This past Friday, the Bureau of Labor Statistics issued a press release stating the monthly employment figures for May 2016. The headlines following the report’s release emphasized that the U.S. economy created only 38,000 additional jobs in May 2016. This figure promptly caused the stock market to fall. The next day’s Wall Street Journal front page led with a story declaring “Weak Hiring Pushes Back Fed’s Plans” for an interest rate hike. But the true figure was much higher than 38,000 jobs.
The actual number of additional jobs created between April and May was 651,000! This much larger figure is calculated using numbers found in table B1’s top line located on the press release’s 28th page.
Why did the government report indicate that 38,000 and not 651,000 more people were employed in May than in April? The 38,000 figure is seasonally adjusted, while the 651,000 figure is not.
The seasonal adjustment process, called X-13ARIMA-SEATS, is quite complex. However, the general idea is simple. The number of people employed jumps up and down dramatically from month to month. The adjustments smooth the data series to show long-term trends in the labor market by accounting for the regular pattern occurring each year because of things like weather, major holidays and school schedules.
To emphasize the long-term trend, the government increases the employment figures in low employment months, such as January, February and March, and reduces them in high employment months, like May and June.
However, highlighting seasonally adjusted data is problematic, because people in businesses like construction, transportation and manufacturing do not look at seasonally adjusted conditions when hiring and firing. Instead, their decisions are based on the current season, weather and business conditions. Seasonally adjusting the data eliminates some of the very patterns that managers watch when making hiring decisions.
If I just hired 1,000 new people for my factory and then read that only 38,000 people were hired nationwide, I would start wondering if I had made a big mistake. The 651,000 figure would give me a very different impression.
Is the 651,000 figure low for May compared with the past few years? Yes, it is. But it’s too soon to panic and declare the end of the U.S. economic expansion.