Chance News 40
Quotation
When I worked at the Labor Party think tank, trying to talk about these issues [oppression of Muslim women], people always accused me of failing to back up my arguments with data. But hard numbers were completely unavailable. When I tried to find out about honor killings, for instance-how many girls were killed every year in Holland by their fathers and brothers because of their precious family honor-civil servants at the Ministry of Justice would tell me, "We don't register murders based on that category of motivation. It would stigmatize one group in society." The Dutch government registered the number of drug-related killings and traffic accidents every year, but not the number of honor killings, because no Dutch official wanted to recognize that this kind of murder happened on a regular basis.
Infidel, Free Press: New York NY
pages 295-296.
Suggested by Steve Simon
Forsooth
From the Independent, 13/09/08:
Last week, a formatting error led to us inadvertently suggesting that there was a one in 1,019 chance of the world ending before this edition. That should have read, er, one in 1019 – rather less likely. Sorry. Feel free to remove the crash helmet.
Suggested by Gareth Hagger-Johnson
From SEED Magazine (Sept/Oct 2008 issue pg 89)
If you play golf, you could be adding five years to your life. A new study
shows that the death rate for Scandanavian golfers is 40% lower than for those who don't golf. The reason may be simple: Golfers walk, spend time outdoors, and developing social relationships. The social interaction can be especially important for the older age groups. Researchers have not ruled out the possibility that golfers simply live healthy lives in
general, but they believe that the game itself does have health benefits.
Submitted by William Montante
We're off by a factor of a lot.--Tony Miller, founder of Carol.com a company that hoped to sell about 200 healthcare policies a month but after eight months sold but a total of 160.
Understanding Uncertainty
The website Understanding Uncertainty is maintained by David Spiegelhalter at Cambridge University. The website provides modules that analize uncertainty issues written by Spiegelhalter with help by others. So far they have provided the following modules:
Coincidences
National Lottery
Premier League
What is Probability?
Risk in the media
How long are you going to live?
When they are completed, a more detailed discussion appears in Cambridge Math Journal Plus
The most recent issue of Plus includes the article uncertainty: How long will you live? by Mike Pearson and David Spiegelhalter. As in previous modules this module includes elegent animations provided by Mike Pearson.
The data for this module consists of UK interim life tables 1982-2006.
In the first animation we see the age and the % Hazard = Chance of death before next birthday and the age of the individual lives evolve through time.
In the second animation we can find the chance of death before our next birthday. We put in 83 and found that there is a 9.1% chance that we will die before our next birthday,
From the third animation we find that our life expectance is 89. However in the fourth animation we find that, since we are:
Non smoker
5 day a week Fruit/Veg
Moderate Alcohol
Physically Active
our Expected Age of death is 91
We think you will enjoy this and other of their modules.
Submitted by Laurie Snell
Are bad models to blame?
How Wall Street Lied to Its Computers, Saul Hansell, New York Times Technology Blog, September 18, 2008.
There is a lot of speculation on why major Wall Street firms are reeling from piles of bad debt. These firms hire some of the best and brightest financial experts. These experts are supposed to manage risk to avoid this sort of problem.
There is some speculation that this is just bad luck.
the level of financial distress is “the equivalent of the 100-year flood,” in the words of Leslie Rahl, the president of Capital Market Risk Advisors, a consulting firm.
But that's only part of the story. Some of the blame comes from bad models.
"There was a willful designing of the systems to measure the risks in a certain way that would not necessarily pick up all the right risks," said Gregg Berman, the co-head of the risk-management group at RiskMetrics, a software company spun out of JPMorgan.
Why would anyone deliberately use a bad model?
Lying to your risk-management computer is like lying to your doctor. You just aren’t going to get the help you really need.
This may be a case of the fox being in charge of the hen house. Higher risk investments are attractive during good times because they offer greater levels of return than low risk investments. They have to offer greater returns, because they wouldn't be able to attract investors otherwise. But these greater returns will disappear if you have to hold back a large financial reserve to cover the downside risk. So there is a great temptation to pretend that high risk investments are really not high risk.
How were these models flawed? One thought is that they used the wrong time horizon.
One way they did this, Mr. Berman said, was to make sure the computer models looked at several years of trading history instead of just the last few months. The most important models calculate a measure known as Value at Risk — the amount of money you might lose in the worst plausible situation. They try to figure out what that worst case is by looking at how volatile markets have been in the past.
But since the markets were placid for several years (as mortgage bankers busily lent money to anyone with a pulse), the computers were slow to say that risk had increased as defaults started to rise.
It was like a weather forecaster in Houston last weekend talking about the onset of Hurricane Ike by giving the average wind speed for the previous month.
Another problem is that many new investments are far more complex than in the past.
"New products, by definition, carry more risk,” [Ms. Rahl] said. The models should penalize investments that are complex, hard to understand and infrequently traded, she said. They didn’t.
Submitted by Steve Simon
Questions
1. This is just one of many examples of someone being pressured to produce results that skew the numbers in favor of a pre-ordained conclusion. What other examples are there?
2. What protections need to be put into place to encourage the use of more accurate models?