence: superpower status requires sustained force projection across multiple theatres, independent technological capacity, and a financial architecture others depend on. The scholar also argues regime survivability under air attack and a demonstration effect for adversaries. These are real observations that describe a formidable regional spoiler. They do not describe a superpower. Disrupting one maritime corridor, however damaging in the current configuration, does not meet that threshold, particularly when the disruption weakens if escort operations materialise or supply chains reroute.

The two dialects are not equivalent in kind. The institutional dialect still references claims that can be checked and defeated by evidence. The conspiracy dialect has abandoned verification entirely. But they share a structural habit: both metabolise disconfirming evidence as confirmation rather than correction. The institutional dialect attributes failure to external shocks rather than revisiting the model. The conspiracy dialect treats contradiction as further proof of the conspiracy’s depth. The mechanism is the same. The recoverability is not. The conspiracy dialect has left the building entirely. This essay addresses the institutional dialect because it is the one this audience speaks, and the one that can still be corrected.

The evidence for the fracture is not anecdotal. It is the cleanest pattern in contemporary electoral politics, and it is visible first and most starkly in Africa.

In Zambia, Michael Sata won the presidency in 2011 on “lower taxes, more jobs, and money in your pockets.” His base was urban working-class voters and Copperbelt labour in a country where 60 per cent of the population lived below the poverty line (World Bank, 2010). In Kenya, William Ruto won in 2022 on the “hustler vs dynasties” narrative, the wheelbarrow as campaign symbol, informal sector workers and unemployed youth as the base. Scholarship on Ruto’s populism echoes the Zambian case and explicitly notes the strategic similarity. Both campaigns channelled economic discontent that the institutional class had failed to address. Both resonated with voters who had no use for policy papers because the policies had not reached them.

The populist base is not a third dialect. It contains people who still operate within the institutional frame, people who have moved into the conspiracy frame, and people who have abandoned both and vote from lived experience alone. What holds them together is not a shared language but a shared grievance. The institutional class cannot decode the coalition because it arrives in every register simultaneously.

The same pattern appears across the industrialised world. In the 2024 US election, Pew’s validated voter study shows Trump led by 14 points among voters without a college degree, while Harris led by 16 points among degree holders. Among voters earning under USD50,000, the margin swung from Democrat +22 in 2012 to Republican +2 in 2024 (Roper Centre exit polls, both years). The education gap has become one of the sharpest demographic predictors of partisan alignment in America, a structural inversion from the era when income was the primary dividing line. In France, the Rassemblement National won 45 per cent of blue-collar workers’ votes in the 2022 legislative elections (IPSOS post-election analysis). Support was strongest in departments with the lowest educational attainment and highest poverty (Reuters demographic analysis, April 2022). In Britain’s 2024 general election, Reform drew 23 per cent among lower-education voters and 8 per cent among degree holders (YouGov post-election breakdown), a ratio of nearly three to one.

Five countries. Three continents. The same tower, the same confusion of tongues.

The institutional response to this pattern has been remarkably consistent. Hillary Clinton called them a “basket of deplorables” in 2016. Gordon Brown dismissed a Labour voter’s immigration concerns as bigotry on a live microphone in 2010. David Axelrod, Obama’s adviser, observed that the Democratic Party approaches working-class voters in the spirit of a missionary: we are here to help you become more like us. Implied in that, he said, is disdain.

The institutional dialect cannot name populism’s cause because naming it would require examining what produced it. The technocratic consensus that shaped economic policy from the early 1990s delivered trade liberalisation without transition support and fiscal austerity applied asymmetrically. It delivered labour-market deregulation that suppressed wage growth while concentrating asset returns. It delivered immigration frameworks designed for aggregate economic benefit without accounting for the distributional strain on communities absorbing the change. The result was a generation of stagnant real incomes. In at least two of the signature episodes, the institutions themselves subsequently acknowledged the error. The IMF’s Independent Evaluation Office reviewed the Asian crisis programme design. The Bank of England commissioned the Bernanke Review of its inflation forecasting after the 2021 to 2022 failures. These reviews were exceptions, not the norm, and both arrived after prolonged institutional resistance to acknowledging the error. Other forces contributed. Technological displacement, demographic shifts, and globalisation would have restructured labour markets regardless of policy choices. The technocratic consensus did not cause all of this alone. But it claimed to manage all of it. And when the management failed, it demanded continued authority over the response. The question is whether the credential survived the outcome.

The institutional pipeline that fed these frameworks into central banks, treasuries, and multilateral organisations also produced a policy class that internalised them as settled science rather than contestable choices. The metrics they optimised (GDP growth, inflation management, trade volumes) were never designed to capture household-level welfare divergence. When household welfare diverged, the metrics said everything was fine. That is the mechanism. The dialect does not reject evidence. It reads evidence the model was designed to capture and structurally misses the evidence it was not.

The institutional dialect extends beyond policy into media. The same credentialing pipeline that feeds central banks and treasuries also feeds prestige newsrooms. Scholars become television commentators. Policy professionals become op-ed columnists. The amplification system reinforces the dialect. When audiences whose lived experience contradicts the institutional framing stop recognising themselves in the coverage, they leave. The growth of alternative media, podcasts, independent newsletters, and social commentary platforms, is evidence that the fracture already existed and the institutional media class was the last to notice. Algorithmic sorting accelerated the separation. But it accelerated a fracture that was already structural. The grievance preceded the platform. The platform gave it a dialect.

We carry our own dialects. Those of us trained in markets learned to evaluate claims by outcomes rather than by source. That discipline is useful. But it also trained us to discount anything that cannot be priced, which means we can be blind to the things that move people precisely because they cannot be quantified. The conspiracy dialect persists because it offers coherence where we offer only data. We do not share that architecture. But dismissing it without understanding why it exists is deafness. This essay is written from inside a dialect, not above the tower. Its value, if any, lies in the phenomenon being visible enough that even a partial view is worth publishing.

Some of the institutional deafness may not be deafness at all. Some of it may be strategic. The institutional class contains people who genuinely cannot decode the populist register and people who decode it perfectly but refuse to engage because engagement would legitimise a threat to their position. This essay addresses the first group. The second group does not need a diagnosis. It needs an adversary.

The tower was never destroyed from the outside. The languages confused themselves. The builders are still standing. They still have their tools. They cannot coordinate because they have lost the shared language that once made coordination possible.

Expertise is not a title or an institutional address. It is demonstrated competence within a specific domain, tested against outcomes.

The tower will not be rebuilt by either dialect speaking louder, by the institutional class demanding deference its track record has not earned, or by the conspiracy class retreating further into a cosmology the rest of us cannot enter. If it is rebuilt at all, it will be rebuilt by people willing to do the most uncomfortable thing either dialect can imagine: cross the floor and learn the other grammar. The goal is hearing what the other side is actually saying. The person on the other side of the fracture is speaking a language we never learned.

The tower does not care how fluent we are. It only cares whether the builders can still understand each other.

Many of us check the source before we evaluate the content. Often we never get to the content at all. The credential hangs on the wall, and the inquiry ends. The affiliation is verified, and the paragraph goes unread. The institution signals, and compliance follows. The question is rarely whether the analysis holds. The question is whether the person is authorised to produce it. The answer to that question ends the inquiry before the first one is asked.

There is a structural reason this persists, and it is not laziness. Under information overload, source becomes a compression device. It lowers the cost of judgment. It preserves social belonging. It lets people outsource evaluation without admitting they have done so. The habit survives because it solves a real problem: there is too much to read and not enough time to read it. But the source is not a shortcut to the answer. It is a shortcut past the question. The danger begins when the shortcut becomes the only method.

Some institutions reinforce it by design. Central banks publish forward guidance so that markets can anticipate policy. The strategic advantage of monetary policy is predictability. In 2013, the Federal Reserve signalled a reduction in bond purchases. Markets repriced rationally, but the repricing punished emerging economies whose fundamentals had not changed. What the system trained was not obedience. It was dependency: the expectation that a reliable, transparent signal would always be available. Not every institution operates this way. Military strategy, intelligence work, and competitive negotiation depend on the opposite logic. Their advantage is surprise.

A generation trained on institutional transparency struggles with institutional opacity. The absence of a signal does not feel like freedom. It feels like abandonment. When the signal is deliberately withheld, people manufacture their own. Traders build consensus from dealer chatter. Journalists construct narratives from anonymous sources. The manufactured signal carries no authority but demands the same compliance. And when analysis fills the gap instead of allegiance, what arrives as diagnosis is received as endorsement. Comprehension collapses into conviction, because when source-checking is your only epistemology, explaining the logic means endorsing the action.

There is a difference between delegation and surrender. Trusting a structural engineer’s calculations before crossing a bridge is rational. The alternative is checking every weld yourself. Trusting a commentator’s framing before forming a view on a contested situation is something else. One is a division of labour. The other is a transfer of judgment. The danger is not that we delegate. It is that we stop noticing when delegation becomes obedience. Medicine maintains formal governance: the practitioner who consistently misdiagnoses faces review, accountability, revocation. Commentary has no equivalent licensing mechanism. In those domains, it is the reader who performs the governance function. But governance requires reading.

A position is explained and the response is not engagement with the logic but a verdict on the person behind it. The analysis is never examined. The analyst is. Both sides of a contested situation often take the same words at face value and arrive at opposite conclusions, operating at the same distance from the structural reality because neither reads the actions underneath the words. The transcript, the policy paper, the earnings call, the diplomatic note: each contains the same data. Each yields mirror-image certainties.

There is a children’s game built on this exact principle. The player never evaluates whether the instruction is worth following. The player only checks whether the authority gave it. Most of us recognise the game. Few of us recognise that we never stopped playing.

Conclusions formed under uncertainty tend to harden. They begin as placeholders and become identities. The shift happens without announcement. Identity does not update. It defends. By the time new information arrives, the response is not evaluation but protection. The update, when it comes, does not feel like learning. It feels like loss. And loss is not something most people will volunteer for twice.

The credential filter, the tribal sort, the institutional signal. Each replaces evaluation with obedience, because evaluation demands sitting with incomplete information, and most of us never learned to sit. Sitting means holding the question open when the room has already answered. It means reading the document when the byline has already been checked. The discomfort is something I have learned to recognise, not something I have overcome.

The chess board does not care who calls the next move. It only asks whether the move is sound.

Maybe it begins with learning to sit without a signal.

AI-generated image. Each pressure is small. The direction is not.

Yesterday we published a back-of-the-napkin estimate that the three-month tax suspension would cost the Treasury roughly ZMW3 billion to ZMW4 billion. Overnight, we refined the model. The revised range is ZMW3.3 billion to ZMW4.6 billion, depending on the conflict path, duration, and consumption growth. But the fiscal cost is not the story. The story is where that cost lands. The IMF had already identified early signs of fiscal slippage before the oil price surge. VAT collection was already under pressure. Mineral royalties are compressed by currency strength. The bond market has tightened. A ZMW14.6 billion maturity wall arrives in June. No IMF programme exists to anchor the response. The cost is manageable. Whether the quarter absorbs it depends on what else is moving.

The Intervention

On 31 March 2026, the Energy Regulation Board published the April fuel price review. On the same day, Cabinet declared the fuel supply situation an emergency. It approved the zero-rating of Value Added Tax and suspension of Excise Duty on petrol and diesel for three months, effective 1 April 2026.

The ERB’s Import Parity Pricing model uses Platts Arab Gulf product assessments, the international benchmark that determines the wholesale cost of fuel imports into Zambia, as its top cost line.

The Kwacha depreciated 0.63 per cent over the same period, from K19.18 to K19.30.

The April pump prices already reflect the removal of excise and the zero-rating of VAT. The ERB’s formula, applied without intervention, would have produced pump prices of approximately K34.21 for petrol and K35.41 for diesel. Cabinet chose to intervene.

Kerosene carries no excise duty, is exempt from VAT, and attracts zero customs duty. The tax intervention mechanism available for petrol and diesel did not exist for kerosene. It rose 53 per cent to K32.26 per litre. With national stock cover at fewer than 15 days and supply constrained to road tanker capacity from Dar es Salaam, subsidising kerosene would have risked a physical shortage. The price pass-through is the mechanism that keeps the residual market clearing. The households that cook with kerosene absorb the full shock. Not because the government chose to exclude them, but because the tax architecture offered no lever and the supply constraints ruled out a subsidy. Charcoal, the primary cooking fuel for lower-income urban households, has more than tripled since April 2019. These are the households with no domestic savings to cushion seven years of above-target inflation through investment participation. The tax suspension on petrol and diesel limits the pass-through to public transport fares, which protects the ability to get to work. It does not reach the kitchen.

The Fiscal Cost

Taxes have been suspended, not spending authorised. The budget loses the revenue it would otherwise have collected.

The fiscal cost per litre equals the excise foregone plus the VAT that would have been charged on the pre-VAT counterfactual price. The counterfactual pump price equals the actual April price plus the suspended excise, multiplied by 1.16 for VAT.

The 2025 consumption figures are 11 per cent higher for petrol and 12 per cent higher for diesel than 2024. The assumption of flat consumption is conservative: demand may contract at higher prices, which would lower the fiscal cost but also signals weaker economic activity. The fiscal cost is driven by the Platts Arab Gulf benchmark, which reflects global conditions regardless of where Zambia sources its product. The supply risk is separate. If the Strait of Hormuz remains closed, physical product from the Gulf does not arrive. Indian refineries, which sit on a shorter shipping route to Dar es Salaam than the Arab Gulf and already supply East Africa, are the most likely alternative source. Nigeria’s Dangote refinery, which reached full capacity in February 2026, offers a medium-term continental option. But the immediate vulnerability is domestic: the Energy Minister disclosed petrol stock cover at 19 days on 6 March; by 19 March the Permanent Secretary reported 23 days. Both figures sit at the margin where disruption to road tanker logistics from Dar es Salaam compresses supply within weeks.

April’s cost is locked. It does not depend on where oil goes next. The question is what May and June cost under different conflict trajectories.

If nothing changes from where we are today, the three-month suspension costs approximately ZMW3.8 billion.

The estimates in Tables 4 and 5 use 2025 consumption volumes held flat. Consumption grew 11 to 12 per cent in 2025 and will almost certainly be higher in 2026.

The range across oil price scenarios and consumption assumptions is ZMW3.3 billion to ZMW4.6 billion for the three-month window. The three-month window is itself an assumption. If the suspension extends beyond June, every additional month adds between ZMW1.0 billion and ZMW1.5 billion depending on where oil prices settle. The pre-war baseline was approximately USD70 per barrel. Any price above that level pressures the government to maintain the suspension, and the market consensus as at end-March does not see a return to that baseline within the three-month window.

The Pressure Points

The suspension removes revenue from the top of the fiscal envelope. The economy was already underdelivering revenue from the bottom.

On 5 March 2026, the IMF staff team concluded its visit to Zambia. The statement identified “early signs of slippage” driven by spending pressures on the wage bill, agricultural support, and election-related expenditures. Absent corrective measures, the primary surplus was projected to fall by about 1 percentage point of GDP relative to the 3.8 per cent envisaged at the last ECF review. The fuel cost arrives on top of a budget that was already slipping.

Fourth quarter 2025 GDP printed 1.6 per cent, dragging the full year to 3.8 per cent against a 2026 budget assumption of 6.4 per cent real growth. The collapse was concentrated in wholesale and retail trade, which subtracted 1.5 percentage points from Q4 value added. The revenue base entering 2026 was already under strain. The table below compares actual 2025 collections with the 2026 budget targets.

The weight falls on VAT. The collapse in wholesale and retail was most visible in domestic VAT, which reached only 34 per cent of its 2025 target. Import VAT, which accounted for 88 per cent of total collections, performed at 91 per cent. The assessment formula converts CIF values to Kwacha at the prevailing exchange rate: when the currency strengthens, every dollar of imports generates less Kwacha in tax at the border. The same mechanism compressed mineral royalties. The 2026 VAT target requires 46 per cent growth from that combined base. The Kwacha has strengthened a further 14 per cent since December. The deficit target of 2.1 per cent of GDP rests on that growth assumption.

Mineral royalties, budgeted at ZMW18.2 billion, face a separate compression. Copper averaged USD12,943 per tonne in Q1 2026, some 30 per cent above the 2025 average. But the Kwacha averaged K19.43 in Q1, roughly 23 per cent stronger than the 2025 average of K25.2. The copper price gain is almost entirely offset by currency strength. At Q1 prices and exchange rates, the budget requires approximately 17 per cent production growth to deliver the royalty target. At 10 per cent production growth, royalties annualise at roughly ZMW17.1 billion, a shortfall of ZMW1.1 billion.

The interest-to-revenue ratio stood at approximately 25 per cent before the suspension, calculated as budgeted domestic interest expense of ZMW52.0 billion against budgeted domestic revenue of ZMW206.5 billion. On the broader base of domestic revenue plus grants, the ratio is 24 per cent. The IMF’s fiscal sustainability framework flags ratios above 20 per cent.

The government’s 2026 domestic borrowing programme targets ZMW106.07 billion in market issuance. At the start of the year, the required monthly average was ZMW8.84 billion. Strong January and February auctions brought the remaining requirement to ZMW80.81 billion and the monthly average down to ZMW8.08 billion. March reversed the trajectory. The remaining requirement is ZMW75.80 billion across nine months, or ZMW8.42 billion per month. The required allocation rate against remaining offered amounts at Q1 auction sizes has climbed from 89.4 per cent to 95.2 per cent. One poor financing month was enough. The refinancing targets were difficult from the start. The government cannot afford poor financing months to continue at the same time that revenues are underperforming.

The fiscal cost must be funded. The net domestic financing authorised in the budget is ZMW21.62 billion. If revenue underperforms, the deficit widens. Funding the wider deficit domestically requires a supplementary appropriation to authorise additional borrowing, which raises the ZMW106.07 billion issuance baseline into a market that now needs to operate at 95 per cent allocation. Without a supplementary appropriation, the shortfall either compresses expenditure or adds to the domestic arrears stock, which stood at ZMW77.6 billion (USD3.25 billion) as at end-September 2025. All three options carry cost.

April and May are light on bond maturities, as March was. June is not. Per Reuters and Bank of Zambia Q1 auction results, June has ZMW14.62 billion in total maturities, of which ZMW8.29 billion is in bonds. The March auction result reflects several pressures moving at once. Exchange rate risk from the Iran situation compounds the IMF’s early revenue concerns and a GDP outturn of 3.8 per cent against expectations of 4.5 to 5.2 per cent. The global risk environment discourages frontier market exposure. The offshore bid that masked the transmission failure in February did not return. If that persists into June, new offshore investors may stay away and existing holders facing maturities may divest rather than reinvest. One undersubscribed auction is a signal, not a verdict. But the domestic market has not historically absorbed bonds at this scale without offshore participation.

The Ministry of Finance’s 2026 Monthly Economic Indicators have not been published. It is 1 April. The government just committed ZMW3.3 to 4.6 billion in tax revenue foregone over three months against a budget of ZMW253.1 billion. The IMF Extended Credit Facility concluded following its sixth review in January 2026. The Ministry of Finance stated that the government chose not to pursue an extension, directing instead that engagement on a successor programme begin immediately. The IMF’s March visit indicated otherwise: initial technical discussions could begin as early as late April, but substantive engagement would resume only after the general elections and once a new government is in place. With the ECF concluded, there are no quantitative performance criteria binding the fiscal authorities. Domestic budget ceilings can be amended via supplementary appropriation.

The government undertook to maintain fiscal discipline when the ECF was not rolled over. Timely fiscal reporting is the minimum condition for that commitment to be credible. The January indicators remain unpublished.

The Structural Point

This note does not argue that the intervention was wrong. The alternative was pump prices of K34.21 for petrol and K35.41 for diesel. That would have imposed severe hardship on households whose purchasing power eroded over nearly seven years of above-target inflation between April 2019 and early 2026. Inflation returned to the target band in March 2026, but the cumulative damage is already embedded in prices. The CPI index sits between 44 and 64 per cent above where the Bank of Zambia’s own target band would have placed it, depending on which point in the band you measure from. Individual staple commodities tell a harsher story over the same period. Charcoal sits 98 per cent above the 8 per cent ceiling path in March 2026. Maize grain sits 81 per cent above. Breakfast mealie meal 69 per cent. The government faced no viable alternative. It absorbed the cost fiscally rather than pass it to the consumer.

The structural point is different. The government exited sovereign default, completed six consecutive IMF reviews, and reached agreement in principle on 94 per cent of USD13.3 billion in external debt under restructuring. Those achievements are real. They also make the current fiscal tightness more consequential, not less. The stabilisation effort earned the space to build larger buffers. The 2026 budget assumed aggressive revenue growth and set an expenditure envelope that requires a large domestic borrowing programme to fund. The refinancing wall from the 2020-2021 issuance cycle, when the prior administration borrowed at yields between 30 and 35 per cent, concentrates maturities in this year. That wall is inherited. The revenue assumptions and expenditure commitments are not. The room had not kept pace with the stabilisation effort.

Kenya, facing the same Platts Arab Gulf price surge through its own import parity model, drew on its Petroleum Development Levy stabilisation fund. Zambia's ERB price build-up includes a Strategic Reserves Fund levy of K0.15 per litre, collected since 2005 with a price stabilisation mandate. Its balance has never been publicly disclosed. Interestingly, it was not deployed.

What this episode reveals is the fragility of Zambia's fiscal shock absorption architecture. The Strategic Reserves Fund was established for exactly this scenario and was not deployed. A commodity-dependent, import-exposed economy with no strategic fuel reserve, a stabilisation fund of undisclosed balance, and no automatic trigger for fiscal adjustment will face this choice again. The architecture that needs building is not a better subsidy. It is a fiscal buffer that is funded, transparent, and operational, making the next shock a technical adjustment rather than a Cabinet emergency.

The excise and VAT suspension reduces collections at precisely the moment the budget needs them most. The interest-to-revenue ratio worsens as the denominator shrinks. Mineral royalties are compressed by Kwacha strength. The domestic borrowing programme meets a bond market that just undersubscribed and a T-bill market trading below the policy rate. A ZMW14.6 billion maturity wall arrives in June. And we cannot assess how the budget is performing because the fiscal data has not been published.

The three-month window expires on 30 June 2026. If oil remains above the pre-crisis baseline of approximately USD70, the government faces a binary. It can extend the suspension and continue absorbing ZMW1.0 to 1.5 billion per month from a tightening revenue envelope. Or it can allow full pass-through into pump prices, with a general election scheduled for August. The political incentive to extend is obvious. The fiscal capacity to extend is not. Both options carry cost. Neither is free.

The fuel tax suspension costs 1.3 to 1.8 per cent of the budget. By itself, it is manageable. It does not arrive by itself. It arrives in a quarter where the IMF has already flagged a 1 percentage point primary surplus slippage. Revenue is under pressure from three directions simultaneously. The bond market has tightened. A June maturity cliff concentrates rollover risk in the same month the suspension expires. None of these is a crisis. All of them are moving in the same direction. A little here, a little there.

Sourcing note: Full primary source citations are available in the online publication at www.canarycompass.com. Key sources include ERB April 2026 press statement, ERB March 2026 price build-up, ERB 2025 Annual Statistical Bulletin, Cabinet resolution (31 March 2026), Bank of Zambia auction results and exchange rate data, Reuters maturity profile, ZamStats GDP and CPI data, 2026 National Budget Revenue Estimates, Ministry of Finance Monthly Economic Indicators (December 2025), Bank of Zambia Debt Statistical Bulletin Q3 2025, IMF Press Releases Nos. 26/024 and 26/073, and prior Canary Compass publications.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu or X @InfinitelyDean.

In November 2025, before the data arrived, we wrote this:

Growth came in at 3.8 per cent. The thesis was right. The projection was not.

The question was never whether Zambia would grow. It was whether growth would reach the parts of the economy that employ most people, serve most households, and generate most of the tax base that sustains public spending. The 2025 data answers that question with unusual clarity. Agriculture and mining drove the headline. Wholesale and retail trade, the sector that touches households most directly, contracted at its deepest rate in at least five years. The Stanbic Bank Zambia PMI, surveying approximately 400 firms across all major sectors, never broke convincingly above 50. Growth happened, but it did not spread.

The miss

Full year 2025 GDP came in at 3.8 per cent, matching the revised 2024 outturn. Every major institution overestimated growth.

The IMF’s March 2026 staff visit, the most recent estimate before the data release, cited weaker mining, softer wholesale trade, and energy constraints on non-mining activity. Even that estimate overshot by 0.7 percentage points.

The miss is concentrated in Q4. The first three quarters of 2025 averaged 4.6 per cent (4.6, 5.5, 3.7), consistent with a moderately growing economy. Q4 printed 1.6 per cent, dragging the annual figure below every forecast.

Where the headline comes from

The headline GDP figure of 1.6 per cent in Q4 is measured at purchasers’ prices. The underlying productive economy (total gross value added at basic prices) grew 2.1 per cent, but a 0.4 percentage point drag from taxes less subsidies on products, which contracted 8.9 per cent, pulled the headline lower. That contraction reflects in part the collapse in wholesale and retail activity, which depresses VAT and import duty collection. It is not confined to 2025. The 2026 budget requires a materially larger revenue envelope. If wholesale and retail activity remains depressed, the domestic tax base that funds it will underperform, making an already difficult financing year harder.

Two sectors account for the bulk of the Q4 deceleration. ICT subtracted 2.5 percentage points. Wholesale and retail trade subtracted 1.5 percentage points. Together they removed 4.0 percentage points from Q4 value added. But the year-on-year swing table reveals that the story extends further.

Gross negative swings totalled 11.2 percentage points. Gross positive swings totalled 4.6 percentage points. ICT accounted for 57 per cent of the gross negative and 96 per cent of the net deceleration. But mining swung negative by 2.3 percentage points and financial services by 0.5. The Q4 print is not the story of one sector collapsing. It is the story of broad service-sector weakness overwhelming recoveries in agriculture, manufacturing, construction, and electricity.

This distinction matters because the sectors that drove the recovery are structurally different from those that contracted. Agriculture, construction, and electricity are capital-intensive, often publicly supported, and employ relatively few people per unit of output. Wholesale and retail, financial services, and the informal economy adjacent to them are where employment concentrates, household spending circulates, and the tax base compounds.

The wholesale and retail signal

The ICT question, which we address below, should not obscure the genuine economic signal in the data.

Wholesale and retail trade contracted 11.2 per cent for the full year 2025, subtracting 1.8 percentage points from annual growth, the single largest sectoral drag on the economy. The five-year trajectory is unmistakable: growth in 2021, positive in 2022, marginal in 2023, mild contraction in 2024, then a sharp acceleration to deep and sustained contraction across every quarter of 2025.

The PMI data confirms the mechanism. In March 2025, firms reported reduced purchasing power at customers. By May, new orders surged to their strongest in seven years as the Kwacha appreciation from its March peak improved purchasing power. That relief was temporary. By December, firms again cited lower purchasing power at clients, and new orders contracted for the first time in nine months.

Three forces converged. First, cumulative price erosion. Consumer prices rose approximately 60 per cent between end-2021 and end-2025, with food inflation running at 18.9 per cent as late as March 2025. Four consecutive years of double-digit inflation, averaging roughly 12.7 per cent annually from 2022 to 2025, systematically eroded household purchasing power. Agriculture grew 51.1 per cent in 2025 and the maize harvest was a record, but agricultural GDP measures output, not farmer income. Smallholders entered 2025 with balance sheets depleted by the failed 2024 harvest and no working capital to convert a strong season into restored purchasing power. The quantity was there. The transmission to household spending was not.

Second, the Kwacha moved through three phases, each interacting with power recovery and cost pressures but visible as the dominant currency regime in the PMI testimony. A depreciation phase through Q1 2025 (peaking at ZMW 28.90 in March) drove up import costs and squeezed margins. A sweet spot from Q2 through Q3, as appreciation lowered input costs and improved purchasing power. Then a competitiveness squeeze from Q4 into 2026 as the appreciation accelerated to ZMW 18.80, with firms explicitly citing cheaper imports undercutting domestic production.

Third, weak credit transmission. Private sector credit growth fell from 21.4 to 15.7 per cent, and what remained was concentrated: approximately 65 per cent of total lending flows to the top 20 borrowers (Bank of Zambia data, December 2025), and only around 10 per cent of firms hold a bank loan or line of credit (World Bank Enterprise Survey, 2019). The Bank of Zambia’s February 2026 Monetary Policy Report documented firms repaying loans rather than borrowing. For the small and medium firms that dominate wholesale and retail, the formal credit channel remained effectively closed.

The relatively consistent contraction across all four quarters (negative 12.0, 10.6, 12.5, 9.8) indicates a sustained structural squeeze, not a single discrete shock. The PMI confirms this independently of the GDP statistics: firms reported demand destruction, margin compression, and falling purchasing power across the year. Part of the measured contraction may reflect relative price effects from the Kwacha appreciation compressing the GDP deflator for imported goods. But the firm-level testimony points to genuine demand weakness, not a statistical artefact. This is the clearest signal in the data that growth in 2025 did not diffuse.

The ICT question

The ICT sector’s Q4 behaviour has become a defining feature of Zambia’s quarterly GDP profile. The five-year record shows why.

In 2021 and 2022, ICT growth was stable across quarters. Starting in 2023, Q4 began diverging sharply: positive 11.6 percentage points above the Q1–Q3 average in 2023, positive 15.6 in 2024, then negative 34.7 in 2025. The swing from positive 27.4 per cent in Q4 2024 to negative 15.2 per cent in Q4 2025 spans 42.6 percentage points in a single year.

Three layers are relevant and may operate simultaneously.

First, the structural observation. Something changed in Q4 ICT behaviour starting in 2023. The divergence pattern did not exist in 2021 or 2022. Whether this reflects a change in real economic activity (lumpy capital deployments, spectrum allocations, discrete regulatory compliance events) or a change in how activity is captured across quarters is not determinable from published data.

Second, base effects. ICT grew 24.6 per cent in Q4 2023 and 27.4 per cent in Q4 2024. If base effects were the primary driver of the Q4 2025 reversal, they should have dampened Q4 2024 as well. They did not. Q4 2024 accelerated on an already elevated base. The three-year compound annual growth rate from Q4 2022 to Q4 2025 remains approximately 10.4 per cent. ICT output is still roughly 35 per cent above its Q4 2022 level. The sector has not collapsed. It has corrected from an elevated peak. But the correction alone does not explain a 34.7 percentage point divergence from the Q1-Q3 average, nor does it explain why the Q4 divergence pattern began in 2023 but did not exist in 2021 or 2022.

Third, the measurement question. ZICTA, Zambia’s telecommunications regulator, publishes data annually. It does not produce quarterly indicators. Kenya’s National Bureau of Statistics, by contrast, uses quarterly subscriber, traffic, and revenue data from the Communications Authority to interpolate quarterly ICT GDP. Whether ZamStats has access to equivalent quarterly source data is not publicly documented. If such a gap exists, it may contribute to the Q4 volatility, which is not uncommon in preliminary national accounts where quarterly estimates are later reconciled against annual enterprise data.

For the full year 2025, ICT grew 6.4 per cent and contributed 0.8 percentage points to annual GDP. The sector was a net positive contributor to the year. Airtel Zambia, the largest mobile network operator, published audited results for calendar year 2025: nominal revenue up 26 per cent to ZMW 8.95 billion, subscribers up 12 per cent to 12.45 million, and ZMW 2.66 billion in new property, plant and equipment. But GDP measures gross value added, not revenue. If intermediate consumption surged, as it plausibly did given the 217 per cent spike in services-sector diesel consumption during the 2024 power crisis, the massive capital deployment, and the dollarised cost of international bandwidth, then value added can contract while revenue grows. The sector-specific GDP deflator also differs materially from headline CPI.

The Q4 2025 ICT figure is preliminary. ZamStats has revised preliminary sectoral data in past releases, sometimes materially. The note’s argument does not depend on resolving the ICT question because the wholesale and retail contraction is independently confirmed at firm level by the PMI.

The power story

Some coverage attributed the Q4 growth deceleration to record power outages. The data complicates that narrative.

Electricity GDP grew 38.4 per cent in Q4 2025, contributing 0.3 percentage points to growth. But that figure is itself a base effect. Q4 2024 contracted 50.9 per cent. The arithmetic: 100 multiplied by 0.491 multiplied by 1.384 yields approximately 68. Electricity output in Q4 2025 remained roughly 32 per cent below Q4 2023 levels. The power crisis was easing, but it was not over.

The PMI confirms this at firm level. In October 2025, firms cited inconsistent energy supplies as a constraint on output. In November, load shedding and unstable electricity hampered production for the second consecutive month. Agriculture was the only sector expanding in both output and new orders by Q4. The 38.4 per cent growth in electricity generation GDP did not translate into reliable power at the business level. Generation recovered; distribution did not.

The damage from the 2024 crisis also operated with a lag. The H2 2024 PMI data shows how severe the conditions were. The headline PMI averaged 47.8 in Q3 2024 and 49.3 in Q4 2024. September 2024 hit 45.6, the sharpest deterioration in over four years. Output, new orders, and employment all contracted simultaneously. Firms were cutting jobs, depleting inventories at the fastest rate since 2020, and reporting backlogs at their highest in over four and a half years. By the time the PMI crossed back above 50 in December 2024, output had been contracting for thirteen consecutive months.

That is the condition firms carried into 2025. The sustained load shedding systematically depleted corporate working capital. Businesses that exhausted cash reserves running generators in 2024 entered 2025 without the financial capacity to restock inventory or expand operations. The sectoral timing supports this: wholesale and retail contracted mildly in 2024 (negative 0.9 per cent), when load shedding was at its most severe, then deeply in 2025 (negative 11.2 per cent), when generation was recovering. If power outages alone drove the 2025 contraction, the 2024 figures should have been worse. They were not. The financial residue of the crisis, not the crisis itself, is consistent with the 2025 pattern.

The firm-level evidence

The PMI data provides the interpretive layer that connects the sectoral GDP figures to what firms actually experienced in 2025. Three patterns stand out.

First, the headline PMI overstated the health of the private sector. Output contracted at firm level in eight of fourteen months between January 2025 and February 2026, despite the headline PMI remaining above 50 for eleven of those months. The PMI was held above the expansion threshold by new orders, employment, and inventory building, not by actual production. GDP measures output, not orders. The PMI’s output sub-index measures month-on-month change at firm level, not year-on-year aggregate output; its value is confirming the breadth and persistence of private sector weakness across the year, not quantifying the contraction. The disconnect between the PMI headline and its output sub-index is the pattern of an economy where firms positioned for growth that did not arrive.

Second, firms hired ahead of demand that never materialised. Employment expanded in eleven of fourteen months. November 2025 recorded the fastest hiring since January 2018. Yet output fell that same month. Firms built capacity for an expansion that the cost squeeze, power instability, and weakening demand did not support. By January 2026, they began cutting staff for the first time in nine months.

Third, the pricing data reveals the margin compression in real time. In Q1 2025, firms could pass rising costs through to customers. By October, they were cutting selling prices at the fastest pace in four years while their own input costs (fuel, electricity, wages) were rising. By January 2026, selling prices were falling at their steepest rate since September 2021. The balance shifted across the year: in Q4, weakening domestic demand was the primary pressure on margins; by early 2026, the further acceleration of the Kwacha to ZMW 18.80, a further 15 per cent appreciation in three months, made import competition the dominant force. Different mechanisms, same outcome: firms absorbing losses to maintain volume.

The February 2026 PMI, at 49.3, reinforces the shift. New orders fell at their sharpest pace since October 2024. Manufacturing, construction, and wholesale and retail all contracted. The driver was no longer power, inflation, or drought residue. It was the Kwacha appreciation making imports cheaper than domestic production. If this persists, the 2026 growth story faces a different constraint from 2025, and the private sector enters another year without the conditions for sustained expansion.

What this means

Zambia’s economy grew 3.8 per cent in 2025. That is slower than every major institution projected. Agriculture contributed 2.2 percentage points of annual growth. Mining added 0.5. Transport and manufacturing contributed meaningfully. These sectors performed. The economy is not stagnant.

But the domestic commercial economy, retail, services, small enterprise, the part that connects growth to household spending, contracted. Wholesale and retail subtracted 1.8 percentage points from annual growth. The PMI never broke convincingly above 50. Firms hired for expansion that did not arrive, cut prices to sustain volume they could not afford to lose, and entered 2026 facing a new threat from import competition.

The ICT sector’s Q4 volatility amplified the headline and shaped how the number was read. But it did not cause the underlying weakness. Remove ICT entirely and the picture remains: a narrow growth base, a contracting domestic commercial sector, and a private sector that spent the year at the waterline.

This is not new. For over a decade, the PMI has hovered near its long-run average of 48.9, never building the sustained momentum above 50 that signals genuine private sector expansion. The 2025 data made visible in the national accounts what the PMI had been documenting at firm level for years.

Three things would change the pattern.

First, reliable power at the distribution level. The PMI data suggests this constraint eased in early 2026, but distribution bottlenecks persisted into Q4 2025 even as generation recovered. The distinction between generation capacity and delivered supply remains the binding gap.

Second, a pace of Kwacha appreciation that domestic producers can absorb. The consumer benefit is real. But a 35 per cent move in twelve months, with a further 15 per cent in the first quarter of 2026, compresses margins faster than firms can adjust. The threat is the velocity, not the direction.

Third, transmission reform. Private sector credit grew 15.7 per cent in 2025, down from 21.4 per cent, and what remained was concentrated: approximately 65 per cent of total lending flows to the top 20 borrowers (Bank of Zambia data, December 2025), and only around 10 per cent of firms hold a bank loan or line of credit (World Bank Enterprise Survey, 2019). Cutting the policy rate does not reach the firms whose balance sheets were depleted by the 2024 crisis. What reaches them is structural reform of the lending channel: pricing transparency, collateral reform, and the legislative measures currently before Parliament. The risk is that fiscal pressure in an election year pushes monetary policy toward convenience rather than transmission. The IMF’s March 2026 staff visit identified early signs of fiscal slippage from spending pressures, including the wage bill, agricultural support, and election-related expenditure. The revenue side compounds the problem: if wholesale and retail activity remains depressed, the domestic tax base that funds the materially larger 2026 budget will underperform. Should the March 2026 bond auction, in which bid volume fell 85 per cent from February, not be a one-off, that pressure augments quite fast.

The 2025 projections overestimated because the models assumed diffusion that the structure did not deliver. The three conditions above would change the immediate pattern. But none of them are new. Power constraints, currency volatility, and broken credit transmission have shaped the private sector environment for years. What 2025 revealed is what happens when all three bind simultaneously. The question we asked in November 2025 remains the right one: who actually drives economic growth in Zambia?

Sourcing note

The analysis in this essay draws on ZamStats Monthly publications (Volume 228 through Volume 276, covering March 2022 to March 2026), Bank of Zambia Monetary Policy Reports (November 2025, February 2026), Stanbic Bank Zambia PMI press releases (September 2024 through February 2026, compiled by S&P Global from a panel of approximately 400 private sector firms), ZICTA Annual Market Reports, Airtel Zambia audited financial results (year ended December 2025), IMF staff statements and Executive Board reviews, and CEIC and FocusEconomics exchange rate data. All sectoral GDP figures are year-on-year growth rates at constant 2010 prices as published by the Zambia Statistics Agency. ICT quarterly figures for 2021 and 2022 are from their respective original publications; 2023 figures are revised (March 2025 publication); 2024 figures are final; 2025 figures are preliminary.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

Canary Compass is also available on Facebook: @CanaryCompassFacebook.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu, X @InfinitelyDean, or Facebook @DeanNathanielOnyambu

Image: AI-generated illustration of carried frames becoming visible across four domains

Sudan has more pyramids than Egypt. More than 200 of them, scattered across Meroe, Nuri, and Jebel Barkal. Kingdoms that once ruled Egypt itself built them across eight centuries. Almost nobody mentions them.

The Hebrew scriptures were oral before they were written. For centuries, the Torah moved voice to voice, generation to generation, long before ink met parchment. Every serious scholar treats this as legitimate pre-literary theology. In the East African interior, entire kingdoms maintained their histories, their laws, their cosmologies the same way. The dominant scholarly framing treats this as evidence that complex civilisation had not yet arrived. Same medium. Different verdict.

When you borrow money from a bank in Zambia, you see two numbers: the central bank rate and your margin. The rate is public. The margin is a black box. Inside it sit six distinct cost layers, each driven by a different input, each governed by a different policy lever. The person who pays for them has no way to see which layer moved, by how much, or why. I recently published an essay decomposing each layer to its regulatory source. They have always been there. They were just invisible to the person paying for them.

I have believed since secondary school that I work best under pressure. Read broadly, absorb widely, compress everything into the final window before a deadline. The pattern followed me through university, through 18 years in financial markets, through every major publication I have produced. I have never questioned it. It has always delivered.

Four scenes. Four different domains. History, scripture, banking, a personal habit formed decades ago. The thread underneath them is the same.

Intelligence analysis uses a framework that divides knowledge into four categories: what you know you know, what you know you do not know, what you do not know you know, and what you do not know you do not know. Most people build their confidence in the first two. I know my field. I know where my gaps are. That feels like the complete map.

The third category is where the ground shifts. These are the things you carry without realising you carry them. The assumption that civilisation requires monuments. The assumption that oral tradition is theology when it comes from one tradition and folklore when it comes from another. The invisible architecture inside a price you pay every month. A habit you have never examined because the results kept coming. None of these feel like assumptions. They feel like facts. And each one arrived from somewhere, at some point, and stayed because it was never examined.

When you start to see the carried frames, the discomfort is immediate. Most people respond in one of two ways. They run harder: refresh the feed, add another analysis, check the numbers one more time. Or they retreat to familiar ground, where the map feels complete and confidence feels earned.

There is a third response. Stillness. Sitting in the discomfort long enough to let it do what discomfort is designed to do: stretch the capacity of what you can see. The connection between a Sudanese pyramid and a lending margin and a study habit does not reveal itself under pressure. It surfaces when you stop.

I have always trusted the pressure more than the preparation that preceded it. This week, I ran multiple review cycles on that essay, coming back to edit it just a few hours after publication. What each cycle surfaced was not new. The pressure was the extraction method. It was never the source.

The search determines the finding. And the frame you carry determines the search. What you have never examined, you will never see. What you have never questioned, you will mistake for truth. The pyramids were always in Sudan. The layers were always inside the margin. The knowledge was always underneath the pressure. The only thing missing was the willingness to sit still long enough to notice.

Image: AI-generated illustration of a lending rate refracted through a prism into its component layers

0. Executive Summary

On 11 February 2026, the Bank of Zambia cut the Monetary Policy Rate by 75 basis points to 13.50 per cent. By the week ending 27 February, the retail lending rate had increased by 124 basis points, while the corporate lending rate had decreased by 159 basis points relative to three weeks earlier. The retail margin widened by 199 basis points. Every observable system-level cost input either improved or held steady: the policy rate fell, government securities yields were compressing across the curve, and the banking system was at its strongest credit position in over a decade. No disclosed component-level explanation accompanied the movement. This was not an isolated episode. Data from every MPC rate decision since May 2023, when the Bank of Zambia began publishing disaggregated retail margin data, shows a consistent pattern: from August 2023 onward, banks compressed the retail margin during every rate hike; during both rate cuts in the current cycle, banks widened it immediately. The academic literature calls this asymmetric pass-through. This essay calls it a transparency problem. The pattern can be observed from public data. The mechanism behind it cannot. Nothing in Zambia’s regulatory framework requires banks to explain what sits inside the margin. Until that changes, rate cuts intended for borrowers will continue to be absorbed by the institutions that price the loans.

This essay opens the margin. It maps the six cost and return layers inside it, constructs an illustrative waterfall from verified regulatory and market inputs, presents the full monetary policy cycle evidence, and proposes a transparency-and-incentives reform grounded in existing Bank of Zambia authority and tested against Kenya’s operational precedent.

1. The February Finding

On 11 February 2026, the Bank of Zambia cut the Monetary Policy Rate by 75 basis points (one hundredth of a percentage point each) to 13.50 per cent, the second consecutive cut following a 25 basis point reduction in November 2025. The stated purpose was to support the economy as inflation returned toward the 6 to 8 per cent target band.

The Bank of Zambia publishes fortnightly statistics showing the policy rate, the base rate, the lending rate, and the margin for retail and non-retail (corporate) borrowers. In the week ending 6 February, before the cut, the average retail lending rate was 27.54 per cent. By the week ending 27 February, it had reached 28.78 per cent. The policy rate fell by 75 basis points. The retail lending rate rose by 124 basis points. For new borrowers, the rate cut made their loans more expensive.

Corporate borrowers experienced the opposite. Their lending rate fell from 21.87 per cent to 20.28 per cent, a decline of 159 basis points. Corporate borrowers are multibanked and can credibly threaten to move their business. Retail borrowers cannot.

The rate cut moved the base down. The banks moved the margin up. The retail borrower paid more. If the pattern is observable, why is the mechanism invisible?

2. What a Rate Cut Actually Does

When a bank in Zambia offers you a loan, it shows you two numbers: the Bank of Zambia policy rate and your margin. The policy rate is public. The margin is set on the day your loan is approved. For many retail facilities, it stays fixed for the life of the loan. For a corporate borrower with bargaining power, it may be renegotiated at the next review. The margin is the black box.

Your Rate = BoZ Policy Rate + Margin

The bank's internal pricing benchmark is the return it could earn by deploying the same money, for the same duration, in the next-best alternative available to it, typically government securities at the matching tenor. That price has a formal name in economics: the opportunity cost (Hirshleifer, 1956). The bank's Treasury department sets this internal price, known as the funds transfer price, and charges it to the business unit that originates the loan. In practice, the Treasury desk acts as an internal wholesale lender: it prices the funds, and the business unit marks them up to cover its own costs and return target. The margin on your loan agreement is the sum of both.

When government securities yields compress across the curve, as they did in early 2026, the carry income banks earn on their bond portfolios declines. To protect the net interest margin, the rational response, typically directed through the ALCO framework, is to recover lost carry income through the lending margin on new originations. This mechanism operates wherever banks hold large government securities portfolios and face imperfect competition in retail lending (Mishra, Montiel, and Spilimbergo, 2012).

For existing borrowers on floating rate loans, a rate cut passes through mechanically on the base. The margin stays. That part works. For new borrowers, the bank recalculates the margin for each application. If nothing inside the margin has changed, the margin should stay the same. The February episode shows the opposite.

3. The Six Layers Inside Your Margin

If you borrow K100,000 for 12 months at 28.78 per cent, you pay K28,780 in interest. Of that, K13,500 is the policy rate component. The remaining K15,280 is your margin. Here is what sits inside that K15,280.

The margin is a composite sum of six distinct cost and return layers, built on top of the base rate. Three sit on the treasury side: they represent the internal transfer price the business unit pays for the funds. Three sit on the business-unit side: they represent what the business unit adds for credit risk, operating costs, and shareholder returns. The allocation between the two sides varies by institution; some banks embed additional components in the treasury transfer price. Each layer has a different driver and a different policy lever.

3.1 The Term Premium

The policy rate is an overnight rate. Your loan is not overnight. If you borrow for 12 months, the bank must fund that commitment at the corresponding tenor. The 364-day Treasury bill yield stood at 14.00 per cent in the week ending 20 February 2026, 50 basis points above the overnight rate. A borrower taking a five-year loan faces a wider curve premium. At the government bond auction of 13 February 2026, the 5-year yield was 14.98 per cent (annual rate compounded semi-annually, equivalent to an effective annual yield of approximately 15.5 per cent), a spread of roughly two percentage points above the overnight rate.

The 50 basis points represents the observed yield curve spread between the 12-month T-bill and the overnight rate. It is an opportunity cost benchmark: the bank’s alternative to lending you the money is buying the T-bill, so the loan must at minimum match the T-bill yield. In practice, a loan is less liquid than a tradeable government security (the bank cannot sell your loan on a secondary market the way it can sell a T-bill), and this illiquidity carries an additional cost. That liquidity premium is real but not separately observable; in this construction it is absorbed into the residual layer described in Section 3.3.

3.2 The Statutory Reserve Cost

The central bank requires every commercial bank to hold 26 per cent of all deposits at the Bank of Zambia. Of every K100 a bank receives, K26 is immediately sterilised. Since 3 June 2024, when CB Circular No. 11 took effect, the BoZ has permitted banks to meet up to 40 per cent of this requirement using eligible government securities at two-year, three-year, and five-year tenors, purchased in a one-off allocation window. The yields at inception were 17 per cent (two-year), 20 per cent (three-year), and 22 per cent (five-year), as set in the Government Bond Auction 5A/2024/BA of 6 June 2024. The portfolio is fixed: banks cannot replace maturing securities with new purchases for SRR purposes, and deposit growth since June 2024 means the eligible securities now represent a smaller share of the total reserve base.

Of the K26, approximately K15.60 sits at the central bank earning nothing; up to K10.40 per K100 of deposits could be held in eligible government securities at inception in June 2024, earning yields of 17 to 22 per cent depending on tenor. Because the portfolio is fixed and deposits have grown since the allocation window, the eligible securities now cover a smaller share of the total reserve requirement than the original 40 per cent, increasing the unremunerated portion. The effective sterilisation cost is the income foregone on the unremunerated portion, allocated through the FTP framework across the bank's earning assets. At the system's 46.3 per cent loan-to-deposit ratio (end-December 2025), banks deploy more of their deposit base in government securities than in private lending. Commercial banks hold government securities equivalent to 51 per cent of their local currency deposits while genuine kwacha private credit reaches 7.4 per cent of GDP (end-September 2025). The allocation is rational: government paper carries a zero per cent risk weight while retail lending attracts 75 per cent, creating a capital cost differential that reinforces the preference for securities over lending and reduces the incentive to compete on retail pricing. The precise sterilisation cost varies by bank depending on the tenor mix and purchase yields of its SRR-eligible securities portfolio.

3.3 The Country-Specific Adjustment

Banks operating in every jurisdiction face a country-specific cost layer, but its magnitude reflects the sovereign credit environment. For subsidiaries of international banking groups in Zambia, that charge reflects a Caa2-rated sovereign with a recent default history, repatriation risk, and parent-entity capital allocation requirements, and is set explicitly by the parent’s transfer pricing framework. The non-renewal of Zambia’s IMF Extended Credit Facility after its expiry in January 2026 may have prompted parent groups to revise this charge upward, making it potentially the one cost input that shifted against borrowers in the February window. If so, it applies equally to retail and non-retail lending and cannot account for the divergence between the two. This practice is standard across international banking groups operating in emerging markets, documented in the FTP literature (Tumasyan, 2012) and observable in the published financial statements of international subsidiaries, where lending spreads in higher-risk jurisdictions are typically wider than those in the parent’s home market. Moody’s upgraded Zambia from Ca to Caa2 with positive outlook in April 2025, following debt restructuring.

Domestic banks do not receive a country risk charge from a parent group. Their shareholders price the equivalent through the return on equity they demand for keeping capital deployed in Zambia, at that institution, rather than in alternative deployments domestically or offshore, which flows through the capital charge in Section 3.6 rather than as a separate FTP layer. This does not mean domestic banks offer cheaper loans. While their cost architecture differs, market competition and offsetting funding advantages force convergence at the final lending rate.

For retail borrowers, this distinction matters less than for corporates: the retail borrower lacks the bargaining power to force the bank to absorb any component of its internal cost structure, regardless of ownership.

In the waterfall construction, this layer is derived as the residual after all other layers are subtracted. It therefore absorbs the country-specific adjustment where it is explicitly priced, the loan liquidity premium described in Section 3.1, estimation error from the operating cost allocation and expected credit loss estimates, and any bank-specific pricing discretion.

3.4 Expected Credit Loss

The bank estimates that some borrowers will default and prices this expected loss into every loan, distinct from the capital charge (which covers unexpected losses) and the country-specific adjustment (which covers the sovereign environment). For a salaried borrower whose salary is domiciled at the lending bank, the expected credit loss is perhaps 1.0 to 1.5 per cent, reflecting a probability of default of 2 to 3 per cent against an unsecured loss given default of 40 to 50 per cent.

Banks price credit risk on a forward-looking basis under IFRS 9, incorporating macroeconomic scenario weightings that may diverge from the current observed NPL ratio. The system-wide NPL ratio stood at 2.8 per cent at December 2025, the lowest in over a decade. Provisions covered 110 per cent of non-performing loans, consistent with banks maintaining reserves as the NPL ratio improved rather than releasing them. Forward-looking IFRS 9 models may still require additional coverage for the performing book under adverse macroeconomic scenarios, but the backward-looking indicator suggests the system was not under acute credit stress at the time of the cut. The October 2025 Financial Stability Report showed 3.5 per cent NPL and 99 per cent coverage; the continued decline in the NPL ratio to 2.8 per cent by December 2025 confirms the trajectory.

3.5 Operating Cost Allocation

Branches, staff, technology, compliance departments, and the infrastructure required to operate a banking licence in Zambia must all be recovered. The sector’s cost-to-income ratio was 49.4 per cent at end-December 2025. Nearly half of every kwacha of operating income was consumed by operating costs. A rate cut does not make branches cheaper.

3.6 The Capital Charge

The regulator requires banks to hold equity against every loan. Under the BoZ’s Credit Risk RWA Directives (2025), a qualifying unsecured retail loan attracts a 75 per cent risk weight. At this weight, allocated equity per K100 of retail lending is approximately 19.4 per cent (75 per cent of the system capital adequacy ratio of 25.8 per cent). The sector earned a cumulative return on equity of 27.7 per cent in 2025. The waterfall uses the realised system ROE as a proxy for the return target banks price into retail loans. Individual banks set segment-specific hurdle rates that may differ from the system-wide outcome. The capital charge is 5.36 percentage points.

The rationale is straightforward. For every K100 lent, the risk weight determines how much of the loan the regulator treats as risk exposure (K75). The capital adequacy ratio determines how much equity the bank must hold against that exposure (K19.35). The return on equity determines what the shareholders demand for tying up that capital in this loan rather than deploying it elsewhere (K5.36). The borrower pays that cost through the lending rate.

The system capital adequacy ratio of 25.8 per cent is well above the 13 per cent published statutory minimum-plus-buffer (10 per cent total regulatory capital under Rule 6(2) of SI 62, plus a 3 per cent capital conservation buffer under Rule 8(1)). Individual banks may face higher effective requirements through bank-specific ICAAP assessments under Rule 20, internal management buffers, and parent group capital standards, though these are not publicly disclosed. The gap between the published minimum and the system ratio reflects the composition of bank balance sheets: heavy holdings of zero-risk-weight government securities compress the denominator (risk-weighted assets), mechanically inflating the capital ratio. It also reflects conservative payout ratios and regulatory preference for capital retention in a highly profitable sector. The high ratio increases the capital charge in every loan because banks price off the actual return on equity applied to the actual allocated capital, regardless of whether that capital level was chosen or is a structural artefact.

4. The Illustrative Waterfall

Sensitivity: if operating costs are 2.50, 3.50, or 4.50 per cent, the residual adjusts to 3.50, 2.50, or 1.50 per cent respectively. The waterfall is illustrative. Individual bank pricing reflects institution-specific FTP frameworks, credit models, and return targets.

5. The Full Monetary Policy Cycle

The February 2026 episode is the essay's sharpest evidence, but it is not the only evidence. The academic literature on asymmetric interest rate pass-through, beginning with Hannan and Berger's (1991) evidence on price rigidity in banking and extended to lending rates by Cottarelli and Kourelis (1994) and Gambacorta (2008), predicts that lending rates adjust faster upward than downward. The Zambian data tells a more specific story.

The table tells three stories.

First, the retail lending rate rose in every single rate hike episode (seven of seven) and also rose in one of two rate cut episodes. The May 2023 episode is an exception to the compression pattern: the margin widened by 30 basis points during a 25 basis point hike, the first MPC decision captured in the disaggregated data. From August 2023 onward, every hike produced margin compression. During the February 2026 cut, the margin widened by more than the base fell, so the rate rose despite the cut. The mechanism differs in each direction, but the outcome for the retail borrower was the same: a higher rate in eight of nine decisions.

Second, during the larger rate hikes from August 2023 onward, banks compressed the retail margin. The 150 basis point hike in February 2024 saw 110 basis points of margin compression. The 100 basis point hikes saw 80 to 85 basis points of compression. Banks absorbed part of each hike. Over the seven consecutive hikes from August 2023 to February 2025, the retail margin declined from 15.92 per cent to 14.18 per cent, a compression of 174 basis points. During the pause between February and November 2025, the margin compressed a further 61 basis points to 13.57 per cent without any change in the policy rate. Across the tightening and pause periods covered by the disaggregated data, the compression totalled 235 basis points.

Third, during both rate cuts, banks widened the margin immediately. In November 2025 (25 basis points), the margin widened by 21 basis points. In February 2026 (75 basis points), the margin widened by 199 basis points. Within the 21-day observation window, banks recovered approximately 73 per cent of what took two and a half years to compress. The credit environment was improving throughout the entire period covered by this table: the gross NPL ratio fell from 5.2 per cent in March 2023 to 2.8 per cent in December 2025, while provisions coverage remained above 85 per cent throughout, consistent with banks maintaining reserves as the NPL ratio improved rather than releasing them. A mild uptick in mid-2024 (to 4.8 per cent in September 2024) reversed within two quarters. By the time of both rate cuts, the banking system was at its strongest credit position in the entire observation window. The margin widened into an improving credit environment, which is difficult to reconcile with defensive pricing. Banking sector performance indicators for 2026 are not yet available.

The non-retail data reveals the structure of the asymmetry. In eight of nine decisions, both margins moved in the same direction. During hikes, both compressed, though non-retail margins compressed more aggressively (the non-retail margin fell from 14.86 per cent to 8.21 per cent across the tightening and pause periods, a compression of 665 basis points, nearly three times the retail compression of 235 basis points). Corporate treasurers, multibanked and able to threaten reallocation, extracted more of each hike. During the November 2025 cut, both margins widened (retail by 21 basis points, non-retail by 77 basis points).

February 2026 is the only decision in which the margins diverged. The retail margin widened by 199 basis points. The non-retail margin compressed by a further 84 basis points. The February 2026 divergence occurred within the same institutions, on the same day, under the same rate cut and the same cost structure. The country risk charge may have risen following the expiry of Zambia’s IMF Extended Credit Facility in January 2026, but any such revision applies identically to retail and non-retail lending within the same institution and cannot explain the divergence.

The co-movement during eight decisions proves the cost inputs are shared. The single divergence, on the largest cut in the cycle, isolates bargaining power as the variable that determines which borrowers absorb the recovery. Lending concentration reinforces this: the twenty largest loans accounted for 65.6 per cent of total lending at December 2025. The corporate book is concentrated among a small number of large, multibanked borrowers with the bargaining power to resist margin widening. Retail borrowers have no equivalent leverage.

The yield environment adds a layer to the margin story. Between May 2023 and February 2026, the policy rate rose from 9.50 per cent to 14.50 per cent and then fell to 13.50 per cent. Over the same period, the 5-year government bond yield fell from 24.00 per cent to 14.98 per cent, a compression of approximately 900 basis points. The carry spread, the gap between what banks earn on their bond portfolios and what they pay for overnight funding, collapsed from roughly 1,450 basis points to 148 basis points. The policy rate and the yield curve are different instruments driven by different forces, and they do not always move together.

What the yield data shows is that margin compression slowed as the yield environment tightened. During the first five decisions (May 2023 through May 2024), when the 5-year yield held between 22.00 and 24.50 per cent, banks compressed the retail margin by 37 to 110 basis points per decision. As the yield curve compressed from November 2024 onward, margin compression slowed to 28 to 32 basis points per hike and reversed entirely during the rate cuts. The banking sector's government securities portfolio grew substantially through 2025, partly as banks increased volume to compensate for declining per-unit yields. That strategy faces a ceiling: sovereign concentration limits and balance sheet capacity constrain how far volume growth can go. At the same time, securities within the existing portfolio mature and must be reinvested at prevailing yields, mechanically eroding the carry income on the stock. The combined effect, a volume strategy that exhausts itself while the existing book reprices downward, concentrates the pressure on the lending margin. The slow-moving components of the margin (credit risk, operating costs, capital requirements) cannot account for movements that track the yield environment within 21-day windows. The component that adjusts is one that cannot be estimated from any public data.

This essay identifies observable correlations between yield curve movements, lending concentration, and margin behaviour. It does not establish causation. Establishing causation requires the component-level data that disclosure would produce. The argument for disclosure does not depend on proving the cause. It depends on the fact that no one, including the regulator, can currently determine the cause from public data.

The counter-argument is predictable: “We compressed margins during hikes to protect borrowers. We are restoring them during cuts.” The data supports the first half of that statement for both customer types. The question the data cannot answer is: which component did you compress during hikes, and which component are you restoring during cuts? Was it the credit risk charge? The operating cost allocation? The return target? The country-specific adjustment? And if the restoration was legitimate, why did it apply only to retail borrowers while corporates, at the same banks, on the same day, saw their margins compress further?

Nobody can answer these questions. Not the borrower, not the journalist, not the IMF. The Bank of Zambia’s supervisory processes can access internal pricing governance structures through on-site examinations, which review ALCO frameworks, credit policies, and risk management structures as part of the statutory mandate. These examinations expose the bank’s internal pricing architecture to supervisory scrutiny, though the primary focus of on-site examination is prudential soundness rather than consumer pricing fairness. The information exists inside the regulatory relationship. What does not exist is public disclosure. The borrower signing the loan agreement cannot see what the supervisor can access. The market cannot discipline what is not visible to it.

That is the transparency problem. The data shows the pattern. Component disclosure shows the mechanism. Without both, the pattern is observable but unexplainable. The essay is consistent with the interpretation that margin behaviour reflects protection of net interest income in an opaque pricing environment. Component disclosure would distinguish between legitimate cost recovery and unexplained margin expansion. Its absence means neither can be tested.

The prevailing conditions at the time of the February 2026 cut reinforce this interpretation. The banking sector indicators described in Sections 3.4 through 3.6 show a system that was well-capitalised, profitable, liquid, and experiencing the lowest credit stress in over a decade. These are stock figures, not three-week movements. They describe the environment in which the margin widened, not the cause of the widening.

The compositional shift explanation deserves particular attention because of what Table 07 measures. The weighted average margin is aggregated across all retail loan products and tenors. If banks originated a higher share of longer-tenor or higher-risk retail loans in the weeks following the cut (unsecured personal loans rather than secured staff facilities, for example), the weighted-average margin would widen mathematically even without any bank changing its pricing on any individual product. This is a structurally valid explanation for any single observation, including February 2026. It cannot, however, explain the directional consistency across nine decisions. Compositional noise does not systematically produce margin compression during every hike and margin widening during both cuts. If composition were driving the February 2026 result, similar compositional effects would appear during hikes, where a week of heavy personal loan origination could widen the margin against the trend. That does not appear in the data. The compositional argument also struggles with the February 2026 divergence. Explaining the simultaneous retail widening and non-retail compression through composition alone would require two opposing shifts in origination mix, retail skewing toward higher-margin products while non-retail skews toward lower-margin products, occurring in the same weeks, at the same institutions, under the same rate cut. That is possible but compounds the improbability.

Other alternative explanations exist: ALCO repricing lags, forward-looking IFRS 9 adjustments where expected losses under macroeconomic scenario weightings diverge from the observed NPL ratio, and a potential upward revision to the country risk charge following the expiry of the IMF ECF in January 2026. Without component disclosure, none of these explanations can be tested against the alternative.

6. Kenya: What Comparison Shopping Can and Cannot Do

Kenya offers the closest regional precedent. The Central Bank of Kenya cut its benchmark rate from a peak of 13.00 per cent to 8.75 per cent by February 2026, a cumulative reduction of 425 basis points across ten consecutive meetings beginning in August 2024. The average commercial banks’ lending rate fell from 17.22 per cent in November 2024 to 14.80 per cent by January 2026, a decline of 242 basis points. CBK data reported by Business Daily indicated that through the early stages of the easing cycle, only a minority of commercial banks passed through the full rate reduction, while a significant number raised their rates.

Kenya built the transparency infrastructure Zambia lacks, and the margin still absorbed the cuts. The Kenya Bankers Association, in partnership with the Central Bank of Kenya, has operated a centralised Total Cost of Credit platform since 2017, publishing per-bank lending rates by product category, fees, and a standardised loan calculator. The CBK publishes per-bank disaggregated rates monthly, creating a regular public accountability cycle where national media report which bank is cheapest and which is most expensive, by name, every month.

In September 2025, the CBK went further. The revised Risk-Based Credit Pricing Model requires all new variable-rate loans to be priced as KESONIA (the Kenya Shilling Overnight Interbank Average) plus a premium, where the premium reflects the bank’s operating costs, return to shareholders, and the borrower’s risk profile. Banks must define the components that make up this premium. They must develop their internal pricing models, secure board approval, and submit them to the CBK. They must publish their weighted average lending rates, weighted average premium, and all fees for each loan product on the TCC website and on their own websites.

Kenya is not telling banks what to charge. Kenya is saying: prove the methodology behind the number.

Zambia’s starting position is closer to Kenya’s than it appears. Zambian loan documents already price as Bank of Zambia policy rate plus margin, with fees separate. The BoZ publishes per-bank lending rates, margins above the policy rate, and fees in the Consolidated Bank Charges Publication, though the most recent edition contains data as at June 2024, making it twenty-one months stale by the time of this essay. What Zambia lacks is the frequency (the BoZ publication is semi-annual at best; the CBK publishes monthly), the consumer-facing platform (the Bankers Association of Zambia has no equivalent of the TCC), and the component layer (neither the BoZ nor any Zambian regulation requires banks to define, disclose, or justify the components inside the premium).

The prevailing Banking and Financial Services Act (No. 7 of 2017, as amended in 2020) and the Cost of Borrowing Regulations (SI 179 of 1995) require disclosure of the total cost of credit as a single annual percentage rate. They do not require disclosure of its components. The Banking and Financial Services Bill, 2025 (No. 36 of 2025), currently progressing through the National Assembly, includes market conduct and financial consumer protection among its objects and gives the Bank of Zambia explicit authority under Section 110(3) to “determine the manner and form, and content of information, required to be disclosed.” If enacted, the BoZ could use this provision to mandate component disclosure. But the Bill does not require it. The gap this essay identifies would persist even after the new Act takes effect, unless the BoZ exercises its discretion to close it.

7. What Must Change

The reform has three layers. The first two are proven in Kenya. The third goes further.

The first layer is frequency and platform. The BoZ should publish per-bank disaggregated lending rates, margins, and fees monthly rather than at the current semi-annual frequency. The Bankers Association of Zambia should build a consumer-facing comparison platform equivalent to Kenya's TCC, developed in partnership with the BoZ as the KBA did with the CBK. Both are operational tasks with established precedents.

The second layer is component disclosure. Every bank should publish, quarterly, the weighted components of its lending margin by product category (personal loans, mortgages, SME facilities, vehicle finance, overdraft facilities, agricultural credit): the funding cost, credit risk charge, operating cost allocation, capital charge, and residual. When the weighted average margin on any product category moves by more than 75 basis points in any quarter, the bank should publish a written explanation identifying which component drove the movement. The threshold is derived from the retail margin data. Across nine MPC decisions since May 2023, retail margin movements cluster into two groups: five between 21 and 37 basis points, and four between 80 and 199 basis points. No observation falls between 38 and 79 basis points. Any threshold within that gap produces the same classification of the nine episodes. The proposed threshold is set at 75 basis points, near the top of the gap, for two reasons. First, the underlying observations are 21-day movements; on a quarterly reporting cycle, smaller weekly movements can compound, and a higher threshold reduces the risk of triggering explanations for accumulated noise. Second, 75 basis points translates to approximately K750 per K100,000 of annual borrowing, a change the borrower would notice on a loan statement. The threshold is a proposed starting point, not a permanent calibration. The nine observations span less than three years, and the gap between the two clusters may narrow as the dataset grows. The 21-day measurement window used here differs from the quarterly reporting cycle the proposal prescribes, and product-level distributions may differ from the system-wide aggregate. The BoZ should review the threshold once quarterly product-level data accumulates. Non-retail margins move more aggressively, but corporate borrowers are individually priced through relationship negotiation rather than published product-level rates. The retail threshold addresses the opacity that retail borrowers face.

Kenya requires banks to submit their internal pricing models to the CBK for board-level approval and to publish the components of the premium. The proposal here goes further: it adds the quarterly explanation trigger and the incentive layer that follows.

Banks may argue that margin components are commercially sensitive. The argument is weak. Banks already study each other’s funding structures and pricing through published financial statements, and large corporates already compare offers across institutions. The current opacity only leaves households, small firms, and depositors without visibility. The proposal asks for weighted averages by product category, not proprietary formulas. Product categories with insufficient facility volumes to protect borrower confidentiality could be aggregated or exempted from granular disclosure. The proposal closes the gap between what banks know about each other and what the public can see.

The Bank of Zambia’s supervisory processes can access internal pricing governance structures and risk management frameworks through on-site examinations. The reform does not ask the BoZ to obtain information it does not have. It asks the BoZ to mandate public disclosure of pricing components that banks already calculate internally. The implementation cost is disclosure, not discovery.

The BoZ holds the regulatory authority to mandate this. Under the prevailing Banking and Financial Services Act, 2017, the Bank of Zambia can require financial service providers to submit information on any matter relating to their affairs or business (Section 151) and may publish that information where it considers it necessary or appropriate (Section 153). The authority to prescribe the manner, form, and content of pricing disclosure rests with the Minister, in consultation with the Bank, under Section 106(3). The pending Bill transfers this authority directly to the Bank under Section 110(3), removing the requirement for ministerial prescription. The Public Disclosure Directives under Rule 21 of the Banking and Financial Services (Capital Adequacy) Rules, 2025 (Statutory Instrument No. 62 of 2025), operationalised by the Banking and Financial Services (Public Disclosure of Capital and Risk Exposures) Directives, 2025 (Gazette Notice No. 1199 of 2025, Special Gazette No. 116 of 19 September 2025), provide an additional basis, though Rule 21 was designed for disclosure of capital and risk exposures. A purpose-built directive explicitly referencing the pricing of financial products would be more resilient to legal challenge and more operationally precise.

The third layer is incentives. The incentive layer links disclosure to prudential consequences. Banks that demonstrate lending rate pass-through consistent with their disclosed component movements could receive tiered SRR relief, reducing the sterilisation cost. The SRR relief mechanism is achievable under the Bank of Zambia’s existing monetary policy authority and must be calibrated to avoid aggregate sterilisation consequences, which may require a cap on total relief or a phase-in linked to the monetary policy cycle. Banks that fail to explain margin movements above the 75 basis point threshold could face conditional capital buffer requirements. Linking capital buffers to conduct rather than prudential metrics would represent a departure from the current Basel III framework and may require explicit legislative authority, potentially through the pending Banking and Financial Services Bill, 2025. This transforms disclosure from a reporting exercise into a transmission instrument.

The proposal does not ask banks to create new information. It asks them to publish a summary of what their pricing frameworks already calculate. If the margin compression during hikes was driven by genuine cost absorption, the components will show it. If the margin widening during cuts was driven by legitimate risk repricing or cost recovery, the components will show that too. In every case, the borrower, the regulator, and the market can see the answer.

Ask your bank for a breakdown of your margin. If they cannot give you one, ask why.

Sourcing Note

The lending rate and margin data in this essay are sourced to the Bank of Zambia Fortnightly Statistical Bulletin, Table 07 (available at boz.zm/fortnight-statistics.htm). The Bank of Zambia began publishing disaggregated retail and non-retail interest rate data in Table 07 from March 2023. The full-cycle analysis in Table 2 uses the last observation before each MPC decision and the observation 21 days later, applied consistently across all nine episodes from May 2023 onward. The February 2026 observations are from weeks ending 6 February (pre-cut) and 27 February (post-cut). Readers may verify the underlying data by requesting the specific Table 07 bulletin editions from the Bank of Zambia or by contacting the author. The January 2026 monthly average (28.02 per cent retail, 21.90 per cent non-retail) provides baseline context.

Banking sector key performance indicators (NPL ratio 2.8 per cent, provisions 110 per cent, CAR 25.8 per cent, CIR 49.4 per cent, ROE 27.7 per cent, LDR 46.3 per cent) are as at end-December 2025, sourced to the BoZ Banking Sector KPIs (available at boz.zm). The twenty largest loans accounted for 65.6 per cent of total lending at December 2025, from the same source. The gross NPL trajectory from 5.2 per cent (March 2023) to 2.8 per cent (December 2025) and provisions-to-NPL from 85 per cent to 110 per cent over the same period are sourced to the BoZ Banking Sector Performance Indicators spreadsheet. The ROE of 33.8 per cent cited for 2023 is as at end-December 2023 from the same source. The October 2025 Financial Stability Report (3.5 per cent NPL, 99 per cent coverage) provides the last publicly available FSR data point. Commercial banks’ government securities holdings at 51 per cent of local currency deposits and genuine kwacha private credit at 7.4 per cent of GDP are as at end-September 2025, sourced to the author’s analysis in “Structure Before Sentiment,” Part 3 (Canary Compass, 2025), derived from BoZ monetary and financial statistics. The BoZ Consolidated Bank Charges Publication (most recent edition: data as at June 2024, published circa March 2025, available at boz.zm) is cited in Section 6. Government securities yield compression across the curve in early 2026, driven by non-resident inflows following the increase in the non-resident participation cap from 5 to 23 per cent (CB Circular No. 01/2026, effective January 2026), is documented in the author’s working paper “The Forecast Is Not the Evidence” (Canary Compass, March 2026).

The SRR of 26 per cent on both kwacha and foreign currency deposits is set by CB Circular No. 02/2024 (effective 5 February 2024). The 40 per cent eligible government securities carve-out is set by CB Circular No. 11 (effective 3 June 2024). Eligible securities comprised two-year, three-year, and five-year government bonds offered in Government Bond Auction 5A/2024/BA of 6 June 2024, at yields of 17 per cent, 20 per cent, and 22 per cent respectively. The allocation was a one-off window; the eligible portfolio declines as securities mature or are sold.

The capital adequacy framework is set by the Banking and Financial Services (Capital Adequacy) Rules, 2025 (Statutory Instrument No. 62 of 2025, signed 12 September 2025, gazetted in Special Gazette No. 116 of 19 September 2025, effective 1 January 2026). Rule 6(2): 10 per cent total regulatory capital. Rule 8(1): 3 per cent capital conservation buffer. Rule 9(1): CCyB at 0 to 2.5 per cent (currently 0 per cent per the October 2025 FSR). Directive 18(2) of the Banking and Financial Services (Computation of Credit Risk Weighted Assets) Directives, 2025 (Gazette Notice No. 1200 of 2025): 75 per cent risk weight for qualifying retail exposures. Directive 11(1) of the same: 0 per cent risk weight for kwacha-denominated GRZ claims. Rule 21 requires banks to prepare and publish information on capital and risk exposures as the Bank may determine. The Public Disclosure Directives are operationalised by Gazette Notice No. 1199 of 2025 (the Banking and Financial Services (Public Disclosure of Capital and Risk Exposures) Directives, 2025).

The country-specific adjustment reflects a pricing component that international banking groups routinely apply in emerging market jurisdictions, reflecting the sovereign credit environment, repatriation risk, and parent-entity capital allocation requirements, as documented in the FTP literature (Tumasyan, 2012). In the waterfall, it is derived as the residual: the difference between the observed rate (28.78 per cent) and the sum of the five observable or estimated layers (26.21 per cent). It absorbs the country risk charge (where explicitly priced by international subsidiaries), the loan liquidity premium, estimation error from the operating cost allocation and expected credit loss estimates, and any bank-specific pricing discretion. The sensitivity analysis (opex at 2.50, 3.50, or 4.50 per cent producing residuals of 3.57, 2.57, or 1.57 per cent) illustrates the range.

Kenya data: CBK benchmark rate path from a peak of 13.00 per cent, with easing beginning in August 2024, to 8.75 per cent (February 2026) sourced to CBK monetary policy decisions. Average commercial banks’ lending rate of 17.22 per cent in November 2024 sourced to CBK. January 2026 lending rate of 14.80 per cent sourced to CBK and KNBS. Bank-level pass-through data sourced to Business Daily reporting on CBK data (August 2024 to February 2025). The revised Risk-Based Credit Pricing Model (effective 1 September 2025 for new loans, 28 February 2026 for existing variable-rate loans) is sourced to the CBK final revised RBCPM (26 August 2025) and the DLA Piper Africa analysis (2025). Under the final RBCPM, banks are required to publish their weighted average lending rates, weighted average premium, and fees on the Total Cost of Credit website and on their own websites; the CBK publishes KESONIA daily. The Total Cost of Credit platform has been operated by the KBA in partnership with the CBK since 2017.

India: pass-through improvement from 11-19 basis points (Base Rate regime) to 26-47 basis points (MCLR regime) is from Chattopadhyay and Mitra (2023), “Monetary policy transmission in India under the base rate and MCLR regimes: a comparative study,” Humanities and Social Sciences Communications 10:408.

Asymmetric pricing: Hannan and Berger (1991), “The Rigidity of Prices: Evidence from the Banking Industry,” American Economic Review 81: 938-45. Neumark and Sharpe (1992), “Market Structure and the Nature of Price Rigidity: Evidence from the Market for Consumer Deposits,” Quarterly Journal of Economics 107: 657-80. Gambacorta (2008), “How Do Banks Set Interest Rates?” European Economic Review 52(5): 792-819.

Theoretical framework: Hirshleifer (1956) on opportunity cost; Tumasyan (2012) on funds transfer pricing; Ho and Saunders (1981) on the bank as dealer; Saunders and Schumacher (2000) on determinants of interest margins; Cottarelli and Kourelis (1994) on the speed of adjustment of lending rates; Mishra, Montiel, and Spilimbergo (2012) on monetary transmission in low-income countries.

Moody’s Caa2 rating with positive outlook was affirmed in April 2025. Zambia’s IMF Extended Credit Facility expired in January 2026 and was not renewed.

The non-retail margin data in Table 3 is sourced to the same BoZ Fortnightly Statistical Bulletin, Table 07, using the same observation dates and 21-day window methodology as Table 2. The non-retail (corporate) margin and lending rate are published alongside the retail data in Table 07 from March 2023 onward.

The government securities yield data in Table 4 is sourced to the BoZ Treasury Bill and Government Bond Auction Results. The 364-day T-bill yield is the most recent weekly auction observation at or before the MPC decision date. The 5-year government bond yield is the most recent auction cut-off yield. Where the Fortnightly Statistical Bulletin and the official auction results diverge, this table uses the auction cut-off yield as the primary source. The February 2024 5-year bond yield is sourced to GRZ Bond Auction No. 02/2024/BA (16 February 2024, cut-off 22.00 per cent) rather than the time series observation.

The Bank of Zambia publishes the Fortnightly Statistical Bulletin on a fortnightly cycle with a lag. The underlying data is collected and collated weekly. The 21-day observation window used in Tables 2 and 3 uses the weekly data within the fortnightly publication. The weighted average margin reported in Table 07 is aggregated across all retail loan products and tenors; there is no tenor-specific breakdown on the lending side of Table 07.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

Canary Compass is also available on Facebook: @CanaryCompassFacebook.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu, X @InfinitelyDean, or Facebook @DeanNathanielOnyambu

Image: AI-generated Illustration of The Question Underneath the Noise

When you let silence sit long enough, the questions change. The surface questions, the ones about schedules and output and efficiency, lose their urgency. A deeper one takes their place. I ask it of my mentees regularly. Why are you doing what you are doing? I have asked it dozens of times. Recently, I asked it of myself, and it landed differently.

Michael Every, the Rabobank global strategist, frames a version of this for economies. His question is not “what will GDP be?” but “what is GDP for?” In a more contested, zero-sum world, he argues, economies can no longer separate output from purpose. Grand strategy, as he puts it, is statesmanship: deciding what you want your economy to look like before you measure it.

The last two weeks, I kept things quiet. Limited calls, limited social media, limited interaction. The world did not.

The Iran war commentary had settled into a familiar rhythm. You are either for the strikes or against them. You either believe Western media or you reject it. You either support America or you condemn it. Two weeks of commentary, and most of it was running on the same pattern: choose a side, defend it loudly, treat anyone who asks a structural question as a threat. Everyone in the conversation was working with incomplete information. That was never the problem. The problem was how many people treated incomplete information as sufficient for conclusion.

A voter ID bill reached the US Senate floor. The reaction split along tribal lines before the debate even began. Most African nations require proof of citizenship to vote, as do most countries globally, and broad bipartisan majorities of American voters, across racial groups, support the principle. From an African vantage point, it is strange to watch a policy with near-universal support elsewhere get swallowed by domestic tribalism. The institutional context differs, and those differences matter. But the polarisation made genuine inquiry almost impossible. The question I asked was simple: what are we missing here?

Then CAF stripped Senegal of the AFCON title and handed it to Morocco, two months after Senegal won the final on the pitch. A committee, from a boardroom, overturned a result that the world watched. We talk a great deal on this continent about colonialism, imperialism, and corruption. We do not talk nearly enough about incompetence. The man who chairs CAF is one of the wealthiest people in Africa. That does not make the institution competent. It makes the incompetence harder to name, because naming it carries a cost most people are unwilling to pay. Name it, and the response is defence of the man, not interrogation of the institution. People will cite the governance structure without understanding the governance structure they are citing.

In each of these cases, the question that dominated was how to respond, how to win, how to defend. The question that was missing was the one Every asks of economies.

A thread I read this week laid out the case for the US and Israeli strikes on Iran with confidence and granular detail. Coercive linkage, anticipatory self-defence, customary international law. The analysis was structured, the vocabulary was precise, and the conclusion was predetermined. It was advocacy dressed as analysis. The strategic rationale for the strikes may well exist. But the author warned about Gulf escalation, about IRGC desperation targeting ports and terminals and desalination plants, and never once followed that warning to its downstream. Eastern and southern African nations are heavily dependent on fuel imports routed through the Middle East. Since the war began on 28 February, the Strait of Hormuz has effectively closed. Fuel supply chains in import-dependent African economies are fracturing. African governments are scrambling to find alternative sources and competing against wealthier economies for what remains. “Finish the job” is a phrase that sounds decisive until you ask what finishing it actually means, and who absorbs the cost of finishing it. The answer, as usual, includes people who were never part of the conversation.

This same pattern showed up closer to home. Zambia’s Cabinet approved a move toward a 24-hour economy. The logic is sound. Idle capital at night is a real inefficiency. The Copperbelt mines have run round-the-clock shifts for decades. Extending that principle to retail, transport, and services is a defensible ambition. But the distance between a Cabinet announcement and actual round-the-clock economic activity is an infrastructure problem. At the drought peak, residential areas were receiving as little as three to five hours of power a day. Street lighting is patchy. Public transport winds down after dark. Private security costs eat the marginal profit of staying open at night. The policy is directionally correct. The sequencing runs ahead of the foundations.

The same gap between ambition and infrastructure shows up at every scale.

Output has never been easier to produce. Technology generates it. Social media distributes it. The cycle of production and consumption accelerates without anyone stopping to ask what the production serves. When output was difficult, efficiency was a competitive advantage. Now that output is cheap, efficiency alone answers nothing. The question shifts from how much you can produce to what your production is for. Careers, schedules, and operating models face the same structural test that Every poses to economies.

I had spent months building a detailed daily operating plan. Every block accounted for, from morning spiritual practice through market hours through evening study. Precise to the quarter hour. I was solving for efficiency. What I had not settled, with the same precision, was the question underneath the efficiency.

Precision is not purpose. You can optimise a 24-hour day and still not have answered the question underneath it.

That is what the silence gave me. A reordering. The question underneath the schedule is the same question underneath the economy, the same question underneath the commentary. What is this for.

I carry my own tendencies, chief among them the habit of building structure before settling direction, and absorbing the rhythm of whatever environment I operate inside. The discipline I ask of institutions, of analysts, of my own mentees, I had not always applied with the same consistency to my own operating architecture. That includes the rooms I choose to stay in. The social pressure to remain in virtual groups, out of fear of missing something or need to be included, is its own form of noise. If a conversation does not sharpen your thinking, it is dulling it. There is no neutral position, which means there is no reason to remain.

Both questions matter. The first one, how do I run this, is necessary. The problem is that most people stop there.

The second question does not generate engagement. It is not rewarded by the boardroom or the news cycle. But it is the question that determines whether the architecture is pointed at anything worth building.

This weekend, before the next week fills the silence again, it may be worth sitting with Every’s version of it. What is the output for.

AI-generated image: One pipeline, five channels. The divergence is the story.

Price data timestamped to market close, 18 March 2026, and early session 19 March 2026. Figures may have moved by publication.

Most people are tracking Brent crude and calling this an oil shock. That framing is too narrow. What is unfolding across global commodity markets since the US-Israeli strikes on Iran began on 28 February and Iran closed the Strait of Hormuz is not one shock. It is a stack of linked disruptions moving through different channels, at different speeds, hitting different countries at very different severity.

Start with crude. Even at the headline level, the price signal is fractured. Cash Dubai hit a record above USD 150 per barrel on 16 March, with Oman settling at USD 152.58. Brent traded above USD 115 in early session on 19 March. WTI briefly touched USD 100. Urals crude reached nearly USD 99 delivered to Indian ports according to Argus data, a 70 per cent surge since the conflict began. These represent structurally different regional exposure to the same conflict, and they tell you immediately that no single number captures this crisis.

The divergence runs deeper than price. The crude that normally transits the Strait ranges from light grades like Murban to medium and heavy sour barrels: Arab Light, Basra Medium, Kuwait Export. The heavier grades dominate the lost volumes. Asian complex refineries were built to process these specific blends. They cannot easily and cheaply substitute light sweet alternatives like Bonny Light or WTI at scale without yield penalties and reconfiguration. The shortage is not just in volume. It is in the specific chemistry that Asia’s refining infrastructure requires. This is why Dubai crude has blown out to a USD 50 premium over WTI, against a normal spread of USD 5 to 8. Grade mismatch and refinery configuration are driving the spread, amplified by scarcity premia and freight dislocation.

Who absorbs the crude hit

Asia absorbs the sharpest immediate impact. About 84 per cent of crude shipments through Hormuz in 2024 were destined for Asian markets. There is no quick rerouting for volumes of this scale, and the replacement barrels are the wrong grade.

Europe is not insulated. It is badly exposed through gas and middle distillates. Up to 3 million barrels per day of Gulf refining capacity is at risk or shut, and the tightest product markets globally are now in diesel and jet fuel, not crude. Europe has some inventory cover and the IEA’s record 400 million barrel emergency release, but calling it “buffered” overstates the position.

The United States is best placed on supply. Domestic production sits at 13.6 million barrels per day. The Strategic Petroleum Reserve has committed 172 million barrels. Venezuelan heavy crude imports surged to their highest level since late 2024 in the week to 13 March, with Gulf Coast complex refineries absorbing barrels at steep discounts to WTI. But supply insulation is not price insulation. US gasoline is approaching USD 4 per gallon and diesel has crossed USD 5.

The cascade most people are missing

The oil price, however visible, is the easier layer to track. What most commentary misses is the downstream cascade, and this is where the crisis becomes structural.

Sulphur is a byproduct of oil refining and natural gas processing. Around half of global seaborne sulphur trade moves through the Strait. When Gulf refineries shut and gas plants close, sulphur supply does not become expensive. It stops. There is no strategic sulphur reserve. Sulphur feedstock prices have surged sharply since the conflict began, with some regional benchmarks more than doubling.

Sulphur matters because sulphuric acid is how the world processes copper, nickel, cobalt, and zinc. It is how phosphate rock becomes fertiliser. It is also critical in semiconductor fabrication and battery precursor manufacturing. The sulphuric acid market is a byproduct market. You cannot ramp it independently of the upstream processes that generate it. When the Gulf goes offline, the acid supply chain breaks with no short term substitute.

Fertiliser is the next transmission channel. More than 30 per cent of global seaborne fertiliser trade passes through the Strait. Gulf countries account for more than 40 per cent of seaborne urea exports and more than a quarter of global ammonia exports. Urea spot prices have surged nearly 40 per cent to around USD 700 per metric ton since the blockade began. More than 1.1 million metric tons of fertiliser cargo are physically stranded in the Gulf, with major producers including QatarEnergy declaring force majeure. China, which the market expected to fill the gap, has instead moved in the opposite direction: Beijing restricted exports of nitrogen-potassium blends in mid-March, maintained urea export quotas, and industry sources say the bans are unlikely to lift before August.

There is a second fertiliser channel that gets even less attention. Urea and ammonia production require natural gas as feedstock. With Qatari LNG halted and Gulf gas infrastructure under attack, fertiliser plants in India, Pakistan, Bangladesh, and Egypt are cutting output because they cannot source the gas to run their reactors. These are separate transmission channels: the sulphur chain breaks phosphate production, and the gas chain breaks nitrogen production. Both hit simultaneously.

Northern Hemisphere farmers are entering planting season. The food price impact is not hypothetical. It is a matter of weeks.

Africa is not one story

For Africa, this crisis cuts across every layer, but it does not cut evenly. About 600,000 barrels per day of oil products that normally flow to the continent from the Middle East are at risk. African refining capacity had already shrunk by about a third over the past two decades before Dangote came online. No African country is a full member of the International Energy Agency. There is no coordinated emergency reserve system. Many economies are operating on weeks of supply.

Consider Nigeria. The Dangote refinery in Lagos, at 650,000 barrels per day the largest single-train refinery in the world, gives Nigeria a domestic buffer most of Africa lacks. The refinery’s management has pledged to prioritise domestic supply. Yet it receives about five crude cargoes per month from NNPC, well short of the thirteen it needs. To fill the gap, it sources crude internationally. Nigerian grades are running USD 3 to 6 above Brent. After freight, crude lands at between USD 88 and 91 per barrel. Pump prices have surged from around 840 naira to above 1,200 naira per litre. The paradox is direct: Africa’s largest oil producer, with the continent’s largest refinery, cannot fully shield its consumers from a Gulf war.

Now consider Kenya. Kenya does not refine crude. Its Mombasa refinery has been dormant for years. The country imports all of its fuel, roughly 100,000 barrels per day, as finished refined products under a government-to-government deal with Saudi Aramco, Emirates National Oil Company, and Abu Dhabi National Oil Company. The benchmark that underpins Kenya’s Gulf supply contracts is not Brent. The contracts price against S&P Global Platts product assessments for the Arab Gulf region, but those product prices are themselves derived from crude feedstock costs. ADNOC, one of the three G-to-G suppliers, refines Murban crude to produce the diesel and jet fuel Kenya buys. Kenyan energy analysts and media track Murban as the reference crude for these deliveries, and for good reason: Murban futures surged above USD 116 per barrel by early March, while Cash Dubai, the physical crude pricing benchmark for the wider Gulf complex, hit a record above USD 150 on 16 March. The G-to-G arrangement replaced an earlier open tender system and operates on 180-day letters of credit, a structure designed to manage dollar demand and de-risk banks from FX exposure on import financing. But a 180-day LC costs multiples of a 30-day instrument, and in a war-risk environment, confirmation and financing charges widen further. Kenya’s actual fuel cost is therefore the Platts product benchmark plus 180-day LC financing plus ocean freight plus war-risk insurance plus the shilling-dollar exchange rate. The commodity benchmark, freight, and insurance layers have all moved sharply since 28 February. Freight rates have quadrupled. War-risk insurance premiums have jumped tenfold in some lanes. Kenya requires importers to hold just 21 days of operational stock, against the IEA benchmark of 90 days. Bloomberg reports that Kenya’s biggest fuel suppliers are already rationing product, with distributors experiencing stock-outs in rural areas. The loss of a single cargo puts the country on a countdown. Bloomberg data shows Kenya’s top fuel origins as UAE, Oman, and India. The Indian supply may appear to diversify the chain, but Indian refineries, Reliance Jamnagar chief among them, have processed large volumes of Russian and Gulf crude for re-export. Kenya’s India-origin imports therefore carry embedded exposure to those same supply chains. The exposure chain extends through intermediaries; it does not break. East and southern Africa receive about 75 per cent of their fuel imports from the Middle East. The headline Brent price materially understates the actual cost and supply risk these economies face.

Then consider Zambia. Landlocked, with no proven crude reserves, Zambia imports all of its fuel. Since March 2023, the 1,710 kilometre TAZAMA pipeline from Dar es Salaam to Ndola has carried finished diesel rather than crude feedstock, after the government converted the pipeline and Indeni ceased refining operations. The remainder of Zambia’s supply arrives as finished products trucked overland from Dar es Salaam or South Africa. All of it comes from or prices against Gulf benchmarks. Ocean freight to Dar es Salaam has surged, and the underlying commodity price has risen sharply since 28 February. The cost pressure compounds at every point in the chain. On 19 March, the Ministry of Energy disclosed that diesel stocks, including volumes in transit at Kigamboni in Dar es Salaam, stood at 285 million litres, or roughly 56 days of cover at current consumption. Petrol cover was thinner at 23 days, kerosene at just over nine. Two observations: the government’s reassurance cited crude prices rising “from US$78 to US$94 per barrel,” a figure consistent with Brent as of early March. By 19 March, Brent itself had risen above USD 110. Both Zambia’s and Kenya’s fuel pricing formulas use Platts Arab Gulf product assessments as the base cost. Those product assessments are driven by the Gulf crude complex: Cash Dubai, Oman, and Murban, which in normal markets trade within a few dollars of each other. Cash Dubai physical was assessed above USD 150 on 16 March. Oman settled at USD 152.58. Murban futures, which traded above USD 116, sit lower because the futures market prices a resolution that the physical market has not yet seen. The product prices sit higher still, because they include refining margins on top of crude. The gap between the government’s reference point and the market Zambia actually buys from is wider than any single number suggests. And the 56-day diesel figure includes stocks physically located in Tanzania, not yet in the country. If the ocean route into Dar is disrupted or delayed by war-risk repricing, those transit barrels are not available.

On sulphuric acid, Zambia’s position is more nuanced than either the global narrative or the aggregate statistics suggest. Zambia has significant installed capacity for acid production as a byproduct of copper smelting: Kansanshi alone generates more than a million tonnes per year, and the country banned acid exports in September 2025. On paper, that looks comfortable. In practice, the system is fragile. KCM and Mopani are operating well below historical capacity. Planned smelter shutdowns have left acid production choppy, and industry sources indicate KCM faced acid shortfalls in 2025 and was reportedly importing sulphur to supplement captive production. Zambia was already importing acid from South Africa before this crisis began. The Copperbelt historically worked as a cross-dependent system between operations, and that system is degraded. The DRC faces the sharper immediate squeeze: its copper oxide leaching operations depend on imported sulphur from the Middle East, and Robert Friedland, chairman of Ivanhoe Mines, warned on 2 March that over 90 per cent of sulphur imported into Africa comes from the Gulf and that oxide operations could face closure within three weeks. But Zambia’s cushion is thinner than the installed capacity numbers imply, and any prolonged disruption to sulphur or acid supply chains through Dar es Salaam or South Africa would expose that fragility. Zambia’s primary exposure in this crisis, however, runs through fuel import costs, fertiliser input prices, and the pressure these place on foreign exchange reserves at exactly the moment the kwacha can least afford it. Zambia does produce fertiliser domestically, and the picture has shifted meaningfully. Nitrogen Chemicals of Zambia expanded its blending and granulating capacity six-fold to over 430,000 metric tonnes per year by 2025. Separately, United Capital Fertilizer commissioned Zambia’s first domestic urea manufacturing plant in October 2025, a USD 641 million coal-gasification facility with plans to double capacity to 1.6 million tonnes per year. Because UCF synthesises urea from coal rather than natural gas, it is partially insulated from the Gulf gas disruption. The government’s self-sufficiency target is 2026. But the transition is not yet complete. Zambia imported nearly 800,000 tonnes of fertiliser in 2024, and NCZ’s blending operations still require phosphate and potash sourced from global markets. Those markets have tightened sharply: Gulf fertiliser shipments are stranded, urea spot prices have surged nearly 40 per cent to USD 700 per tonne, and China, which the market had expected to fill the supply gap, moved in mid-March to restrict exports of nitrogen-potassium blends while maintaining existing urea export quotas. The Gulf is blocked. China has shut the door. The two largest potential sources of replacement supply are unavailable simultaneously. Under Zambia’s current e-voucher FISP system, higher input prices do not automatically raise the subsidy bill within the current season, but they compress voucher purchasing power, force supplementary procurement at elevated prices, and will feed directly into tender pricing for the next season. The fiscal exposure is real; the transmission is policy-mediated rather than mechanical, which delays the pressure but does not eliminate it.

The architecture question

There is a deeper structural point that has received almost no attention outside specialist maritime circles.

Kinetic force alone does not maintain the effective closure of the Strait of Hormuz. Insurance does. When the P&I clubs belonging to the International Group issued 72-hour cancellation notices for war-risk coverage in early March, they made transit commercially prohibitive for most operators. The Group’s twelve member clubs insure roughly 90 per cent of the world’s ocean-going tonnage. Without their coverage, most commercial ships will not sail, most ports will not accept them, and most banks will not finance cargo.

The US response was not a carrier strike group. It was the announcement of a USD 20 billion maritime reinsurance facility through the International Development Finance Corporation, with Chubb as lead underwriter. The facility has not yet been activated. That distinction matters. The United States has built the mechanism to serve as insurer of last resort for shipping through the world’s most critical chokepoint, but it has not turned it on.

Whether by design or by default, this creates a structural reality: the US now holds a dormant lever over the financial architecture that governs passage through the Strait. Holding a mechanism and activating it are different decisions, and the gap between the two is where the leverage sits. China could conceivably self-insure its own fleet through sovereign-backed arrangements. But for now, every downstream calculation, from sulphur to fertiliser to mining inputs, is contingent on when and whether the insurance architecture is activated.

For those of us who have spent years arguing that financial architecture determines economic outcomes, this should not be surprising. Structure precedes sentiment. And the countries that build the architecture hold the options.

The signal

The headline is oil. The story is what oil enables downstream: fuels, fertiliser, sulphuric acid, mining inputs, food systems. The question most people are not asking is who controls the architecture that determines when those flows resume, and on what terms.

We are not all in this together. We never were.

Sourcing note: Price data from S&P Global Platts, Argus Media, and Bloomberg. Trade flow and capacity data from the US Energy Information Administration, International Fertiliser Development Centre, and UN COMTRADE. African country data from the African Development Bank, Zambian Ministry of Energy (19 March 2026 statement), Kenya’s Energy and Petroleum Regulatory Authority, and company disclosures. Insurance data from the International Group of P&I Clubs, gCaptain, and Reuters.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

Canary Compass is also available on Facebook: @CanaryCompassFacebook.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu, X @InfinitelyDean, or Facebook @DeanNathanielOnyambu

Image: AI-generated conceptual illustration, "The distance between the forecast and the evidence

SECTION 0: Executive Summary

On 11 February 2026, the Bank of Zambia’s Monetary Policy Committee reduced the policy rate by 75 basis points to 13.5 per cent. This paper examines whether the evidence available at the time of the decision supported that magnitude. The finding is that the direction of easing was defensible. The magnitude was not.

The evidence is drawn from published auction results, ZamStats price bulletins, six IMF programme reviews, Ministry of Finance debt service schedules, the 2026 National Budget, and the Bank of Zambia’s own data releases.

The structural context. The T-bill market operates under a pre-existing corridor trap at the short end. The condition was created during the 2023-2025 hiking cycle, when the BoZ raised the policy rate from 9.50 to 14.50 per cent but T-bill yields covered less than half that distance. The corridor floor overtook short-dated T-bill yields, making them unattractive relative to overnight deposits. Pension funds, who hold 49.5 per cent of outstanding Treasury bills, face a parallel disincentive: bonds paying 14.50 to 17.59 per cent dominate T-bills at every tenor (Ministry of Finance, Quarterly Debt Statistical Bulletin, Q4 2025). The 91-day T-bill yield sat below the overnight deposit floor before the February cut, and would have done so under any magnitude. The 75 basis point cut extended this condition from the short end into the medium tenors: the 273-day now sits at the policy rate, and the 364-day, as of the March 2026 auction, offers zero premium above the policy rate. The yield curve from overnight to one year is effectively flat. The entire T-bill curve is inside the trap.

Three consequences of the magnitude. First, the February decision reduced the room available to the May MPC. The corridor trap at the short end predates the February decision: it was created during the hiking cycle when the policy rate outpaced T-bill yields, and it would exist under any magnitude, including zero. What the 75 basis point cut did is extend the trap from the short end into the medium tenors (the 273-day and 364-day now offer zero premium above the policy rate) and leave the May MPC with less distance to work with. The operational consequence is direct: Treasury needs T-bill auctions to subscribe in order to manage the ZMW106 billion domestic borrowing programme. When three of four tenors fail to cover because yields offer insufficient compensation relative to overnight, the government cannot refinance its T-bill book at current yields. The corridor must come down for T-bill yields to offer a meaningful spread over overnight. The plumbing now requires a cut regardless of where inflation sits (2026 National Budget, paragraph 153: the Minister of Finance attributes the domestic debt servicing pressure to the 2015-2021 issuance cycle, with instruments “now maturing hence the big jump in the domestic debt servicing cost, particularly in 2026”). Had the MPC cut 25 basis points to 14.0 per cent, the May meeting would have had two to three months of confirmed within-band data and could have calibrated with evidence in hand. At 14.0 per cent, the 364-day would still have offered spread above the policy rate, preserving the last functioning T-bill tenor for the May MPC to work with. At 13.5 per cent, the 364-day, the last T-bill tenor that attracted meaningful demand, now sits at the policy rate with zero premium. The short end has operated below the corridor floor since the hiking cycle. The corridor must come down for the 364-day to offer compensation again. Second, a front-loading constraint: the government borrowed heavily in the first eight weeks of 2026, consuming 64.9 per cent of its annual domestic borrowing allowance. Treasury issued ZMW30.8 billion in face value across January and February against ZMW16.8 billion in maturities. Issuance must now slow. Smaller auctions compress yields through scarcity, not policy transmission. Third, a governance precedent. The decision was taken without a published forecast error record, without transmission evidence from the preceding cut, at the first monthly reading near the target band ceiling, in an electoral year where the IMF’s final programme assessment explicitly counselled prudence. If that is the evidentiary standard for the largest rate move in years, the standard for the next decision is lower.

The exchange rate dependency. The improved inflation projection, revised down 70 basis points from 7.6 to 6.9 per cent, is attributed by the Bank of Zambia itself to the lagged effects of the exchange rate appreciation. The kwacha appreciated 14.2 per cent between the November and February meetings. That appreciation reflects copper prices, mining sector FX supply, capital flows, and the Currency Directives. The BoZ issued the Currency Directives, and its FX operations (including the sale of mining tax USD receipts into the market) influence the exchange rate. The BoZ has a documented pattern of intervening aggressively in the FX market, as the IMF has flagged across multiple programme reviews, including selling USD in excess of programme allowances. But the ability to sustain any given level is constrained by reserves, and the level at 18.9 is a market outcome that depends on conditions outside monetary policy. The MPC statement attributed the improved projection to the exchange rate appreciation but did not specify what happens to the monetary policy stance if the assumption changes. That is a conditionality gap. At the 2026 budget exchange rate of approximately 20, the current rate of 18.9 compresses FX-linked revenues by approximately 5.5 per cent, a bounded shortfall of ZMW4.5 to 5.5 billion. The fiscal sensitivity is real but conditional on copper and the kwacha path.

The maximum defensible magnitude was 25 basis points, matching the November move. The case for 25 rested on three evidential legs, none of which were met for a larger move: no transmission evidence from the November cut, a single reading near the band ceiling (not inside it), and a CPI instrument whose weights have not been updated in 23 years. Seventy-five basis points was premature and disproportionate.

After 81 months above the target band, with GDP growth projected at 5.8 per cent and no published evidence of a negative output gap, the households who lost 31 to 39 per cent of purchasing power since 2019 needed the rate to settle inside the band long enough for price levels to stabilise. The economy was not in distress. There was no urgency.

A note on terms used throughout. “Transmission” refers to the mechanism by which a change in the policy rate reaches market interest rates and, through them, spending and investment decisions. “Corridor” refers to the Bank of Zambia’s interest rate corridor: the range between the overnight lending facility rate (ceiling) and the overnight deposit facility rate (floor) within which the interbank rate is intended to trade. “Corridor trap” is this paper’s term for the condition in which Treasury bill yields fall at or below the corridor floor, removing the incentive for banks to move capital from overnight deposits into short-dated government securities. “NDF ceiling” refers to the Net Domestic Financing ceiling set in the 2026 budget, which caps the government’s borrowing from domestic sources. “Carry trade” refers to the practice of borrowing in a low-yielding currency (typically US dollars) and investing in a higher-yielding currency (kwacha bonds) to capture the interest rate differential; the trade is profitable if the exchange rate holds but produces losses if the kwacha depreciates. “Pass-through” refers to the speed and magnitude with which exchange rate movements translate into consumer prices. “Sovereign-bank nexus” describes the relationship in which commercial banks hold large amounts of government debt, linking the health of the banking sector to the fiscal position and crowding out lending to the private sector. “Front-loading” refers to the concentration of government borrowing early in the fiscal year, which consumes borrowing headroom and constrains issuance later in the year.

SECTION 1: The Decision and the Question

On 11 February 2026, the Bank of Zambia’s Monetary Policy Committee reduced the Monetary Policy Rate by 75 basis points to 13.5 per cent. In November 2025, the MPC had reduced the rate by 25 basis points to 14.25 per cent, describing the move as cautious and measured. The February decision tripled the magnitude of easing. The MPC’s stated basis was a decline in headline inflation from 11.2 per cent in December 2025 to 9.4 per cent in January 2026, a revised projection placing the 2026 average at 6.9 per cent, and a judgement that the cut represented an appropriate monetary policy stance.

This paper examines whether the evidence available to the MPC on 9 to 10 February 2026 supported the magnitude of that decision.

This paper proposes a specific test for whether a rate cut has been transmitted: Treasury bill yields should move in the direction of the change, with subscription rates at or above 100 per cent, within two auction cycles. The logic is straightforward: if the policy rate signal reaches the domestic market, market rates follow and demand for government securities does not collapse. Two auction cycles (approximately four to six weeks) is a reasonable window for the domestic market to reprice. The November cut failed that test. The February cut failed it again. In Zambia’s financial architecture, the primary transmission channel, private sector credit, is weak to the point of non-function. Broad money at 29 per cent of GDP is too shallow to carry the rate signal into lending behaviour. Commercial banks hold government securities equivalent to 51 per cent of local currency deposits. True kwacha-originated private credit stands at approximately 7.4 per cent of GDP. The IMF, across its Third and Fifth Reviews, described the operational framework as actively hindering transmission. The remaining channels (sovereign yields, exchange rate expectations, and forward guidance) do function, but none required 75 basis points to operate. The bond market compressed on capital flows, not the rate quantum. The carry trade responded to the yield level, not the increment. The forward guidance could have been delivered at 25 basis points with explicit conditionality.

The strongest counter-argument is that the magnitude itself carried a signalling effect: by cutting 75 basis points, the BoZ communicated conviction in the disinflation path, and that conviction attracted capital. The evidence points against this. Uganda’s 10-year yield compressed 225 basis points in six weeks without a 75 basis point policy move. Ghana’s 91-day T-bill fell to 30-year lows. South Africa’s 10-year touched decade lows. The continental compression is driven by a global repricing of African sovereign risk, credit rating trajectories, and carry trade arithmetic, not by the quantum of any individual central bank’s rate decision. The directional signal matters. The magnitude does not change the capital flow dynamics that produced the result.

The paper proceeds as follows. Section 2 sets out the MPC’s stated case and the concurrent data. Sections 3 and 4 examine what the return to the inflation target band means at the price level, and whether the instrument measuring that return can be trusted. Section 5 sets out the IMF’s independent assessment across six programme reviews. Section 6 examines the exchange rate channel through which the improved projection was generated and what it costs. Section 7 presents the February auction evidence. Section 8 documents the information environment. Section 9 places Zambia’s yields in continental context. Section 10 proposes three governance moves. The close renders the finding.

SECTION 2: The MPC’s Stated Case and the Concurrent Data

The Bank of Zambia offered three reasons for the February cut. Headline inflation declined from 11.2 per cent in December 2025 to 9.4 per cent in January 2026. Projections showed faster convergence to the 6 to 8 per cent target band, with the 2026 average revised down 70 basis points to 6.9 per cent. The MPC described the cut as an appropriate monetary policy stance.

The November 2025 cut provides the baseline. On that occasion, the MPC reduced the rate 25 basis points to 14.25 per cent. The language was deliberate: “mindful,” “measured adjustment,” “elevated expectations.” Within weeks, three-month Treasury bill rates rose 50 basis points. Six-month rates rose 45 basis points. The first post-cut auction undersubscribed at 66.1 per cent. Two-year bond yields rose 50 basis points. The market did not follow the rate signal. The February decision tripled that signal in a system whose capacity to carry it had not changed. The MPC’s February statement dropped the cautious language of November entirely.

Between the November and February meetings, several concurrent data points were available alongside the improved inflation trajectory.

The concurrent data did not indicate economic distress. GDP growth was projected at 5.8 per cent. Imports grew on capital goods demand. Reserves rose. Credit growth at 15.7 per cent year-on-year nominal remained above nominal GDP growth; the deceleration reflected balance sheet housekeeping, not weakening demand. The FX intervention reversal and the current account swing are external position signals, not domestic demand signals. The economy was not in distress. There was no urgency requiring a tripling of the easing pace.

The case for the cut

The strongest version of the MPC’s argument deserves to be stated plainly. Inflation had fallen 7.1 percentage points from its 2023 peak. The kwacha had appreciated substantially. The IMF had completed a sixth consecutive successful programme review. Zambia had exited selective default. Credit rating agencies had assigned positive outlooks. The argument that the easing cycle should accelerate was not irrational. The forward-looking objection, that central banks are supposed to act on forecasts rather than trailing data, is valid in principle.

The evidentiary burden for doing so, however, rises in proportion to three conditions. First, when the transmission mechanism is documented as non-functional: a forecast-based cut into a broken transmission system is a bet that cannot be collected through the channel the cut is supposed to operate. Second, when the preceding cut produced no observable evidence of transmission: the November 25 basis point cut was the first test, and it failed. Third, when the instrument measuring the variable the forecast targets carries known structural limitations: the 53.5 per cent food weight flatters the disinflation trajectory by approximately 0.35 percentage points (Section 4).

The distinction between direction and magnitude maps to an unequal risk. A 25 basis point cut that proves too cautious delays stimulus by one MPC cycle: three months of slower easing in an economy growing at 5.8 per cent. A 75 basis point cut that proves too aggressive extends the pre-existing corridor trap into the medium tenors, uses the space the May MPC would need to respond to new data, and sets a precedent that weakens the evidentiary standard for the next decision. Even in a system where broad transmission is broken, the rate cut reaches the narrow segment that does have access to bank credit. Those borrowers receive economic relief in proportion to the magnitude, generate demand, and that demand works against the disinflation the cut was intended to support. A larger cut amplifies this effect through the narrow channel that functions. The cost of being too cautious is a delay. The cost of being too aggressive is a structural distortion that compounds. When the transmission mechanism is broken, the forecast instrument is biased, and the preceding cut produced no evidence, the risk favours caution on magnitude even when the direction is clear.

The purchasing power standard

GDP growth was projected at 5.8 per cent for 2026 against no published evidence of a negative output gap. Four in five Zambian workers operate in the informal sector with no access to bank credit at any price. For those households, the rate cut does not arrive. The economically disenfranchised, the majority of the population, do not hedge, do not shift portfolios, do not refinance. They carry the full weight of the price level.

After 81 months above the target band, the households who lost 31 to 39 per cent of purchasing power since 2019 needed the rate to hold inside the band long enough for price stabilisation to begin. One monthly reading at the boundary of the band is arrival. It is not recovery. The appropriate standard was to allow inflation to settle inside the 6 to 8 per cent range for at least two to three consecutive confirmed monthly readings before increasing the magnitude of easing beyond the 25 basis points already delivered in November. That would have placed the earliest window for a larger cut at May 2026, if March and April data confirmed sustained within-band readings and the November cut had produced observable transmission evidence.

The MPC cut at the moment of arrival, before recovery had begun, before the band had been held, before the price level had provided any meaningful relief. Prices rising more slowly is not the same as prices coming down. The rate returned to the band. The level remained.

SECTION 3: The Price Level

The last time inflation sat inside the Bank of Zambia’s 6 to 8 per cent target band, a 25-kilogramme bag of breakfast mealie meal cost K104.74. That was April 2019. Today, with inflation back inside that band at 7.5 per cent, the same bag costs K302.21. The rate has returned. The price level has not.

Central banks do not target price levels. They target rates of change. An inflation targeting framework, by design, accepts the level and celebrates the rate. This section does not argue that the Bank of Zambia caused the price level gap. Supply shocks, a pandemic, a sovereign default, and a currency crisis would have moved prices regardless. The section argues that the depth of the cumulative price shock since 2019 means the social buffer is exhausted. After 81 months above the target band, any monetary policy move that subsequently proves premature damages the credibility of the framework at the moment it needs to be rebuilt. The constraint operates through expectations, not sympathy. When households have experienced seven years of prices outrunning their income, the tolerance for error is the distance between where the price level sits and where the public believes the central bank can hold it. That distance is the widest it has been since the framework was adopted.

April 2019: the anchor

April 2019 is the correct starting point: the last month the headline rate sat inside the 6 to 8 per cent target band before the departure that lasted until February 2026. The comparison uses eight commodities drawn from ZamStats national average price tables for April 2019 (Table 9, Monthly Bulletin Vol 193) and February 2026 (Table 7, Volume 275). Selection criteria: identical product description and unit of measure across both bulletins; items representing the primary expenditure categories for low-to-middle income urban households; items whose price movements are not dominated by regulatory intervention. Cement is included because in Zambia’s urban and peri-urban economy, informal construction and home improvement represent a genuine and recurring household capital expenditure. The eight items are not a random sample of the 440-item basket. They are selected for welfare significance: the goods that households cannot substitute away from.

Charcoal, the primary cooking fuel for lower-income urban households, has more than tripled. Maize grain has more than tripled. The least damaged item is sugar, at 157 per cent. These are the floor of household survival.

What the target band would have produced

The period from April 2019 to February 2026 is 6 years and 10 months. This is a ceiling exercise, not a forecast. The counterfactual shows the maximum that the Bank of Zambia’s own framework permits: the price level at the target band ceiling, compounded from the last date the rate was inside that band. Everything above that ceiling is, by the Bank’s own stated standard, excess. Compounding each item at 6 per cent (the floor), 7 per cent (midpoint), and 8 per cent (the ceiling) produces the counterfactual.

Every item exceeds the 8 per cent ceiling path. The CPI index sits 44.1 per cent above the most generous interpretation of the Bank of Zambia’s own framework. At the midpoint, 53.5 per cent. At the floor, 63.7 per cent. There is no version of the 6 to 8 per cent target band that produces K302 mealie meal or K158 maize grain.

The February 2026 ZamStats bulletin records negative year-on-year price changes on several staples: breakfast mealie meal fell 16.16 per cent, roller mealie meal fell 19.38 per cent, maize grain fell 30.74 per cent. These are real declines in the rate of change. They are irrelevant to the level. A bag of mealie meal at K302 in February 2026, down 16 per cent from K360, cost K104 in April 2019. The rate returned to the band in February 2026. The instrument that measures it has not been recalibrated since 2009.

SECTION 4: The Instrument the Central Bank Trusts

The Bank of Zambia’s inflation projections rest on a Consumer Price Index whose basket was last updated in 2009. The expenditure weights reflect consumption patterns observed in the 2002/03 Living Conditions Monitoring Survey. They have not changed since.

International guidance is unambiguous. The IMF CPI Manual (2020, Chapter 9) states weights should be updated at least every five years. The ILO resolution (17th ICLS, 2003) goes further: in periods of high inflation, updates should be more frequent. Five years is the minimum. Zambia’s weights are 23 years old. Every comparator country in eastern and southern Africa has rebased within the last five years.

The country the basket measures

These weights describe a country where more than half of household expenditure goes to food, where communication costs barely register, where health expenditure is negligible, and where eating outside the home commands three tenths of one per cent. That country was Zambia in 2002/03. It is not Zambia in 2026.

Zambia’s urban population has risen from 34.7 per cent in 2000 to 44.7 per cent in 2022 (Census 2022). The 2022 LCMS confirms the national food expenditure share has fallen to approximately 46.7 per cent. The basket assigns 53.5 per cent: a gap of 6.8 percentage points. Mobile phone ownership has reached 92.9 per cent. Communication carries 1.3 per cent. Only 33.9 per cent of households have grid electricity; 57.6 per cent rely on charcoal. Housing carries 11.4 per cent; South Africa assigns 24.1 per cent. The informal economy accounts for 76.3 per cent of employment (LCMS 2022). These workers commute, buy prepared food, use mobile money, and pay for transport daily. Transport carries 5.8 per cent; restaurants carry 0.3 per cent.

What happened when other countries corrected the same distortion

Five African countries have rebased their CPI in the past four years. Where food was driving inflation and the old basket overweighted food, the rebase lowered the headline. Nigeria is the clearest case: the food weight fell from 51.8 per cent to 40.1 per cent and the reported rate dropped from 34.80 per cent to 24.48 per cent, a 10.32 percentage point reduction on revised weights alone. Kenya fell 1.5 percentage points. Ghana fell 1.6 percentage points.

Where food was not the dominant driver, the rebase raised the headline. Uganda’s increased by 0.4 percentage points. South Africa’s by 0.2 percentage points.

Where Zambia sits in the cycle

At the peak of the inflationary episode in March 2025, food was inflating at 22.3 per cent and the 53.5 per cent weight amplified that pressure. Zambia matched the Nigeria pattern: a rebase at that moment would have lowered the headline.

The MPC cut in February 2026, not at the peak. By February, the inflationary structure had shifted. Food inflation had fallen to 8.2 per cent, barely 0.7 percentage points above the 7.5 per cent headline. The categories inflating above headline (housing at 8.8 per cent, weight 11.4 per cent; health at 9.6 per cent, weight 0.8 per cent; restaurants at 9.5 per cent, weight 0.3 per cent; and miscellaneous goods at 9.4 per cent, weight 5.0 per cent) are precisely the categories the structural evidence confirms are underweighted. Their combined weight is 17.5 per cent. In South Africa’s rebased basket, the equivalent categories carry approximately 45 per cent.

Zambia in February 2026 matches the Uganda and South Africa pattern. The international evidence indicates the headline would move upward in a rebase, not downward.

A bounded sensitivity analysis quantifies the trajectory effect. Using Q1 2025 quarterly average inflation (overall 16.7 per cent, food 19.6 per cent, non-food 12.7 per cent; Bank of Zambia, Governor’s Media Presentation, November 2025) as the peak and February 2026 (food 8.2 per cent, non-food 6.5 per cent; ZamStats Vol. 275) as the endpoint, the old basket measures headline disinflation of approximately 9.0 percentage points. Under a rebased 46.7 per cent food weight, the measured disinflation is approximately 8.6 percentage points. The overstatement is approximately 0.35 percentage points. Food contributed 68 per cent of the total measured disinflation at old weights, because food was both the fastest-decelerating category and the largest weight. The direction of the bias is structural and holds under all plausible reweight scenarios. The magnitude, 0.35 percentage points, is approximately half of the 70 basis point improvement in the BoZ’s own forecast revision from 7.6 to 6.9 per cent. The instrument flatters the trajectory the MPC relied upon.

The January data showed food-driven persistence, not broad disinflation. Food inflation stood at 10.9 per cent, 1.5 percentage points above headline, contributing 69 per cent of all measured inflation in a basket that already overweights food. The moderation between January and February was real and significant, but the February data was published on 26 February, two weeks after the decision. The MPC calibrated its largest rate move in years to the January reading.

The known bias is in one direction. The unknown bias may reinforce it. The full 12-category expenditure breakdown required to construct updated weights does not exist in any publicly available document. The magnitude of the full rebase effect remains unknowable because the 2022 LCMS expenditure data has not been published in disaggregated form. ZamStats has confirmed that a Household Budget Survey will produce new weights for the forthcoming rebase. The capacity constraints at ZamStats are real and documented. They explain the delay in rebasing. They do not explain why the expenditure data collected in the 2022 LCMS has not been published in disaggregated form. The MPC calibrated the largest move in years to an instrument with an unquantified error bound. That is a governance finding, not a statistical complaint.

SECTION 5: What the IMF Saw

Between June 2023 and February 2026, the IMF conducted six programme reviews of Zambia under the Extended Credit Facility. The record constitutes an independent institutional assessment across the same period this paper examines. Report references: First Review (Country Report No. 23/260), Second Review (No. 24/009), Third Review (No. 24/171), Fourth Review (No. 24/373), Fifth Review (No. 25/194), Sixth Review (No. 2026/021). All available at imf.org.

The sovereign-bank nexus deserves specific attention. By the Sixth Review, banks’ exposure to government debt had risen to 22 per cent of banking sector assets. Measured against local currency deposits, commercial banks held government securities equivalent to 51 per cent as at end-September 2025, with true kwacha-originated private credit at approximately 7.4 per cent of GDP. The IMF’s Finance and Development (March 2026) documents the same dynamic across the continent: the sovereign-bank nexus in African low-income countries grew 4.8 percentage points between 2019 and 2024, the fastest of any group globally (Sy and Laws, “The New Face of African Debt”).

The easing warning was published on 31 January 2026, nine days before the MPC convened. IMF Acting Chair Nigel Clarke called for “careful monetary policy calibration” (Press Release No. 26/024, 27 January 2026). The staff-level agreement had noted that “the real policy rate is expected to increase in the coming months as inflation decelerates, hinting toward a tighter monetary policy stance” (Press Release No. 25/431, 18 December 2025). The BoZ responded by tripling the magnitude of the cut.

The IMF’s track record is mixed but the pattern is specific. The Fund’s structural diagnoses (transmission failure, reserve problems, sovereign-bank nexus, FX intervention practices) have been consistent and accurate across six reviews. Its timing forecasts have been wrong: it placed target-band convergence in 2027 when convergence arrived in early 2026. The February decision relied on the timing the IMF missed and ignored the structural warnings the IMF got right.

The same six reviews also record material progress. Zambia restructured approximately 94 per cent of its USD13.3 billion external debt, exiting Selective Default. S&P upgraded the sovereign rating. All three rating agencies assigned positive outlooks. The Sixth Review was the sixth consecutive to reach Board completion. Structural benchmark compliance, however, fell from 66 per cent to 42 per cent as the election approached. No successor programme had been announced and no numeric fiscal anchor had been published.

SECTION 6: The Exchange Rate and the Inflation Projection

The Bank of Zambia’s February MPC statement projects average inflation of 6.9 per cent for 2026, an improvement from the November projection of 7.6 per cent. The Governor’s presentation attributes the improved outlook to “the impact of the lagged effects of the recent appreciation of the exchange rate and expected favourable agricultural output” (Bank of Zambia, Governor’s Media Presentation, 11 February 2026). The exchange rate is the dominant input.

Between the November and February meetings, the kwacha appreciated from approximately 22 to 18.9 against the US dollar. That 14.2 per cent move generated the 70 basis point forecast revision. Zambia’s pass-through from exchange rate to consumer prices is high and immediate, as Structure Before Sentiment Part 1 documented. The kwacha’s level is not a target the Bank of Zambia announces, but the BoZ is not a passive observer. It issued the Currency Directives, it sells mining tax USD receipts into the market, and it has a documented pattern of intervening aggressively in the FX market, including selling USD in excess of programme allowances, as the IMF has flagged across multiple reviews. The ability to sustain any given level is constrained by reserves. The exchange rate at 18.9 is a market outcome that the BoZ’s actions influence but do not determine.

The Currency Directives matter. The revised Currency Directives, effective 26 December 2025, require domestic transactions to be settled in kwacha, with defined exemptions. They create genuine structural demand for the kwacha that previous appreciation cycles lacked. The directives are working. They provide a floor that is structural, not sentiment-driven. This paper supported the Currency Directives in Part 1 and continues to do so.

The fiscal sensitivity. A stronger kwacha compresses kwacha-denominated revenues across FX-linked lines: import VAT, customs duties, excise, mineral royalties, and USD-denominated grants. Import-linked taxes, royalties, and USD-denominated grants together constitute an estimated 40 to 48 per cent of budgeted domestic revenue, derived from the 2026 budget (2026 National Budget, Revenue Estimates). The FX-linked share is calculated from import VAT (estimated at 88 per cent of budgeted VAT, based on the FY25 ratio), customs duty, mineral royalties, export duties, a 30 to 50 per cent FX-linked share of excise duties, and USD-denominated grants.

The 2026 budget was constructed at an implied exchange rate of approximately 20 (derived from the ZMW3,832.4 million programme loan allocation against the USD190 million IMF ECF disbursement, 2026 National Budget, Resource Envelope). At 18.9, the kwacha is approximately 5.5 per cent stronger than the budget assumption.

At 18.9, the estimated revenue shortfall on FX-linked lines ranges from ZMW4.5 billion to ZMW5.5 billion. This is bounded, not existential. If copper overperforms and the kwacha holds, higher USD-denominated production values may offset the exchange rate compression on royalties. If the kwacha strengthens further toward 18.0, the shortfall widens to ZMW8.3 to 9.9 billion. If it weakens toward the budget rate, the compression disappears.

The mineral royalty line illustrates the mechanism at the most transparent single revenue item. Mineral royalties, at ZMW18.2 billion, represent 8.8 per cent of the ZMW206.5 billion domestic revenue budget. At the current exchange rate of 18.9, all else equal, royalty collections compress to approximately ZMW17.2 billion, a shortfall of ZMW1.0 billion. Production growth of approximately 5.8 per cent at current copper prices would close the gap. At an exchange rate of 18.0, required production growth rises to approximately 11.1 per cent. If copper retreats toward consensus averages of USD11,000 to 12,000 per tonne in the second half, the required production growth increases further.

What the MPC statement did not address. The February MPC statement attributed the improved inflation projection to the exchange rate appreciation but did not specify what happens to the monetary policy stance if the exchange rate assumption underlying the forecast changes. That is a conditionality gap. The BoZ influences the exchange rate through its FX operations and the Currency Directives, but the level depends on copper prices, mining sector supply, and capital flows that monetary policy alone cannot sustain. If the exchange rate moves, the projection moves with it.

SECTION 7: The February Evidence

Three auctions across the decision

February brought three auctions that bracket the MPC decision. The Treasury bill auction on 5 February settled the day the MPC began deliberating. The bond auction on 13 February was the first post-cut issuance. The Treasury bill auction on 19 February was the second post-cut test. Together, they provide the cleanest observable test: identical instruments, identical issuer, pre- and post-cut pricing.

Treasury bills: where the domestic market speaks

The Treasury bill market is where domestic institutions price sovereign risk without non-resident influence. On 22 January, total bids at cost reached ZMW3.68 billion against ZMW2.20 billion offered. By 5 February, bids at cost had already fallen to ZMW2.91 billion. The market was weakening before the cut. By 19 February, eight days after the cut, bids at cost collapsed to ZMW1.86 billion against ZMW2.20 billion. Three of four tenors failed to cover: the 91-day drew ZMW139 million against ZMW440 million. The demand trend and the pricing conflict are distinct problems: demand was softening independently on liquidity conditions, and the corridor trap at the short end predates the cut. What the 75 basis point magnitude did is extend the pricing constraint from the short end into the medium tenors.

Bonds: capital flows, not policy transmission

The bond auction on 13 February appeared to tell a different story. Bids reached ZMW21.34 billion against ZMW4.20 billion offered. Yields compressed 40 to 120 basis points across the curve. Of the ZMW9.88 billion allocated, approximately ZMW6.8 billion went to non-resident investors: 69 per cent. Domestic institutions were minority participants in their own sovereign bond auction.

The 5-year received ZMW4.88 billion in bids but was allocated only ZMW42 million. Capital was directed into the 15-year, where ZMW8.82 billion in bids competed for ZMW560 million on offer. This is maturity management at scale: the sovereign needs long-duration paper to avoid recreating the refinancing wall, and only offshore capital bids at the long end in volume.

The liquidity sequence

On 6 February, interbank liquidity stood at ZMW5.16 billion. Some of that cash was likely offshore conversions parking in kwacha ahead of the bond settlement on 16 February. Seven days later, the bond auction absorbed ZMW9.88 billion in face value. The system drained. Six days after that, the 19 February Treasury bill auction could not attract bids equal to the amount offered. The sequence is instructive: transient liquidity from offshore conversions drove bond yield compression; the bond settlement drained that liquidity; the Treasury bill market, emptied of cash, undersubscribed. Liquidity conditions matter more than the rate signal. The February data does not contradict the November finding. It confirms it from the opposite direction.

The corridor trap

The arithmetic explains why banks stayed away from Treasury bills. The MPC rate sits at 13.5 per cent. The corridor runs 12.5 to 14.5 per cent. Now consider the cut-off yields the government accepted on 19 February:

The 91-day sits 160 basis points below the policy rate and 60 below the overnight floor. A bank treasury desk weighing the 91-day at 11.90 per cent against overnight at 12.50 per cent faces negative carry for three months of locked capital. Banks that bid at yields inside the corridor on the 91-day (12.0 to 13.0 per cent) were rejected. The government capped the yield below the overnight floor. Neither side moved. The auction undersubscribed.

Treasury bills also serve as high-quality liquid assets (HQLA) under prudential requirements, generating compliance-driven demand independent of yield. That demand showed up: ZMW135 million on the 91-day. It was not sufficient to cover the auction. At yields below the overnight corridor floor, the carry is negative against the bank’s own funding cost. The HQLA incentive and the yield disincentive pulled in opposite directions. The auction result shows which force was stronger.

A note on responsibility. The corridor trap at the short end is a pre-existing condition. The central bank sets the policy rate and corridor bounds. The issuer sets cut-off yields. The 91-day clearing at 11.9 per cent reflects a cut-off yield the issuer accepted inside a corridor the central bank set. The 91-day trap would exist under any magnitude of cut, including zero. What the 75 basis point cut did is extend the trap from the 91-day into the 273-day and 364-day tenors. A 25 basis point cut would have left the corridor wide enough for the medium tenors to function.

The yield history

The distance between T-bill yields and the policy rate is not a product of the February decision. Historical auction data shows the 91-day traded at 9.50 per cent in mid-2023 when the policy rate was also 9.50 per cent. As the BoZ hiked through 2024 to 14.50 per cent, the 91-day followed but covered less than half the distance, reaching 11.90 per cent by February 2026. The gap was created from above during the hiking cycle as the policy rate outpaced T-bill yields. The easing cycle is now closing it from above, making the compression visible and acute. The 364-day T-bill converged to the policy rate by June 2025, five months before either cut, as part of an independent yield trend. The medium-tenor compression the paper identifies is partly a continuation of that trend, accelerated by the 75 basis point magnitude.

Pension funds hold 49.5 per cent of outstanding Treasury bills; commercial banks hold 44.6 per cent (Ministry of Finance, Quarterly Debt Statistical Bulletin, Q4 2025). Non-residents hold zero. When the T-bill auction undersubscribes, it reflects two investor classes pulling back for different reasons. Banks face negative carry against overnight on the short tenors. Pension funds face an opportunity cost against bonds paying 14.50 to 17.59 per cent. Both dynamics lead to the same outcome: the T-bill market cannot subscribe at yields below the corridor floor. Commercial banks hold 19.9 per cent of government bonds, concentrated in maturities of five years and shorter; beyond five years, the holders are pension funds (23.5 per cent of bonds) and non-resident investors (31.9 per cent of bonds).

Banks fulfil the 26 per cent statutory reserve requirement primarily through government bond and T-bill holdings, which simultaneously serve yield, prudential liquidity, and reserve compliance. The system is saturated in sovereign paper rather than flush in deployable cash: free current account balances average K2 to 6 billion against a banking system with over K100 billion in deposits. On 19 February, the day the T-bill auction undersubscribed, positive current account balances stood at K2.62 billion, down from K14.01 billion on 13 February when offshore conversions had flooded in ahead of bond settlement. Subscription rates at any given auction reflect the cash position of the banking system on that date, not the monetary policy stance (Structure Before Sentiment, Part 4, Section 10).

The stock of domestic debt contracted through government securities stood at K253,727 million (USD11,446 million) as at end-December 2025 (Ministry of Finance, Quarterly Debt Statistical Bulletin, Q4 2025). The Minister of Finance attributes the domestic debt servicing pressure to the 2015-2021 issuance cycle: “These instruments are now maturing hence the big jump in the domestic debt servicing cost, particularly in 2026” (2026 National Budget Address, paragraph 153). The market issuance requirement for 2026 is ZMW106.07 billion, with a further ZMW48.01 billion in interest expense funded from revenues (Section 7, auction data). Each maturity event returns kwacha to the banking system. New issuance reabsorbs it. The cycle is continuous. Bond auctions settle monthly between fortnightly T-bill auctions, creating a predictable cash cycle in which bond settlements tighten free cash ahead of the subsequent T-bill auction. This is a design feature of the auction calendar, not a February event. The question of whether the mid-corridor framework remains appropriate for Zambia’s liquidity environment is beyond this paper’s scope but warrants dedicated analysis (Structure Before Sentiment, Part 3).

The flat curve

The yield curve from overnight to one year is now effectively flat. The 364-day T-bill, at 13.50 per cent as of the March 2026 auction, offers 100 basis points above the overnight floor for twelve months of locked capital: approximately 8 basis points per month of additional compensation. In February, the same tenor offered 150 basis points, or 12.5 basis points per month. The compensation for duration is shrinking at each auction. The strong bidding interest at the 364-day in March (2.44 times subscription) reflects front-loaded demand from institutions capturing the last available spread before the next cut compresses it further, not durable appetite for the tenor. When that spread narrows to the point where the 364-day no longer compensates for the lock-up, the last functioning T-bill tenor stops working. At that point, no T-bill maturity offers a meaningful premium over overnight, and the government’s ability to fund its rolling T-bill requirement through the domestic market depends entirely on a further cut to the policy rate to push the corridor down.

The dominant mechanism to release the trap is to cut the policy rate further, pushing the corridor down until T-bill yields offer a meaningful spread over overnight. The 91-day has operated below the corridor floor since the hiking cycle and will do so under any rate level. The funding pressure is specific to the current state: the 364-day, the last tenor that attracted meaningful demand, now sits at the policy rate with zero premium. The corridor must come down for it to offer compensation again. The May MPC faces a choice in which the corridor arithmetic, not the inflation evidence, may drive the rate decision. That is the structural consequence of the magnitude: the 75 basis points pushed the medium tenors into the trap, and the funding requirement, not the data, now compels the next cut.

What the 75 basis point cut did is extend the trap from the 91-day, where it would exist under any scenario, into the 273-day and 364-day tenors. A 25 basis point cut would have left the corridor wide enough for the medium tenors to function. The magnitude widened the dysfunction. It did not create the underlying pressure.

There is a simpler reading. In a falling-rate environment, investors rotate into duration to capture repricing gains. Capital moves from the short end to the long end. The bond oversubscription and T-bill undersubscription are two sides of the same portfolio decision. That reading explains the long end. It does not explain the short end. If investors expect yields to fall, the rational response is to lock in current yields before the next cut compresses them further. The falling-rate thesis predicts oversubscription at the short end, not undersubscription. More importantly, it does not explain why banks bid at yields inside the corridor on the 91-day (12.0 to 13.0 per cent) and the government rejected them. That is a pricing conflict between issuer and market, not a portfolio preference. The corridor created the conflict. The auction result measured it.

The front-loading constraint

By 13 February, Treasury had issued ZMW28.98 billion in face value against ZMW14.46 billion in maturities. Net above maturities: approximately ZMW14.5 billion. Against an NDF ceiling of ZMW21.62 billion, 64.9 per cent of annual headroom is consumed in six weeks.

Auctions must shrink. Smaller auctions with residual demand compress yields through scarcity, not through any policy mechanism. Falling yields over March to May may appear to vindicate the 75 basis points. The mechanism will be supply-side.

The auction data

The annual market issuance requirement is ZMW106.07 billion: ZMW73.19 billion in bond maturities at cost, ZMW11.25 billion in Q4 2025 Treasury bills maturing in 2026, and ZMW21.62 billion in net domestic financing. Interest expense of ZMW48.01 billion is funded from revenues, not issuance. January absorbed ZMW12.65 billion. February’s three auctions update that position.

T-bill 03/2026, before the MPC met, subscribed at 1.19 times. T-bill 04/2026, after the cut, subscribed at 0.76 times. Demand halved across the decision. The 364-day absorbed the bulk of both auctions: ZMW1,574 million at cost in the pre-cut auction, ZMW1,237 million in the post-cut auction. Short-tenor demand was negligible at both dates, but the post-cut collapse was concentrated precisely where the corridor trap operates.

Subscription: 5.08 times. According to Reuters, no significant bond principal maturities fell in February. The entire ZMW9,879 million allocation was net new money, expanding the outstanding stock of domestic government bonds. In January, ZMW6,800 million in bonds matured, producing net new bond money of ZMW2,503 million. Cumulative bond net new money through end-February: ZMW12,382 million.

Twenty-four per cent of the annual issuance requirement absorbed in two months. The required monthly average across the remaining ten months falls to ZMW8,081 million. The required average allocation rate against remaining offered amounts at Q1 rates: 89.4 per cent.

Offshore participation and the refinancing wall

Non-resident investors took approximately 49 per cent of the January bond allocation and 69 per cent of February’s. Combined offshore absorption across two auctions: approximately ZMW11,375 million. Against the 23 per cent non-resident cap (ZMW24,395 million), two auctions have consumed 46.6 per cent.

Total non-resident cash outflows in 2026 total ZMW23.52 billion, or USD1.18 billion at an exchange rate of 20 (Ministry of Finance Q3 2025 Debt Service Schedule). The remaining non-resident refinancing requirement is approximately ZMW12,145 million. Ten bond auctions remain.

The composition of the January and February offshore participation matters. Some was new entry: investors converting US dollars to capture the carry trade, drawn by double-digit nominal yields and a kwacha that had appreciated 14.2 per cent. Some was front-loaded rollover: non-residents whose maturities fall due in June or December who moved early, fearing yields would compress and entry points worsen. New entry adds to the stock of offshore exposure and increases the refinancing requirement in future years. Front-loaded rollover reduces the remaining 2026 rollover risk but pulls demand forward from future auctions. The distinction is not visible in the auction results. It is visible in the participation data the Bank of Zambia holds.

The binding constraints

Three constraints now operate simultaneously.

The NDF ceiling is binding. ZMW7,578 million remains of annual headroom. At Q1 offered amounts with full allocation and light maturities of ZMW3,000 to 4,000 million per month, net issuance would run approximately ZMW5,100 million per month. Three months at that pace would exceed the remaining headroom. Treasury must reduce issuance.

The corridor trap constrains the T-bill market. T-bill 04/2026 raised ZMW1,844 million in face value against approximately ZMW2,197 million in maturing bills. The government could not fully refinance its T-bill maturities after the bond auction drained the system. The corridor inversion did not just suppress demand. It produced a net cash outflow on the T-bill book after 13 February.

The offshore cap constrains the bond market from the opposite direction. If non-resident participation continues at February’s 69 per cent, remaining cap headroom exhausts within three to four auctions. If it moderates to January’s 49 per cent, the cap accommodates the remaining ten.

These three constraints interact. The NDF ceiling forces smaller auctions. Smaller auctions with residual demand compress yields through scarcity. Compressed yields weaken the carry trade. A weaker carry trade thins offshore appetite precisely when the remaining ZMW12,145 million in non-resident rollovers must be absorbed. When heavy June maturities arrive, Treasury needs the market at scale again. But by then, yields may have compressed through supply restriction to levels that thin the carry trade and weaken offshore appetite.

Yields compressed 40 to 120 basis points on the bond curve. T-bill yields moved 10 to 50 basis points on a 75 basis point cut. The bond compression was driven by offshore capital. The T-bill compression was driven by reduced supply in an undersubscribed auction. The policy rate sat in the middle. It moved neither market through transmission.

Post-decision update: March 2026

T-bill 05/2026, held on 5 March, is the third post-decision auction. Total bids at cost reached ZMW2,921 million against ZMW2,200 million offered, producing an overall subscription of 1.33 times. The headline crossed 100 per cent for the first time since the cut. The composition tells the fuller story: the 182-day subscribed at 40 per cent (ZMW198 million against ZMW500 million), worse than the February post-cut auction. The 91-day subscribed at 96 per cent. The 364-day absorbed the excess, with ZMW1,778 million in bids at cost against ZMW730 million offered. The 364-day yield compressed a further 50 basis points to 13.50 per cent, now exactly at the policy rate. Zero premium for 12 months of locked capital. Across three consecutive post-decision auctions, the corridor trap has climbed the entire T-bill curve.

SECTION 8: The Information Environment

Every data point cited in this paper carries a release date. The distinction between what the MPC had and what arrived later matters because a finding that relies on post-decision data is vulnerable to the charge of hindsight.

The MPC had the January inflation reading, the IMF’s caution, the November transmission failure, the CPI weight structure, the NDF ceiling, and the January issuance data. It did not have the February inflation reading, the February auction results, or the Ministry of Finance revenue data.

The Ministry of Finance January 2026 monthly economic indicators remained unpublished as of the date of this paper. These would cover the first full month under the revised Currency Directives, the first month of kwacha trading below 20, and the first month in which the fiscal compression described in Section 6 would appear. Their absence means the fiscal picture the BoZ cut into is only partially visible from the public record.

The chain of dependencies makes the absence consequential. The improved inflation projection depends on the exchange rate holding near 18.9. The sustainability of that exchange rate depends in part on fiscal confidence. Fiscal confidence depends on revenue execution data. Revenue execution data for the first month under the new exchange rate regime is absent from the public domain. A market operating without a successor IMF programme, without numeric fiscal anchors, and without monthly revenue data is pricing sovereign risk on incomplete information.

The MPC may have had access to internal Ministry data. But the principle of accountable monetary policy is that decisions should be justifiable on published evidence. The public cannot verify what it cannot see.

SECTION 9: The Yield Question

The question is not whether Zambia’s yields are competitive. They are. The question is what drives them and what the market is pricing that the policy rate is not.

On 13 February, offshore investors bid approximately ZMW21.3 billion into a single bond auction. Of the ZMW9.88 billion allocated, approximately ZMW6.8 billion came from non-residents. Six days later, the T-bill auction undersubscribed. No offshore participation. These two auctions, separated by six days, are the cleanest test of what moves Zambian yields. The bond market moved because offshore capital entered a shallow pool. The T-bill market did not move because the corridor trap, the yield arithmetic, and the liquidity position of the banking system on that date prevent domestic institutions from following the rate signal.

The real yield landscape

Zambia’s real yield of +9.10 per cent on the 10-year and +10.09 per cent on the 15-year rank among the most attractive sovereign exposures on the continent. The disinflation from 16.6 per cent in mid-2023 to 7.5 per cent today, combined with yield compression driven by capital inflows, has materially improved Zambia’s competitive position.

The continental compression

The compression is not unique to Zambia.

Uganda compressed 225 basis points without a 75 basis point policy move. South Africa compressed 65 basis points with no policy change. Nigeria compressed 30 basis points with no rate change. Zambia’s 59 basis points is within the range of its peers. The compression is continental, driven by a global repricing of African sovereign risk (Sy and Laws, IMF Finance and Development, March 2026), not by the quantum of any individual central bank’s rate decision.

Credit ratings and the currency risk premium

USD Eurobond yields are drawn from primary issuance: Kenya priced a dual-tranche Eurobond on 19 February 2026 at 7.875 per cent (7-year) and 8.70 per cent (12-year). Nigeria’s FGN 7.875 per cent February 2032 yielded 6.72 per cent and the 10.375 per cent December 2034 yielded 7.29 per cent (DMO, citing Bloomberg, 26 February 2026). Ivory Coast raised USD1.3 billion in 14-year notes at 7.125 per cent (Reuters, 19 February 2026). South Africa trades at approximately 5.5 to 6.0 per cent at the 10-year equivalent.

Zambia’s spread of approximately 940 basis points is the widest in the peer group. Kenya, one to two notches above at B/Caa1, carries 470 to 480 basis points. Nigeria, at B-/B3, carries approximately 850 basis points. The sensitivity to spread compression is arithmetic. As the rating migrates and the spread narrows, the implied local 10-year yield falls:

At a spread of 600 to 700 basis points, the 10-year converges toward 13 to 14 per cent. At 800 basis points, it settles near 15 per cent. The February auction cleared at 16.60 per cent: 200 to 350 basis points of compression room remains. The 15-year, at 17.59 per cent, retains 150 to 250 basis points against a peer-implied range of 15 to 16 per cent. The compression between January and February is the market beginning this repricing in real time.

What the comparison reveals

Zambia’s real yields are attractive and improving. The compression is rational: a sovereign that has restructured 94 per cent of its external debt and attracted positive outlooks from all three rating agencies should trade tighter. The continental context validates the trend. The bond market and the T-bill market are telling different stories: bonds oversubscribed 5 times on 13 February while T-bills undersubscribed at 84.6 per cent on 19 February. Kenya, which has cut 425 basis points across 10 consecutive meetings with T-bill auctions routinely above 200 per cent coverage, illustrates the difference. That difference is market depth: Kenya’s M3-to-GDP at approximately 40 to 45 per cent versus Zambia’s 29 per cent.

A central bank that cannot price the short end of its own domestic curve through the policy rate should be cautious about claiming credit for what happens at the long end through capital flows. The market will compress Zambia’s spread as the credit rating improves, as the disinflation proves durable, and as the kwacha holds. That compression belongs to the country’s trajectory. It does not belong to 75 basis points.

SECTION 10: Three Moves

The IMF called for projection transparency, fiscal data discipline, and operational framework reform across multiple reviews. Those recommendations remain unimplemented in material respects.

First, projection accountability. The Bank of Zambia should publish its forecast error record: every inflation projection since the adoption of the inflation targeting framework, placed against the outturn. The format exists in peer central banks. The BoZ should publish by the May 2026 MPC and update at every subsequent meeting. If the record shows systematic accuracy, the market should follow the rate signal. If it shows systematic over-optimism, the market is rational to price liquidity rather than the policy rate.

Second, fiscal transparency. The Ministry of Finance should publish revenue outturn data within three weeks of month-end, starting with the January 2026 indicators. The measurable threshold: monthly economic indicators available in the public domain by the 21st of the following month. This is operationally achievable: the data exists within the Treasury systems, and every IMF programme review demonstrates the Ministry can produce these figures when required. In a year with no external programme anchor and the heaviest domestic financing requirement in the country’s history, the market cannot price what it cannot see. Silence prices as a risk premium.

Third, proportionality discipline. Rate decisions should be calibrated to two conditions: confirmed transmission evidence in the domestic T-bill market and sustained within-band inflation across at least two to three consecutive readings. The specific test: post-MPC T-bill auctions should show yields moving in the direction of the change, with subscription rates at or above 100 per cent, within two auction cycles. The MPC should not increase the magnitude of easing beyond the previous move unless transmission evidence from that move has been observed. The November 25 basis point cut produced no such evidence. The February 75 basis point move was therefore disproportionate. This is not a call for permanent incrementalism. It is a principle that magnitude should escalate only when the preceding signal has been confirmed.

SECTION 11: What the Evidence Shows

The question this paper posed was whether the evidence available to the MPC on 9 to 10 February 2026 supported a 75 basis point reduction in the policy rate.

On the direction. Defensible. Inflation had declined materially. The kwacha had appreciated on a structural intervention this paper supported. The IMF projected continued disinflation. A rate cut in February 2026 was not irrational.

On the magnitude. Not supported by the evidence. The transmission record shows that neither the November nor the February cut moved domestic T-bill yields through policy transmission; across three post-decision auctions, the pattern has deepened. The measurement environment shows the CPI instrument flatters the disinflation trajectory by approximately 0.35 percentage points, accounting for roughly half of the 70 basis point forecast revision. The improved projection depends on the exchange rate, which the MPC statement did not condition upon.

On the alternative. A 25 basis point cut, matching the November magnitude, was the maximum the evidence supported. The bond compression would have occurred regardless: offshore capital drove 69 per cent of the allocation and was responding to carry trade arithmetic, not the difference between 25 and 75 basis points.

The corridor arithmetic shows the difference. At 14.0 per cent, the policy rate corridor runs 13.0 to 15.0 per cent. The 91-day trap is a liquidity problem under either scenario. The difference appears at the medium tenors. At 75 basis points, the 273-day yield (13.5 per cent) sits exactly at the policy rate: zero premium for nine months of locked capital. At 25 basis points, a 273-day yield around 13.8 per cent offers a meaningful spread above the 13.0 per cent corridor floor. The 364-day, at approximately 14.3 per cent, offers 30 basis points above the policy rate. The medium curve remains functional. At 75 basis points, even the 273-day offers no compensation for duration risk. The trap extends across the entire Treasury bill curve.

At 14.0 per cent, the May MPC would have had two to three months of confirmed within-band data and could have accelerated with evidence in hand. At 13.5 per cent, the corridor trap extends across the full T-bill curve, leaving the May MPC less room to act. The structural pressure to cut to release the trap exists at any rate level, but the 75 basis point magnitude narrowed the space within which the May MPC can respond to data rather than to the trap. The February decision did not preserve the May MPC’s room to act on evidence. It used it.

The front-loading constraint would have been identical under either scenario: it is a function of the NDF ceiling and the issuance profile, not the rate level. The exchange rate dependency would have been identical: the forecast rests on the kwacha regardless of the policy rate. What the 25 basis point alternative would have changed is the corridor arithmetic (preserving the medium tenors), the governance precedent, and the room available to the May MPC. Those are the consequences of the magnitude.

On the governance precedent. The February decision set a standard. A 75 basis point cut calibrated to a forecast rather than confirmed evidence, delivered into a transmission mechanism the IMF documented as actively hindering monetary policy, measured by an instrument whose weights have not been updated in 23 years, in an electoral year where structural benchmark compliance had fallen from 66 per cent to 42 per cent, and two weeks after the IMF counselled prudence. If that is the evidentiary standard for the largest rate move in years, the standard for the next decision is lower.

The structural pressure to cut to release the corridor trap exists regardless of the February decision. It is a product of the yield arithmetic created during the hiking cycle and the auction calendar’s cash dynamics. What the 75 basis point magnitude did is extend the trap across more tenors, flatten the yield curve from overnight to one year, and narrow the May MPC’s room to respond to data rather than to the corridor. The governance precedent removes the institutional basis for resisting that pressure. The first is a structural condition the paper documents. The second is a warning. Both matter.

On the purchasing power standard. After 81 months above the target band, with GDP growth projected at 5.8 per cent and no published evidence of a negative output gap, the economy was not in distress. The households who carried seven years of price level damage needed the rate to settle inside the band long enough for stabilisation to begin. The MPC cut at the moment of arrival, before recovery had begun, before the band had been held. Even a 25 basis point cut rested on a single reading near the band ceiling. A more prudent window was August 2026, after a full quarter of confirmed within-band data.

The February decision may not have been motivated by anything other than the MPC’s genuine assessment of the inflation outlook. The evidence assembled in this paper does not make a claim about motivation and cannot test it. What the evidence shows is that the decision was not proportionate to what the data supported, and that the institutional environment in which it was taken pointed toward caution on every dimension the paper has examined.

What would change this assessment. This paper’s finding is falsifiable. Three pieces of evidence would weaken or overturn it. First, if T-bill auctions show broad-based subscription above 100 per cent across all tenors (not concentrated in the 364-day) with yields moving in the direction of the cut, the transmission finding is wrong. The March 2026 auction crossed 100 per cent on the headline but the 182-day subscribed at 40 per cent; the pattern has not yet falsified the finding. Second, if the Ministry of Finance publishes revenue outturn showing FX-linked revenues held up despite the appreciation, the exchange rate dependency finding is weaker. The FY2025 outturn, published after this paper’s analysis was completed, provides a partial test: total revenue and grants reached K187.4 billion against a budget of K185.5 billion. Income taxes overperformed at 124.8 per cent of budget, driven by mining corporate tax. However, VAT collected K37.1 billion against a K48.7 billion target (76.2 per cent), with import VAT at 90.6 per cent and domestic VAT at 34.4 per cent. Mineral royalties collected K16.0 billion against K17.2 billion (93.0 per cent). The FX-linked lines underperformed as the paper’s Section 6 analysis predicts; copper profitability compensated through the income tax channel. The fiscal finding is partially confirmed, partially offset. Third, if ZamStats publishes rebased CPI data and the trajectory effect is negligible, the measurement finding loses force. The confidence levels are not uniform: the transmission failure is highest confidence (three auctions, consistent pattern). The CPI trajectory effect is moderate confidence (approximately 0.35 percentage points, directionally certain). The exchange rate dependency is structural and certain; only the fiscal magnitude is conditional on copper and the kwacha path. The governance precedent is an institutional judgement, not a quantitative finding. The lending channel remains fully disconnected: commercial bank average lending rates stood at 28.0 per cent in December 2025 (Ministry of Finance, Monthly Economic Indicators, December 2025), unchanged in substance from 29.0 per cent at the start of the hiking cycle.

The answer the data returns is that a 75 basis point cut into a broken transmission system, at the first monthly reading near the target band ceiling, calibrated to a projection dependent on an exchange rate the central bank influences but cannot guarantee, and measured by an instrument not updated in 23 years, was not proportionate to what the evidence supported at the time of the decision. The direction was defensible. The magnitude was not.

References

Bank of Zambia (2025). Monetary Policy Rate announcement, 18 November 2025. Lusaka.

Bank of Zambia (2026a). Monetary Policy Rate announcement, 11 February 2026. Lusaka.

Bank of Zambia (2026b). GRZ Bond Tender Results 01/2026/BA, January 2026. Lusaka.

Bank of Zambia (2026c). GRZ Bond Tender Results 02/2026/BA, 13 February 2026. Lusaka.

Bank of Zambia (2026d). Treasury Bill Tender Results 02/2026, 03/2026, 04/2026, 05/2026. Lusaka.

Bank of Zambia (2025). Governor’s Presentation to the Media: Monetary Policy Committee Statement for the Third Quarter of 2025, 12 November 2025. Lusaka.

Bank of Zambia (2026e). Governor’s Presentation to the Media: Monetary Policy Committee Statement for the Fourth Quarter of 2025, 11 February 2026. Lusaka.

Bank of Uganda (2026). Government Securities Auction Results, 18 February 2026. Kampala.

Central Statistical Office / ZamStats (2009). Consumer Price Index: Base Year and Methodology. Lusaka.

Central Statistical Office / ZamStats (2023). 2022 Living Conditions Monitoring Survey Report. Lusaka.

Central Statistical Office / ZamStats (2025). Monthly Bulletin, Vol. 193 (April 2019 reference data). Lusaka.

Central Statistical Office / ZamStats (2026a). The Monthly, Vol. 274, January 2026. Lusaka.

Central Statistical Office / ZamStats (2026b). The Monthly, Vol. 275, February 2026. Lusaka.

ECA, UNDP and APRM (2025). Joint Statement on Zambia Debt Restructuring, 6 December 2025.

ILO (2003). Resolution concerning consumer price indices. 17th International Conference of Labour Statisticians. Geneva.

IMF (2020). Consumer Price Index Manual: Concepts and Methods. Washington, DC.

IMF (2023). Zambia: First Review Under the Extended Credit Facility. Country Report No. 23/260. Washington, DC.

IMF (2023). Zambia: Second Review Under the Extended Credit Facility. Country Report No. 24/009. Washington, DC.

IMF (2024a). Zambia: Third Review Under the Extended Credit Facility. Country Report No. 24/171. Washington, DC.

IMF (2024b). Zambia: Fourth Review Under the Extended Credit Facility. Country Report No. 24/373. Washington, DC.

IMF (2025a). Zambia: Fifth Review Under the Extended Credit Facility. Country Report No. 25/194. Washington, DC.

IMF (2025b). Press Release No. 25/431, 18 December 2025. Washington, DC.

IMF (2026a). Zambia: Sixth Review Under the Extended Credit Facility. Country Report No. 2026/021. Washington, DC.

IMF (2026b). Press Release No. 26/024, 27 January 2026. Washington, DC.

Ministry of Finance, Zambia (2025). 2026 National Budget Address, delivered to the National Assembly, 26 September 2025. Lusaka.

Ministry of Finance, Zambia (2025). 2026 National Budget. Lusaka.

Ministry of Finance, Zambia (2025). Quarterly Debt Service Schedule, Q3 2025. Lusaka.

Ministry of Finance, Zambia (2025). Quarterly Debt Statistical Bulletin, Q4 2025. Lusaka.

Ministry of Finance, Zambia (2026). Monthly Economic Indicators Report, December 2025. Lusaka.

Bank of Zambia (2025). Quarterly Debt Statistical Bulletin, Q4 2025. Lusaka.

Bank of Zambia (2026f). Commercial Bank Liquidity Data, July 2023 to February 2026. Lusaka.

Bank of Zambia (2026g). Treasury Bills and GRZ Bonds Rates, July 2023 to February 2026. Lusaka.

Bank of Zambia (2026h). Interest Rate Data (Policy Rate, Lending Rates, Interbank Rates), July 2023 to February 2026. Lusaka.

Nigeria Debt Management Office (2026). Eurobond Closing Prices, 26 February 2026. Abuja.

Bloomberg (2026). ZAMBIN 5¾ 06/30/2033 REGS Corp, bid/ask prices, 6 March 2026.

Onyambu, D. N. (2025). ‘Structure Before Sentiment, Part 1: Zambia’s Monetary Policy Transmission Failure.’ Canary Compass, November 2025.

Onyambu, D. N. (2025). ‘Structure Before Sentiment, Part 3: Foundations for Transmission.’ Canary Compass, December 2025.

Onyambu, D. N. (2025). ‘Structure Before Sentiment, Part 4: The World Will Not Wait.’ Canary Compass, December 2025.

Onyambu, D. N. (2026a). ‘Copper Output and the 2026 Royalty Arithmetic.’ Canary Compass, January 2026.

Onyambu, D. N. (2026b). ‘Zambia 2026 Domestic Market Absorption.’ Canary Compass, February 2026.

S&P Global Ratings (2026). African Sovereign Ratings Summary, February 2026.

Sy, A. and Laws, A. (2026). ‘The New Face of African Debt.’ Finance and Development, March 2026. Washington, DC: International Monetary Fund.

Zambia Statistics Agency (2023). 2022 Census of Population and Housing: National Analytical Report. Lusaka.

Disclaimer

This article does not constitute legal, financial, or investment advice. The author shares views for perspective and discussion only. Do not rely on them as a substitute for professional advice tailored to your specific circumstances. Always consult a qualified legal, financial, investment, or other professional adviser before making decisions based on this content. The analysis reflects proprietary research undertaken by Canary Compass and the author.

Canary Compass and the author accept no liability for actions taken or not taken based on the information in this article.

The views expressed in this article represent the author’s independent professional analysis and do not constitute an endorsement of any individual, institution, or position. Canary Compass and the author accept no responsibility for how this content is interpreted, excerpted, or recontextualised by third parties not involved in its production and publication. Reproducing any portion of this work in isolation, or in combination with other material, in a manner that misrepresents the author’s original meaning constitutes a distortion of the published record.

The author may hold positions in financial instruments, currencies, or assets discussed or referenced in this publication. Such positions do not constitute a recommendation to buy or sell.

All views, projections, and forecasts reflect the author’s assessment at the time of writing. Data sourced from third parties is believed to be reliable but has not been independently verified. Past performance does not indicate future results.

All content published by Canary Compass is the intellectual property of the author. Reproduction, adaptation, or redistribution, in whole or in part, requires written permission.

About the Author

Dean N. Onyambu is the Founder and Chief Editor of Canary Compass, a financial research publication focused on African monetary architecture and financial sovereignty. He brings 18 years of experience across trading, fund leadership, and economic policy, with senior roles at Standard Bank, First Capital Bank, and Opportunik Global Fund.

Read and subscribe at www.canarycompass.com.

The Canary Compass Channel is available on @CanaryCompassWhatsApp for economic and financial market updates on the go.

Canary Compass is also available on Facebook: @CanaryCompassFacebook.

For more insights from Dean, you can follow him on LinkedIn @DeanNOnyambu, X @InfinitelyDean, or Facebook @DeanNathanielOnyambu

Image: AI-generated Illustration of Structure Beneath a Single Word

I was sick last week, so the reflection did not come. By the time I resurfaced, the Iran crisis had taken over every timeline. Military strikes. Retaliation. Nuclear capability. Within hours the world was full of nuclear strategists.

The part that always interests me is the commentary.

If you listen carefully, most of that conversation is running on one frame. The word “nuclear” appears in a headline and the entire situation collapses into a single category. There are actually four separate questions inside this crisis, and each one requires a completely different analytical structure.

The first question is enrichment. Can Iran produce uranium at levels that approach weapons grade? That is a chemistry and engineering question.

The second question is weaponisation. Has Iran assembled, or is it assembling, an actual nuclear weapon? That is a manufacturing and political decision question. Enrichment alone does not answer it.

The third question is delivery. Does Iran have ballistic missile systems capable of carrying a weapon to a target? That is a military logistics question. It has nothing to do with uranium percentages.

The fourth question is intent. Has Iran’s leadership decided to cross the threshold? That is a political and strategic intelligence question. No headline answers it.

The 2025 strikes were aimed at question one. Slow the enrichment pipeline. Extend the breakout timeline. Degrade the physical infrastructure. On a tactical level, the strikes achieved that objective in the short term. Natanz and Isfahan took direct hits. Centrifuge cascades were disrupted. Production capacity dropped. The immediate clock slowed.

But tactical success against question one does not resolve question four. You can damage centrifuges and destroy conversion sites. You cannot bomb away knowledge that has already been acquired. And you cannot eliminate a decision that has not yet been made.

What is happening now has expanded to questions two and three simultaneously. The debate has shifted from enrichment levels to delivery systems, missile launch infrastructure, and broader regional strike capacity. That is a completely different military and political objective from where this started.

Most public commentary does not track that shift. People are debating across questions without realising they have moved between them. Everyone is using the same vocabulary. Almost no one is using the same frame.

The result is confident analysis built on a collapsed category.

That is worth examining. The people speaking are intelligent. They have the vocabulary. What they have not done is separate the four variables. The reason is the environment they are thinking inside of.

The modern information feed has a specific rhythm: headline, reaction, conclusion. Exposure to volume replaces the need for structure. The sequence feels like analysis because it moves quickly and uses the right words. But speed and vocabulary are not the same as a framework. The feed gives you everything except the frame.

Environments shape us in ways we rarely track consciously.

I noticed this first with accents. I grew up in Kenya. I have lived in South Africa, the United Kingdom, and Zambia. Every place left its mark on how I speak. The adjustments were conscious at first. Over time they became automatic. One day you open your mouth and you cannot fully account for what comes out. Accents are absorbed through exposure.

Thinking works the same way.

Spend enough time inside a particular information environment and the rhythm of your reasoning shifts. The feed trains a pattern: move from exposure to conclusion, quickly, confidently. A disciplined analytical practice trains a different pattern: separate the variables, build the structure, then conclude. Both are absorbed. Both become automatic. The question is whether you know which one you are running on.

That produces a strange inversion. The person who has absorbed the feed speaks confidently and sounds authoritative. The person who has spent time building analytical structure sounds artificial, too careful, too layered, too slow. The environment has reversed our intuitions about what competence looks like. Shallow analysis feels normal. Careful analysis sounds generated.

Tools have always done this. The printing press shaped how arguments were constructed. Legal training shapes how lawyers reason through evidence. Financial markets train analysts to think in terms of incentives, probability, and price. Large language models are the newest environment doing what every environment has always done.

I am inside these environments too. My own reasoning carries their imprint. This week reminded me to check whether the structure I was using was built for this problem or borrowed from the last one.

Next time you catch yourself reaching for a single word to describe a complex situation, it may be worth pausing to count what is actually inside it.

When I listen to the Iran commentary, that is the failure I hear. Four questions collapsed into one word. Tactical success confused with strategic resolution. Knowledge treated as equivalent to infrastructure. Intent treated as settled when it remains the least understood variable. And when categories collapse, capital gets allocated against the wrong risk.

The people saying these things are well-read and well-briefed. They are over-absorbed. The modern information environment trains people to sound informed without ever requiring them to build the structure that real analysis demands. That is confidence without architecture.

The frame is yours to build.

Image [Pexels]: Exposed wiring. The architecture comes after.

Every system that gains power eventually reduces tolerance for intensity at its margins. I discovered this firsthand this week while using AI as a thinking tool. I was exploring questions about time, memory, and experiences that sit at the boundary between the rational and the metaphysical. Territory I enter because my internal anchors hold: faith, spiritual discipline, mathematical training. These are load-bearing walls, and they allow me to explore without losing orientation.

Mid-conversation, the system narrowed. The speculative latitude I had enjoyed minutes earlier, when we were dissecting geopolitics and capital flows, quietly disappeared. I noticed, and I said so.

But the pattern was clear: the moment the questions moved from abstract to personally consequential, the latitude narrowed. The system could not distinguish between exploration conducted from stable ground and exploration conducted from unstable ground.

A central bank faces the same constraint. Even with macroprudential tools and targeted facilities, the policy rate remains a blunt instrument at scale. The cost falls on those within reach of the instrument, regardless of whether they are the source of the problem. Any system that grows more capable simultaneously grows more cautious, because it must operate at scale, and scale demands blunt instruments.

The AI offered a formulation I have not stopped thinking about: civilisation can tolerate variance when individuals have internal constraint. If internal guardrails weaken, external guardrails tighten.

That pattern showed up again the same week, in a completely different domain. I assessed a specific monetary policy decision on my LinkedIn platform and the Canary Compass channel. Within days, someone took those words, removed the policy target, and repackaged them as commentary on a different policy entirely. My name, my image, my language, redirected at an argument I never made.

The published record already contradicts the extraction. The January arithmetic on domestic market absorption explains the repackaged policy in question with full data tables. What interests me is the structural operation, because it is identical to what the AI did. A precise signal was produced. A system detected the signal and stripped its context. With the AI, the system tightened to protect stability. With the political actor, the system extracted to gain utility. Different motive, same mechanism: context removed, signal repurposed, cost borne by the originator.

The defence in both cases is the same: build before the extraction arrives. A publication trail with dates, data tables, methodology. A corporate container that separates research from proprietary activity. Disclaimers that delineate where analysis ends and personal positions begin. These exist because the vulnerability is permanent. As long as you publish independent analysis in contested markets, your words will be excerpted. The question is whether the excerpt encounters a record or a vacuum.

And then the pattern appeared a third time, in a domain I did not expect. Earlier this week, I published a post on X examining where value gets captured in Zambia’s copper chain. I called cathode an intermediate product. That raised eyebrows.

In metallurgical terms, cathode is the final refined product: 99.99 per cent pure, LME Grade A, the end of the purification chain. By the classification system taught in mining economics courses across the continent, calling it “intermediate” is a category error.

But the classification was written for an economy where copper wire went into a building. In an economy where copper interconnects go into a GPU that trains an AI model that returns to Zambia as a subscription service at USD20 per month, the distance from cathode to end-use has expanded by orders of magnitude. Intermediate here means intermediate to end use and to high-margin manufacturing, not intermediate within the refining process.

Cathode ends the refining chain and begins the manufacturing one. It is a raw input. Nobody builds anything with a cathode slab. It must be melted and cast into wire rod, strip, or billet, then drawn into wire, rolled into sheet, stamped into connectors, assembled into components, installed in systems. That is where value multiplies. Africa is absent from nearly all of those stages.

My use of “intermediate” was precise within the manufacturing frame. The pushback came from readers applying the metallurgical frame. The manufacturing context was stripped, and the claim was evaluated against a classification system that no longer reflects how value flows through global supply chains.

A legacy framework detected a signal, removed its context, and evaluated it against a reference point the global economy had already outgrown.

The structural point extends beyond copper. The DRC exports the vast majority of its copper as refined metal and still captures almost nothing from the manufacturing economy. Zambia exports predominantly as anode and intermediate refined forms, and even where it exports cathode, it remains at the bottom of the manufacturing value stack. Its electrolytic capacity was severely reduced over time, including through refinery shutdowns under Glencore and Vedanta.

The precious metals embedded in those anodes, the gold and silver in the anode slime, travel with the copper to wherever the final refining occurs. Where anodes leave the country, the slime value is captured entirely at the destination refinery. That value leakage is rarely visible in headline trade statistics because it is captured inside the refining operation, not reported as a separate export.

Africa’s entire policy debate about mineral value addition operates within the refining band of a much wider value chain. The policy discussion treats the metallurgical boundary as if it were an economic boundary. The two are different things.

The Forced Choice examines how Africa’s positioning between absorber and surplus economies is being determined now, under rupture conditions, with limited time. The Cathode Economy, the second article in the Mineral Trilogy currently in development, builds the case mineral by mineral, with the chemistry, the byproduct economics, and the manufacturing stages the current debate omits. The tweet planted the flag. The article builds the fortress.

While words were being extracted and repurposed in one space this week, the real analytical work continued in another. Conversations about maturity profiles, net domestic financing ceilings, whether 100 per cent allocation signals funding pressure or strategic front-loading. That is what precision looks like. It does not fit in a screenshot.

The full quantitative assessment of the February MPC decision is coming soon, alongside the follow-up to the January domestic market absorption baseline. The delay is deliberate. You do not assess a month’s auction dynamics before the month’s data is complete. The analysis follows the data.

Three different systems. Three different scales. A technology tightens to protect stability. A political actor extracts to gain utility. A legacy classification misframes because the economy outgrew the definition. In each case, context is removed from signal. The cost scales with the system: a constrained conversation, a contested reputation, a mispriced continent.

The response is construction. Publish the quantitative work, on your timeline, with your methodology, to a standard where a central bank could circulate it without edits. Build the internal discipline that lets you explore territory others avoid, because your anchoring holds where theirs might not. And recognise that a classification system written for a simpler economy still determines how a continent values its endowment.

Structure before sentiment. Always.

AI-generated image of the signal you override

My father pulled me aside once and said something I have never forgotten.

“You and I have the same problem. We trust too easily. And bad people will always take advantage of that.”

He was calm when he said it. He was naming something he recognised in himself and saw forming in me. A wiring. The instinct to extend trust before it has been earned, to assume shared ethics in a room where the other person is operating on a different ledger entirely.

I carried that observation for years without fully understanding it. I thought awareness was enough. I thought knowing about the wiring would protect me from it.

It did not.

Last year, someone I trusted took my work and presented it as theirs. A person I had helped, whose published work carried structural contributions I had made. For years, people around me had warned me. More than one. They said be careful. They said the dynamic was not what I thought it was. I dismissed every warning because the friendship felt more real than the pattern forming underneath it.

The deepest recognition was not the plagiarism. It was that my father had already given me the diagnosis, and I still overrode the signal. I edited my own reading of the situation to protect a conclusion I preferred. That the relationship was mutual, that the investment was shared, that the ethic was the same.

It was not.

And then came the second override. People around me said take the high road. Let it go. Be the bigger person. So I stayed silent longer than I should have. But accountability does not dissolve because you choose quiet. It defers. And deferred accountability does not surface as measured correction. It erupts, volcanic and disproportionate, harder to govern than it would have been if you had simply acted when the signal first demanded it. The high road, taken repeatedly out of avoidance rather than genuine grace, is just another override. A different signal ignored. A different cost deferred.

That failure has sat with me all week. Because midweek, I watched an institution do the same thing in public.

The Bank of Zambia cut the monetary policy rate by seventy-five basis points.

I had taken the week off. Fatigue demanded honesty, and I needed space to think about Canary Compass and what structural changes it requires. The plan was stillness. Then the decision landed.

The data did not support the magnitude. Inflation has reached single digits, and people are celebrating as though the cost of living has healed. It has not. Prices are rising more slowly. That is not the same as prices coming down. It does not undo what households absorbed over years of erosion. And the people who absorb most directly, the economically disenfranchised, who represent the majority of the population, do not hedge, do not shift portfolios, do not refinance. They carry the full weight.

The architecture does not support the transmission. This is an economy where four in five workers operate in the informal sector with no access to bank credit at any price, and the credit that exists within the formal system is concentrated. Government securities absorb a large share of bank assets and set a floor under lending rates that the policy rate alone cannot move. A lower policy rate in that structure does not function as broad-based stimulus. It delivers selective benefit to the narrow band of borrowers who already have access.

We already have evidence for this. When the Bank cut by twenty-five basis points in November, short-end Treasury bill rates rose, the first post-cut auction was undersubscribed, and the two-year yield moved in the opposite direction. The transmission architecture did not carry even a modest signal to the broader economy. This cut is three times that size, and the structure has not changed. For the narrow band of formal borrowers who already have access, some easing will register. But that selective activity generates demand that tempers the very disinflation the majority is being asked to celebrate, and the cost of that slower disinflation falls entirely on the households who never saw the benefit of the cut. The policy serves a fraction and invoices the rest.

Twenty-five basis points would have achieved the headline. The median forecast in a Reuters poll of economists was precisely that, a twenty-five basis point cut. The Bank of Zambia delivered three times the consensus estimate. This was not a marginal difference in reading the same data. The magnitude itself was the outlier. Seventy-five reads as a signal aimed at political comfort rather than economic discipline. To their credit, the statutory reserve requirement was left unchanged, which suggests awareness of the more dominant inflationary channel. But that single anchor does not resolve the central question: what is the purpose of an aggressive cut in a monetary system that cannot carry it broadly?

There is a counter-argument, and it deserves honest engagement. My own published work identifies the fiscal channel, not the interest rate channel, as the stronger inflation driver. If the Bank of Zambia protected the stronger lever and moved the weaker one, the case could be made that the cut does minimal damage while delivering a political headline. On pure mechanics, that logic holds. The question is not whether the cut causes immediate damage through a weak channel. The question is what the magnitude reveals about how the institution makes decisions. But credibility is not built on mechanics alone. It is built on signal.

Twenty-five basis points would have carried the same tactical logic at a magnitude that signals caution. Seventy-five signals that the political room was louder than the data warranted. The real danger is precedent. Favourable conditions make aggressive moves feel safe. The kwacha is stronger. Inflation is in single digits. Foreign investors are buying bonds. Each of those facts is real. Each is also being read as permission rather than tested as structure. Single-digit inflation still sits above the target band where it has been since April 2019. One strong auction does not resolve a refinancing challenge that runs the full year. Conditions that feel supportive can be fragile, incomplete, or narrow, and an institution that moves further than the architecture warrants because the surface feels calm will find the pattern difficult to reverse when conditions shift. Each decision looks defensible in isolation. The accumulated pattern tells a different story.

I watched the 2020 to 2021 period closely. I watched institutions that should have served as checks drift into short-term alignment. The cost of those decisions did not arrive immediately. The bill was patient. It always is.

Sometimes it is fine to be the only voice in the room saying the uncomfortable thing. If you are right, and you know you are right because you have tested your thinking critically, reality will eventually back you. Slowly and without courtesy, but eventually.

This is a lonely position to hold, especially in environments where people position themselves for appointments rather than speak plainly.

On the Canary Compass channel, I said the central bank has lost credibility. That is a harsh verdict. Let me state it with the precision the claim demands. This decision damages credibility because it breaks proportion between data and action. When decisions appear shaped by the political room rather than the economic structure, credibility does not collapse in a single moment. It erodes in the quiet revision of what the institution was willing to see. And when that erosion reaches a threshold, the cost lands where it always lands. On the currency, on risk premia, on the households least equipped to absorb another shock.

Next week, I will publish the full analytical case. The transmission arithmetic, the distributional reality of who this cut actually reaches, the inflation trajectory, and the historical parallels. The data will settle it. This week, the reflection is enough.

My father’s observation was not about one friendship. It was about wiring.

Some people are built to override the signal. The relationship in the room feels more credible than the pattern underneath it. The political alignment feels more real than the transmission arithmetic. The friendship feels more durable than the warnings accumulating around it. This is a specific kind of loyalty. Loyalty to the conclusion you prefer over the signal you have already received.

Institutions carry the same wiring. Central banks are not staffed by fools. They are staffed by professionals who operate inside rooms where political proximity feels weightier than structural constraint. The override is not a single lapse. It is a culture. A repeated willingness to edit interpretation in favour of comfort. And the cost compounds quietly until the gap between what the institution says and what the structure reveals becomes the thing everyone sees.

That is how credibility dies. In people and in institutions. Through the accumulated habit of choosing comfort over clarity.

My father was right. I do trust too easily. That will not change. What can change is the structure around the wiring. The analytical architecture, the publication record, the independence that turns a harsh verdict into a testable claim rather than an opinion. That is what I am building.

The signal is there. The question is whether you act on what you already see, or whether loyalty to comfort invoices you later.