SINE Strategy Wiki

Coffee — Minimum Viable Context (MVC), v0.1

31 min read

Status: v0.1, 2026-08-18. Built on request of Tell Münzing as an experiment in a new artifact type: a Minimum Viable Context — the smallest body of shared context that lets a strategy conversation about financial markets proceed without either side guessing. His framing (Slack, 2026-08-18): “the lack of a common context. My world: I need to get you into the context and relationships.” The test case he chose: the GIZ/WHH approach in coffee.

The design principle is that this page’s job is to be wrong in visible places. It is a working instrument, not a finding. Every cell that is guessed is marked. The register of holes at the end is the most valuable part.

Provenance. Facts here come from six commissioned research notes filed on 2026-08-18 in sources/research/ (…-global-coffee-platform.md, …-welthungerhilfe.md, …-giz-leadership-private-finance.md, …-coffee-value-chain-data.md, …-agri-coffee-finance-instruments.md, …-coffee-asset-owners-principals.md). Those notes are research-grade web research, not primary sources — figures need primary verification before external use. Statements made by Tell Münzing in the 2026-08-18 Slack conversation are marked ⟨Tell Münzing, Slack 2026-08-18⟩ and are not corroborated unless a source is also cited. My own reasoning is marked ⟨Inference: …⟩.

1. The spine: principals sit at both ends, and SINE is currently in the middle

Tell’s thesis: “we need to shift closer to actual asset owners and principals.” ⟨Tell Münzing, Slack 2026-08-18⟩

⟨Inference: in coffee the principals are at both ends of the chain, not one. The smallholder owns land, trees and harvest — a principal with a tiny asset and no agency. Pension funds, insurers’ balance sheets and the roasters’ owners are principals with large assets and remote sight. Everyone in between — GCP, WBCSD, GIZ, WHH, traders, auditors, certifiers, and SINE today — is an agent, acting on someone else’s mandate and someone else’s money. The move Tell names is therefore not one move but two, in opposite directions, and the thing that connects them is a verified primary-data stream. That is the candidate spine of the finance strategy.⟩

The ownership research supports the “large remote principal” end concretely: coffee’s buy side is unusually family- and state-controlled, not dispersed. Melitta 100% Bentz family; Lavazza 100% family via Finlav; Tchibo 100% Maxingvest (Michael and Wolfgang Herz); ECOM ~92% Esteve family; Louis Dreyfus 55% family Akira Trust / 45% Abu Dhabi’s ADQ; Olam Agri 80.01% Saudi SALIC (a PIF subsidiary); NKG family-controlled; JDE Peet’s JAB-controlled and now being acquired by Keurig Dr Pepper. Nestlé is the exception, dispersed with BlackRock ~5% (voting capped at 5%). sources/research/2026-08-18-coffee-asset-owners-principals.md

⟨Inference: this changes the addressing problem. A strategy aimed at “the capital markets” mostly mis-addresses coffee, because most of the buy side is not answerable to capital markets. It is answerable to a named family or a sovereign fund — a small number of identifiable principals with long horizons and no quarterly disclosure pressure. Long horizons cut both ways: patient capital is reachable by an argument about durable resource security, and immune to an argument about next quarter’s ESG score. This is an argument for approaching principals through relationships rather than through disclosure regimes.⟩

2. The chain, stage by stage

Five columns per stage, per Tell’s correction: (1) primary data — who holds it, what is missing; (2) the claim that must be made believable, and to whom; (3) SINE’s five instruments applied — governance approach, data collaboration, primary data, standards, open-source code; (4) economics — how the value proposition and business case shift, and what business-model evolution that opens; (5) finance touchpoint — who pays, which claim gates the money, what is unverifiable today.

2.0 Nature / origin

  1. Primary data. Held by nobody as an owner. Proxied by satellite (deforestation baselines against the EUDR cut-off of 31 Dec 2020), remote-sensing vendors, and public forest datasets. Missing: soil carbon, below-ground biomass, water balance, on-farm biodiversity — none routinely measured at plot level.
  2. Claim. “This hectare was not deforested after 2020, and its ecological state is improving.” Owed to EU operators (EUDR), to carbon standards (Verra VCS, Gold Standard, Plan Vivo — all requiring additionality and permanence, typically enforced via a withheld buffer pool), and to insurers. sources/research/2026-08-18-agri-coffee-finance-instruments.md
  3. SINE instruments. ⟨Inference: this is the stage where SINE’s governance approach matters most and its code matters least. Nature generates no data by itself; someone decides what gets measured, by whom, and who may see it. That is a commons-design question in Ostrom’s sense (→ Ostrom’s commons design principles), not a software question.⟩
  4. Economics. ⟨Inference: nature is the only stage where the asset can appreciate rather than depreciate, which is why permanence claims are worth money. Permanence is a time claim and time claims are exactly what a ledger-style record can carry and a snapshot certificate cannot (→ ledger-based carbon accounting).⟩
  5. Finance touchpoint. Carbon and biodiversity payments. Confirmed coffee case: Uganda Carbon Coffee Project with ECOTRUST under Plan Vivo, 3,000+ farmers; Efico/Aldea Coffee buying Plan Vivo credits tied to named farmers’ agroforestry adoption. Unverifiable today: the measurement protocol itself — biomass/soil sampling frequency and auditor were not retrievable for either project. sources/research/2026-08-18-agri-coffee-finance-instruments.md

2.1 Smallholder farm

  1. Primary data. ~12.5m farms globally, mostly under 2 ha. Holds plot boundaries and tenure (often informal), planting density, variety, input use, harvest volume, picking dates — largely undigitised, on paper or in memory. Cannot see: downstream roast profile, blend composition, retail price, end-buyer identity, or what premium its own coffee earned. sources/research/2026-08-18-coffee-value-chain-data.md
  2. Claim. Four at once, to four different audiences: my plot is where I say it is and was not deforested (to the EU operator); my inputs were what I say they were (to the PCF calculator); my harvest exists and is stored (to the lender); my loss occurred (to the insurer).
  3. SINE instruments. ⟨Inference: all five bear here, and this is the stage where the farmer-centric data governance line SINE is already bidding on (→ GIZ Tender 10036847, incl. the open-source data-wallet prototype) meets the finance question. The data wallet is not primarily a compliance tool; it is the farmer’s instrument for holding a claim that someone else needs. Whoever holds the claim can charge for it.⟩
  4. Economics. Farmers capture ~3–10% of final retail price; a German-market margin study puts the farmer at ~€0.41/kg against a retailer’s €1.39/kg and a roaster’s €0.89/kg. The 2.5%–10% spread across sources is genuine methodological disagreement, not resolved. sources/research/2026-08-18-coffee-value-chain-data.md ⟨Inference: the business-model shift available here is not “a better price for the same coffee.” It is the farmer becoming a supplier of two products — coffee, and a verified data record about the coffee — with different buyers for each. That is the business-model evolution Tell asked the MVC to make visible, and it is the one SINE’s instruments actually create.⟩
  5. Finance touchpoint — and the hinge of the whole map. Warrantage. See §3.

2.2 Producer organisation / cooperative

  1. Primary data. Aggregated member lists, plot GPS points where collected for certification, cherry delivery records, prices paid per farmer. Missing: verified deforestation status for every member plot; harmonised digital records (much still paper); farm-level input and emissions data.
  2. Claim. “I can vouch for all my members.” ⟨Inference: this is the single most load-bearing and least verified claim in the chain — the co-op is asked to aggregate a guarantee it has no instrument to check.⟩
  3. SINE instruments. ⟨Inference: data collaboration in its literal sense. A co-op is already a data-pooling institution with a membership rulebook; it is the nearest thing in the chain to a functioning commons. The unglamorous consequence is that SINE’s governance approach has an existing institutional host here and does not need to invent one.⟩
  4. Economics. No consolidated margin figure found; capture varies with co-op governance and direct-trade access. ⟨Inference: aggregation is the co-op’s only source of pricing power, and verified aggregate data is a second form of the same thing.⟩
  5. Finance touchpoint. Pre-harvest and working-capital lending against offtake contracts, not against land: Root Capital ($200k–$2m loans, repayment structured to harvest cycles, purchase order as the collateral instrument; 53% of borrowers are farmer-owned associations, coffee the primary crop for roughly half); Shared Interest (Capucas, Honduras, since 2015 — terms undisclosed); Incofin ($48m+, 14 countries). sources/research/2026-08-18-agri-coffee-finance-instruments.md

2.3 Washing station / wet mill

  1. Primary data. Cherry intake by farmer/weight/date, processing method, fermentation time, water source, moisture, lot sheets. Described in the sources as “the most critical traceability node” because it is the first point of aggregation across farmers. Missing: per-plot geolocation unless deliberately collected — and at larger facilities, cherries from multiple farmers are routinely commingled, destroying individual-lot traceability. sources/research/2026-08-18-coffee-value-chain-data.md
  2. Claim. “This lot is composed of these farmers’ cherries in these proportions.”
  3. SINE instruments. ⟨Inference: the wet mill is where standards and code earn their place, because commingling is a mass-balance problem, and mass balance is precisely the point where a claim about a physical mass becomes a claim about an accounting convention. This is the same structural problem the wiki already holds under ledger-based carbon accounting: chain-of-custody models are the part the ICC review left unresolved. Coffee’s wet mill is a concrete instance of that unresolved thing.⟩
  4. Economics. No stage-specific margin found; captured inside co-op or exporter margin.
  5. Finance touchpoint. ⟨Inference: none directly today. The wet mill is invisible to finance while being the node that determines whether any downstream claim can be substantiated at all — a structural mismatch worth naming.⟩

2.4 Dry mill / exporter

  1. Primary data. Parchment reports, cupping/quality grades, export lot identity, HS codes, buyer contracts and FOB prices. Missing: verified farm-level deforestation compliance unless separately sourced.
  2. Claim. “This export lot is graded as stated and legally compliant.”
  3. SINE instruments. ⟨Inference: standards — this is where a national data infrastructure either exists or does not, and where GIZ’s convening power is real.⟩
  4. Economics. Exporter margin ~€0.29/kg.
  5. Finance touchpoint. Export credit; the FMO-led $155m long-term working-capital facility for ECOM with BII, BIO, DEG and PROPARCO, explicitly for coffee/cocoa/cotton sourcing and targeting smallholder livelihoods in Ghana and Côte d’Ivoire. sources/research/2026-08-18-coffee-asset-owners-principals.md

2.5 International trader

  1. Primary data. The richest position in the chain: consolidated multi-origin volumes, quality labs, hedging positions, and increasingly proprietary sustainability platforms — Sucafina’s IMPACT programme runs 11 mandatory plus 71 improvement indicators, with a Farmer Connect / IBM Food Trust blockchain pilot. Dominant firms: NKG (export mills, warehouses and quality labs in 27+ countries), Volcafe (ED&F Man), Olam/ofi, ECOM, Louis Dreyfus, Sucafina. Precise market shares not found. sources/research/2026-08-18-coffee-value-chain-data.md
  2. Claim. “My book is compliant and my sustainability performance is improving.”
  3. SINE instruments. ⟨Inference: the trader is SINE’s hardest and most important counterparty, because the trader is the one actor that already has the data and monetises its asymmetry. SINE’s data collaboration is a direct threat to informational rent and simultaneously the only route to the trader’s own cheaper capital. The strategic question is whether confidentiality-preserving verification is enough to make that trade acceptable — which is exactly the hypothesis already on file (“confidentiality solves the entry problem and damages the dissemination mechanism”, → coordination vs. bargaining failure). Coffee’s trader stage is a live test of it.⟩
  4. Economics. Trader margin ~€0.26/kg — the lowest of any measured stage. ⟨Inference: worth dwelling on. The actor with the most data and the most market power earns the thinnest per-kilo margin, because the business is volume and risk management, not value capture per unit. It follows that a data proposition to a trader must be argued as cost of capital and risk, never as margin.⟩
  5. Finance touchpoint. Sustainability-linked revolving credit: ECOM $600m (a $400m one-year tranche at 125bp plus a $200m three-year tranche at 175bp, ABN AMRO); Olam $500m, 15 banks, pricing tied to annual performance against 50 ESG criteria assessed by Sustainalytics. What is unverifiable today, and this is the finding: the claim gating that money is an ESG rating, not primary data. Nothing in the pricing mechanism touches a farm. sources/research/2026-08-18-agri-coffee-finance-instruments.md

2.6 Roaster

  1. Primary data. Blend formulation, roast profiles, brand claims, prices paid to traders, chain-of-custody or mass-balance certification records. Missing: independent confirmation of farm-level deforestation and legality beyond what certifiers or traders attest — “the crux of the EUDR compliance gap.” Top five roasters ≈55–60% of global roasted/soluble volume (indicative, not audited).
  2. Claim. “Everything I sell is legal, deforestation-free, and as sustainable as I say.” Owed to the EU (as operator, where the roaster is the importer of record for direct sourcing), to consumers, and to owners.
  3. SINE instruments. ⟨Inference: the roaster carries the legal liability and holds the least upstream data — the widest gap between exposure and sight anywhere in the chain. That gap, not a sustainability ambition, is the commercial opening for verification. Note also that a roaster’s governance interest is the opposite of the trader’s: the roaster wants the data pooled and the trader wants it held.⟩
  4. Economics. Roaster margin ~€0.89/kg, second-highest stage. ⟨Inference: the roaster can afford to pay for verification and, unlike the trader, has margin from which to pay.⟩
  5. Finance touchpoint — the unoccupied space. No documented case was found anywhere in which a roaster’s own cost of capital is contractually indexed to verified farm-level data. Every confirmed sustainability-linked facility sits at trader/exporter level. sources/research/2026-08-18-agri-coffee-finance-instruments.md ⟨Inference: this is the clearest white space on the map, and it is at the end of the chain where the principals are family owners and the margin exists. If SINE wants one demonstrable financial-market claim, this is the candidate: replace a rating-based proxy with a verified primary-data trigger, once, at roaster level.⟩

2.7 Retail / food service

  1. Primary data. Consumer sales, loyalty and pricing data. Holds nothing upstream that the brand did not pass on.
  2. Claim. Shelf-level sustainability claims; Rainforest Alliance is now extending certification and traceability requirements to retail and food-service buyers.
  3. SINE instruments. ⟨Inference: standards, defensively. Retail is where claims get litigated and where greenwashing liability lands.⟩
  4. Economics. Highest margin per kg (~€1.39/kg).
  5. Finance touchpoint. ⟨Inference: none specific to coffee found; retail finance is balance-sheet finance and does not reach into the chain.⟩

2.8 Consumer

Holds purchase and consumption choice, no chain data. ⟨Inference: the MVC should resist treating the consumer as a lever. Every stage above prices its claim for a regulator, a lender or an owner, not for a shopper.⟩

2.9 Cycle closure / waste

Spent grounds, husk, packaging. A genuine hole: the research found no source at all on data holders or data practice for coffee’s circular streams — not an omission but an absence. ⟨Inference: notable given that Tell’s opening framing runs explicitly “till the end consumption phase and closure of the resource cycle.” The stage he named as the endpoint is the stage with no data infrastructure whatsoever.⟩

3. The hinge: why warrantage is the right place to start, for a reason Tell did not state

⟨Canonical page for the instrument itself — mechanism, preconditions, evidence, limits, open items: Warrantage. What follows is the strategic argument, which stays here.⟩

Tell: warrantage “is localised finance that is close to the credibility and credit claim.” ⟨Tell Münzing, Slack 2026-08-18⟩

The mechanism: farmers deposit non-perishable stored produce in a warehouse instead of selling into the harvest-price trough, and borrow against it; repayment on later sale at a higher price. The warehouse is dual-locked, keys held jointly by the bank and the farmer group — that padlock is the verification mechanism. Pioneered in Niger in 1999 with FAO support, extended to Burkina Faso, Mali and Senegal. Three documented preconditions: a functioning farmer association, a willing local bank or MFI, a secure store. A Niger evaluation reports participating farmers’ incomes up 19–113% within six months and yields up 44–120%. Coffee-specific application is documented in Mbinga District, Tanzania. sources/research/2026-08-18-agri-coffee-finance-instruments.md

Now the part that makes it structural rather than anecdotal — land tenure:

  • Ethiopia: land is constitutionally state-owned (Art. 40). Farmers hold inheritable but non-transferable usufruct; they cannot sell, exchange or mortgage land. Coffee smallholders — 95% of national output — are inside that regime.
  • Vietnam: Land Use Rights Certificates can be mortgaged; banks lend 70–85% of appraised collateral value.
  • Uganda, Brazil, Colombia: not retrieved. A hole. sources/research/2026-08-18-coffee-asset-owners-principals.md

⟨Inference: warrantage is not a preference for the local; it is what is left when land cannot be pledged. Where the immovable asset is unmortgageable, the only pledgeable asset is the movable harvest — and the only available verification technology is a physical padlock, which caps the instrument at the size of a warehouse someone can walk to. That is why this finance is local: not by design, but by the reach of its verification mechanism.

This gives SINE’s proposition its sharpest possible formulation in this domain: a verified data stream is a claim-verification technology that can do what the padlock does, without the padlock’s radius. If the claim “this harvest exists, is of this quality, is stored here, and is not pledged twice” can be made believable by verifiable data rather than by joint physical custody, the instrument stops being local. That is the same competence the wiki already identifies as SINE’s core — verification — reappearing in a place nobody has connected it to (→ sources/discussions/2026-08-15-strategiegespraech-themenfelder.md, Martin Pompéry: “SINEs Kompetenzbündel ist im Kern eine Verifikationskompetenz”).

It also reframes the farmer-focus question. Tell says the farmer focus is interesting because “regenerative agriculture, biodiversity and carbon permanence also need the local context.” The tenure finding says something stronger and less comfortable: for a large share of the world’s coffee farmers the asset they own is legally unfinanceable, so the harvest and the record about it are the only financeable assets they have. Making that record credible is therefore not a sustainability intervention. It is the only route to the farmer holding a bankable asset at all.⟩

Caveat against over-reading. Warrantage’s documented default handling is that stored grain is sold to recover the loan — the collateral is physically present and liquidatable. A data-verified claim is not. ⟨Inference: substituting verification for custody moves the risk from theft to fraud, and no source here says that trade is favourable. It is a hypothesis with a named failure mode, not a conclusion.⟩ No loan-to-value figure and no aggregate default rate for warrantage could be found — both material to the argument and both open.

4. Actor register

Per Tell’s emphasis on “mandate, finance and relationships that hold the power to transform,” each actor is recorded by mandate / money and horizon / relationships / principal-or-agent and distance to the asset.

Global Coffee Platform (GCP) — the invitation

  • Mandate. Multi-stakeholder membership association, 160+ members, formed March 2016 by merging the 4C Association with IDH’s Sustainable Coffee Program; Geneva seat, Bonn secretariat (sources inconsistent); Executive Director Annette Pensel. Board of up to 16, 13 elected: 4 producer, 2 trade, 4 industry, 3 civil society, with the ICO as observer.
  • Money and horizon. Member fees plus co-funding from IDH, BMZ via GIZ (which funded the Kenya country platform), Rainforest Alliance, ISEAL Innovation Fund. No figures found. Horizon: institutional, indefinite.
  • Relationships. Nestlé, JDE Peet’s, Keurig Dr Pepper, illycaffè, Melitta, Tesco, Westrock, Julius Meinl; traders ECOM, Louis Dreyfus, Volcafe, Sucafina, Walter Matter; NGOs Rainforest Alliance, Conservation International, Fair Trade USA, Enveritas. Both NKG and HRNS are members (NKG rejoined December 2022).
  • Principal or agent. Agent, far from the asset — but the convening agent. Per Tell, conveners like GCP “provide the initial invitation.”
  • What it already has, and what it lacks. GCP owns a Coffee Data Standard — 15 common farm-level indicators, built 2017–18 with COSA, Rainforest Alliance and WaterWatch, now underpinning the Delta Project with the ICO, Better Cotton and ICAC and aiming at government endorsement. It also runs Collective Reporting (2024: 11 reporting roasters/retailers, sustainable volume up from 639,710 MT in 2018 to 1,738,800 MT) and ~10 country platforms, of which 7 are named — Brazil, Honduras, Indonesia, Kenya, Uganda, Vietnam, Ethiopia. No GCP position on EUDR could be located, and no substantive statement on data ownership or governance of the aggregated data. sources/research/2026-08-18-global-coffee-platform.md
  • ⟨Inference: this is the most consequential single finding for SINE’s entry. GCP is a convener that already holds a farm-level data standard with a route to government endorsement — and, as far as this research can see, no answer to who owns the data, who may see it, or how a claim under it is verified. That is precisely and only SINE’s layer. SINE does not need to build a coffee data standard; one exists, inside the institution that issues the invitation, missing the governance and verification layer SINE builds. The corollary is uncomfortable and should be stated: GCP is also the actor best positioned to build that layer itself or to buy it from someone else.⟩

GIZ — the reach

See GIZ. Relevant additions from this research: SAFE (Sustainable Agriculture for Forest Ecosystems) runs in 10 countries including Uganda, Vietnam, Brazil and DRC, “expands traceability systems,” targets 11,000 smallholders and 121,700 ha, co-funded by the EU, BMZ and the Netherlands; GIZ-SAFE coordinates the Team Europe Initiative “Zero Deforestation Hub” supporting EUDR implementation. Private-capital instruments: develoPPP (up to €2m per project, partner covers ≥50%, implemented with DEG Impulse), an MoU with Allianz Trade on export credit guarantees, green loan/bond advisory, and the IKI project “Transformative Climate Finance — Shifting Financial Flows and Private Sector Mobilisation.” No GIZ-wide transition-finance unit found. sources/research/2026-08-18-giz-leadership-private-finance.md

  • Money and horizon. Commissioned, project-tied, short — per Tell, GIZ and WHH “run on project or short term finance on which they invite and work.”
  • Principal or agent. Agent, explicitly and by construction: GIZ acts on BMZ’s mandate with BMZ’s money.
  • ⟨Inference: GIZ appears twice in this map from opposite directions — as co-funder of GCP’s Kenya platform and as commissioner of the data-governance tender SINE is bidding for. It is the connective tissue between the invitation and the substance, and it cannot supply either.⟩

Welthungerhilfe (WHH) — the network

  • Mandate. Deutsche Welthungerhilfe e.V., founded 14 Dec 1962; ~37 countries currently, 647 overseas projects; €5.07bn disbursed cumulatively.
  • Money and horizon. 2024: €86.5m private donations against €291.9m institutional grants (~23% / 77%), of which German federal funds (BMZ + Auswärtiges Amt) were just over half. Donors include BMZ, AA, EU/ECHO, FCDO, FAO, GIZ, KfW, WFP, USAID, UN agencies. Horizon: grant cycles. ⟨Inference: this quantifies Tell’s point exactly — 77% of WHH’s money is earmarked and time-boxed, so WHH can invite and implement but cannot commit beyond a grant.⟩
  • Relationships. Coffee presence is real but small and corporate-partnered: Coffee Circle has funded WHH work in Ethiopia’s Limu region since Nov 2014 (800+ coffee farmers, washing facilities, nurseries in Jimma); a Bblo Kawa cooperative coffee project in eastern DRC (2023–2025). Climate programmes SRAPLEA (Ethiopia/Kenya/Uganda) and Green Evolution (South Asia).
  • Principal or agent. Agent, close to the asset — WHH’s distinctive value in this map is proximity, not capital.
  • A contradiction to hold open. See §7. sources/research/2026-08-18-welthungerhilfe.md

Traders, roasters, owners

Covered in §1 and §2.5–2.6. The register entry that matters: their money is patient and their principals are named.

Institutional investors and insurers — the far principals

  • PRI Spring: 240 investors, >$19T AUM (2025), 96 actively engaging investees; signatories include AP2, AP3, AP7, Brunel Pension Partnership, Ethos, CDPQ, Bayerische Versorgungskammer, Cardano. Also Nature Action 100 (forest-risk commodities explicitly include coffee), the Investor Policy Dialogue on Deforestation (PRI/Ceres), and FAIRR. What none of them could be shown to have: a specified list of the data they actually demand, or a stated account of what is missing. That gap was searched for and not found.
  • Insurers: the smallholder financing gap exceeds $200bn/year; insurance covered under 3% of losses in recent Asia-Pacific floods and droughts (an 86% protection gap); the BIS/IAIS Financial Stability Institute names data infrastructure as a systemic constraint on reinsurance pricing and underwriting for climate risk; Swiss Re promotes parametric cover to reach smallholders. No coffee-specific reinsurer statement found. sources/research/2026-08-18-coffee-asset-owners-principals.md
  • ⟨Inference: the BIS/IAIS line is the strongest single piece of external corroboration in this map for SINE’s whole premise — a financial-stability body stating that the binding constraint on pricing climate risk is data infrastructure. It also vindicates Martin Pompéry’s 2026-08-15 argument for putting insurance first: insurability precedes financeability, and the problem is informational rather than capital. Coffee then supplies the concrete instance, because parametric insurance’s central adoption barrier is basis risk — the mismatch between an index reading at a reference station and one farmer’s actual loss — and basis risk is a primary-data resolution problem in the most literal sense. Reduce the distance between the measurement and the loss and the instrument works. Only Colombia was found as a coffee-specific parametric case; East and West African coffee, not confirmed.⟩
  • ⟨2026-10-14 — the industry’s own account, and it complicates the one above.⟩ Sounding the Alarm, an anonymous memo by senior UK/European (re)insurance professionals, is the first primary voice from inside this actor class in the wiki. It does not say the data is missing. It says the models exist, vary greatly in quality between firms, are commercially adjusted downward where a higher risk estimate would cost margin or market share, and that the industry “collectively rel[ies] on a small number of base case models” because “internal models are very costly to develop and robust external data comes at a steep price” (→ the NatCat model layer). It names farming communities and the agricultural supply chain among those it expects to be hit hardest, flagging that they “represent a small part of the insurance market”, and states the industry’s intended response as reprice → restrict → withdraw (→ the protection gap). ⟨Inference: two things follow for this map, and neither is comfortable. First, the BIS/IAIS line above is not refuted but it is no longer the whole constraint — missing infrastructure and deliberately shaded projections are different problems, and only the first is fixed by building infrastructure. Both are held open. Second, the insurance-first thesis has to be stated more narrowly than §1 states it. Better information does not make an uninsurable risk insurable; this memo describes an industry whose response to a sharper view of risk is a faster exit. What better information does is move the line — it lets an individual farm with demonstrable adaptation be priced apart from its region, which is a claim about granularity, not about aggregate insurability. That narrower claim is the one this map can defend, and it happens to be the same claim basis risk already makes. Caveat on the memo itself: anonymous, unevidenced, from a party with an interest in shifting cost to government, and about Europe and North America rather than coffee origins. It still leaves no coffee-specific reinsurer statement on file.⟩

Blended finance — the de-risking layer

IDH Farmfit Fund: €100m closed-end blended vehicle, tranched risk-return with first-loss guarantees from public actors, plus a €30m Business Support Facility; backed by Unilever, Mondelez, Jacobs Douwe Egberts, and government funders from the US, UK, Denmark and Netherlands; FMO and Rabobank each put in €20m; a Dutch government mid-term review exists (Oct 2025). ⟨Inference: IDH appears twice — as GCP’s founding co-funder and platform funder, and as the manager of coffee’s largest de-risking vehicle with a coffee roaster (JDE) among its backers. IDH is a hub the wiki has no page for and probably should.⟩

5. People register

Role / relationship path to us / adjacency each person opens. Per Tell: people “also provide adjacency that enables new alliances and innovation.”

  • Mathias Mogge — Secretary General and Vorstandsvorsitzender, Welthungerhilfe, elected Sept 2018; agricultural engineer and environmental scientist; at WHH since 1998, Executive Director Programs from 2010. Path: direct — Tell has spoken with him; Mogge “wants to learn about our approach and how it can support their work e.g. Warrantage.” ⟨Tell Münzing, Slack 2026-08-18⟩ Adjacency: rural finance in the Sahel, and the WHH country network. ⟨Inference: also the most useful sceptic available — WHH published critical commentary on corporate power in the digital agri-food system, which means the governance question will be asked seriously rather than waved through.⟩
  • Ingrid-Gabriela Hoven — GIZ Managing Director since Oct 2020. Portfolio: Sector and Global Programmes, Sectoral Department, Finance, Evaluation, Corporate Security, and Europe/Mediterranean/Central Asia. Before: BMZ Director-General Global Issues and Commissioner for Climate Policy and Finance (2014–2020); World Bank Executive Director for Germany 2010–2014. Path: via GIZ, unspecified. Adjacency: the multilateral climate-finance world and the World Bank.
  • Anna Sophie Herken — GIZ Managing Director since 15 Aug 2023. Portfolio: Auditing, Digital Transformation and IT Solutions, Legal Affairs and Insurance, Revision, and oversight of Africa. Before: Business Division Head at Allianz Asset Management, board member Allianz Life (USA), CFO HPC Germany, MD Hertie School; earlier World Bank and EBRD. Yale Executive Education in Sustainable Finance and Investment (2022). On her own account: “I understand both worlds – the private sector and the public sector.” ⟨Inference: read the portfolio again — digital transformation, insurance, and Africa in one board mandate, held by someone who ran money at Allianz. If the insurance-first thesis is right, Herken is the single best-matched individual in this entire map, and the match is structural rather than personal: it is her formal portfolio, not her stated interest.⟩
  • Michael Opitz — MD HRNS 2005–2026, 21 years; see Michael Opitz. Adjacency: the coffee-origin implementation world and the Neumann orbit. Stepped down 2026.
  • Stefan Dierks — Director Sustainability, Melitta; 17 years Tchibo; see Stefan Dierks. ⟨Inference: the only person on file who sits inside a family-owned roaster — i.e. inside one of the named principals of §1 — and he originated the GIZ tender constellation. Adjacency: the German family-roaster cluster (Melitta, Tchibo), plus BAUM e.V. and his International Bankers Forum chapter on Mittelstand and transformation, which is a finance adjacency the wiki has not used.⟩
  • Annette Pensel — Executive Director, GCP. Path: none recorded. A hole.
  • Michael R. Neumann — chairs the HRNS board; family behind NKG. ⟨Inference: a rare single point where a large trader’s owning family and a development implementer meet. Untested.⟩

6. What shifts in the economics — the cross-cutting answer

Tell’s correction was that the MVC missed the economics: “how value propositions and business case are shifting which allows business model evolution / transformation as we do our work.”

Two numbers from the research carry most of the answer.

One — the variance is the business case. A 2025 meta-analytical review puts coffee production’s carbon footprint at a median 2.18 ± 2.04 kg CO2e per kg green beans, with nitrogen fertiliser manufacture and application responsible for 65–100% of emissions. Practice combines primary records with secondary databases such as ecoinvent, and sources consistently state the core problem is that farm-level activity data is rarely collected systematically, with the gap compounding at each upstream tier. sources/research/2026-08-18-coffee-value-chain-data.md ⟨Inference: an uncertainty band roughly as wide as the median means the number cannot price anything — not a loan margin, not a premium, not an insurance trigger. And the dominant emission source is a farm input the farmer already knows the quantity of. So the value proposition shift is precise and narrow: primary data does not make the footprint lower, it makes it narrower, and a narrow number is the first one a financial instrument can attach to. That is the sentence that connects SINE’s primary-data instrument to a financial-market outcome without hand-waving.⟩

Two — the margin is at the ends, the data is in the middle. Retail €1.39/kg, roaster €0.89/kg, farmer €0.41/kg, exporter €0.29/kg, trader €0.26/kg. ⟨Inference: the actor with the richest data has the thinnest margin, and the actors with margin have the least data. Every business-model proposition in this chain is therefore a trade across stages, not an improvement within one — which is exactly the condition under which a data-collaboration institution rather than a product is the right vehicle. It is also the condition under which value capture is a bargaining fight, not a coordination problem (→ coordination vs. bargaining failure).⟩

Timing. EUDR was due 30 Dec 2024, delayed to 30 Dec 2025, then in December 2025 delayed again to 30 Dec 2026 for medium and large operators and 30 Jun 2027 for micro and small. Obligations on the operator placing coffee on the EU market first: a due diligence statement with plot-level geolocation (point below 4 ha, polygon at or above 4 ha), proof of no deforestation after 31 Dec 2020, supplier and HS-code data, a documented risk assessment concluding negligible risk, and a signed compliance statement. sources/research/2026-08-18-coffee-value-chain-data.md ⟨Inference: the GIZ tender SINE is bidding for runs 15.10.2026–15.10.2027 — across the compliance date for medium and large operators and up to the threshold for small ones. Whatever it produces lands in the exact window when the obligation becomes real. Also worth noting soberly: the regulation has now been postponed twice, so any strategy whose business case rests on the EUDR date should carry a third-delay scenario.⟩

7. Contradictions and unverified claims, held open rather than resolved

  1. WHH and warrantage. Tell reports a conversation with Mogge about “an approach he brought from Niger to Burkina.” No source found attributes warrantage implementation to WHH, in Niger, Burkina Faso or anywhere. All located sourcing credits FAO (Niger 1999, extended regionally), Oxfam and the World Bank; IFAD co-funded a nine-country study. A French-language video on “warrantage équitable au Niger et au Burkina Faso” exists but its producer could not be confirmed as WHH. No WHH VSLA, cooperative-credit or microinsurance programme was found either. sources/research/2026-08-18-welthungerhilfe.md ⟨Inference: three readings stay open — Mogge described an approach WHH participates in without being its author; WHH’s involvement is real and simply unpublished; or the “Niger to Burkina” transfer is Mogge’s personal professional history rather than a WHH programme. The distinction matters because it determines whether WHH is a warrantage partner or a warrantage audience. This resolves only from Mogge’s own account — hence the standing request for those conversation notes.⟩
  2. GIZ succession. Tell states Hoven is “probably becoming the next Vorstandsvorsitzender” of GIZ. Thorsten Schäfer-Gümbel has chaired the GIZ Management Board since November 2022 and no public reporting of any succession was found. Recorded as Tell’s non-public information, not as a sourced fact. sources/research/2026-08-18-giz-leadership-private-finance.md
  3. Farmer share of retail price. 2.5% to 10% across sources — genuine methodological disagreement (different countries, direct-trade vs. commodity chains). Not resolved.
  4. GCP seat. Geneva (registered) vs. Bonn (secretariat); the statutes PDF could not be read.
  5. Roaster and trader market shares. The Nestlé “>20%” and JDE Peet’s + Keurig “19% combined” figures come from an aggregated search summary, not audited market research. Trader shares not found at all.
  6. NKG revenue. A third-party estimate of $422.7m is inconsistent with 60+ business units and a $1.1bn syndicated credit facility (Nov 2025, 18 banks). Treat as unreliable.

8. Register of holes

Holes SINE can close by research:

  • The full list of GCP’s 10 country platforms (7 named), its named board members, its fee schedule and budget, and above all the substance of the Coffee Data Standard’s governance page — the single highest-value follow-up on this page.
  • Whether GCP has any EUDR position at all.
  • Warrantage loan-to-value ratios and aggregate default rates.
  • Land tenure and collateralisability for coffee smallholders in Uganda, Brazil and Colombia — Uganda especially, as a GIZ tender pilot country.
  • What PRI Spring, Nature Action 100, IPDD and FAIRR actually ask for, and what they say is missing.
  • Data practice in coffee’s waste and circularity stage — nothing found.
  • IDH as an institution (funder of GCP, manager of Farmfit) — no wiki page.
  • Sucafina’s ownership; the current commercial scale of Enveritas and Bext360.

Holes only Tell can close:

  • The FAO warrantage document. The URL arrives truncated on my side both times. Backticks or just the 36-character UUID will do. Three candidate bitstreams are already identified: 68dfb1ea-126e-46f7-a9e0-c041cb468a86 (“Manuel de formation sur le warrantage au Niger”), b819a8ec-f244-49e8-bd17-ef05d5418841 (“Fiche d’expérience Niger: Le warrantage”), d2e51088-ae67-4fcb-9c19-18bc032c5a96 (untitled, 7.1M).
  • The Mogge conversation — notes or transcript. §7.1 does not resolve without it.
  • The finance-side names. This map has a populated development and industry layer and an empty finance layer: not one named individual at a bank, insurer, asset owner or finance association. That is the layer the whole strategy is about, and it is the layer only Tell’s relationships can populate. Slack attachments do not reach me — text only.

Holes that are structural, i.e. someone’s job to fix and nobody’s yet:

  • Nobody in the chain can verify the co-op’s claim to vouch for all its members (§2.2).
  • The wet mill determines whether any downstream claim can be substantiated and is invisible to finance (§2.3).
  • No instrument anywhere prices a roaster’s capital against verified farm data (§2.6).

Cross-references

Graph View

Backlinks