By Chris Logan and Srivathsan Karanai Margan
The modern insurance industry has evolved over the past four centuries, shaped by geopolitical, economic, regulatory, and technological events to become what it is today. Insurers have established fundamental principles and procedures that are meticulously followed when designing products, classifying risk, accepting risk, and servicing and settling claims. Principles such as the law of large numbers, risk pooling, and premium adequacy became inviolable constructs of insurance products as the offerings evolved to meet new realities.
The rise of the millennial generation as a major consumer force, however, concurrent with the emergence of technologies such as ubiquitous mobile phone usage, cloud computing, big data, and artificial intelligence, is causing a radical shift in the product lines of some insurers. Underpinning this shift is millennials’ self-directed, tech-savvy approach to life and their expectation of hyper-personalization in products and services. Insurers and insurtech companies are responding with new service offerings that aim to reinvent product lines, pricing, and the customer service experience, from peer-to-peer coverage models to well-being-focused connected insurance products and on-demand insurance.
On-demand insurance allows customers to use their smartphones to buy insurance just when they need it, usually for a specific possession. It is receiving a great deal of attention because of the simplification of the customer service experience and the flexibility it allows in coverage and pricing. While the on-demand insurance market remains in its infancy, it presents a conundrum to the insurance industry. Skeptics fear that on-demand insurance overlooks perilous consequences that could disturb some of the fundamental pricing principles of the industry, while optimists predict that it is another inflection point in the history of insurance that is likely to shape all insurance offerings in the future.
What Is On-Demand Insurance?
Technically on-demand insurance isn’t new. Insurers have long sold insurance products such as travel and personal accident insurance on an as-needed basis. However, newer forms of on-demand insurance are receiving extra scrutiny because products are being made available for other types of personal insurance, such as personal belongings, home-share and ride-share risks, auto insurance, smartphones and renters insurance, and income protection, in which usage and behavior components are factors that must be reckoned with.
On-demand insurance providers not only sell and service these products via smartphones in a direct-to-customer mode, they also to identify micro-moments of relevance to insurance, adapt personal and contextual requirements, and position their products in an individualized way. This is particularly appealing to the millennial generation, many of whom have delayed owning assets and are invested in the concept of a sharing economy that thrives on access rather than ownership. They not only seek insurance coverage only during the periods of usage or for the specific items they wish to insure, but question the depersonalized pricing principles followed by insurers in cross-subsidizing high-risk customers with profits from low-risk customers like themselves, which makes the promise of individualized pricing appealing to them.
On-Demand: A Misadventure
While on-demand insurance models are emerging, a dominant design has yet to evolve. However, some of the characteristics of the existing models, such as personalized premium rates and short periods of coverage, appear to be impelled more by their marketing potential than by actuarial propositions, raising questions about the long-term sustainability of these products.
One potential flaw in on-demand policies is the collapse of the law of large numbers. Assessing and pricing pure risk—which event, how and when it will occur, and the loss effect it may cause—is always a matter of intelligent guessing. It is impractical, however, to predict the probable claim occurrence for one insured within a specific period. But if a larger number of individuals are pooled, the law of large numbers means the average loss exposure for the group can be accurately assessed.
To make the guess reliable, insurers pool customers based on the similarity of their risk profiles, analyze their past histories, and predict the expected long-run loss frequency and severity. When certain risk factors of the customers within the pools don’t match, the cost of the loss associated with the risk factor is borne by all the customers in the pool, i.e., there is cross-subsidization between those with higher risk and those with lower risk.
Having customer pools of larger size, therefore, helps insurers to diversify and average the overall risk. The greater the size of the pool, the greater the chance that aggregate surpluses will match or exceed the aggregated deficits. On-demand insurance, however, is a bespoke proposition that appeals only to a small segment of customers. Hence, with smaller customer pools, an estimate of the expected loss for each risk factor is unlikely to be accurate, and it is likely that aggregate surpluses will not match or exceed the aggregated deficits. As a result, reinsurance or additional shareholder capital will be required to offset the risk.
This means that contrary to the sales pitch that on-demand products are charged at personalized premium rates that take into account context, need, usage, and behavior of the customer, on-demand premiums are unlikely to reflect an insured’s true risk. Without the advantage of large numbers, price setting and capital adequacy of insurers may be at risk of collapse. The cost of capital may even make the personalization-driven on-demand insurance market a loss-making venture.
In addition to large numbers to increase the scope of diversification and improve pricing accuracy, insurers generally offer coverage for a period of a year or more to avoid the problem of risk peaks. Typically, policy terms comprise periods of risk peaks and risk troughs, indicating the periods when the probability of the occurrence of a risk event are high and low, respectively. On-demand insurance on the other hand gives customers the option to purchase insurance coverage when it is actually needed and for the exact period required. This means that customers could switch on insurance coverage just before they go through a risk peak and switch it off immediately thereafter.
If all the policies in an insurer’s portfolio are purchased only to cover risk-peak periods, the average risk of the portfolio could increase. Because on-demand insurance is a new offering, insurers will not have historical data and may find it difficult to assess the short-term expected loss for each risk factor and insured. Furthermore, as the learning time, which is the customer-in-force time, is lower for on-demand products, it may take longer for insurers to reasonably assess the short-term expected loss for each risk factor and customer.
Considering that if customers switch on the insurance coverage just before they go through a risk peak, the causal chain of events from the point of insurance purchase to that of risk occurrence is shortened, which can lead to claims becoming less random. In extreme cases where a claim is almost certain, the absence of the uncertainty of risk would make insurance an invalid concept and the policy may resemble a sinking fund. An unbalanced portfolio entirely skewed toward risk peaks is potentially a death zone for insurers.
Another potential problem for on-demand products is premium inadequacy. The premiums insurers charge normally are set to recover three costs: the expected value of insured losses, operational expenses, and profits. While consumers may expect premiums to scale linearly with the period of risk exposure, this often is not the case, as the expected value of insured losses is unlikely to vary linearly.
Operational expenses, primarily related to selling polices and managing claims, normally are related to the number of policies and not to the aggregate period of exposure. Hence, the ratio of operational expenses to period of exposure gets higher with smaller periods of exposure, which has been the main hurdle to offering on-demand insurance in the past. On-demand insurers are looking to break this relationship by using online self-checkout and bots to assist with sales and claims. While the cost to maintain these services is small, the upfront build costs are substantial.
Profits required to compensate for shareholder capital will be higher for undiversified risk. For example, consider the ratio of sum insured to premium scales for a typical policy: The sum insured stays the same with shorter risk exposure periods but the premium is reduced. Based on this ratio, on-demand coverage looks more like reinsurance than insurance—where the required profit needs to exceed the expected value of insured losses due to the significant amount of shareholder capital required.
While on-demand insurance for a month costs more than one-twelfth of an equivalent annual policy, there is a limit to how much consumers will pay for the additional convenience of an on-demand policy over a traditional annual policy. Until on-demand insurers reach a substantial number of policies-in-force, premiums may be inadequate to recover the three core costs.
On-Demand: A Profitable Venture
When traditional insurance models were constructed, obtaining accurate information about an individual customer’s profile to plot the risk signals was costly. The advent of new technologies and connected devices, however, has reduced the cost of obtaining this information and made it easier for insurers to continuously access the types of “hot data” regarding the object of risk, its usage, and user behavior. Insurers can continously monitor customers to mitigate risk and engage with them, suggesting a new future of smart insurance products that defy traditional models.
New technologies and techniques also are allowing insurers to unpool risk for the segment of one, where each customer becomes her or his own segment. In the past, insurers had to depend on large numbers because they were unable to ascertain the risks associated with a single customer with reasonable fairness. Today, modern mathematical techniques such as independent component analysis mean insurers no longer need to pool similar customers, but a reasonable number of customers are adequate to assess the expected loss for each risk factor or compound risk factors.
Accurate and continuous information provided by connected devices helps insurers eliminate information asymmetry. Insurers can prevent or objectivize subjective risks and calculate the cost to cover those risks. Consequently, the number of customers needed to assess the expected loss for each risk factor or compound risk factors is reduced. For instance, when a customer opts for on-demand insurance, insurers are able to immediately check industry, regulatory, and third-party databases to verify the customer’s credentials. With the help of connected devices, they are able to accurately assess usage and behavioral risks to predict the probability of risk and even engage with customers regarding behavioral changes to prevent risk. This helps insurers unpool the systemic and behavioral risk components to arrive at a segment of one and to price customers individually.
While on-demand insurance may be overly subject to risk peaks, certain customer behavior and experience-driven choices may in fact help insurers tunnel through the risk peaks and benefit from them. Traditional insurance provides coverage for both risk peaks and risk troughs; hence, the risk peaks in on-demand insurance are just a subset of the coverage provided earlier. Based the assumption that all the customers who purchase on-demand insurance do so only during risk peaks, the claims frequency will be higher and hence the experience relatively less variable. This suggests it may take less time for an on-demand insurer to accurately assess the short-term expected loss for each risk factor and customer.
Further, by eliminating annual renewals, on-demand providers remove an important milestone when the insured assesses the cost effectiveness of their policy and decides on remaining with the same insurer. As a result, when risk peaks prompt consumers to take up coverage with app-driven contextualization, on-demand providers may be better placed to nudge customers and increase persistency. While some customers will be seeking coverage only for a short period, many others may intend to continue on-demand coverage well beyond the traditional contract period of one year. Customers may be attracted to the convenience and user experience of handling their insurance online or they may value not being locked-in to a contract.
There is also the potential that on-demand insurance may lower only the premium bar but not profit. By lowering the absolute premium amount, on-demand insurers are able to access an underserved market: people who need insurance only for a specific period or for one specific item rather than coverage for a gamut of risks for an extended period. It is likely that many of these people currently self-insure rather than opting for traditional insurance by paying a relatively expensive premium.
Importantly, on-demand insurers are able to offer products without sacrificing profitability by partnering with insurers or reinsurers. This enables on-demand providers to reduce the risk of mispricing and the cost of diversifying their risk. While these partnerships benefit the on-demand providers by minimizing cash burn, for traditional insurers and reinsurers it is a low-risk opportunity to explore and tap a new market space. Further, through partnerships, on-demand risk can be folded into the insurance partner’s traditional book, thus reducing the level of expensive shareholder capital or reinsurance buffer required. The cost of providing this diversification would be quite small for the partnering insurer.
Risk peaks aside, the consumers of on-demand insurance are in fact willing to pay more for insurance, as they are trading premium costs for convenience. In addition, on-demand insurers are front-loading expenses, thereby reducing the long-term policy acquisition costs. And with shorter periods of risk exposure, on-demand insurers have greater flexibility to re-rate. Hence, if surpluses are running low, on-demand insurers can increase prices rather than rely on longer-term reversion to the mean. For on-demand products, then, it is only the absolute value of the premium that appears to be lower but not the profits, which may in fact be higher than that of traditional products.
The Future of On-Demand Insurance
The insurance landscape is transforming. Insurers are atomizing products, bundling them based on individual requirements, developing new engagement models, and pricing each offering accordingly. In creating these new products, insurers are not depending only on traditional risk models. They are revisiting their existing rating algorithms to suit these product and service offerings.
Backed by insurers and reinsurers, on-demand providers currently are in an expansion mode and are focusing on market penetration and growth. Dominant designs for on-demand insurance offerings will emerge as providers start focusing on generating profits.
The core of on-demand insurance is the economic concept of utility—giving customers utmost satisfaction and immediate gratification. While this concept is well-suited for consumables, its applicability and sustainability for financial instruments such as insurance has to be viewed with a tinge of skepticism. The concept of personalized pricing may work until the insureds demonstrate optimal levels of usage and behavior and the uncertainty is within tolerable limits. Customers who fail to conform to the standards may be classified as high-risk, and may be subjected to extreme fair discrimination by insurers. This may force those customers to bear the entire cost of the risk they represent and, consequently, the cost per unit of risk could be priced at prohibitively exorbitant levels. In such instances, regulators may have to act proactively to protect customers’ interests.
While on-demand insurance providers are pitching to millennials comfortable in the connected and sharing economy, on-demand insurance may also attract less tech-savvy, less wealthy consumers who aren’t able to afford traditional insurance. To the contrary, the behavior of millennials may change as they start earning more and owning more assets. It is also possible that given the higher absolute premiums of on-demand products, they could return to traditional insurance coverage.
Though the perception in the market is that on-demand insurance is an industry disruption and a seismic shift, it also can be viewed as an inclusive expansion of an existing idea to additional products facilitated by emerging technologies. On-demand insurance does not disrupt the fundamental pricing principles of insurance, and if properly positioned, it potentially could turn out to be quite profitable for insurers. Nevertheless, insurers need to be careful to avoid missteps that could potentially increase costs, dent profits, and lengthen the payback period.
CHRIS LOGAN is the founder of the peer-to-peer insurer www.PeerCover.co.nz, a donation-matching club for crisis crowdfunding campaigns.
SRIVATHSAN KARANAI MARGAN works as an insurance domain consultant in Insurance & Healthcare Innovation Lab at Tata Consultancy Services Ltd.