The Arithmetic Doesn’t Close

I. The season that did not open

In June, in Abidjan, the presidents of Ghana and Côte d’Ivoire agreed that their 2026/27 cocoa seasons would open on the same day: 1 September. Côte d’Ivoire opened on time, at 1,200 CFA francs a kilogramme. More than three weeks later, Ghana had announced neither a price nor a date on which buying would begin. When the price came, on 25 September, it was GH¢2,650 a bag, 2.4% more than February’s, and the season opened the same day.

The delay was not paperwork. In mid-September the licensed buying companies, which buy the crop from farmers, said COCOBOD still owed them about GH¢4 billion for last season’s beans, that banks had stopped lending to them, and that nobody had told them how the new season would be paid for. “We are in the second week of September,” the Chamber of Cocoa Marketers said. “Nobody has called us to tell us anything about what is on.” The terms of COCOBOD’s GH¢16.3 billion domestic financing programme, set out in a government presentation to investors, were reported only on 24 September. The price and the money were late for the same reason: the sum behind them does not resolve.

Ghana has committed itself to three things at once. Under the Ghana Cocoa Board Act, 2026, signed on 26 August, farmers are guaranteed at least 70% of the gross free-on-board export price. At least half the crop is to be ground inside Ghana. The season is to be financed in cedis, by notes sold to local investors and secured on the proceeds of forward bean sales, replacing the offshore syndicated loan that funded the crop from 1992 until it could not be raised for 2024/25. And the crop all three promises apply to is shrinking: by at least 16% on COCOBOD’s forecast, and by 18% to 38% on the forecast of the man who sells it.

Each commitment is defensible on its own, and several of the Act’s reforms are overdue: ring-fencing legacy debt, putting the Producer Price Review Committee on a statutory footing, ending a dependence on a dollar loan. This is not a case against reform. It is a case about arithmetic. The processing mandate removes beans that would otherwise secure the financing. The shrinking crop reduces the base for both. The 70% floor raises the cash cost of the crop the financing must buy. The three are individually sound and, on a shrinking crop, collectively unfunded. A season that did not open on time is the first visible symptom.

One proportion is worth holding onto. Beans are 5–7% of the retail price of a chocolate bar, and Mondelez alone reported $36.4 billion of revenue in 2024, roughly eight times Ghana’s annual cocoa export earnings. Ghana is legislating about its share of a small slice.

II. Why Ghana has a cocoa board at all

Cocoa arrived in the Gold Coast in 1879. Tetteh Quarshie, a blacksmith, brought seed pods back from Fernando Po and established the first successful farm at Akwapim-Mampong. By 1911 the Gold Coast was the world’s largest producer, built by smallholders rather than plantations: a very large number of very small farmers facing a very small number of very large buyers.

In 1936/37 the asymmetry reached its logical end. Thirteen European firms shipped approximately 98% of Gold Coast cocoa exports. Twelve of them, accounting for about 94%, signed a Buying Agreement that fixed prices and allocated quotas between them. The United Africa Company, Unilever’s trading arm, was by far the most important shipper; Cadbury Brothers, buying with J. S. Fry & Sons, was the largest manufacturer-buyer. The figures come from the Nowell Commission’s report, Cmd. 5845 of 1938, paragraphs 28 to 31.

The farmers’ answer was the cocoa hold-up. From October 1937 to April 1938, farmers, brokers, chiefs, market women and transport workers refused to sell the crop and boycotted European goods. By February 1938 sales were down 90% year on year. Nigeria, under the same agreement, did not follow.

The Commission recommended replacing European brokerage with what it called a “monopsonistic farmer’s co-operative”. That phrase is COCOBOD’s charter. The West African Produce Control Board bought the crop from 1940 to 1946; the Gold Coast Cocoa Marketing Board succeeded it in 1947. Akuafo Hall at Legon — Farmers’ Hall — was built with farmer money channelled through the board, which makes it either a monument or a receipt.

The institution exists, in other words, because a handful of firms once controlled the crop and the state was built to break them. Keep that in mind for section VI.

III. Eighty years of the same lever

Every Ghanaian government since independence has had one instrument it could use without legislating a tax: the farmer’s share of the export price. All of them have used it.

Under Nkrumah, the producer price fell from 224 to 187 new cedi a tonne between 1961 and 1964; in real terms it roughly halved by 1965–66. Cocoa money helped build Akosombo and Tema, and the farmer’s margin is where it came from.

By 1983 the lever had been pulled to destruction. The official exchange rate was ₵2.75 to the dollar against a parallel rate of roughly ₵80 — a twenty-nine-fold gap, and the most efficient tax ever levied on a Ghanaian farmer, because it needed no collector. Recorded output fell from over 550,000 tonnes in 1965 to about 158,000 tonnes in 1983/84, down 71%, and many of the missing beans walked across the border to Côte d’Ivoire and Togo.

The Economic Recovery Programme reversed the lever. The producer share of the FOB price rose from 25% in 1986/87 to 54% in 1997/98, and licensed buying companies were readmitted in 1992, though the Cocoa Marketing Company kept the export monopoly. Production reached a record 1.04 million tonnes in 2010/11, and from 2000/01 to 2014/15 the farmgate price averaged 57% of the ICCO daily price.

One caution belongs in the text: these percentages are not measured against the same thing. The early figures are shares of the world price, the 1980s and 1990s figures shares of FOB, the 2000s average a share of the ICCO daily price, and figures from 2023 shares of gross FOB. They are comparable in direction, not in level.

The direction is the point. The share moves whenever the state needs money. A statutory floor is an attempt to stop that by law. Its first test is this season.

IV. Three promises, one crop

1. The 70% floor

The floor is real, and for the first time it is law. The difficulty is that it is a share of a number COCOBOD does not publish, and that the Act’s promise to adjust the price automatically, with world prices, the exchange rate and “other relevant factors”, comes with no published formula.

Work out what 70% has to clear. At GH¢11.62 to the dollar, paying last season’s GH¢41,392 a tonne required COCOBOD to achieve an export price of $5,091 a tonne. Paying a living income — GH¢45,400 a tonne, Fairtrade International’s April 2026 reference price for Ghana, modelled on a 1.3-hectare farm yielding 870kg for a household of five — requires $5,584.

The 2026/27 price, announced on 25 September, is GH¢2,650 a bag: 2.4% more than February’s GH¢2,587, or GH¢42,400 a tonne. That is less than half the 5.8% rise reported beforehand to be under consideration, and GH¢3,000 a tonne, or 6.6%, short of a living income. It would be exactly 70% of gross FOB if COCOBOD’s realised price for the season were $5,215 a tonne.

This time COCOBOD said more than usual. It stated that the new price is 71.18% of the season’s realised gross FOB value. Two published numbers fix a third: the realised price is about GH¢59,570 a tonne, or roughly $5,130 at the day’s exchange rate. On COCOBOD’s own figures, then, the legal minimum is about GH¢2,606 a bag, and farmers are being paid GH¢44 a bag above it, a margin of 1.7%. COCOBOD did not publish the dollar figure, the exchange rate it used, or how much of the crop it has sold forward and at what price.

Here the next step is easy and wrong. On 25 September New York cocoa futures stood at $5,538 a tonne, 19.8% lower than a year earlier and about $410 above that implied $5,130. The gap is not in itself a sign of anything amiss. COCOBOD does not sell at the day’s futures price. It sells forward, months ahead, so the price it realises reflects contracts struck earlier, when the market was at times far lower: in February it reported achieving $4,200 a tonne. What the gap does show is that the floor now rests on a number only COCOBOD can see. What can be said precisely is this: if COCOBOD’s realised price for 2026/27 is above $5,215 a tonne, GH¢2,650 is below the legal minimum. COCOBOD says it is not. Nobody outside COCOBOD can check.

Two more things follow from the same sum. First, at 70% neither measure of the export price reaches a living income: 70% of the $5,538 futures price, converted, is GH¢2,814 a bag, and 70% of COCOBOD’s implied realised price is about GH¢2,606, against a living-income reference of GH¢2,837.50. Second, the bar moves with the currency, because the floor is a share of a dollar price and farmers are paid in cedis. At GH¢11.34 to the dollar on 2 September, paying February’s price needed $5,214. At GH¢11.62 on 24 September it needed $5,091. Each 10 pesewas on the exchange rate moves the bar by about $45. A stronger cedi makes the floor harder to honour; a weaker one makes it easier. Nobody designed that.

2. Half the crop, ground at home

Ghana’s installed grinding capacity is about 505,000 tonnes a year, of which Barry Callebaut, Cargill and ofi hold 175,000. Actual grindings are around 210,000 tonnes — 41.6% utilisation — and Cargill stopped grinding in Ghana in September 2025. The state-owned Cocoa Processing Company runs at roughly a fifth of its capacity. The 210,000 figure is the US Department of Agriculture’s estimate; the International Cocoa Organization puts 2023/24 at 194,000, so the current figure is probably lower still.

Set the mandate against the Cocoa Marketing Company’s own forecast range:

2026/27 crop	50% mandate	Extra grinding needed over ~210,000t	Utilisation needed	Beans left to export
620,000t (−18%)	310,000t	+100,000t (+48%)	61.4%	310,000t
470,000t (−38%)	235,000t	+25,000t (+12%)	46.5%	235,000t

That table runs the opposite way to intuition. The worse the harvest, the easier the mandate becomes — and the deeper the fall in exports. Last season Ghana had about 550,000 tonnes of beans to sell abroad: a 760,000-tonne crop, less roughly 210,000 tonnes ground at home. At the bottom of the forecast range, with the mandate applied in full, that falls to 235,000 tonnes, down 57%.

A worse harvest would make Ghana’s grinding rule easier to meet, and more than halve the beans left to export

Ghana’s cocoa crop, thousand tonnes. 2025/26 is split at today’s grind; 2026/27 at the minimum the 50% rule would need if applied in full to the Cocoa Marketing Company’s forecast range.

Ground in Ghana
Left to export as beans

Scenario, not a forecast: the Ghana Cocoa Board Act, 2026 phases the 50% minimum in over a transition period. Crop figures: Cocoa Marketing Company managing director, 3 September 2026 (COCOBOD had targeted 650,000t for 2025/26). Today’s grind: USDA estimate, 41.6% of 504,780t capacity; ICCO’s last published figure is 194,000t (2023/24). Figures as at 25 September 2026.

Source: Cocoa Marketing Company Ghana via Reuters; USDA FAS; ICCO; Ghana Cocoa Board Act, 2026

Show the data

Scenario	Ground in Ghana (thousand tonnes)	Left to export as beans (thousand tonnes)
2025/26 crop: 760	210	550
2026/27 high case: 620	310	310
2026/27 low case: 470	235	235

Download the data (CSV)

Two caveats. The Act phases the mandate in, so this is a scenario, not a forecast. And most of the export fall is the harvest, not the rule: in the low case, 290,000 tonnes of the 315,000-tonne decline comes from the smaller crop.

The collision is in what the beans are for. The notes that fund the season are secured on receivables from forward export sales. A tonne reserved for local grinding is a tonne that cannot secure them. Meanwhile local grinders are negotiating, aiming to settle by the end of September, for discounts on main-crop beans and an end to the requirement that they put up 80% collateral in advance. The first is revenue COCOBOD would forgo, the second risk it would take on — both from an institution borrowing against its bean sales to buy the crop. And the cocoa ground would have to sell into a market where European grindings in the first half of 2026 were 7.8% lower than a year earlier.

3. The notes

The syndicated loan raised $28 billion from 1992 and could not be arranged for 2024/25. Its replacement, as set out to investors and reported on 24 September, is a GH¢16.3 billion ($1.4 billion) programme, opening with GH¢4 billion of commercial paper of up to 270 days to buy the crop and a GH¢2.3 billion bond of up to five years to refinance COCOBOD’s existing short-term debt. Repayment is backed by export receivables assigned from selected forward sales contracts. Banks, pension funds, insurers, wealthy individuals and international cocoa buyers may all subscribe.

Set that against what the season costs. At GH¢2,650 a bag, buying the crop alone costs GH¢19.9 billion at 470,000 tonnes and GH¢26.3 billion at 620,000 tonnes. Reuters had put the season’s funding need at about GH¢26 billion with the larger price rise then under discussion, and COCOBOD’s chief executive, Randy Abbey, has said that if the crop needs that much, the board intends to raise about half through the notes. Where the other half comes from has not been published.

In August, with Treasury bill auctions oversubscribed by 79–88%, the market looked like the easy part. By September it looked harder: the auction reported on 20 September missed its GH¢4.1 billion target for the first time in more than two months, investors were reported to be demanding higher returns for cocoa paper, and inflation had risen to 5.0% in August from 4.6% in July.

On cost, the case for the switch is better than it looked a month ago, and that should be said plainly. No 270-day bill is auctioned, but interpolating between the 182-day bill at 6.48% and the 364-day bill at 9.98% gives about 8.2%. Add 300 to 600 basis points for an issuer that restructured its own cocoa bills in 2023, and a COCOBOD note should price at roughly 11–14%. The last dollar loans cost roughly 6–8%. Borrowing in cedis is cheaper only if the cedi weakens by more than the difference — somewhere between 3 and 8 percentage points a year. The Bank of Ghana puts the cedi’s fall over the first nine months of 2026 at 9.5%, from GH¢10.45 to GH¢11.55 to the dollar, which clears even the top of that range. On this year’s currency path, borrowing in cedis would have been the cheaper choice. The depreciation that matters is the one over the life of the notes, which nobody knows — and no coupon has been published.

4. The crop

COCOBOD forecast a fall of at least 16%. On 3 September, Wisdom Kofi Dogbey, managing director of the Cocoa Marketing Company — COCOBOD’s own selling arm — told a conference in Singapore it could be 18% to 38%, from 760,000 tonnes to between 470,000 and 620,000. He cited ageing trees, disease, weather, the alternate-bearing cycle and, in his words, “pollination that is failing for various reasons. We haven’t seen that in 20 years.”

That 760,000-tonne baseline does quiet work of its own. COCOBOD had targeted 650,000 tonnes for 2025/26; reporting citing the USDA said 750,000. COCOBOD’s own marketing chief has now effectively adopted the higher figure. That is the closest thing to a resolution on the record; it is not a reconciliation.

There is a second possible explanation for the gap that nobody has ruled out. On 4 March 2026, Côte d’Ivoire cut its farmgate price by 57%, to 1,200 CFA francs a kilogramme, and it opened the new season at the same price. Ghana’s price is now more than $1,500 a tonne above Côte d’Ivoire’s, against COCOBOD’s own stated threshold of about $400 a tonne for smuggling to take off — and the new price widened the gap, in the year the two governments committed in Abidjan to harmonise their prices. COCOBOD’s Director of Special Services has said: “We were fighting smuggling of Ghana’s cocoa to Ivory Coast; now the reverse is the situation.” Each tonne of Ivorian cocoa that entered Ghana’s buying system at February’s price transferred about GH¢22,192 of Ghanaian public money — GH¢41,392 paid, less the GH¢19,200 those beans are worth across the border (GH¢1,200 a bag, times 16 bags). At the new price the transfer is GH¢23,200 a tonne.

Is part of Ghana’s reported crop Ivorian? That is a testable hypothesis, not a finding. Regional purchase data would settle it, and none has been published.

V. Where the money goes

To see why 70% is hard to pay, look at the last time COCOBOD broke down the export price. Its Turnaround Strategy of June 2023, Table 6, built up a gross FOB of GH¢25,300 a tonne like this: farmer 64.60%; industry costs — roads, CODAPEC, fertiliser, jute sacks, swollen-shoot rehabilitation — 22.11%; licensed buyers’ margin 7.45%; COCOBOD operations 2.38%; crop financing 2.10%; hauliers 1.62%; disinfestation, grading and sealing 1.56%; internal marketing 0.87%.

As published, those shares add up to 102.7%. The discrepancy has never been explained.

The shape is still clear. The non-farmer costs have historically taken 35–38% of gross FOB. In February 2026 COCOBOD chose to pay farmers 90%, leaving about a tenth to fund all of them. On a 750,000-tonne crop, that is a structural shortfall on the order of GH¢9 billion. In February COCOBOD owed the buying companies more than GH¢10 billion. The arrears were not an administrative failure. They were the arithmetic. Farmers went unpaid because the sum did not close, and the sum did not close because of a decision announced as generosity.

COCOBOD has since released GH¢3.62 billion to the buyers in February and March and GH¢2.6 billion in July. The buyers’ chamber now puts what is still owed at about GH¢4 billion. The fall roughly matches the payments, though no ledger has been published, and the remainder is one reason the new season was late.

Take the buyers’ own numbers. A licensed buyer’s margin is about 8.3% of the producer price. The chamber says some members are borrowing at up to 40% while they wait to be paid, and its chief executive, Victus Dzah, says COCOBOD can “sit on” payment for “five months, three months, eight months”. For a buyer that borrowed the full price of its beans, five months at 40% is 16.7% of the price of every bag — twice the margin earned on it. Banks can do that sum too; the chamber says they have stopped lending. On 14 August COCOBOD banned buying from farmers on credit, with licence revocation for repeat offenders. That turned a financing problem into a cash requirement in the season cash became hardest to find. The government is reported to be considering a bridge facility of about GH¢2 billion for the buyers.

VI. The circle closes

In 1936/37, thirteen firms shipped 98% of Gold Coast cocoa, and twelve of them fixed the price between themselves. The board Ghana inherited was built to end that.

Today about 20 of Ghana’s 62 registered licensed buying companies are active, and five buyers backed by international traders — ofi, Ecom, Cargill Kokoo, Barry Callebaut’s Nyonkopa and Touton — account for roughly 60% of internal purchases. The Produce Buying Company, the listed successor to the state’s own buying arm, has gone from GH¢756.6 million of parent revenue to GH¢10.1 million, a fall of 98.7%, and carries GH¢673 million of debt, a GH¢495.4 million court attachment and GH¢24 million owed to farmers for more than 9,000 bags.

The concentration never went away. It reassembled inside the system built to prevent it, and the state’s own former buying arm has effectively stopped competing.

The objection deserves to be stated at full strength. Five competing buyers are not a cartel of twelve. Today’s buyers are price-takers: the farmgate price is set by the state, not negotiated, so they cannot do what the 1937 firms did. They bring working capital Ghana’s banks cannot or will not provide, on terms no local buyer can match, and several have invested in grinding capacity the country needs. The mechanism of 1937 is absent.

The exposure is not. When the state cannot fund the crop, whoever can will set the terms on which it moves — not by fixing a price, but by deciding whether and when to advance money against it. COCOBOD’s chief executive expects the notes to raise about half of what the season needs, and the notes are open to international cocoa buyers. The firms that buy most of the crop may soon be among those lending against it too.

Meanwhile the land is changing hands. In the first half of 2026 Ghana’s gold export earnings rose by $4.2 billion and cocoa’s by $0.1 billion; gold is now 68.7% of exports, cocoa 12.1%, and in July COCOBOD’s Western South manager put cocoa land lost to mining at more than 100,000 acres. The Act answers with up to 20 years’ imprisonment for converting a cocoa farm without approval — a prison term asked to do the work of a price.

VII. What would settle it

None of this needs an inquiry. Every open question here would be closed by an announcement, most of them by a single number.

The realised export price, and the forward book behind it. It sets the farmgate price, the buyers’ margin, COCOBOD’s cash flow and the credit quality of the notes. With the new price COCOBOD disclosed it only as a ratio, the farmer’s 71.18% share, not as a dollar figure, and it has not said how much of the 2026/27 crop it has already sold forward, or at what price. Côte d’Ivoire has said it contracted about 1.1 million tonnes between March and June.
The formula above the floor. The 2026/27 price is now known; how it was reached is not. A number without a formula is this year’s politics, not a pricing regime.
The buyers’ margin. COCOBOD said on 25 September that buyers’ margins, hauliers’ rates and its other fees were unchanged. It did not say what they are.
The coupon and allocation of the notes — and where the other half of the season’s money is coming from.
The forensic audit the government ordered in February 2026, still unpublished seven months later.
Regional purchase data, which would show whether Ivorian beans are inflating Ghana’s reported crop.

In 1937 the answers came only after farmers stopped selling for six months and a Royal Commission was sent to find out why. The 2026 Act contains a quieter remedy — audited accounts laid before Parliament — if it is implemented as written.

In 1937 the farmers refused to sell. This September, for more than three weeks, they had no price to sell at. When the price came, COCOBOD answered the question the Act puts to it — what is it selling the crop for? — with a ratio rather than a number. Until it publishes the number, nobody, whether farmer, buyer, lender or legislator, can check whether the new law is being kept.

Sources

Primary and news (accessed 25–26 September 2026 unless dated)

Background research. The historical price series, grinding capacity, Produce Buying Company accounts, gold export figures and the 2023 cost stack are drawn from the author’s earlier research notes on Ghana’s cocoa sector, which cite their own primary sources. Details are available on request.