The End of Insurance as We Know It

by Rob Galbraith

  7.For Glass House, the lower premium of $600 offered by Clueless Co. will attract them to switch from the higher premium of Genius Inc.

  8.While Brick House is leaving Clueless Co. and costing them $300 in profit, Glass House is switching to Clueless Co.

  9.Unfortunately, since the true loss cost of Glass House is $700, Clueless Co. is managing to attract unprofitable business at a loss of $100.

  This is known as adverse selection; Clueless Co. is attracting unprofitable risks because they are unable to properly segment between Brick House and Glass House while Genius Inc. is only attracting profitable business. As Clueless Co. loses profitable Brick House for unprofitable Glass House, they must adjust their premiums upward to reflect the higher loss costs of $700, which makes them even more unattractive to Brick House. By contrast, Genius Inc. is shedding unprofitable Glass House and gaining profitable Brick House, putting downward pressure on their rate indications and allowing them to lower premiums. Over time, Competitor B has much lower premiums than Competitor A but is still able to write business profitably due to superior risk segmentation.

  FILLING THE POOL

  How do you know if a subgroup of risks is large enough to consider its loss propensity to be statistically significant from another group? This is determined by the actuarial concept of credibility, which is the theoretical minimum group size that can be used to segment between different risk pools or subgroups with a high degree of confidence. Taking credibility into account, risk segments are defined using actuarial (statistical) techniques such as Generalized Linear Modeling (GLM.) Traditionally, these techniques use observable characteristics (a/k/a variables or rating factors) that are correlated with losses. Examples include age, gender, marital status, location, valuation, and many (many, many, many) others. Some variables are highly correlated with losses and some are not, and statistics determines which are significant (and therefore useful) and which are not.

  As discussed previously, variables that are highly correlated with losses are not 100% guarantees. For example, some 16-year olds are exceptional drivers while some 35-year olds are terrible drivers. However, in the aggregate, 16-year old drivers tend to have much higher auto losses than 35-year old drivers. There is a strong statistical correlation between age and auto losses, so the variable is highly predictive. Also as discussed before, with the advent of telematics, insurance carriers can now go beyond strong statistical correlations between proxy variables such as age and directly “observe” driving behaviors of each driver. This allows carriers for the first time to “see” which 16-year olds are good drivers and which are not.

  TAKE A DIVE

  In the past few years, there has been a lot of press about “earning your own rate” or the personalization or individualization of insurance premiums. This is at odds with traditional approaches. Insureds want more flexibility in terms of coverages, contract terms, and price points; but insurance is fundamentally based on this concept of pooling to evaluate, manage and price risk. In summary, pooling is grouping individuals with other individuals with similar characteristics to leverage the law of large numbers in order to diversify risks and lower costs (and premiums).

  Is personalization possible or fundamentally at odds with the concept of pooling risk, which lies at the heart of insurance? Many large carriers tout their ability to accurately price each risk through a highly refined pricing plan. These plans use a large number of variables to achieve billions of pricing cells. Think of pricing cells as potential “buckets” that risks could theoretically be placed in based on their unique intersection of multiple characteristics. With that many pricing cells and only thousands or, at best, a few million insureds, clearly there are more potential cells or buckets that insureds could be placed in than are actually used.

  Is a rating plan consisting of billions of pricing cells consistent with individualization? Yes and no:

  •You could be the only insured that falls into a particular pricing cell (yes)

  •You could also be placed with hundreds or even thousands of insured with the same or very similar characteristics (no)

  Simply because you are in a pricing cell by yourself, the idea of “earning your own rate” suggests some ability to demonstrate through your actions the ability to lower (or raise) your premium. Perhaps the best example of the “earn your own rate” concept are the auto insurance providers that charge premium by miles driven rather than a time period (such as the standard 6-month policy). In these instances where carriers charge a per-mile rate rather than a time-based premium, the insured can directly impact their premium by deciding how much (or how little) to drive.

  JUMPING OFF THE DEEP END

  Going beyond this form of individualization, what about the ability for insureds to select their own riskpool? At its essence, this is a return to the foundational roots of insurance where insureds personally knew each other and agreed to indemnify each other in the event that one or more of them incurred a loss. One historical example of this is USAA, which was formed back in 1922 by 25 Army officers stationed at Fort Sam Houston in San Antonio. At the time, military officers were considered a poor risk by insurance carriers of the day, so the 25 officers started their own arrangement to insure each other in the event of an auto accident.[220] The scheme worked pretty well. Today, USAA has over 12 million members and has insurance revenues of over $20B.[221]

  Modern examples are generally classified as “peer-to-peer insurance” arrangements and are very similar to peer-to-peer lending arrangements such as LendingClub or Prosper. In P2P lending individuals and business seek loans from individuals known as investors. The investors pool their funds together to provide a large enough source of capital to provide larger loans than any of them individually could provide on their own. The platforms or marketplaces typically take a small percentage commission such as 1% on each loan origination. Benefits to borrowers include potentially lower interest rates and more flexible terms than would be offered by a traditional bank. Benefits to investors include a way to earn greater returns than a traditional savings account and a way to diversify their investments from traditional stocks, bonds and mutual funds, plus the opportunity to direct their lending to socially-minded startups.[222]

  Could this P2P model translate to the insurance space? Some companies are already trying to make it work. Friendsurance, based in Germany, works with insurance providers but can pool risks together in a way that has a peer-to-peer element. Some of the money paid by consumers goes to the insurance company for protection but the rest goes into a separate pot to insure the group (on average around 10 people). If no damage occurs, some money is returned to the customer from the pot reserved for the group. If damage does occur, there is no increase in premium and no additional deductible is owed: it instead is paid from the pot. How does Friendsurance make money as an independent broker? They earn commissions from the insurance carriers that participate.

  Another startup insurance business model in the P2P world is called VouchForMe (formerly InsurePal). VouchForMe leverages blockchain and social proof endorsements that allows insureds to ask up to 10 friends who know them vouch for them by making a financial commitment to back them. In turn, those insureds earn discounts on their insurance premiums with the more endorsements they receive, and those who agree to endorse their friends also receive tokens that can be used for discounts, services or exchanged for cash.[223]

  What are some potential benefits of a P2P insurance model?

  More information

  •By selecting individuals, small businesses or other insureds to create your own risk pool, you may have more information on their loss propensity

  •A traditional insurance carrier only has the information gathered from the insured and third-party data from the application process

  Less fraud

  •A pool that has some familiarity may be less likely to submit phantom or inflated claims

  •In theory, a group with direct, personal knowledge of each ot
her or at least a sense of shared identity is less likely to knowingly enrich themselves at the expense of other members of the group

  Reduced expenses

  •By recruiting others to join the risk pool, the costs associated with advertising and marketing in order to attract new customers may be reduced

  •Customer retention may be improved through the affinity to the group, reducing the incentive to shop around

  These P2P arrangements can also “lower the expense floor” for insuring lower-valued items. Traditionally lower-value items are prohibitively expensive to cover due to large expenses involved. P2P can facilitate customization as the group can agree to practically any terms.

  WHY RISK SHARING COULD BE THE NEXT RIDE SHARING

  Another possibility for reducing expenses in the insurance market related to regulation is if these products are designed to be quasi-insurance.These are a class of products which share similar risk transfer characteristics while stopping short of being true insurance contracts governed by state insurance regulators. (Think back to extended warranty contracts.) Such products would have to be different than the traditional “take it or leave it” contract of adhesion that require approval by DOIs. In short, while peer-to-peer insurance is a tiny fraction of the market today and well behind the development of peer-to-peer lending, the upside potential is quite attractive and could allow customers to truly choose their own riskpool.

  What are agencies, carriers, insurtech firms, startups and VC investors seeking? All want to see major improvements in the status quo. This can be achieved by creating large efficiency gains and new products and services that provide a needed upgrade in customer experience. The first three parts of this book have described the broad problems in the P&C insurance industry generally, and highlighted some major technologies that could close the existing gaps. Is merely closing known gaps in the current insurance model thinking big enough? Is it possible to describe an alternative business model that goes beyond what most P&C industry experts predict, yet could be entirely plausible and has some precedent in other industries?

  Does disruption of the taxi industry by ride-sharing companies provide a close parallel to P&C insurance? Here are some similarities:

  1.Both exist to solve a common problem that consumers need a solution for

  2.To protect consumers, a regulatory framework was created and evolved over time

  3.The presence of fairly strict regulations provide some meaningful benefits to consumers

  4.Regulations also led to artificial barriers to entry and stifled innovation while serving to protect existing players from a slew of competitors

  5.This stagnation led to complacency among providers and a less-than-ideal customer experience

  6.Purchase decisions were made less as affirmative choice but rather as an obligation due to a lack of good alternatives

  MOVING CHEESE

  Technological advances and the rise of an entire new business paradigm - ride sharing - was the key to overcome a stagnant industry that for decades failed to adequately meet customer needs. As a result, the new alternative business model was rapidly embraced as a superior alternative by many consumers. Ride sharing quickly has quickly become the preferred way to achieve customer short-term transportation objectives and has established itself as the new normal.[224]

  If ride sharing was successful enough to overtake the taxi industry in terms of average daily ridership in New York City,[225] could a new “risk sharing” business model do the same to disrupt the insurance industry in less than a decade? Let’s re-examine some of the reasons why the insurance industry exists in the first place and has persisted for so long.

  Customer needs

  •Risk aversion

  •Exposure to large financial loss

  •Lack of ability to self-fund to recover from large financial loss (“self-insure”)

  Stable funding source to ensure money is available to cover losses that occur

  •Mechanism to create “pool” of funding to cover losses by a few, taking advantage of the law of large numbers

  •Specialization of actuaries, underwriters, and claims adjusters to ensure premiums are sufficient, terms & conditions of contracts are appropriate given the rates and that fraud is minimized while quickly settling legitimate claims

  Enforcement mechanisms to build trust in system

  •Creation of binding contracts enforceable by legal system to overcome absence of trust with third-party to pay insured claims after premiums paid up front

  •Representing insureds when harmed by a third-party that is covered by another insurance carrier and facilitating settlement between the parties

  •Claims adjustment process to ensure legitimacy of losses and prevent fraud

  •Regulation to ensure insurer has adequate funding to pay claim obligations

  A key question here in the 21st century: can these important tenets that underpin the industry we know as P&C insurance be accomplished a different way today with the advent of new technologies?

  •Sensors and cloud storage to generate and store “big data” to provide a granular and continuous stream of data to monitor for losses

  •The combination of data science, artificial intelligence and machine learning to quickly identify changes in state and report or even prevent losses

  •Crowdfunding through social media and P2P networks

  •Blockchain and parametric triggers to create “smart contracts” that resolve ambiguity, reduce expenses and time from FNOL to claims settlement (ideally, almost instantaneously) and build trust

  RISK TRANSFER RE-IMAGINED

  No doubt the devil is in the details, but the idea is that conceptually an aspiring entrepreneur (or, more likely, a set of entrepreneurs working independently but increasingly in a coordinated fashion) could build a “risk transfer ecosystem” that has insurance-like characteristics (the benefits) without some, most or all of the associated pain points (the drawbacks). In other words, the development of a new market consisting of insurance-like products that retain the positive elements while removing some or all of the drawbacks of our current insurance ecosystem. The idea of risk sharing is an attempt to outline what one such alternative scheme might look like. The more options that can be conceived, the more likely one of them will become not just a reality but become the dominant risk transfer paradigm in the latter half of the 21st century.

  1

  CHAPTER 23 - PLACING BETS IN THE INSURTECH CASINO: A GUIDE

  BETTING STRATEGY

  Whether you are new to insurtech or closely follow industry developments, a quick Google search reveals an overwhelming amount of information: press releases, industry trade publications, social media posts and a proliferation of conferences and meetups. How does a savvy VC investor, startup founder, C-suite executive at a traditional carrier, an agency owner, third-party provider to the industry and others sort through what is hype and what is real? This chapter is intended to provide a framework for thinking about the current world as it rapidly evolves. Regardless of the lens you look through, the goal of this book is to set you up for success in gauging future developments in insurtech, deciding which few of the thousands of new announcements weekly are worth your time and attention to understand at a deeper level, and confirm (or alter) your current view of the P&C insurance world.

  One mental model that can yield valuable insights is the classic 2x2 grid common among many consulting firms that represent two axis: the size of the opportunity being evaluated and the scope of the opportunity being evaluated.

  Both scales evaluate potential new products or services enabled by insurtech on whether it is enhances or disrupts the current insurance paradigm in the following ways:

  The size of disruption scale

  •Does a new product/service improve on existing methods and processes (enhances) or does it replace them with new methods and processes (disrupts)?

  •Does the new product/service build on those offered by traditio
nal players (enhances) with traditional players or compete with them (disrupts)?

  The scope of disruption scale

  What is/are the part(s) of the insurance ecosystem that will be impacted by this new product/service? Is this a course correction or paradigm shift? Consider the following areas that may be impacted:

  •Distribution

  •Marketing and lead generation

  •Quoting and the pricing process

  •Binding coverage (issuing the policy)

  •Servicing, back office and underwriting

  •Claims

  •Compliance

  •Agility and speed to market

  After rating the size and scope of the potential disruption, you need to perform an initial evaluation of the potential success of the new product/service, then periodically return to confirm or re-evaluate your initial assessment.

  INCREASING YOUR ODDS

  What factors should be part of your assessment? Every investor and founder has their own way of approaching this process. I do not have a template to offer to print off and fill out, only some thoughts on what should be part of any assessment. Here are some questions that should be addressed:

  Amount of competition - is this a “blue ocean” idea or yet another entrant in an already crowded field?

  •Do not rely on snap judgments: test your thinking by consulting with other experts inside and outside of your particular discipline.

  •There are often early-mover advantages in insurance: being first may not be critical but after there is a certain amount of industry adoption, it becomes much more difficult to change existing products and services unless there are significant pain points with those solutions (not simply annoyances or incremental improvements).