Whether or not the U.S. economy is tanking may be a question of whether statisticians at the Bureau of Economic Analysis have run the numbers right.
In the wake of a report that showed the economy nearly stalled out in the first quarter, analysts outside the bureau, including those in the Federal Reserve System, are questioning whether the bureau is correctly accounting for seasonal fluctuations in the data.
On Monday, researchers at the Federal Reserve Bank of San Francisco posted a report finding that the bureau’s adjustments are failing to catch a seasonal pattern in the first quarter, an oversight that “may understate the true strength of the economy.”
Their research joins recent analyses from the Federal Reserve Board of Governors, the Philadelphia Fed and Barclays, among others, that raise questions about whether the economic weakness in the first quarter is a statistical artifact, rather than a serious sign that the U.S. may be at risk of another recession this year.
The question is important, wrote Philadelphia Fed economist Tom Stark on Thursday, “because economists must decide whether an observed slowdown in first-quarter growth represents residual seasonality or the start of a cyclical slowdown.”
Nicole Mayerhauser, the chief of the bureau’s National Income and Wealth Division, responded that the agency “is aware of the potential for residual seasonality in GDP and its components, and we are continually looking for ways to minimize this phenomenon.”
The problem is that the gross domestic product figures are adjusted to remove predictable seasonal movements that make it difficult to assess ongoing trends. For example, economic activity usually slows sharply during the first quarter, as the frenzied shopping of the holiday season gives way to the slow winter months. Then, in the second quarter, the weather warms up and construction picks up, boosting commerce.
The bureau smooths out those variations to make different quarters comparable to each other, by using seasonally-adjusted source data, such as retail sales, construction and exports.
In theory, there should be no differences in average growth by quarter of the year.
In practice, however, there are differences. Between 2000 and 2014, the San Francisco Fed found, growth in the first quarter was 2.3 percentage points lower than the rest of the year on average.
The researchers speculate that discrepancy might be due to the fact that insignificant seasonal patterns in the source data might become significant when the bureau aggregates them, or to the possibility that patterns that are not evident in monthly data show up in quarterly GDP data.
To address those concerns, the San Francisco Fed researchers performed a “second round of seasonal adjustment” on the GDP data to account for the residual seasonal patterns.
With that second seasonal adjustment in place, they found, first-quarter growth would have been at a 1.8 percent rate, rather than the 0.2 percent in the bureau’s first estimate.
That analysis basically mirrored what Stark and Barclays’ economist found.
Staff at the Federal Reserve Board of Governors in Washington, however, released an analysis Thursday in which they did “not find convincing evidence of material residual seasonality in GDP in recent years.”
Instead, a group of researchers wrote, “the first-quarter weakness appears to be driven by a couple of outlier years and by soft readings in a varying subset of underlying components.”
Their analysis, however, only covers the years 2010-2014.
A few developments in recent years could make it more difficult to tease out background seasonal variation.
One is the steep first-quarter contraction in 2009, when the economy was in a free fall as a result of the financial crisis. Another is the unusually severe winters the U.S. has experienced recently.
Both those factors, however, would lead the seasonal adjustments calculated by the bureau to overstate growth, not understate it, the San Francisco Fed paper noted. It is possible that the underlying, long-running problems with the seasonal corrections offset the overstatement caused by the crisis and harsh weather.
The reality will come to light in the months ahead, as economic output either recovers or continues to stagnate.
In the meantime, the bureau said, it’s reviewing the possibility of residual seasonality in its calculations. It is currently examining the possibility that seasonal differences in government defense spending are not adequately accounted for.
