Imagine the following scene: NBC reporter Andrea Mitchell and her husband, Alan Greenspan, decide to take advantage of the post-inauguration lull and get away from Washington. A few days at a Colorado mountain resort will be wonderful.
They arrive at Aspen. After reaching their room, Mitchell changes into something suitably fashionable for socializing. But her lips purse in frustration when she sees her husband pull on old blue jeans and galoshes and grab a flashlight and measuring tape from his bag. “Oh, Alan,” she protests, “You aren’t going to check out a sewage treatment plant again?”
Sound improbable? Keep in mind that Federal Reserve Chairman Greenspan has a reputation for knowing every possible indicator of how the U.S. economy is performing, no matter how obscure. And Colorado tourism officials have just added a new stat to the list: daily flow rates for sewage treatment plants at ski resorts.
Their problem is that resort operators won’t disclose exactly how many lift tickets they sell nor how many rooms are rented or meals sold. This is proprietary information that might be useful to competitors in the dog-eat-dog recreation business.
Colorado officials, however, want some idea of how the crucial ski industry is doing. If you cannot count paying skiers, count toilet flushes. At places like Aspen or Snowmass, the two are highly related. Compare wastewater flows to comparable periods in prior years and you get a rough indication of how the recreation industry is doing.
An amusing scenario, but is there a broader lesson? Yes: It is that measuring economic activity is difficult, and in many ways it is getting harder as we move to a more decentralized service economy.
Counting automobiles, locomotives or tons of steel is not hard. Counting lines of computer code, Internet auction trades and shiatsu massages is more difficult. Yet failure to count correctly can lead to erroneous conclusions about how the economy is performing.
Economists were troubled by a decline in productivity growth rates that took place about 1973. By the mid-1990s many were convinced that productivity growth was surging back but that the official tabulations were missing some growth. Some discussion centered on the degree to which erroneous price index calculations produced a downward bias in measured output and household consumption. Adjustments were made to the way the Bureau of Labor Statistics tabulates the Consumer Price Index, but the broader question of how to measure effectively the new economy remains under study.
Sometimes a common sense observation corrects mistakes in official statistics. After the Soviet Union collapsed nearly a decade ago, official numbers showed that output in Russia was down by more than half. There was much hand-wringing about what to do. Then someone noted that while production of electricity was down, it had not dropped nearly as much as tabulated output. As with skiers and wastewater, there is a rough correlation between gross domestic product and energy use.
Closer checking showed that the observant analyst was correct. Output had dropped, but not nearly as much as Russian government numbers showed. Much production was not reported because of tax evasion reasons. The true situation was bad but not catastrophic.
The overall lesson is to remember that measuring economic activity is a very imperfect science. Most published statistics are fairly accurate, but they should not be treated as exact or infallible. Wherever possible, it is best to look at a number of indicators from different sources, and like a jury, weigh the preponderance of evidence before reaching a hard conclusion.
One final note. Even the most carefully tabulated indicators are subject to the constraints of sample size. The Current Population Survey carried out each month in the United States is perhaps the most meticulously designed and administered survey in the world. It is used to produce unemployment rate estimates every month and annual estimates of household incomes.
But the only questions 50,000 out of the more than 100 million households in the United States. As a result, its tabulations of how specific population subgroups fare are of limited usefulness.
For example, if you want to use CPS data to determine how Native Americans in Minnesota are doing, save your time. There are only two Minnesota Native American households in the national sample of 50,000. Any conclusions you draw will be essentially meaningless.
Why such a small sample, you ask? The sample size was reduced sharply early in the Reagan administration for budget reasons and no Congress or president in the succeeding two decades has seen fit to restore it. So while information about our economy and society is both scarce and valuable, there is no lobby for any greater effort.
© 2001 Edward Lotterman
Chanarambie Consulting, Inc.