Sealing the envelope: catastrophe modeling companies now hire armies of actuaries, statisticians and software engineers to help clients predict the scale of potential losses.[ILLUSTRATION OMITTED]
There was a time when catastrophe losses were derived from simple back-of-the envelope calculations.
There was a time when insurance companies relied solely on 19th century Sanborn maps Sanborn Maps were originally created for assessing fire insurance liability in urbanized areas in the United States. The maps include detailed information regarding town and building information in approximately 12,000 U.S. towns and cities from 1867 to 1970. to assess values.
There was a time when there was no organization called the International Society of Catastrophe Managers.
There was a time when hurricanes never inflicted more than $1 billion worth of damage.
There was a time when CAT models were ruled by deterministic and not probabilistic (probability) probabilistic - Relating to, or governed by, probability. The behaviour of a probabilistic system cannot be predicted exactly but the probability of certain behaviours is known. Such systems may be simulated using pseudorandom numbers. criteria.
That was a time when Richard L. Clinton, then an underwriter, was at the beginning of what's turning out to be one of the longest careers in the catastrophe modeling
Today catastrophe modeling firms, with EQECAT, AIR Worldwide and RMS (1) (Record Management Services) A file management system used in VAXs.
(2) (Root Mean Square) A method used to measure electrical output in volts and watts.
1. RMS - Record Management Services.
2. among the most well-known, are packed to the gills with engineers, mathematicians, statisticians Statisticians or people who made notable contributions to the theories of statistics, or related aspects of probability, or machine learning: A to E
"The big area I've seen is that the industry has gone from a handful of people to a major occupation and a career path," says Clinton, president of EQECAT.
Clinton, who started as an underwriter with Chubb nearly 40 years ago, isn't kidding when he says hurricane metrics used to be calculated on the back of envelopes. That was when underwriters simply added a few percentage points to the catastrophe premium paid by risk managers to account for the average annual increase in the probable maximum loss Probable Maximum Loss (PML)
The anticipated value of the largest loss that could result from the destruction and the loss of use of property, given the normal functioning of protective features (firewalls, sprinklers, and a responsive fire department, among others, in the sustained from one hurricane season Hurricane season refers to a period in a year when hurricanes usually form. For more information see: Tropical cyclone#Times of formation.
For a lists of past seasons, see:
That simple arithmetic changed after three watershed events, says Clinton: Hurricane Hugo Hurricane Hugo was a destructive Category 5 hurricane that struck Guadeloupe, Montserrat, Puerto Rico, St. Croix, South Carolina and North Carolina in September of the 1989 Atlantic hurricane season, killing 82 people. It also left 56,000 homeless. in 1989, which caused more than $4.1 billion in insured losses, followed by Hurricane Andrew This article is about the 1992 hurricane; there was also a Tropical Storm Andrew during the 1986 Atlantic hurricane season.
Hurricane Andrew is the second-most-destructive hurricane in U.S. history, and the last of three Category 5 hurricanes that made U.S. in 1992, which caused $15.5 billion in insured losses, followed by the Northridge earthquake The Northridge earthquake occurred on January 17, 1994 at 4:31 AM Pacific Standard Time in the city of Los Angeles, California. The earthquake had a "strong" moment magnitude of 6. in 1994, which caused $12.5 billion in insured losses. Back then, those kinds of losses were considered astronomical and the trio of disasters ratcheted up the maximum losses that could be sustained by any one insurance carrier.
The losses were so high, in fact, that several insurance underwriters went belly up. The scale of those hurricane and earthquake losses brought into perspective the importance of accurately predicting what could be inflicted by any one event, says Clinton.
If risk managers had an accurate guide--indeed even a semi-accurate guide--to how extensive losses could be, it could go a long way to helping them allocate their employer's capital more efficiently.
Still, for all the computing, academic and intellectual firepower packed into modeling firms today, the catastrophe models are just that: Beautiful perhaps, but somehow never quite good enough.
A pair of hurricanes, Ike and Gustav, which struck last summer, once more reminded the modeling firms, insurance carriers and corporate insurance buyers of the models' fallibility fal·li·ble
1. Capable of making an error: Humans are only fallible.
2. Tending or likely to be erroneous: fallible hypotheses. .
EQECAT's models, like most catastrophe models, were accurate in modeling losses from wind but came up short when it came to accounting for wave height. "What we found was that we had a good correlation between what we were modeling and the loss but it didn't explain the losses and issue with wave height," says Clinton. "So we went back and found that we missed some of the big waves."
Combining wave height and wave duration data with other weather factors helps modelers parse through hurricanes with much finer resolution. That affects how purchasers buy hurricane catastrophe insurance, says Clinton, who's been with EQECAT since 1994.
Catastrophe models have come in for some criticism over the past few seasons, but it's also the case that risk managers don't always use the models to their rill potential, says Clinton. If they did, risk managers would find themselves making better decisions with regard to the efficient use of their available capital.
Faster computers and more advanced software have made the models easier to use over the past 15 years, says Clinton. At the same time, the models are more sophisticated because they contain more data about more items. As a result, models require risk managers to be more sophisticated in their grasp of losses associated with catastrophes.
Clinton says that the best way the modeling companies can help risk managers is to give them more perspective with regard to the risk and its relative loss potential to an organization. Risk managers can also benefit from catastrophe models that tell them how the nature of the risk is changing. "That's our responsibility--to give them perspective of the loss," says Clinton.
CYRIL TUOHY is managing editor of Risk & Insurance[R]. He can be reached at email@example.com.