Tuesday, June 17, 2008

Statistics as a government plot 

There are two kinds of statistics: the kind you look up and the kind you make up

I got an email after last week's appearance on the David Strom show that accused me of lying because I was using official data. Here's a snippet:
The Government data you cite is a fraud. Government lies on unemployment numbers, GDP, CPI, inflation and any other numbers should be suspect. Inflation is understated because the government doesn�t want to pay out Cola�s to Federal Employee retirees, SSRI recipients and other retirees. Think of the cost. Unemployment is deliberately understated and they don�t count people who stopped looking for work. John Williams who writes Government Shadow Statistics gives you accurate numbers and received praise for his work. We are headed for the worst Recession and Depression we will face in our lifetime. The big banks in New York are insolvent.
This is not the first time I have heard this, and I'm familiar with John Williams' work. Barry Ritholz has written not too long ago about the "disappearance" of M3, suggesting sinister intent. (I would argue for Goodhart's Law instead -- if M3 ever worked enough to use it, targeting it would cost us the information content in M3. So Barry should be happy they don't publish it, because that keeps its forecasting power intact!) And it's not new -- counting employment has been hard for decades, and so has been kvetching about the counters. David Leonhardt gives his usual sensible look at how people misperceive CPI.

Now many of us have favorite statistics, adjustments we make to data, that we think help us with forecasting or explaining what is happening currently in the economy. And between revisions and the results of budget cuts, most of us have gone through periods where data we used to look at disappears. Our local economic indicator series used to rely on a measure of local electricity hookups. When Xcel Energy bought Northern States Power we lost that series because the office that collected it was moved from Minnesota to Colorado. Nobody of course thought this sinister (except maybe those folks who worked in the Minny office -- one of whom was a former student that helped us get the data!) and nobody complains. So when government actually makes a choice to cut a budget and save a few dollars -- or suggests outsourcing or raising private funds to improve data collection -- I think that's something we should support.

People will believe what they want, and some will sift and sort data to fit a story line they already have. For example, Lee Egerstrom seems determined to look at data and decide things must be bad: He's not necessarily right or wrong, but reasonable people can look at the same data and see two different things. Thus is the nature of turning points in the business cycle, and in making any argument you want to de-emphasize or discredit the data of the other side. But some people seem determined to view government data with more than just a good dose of scientific skepticism when it doesn't fit the story they already believe.