Insuring the Sharing Economy
Uberrima fides means “utmost good faith” or, more simply, “trust”. Two-way trust lies at the heart of the business models of both the insurance industry and the sharing economy. ‘Trust’ uniquely binds the two. When policyholders pay a premium they trust that insurers will honor their promise to pay at claim time. Insurers trust that policyholders will truthfully disclose relevant information at the time of both underwriting and claim submission. When somebody rents out their home for a few days to a stranger using Airbnb, they trust that this stranger will look after their property. When somebody takes a ride downtown courtesy of Uber they trust that the driver will get them there safely and hassle-free. Both those providing and receiving the peer-to-peer service trust that the tech-based intermediary that matched them up has policies in place to effectively deal with things going wrong. Whether such events are tragic accidents or malicious, everyone trusts that the relevant party within the transaction has adequate insurance coverage and that lawmakers have adequately anticipated these risks and mandated beforehand which party should carry what kinds of coverage. Most of the time, this trust is well-placed and this new economy functions seamlessly. But when it does break down, it does so with unfortunate consequences.
The “move fast and break things” mantra of Silicon Valley is at odds with the slow and cautious approach of the insurance industry. With traditional insurance you know what you’re insuring, who is using it and why. For auto coverage you know the insured’s age, address, vehicle and driving record, and you know if they will be using it for weekend drives or commuting to work. For homeowners coverage you know location, construction type, characteristics of the residents and that they will be living in the property most of the year. The fine rating details may vary from policy to policy, but the broad risk profile is consistent both over time and between policies. Even in commercial auto policies you know the fleet of vehicles being insured and possibly the pool of drivers, even if you can’t know who will be driving at any point in time.
Those assumptions break down in the sharing economy, where individuals act as micropreneurs, switching their assets seamlessly between personal and commercial use. Risk in the sharing economy is somewhat similar to a landlord’s policy or a home business endorsement on a homeowner’s policy, where a traditionally personal lines asset is used for income generation. But even these personal-commercial hybrid risk profiles tend to be both standardized in their own right and constant over time. A rental property remains so throughout the year, with tenants changing at most once or twice a year and with standard risk mitigation measures in place, like reference checking and bond requirements. The same properties (or vehicles) entering the sharing economy can see personal and commercial uses being juggled day in, day out. This new breed of mixed use asset is being facilitated through the recent rise of the smartphone app and tech juggernauts acting as brokers that instantly match service seekers with service providers.
Hundreds of sharing economy startups have launched online in the past six years. The following are some of the better-known firms:
- Homesharing: Airbnb; VRBO, HomeAway, Wimdu
- Transportation Network Companies (TNCs), aka Ridesharing: Uber, Lyft, Sidecar, Hailo
- Carsharing: RelayRides, Getaround, FlightCar
- Care: DogVacay.com (dog care), care.com (child care, home care, senior care, pet care)
- Other: Taskrabbit (odd jobs and errands), SnapGoods (possession sharing), EatWith (dining with strangers)
Businesses like ZipCar and car2go offer essentially by-the-hour rental cars leased from a dedicated corporate entity and do not offer truly peer-to-peer services, so aren’t included above.
Insuring the sharing (or peer-to-peer) economy requires a unique insurance product design that bridges personal and commercial insurance, as well as a pricing methodology that is responsive to the mix of personal and commercial exposure varying day-to-day or minute-to-minute. The unique risk profile of using personal assets for peer-to-peer income generation on a large scale, facilitated by a technological intermediary, gives rise to two different issues. Firstly, an insurance gap arises because personal lines policies generally won’t pay claims if the asset was being used for income generation at the time a claim is incurred. A separate policy, probably a commercial lines one, would be required, which can be an expensive and onerous requirement for a micropreneur. Secondly, even where insurers are prepared to cover these periods of exposure, the question of how to price this coverage is tricky. Historical experience is little help as a pricing guide for the new and unique risks presented by the sharing economy.
Unique Risk Profiles
It is important to isolate the underlying reasons for why the risk exposure for each type of peer-to-peer business is different from that of a standard personal lines policy covering the same asset.
Homesharing services like Airbnb generally create a greater property risk than is priced into a traditional homeowner’s policy, resulting in denial of claims arising from or because of a ‘guest’. The increased risk mainly arises from having strangers occupy and use the property without the owner’s supervision. The exact accommodation a guest may rent varies from the entire property to just a couch for the night. Guests may intentionally steal or destroy the property or simply act more carelessly than they would with their own property.
A 2011 incident referred to as ‘Ransackgate’ involved a woman renting out her apartment in San Francisco’s Mission district on Airbnb. The guests vandalized the property, burning much of her possessions to ash, as well as stealing birth certificates, social security numbers and credit cards that were kept in a safe on the premises. This prompted Airbnb to create its $50k (now $1m) Host Guarantee and offer it free of charge to hosts in the United States and now several other countries.
Transportation Network Companies (TNCs)
TNCs (or ridesharing companies) involve a taxi-like, or limousine-like, service where drivers respond to requests on their smartphone and transport passengers from one destination to another. The TNCs (such as Uber, Lyft or SideCar) sign on drivers as independent contractors and not employees of the TNC. The TNC provides the connection infrastructure, payment processing and branding that drivers rely on to attract passengers.
The increased risk for TNC drivers working for companies like Uber, Lyft and Sidecar arises for quite a different reason to that of homesharing. The guests in a TNC service are chauffeured rather than left unsupervised, so malicious damage to, or theft from, the vehicle is unlikely to be a problem. The increased auto risk simply arises from being on the road for a longer period of time (and proportional increase in liability and collision risk) when your job involves driving for much longer periods than would be the case under an equivalent purely personal use vehicle. There is also an increased risk from travelling through a greater variety of neighborhoods that the driver may not be familiar with.
There are many types of TNC drivers, from those who drive a 40-50 hour-per-week full-time job, to those who opportunistically offer rides occasionally when they happen to be travelling for personal reasons and check their app to see if they can pick up someone travelling in the same direction. The nature of risk exposure from TNC risks is not qualitatively different from that of a personal lines policy, only the duration (or distance travelled) for which they are exposed.
Carsharing services like Getaround and RelayRides involve an individual advertising their car on a smartphone service, which can then be rented by other individuals on a short term basis. As a cross between Transportation Network Companies (without the chauffeur component) and homesharing services, carsharing services suffer from the problems of both groups. These vehicles are on the road for longer than pure personal vehicles are, and they are also in the possession of strangers, who may act maliciously or more carelessly than the owner would. A high demand vehicle in a city center location could have six different drivers per day…
When you rent out your vehicle under RelayRides and Getaround, your personal lines policy ceases to be exposed, with the RelayRides and Getaround commercial policy becoming primary with a Combined Single Limit of $1m. The insurance market had never seen this type of exposure before (outside of traditional rental car businesses), so even when insurance is offered it tends to be priced somewhat conservatively as if it were a commercial livery policy. More commonly, insurance is not obtainable at all.
The business model behind TaskRabbit is that when someone needs a quick errand run (such as picking up dry cleaning), they will advertise it on the app, with an individual (known as a Tasker) in the area bidding on it, such as offering to pick up your clothes and deliver them to your house for a $5 fee, while another person/Tasker beats that offer with a $4 bid. On the surface, an errand running service seems to have a less problematic business model than Airbnb or Uber. After all, they aren’t leasing property, nor are they transporting people on public roads. However, if the Tasker is driving while running an errand, say to pick up your dry cleaning, then the personal lines auto insurer usually won’t pay, since their vehicle is being used for income generation. Knowing this, it’s much easier for a TaskRabbit courier to equivocate at claim time, saying they happened to be driving for personal use at just that time rather than running someone else’s errand, than it is for an Uber driver or Airbnb host to similarly claim that damage arose purely from personal use. A larger risk exists for liability and vicarious liability if people/property are damaged during the course of a Tasker performing an errand.
Peer-to-peer possession sharing, like that offered by SnapGoods, is probably the most problematic offering in the peer-to-peer economy from a risk pricing perspective. If one person lends another person a ladder for a fee and the ladder breaks, who is liable for the resulting damage? How can such a liability be priced in advance? When faced with the liability risk of one stranger lending another stranger any variety of household tools like a ladder, a drill, a corkscrew or a chainsaw in any possible state of repair or disrepair, other peer-to-peer arrangements like Airbnb and RelayRides start to look extremely standardized and predictable by comparison.
Current Insurance Arrangements
In one sense, many of the risks presented by the sharing economy have always existed. They have just been unknown and absorbed into the general risk pool. Joel Laucher (CDI) commented:
It seems so new, and yet we know it has probably been going on for some time, so it’s not brand new. I think it’s just that size of the enterprise has grown to a point where it has vaulted into view. The exposure’s been there and it has been absorbed without anyone noticing. There haven’t been any awful consequences that we’ve heard about before the TNC activities hit the news as a result of accidents involving fatalities and injuries.
Robert Passmore (PCI) expressed similar sentiments:
People have had vacation homes forever and rented them out part of the season and used them part of the season. Insurance products have been adapted to them…..Like being a handyman; TaskRabbit is providing a more formal marketplace for something that has always been available informally. The risk has always been there. If you’re working as a handyman there’s a possibility you could make a mistake and something bad could happen. I don’t think that’s changed. I think they could grow a lot more because of the ease of use of the marketplace. The smartphone is just a boon for this kind of stuff. Before you had to hear about a handyman by word of mouth.
In many of these instances, the admitted market won’t even accept the risk, and coverage needs to be sought from the Excess and Surplus (E&S) market. In most states even accessing the E&S market first requires a licensed agent to conduct a due diligence search from admitted carriers in the state to try and accept the risks. Upon three rejections, the agent is allowed to access the E&S market via a surplus lines broker licensed in the state. John Clarke (James River) explains how critical the E&S market was to the fledgling sharing economy:
The creation of coverage for the TNC industry (ridesharing coverage) is a great example of the surplus lines market at work. Last year, as the new industry saw a large amount of growth, they (the ridesharing companies) were deciding to buy, add or endorse coverage related to UM, UIM, expanding limits, changing from contingent to primary coverage and making all kinds of the other coverage changes. Frankly, you have to have the flexibility of surplus lines to keep up with something that’s evolving this fast. We could make changes for them rapidly. The admitted market just doesn’t have that flexibility, even if they wanted to do so.
These exposure types do have precedents from the old economy though. A pizza delivery driver using their own vehicle wouldn’t be covered by their personal lines policy during work periods, so the pizza chain’s commercial auto policy could cover these non-owned autos for the specific times they are being used commercially. A landlord’s policy or home business endorsements are other existing examples of personal/commercial lines hybrids.
Some of the ridesharing companies have an E&S policy covering their drivers whilst they are operating in a ridesharing capacity with a commercial liability limit typically of $1m. However, up until recently they were engaged in a disagreement with admitted insurers in most states on the risk profiles of their drivers and hence the price of the commercial coverage. TNCs argued that their full-time and part-time drivers have risk profiles more similar to that of a personal lines risk or a less active livery service. Insurers, however, argued that the ridesharing companies’ drivers are more similar to taxis, requiring higher premiums than other types of livery, and much, much higher than a personal auto policy.
There have been reports in the news that TNC-related claims are probably still being reported as personal auto claims. Joel Laucher (CDI) commented:
The personal auto carriers are probably paying some costs that they didn’t account for in their pricing and are paying for coverage that they think they’ve excluded.
Personal auto insurers don’t seem too concerned by the rise of ridesharing, with Joel Laucher further referring to conversations with carriers:
I was surprised on the ridesharing part, that companies didn’t come in and immediately strengthen their exclusions on the livery….No one’s really told us, well our first question now during a claims investigation is “Are you driving for a TNC?”. They have kind of said that they haven’t really changed their practices. It doesn’t really seem prudent, at least in these areas where you know there is a lot of activity.
Laura Maxwell (Pinnacle) reiterated this view:
Insurance companies need to start working on their underwriting rules and policy exclusions. I don’t see that happening.
The lack of proactivity, initiative or innovation by the insurance industry was a consistent theme that resonated through each interview. Dave Cummings (ISO) elaborates:
From my point of view, [the insurance industry] has historically not been at the forefront of emerging technologies and changing conditions. Many new exposures are initially excluded and the sense of urgency is not immediately recognized. However, over the past few years, we are seeing some insurers addressing these issues differently and embracing change and innovation. This is a very encouraging trend for the entire industry.
The reluctance of carriers to take decisive action can probably be attributed to two things:
Firstly, lack of a clear opportunity, with both the small size of the potential premium pool and the high uncertainty around its size. Joel Laucher (CDI) elaborated on this:
TNCs themselves play it pretty close to the vest in terms of how many drivers they have and so there’s kind of a lack of information about how many exposures are out there. Insurers don’t know how many TNC drivers they have on their books now. And they don’t know if there are enough TNC drivers to make it a market that they want to get active in. I guess they don’t know how much it might grow and if they are going to miss the boat if they don’t get out there. Insurers really want to see something worth their time before they spend much energy on it.
Laura Maxwell (Pinnacle) made a similar observation about the observed to-date small size of the ridesharing opportunity, and by extension the additional risk, in a report for the Colorado DOI:
[The report] looked at how many extra miles rideshare drivers are going to drive compared to how many miles are already in the personal auto system. And they came up with pretty much none. It is such a small percentage at this time.
Secondly, the elephant in the room is simply the difference in culture between the insurance industry and the tech startup industry, with Sam Zaid (Getaround) elaborating:
I think insurance [culture] is very entrenched and slow moving. That’s sort of the nature of something where you have a lot of risk. If something is risky and you iterate very quickly, odds are you are going to lose that game. If you have something that really works and covers all the risk, it can be scary to go in a new direction. Historically the rate at which industries formed and shaped has been a lot slower than it is today. Insurance companies are iterating at that previous pace.
In mitigating the risk of stepping into the unknown, the actuarial profession’s default approach is to first amass data. The more data the better, and don’t come back until you’ve got it. Dave Cummings (ISO) commented on ridesharing data collection:
Personally, I’d love to have as much of the data as possible. The more data we have on the risk the more accurately we can price. There are opportunities in data here which could enable some very interesting pricing and could respond well to the types of exposures and risks. This data would be different than the data we traditionally collect. However, it may be data ridesharing companies are reluctant to share. The fundamental questions we want to answer regarding risk are; how often a driver is using their vehicle for personal use vs. ridesharing, how many miles are they driving, where are they driving when working and when are they driving for the ridesharing company? Understanding if a driver is operating in an urban environment at night versus daytime in a rural setting, gives us an opportunity to think about and analyze the risk holistically.
And on homesharing data collection:
I’d like to see more data collection that speaks to the exposure, seeing as how the exposure is a little different that a traditional homeowner or renter exposure. A lot of data is being collected by other sources and relates to the home’s usage and exposure. One useful piece of information would be a better understanding of the owner or renter to relate that to other aspects of the risk. Knowing who is hosting and who is occupying the space during could better define the aspects of risk, specifically to the individuals involved.
And touching on setting up a sharing economy central database:
A central repository to identify those who are drivers for ridesharing, leasers of homesharing, or participants of carsharing could be beneficial across the entire industry. The individuals will benefit from receiving proper coverage and ensuring there are no gaps in the coverage. Insurers will benefit by better pricing and classifying the risks they choose to write. Creating a mechanism where insured and insurers are able to communicate openly about coverage would help ensure coverage exists from the insurer’s side and adequate coverage is received by the insured.
Mariel Devesa (Farmers Insurance Group) also spoke on the challenges of pricing ridesharing in the absence of good quality data:
Data is key for us to price appropriately. As this is a very new industry, we used available data to price. As we learn more, by gathering actual data about our specific drivers and start understanding our drivers’ behaviors better, we’ll be able to improve. With more data, everything will get better.
A catch-22 arises from the insurance industry’s desire for data as a prerequisite to offering dedicated insurance products for the sharing economy. This has led to much frustration from entrepreneurs unable to launch their sharing economy ventures. Shelby Clark (peers.org) recounts his experience in launching RelayRides:
When we were trying to launch RelayRides, [one carrier] strung us along for six months and then they said, ‘We really want to write this policy but we just need some data so why don’t you come back after six months of operations and we’d be happy to take a look at this.’ For us, that was really not helpful at all and they should have told us this six months ago. How are we supposed to get the data if we can’t operate without insurance? We didn’t go back to them. If you don’t take a chance you lose the business.
The RelayRides experience was far from unique. Sam Zaid (Getaround) also detailed the difficulties he faced securing an insurance arrangement before being able to launch his company’s carsharing business:
We talked to agents and brokers. Many of them told us they could help but all they offered us were off-the-shelf products. We resorted to calling VPs from different insurance companies directly—we probably reached out to between 50 and 100 different contacts, either through warm introductions or cold calls. One or two insurers came to the table but, still, we did not get a deal done. That process was probably about 12 months. I guess you could argue that a car owner’s insurance would apply but that wasn’t proven. Since the car owners themselves weren’t driving, we felt we needed a group insurance solution to protect our community. Once we finally secured insurance, we launched.
Other startups haven’t been so cautious, instead preferring to launch anyway, expecting (or hoping) their personal insurance policies would simply cover claims due to ambiguities in policy wording that meant, in many cases, sharing economy activities weren’t explicitly excluded. With an industry that has been very slow to offer specialist coverage, or even tighten up exclusion ambiguities in existing personal lines coverage. The only real option for such startups has just been to launch and hope for the best, with the expectation that once enough data is collected, proper insurance solutions will be developed.
In October 2014, five and a half years after Uber was founded, Erie Insurance launched what they touted to be a first-of-its kind coverage specifically designed to protect TNC drivers. As best we can tell from examining the publicly available filings, Erie have taken their personal auto business use endorsement, like that used by pizza delivery drivers, and removed the livery exclusion, while keeping the existing pricing structure in place. Their flat business use endorsement remains at 12% or 20%, depending on whether the annual number of miles driven is less than or greater than 12,500, but is not sensitive to the proportion of the driver’s time that is split between personal driving and driving for hire.
Erie’s effort was followed by Farmers Insurance Group launching a ridesharing specific endorsement to their personal auto policy in Colorado. Available from February 2015, the Farmers endorsement extends personal lines coverage to Colorado’s legally required limits when a ridesharing app is turned on but no passengers have yet been accepted (commonly referred to as Period 1). The endorsement ceases when a ride has been accepted, as the TNC’s group commercial policy should then become primary. Press releases suggested the endorsement was priced at an average of 25% loading.
In early 2012, in response to ‘Ransackgate’, Airbnb started offering a Host Guarantee Policy to hosts living in qualified countries. The Host Guarantee Policy, underwritten by a Lloyds of London syndicate, provides a $1m limit. This only covers deliberate property damage by a guest and does not cover accidental property damage nor liability, applies in excess of any primary policy, applies only after seeking and failing to recover from the malicious guest himself and needs to be reported the sooner of 14 days after check out or at the start of the next rental. This coverage could be considered pretty restrictive and possibly lead to an insurance gap problem. New York lawmakers have picked up on this, urging the State Superintendent in September 2014 to investigate.
In response to this gap in liability coverage, on 20th November 2014 Airbnb announced the introduction of Host Protection Insurance. Effective 15th January 2015, it will automatically provide $1m liability coverage to hosts within the US in excess of their primary coverage.
HomeAway (another homesharing service) offers primary commercial coverage to their members through a program called Assure that they write in partnership with P&C broker CBIZ Insurance Services.
Like Airbnb’s Host Guarantee and Host Protection Insurance, most homesharing coverage developments have been initiated by the homesharing websites themselves. Insurers have been slower to address homesharing coverage gaps than they have been for ridesharing. Joel Laucher (CDI) commented:
There’s a big difference between homesharing and ridesharing. I think it’s fair to assume that the insurers’ intent was to not cover ridesharing exposure at all. On the homesharing side it’s not that definitive that companies didn’t want to write that coverage. There’s an exclusion in the liability section that indicates ‘We don’t cover any rental of the home except on an occasional basis’. Because it is not an absolute exclusion, there’s clearly some level of this exposure permitted, so the coverage issues will probably evolve a little more slowly.
This sentiment was mirrored by Dave Cummings (ISO):
We are devoting more attention ourselves to the homesharing side of the issue. I think there’s more to come there. The insurance side hasn’t bubbled up in the same way as media or public awareness has when compared to ridesharing in the last year and a half. To some degree I think that might simply be driven by where public attention is going and where regulatory attention is going. Certainly ridesharing has been increasingly active. Homesharing hasn’t been receiving that level of activity, at least not yet.
And by Robert Passmore (PCI):
We haven’t seen much regulation or legislation on the Airbnb’s of the world. Most of the discussion about Airbnb has been around zoning and taxes, things like that that don’t come so much into the insurance realm. Airbnb has taken a different approach. They’ve come along a little bit quicker, perhaps taking what’s happened with the TNCs as a cautionary tale.
Joel Laucher (CDI) also elaborated on the greater difficulties home insurers have than auto insurers in even identifying that insured’s assets are being used in the sharing economy:
Auto of course has a high frequency of accidents so an insurer is more likely to find something out about its risk. Home insurance, unless something pretty bad goes wrong and there is an injury, nobody knows about the sharing activity. You’re not going to have enough of the frequency to give you much of a signal to tell you something is going wrong. I think sharing-specific coverages or exclusions will be slower in developing – it may still be some time before insurers really get concerned or even figure out how to monitor exposure on the homesharing side. There’s got to be enough frequency for them to catch on to the exposure. Until that happens, everything is just absorbed into the regular loss pool.
As of late 2014, four insurers offered a business owners policy (BOP) specifically for people offering their homes for short term rental on sites like Airbnb and HomeAway. Officially structured as a BOP, they are designed to replace the homeowners or renters policy that an occupant would normally have. To take one of these four as an example, Proper Insurance Services offers $1m in commercial general liability, $1m in personal liability, building damage coverage, personal property coverage and lost income. One key difference between this coverage type and that of HomeAway’s is that this coverage is offered at a flat premium (many multiples that of a pure personal lines policy) regardless of how often the property is rented, rendering it uneconomical for the very occasional host.
RelayRides and Getaround sought insurance on behalf of their pool of available vehicles from more traditional admitted carriers.
Ridesharing companies don’t generally provide blanket commercial insurance coverage. Their business model becomes much simpler, and their liability much reduced, if they are facilitators or matchers of service seekers to service providers, not as providers themselves. If each micropreneur were responsible for their own insurance coverage, sharing companies would fall back to the much less risky position of being a tech company simply providing an online matching service. Insurance is a very complicated and compliance-driven area that falls outside their prime competency.
As described earlier, the first problem is disagreement between the peer-to-peer companies and insurers on whether the true risk profiles are more similar to that of personal lines policies or commercial policies. There isn’t enough data to determine which view is closer to the truth. The second problem is that the part-time personal / part-time commercial nature of these risks makes it problematic to even determine when the asset has moved between these two states.
The ultimate solution may be a hybrid personal and commercial policy, switching between these coverage types as appropriate. Determining where and when the pendulum should swing between these two bounds, and pricing it accordingly, is the challenge facing peer-to-peer networks, insurers and regulators.
Regulators have been more proactive in addressing sharing economy coverage gaps than insurers themselves have been. The California Commissioner, Dave Jones, has been active writing letters to the public utilities commission, holding hearings, moderating an educational event at a recent NAIC meeting and chairing a sharing economy working group. Joel Laucher (CDI) commented:
Commissioner Jones is very interested because he sees a huge exposure, a chance for people to be injured and not compensated, when clearly these are exposures that should be covered. He wants to exert his influence as much as he can to see that the sharing economy industry is taking appropriate responsibility and that the insurance market is responding with relevant products. That’s what insurance is all about. But I think we found that much of the industry is kind of sitting back and watching to see where this will go.
Amy Gibbs (ANZIIF) was more critical of this ‘wait and see’ approach:
Sitting back and waiting to see what happens has not worked for other industries, neither has dismissing the technology as fad. Those that do embrace digital early will stand a good chance of becoming market leaders, so the potential benefit may outweigh the risk.
In October 2014, San Francisco passed a law, becoming effective February 2015, legalizing property rentals for less than 90 days for city residents renting out their property, but requiring the collection of hotel taxes and a minimum $500k liability insurance coverage. This law has been dubbed the ‘Airbnb law’ by competitors like HomeAway because legalizing only rentals for resident hosts in the city disadvantages HomeAway’s customer base that is weighted more toward out of town owners that list their San Francisco properties for short term rentals on a full-time basis.
In September 2013 the California Public Utilities Commission (CPUC) passed a law labelling ridesharing companies as ‘Transportation Network Companies’ (TNC) and that all drivers operating in California must carry $1m commercial liability insurance effective when the vehicle is operating as a livery vehicle. At that time, no guidance was given on when the personal lines coverage should give way to the commercial coverage or vice versa.
The ambiguity of coverage came to a head on 31st December 2013 when Syed Muzaffer, a 57 year old Uber driver, tragically hit and killed six year old Sophia Liu, as well as injuring her mother and brother in the Tenderloin neighborhood of San Francisco.
The driver was not carrying a passenger, nor responding to a passenger request, but did have the Uber app turned on. The driver’s personal lines insurer denied liability, arguing that the app being turned on was enough to classify the vehicle as being used for a commercial purpose at that time. Uber also denied liability, arguing that with the driver not carrying a passenger, nor responding to one, the fact that the app happened to be turned on was not enough for it to be classed as commercial use.
This insurance gap was picked up in the media and political discourse, with the CPUC mandated to devise a solution. The public dialogue of liability in the wake of Sophia Liu’s death led to the formulation of a three period system promulgated by assembly bill AB 2293, which goes into effect July 2015:
- Period 1 – driver turns app on waiting for a passenger match;
- Period 2 – match accepted, driver en route but passenger not yet picked up; and
- Period 3 – passenger in the vehicle until passenger exits the vehicle.
The Sophia Liu incident occurred under a period 1 exposure, but the bill passed in September 2013 did not explicitly specify if the commercial insurance requirement was to apply under period 1 (or any other period). TNCs are generally in agreement that the commercial liability requirement applies for periods 2 and 3. Dave Jones, Insurance Commissioner for California, and Benjamin Lawsky, Superintendent for New York Department of Financial Services, told the press in January/February 2014 that they had concerns about the insurance gap in period 1.
Robert Passmore (PCI) observed that this incident led to a rapid closing of insurance gaps:
The TNCs themselves went from offering little or nothing in the way of insurance coverage, and they’ve incrementally increased that over the last year and a half or so. The discussion has become more about a couple of narrow periods of time where there are gaps rather than no coverage whatsoever.
Robert Passmore (PCI) also opined that rather than overly prescriptive regulation, the best approach to further close insurance gaps is legislating simple, clearly defined requirements but leaving the ‘how’ up to industry innovation, combined with adequate disclosures to drivers:
Our position is pretty simple. The best way to support innovation is to have some clear, very basic insurance rules that say when you’re making yourselves available you need to have specific insurance coverage that applies. We want to leave the door open for insurance companies to innovate and offer a personal lines product with an endorsement to cover those kinds of exposures. You can leave the door open to all sorts of things, coverage purchased by the driver, coverage purchased by the TNC company or combinations thereof. …People that sign up for the program need to get some information about here’s the insurance that you need to get or here’s what we provide for you and here’s some information about the personal lines policy as it is unlikely to provide coverage for you. We think some basic disclosure when you sign up is very important. That’s when you need the information, when you’re deciding to enter into this activity.
When we asked Frank Chang (Uber) what he wishes the insurance industry would do to make ridesharing more accessible, he responded:
There’s a huge opportunity for premium for the companies who can construct a seamless product for period 1.
On 14th March 2014 Uber announced that effective immediately they would provide$50k/$100k/$25k of coverage during period 1. When it becomes effective in July 2015, AB 2293 will require a minimum $50k/$100k/$30k of insurance coverage in period 1, while periods 2 and 3 remain at the $1m limit requirement. Further, the TNC coverage is to be primary.
For context, other CPUC mandated minimums include:
- $15k/$30k/$5k for personal auto
- $750k commercial liability for up to seven passengers (charter-party)
- $1.5m liability for up to 15 passengers (charter-party)
- $5m liability for 16 or more passengers (charter-party)
The California state minimums for taxicabs mirror that of personal auto (15/30/5). However, in California, cities and counties regulate taxis, not the CPUC, and each typically imposes their own higher minimums. San Francisco, for instance, imposes a $1m minimum.
Although California is leading the way on TNC regulation, and therefore has the most relevance for framing insurance product development, other states are making similar strides. Colorado has passed SB 14-125 and other municipalities are following Colorado’s lead. These cities & states are likely to adopt regulation similar to, if not identical to, AB 2293 in California, with period 1 requiring a limit greater than normal personal lines coverage but not quite as onerous as that required for periods 2 and 3. It will be up to personal lines insurers whether they want to cover or exclude that period 1 exposure.
Robert Passmore (PCI) said that the city-by-city patchwork of insurance requirements is a challenge for the industry, but is ultimately a public policy issue:
The TNCs say they are not a taxi service, but what they do very closely resembles what taxis and limos do. If you look at insurance requirements for those kinds of services, they are all over the map, partially because they are set sometimes to the local level. Some places you don’t have to have limits that are any higher than personal auto has. Other places it is as high as $5m. The consensus in the industry is that that’s a public policy issue for the individual states to decide how much they want to require. We don’t take a position on how much. We take a position on what is primary and specific so it doesn’t leave any gaps between what the driver would have on their personal policy and what the TNC-specific insurance is providing.
In the next 18 months the National Association of Insurance Commissioners (NAIC) is likely to adopt model laws for all the states to subscribe to, making the TNC category a described line of insurance alongside taxis, livery and charter parties.
This period-based approach, where each of the TNCs hold their own excess policies, is complicated by the fact that drivers can have multiple apps turned on at once (e.g. Uber, Lyft and SideCar apps all active on the smartphone), with drivers wanting to access the largest pool of potential passengers they can, rather than limiting themselves to one brand. Which TNC’s coverage would apply in this case, with multiple apps active but no specific passenger having been accepted? The more entities that are potentially liable, the more likely that none will ultimately be held liable because each can convincingly argue that ‘someone else’ is. This is the ‘diffusion of responsibility’ principle at play.
Cities in the US and around the world such as Omaha and Berlin are trying to make it illegal for TNCs to operate. This is partly due to the insurance gap problem, and partly successful lobbying from the taxi industries trying to address a competitive threat.
The NAIC has formed a sharing economy working group. Some of the aims of the group are to create a common language around the TNC exposure, for example about what the periods are, and share developments in terms of the coverage requirements that the states have developed. Joel Laucher (CDI) commented about the objectives of the working group:
I think a lot of it will be about the sharing of information about what is going on in the states on the legislative or regulatory fronts, identifying and clarifying the exposures involved and the coverage gaps, communicating with the industry and consumer representatives that are there about these new exposures and getting input about how to address them. Our commissioner really wanted there to be a forum to address these issues in a more orderly and comprehensive way.
The TNCs themselves have a very strong incentive to work with the regulatory and legislative process, if only to achieve uniformity of coverage across jurisdictions. Joel Laucher (CDI) commented:
I’m pretty sure the TNCs will want to have some level of consistency across the country so that they don’t have 40 different policies at different underlying limits in different states. They would probably like to have a national coverage policy that has the same underlying coverage by period. They were very engaged here in California and Colorado, and the period 1 underlying limits are the same in both states.
Crafting regulation pertaining to how insurers can act towards policyholders and members of the public who choose to participate in the sharing economy is a good first step, but then education and enforcement is required. Sam Zaid (Getaround) said:
We continue to see a few cases where a car owner’s insurance company refuses to renew their personal auto policy. This is usually pretty straight-forward to resolve although we often have to contact their insurance provider and educate them. The consistent response we receive is that they are unclear on policy and regulation. A lot of it is just an education process. Insurance is such a distributed and decentralized industry that you always have agents that are unclear on their own carrier’s official policies.
Often overlooked, operational growing pains rarely grab headlines the way regulatory and product design challenges do, but they are no less real, as explained by John Clarke (James River):
The TNC auto coverage has been a challenge simply because of the sheer amount of industry growth. We have people 24 hours-a-day setting up new ridesharing claims. We’ve established large teams in Scottsdale, Arizona as well as our home base in Richmond, Virginia to deal with the claim volume. The growth in those teams is not stopping any time soon. The growth of these businesses and the numbers of rides, the number of drivers and the number of miles these firms are rolling every day far exceeds what they could have guessed what they were going to do a year ago and certainly two years ago. This is a frequency driven business. There are the occasional large losses that generate headlines but the quiet headline is the sheer volume of very small claims as you would expect in an urban environment. Most may be small claims but there are a lot of them, and we have to be able to meet the service demands of these clients.
These new types of risks will necessitate a new type of insurance coverage. Tweaking some policy wording or trying to retrofit an existing insurance product just won’t be enough. In insurance, as in everything, necessity is the mother of invention. Sam Zaid (Getaround) opined on this:
Insurance is typically supporting the business so it subtends many other industries. Insurance usually follows something—there’s this new risk so insurance fills a gap. All this innovation is being driven by disruption of industries that have said, ‘We have a new risk profile and the insurance products that exist don’t cut the mustard.
The industry has not reached a consensus on TNC insurance product design. Some are advocating modeling it on personal auto, whilst others are advocating a commercial auto approach. Laura Maxwell (Pinnacle) explains:
I think personal auto can rate so much better than if you rate it as commercial auto. You can just get so much more detail from personal auto.
Mariel Devesa (Farmers Insurance Group) also argues that personal auto provides a better template for ridesharing coverage:
We looked at the underlying usage of the vehicle, what it is being used for, and how consumers are interacting with the TNCs. What we are seeing is that the majority of the time drivers are using their personal vehicles for personal use. Our position, currently, is that period 1 is an extension of that personal use and therefore would fall under a personal use policy.
Dave Cummings (ISO) agrees that the sharing economy isn’t going away, nor the insurance gaps that attend it:
I believe the sharing economy will continue to grow. It’s a new business model that, to some degree, blurs the distinction between personal and commercial exposures. As a result, there’s a big insurance coverage issue that needs to be handled and addressed. Personal lines insurers will need to be part of the solution and need to accommodate in some way. I believe that we’re only seeing the tip of the iceberg of what these issues may become. What we are seeing is a pattern where technology and connectedness are enabling an entrepreneurial model that wasn’t previously possible. Other innovations are likely beyond the sharing economy. There’s likely to be additional innovations where you see interactions and people thinking of ways they can commercialize their assets, their belongings, and their time in ways that are going to create a different type of business model again. This could produce different types of insurance exposures that we need to be ready to adapt to. We are currently writing coverage so that the industry as a whole continues to grow and address and enable these economic developments quickly and effectively.
Our view is that the distinction between personal auto and commercial auto is artificial and unnecessary. While it has been historically convenient to treat them separately, that notion is becoming outdated with the sudden ubiquity of peer-to-peer services that blur the line between commercial and personal. With separate customer bases and different drivers of claim experience, it has historically been convenient for carriers to separate product management, pricing and distribution networks into personal and commercial streams. Any middle ground between the two streams, such as personal vehicles being used for occasional commercial use (e.g. evening pizza delivery), has carried such little exposure that it wasn’t worth deviating from the binary personal/commercial structural split. It was easier to just add an endorsement to either the personal or commercial policy templates to cater for these infrequent edge cases. With TNCs now a growing segment firmly occupying that middle ground between personal and commercial, it makes sense to break free of the binary product template that has been a convenient way to segment the auto insurance market for so long. After all, there’s no inherent reason why you can’t segment commercial auto as granularly as personal auto. It just hasn’t historically been convenient to do so. Those edge cases are now becoming so common that the old binary split is now best thought of as a continuum.
We believe one of the simpler solutions to insuring TNC drivers is to adopt a Usage Based Insurance (UBI) philosophy priced with a personal lines rate plan and then applied to commercial or hybrid usage. First, the characteristics of the driver and the vehicle (age, sex, zip code, credit score etc.) would be used to compute the premium as if it were a plain vanilla personal lines auto policy and then broken down to cost per-mile. In addition to the standard personal auto premium, this cost per-mile is charged to each driver on a quarterly basis based on the number of miles they are actually driving in their capacity as a TNC driver over and above their personal driving. Depending on the jurisdiction, and the resulting limit requirements, this cost per-mile can be scaled up by the increased limit factors appropriate to the limit that applies to the period in which that ‘TNC mile’ falls, all automatically recorded by the app and reported by the TNC to the insurer. This places the cost of insurance back on the driver, while ensuring the TNC itself is complicit in accurately recording and reporting each driver’s risk exposure.
As a simple example, if based on driver and vehicle characteristics the personal auto premium is $500, assuming an average of 10,000 miles and minimum personal auto limits, then the effective cost per ‘personal mile’ is 5 cents. If the increased limit factor to meet mandated limits when driving in a TNC capacity (defined as being when the app is turned on, say) is 2.0, then the cost per ‘TNC mile’ is 10 cents. The total premium for that driver then becomes $500 plus 10 cents times the number of miles driven while the app is turned on, the mile count being automatically reported by the TNC to the insurer.
This per-mile pricing would require the insurer to have a relationship with the TNC. It would be very difficult for an insurer to unilaterally insure a TNC driver, hoping to differentiate pricing periods between TNC miles and personal use miles and identify if a particular claim occurred on TNC time or personal time. It would, however, also directly address the ‘insurance fraud’ argument made by the taxi industry against TNCs that drivers have an incentive to leave their apps turned on even when having no intention of picking up passengers because of the benefit from increased insurance coverage. Paying per-mile for the increased coverage would disincentivize drivers from triggering the app unnecessarily.
We would also expect such product innovations to incorporate other developments like Pay How You Drive (PHYD) telematics, social media-based rating and transitioning from agent-based distribution to the pure-play online distribution increasingly expected by the millennial generation, but we won’t dwell on these developments in this paper as they are not specific to insuring the sharing economy.
On January 28, 2015 Dave Jones, the California Insurance Commissioner, announced approval of a new insurance endorsement for UberX drivers that have their vehicles insured by Metromile (a per-mile personal auto insurance MGA) to obtain period 1 coverage.
Metromile is leading the way to expand the insurance coverage available to UberX drivers and passengers… We encourage other insurance companies to offer insurance coverage to California drivers who drive for UberX and other transportation network companies.
Frank Chang (Uber) had this to say on the proactivity of personal lines insurance carriers and UBI:
There has been response from a limited number of players who are set up to build ridesharing insurance products. In the news, Metromile, USAA and Farmers have products for period 1. There is a Virginia filing from GEICO that covers all three periods. Definitely UBI is the best solution, so we’re glad for the partnership with Metromile.
Dave Cummings (ISO) framed the opportunities around using apps for data capture:
Due to the advances in the technology, the apps on our phone that we are already using, provides an opportunity for us to leverage new data that wasn’t available even five years ago. This will further help us seek risk based pricing by getting more and better data about the true exposure and risk.
There are many opportunities like determining how many miles are being driven, where ridesharing drivers are operating their vehicle. Again, this will shed light not only on how they are driving, but what driving conditions they are operating. Are they in rush hour traffic? Are they driving in a snow storm? These are just some questions we can seek to answer with technology advances.
Mandating the capture of detailed usage data like this will facilitate, in the long term, better pricing models specifically for TNC usage, ending the debate over whether TNC miles are closer in risk to personal use, limousine use or taxi use.
This approach also solves the problem of coverage questions when drivers have multiple apps turned on at once. By passing the responsibility of coverage back from TNCs to the driver, the driver’s own insurer will be liable regardless of how many apps happen to be turned on at the time of an incident. To avoid being double or triple charged this ‘per-mile’ premium for each ‘TNC mile’ driven with multiple apps turned on, the TNCs will just need to ensure that they report the exact time periods the app was on so that the insurer can identify and remove any potential double counting across TNCs. TNCs operating in California are already required to maintain “waybills”, which are records of all trips taken by each driver, which can be inspected by the CPUC on demand.
While Usage Based Insurance may be an ideal structure for ridesharing, for other sharing economy business models like carsharing, it is a virtual necessity. Sam Zaid (Getaround) relates his experience in finding a workable insurance solution at the dawn of the carsharing economy:
We need usage-based insurance for our model as we can’t control when an owner makes their car available for rent. We also require a group policy format so that one policy could cover two different sets of parties; drivers and owners. It’s kind of a hybrid commercial/personal model. Because we were creating this novel thing that is also priced in a different way, folks had to get their heads around a lot of different things. If you didn’t have the right senior people at the table, it was just never going to happen. You need to rate things differently and think about things a little differently. Once you’ve made the initial investment to figure it out, it’s not so bad. You’ve got that foundation and you can start to really explore new things. But you have to find a carrier willing to innovate—not many carriers are willing to do that.
Sam Zaid (Getaround) also elaborated on another insurance model that would structurally align with the carsharing model perfectly, where coverage is purchased by and follows the driver, not the vehicle:
It would make a lot of sense if insurance followed the driver and was all usage-based because, structurally, that aligns with our business and probably all the new TNCs. Any person with a driver’s license should have an insurance rating factor and when they hop in a car, you combine the car’s rating factor to compute the base insurance rate they pay. If you’re a risky driver then it’s higher. If you’re driving in San Francisco vs New York then the rate changes.
The difficulty of pricing for homesharing within the framework of existing homeowners policy endorsements is quite problematic, with Joel Laucher (CDI) commenting:
In auto there’s a structure to get separate charges for this type of activity through a class plan. Homeowners isn’t really set up pricing-wise generally to allow for special events or circumstances like occasional renting. But there are always answers or similar situations out there once you start looking around. Vacation rentals and special events aren’t new concepts.
Airbnb’s Host Guarantee and Host Protection Insurance aren’t without their critics, but we believe these are significant steps in the right direction. Including this coverage automatically, and embedding the cost in Airbnb’s listing fee, ensures that even occasional hosts can be covered at a reasonable cost. Contrast this against the Proper Insurance approach, and those like it, which are sold direct to hosts and have a fixed annual rate that is insensitive to how much time the property is rented as opposed to occupied by the primary resident. The Proper Insurance pricing structure is understandable since the alternative would be to somehow price a usage based insurance product with commensurate administrative overhead, but can be cost-prohibitive for part-time hosts.
The Global Sharing Economy
Much of the discussion about the sharing economy to date has been very US-focused. Some of the major sharing economy players such as Uber and Airbnb have expanded globally, offering much the same service model, and met with as varied a reaction as they have in the US. Some countries have outright banned them, while some have accepted them and regulated them. But the common problem in every country faced with sharing economy entrants is dealing with newly created gaps in insurance coverage.
An international view on insuring the sharing economy provides a fresh perspective on how possible solutions could be approached in the US. Graeme Adams (Finity Australia) commented on how Australia has embraced carsharing:
Here we’ve got GoGet, which is renting a car by the hour [similar to ZipCar]. It’s been embraced by local councils. Local councils now are providing locations on street corners for shared cars. Developers are now doing deals with councils so that the design of a block of units would include spaces for shared cars. Clearly it’s in the interests for the developer because one developer that’s developing a development here called Central Park, they’ve allowed 44 spaces for GoGet. That means they don’t have to provide one or two car spaces per unit, so it’s actually good for the developer. We’ve even got the state government using GoGet or shared cars rather than having their own state fleet.
The insurance regulatory regime in countries like the UK and Australia is quite different from that of US states. Insurance product design and pricing in these countries can be modified and iterated with the same freedom as most other industries, with mainly reserving and solvency requirements of insurers being heavily regulated. For most classes; product design, forms and pricing changes don’t need to be filed and approved. The ability for insurers to change their products and pricing structures in response to external stimuli such as the rise of ridesharing and homesharing means that, as long as the sharing service itself abides by applicable regulations, the idea that there could be any systemic coverage gap in a competitive and responsive insurance market is seen as a quaint notion. Graeme Adams (Finity Australia) explains:
Have you given the insurance company what they need to understand the risk they want to take on? Have you declared that you are a ridersharer? When you take out insurance you need to declare if it’s going to be for a business purpose or private use. If you say business purpose then they’ll say, ‘Well what’s your business, plumber or electrician?’ ‘Well, no I do rideshare’. Then they should say ‘well how many times a week?’, ‘how many kilometres?’ etc and they should have a premium for it. Now in some cases they may not. They may just have a standard uplift of 20% because it’s business use. It depends on how sophisticated they are. If you have said ‘yep, that’s fine’ and paid your additional premium I don’t see what the problem is because the insurer has properly assessed the risk and there’s been an appropriate premium struck between the insurer and the insured. So it’s not a problem.
The same goes with homesharing. Have you advised your insurer that you want to have others that you don’t know living in your house for a time? The insurer might say ‘that’s fine we’ll hit you with another premium’. They might say, ‘Look, we’re going to offer you a landlord’s policy for three months while you have Airbnb clients staying’. Again I don’t see a problem with that because the insured and insurer have discussed the appropriate risk and they’ve struck an appropriate premium and they have an opportunity to buy an appropriate product for the insured or landlord.
Jim McNichols (Greenlight Re) provided a forward-looking insight into the direction insurance is taking and the cultural challenges it poses:
I firmly believe, no, I know, that bitcoin, driverless cars, electric cars, drones, they’re coming! We will have synthetic currency. We will have driverless cars. I can tell you that, as a certainty it will happen. The ultimate question is when will the regulatory environment allow it and when will society and insurance catch up with them?
I am from the baby boom generation and we own our cars, homes, albums, CD’s and highend electronics. If you contrast the way that I approach my work and view the economic landscape with how millennials do, it may as well be medieval versus modern. Millennials avoid ownership but do require efficient access (to cars, homes, music, equipment, etc…). Much of this change is going to be forced by the new generation of consumers by not only expecting it to be this way but rather demanding it be this way. As a baby boomer, my mindset is that the efficiencies of the sharing economy may not make much difference on the margins, whereas a millennial is going to think ‘Look, this is the only way it should be done.’ There are four generations currently obtaining homeowners and auto insurance with very different perspectives as to what it is and how it is supposed to perform.
Many of the innovations and cultural clashes touched on by Jim McNichols will inevitably have to be addressed by the insurance industry as new risk exposures and business models that are today inconceivable eventually become commonplace. In the next section we explore how the insurance business model itself becomes one such arena, with well-funded disruptors clashing with incumbents on the battleground of peer-to-peer insurance.
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.