Ghana’s gold-to-dollar trade and its public cost

Ghana buys local gold in cedis and sells it abroad for dollars, aiming to strengthen its currency and rebuild reserves. GoldBod, the public gold agency, reported $1.315 billion in foreign exchange generated in August 2026 and projected $1.4 billion for September. Those inflows matter. So does the amount Ghana spends to obtain them. ^[1][2]

An IMF study estimates that the central bank’s gold-buying programme generated more than $1.7 billion in losses in 2025, combining charges and valuation effects. GoldBod has meanwhile defended its own positive financial results. These figures concern different accounts. With responsibilities shifting since July, following the purchases, the proceeds and the public funding is essential to understanding the reform. ^[2][3][5]

Paying in cedis for an internationally tradable asset

The Bank of Ghana introduced its Domestic Gold Purchase Programme, or DGPP, in 2021. GoldBod, formally the Ghana Gold Board, was established in 2025. The older programme expanded as Ghana faced pressure on its reserves and exchange rate and lost access to international capital markets in 2022. Buying locally produced gold in cedis offered a way to obtain an asset that could be sold abroad. ^[2]

There were two important routes. Some gold was refined and held as monetary gold, part of the central bank’s reserve assets. Other purchases, particularly artisanal and small-scale miners’ doré (an alloy requiring further refining) were exported for foreign currency. The dollars could remain in reserves or be sold to domestic banks. Gold purchases, export proceeds and the reserves ultimately retained are therefore different quantities. ^[2]

The central bank’s accounts make another useful distinction. Doré acquired to generate foreign exchange is sold as quickly as possible, rather than held for a speculative gain. A rising international gold price does not automatically produce a trading profit: the domestic purchase price can rise too. ^[4]

Someone also has to finance the interval between buying the metal and collecting the export proceeds. A GoldBod agreement with aggregator Bawa-Rock, dated 23 June 2025, illustrates the arrangement. It provided for revolving trading capital of up to 2 billion cedis, with the aggregator advancing funds to local buyers at its own risk. That was a contractual ceiling, not evidence that the full amount had been disbursed. The document provided for a one-year renewable licence; it does not establish the terms applicable today. ^[12]

The potential benefit is tangible. A producer can receive local currency without waiting for the international transaction to settle. But the public system or its partners must fund that gap, verify the gold and sell it on acceptable terms. Export receipts measure the proceeds at one end of this chain. They do not deduct the resources committed at the other.

August’s dollars, split between banks and reserves

GoldBod’s August announcement splits the reported foreign exchange into $668.21 million sold to commercial banks, through spot sales and funded forward arrangements, and $646.59 million made available to the Bank of Ghana for reserve accumulation. The components total $1,314.80 million, consistent with the rounded $1.315 billion headline. ^[1]

It would be wrong to describe the entire amount as an increase in official reserves. Even the allocation to the central bank does not establish the net change in its reserves: other receipts, payments and valuation movements must also be considered. Neither allocation measures a commercial profit.

As of 10 September, $1.4 billion for September remains a target. August’s figures are the agency’s reported results, not an audited monthly account that we have independently reconciled to bank settlements. ^[1]

How a dollar can cost more than its recorded value

The IMF study identifies several sources of losses in the former programme: service and assay fees, discounts allowed to export buyers, and especially the difference between exchange rates. Domestic gold purchases could use the forex-bureau rate, while the central bank recorded the transactions at its reference rate. ^[2]

Consider a wholly hypothetical transaction. A parcel worth $1 million internationally is purchased for 12 million cedis, using an exchange rate of 12 cedis per dollar. It is then sold for the full $1 million. At a reference rate of 10 cedis per dollar, those proceeds enter the accounts at 10 million cedis. There is a 2 million cedi gap even though no dollar is missing from the sale.

What happens next determines how to interpret that gap. If the dollars remain in reserves, the institution holds an external asset whose cedi value depends on the exchange rate used. If it actually sells the dollars for 10 million cedis after spending 12 million to acquire them, it has failed to recover its outlay, before any fees. In this example, the recipient of the dollars obtains better terms than the public buyer paid to source them.

The IMF explicitly notes that some losses reflect valuation effects rather than economic costs. A headline loss cannot therefore be described as an equivalent amount of missing cash. That qualification does not remove the need to examine the transfers between gold sellers, public institutions and recipients of foreign currency. ^[2]

This is also a pricing problem. Attracting gold into the official channel requires terms sellers find competitive. Supplying banks with inexpensive dollars requires attractive terms at the other end. Where the two prices fail to cover the cost of the chain, somebody has to absorb the difference.

Losses in the IMF study and central bank accounts

Guillaume Nolin’s IMF study, published in August 2026, estimates that DGPP losses exceeded $1.7 billion in 2025, or 1.5% of GDP, almost entirely associated with doré purchases under Gold-for-Reserves. That estimate excludes the cost of monetary sterilisation: operations used to withdraw liquidity created in the process. ^[2]

Note 13 of the Bank of Ghana’s 2025 financial statements reports a different measure: 9.053 billion cedis of net losses on gold deals, comprising 8.850 billion under Gold-for-Reserves and 0.203 billion under Gold-for-Oil. The latter was a separate fuel-import programme discontinued in 2025. These figures are rounded here to the nearest million cedis. ^[4]

Crucially, the same note says the Gold-for-Reserves result includes 7.99 billion cedis of realised fair-value gains on monetary gold, reclassified into profit or loss after the metal was sold. This is a net figure incorporating more than doré trading losses. ^[4]

We have not established a complete reconciliation between that accounting line and the IMF estimate. Converting one into the other at an exchange rate chosen retrospectively would not resolve differences in coverage and measurement. Adding the two would be equally misleading.

The documents establish substantial costs. They do not justify pretending that a single, fully reconciled measure of the policy’s public cost is already available.

The grants behind GoldBod’s surplus

On 19 August, GoldBod defended its position as a buying agent for the central bank under the former arrangement. Its chief executive said the agency did not set gold resale prices or negotiate offtake agreements. That is a relevant distinction about responsibility, but it is not an economic assessment of the entire programme. ^[5]

The accounts show the relationship more concretely. GoldBod’s note 15 records 558.1 million cedis in Bank of Ghana service charges, within 970.8 million cedis of non-tax revenue in 2025. A fee can generate income for GoldBod while adding to the cost of the central bank-funded operation. Looking at the public institutions together, that internal transfer does not become new income from outside the public sector. ^[3]

The size of the reported surplus requires a second qualification. GoldBod’s statement of financial performance records 5.444 billion cedis after exceptional items, but its revenue includes a 4.548 billion cedi government grant. The report describes this funding as revolving trading capital for gold purchasing and trading. It is public funding, not a margin earned by selling gold. ^[3]

GoldBod also earned service and other income, whose origin matters when assessing the wider programme. An agency paid to perform a service can report a surplus while the operation it serves incurs a loss. Conversely, a loss on the wider programme does not by itself establish mismanagement at the agency.

There is a further reporting discrepancy: the management commentary and the statement of financial performance in the same annual report show different surplus figures. The commentary reports GHS5,457,413,825 after exceptional items, against GHS5,443,990,347 in the financial statement, a difference of GHS13,423,478. We use the latter figure; the report does not explain the discrepancy. ^[3]

The July transfer changes responsibility for the cost

The institutional change came before the financing model launched on 3 August. The IMF study describes a transfer of DGPP activities to GoldBod from 1 July 2026 under a memorandum of understanding. The central bank is to withdraw from the programme’s financial risks while remaining a fiscal agent and a recipient of monetary gold for its reserves. The arrangement provides for the government to assume the costs transparently. ^[2]

This moves a policy activity out of the central bank’s accounts. The IMF calls the associated risks quasi-fiscal: they resemble the costs of a government intervention but were being carried by the monetary authority.

The transfer could improve budgetary visibility and accountability. It is not, by itself, a saving for the public sector. A genuine reduction requires better purchase and sale terms, lower fees, cheaper finance or smaller losses on comparable operations.

The study does identify progress. Costs relative to gold purchased fell from 14.5% in 2025 to 11.4% in the first quarter of 2026, helped by lower charges, with a target of 5%. The first-quarter observation is not a September reading, and the target is not an achieved result. Those percentages also cannot be directly compared with the 17% cited elsewhere in the paper: that ratio uses gold sold, rather than purchased, as its denominator. ^[2]

A separate IMF publication, country report 26/212, gives ratios of 15.3%, then 11.7% in the first quarter of 2026, relative to gross gold purchases. Both documents indicate falling costs, but their reported levels differ. The passages consulted do not explain that difference. We retain each document’s figures with their attribution, rather than combining them. ^[2][13]

Funding access is now an explicit credit decision

Large international sales do not rule out working-capital problems earlier in the chain. On 24 August, Reuters reported delays of up to three weeks for some gold buyers, citing anonymous industry sources. GoldBod subsequently rejected claims of a funding crisis, distinguishing its advances to aggregators from the credit those aggregators provide to downstream buyers. These accounts do not establish how widespread the delays were. ^[7][8]

An official directive dated 22 July sets out the financing rules. Eligible buyers seeking GoldBod funds through an aggregator must undergo identity and credit checks, execute an agreement and provide an acceptable form of security. That security must cover between 10% and 50% of the approved financing, depending on the credit assessment. Existing participants were required to settle their old financing accounts and regularise their participation by 1 August. ^[6]

Security may take the form of a bank guarantee or insurance bond; it need not be an equivalent cash deposit. A licence to buy gold therefore does not automatically entitle its holder to a public advance. ^[6]

Tighter safeguards can protect public funds while making finance harder to obtain for particular traders. Distinguishing that effect from a general liquidity shortage requires information on applications, refusals, actual disbursement delays and advances outstanding. The dollar proceeds reported at the end of the chain cannot answer those questions.

More reserves also mean more resources committed

The industrial mining channel is changing too. An agreement announced on 25 June provides for the purchase of 30% of large-scale mines’ gold output from 1 July, in cedis at the central bank’s reference exchange rate and at a 0.55% discount. After refining and certification, the gold is intended for reserves. These are agreed terms, not evidence of the quantities subsequently delivered, and they should not be confused with artisanal purchases. ^[9]

The government’s policy targets reserves covering 15 months of imports by 2028. The IMF study estimates adequate coverage at six months. That is not a universal ceiling, but the difference raises an allocation question: how much extra protection does each additional reserve asset provide, compared with other possible uses of the resources? ^[2][11]

Financing matters alongside reserve size. Where central bank purchases increase cedi liquidity, the bank may withdraw part of it by issuing interest-bearing instruments. Those sterilisation operations carry a cost. Selling dollars back to banks against cedis can also withdraw liquidity. It would therefore be wrong to attribute all monetary-policy expenses to gold purchases, even though reserve accumulation can require additional spending on liquidity management. ^[2][4]

Processing the gold locally brings its own bill. A directive dated 24 August, effective 1 September, requires self-financing aggregators and their approved offtakers to have the relevant gold refined in Ghana before export. It assigns refining charges to the aggregator or offtaker according to their commercial agreement, with payment before export. The requirement seeks to retain activity locally. It does not make that processing free, and the directive is not evidence of the results already achieved. ^[10]

The cost of greater monetary security

The strongest defence of the programme is macroeconomic. Access to foreign currency and a stronger reserve buffer can be worth more to a country than a trader’s margin. The IMF study considers the programme’s expansion a plausible contributor to stabilisation. The Bank of Ghana reports that the cedi appreciated by 40.7% against the dollar in 2025, while also identifying fiscal and monetary policy as important to the recovery in stability. Gold alone cannot be credited with the outcome. ^[2][4]

A central bank should not be judged as though its only task were to maximise annual profit. But policy objectives do not remove the obligation to measure the cost of pursuing them. A national assessment must also consider extraction’s environmental effects. The IMF study highlights pressure on farmland and water resources; environmental traceability needs to be assessed throughout the chain, including purchases through the official channel. ^[2]

The documents establish sizeable foreign exchange flows, losses in the former programme and a change in institutional responsibilities. They do not yet provide a consolidated account of the new system that reconciles purchases, proceeds, inventories, fees and public support. That is what would distinguish a genuine improvement from a shift in where the cost appears.

Ghana may reasonably choose to pay for more secure access to dollars. It still needs to show the price, who benefits and which risks remain public. Neither export receipts nor one institution’s surplus provides that answer on its own.

Further reading: Ghana’s cocoa financing and cash-flow crisis examines the relationship between domestic purchase prices and funding. Our guide to the gold market explains the metal’s uses and price formation.

Sources and documents

1. Ghana Gold Board · 31 August 2026; web account on 1 September. FX Generation and Sales Update : September 2026. Original one-page release. Reported August figures and September target, not an independently audited monthly account. The two allocations sum to $1,314.80m, rounded to $1.315bn. Web announcement.
2. International Monetary Fund · Guillaume Nolin · 7 August 2026; completed 10 July. Lessons from the Bank of Ghana’s Domestic Gold Purchase Programme : Selected Issues Paper 2026/084. Authored staff study, not an audit of the new August model. Printed pp. 5–7: channels, context and flows; p. 8, paragraphs 13–14: losses, valuation, transfer and ratios; p. 9, paragraphs 15–16: reserves and traceability. The 17% figure uses doré sold; 14.5% and 11.4% use gold purchased. The 5% figure is a target. The same study appears in country report 26/213, not an independent confirmation. Publication record.
3. Ghana Gold Board · year ended 31 December 2025; publication announced 29 April 2026. Annual Report and Audited Financial Statements 2025. Printed p. 16 (PDF page 19): financial performance; p. 39 (PDF 42), notes 15–17: BoG charges, grant and finance income. P. 5 (PDF 8): management commentary, whose surplus differs from the financial statement by GHS13,423,478. Figures used here come from the financial statement. A GHS1 difference between the component sum and printed revenue total does not affect the rounded figures and is noted here. Publication page.
4. Bank of Ghana · financial year 2025; accounts approved 29 April 2026. Annual Report and Financial Statements 2025 : note 13 and foreword. Note 13, financial-statement p. 63 (PDF page 125): figures in thousands of GHS, converted here to billions. The net loss includes realised monetary-gold gains. Foreword pp. 1–2 (PDF 9–10): 2025 macro outcomes and monetary-policy costs. No complete reconciliation to the IMF estimate established.
5. Ghana Gold Board · 19 August 2026. GoldBod CEO Explains Why BoG’s US$1.7bn DGPP Loss Is Not a GoldBod Loss. Management’s position, an interested source. Supports its account of acting as buying agent rather than running the BoG programme. Effectiveness and responsibility claims are not independent findings; financial figures are checked separately in source 3.
6. Ghana Gold Board · Compliance Directorate · 22 July 2026; regularisation required by 1 August. Mandatory Procedures for Accessing Gold Purchase Financing through Aggregators. Original four-page directive. Pp. 1–2: KYC, credit checks, agreements and security covering 10–50% of financing; p. 2: existing accounts; p. 3: no automatic entitlement to funding. Coverage is neither an expected loss nor necessarily an equivalent cash deposit.
7. Reuters · Maxwell Akalaare Adombila and Emmanuel Bruce · 24 August 2026. Ghana’s GoldBod buyers hit by funding delays, sources say. Full Reuters dispatch read in its attributed Kitco republication. Delays reported by industry sources, including five anonymous sources; not a comprehensive measure of applications or disbursements. GoldBod’s subsequent response is included through source 8.
8. MyJoyOnline / Multimedia Group · 24 August 2026, 9:31pm as shown on the page. GoldBod rejects funding crisis claims, says gold purchases remain on course. Account of GoldBod’s response, not an audit of liquidity. Distinguishes direct aggregator funding from downstream buyers’ credit. The underlying regulatory requirements are cross-checked against the official directive in source 6.
9. Ghana Gold Board · 25 June 2026; announced effective date 1 July. 30% Gold Offtake Deal with Large-Scale Mines. Announced terms: 30% of large-scale output, 0.55% discount, payment in GHS at the BoG rate, refining and reserve destination. The IMF study also describes these terms in paragraph 15. No actual deliveries or final refining costs are inferred.
10. Ghana Gold Board · Compliance Directorate · directive dated 24 August 2026, listed on 26 August; effective 1 September. Mandatory Local Refining of Gold Doré Prior to Export. Original two-page directive. Covers Self-Financing Aggregators and their approved offtakers; section 3 assigns refining costs under their commercial arrangements, payable before export. A rule does not establish the results of its implementation.
11. Ministry of Finance, Ghana · 27 February 2026. Parliament Passes Ghana’s First Accelerated National Reserve Accumulation Policy. Government account of GANRAP and its target of 15 months of import cover in 2028. A policy target, not an achieved result. The six-month comparator is the IMF assessment in source 2, not a universal rule.
12. Ghana Gold Board / Bawa-Rock Company Limited · 23 June 2025. Aggregator Agreement : clauses 2, 4 and 5.1.3. Scanned 24-page agreement. PDF pages 2–6 visually read: date, one-year renewable licence, GHS2bn advance ceiling and downstream financing at the aggregator’s risk. Historical illustration, not evidence of full disbursement or renewal. Targeted reading of cited clauses, not a complete legal review.
13. International Monetary Fund · 4 August 2026; report completed 10 July. Ghana, Article IV consultation and sixth review, country report 26/212. Printed p. 37 (PDF 42), paragraph 83: loss ratios of 15.3%, then 11.7% in Q1 2026, relative to gross gold purchases. These levels differ from the authored study in source 2; the passages do not supply a reconciliation. Paragraphs 81 and 84 address the risk transfer and financial-reporting framework.