There's one aspect of today's ISM report that probably won't get the attention it deserves - the surprisingly close correlation between the ISM index and asset returns.
Consider the difference between annual returns on UK equities and 10 year gilts. Since 1994 this has had a correlation coefficient with the ISM index of 0.59. A regression of relative returns upon the ISM index gives an R-squared of 34.5 per cent. That's quite hefty , considering the volatility. The regression implies that an ISM index below 51.7 (today's reading was 51.4) is associated with gilts out-performing equities.
Results for US equity and bond returns are slightly weaker - an R-squared of 29.5 per cent and correlation of 0.55. And ISM readings below 49.5 are associated with Treasuries out-performing equities.
There are, I think, two messages here.
1. We shouldn't be too surprised by the recent fall in Treasury and gilt yields. It's what you'd expect, given the weakness in US manufacturing; relative equity-bond returns in both countries have been within one-third of a standard error of the ISM regression's prediction.
2. Asset returns are surprisingly cyclical; a strong manufacturing sector is good for equities, a weak one bad. This suggests investors' appetite for risk is closely correlated with manufacturing conditions. The same things that cause US manufacturers to feel good also cause investors to buy shares rather than bonds.
The interesting thing here is that this relationship is a relatively new one; before the 1990s, the link between the ISM and relative returns was weak. As manufacturing has become less important for the US economy, it's become more important for the markets.
Fascinating, congrats on a great blog.
Posted by: edward | June 03, 2005 at 09:34 AM