Original 2023 proposal and project · Notes and sources
Revisiting a climate-insurance proposal and the work behind a promise of protection
By Héctor González
When I return to our climate-insurance proposal, I think about the work that would have happened after a farmer enrolled. Someone would need to explain the coverage, answer questions during a difficult season, and help when a payment failed to arrive. That work would need funding even when the farmer had no new policy to buy.
In 2023, Alejandro Carrillo and I developed the Climate Resilience Platform, a proposal to improve access to agricultural insurance for smallholder farmers in Mexico. We wanted to simplify enrollment, provide training, and use weather-related indicators to trigger payments after an insured event.
We received an award, presented at Big Ideas, and also presented the proposal to the Milken Institute. I remember considerable interest from Milken. We never pursued the project into a pilot.
Looking back, I want to give the proposal a fair account of both its ambition and its unfinished work. The interest it received did not test whether we could sustain the service.
Our original documents already recognized partnership, regulatory, and risk-model challenges. The question now is how far the proposed answers went—and what remained between the design and a service a farmer could depend on.
Financing that service would require several commitments. A grant could help build the platform. Premiums, insurer resources, and reinsurance would support the insurance obligation. Payments to CRP would have to sustain the service around it. A smart contract could help execute a transfer, but someone would still have to supply the money and remain responsible for paying. Reinsurance would not remove the insurer's obligation to its policyholder. [1]
A widespread drought makes those responsibilities more demanding. Many farmers could need payments together. The insurer would need liquidity, while the platform could face more questions and service problems at the same time. The people answering those questions would still need to be paid.
Our proposal suggested a commission of 7–10 percent, with references to both premiums and underwriting profits. We left the precise arrangement to negotiation. Revisiting that wording, I see how much it left unsettled. The amount available to run the platform would depend on the fee's base, the policies sold, the work promised, and whether revenue continued during a bad year.
The subsequent financing review found Mexican agricultural distribution compensation close to the range we proposed. That gives the assumption some grounding. It does not establish that such compensation would pay for technology, enrollment, training, and continuing support together. [2]
We also used hypothetical budgets to examine how a service could run short of money despite collecting commissions. The calculations helped identify the conditions a contract would need to meet. They did not measure our costs or establish what an insurer would pay. The figures and their assumptions are in the notes. [3]
The conclusion I take from that work is clear: we did not demonstrate that the proposed revenue could sustain the service we imagined.
An insurer might still have a reason to buy it. CRP could conceivably reach farmers the insurer would otherwise miss, improve service, or replace expensive administrative work. Those possibilities would need comparison with what the insurer could do through its own staff, agents, and existing systems.
Published insurer accounts provided real expenses and financial results, but did not isolate the cost of our proposed service or the profit from additional customers in the intended population. They could not turn the illustrative budget into a demonstrated commercial case. [4][5]
I would judge a separate service contract by evidence of what the buyer would pay and what the promised service would cost.
One possible arrangement examined in this retrospective would assign routine enrollment and policy support to an insurer or insurance-fund network, seek FIRA support for eligible financial education, and look for state funding for additional access services. This was a later reconstruction, not an agreement we reached in 2023.
There are institutional routes through which parts of that arrangement could be investigated. Insurance-fund integrators can contract services, although they cannot themselves conduct insurance. FIRA includes insurance within eligible financial education, subject to its requirements and budgets. The review did not establish a recurring state purchase of CRP's proposed outreach. None of those routes amounted to a commitment to us. [6][7][8]
Who pays would also influence whom we served. A lender has reasons to support borrowers. An insurance fund serves members. A commercial purchaser has reasons to protect its suppliers. Following those relationships might make delivery easier while leaving out smallholders beyond them.
That possibility matters to how I judge the idea. If the financing worked only after we moved toward farmers who were already easier to reach, we would need to acknowledge the change. A viable business for a different population would leave the original inclusion question unanswered.
The insurance itself poses another question: how closely would the signal used for payment describe the loss a farmer experienced?
Our product notes suggested using CHIRPS to monitor soil moisture. CHIRPS estimates rainfall. Connecting that rainfall to soil conditions and crop losses requires further information and assumptions. Correcting the description is a small part of a much larger task: establishing whether a trigger offers useful protection for a particular crop and place. [9]
A farmer could enroll, pay, understand the policy, and still suffer a damaging season without the index reaching its threshold. A payment could also arrive where damage was limited. This mismatch, known as basis risk, would shape the protection the policy actually offered. Explaining it clearly would be part of the service, but clear explanation alone would not make inadequate coverage useful.
Our notes also proposed payment within 24 hours after trigger verification. The farmer's wait would begin earlier, with the weather event and the time needed for measurement, data publication, and verification. It could end later, when the money became accessible. We never ran a pilot measuring that experience.
Enrollment and training bring their own limits. An ideas42 brief based on interviews in two Mexican communities identified confusion about claims alongside affordability and product suitability. Better information could help someone understand a policy without making it affordable or appropriate. [10]
A farmer who understood the coverage and declined it would not necessarily represent a failure of education. The decision might reveal a problem with the offer. Yet explaining that offer would still take time and money, even when it produced no sale.
I would also ask what our proposed smart contracts improved compared with an ordinary administration and payment system. Data corrections, privacy, reliability, access, and complaints would still require practical answers.
The absence of a pilot leaves important limits on this retrospective. The materials reviewed do not establish validated pricing and triggers, committed insurance and reinsurance terms, a recurring buyer, or measured delivery costs. We did not demonstrate improvements in enrollment, retention, payout access, or farmer outcomes.
Those gaps do not prove that the proposal could never work. They tell me where confidence would need to come from before returning to it: a defined product and population, evidence on losses and service needs, and institutions willing to fund and carry the obligations assigned to them. The comparison would need simpler insurance arrangements and other ways of protecting farmers, including measures that reduce their exposure to loss.
That leaves me with a conditional assessment: the financing review found plausible institutional routes and relevant compensation practices, but did not demonstrate a sustainable service or dependable farmer protection.
If I returned to the project, I would begin with what a farmer could reasonably expect after enrolling, who would be responsible for providing it, and how that responsibility would remain funded. A faster payout would matter as part of that larger promise.
