Six risks are named repeatedly in the sector’s own literature. Four more are not named, and have closed as many plants.
The named register
ComCashew lists six major risks in RCN processing. The list is short, specific, and drawn from failures actually observed:
| # | Risk | What it looks like on your P&L |
|---|---|---|
| 1 | Supply and cost of high-quality RCN | You buy at import parity, or you buy 60% of your requirement and run the plant at 60% utilisation |
| 2 | Reliability and traceability of the supply chain | A buyer audit you cannot pass; a claim you cannot trace to a lot |
| 3 | Fluctuating market prices and trends | The 2018 crash and the 2024 squeeze, in both directions |
| 4 | Exchange rates and unstable local currencies | Revenue in USD, costs in local currency, and a devaluation between purchase and shipment |
| 5 | Non-existent or unstable government policy | An export levy imposed, removed, or announced and never enforced. A business case built on an unenforced levy is built on air |
| 6 | Labour scarcity and wage increases | Absenteeism is the largest documented operational challenge in this industry |
On the last: Mozambican processors have said they could run their factories with 60% of current staff if absenteeism were eradicated. That is a labour cost roughly 40% above what the work requires, and it does not appear as a line item anywhere in a model project report.
The risks the list omits
Power. Nigerian processing has been documented running below installed capacity because power supply to industry was unreliable. The cost is downtime and quality, not tariff.
Fire. Shell dust, a shell-fired boiler, jute bags, and a warehouse holding a year of raw material in one building.
Quality claims. Documented rejections include organic kernels from Togo stopped in the Netherlands for aflatoxin, and a Vietnamese shipment stopped in Italy. No aggregate rejection statistics by origin exist, which means you cannot benchmark your own claim rate against anything.
Key person. Your production manager and quality manager between them hold most of the operating knowledge that is not written down.
Counterparty default. In the 2018 crash the failure mode was counterparty default and forced renegotiation. Your forward sale is worth exactly what your buyer is worth.
Climate and crop. Mozambican production has swung between roughly 63,000 t and 113,000 t within five years. Outturn moves too, and a decline of 2.0–2.5 lbs across key origins is a hidden cost increase of roughly US$55–68 per tonne on top of the headline price — at about US$27/t per pound of outturn.
Mitigation, and what survives it
| Risk | Primary mitigation | What remains |
|---|---|---|
| RCN supply and cost | Multi-origin sourcing; farmer linkage; tranched buying against sold kernel | Origin-wide crop failure — you cannot buy what did not grow |
| Traceability | Lot coding to the stack; supplier scorecards; annual mock recall | Fraud upstream of your gate |
| Price volatility | Back-to-back matching; price bands set from conversion cost | Everything unmatched |
| FX | Borrow in the currency you sell in; price RCN in USD-linked terms where the market allows | Local-currency cost inflation after devaluation |
| Government policy | Join the processors’ association; model the business without the subsidy | Policy reversal on a political timetable |
| Labour and absenteeism | Transport, canteen, productivity bonus, satellite stations; retention through the shutdown | A regional wage shift |
| Power | Generator sized to hold drying and humidification through an outage — not the whole plant | Extended grid failure during a drying batch |
| Fire | Compartmentation, extinguishers, hot-work permits, shell-dust housekeeping, boiler discipline | Total loss — insure it |
| Quality claim | Retention samples; pre-shipment sample discipline; remediation allocated in the contract | Claims arising after your control ends |
| Key person | Written SOPs by stage; a deputy for each named function | Simultaneous departure at season start |
| Counterparty default | Credit-insured forwards; letter of credit at sight for new buyers; concentration limits | A default inside your deductible |
| Climate and crop | Multi-origin; conservative volume plans | The season itself |
Note the generator row. Sizing a generator for the whole plant is expensive and usually unnecessary; sizing it to carry the batch stages through an outage is the decision that protects product already in process. A drying batch interrupted mid-cycle is a quality problem, not just a delay.
Insurance: stock throughput is the right structure
Stock throughput (STP) covers all-risks physical loss or damage across the whole chain — raw materials, work in progress and finished goods; international and inland transits by any mode; goods in storage, processing or manufacturing; and all locations, including unnamed third-party warehouses. Perils include fire, theft, natural catastrophe, earthquake, windstorm and flood.
Two features matter more than the peril list. Goods are insured at selling price rather than material value, so the policy captures margin rather than only input cost. And it removes the classic gap where a marine cargo policy and a property policy each disclaim a loss depending on exactly where it occurred.
Around it you still need:
- Fire, property and machinery breakdown for buildings, boiler and lines — the boiler usually carrying a statutory inspection regime as well
- Product recall and contamination, whose trigger events are your aflatoxin and Listeria exposures
- Product liability — cashew is a declared major allergen in the EU, so mislabelling and cross-contact are live exposures
- Export credit insurance against buyer non-payment, insolvency and political risk
Marine cargo is subsumed into STP if you buy STP, and bought separately if you do not.
The business-interruption trap
Stock throughput explicitly excludes business interruption, and that exclusion is where seasonal processors get hurt.
Work the example. A fire destroys your warehouse in month two of a twelve-month cycle, three weeks after the buying season closed.
STP pays for the stock, at selling price. So far so good.
But the crop is gone. There is no RCN to buy at any price until the next harvest, ten months away. The plant cannot run, your buyers go elsewhere, your trained peelers leave, and your fixed costs continue throughout.
A business-interruption policy with a 12-month indemnity period measured from the date of loss expires at roughly the moment you become able to buy nuts again — and before you have converted a single tonne of the replacement crop into invoiced kernel. It pays for the wrong ten months.
The fix is to negotiate the indemnity period against the harvest calendar rather than the calendar year: long enough to cover the dead months plus a full buying window plus the conversion and collection cycle that follows it. That is an 18- to 24-month indemnity period in most origins, and it is worth more than the premium saving on a 12-month one.