Insurance on the Blockchain

I wrote this in 2016, it’s still valid …

For months, I’ve been conjuring up radical visions of a brave new decentralized world with use cases from robotic ants that explore the walls of buildings to what I call “programmable land,” to hospitals that run themselves without administrators and people voting on their cell phones. I’ve tried to think of a way to radically reshape insurance. Aside from the standard disintermediation story, I’ve been drawing a blank for months.

Until last week, when I hit upon an idea that may well change the nature of insurance within a decade. I explained it to some people in the insurance industry yesterday, and they took three pages of hand-written notes. I want to document it here so everyone can get the idea.

Insurance modules
Think of the market for insurance. At the 30,000-foot level, it’s millions and millions of contracts that specify how much a buyer pays in premiums and how much an insurance company pays under what circumstances. These contracts are often complex and sticky — they are custom-made for each customer and each situation. Try making a change to one of your insurance contracts! It can take about as much effort as getting a new one altogether.

Now think of it differently. Think of the market for insurance products as broken into small modules, like modules that can fit together, each of which fundamentally represents an investment. There’s an investor on one side; on the other side is a customer paying a premium, providing an income stream. And in the middle, there’s a contract specifying the nature of the risk and payment under various circumstances. Think of these as modules of different colors, sizes, and types. In reality, there aren’t that many different kinds — a few dozen basic types should describe most situations. A policy, then, could be a particular combination of these modules to suit the requirements of a given customer at a given time — sort of like a customer putting the balls into a sack. We could replace our collections of contracts with virtual sacks of these balls.

So here’s the idea: Create tokens that represent small units of insurance, and then let customers come along and put all the units they want into their “insurance wallet,” which represents their risk portfolio. A smallest unit, for example, could have a $10 monthly premium and pay out as much as, say, $1,000 if specified conditions are met. Now put these tokens into a marketplace where all investors and consumers can see them. This creates a liquid market with low spreads and no middlemen.

Of course, people wouldn’t choose these small units by hand — they would specify what coverage they want, and software agents can buy the tokens they need until they assemble the risk portfolio they want. An investor may be only interested in insuring something like mortgage prepayments, while a buyer may be interested in insuring his house from many possible adverse events. Many actors in the market determine the price at all times, and there are ways to signal that you’re looking for a particular module and what you’ll pay for it if someone creates it. The tokens trade at market price at all times, depending on various risk factors that can move prices up or down.

Insure-Bits
As a token gesture, I’m going to coin a new term: insure-bits. An insure-bit is a cryptographic token that represents an indivisible unit of insurance of some kind. Presumably, a consortium would taxonomize and standardize the formats for these tokens, so they can be discovered by algorithms and unambiguously represent legal contractual agreements. Taking a step further, they would actually be legal contractual agreements. Customers and investors would aggregate insure-bits on a common exchange to build portfolios that suit them.

Once we have an insure-bit infrastructure and ecosystem, we get real-time dynamic allocation. That’s right — you could automatically have the exact combination of insurance you need at each moment, and that mix of insure-bits can vary according to circumstances. Your insurance bot could increase your homeowner’s insurance as it learns about potential storms, or it could add to your travel insurance if it knows you’ve just rented a motorcycle. If the value of one of your assets goes up (as determined by an oracle — see below), that could trigger the purchase of more insure-bits. Keep in mind that dynamic allocation also involves dynamic pricing — you’ll find that the insurance you want to buy in a hurry may be significantly more expensive than before, when you didn’t need it.

The Google-ization of Insurance
In essence, this is what Google did to advertising. While advertisers were buying big media campaigns that included banner ads as part of the mix, Google broke every single search word down into its own little market and sold the words at market price. This is the atomization of advertising — by breaking everything down into the smallest possible unit, it’s easy for buyers to purchase what they want when they want it, and they can specify in fairly sophisticated ways how much they are willing to pay for various words under various conditions. Whatever insure-bits you have in your portfolio at the exact time of the fire, or the car accident, or the cancellation of your flight determines your payoff.

Similarly, the obligations these tiny tokens represent can be traded on the back end as well. They can be assembled, modeled, tested, collateralized, and securitized in real time to meet the needs of various investors. The fact that both investors and customers are constantly trading these things has no impact on the market, since a double-spend is impossible and we can determine exactly the conditions that trigger the payouts to the millisecond. Investors may have to put up large deposits to ensure payouts, or they can securitize their holdings with tokens that represent real assets. None of this is particularly different from the way things are done today, but insure-bits moves everything to a real-time basis, which will automate and improve the industry significantly.

This could happen in many other industries. Stop making custom products. Just make a number of identical products, turn them into smart tokens with all their rules attached, add the regulatory code, and let the market buy and trade them at will (think: mortgages, tax liens, bonds, etc.). Anything that is currently traded in large lumps could be traded as groups of tokens that represent the smallest unit of value in the lump. This increase in granularity requires less custom work and no middlemen, making markets more efficient.

I haven’t spent much time on it, but I think this general concept eliminates the need for most reinsurance as well. There will always be one-off cases for special insurance, but in general most of the basic insurance and re-insurance use cases can be met with combinations of standard insure-bits.

We’re Going to Need a Better Oracle
Which brings me to the oracle problem. People in the insurance business know that there is still no clean, decisive definition of a car accident. If smart contracts are going to pay off when triggered by various events and conditions in the real world, we’ll need oracle services to put that information onto the blockchain.

I won’t go into the details on oracles, because my friend Matt Liston has written a short piece on oracles that will get you started. Oracles will be hard to design, execute, and integrate with smart contracts, but they’ll be important. We’ll start with simple things and working toward more complex solutions. Insurance companies should be putting a lot of effort into oracle services — we’re going to need a lot of them, as each little atom of insurance will have its corresponding oracle to trigger payment unambiguously.

We’ll probably also need a better definition of time. If my car is in an accident, my personal data locker could immediately try to buy more insurance before the market is aware of the event. Cases like this will bring in a new, connected infrastructure that ties oracles, sensors, cameras, published information, witnesses, and the Internet of Things all together using a common timeline that helps determine what happens when. This is another set of challenges we’ll need to solve before the insure-bit ecosystem is fully functional. But it sure beats the ecosystem we have today.

Who wins and who loses in this scenario? Both people buying insurance and those providing risk capital gain. The losers are the large, branded insurance companies with cartoon mascots that like to keep the print very small and the contracts very long.

Do you see any business opportunities here? Could this be a chance for insurance companies to re-invent themselves on the side of the customer, rather than extracting rents as intermediaries?

Summary
There are a lot of moving parts to this approach, but it’s very doable. It’s probably 95 percent more efficient than today’s insurance market, and possibly several times better for everyone except the intermediaries. It applies to plenty of other industries as well. The concept of tiny insurance tokens explodes the industry, from monolithic paper contracts to agile smart contracts that are infinitely adjustable both to buyers and sellers. A crazy idea like this could upend the insurance sector sooner than the CEOs of many large insurance companies may think.