June 25, 2021
Before Lloyd’s of London was hatched in a Tower Street coffee house, exporters would transport goods via a flotilla of ships to prevent a total loss of cargo in the case of a weather disaster. Lloyd’s began underwriting insurance for such events, obviating the need for businesses to spend resources diversifying their voyage risk. Lloyd’s core businesses grew predictably until the 1906 San Francisco earthquake, when it paid out claims far in excess of the premiums it had collected from policyholders. Reinsurance, which insured insurers against catastrophic losses incurred during rare events, began to truly emerge. The risk that any one insurance company would be rendered insolvent by a cascade of claims was further ameliorated by the growth of insurance-linked securities, or securitization. Underwriters could now package their insurance contracts, which entailed a stream of cash flows (premiums) from policyholders, and transfer them from their balance sheet to investors. Insuretech start-ups came along after the financial crisis to take another crack at the insurance code. Companies like Lemonade and Oscar Health incorporated algorithms to automate some functions and built out pleasant user interfaces for policyholders. As the decades pass, upstarts make incremental improvements to the way that companies, investors, and individuals hedge uncertainty. The perennial promise to disrupt the insurance industry has evolved into a trope, one that may hold some water. To better understand how decentralized protocols might empower more robust leaps in innovation, we need to understand the constraints on the incumbent insurance providers and their customers.
On one side of the insurance equation is a coefficient representing the insurable event and a variable representing the probability of that event occurring. Sometimes the probability is obvious and easy to plug into the risk model. Other times the data is sparse and the probability less confidently assigned. In the clearer case, insurance markets are supplied by many participants. In the more opaque, there exists either an insurance oligopoly (and oligopoly pricing) or a complete dearth in supply. So, the first constraint on providing insurance for a given risk category is obtaining sufficient data to get comfortable.
Granular data will illustrate a clearer scope of the risk being considered for coverage, but can be difficult and expensive to acquire. Once the insurer has data-driven clarity, it can price risk efficiently. The risk an underwriter assigns to an event is the assumption that the entire equation hinges on. The less confident an underwriter is in assigning this probability, the more expensive the policy will be, all else equal (or coverage will simply not be offered). By pricing premiums appropriately, an insurance provider will simultaneously attract prospective policyholders and collect premiums in excess of the claims paid out.
A catastrophic event, on the other hand, can push insurers to insolvency if the pool of risk is too correlated. The second constraint on providing insurance for a given risk category is tied to the first: without attracting a broad set of policyholders, risk cannot be sufficiently diversified to absorb a tail event that triggers the payout of a large claim.
Furthermore, insurers will underwrite more broadly if they can offload some risk to other insurers and/or investors. The third constraint on providing insurance for a given risk category is the extent to which underwriters can syndicate risk among third parties. If the claims risk associated with insurance contracts can be divvied up and made accessible to investors/other institutions, the additional liquidity will trickle down to a lower cost of capital (premiums) for policyholders. More policies can be written for more policyholders conducting different lines of business.
The last constraint we will consider has to do with the actual claims process. Once those involved in a contract reach this stage, an undesirable event has occurred and the counter-parties’ incentives are misaligned, mostly. To the extent that an insurance underwriter’s reputation and thus future earning power hinges on trust, it will strongly consider paying out claims if the conditions of the contract’s payout are met. However, bloated claims departments suggest that insurers seek to minimize payouts when claims are submitted. This is clearly not optimal for a policyholder who has incurred a casualty and needs cash more than ever to continue operating. The final constraint we consider is the misalignment between policyholders and claims departments. The need for and general lack of trust between these counter-parties paralyzes the potential demand for and thus supply of coverage policies.
To summarize, the constraints on providing insurance for a given risk type include:
Collecting enough quality data
Attracting a diversified set of policyholders
Syndicating risk adequately across underwriters/investors
Generating demand from policyholders despite general untrustworthiness of claims process
Clearly these constraints are not isolated, but reflexive, with each feeding into the others.
Now that we have established a taxonomy of the constraints on providing insurance, let’s dive into why crypto can enable anything different or better for existing corporations keen on innovating. To me, embedding insurance contracts into crypto-economic protocols can help hurdle each of the four aforementioned constraints.
Vast oracle networks can be incentivized to feed insurance contracts with a breadth and quantity of data that was previously infeasible.
Crypto accounts and wallets are universally accessible to anyone with a smartphone. Assuming oracle networks exist locally, individuals and businesses can access insurance contracts underwritten by anyone globally.
By representing insurance contracts via non-fungible tokens on a blockchain, insurance companies can access international liquidity and offset some of their risk.
Oracle networks and accessible banking will catalyze more insurance to be underwritten for more prospective policyholders. Parametric smart contracts will execute without human input, mitigating the misalignment between claims departments and policyholders. The reliable execution of claims will attract demand for insurance at a scale yet to be seen.
Let’s be a fly on the wall in a board meeting for a blue-chip insurer, say AIG. AIG’s C-level suite recognizes that the price of insurance premiums have outpaced most other goods and services and life has been good. Covid screwed up some models and the company took a hit on its 10-K but business is resuming as predicted. As an insurance company does, AIG seeks to increase returns in the new year. It will explore ways to broaden its insurance offerings and price existing offerings more tightly. The Chief Risk Officer explains a report published by an investment bank on Nexus Mutual, a decentralized insurance protocol built on Ethereum. The protocol currently underwrites coverage for smart contract failure, paying out for faulty code, governance mismanagement, and economic design. Although the protocol does not yet offer traditional insurance vehicles, the transition is inevitable, she says. She goes on to illustrate how protocols like Arbol and Etherisc offer parametric crop insurance, flight delay insurance, supply chain insurance, life insurance, and disaster insurance.
The CEO, simultaneously worried and intrigued, asks why AIG would want to fiddle with systems that relinquish the company’s control. The CRO explains:
Streamlining Expenses
First off, the company allocates significant resources to its actuarial, claims, compliance, and legal departments. The bloat in these departments has grown relatively slowly compared to the historic revenue booked through premiums but Covid has thrown the balance. Either integrating with and underwriting on an existing decentralized protocol or building out its own app could significantly cut expenses.
Actuarial data is expensive and difficult to collect. Insurers must build out internal departments of a size commensurate with the number of insurance risks they seek to provide coverage for. Existing incentives stymie exchange of data among insurers (See: Blockchain Commercial Solutions Pt.1 - Oracles). Once collected, the risk data becomes stale quickly, requiring companies to spend ongoing dollars to be comfortable rolling and issuing new contracts at an efficient price.
Many decentralized (insurance and other) protocols leverage decentralized oracle networks for the collection of and refinement of data. These data sources can be individuals or companies and are incentivized to provide accurate data via their APIs. A corporation can customize the oracle network it uses and choose one that suits its purposes best. For instance, if AIG wanted to provide supply chain delay insurance on routes through the Suez Canal, it could elect for a localized oracle network that had a long-standing reputation of reporting accurate data in Northeastern Africa. To contrast, for provision of life insurance in Europe, AIG could look for a broader set of oracles who report on death certificates across the Eurozone. To further enhance the data quality, AIG can filter oracles based on the amount of value they have bonded, or at risk. The higher the bond, and thus the higher the value at risk for the nodes, the more likely that data is to be accurate. The point here is that AIG can more efficiently collect broader data from external providers who are even more incentivized to provide good data for risk equations than AIG’s internal actuarial and risk departments. This has two benefits to AIG 1. Cut payroll expenses 2. Expand to new insurance offerings and more efficiently price existing offerings
Bloat across the compliance and legal department is a familiar narrative across finance. Bloat across claims departments is a familiar one across insurers. The CRO shows that by leveraging oracle networks to achieve high-fidelity data the company can begin offering parametric insurance contracts. These “hybrid smart contracts” ingest oracle data, compare it to a codified threshold, and release value accordingly, automatically. For those insurance lines that are amenable to these smart contracts, the concept of claims and the associated departments disappear. For the majority of insurance types, namely those that are unstandardized and nuanced, fully parametric contracts are not possible. But finding a middle ground between a fully automated execution that obviates the need for claims processing and a slow and adversarial execution with claims management is possible. Nexus Mutual underwrites contracts and effectively sells off both the claims risk and governance to syndicates, or investors via issuance of a token. In this capacity, it can focus on its core business, structuring contracts, without absorbing the full claims risk or expenses associated with it. By offloading claims management, payroll and potential legal liabilities associated with it shrink. By shrinking payroll and circumventing legal complication, compliance and legal expenses shrink. The savings here can be sluiced into other productive projects. The CEO begins to see that by piggybacking on information-age infrastructure and relinquishing control of certain stages of the insurance process, AIG might in fact become a more profitable and robust business.
Demand - Expanding Access to Policyholders and Investors
Structured products (Credit Default Swaps, Life Insurance Pools etc.) and the syndication of claims risk enable AIG to do business without incurring the entire risk of that business. Structured products and syndicates are often only accessible by institutional, sophisticated investors. The shallow investor/syndicate base is a testament to lack of education and access for non-institutional investors. A US retail investor cannot get exposure to insurance assets via their brokerage accounts. Nor can he visit the local AIG branch and ask for an allocation to a tranche of credit default swaps on junk bonds. He might be able to access a life insurance annuity. If the access to insurance assets is constrained in developed nations, it is nonexistent in those where even a bank account is difficult to access.
The CRO explains that cryptocurrency wallets enable any individual with wifi and a smartphone to send, receive, and invest in digital assets. So if an insurance contract exists on a blockchain, a farmer in a developing nation can purchase coverage. Equally as important, to codify contracts on a smart contract blockchain is to access the universe of investors who have cryptocurrency wallets. Crypto-economic protocols allow corporations to access an unprecedented level of demand and liquidity for their offerings.
Asset originators could represent their digital insurance contracts as non-fungible tokens on Ethereum, a layer 1 blockchain. The NFT, representing a stream of future cash flows, could be locked in a smart contract that releases digital value to the policyholder in the case that a certain conditional threshold is met. At this point, the digital asset would be freely tradeable among investors who have an Ethereum wallet. With 160MM Ethereum wallets and counting, the implications of bridging insurance assets to blockchains are astounding. If AIG fails to dip into this vast investor pool, one of their competitors ultimately will. They’d prefer to lead cautiously than to chase their peer set.
Parametric Insurance - Epitrust
The CRO pivots to how AIG might reconsider the traditional claims model. This model has been tweaked and iterated but has remained largely the same despite the technological leaps the internet and mobile enabled. Historically a point of contention for insurers, the claims process is inherently adversarial and can be expensive. The immediate costs accrue from the overhead for the department and the legal expenses incurred upon claims litigation. The longer-term intangible costs concern reputation, trust, and the ongoing ability to generate new business and retain existing business. She explains that Covid has ushered in a new degree of mistrust on behalf of policyholders. Insurers broadly attempted to define force majeure their own way and avoid concentrated claims payouts. She believes that the role and robustness of trust in business is changing. Retail investors were shocked when Robinhood shut down single sides of certain equity markets because it did not suit their interests. Web 3.0 consumers are growing wary of rent-seeking intermediaries who inhibit self-sovereignty of assets. Customers will continue to recoil from trusting brands and lean into trustless non-custodial solutions that simply execute as codified in the contract between the counter-parties.
Insurance is no exception. Incentives in the claims process are misaligned and existing and prospective customers know this. When a process is plagued by a misalignment of incentives, the role of trust increases. It is not during the doldrums but at the edge, during crises, that these trust relationships are strained. AIG lost face during the mortgage crisis because it could not maintain enough solvency to cover its liabilities without external stimulus. During similarly catastrophic events, providential insurers might instruct their claims departments to execute claims fairly reliably. (Lloyd’s of London instructed its claims department to generously pay claims to policyholders during the 1906 San Francisco Earthquake.) While such forward-looking execution might be self-preserving over the long run, many insurers will elect for a more short-term strategy and violate trust. As consumers become increasingly sensitive to systems requiring trust, they’ll converge on solutions that, all else equal, minimize the need for trust. AIG could capitalize on this shifting tide of sentiment and build out more trustless solutions to increase market share.
If AIG were to integrate parametric insurance contracts, where claims were executed automatically and where data feeds were generated by nodes external to AIG, it would benefit in a variety of ways. First, and most simply, AIG would capture the new base of consumers and businesses concerned with unilateral exploitation of the claims process. Relinquishing execution of claims to code will resonate well with many groups keen on more trustless execution: the crypto-savvy, parties who’ve been shorted by claims processes in the past (most everyone), and entrepreneurs (farmers, merchants) who’ve simply avoided the oligopolistic insurance offerings in their geographies. Second, AIG could widen its moat by throwing a network effect flywheel into play. Insurance companies do not as obviously benefit from network effects like other Web 2.0 vendors do. The policyholders and investors do indeed benefit from the depth of the respective pools, but the benefit is more linear. To be one of the first insurers to build easily accessible and liquid contracts on a smart contract blockchain would enable AIG to capture network effects similar to those of decentralized financial applications like Maker and Uniswap. A leap in demand from policyholders and stickiness of capital should more than offset the loss of control that AIG would incur. Third, even if AIG were not keen on taking the risk of being a first-mover, parametric insurance will become the new norm. It should prefer to be proactive and seize the opportunity rather than allow a competitor to stake its claim on the frontier. Chasing is inherently defensive. Although AIG is a mature, cash-flowing business, management should recognize that smart contracts and crypto-economic protocols will usher in a paradigmatic shift in consumer expectations.
Through-Line: What is the takeaway for incumbent insurers?
New technologies always affect incumbency. Most mature businesses will rest on their laurels and continue to invest in their cash cows, either refusing to acknowledge or actively lobbying against proliferation of decentralized technologies. Traditional business units have delivered excellent returns to stakeholders, after all. So why fix it if it ain’t broke? But the innovator’s dilemma is all the more stultifying when a genuine paradigm shift is afoot.
We at Vulpine believe that the intersection of Main Street and blockchains is under-considered. Insurance is one of the purest propositions for this accretive interbreeding. Smart contracts can and will alter the state of insurance. Expenses can be minimized by offloading data collection, compliance, and thus legal. Through access to more and better data via oracle networks, insurers can underwrite more policies for broader groups of policyholders. By codifying digital insurance contracts as “hybrid smart contracts” that are freely tradeable, liquidity for these instruments eclipses historical depth, enabling more issuance. Finally, and maybe most importantly, the Overton window has shifted: consumers and businesses have been burned by trusted counter-parties enough times. A burgeoning appetite for more trustless solutions implies that returns will asymmetrically accrue to those supplying them. Insurance is a fundamental building block to capital accumulation and most of the world remains without it. This rare confluence of technologies will enable entrepreneurs who’ve been paralyzed by uncertainty to access insurance for the first time. It is up to existing insurers to cooperate with and leverage decentralized solutions to do so, lest they end up competing directly with them.