Commodities are usually advertised as the ‘orthogonal asset class’ by the financial industry,1 based on the findings of several academic studies. Among the first, Gorton and Rouwenhorst (2005) show that commodity future contracts have the same average returns as equities along with a negative correlation, but present less volatile returns. In other words, commodities provide excellent diversification benefits because commodity returns grow when returns of bonds and equity decrease. Commodity futures are even to earn above average returns when equity earn below average returns. Several papers argue that the correlations between equity and commodity tend to fall in turbulent periods, an asymmetric pattern attributed to investors' flight-to-quality strategy (Chong and Miffre 2005).
These patterns have attracted a large number of financial investors seeking to diversify their portfolio (see Bichetti and Maystre 2012). The number of futures and option contracts outstanding on commodity exchanges has increased fivefold between 2003 and 2012 and investors with a motive of physical hedging represent now less than 30% of positions, according to the Commodity Futures Trading Commission.
Curbing commodity speculation
As a direct result, the Food and Agriculture Organization prompted governments to curb commodity speculation. Since then, reforms to move over-the-counter commodity derivatives to exchanges and improve data reporting have been adopted in the US and in Europe. While these reforms are clearly a step towards less opaque markets, they do not address issues related to cross-market linkages between traditional assets and commodities. One explanation may be, as shown by Evenett and Jenny (2012), that there is no clear empirical evidence of the impact of financial inflows and speculation on commodity prices.
In recent research (Delatte and Lopez 2013), we argue that many of the previous empirical findings reflect the misspecification of the models used rather than evidence from the data. As an alternative, we adopt a more flexible approach to model the dependence structure between commodity and stock markets from January 1990 to February 2012.
A novel approach in the commodity studies
We assess the evolution of synchrony between the traditional asset and commodity markets. Our approach to measure this co-movement should be able to answer two empirical questions:
- First, which statistical tool should be used to measure the co-movement between two asset classes?
Investors are usually more concerned by downside risks and their potential losses than the upside. Technically, it means that individual returns are not normally distributed which disqualifies the standard correlation as an appropriate measure of synchrony between asset returns.
- Second, is the relation between these returns stable over time?
For example, when markets are bearish, i.e. the returns are negative, is the co-movement of commodity and equity the same as in bullish markets? Most of the existing studies have answered the questions by imposing restrictive conditions on the joint distribution of the returns, i.e. they have imposed the distribution of the co-movement process. We argue that such an assumption lead to biased results.
In our paper, we employ Patton’s (2006) time-varying copula approach. This measure of financial markets co-movements presents several advantages and addresses some of the points raised above.2 The key point is that it allows us to be agnostic when identifying the dependence structure existing between the returns of equity and commodity futures for the past 20 years. That is, we select the dependence structure that describes best the co-movement of the returns, among the three following cases:
- The co-movement can either be symmetric and occur most of the time.
- The co-movement can be symmetric or asymmetric and occur mostly during extreme events.
- Or it can occur mostly during extreme and negative events.
The strength of the relationship can also vary during the period for each case. To our knowledge, no academic work uses time varying copula to model the co-movement between commodities and traditional assets.
As half of the exposure to commodity price movements is based on investment in commodity indices,3 we first investigate the dependence between the total returns of the two most popular commodity indices and their sub-indices on agriculture, industrial metal, and energy, the Goldman Sachs Commodity Index and the Dow Jones UBS Commodity Index and four major equity indices, CAC40, DAX30, FTSE100 and SP500. Second, we examine the cross-linkage between equity indices and individual commodity futures covering industrial metal, agricultural and energy markets to investigate the presence of heterogeneity across different markets.
The co-movement is symmetric and occurs most of the time
Our analysis uncovers three important stylised facts:
- The dependence between commodities (future or indices returns) and (US and European) stock index returns is time varying, symmetric and occurs most of the time.
In sum, we find that the underlying relation between commodity and equity markets remains the same whether the market bearish or bullish. What changes is the strength of that relationship.
- Our results also show that some the findings previously reported in the literature may not be robust.
By restricting the relationship between returns to be constant over time, some studies obtained spurious evidence of tail-dependence that is the co-movement occurs only during extreme events. Similarly, by imposing a tail-dependence, some works obtained false evidence of asymmetry.
- We find that the degree of dependence changes over time: the co-movement between industrial metals and equity markets strengthens as early as 2003, then spreads to all commodity classes with the global financial crisis after fall 2008 (see Figure 1).
The two categories mostly integrated to equity markets are energy and industrial metals.
Figure 1. Evolution of the dependence parameter between SP 500 and:
What can be the driver of this growing integration? As our approach is not structural, we can only formulate hypotheses. This growing dependence may be due to liquidity constraints faced by investors in the midst of the crisis. They have proceeded to fire-sale liquidation of assets to restore their balance sheet, a strategy triggering spillover to all asset markets. In the aftermath of the subprime crisis, investors have deleveraged for a long while because their balance sheets have remained seriously impaired long after the initial shock. The fact that the deleveraging has dragged on may explain that the dependence measure is still high at the end of our estimation period (February 2012).
In sum while the arrival of investors in the commodity market has most probably increased market liquidity when global liquidity was abundant, the growing cross-market linkages have made the commodity market vulnerable to liquidity spirals. Our results suggest that an appropriate market monitoring is needed on the top of the current reforms. Regulators must pay special attention to the new investment vehicles that provide a cheap access to the commodity market, such as index-funds, Exchange Traded Commodity etc.
Our findings have obvious portfolio implications. Unlike asymmetric relation or negative correlation, the symmetric relation between equity and commodity returns, especially when it is close to zero, does not allow for a ‘positive’ diversification of a portfolio, as it is less informative regarding the behavior of one set of returns relative to the other. Yet, the (on average) lack of relation between equity and commodity returns between 1990 and 2008, argues in favor of the commodity futures' risk diversification benefits for that period, with the exception of industrial metals whose sub-index shows growing co-movements with equity markets in 2003. However, it is rather difficult to defend the idea of diversification since 2008 as we clearly show that, for the remaining of the sample, the returns of equities and commodities (index, sub-index and most of the markets) have become more integrated.
Maystre, Nicolas and David Bicchetti (2012), “Are commodity derivatives good or bad? New evidence of high-frequency data”, VoxEU.org, 5 April.
Chong, J and J Miffre (2010), "Conditional correlation and volatility in commodity futures and traditional asset markets", Journal of Alternative Investments 12 (61), 75.
Delatte, A L and C Lopez (2013), “Commodity and Equity Markets: Some Stylized Facts from a Copula Approach”, CEPR Discussion Paper 9558.
Evenett, S, and F Jenny (2012), “Trade, Competition, and the Pricing of Commodities”, London, Centre for Economic Policy Research.
Gorton, G and KG Rouwenhorst (2006), "Facts and fantasies about commodity futures", Financial Analysts Journal 62(2), 47-68.
Patton, AJ (2006), "Estimation of Multivariate Models for Time Series of Possibly Di erent Lengths", Journal of Applied Econometrics 21(2), 147-173.
1 See for example “An investor Guide To Commodities”, Deutsche Bank, 2005.
2 Patton (2006) extended the standard copula approach to the conditional case. His measure of time dependence allows to capture any change in the strength of the co-movement, while keeping the underlying structure of that relation unchanged.
3 Commodity indices are financial vehicles that track commodity futures in a single measure calculated as a weighted average of selected commodity prices.