Markit Composite PMI: Yet Another Diffusion Index Incorrectly Showing Strength

The Markit USA flash (preliminary) composite PMI shows strength this month, as have other diffusion index reports including ISM and the Fed regional reports. Delivery delays contributed to manufacturing strength. Those delays are a hurricane impact, as are reports of hiring.

The Markit Composite PMI came in at 55.7, ahead of the Econoday consensus estimate of 55.1. Econoday often makes a comment about these diffusion indexes not matching actual production, but not today.

Advance indications are picking up increasing signals of growth including the PMI composite which, at 55.7 in the October flash, easily tops Econoday's high forecast. Manufacturing is contributing strongly to the rise, at 54.5 which is an 8-month high and up strongly from September. Services are also contributing, at 55.9 for also a sizable gain.Strength in the manufacturing sample is centered in new orders and employment. Of special note are unusual delivery delays, which help lift the composite indexes and are the result of lingering disruptions and stretched workloads following Hurricanes Harvey and Irma. Orders and employment are moderating for the service sample but not the 6-month outlook which is improving. Price data show moderating but still strong pressures for inputs but slowing traction for selling prices.Markit Economic's samples have been reporting significantly less strength than other private and regional reports including the ISM where both the manufacturing and non-manufacturing composites are near or over the 60 mark. The uptick for Markit's data offers confirmation perhaps of strength in other reports and does point to a solid close for the 2017 economy. Note that private and regional reports are based on small sample sizes and offer no more than a directional signal for economic change.

Unusual Delays

Unusual delays and unusual hiring are hurricane impacts. The uptick confirms nothing. These diffusion indexes have been wrong for over a year.

Industrial Production and Capacity Utilization

The above chart from the Fed's Industrial Production and Capacity Utilization report.

Regional Report Nonsense

Last week, the Empire State Manufacturing Survey, a diffusion index, came in at a blistering 30.2 yesterday.

Econoday offered this comment, emphasis mine:

"Today's results point to similar strength for Thursday's Philly Fed report which, like this one, has been posting unusually strong results all year. It's important to remember that diffusion indexes offer only rough assessments of activity and in Empire State and Philly Fed are based on relatively small samples where responses are always voluntary. And the rare strength of these samples has yet to be matched by the government's factory data."

Reality Check

Also last week, I noted the Philadelphia Fed Manufacturing Survey Shows Employment Index Highest in 48 Years. Here's a reality check.

Change in Manufacturing Employment

The BLS reported a decline of 1,000 manufacturing jobs in September. We do not have the October numbers yet, but it is safe to say manufacturing will not show the strongest gain in 48 years.

What's going on?

With diffusion indexes, a company hiring one employee counts as much as another firing 300 or even 3,000. Factor in a small sample size and the ability of companies to not waste time responding to surveys and the reports might look like (and do look like) complete nonsense.

Responding to these reports is voluntary. I also wonder if companies with negative hiring or outsourcing are reluctant to admit that for fear of incurring the wrath of Trump.

Regardless, something is seriously wrong with ISM reports and nearly every regional Fed manufacturing report compared to actual production.

Mike "Mish" Shedlock

false