Archive of payroll jobs outlook discussion:

March 2005 -- The February Employment Report: The strong increase in payroll employment by 262,000 jobs in February continued the solid jobs gains observed for most of the past year and a half. Payroll employment has increased by 3 million jobs since May of 2003 -- the trough of the payroll employment series following the recession. The unemployment rate rose slightly to 5.4 percent from 5.2 percent in January.

All in all, the report is a good one -- continued solid growth in jobs -- but not too strong of a report that would rattle financial markets because of concerns that the Fed would have to raise interest rates faster or more than previously expected. The low wage growth numbers in the report helped to alleviate concerns in financial markets about inflationary pressures. In fact, the bond and stock markets rallied following the report.

Because of revisions to the December data, the report also provides "final" information on the performance of payroll jobs relative to projection ranges that were made in October 2003 (before any solid evidence of payroll jobs growth had even been observed). The chart at right shows that the actual payroll jobs performance is almost exactly in the middle of the projection ranges -- actually slightly better than the middle (in the 57th percentile). That is an incredibly accurate projection for the payroll jobs performance of the past year -- a projection made before any solid evidence of jobs growth had occurred.

The previous month's report (for January, released in February) had included the benchmark revisions for the March 2003 to March 2004 period. Jobs growth in the revised data was higher than previously reported -- by about 200,000 jobs. The payroll jobs level in January 2005 is now higher -- by 119,000 than it was in January 2001 when President Bush took office.

Last month's report also was pretty much the final word on the benchmark adjustment for jobs at the business cycle turning point: In August 2003 I wrote a widely-circulated paper that cited the historical observation that the payroll jobs numbers typically were biased downward during early periods of a business cycle expansion. (See discussion from last October below.) The report provided the final word on this issue. The upward revision to jobs growth during the March 2003 to March 2004 annnounced by the Bureau of Labor Statistics (BLS) does in fact conform to the historical observation for a cyclical downward bias in an early expansion, the number is small relative to what would have been suggested by prior historical relationships. So, the jury has returned with a verdict of guilty, but in a sense to a lesser charge.

The general point from last year's research still applies. As the discussions and research presented below revealed, we must be careful how we interpret the initial real time payroll jobs estimates from the BLS, especially in early expansion periods. The new data now show that the turning point for payroll jobs was earlier than previously estimated, with the trough now in May 2003 -- and not in August 2003 as the prior estimates had shown. That is a closer match with the turning point for the unemployment rate, a point that was made earlier. The bottom line is that the month-to-month preliminary payroll jobs numbers aren't as reliable around business cycle transitions as we always used to think they were. At times of business cycle transition, the unemployment rate may be a more reliable indicator as its real time turning points and trends are reliable and have been preserved through subsequent revisions. Also, perhaps what we need to look at something like a centered 3-month moving average of the jobs change numbers as the "better" measure, instead of placing so much emphasis on unreliable month-to-month estimates that we see in real time. (A similar approach is used for weekly unemployment insurance claims, using a 4-week average.)

Here's what was said in October 2003: Here's a chart showing the likely projection ranges for payroll jobs growth over the next year. The chart is based on the estimated historical relationship between real GDP growth and payroll jobs, including accounting for productivity trends, and accelerated cyclical productivity growth during early expansion periods. The chart is derived from Monte Carlo simulation using the Blue Chip projection for real GDP growth, and allowing for the error distribution for (1) real GDP forecasts, (2) the relationship between real GDP growth and payroll jobs, and assuming a gradual return to trend productivity growth from the elevated cyclical growth of the past 6 quarters. The chart shows that the central expectation is for payroll jobs gains of around 2.3 million by the end of 2004, with a 70 percent probability that more than 1.1 million jobs will be created. Click here for the estimated GDP-Jobs equation I used.

A document with a discussion of the outlook for jobs growth that uses the chart: "Stronger Economy - And Jobs Growth - Expected Through 2004"

From October 2003: THE JURY IS STILL OUT: On Friday October 3rd, 2003 BLS announced its preliminary estimate that the payroll jobs benchmark revision for March 2003 would result in a downward revision of 145,000, or about -0.1 percent of the total payroll jobs estimate; this estimate does not confirm the view of the above paper, but it doesn't yet reject it either. The jury is still out-all the evidence isn't in yet -- as current months' job growth data (e.g. post-March 2003) will yet be subject to revision from the benchmark revision that will occur next year-and we won't get a preliminary view of that til October 2004 and the final data til Feb. 2005. That is, in fact, what happened in 1992 -- the 1992 benchmark was a small negative revision, and then the 1993 benchmark revision was positive, leading to stronger estimated payroll jobs growth during 1992.

As to the general situation, the unemployment rate appears to have started trending down -- and its movements tend to be preserved through revisions. Some good signs exist-but we will need to see continued declines in the unemployment rate and sustained increases in the payroll jobs numbers (even if they are preliminary, e.g. real time estimates of jobs gains at 150,000 a month might end up being 200,000 a month after benchmarking, but it is highly unlikely 50,000 a month could become 200,000 a month). The outlook for real GDP growth in the 3-1/2 to 4-1/2 percent range over the next year bodes well for the outlook for growth in payroll jobs. How many net jobs will be gained for a given growth path for real GDP will depend on how strong productivity growth will be-whether we continue at the high rates of the cyclical surge of the past year and a half, whether we revert to a post-1995 trend rate, or if we revert to a pre-1995 trend rate. For economic analysts and the media, though, the general cautions expressed in the paper still apply. We've tended to treat the flow of payroll jobs data as a clean, real-time series with a clear signal. But there's a lot more noise in the series than we've typically thought, especially near the turning points of the business cycle-and we need to be careful how much weight we place on any given month's-or set of months'-payroll jobs data.

Return to top