fat tails tail dependence micro correlations fat tails cooke and kousky nsf# 0960865 ...
TRANSCRIPT
Fat Tails
Tail Dependence
Micro Correlations
Fat Tails
Cooke and Kousky nsf# 0960865
http://www.rff.org/Events/Pages/Introduction-Climate-Change-Extreme-Events.aspx
“Nice” distributions don’t surprise
Tallest so far
Next record
Someday, you’ll meet a taller person
Next record
But Not:
Catastrophes are Different!
Stock market
Hurricanes
Insurance
Nice distribution Women’s height [cm]
Bigger is less bigger Mean excess
Average height above heloise
Mean excess curve decreasing
Fat Tails: Natural disasters
Average above Heat1
Mean excess curve increasing
Worse is more worse Mean Excess
Fat Tail heuristics
US crop insurance claims mean excess
Variance is finite
Variance is Infinite
US crop insurance claims running average
Historical averages ‘average out’
Fat Tail Heuristics
US Flood Claims per $ Income by County and Year
Historical averages just keep growing
Katrina cost 100$B
What’s the chance that the Next Katrina will cost >200$B?
Probability that next extreme > 2 x current extreme
100 samples 2500 samples
Thin tail (exponential)
0.02 0.0008
Super Fat 0.5 0.5
Ask someone from St. Tammany County, LA:
‘After Katrina, flood loss claims in your county
totaled $240 per dollar income (2000 dollars); in the next flood at least as bad as Katrina, what do
you expect your (2000) dollar loss per dollar income to be?”
Answer: $4,000
US Flood Claims per $ Income by County and Year
Sobering Data
• Tail Risk Show
http://www.rff.org/News/Features/Pages/Understanding-Fat-Tailed-Distributions-and-What-They-Mean-for-Policy.aspx
Background on Fat Tail Distributions