Home / Sharing Economy / Peer to Peer Insurance

Peer to Peer Insurance

 

Dion Oryzak

Dion Oryzak

 

In the previous post we discussed the rise of new peer-to-peer micro-commercial insurance risks. These may disrupt the industries they are attempting to displace (like taxis or hotels) but, apart from necessitating rewording of some policy documents, they don’t impact the insurance business model itself. Since insurers are free to accept or reject these risks depending on their confidence in being able to price or underwrite them there isn’t any existential risk to insurers. However, true peer-to-peer insurance, which we have not yet seen, could disrupt the insurance industry as forcefully as TNCs have disrupted the taxi industry. But is peer-to-peer insurance even possible?

Many startups are styling themselves as “peer-to-peer insurance”. Friendsurance, was founded in Germany in 2010. Similar models launched in the UK include Bought By Many launched in 2012 and Guevera launched in 2014. These models are more a form of insurance ‘group buying’, like Groupon in the US, or One Big Switch in Australia, than a true peer-to-peer business. In these models the risks arising from groups of ‘friends’ are transferred to insurance carriers with whom Friendsurance, Guevara or Bought By Many have partnered on favorable terms. These models differ from traditional insurance in that customers are placed into risk pools of ‘like’ customers, with ‘No Claim Discounts’ or rebates then earned at the pool level, rather than the individual policy level. The defining characteristic of peer-to-peer models like ridesharing or homesharing is that there are micropreneurs earning an income by providing a service (such as livery or accommodation) to customers. There are no micropreneurs in the Friendsurance, Guevara and Bought By Many business models. There are only passive pools of customers who are somewhat affected by the claims experience of other customers in their pool. Hence we would argue that the ‘peer-to-peer insurance’ label attributed to these companies is a misnomer.

The closest we have come to seeing a genuine large scale peer-to-peer risk transfer arrangement isn’t insurance related at all, but is actually peer-to-peer lending, pioneered by Zopa in the UK and today dominated by Lending Club in the US. Lending Club listed on the NYSE to much fanfare on 11th December 2014 with a valuation of $9 billion. The service connects ’investors’ or lenders with borrowers directly, effectively disintermediating banks. At the date of its IPO, Lending Club was licensed to lend to individuals in 45 states and accept investors in 27 states. The listing was a major milestone in the maturation of peer-to-peer lending, having previously been dominated by startups, in the same way that Facebook’s IPO in 2012 legitimized the social media business model. Lending Club’s growth has validated its inherent advantage over the “legacy infrastructure” and “incumbent inertia” of large banks. Marc Jacobs, the founder of OnDeck, a competitor to Lending Club, summed up the opportunity quite succinctly:

It sounds retro to say the Internet has arrived. But financial services are really the last massive market that is technology-based but remains rooted in systems from the 1980s and 1990s, before the Internet disrupted everything.

In many ways, peer-to-peer insurance is a natural extension of peer-to-peer lending. Let’s now speculate how a genuine peer-to-peer insurance arrangement might work, where one individual directly insures another individual (or more likely a group of individuals insures another individual) without using the traditional insurance corporation as the intermediary. In the following discussion we make very little reference to specific federal and state laws, for two reasons. Firstly, laws affecting the sharing economy are malleable and constantly in flux. Secondly, we don’t want this document to be construed in any way as legal advice. It is far more useful and readable to stick to a general discussion of the business model than to delve into such specifics as how the Gramm-Leach-Bliley Act (as currently applied) impacts privacy policy or how registration requirements of the Securities Act would impact the process of securitizing insurance-backed notes.

So in that spirit, in the middle you would have an entity (the ‘central entity’) that provides the electronic infrastructure in the form of apps, a large database, an online interface and a payment clearing house. As with other brokering models, like Uber or Airbnb, it doesn’t directly provide the service it advertises but is a facilitator of this service, matching an individual service provider with an individual service seeker. The central entity (probably) takes no risk onto its own balance sheet, but takes a fee on each transaction it facilitates. Joel Laucher (CDI) agreed with this view, stating:

The first thing we would be concerned about is who is controlling the funds? Maybe you would need a licensed administrator. Maybe the peers are just signing a pledge or a surety to offer up the funds when a participant has a loss.  Is that money really readily available? It’s all about their fiduciary responsibilities. And you’d have to have a group big enough or fund large enough to pay out a major claim and still exist after it had one loss.

On one side of this central entity are the risks to be insured. Similar to online insurance quoting today, customers would enter their details into an online interface provided by the central entity and receive a quoted premium. In a reverse auction arrangement, the individual might bid the premium they are prepared to pay, which can be accepted or rejected by individual underwriters on the other side of the transaction. It is these individual underwriters who are the micropreneurs, the insurance equivalent of an Uber driver. Although Lending Club’s pricing model involves setting interest rates for particular credit tranches in advance, its competitor Prosper, started with an auction pricing approach, where an applicant’s interest rate falls as lenders bid to invest in that loan. This, like other aspects of their business model, subsequently evolved into one more closely resembling Lending Club’s. A reverse auction pricing model would be impossible under all US states’ pricing regulation, but could be a viable model overseas.

Like Lending Club’s lenders, these underwriters would be akin to amateur or semi-professional financial derivatives traders. They would lodge capital with the central entity, like an initial margin, and then determine what risks they are prepared to take onto their personal ‘balance sheet’. Like amateur derivatives traders moving into and out of positions based on technical or fundamental indicators, they would monitor their portfolio of auto, home and other P&C risks, growing in desired market segments and running off others. Like current day employed insurance portfolio managers, these underwriters would earn premium in proportion to the risks they are exposed to and suffer claim losses accordingly. They would decide what lines of business they want to ‘dabble’ in and how best to structure their own portfolio to achieve suitable diversification. The key difference is that their personal capital is at risk.

To achieve sufficient diversification, each of these underwriters would only ever be able to take small slivers of any individual risk (like 0.01% of a home insurance risk).  Due to this small exposure an underwriter would have to any individual risk, it probably wouldn’t be possible for each underwriter to manually inspect the profile of all risks they absorb onto their balance sheet. Maintaining the privacy of the insureds could prevent this from ever happening.

The underwriters would probably address this information limitation in one of two ways: rule based acceptance; or syndicate based acceptance. Under rule based acceptance the underwriter specifies some predetermined risk acceptance rules, with acceptance/rejection then being automatic. For instance, they might specify that they will accept 0.01% of any auto risk from people with clean driving records, capped at 100,000 policies per city, say, to achieve geographic diversification.

Syndicate based underwriting would follow the Lloyds of London model, where a group of individual underwriters follow (or appoint) a lead underwriter. The lead underwriter spends more time manually inspecting each risk and then has the power to bind all the individual underwriters in that syndicate to those risks they deem acceptable. In return for the extra effort selecting and managing the portfolio, they take a larger, but pre-specified, cut of the profit from that syndicate.

Underwriters would need to be able to sell their risk portfolios to other underwriters, either to withdraw their capital, to limit their own risk exposure or for regulatory and solvency reasons. Underwriters’ positions would need to be valued as frequently as possible (at least daily), both for the purpose of determining a fair transfer price between underwriters (even if only advisory) and for determining individual solvency. With traditional financial traders, determining a P&L, solvency and hence margin requirements at any point in time is relatively easy through marking to market, however, for peer-to-peer insurance a very sophisticated, and largely automated, valuation and capital model would be necessary. Conceivably, this could use existing actuarial reserving and DFA models but be much more automated with the use of sophisticated machine learning. New techniques would inevitably need to be developed to cope with both the new business model and the extremely short time frames required, i.e. even just moving from a quarterly reserving basis to a daily one would be problematic for most actuarial reserving techniques. One could imagine requiring a very large correlation matrix, capturing every risk in the system to determine and allocate appropriate diversification benefits to each underwriter. The diversification benefit would be different for each underwriter based on their own mix of risks by geography, line of business and other factors.

Just like present day insurers, individual underwriters would also need to select the asset mix in which capital, lodged to back their liabilities, is to be invested. This need be no more complex than the process employees go through today with their 401(k) plans, allocating their fund mix between cash, domestic equities, international equities, listed property etc. In the interests of simplicity, to ensure that underwriters focus more on the liability side than playing the asset side, let’s assume that the platform offers only two options, a risk free cash account and an S&P 500 index fund, where the allocation between these two must sum to 100%. Any fluctuations in the S&P 500 fund should be marked to market in real time and reflected in the underwriter’s P&L. Similarly, asset volatility would need to form part of the capital requirements model.

Although we would expect these micropreneur underwriters to be more technically savvy than the average person, understanding the complex relationships between their underwriting decisions, asset allocation, diversification measures and their capital requirements will be challenging. How many professional underwriters today have a thorough understanding of how each decision they make impacts their carrier’s capital requirements? Communicated effectively, the capital model would convey to each underwriter what their ‘risk budget’ is and allow them to ‘allocate’ that budget accordingly. Taking risk in one area (say underwriting risk) eats into their risk budget, limiting their capacity to take risk in another area (say asset risk). Derivative traders today face a slightly simpler version of this mechanism where their margin requirements change dynamically as they open and close positions and as market prices shift.

Anyone following along with this description of what is effectively an online trading platform provided by a central entity might see similarities with the defunct Enron Online (EOL) an online energy trading platform provided by the Enron Corporation. In brief, EOL allowed commodity traders (particularly natural gas traders) to trade directly with Enron as the market maker. This utilized a one-to-many trading model, as opposed to the many-to-many model used by the NYSE for instance. This first-of-its-kind platform quickly dominated the commodity trading market with its ease of use, with the EOL platform claiming a 60% share of the world’s natural gas trading volumes. This model was riddled with problems. The FERC investigation into Enron after its collapse concluded that ‘like a casino, Enron had the “house” advantage by trading on EOL in energy markets’, that ‘Simply put, the use of EOL enabled Enron to post any price it wanted’, ‘The overall evidence supports the conclusion that trading abuses and manipulation occurred on EOL’.

There are a lot of learnings from Enron Online that should to be applied to any web based peer-to-peer insurance platform. Some of these are of the ‘What did they do right?’ variety but many more are ‘What did they do wrong?’ Some learnings include:

  • Don’t allow the exchange to trade on its own account. Uber and Airbnb don’t compete with their own partners (drivers or hosts) by operating ridesharing cars or buying properties to rent out. They act purely as a many-to-many exchange, which limits conflicts of interest. This doesn’t mean that the platform can’t participate in the risks and profits too. In fact the originate-to-distribute mortgage securitization model, where originators have ‘no skin in the game’, disincentivizes prudent risk selection (to the extent that the platform manages or influences this). The separation of writer and ultimate financial bearer of risk leads to its own conflict and in fact was one of the leading causes of the ‘07-’08 financial crisis. Some form of risk retention or risk sharing by the platform would probably be desirable. However this is structured, the key philosophy is that the peer-to-peer platform be a partner to its micropreneur underwriters, not counterparty to them.
  • Disincentivize trading and speculation. Since the purpose of the platform is to allow individuals with capital to absorb real world auto and home risks of other individuals, there shouldn’t be any need to trade or speculate. Trading should really only be necessary to manage or withdraw capital. Uber and Airbnb don’t allow individuals to buy up large blocks of ridesharing or homesharing time in the hope of reselling it later for a profit (like the business model of hotels.com for example).
  • Ensure only simple, liquid, well known asset classes are allowed. Part of Enron’s dubious accounting practices involved marking to market ‘washed’ illiquid assets to manipulate paper profit. Allowing only very simple, liquid and transparent asset classes (like a cash account and an S&P 500 index fund) for underwriters to park their capital limits the ability for any party to manipulate their financial position through trading and washing.
  • The valuation and capital requirements models should be as transparent as possible. Ideally the regulator would have full view of the inner workings of the model, but the parameters and capital requirement formula should also be transparent enough to the public for a knowledgeable individual underwriter to approximately reproduce their imposed liability valuation and capital requirement from information they know about their own portfolio.
  • Strong whistleblower protections. Whistleblowing was critical in uncovering the Enron fraud. In practice, protections for whistleblowers are often inconsistently applied. If you’re going to encourage whisteblowing (a la “If you see something, say something”), don’t send mixed messages by vilifying whistleblowers.

The rise of peer-to-peer insurance would see a reversal of industry consolidation taking place over the past couple of decades. Dave Cummings (ISO) suggested:

The trend in the industry over the past 20 years, particularly in personal lines insurance, has been to consolidate. In personal auto there are far fewer insurers in the market today than there were even 10 years back. Companies have continued to grow organically in addition to the consolidation. If peer-to-peer insurance really breaks into the market, there is potential to reversing of that trend. If this market were to grow, it could take back some of the market share the largest insurers have been able to consolidate. If so, it would be a change to the balance and competitiveness of the market.

Hurdles to Implementation

We believe there are five main obstacles to the above business model becoming a reality: technical; consumer acceptance; privacy; regulation; and industry inertia.

Technical

Current peer-to-peer arrangements are technologically quite simple (compared to an insurance operation). Airbnb, Uber and eBay are just sophisticated online bulletin boards, with payment processing and a feedback rating system to keep participants (mostly) honest when dealing with strangers.

Lending Club’s platform provides a good starting point for thinking about the peer-to-peer insurance platform. Lending Club pulls credit reports, summarizes information about prospective borrowers for investors to review and has a messaging capability to enable investors to ask borrowers specific questions about their financial position. Once loans are issued, each investor is able to track payments and defaults from borrowers in their portfolio.

The technical hurdles for true peer-to-peer insurance are much, much greater than other peer-to-peer services, even that of peer-to-peer lending. Any large scale external event, from hurricanes to terrorist attacks, needs to be reflected in claim valuations in real time. Just automatically valuing each insurance risk each moment, determining diversification benefits and capital requirements would necessitate automating reserving, catastrophe and capital models while maintaining at least as much accuracy as their currently labour intensive versions today.

From this point of view, you could almost say that automating away the entire actuarial services industry is a prerequisite for the viability of true peer-to-peer insurance. However, you’d still need actuaries to build and review the models being used and explain their workings to regulators. The fact that they would operate automatically day to day, or minute to minute, isn’t too far removed from current practice where reserving spreadsheets are automatically updated each quarter with new input data. This update cycle would just need to be shrunk from quarters to seconds. Even if large scale machine learning infrastructure that is able to accommodate processing this volume of information in such tight timeframes isn’t quite there today, it certainly will be in the near future.

Pricing without any experience to draw on presents a technical hurdle, albeit one not at all unique to the peer-to-peer business model. Dave Cummings (ISO) suggests:

Pricing without prior experience is a significant hurdle.  New carriers will need to acquire data and insurance knowledge related to the risks they plan to take on.  However, without older legacy systems holding them back, they get the opportunity to start with more sophisticated pricing models and more granular, data driven underwriting.  Additionally, they have the opportunity to embrace technology and enable them to do more with fundamental pricing, underwriting and claims handling. A significant portion of the segment invasion comes from this flexibility.

Consumer acceptance

Dave Cummings (ISO) suggested that financial stability would weigh foremost when prospective policyholders consider peer-to-peer insurance:

I would expect many people would first want to ensure that the peer-to-peer insurance has the financial backing it needs to cover policies. It’s hard to know how much that enters into people’s minds. I do wonder how many tech-savvy consumers are aware of or concerned with the financial stability of their insurer. I’m guessing that they may not place as much emphasis, so it’s something that may or may not be an issue that consumers think about. If they are comfortable with the financial stability and claims handling process, then I would expect that there would be many who would embrace this concept. It’s an attractive business model in many ways. It is something that seems to speak to some of the sentiments in the consumer base about insurance companies, and it does have a startup entrepreneurial feel to it that many consumers would look on positively as long as that basic threshold of meeting the expectation of financial and claims handling is going to be met.

Amy Gibbs (ANZIIF) further opined:

We know from the digital disruption of other industries, such as with the entertainment industry, that the underlying technology is attractive to consumers who want to take more control and circumvent systems they see as being unfair or overly costly. Once the systems have been worked out in a technical sense, such as with Friendsurance or Peercover, the conversation changes, not to whether customers will use the new technology, but which provider of the new technology to use, and then more traditional evaluation comes into play – which provider is trustworthy, works the best or simply survives or outperforms the others. While Napster might have been shut down, its closure did not protect the music industry from countless other groups providing the same technology to consumers. When it comes to insurance, the idea of avoiding traditional insurance companies with their less than positive reputation (whether fairly or unfairly earned) is going to remain attractive to consumers.

While there are definitely technical hurdles for peer-to-peer insurance to cross, I think that it will be the social and cultural ones that will prove more difficult. With many insurers hesitating to even dip a toe in the water, it will be entrepreneurs from outside the industry that pave the way technologically speaking, and these groups won’t have the wealth of knowledge – and safeguards – that the established insurance industry has.

Peer-to-peer lending and crowd sourcing technology already show that people are willing to take on the risk of trusting relatively new technology when it comes to their finances. Removing the alleged bad guy from an equation – be that big business, banks or insurers – is a powerful incentive for people and small business who want a fair go. For smaller insurance needs I think people will be very interested, particularly if it means they can afford to insure things they would not normally insure, or would deliberately underinsure for financial reasons. Equally, peer-to-peer insurance will open the door to niche insurance possibilities that consumers simply cannot get access to or afford, such as ‘Bought by Many’.

Privacy

The privacy implications are very different between using a peer-to-peer service for transport, accommodation or errands as opposed to using one for insurance. When you use Uber, Airbnb or Taskrabbit you provide your name, address, email, phone number and pay with a credit card. You are revealing about as much about yourself as you do when you buy a book off Amazon, so privacy isn’t a prime consideration. But when you buy insurance you need to reveal a raft of personal information including criminal history, credit score and even biometric information in the case of health insurance. We may accept giving this information to a large faceless corporation with no personal agenda beyond taking our money and making a profit, but when the person on the other side of the transaction is a micropreneur underwriter (or many, many micropreneurs if each takes only 0.01% of your risk), then privacy becomes much more of an issue. Although de-identified, the micropreneur reviewing your insurance application might be your neighbor, your boss, your mother-in-law or your parole officer.

The peer-to-peer lending model has already tackled this privacy issue. Individuals apply for loans on the online platform, where they input their credit score, income, financial position and intended use of the borrowed funds. The platform assigns a risk profile, which investors can review and then either lend or not based on criteria the investor chooses to screen for or against. Lenders and borrowers converse with each other to discuss financial position, but personally identifiable information is not (or should not) be shared.

Alternatively, if privacy concerns become such that amateur underwriters can’t view and analyze insured’s information at all how can they underwrite the risk?

The two broad answers, mentioned earlier, involve:

  • De-identifying and aggregating the information to allow underwriters to analyze the aggregated data and then formulate their own rule based approach to underwriting, such as accepting no one with a credit score below 600; and/or
  • Joining a syndicate and allowing a lead underwriter to manage the risk selection for you. The lead underwriter would act like underwriters today, being similarly licensed and bound by privacy requirements, so that they would have access to enough personal information to evaluate the risk of each applicant, but no more.

With appropriate limitations and licensing in place, we don’t think this privacy hurdle, even today, is a showstopper for this peer-to-peer insurance model.

Regulation

Like all new forms of peer-to-peer business models, industry-specific regulation would need to be rewritten to accommodate this new business model. It’s impossible to determine in advance how this regulation would apply, especially considering the process of writing regulations is itself the result of industry consultation, political compromise and a hearty dose of lobbying. The evolution of regulation in the face of similar business models, however, provides a good guide to how regulation of peer-to-peer insurance would evolve.

Dave Cummings (ISO) suggests that startup entrepreneurs considering entering this space shouldn’t underestimate the regulatory hurdles:

I would expect that they need to go through similar regulatory and licensing processes, which are significant. That’s going to be a challenge and far from trivial. More generally, it seems there are a few things this sharing economy has highlighted. The companies going forward based on an interesting technology or business model may be slower to recognize the impact of regulation on insurance. It’s something that they need to be aware of and they need to address early. I’d say generally regulators are supportive of new companies entering the market. [Startups] have that on their side as long as they have the right structure in place like financial stability, as well as understanding rate and market conduct regulation.

The two main groups of parties to the peer-to-peer insurance transaction are the underwriters and the insureds. Relationships with underwriters, essentially being individual investors, would most likely be regulated by the SEC, while relationships with insureds would likely be governed by each state’s existing Insurance Departments. Like lenders and borrowers in the Lending Club model, the pool of underwriters and insureds would likely span many states on both sides of the transaction. In fact the principle of geographic risk diversification would make this many-to-many relationship by state desirable even as it makes it much more complex to regulate.

The underwriters would be in a very similar position to the lenders in the Lending Club model. In fact Lending Club investors can inspect individual applications for loans, ask each prospective borrower questions about their financial position and then decide on a case by case basis which loans to invest in. Lending Club CEO Randolph Laplanche described their regulatory framework:

In our case we are selling an investment to an investor, so it’s regulated by the SEC [Securities and Exchange Commission]. The investment isn’t guaranteed. The investors can ask Lending Club for their money back and get it on the normal monetization schedule of the loan. There’s no risk of a run on Lending Club like there is risk of a run on a bank. For that reason there is not FDIC [Federal Deposit Insurance Corporation]-imposed reserve requirements.

Assuming the underwriter’s funds also would not be ‘at call’ we speculate a similar regulatory framework to that governing Lending Club’s investors would apply. One of the most comprehensive summaries of the regulatory framework for peer-to-peer lending services that we could find freely available online can be found here . Underwriters would only be able to withdraw funds once their claims backed by their funds had sufficiently run off or their liabilities were sold to another party.

We see no reason why the regulation governing the insured’s interest in peer-to-peer risk transfer be different to that governing their relationship with insurance carriers today. First and foremost, reserves sufficient to pay claims need to be held. It goes without saying that the threat of a bad review on an eBay-style feedback rating system won’t be enough to entice micropreneurs to turn over all their worldly assets in the event that their initial ‘margin’ proves insufficient.

You would need to have fairly stringent up-front capital requirements equal to, say, the 99th percentile of the expected claims distribution after an allocated diversification benefit (analogous to margin requirements in derivatives trading, say) to mitigate default risk, combined with mandatory catastrophe reinsurance. As long as these parameters were set appropriately, there is no reason the risk of default need be greater under a peer-to-peer arrangement than under a traditional insurance arrangement.

Dave Cummings (ISO) suggests:

I think there are obviously some issues that need to be addressed. Starting an insurance company in your basement is a very different thing. We need to ensure that as the company or program develops that they have the financial resources necessary, which is different from being able to develop a cool app. We need to make sure, as they grow, they’ve got the right expertise and information to make sure they are prepared to bear the risk that they are going to take on.

Amy Gibbs (ANZIIF) commented on the evolution of consumer protection legislation in peer-to-peer insurance:

It will also be hard for regulatory bodies and national law to accommodate new technology. Consumer protection under these circumstances will prove hard. It’s one thing to peer network your music downloads, but quite another when both your money and assets are at risk. That said, regulation will (eventually) have to keep up with the use of the people. Whether it will do that in time to avoid a potential financial disaster remains to be seen.

The second area for regulators interested in consumer protection to consider would be pricing. To be viable, consumers would need to, on average at least, pay less for insurance under a peer-to-peer arrangement than under traditional channels. Cost savings are a common theme in peer-to-peer business models. Just compare TNC vs taxi pricing and Airbnb vs hotel pricing. The best indicator for the cost savings that would likely arise from peer-to-peer insurance again stems from Lending Club’s experience. Their ratio of expenses to loan value is less than 2 percent compared to banks’ ratio of between 5 to 7 percent, largely due to Lending Club having more automated and streamlined processes than banks and not needing to maintain a branch network. We strongly believe a similar efficiency dividend would be realized in the insurance market, particularly when comparing agent-based distribution to a pure-play online distribution.

Industry Inertia

As a broad generalization, technological innovations originate (or are at least first commercialized) in the US and are subsequently exported to other countries, eg Uber, Airbnb, Apple, Google and Microsoft. The opposite usually occurs in financial services, with US innovation generally lagging that of other countries.

Examples of overseas innovations that were slow to be adopted, or haven’t yet been adopted, in the US in insurance include property level homeowners pricing, common use of GLMs, demand modeling, price optimization and the widespread transition from agent-based to direct online transactions. Similar examples of the US being a late adopter in banking include free overnight peer-to-peer fund transfers between any bank, chips in credit and debit cards to prevent fraud, contactless payment and the abolition of paper checks. Even US payment innovations like PayPal and the contactless ‘Apple Pay’ were essentially non-banking workarounds developed to provide the same payment functionality that had already existed for over a decade in personal banking in many countries outside the US, such as direct transfer and PayPass.

In the US, tech companies tend to be fast moving and agile, while insurance companies tend to be risk averse and compliance driven. What happens when you have a new world tech-based solution encroaching into an old world industry? It’s a case of an unstoppable force meeting an immovable object.

The insurance industry’s default course of action of sitting back and waiting to see what happens has not worked well for other industries disrupted by peer-to-peer technology, such as the music and entertainment industry. We believe this could go one of two ways. Just as the hotel industry, through the Hotel Trades Council, has preferred to let regulators wage war on Airbnb rather than expending energy doing so itself, so too would the insurance industry find this an effective first line of defense. As peer-to-peer insurance would represent a true existential threat to the insurance industry, lobbying of regulators by the industry to maintain the status quo could easily kill peer-to-peer insurance in the US before it can even start.

The second possibility, which would become increasingly likely if the default response to neutralize the threat fails, is that the industry pivots, embracing the peer-to-peer model, positioning itself for lead underwriter roles in ‘peer-to-peer’ insurance syndicates (as described earlier) and hence taking on members of the public merely as passive investors. The composition of Lending Club’s ‘investors’ followed this trajectory. Initially the funding base consisted of individuals lending as little as $25, but now only one third of funds are from individuals investors, with the rest coming from mutual funds and institutional investors who don’t micromanage every loan application.

Many other peer-to-peer businesses have become dominated by large established players once the opportunity (or threat to their legacy business model) was recognized. Avis acquired ZipCar in 2010, effectively a by-the-hour self-serve rental car service using cars conveniently scattered throughout participating cities. Mercedes-owned Daimler expanded its car2go service in 2009 which allows users to hop very short distances in a car without needing to return the car to its original location, effectively being a cross between Zipcar and the bike share infrastructure appearing around the world. In 2011 General Motors even invested $3 million in RelayRides. This model is analogous to creating a platform for crowd funded startup insurance trusts to operate, with the role of traditional insurance carriers morphing into that of managing these startup trusts.

So, is peer-to-peer risk transfer feasible? Could “insuring the sharing economy” really give way to “sharing the insurance economy”? We suspect asking the insurance industry this question would be like asking the Taxi Federation five years ago if they thought app-based ridesharing was feasible. The safest prediction we can make is that any entrants into this market will following in the footsteps of other disruptors, possibly asking forgiveness, but never asking permission.

Global Perspectives on Peer-to-Peer Insurance

Taking an international perspective on peer-to-peer insurance can be useful in understanding if, or how, it could be implemented in the US. The entire value proposition of peer-to-peer is that price savings can be achieved by disintermediating an inefficient, legacy-driven middleman. Ironically, the biggest force that could see peer-to-peer insurance thrive in the US, could be the very force keeping it out of overseas markets. We are referring to relaxing regulation and letting competitive forces drive product design and pricing. While US auto expense ratios are typically around 25%-30%, competitive forces in Australia, for instance, had driven expense ratios down to 10% decades ago. This has been achieved by significant automation and the dominance of online direct sales. In an already lean environment it is hard to see how a peer-to-peer platform could gain a cost advantage over existing players. Graeme Adams (Finity Australia) explains:

The industry in Australia has been direct for a long, long time. They switched into internet channels and electronic commerce. Branches and even telephone centres are a thing of the past. The leading car insurer has an expense ratio on their car insurance of around 10% but they also have massive buying power so they can get cars fixed cheaper than most other insurers, let alone an individual. So if you have peer-to-peer insurance on car insurance, how could that beat an expense ratio of 10%? What is the real saving they get in terms of the premium they pay? There is a cost to manage the enormous complexity when 200-300 people are effectively paying the claim.

There has, however, been somewhat of a resurgence in mutuals and buying groups overseas. Graeme Adams (Finity Australia) explains:

Buying groups are getting quite a leg up here. One Big Switch has got 630,000 members now from a standing start three years ago. That’s a lot. Another, Capricorn, is a discretionary mutual. They don’t provide insurance, they provide what they call ‘protection’. The thing with a discretionary mutual is they are not obligated to pay out a claim under a policy. It’s at their discretion that they pay a claim. Maybe there could well be a resurgence in mutuals because they have cheaper capital and don’t have to make a commercial profit. It’s particularly an issue as insurance becomes more expensive here.  It’s becoming more expensive for a whole host of reasons. It’s on more of a sustainable footing now. Also we understand the risk better. We understand flood particularly, earthquake, other natural peril risks are well reflected in premiums down to individual properties.

Dave Cummings (ISO) agrees with this comparison:

In many ways it’s a reinvention or an older concept. This could be analogous to mutual insurance, as it started years ago. The idea of groups coming together to self-identify and to start to provide means for insurance.  It’s interesting to see how we are resurrecting an idea that originated over 100 years ago due to modern circumstances.

From this perspective, peer-to-peer insurance isn’t anything new. It’s really just a resurgence of mutuals that have been with us since the dawn of the insurance industry, only this time with a flashy new app.


In preparing this post we interviewed the following people.

  • Joel Laucher, Deputy Commissioner, California Department of Insurance (CDI)
  • Frank Chang, Lead Actuary, Uber
  • John Clarke, Senior VP Marketing, James River Insurance Company
  • Sam Zaid, CEO and Founder, Getaround
  • Shelby Clark, Executive Director, peers.org, (also Founder and ex-CEO of RelayRides)
  • Dave Cummings, Senior VP Personal Lines, ISO
  • Mariel Devesa, Head of Innovation, Farmers Insurance Group
  • Robert Passmore, Senior Director of Personal Lines Policy, Property Casualty Insurers Association of America (PCI)
  • Jim McNichols, Chief Actuarial Officer, Greenlight Re
  • Laura Maxwell, Consultant, Pinnacle Actuarial Resources
  • Graeme Adams, Principal, Finity Consulting, Australia (and ex-Head of Product & Underwriting at IAG)
  • Dr. Amy Gibbs, Digital Communications Manager, ANZIIF

 The quotes attributed to each interviewee throughout this paper were spoken extemporaneously and do not necessarily represent the views of the organization they work for. We thank them immensely for their time, input and expertise.


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>