, , , , ,

I was going to write about Greece this morning, but a closely watched gauge of American manufacturing has just turned in such a totally weird reading (again), and is prompting such excitement (again) that the cold water bucket needs to come out now.

The gauge is the Institute for Supply Management’s (ISM) monthly Report on (manufacturing) Business, which looks at many different measures of manufacturing performance. I’ve written previously that both its overall headline reading and its sub-gauge of output have no relationship to the Federal Reserve’s industrial production index – which is much broader and otherwise more methodologically sound. So I’m not at all surprised that today’s ISM production figures show that this gap is still intact, along with the ISM’s pattern of reporting a much rosier picture than the Fed.

The new ISM production figure (for June) was 54.0, which is solidly in “solid” territory. (All ISM data over the 50 level indicate expansion.) We won’t get the Fed’s June figure for about two weeks (one reason economists and investors tend to pay more attention to the ISM). But in May, according to the Fed, inflation-adjusted manufacturing output shrank from April’s levels by 0.21 percent. By contrast, the ISM May production reading of 54.5 pointed to even “solid-er” growth.

Nor was May an aberration. Here are the Fed’s monthly real manufacturing output change figures for this year, with the ISM’s production number for that month in parentheses:

January -0.66 percent (56.5)

February -0.18 percent (53.7)

March +0.31 percent (53.8)

April +0.13 percent (56.0)

Sometimes, ISM defenders argue that it does a better job tracking changes in the direction of manufacturing’s performance rather than revealing the sector’s status at any given point in time. But these numbers, as well as earlier historical data I’ve analyzed, show nothing of the kind. And my explanation remains intact: The ISM’s methodology suffers from “survivorship bias” – i.e., it (tries to) measure the performance of the existing manufacturing base in any given month without accounting for how the size of the base may have changed over time.  The Fed tries to measure actual production, adjusted for price.

Therefore, the lessons are clear for anyone who’s following manufacturing. If you want something close to the real deal, look to the Fed’s industrial production data. If a shot of hopium is more your style, nothing beats the ISM.