Reserves for auto liability are supposed to be some of the easiest to analyze. Paid patterns are very stable, there are large numbers of claims, and duration is relatively short.
If you booked your auto reserves short, it was probably because you chose to rather than because you missed a changing trend. And even then, you’d have to correct it by the next year.
However, something pretty dramatic has happened in the auto world over the last ten years. Payment patterns keep getting longer.
This has really wreaked havoc with reserving and, collaterally, pricing. If the actuaries didn’t understand that claims are paying more slowly, then they would see stable paids year over year, assume required reserves were the same as last year, and they could thus keep prices flat.
But, if those stable paids were due to a change in the payment pattern, then ultimate reserves will be higher and more pricing would be needed.
This is a big part of how the industry got so far behind on pricing. Sure, the increased complexity of parts and inflation at body shops played a role, but that’s the physical damage component.
What I will show here is the changing liability tail has been just as prominent a problem for auto insurers.
More Conservative Reserves?
When you look at the big auto insurers, you find a repeating pattern. They used to have much higher first year paid to incurreds.
The table below is first year paid to current incurred. Normally, I prefer to look at initial incurred, but that is a better indicator of how companies choose to reserve. If we want to see if patterns are changing, we should look at the developed reserves.
Couple of notes: first, each company has different levels because of different business mixes, so focus on the directional change rather than the absolute levels of each company. Second, Covid messed up the ratios for 20 and 21, so skip over those.
| Initial P/Current I | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 |
| State Farm | 41.4% | 41.9% | 42.2% | 41.6% | 40.3% | 39.4% | 36.1% | 34.1% | 34.7% | 35.1% |
| Progressive | 49.5% | 46.6% | 46.3% | 45.5% | 45.2% | 45.8% | 43.3% | 40.7% | 40.5% | 40.8% |
| Allstate | 38.6% | 41.5% | 40.2% | 37.4% | 36.6% | 36.3% | 35.0% | 31.0% | 32.6% | 32.0% |
Normally, one would look at that and say, wow, reserves have gotten so much more conservative since 2016. If paid patterns are stable, then a declining P/I suggests more IBNR is being put up and the reserves will develop favorably over time.
The problem is those future reserve releases never emerged. More on that in a minute.
Let’s first look at State Farm and Allstate since they have more mature books. They each showed 3-4 points lower P/I between ’16-’19 and another 4 points from ’19-’23.
So all in, it’s about a point a year drop. That’s a dramatic change. If it declined even half that amount, it would be a big change.
Progressive moved a little slower from ’16-’19, but has seen similar changes since. They also have a different driver mix than the others, which could be seeing somewhat different trends.
But all in all, there is a definitive move towards putting up much more initial IBNR for auto liability than in the past.
We have two possible explanations for this. One is that companies are reserving more conservatively. The other is that the tail changed and claims are paying out slower.
Let’s first look at the latter. Here is the same info as above, except now I will use the third year paids instead of first year (so the paids as of YE 2019 for the 2017 AY).
| 3rd Year P/I | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 |
| State Farm | 83.4% | 84.4% | 83.4% | 82.7% | 81.2% | 79.6% | 79.4% | 79.7% |
| Progressive | 90.7% | 89.7% | 89.4% | 89.7% | 89.2% | 89.8% | 88.1% | 89.4% |
| Allstate | 83.1% | 81.5% | 79.4% | 78.1% | 78.0% | 79.4% | 80.5% | 82.1% |
We can see after three years, the P/I trends are largely stable (though less so at State Farm) as we would expect. This suggests that claims that used to be paid in less than 12 months are now extending out to two or three years.
Hopefully, I’ve convinced you that the payment pattern has changed, but just to be sure, let’s check whether the better initial ratios are signaling that the reserves are more conservative.
Reserve Development
Here is the reserve development over time for the same companies.
| Adverse Development By AY | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 |
| State Farm | 192K | 1.2M | 510K | -137K | 195K | 536K | 146K | 2.4M | 1.3M |
| Progressive | -130K | 73K | 154K | 79K | 169K | 42K | 14K | 497K | 527K |
| Allstate | -16K | -23K | -59K | 155K | 272K | 391K | 112K | 1.1M | 87K |
Wow, that’s pretty terrible. You wouldn’t expect a line like auto to have such consistent adverse development. Sure, most years the numbers are small, but 22 of 27 AYs across the three companies have developed adversely.
So that definitely answers the question about why the P/I ratios have improved. They aren’t a result of extra conservatism to drive future releases. Rather, they were a realization of a longer paid tail.
IBNR Trends
Before getting into why this is happening and how long it may continue, let’s look at one other metric – initial IBNR.
If the tail is getting longer, then we should see one of two things. If the companies have their arms around the trend, then we would see higher levels of case with little change in IBNR.
If the companies still haven’t figured out what the new payment patterns are, then they will put up more IBNR initially to reflect the uncertainty.
| Initial IBNR/I | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 |
| State Farm | 31.1% | 27.5% | 32.9% | 37.6% | 38.4% | 35.2% | 42.3% | 31.9% | 39.0% | 36.8% |
| Progressive | 13.5% | 12.5% | 13.0% | 14.6% | 14.9% | 14.4% | 16.9% | 13.8% | 15.5% | 19.1% |
| Allstate | 12.4% | 11.1% | 12.3% | 13.0% | 13.7% | 14.2% | 18.4% | 16.1% | 19.1% | 22.8% |
As you can see, first year IBNR to incurred ratios have risen significantly (skip over ’20-21 again), especially at Progressive and Allstate. The companies have consistently underestimated how much the tail has lengthened and have to keep chasing the trend.
This is why we see continued adverse development. They thought they had increased the IBNR enough, but keep finding out they’re wrong.
The actuaries are hopelessly behind trend and keep throwing more IBNR at it hoping to eventually get it right.
So are we there yet? Who knows? One would hope so, but you probably would have hoped so five years ago too.
A lot of this depends on why the tail has gotten longer.
Attorney Representation
I’m sure there are some other reasons, but the most obvious one is increased attorney representation. While I don’t have stats handy, there is plenty of evidence that more people are lawyering up when they’re involved in an accident.
As more drivers with claims use lawyers, naturally, cases take longer to settle. Thus, the first year paids are lower, but the ultimate losses are higher, which results in the adverse development.
Your guess is as good as mine as to what the limit is on litigation activity. I think there are now more ads for auto injury lawyers than auto insurance!
It’s reasonable to assume we must be approaching the peak, but I think this is a trend where you’d rather call it after the inflection than hope for the best by assuming it can’t get worse.
It’s certainly possible that the spike in IBNR levels over the last two years is overkill and won’t ultimately be needed but, remember, 2021 and 22 were the worst adverse development of the cycle, so that doesn’t seem like a good bet.
Pricing Uncertainty and the Growth Challenge
One final point worth addressing. As long as payment trends are unstable, it’s very difficult for an insurer to be confident in its pricing.
Yes, I know prices have gone up a lot, but, as we see from the results above, that hasn’t helped combineds much yet.
For insurers to be willing to enter growth mode again, they have to be confident they have fully captured the change in trend. That just doesn’t seem possible at this time.
It may be reasonable to feel better about physical damage inflation, but the bodily injury situation is just too speculative at this time.
If you turn on the growth now, just because severity has improved for a couple of quarters, you’re taking a big gamble on what the legal environment is like three years from now.
And if you’re growing by giving a price cut to someone who improved from a 110 CR to a 105 and hopefully is going to 100, well, they don’t deserve a price cut. We are still far from even average combineds, let alone peak.
What makes this different from past cycle bottoms is those were dealing with temporary shocks to loss trend. So once those passed through and rate increases were achieved, it made sense to accelerate growth.
But this time, you are pricing with much less certainty that the trend has passed. So if you grow, you can really impair your franchise when you have to jack those customers rates up a year out.
One other thing to consider. In most hard markets, price and terms improve. Nobody is changing terms on an auto policy that I am aware of.
I’m not even sure what terms you can come up with that would prevent someone from using a lawyer. It’s not like you can exclude certain types of accidents.
So if trend is permanently higher, your only tool is price which suggests one shouldn’t look to grow until they get their CRs all the way back to normal.