January 24, 2017 | By Devie Mohan
The insurance industry has been historically slow to embrace technology, lagging behind even the banking sector. This attitude is understandable — the industry relies heavily on historical data, which is generally not available for new technology, and the industry is immensely risk-averse, as even one failure to live up to their commitments could be devastating to an insurer.
Technology is putting pressure on the insurance industry from three sides.
New customer demand
The first is customers, who have grown accustomed to an easy, Facebook-like experience in interacting with large service providers. Current insurance products are far too generalized and one-size-fits-all to appeal to a customer base that is expecting easily individualized products. Technology like usage-based insurance can make a provider significantly more appealing in this respect by making it possible to only pay the premium for risk actually taken; wearables can revolutionize the healthcare insurance market by allowing for truly personalized pricing.
The second source of pressure comes from competitors. Not only will consumers be more likely to give their business to a digital-native insurer, but entire new kinds of exposure are opening that will give a challenger an opportunity to strike. The cybersecurity market is growing everywhere, along with the pressure to contain and manage the risk better, yet traditional insurers are slow to make convincing offers to threatened customers. In addition, the blockchain is making a more decentralized market possible: While insurers could so far count on the immense need for capital as a barrier to entry, the blockchain could finally bring the transparency and reliability needed to make dynamic, small-scale insurance underwriting possible.
Lastly, technology provides new avenues to cut costs in internal processes and pricing products by making available huge sources of data and enabling its more efficient analysis. Insurers currently spend a lot of money on services that aren’t in their core specialty- processing claims, detecting fraud or manually assessing new risk. New algorithms for predicting risk, for example using machine learning, will allow for vast automation of the underwriting process, and managing contracts and identities with the blockchain will reduce the resources needed for fraud detection. New diagnostic technology, like wearables for healthcare or GPS trackers for cars, is bringing a new wealth of data that may balance the lack of historical data that is currently keeping insurers in as-is mode.
With those pressures, however, come a number of important opportunities in three areas: underwriting automation, connected devices and cybersecurity
Underwriting Automation Data
Automation in the insurance industry can make underwriting both more efficient and more precise, with different lines offering different opportunities for automation.
Insurers are currently using automation primarily to support underwriters and aid in triage, with only a fifth saying their primary objective is to fully automate the process. What kind of automation is possible varies between business lines, but even in the most advanced segment, personal lines, only 42% of insurers say they have “mastered or almost mastered” automation. At the bottom end, life insurance, 80% of insurers say they are struggling or just getting started with automation.
Insurers are focusing on personal lines and small and mid-market commercial to expand their automated underwriting capacities, with more than 40% saying they will increase their spending in each field. However, in line with being late adopters, it is estimated that only 10% of insurers will have an algorithmic business strategy in 2019 that makes use of more advanced techniques like machine learning, which could make automation viable for more involved lines like health.
For most policies in motor, home and life, an underwriter reviews between eight and 15 factors. Current automation systems for life insurance have similarly small data requirements, with around half the systems drilling down into no more than 10 questions, and a third of them asking as many as 60 follow-up questions. Most systems incorporate lab data and prescriptions databases. These amounts of data are small compared with what a sophisticated automated system could use to assess risk.
As a naturally data- and analysis-heavy industry, insurance stands to profit from advances both in the sophistication of automation and in its affordability. As an industry that is also conservative and late to adopt technology, it faces the risk of being outflanked by a less risk-averse challenger that’s willing to bet on automation skills.
Insurers have for more than 25 years used primitive systems to fully automate small-scale risk in simple lines (for example, travel insurance), or to aid their underwriters by more effectively triaging requests and directing them to the underwriter that’s best suited for them, or to do some preliminary analysis. These systems generally rely on simple rules and are seen as supporting underwriters. As automation products become easier and cheaper to implement, and new decentralized technologies like blockchain make small-scale underwriting more transparent and available, we can expect their share to increase incrementally.
More importantly, insurers are also facing a new wealth of data both for historical risk research and for better assessment of new risk that could fundamentally change the way risk is priced. However, traditional systems are not equipped to deal with these amounts of data, and few insurers are ready to implement the machine learning technology that would be. The problem is that modern machine-learning can produce results but cannot generally explain them. Policy underwriters are naturally skeptical of underwriting risk based on a technology that provides no justification for a pricing beyond the rigor of its setup and the vastness of the data it has been trained on. However, insurers already use fundamentally similar systems for assessing their underwriters’ competence—if a junior underwriter repeatedly prices a risk outside of the usual range the same way a more senior underwriter would, the junior underwriter will be allowed to price those risks without supervision. If insurers can learn to trust this approach with technology, too, they will embrace machine learning.
Underwriting automation will become a significant field of innovation around both reducing staffing and coping with the new amounts of data, with each business line requiring its proper automation technology. As risk assessment algorithms become more reliable and executives more confident in them, they will be able to make low-level underwriting both cheaper and more consistent. As new sources of data for risk analysis become available, insurers will have to use machine learning algorithms to be able to make sense of the vast amounts of data.
Connected Devices Data
Connected devices in insurance describes the network of smartphones, wearables, home diagnostics and other internet-connected devices that form one of the fastest growing spaces within insurtech. This stands to make available a new wealth of data for insurers to handle better pricing and encouragement of risk-decreasing customer behavior.
Wearables and Diagnostics
87.7 million U.S. adults, or about 38%, are expected to be using a wearable device in 2019, a growth mainly fueled by smartwatches and wristbands. VCs invested around $3 billion in IoT startups worldwide in 2015, and 38 million European and North American households are expected to have a smart thermostat in 2018, with two-thirds of those lying in North America. Nearly two-thirds of consumers already own or plan to purchase an in-home IoT device in the next five years.
Only 3% of insurers are already making use of wearable devices, and less than a fourth are developing a strategy for them, even though 60% of insurance executives believe that wearable technologies will be adopted broadly by the industry.
Telematics in cars allow insurers to track driving patterns of their customers. The advent of cheap GPS devices has made this technology ready for widespread adoption with usage-based insurance (UBI) and dynamically adjusted premiums. More than 15% of the U.K. car insurance market is usage-based, and Progressive alone has more than 4 million UBI customers in the U.K. In the U.S., there are around 5 million UBI policies in effect, and approximately 70% of all auto insurance carriers in the U.S. are expected to use UBI by 2020, with more than 26% of all motor policies being usage-based. Usage-based programs on average lead to a 57% decrease in total claims cost.
Health insurance tech startups raised more than $1.2 billion in venture funding in 2015, more than twice as much as in 2014, and making up almost half of the $2.6 billion in venture funding that was raised by insurance tech startups overall. Insurers themselves have committed more than $1 billion to investments in startups, and many of them have established their own in-house venture capital funds to exploit IoT and ready themselves for new markets.
58% percent of smartphone users in the U.S. have downloaded a health-related app, and around 41% have more than five health-related apps, generating data that insurance providers could use to fine-tune their individual premium pricing and encourage low-risk customer behavior. The first insurance company to offer discounts to customers using technology aids for better living was John Hancock in 2015. Other companies in the U.S. and elsewhere have since followed suit, offering as much as a 15% premium discount.
The number of connected devices is projected to grow by 35% each year over the coming years. This creates a new wealth of data, which insurers see as important but do not know how to tackle. To understand how insurers can approach the issue, we must look at the health insurance industry, which is at the forefront of integrating wearable tech and makes up for about half of all insurance tech investment.
Most of the efforts to integrate technology by insurers are simple and mainly designed as promotions, like awarding credits for a number of steps taken: this is a far cry from what big data could do for adaptive premium pricing based on comprehensive health data for each customer. The problem is likely a skepticism toward new technology for which no historical experience is available.
The other major industry using connected devices is car insurance. Here discounts are given to customers who drive less and more safely than others, and the benefits so far have been clear: a 57% reduction in claims. It remains to be seen how much of this reduction will turn out be a temporary Hawthorne effect, but it is sizable enough to pique interest everywhere. The major problem is that so far insurers do not penalize worse-than-average drivers, and it is unclear to what extent customer will accept self-tracking as mandatory or de facto mandatory by pricing. The same issues will also have to be faced by other insurance industries moving to integrate IoT.
Insurers agree that the Internet of Things and wearables will play a major role for the industry but have so far only used them in often-gimmicky promotional efforts, hindered by the fact that they cannot penalize customers for risk-increasing behavior. The health insurance market is the main point of investment for insurance tech, but the rise of smart devices everywhere makes innovation possible in all parts of life. The first insurer to overcome the regulatory hurdles and offer truly adaptive and responsive insurance that is not limited to one or two factors but embraces big data will have a strong first-mover advantage.
The cyber insurance market grows each year both in size and import but is insufficiently understood and served by insurance providers, who so far have few technological options to contain, predict and address cyber risk.
Risk levels and market size
Estimates for the yearly cost of cybercrime vary from €330 billion to €506 billion. The cost will increase as businesses and their supply chains become more digitally integrated. In the past three years, the average economic impact of cybercrime per organization in the U.S. has risen from $11.6 million to $15.4 million. The biggest share of this impact comes from the cost of business disruption. The global market for cyber insurance is estimated to rise to $20 billion in premiums by 2025.
Customer awareness and adoption
Businesses are insufficiently insured and informed around cyber risk. Around 40% of Fortune 500 businesses currently have insurance against cyber incidents, but generally not enough to cover their full exposure. In the U.S., 24% of all business have some form of cyber insurance. 48% of enterprise customers say they lack the necessary understanding of the complexity of cyber risks to better prepare against them.
Available products and expertise
Of the 10 largest insurers, only five offer standalone cyber coverage. While 90% of all insurance underwriters offer cyber insurance as an add-on to other products, more than 50% do not have any dedicated underwriters for cyber risk and rely on underwriters for other lines. Consequently, 70% of insurance brokers claim there is little to no clarity about what is covered in cyber products.