Saturday, November 30, 2013

What is your flood risk tolerance?

I recently read an article in the December 2013 issue of Money magazine titled "After the Flood", which chronicled the financial difficulties of a Staten Island couple in the wake of Hurricane Sandy. Sandy made landfall on October 29, 2012 and when it was all said and done, caused $70 billion in damages across 24 states. The couple stayed put during the storm, which inundated their house up to the second floor and floated their cars several blocks down the street. Their home value decreased from $303,000 to $204,000 post-Sandy (likely due to flood damage and higher cost to insure) and they still owe $275,000 on their mortgage - putting them literally and figuratively underwater. Even worse for the couple (but perhaps better in the long term), new post-Sandy FEMA building standards will require the couple to elevate their home by 10 feet to place it above the new flood elevation - a very costly proposition which they cannot afford. These are the costs and risks of living in flood prone areas.


Predicted and observed tide levels at the Battery, NY during Hurricane Sandy. October 29, 2012 was the highest ever recorded tide at this location due to coincidence of high tide and peak storm surge.

The Money article provides more details on the financial struggles of the couple and the steps they are taking to get their lives back on track. The article got me thinking about our attitudes towards risk and the value we place on home ownership, even if we do live in disaster-prone, yet desirable, areas. This was highlighted the most clearly by the very last paragraph of the article, which noted:
Still, they're eager to move on and to finally be able to put Hurricane Sandy behind them. Says Motherway (the wife): "You just hope it's the kind of storm that happens only once every 100 years."
As an engineer, I consider flood risk from a statistical perspective and not one based on wishful thinking - if you live in a flood zone, you are playing the odds and betting that either: (1) you will not encounter a storm severe enough to flood you, (2) you have adequate insurance to cover your losses, or (3) the state and/or federal government will come to your aid in the event of a natural disaster. For example, if you live in a low-lying area behind a levee that provides 100-year flood protection, you are presumably protected from all less severe flood events (e.g., 10-year, 50-year, etc); however, for any storm more severe than a 100-year event, you will be flooded. That is a level of risk that many Americans live with every day. The problem with the terminology "the 100-year flood", as illustrated by the Staten Island couple's statement, is that it implies such a storm will occur only once every 100 years. More accurately stated, the 100-year flood is an event that has a 1% chance of occurrence in any individual year (the "1% annual chance" terminology is preferred by risk analysts). Over a very long period of time, this severity of storm will indeed occur, on average, once every 100 years, but over the short term it could occur more or less frequently. For example, during El Ninos in California, we can have back-to-back severe weather events (e.g., rain, water level, and wave) during a single winter storm season, separated by years of calmer conditions.

Using a statistical concept known as "encounter probability", we can estimate the likelihood of experiencing a storm of a particular severity over a specified time frame. For example, the commonly quoted statistic that the San Francisco Bay area has a 63% chance of experiencing a 6.7 magnitude earthquake or greater over the next 30 years is a type of encounter probability analysis. In terms of flooding, there is about a 25% chance that a homeowner will experience 100-year flooding or greater over the lifespan of a standard 30-year mortgage. Now, I don't know about you, but that seems like a pretty high likelihood to me! It's actually a bit frightening to think that our nation's flood protection infrastructure in any one location generally has a 25% chance of being overwhelmed by flood waters over such a short length of time. Using the concept of encounter probability, I produced a set of curves (shown below), which illustrates the likelihood of experiencing a storm of a given severity over a specified length of time. By selecting a storm severity curve and project lifespan, the likelihood of flooding (encounter probability) can be determined. For example, using the green curve below, there is an 80% chance of experiencing a 25-year flood or greater over a 40 year time span.

Encounter probability of an extreme storm event over a given duration.

FEMA's National Flood Insurance Program uses the 100-year floodplain as the standard to assess whether or not a homeowner is required to purchase flood insurance. Most mortgage companies require borrowers to purchase flood insurance to protect their loans until homeowners pay off their mortgages (after that, insurance is recommended, but not required). I calculated the encounter probabilities for storms of different severity over a standard 30-year mortgage period and a 75-year lifetime of a typical American (see table below). As you can see, the chances of experiencing a moderate to strong storm (say, 25 to 50-year event) over these time periods are quite high. For reference, I also estimated the encounter probability for two recent extreme storms - Hurricanes Katrina (300 to 400-year event) and Sandy (1,000 to 1,500-year event). Estimates of the exact frequency of these types of storms are uncertain given our relatively short period of record, which is why I show a range of likelihood. Based on these estimates, it seems fair to say that Hurricane Sandy was a "once in a lifetime" storm, whereas Hurricane Katrina, while rare, has a reasonable chance of recurrence (approximately 20%) during our lifetimes (especially considering the effects of climate change, which could increase sea levels and storm frequency and intensity - but that's another topic!).

Likelihood of flooding (encounter probability) for different storm severities over a 30-yr mortgage period and 75-yr lifetime:


Return Period
(yrs)
30-yr Mortgage
Encounter Probability
75-yr Lifetime
Encounter Probability
5
100%
100%
10
95%
100%
25
70%
95%
50
45%
78%
100
26%
53%
200
14%
31%
Katrina
7-10%
17-22%
Sandy
2-3%
5-7%

I polled my girlfriend, Allison, and her sister, Tina, to see what level of risk they would take on assuming a 30 year habitation in their home. Tina indicated she would assume a 20% risk of flooding (comparable to FEMA's acceptable risk level) and Allison would accept a 10% risk (on the order of a Katrina-type storm). I plotted those points on the figure below (blue and red triangles) and they roughly translate to the 125 and 300-year flood events, respectively. Personally, I would tolerate a much lower risk of flooding, probably more along the lines of 5% (a 600-year flood event). This indicates that all three of us would not feel comfortable owning property within FEMA's 100-year floodplain.

Personal risk tolerance relative to likelihood of flooding during a 30-year mortgage period.

So what is your risk tolerance? If you purchased a home in a river or coastal flood prone area, what likelihood of flooding would you accept? It's not a simple question - what other factors would you consider?


References:
Money Magazine
http://money.cnn.com/video/pf/2013/10/28/pf-superstorm-sandy-staten-island-home.moneymag/

Climate Central
http://www.climatecentral.org/news/sea-level-rise-has-already-doubled-risk-of-sandy-level-flooding-in-nyc-16434

Significance Magazine
http://www.significancemagazine.org/details/webexclusive/3004001/Hurricane-Sandy-a-look-back.html

The Lens
http://thelensnola.org/2013/11/05/armoring-overtopped-levees-against-erosion-and-collapse-will-grass-cut-it/