The not-so-secret sauce of ISO Loss Costs

October 21, 2020 / 4 min read

Loss costs are a key ingredient in helping insurers price policies. No insurer (that we know of) owns a crystal ball that can foretell the losses on any given policy they write. But they can access the next best thing: a robust statistical database and models that they can leverage to help guide their predictions of future losses.

As most statisticians will tell you, when it comes to data, you can’t have too much of a good thing. But even large, national insurers often have just a fraction of data when compared to the entire addressable market.

2020-10 inline

That's where Verisk comes in.

Our statistical database of insurance premium and loss data is one of the world's largest. By harnessing the power of aggregated data, we’re able to offer insurers a critical ingredient—a reliable benchmark informed by the collective experience of hundreds of insurers across multiple lines of business. This database can provide broad, statistically robust insights into future loss costs in a way that individual insurer’s data assets often can’t.

But simply collecting data isn’t enough. Instead, we subject it to rigorous quality controls to help ensure its validity, consistency, reliability, and accuracy. Our staff of actuaries, data scientists, and insurance experts then leverage a number of sophisticated techniques to help yield useful, predictive results. Think of these techniques as the not-so-secret sauce that helps transform raw experience data into something of much greater value to end users. These techniques fall under four broad categories: loss development, loss-adjustment expenses, loss cost trends, and external factors.

Loss development

We want to use the most recent available data as the basis for our loss costs because it best reflects current social, legal, technological, and other factors that affect losses. But because it takes time to learn about, settle, and pay claims, the most recent data is very often immature. That’s why we use a loss development model to project early estimates of losses to their ultimate settlement level. To do this, we examine historical patterns in changes to loss estimates from an early evaluation date—shortly after the end of a given accident year—to the time, several or many years later, when the loss is mostly likely to be settled and paid. We also calculate loss development factors that allow us to project the data from a number of recent years to the ultimate settlement level. We use this projected—or developed—data as the basis for the rest of our calculations.

Loss-adjustment expenses

In addition to paying claims, insurers often pay a variety of expenses related to claim settlements, such as legal fees. Many insurers allocate some of those costs to the particular claim. Other costs sometime appear as overhead. Verisk collects data on allocated and unallocated loss-adjustment expenses and reflect those expenses in our loss costs.

Loss cost trends

Losses adjusted by loss-development factors and loaded to include loss-adjustment expenses tend to give the best estimates of the costs an insurer will likely ultimately pay for past policies. But an insurer still needs estimates of what future losses might be. To produce those, Verisk builds models to reflect two components of loss cost trends— claim frequency and claim severity—by examining recent historical patterns in losses to understand the effects of changes in the number of claims per unit of exposure (the frequency) and in the average cost per claim (the severity). Accounting for these trends in our loss costs can help more accurately anticipate what losses may be on policies that will be written.

External factors

Because losses rarely occur in a vacuum, we also model the potential impact of external factors—such as shifts in the macroeconomic environment, regulatory changes, and behavioral shifts—on losses. For example, for auto insurance, we look at changes in speed limits, road conditions, traffic density, gasoline prices, the extent of driver education, and patterns of drunk driving. We closely monitor the regulatory and legal landscape for developments that could potentially impact future losses.

For each line of insurance we support, we perform thousands of separate reviews per year to inform our loss cost estimates, freeing insurers to focus often stretched analytical efforts on more strategic initiatives. This business intelligence helps us develop methods to appropriately adjust the developed losses and loss-adjustment expenses to the future period for which cost information is needed.

An investment in the future

Mix these four techniques together and you get solid data to help you determine prices based on coverage, state, territory, class, policy limit, deductible, and many other categories. More than that, you can access estimates based on one of the largest, most credible set of insurance statistics in the world, produced by some of the industry’s most well-respected actuaries and insurance experts.

Learn more about ISO Loss Costs.