A former governor of the Reserve Bank of India (RBI) once famously said, “Everywhere in the world the future is uncertain. In India, the past too is uncertain.” He was referring to the vast amount of unreliable data routinely used to assess the health of the Indian economy. When it comes to the health of people, which the insurance companies are interested to know before they underwrite their life and health policies, the problem gets even messier.
Data in India never accurately reflects reality. We may say mathematics is the only truth, but statistics—a branch of mathematics—can spin any story.
Insurers like to explore the truth about the pain points of life: how long we live, when do we die, which illnesses we will suffer from…. All existential questions that lesser mortals will struggle to answer. But underwriters in insurance companies believe they can find these truths, and unfortunately they depend on past data.
Underwriters are a key pillar of the insurance business. The other pillar is the salesperson. But there is a saas-bahu (mother-in-law and daughter-in-law) kind of tension between them. Underwriters believe that salespersons bring bad risks to the book, and salespersons think the only bad risk is losing business to competitors.
Having worked on both sides of the fence, I wonder why we continue to conduct the insurance business—be it underwriting or sales—the same way as before even though the world has long moved on.
The job of underwriting is to ensure that the price fixed by actuaries for the insurance product, after deducting expenses, is enough to pay for the claims. They want customers to pay the right price for the right risk. If they don’t, a free floating price can bankrupt any business.
In any normal product, price is determined after adding an element of profit to the costs. Costs are relatively easy to determine, by adding the production, administration, and distribution expenses and dividing them by the estimated volume of sales.
But in insurance, what is the production cost? What do we actually produce? We produce a bunch of promises.
A policy is a promise to pay a certain amount on an uncertain event. Now, if the event is uncertain, it’s anybody’s guess on how to deal with its likely cost.
Underwriters believe that if we ask people a few questions, they will answer truthfully. This is the problem. Not many applicants for insurance are like Mahatma Gandhi or Mother Teresa. Sometimes they lie about their health, family history, and other details. Worse, even if they tell the truth, underwriters may not know how to interpret it.
There have been many hilarious instances with me. I referred a case where a senior medical doctor, who wanted to take insurance, dutifully mentioned a minor health condition that seemingly had no bearing on the risk to be insured. His case was quickly rejected. When the doctor contested the decision with medical evidence, the underwriting team—which had hardly any medical knowledge—promptly referred the matter for second opinion to a veterinary doctor. When the incensed applicant threatened to sue the company for gross impropriety, they quietly accepted the proposal without a murmur and, of course, there was no claim. The ground reality is that the practice of underwriting in India is nearly reduced to a pseudo-science. Not because of underwriters who diligently follow their professional rule book, but because of the system in which the businesses operate.
In another instance, knowing the importance of data-based underwriting, I diligently collected relevant data to present a logical case for fair pricing in health insurance for a large group. But the bored manager brushed aside my presentation and simply asked how much he should drop the price to clinch the deal from his rivals. He was plainly not interested in any analytics.
Maybe he was practical. He knew data in India is rubbish. Worse, even if it’s accurate, it’s ineffective in predicting the future, because of the interplay of mutually dependent variables that sway the health care costs.
If past health data sourced from insurance applications is ineffective, what about current information from lab tests when companies send their prospective customers to diagnostic clinics before insuring them?
This is another charade, especially in smaller cities and towns. If you believe that all the medical reports that underwriters examine every day with obsessive fervor before they accept a risk, are genuine, you are naïve. The applicant never went for the test in the first place. Though there’s a process of photo identity, it’s a farce. The agent manages a proxy test and the customer foolishly believes that by not going to the lab himself, she is saving precious time.
The question here is not about ethics in insurance or medicine. It’s about the futility of trying to solve a statistically difficult problem with scarcely reliable data. Why take underwriting seriously, when you don’t know whether the historical data is true or not? The traditional approach is not very helpful in evaluating risk.
Once, a reputed global society of actuaries undertook a study on the efficacy of underwriting in reducing losses in insurance. What would happen if we completely do away with underwriting? The study revealed that the cost of claims would go up, but the companies would make more profits because they saved huge expenses of underwriting and brought in more customers who preferred the hassle-free process they introduced. The gains of no underwriting outweighed the losses of adverse selection (where bad risks enter the book and cause more claims).
But how to price accurately for the claims?
Newer technology and big data provide some answers. You can move away from traditional underwriting based on historical data. Social media information, wearable devices for health and telematics for vehicle insurances provide a more dependable picture of risks involved. If you add the algorithms of dynamic pricing of advanced economics, you will have a new-age insurance company that asks you no questions but knows the right answers about how much to charge you as premium.