Missed the introductory post? Click HERE to learn more about iansbnr.com!

Following the hurricanes of 2005 (KRW), investors began to demand more disclosure from companies about their exposure to natural catastrophes. At the same time, the ratings agencies implemented stricter requirements for cat exposure, largely based on limits around the 1 in 100 occurrence PML.

Naturally, investors began to ask what those 1 in 100 PMLs were. Some companies disclosed them promptly while others held out for a few years, but eventually 1 in 100 disclosures became the norm.

Unfortunately, there were many flaws in these disclosures that made them difficult to compare across companies (I will likely say more about this in a future entry). I have also long argued that occurrence PMLs aren’t nearly as useful as aggregate PMLs. When companies experience stress due to cat losses, it is usually from an accumulation of events rather than a single loss.

Which brings us to the main topic of this entry…what is the purpose of PMLs? It is not for investor disclosure but rather to manage balance sheet risk. Most companies manage to what the rating agencies require which is a single point estimate aka the 1 in 100 as a % of capital. What I will argue below is that is a poor risk management approach.

The PML risk tolerance should NOT be based on a percentage of capital at risk. Instead, it should be based on two items:

  1. How much capital can I lose before I would have to raise capital?
  2. How long does it take me to earn back my loss?

Capital Risk

When Irma was approaching Miami, apparently the mood amongst a number of reinsurers at Monte Carlo was one of fear, according to subsequent reports. I naively asked some executives, “why”? After all, insurers base their entire risk model on being ready for the 1 in 250 in Miami. It would have clearly been a blow to book value and excess capital but the doors should have been open the next day and ready to go.

The answers I got really floored me. Many companies “plan” for the most obvious risk on their books (the Great Miami hurricane) by planning to issue capital post event! Um, what? In other words, they have no plan! 😲Thus, the great sighs of relief when Irma steered westward!

I don’t know what to call this other than malfeasance. It is an abdication of responsibility by the board and this risk is certainly not spelled out anywhere in a 10-K. Trust me, I would have noticed a risk factor that said “if a 1 in 250 hurricane happens and we lose money in accordance with our disclosed risk tolerance, we will still be facing ratings downgrades and have to raise desperation capital on unattractive terms.”

Conclusion #1

A company’s cat risk tolerance should be such that they DO NOT need to raise capital post an expected tail event. Note, this implies that PML limits should adjust real time with changes in excess capital.

Payback period

The title of this post mentions “dynamic PML limits”. What does that even mean? It means your limits should change based on the company’s circumstances rather than be set once and subsequently forgotten.

Let’s do an example. If my cat book is earning a 20% ROE, I can perhaps be comfortable exposing as much as 40% of my capital to a 1 in 250. I will generate earnings equal to 20% of my capital (if I am a pure play cat writer) if the rest of the year is “normal” and another 20% the year after that. So my capital is fully restored one year post event.

Now let’s suppose an environment closer to today. ROEs are perhaps 10% on cat, maybe even lower. Now, that same 40% 1 in 250 is clearly irresponsible as it takes three full years to get back to neutral post event (if there is no post event pricing response). Fortunately, nobody appears to be running at 40% PMLs anymore (or at least that’s what they tell themselves)!

Still, the math is clear that if the ROE has been cut in half, the PML tolerance should be cut in half to say 20%. While some companies have lowered their PMLs in recent years because of lower returns, I have never heard it expressed as a risk decision tied to payback period.

Rather, they say “pricing on new business has come down so we decided to write less” thus lowering their deployed PML (but not necessarily their tolerance). We have also seen PMLs come down as a result of “cheap retro” from non-traditional players which may not be sustainable (as evidenced by recent events) and is more of an arbitrage decision than a statement about their net tolerance.

My position is companies should lower their tolerance when return prospects are lower, regardless of what they actually choose to deploy based on underwriting decisions. This leads to

Conclusion #2

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

That is today’s topic. Agree? Disagree? Feel free to leave comments below and if you want to read more musings like this in the future, click the subscribe button on the right!

11 thoughts on “The Case for Dynamic PML Limits”

  1. Ian,
    Great blog! It looks great and I really l look forward to seeing future posts. Regarding this post: as usual, you raise valid points that I agree with. Only comment: companies’ ability to adjust their exposure real time needs to be balanced with their desire to provide consistent capacity to their clients. While their exposure can be partly managed via ceded reinsurance and ILS purchases, adding and withdrawing capacity might be seen unfavorably by brokers and clients.

    Thanks for the insights and I will tell others about it
    Jay

    1. Thanks Jay! Very good point. As you say, this can be managed on the back end to avoid disruption but that is a valid concern. I should also probably clarify real time doesn’t necessarily mean daily. Perhaps it’s quarterly with enough lead time built in before the next renewal to plan accordingly so to minimize disruption? Maybe it’s also a good way to create discipline to force you to get off or reduce line size to your worst clients to get back within your lower tolerance as returns come down.

      I guess last thought is it’s also pretty disruptive when you do things the “typical” way and end up undercapitalized post a large event and have to take drastic actions on your renewal book to protect your rating!

      Finally, you win a free T shirt for being the first comment! (disclaimer: there may never be an iansbnr T shirt, but if there ever is, you definitely get one)

  2. Huzzah, huzzah! Looks like someone is on his way to putting me out of business.

  3. Ian – what exactly does “semi-retired” mean as it looks like you continue to be fully engaged? Greg

  4. Ian,
    I don’t mean to intrude on the awkward and uncomfortable mutual admiration that you and Josh share, but I need to ask: what the hell are you talking about?

    Risk appetites are part of company’s overall risk management framework. PMLs are only one of the metrics companies use to determine the appropriate level of catastrophe risk. The fact that they utilize less when projected returns are lower, is itself evidence of dynamic risk management in practice.

    Assembling an optimized portfolio is an inherently dynamic process, and using all tools available is essential, including “cheap retro”.

    As far as the concept of payback period, that suggests that managements should believe that years with consecutive losses don’t occur in assembling their portfolio. I would far prefer to ascertain whether I am being paid fairly or above the odds for the risk I take on, subject to all of the risk management constraints underpinning how much I am willing to put at risk in light of my capital.

    Finally, anyone who builds a portfolio on the basis that they will be able to raise capital post-event should be sentenced to six months of watching Keeping Up With the Kardashians. (By the way, I know the real story on Ye and the Miami apartment, but I’m under an NDA).

    1. Oh, our first troll! Yes! Every board needs a good troll to drive comment activity. Thank you! You will definitely get a free (non-existent) T shirt!

      And to address any confusion, no I don’t believe payback is the right way to look at things, but I know too many companies did indeed adopt that approach.

      1. In what kind of tops-turvy world does a man get attacked for creating the maximum amount of shareholder value possible? I have never heard of this Weston Hicks character, if in fact that’s his real name, but I believe this blog has become some sort of Progressive Green New Deal worker’s paradise.

        1. Maximizing shareholder value? You don’t think you could have got someone in Japan to pay $75 if you shopped around more? But no, you were in such a rush to binge watch every season of the Kardashians that you settled for $68. Unbelievable!

          PS: There are other posts available to critique. Feel free to tell me how bad the Gallagher for Willis idea is!

  5. Wow. This looks like a fun place to hang out. Great guests, interesting conversation and a fantastic host to keep it all moving. All we need is an open bar! Cheers and best wishes from Miami Beach.

  6. Ian,

    You have identified a key issue that all companies should think about and manage. There is a world of difference between assuming cat risk when embedded profitability is high and when it is low. We agree with your conclusions 100%!

    Glad to see you doing this blog!

Comments are closed.