Mixed Signals: Which Is More Relevant - The ISM Manufacturing Or Markit PMI Survey

A notable divergence has emerged between two key economic sentiment indicators: the ISM manufacturing survey and IHS-Markit PMI, which sent opposite signals of manufacturing activity in September, prompting some - such as Bank of America - to wonder which one is a more accurate indicator of the economy's current state.

As BofA's Michelle Meyer writes in a Friday report, "both have pros and cons and in the end we find that the most prudent course of action is to monitor both along with a range of other manufacturing indicators." That said, on balance "the data show a weakening manufacturing sector but not a collapse as the ISM survey implied."

Which indicator is better?

As BofA notes, markets were roiled last week as the much-watched ISM manufacturing survey fell further into contraction territory. The index declined from 49.1 in August to 47.8 in September, a ten year low.

In the ten minutes following the release, equity markets tumbled 0.5% before closing down 1.2% on the day.

Meanwhile, another survey of manufacturing - the IHS-Markit PMI - showed that the manufacturing sector remained in  expansionary territory in September.

Although the two measures often diverge in terms of magnitude, it is rare for the two signals to get crossed (though not in China, where the official PMI survey has traditionally diverged and generally been weaker than the Caixing PMI). In the US, BofA calculates that the two have only sent opposite signals roughly 13% of the time since IHS-Markit's first print in May 2007. This divergence and the difference in the equity market response to the two indicators raise the question: which measure is the better gauge of manufacturing activity?

One or the other?

To answer this, BofA looked at the component indexes that make up both the ISM and IHS-Markit PMI. These indexes measure the month-over-month change of the following: production, new orders, employment, inventories and supplier deliveries. The bank's economists then regressed the component index from each survey on the % mom change in the comparable "hard data".

Fast forwarding to the results, BofA finds the coefficient on each survey index is statistically significant at the 5% level. Reducing the sample, however, results in only the employment and supplier deliveries components having statistically significant coefficients for both ISM and IHS-Markit PMI, suggesting that the 2008/09 recession skews the data. As expected, most of the models have a relatively low fit in the full sample and a nearly nonexistent fit in the reduced sample, which is somewhat problematic. There is one exception; the employment model has a reasonably strong fit over the full sample and a decent fit in the reduced sample. BofA thinks the consistency in the results across samples and measures suggests that it'd be prudent to look at both surveys when assessing the health of the manufacturing sector.

Two, it's the magic number

To test whether or not looking at both measures is better than focusing on one, the bank ran the same exercise as above, only this time it treated the average of the two survey component indexes as its independent variable. Comparing the fit of all three models across the full sample period, we find that this approach yields a slightly better fit for most measures (Chart 3 above). This again, suggests that one shouldn't discount the signal from either.

A manufacturing downturn any way you slice it

No matter what weight one assigns to either of the two surveys, "averaging out between the two surveys implies that the manufacturing sector has weakened but perhaps not to the extent suggested by the ISM measure." In terms of assessing the state of the manufacturing sector, it is also helpful to look at a broader range of indicators. Looking back at how the data have evolved over the past year shows a substantial weakening in both survey-based measures and official statistics on the sector (see table and chart below). This is unsurprising given that the industry has faced a slew of headwinds ranging from broad factors like the trade war and fading fiscal stimulus, to industry specific factors like the Boeing 737 max groundings.

Where's the sector headed? According to BofA, it ultimately boils down to future of the US-China trade war. While it's likely that a couple of industry specific factors-the Boeing groundings and the General Motors United auto workers strike-prove to be transitory, the bank expects the broader trade war drag to persist. Therefore, it concludes, "the sector will likely continue to face an uphill battle."


BofA's dour outlook on the manufacturing sector serves to magnify the focus on other sectors of the economy. As Michelle Meyer notes, she has argued time and again that the consumer remains strong and supportive of continued growth in the economy (even though Morgan Stanley is not so sure). However, even BofA admits that a persistent downturn in the manufacturing sector risks having a spillover effect into other areas of the economy. Indeed, we are now seeing evidence that the trade war has started to affect industries outside of manufacturing as indicated by the September ISM non-manufacturing index, which - both in Europe and the US - disappointed expectations and fell to a 3-year low (see below). We have also seen both total and service employment growth decelerate significantly from last year's torrid pace.

The bottom line concern here is that these trends worsen as the US plans to raise tariffs by 5% on roughly $250bn of Chinese goods imports on Oct 15 and impose 15% tariffs on $160-170bn worth of goods on Dec 15.

As Meyer concludes, "trade talks between China and the US next week loom large. We don't have high expectations." And while Larry Kudlow managed to stage another impress short squeeze, the latest report from Bloomberg confirms that neither China nor the US have high expectations of any favorable announcement this week either...