TandaPay and Peer-to-Peer Insurance

|

Five years ago, I wrote an article entitled "Cryptoinsurance," describing the possibility of a peer-to-peer insurance system implemented through cryptocurrency. I have recently found out about a company called TandaPay, which is seeking to implement a version of decentralized cryptoinsurance and was awarded its first patent last month.

TandaPay's vision of how cryptoinsurance might work is fairly simple (but is explained in more detail by founder Joshua Davis here). Individuals may choose to join a mutual insurance group, containing others they know and trust, by monthly contributing funds in the form of cryptocurrency. If a member of the group makes a claim, a designated secretary determines whether it is valid, and members of the group are then supposed to allow their funds to be released to the claimant. However, any member of the group may choose to refuse to release their funds, in which case that member is banned from further participation in the group. If claims fall short of contributions in any month, then the difference is refunded. If, on the other hand, claims exceed contributions, then the contributions are divided to claimants in proportion to approved claims.

One interesting aspect of this arrangement is the principle that all premiums are distributed to policyholders every period, either in the form of claims or as refunds when there is money left over. Nothing in this principle is dependent on the crypto-aspects of the broader proposal, and indeed Davis of has argued (see here, half way down) that full distribution of premiums might help overcome some difficulties in existing insurance markets. If there are a sufficient number of policyholders, one could even imagine a contract in which the insurer is required to spend all premiums on claims each policy period. Insurers have to make money, of course, so they would charge an administrative fee separate from the premiums.

In existing insurance markets, the primary function of the legal system is to prevent insurance companies from taking advantage of policyholders. It is widely understood that insurance companies will pay as little as they can get away with, which often means underpayment but sometimes means that they make payments on claims that the parties would have excluded had it been possible to contract ex ante concerning the precise scenario. In a regime in which insurers must pay out all premiums each period as claims, the insurer and the insured are no longer in an inherently adversarial relationship with insureds. An insured would look for an insurer who has a reputation for fairly judging claims and who charges relatively low administrative fees. In such a regime, the appropriate role of the legal system ought to be quite limited, focusing solely on whether the insurer in fact has distributed the premiums, whether any bribery has occurred, and whether the insurer has made proper disclosures, clearly identifying the portion of the payment that will be kept by the insurer as an administrative fee.

This model thus provides the insured with a different form of assurance than the conventional insurance model. In the conventional insurance model, the insured knows that the contract promises certain payments in the event of a loss and that the insured can sue if the insurer fails to pay. In this alternative version of the mutual model, the insured knows that the insurance company's incentive is to assess claims fairly, but it may be hard to predict precisely how much will be paid on a claim that is clearly covered by the insurance contract. The most significant advantage of the model is that it may have lower transactions costs, both because underwriting is unnecessary (only filed claims need to be examined) and because of the lack of need for enforcement in the courts, at least if the legal system respects a contractual term giving the insurance company final say over how to distribute premiums to claimants. The case for such respect is that the insurer is serving a role as arbitrator among competing insurance claims.

The required-payout model could work reasonably in a context like automobile insurance in which claims are fairly predictable (even if there is a small risk of a very bad month in which claimants will receive less than they would have expected). It could, however, also be adapted to a context in which there might be highly correlated losses. For example, homeowners in a particular geographic region might purchase a policy where the insurer promises to use the premiums to buy catastrophe bonds relating to that region. In a period in which a catastrophe such as a hurricane occurs, the insurer's job would be to collect on the catastrophe bonds and then distribute the money to insureds. An insurer would give more money to an insured suffering a larger loss, taking into account how much the insured contributed and how large the risk was that the insured faced ex ante. For example, if two insureds had contributed the same premium, an insured with a house on stilts would receive more than an insured whose house was sure to be flooded in a rainstorm. An insurer with a reputation for not making such adjustments would end up with an insurance pool of only bad risks.

TandaPay does not work quite like this—it relies on individual group members to police insurance payments. That approach, however, reinforces that insurance requires trust, but the trust can come in many existing forms. With conventional insurance, insureds must trust that the legal system will require insurers to meet their obligations. In the mutual approach that I have described, insureds recognize that insurers have incentives to compete on price and on fairness, thus trusting market incentives. And in TandaPay, insureds must trust the fellow insureds in their group, taking into account that an insured who defects and refuses to pay may suffer in the future, both reputationally and by being barred from further participation in the insurance scheme.

NEXT: ACLU's Argument Against Puerto Rico's "Fake News" Ban

Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Report abuses.

  1. Interesting concept. As you say, it depends on trust and reputation. Makes me wonder exactly why we trust current insurance companies; I know in my case it is not because government keeps them in check — I believe government intervention in insurance markets only makes them less efficient and more expensive. Maybe I trust them because (a) I don’t have much choice, I have to choose one or another,and (b) they’ve been in business a long time.

  2. If you crash your car, or there is damage to your house, do you really want your friends to have to pay for it?

    1. You can always choose to not make an insurance claim, same as now.

    2. How do you account for the fact that 1/3 of all GoFundMe campaigns are medical fundraisers? Are breadfund groups just aberrations or are they valid forms of insurance?
      https://www.shareable.net/dutch-collective-broodfonds-provides-gift-based-health-insurance-for-freelancers/

      1. You get insurance because you don’t want to have to rely on GoFundMe campaigns for medical expenses. And seriously – “breadfunds”? You are talking about really niche cases.

        But you unintentionally bring up a key point why TandaPay will not succeed: complexity. People go to GoFundMe because its easy to setup and use. The same goes for buying Auto insurance in the United States. It took me twenty minutes a few hours ago. And now I don’t think about it unless I actually get into an accident or renew in six months.

        You are trying to solve a problem no one is having. And doing it by using complex and time consuming technology.

      2. Interesting concept, but doesn’t the required-payout-every-period limit the amount of value that can be insured?

        Right now my annual homeowners premium is about the same as our auto insurance. I have had a number of medium-sized auto insurance claims over the years, in contrast to literally zero on my homeowner’s policy. But if my house burned to the ground, I would have an orders-of-magnitude larger claim than any of my auto claims. I don’t see how a required disbursement regime would allow the insurance scheme to build up enough reserves for a rarely occurring, expensive claim.

  3. Individuals may choose to join a mutual insurance group, containing others they know and trust, by monthly contributing funds in the form of cryptocurrency. If a member of the group makes a claim, a designated secretary determines whether it is valid, and members of the group are then supposed to allow their funds to be released to the claimant.

    Are we really talking about an insurance arrangement with only a handful of participants? How many people do you know and trust enough to join such a pool with?

    The most significant advantage of the model is that it may have lower transactions costs, both because underwriting is unnecessary

    I don’t see how underwriting is unnecessary, except in the most literal sense that you don’t have to do it, if you don’t mind your scheme falling apart.

    and because of the lack of need for enforcement in the courts, at least if the legal system respects a contractual term giving the insurance company final say over how to distribute premiums to claimants. The case for such respect is that the insurer is serving a role as arbitrator among competing insurance claims.

    And why would a customer agree to such a provision? We could say that normal insurance wouldn’t need enforcement in the courts if the policies contained such a provision. Wonder why they don’t.

    And notice too that conventional insurance companies also have incentives to develop a reputation for fairness in handling claims. Sure, they can be stingy, or worse, but in the mutual model there is also an incentive not to pay a lot, so that more money is refunded to members.

    Also, this business of paying out all the money every month just looks silly. Why not have a reserve, so if one month has no claims and the next one has a lot you can still meet them all. Why should whether I get my claim paid depend on how many other claims come in during the same month?

    1. “Also, this business of paying out all the money every month just looks silly. Why not have a reserve, so if one month has no claims and the next one has a lot you can still meet them all. Why should whether I get my claim paid depend on how many other claims come in during the same month?”

      Because that would violate the law and we don’t want to create groups that violate the law:
      https://medium.com/predict/the-law-and-smart-contracts-9611e9d7e783

      1. Thanks for responding, Joshua, but IANAL, so I don’t quite get the violation here – not that I question your statement.

        But if there is no reserve then the scheme doesn’t seem very sensible to me, especially since you are apparently contemplating that the groups will be small, so you can expect a fair amount of variance in total claims.

        Indeed, Michael writes that,

        In existing insurance markets, the primary function of the legal system is to prevent insurance companies from taking advantage of policyholders.

        One way the system does that is by requiring insurers to maintain some level of reserves, to assure that claims can be paid.

  4. Another problem seems to be that the member is not getting the insurance paid for.

    Say my expected losses are $50/month, so that, plus some amount for administration, is my monthly premium.

    But of course I don’t have a $50 claim every month. Instead, I have a $1200 claim once every two years. And if I’m mildly unlucky there are enough other claims that month that I don’t get my $1200. So I’m not getting the insurance I’m paying for.

    It’s possible I don’t understand the scheme, but it looks like a bad deal to me.

    1. Concur. It is a bad deal.

      The P2P insurance relies on trust. I am more of the ‘trust but verify’ kind of guy whenever it comes to money.

      P2P lending is a little different, but not that much different.

      1. One important difference is that with lending you know what your risk is – the amount you lend.

        With insurance your risk is the amount of your possible claim, which you don’t know before the damage occurs.

  5. which is seeking to implement a version of decentralized cryptoinsurance and was awarded its first patent last month.

    There were 391,103 patents issued in the US in 2018. Hell – my husband has a patent. Who cares?

    TandaPay’s vision of how cryptoinsurance might work is fairly simple

    No one will look at the articles on medium (which should be your first red flag) and think TandaPay is “simple”. And it employs blockchain (second red flag) which adds complexity and is not an efficient system for this use case (or really any as companies are finding out)

  6. Wait, so this is like a Lloyd’s of London syndicate–which has been going for literally hundreds of years–or a mutual insurer, but there is no reserve to pay claims and no guarantee that the individuals who are “names” in the scheme will pay? Each participant is individually assessed upon a claim being made, and they have the choice to participate or leave the scheme? And if they refuse to pay for any reason at all, I have no recourse as the claimant?

    So if my house burns down and it will cost $350K to replace–a valid claim that would result in a $10k assessment to each scheme member–each scheme member (who was paying, say, $50/month), could just choose not to pay my claim and the only punishment is being denied the opportunity to participate in the scheme in the future? Or did they pre-agree to pay up to $X up-front in case of a claim?

    Is the only punishment being excluded from future TandaPay deals like this? Why wouldn’t scheme members just default on any large claim and start a new scheme?

    While I’m sure it’s legally interesting, as insurance it just seems like a worse deal for consumers. And, as described, it is deliberately structured to avoid laws that exist for a reason–to protect policyholders. One may hate insurance companies, but the US has some of the strongest insurance regulation in the world to ensure that insurers don’t go bust leaving the insureds unprotected.

    As described, the advantage is: “the insured knows that the insurance company’s incentive is to assess claims fairly, but it may be hard to predict precisely how much will be paid on a claim that is clearly covered by the insurance contract.” I don’t buy insurance because I want to roll the dice on the amount of coverage I have when I make a claim. A fairly assessed claim on which I am paid $0 (or a tiny fraction of my valid claim amount) is not a very useful service. An insurance company very well might deny a valid claim. But here it seems I have no recourse at all.

    1. All true.

      And your example is precisely the kind of thing we want to insure against – a low probability high cost event.

    2. My understanding from reading around the web is that a tanda is designed for relatively small expenses that would still be difficult for people with little-to-no savings. Think hundreds of dollars, not hundreds of thousands.
      If you look at the examples given in the article links, you’ll see they are often expenses less than $500.

      At those levels, in a community with high trust AND strong extra-legal enforcement mechanisms (social, not hitmen), it might work for a while.

      But as a replacement for traditional insurance? I agree completely; this is an absurdity.

  7. Yes, the whole point of insurance is it converts a stream of unpredictable rare big losses into a stream of small losses which are bigger in the aggregate, but predictable. And it only works because the insurer gets such a large pool together that even the rare-to-the-individual losses are predictable across the pool.

    In this scheme, unless you have a truly massive number of trusted friends, the rare big events are going to cluster unpredictably, leading to unreliable delivery of the only value the scheme is intended to deliver.

    And unless you also do something like reinsurance to address *nonrandom* correlated claims, the scheme also fails to deliver when a big-loss event impacts a significant chunk of group members because of something they have in common like a profession, or geographic location. Which trusted-friends groups tend to have.

Please to post comments