Threading the Needle
Since data allows us to price risk more accurately and, at the same time allows us to offer incentives for appropriate behaviour, it is a very useful tool for insurers looking to achieve optimal risk pooling. However, if we take it too far we risk ending up in a surveillance society.
This is a link-enhanced version of an article that first appeared in the Mint. You can read the original here. If you would like to receive these articles in your inbox every week please consider subscribing by clicking on this link.
Insurance is a human innovation that ensures that in moments of crisis, no one individual has to bear the weight of misfortune alone. Pooling risk has allowed us to contract, innovate and prosper, but with the ubiquitous availability of granular data, that assurance is starting to be replaced by something colder and more precise.
The modern insurance industry dates back to a time when merchants gathered in coffee houses (such as Lloyd’s) to collectively insure their vessels against sea-faring risks. Since then, it has expanded into other forms—health, property and life insurance—while remaining true to the fundamental economic principle of risk pooling.
Pricing
The risks that individuals face can be quantified and priced. What cannot be foreseen is when misfortune will occur and at what scale. Individuals who worry about the downside of certain risks can join together to share them so that they can support those among them who actually end up suffering harm.
When this pooling of risk is aggregated across a wide enough population base, insurance companies are able to forecast the frequency and severity of adverse events with enough accuracy to be able to set premiums at levels that are sustainable for all participants. This is the economic basis of the modern insurance industry.
Having said that, it is critical to price that risk accurately. Take, for instance, adverse events linked to the irresponsible behaviour of high-risk individuals. If insurance companies charge a low flat price, low-risk policyholders will end up subsidizing higher-risk policyholders. If the cost of an insurance premium is low relative to the risk, high-risk participants will ‘free-ride’ on that backstop, engaging in risky behaviour secure in the knowledge that any loss they suffer will be compensated. If, on the other hand, premiums are priced too high, low-risk policyholders will exit the market, making the entire exercise of pooling risk ineffective.
Measurement
To address this, insurance companies are constantly trying to come up with new and better ways of measuring risk. They know that unless they can granularly measure the risk they are insuring against, they will not be able to price it in a way that strikes the right balance. Thanks to advances in digital technologies, they finally have a way to do that. Affordable devices like Fitbit, Oura and Whoop now provide streams of personal health data in real time, while most modern vehicles have built-in telemetry that offers similar insights into driving behaviour. And insurers have begun to capitalize on this.
In 2024, after India’s Insurance Regulatory and Development Authority (IRDA) approved telematics-based policies, insurers began using black-box devices to collect real-time driving data to assess people’s behaviour behind the steering wheel. As a result, safe drivers can now receive discounts of up to 25% on their premiums based on this driving data.
Health insurers have also begun to integrate data from wearable devices into their calculations. They now offer discounts of up to 15% to policyholders who consistently maintain more than 10,000 daily steps while disincentivizing smokers with surcharges of as much as 20%.
A price, according to economists, is a ‘signal wrapped in an incentive.’ Risk is no different. Granular data allows us to price risk more accurately while offering incentives for less risky behaviour. This data-driven approach ,therefore helps insurers achieve optimal risk pooling.
Threading the Needle
Having said that, it is possible to go too far with this approach. If, for instance, we start pricing the risk of circumstances over which the policyholder has little control, such as pre-existing genetic conditions or the neighbourhood in which they are constrained to live, we will, instead of incentivizing low-risk behaviour, be punishing policyholders for mere accidents of birth. If we continue down this path, we will end up creating a risk-evaluation society where our premiums are constantly being adjusted based on our every action.
If anything, data will only become more easily available and at a far more granular level. With the rampant proliferation of sensors, we are headed for a world where everything we do will be tracked every minute of the day. While this will, no doubt, benefit us in countless different ways, the availability of granular data at scale will allow insurers to assess us with unprecedented accuracy. Once all this data is funnelled into their risk prediction algorithms, they will eventually be able to predict with reasonable statistical certainty what is likely to happen to us and when. When that happens, the premiums we have to pay insurers will not be a form of risk pooling but a pre-payment for nearly certain misfortune.
We need to find a way to thread this needle. We should be able to obtain sufficient insights to disincentivize free riders, but not so much that we evolve into a round-the-clock surveillance society. This will require us to establish clear boundaries on how insurers can use data and ensure that risk pricing doesn’t turn into some sort of punitive system for undesirable activities. This should also extend to restrictions on how we use sensitive data—such as genetic information—and limitations on how health-tracking information, geographic risk assessments and the like are used.
Insurance has always been about balancing the scales of risk. With the abundance of data that the world now has, we may need to relocate the fulcrum.
