The Case For Dynamic PML Limits, Revisited

The very first topic I wrote about when I launched the blog was catastrophe risk management. In the original post, I made the case that a company’s cat risk tolerance should be a function of market conditions and their capital position, not just a measure of amount at risk such as a PML target.

That post was made in an environment where ROEs on cat business were low and my advice was that companies should reduce their PML allowance to acknowledge the lower returns available.

We are now in the opposite situation. Returns are very high and thus companies should be willing to take risk above their historical limits to capitalize on the opportunity.

Let’s review…

The Two Guiding Principles

I outlined two rules to follow at that time.

No Post Event Capital Raise

First, a company’s cat risk tolerance should be such that they do not need to raise capital post an expected tail event.

This is much more relevant than a percentage of capital target as we have learned that a loss less than a stated limit isn’t always enough to avoid a capital raise. Also, we have seen losses far less frequent than 1 in 100 or 1 in 250 exceed the percentage loss estimate.

If you want to gain the confidence of investors to tolerate the volatility of cat risk, you need to be able to assure them you never need to raise defensive capital (at least not from a 1 in 250 occurrence event).

As a reminder of the perils of relying on “just in time” capital, look no further than Silicon Valley Bank’s inability to compete its equity raise that it thought would solve its problems.

You can’t count on capital markets to be open or friendly, so your cat tolerance needs to assume you can trade forward without external capital.

The one exception would be if you purposely kept a low debt/capital and communicated openly that you would raise senior debt to a more peer like level after a large cat loss.

One other tool I liked that used to be more common was contingent equity. This was like convertible debt but the conversion was at the company’s discretion, not the investor’s. This bypassed the capital markets risk from volatile markets as the company could have certainty about equity terms but only exercise if needed.

Incorporate Payback Period

Second, a company’s cat risk tolerance should reflect the current returns in the market and thus the time it takes to earn back losses.

This means if you can only write cat at a 10% ROE, you should take half the risk relative to a market where you can earn a 20% ROE. In other words, rate increases earnings power so you can make back your losses quicker in a higher priced market which means you can afford to take more risk.

If you are not going to raise capital, then you need to be able to restore your previous capital position in a reasonable time like 1-2 years to keep the rating agencies at bay.

For example, if cat returns are 12.5%, it’s reasonable to expose 25% of capital to a tail event. In this case, you would have restored your capital to your 1/1/23 starting point by year end 2024 (two years of “normal” losses would grow equity 25% offset by the 25% large cat loss).

While some may be tempted to assume a post event hard market and take greater risk, I think the past 15 years has shown that comes true a lot less often than hoped for.

Of course, this approach relies on a number of assumptions. First, it assumes the following year is a normal year which is by no means assured. Second, it speaks of occurrence PML when often trouble comes from an accumulation of events. For that reason, I prefer to look at aggregate PMLs.

Today’s Opportunity

So what should a reinsurer do today? If you believe ROEs on cat today are 20+%, you can argue for a 40% 1 in 250 PML.

While that sounds high, you can earn it back fairly quickly, especially if you get lucky and the next year is a no loss year. And, honestly, this environment is one where you could feel pretty confident a big event would lead to another leg up in pricing.

However, as noted, nothing precludes a second bad year from happening and you don’t earn the losses back. So you better be certain you won’t face a downgrade if you have a big loss and then only breakeven the following year.

If one were to run a PML this high, it makes sense to have strong second event retro to manage your aggregate curve as multiple events can really harm you.

This brings up another issue. There is always some outlaw who thinks it’s brilliant to max out all their peak zones to recycle their capital and maximize leverage. This will obviously impact your aggregate curve, so if you’re going to take this approach, you need to lower the occurrence limit to keep your aggregate exposure sane.

The Biggest Risk

Remember, the biggest risk in a hard market is not being able to participate because your capital is impaired. So while I do advocate increasing risk tolerances in a hard market, you don’t want to be such an outlier in a heavy loss year that you can’t grow into the further hardening of pricing.

It goes without saying that if you lose 40% of capital in an event, you may not need to raise capital, but you’ll have a hard time participating in the bigger opportunity the following January because you’re only going to be able to maintain your renewal book.

You also are putting a lot of trust in your models. If losses exceed the model, your expected 40% may be real world 50%.

You’re also effectively betting on no large quakes. If you lose 30% in a quake in the first half, you’re going to be sweating wind season really hard or scrambling to buy expensive retro.

If you write most of your wind book at June 1, you can drop participations, but if you took on a lot of that risk in January and a CA quake happens in April, you’re pretty locked in.

So before blindly adopting my advice, you obviously want to think through all these issues and adjust your risk taking appropriately.

Investor Disclosures

So you’ve decided to increase your cat risk. How do you communicate this to investors? You can say “we’ve increased our 1 in 250 PML from a maximum of 25% of capital to a maximum of 35%” and be done with it. But that would be lazy. It is not very helpful.

A better approach would be to explain why you raised your limit and what your framework is. For example, “when assessing our tolerance for cat risk, we incorporate our outlook for expected profitability and the potential volatility of that profit. We balance this with our current view of excess capital. Our framework leads us to believe that after a 1 in 250 event, we could restore book value no later than the end of next year and we would not need to raise any new capital.”

An even better disclosure would explain what a 1 in 250 event looks like. “Based on our expected market share of a number of potential global cat events, we believe it would take at least a $125B industry loss for us to lose above 35% of our capital to one event. Alternately, industry aggregate losses for the year would have to be above $200B.”

This would mean investors don’t have to guess whether your modeled event played out as expected. They can instead follow the industry loss estimates and estimate your share of the loss.

One more ideal statement would be “we have bought ILWs at a $250B aggregate level which would cap our loss at 40-45% of capital. We have issued contingent equity that can be exercised if necessary in that scenario.”

This means investors at least know they will not face a dilutive equity raise and should come out about even over the two year period. While this may lag other companies who took smaller losses, it at least avoids a painful spiral.

Final Thoughts

Obviously, every company will come to a different conclusion around how much to raise their cat tolerance in this market. I am not suggesting everyone should look at a two year payback window. That’s admittedly an arbitrary endpoint.

However, every company writing cat should increase their tolerances to some degree. And the first rule is more important to remember than the second…don’t get so greedy that you increase your odds of a dilutive capital raise if a major loss happens.