In the first quarter of 2014, the government sent shivers through global financial markets by reporting a recession-like contraction in the U.S. economy of 2.1 percent. Despite severe winter storms and record-breaking cold, economists still could not account for the deep plunge in growth. Even Federal Reserve officials worried something worse than just the effects of bad weather could be amiss in the economy.
Turns out, it was the data itself and how the government calculates gross domestic product that were actually amiss. The Bureau of Economic Analysis, the federal agency responsible for GDP, now says growth was actually 1.2 percentage points higher in the quarter for a more manageable and understandable decline of 0.9 percent.
The bureau announced Thursday its regular annual revisions of the past three years and confirmed a series of stories in CNBC that raised questions about the seasonal pattern of growth. Overall, the government revised down GDP over the three years ending in 2014 by a modest 0.3 percent but also took the first steps in changing the way it calculates GDP to account for anomalies in the quarterly pattern of growth.
Read MoreFirst reading on Q2 US GDP at 2.3%
From 2012 to 2014 growth actually averaged 2.1 percent, compared with the originally reported 2.4 percent. The bulk of the downward revision came in 2013, where GDP is now reported to have increased a meager 1.5 percent, down 0.7 from the prior level. The years 2012 and 2014 barely budged.
Bureau officials, in a briefing to reporters earlier this week, said the overall change was not out of line with prior revisions. However, they said the new numbers resulted from a series of initial steps they have taken to correct problems in how GDP is calculated, including seasonally adjust federal defense spending, and how it calculates inflation.
CNBC found that first quarter GDP has run substantially below prior quarters for both the past six years of the expansion and for as long as 30 years, leading to what appeared to be outsized market anxiety over slowdowns that could be more statistical than real. Several economists confirmed CNBC's findings.
In its report, the bureau revealed a new problem that CNBC had not found: Overstatement of growth in the third quarter for at least the past three years. In fact, the biggest change over the three-year period was a 2 percentage point downward revision to the third quarter of 2012 to show just a half-point of growth, and a 1.5 percentage point downward adjustment to the third quarter of 2013 to show 3 percent growth. In both quarters, the biggest subtraction came from lower federal defense spending.
But there are likely more revisions to come. The bureau said it had only completed the first of three phases of reviewing the GDP numbers and planned to go back further than just three years.
The revised GDP data show discrepancies between quarters still exist but the quarters have all moved closer toward the overall average of the three-year period. For example, first quarter growth, which was revised up by 0.2 percentage points over the period, is now 0.5 percent closer to the three-year average than initially reported. However, it is still by far the weakest quarter. That's in part because, with the broader downward revision to 2013, first quarter growth was reduced by 0.8 point to now stand at 1.9 percent.
The new methods could mean more muted variations in quarters—and less unwarranted market anxiety. When properly adjusted for seasonal variations, economists say, GDP data should not show big quarterly variations over long periods of times.
In the current phase of its review, the bureau used new seasonal adjustments for federal defense spending and for consumer purchases of services. It also updated the seasonal adjustments it makes to a variety of data, including construction, sales and inventories. It worked with other agencies that provide data for the massive GDP calculation to begin seasonally adjusting some reports that previously were not.
By sectors, the biggest contributions to the downward revisions for the three-year period came first from consumer spending and second from government spending. For consumer spending, there were changes in almost every sub-category, but the biggest was in financial services, mostly the result of using a new method for calculating price changes, or inflation, in the sector.
In government, the biggest change came from downward revision to state and local spending, which a bureau economist said was simply the result of lagged data. However, another part of the revision came from new seasonal adjustments to federal defense spending. CNBC had pointed to that category as an area with seasonal patterns that the government might not be accurately taking account of.
In phase two, the bureau said it will conduct a component-by-component review of the detailed series that measure GDP and gross domestic income, working with other agencies to make any needed adjustments. In phase three, it will try to increase transparency around the data by developing a nonseasonally adjusted set of statistics, so the public can better see how the numbers are adjusted.