The Coffee House Theorem
In 1688, Edward Lloyd opened a coffee house on Tower Street in the City of London. It was not remarkable for its coffee. It was remarkable for what happened inside it. Ship captains brought news of voyages. Merchants brought cargo manifests. Underwriters brought capital and judgement. Within a generation, Lloyd’s Coffee House had become the most efficient insurance market in Europe — not because anyone designed it to be, but because the room solved a problem that the Nobel laureate Oliver Williamson would not formalise for another three centuries.
That problem was transaction costs. Williamson, building on Ronald Coase’s foundational 1937 insight that firms exist because markets are expensive to use, argued that the critical variable determining how economic activity is organised is asset specificity — the degree to which an investment is tailored to a particular transaction. When asset specificity is low, spot markets work. When it is high, the parties become locked into a bilateral dependency that invites opportunism, renegotiation and hold-up. The rational response is vertical integration, or at least a relational governance structure that substitutes trust and reputation for the anonymity of the market.
Lloyd’s Coffee House was exactly such a structure. Maritime insurance in the seventeenth century was characterised by extreme asset specificity. Each voyage was unique. Each hull, each cargo, each route carried risks that could not be standardised. The underwriter who wrote a policy on a ship bound for the Spice Islands was making a relationship-specific investment — his assessment of the captain, the vessel and the season could not be transferred to another transaction without loss. The coffee house reduced the transaction costs of this inherently bespoke market by concentrating information, reputation and counterparties in a single room.
Three centuries later, the reinsurance market — the market where insurers themselves buy protection against catastrophic loss — retains much of this relational architecture. Reinsurance treaties are negotiated bilaterally, often renewed annually over decades of continuous relationship. The cedent (the insurer buying protection) invests in explaining its book of business; the reinsurer invests in understanding it. Claims on long-tail lines — asbestos, environmental liability, directors’ and officers’ coverage — can take twenty or thirty years to settle. The relationship-specific capital embedded in a mature reinsurance treaty is enormous, and it is precisely this asset specificity that keeps the market concentrated.
A Herfindahl-Hirschman Index calculation across the global reinsurance market reveals a moderately concentrated oligopoly. Munich Re, Swiss Re, Hannover Re, SCOR and Berkshire Hathaway’s reinsurance operations collectively command a substantial share of global premiums. This is not an accident of history. It is a structural consequence of Williamson’s logic: when asset specificity is high, the market gravitates toward fewer, larger counterparties capable of bearing the relationship-specific investments that the business demands.
Enter tokenisation. In June 2026, Oxbridge Re’s subsidiary SurancePlus launched three tokenised reinsurance securities on Solana — Series A, B and C — offering qualified investors contractual exposure to excess-of-loss reinsurance risk, with targeted annualised returns ranging from nineteen to two hundred and forty-three per cent assuming no underwriting losses. The three securities brought SurancePlus’s total to five tokenised offerings for the year. This week, Oxbridge attends the Rendez-Vous de Septembre in Monte Carlo, the annual gathering where the global reinsurance market sets its terms for the year ahead. That a blockchain-native firm now walks the same corridors as Munich Re and Swiss Re is not a curiosity. It is a test of Williamson’s theorem.
The proposition is seductive. By representing the economic exposures embodied in reinsurance contracts as digital tokens on a distributed ledger, proponents argue that the market can achieve what Lloyd’s Coffee House achieved, but at global scale and near-zero marginal cost. Fractionalisation would allow institutional and retail investors to take precisely calibrated slices of catastrophe risk. Smart contracts would automate claims triggers, reducing settlement times from months to seconds. Liquidity, that perennial scarcity in reinsurance, would emerge from the depth and diversity of the token market.
The catastrophe bond market has already moved partway in this direction. Outstanding issuance reached a record of nearly sixty-six billion dollars by mid-2026, with first-half volume of nearly eighteen billion dollars across eighty-three transactions. Cat bonds have demonstrated that capital markets can absorb peak perils — Florida hurricane, California earthquake, Japanese typhoon — without the relational infrastructure of traditional reinsurance. Parametric triggers, which pay on the occurrence of a defined physical event rather than on demonstrated loss, sidestep the most asset-specific element of the reinsurance relationship: claims adjustment. If the wind speed exceeds a threshold, the bond pays. No adjuster, no negotiation, no relationship required.
But Le Chatelier’s principle warns that when a constraint is relaxed in one part of a system, the stress does not vanish — it migrates. The bespoke underwriting judgement that a reinsurer brings to a complex casualty treaty cannot be encoded in a smart contract without losing the very thing that makes it valuable. A token can represent the cash flows of such a contract. It cannot represent the governance.
Real options theory sharpens the point. Tokenisation creates optionality for the cedent: the ability to restructure a risk portfolio dynamically, to exit positions that a traditional treaty would have locked in for a renewal cycle, to access capital from counterparties who would never have appeared at a brokers’ market dinner. But optionality is not free. The premium is paid by the reinsurer, who loses the relationship rents that compensated for the asset-specific investments in understanding the cedent’s book. If those rents disappear, the incentive to make the investment disappears with them — and with it, the deep underwriting expertise that distinguishes reinsurance from mere risk warehousing.
The fat-tail problem compounds the difficulty. Reinsurance exists precisely because the distribution of catastrophic loss is not Gaussian. It is heavy-tailed, spatially correlated and subject to regime change. Expected shortfall — the average loss in the worst percentile of outcomes — is the metric that matters, not variance. The question for tokenised markets is whether they can handle correlation in the tail. Lloyd’s own history provides the most instructive, and most thematically apt, answer. In the late 1980s, the London Market Excess of Loss spiral — a chain of inter-syndicate retrocession in which syndicates reinsured each other’s catastrophe exposures in layer upon layer of excess-of-loss contracts — created what participants believed was a sophisticated mechanism for spreading risk widely across the market.
When Piper Alpha, Hurricane Hugo and a succession of catastrophes struck between 1988 and 1992, losses did not disperse. They amplified. Each claim spiralled through the retrocession chain, multiplying at every turn. Lloyd’s lost billions of pounds — estimates range from eight to fourteen billion, depending on the accounting boundary — and came within sight of institutional collapse. The Names — individual investors trading with unlimited personal liability — were ruined not by the catastrophes themselves but by a market structure that had manufactured the illusion of diversification while concentrating correlation. Tokenisation, which promises to fractionalise and distribute reinsurance risk across a broader investor base, must answer the question that the LMX spiral posed and the coffee house had already solved: does the governance structure reveal correlation or conceal it?
Put-call parity offers a final lens. Purchasing reinsurance protection is economically equivalent to buying a put option on catastrophic loss — the right, but not the obligation, to transfer the downside beyond a retention level. Tokenisation does not alter this fundamental symmetry. It changes the mechanism of exchange, not the structure of the payoff. But Williamson’s central insight was that mechanisms matter — that the governance structure through which a transaction is conducted is not a neutral conduit but shapes the incentives, the information flows, and ultimately the risks themselves. A put option written in a coffee house, where the underwriter has looked the broker in the eye and interrogated the slip, is not the same instrument as a put option minted on a blockchain, where the counterparty is an algorithm and the slip is a smart contract.
None of this is an argument against tokenisation in reinsurance. It is an argument for precision about what tokenisation can and cannot do. It can reduce frictional costs. It can broaden participation. It can accelerate settlement. What it cannot do — what no technology has yet done — is eliminate asset specificity. The tailored judgement, the relational trust, the long-horizon commitment that distinguish reinsurance from commodity risk transfer are not transaction costs to be engineered away. They are the product itself.
Edward Lloyd would have understood. His coffee house did not succeed by eliminating the need for judgement. It succeeded by creating the conditions in which judgement could be exercised efficiently. The room was the technology. Three hundred and thirty-eight years later, the question is whether a distributed ledger can be a room — or whether some markets still need walls, and a bell, and the sound of a name being called across a crowded floor.