A recent article in the magazine that publishes this blog (“With Data, You Get What You Pay For”) points out that virtually all the big-picture economic numbers that influence public policy and business decisions are generated by three federal agencies: the U.S. Census Bureau, the U.S. Bureau of Economic Analysis, and the U.S. Bureau of Labor Statistics (BLS). “Some of the biggest users of the data,” the author, Matthew Philips notes, “are the in-house economists at big companies.”
Philips makes the case that the economic analysis agencies do vital work. At the very least, he suggests, they shouldn’t have their budgets cut, as some in Congress have proposed.
I couldn’t agree more. But let me add an additional point: Just as important as the data is how it is used—or misused, as is often the case. Politicians, journalists, and markets focus on the “blips” in the economy rather than on longer-term economic trends.
Turn on CNBC before any announcement, such as the BLS labor market report, which gives the previous month’s jobs and unemployment numbers, and you’ll see countless guests trying to predict the numbers and offering opinions on whether such numbers would be cause for celebration or concern.
The monthly numbers are important, but it’s also important to remember that each set of numbers provides only a snapshot. The U.S. labor force, for example, includes more than 154 million people. When BLS reports a net gain of 115,000 jobs, as it did for April, we are talking about less than one tenth of 1 percent of the labor force. Moreover, the number will be revised over time as more data are collected in the weeks ahead. Given that the revisions can often be 30,000 to 50,000 per month, up or down, it seems like we could be chasing rounding errors rather than significant movements.
While the BLS and the other economic analysis agencies do an excellent job, business leaders need to focus more on long-term moving averages. The American Institute for Economic Research does just that in its monthly business-cycle report, using 24 different data sets.
Small monthly changes in very large numbers can be misleading. Keep the blips in perspective.