Whatever Happened to the 100-Year
Event?
They're big, we know them when we see them, and we remember them. By STEVEN SULLIVAN

B

y September 16, 2013, more than 4,500 square miles of Colorado were under water. That’s an area roughly the size of Delaware by some estimates; others compared it to Connecticut.
The territory on the eastern slope of the Rocky
Mountains had already been subjected to six straight days and 17 inches of rain, more than five inches above the annual average for the area. More than
1,200 people were missing, 19,000 homes were damaged or destroyed, 12,000 people evacuated. Fatalities were still unknown, but expected to be numerous. Television images showed raging torrents in streams and creeks that rivaled the last few hundred yards of the Niagara
River before the Falls.
Colorado was experiencing not just a 100-year event; this was a 1,000-year event, a flood that even the National Weather
Service characterized as not only historic, but biblical.
That turns out to be a pretty good characterization of a
100-year event. They’re big, we know them when we see them, and we remember them: Hurricane Katrina in 2005; Super
Storm Sandy in 2012; the tornado that devastated Moore,
Oklahoma in 2013; the entire wildfire summer of 2012. And if one of them takes your loved ones, your home, or your possessions, it doesn’t really matter if it makes the record books.

26

ACTUARIAL REVIEW

NOVEMBER/DECEMBER 2013 • WWW.CASACT.ORG

Misleading Term

percent annual probability of occurring at a given location. For

The term “100-year event” (or whatever number you want to

example, if Miami, Florida, is impacted by a 100-year hurri-

attach to it) expresses a probability rather than a certainty.

cane event in 2013, this doesn’t mean the next 100-year hur-

Events this extreme are commonly measured by how likely

ricane in Miami will occur in 2113; the probability for another

they are to happen. In the case of Boulder, Colorado, a flood of

hurricane in Miami in 2014 of the same intensity remains one

this magnitude is expected to occur only once in a thousand

percent. Hundred-year events in consecutive years are rare,

years. Lesser catastrophes may happen more frequently, say

but certainly not impossible.”

once every 100 years. But these are only guesses. Projections.

“We need to make sure we’re communicating that we

There’s nothing to say that another 1,000-year event won’t

don’t mean this is going to happen only once every 100 years,”

happen next year. Or next week. Or not for another 2,000 years.

adds Mary Frances Miller, an actuary with Select Actuarial

The term became popular in 1973 when the National

Services in Nashville. “It’s possible to get two in a row. Highly

Flood Insurance Program (NFIP) needed a standard to mea-

unlikely, but possible. And we need to communicate that

sure flooding across the country. Some areas are more prone

it’s based on a model. It’s not like we’re absolutely confident

to flooding than others, some flood more frequently than oth-

that the probability is one in 100. It might really be one in 60.

ers, and some areas of the country haven’t kept records long

Or one in 200. We can’t confidently say it’s not one in 10. It’s

enough to ensure statistical accuracy for prediction. Neverthe-

unlikely, and it’s out on the I-don’t-really-know- how-unlikely end of the scale.”

less, the NFIP was mandated to map all

In other words, there’s nothing

the flood plains in the country. Bringing

certain about predicting any extreme

all these factors together, along with a number of different ways for measuring flood magnitude, it compromised on the 100-year frequency as a standard.

“We need to make sure we’re communicating

both the general public and insurance professionals,” says Mark Bove, senior research meteorologist with Munich
Re America. “A ‘100-year event’ refers to a natural catastrophe that has one

WWW.CASACT.ORG • NOVEMBER/DECEMBER 2013

are looking for. Certainty makes them feel confident, and that confidence can get them into trouble.

“The term ‘100-year event’ is misleading and is often misinterpreted by

event. Yet certainty is what many people

that we don’t mean this is going to happen only

Probably Questionable Behavior
Dr. Howard Kunreuther and Erwann

once every 100 years.”

Michel-Kerjan, both of the Wharton

—Mary Frances Miller

nia, wrote an op-ed piece in The New

School at the University of Pennsylva-

ACTUARIAL REVIEW

27

York Times (“Paying for Future Catastrophes”) after Super Storm Sandy. “Our research shows,” they said, “that half of all policyholders cancel their flood coverage after only three or four years.
Why? Because they paid premiums without getting anything in return and are likely to think ‘Bad investment!’ But insurance is a safety net, not a bet.”
Miller tells the story of a man in Miami who was rebuilding in the wake of
Hurricane Andrew in 1992, at that time the worst storm in Florida history. He

Wilderness is prime real estate for those who can afford it, and those who can afford it build expensive houses, communities, and businesses in what was once pristine prairie or forest.…So even if wildfires aren’t becoming more frequent, they’re certainly becoming more costly and destructive from a human standpoint.

was installing a set of glass doors that clearly weren’t up to code and would never withstand another hurricane. When asked why he was doing this (aside from saving money, of course), he replied that since Hurricane Andrew was a one-in-25-year storm, and the doors were designed to last for maybe 20 years, there really wasn’t anything to worry about.
“We chuckle,” says Miller, “but that just indicates a profound misunderstanding of probability.”
The other extreme is becoming too cautious. The classic example of this is flying vs. driving. Plane crashes are extremely rare, but they’re usually spectacular when they happen.
Traffic accidents happen every day—many times every day.
Yet people tend to be more afraid of dying in a plane crash than a car accident, even though their chances of dying in a car are much higher.
Yet somehow we (or at least the media and politicians) seem to admire the people who refuse to retreat, who are determined to rebuild in the same flood plain after they’ve been wiped out. The ones who decide not to make the same mistake twice are viewed as quitters. So what is the proper response to a 100-year event?
There’s no easy answer to that question. It may be one thing if you’re a homeowner living in New Orleans or Miami or Boulder. It’s another thing if you’re an insurance company covering properties in those areas.
Where the People Are
There’s a difference between a 100-year meteorological event and a 100-year insurance loss event, says Mark Bove. “The former deals with the frequency and severity of the hazard at a given location, while the latter is largely influenced by the built

28

ACTUARIAL REVIEW

human environment impacted by an event. It is quite possible to have a 100-year (or longer return period) event occur in an unpopulated area, causing little to no insured loss, while a weather event of moderate severity, well below a 100-year return period, could cause a 100-year insurance loss if it impacts a densely populated area with high insurance penetrations.”
“It all boils down to people,” says Tom Jeffery, senior principal scientist at CoreLogic in Madison, Wisconsin. “People who live in coastal areas face a known risk, some more than others. If they continue to live there, there is going to be a recurring cost of damage.”
It’s too early at this writing to say how bad the 2013 wildfire season was. Though memory of the more than 6 million acres that burned in 2012 may have faded, the more than 3 million scorched acres so far in 2013 seem bad enough.
There have probably been even worse fire seasons before recorded history, when lightning strikes ignited hundreds of square miles of unpopulated wilderness and burned until the rain eventually extinguished the fire. Even when Native
Americans began to settle that wilderness, their nomadic lifestyle enabled them to pull up stakes and move whenever fire appeared on the horizon and threatened their settlements.
That was their fire insurance.
It’s not so easy for modern settlers. Wilderness is prime real estate for those who can afford it, and those who can afford it build expensive houses, communities, and businesses in what was once pristine prairie or forest. None of it is movable. And in addition to those random lightning strikes, all those people can accidentally or deliberately start fires with a careless match or a spark from a piece of machinery. So even if

NOVEMBER/DECEMBER 2013 • WWW.CASACT.ORG

wildfires aren’t becoming more frequent, they’re certainly becoming more costly and destructive from a human standpoint.
This extends to other types of natural disasters, too.
The numbers are daunting: Estimates for the cost of
Hurricane Katrina range from $108 billion to $150 billion with more than 1,800 dead; Super Storm Sandy cost $65 billion and killed 285; Hurricane Andrew destroyed $26.5 billion in property and caused 65 fatalities.
“The incidence of extreme events is far more frequent,” say Kunreuther and Michel-Kerjan in their New York Times article. “Twenty of the 30 most expensive insured catastrophes worldwide from 1970 to 2011 have occurred since 2001—and
13 of them were in the United States. Aside from the 9/11 terrorist attacks, all were natural disasters. The increase is most likely because of the location in high-risk areas of more people and more valuable properties, along with a changing climate.”
An Understood Currency
Correctly assessing that risk is the work of catastrophe modelers. Using high-speed computers, modelers have learned how to create complex programs that combine data of past occurrences and emerging science to come up with a reasonable probability for how many extreme events might occur in a given time frame, how severe they might be, and the cost of the damage they might inflict.
According to Kay Cleary, an actuary and director at the modeling company Risk Management Solutions (RMS), catastrophe models have come a long way since they were first used in the 1980s to provide point estimates—like the 100-year probable maximum loss (PML). But no matter how sophisticated they are, they still can’t provide a straightforward right or wrong answer.
“The 100-year PML just gives you a probability and a number,” says Cleary. “It would be nice if you could just draw a couple of lines and say ‘That’s my range of uncertainty.’ But it doesn’t work quite that way because of the complexity of what goes into it. With the advances in computer speed and science, you can now drill down and determine the amount of trouble you’ll be in, assuming you are in trouble. You can do more sensitivity testing by varying some of the assumptions and seeing how it shifts things around. You can learn a lot more, a lot easier, and a lot faster by having more numbers to look at than just a point estimate of PML.”
What models do provide, Cleary says, is an agreed-upon, understood currency that enables insurers to quantify the risks

WWW.CASACT.ORG • NOVEMBER/DECEMBER 2013

and trade them in the marketplace. “We hope the number is close to reality, but even so, it’s an agreed-upon amount that allows things to happen. The fact that it’s exactly right or exactly wrong is obviously important, but what it mainly does is enable us to do business. It’s kind of like a stock price. What is a stock price, really, but an agreed-upon transaction number?
It’s not a real number.”
According to David Lalonde, a senior vice president with the modeling firm AIR Worldwide, “the one percent exceedance probability loss (or the 100-year return period loss) is estimated to be just over $200 billion—a figure that could be driven by an active and severe U.S. hurricane season or by a combination of different perils in different regions, as was the case in 2011. AIR estimates the average annual loss (AAL) from natural catastrophes is $59 billion, in line with global catastrophes losses from 2012, which are estimated to be around $58 billion.” So why are we seeing more of these events that are getting so destructive and so costly? It’s tempting, of course, to blame global warming. But scientists and modelers are cautious.
Mark Bove at Munich Re says we still don’t know enough to be certain, but a warmer, moister atmosphere could be

ACTUARIAL REVIEW

29

conducive to more tornado and hail events across the eastern

next five years is a little bit higher than that. We don’t make an

two-thirds of the United States. Climate change may or may

explicit adjustment for it, but we think there is some implicit

not produce more frequent extreme events but “those that

acknowledgement of those impacts.”

form could become more intense, reducing the return period

“Attributing every weather anomaly to manmade climate change—other than the higher temperatures the global

of severe hurricanes.”
“Climate change makes the model more complicated,”

warming phenomenon is named for—is a high-stakes gamble,

says Mary Frances Miller. “I don’t know if it makes hurricanes

rooted more in politics than in science,” says Nate Silver in his

more or less frequent, there’s argument about that. But it cer-

2012 book, The Signal and the Noise: Why So Many Predic-

tainly makes the modeling more difficult.”

tions Fail—But Some Don’t. “There is little consensus about

Global warming, or climate change, is a long-term phe-

the ways that climate change might manifest itself other than

nomenon, adds Kay Cleary, and models are better suited to

through temperature increases and probably rising sea levels.

much shorter-term projections—two to five years—so models

The greenhouse effect almost certainly exists and will be

typically don’t reflect it. “We don’t know enough to do that re-

exacerbated by manmade CO2 emissions. This is very likely to

sponsibly. However, to the degree that there have been trends

make the planet warmer. The impacts of this are uncertain, but

recently and the impact of global warming is felt, we do look at

are weighted toward unfavorable outcomes.”
Improved models now account for a previously unrecog-

that information in light of the medium-term rate.”
There’s a controversy in hurricane modeling between the

nized peril: storm surge. Storm surge is a hybrid, neither wind

long-term rate and the medium-term rate, Cleary explains.

nor flood. Even relatively weak wind events can push huge

“The long-term

volumes of water to surge over the land, causing billions of

is based on the

dollars of damage.

112 years of data

“It has to do with bathymetry, the study of the underwater

that we have. The

landscape,” Cleary says. “Water pushing against the ocean

medium-term

floor starts to build up and inundates the land, creating a surge

acknowledges that

flood. Initially, it made a lot of sense to have surge factor off

there are cycles.

the wind, but we found that surge has its own characteristics

Right now we’re in

not necessarily related to the wind. We upgraded our model to

a somewhat higher

account for not only the severity of what can happen, but also

than long-term

the types of things to worry about.”

average cycle. If you look at the 112

Paying for the Risk

year average, what

So who pays for all this? That, of course, is what insurance

we expect in the

companies are for, but many private-sector insurers have

“Water pushing against the ocean floor starts to build up and inundates the land, creating a surge flood. Initially, it made a lot of sense to have surge factor off the wind, but we found that surge has its own characteristics not necessarily related to the wind.” —Kay Cleary
30

ACTUARIAL REVIEW

NOVEMBER/DECEMBER 2013 • WWW.CASACT.ORG

increased premiums in high-risk areas to make it virtually impossible for some people who want insurance to afford it.
Some carriers have exited certain markets altogether. And most carriers depend on reinsurers—like Munich Re and
Swiss Re—to share much, if not most, of the risk.
But there are other mechanisms as well. States prone to natural disasters, particularly along the coasts, fund wind pools that offer high-risk coverage at lower rates. Wind pools grew out of the FAIR (Fair Access to Insurance Requirements) plans that were formed in 1960s when many insurers pulled

Politically, says Tom Jeffery, mitigation can be a can of worms because it ultimately impinges on individuals’ rights to live and own property wherever they want to.

out of areas hit hard by rioting and civil unrest. One of the largest wind pools, Florida’s Citizen’s Property Insurance

property rather than the property owner. The cost of the loan,

Corp., covers nearly $500 billion in assets. Most wind pools,

he says, will be less than the benefits from the lower insurance

however, aren’t nearly so well-endowed and must resort to the

premium.

reinsurance market as well.
In 1968, the federal government instituted the NFIP to

Even though risk-based rates make economic sense for those who can afford them, they’ll still end up pricing many

subsidize premiums for people living in flood plains. But

low-income homeowners out of the market. “Our response to

claims from the disastrous 2005 and 2008 hurricane seasons

that is to help them out by using vouchers,” says Kunreuther,

forced the NFIP to borrow $18 billion from the U.S. Treasury,

similar to the Supplemental Nutrition Assistance Program.

which it hadn’t yet repaid when Super Storm Sandy hit in

They would receive the means-tested voucher, as well as a

2012.

loan, only if they agree to mitigate and make their homes

In July 2012, Congress passed the aptly named Biggert-

safer. “Mitigation will bring their premium down so much that

Waters Flood Insurance Reform Act, which attempts to make

the actual magnitude of the voucher, even if they pay for the

the NFIP more financially stable and ensure that flood insur-

loan themselves, would be less than just paying the difference

ance rates more accurately reflect the real risk of flooding.

between what they can afford and the much higher insurance

The act institutes changes in flood hazard mapping and flood

premium.”

plain management, but primarily it gradually eliminates the

Politically, says Tom Jeffery, mitigation can be a can of

subsidies that enabled many insureds to pay lower premiums

worms because it ultimately impinges on individuals’ rights to

than are reflected by their risk.

live and own property wherever they want to. And it’s not just

“The Biggert-Waters Act is moving exactly in the right

about the cost of individual insurance coverage. The levees

direction,” says Howard Kunreuther. “It’s the first piece of

that were rebuilt to protect New Orleans from another Katrina-

Congressional legislation that I’m aware of that has risk-based

level storm cost in excess of $14 billion. Estimates for building

rates, not on all homes but on second homes and on repeti-

a sea wall that would protect New York from another Sandy

tive-flooding homes. With risk-based rates you can give a rate

run as high as $23 billion.

reduction when people mitigate.”

“It’s an awfully big undertaking,” he says. “In an area like
New York, it certainly would offer tremendous protection and

Mitigating Circumstances

reduce or eliminate the tens of millions of dollars of damage

Risk mitigation is something that Kunreuther has been

that could come in the future. And that’s the question every-

championing for years, though he admits it can be a tough

body wants answered: What can we expect in the future? No-

sell because it can be expensive for homeowners. The trick, he

body knows. We can project based on what’s happened in the

says, is to give people a financial incentive to relocate or retro-

past, but nobody knows if it’ll happen next year, in 10 years, or

fit their homes to better withstand wind and water damage. He

in the next 100 years. All we know for sure is that there will be

advocates long-term loans and low-cost, long-term insur-

another one sometime in the future.”

ance—perhaps as long as 20 years—that would be tied to the

WWW.CASACT.ORG • NOVEMBER/DECEMBER 2013

Steven Sullivan is a writer living in Baltimore, Md. ●

ACTUARIAL REVIEW

31