The Era of Reserve Releases Has Ended

The Axis and Markel reserve charges over the last week weren’t the first of this cycle and, most assuredly, won’t be the last.

But I do think they serve as something of a marker. We have gone through a generation of reserves nearly always being favorable. That time appears to be over.

We are now entering a period where there will be more mystery, and with it more attention, around the impact of reserve changes on earnings. Let’s face it, investors would rather see a 92 combined ratio (CR) from a 96 accident year (AY) and 4 points of releases than a 90 AY with 2 points of additions.

While we can argue how much that really should matter, what clearly does matter is the direction of the change. If that company with the 96 AY only has two points of releases now, that’s a problem. Even if the 96 improves to a 94, if the releases disappear, that’s an even bigger problem.

If releases slip from 4 points in recent times to 2 points in the first half this year and 0 in the second half, investors will worry that next year will be 2 points of adverse, or even a big one time charge. Some gradual improvement in the accident year won’t be enough to overcome that concern.

What’s Changed?

In normal times, the peak of reserve releases tend to come from accident years 3-6 years old. There is then a smaller contribution from the 7+ year vintages and a small assist from the most recent years that need more seasoning before the releases accelerate.

We have seen many companies suggest that casualty reserves from 2015-2019 AYs are (or were) inadequate. These are 4-8 year old reserves at YE23 and thus should have provided ample releases.

Instead, they have required strengthening. This is a key negative delta to overall releases. This pattern likely would have continued into 2020 but…Covid.

The Covid years appear to be very light on claims, but there is debate on whether the claims have simply been delayed or disappeared completely due to the decline in economic activity.

Some companies will choose to release sooner than not, while others will take a wait and see approach. There is a good chance these years will prove redundant though.

So what’s there to worry about? We’re back on the path of releases!

Except…remember auto? Lack of activity followed up by tidal wave of activity. But for auto, it only affected new business. With casualty, there is good reason to believe that tidal wave of activity will impact all open claims from 2015-21 plus newer cases from 22 onwards.

While pricing has improved over the last five years, it is an open question whether it is enough to fully adjust for social inflation.

Lack of Conservatism

I would feel a whole lot better about post-Covid casualty reserves if companies were doing what they did in 2002-2005. Back then, they set their accident picks closer to the developed loss ratios from the bad years to build cushion in for future inflation.

This is what laid the foundation for the tidal wave releases of later years. The combination of high picks and low inflation allowed for large future releases.

Conversely, recent year AY picks are similar to the starting points of five to ten years ago, even though those years were set far too low in hindsight.

There is certainly an argument for that based on higher pricing achieved in recent years, but that is a guesstimate and also assumes that the bad years have now been fully reserved, which is a wish more than a certainty.

More likely, the actuaries are stuck in their ways and tend to keep picks the same regardless of changes in the environment. The lack of conservative picks during a period of high loss inflation suggests these years will not experience the same degree of redundancy as those of 2002-2014.

That means you’re really betting on those Covid years being massively redundant. There is no place else for the releases to come from.

Litigation Funding

I have admittedly skipped over an important element of the story. Why did casualty years start to deteriorate beginning in 2015? Was it just the cumulative effect of lower pricing?

That is a part of it, sure, but there is another, larger story – litigation funding. It can’t be coincidence that 2015 is when we saw mentions of litigation funding start to pop up in the insurance world.

How do I remember it was 2015? I don’t, other than directionally, but I found a way to confirm it. Burford is the largest company involved in litigation funding. Here’s a stock chart of Burford in 2015 (forgive the small print).

As you can see, it was up more than 50%. Guess what it did in the three years before that?

That’s right. Nothing. The gains began in 2015, coincident with the first year of poor casualty reserves. Guess how it did in the years after 2015?

Even better! It went up 15X!!! This was one of the hottest stocks around (it ran into problems later in 2019 due to accounting concerns). If you don’t think this correlation implies causation, I don’t know how else to convince you.

FYI, after the fall from the accounting headlines and Covid gumming up the courts, the stock went flat through 21 and 22.

But guess what it’s done in the last 12 months? It’s doubled. How does that make you feel about post Covid casualty reserves???

The Compounding Effect

Here’s the real problem with not setting picks on recent years with caution. Let’s assume you set initial picks at 65 in 2015-19 and now 2019 has deteriorated to 80. If pricing has doubled since then and loss trend is up 50%, you can argue your expected LR is now 60 for 2024 so you’re being conservative using 65.

The problem is what if 2019 keeps developing to an 88? And let’s say you overestimated the Covid benefit and trend is really up 70% since 2019. Guess what? You’re writing new business at a 75 instead of 60.

To help with the math, the 50% increase assumed 8.5% loss trend CAGR where the 70% is 11%. If you don’t think it can be 11%, look at that Burford stock chart again.

Also, remember the levered effect if you’re writing excess. Higher underlying loss inflation increases both frequency and severity in excess casualty.

If you are trying to figure out which companies to be nervous about, start with those with the biggest excess casualty books. They have been swimming in excess profits for a long time, but that is going to unwind. They have the biggest litigation funding risk.

The Earnings Impact

So, let’s add this all up. Reserves prior to 15 are too small to move the bar. AYs 15-19 are bare and maybe still short. Covid releases may continue, especially for those who have been slow to release them so that’s one positive. AYs 22-24 certainly won’t release anything near term, if ever.

But there’s still a few more things to consider. How many times have you heard companies mention they recognize bad news before good? This means that they wait a few years before releasing from the most recent AYs, but will put up extra case for any bad claims news right away.

This hasn’t mattered a lot in the past, as the practical effect many times was for companies to release a little extra (wink, wink) from older years to net out the bad news. That is no longer an option.

Also, when AYs were set with an anticipation of future releases, then some bad news was effectively already in the numbers so you didn’t have to recognize every blip. But if current picks are more down the middle, then there is less room to absorb any bad news without it affecting earnings.

Put that all together and overall reserves will be a net drag on earnings for the next few years. At best, they will be neutral.

Finally, consider that, at some point, companies will bite the bullet and start setting new AYs with more cushion. That means earnings estimates for 25, 26, and beyond are likely too high.

Silver Lining

OK, this was a bit gloomy, but let’s end on a bit of a positive note. The industry got too comfortable with perpetual favorable development. With less margin for error, underwriting standards will need to rise and discipline improve.

From an investor standpoint, once the initial pain works it way through, the lower earnings will be worth a higher multiple. Reserve releases, all else equal, are worth 1X book. Accident year earnings are worth a P/E multiple that is usually a healthy premium to book.

Earnings quality is better if the 92 CR mentioned above is something like 93 AY – 1 point of releases rather than the 96 – 4.

But we got to cross through the valley first before we get to the next summit.