Correlation of ISM employment indices and changes in non-farm payrolls

3 Jun 2010 by Jim Fickett.

There is a clear correlation between the monthly change in non-farm payrolls, on the one hand, and the levels of the ISM manufacturing employment index and non-manufacturing employment index, on the other. However the month-to-month noise in all three series is so high that the predictive ability of the ISM indices for any one month is quite low.

The ISM released the non-manufacturing index today, and a number of news stories picked up on the fact that the employment sub-index was finally above 50. From the Washington Post:

Fewer people are filing claims for unemployment aid, new jobs are showing up in service industries, and companies are squeezing all they can from lean staffs and may need to hire soon.

Hopes for the job market brightened Thursday

[emphasis added]

So how much can one really tell from the ISM employment sub-indices in manufacturing and services? The following graph shows those two indices and the month-to-month change in non-farm payrolls, from the BLS (all data through April; click for larger image):

While it is clear that there is a strong correlation, there is also a great deal of noise. And some misleading signals, as well. Note, for example, that in the middle of 2008, when the downward trend in jobs was starting to really accelerate, there was a big spike up in the ISM manufacturing employment index.

All in all, it is not worth spending much time trying to read the tea leaves on this one.