[HN Gopher] The Business of Betting on Catastrophe
___________________________________________________________________
The Business of Betting on Catastrophe
Author : anarbadalov
Score : 57 points
Date : 2025-06-23 13:04 UTC (3 days ago)
(HTM) web link (thereader.mitpress.mit.edu)
(TXT) w3m dump (thereader.mitpress.mit.edu)
| MarkusQ wrote:
| This reminds me of the predator hierarchy (for example, see
| Colinvaux's "Why Big Fierce Animals are Rare"): the reinsurers
| spread the risk from various insurers and for various
| catastrophes around among a pool of meta-insurers. But this pool
| is necessarily smaller than that of primary insurers, and their
| risks more likely to be correlated (catastrophes can cause other
| catastrophes, and multiply primary insurers can be affected by
| the same catastrophe).
|
| For that matter, I'm also reminded of credit default swaps, and
| Lehrer's "We Will All Go Together When We Go."
| falseprofit wrote:
| One aspect worth pointing out is that ILS are transferring
| insurance risk outside of the insurance industry. Appetite has
| gone up and down but e.g. hedge funds would normally not be
| available to assume insurance risk otherwise.
| kqr wrote:
| Reinsurance does not only spread risk by pooling multiple
| insurers, but also smears out the impact of catastrophes
| geographically and temporally: big events in one year, in one
| part of the world result in more expensive reinsurance all over
| the globe for a few years forward, as reinsurers collectively
| stock up on capital again.
|
| So while locally catastrophes can cause other catastrophes, for
| the most part earthquakes in Thailand does not trigger
| wildfires in Texas. Nor does a hurricane in Florida one year
| cause more hurricanes in Florida the next year.
| simicd wrote:
| It's correct that the number of reinsurers is smaller than that
| of primary insurers. But the risk born by reinsurers is less
| correlated, not more. Any given primary insurer has risk
| clusters (domestic market, line of business, etc.). If a large
| catastrophe happens in their domestic market they might go bust
| but what are the chances that it happens simultaneously to all
| markets globally?
|
| Say you're a primary home insurer in the US. If a hurricane
| hits you might not have enough capital to rebuild all the
| homes. A reinsurer which is also covering Europe, Asia, LatAm,
| etc. is less likely to go bankrupt. The reinsurer can cross-
| subsidize and use the insurance premiums from other regions to
| pay out the claims from the US market. All that matters is that
| on average the loss probabilities and severities are estimated
| correctly.
|
| And this is just using one line of business as example,
| reinsurers are covering property, casualty, life and health
| which add extra layers of diversification.
| Onavo wrote:
| Not sure what the controversy here is. Catastrophe risk is the
| bread and butter of property insurance.
| falseprofit wrote:
| Felt like the article ended before a thesis statement.
| gwern wrote:
| > She is the author of "Investable! When Pandemic Risk Meets
| Speculative Finance - A Cautionary Tale," from which this
| article is adapted.
|
| So I think structurally, the conclusion here is that 'cat
| bonds are an example of how insurers can work with abstract
| risks, and so any risk (such as global pandemic) could be
| worked with this way', and the rest of the book then examines
| how people are trying to actually do so with pandemic risk.
| PaulHoule wrote:
| But as traditional insurance, not cat bonds.
| munificent wrote:
| _> New "efficiency features" regularly get introduced in ILS and
| written into contracts. One of the most transformative has been
| the use of parametrics. Unlike traditional insurance, which
| calculates payouts based on actual losses (what's called
| indemnity), parametric insurance uses preset triggers to
| determine whether money gets released. During an interview, a
| London-based parametric expert gave me this example of a
| parametric scenario: If, during a hurricane, wind speeds off the
| Florida coast hit a predetermined trigger speed -- say 175 mph --
| at a trigger distance of two miles offshore within a preset
| longitude and latitude grid, the payout is, in theory, immediate.
| No actual damage need occur; the trigger measures just need to be
| met._
|
| Wow, that is absolutely _begging_ for exploitation.
|
| Whoever controls the authority reporting these figures now
| controls whether these bonds pay out. That in turn means that
| whoever holds those bonds has a huge financial incentive to
| manipulate what that authority says.
|
| Put another way, if you're holding a bond that will cost you $100
| million if a hurricane windspeed hits 175 MPH, then you have $99
| million bucks that are worth spending trying to get the NOAA to
| say anything but that.
| gruez wrote:
| The same incentive exists for economic figures (inflation
| linked bonds) and market prices (cash settled derivatives), and
| it's seemingly not an issue, and those are far easier to game
| than physical measurements like wind speed or whatever.
| jallmann wrote:
| Manipulation of reported economic numbers has been an issue
| in the past, see LIBOR.
| krisoft wrote:
| > and those are far easier to game than physical measurements
| like wind speed or whatever.
|
| I'm not so sure about that. I bet that we could tamper with
| an anemometer somewhere out in a field. Easiest is to put
| brushless motor with a propeller next to it and blow propwash
| on it. More technically difficult is to tamper with the
| signal between the sensor and the station, or MitM the
| station.
|
| If you are careful and only modifying the measurments when
| the weather is already crummy they might not even suspect.
| singleshot_ wrote:
| Replace the anemometer cups with slightly larger ones.
| 56544562 wrote:
| > and it's seemingly not an issue
|
| Sounds like you do see an issue after all. Is it obfuscated?
| Hidden in complexity? Is it artificial to appear as one thing
| while indirectly, as a side effect, leading to another?
| jowea wrote:
| There have some governments outside of the developed world
| accused of manipulating inflation numbers.
| kqr wrote:
| Sure. Getting reliable data all parties can agree on is one of
| the biggest challenges to parametric insurance. The data
| sources I hear about most often are US governmental agencies -
| and this is a problem, since the US political system is not
| stable enough to finance its governmental agencies reliably.
| (Most recently I recall concerns around budget and staffing
| cuts for NOAA and USGS.)
|
| That said, sane practice for parametric insurance is to have
| redundancy in data sources, and an agreed procedure for
| settling differences in conclusions resulting from relying on
| either of them alone.
| soulofmischief wrote:
| Goodhart's law continues to ring true.
| dylan604 wrote:
| Taking your point of view now makes sense on why to defund
| NOAA, fire everyone that's not going to toe the line, and then
| have those that will parrot the necessary info to keep from
| paying out. Make Weather Great Again, just doesn't lend itself
| to a hat though
| the_pwner224 wrote:
| People have already done this with NWS weather equipment for
| federal farm drought insurance:
| https://coloradosun.com/2024/09/08/patrich-esch-ed-dean-jage...
| bgnn wrote:
| In Turkey the mandatory earthquake insurance for homeowners is
| owned by the government. It triggers parametrically (> 7
| magnitude). In one case at least [1] the government office
| responsible for announcing the magnitude, AFAD, declared it
| lower than this threshold although other countries and Turkish
| research institues measured it as 7.0 . At the end the
| insurance payout was so much more limited even for people who
| kist their house and loved ones.
|
| As wiki page mentions in notes section AFAD declared this a 6.6
| magnitude earthquake although it was 7.0 . [1]
| https://en.m.wikipedia.org/wiki/2020_Aegean_Sea_earthquake
| paxys wrote:
| As if pandemics weren't already political enough. Let's get large
| corporations, investment funds and billionaires involved and give
| them direct stake in declaring what is or isn't a pandemic, how
| many deaths have happened in a certain area, what was the cause
| of death etc. That should end well.
| leemelone wrote:
| This is the dumbest idea I've read about in a long time.
| danielfoster wrote:
| I like to think I'm somewhat intelligent, but there's something I
| don't understand here. The article cites an example of pandemic
| bond holders receiving a return of 40% over 3 years and these
| bonds being a useful way for the issuer to secure needed funds in
| the event of a pandemic. Unless a pandemic happens every ~8
| years, isn't this a ridiculous and unsustainable risk premium to
| pay?
| HillRat wrote:
| The class B bonds paid roughly 11% over LIBOR, so about 40%
| over three years, against the risk of a viral outbreak for five
| different families, defined as At least two countries
| experiencing at least 250 fatal cases increasing over twelve
| weeks, so the trigger did not have to be as globally-
| significant as COVID-19 turned out to be. That's a pretty
| aggressive coupon, but the chance of a regional outbreak was
| also pretty high.
| danielfoster wrote:
| Makes sense now, thank you! I feel like the author should
| have mentioned this.
| dmurray wrote:
| Based on that description it would have been triggered by
| COVID-19, swine flu in 2009, and I think just missed out
| (depending on the fine print) on SARS in 2002. That's two or
| three in 18 years, so losing your money once every eight
| years is not far off the recent performance of this kind of
| bond.
| 21secondstogo wrote:
| CTO at an ILS fund. Cat bonds are essentially securitised
| versions of fully collateralised reinsurance contracts where the
| premium is the coupon plus the return on collateral. A benefit
| being that you can trade them. They're not usually used for
| speculation as stated in the article - investors are typically
| pension funds looking for investments that are uncorrelated to
| traditional financial market risk. e.g. on a US hurricane exposed
| cat bond you may only lose money if a huge hurricane blows
| through Florida, no matter what credit and equities are doing.
| It's true that a lot of the deal sourcing is relationship-driven,
| but there is a good amount of data-driven tech involved in
| overlaying the insured's past claims and underwriting data on top
| of simulated catastrophe model output, applying your own view of
| climate, vendor model adjustments, hurricane activity etc.
___________________________________________________________________
(page generated 2025-06-26 23:00 UTC)