How many tonnes of cocoa beans will the Ivory Coast produce in 2026?

TL;DR

I forecast Côte d’Ivoire will produce 1.86 million metric tonnes of cocoa beans in calendar year 2026. My central 80% interval is 1.65–2.05 million tonnes, with a 17% chance of exceeding 2.0 million. The strong 2025/26 crop is largely offset by the delayed and smaller crop contributing to late 2026.

Context

The hard part is timing. ICCO normally reports production for cocoa years running from October 1 through September 30, while FAOSTAT defines this indicator as the amount produced during the stated year (ICCO statistics methodology; FAOSTAT metadata via Our World in Data). The resolution rules favor a genuine January–December figure if one is available.

Most of the crop ending in 2026 is now observable. Port arrivals reached 1.999 million tonnes by August 23, 2026, but roughly 1.1 million tonnes of that flow belonged to October–December 2025 (CocoaIntel, August 24, 2026; Reuters, August 26, 2026). The remaining question is how much of the delayed new crop will be harvested and recorded before December 31.

Evidence

The historical backbone is FAOSTAT's annual cocoa-bean production series. It contains 64 observations, covers 1961–2024, uses metric tonnes, and was last updated on February 25, 2026. FAO describes the measure as production during the year (FAOSTAT via Our World in Data). The full history is below.

The 2015–2024 reference class averaged 2.031 million tonnes, with a sample standard deviation of 236,000 tonnes. Production reached 2.359 million in 2022, then fell to 1.822 million in 2023 and recovered only to 1.890 million in 2024 (FAOSTAT via Our World in Data). The history supports wide year-to-year variation, but it receives less weight than the detailed 2026 evidence.

The national calendar-year series is especially useful. ANStat reported 1,890,774.9 tonnes for 2024, almost identical to FAOSTAT's 1,890,442, and 1,840,311.5 tonnes for 2025 (ANStat, April 1, 2026). This close match makes the national series a better guide to eventual resolution than a private crop-year forecast.

ANStat also reported that first-quarter 2026 cocoa production rose 43.6% year over year (ANStat, July 24, 2026). The first-quarter 2025 total was 334,221.3 tonnes, so the reported growth implies about 479,942 tonnes in January–March 2026 (ANStat, 2025 first-quarter table). This confirms a strong start, but cocoa production is heavily backloaded: only 554,897 tonnes were produced in the first half of 2025, leaving about 1.285 million for July–December (ANStat, December 2025 total; ANStat, June 2025 table).

The nearly completed 2025/26 crop is large. The regulator forecast total production of 2.0–2.1 million tonnes on May 21, after port arrivals had exceeded 1.7 million by May 11 (Reuters, May 21, 2026). Arrivals subsequently reached 1.999 million by August 23 and had slowed to about 3,000 tonnes in the latest week (CocoaIntel, August 24, 2026). These figures put the January–September 2026 contribution from the old crop near 0.9–1.0 million tonnes.

The new crop supplies the other half. Reuters reported on August 26 that it was delayed by eight to ten weeks, with September arrivals expected below 15,000 tonnes per week and October below 25,000. The same sources still expected about 900,000 tonnes to reach ports during October–December 2026, compared with an updated 1.1 million in the same months of 2025 (Reuters, August 26, 2026). This direct fourth-quarter estimate matters more than broad forecasts for the whole 2026/27 season.

Field evidence gives that estimate a downside tilt. Four pod counters and five exporters put the September 2026–February 2027 main crop at 1.35–1.45 million tonnes, down from about 1.6 million, after more than 20% of flowers and young pods died between May and June. Excess moisture also encouraged black-pod disease (Reuters via Business Recorder, July 12, 2026). NOAA then assigned greater than 90% probability to a very strong El Niño during Northern Hemisphere autumn and winter 2026/27 (NOAA, August 13, 2026). I treat that as added tail risk, not a second automatic crop deduction, because the later Reuters arrival forecast already reflects poor field conditions.

Port arrivals are not the resolving statistic. Rejected trucks can be counted at several exporters, while stocks, cross-border flows, local processing and harvest-to-delivery timing can push official production either above or below reconstructed arrivals. In one documented comparison, exporters counted 1.624 million tonnes while the regulator counted 1.540 million, a gap of roughly 5% (Reuters via TradingView). I therefore do not apply a fixed arrivals-to-production multiplier.

The later government planning anchor was 1.8 million tonnes for 2026. The same document assumed only 1.71 million for 2025, versus the eventual national figure of 1.840 million, so the plan has recently been conservative (Côte d’Ivoire Citizen Budget 2026; ANStat 2025 result).

I combined four regimes. The main calendar-year regime has 70% weight and is centered at 1.86 million tonnes. A delayed or weaker late crop has 15% weight and is centered at 1.64 million. A favorable delivery or high official-statistic regime has 10% weight and is centered at 2.04 million. The permitted 2025/26 crop-year proxy receives 5% weight and is centered at 2.05 million. Heavy-tailed component distributions allow for revisions and weather surprises.

The resulting distribution has a mean of 1,854,500 tonnes, a median of 1,860,760, and a standard deviation of about 159,000. Its 10th–90th percentile interval is 1.647–2.051 million tonnes, while the 5th–95th interval is 1.576–2.101 million. It assigns 15% probability below 1.7 million, 17% above 2.0 million, 5% above 2.1 million, and 1% above 2.2 million.

What's non-obvious

The regulator's 2.0–2.1 million-tonne forecast is not the calendar-year answer. A large part of that crop arrived or was produced in late 2025. Subtracting that block before adding the weaker late-2026 crop moves the estimate well below 2 million (Reuters, May 21, 2026; Reuters, August 26, 2026).

The bearish story can also go too far. An eight-to-ten-week delay does not imply that only a few hundred thousand tonnes will arrive by year-end. The same late-August report that described the delay forecast about 900,000 tonnes for October–December, within the normal 800,000–1.0 million range (Reuters, August 26, 2026). The evidence supports a weak and late crop, not a collapse.

Uncertainties

• The 900,000-tonne October–December estimate is still a forecast. November and December will determine much of the final result (Reuters, August 26, 2026).
• Production, exporter arrivals and regulator arrivals are different measures. Stock movements, duplicate counts, informal trade and recording delays could shift the eventual figure by several percentage points (Reuters via TradingView).
• ICCO's newest detailed country tables were not publicly accessible at the cutoff. Its May 29 bulletin warns that estimates can be revised, while the public statistics page states that country production data are revised quarterly (ICCO, May 29, 2026; ICCO statistics).
• The final source may apply a crop-year convention despite the calendar-year wording. I assign that outcome only 5% because ANStat already publishes twelve-month totals and FAOSTAT explicitly maintains an annual production series (ANStat; FAOSTAT via Our World in Data).