Economists Are Having a Huge Debate About the Real Reason First-Quarter Growth Was So Slow

"Residual seasonality" skews quarterly GDP figures

Wind-Driven Snow Blankets East Coast, Shuts Airports

Snow was blamed for this year's weak first-quarter figures.

Photographer: Peter Foley/Bloomberg

The advance estimate of U.S. growth for the first three months of the year showed the economy expanded by just 0.2 percent on an annualized basis, far shy of the 1 percent median forecast.

Temporary factors such as larger-than-average snowfall and the West Coast port strike weighed on activity, but this is not the first time in recent memory that first-quarter growth disappointed by a substantial margin.

Since 2010, the rate of economic expansion in the first three months of the year has paled in comparison to the average rate for the following three quarters, leaving economists searching for answers about whether something's awry in how these figures are calculated.

This week, economists from Wall Street and the Federal Reserve battled it out over whether the first-quarter growth statistics are fundamentally flawed.

Barclays Chief U.S. Economist Michael Gapen and economist Jesse Hurwitz think they've solved one of the biggest mysteries about the U.S. economy. As they put it in a note to clients:

We find that Q1 GDP growth has systematically underperformed since the 2008-09 recession, with growth in Q1 averaging 1.9pp less than the remaining quarters in the same calendar year. This residual seasonality in US GDP data was apparent prior to the crisis but has become significantly more pronounced since.


Gapen and Hurwitz figure that the "residual seasonality" that has weighed upon first-quarter growth figures can be traced back to the way the Bureau of Economic Analysis prepares the U.S. quarterly national income and product accounts. They cite the example of warehouse construction—a component of investment in nonresidential structures—to demonstrate the three ways in which quarterly growth figures can be skewed. The BEA receives nominal, seasonally adjusted monthly figures on construction spending from the Census Bureau.

Since 2010, investment in nonresidential structures has contracted by an average of 10.9 percent in the first quarter while rising an average of 9.4 percent in the following three quarters, thanks to deficiencies in how the Census Bureau compiles and seasonally adjusts the data.

1) Time aggregation: The Census Bureau generates the quarterly estimate for warehouse construction by adding up seasonally adjusted figures from the Value of Construction Put in Place report, which is released on a monthly basis. This method of tallying up seasonally adjusted months of results, however, does not necessarily produce an accurate seasonally adjusted quarterly figure. In other words, quarterly warehouse construction can be more—or less—than the sum of its seasonally adjusted parts.

2) Cross-sectional aggregation: Warehouse construction is only one of many components that make up the "investment in nonresidential structures" line item of gross domestic product. These different variables can have small seasonal variances that are individually insignificant but have a much larger seasonal effect when the group is presented as a cohesive whole.

3) Price adjustment: The BEA takes nominal warehouse investment and deflates it by its respective producer price index. Though both of these metrics may be seasonally adjusted, combining these can also “contaminate” reported growth figures, according to the economists.

“These three sources of residual seasonality suggest that the contrast between repeatedly soft Q1 growth and activity in remaining quarters likely arises from a combination of poor or incomplete seasonal adjustment of the underlying source data and the process by which the BEA converts that source data into the NIPA [national income and product accounts] data,” they assert.

The national income and product accounts are measures of different components of GDP that are compiled to produce the quarterly economic growth figures.

According to Barclays, the component that displays the most residual seasonality—that is, where the shortcomings of the current seasonal adjustment methodology are most evident—is structures investment, which helps explain why reported first-quarter growth comes in lower than the following quarters.

By its nature, construction activity is more affected by severe weather than other areas of the economy.

In addition, the economists claim that the sample size for the construction series is too small and difficult to smooth given the boom-and-bust cycle of the 2000s.

Exports and defense spending also have prominent seasonal trends that do not appear to be accurately captured by the current adjustment process, the Barclays duo argues.

After reapplying a seasonal adjustment filter, the economists find that growth in the first quarter has been higher than officially reported in the postrecession era, and that third-quarter growth statistics, in particular, have been artificially high.


In fact, using their method of adjustment, Gapen and Hurwitz estimate that first-quarter growth in 2015 was actually 1.8 percent, well above the advance estimate of 0.2 percent.

While that still represents a slowdown in the rate of economic expansion, the U.S. economy was also buffeted with one-off shocks in the first three months of the year, namely, the strike at the West Coast ports and above-average levels of snowfall.

Economists at the Board of Governors of the Federal Reserve System, however, beg to differ with this analysis.

Though "there is something a bit unusual" about the gap between first-quarter growth and the economy's performance in subsequent quarters, there isn't a smoking gun to hang one's hat on, they write.

However, the Fed economists do acknowledge that some of the components of GDP highlighted by the Barclays economists do indeed display a seasonal pattern despite the Bureau of Economic Analysis's attempt to strip out that effect.

"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," they write.

Gapen and Hurwitz then offered this re-rebuttal to the claim that the Q1 growth numbers do not contain statistically significant residual seasonality. "Although [the Federal Reserve Board economists] do not find the difference statistically significant at the 95% confidence level, had they chosen a confidence interval such as 90%, they would have found statistically significant residual seasonality."

Their primary conclusion was supported by Tom Stark, an economist at the Federal Reserve Bank of Philadelphia, who examined the economy's relative performance in the first three months of the year over the past 30 years.

"It seems clear that the BEA’s averages for quarterly real GDP growth vary according to a consistent pattern over the four quarters of the year, suggesting that the BEA’s seasonal adjustment procedures are not filtering out all the intra-year movements in the data," he wrote. "This first-quarter effect is economically large and statistically significant."

Residual seasonality will continue to affect quarterly growth figures if the Census Bureau and the Bureau of Economic Analysis fail to change their methodology, the Barclays economists warned. In the meantime, they suggest looking at other economic indicators—such as unemployment and manufacturing activity—to gauge the true direction of the U.S. economy.

The residual seasonality in US GDP data is one reason why we continue to put significant weight on other high frequency economic indicators, including the monthly employment report and ISM surveys. These indicators are easier to prepare, are revised less frequently, and are an important tool for taking the pulse of the US economy in real time. We see them as likely to be more informative about the pace of activity and state of the economic cycle than GDP.

Stark, for his part, says other relatively obscure gauges of quarterly economic activity provide superior insight into how the U.S. economy fares in the first three months of the year:

Model-based estimates of real GDP that draw on the BEA’s official measures for real GDP and real gross domestic income (GDI), such as the Federal Reserve Bank of Philadelphia’s GDPplus, display little evidence of a first-quarter effect, suggesting that they provide better readings on first-quarter growth than the BEA’s estimate of real GDP does.

Before it's here, it's on the Bloomberg Terminal. LEARN MORE