Many businesses are now becoming data businesses. But there are also businesses that have always been data businesses. One such “traditional” data business is the insurance business. But insurance companies are now moving to a new era of customer-oriented behavior data. This is a significant step, because such changes can make the insurance business better and fairer. But it also raises concerns like privacy and personal rights. User-held data models can help strike that balance.
The insurance business has always been based on data and probabilities. The simplest concept of insurance emerged when people in a village put some money together in case a fire broke out in one’s house. Basically, an insurance company tries to estimate the propensity for a certain event and then collect insurance payments to cover those costs and some profit. In practice, this means collection and assessment of all kinds of data for events like fires, traffic accidents, diseases, deaths, earthquakes, burglaries and many other things.
We can simplify things to say that the insurance business has two main factors to consider for each case: 1) general propensities, like how many fires, burglaries, traffic accidents, or cancer cases there are as a whole, and 2) customer-specific data on how a specific customer’s propensities compare to the general statistics, e.g., whether a customer has higher probability of encountering a traffic accident or being diagnosed with cancer. An old story claims that English football players have the most expensive car insurance, because they own expensive cars and have a track record of damagingthem quite often.
Customer-specific data has mainly been demographics data such as age, gender, living area, occupation, and some life habits (do you smoke, drink alcohol , go skydiving, etc). But now, it is possible to get much more specific data such as wearables or genetic data. This can help insurance companies to tailor pricing more accurately for each individual. But it also raises many questions.
Technical, legal and ethical questions
The use of personal data for insurance purposes is a complex question with various technical, legal and ethical aspects.
The typical ethical discussion has to do with what kinds of data insurance providers should be able to have access to. While these kinds of discussions have yet to be definitively settled once and for all, there seems to be a kind of simplified consensus that it’s okay to use data from things you can control personally, like how you exercise or how safely you drive, but not from things you cannot control, like your genetic data.
For example, in the US, the Genetic Information Nondiscrimination Act (GINA) prohibits health insurance companies from using genetic information to make coverage or rate decisions. In the EU, the Council of Europe has a recommendation for the use of personal health data for insurance purposes. But interpretations of these also vary, and some states have exceptions, so the legal issues are not that simple, either.
Lifestyle and behavior data is now included in some insurance policies, and most probably, this trend will continue. Some insurance companies offer wearables to their customers to get a lower premium if they exercise regularly. For example, John Hancock’s Vitality program offers Apple Watches for this purpose to their customers. Often insurance companies also filter (somehow) which customers are qualified for this kind of program. Driving habits are another good example of personal data being harvested for setting insurance premium prices.
But wearables and driving data raises many questions. Maybe we can say it is fair that people’s behavior can influence their insurance premiums – but how much data do people really want and need to share with insurance companies? What details are really relevant, and who gets to decide? Could it be enough to share just some data analytics results?
For example, you might feel it’s okay for the insurance company to know how much you exercise, whether you live in high or low-risk areas, whether you often make fast accelerations and brakings, and whether your circadian rhythm is regular. And that might be enough data for them to set the premium level – they don’t really need to know, say, your location, heart rate and sleeping patterns.
Insurance companies are already recognizing the issue with personal data. Many of them want to offer services to their customers to help them live healthier and better lives. If people are healthier, less stressed, and think about their life habits, it means fewer claims and compensation payments for the insurance companies. But they also know that data privacy regulations such as GDPR or California’s CCPA make it more complex to collect personal data – and, perhaps more to the point, keeping personal data also creates significant liability risks for them.
That’s why insurance companies have become interested in user-held data models. Basically, the user-held data model means that users can collect their wearables, driving and daily lifestyle data, keep the data in their personal storage, process it with insurance analytics, and then share only the results of the analytics with the insurance company (rather than the raw data itself) if they want to get special pricing. For insurance, this is much simpler and cheaper than collecting a lot of sensitive raw data from customers.
Some may worry that users might fake some data results, but it has always been an issue with insurance data; some people lie on questionnaires about, say, their smoking or drinking habits to get a better premium deal. Smart data models can ensure results are reasonably reliable, while those who intentionally mislead insurance companies risk being caught and charged with fraud.
New models are needed
We’re still in the early days with user-held data models, but there will be an insurtech hackathon in a couple of months held by a large insurance company and leading tech companies to develop applications on user-held data for insurance purposes. Several large insurance companies have also started piloting and testing user-held data models. Many insurance companies believe that having more data is of value for themselves as well as their customers, but they also realize that the Wild West time of data is starting to fade and they need models for ethical data usage that respects the rights of their customers.
The insurance business has always lived on data and it must actively find new models. Personal user-generated data will push the insurance business and its pricing mechanisms towards a new era. At the same time, regulation and ethical standards are crucial to keep the insurance business fair and available for all kinds of people. It is essential that insurance companies invest in data models that give control to their customers that can impact their premium prices while also protecting their privacy and rights.