The Institute for Supply Management has for many years conducted a survey of purchasing managers in the manufacturing sector, asking questions such as whether output and employment are higher or lower than in the preceding month. From these answers it constructs an index, whose value is below 50 when reports of contraction outnumber those for expansion. I noted with concern when the ISM manufacturing index slipped below 50 in early January, and voiced my relief when it was reported back above 50 on February 1.
But all the attention yesterday was on a separate index ISM prepares for the non-manufacturing sector, whose latest values certainly look scary when you plot them.
This is not a number that most of us had been paying much attention to, until it fell off the cliff yesterday. For one thing, there is not as much of a history with this series to allow us to put the drop in context. The earlier sharp decline in 2003 seems to have been a false alarm, and the other previous big drop arrived at the end of the 2001 recession.
The sectors covered by the ISM nonmanufacturing index include construction, financial services, and real estate, which we knew faced some big problems, though it’s surprising that these would have suddenly shown up in January. It also includes the entertainment industry, which was perhaps affected by the writers’ strike, though if that’s the explanation it’s hard to square with a sudden drop in January. Retail trade is also covered, and perhaps the ISM nonmanufacturing numbers are another harbinger that next week’s retail sales numbers will be disastrous.
For my tastes, Jeffrey Miller offered some very useful insights (as he invariably does). Jeffrey wrote yesterday:
Today’s [stock market] trading was sparked by the ISM services report, something that has not attracted much attention in recent years. The Market has focused more on the traditional ISM manufacturing survey, partly because it has a longer history, and a clear link to GDP. For those who have forgotten that report, released two trading days ago, it suggested GDP growth of 3% as of mid-January, the time of the survey….
We have tested the ISM service series against employment changes and other economic data, and we find it to be pretty good. When comparing it to the ISM manufacturing index we discovered that it did not add much information. We now have a single data point where there is a significant divergence. This would be nice to test, but there are not many other divergences to use.
Our review of today’s news did not find anyone else who highlighted this statistical fact, or wondered about the meaning.
Excellent observation, Jeffrey, and I welcome input from anyone who has a good answer. But in the mean time, I’ll agree with the crowd that this is a very ugly-looking move in that series, whatever it ultimately proves to mean.