Is there anything good to say about today’s report from the Bureau of Labor Statistics that the U.S. unemployment rate jumped up to 6.1% while seasonally adjusted nonfarm payrolls declined by another 84,000 jobs? Well, here’s one thing. It gives us some real clarity as to just where the economy stands.

Civilian unemployment rate,
from FRED, with NBER recessions as shaded regions.

Sure looks like a recession when you inspect a graph the unemployment rate, doesn’t it? And it also looks like a recession from the perspective of a model of unemployment dynamics that I published in 2005. If you use that model to analyze the latest unemployment numbers, you’d calculate the current probability of being in a recession at 95%.

Probability that the economy is in either a mild or severe recession at indicated date, as inferred on the basis of the full sample of revised data on the unemployment rate available as of September 2008 calculated using the model in Hamilton (2005).

As I noted when I discussed that model last month, one reason that the above graph seems to be able to identify recessions so clearly is that it uses the full sample of data available today to infer what was the situation at each historical date. If instead you try to base a call only on the data available at the time, the inference is much choppier. Even so, a 95% probability is not likely to be a miss.

Probability that the economy is in either a mild or severe recession at indicated date, as inferred on the basis of data (as currently revised) on the unemployment rate through the indicated date calculated using the model in Hamilton (2005).

That model distinguishes between a mild recession and a severe recession, with the graphs above combining the two. In fact, the August unemployment report leads to a 14% probability that we just entered the “severe contraction” phase. The last time we had a one-month filter probability of that regime higher than that was October of 1982.

Probability that the economy is in a severe recession at indicated date, as inferred on the basis of data (as currently revised) on the unemployment rate through the indicated date calculated using the model in Hamilton (2005).

Last month I also discussed some thresholds for recognizing a recession recently proposed by
UCLA Professor Ed Leamer. Leamer says it usually means recession if the unemployment rate has jumped up by 0.8% within the last six months. When Leamer proposed that criterion two months ago, the 6-month increase of 0.5% seemed to leave us well short of the threshold. Today’s numbers imply a 6-month increase of 1.3%, shooting past it pretty definitively.

The 6-month change in civilian unemployment rate,
from FRED, with NBER recessions as shaded regions and dashed line at +0.8 threshold.

Leamer also said it would be a recession if the 6-month change in the measure of civilian employment based on the BLS household survey fell by more than 0.4%. Today’s number of -0.354% would technically fall short of that, if you’re determined to split hairs more finely than the allowable pixels in the graph below.

100 times the 6-month change in natural log of civilian employment,
from FRED, with NBER recessions as shaded regions and dashed line at -0.4 threshold.

Finally, Leamer said he’d call it a recession if the 6-month change in nonfarm payrolls fell by over 0.5%. Today’s NFP report, while disappointing, still leaves us at only -0.3%. Whew! That was a close call, no?

100 times the 6-month change in natural log of seasonally adjusted nonfarm payroll employment, from FRED, with NBER recessions as shaded regions and dashed line at -0.5% threshold.

No. I think we’re better off looking at the 12-month rather than 6-month change in nonfarm payrolls. And the 12-month change, now at -0.2%, is clearly a recession-type number:

100 times the 12-month change in natural log of seasonally adjusted nonfarm payroll employment, from FRED,
with NBER recessions as shaded regions and dashed line at 0.0 threshold.

And, by the way, calling it a recession when the 12-month change in nonfarm payrolls becomes negative appears to be pretty robust even given the revisions we know are likely to come in these data later.

100 times the 12-month change in natural log of seasonally adjusted nonfarm payroll employment as it would actually have been reported at any given date, from ALFRED,
with NBER recessions as shaded regions and dashed line at 0.0 threshold.

So I don’t see any way to slice today’s report other than to say, at least as far as the employment numbers are concerned, the U.S. is now definitely in a recession.

And if you won’t take my word for it, you can hear pretty much the same thing from
Brad DeLong,
Justin Fox,
William Polley,
Paul Krugman,
and the various economists quoted by WSJ Real Time.

Technorati Tags: employment,
recession probability,
recession indicator index,
macroeconomics,
economics

Last 5 posts by James Hamilton

• Receiver operating characteristics curve - November 18th, 2009
• Commodity inflation - November 15th, 2009
• Will rising oil prices derail the recovery? - November 10th, 2009
• Consequences of the Lehman failure - November 7th, 2009
• Current economic conditions - November 4th, 2009