Facts behind Insurance

The Target parking lot in Edison on Nov.27,2015. (Yiyi Zhang)

There are risks everywhere. So we purchase insurances for many things. In the insurance industry, there are many things that most people are not aware of.

Insurance is divided into two main parts: life and property and casualty (P&C). Generally, the life side is more stable. Meanwhile, the property and casualty side includes many probabilities that accidents may occur such as car accidents and natural calamities. These are hard to predict.

Insurance companies collect premiums from customers. Some of the money will be invested into their development, and other money will be put into account as reserve which may be used for the coming losses.

Insurance companies should have relatively accurate predictions for the aggregate losses of all their policies. For different insurance products, the companies hire groups of actuaries to work on various models and predict losses that may occur.

There are usually three steps to estimate aggregate losses. Estimating frequency is the first step. Frequency means the number of accidents that may happen in the future. Then the companies should estimate severity, which means the loss amount that one accident causes. Finally, combine the frequency with the severity and you will get the prediction of the whole loss.

To build these models, it requires not only many statistical and mathematical methods — but also past data — because they can increase the accuracy of predictions by testing these models with the past data.

However, Srinivasa Ramanujam, who teaches actuarial science at Columbia University, said that many insurance companies are short of these data or the data were modified by either truncation or censoring.

Almost all the insurance policies have deductibles. How does the deductible work?

For example, one people has a car insurance with $500 deductible. Unfortunately, a car accident happened to him and caused $400 loss. The insurance company will not pay for the loss since it was within the deductible amount. If the loss was $2,000, he would be paying $500 and the insurance company should take care of the rest loss.

As a result, people usually don’t tell their insurance companies about their small losses. Because if they do, the company will enter the information into companies’ systems and their premium will be increased next year. They won’t be able to get the loss from insurance company after all.

However, these small losses did occur so the data has been truncated.

Some policies have limits. If an accident occurs and causes loss, the insurance company will not pay for the amount of loss that exceeds the limit. Insurance companies only record the maximum loss they paid. This is called censoring.

Because of truncation and censoring, the past data is not complete. It leads to more challenges to loss predictions.

This is information about the whole picture. Different people pay for different amounts of premiums. What do the differences relate to? Car insurance, for example, is related to drivers’ age, gender, cars’ makes and models, and so on.

Steven Armstrong is the chief pricing actuary for Personal Lines, Mortgage Services and Crop at QBE North America. He said that there was research about this decades ago. It shows that credit score has the strongest correlation with individual loss among all the factors such as gender and age. Since then, most insurance companies started to use credit score to calculate premiums, but the causality of the correlation was unknown.

As technology develops, usage-based insurance (UBI) becomes more and more popular.

Progressive’s snapshot program is UBI. It provides customers a plug-in device to collect their driving data, including their driving times, frequencies, swerving speed and so on. Progressive receives much valuable data through this program in order to build better models. It also tells how good you are as a driver. In return, the company gives “good drivers” discounts based on the data they collected from the device. The company will not increase premiums of other customers.

This device can also collect the data that where do customers drive their cars but the insurance company can not access these data. The information of customers’ locations is either stored in Google database or Apple database. If insurance companies can access these data, they would be able to predict losses more accurately. However, privacy remains an issue.

Armstrong believes that UBI is the future of rate making.

Shaun Wang is a member of International Association of Insurance Supervisors. This association wants to build a global model that can evaluate all the insurance companies. Their final goal is to achieve global financial stability.

Wang gave a speech discussing worldwide insurance regulation at Columbia University on Nov. 23. He tried to use 12 insurance companies’ data to build a international standard for insurance companies.

“No matter what they (IAIS) come up with, there won’t be the right formula,” Wang said. “ Companies are different. Risks from one country to another are different. Even car insurance, in Germany is very different from in the U.S. That’s why we need academics to involve.”

Insurance is an essential part in our life, because it distributes risks and provides financial stability. Facts talked above is just the tip of the iceberg. The business behind insurance is far more complicated than we usually think about.

A single golf clap? Or a long standing ovation?

By clapping more or less, you can signal to us which stories really stand out.