This is the money pillar. It answers three questions: does a cashew orchard pay, does processing pay more, and how is either financed. The published Ghana cashew budgets are old: the most recent full farm budget, from a 2009-10 survey of 140 farmers in Wenchi, found an internal rate of return of about 44 percent on a one-hectare plantation at 2010 prices,1 so the orchard figures here are rebuilt from those benchmarks at 2025-2026 prices and should be re-priced with a grower’s own costs before any decision. What is solid is the shape of the opportunity: the value added by processing, the cost gap that holds it back, and the finance now lining up behind it.
HOW TO READ THE NUMBERS IN THIS PILLAR: Prices are 2025/26 and 2026 figures. Where a cost comes from an older study, such as the 2009-10 Wenchi farm budget or the 2020 processing-cost benchmark, its year is stated next to it.
Source: TCDA (2025); African Cashew Alliance (2026).
At 2026 prices a kilogram of W320 kernel from an African processor sells for about US$7.50 against about US$1.07 for a kilogram of raw nuts at the farm-gate floor. The 4.8 kilograms of nuts needed for that kilogram of kernel cost about US$5.10, before any processing cost.
The central financial fact of Ghanaian cashew is the gap between what the country sells and what the market pays. In April to June 2026 Ghanaian raw nuts sold at about US$1,100 to 1,280 a tonne at the farm gate and US$1,560 to 1,580 delivered to India or Vietnam, while W320 kernels from African processors sold at about US$7,400 to 7,800 a tonne free-on-board.2 A kilogram of kernel therefore sells for about six to seven times a kilogram of raw nuts at the farm gate, but because it takes about 4.8 kilograms of raw nuts to make one kilogram of kernel, a tonne of nuts turned into kernels is worth only about 1.2 to 1.5 times its farm-gate value, before processing costs.34 Add roasting, branding and retail and the gap widens further.
This is why the whole strategic push is towards processing: the sector could earn over US$660 million a year with full value addition, an estimate first put forward by Cashew Watch Ghana and since repeated by the Tree Crops Development Authority,56 against about US$298 million earned in 2025;7 the gap of roughly US$360 million is extra export revenue at stake, before the cost of processing.
What this means. The premium shows where the business is, and its limits. A raw-nut trader earns a thin margin on a low-value commodity; a processor sells a kernel worth about seven times the raw nut a kilogram, but must buy about 4.8 kilograms of nuts for each kilogram, pay to process them, and sell a good share of whole kernels to come out ahead. The by-products can pay again on top. The strategic question for the sector is how to keep more nuts at home and process them at a profit.
A cashew orchard is a long-term asset with a slow start. It earns nothing for the first two years, gives first nuts in years three to four, and reaches full bearing around year seven, after which it can produce for decades. The immature-year cash-flow gap (covered in Pillar 4) is bridged with intercrops. Two complete Ghana returns are on record, both historical: the Cashew Development Project reported a financial internal rate of return of about 35 percent for production and processing together,8 and a 2009-10 survey of Wenchi farmers put the return on a one-hectare plantation at about 44 percent.1 Neither can be read across to today’s prices.
The table below prices a mature hectare at 2026 prices. Labour quantities come from the 2009-10 Wenchi budget,1 valued at the 2026 national daily minimum wage of GH¢21.77.9
Yield decides the return| Line (per hectare, mature) | Value (2026) | Basis |
|---|---|---|
| Yield | 550 kg/ha (good practice); 200 kg/ha (old seedling orchard) | achieved with improved material and good practice; 200 kg/ha was the yield in project areas before the Cashew Development Project of the 2000s; most farms now average about 400 to 500 kg/ha10 |
| Farm-gate price | GH¢12 / kg | TCDA 2025/26 minimum; GH¢12 to 14 paid in April and May 2026 |
| Gross revenue | ~GH¢6,600 / ha (GH¢2,400 on an old orchard) | yield x floor price |
| Labour cost | ~GH¢3,810 / ha | 175 person-days a year (weeding, fire belts, pruning, spraying, harvest, drying) at GH¢21.77 a day |
| Margin before inputs | ~GH¢2,790 / ha | revenue minus labour; fertiliser, sprays and sacks come out of this |
Sources: Wongnaa and Awunyo-Vitor (2013); Tree Crops Development Authority (2025); Ankrah (2025); Ministry of Food and Agriculture (2012).
Yield, more than price, decides whether an orchard pays. At the GH¢12 floor a well-managed hectare covers hired labour at the legal minimum wage and leaves about GH¢2,790 for inputs and profit, while an old seedling orchard at 200 kilograms a hectare earns about GH¢2,400, less than the labour it would take to hire. That is why most smallholders run cashew on family labour. Day rates in the cashew belt are usually above the legal minimum, so price a plan with a quote from your own district. The accompanying cost model sets out the 2009-10 budget in its own year and a separate 2026 price test.
Sell farm-economics data. A farm-economics data service that collects real per-acre costs and yields owns information every farmer, aggregator and lender needs (Pillar 7).
Use idle capacity first. About two-thirds of Ghana's installed processing capacity sat unused in 2025 for want of feedstock and finance, so an entrant who can secure nuts and working capital can put existing plant to work before building new.
Costs and prices also move with the wider economy, in both directions. Inflation eased from 23.5 percent in January 2025 to 3.2 percent in March 2026, the lowest reading since the consumer price index was rebased in 2021, then edged back up to 5.0 percent by August 2026, with food inflation at 3.0 percent.111213 Much of what a cashew business buys is imported, including fertiliser, fungicides and insecticides, shelling and grading machinery, spare parts, packaging and fuel, so its cedi cost follows the exchange rate. The cedi lost about 9.5 percent against the US dollar in the first nine months of 2026,14 and the World Bank expects fertiliser prices to rise about 31 percent in 2026, with urea about 60 percent above 2025.15
The exchange rate cuts both ways in cashew, because raw nuts and kernels are priced in US dollars. A weaker cedi raises the cedi cost of imported inputs and machinery, but it also raises the cedi value of export sales, and the TCDA converts its dollar reference price into the farm-gate floor at a six-month average exchange rate, so the floor follows the cedi with a lag. Borrowing has become cheaper: the Bank of Ghana held its policy rate at 14 percent in September 2026, and the average bank lending rate fell to about 15.9 percent in August 2026.16 A cost build that holds one season can be wrong the next, and a fall in the kernel price can arrive in the same year as a rise in input costs. Re-cost inputs every season, keep a buffer against both a price fall and a cost rise, and prefer supply and sales contracts that fix the price or tie it to costs.
Source: Wittern et al. (2023), 2020 data.
In 2020 a model semi-mechanised plant in Ghana, Côte d'Ivoire or Benin spent about US$350 a tonne of raw nuts on processing against about US$210 in Vietnam, and the raw nuts themselves were about four-fifths of total cost. Cheaper finance, larger scale and a high whole-kernel share are what close that gap.
Processing sells the kernel at about seven times the raw-nut price per kilogram, but the margin after buying nuts and processing is small or negative: an older industry estimate put African gross margins at 30 to 40 percent,17 whereas a model of Ghanaian processors found long-run economic losses in every year from 2015 to 2020.4 Ghana processes only a fraction of its crop, and the reasons are well documented.
First, cost: in 2020 a model semi-mechanised plant spent about US$350 a tonne of raw nuts on processing in Ghana, Côte d’Ivoire and Benin against about US$210 in Vietnam,4 and in 2026 building a 20,000-tonne plant cost about US$9.2 million in Ghana against US$5.3 million in Vietnam.18 The raw nuts themselves are about four-fifths of a processor’s total cost. Second, finance: Ghanaian processors borrowed at 30 to 35 percent in 2018, against 7 to 8 percent in India and Vietnam and 1 to 2 percent for large international buyers;19 by 2026 the gap had narrowed but remained wide, at about 20 percent in Ghana against about 9 percent in Vietnam,18 and financial costs in Ghana run about 83 percent higher than in Vietnam.4 Third, working capital: a processor must buy a whole year’s nuts in the three-to-four-month harvest window. Fourth, scale: in West African benchmarks, plants above 10,000 tonnes processed at about US$2.55 a kilogram against US$3.55 for plants under 5,000 tonnes,20 and a 2026 Ghana study uses a 20,000-tonne plant as its competitive reference.18
Market signal: the four barriers are the diligence questions. The four barriers are also the diligence questions. Before backing any processor, ask what it costs to process a tonne against the Vietnam benchmark, what rate it borrows at and whether the loan is guaranteed, how it funds the harvest-window buy, and whether it runs at a scale that can compete. A plant with good answers to those four is a business; one with a machine but no answer to finance and feedstock is the stranded capacity described below.
The result is idle capacity. Ghana’s twelve registered processors had about 46,000 tonnes of installed capacity, of which only 10,000 to 12,000 tonnes was used in 2019 to 2022;21 in 2025 about 15,000 tonnes of raw nuts were processed, roughly a third of that capacity, and only one major industrial plant was operating.18
Even Ghana’s largest processor, Usibras, secured only about 7,000 tonnes of raw nuts for its 35,000-tonne plant in 2025 and threatened to move to Côte d’Ivoire,22 although it was still operating in Ghana in mid-2026.23 The machines exist; the feedstock and finance do not reach them. That is the problem the announced raw-export ban and the new finance facilities are meant to fix.
Source: Wittern et al. (2023); ComCashew (2024); Nitidae (2026).
Twelve registered processors had about 46,000 tonnes of installed capacity (2022), of which about 12,000 tonnes was used; in 2025 about 15,000 tonnes of raw nuts were processed. The machines exist; the nuts and the money do not reach them.
A cashew nut is more than its kernel. The shell, about 70 percent of the nut’s weight, contains cashew nut shell liquid (CNSL): the liquid makes up about 30 to 35 percent of the shell’s weight,24 used in brake linings, resins, paints and bio-based chemicals. By-product sales can cut a processor’s costs by up to 20 percent, and the shell itself can fuel the roasting.1724 Ghana already has a commercial CNSL extractor in the Bono region.
The cashew apple, the fleshy fruit the nut hangs from, is the larger waste: under 10 percent of cashew apples are used,25 even though it makes juice, gin, jam and animal feed. Mim Cashew, which stopped kernel processing in 2020, makes cashew-apple spirits,25 and recent government initiatives are promoting cashew-apple beverages in the Bono and Bono East regions.
By-products turn a cost into two revenue lines. CNSL and cardanol extraction (following the Bono-region model) both cuts processing cost and adds a chemical-industry revenue stream, though it only pays at volume. Cashew-apple juice, gin and dried-fruit products use a near-zero-cost feedstock that is more than 90 percent wasted. The barriers are perishability (Brazil handles it with cold storage at about minus 17 degrees) and skills, while the raw material is plentiful. For a processor, integrating shell and apple use at the plant-planning stage can narrow a loss or widen a thin margin.
Source: Cashew Cost and Returns Model v2 (2026 prices).
Swinging each driver across its sourced range, the kernel price (GH¢61 to 83 a kilogram) and the processing cost (US$210 to 570 a tonne) move the margin most, then the raw-nut price (GH¢12 to 14) and the factory kernel yield (20 to 22 percent). The base case sits at about minus GH¢1,430 a tonne of raw nuts.
Before asking how to cut cost, an operator should know which numbers actually move the result. The chart below takes the processing margin at 2026 prices and swings each driver across a sourced range, holding the others at their base. The kernel sale price moves the margin most, followed by the processing cost (from Vietnam's level to Ghana's own 2020 cost at low capacity use), then the raw-nut buy price and the factory kernel yield. For a processor, this means grade and price discipline (selling a high share of whole kernels) comes first; running cost and the buy price then decide whether any margin is left.
The base case uses 2026 prices: raw nuts bought at GH¢13 a kilogram (the middle of the GH¢12 to 14 paid in April and May 2026), a factory kernel yield of 21 percent from a 46 to 47 pound outturn, a blended kernel price of about GH¢72 a kilogram (June 2026 grade prices in an assumed mix of 65 percent wholes, 15 percent splits and 20 percent pieces), a shell credit of GH¢392 and the 2020 processing-cost benchmark of GH¢3,920 a tonne. On those numbers a processor loses about GH¢1,400 on every tonne of raw nuts, and the most it could pay for raw nuts and still break even is about GH¢11.60 a kilogram, below the floor. The margin turns positive only when the plant buys at the GH¢12 floor and sells a larger share of whole kernels: at the floor, with 48-pound nuts and every kernel priced as W320 (about GH¢84 a kilogram), it earns about GH¢2,600 a tonne. That matches the African Cashew Alliance’s reading that raw nuts bought in April 2026 could not be turned into forward kernel sales at a profit, and that Ghanaian processors faced a difficult 2026 season.226
What this means. Because kernel price leads, the plant that wins invests in correct steaming, humidification and trained cutting to lift the whole-kernel share, and in grade discipline to sell into the premium W-grades. Processing cost comes next, so scale and cheap energy matter, and the buy price then decides whether any margin is left.
Move the levers yourself. The defaults are 2026 prices (April to June); enter your own quotes for the kernel price, the kernel yield, the raw-nut buy price, the processing cost and the finance rate, then swing them one at a time. Kernel price moves the margin most, then processing cost, the raw-nut buy price and kernel yield. At 2026 prices the base case loses about GH¢1,400 a tonne of raw nuts; it turns positive only with a floor-price buy and a high share of whole kernels.
Slide the drivers that decide a tonne of raw nuts turned into kernel: the price the kernel sells at, how much kernel comes out of the nut, what the raw nuts cost, what it costs to process, and the finance rate on the harvest-window buy.
Defaults: 2026 prices (April to June); enter your own quotes. Kernels at US$6,430 a tonne, the blended price of June 2026 grade prices in an assumed mix of 65 percent wholes, 15 percent splits and 20 percent pieces (the W320 upper case is US$7,500); a factory kernel yield of 21 percent at a 46 to 47 pound outturn; raw nuts at US$1,160 a tonne (GH¢13 a kilogram at GH¢11.20, the middle of the GH¢12 to 14 paid in April and May 2026); a processing cost of US$350 a tonne of raw nuts (2020 benchmark, which already includes finance at about 20 percent); and a shell credit of GH¢392 a tonne. Money is converted at about GH¢11.20 to the US dollar, the Bank of Ghana rate in mid-June 2026.
The same model, swung across the sourced ranges behind Figure 5.4. Each bar is how far the gross margin moves from the low end of one driver's range to the high end, holding the others where you left them.
The blended price of the kernels the plant sells. Selling a high share of whole kernels and holding grade discipline is the strongest single lever on the margin. Margin from -US$327 to +US$78 a tonne.
From Vietnam's level of about US$210 a tonne of raw nuts to Ghana's own 2020 cost at low capacity use, about US$570 (Wittern et al., 2023). Scale, capacity use and cheap energy decide where a plant sits. Margin from -US$345 to +US$15 a tonne.
From the TCDA floor to the top of the April to May 2026 market. Raw nuts are about four-fifths of a processor's total cost, so buying early at the floor matters. Margin from -US$215 to -US$36 a tonne.
About a 44 to 49 pound outturn. Correct steaming, humidification and trained cutting protect the yield and the whole-kernel share. Margin from -US$189 to -US$60 a tonne.
From a raw shell sale to shell-oil extraction worth about 8 to 10 percent of the raw-nut cost (Nitidae, 2023). By-products narrow a loss but rarely close it alone. Margin from -US$125 to -US$44 a tonne.
Ranges follow the Sensitivity sheet of the Cashew Cost and Returns Model v2: kernel price plus or minus 15 percent; processing cost US$210 (Vietnam) to US$570 (Ghana's 2020 cost at low capacity use, Wittern et al., 2023); raw nuts GH¢12 to 14 a kilogram (African Cashew Alliance, 2026); factory kernel yield 20 to 22 percent; shell and CNSL credit GH¢392 to 1,300 a tonne (Nitidae, 2023).
The section above explains why Ghanaian processing is expensive; this one asks the operator’s question: which of those costs can actually be cut, with what lever, and by roughly how much. The savings below are estimates from published benchmarks, including a 2026 cost study of Ghanaian processing,18 and each comes with a way to check it against your own numbers. They are ordered by where the money sits: raw material and finance first, because they dominate the cost.
The two that move the result most are finance (a guarantee that cuts the borrowing rate on the harvest buy) and whole-kernel yield (more wholes lift the blended kernel price, the biggest driver in Figure 5.4), because raw nuts, bought with borrowed money, are about four-fifths of a processor's cost. Check each against your own quotes in the accompanying cost model before committing capital.
| Cost line | Base (2026) | The lever | Estimated saving | How to check it |
|---|---|---|---|---|
| Working-capital finance | about 16 to 20% interest on the harvest-window buy in 2026, with some banks charging up to 30% (raw material is about 80% of cost) | stack a GIRSAL guarantee (up to 70% of the loan) and time the buy against warehouse receipts | rate cut towards the Ghana Reference Rate (10.61% in August 2026)27 plus a smaller risk premium | get a GIRSAL-backed loan quote against an unguaranteed one |
| Process energy | grid or diesel heat for steaming, drying and roasting (grid tariffs rose 3.49 percent for all customer classes from July 2026, and high power costs were a main complaint of the largest processor in 2025)2822 | burn the cashew shells and use CNSL as process heat instead of buying fuel | up to ~20% of processing cost (with by-products) | compare a shell-fuel trial against the current energy bill |
| Whole-kernel yield | breakage downgrades kernels well below whole W-grades | correct steaming and humidification plus trained cutting to lift the whole-kernel share | about GH¢400 to 700 more per tonne of raw nuts for each 10 points more wholes (June 2026 grade prices) | run a cut-test breakage rate before and after |
| Labour and siting | manual shelling and grading labour at a single plant | satellite village stations near the Bono and Ahafo supply (the model Mim Cashew used until 2020) | lower unit labour and inbound freight | a time-and-motion check at a pilot station |
| Raw-nut purchase | the GH¢12/kg 2025/26 floor plus the aggregator’s margin, in a market where farm-gate prices swung from about GH¢20/kg in February 2025 to GH¢7 by May18 | buy direct through a cooperative and the commodity exchange on an outturn-based grade | the aggregation-layer margin | track a direct price against an agent price over a season |
| Farm post-harvest loss | 3 to 4 day drying, and no store for about 76% of farmers surveyed in 2009-10 | dry on raised mats to 10% moisture or below and store in clean jute on pallets | higher outturn and price, fewer rejections | measure moisture and kernel outturn before and after |
Sources: Nitidae (2026); Nettey (2026); Arthur (2026); Citi Newsroom (2025); Gyedu-Akoto et al. (2014); June 2026 grade prices.
The finance gap that has starved processing is being addressed by a stack of instruments a serious entrant should learn and combine.
The most important move for a would-be processor is to stack these: use a GIRSAL guarantee (covering up to 70 percent of the loan principal) to make a bank comfortable, add a Free Zone or agro-processing tax concession to protect the margin, then apply for the Tree Crop Diversification Project’s matching grants or a bank line under GCB Bank’s agribusiness framework for scale. Because Ghanaian processors still borrow at about 20 percent against about 9 percent in Vietnam,18 de-risking the loan is often what decides whether a plant survives.
Buy a year of nuts in one window| Instrument | What it offers | Best used for |
|---|---|---|
| GIRSAL credit guarantee | covers up to 70% of a loan's principal; favourable risk-weight; accepted as collateral | de-risking harvest-window working capital |
| World Bank tree-crop facility | Ghana Tree Crop Diversification Project: US$200m IDA credit approved in 2023, including matching grants for about 185 processing SMEs2930 | production, processing and trade scale-up |
| GCB Bank framework | a framework launched in 2025 with an initial commitment of up to GH¢2.5bn31 | processor and aggregator lending |
| Free Zones | 10-yr corporate-tax holiday + 0% equipment duty if >70% exported | export-focused processing plants |
| GIPA agro-processing | 5-year agro-processing concession under the Income Tax Act, then a rate that depends on location (Act 896); GIPA duty exemptions on plant and machinery | domestic-and-export processors |
Sources: GIRSAL; World Bank (2023); Business and Financial Times (2025); GCB Bank (2025); Free Zones Act (1995); PwC (2026).
What this means. The economics point to a policy choice. A peer-reviewed model of Ghanaian cashew found that a processor subsidy is the most appropriate way to grow processing, while a 12.5 percent raw-export tax, simulated on 2020 prices, hurts farmers most (an average loss of about US$93 a year per smallholder).4 For an investor this means the better bet is to secure feedstock and working capital first and use existing idle capacity where possible, while the finance stack and incentives are being rolled out, and to press for a processor subsidy and guaranteed off-take alongside any export restriction.
MARKET SIGNAL: AI AND DIGITAL OPENING. The money pillar is where digital tools close the two gaps that sink cashew ventures: records and finance. Feasible now: a simple farm-and-plant record app to build the current cost data the sector lacks, mobile money (MTN MoMo) for transparent farmer payments and harvest-window float, and digital credit scoring that lets a GIRSAL-backed lender price a loan on real cash flow rather than collateral alone. Ghanaian platforms such as AgroCenta already bundle payment and credit for smallholders. Commercial-only for now: automated grading lines with camera-read outturn. The opening is a farm-economics data service that both collects the field figures and feeds them into finance (Pillar 7).
Paying more for raw nuts than the kernels made from them will sell for, once processing cost is added.
In April 2026 raw nuts bought at the farm gate could not be turned into forward kernel sales at a profit,2 and at GH¢13 a kilogram and a blended kernel price the margin is about minus GH¢1,400 a tonne.
Processors and aggregators who buy at the harvest peak without a forward kernel price or a grade record.
Work the buy price back from the forward kernel price and your own grade mix, buy early at the floor, and sell part of the output forward; the opening is a price-and-grade data service for processors.
A processor must buy a whole year of nuts in a three to four month window, at about 20 percent interest in 2026, then hold overheads until kernels sell.
Raw material is about four-fifths of processing cost, and Ghanaian processors borrowed at about 20 percent in 2026 against about 9 percent in Vietnam.418
Under-capitalised processors, who then run below capacity or fold.
Use a GIRSAL guarantee and the World Bank project's matching grants to secure harvest-window capital, with a Free Zone or agro-processing tax concession to protect the margin; the opening is structured trade finance.
At the GH¢12 floor and 2026 wages, a well-managed hectare at 550 kilograms grosses about GH¢6,600 and leaves about GH¢2,790 after hired labour for inputs and profit; an old orchard at 200 kilograms grosses about GH¢2,400, less than the labour it would take to hire.
A kilogram of W320 kernel sells for about seven times a kilogram of raw nuts, but it takes about 4.8 kilograms of nuts to make, so a tonne of nuts turned into kernels is worth only about 1.2 to 1.5 times its farm-gate value before processing costs.
At 2026 prices the base-case processor loses about GH¢1,400 a tonne of raw nuts; the break-even raw-nut price is about GH¢11.60 a kilogram, below the floor, and the margin turns positive only with a floor-price buy and a high share of whole kernels.
Ghana processed about 15,000 tonnes of raw nuts in 2025, roughly a third of installed capacity, held back by processing cost, finance at about 20 percent against about 9 percent in Vietnam, the harvest-window working-capital need and small scale.
An orchard paid well on 2009-10 prices; at the 2025/26 floor of GH¢12 a kilogram it covers hired labour at about 320 kilograms a hectare and leaves real room for inputs and profit only as yields approach the 550 kilograms a hectare of good practice. Processing sells the kernel at about seven times the raw price per kilogram, and the by-products pay again, but only if a plant has scale, affordable finance and year-round feedstock. The finance stack (GIRSAL guarantees, the World Bank tree-crop project, GCB, and Free Zone or agro-processing tax concessions) can narrow the gap. The first move is to secure feedstock and de-risk the working capital before adding capacity.
Where this connects. The processing premium here is the value the raw-export trade of Pillar 2 gives away, and the orchard economics rest on the slow-bearing tree and the intercrops of Pillar 4. The finance stack that de-risks the harvest-window buy returns in the entry-point map of Pillar 7, and the by-products, grading and branding that lift the margin are where a processor reaches up the value chain of Pillar 6. The thread through all of them: secure feedstock and de-risk the working capital before adding capacity.