For a second year in a row, the first-quarter GDP estimate indicated a contracting economy—a fact many economists find disconcerting, since employment growth remained strong and wage growth finally started to pick up.
For the second year in a row, the first-quarter GDP estimate for the United States indicated a contracting economy—a fact that many economists find disconcerting, since during the same period, employment growth remained strong and wage growth finally started to pick up. The focus is now on whether the GDP accounting needs improvements.
GDP’s first-quarter 0.2 percent contraction left us scratching our heads, just as when the first quarter of 2014 showed a contraction of 2.1 percent, on how an economy that by many measures appears to be growing actually shrunk. True, both of these first quarters experienced abnormally cold weather, and some factors in early 2015 negatively impacted exports, including a labor issue at the West Coast ports, a relatively high US dollar, and slower growth among some of our trading partners. But even with these headwinds, these contractions seem out of sync with reality, leading to the question: Is there a problem with how the government’s statisticians calculate US GDP?
Researchers quickly closed in on the “seasonal adjustment” process as the most likely culprit. The government publishes most official statistical series with the impact of regularly occurring patterns removed to show changes in the underlying trends. For example, the Census Bureau adjusts automotive production in July as factories shut down to retool for new models, and it adjusts heating oil production to account for the regularly observed September increases in anticipation of the winter heating season.1 The Bureau of Labor Statistics (BLS) adjusts the individual components of the consumer price index for a wide variety of factors that can cause seasonal variation in prices, such as changing climatic conditions, production cycles, model changeovers, holidays, and sales events. As an example of a regularly occurring pattern, BLS points to oranges, which can be purchased year-round but whose prices are significantly higher in the summer months, when the major sources of supply are between harvests.2
In constructing the GDP estimate, the Bureau of Economic Analysis (BEA) draws on data from many sources, much of which each collecting agency seasonally adjusts, with the BEA seasonally adjusting the remainder if the time series is sufficiently long. Both the BEA and the source data agencies review and update their seasonal adjustment procedures regularly, but even with regular reviews and updates, there are several ways in which uncorrected seasonal patterns can remain in the GDP estimate. These include seasonal patterns in the aggregates, not apparent in the individual components; patterns at the quarterly level not present in the underlying monthly data; and patterns that appear in the deflated estimates and that are not apparent in the nominal estimates or the prices.3
The impact of this “residual seasonality” on the quarterly estimates may be substantial. A recent study by researchers at the Federal Reserve Bank of San Francisco applied a second round of seasonal adjustment to the published real GDP data and concluded that analysts have underestimated first-quarter GDP since the late 1990s—and that the underestimation has risen in recent years to about 1.5 percentage points. Therefore, according to this analysis, first-quarter GDP should have shown weak, but firmly positive, growth.4
Figure 1 illustrates the seeming mismatch between the quarterly GDP estimates and another indicator of economic health: employment. As the US economy entered recession in early 2001 and again at the end of 2007, employers reacted quickly to reduce headcount. Even when output began to rise as the economy shifted into expansion, employment continued to fall, giving rise to the concept of the “jobless recovery.” But eventually employment began to increase in both of these expansions. Where the first quarters of 2014 and 2015 stand in stark contrast to expectations is in the strength of employment growth, even as GDP apparently contracted or stalled.
The pickup in wage growth was another signal that the first quarter of 2015 was stronger than the GDP estimates portray.
The pickup in wage growth was another signal that the first quarter of 2015 was stronger than the GDP estimates portray. Much attention is generally given to the monthly changes in average hourly wages published alongside BLS’s monthly employment estimates. However, changes in the average hourly earnings mask changes in industry and occupational shifts; another BLS series, the quarterly Employment Cost Index (ECI), corrects for this by measuring the change in labor cost, free from the influence of employment shifts among occupations and industries.
The ECI tracks the two major components of labor costs: wages and salaries (approximately 70 percent of compensation costs) and benefits (the remaining 30 percent). As shown in figure 2, the ECI for wages and salaries has accelerated in recent quarters after remaining relatively constant for most of the recovery.
Even if some unaccounted-for residual seasonality depressed first-quarter GDP, it was still a weak quarter. Although the labor market remained strong, other indicators—such as the large decrease in exports—constrained growth. The combination of a strong dollar and slower growth among some US trading partners caused exports to decline in both real and nominal terms. Figure 3 shows the percentage change in nominal exports during the first quarter of 2015. Overall, exports declined 4.2 percent, with the decrease more than accounted for by the drop in goods exports. Indeed, services exports actually increased. On a country basis, the five largest US export markets, accounting for 45 percent of total exports, all declined. The declines were particularly large in goods, with only Japan showing a higher percentage drop in services.
The combination of a strong dollar and slower growth among some US trading partners caused exports to decline in both real and nominal terms.
Other weaknesses in the first quarter include consumers who did not take advantage of lower gas prices and higher incomes to spend more—instead, they paid credit card debt and increased savings, although they did take out more in auto and student loans. Mortgage debt, the largest debt category, remained unchanged from the fourth quarter.5 Business investment also slowed, as declines in structural investment overwhelmed the increased spending on equipment and intellectual property (primarily software, and research and development). Residential investment still appears to be sputtering along, with no new upward shift in trajectory.
Although factors causing the first-quarter weakness (whatever its magnitude) may spill over into the second quarter of 2015, the United States will likely experience stronger growth in the second half of the year. The Federal Open Market Committee of the Federal Reserve Board continues to monitor conditions, looking to judge the correct point to begin to raise interest rates. There are no major imbalances present—at least imbalances within US borders. Unfortunately, there is no shortage of external situations that might spiral out of control and disrupt world growth.
After the BEA issues its regular annual revisions in July, which may include a change to some of the underlying seasonal adjustments, we might have a clearer idea of where we have been—a necessary prerequisite to better contemplating future US prospects.