London, 25 August 2010

U.S. crude oil futures are again sinking in tandem with U.S. equity markets -- sparking another round in the now familiar debate about whether crude prices are reacting to poor fundamentals (rising inventories and disappointing demand growth) or reacting willy-nilly to negative sentiment about the macro picture.

The most realistic explanation combines elements of both. Crude oil prices could rise strongly in 2009 and early 2010 despite plentiful inventories as long as investors could look through the "temporary" surplus to focus on a medium term picture of strong demand growth which would drawn down stocks and lead to a tight or at least balanced market.

The anticipatory element of "expected fundamentals" trumped the actual supply-demand-inventory balance. But once that optimism began to fade, the reality of high stocks and moderate demand growth began to weigh down on prices. The crumpling of oil prices and equities since the start of Q2 both reflected and propagated those fading hopes.

In recent months, crude oil prices have been tied more closely than ever before to daily changes in U.S. equities. The question is whether the phenomenon is temporary or marks a more enduring shift in pricing behaviour. Is oil becoming more "macro" and more "financial", or is this just a passing phase before normal pricing patterns reassert themselves? If it is a passing phase, how long will it last? And what level of correlation will prevail when the market reverts to "normal"?

HISTORIC CORRELATIONS

In a thoughtful report published this month, Barclays Capital argues there is "scant evidence" commodities have been "financialised" or that there has been a "structural change in correlations with other assets" (Barclays Commodity Investor, August 2010).

Barclays examines 3-month, 6-month and 12-month correlations between the Standard and Poor's Goldman Sachs Commodity Index and the S&P500 U.S. equity index since 1970.

Correlations have always been unstable, and have become more so over time, but Barclays finds the series is mean reverting with a constant mean of just -0.03. Periods of strong positive and strong negative correlation have become more pronounced since 2000, but the series is still mean-reverting.

The authors admit the recent period of positive correlation between commodities and equities is "unusual in its strength and duration, but then so was the crisis that preceded it". They conclude "it seems premature to draw conclusions about permanent changes in asset class behaviour based on the extreme events of the past two years".

But for at least one component of the GSCI, crude oil, it is hard to deny there has been a profound change in pricing behaviour since the fall of Lehman Brothers in the autumn of 2008. The attached chart shows the 30-day moving average correlation between front-month oil prices and the S&P 500 index since 1990.

The structural break from late 2008 onwards (through both the downswing and subsequent recovery) is unmistakable; correlation coefficients from late 2008 onwards do not even appear to belong to the same series as the earlier data.

Historically, correlation coefficients cycled between +/- 0.4 and exceptionally +/- 0.6. But since late 2008, the coefficient has never been lower than +0.2, almost always above +0.4, and much of the time above +0.6. Strong positive correlation has become even more pronounced and consistent in 2010. Even as the immediate crisis fades, the correlation remains strong and consistent, and shows no sign of weakening.

STRUCTURAL BREAK -- OR NOT?

Is it premature to conclude there has been structural shift in oil pricing? In one sense, yes. The authors point out that the deepest macro crisis in 80 years can be expected to produce unusual and persistent shifts in asset correlations, but that does not mean they will prove permanent. Only if they persist well into the recovery would such a conclusion be warranted.

But there is more to this shift than just a macro crisis. In fact increasing correlations pre-date the Lehman Moment. Correlations between a host of other commodities (including cocoa, coffee, industrial metals and livestock) and U.S. equity markets have been rising since 2005. In most cases, the correlation is much lower than between oil and equities, but in each one the trend is clear. Heightened correlations between oil and equities, and between different commodity markets themselves, are consistent with a general increase in the degree of correlation across the equity markets and other asset classes. The dispersion in equity returns has fallen to its lowest level since 1987, according to recent research reports, creating problems for statistical arbitrage programmes and stock pickers alike. Oil seems to have become firmly part of the riskon/risk-off trade. Clearly many of the factors driving oil prices (such as forecast economic growth, interest rates, liquidity and appetite for risk) also drive pricing of other assets. So some degree of correlation should be expected.

The more oil prices become "forward-looking" (and therefore assets rather than raw materials) rather than based on the immediate supply-demand-inventory balance, the more correlated they are likely to become. The same expectations which shape oil prices shape other asset classes.

It is also likely oil prices will become more integrated with other asset prices as the commodity markets become mainstream. Increasingly, the participants in oil markets are the same as those in other asset classes such as equities. Shared participation helps tie oil prices into other asset markets.

The same integration is already apparent and accepted across geographies with the increasing integration of equity markets in the advanced industrial economies and emerging markets.

There is no reason why oil markets should become any less integrated as they become more mainstream. If there is a structural break in the way oil prices behave, it will only be proven in retrospect. But I would argue the recent "excursion" away from normal correlation patterns has already been big enough and lasted long enough that it is not safe to reject the hypothesis there has been no break in the series. Investors and traders should at least keep an open mind and recognise that pricing behaviour may have changed.

Inevitably, the current exceptionally high level of correlation will not last forever. At some point, a war in the Middle East, a hurricane, a growth spurt in China, or some other factor will cause oil and energy markets to diverge, and the correlation coefficient will fall. Oil prices are a mixture of common factors with equities and idiosyncratic ones specific to the commodity itself.

But it is not unreasonable to conjecture that there has been a structural shift and the constant mean correlation between oil and equities will be higher than -0.03 in the years ahead.

It seems unlikely oil prices will again de-link almost completely from other asset classes (on average). Commmon factors will be more important than in the past. How much more important remains to be seen.

Ends --

By John Kemp, Reuters market analyst - for Commodities Now.

The views expressed are his own.