Nonfarm payroll employment grew by 243,000 in February, and the unemployment rate was little changed at 4.8 percent, the Bureau of Labor Statistics (BLS) of the U.S. Department of Labor reported today. Stock futures rose on the announcement, which was better than the expected 210,000 increase. Apparently we are close enough to the end of Fed Funds rate hikes that good news is interpreted as good news again.
For those of us geeks that read the employment report, we know that the data are sometimes misleading. First of all, they are not especially accurate. At an unemployment rate of around 5.5 percent, the 90-percent confidence interval for the monthly change in unemployment is about +/- 280,000. So we see that at an estimate of 243,000 jobs, the BLS is 90 percent sure the actual number of jobs created was between a loss of 37,000 and a gain of 523,000. For some reason, that doesn’t make for as good a headline as a 243,000 gain. People might question whether it is worth all that money collecting the information if we are that uncertain about it. It is especially relevant considering the market’s reaction was based on the 33,000-job surprise. If the 90 percent confidence interval is +/- 280,000 jobs, how sure can we be that the number of new jobs was actually higher than expected? I’m sure the math could be done on this, but suffice to say we can only be a little over 50% sure that the number of jobs created was more than economists thought it would be.
Another issue with the reported data (given on a month over month basis) is the seasonal adjustments made. As the BLS points out, “Over the course of a year, the size of the nation’s labor force and the levels of employment and unemployment undergo sharp fluctuations due to such seasonal events as changes in weather, reduced or expanded production, harvests, major holidays, and the opening and closing of schools. The effect of such seasonal variation can be very large; seasonal fluctuations may account for as much as 95 percent of the month-to-month changes in unemployment. Because these seasonal events follow a more or less regular pattern each year, their influence on statistical trends can be eliminated by adjusting the statistics from month to month.” This is fine as long as the weather or number of students leaving school each year is relatively constant as a percentage of the labor force. Otherwise, the seasonal adjustment could introduce an error. The BLS knows this, and makes several revisions to the data over time, resulting in a final tally that is pretty reliable. However, much less is made over these revisions, as they are considered old news by then.
So what are we to do with the information? Our preference is to look at the change in jobs on a year/year basis. This has several advantages. First, because the year/year number is larger than month/month, the relative size of any mis-estimation is smaller. Second, we can use non-seasonally adjusted data because the last we checked February was in winter both in 2005 and 2006. Finally, it makes use of all of the revisions made to last February’s data. What does this analysis tell us?
Between February 2005 and February 2006 the economy added 2,070,000 jobs, slightly more than the 2,051,000 reported in January. As a percentage of starting employment, the increase was 1.6 percent Y/Y, compared to 1.5 percent in January. As the chart below shows, it is a decent level and improving from recent (Katrina-related) lows. In this case, the market reaction appears justified.
Like this article? Why not try out: