Yes, sure. So let me try to take that in pieces. So the 9% we referenced is for GL. That was in the overall casualty of 8% and then 8.5% we had talked about, excluding workers' comp. So the 9% is GL specific. And as we said in the beginning of the year, that didn't assume any frequency change. So you could assume then that, that 9% is entire severity driven. But now if you look back, because obviously, historical loss trends -- actual historical loss trends influence how you think about forward loss trends. And if you were to look back, and I think we've been as transparent as anybody in terms of our expected loss trends, and we've been doing that for several years, if you look back over the last four or five accident year, so let's just focus on the period that is in everybody's focus right now, which is that '20 to '24 period. Our expected loss trend over that time was about 4% in 2020. It's now at 9%. But if you look on average, from '20 through '23, our average expected loss trends was about 5% over that timeframe. Our current evaluation of those years, and I'm careful not to call it loss trend, because technically, it's not going to be, if you just think about the -- what we're seeing in terms of average change in frequency and severity, which gives your average loss cost change, what we've seen over that time period on an updated basis would say rather than the 5% we assumed on average, it's closer to 6.5%. And again, I think it's important to understand the pieces there, and you heard us refer to frequencies, that 6.5% is made up of frequency, which has been running better by about 4%, give or take. And then severity trends have been running closer to 10%. So you blend those two together to get to 6.5%. So when you look at that and then look at where we have 2024 at 9% with no assumed frequency drop, we think that we feel pretty good about that. But then also remember, you're taking that 9% now and applying it to a higher starting point because your base years is the last -- more recent accident years, which have moved higher, you're applying that 9% forward trend to a higher starting point. And that's how we pull the pieces together. And we think that's reasonable and appropriate, and in line with what the more recent years have been telling us. With regard to history, the longer-term history, Mike, I think this has been somewhat unprecedented in that the entire diagonal has been moving. All accident years and all evaluation periods have shown a movement higher in terms of severity emergence. And again, I think that's just another data point to suggest that this is a more widespread industry phenomenon.