Brazil tops world exports of coffee, sugar, and soybeans. Explore what a bad season would expose.
1The setup
Brazil is the world's largest exporter of coffee (34.5%), sugar (47.4%), and soybeans (53.6%) — shares of global export value, 2024 vintage. When one country's harvests feed that much of world trade, a single bad season travels fast. This guide walks through what the monitor is seeing right now, how exposed each market is, and lets you model a disruption yourself.
Market check · Polymarket
What the market thinks
Peak Severe Drought in Minas Gerais during the Coffee Growing Season?
Polymarket · live contract
under 5%
11%
5% to 19.99%
23%
20% to 39.99%
33%
40% to 59.99%
20%
60% or more
21%
The market's favorite outcome is “20% to 39.99%” at 33% — think the real chance is higher or lower? Take a side and we'll track your call against the live market as it moves.
Your take lives on this device only. Illustrative — not a bet, not advice.
The same drought market shown on the Brazil coffee page — severe-drought coverage in Minas Gerais, Brazil's coffee heartland. It covers neither sugar nor soybeans, and it is not a contract on crop prices.
Illustrative only — not financial advice, not a recommendation.
2Right now
☕Coffee
QUIET
Stress 2/10
›Cerrado Mineiro: week-ahead forecast shows severe deficit rainfall
›Garzon: extremely wet rainfall — 207% of the 30-day normal
›Garzon: week-ahead forecast shows wet rainfall
🍩Sugar
ELEVATED
Stress 7/10
›price +7.0% in 5d
›Ribeirao Preto: extremely wet rainfall — 316% of the 30-day normal
›Ribeirao Preto: week-ahead forecast shows severe deficit rainfall
🫘Soybeans
QUIET
Stress 3/10
›Sorriso: week-ahead forecast shows severe deficit rainfall
›Rosario: severe deficit rainfall — 32% of the 30-day normal
›Harbin: severe deficit rainfall — 36% of the 30-day normal
Monitor data as of Sep 24, 2026. Scores are observed readings, not forecasts.
3The exposure
Brazil
34.5%
Vietnam
12.1%
Colombia
10.9%
Indonesia
4.6%
Ethiopia
4.3%
Share of global export value (USD) — 2024 vintage. Amber = the scenario focus. Ranked by share.
How the shock travels
Follow the shock downstream — from the coffee market, through processors, to the shelf. These are the usual stages, not this scenario's outcome: exposure at each step, never a prediction.
Day 0: coffee futures spike
timing not established
Lag basis: Day 0 is the shock itself; no lag applies.
green (unroasted) coffee beans
Roasters and manufacturers buy green coffee and reset list prices
timing not established
Lag basis: lag not established in sources reviewed
bagged ground coffeeinstant coffee
Retail shelf and coffee shops
timing not established
Lag basis: lag not established in sources reviewed
grocery-store coffeecoffee-shop drinks
Starbucks
SBUX
Green coffee is a core input: as of Sep 29, 2024, Starbucks had committed to purchasing $180 million of green coffee under fixed-price contracts and an estimated $929 million under price-to-be-fixed contracts (FY2024 10-K).
Coffee is a reported product category: in 9M 2024, coffee sales grew at a mid-single-digit rate, 'driven by Nescafé, particularly in ready-to-drink offerings' (Nestlé 9M 2024 sales release).
No single corridor — share not applicable — Brazilian coffee exports load at Santos and other Atlantic ports; no single strait or canal applies. No route share applies -- this is not missing research.
How we map this
Exporter and importer shares are 2024 vintage · annual data, 1–2y lag from UN Comtrade and OEC (BACI/CEPII); an importer counts as exposed when ≥20% of its coffee imports come from a shocked exporter. The shock geography is the scenario’s own focus selection. Annual data with a 1–2 year reporting lag — the structure moves slowly, the prices don’t.
4Adjust the assumptions
70%
Historical check: The closest episode, Brazil drought, 2024–25, saw coffee futures move +95.7% (extreme band); your 70% disruption setting sits in the large band.
What share of Brazil’s coffee exports is disrupted.
Resolution: Share of the focus exporter's shipments disrupted
Share of world exports
0%50%100% of world exports
Brazil coffee — 34.5% of world Coffee exports (Share of global export value (USD) — 2024 vintage)
Exposed under your assumptions: 24.2% (34.5% × 70% severity)
The hatched area is modeled from your assumptions, not a measured outcome.
Exposed ≠ lost — affected trade can reroute, draw stocks, or substitute. This sizes the exposure, not the damage.
90d
Historical check: The closest episode, Brazil drought, 2024–25, lasted 365 days (long); your 90-day duration sits in the medium band.
Resolution: Days the disruption lasts
60d
Your assumption — not a measured buffer.
Resolution: Assumed days of stockpile cover — an assumption, not a measured buffer
Disruption timeline
090180270365 days
Disruption: 90 days
Assumed cover: 60 days — your assumptionYour assumption
30 days buyers can't ride out — not automatically a physical shortage
What this would have meant
At these settings, about 24.2% of the world's coffee exports would be exposed to the disruption, and the disruption runs 90 days against 60 days of assumed stockpile cover, leaving about 30 days beyond what the assumed stockpile covers.
The closest recorded episode, Brazil drought, 2024–25, saw coffee futures rise +95.7% in 425 days (Jan 2024 → Mar 2025), and for consumers: ground roast coffee (US) rose +20.6% to +32.8% over 0–12 months. Illustrative — what happened then, not what will happen now.
Your settings aren't an exact match: your 70% disruption setting implies a large move, while the episode's futures move was +95.7% (extreme).
Exposed ≠ lost: affected trade can reroute, draw down stocks, or find substitutes — this number sizes the exposure, not the damage.
Exporter share ≠ spare capacity: rivals may not be able to surge supply to fill the gap.
Stockpile cover of 60 days is your assumption, not a measured buffer.
Results are physical-exposure arithmetic from your stated assumptions — not a price forecast.
Analog check
Your settings look closest to Brazil drought, 2024–25(same commodity, same shock type, similar supply concentration). The episode's measured moves, caveats included: coffee futures +95.7% in 425 days (Jan 2024 → Mar 2025); Ground roast coffee (US): +20.6% to +32.8% over 0–12 months. Vietnam's robusta drought lifted robusta prices in parallel; broad 2024–25 food-at-home inflation also pushed shelf prices..
Only about 11.8¢ of every US consumer food dollar — and 18.5¢ of every food-at-home dollar — reaches the farm (USDA, 2024) — that's a ceiling, not a prediction: margins can absorb a spike or amplify it.
Illustrative — what happened then, not what will happen now. Your scenario differs: your 70% disruption setting implies a large move, while the episode's futures move was +95.7% (extreme).
Commodity:requirement — Both are coffee — non-coffee episodes are excluded outright.
Shock type:2/2 — Both are a drought shock.
Size:0/1 — Your 70% disruption setting band: large; episode futures moved +95.7% (Jan 2024 → Mar 2025) — band: extreme. Severity is % of shipments disrupted, not a price move: this banding is a v1 matching heuristic.
Duration:0/1 — Your disruption: 90-day (medium); episode: 365-day (long) — Apr 2024–Mar 2025 drought window (repo event record brazil-coffee-drought-2024).
Supply concentration:1/1 — Your hazard zone: 34.5% of world exports (high); episode disrupted ~33% (high) — Brazil is about a third of world coffee exports (Conab, Sep 2024 — repo event record brazil-coffee-drought-2024).
Score 3/5 — partial analog (mismatches named above).
El Niño: the climate driver that doesn't always deliver
Recurring — 2009-10, 2015-16, 2023-24
“El Niño moves the weather, not always the price. In 2015–16 wheat fell sharply but tracked outcomes were mixed (rice +7.7%, soy +2.1%) while sugar spiked — fear the mechanism, measure the outcome.”
Price moves were observed around the event — not proof the event caused the entire move.
“Drought trimmed Brazil's crop — but near-empty warehouses turned a bad harvest into a 14-month rally that nearly doubled arabica. When inventories are thin, weather gets priced twice.”
Price moves were observed around the event — not proof the event caused the entire move.