Markit explains the divergence between two U.S. PMI surveys in November.
Two barometers of US manufacturing business conditions moved in different directions in November, but if examined in more detail both tell a similar story of modest expansion of manufacturing output so far in the
Markit’s final manufacturing PMI came in slightly higher than its earlier flash PMI, up from 51.0 in October to a six-month high of 52.8 in November. The ISM’s headline PMI meanwhile fell from 51.7 in October to 49.5, its worst reading for nearly three-and-a-half years.
However, the PMIs are composite indicators derived from various survey questions, and although using the same indexes, the two surveys have different weights for each component. While the headline composite indexes from the two surveys did diverge, the discrepancies are smaller when you look at the subindices. Removing some of the volatility in the ISM numbers also helps.
Caution must also be used in interpreting both surveys due to disruptions to business emanating from the storms.
For example, the Markit Output Index has risen from 50.6 in September to 51.4 in October and 53.5 in November. The ISM Output Index has meanwhile risen from 49.5 in September to 52.4 in October and 53.7 in
Both surveys are therefore signalling an acceleration in output growth. Furthermore, over the last three months the ISM and Markit Output Indexes have averaged 51.9 and 51.8 respectively, signalling almost identical modest trend rates of expansion.
Looking at new orders, the ISM index has shown strong volatility, rising to 54.2 in October before falling back to 50.3 in November. Markit’s index has been more stable. Importantly, over the past three months both the ISM and Markit New Orders Indexes have averaged 52.3, so both surveys are pointing to a modest upturn in demand over the past three months.
However, one of the key divergences between the headline PMI readings for the two surveys is due to the differences in the Stocks of Purchases/Inventories and Employment indexes. The Stocks of Purchases Index is weighted higher in the ISM PMI than in the Markit
equivalent index. Both stocks of purchases and employment are also comparably volatile indexes in the ISM survey.
The ISM survey showed that employment fell in November despite the increase in production. Markit’s survey has shown a more consistent message of ongoing weak job creation alongside similarly meagre
output and order book growth.
Markit survey enjoys higher correlation with official data
There are many factors that can explain divergences between the two surveys, and also reasons why Markit’s survey typically provides a smoother trend, with a lower “noise to signal” ratio. Markit’s survey
panel is nearly twice as large as the ISM’s stated panel size, is very closely mapped against the official structure of the economy and uses a different method of seasonal adjustment, calculating the factors every month instead of once per year.
These methodological differences have a clear impact. When the Output Indexes from the two surveys are compared against the three-month change in official production data (a widely used comparison for survey
and official data), the Markit index has a correlation of 94% compared with 87% for the ISM data (this is based in both cases on the data from mid-2007 onwards, when Markit data were first available).
If one therefore focuses on the recent trends of new orders and output, and looks through some of the volatility in the ISM data, the two surveys paint a similar picture of steady, but unspectacular manufacturing expansion so far in the fourth quarter.