While the markets had bigger worries than the little ISM reports, both headline numbers of the reports were above consensus expectations. The PMI for manufacturing was three tenths lower than last month's number at 50.6%, but well above the consensus of 48.5, which was expecting a drop of almost 2 1/2 points. The NMI for non-manufacturing was nicely up 0.6 to 53.3% while the consensus was expecting a drop of 1.7 to 51%. Both were neatly in the consensus range provided by Econoday with the PMI range of 47 to 51.9% and the NMI range of 49.7 to 56.6%. Looking at the ranges of the consensus, the NMI range as usual is broader. This may reflect that the economists find it harder predicting the non-manufacturing sectors. I do question the one or more economists that thought the NMI would jump nearly 4 points last month.
Overall, the headline numbers were better than expected, but the expectations were based on gloomy forecasts. The US was definitely not alone in the manufacturing sectors contracting as World-Wide Factory Activity showed only four countries having higher index numbers in August and 18 countries having negative change month over month.
Trends Are Our Friends ... Or Not
Employment is one of the sub-indexes that the Macro View of the Markets looks at for trends. These are important issues when considering that this weak recovery could be called a "jobless recovery." The graph below is from the Federal Reserve (FRED - Economic Data) which shows two trends. The first starts at February 2009 and peaks at February 2011 with a strong upward trend. The second trend is a downward trend since this February. Whether the second downward pressure persists is the question, and the latest drop of 1.7 points to 51.8% is a negative indicator against continued employment growth in manufacturing.
The chart below shows the non-manufacturing employment index with a trend line since December 2009. The slope of the trend line has been declining since at least March 2011. Even before the trend line drops to zero, the downward trend is likely to start. That is the question we are facing now with the last two months dropping below trend. If the short term trend continues, this would mean contraction in the non-manufacturing sectors and fewer prospects of decent job growth along with it.
The following two graphs shows the number of commodities with multi-month price increases and the total number of commodities increasing in price for both manufacturing and non-manufacturing sectors. All four series peaked in April or May of this year, creating an upward trend since October 2010 and then a general downward trend since its peak. No trend can continue past the zero-bound limit as in this case. It is important to look at trends, since the index numbers are not completely independent events. One month's numbers are likely to follow closely to the last month's numbers, plus a possible trend factor. The factors and forces causing the respondents attitudes and thus responses in the ISM reports are likely to carry over from one month to another. Even the business cycle theory would have to consider a stochastic process for determining the next stage in the cycle.
The following two graphs, from Fred Economic Data, shows a recent downward trend for manufacturing with the index dropping 3.5 points to 55.5%. But non-manufacturing has reversed its recent down trend and jumped up 7.6 points to 64.2%. Not that it is a perfect correlation, but input prices have shown a positive correlation with the growth of industries. In this month, the NMI increased and so did the price index for non-manufacturing, while the PMI declined along with the price index in manufacturing.
That correlation was only based on one month observations, but I think it is more broadly applicable than that. When the US economy heats up and helps drive worldwide expansion, commodity prices will start rising dramatically. This leads to the economy cooling as a result of higher commodity prices. The weak economy then drives down demand for commodities and then the cycle begins again. Another way of describing this phenomenon is at Our Oil-Constrained Future.
If this model is accurate—and if the ceiling on global oil production really is around 90 mbd and can be expanded only slowly—it means that every time the global economy starts to reach even moderate growth rates, demand for oil will quickly bump up against supply constraints, prices will spike, and we'll be thrown back into recession. Rinse and repeat.
Conclusion and Trends Overall, the ISM reports were better than expected, but fell short of dispelling rumors of a second dip recession. Hopefully, the constraints on economic growth caused by slingshot effects of growth and commodity prices will be solved someday, maybe by increased production in Libya or Iraq. But these constraints will likely persist until the structural rigidity problems of the US are solved.
This post talked about trends in general terms and avoided too much statistical jargon, and we could have explored more about knowing whether a trend has structurally changed or if it was just a slight deviance from the norm or just an anomaly in the data set. These general discussions should be enough to understand that the graph below does not show a trend line, but simply the average over the extreme
long-term data points. The first reaction from a technical perspective is that the scale should be log, at least on the vertical axis. The numbers should have been adjusted by something like the GDP deflator that adjusts for the value of the dollar over time, especially over long periods of time. Even with these adjustments it might not help since the underlying data appears to be nominal stock prices and not total market cap. That is, what is measured is not value but the arbitrary amounts of money traded for a share. A company can control outstanding shares and thus the price of their share through dilution or stock splits or even reverse stock splits. Lastly, a trend line should go through as many points as possible and be close to the data points. This "trend line" only has one point in common with the data set, and the origin point is immaterial.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.