One thing I like to do here is make predictions. When those predictions come true, I am pretty good at reminding you about it! That also requires owning up to it when they don’t come true.
Before earnings season began, I suggested we would see companies take year-end reserve charges, especially if they could spin them away as a show of strength. However, we only saw a few companies admit reserve weakness and most of those were the ones already struggling.
So, what happened? Was I wr-r-r-r- not correct? In the near term, yes. In the long term? To be determined…and likely in my favor. Every data point reported this quarter remained consistent with the thesis that loss trend is getting worse and is not fully reflected in most companies’ reserves.
However, most companies made clear they have no desire to go to confessional yet. Thus, there was a lot of bravado about how “things are bad, but not for us“.
The Pseudo Confession
The one admission that was popular this quarter was the “don’t assume pricing will fall to the bottom line yet“. In other words, even though pricing appears to be ahead of loss costs, you shouldn’t model improving margins. Yes, some of this was the written vs. earned argument, but there was more than that going on.
There were a number of companies who suggested that even though they are in theory earning better margins, they’d rather be prudent and not book it that way until it becomes clear that trend has stabilized. This is certainly a very appropriate stance and who can argue with that?
Me, of course! If you are truly up to date on your recent accident years, then, yes, that is clearly the prudent path to take. The problem lies in if you are behind the curve. For example, if you are booking a 67 knowing it’s probably a 72, then the reason you aren’t showing improvement from pricing is because the actual improvement is off a 72, not the 67.
Let’s do a simple example:
| 2016 | 2017 | 2018 | 2019 | 2020 | |
| Premium | $100 | $100 | $100 | $104 | $110.24 |
| Rate Increase | 0% | 0% | 4% | 6% | |
| Loss $ | $66.60 | $69.26 | $72.03 | $74.91 | $77.91 |
| Loss Trend | 4% | 4% | 4% | 4% | |
| Ultimate Loss Ratio | 66.6% | 69.3% | 72.0% | 72.0% | 70.7% |
| Initial Pick | 67.0% | 67.0% | 67.0% | 67.0% | 67.0% |
As you can see, there is 1.3% of “true” LR improvement in 2020 but since they picked 2019 5% below the ultimate, they can’t show any GAAP improvement in 2020 because they are in catch-up mode.
Thus, my takeaway is what most companies were telling you when they said “we’d rather be cautious in assuming margin improvement next year” was actually “you shouldn’t trust what we booked last year at“. That’s the closest anyone is going to get right now to confessing their reserves are short.
The other way a company could obscure a reserve problem is by offsetting it with releases elsewhere. For example, a company might have shown diminishing, but still positive, total reserve releases and not disclose the change in the moving parts.
Areas like general liability and commercial auto may have flipped to meaningfully adverse and this was offset by accelerated releases out of workers comp. The problem with this approach is the comp redundancy is beginning to shrink due to pricing pressure so one can only play this game for so long.
Grading the Self-Graded Exam
The good news is we are only weeks away from getting 2019 reserve triangles where we can test some of these hypotheses.
One of the simplest ways to assess whether companies have booked recent years short is to look at the initial paid/incurred ratios. If these numbers are getting worse, it’s a pretty good sign that recent years have been booked too light.
Another quick test is to see how the new picks compare to the developed recent years. Take the example above where the company continuously set their pick at 67. If the developed triangle looks something like this, it’s a good sign they are behind the curve.
| 2016 | 2017 | 2018 | 2019 | |
| Initial | 67 | 67 | 67 | 67 |
| 1 Yr Later | 66.9 | 67.7 | 68.8 | |
| 2 Yr Later | 66.8 | 68.5 | ||
| 3 Yr Later | 66.7 |
If a company has seen the prior two accident years develop adversely off the 67 pick, it is awfully optimistic to assume the 67 pick will hold in the recent year given the loss trend pressure has only gotten worse.
Similarly, we can assess if comp releases were accelerated to hide weakness elsewhere if we see the 2018 year had a larger initial release than normal or the 2019 pick is a couple points lower than recent years initial estimates.
Bad Before Good?
One of the things you tend to hear companies say, at least in good times, is the reason reserve releases come out slowly is they acknowledge “bad news before good”. In other words, if there are signs of adverse trends, they add reserves immediately, but if there are signs of good news, they delay incorporating it until they have more evidence.
This would suggest pressure from adverse litigation trends should be showing up pretty quickly in results. And yet – with a few exceptions – it hasn’t. Excess players certainly seem to be playing the “we don’t think it will reach our layer” game (and then, when it does, professing shock) while the primaries don’t want to put up big reserves that could be discoverable.
Bad before good only applies when balance sheets are strong. It’s postpone the bad when reserves are all wrong.
So What’s It All Mean?
Reserves are like a balloon. If you squeeze on one part, it expands somewhere else. The savvy move would be to take your reserve pain now so you can get your current year right sooner and get your earnings growing again.
Unfortunately, many companies have decided to suppress reserve additions because of fear of negative perception. The problem is this pressures forward earning estimates by forcing the pain onto the current accident year. This may work in the short run, but it’s not a wise long term trade.