Tom Wilson: Okay. Yes. Let me €“ I’ll take a shot at it and Mario, you can jump in. So we use €“ I mean we’re a data-driven company, so we use predictive models as you know well, for just about everything. That could be fraud. There could be €“ do we think this claim might end up being severe enough where it gets represented by a lawyer? So it’s important for us to establish a relationship with the customers as possible. It might be, do we think there’s a better way to settle this claim, whether it gets €“ the car gets totaled or we send it to a body shop. So there’s we use predictive analytics throughout the business and obviously largely in claims as well. So we’re always tuning those. We think we’re pretty good at it.
You can’t really take one specific algorithm. But when you look at our claims severities, you can look at them externally. And when you look at absolute dollars, we think we did really well. It’s easier on physical damage, obviously, because you’re just fixing a car bodily injury, it’s like, okay, well, what was the case worth? What’s the average case? That gets a little harder to do. But when in €“ the only weakness in the external stuff is it tends to be a percentage increase over the prior year, which is, of course, we work in absolute dollars. And our models are done in absolute dollars. And so even though it all depends where you start €“ but we like our overall position. Mario, do you want to talk specifically about any models that you’re using now that think you can point to where we’ve updated and increased the value-added?
Mario Rizzo: Yes. The one I’d point to, and I think generally, the statement Tom made about like leveraging all the data and the capabilities we have, but also looking to tune those models and evolve over time. The example I would use would be around bodily injury, both potential loss identification and attorney representation. Given, obviously, the environment around us has evolved pretty significantly over the past couple of years in terms of higher levels of attorney representation and bodily injury claims and just medical inflation, medical consumption and treatment, those kinds of things. So what we’ve been doing is tuning the models to be able to use the components and the data that we gather early on in the claims process, to identify claims where there is, first of all, the potential for an injury.
More importantly, the potential for a major injury given it’s a higher impact accident or things like that. So we can get out ahead of the claim, make contact earlier and manage the claim much more effectively. The same would be true around claims that have the potential, ultimately to be represented by an attorney. Again, creating contact with a third-party claimant and establishing dialogue and communication and leveraging the tools and the models at our disposal to better manage the claim process through the bodily injury claims. So those are just a couple of examples of how we tuned models that we’ve had to adapt to the current environment. And we’re going to have to continue to, as I mentioned earlier, evolve our processes and those models to adapt to the environment over time.
So this is not a static process, and we’re always looking to get better based on the most current information as well as the external environment that we’re operating in.
Elyse Greenspan: Okay. Thank you.
Operator: Thank you. And our next question comes from the line of Andrew Kligerman from Credit Suisse. Your question, please.
Andrew Kligerman: Hey, thanks a lot for getting me in. First question is around social inflation, and in particular, bad price point. And in 2022, I think the court up a lot. And we €“ I would €“ I suspect that had a big impact. As we move into 2023, what’s your thinking about further social inflation issues? Do you think it will get materially worse? Could you give us some measurement around that? And that’s the question.
