Daily Speculations The Web Site of Victor Niederhoffer and Laurel Kenner


The Chairman
Victor Niederhoffer


About Victor Niederhoffer

Write to us at:(address is not clickable)

Ten Variations on Leads and Lags in Industry Groups, by Victor Niederhoffer

In a previous memo, I discussed the Austrian Von Mises and Hayek theory of business cycles which in a nutshell predicts that there will be consistent leading relations whereby capital goods manufacturers turn at the beginning and end of business cycles because they are most sensitive to artificial interest rate changes induced by the monetary authorities. "Interest rate sensitivity increases with temporal distance of the investment subaggregate from final consumption".  The problem with this theory is that it doesn’t take account of rational expectations, i.e., the ability of people to make the correct decisions and learn from past mistakes.

To put some meat on the bones of the more general question of whether there are consistent leads and lags, I turned to the Russell 2000, consisting of 12 sectors. I studied the period from 1995 year end to April 2005 month end, a nice 10-year period, earlier than which I would speculate that all relevance is lost because of learning and changing structures. The first question is which industries were best and worst. And here's a surprise:

Performance of Russell 2000 sectors, 12-95 to 4-30-05:

Sector9.5 yr % Return
Other energy+209.8
Integrated Oil+181.0
Producer Durables+124.4
Financial Services+120.6
Consumer Discretionary+118.6
Auto and Transport+93.8
Consumer Staples+93.5
Health Care+72.3
Materials and Processing+57.8

Who would have thought? Technology was worst and Auto and Transport was up 94%. Energy at the top is an interesting reflection of the boom in last year and a half in oil, but I wonder if the cycles are about to change.

Has there been any consistency in these rankings over time? The Professor's star student Chris Hammond computed the rank correlations as his first summer project here. And I hasten to report that the average one-year correlation is -0.02, a completely random result. And yet the two-year correlations average -0.12, a figure bordering on 4 to 1 against if there were no negative consistency over time. In other words, if an industry is good in 2002, it is likely to be bad in 2004, and if an industry is good in 2003, it's likely to be bad in 2005.

The ranking of the industry returns in 2003 was: Integrated Oil +128, Technology  +63, Producer Durables +59, Health Care +58, Materials and Processing +43, Consumer Discretionary +43, Auto and Transport +41, Other Energy +39, Other +37, Financial Services +35, Utilities +34, Consumer Staples +24. But before you go out to buy the Consumer Staples and short Integrated Oil, note for example the rank correlation between 2001 and 2003 performance of 0.56 and 2002 and 2004 performance of 0.51.

Some correlations of industry performance two years apart:

1996 and 1998   -0.65
1997 and 1999   -0.71
1998 and 2000   -0.65                 
1999 and 2001   +0.31
2000 and 2002   -0.25
2001 and 2003   +0.56
2002 and 2004   +0.51

Industry rankings YTD 2005: Other Energy +3, Utilities 0, Integrated Oil -4, Consumer Staples -5, Consumer Discretionary -9, Other -9, Materials and Processing  -10, Financial Services -12, Health Care -13, Producer Durables -13, Auto and Transport -18, Technology -20.

Variation Seven of Industry Lags:

It requires no co-integration or three-stage least squares to reject the Austrian business cycle theory in the last 10 years. Producer Durables was up each year except 1997 and 2002: 1996 +14, 1997 -12, 1998 +36, 1999 +2, 2000 +2, 2001 +5, 2002 -26, 2003 +59, 2004 +16, YTD 2005 -13.

There seems to be a tendency for producer goods to bounce back the next year. But its decline did not presage a decline in other industries in subsequent years; 1998 and 2003 were great years for all the other sectors.

An Erudite Spec replies:

Interesting question, but I don't recall Austrian economists' discussing stock performance leads and lags by industry. I've taken a look at Roger Garrison's pictoral illustration of the time structure of production, and I think he would say to measure leads and lags in terms of profit or revenue growth rather than stock performance. The Austrians said surprisingly little about stock markets, and I can't recall their saying much nice about them. Most newsletter-writers/gurus/Austrians seem to be doomsayers who say the stock market is dominated by blind fools. However, one paper that at least tangentially addresses the industry lead/lag question is Cross Industry Momentum by Menzly and Ozbas.

Kim Zussman adds:

Trivial observations in the face of great ones:

The momentum literature talks about industry momentum*, and some papers ascribe significant continuation (for 3-12 months, depending on look-back period) to industry. Also industry relationship has some explanatory power. Longer periods exhibit mean reversion; i.e., bottom performers in look-backs of longer than 1-2 year outperform. All of this is older than your study but seems consistent none the less.

One might capture much of this effect, as per your hint, with sector ETFs. Or alternately long bottom performers in recent 3 years which could be called half of DeBondt and Thaler effect.

*This year momentum doing poorly.

For more of Victor Niederhoffer's writings, click here>>>