The Future of Insurance
The future of insurance is embedded. Today, customers increasingly demand convenience and self-service when transacting. While most other industries have risen to that challenge, insurance has lagged behind. Insurers sell insurance to consumers who vary in their risk whereby high-risk consumers are more expensive to insure. As a result, competition among insurers is focused not only on selling more insurance policies, but also on identifying and attracting buyers that are less costly to cover. This is known as a selection market.
In conventional markets, sellers are only concerned about how consumer demand affects how many units they sell and are unconcerned about which customers buy their products. Consumer demand and producer cost are independent objects. For example, the bottom line of an electronics company would be unaffected if Mary or Bob bought their cell phones. However, in selection markets like insurance, consumers vary not only in how much they are willing to pay for a product, but also in how costly they are to the seller. Because both buyers and sellers incur costs in searching for a counterparty, this selection process represents one of the biggest costs of the traditional insurance model.
In the traditional selection process, insurers must acquire customers—an expensive process involving advertising, sales agents’ commissions, and the time-consuming task of underwriting risks associated with each new customer. Together, these costs are known as customer acquisition costs, or CACs, and must be included in the premium charged to the average customer so that the insurer can break even in expectation. Today, most customers are acquired through a switch-and-save model whereby insurers offer lower premiums to customers of other insurers to switch to their services. This model is particularly costly because it relies on costly advertisement campaigns and is detrimental to the insurer’s current customers, who must pay an increasingly high premium for the same level of coverage. As a result, many customers choose to self-insure or opt for suboptimal levels of coverage. From the insurer’s perspective, however, these tend to be the least risky customers. In the long run, switch-and-save is a zero-sum game because ever-increasing advertising budgets are the only way of growing the insurer’s market share.
With customer acquisition costs at effectively zero, premia should reflect only expected claims expenses, taxes and administration costs, so that almost all customers will be willing to acquire insurance. The resulting market is as large, and therefore as low risk, as it can be. Moreover, the switch-and-save model restricts the market to customers that are ex ante searching for insurance. Embedding not only allows for the expansion of these established markets but also can create insurance markets that wouldn’t have otherwise existed by minimizing frictions and offering coverage to potential customers at the point where they are most likely to buy them.
This white paper demonstrates how embedded insurance can greatly reduce the friction of this selection process and transform insurance into a traditional market. It starts by considering the shortcomings of selection markets and historically contextualizing the erosion of trust in the insurance industry, conceived as a trust-based community resource. It then demonstrates how embedding can help rebuild this trust by creating a virtuous cycle whereby convenience and trust act as complements, rather than substitutes. After overviewing the economics of insurance, the white paper explains how the embedded insurance model both shifts the supply curve down and the demand curve up. Subsequently, it shows how these supply and demand curve shifts increase the total surplus and expand the total market for insurance. By using single price versus price discrimination model, it demonstrates who will get this surplus and why. Finally, it concludes by suggesting how embedding could create markets that wouldn’t have existed in its absence by expanding the production possibility frontier.
Section II: the Shortcomings of Selection Markets
The problem of adverse selection notoriously plagues the insurance market.
In the context of insurance, customers choose among the set of contracts offered by the insurer according to their expected probability of needing these services. This knowledge is asymmetric because consumer type is private information and is unobservable by the firm: the customer has more information about their probability of loss and/or distribution of the size of the loss than the insurers who underwrite the policy. This is one of the largest sources of inefficiency in insurance today. For example, individuals who expect higher healthcare costs differentially prefer more generous health insurance plans than those who expect lower costs. Extremely low risk people—such as young, healthy adults—often choose to self-insure and opt out of the market. These concerns are at the centre of debates on Medicare reform and changes in requirements for mandatory auto-insurance.
Economists have considered the problem of adverse selection since Rothschild and Stiglitz (1976). The paper considers two groups, identical in every respect except their risk probability. They show that under certain assumptions, there is no “pooling equilibrium”. In other words, there is no contract which is optimal for both groups. Because of the costs associated with their risk profile, high-risk individuals cause an externality: the low-risk group is worse off than they would be in the absence of the high-risk group. However, the high-risk individuals are no better off than they would be in the absence of the low-risk individuals. Even though the problem of adverse selection has been studied both theoretically and empirically for almost fifty years, the contracting model used by the insurance industry is still greatly subject to these inefficiencies.
We consider the economic consequences of adverse selection by looking at a simple supply and demand graph. The insurance price (PAS) and quantity (QAS) are given by the intersection of the demand and average cost curves, known as the equilibrium point. A market for insurance exists for all prices higher than the equilibrium price. The further to the left, the higher are the margins made by the insurers because the customers are willing to pay more than the cost of supplying the insurance. Moreover, a market does not exist for all prices below the equilibrium price because the cost to insure low-risk customers is higher than their willingness-to-pay, which depends on those customers’ perceptions of their own risk.
Adverse selection affects the average total cost curves by increased expected underwriting loses. The insurer’s costs can be broken into four components, namely: claims, tax, administration and distribution, or customer acquisition cost (‘CAC’). Of these, claims (expected underwriting loss) and tax are also increasing in buyers’ risk type. Overall, unit costs are increasing in buyers’ risk type. Therefore, the average cost curve is downward sloping. Because the expected underwriting losses from insuring low-risk customers are small, the CAC accounts for most of their marginal cost. Note that the marginal cost is lower than the average total cost used in the calculation of the premiums.
Adverse selection in insurance markets can therefore lead to four categories of inefficiencies. First, prices offered to clients do not reflect marginal costs. Assuming that individuals consider costs and benefits when choosing a contract, they will be allocated the wrong plan for them. Second, adverse selection increases premium variability, which leads to risk-sharing losses. Third, adverse selection forces insurers to manipulate their policies to improve their mix of insureds. Finally, the main customer acquisition strategy used by insurers fails to capture lower risk customers that aren’t already in the market for insurance. Only higher risk customers or very risk averse customers that are aware of the need for insurance will actively seek out insurers. Because of the obstacles to the acquisition of insurance, lower risk people will not go out of their way to seek insurance or won’t even consider the potentially low probability of needing insurance in the first place.
Section III: Rebuilding Trustworthiness in the Insurance Industry
Section III(a): the Trust-Based Origins of Insurance
Trust in the insurance industry is at an all-time low. The insulation from risk and loss that insurance services provide has become a privilege that far fewer people can afford. Moreover, those who can afford coverage don’t trust their insurer to act in their best interest. In this section, we will explain (i) how insurance was originally conceived as a community resource based on trust, (ii) how the current insurance model incrementally eroded that trust, and (iii) how embedded insurance can rebuild trustworthiness in the industry.
Insurance is as old as human society itself; it has been an important element of commerce and social economy for almost two thousand years. The first records of insurance date back to Chinese texts from 700BC, where public granaries were used to indemnify members against famine. Members contributed cereal to a communal reserve, which was sold at affordable prices during periods of scarcity or when prices were inflated by local traders when commodities were in short supply. Subsequently, Greek ‘friendly societies’ in 600BC organized for the purpose of extending aid to their unfortunate members from a communal fund and to provide for the expense of a fitting burial for its members. Roman Collegia offered an early prototype of health and life insurance, covering the costs of treatment for the sick and providing care for families of deceased members.
More recently, guilds played an influential role in many European countries from the eleventh to the eighteenth centuries, possessing many of the characteristics of the modern mutual benefit association such as insuring its members against sickness and old age and furnishing indemnity to those who had suffered loss by fire. Records of the Florentine Chamber of Commerce from 1318 show the contractual terms and prices paid to insure goods subject to peril of transportation either on sea or land.
All these arrangements were based on trust-based relational contract: in the absence of regulation, members needed to trust that paying into these arrangements would insure them against the materialization of some future risk.
Section III(b): the Erosion of Trust in the Insurance Industry
The modern insurance marketplace dates back to the founding of Lloyd’s of London in the late 1600s. The establishment was conceived as a meeting place for sailors, merchants, and shipowners to meet and access reliable shipping news and eventually began selling marine insurance. Over the following centuries, the marketplace expanded to a wide portfolio of insurance and reinsurance categories not just in England, but also in the United States.
The business of insurance was brought to the US through the Philadelphia Contributionship, co-founded by Ben Franklin in 1752 to insure houses against loss by fire. It set new standards for construction because it refused to insure properties it considered fire hazards and these standards eventually evolved into both building codes and zoning laws. With the expansion of American commerce following the Civil War, came the concurrent growth of the insurance industry, both in terms of size and diversification. By 1880, per capita insurance spending in the United States equaled that of the United Kingdom. In the absence of effective regulation, fraud scandals became widespread: dividends were declared that had not been earned, reserves were inadequate, advertising claims exaggerated, and office buildings erected that cost more than the total assets of the company.
By WWII, to combat the impact of wage freezes, employers sought to attract workers by offering group life and health insurance as employee benefits. These group offerings were priced so that only large corporates could afford them. These companies, in turn, sought cover only from major insurers, driving smaller providers out of business. Major insurers soon started offering almost indistinguishable products and had to compete on price.
The scale of these major insurers also depersonalized the insurance acquisition process and the relationship between the insurer and the insured came to be intermediated by brokers. Clients customarily had long-term relationships with their brokers, who were often a one-stop shop for all insurance categories. Trust was shifted from the insurer to the broker, who was expected to act in the client’s best interest. Brokers were paid on commission, which could be as high as 200% of the premium; this commission-based model came with a huge risk that the broker would put their earnings ahead of what’s right for their customer.
Section III(c): the Problem with Switch-and-Save
The most profound change in the U.S. insurance industry in recent years has been propelled by the growth of the internet. Instead of meeting with an intermediary, clients increasingly turned to online platforms to purchase insurance. This increased access to information came with important caveats. In the absence of a broker to perform an advisory function, people only search for insurance upon acknowledging that a risk is significant enough to require insuring against. Moreover, instead of turning to one person or company for all of their insurance needs, the internet allowed future customers to compare policy prices and coverage prices at the click of a button. Propelled by these changes in the search for insurance, insurers gradually moved to a “switch-and-save” model. This model, however, both eroded the trust between the insurer and the insured and led to economically suboptimal outcomes for both parties.
Switch-and-save erodes the trust between the insurer and the insured. To break even in expectation, the insurance company must drive up the price charged to its current clients to cover customer acquisition costs. Because of this dynamic, both parties treat their relationship as temporary. YouGov research found that 73% of policyholders say they’re actively shopping around, and only 6% say they intend to “stick with their existing insurer.” From its historical origins, the relationship between insurers and insureds was trust-based. Insurers were incentivized to absorb short-term costs in lieu of tarnishing their reputation. For example, when a major earthquake struck San Francisco in 1906 and destroyed its regional records, Lloyd’s underwriters famously instructed settlements to be paid out solely on the words of the claimants. Reputation was a long-term game and trust was a necessary condition for its maintenance. Switch-and-save, however, turns that model on its head and places a burden on clients to be constantly switching between providers to avoid paying more for the same product. As John Campbell highlighted in his Presidential Address to the American Finance Association in the US mortgage market, customers consistently pay the price for failing to renegotiate their contracts.
In the long-run, switch-and-save is suboptimal even for clients that are willing to switch insurance providers. The process normally involves long forms and having to recall trivial details about your life. Once a policy is signed, customers are unlikely to want to think about insurance again. Economics research has long recorded that consumers in various markets stick to products they have previously purchased and prefer to buy brands they are familiar with even when cheaper alternatives of similar quality are available. This stasis is rooted in switching costs, inattention due to information costs, risk aversion, and quality uncertainty. Moreover, the more options presented to subjects, the more likely they are to stick to defaults or follow the "path of least resistance". Because nobody derives joy from shopping around for insurance and most people are willing to pay a premium to stay put, customers need to be actively nudged to switch.
In the long-run, switch-and-save is also suboptimal for the insurer because it only engages higher-risk customers who are already in the market for insurance. This pool includes three types of customers. First, customers that are required to have insurance, e.g. homeowners applying for a mortgage. Second, risk averse customers who believe they need it. Third, higher risk customers who know that they are better off being insured. This leaves out low risk customers who would either not have searched for insurance or whose reservation price was lower than the premium charged by the insurer. Under the switch-and-save model, insurance providers must then convince a smaller and higher-risk pool of potential customers to forgo their current insurance policies and switch. The entire process is an expensive uphill battle: investment in advertisement translates into higher costs and, therefore, higher premiums paid by insureds. By shifting the average cost curve up, switch-and-save shrinks the total market for insurance. The higher premium charged surpasses some customers’ reservation price, forcing them to self-insure or to choose a suboptimal level of coverage.
Finally, even in a theoretical world where switching is costless and, thus, every customer switches insurance providers at the end of every contractual period, the switch-and-save model would still be a zero-sum game. Insurers would still have to compete over the same limited pool of customers that are already in the market for insurance.
In short, we live in a world where customers increasingly expect more efficient solutions to service provision. For example, a millennial would be hard-pressed to remember the last time they had to visit a physical bank branch. The insurance industry, however, still relies on an obsolete model that is convoluted, prohibitively expensive, and counter-cultural, resulting in consumer resentment and mistrust.
Section III(d): Rebuilding Trust Through Embedded Insurance
While switch-and-save has become the status quo model of insurance provision, it is not inevitable. In this section, we will demonstrate how the embedded insurance model can rebuild trust in the insurance industry by overcoming the problems brought about by switch-and-save. Embedded insurance bundles coverage within the purchase of the product. Instead of selling insurance to customers on an ad hoc basis, it is incorporated as a native feature of the sale process itself. By doing so, embedded insurance bypasses customer acquisition costs by offering coverage at the point of need. By making the search process frictionless, this significantly increases the pool of customers buying insurance. It also counters the adverse selection of marketing-driven acquisition models and so, by adding low-risk insureds to the carrier’s pool, they can offer lower-cost policies to both low- and high-risk customers. This decrease in cost, in turn, shifts down the average cost curve, attracting even more clients who previously self-insured.
The shift to embedded insurance is attractive to the insurer and the insured in both a short- and long-term horizon. It is attractive in the short-term because, through overcoming customer acquisition costs borne by the insurer and search costs borne by the insured, it reduces the price of the premiums and expands the market for insurance. Perhaps most importantly, it is attractive in the long-term because it re-establishes a trust relationship between the insurer and the insured. Because premium pricing no longer needs to accommodate CACs, insureds wouldn’t need to switch providers at the end of every contractual period. Rather than being penalized, customer loyalty would be rewarded because the expansion of the insured pool would shift the insurance premium down, rather than up. Moreover, through bypassing the adverse selection problem to which the switch-and-save model is exposed to, the insurer would have an incentive to build a trust-based relationship with its current customer pool. Probabilistically, there is no reason to believe why the current pool of customers, acquired through embedding, is riskier than a person selected at random from the general population.
Embedded insurance creates a virtuous cycle whereby convenience and trust can be complements, rather than substitutes.
Section IV: Overcoming the Shortcomings of Selection Markets Through Embedded Insurance
In this section, we consider how embedded insurance can overcome the supply and demand problems associated with selection markets.
Section IV(a): the Demand Story
The current process of acquiring insurance is undoubtedly inconvenient. Even when insurance is entirely searched for and acquired online, underwriters don’t submit binding quotes until very late in the process. If the customer wants to find the optimal insurance contract by comparing quotes from different providers, they need to go through this process multiple times. In searching for an optimal insurance contract, customers also incur psychological non-search costs. In sum, shopping for insurance is a painful process, so the more convenient we can make the purchase process, the more customers should be willing to pay for the same level of coverage.
Recent economics research has shown that it isn’t access to information but convenience that truly impacts how we make decisions. Providing calorie information had no effect on meal choice but convenience manipulations, or nudges, had a very significant effect: subjects were more likely to choose lower-calorie meals when it was more convenient to do so. Similarly, retirement savings rates and organ donation rates increase when those choices are made more convenient, for example, by defaulting into participation. By making a socially positive choice convenient, nudges use decision errors that ordinarily hurt people to help them instead. We believe that embedded insurance can have a socially positive impact by nudging customers into buying coverage in historically underinsured markets, such as flood insurance and health insurance.
Because search costs are incurred by the customer, not the insurer, they reduce the demand for insurance. These costs all act like a tax on the insurance eventually acquired, borne by the customer. Therefore, the value of the insurance to the customer must exceed not just the premium paid by the customer but also these additional convenience costs. Embedded insurance removes most of these costs by offering insurance at the point of need, eliminating all search costs because the customer no longer needs to search for insurance policies by themselves.
Embedding at the point of purchase maximises the distance between valuation and perceived cost, or the customer’s reservation because of two behavioural phenomena, namely salience bias and loss aversion. The customer’s maximum valuation of a product happens at the point of purchase, because the importance of this new good to the customer’s endowment is particularly salient. They are, thus, more likely to pay to insure it than at any later point in the future. By making insurance salient at the valuation maximum, embedding also maximises the loss aversion experienced by the customer. The customer is more likely to overweigh the risk of losing the product, increasing their reservation price.
Embedded insurance also mitigates the adverse selection problem by lowering the pooled risk borne by the insurer and the premium paid by the customers. The embedded insurance model expands the potential insurance pool to the entire customer pool of the insurable product. By offering coverage at the point of need, embedding can capture customers who wouldn’t have incurred the search costs of finding an insurance policy after the purchase or wouldn’t even have thought about searching for insurance in the first place. By adding these lower-risk customers to the insurer’s pool, the average cost of insurance decreases and drives down the price charged to customers. This, in turn, attracts even more customers that would have otherwise chosen to self-insure. In short, embedded insurance creates a virtuous cycle whereby the more customers it can capture from a lower risk pool, the more customers are likely to join the market.
Section IV(c): Supply Story
The customer acquisition cost (or CAC) represents a large proportion of the marginal cost of low-risk customers. Embedded insurance drastically reduces the CAC by pairing coverage with the acquisition of the insurable product. The insurer doesn’t have to search for customers because the customer pool of the insurer and the customer pool of the product seller are one and the same. By pairing the sale and the insurance processes, embedding could enable a drastic reduction in insurance premiums. In auto insurance, for example, premiums have been increased by 158% because of CAC. This decrease in cost leads to a downward shift of the average total cost curve and, thus, a downward shift of the supply curve. The insurer is, thus, able to provide coverage at a lower premium and capture a larger segment of the population. The marginal cost incurred by these new customers is lower than the average total cost at the original equilibrium because the average new customer has a lower risk profile. As a result, customers across all risk profiles can be insured at the willingness-to-pay of the incremental, lower-risk customer.
As demonstrated in Image 3, the effect is particularly significant for early adopting monopolies at the new cost curve. As the embedded insurer is the sole provider of insurance to this previously self-insured segment of the population, they gain most of the surplus generated by the reduction in average cost. This is represented by the light blue area above the average cost curve. Moreover, the inclusion of low-risk customers to the insurer’s total pool also decreases the premium paid by the high-risk customers. This is represented by the light blue area below the average cost curve. The surplus internalised by high-risk customers in turn, makes the embedded insurer more attractive than their competitors to customers across all risk profiles.
In short, by reducing customer acquisition costs, embedding reduces adverse selection and, thus, the price of insurance. This attracts a bigger pool of lower-risk customers, mitigates adverse selection, lowers underwriting costs, and benefits both the high- and low-risk customers.
Section IV(d): Social Welfare
Taken together, these supply and demand effects of embedding minimise the costs associated with selection markets. Unlike the "switch-and-save" model, in which everybody loses in the long-run, embedded insurance is a win-win-win: the customer gets coverage faster and at a lower price, the insurer has access to a wider customer pool that exhibits lower risk characteristics, and the greater community benefits from lower risk exposure.
In a world where, in equilibrium, the insurer can negotiate behind the veil of ignorance, insurance is no longer a selection market. If the customer base is wide enough, it could be optimal to charge every customer the same amount. The increase in pool size is valuable to the insurer and attractive to the insured, while producing a net-positive social impact because a larger portion of the overall population is insured against a portfolio of risks. For example, if more people could acquire health insurance at an accessible cost, the rate of health and non-health related debt would decrease. Research on the impact of Massachusetts' health reform found that healthcare coverage reduced the total debt and the fraction of debt that was past due, improved credit scores, and reduced personal bankruptcies. From a social welfare standpoint, this is very significant because medical debt is now the number one source of debt collection in the United States, surpassing debt in collection from credit cards, utilities, auto loans, and other sources combined. This pattern held even as strong economic growth led to declines across all debt categories.
Furthermore, increasing the financial security of communities by means of insurance also stimulates spending in local economies as individuals have more dependable disposable income to spend on goods and services. Finally, increased insurance coverage also has a positive impact in workplace productivity and economic output. In this field alone, it is estimated that uninsured productivity losses reduce U.S. GDP by $260 billion per annum: one study found that uninsured workers missed almost five more days of work than insured workers per year.
Section IV(e): Expanding the Production Possibility Frontier
Finally, embedded insurance is a catalyst for market expansion by expanding the production possibility frontier. In the previous subsections, we explained the supply and demand consequences of embedding in existing insurance markets. We conclude by considering how embedding can play a pivotal role in creating insurance markets that wouldn’t otherwise have existed because customers never thought to insure them. For example, many customers wouldn’t have thought of insuring carbon credits, despite the increasing risks of re-sequestration associated with global warming. If they didn’t even consider the risks associated with the product, they definitely wouldn’t go out of their way to insure it. By highlighting the risks and offering coverage at the point of purchase, therefore, embedding would generate a market that wouldn’t have existed in its absence. This represents an outward shift of the insurance production possibility frontier for many insurance products and for the insurance market as a whole.
We briefly consider its demand and supply-side implications. The availability of insurance in a previously uninsured market will increase the number of players in the market, especially if the associated risks were deemed a deterrent to market entry. It will also shift down the supply curve because some or all of the risk will be shifted from the supplier to the customer, who will be able to choose their optimal level of insurance at the point of purchase. In the case of carbon credits, carbon registries currently lock 10-20% of carbon credits registered by developers in a buffer pool. This significantly reduces their profit margin. A more efficient alternative would allow project developers to register all the carbon credits sequestered by their projects. These credits could be sold at a lower price and the customer could then choose the optimal level of coverage according to their individual risk preferences.
- Einav, Finkelstien, and Mahoney (2021) The IO of Selection Markets.
- Arrow, 1963; Pauly, 1974; Rothschild and Stiglitz, 1976
- Rothschild and Stiglitz (1976)
- For discussion, see Cutler 1996
- Gallup survey reference
- Gallup survey reference
- Vance, W. W. (1908). Early history of insurance law. Columbia Law Review, 8(1), 1-17.
- Bretano, The History and Development of English Guilds
- Ibid. p. 11; Cheyne, Industrial and Social History of England, p. 72
- Bensa, Il Contratto di Assecuratione nel Medio Evo, p. 48
- YouGov Whitepaper, “Better Safe than Sorry”, https://yougov.co.uk/topics/resources/articles-reports/2019/09/09/better-safe-sorry
- Klemperer, 1995; Erdem and Sun, 2001; Dube et al., 2009; Shcherbakov, 2016
- Agnew and Szykman, 2010
- Honka, 2014.
- For overview, see Choi, Laibson, and Madrian, 2005.
- Downs, Lowenstein, and Wisdom (2009)
- Madrian and Shea, 2001
- Johnson and Goldstein, 2003
- Kunreuther, Pauly, and McMorrow, 2013
- Honka, 2014.
- Hu, L., Kaestner, R., Mazumder, B., Miller, S. and Wong, A., 2016. The effect of the Patient Protection and Affordable Care Act Medicaid expansions on financial wellbeing (No. w22170). National Bureau of economic research.
- Mazumder, B. and Miller, S., 2016. The effects of the Massachusetts health reform on household financial distress. American Economic Journal: Economic Policy, 8(3), pp.284-313.
- Kluender, Mahoney, Wong et al. (2021) Medical Debt in the US, 2009-2020, The Journal of the American Medical Association, 326(3):250–256.
- Dupor, B. and Guerrero, R., 2021. The Aggregate and Local Economic Effects of Government Financed Health Care. Economic Inquiry, 59(2), pp.662-670.
- Davis, K., Collins, S.R., Doty, M.M., Ho, A. and Holmgren, A.L., 2005. Health and productivity among US workers. Issue Brief (Commonw Fund), 856(856), pp.1-10.
- Mattke, S., Balakrishnan, A., Bergamo, G. and Newberry, S.J., 2007. A review of methods to measure health-related productivity loss. American Journal of Managed Care, 13(4), p.211.
- Dizioli, A. and Pinheiro, R., 2016. Health insurance as a productive factor. Labour Economics, 40, pp.1-24.