Big data analytics has generated a lot of hype in the financial services industry over the last few years. Many financial service providers claim that it will enable financial inclusion by being able to properly assess the credit risk of people with no credit history, or that by analyzing in a more granular way customers’ profiles, financial products will be better tailored to suit customers’ needs. But there is one aspect that no one has talked about or even envisaged thus far: the possibility that Big data analytics could make private companies subsidize your insurance or your credit.
Let us start by examining how Big data analytics allegedly enables financial service providers to better price risk or tailor financial products to consumers’ needs. By combining various data sets such as delinquencies on a loan with social networking data, the correlations made can enable financial service providers to predict default rates of users having a similar social networking behaviour even if they have no credit history. The same is being done for health insurance, by gathering and combining various sources of data such as connected devices (pedometer, smart watch,…) and social networking behaviour (a study has shown that a user publishing “positive” posts are more likely to be in better health). From the point of view of defending the general interest, there is much to be concerned about. Big data could enable a systematic discrimination of the poorest and most vulnerable part of the population, breaking the socialization of risk and moving to individualized risk based pricing.
But there might be a silver lining after all. In recent years, there have been many breakthroughs in certain fields of study such as psychology, behavioural economics, sociology, neurology and even the birth of entirely new disciplines like psychometrics. To name a few examples, neuroscience of free will is a discipline which looks into human decision making. A number of studies showed that decisions are made in the brain before they reach our consciousness, questioning the very concept and existence of free will. Psychometrics enables researchers to make a link between detailed psychological characteristics of people by matching their responses from a very detailed psychology questionnaire with their social networking behaviour, a research directly put to use by the Brexit “leave” campaign and Donald Trump’s presidential campaign last year.
What does this have to do with private companies subsidizing insurance or credit will you ask… Well, everything! As Big data analytics becomes more and more prevalent and the amount of data available about users grows in availability and scale, combining it all may finally provide insight into one of the most sought-after mysteries of all time: how and why do people act in a certain way. So far, studies which tried to examine why people would drop out of school, why people were more or less likely to commit crime or solve the mystery behind any other human behaviour, had to go through a very long and tedious process of longitudinal collection of masses of data, mostly via questionnaires, and publicly available data (income, age, gender, profession,…) and try to find some meaningful correlations which could explain the observed behavioural differences. But the outcomes were always shaky as it was nearly to impossible to account for all or at least a sufficient number of variables.
That may very well change in the near future. Lets take a very concrete example: how can we measure the impact of advertising of unhealthy food and beverages on eating habits?
Although there is mounting evidence that eating habits are influenced by advertising, culminating in last year’s WHO report, the major food and beverage companies always successfully argued that there are too many “factors” which determine eating habits, and therefore, that it is unfair to point fingers at advertising. But soon, the controversy may be over. Children are going online at an ever younger age, and in the near future, everything they see or do online or even offline (through Internet of Things, connected toys, their smartphones, and other devices which monitor their offline behaviour) will be documented to build up a “profile”, most notably to feed targeted advertising, but ultimately, such data could be put to a million other uses. For instance, data could be collected about which advertisements a child has seen, how many times he/she has clicked on them, or how such advertisements translated into eating habits later in life by comparing eating habits of millions of other grown up children exposed to similar advertisements. It may be possible to finally untangle the complex web of variables which determine behaviour: how much of a person’s eating habits is linked to family models, advertising, media consumption, the shelf placement of products and so forth.
Now imagine what this means. Making people pay higher health insurance premiums can only be justified if and only if you take the view that there is such a thing as 100% free will. But thanks to Big data, in combination with advances in neuroscience and other fields of study mentioned above, it will be possible to assess the various factors which influence your behaviour. In short, Big data analytics may evaluate that advertising is a factor which can explain up to 27% of your eating habits. And should such eating habits explain in turn higher health risk and thereby higher health insurance premiums, it is certainly fair to turn to the companies which are responsible for such eating habits and ask them to pay 27% of the health insurance premium.
The same could be applied to a myriad of other fields. For instance, falling into debt due to excessive/compulsive consumption and it’s relation to advertising, or make employers who pay women lower wages cover the excess credit risk such women present because of their higher risk of defaulting on a loan.
So far, it has mostly been used to maximize the impact of advertising, but Big data might open a huge Pandora’s box, as researchers start grasping just how much we could discover about human behaviour. Making links between advertising and people’s behaviour is only a tiny aspect of a much larger philosophical challenge dealing with the untangling of free will.
Alternatively, Big Data analytics and Artificial Intelligence may spell the end of certain financial products altogether. As individual risk can be more and more accurately measured, to the point of nearly knowing with certainty which risks an individual might face in the future, insurance products will make little sense as individuals could simply directly calculate how much money they need to put aside to cover for the financial cost of facing these risks. Insurance companies might be replaced by automated savings algorithms which set money aside based on risk calculations derived from the mass of Big Data available about millions and millions of individuals in real time, putting thousands of insurance professionals out of a job. This alternative, however, does not examine the underlying reasons behind individual’s risks and makes individuals cover their own risk, which may lead to exclusion of certain parts of the population.
Ultimately, we might find that such insight is dangerous for our societies, violating essential freedoms, individual rights, the necessary belief in free-will to preserve sanity, and that Big data should only be used to monitor global trends and shape the overall environment and context to be more conducive to positive outcomes for society as a whole. But since many private companies, especially financial service providers, seem keen on going down the road of peeping into the individual behaviour of people, there is no reason they should not face the full scale implications of their decision.