This last week has seen wild swings in the markets and certainly the ISM reports did not help the mood of pessimism. Starting Monday, the markets got a significant jump up of almost 140 points on the DJIA, but the mood changed as soon as the July 2011 Manufacturing ISM Report (PMI) was released at 10am. Although Wednesday was up slightly for the day, the July 2011 Non-Manufacturing ISM (NMI) also reversed the up direction at 10am.
The markets were reacting to the headline numbers as both were below the consensus marks of 54.3% versus the actual of 50.9% for PMI and 53% versus 52.7% for NMI. The market also reacted more dramatically on the PMI, presumably because it also missed the consensus range of 52 to 55.4%, as reported by Econoday.
Before the two reports came out, leading economics bloggers became gloomy on the economic outlook, and the report said “optimism is out; pessimism is in.” Professional economists reacted by stating the manufacturing report was very weak and disappointing. Plenty of economics bloggers also wrote about the reports like the following linked list.
- The Capital Spectator: US Manufacturing Activity Slows Sharply In July
- Manufacturing Slows In July, Stoking Slowdown Fears
- Mish's Global Economic Trend Analysis: Gap-and-Crap it Was; ISM Plunges to 50.9, Lowest Level in 2 Years, New Orders Contract; Key Thoughts; Reaction in Gold
- Calculated Risk: ISM Manufacturing index declines in July
- Mish's Global Economic Trend Analysis: ISM says "Business Conditions Flattening Out"; Why Services Number Worse Than It Looks; Unsustainable Conditions
- Calculated Risk: ISM Non-Manufacturing Index indicates slower expansion in July
- Manufacturing Weakens, But It's Not a Death Knell - Seeking Alpha/Calafia Beach Pundit
- A Bright Spot in the ISM Report? - Seeking Alpha/Cullen Roche
Interestingly enough, the last two articles hint that the time to abandon ship has not yet arrived. Calafia Beach Pundit makes an important point that just because the manufacturing sector is slowing in growth, the correlation between GDP growth and PMI is not at the break-even point of 50 but is at 47%. The last manufacturing report stated:
The past relationship between the PMI and the overall economy indicates that the average PMI for January through July (57.6 percent) corresponds to a 5.3 percent increase in real gross domestic product (GDP). In addition, if the PMI for July (50.9 percent) is annualized, it corresponds to a 2.9 percent increase in real GDP annually.
But when looking at the chart by CBP (below), it looks like this business cycle has diverged from the expected outputs. The biggest spikes in PMI have not resulted in the same magnitude of spikes in GDP; more important is the divergence between PMI and GDP during this business cycle. Maybe it was a weaker non-manufacturing sector that did not contribute the same magnitude as other business cycles. Looking at the last major spike in the PMI index at around 2004, it still showed a solid growth in GDP with even the lows well above the 2% mark. This time the first two-quarter GDP growth has just dropped off during the spike to just above no growth. Below is a more detailed (esoteric) discussion about correlations between GDP and the PMI indexes.
[Click all to enlarge]
Roche tells us the bright spot in the ISM report is that inflation fears by hyperinflationistas (my word) should go away now. But the other side of the story is that this just means lack of aggregate demand in the economy. Right now we need an economy more like China's than Japan's, as Roche contrasted, that is higher inflation rates and higher growth rates. No way we can achieve that high growth due to the size of the economy and so much structural rigidity built into the system.
In addition to the fact that the manufacturing price index has dropped a staggering 26.5% over the last four months, both indexes dropped below the 60 mark with a drop of 9 to 59% and 4.3 to 56.6% for manufacturing and non-manufacturing respectively. It seems reasonable that moderate levels of inflation are good for the economy and thus moderately rising prices for the sectors are fine. Even the Fed targets moderate inflation levels of around 2%. Below are graphs of the manufacturing and non-manufacturing commodities up in price that include multi-month commodities up in price and total number of commodities up in price. Manufacturing shows the continued reduction in both categories, but there was a slight increase of total commodities up for non-manufacturing. Still nothing to be worried about for now.
Overall there really is not much of a silver lining in either report. The biggest positive in either report was the Business Activity index in the non-manufacturing report with a gain of 2.7 to 56.1%. Even that may not be sustainable as new orders and employment both dropped 1.9 and 1.6 respectively. Imports increased in both reports, but big deal. As discussed before, it does not hurt our economy overall but this comes with shrinkage in the non-manufacturing New Exports index by 8 to 49%.
The all-important employment indexes were significantly lower last month as manufacturing dropped 6.4 to 53.5% and non-manufacturing eased lower by 1.6 to 52.5%. Below is the updated employment index for non-manufacturing along with its trend line since December 2009. Three other times it has eased below the trend line as drawn, but this looks like a more significant change than previous below trend changes. The whole trend has been tenuous at best. This does not bode well for rapid reduction in the unemployment rate in the short term.
Last month I explored how the ISM headline indexes (PMI, NMI) are calculated. The PMI used five sub-indexes of equal weight, and the NMI used four, which are seasonally adjusted in both reports. According to the ISM website (Reports On Business: Overview), there is a close correlation between the PMI and growth of the economy or specifically GDP.
An update of research originally done by Theodore S. Torda, the late economist for the DOC, shows a close parallel between growth in real Gross Domestic Product (GDP) and the PMI. The index can explain about 60 percent of the annual variation in GDP, with a margin of error that averaged ± .48 percent during the last ten years. George McKittrick, an economist at the DOC, said "Not only does the PMI track well with the overall economy, but the indication provided by ISM data about how widespread changes are, complements analogous government series that show size and direction of change."
I did not get as significant a correlation when doing linear regression analysis on GDP and the PMI, but this may be due to different data sets or adjustments made to the data sets. The Ordinary Least Squares shows that 52 2/3 of the variance in GDP is attributable to changes in the PMI index. The sign of the slope is of the correct sign (positively correlated) and the T-Ratio probability (p-value) is 0.000. Since this OLS produces a failed test for functional form, I tried the regression on log of the PMI. This resulted in an adjusted R^2 rising slightly to 0.548 with a better diagnostic test for functional form. Although there is a high degree of correlation between the PMI and NMI, it still made sense to combine the two in a regression. The NMI variable was significant at the 5% level but failed at the 1% while the PMI retained the 0.000 p-value. This increased the adjusted R^2 to 0.609. Lastly, taking the logs of PMI and NMI gave the same significant levels as the last but increased the adjusted R^2 to 0.631. Below shows the scatter relationship between GDP growth and the PMI index.
The question then should be, how do the other economic indicators compare with these results? Below is a table of some of the results including what was already discussed.
As observed from the table above, none of the other economic indexes are nearly as correlated with GDP. Even the index of leading economic indicators (LEI) resulted in no significant correlation. Since this is a "leading" indicator, using lags resulted in some significance for two to six month lags but still not nearly as strong as correlation with PMI. When I regressed GDP on the independent variables of PMI and LEI, it resulted in higher adjusted R^2 but the slope of the LEI becomes negative. This indicates that as LEI increases it would signify slower GDP growth. Obviously, the opposite sign as expected. Even Housing Starts did not have any significance with as many as 24 lags out. The slope coefficient was of the correct sign (positive) for current index and up to four period lags (months) but none are significant at even the 0.1 level.
Back in August 2010, I asked if the ISM is an overrated index. At least with respect to the GDP, it certainly is not. Some further questions for investors is how GDP growth or lack of growth affects the equity markets or at the sector levels of the economy? Can investors use the ISM reports as an investing strategy, and how effective are such models? For further discussions and links on these questions, look at the section titled Using the ISM Cycle as an Investment Guide in the May ISM reports post.