It's not that the feds - under both parties - are simply pulling number out of hats. They are not making it up. Instead, they use methodologies and models that incorporate happy-faced assumptions in lieu of actually measuring what it is that they are claiming to measure. That's how the Stimulus package can create-or-save millions of jobs, and it's how the birth-death model can add jobs every month in the face of actual measurements that show that jobs are being lost in every category that is actually measured. But most of the time, since the department giving these answers refuses to show its work, it's very difficult to tell exactly both how they are bogus and how bogus they are.
Not so much with the new Census Bureau measurement of retail sales:
The U.S. Census Bureau announced today that advance estimates of U.S. retail and food services sales for January, adjusted for seasonal variation and holiday and trading-day differences, but not for price changes, were $355.8 billion, an increase of 0.5 percent from the previous month and 4.7 percent above January 2009. Total sales for the November 2009 through January 2010 period were up 4.3 percent from the same period a year ago.Now the first thing to realize is that these numbers, despite the precision with which they are measured, are bogus. They are not true. If there's one real, actual number that's proportional to retail sales, it's retail sales taxes collected, and states and cities are uniformly reporting year on year drops in revenue, even as some of them raise rates. To pick an example or three, New York state's revenues are down 6%; Fort Collins, Colorado is down 5%; tiny Columbus, Nebraska is down 10%. The Rockefeller Institute of Government reports that nationwide sales tax revenues are down 4.1% and have been down five straight quarters. Those examples are fairly consistent across the board. In short, there is no way that retail sales rose 4% nationwide while sales tax collections are down 4% and more from Maine to California.
So where does the bogus, the fudge factor, come in? In methodology, of course:
The advance estimates are based on a subsample of the Census Bureau's full retail and food services sample. A stratified random sampling method is used to select approximately 5,000 retail and food services firms whose sales are then weighted and benchmarked to represent the complete universe of over three million retail and food services firms.On the surface this makes sense - you take 5000 stores, measure this year's sales vs. last, and from that extrapolate to the whole world. Except for one thing: none of the thousands of stores that closed during 2009 get counted. It would be one thing if you used the same 5000 stores from year-to-year with $0 factored in for closed stores, but this faulty selection process constantly excludes "bad news" - is it any wonder only good news is left?
Let's give an example of how this methodology can go very wrong. Let's say there are 7000 stores in the US, each with sales of $1, for a total of $7000 in sales** in 2008. In 2009, 2000 of those stores go out of business and those left have have sales of $1.10, for total sales of $5500. The Census Bureau measures the $1.10 per each of its 5000 remaining stores, compares it to the $1 last year, and declares that retail sales have risen 10% when in fact they have dropped more than 20%. I'm not saying the numbers are that wrong, but I am saying they are wrong in that direction and will always be. In a time where we have massive bankruptcies, like today, the numbers are the most wrong.
The question arises, of course, whether this bias is on purpose or the result of incompetence. I think it's the former. If it were mere incompetence, the numbers we get ought to be good or bad a little more randomly. But no one ever got fired for giving good news to his boss. Especially when the boss can use those numbers as proof of the great job he's doing.
* It's not simply that I dislike government; I dislike very much the societal effects of expensive, overweening, and (most of all) crooked government. I can handle bad news. But a government that refuses to give bad news is like a doctor who refuses to tell you that you have cancer. Yes, the effects are the same - only the body is different.
** I'm a little slow, so small, round numbers appeal to me.
(hat tip: Mish)