An in-depth analysis by CNBC of the government's reports on gross domestic product suggests large and persistent errors that should give investors, business executives and policymakers pause in relying on the data for key decisions.
CNBC looked at each quarterly report going back to 1990 and found an average error rate of 1.3 percentage points. So an initial report of 2 percent growth on average later would be revised to 3.3 percent or 0.7 percent.
The research does not show any systematic overstatement or understatement of growth, just persistently large revisions.
CNBC also found:
- The error rates in the second and third estimates of GDP are the same as the first. So despite more time and data, the error rates will be just as large three months after the end of the quarter as they are one month afterward.
- About 30 percent of the time, the government gets the direction of growth wrong. That is, GDP initially shown to be higher than the previous quarter could in fact be lower, and vice versa.
- The error rates haven't improved over the decades despite vast improvements in computing power and communications speeds. The size of the revisions from 2008 to 2013 is the same as those from 1990 to 1995.
The errors have potentially important policy implications. On April 30, 2008, with the Great Recession just gathering steam, the Federal Reserve cut interest rates by one-quarter point to 2 percent. That same day, the Bureau of Economic Analysis, the agency that produces the GDP report, announced that the economy was growing an anemic 0.6 percent but still growing.