Amazon runs an auto insurance program for its delivery service providers, but if those DSPs choose to buy their insurance from competitors instead, Amazon is fine with that.
In fact, Amazon might prefer it if their insurance customers buy coverage from competitors instead, the head actuary of the giant retailer suggested during a session of the Spring Meeting of the Casualty Actuarial Society last month.
Eric Schmidt, who started his career as an actuary at Allstate and later moved on to a product management role at InsurTech Clearcover before landing the lead actuary job at Amazon, also explained why the giant retailer would hesitate to raise the auto insurance premium of a driver that takes a risky but efficient route to “deliver smiles” to buyers of online merchandise.
Schmidt explained that DSPs are the small business owners who contract with Amazon to do all the last mile deliveries. “At the end of the day, the success of the DSP program is critical to Amazon’s mission,” he said, noting that Amazon’s Logistics program and the DSP program were created because FedEx and UPS couldn’t do what Amazon wanted to do for customers.
“We need DSPs to be successful.” Being successful from a financial standpoint means making money—and having access to affordable insurance. “That’s why there is a program through Amazon to offer insurance as a value-added service.”
Highlighting some distinct characteristics of the Amazon insurance program for DSPs, Schmidt first noted that because DSPs are independent contractors, they are not required to purchase the insurance Amazon offers. “In fact, in a lot of ways, if other competitors want to come in and offer a much better rate for our DSPs, that’s actually a really positive thing because that goes to that goal of lowering the overall cost in the last mile ecosystem.”
“So, in that way, you can think of what we’re doing as a backstop—I don’t like thinking about it as a backstop, but in a way [we just] want to make sure that there’s available and affordable insurance for DSPs.”
“We’re a little bit different, as well, from the traditional carriers. Our goal is not to make money on insurance. That’s not why we’re here,” he said, going on to give an example of how the two factors—the customer profiles and the absence of an insurance profit goal—allow Amazon to think about insurance differently.
“If we’ve got a DSP that is used to running one kind of route, [and] we decide we’re going to route them completely differently—let’s say that different route goes through a much riskier area, but we don’t have a mechanism for being able to compensate that DSP differently for taking on that different route. Then we’re going to really think twice about how the insurance program responds to that change…”
“If it’s something that a DSP can’t control, we don’t think it’s necessarily fair to be able to pass that cost on to DSPs. Let’s say that route does change and the insurance premium goes up—at no control to them. That’s a really bad experience to our DSP—to our customer.”
“At the end of the day, the DSP is the customer, and making sure they’re satisfied and that they’re profitable, the success of the DSP program is critical to Amazon’s mission,” he said.
Schmidt was one of four Fellows of the Casualty Actuarial Society who participated in the session titled “Emerging Risks in Mobility,” fielding questions from Panel Moderator Ryan McMahon, vice president of Cambridge Mobile Telematics, about actuarial activities at non-insurance companies like Amazon.
Like Schmidt, Joshua Pyle, senior director and head actuary for DoorDash, talked about the risk implications of rerouting drivers. “There are things like operational efficiencies that we could point to, or batching. If you think about what’s being delivered, things like pizza may not lend themselves to that, but can you batch orders in a way that actually makes routes more efficient or effectively reduces the number of miles driven per delivery? That’s a direct reduction of your risk,” he noted.
Pointing to another factor that changes the commercial auto risk profile of deliveries, he said, “One interesting evolution for us is just modality of delivery. As we think about expansion from food to say convenience or grocery or something else, [then] depending on the geography, you could be delivering this via car, via motorcycle, scooter, even walking,” said the actuary who previously worked for Liberty Mutual, Fireman’s Fund and CSAA Insurance Group, and also spent time working on cyber catastrophe modeling at Symantec and CyberCube.
At Cruise, GM’s self-driving vehicle company, Corey Rousseau, an actuary partner on Cruise’s risk management team, is focused on the mode of delivery moving in the other direction—toward more automated methods. “Probably the No. 1 thing that we have to think about that’s going to impact our trends is the technology improvement.”
With traditional auto, “human driving doesn’t change materially from year to year,” Rousseau said, (noting the exception of the pandemic, which did alter human driving patterns). In the world of self-driving commercial vehicles, change is continuous—and “trying to understand how our tech improves and then, as an actuary, [you’re] trying to think about how you use your historical experience, which becomes far less useful when your technology stack has improved and the vehicle is safer than before,” he said.
“What’s your correct complement of credibility?” he asked, referring to the process actuaries use to weight different pools of loss experience data in their pricing models.
Rousseau also noted that the claims resolution process in the AV space has some unique characteristics. “We have an in-house claims team.” While a lot of traditional insurers have those too, at Cruise, “when an event happens, we [also] have all the data and camera images of everything that happened. [That] will have a big, big impact on our trends compared to the industry,” he said, referring to the likely faster speed of the claims resolution at Cruise.
Cruise also has control over its exposure base, he said, noting that the company can choose what part of its fleet to operate, where to operate it, and at what time of day. The actuaries, by “being in tune with the business plans of the organization,” can derive “a good understanding of where [the] exposure will be, which is sometimes on the [traditional] carrier end of things maybe a bit of a mystery sometimes,” Rousseau said.
Schmidt made a similar observation when describing Amazon business decisions that push the boundaries in terms of delivery time and on-time performance and minimize the costs associated with the last mile network ecosystem. The company, he said, is “constantly tinkering with how to kind of optimally utilize the last mile fleet,” choosing between less routes, higher duration and more routes, lower duration.
For actuaries, “it’s fun to be a part of the conversation proactively to determine [whether you should] make those decisions or not, or how to test out different hypotheses…That’s a completely different conversation than what happens on the traditional insurance side. If you were a traditional carrier insuring Amazon DSPs, you wouldn’t know that there was a shift in the fundamental risk exposure until it actually hit your books. You’re playing catch-up and you’re being reactive [at a traditional carrier] versus on the program side, [where] we’re able to be proactive [in] thinking about the risk evolving and what implications [business decisions have] on the insurance risk.”
Emerging Mobility Risks
McMahon, whose prior experience in the industry was in claims and marketing roles for traditional insurers rather than actuarial work, gave the perspective of his current employer, Cambridge Mobile Telematics, on emerging risks. “As exotic as self-driving cars are for sure—it’s a completely new category of risk with lots of new emerging elements to it—I would argue that we’re all in that environment right now, post-COVID,” where the risk landscape has rapidly changed. “For everyone that prices auto risk, it looks nothing like it did in 2019. Nothing at all, just from a pure telematics standpoint. Distracted driving is up 30 percent from where it was in 2019, which is up from where it was in 2018 and 2017.
“So, that is a completely different risk environment that has no historical trends behind it. And that’s plus inflation and everything else.”
McMahon coaxed his panelists to describe an array of evolving risk and liability challenges in mobility, including those related to navigating through Transportation Network Company regulations, cyber risks for self-driving cars—and around curious onlookers who jump out in front of Cruise vehicles to take selfies with the cars in the middle of the street.
Cruise’s Rousseau also referenced the question of accident liability for owners of autonomous vehicles, who may sit behind the wheel of driverless cars but not drive them—a question he said will be “answered over time as the technology integrates.”
For Cruise, where the business plan is currently focused on last mile delivery services, the answer is clearer, he suggested. “We’re the service provider. The driver—it’s our vehicles, our technology. So, there’s potentially no question of liability for someone else involved.”
Schmidt noted that Amazon’s Logistics program, putting a battalion of DSPs out on the nation’s roads, may be having an impact on personal lines insurance risks. The program was created because “there wasn’t enough capacity in the existing last mile system to be able to deliver what Amazon wanted to deliver to customers—moving from two-day delivery to same-day to two-hour,” he said, noting that Amazon Logistics has grown to the point where last year Amazon overtook FedEx and UPS as the largest logistics operations in the U.S.
It’s hard to know exactly how that’s shifted the way that people live their daily lives, he observed. “How many trips to the store no longer happen because you get your packages dropped off right at your desk?” asked the former Allstate actuary. “Maybe you operate differently, right? So, you’re making different trips. Rather than going to the store, you go to the park or something like that,” changing the nature of personal lines risk, he said.
Actuaries in Mobility Tech: A Good Fit
McMahon started the session by asking the actuaries to describe their experiences with technology companies in trying to bridge cultural knowledge gaps about insurance concepts. He encouraged audience questions, which focused on the skills needed for actuaries who take on important roles at non-insurance companies.
Clive Thompson, vice president of mobility of INSHUR, deals with professionals whose knowledge of the insurance industry ranges from minimal to expert in his role at a niche InsurTech MGA focused on rideshare and delivery markets of the gig economy. INSHUR’s mission “to support drivers who power the world” means customers could be any entrepreneur focusing his or her livelihood and income around a vehicle, he said, noting that his firm is crafting insurance solutions for the DSP-type businesses of Amazon, as well as the individual contractors who drive for Uber and DoorDash.
“I find myself bumping into loads of people who don’t have insurance backgrounds and trying to synthesize difficult insurance concepts. One of our stakeholders is just quite simply our investors. Our investors are VC animals, they’re finance animals, but they don’t necessarily understand what a loss ratio is and why that’s so important.”
INSHUR also seeks insurance carrier partners that have very sophisticated insurance teams. “But if we’re talking a small startup, maybe one of these new kind of last mile delivery solution providers, they don’t have an insurance department, or they don’t understand the insurance needs of their drivers.”
Thompson believes an actuarial background prepares CAS members to answer not-so-difficult questions about trend and reserving patterns, while also “trying to find some really pragmatic solutions” for customers across the mobility spectrum. Giving an example, he said, “It’s amazing to hear just how much people care about ACORD forms in digital format at scale. I know it’s not necessarily what an actuary does day-to-day,” but that’s one of the things that INSHUR actuaries faced. Then, “we still have to procure capacity, bring a product to market, make sure it has a sustainable loss ratio,” he said, referring to INSHUR’s core activities. “But you get to do a little bit of both of that every day.”
Schmidt, noting that the DSP program at Amazon is relatively new, shared an insight about the risk progression of the program that non-actuaries overlooked. “As you would imagine, as we bring on new DSPs, they don’t start off with huge fleets of vans. They start pretty small, and then they grow over time as they get more acquainted with the business. So, it would be really important that you have a mechanism to be able to pick up on additional exposures and get them added to an insurance plan.”
“That seems obvious to everyone in this room, but [at a technology company], somebody has to be there to say, ‘Hey, this is an actual risk. These exposures are growing. Are we making sure we’re capturing all of that exposure?'”
Even though the insight “has nothing to do with actuarial science,” actuaries can add value into tech companies by drawing on their general knowledge of the insurance industry and the nature of risk, he suggested.
At DoorDash, Pyle noted that no one on his 17-member team, which includes five actuaries, has been with the company for more than a year-and-a-half. There were risk management employees across the organization previously, who procured insurance and hired outside actuarial consultants to do some reserving studies. “This really does touch every insurance line,” he said, later revealing that DoorDash set up a captive in 2020, without indicating what type of insurance is written through it.
“We’re really getting to a point we’re switching from risk management to risk mitigation.” Like Cruise, DoorDash has “a short period of a ton of data. We have insights that we can glean from exploratory data analysis, and we’re really starting to engage other groups within the company, whether that’s mapping or the assignment team, the product team, the strategy team, to really turn that into something actionable,” he said.
Later, Pyle spoke about the need to share insurance knowledge throughout this organization. “One thing I’ve found extremely useful is just educate, educate, educate—doing that through formal memos and, importantly, through shared docs where different teams across the company can just jump in, read through, ask a hundred questions about insurance.”
Pyle said he has also found it useful to create very short videos on different insurance concepts, noting that new people are continually joining mapping, operations, finance and other teams, and the same questions will just come up again and again. “That’s certainly true with all these companies. They are growing quite rapidly,” he said, suggesting that videos and shared docs are scalable and efficient education tools.
Schmidt schooled actuaries in nontraditional tech company roles to learn the lingo of internal stakeholders first, and then to work backwards from there to deliver the value of their knowledge to their businesses. “So, for example, there’s no such thing as telematics data at Amazon. It’s safety data,” he said.
Pyle suggested that actuaries are predisposed to be good at nontraditional jobs. “We’re used to teaching ourselves things,” he said, noting that he knew nothing about cyber when he took his previous job in cyber cat modeling.
“I really think it’s about being resourceful, independent, unlocking the value within the organization, because there won’t be that definition of what you can and can’t do. You just have to operate. You have to build relationships, find out what other people do and find out how what you can do to benefit other groups.”
INSHUR’s Thompson said actuaries dealing with non-insurance folks “need to build up a mechanism of patience. “We take for granted how big and complicated insurance is,” he explained. “We live and breathe it every day…I don’t think there’s anything else like it in so many different dimensions,” he said, advising actuaries to take the time to craft complex concepts into simple messages. “If you enjoy that, that’s one of the big challenges,” he said, responding to an audience member who wondered if he had the right skills for a nontraditional actuarial role.
Thompson, who previously worked as a pricing actuary for Zurich and a casualty underwriter at AIG, also shared a sense of what the transition is like going from a big incumbent to small InsurTech. “When we’re in traditional work, there’s a history, or there’s a book of business, and you’re tinkering with that,” he said, contrasting the “tiny wins” of building new products and processes at InsurTechs.
“It’s going to feel so minute, because we’re used to looking at billion-dollar balance sheets,” he said. Even though it isn’t enormous or a problem isn’t yet solved, “you’re not failing; you’re building,” he said, encouraging a different mindset for actuaries who pursue nontraditional roles.
Pyle agreed. “Getting used to imperfection, that’s something that actuaries are not very good at. But you can get used to it with time,” he said, noting that getting accustomed to incremental launches and MVPs—minimum viable product—are part of the mindset shift required.
Was this article valuable?
Here are more articles you may enjoy.