ISM Mistake Denotes Need For Simpler Method

Yesterday the Institute Of Supply Management released the monthly results for the May manufacturing survey. Unfortunately, ISM eventually released the data two more times over the course of the day to correct a mistake in their data.  To wit:

“We apologize for this error. We have recalculated and confirmed that the actual index indicates that the economy is accelerating,” said Bradley J. Holcomb, CPSM, CPSD, chair of the Institute for Supply Management® (ISM®) Manufacturing Business Survey Committee. 'Our research team is analyzing our internal processes to ensure that this doesn’t happen again,' he added.

The May PMI® registered 55.4 percent, an increase of 0.5 percentage point from April’s reading of 54.9 percent, indicating expansion in manufacturing for the 12th consecutive month. 

The error resulted when the software incorrectly used the seasonal adjustment factor from the previous month."

Of course, the immediate question that comes to mind is why only five of the components were affected by the seasonal adjustment calculation?  Why not all of them?

The seasonal adjustment calculation only to serves to transform otherwise very volatile data like this:

Into this:

(Notice that on a non-seasonally adjusted basis manufacturing activity did indeed slow in May which corresponded to the first report released by ISM.  However, two revisions later with the correct seasonal adjustments and the index was improved to account for seasonal tendencies.)

The "assumptions" made to smooth the data have led to much debate in the financial community. The "seasonal adjustments" are derived from the historical tendencies of manufacturing activity over time. For example, we know that there are surges in employment due to holiday shopping patterns that led to temporary hires.  Therefore, the employment data is adjusted to account for this seasonal tendency. However, there is a reasonable argument that the current economic backdrop may have deviated from these historical norms rendering the seasonal adjustment process less reliable.

It is from this basis that I have often discussed using a more simplistic 12-month moving average to smooth the data.  In this manner, we can use the more volatile non-seasonally adjusted data and remove any potential biases, manipulations or "mistakes"from the analysis.  The chart below shows the seasonally adjusted ISM index as compared to the 12-month moving average of the non-seasonally adjusted data. The reason I like this analysis is because we are simply dealing with what "is."

As you can see, there is a 100% fit to the data with only a slight lag due to the averaging of the monthly data as would be expected. The 12-month average also confirms the recent uptick in the manufacturing survey following the revision by ISM.

The next chart, which is quite massive, is the 12-month average of the subcomponents of the ISM Manufacturing Composite Index. 

There has clearly been a pickup in activity in the index since the middle of 2013 which corresponds with expectations of an increase in economic activity.  However, it is also important to note that activity in most cases remains at levels lower than seen in 2010 after the economy surged from the recessionary lows.

We have witnessed surges in economic activity over the last several years that have been unsustainable. Each surge in activity, driven by an inventory restocking cycle, fades as aggregate end demand fails to emerge. Subsequently, economic activity slows to the point where pent-up demand leads to the next restocking cycle.

With government regulations, taxes and poor sales remaining the at the top of the list of concerns by "small businesses" it is not surprising that they continue to work on an "as needed" basis.  The lack of confidence in their outlooks, as witnessed in the latest NFIB Small Business Survey, continues to suggest that employment and fixed investment will continue to run at levels associated with population growth rather than expectations of surging demand due to a vibrant economic recovery.

The "good news" is that the economy does seem to be getting a bit of a lift due to the extremely cold winter weather in the first quarter which slowed activity.  The "bad news"is that this "recovery," like the last few, may again be elusive in nature.

This puts the Federal Reserve in a quandary after admitting that they were a bit to"exuberant" about the economy at the first of the year.  There are several Fed officials that are now calling for hikes in the Fed funds rate even before the current liquidity program is wound down as they see signs of sustained economic recovery.  The tightening monetary policy acts as a drag on economic growth and acting too soon could end the "party" before it even really gets started.

Is using a 12-month moving average of the non-seasonally adjusted data a better method than currently employed by the ISM? There are certainly advantages and detractions from both methods. However, using an average to smooth the data certainly provides greater transparency and credibility to the reported data while eliminating the possibility of "mathematical malfunctions."