It was just in time for Rendez-Vous de Septembre 2024 that Moody’s Corporation announced its acquisition of Praedicat, a leading provider of casualty insurance analytics.
Discussing the deal, which supports Moody’s ambition to build out a comprehensive suite of casualty and liability modeling capabilities, Robert Muir-Woods (pictured) highlighted that Praedicat started life as a project in a research division of RMS, which was acquired by Moody’s in 2021. As a business designed to solve different problems, it required different expertise, he said, and so was established as a separate company, which had joint investments from RMS and RAND.
“The idea was that if you ask a question around quantifying liability risk, what do you need to do to actually quantify that liability risk? And it starts with tracking potential sources of injury or harm as early as possible,” he said. “The first idea was that you would use AI to search and scour literature to find the first indication that something has been identified as a cause of harm.
“Then, once you’ve identified it, you then track what is the potential development pathway of that harm. Is it likely to turn into liability in some form? And then what does the shape of that look like in terms of potential losses? That’s the process behind Praedicat and they are a super smart organisation of about 30 people who have done a fantastic job in that area.”
A key challenge facing the industry today is how to improve the technicality of the liability market, he said. Typically, a company may buy a liability protection product that runs the gamut of a wide range of potential losses, with viable liability losses mixed in with a range of other topics.
The question for the market is how to filter that out and find what to focus on. It’s a question of interest across the market, whether to the insurers themselves, or the reinsurers, both of whom are picking up the losses associated with liability risk. But it’s also a matter of consideration for the risk managers who are buying the coverage and making the decisions about whether or not to purchase additional protection according to where they perceive their potential losses.
“You can see situations - an obvious one is Monsanto and the litigation that focuses on the weed killer called Roundup,” he said. “It's costing an awful lot of money. And the question for the investor, in any company, is whether there is some potential for a major liability loss to emerge? All these areas are part of what Praedicat is exploring so far, and bringing its unique expertise, with great promise for expansion into new markets.”
With some liability suits, the matter is further complicated that the result may not come down solely to a scientific truth. It may come down to what the jury thinks, Muir-Woods said. What is clear is that there is the potential for liability claims to exceed the value of the company.
What is especially interesting about the Praedicat model, he said, is to see first-hand how much earlier risks can be flagged than when they’re making headlines. “It’s something I’ve been impressed with – how [the warning signs] may start off as small notes where somebody has identified a few cases in a hospital etc. You’re scouring back and seeing where there is a new potential source of liability, and then you’re starting to trace it, to track it. And some of the time these signs come to nothing, and other times you start seeing it expanding.”
Also interesting to see is how the development of Praedicat has aligned with the steadily growing interest in the role of modeling as a form of proactive risk management. The market is moving in the direction of more proactive solutions, he said, which is made possible because now things can be done at a much more granular level. Because of capabilities such as AI, it’s now much easier to single out individual risks, and it’s possible to track those risks in a way it simply wasn’t possible to do before.
Muir-Woods has seen a changing emphasis placed on data and data analytics, particularly in the last 20 years. There’s capacity to absorb and employ data in new and meaningful ways, that just wasn’t formally possible. Particularly when it comes to decisions around underwriting, there previously wasn’t the time and there wasn’t the data to model the underlying risk in every case, and so you had to find workarounds.
“It is changing now,” he said. “We can use data on past losses and employ that in some way. We can take a risk like flood risk, which we know varies on a very fine resolution. But there again, there's a potential to do things which were never possible before.”
