KCC’s Clark: industry needs to move on from “archaic” model customization

As reinsurance buyers and reinsurers negotiate the toughest property cat renewal in decades based on their own “view of risk”, Karen Clark has told The Insurer TV that the industry needs to modernize its approach to cat modeling, arguing that not all model vendors are the same.

“With all the advances in the science and computing power we need to adopt, embrace new methodologies [and] new ways of building accurate models,” the cat modeling pioneer said, speaking on The Insurer TV. “And that's what we've done.”

Clark launched the eponymous firm Karen Clark & Company in 2007, after founding what became known as AIR Worldwide in 1987, and developing the industry’s first hurricane model.

Cat modeling accuracy has been a major topic of industry debate, with the increasing frequency and severity of cat events calling into question cat models’ reliability.

Clark said some of the cynicism and skepticism around models’ accuracy – not to mention the fact that the market has essentially been cornered by two firms, RMS and AIR – was among the driving factors that led KCC into modeling.

“The whole reason KCC got into modeling was because of that cynicism and the skepticism when we started KCC,” Clark explained.

“We never planned to be in modeling, we had other initiatives underway, but the industry had lost confidence in the models. And there are a lot of complaints not only in the numbers, but the lack of transparency,” she said.

“So KCC actually went into the modeling business to address the skepticism, and basically to take up the challenge of building more transparent and accurate models,” Clark added.

Clark said one of the ways in which her firm distinguishes itself from its competitors is the persistent, rigorous testing of its models on a daily basis.

“One of the things we do very differently is we test our models every day, so our insurers can test them. We give these footprints for SCS – our clients know within a few weeks or months if our model was accurate or not,” Clark said.

Industry needs to move beyond black box models

The executive said that the near-constant feedback effectively acts as an “accuracy engine” that helps the firm’s models become more accurate over time.

Clark said that most of her firm’s current roster of clients is made up of primary insurers, which she said has largely been driven by the demands of that cohort of clients for models to be “accurate out of the box”.

That generally stands in contrast to reinsurers, who have acknowledged the most widely-used models’ limitations and often rely on teams of scientists and researchers to effectively make customizations to modeled outputs.

“It's almost like the model that they license is like a starting point,” Clark said, saying those companies then have to resort to using their own research to “fix the model”.

“And we think that's very archaic,” Clark said, adding that the industry needs to start raising its expectations over model accuracy.

“The models are self-contained black boxes [where] the only way you can fix it is to apply crude adjustment factors to the output. So that’s what's archaic: the lack of transparency, the inability to make surgical adjustments, which you can do with the KCC model.”

To that end, Clark said with KCC’s model, a company writing solar or green energy insureds can make “surgical” adjustments to model components to reflect the unique risk, rather than making crude tweaks to model outputs.

“So that's what I meant by it being archaic,” Clark said.

She added: “Our models are proven to be more accurate. Our insurance company clients don't want to take our word for it, they can prove it to themselves.”

Loss creep drives appeal of modeled loss transactions

Discussing the difficult ongoing renewal dynamics, Clark said rising litigation of property claims has led to an ever-growing tail for property claims, but that if parties to a transaction are confident in their modeling, modeled loss transactions can result in faster claim payments.

“Even for reinsurers, you want to know a year at least two years after the event, what your final losses are, and there's been this loss creep. Again, it's very focused on Florida or to a great extent focused on Florida. So what do you do about it?”

Clark said companies who enter into transactions using modeled loss numbers based on KCC’s models can mitigate the basis risk between a firm’s modeled loss estimate and its actual losses, while also settling claims much faster.

“The beauty of that to reinsurers and to the ILS market is you can settle the loss within a couple months, not a couple years. And so that prevents a lot of tied up capital. And so there's a lot of potential for this. And again, you need accurate models.”

“I think you could see a resurrection of this type of transaction, because it'll help. Certainly reinsurers and ILS investors will be much happier with this type of transaction, where you can see what the model says, there's complete transparency, and it settles very, very quickly.