Well, it’s turned out to be a light hurricane season (though, yes, it’s not over) so I thought I would, belatedly, join all the others doing their 20 year Hurricane Katrina retrospective.
The main difference is I will tell it primarily from my perspective, which is less about the human tragedy and more about how it changed the insurance world, as well as a little bit of autobiographic detail.
The Soup Bowl and the Levees
I first became aware of the hurricane threat to New Orleans around 2003. I recall reading a Dowling note about research discussing the “soup bowl” nature of New Orleans, meaning the city itself is actually below sea level and thus heavy rains – or worse, “soup” from the surrounding gulf (or lake or river) – could fill the bowl and leave the city underwater.
As an aside, the New Orleans Soup Bowl differs from the Mexico City Jello Bowl which makes the latter particularly susceptible to strong earthquakes.
Anyway, I started asking some questions and was told not to worry because New Orleans is surrounded by levees to prevent the water from getting in. Most people would have taken that as a relief and stopped asking questions.
However, I later came across a report from the Army Corp of Engineers describing how the levees were badly in need of repair and may not survive a severe hurricane. That set off the alarm bells!
Getting The Answers Before The Test
At that point, I was on the hunt for which reinsurers were overweight New Orleans. I was on a trip to Bermuda sometime in 2004 where we met with most of the reinsurers on the island.
In most every meeting, I asked about hot spots outside of Florida and California, with a particular emphasis on New Orleans and New York City (I was worried about a Sandy like storm as well).
I liked to ask questions like this when they weren’t particularly topical. I am sure there was some eye rolling about why I was asking about “off topic” issues like New Orleans when people were focused on Florida or cat prices or whatever the topic du jour was.
The advantage of asking things when they’re not the story of the day is you get more honest answers! People aren’t as guarded about something that doesn’t seem like a big deal at the moment.
When Katrina hit New Orleans, I remembered I had old notes with who told me they had bet heavily on New Orleans. More on that below, but I had the answers everyone in the market wished they had!
A Disrupted Vacation
One thing many people forget is Katrina was supposed to be a nothing burger. It was a Category 1 headed for Florida that was then going to fizzle out in the Gulf before breaking up over the Panhandle.
I remember this well because my wife and I were headed out for a long weekend to Vermont on the day it hit Florida.
We were packing up and about to head out the door when my wife say “hey did you hear about the hurricane?” and I was confused because it was a non event but I went to see what she was talking about.
The news was on and talking about Katrina decided not to turn for the Panhandle and was now heading to New Orleans! My first reaction was probably something I won’t say in print.
My second reaction was to go check the meteorological sites and after I saw that, yeah, this is now a bull’s eye for New Orleans was to tell my wife we may not be going away after all.
I can’t remember the exact specifics of what happened next. I know we still went to Vermont, but it was probably a couple hours later after I crammed as much research as I could about the path and damage potential before leaving. Oh, and making sure I had that notebook with all the info on who said they wrote a lot in New Orleans.
Montpelier
No, we didn’t stop in Montpelier, Vermont (though we drove nearby). I am referring to Montpelier Re.
Montpelier was the company who told me how much they liked New Orleans. It was “well priced” and a “diversifying zone” meaning that, because it wasn’t a peak risk like Florida, you had to hold very little capital against the risk booked there.
That is a good thing when there are no losses. As you’ll see, it’s a pretty terrible thing when there is a big loss.
Now, I was already short Montpelier’s stock (I think it’s OK to reveal that now?) because they had made another bad bet earlier that year. They paid a large special dividend which left them without excess capital heading into hurricane season.
I thought that was a bad idea and was short the stock because any hurricane would have been bad news for them relative to the other reinsurers who could more easily absorb it. I was not directly betting that the way I would get paid on Montpelier was specifically through a New Orleans hurricane.
But that’s what it looked like was going to happen. Now, remember this was all before the Iphone. It was much harder to get real time information, especially in a small town without great cellular service.
Fortunately, I had my Blackberry (which I still miss)! That at least gave me stock quotes and some ability to get weather updates. And a phone line to call my trader and tell him to short more Montpelier.
It had become a working vacation (the first of many).
From Cat 1 to Cat 5
I’m not going to pretend I remember every detail of what happened next but I do remember becoming more and more certain this was the big one for New Orleans and not a false alarm. A fair amount of time that weekend was spent plotting what my trades were for Monday.
I also remember that the initial cat model projections after landfall were hopelessly optimistic. If memory serves, they were in the $10-15B range after landfall. Absurd!
The models got so much wrong that could have easily been anticipated – the levee risk, the record surge risk, the erosion of the swamplands. None of this was in the model.
The one thing that was harder to predict was the role of the Loop Current. What turned Katrina from a 3 to a 5 over the Gulf was the “fortuitous bounce” through the hot water in the Loop Current.
This is extremely hard to model in real time, especially 20 years ago. The meteorological models all missed this initially.
I didn’t know either, at the time, that the Loop Current was why it bombed out to a 5, but I knew what it meant. It meant the levees were going to fail.
And once I realized that and saw the market on Monday was betting on “only” $10-15B of damage, I started selling and selling and selling some more.
I don’t want to overplay it. This was my first year in the hedge fund world and I wasn’t going to bet the farm like I would have had I been doing it for ten years, but it was certainly my big bet for the year.
The market didn’t believe the worst was possible and I did. My mistake, honestly, was not going out on a limb and betting the farm.
The Aftermath
Once the extent of the damage became clear, while the rest of the country was fixated on people being rescued from roofs or stuck homeless in the Superdome, my focus moved to figuring out which other companies did poorly or excelled.
One of the first opportunities to do some reconnaissance was in early September at the annual KBW conference.
To their credit, most of the insurance CEOs made their way to New York rather than hide in Bermuda. Some were even able to reassure investors that they fared relatively well.
Others were either unprepared or unconvincing and their stocks suffered for it as investors began to realize this was a bigger loss than they had previously realized.
Floating Casinos And Other Bad Bets
While it is easy to pick on the cat modelers for not understanding the risk of New Orleans, it’s not like the insurers understood it any better. There were some awful, awful underwriting decisions made that led to losses far worse than any well tuned cat model could have anticipated.
There was your typical poor data quality of buildings not being coded correctly which led to risks being bound that never should have been, as well as purposely downplaying significant risks in the quest to chase premium, not to mention to all the predictable wind vs. water debates on the personal lines.
The fac market, in general, lacked the imagination to consider what would happen to their properties if the levees didn’t hold or if the surge were greater than expected and got hammered.
Then, you had the more understandable, but still unaddressed, issue of national covers for say Walmart or Home Depot that neglected to consider that their New Orleans locations were riskier than those in other parts of the country.
One of the more famous mistakes in this regard was the floating casinos, aka the riverboat casinos which were docked just off land. They were written as if they were typical Vegas style casinos, ignoring the obvious risk related to wind and surge from being parked in the water.
This was compounded by reinsurers who wrote national cat covers without understanding their exposure to poorly modeled non-peak risks like New Orleans. A lot of the “surprise” losses came less from individual risks that performed poorly and more from national covers that had unexpected New Orleans exposure the reinsurers were unaware they had assumed.
Willful Ignorance
I have always been puzzled by the lack of geologists working for insurance companies. If I was able to do some reading and learn about marshland erosion or weakening levees, why didn’t Chief Underwriting Officers hire people to learn about these risks and identify “no go” risks to avoid?
So many events from this century (Sandy, Katrina, Sendai, etc.) were risks previously identified by geologists and/or climatologists yet the industry was seemingly ignorant of them.
You can imagine I have raised this issue many times and CEOs always dismiss it. There are a few like Ren Re and, to some extent, Swiss Re, who research these risks, but for the most part the industry puts their head in the sand and decides to raise price after rather than avoid a loss before.
Capital Management
Probably the biggest insurance lesson from Katrina was seeing how companies managed risk relative to their peers.
Montpelier, as noted, thought there was “free money” in writing non-peak zones. From a mathematical analysis of the rating agency model, they were correct. Diversifying zones could be written for much less capital than peak zones.
Thus, even if the rate on line was lower in a place like New Orleans, it could be a higher modeled ROE. But this analysis ignores one big problem…what if you have a peak level loss in a non-peak zone???
Then, the math works against you in a very painful way. If you lose 25% of your capital in a diversifying zone, then you only have 75% of your capital left to support the peak zones. This means if you thought you had 30% of you capital exposed to your peak, you now have 40% at risk (numerator is the same, the denominator is smaller).
Montpelier (and others, they certainly weren’t alone) learned the hard way that optimizing the capital model increased their real world risk. They had to raise fresh capital to defend their rating. When this happened, I did very well on my short position.
The Hard Market
As most of you know, the flip side of all these losses was cat prices went through the roof. Many reinsurers tried to spin their need for capital as an “offensive raise” to take advantage of the newly hard market.
Rather than admit the rating agencies had a gun to their head to raise capital to preserve their rating, they pretended they were poised to grow in 2006 and shareholders would be rewarded for providing them more capital.
In some cases, this was true, but most of the time the pitch for capital was offensive in another way, as in I was offended by how blatantly they lied to me!
The most hotly debated (though far from most egregious) of these was Ace’s “offensive raise”. Ace didn’t have nearly the losses others did and wasn’t in any financial trouble but, if memory serves, their initial loss estimate developed poorly over time which, when combined with their continuing asbestos losses, consumed most of the capital raised.
Some insist that made it a defensive raise, while others gave them the benefit of the doubt since they weren’t in need of capital, they just wanted some extra cushion.
Most other raises were defensive, though some were able to raise enough to put some into defense and keep a little for offense to grow. Only two companies really stood out as not needing capital at all and were able to go fully on offense – Ren Re and Arch.
Both companies saw significant multiple expansion after Katrina as they were rewarded for not needing capital. They also were able to grow earnings faster than the rest of Bermuda as they were better poised to capitalize on the hard market.
Thus, the most important lesson I took away from Katrina (which was really a reinforcement of what I learned after 9/11) is that the key to being successful writing cat risk isn’t how much money you make in loss free years. It’s how much less do you lose in the heavy loss years.
Has The Industry Learned Anything?
I’ll finish up with a question. Do we think the industry is any better at underwriting large cat events now than it was 20 years ago?
Certainly, the models have improved and data quality is better, so those are positives. But I don’t think the psychology has changed much.
There are still reinsurers who view non-peak zones as free. There are still plenty of companies who would have been decimated last year if Milton hit Tampa, as it almost did.
There are still way too many companies who don’t “trust the science” and incorporate it into their underwriting. I’ve tried to warn you all about Seattle multiple times!
And I didn’t even get to “new” cats like cyber or electromagnetic pulse or drone attack.
The paradox is that while companies tend to learn from their past mistakes, they are terrible at learning how to spot the behavioral patterns that make those mistakes happen again.
So, sure, another Katrina would be handled much better than last time, but the next “nobody could have anticipated this” event – that actually some people did anticipate – will likely be just as bad as Katrina, or even worse.