On the Causality of Commodity Price Movements
Based on the brief description above, both literature and evidence would suggest that the actual causality of agricultural commodity price movements is really multidimensional, and much more complex than often described. What follows will try to assess the role of individual factors and identify their eventual links and the relative weight of their impact.
Each of the several macroeconomic and sectoral influences identified in the previous section played a role in determining agricultural commodity prices. Yet it did so not in isolation, but in a rather complex confluence of developments. At times, volatility tended to dominate the debate, with the financialisation of commodities assuming the prominent explanatory role. At other times, the co-movement of prices, and especially their link to energy price changes, assumed the prominent role. Throughout, the level at which agricultural prices would settle in a new equilibrium and the sustainability of this level in the longer term raised and maintained strong policy and political concerns.
I n retrospect, some of the measures (export bans) and suggestions (strategic or virtual stocks) advanced during the most extreme period of market turbulence seem disproportionate, if not outright irrelevant, to the problem they were supposed to solve. (It is interesting in this respect to note that very limited attention was paid to the fact that the level of rice stocks dropped in 2015 more than it did in 2009. More interesting would be to estimate the potential loss for public expenditure had the idea of grain stock-building been accepted).
Unlike the developments in 2006-08, agricultural commodity prices recently declined independently of some of the factors affecting them. For example, biofuel use is still high, although it is growing at a much slower pace than before, and interest rates are still low, or even negative when adjusted by quantitative-easing policies. Yet energy, fertiliser, and agricultural prices continue to be characterised by a degree of co-movement.
In the search for explanations about the relative weight of the various factors affecting price movements, a recent paper (Baffes and Haniotis 2016) updates previous analysis of six agricultural markets (wheat, maize, rice, soybeans, palm oil, cotton) by incorporating more recent data. In summary form, here are its main conclusions.
From the macroeconomic variables, income growth was found to be negatively associated with agricultural prices, with parameter estimates negative and highly significant. Parameter values of income elasticity ranged within a remarkably tight band (the panel estimate indicates that a 10 per cent increase in the income of low- and middle-income countries reduces the real price of agricultural commodities by about 5 per cent).
Counterintuitive at first sight, this result is consistent with the Prebisch- Singer hypothesis, which states that, as income grows the price ratio of primary commodities over manufactured goods declines (Prebisch 1950; Singer 1950), as well as with Engel’s Law of less-than-unitary income elasticity for food commodities, since the negative sign of income elasticity should be interpreted as the difference between the effect of income on nominal food prices and the effect of income on the deflator (the former is generally lower than one, the latter greater than one).
The impact of a rise in the real interestrate is found to be negative but small on the prices of individual commodities. A weak relationship between interest rates and commodity prices is a common finding in the empirical literature (Gilbert 1989; Baffes 1997; Frankel and Rose 2010; Frankel 2014), although some studies (e.g., Akram 2009; Anzuini and others 2010) found that interest rates had a moderate effect.
The exchange rate, on the other hand, was found to have a negative effect on prices of individual food commodities, consistent with expectations. This confirms conclusions of numerous studies that have highlighted the negative relationship between a US dollar appreciation and commodity prices (see, for example, Lamm 1980; Gardner 1981; Baffes and Dennis 2015, for agriculture; Gilbert 1989; Baffes 1997; Akram 2009, for metals).
Among the sectorial fundamentals, the effect of a rise in the stock-to-use ratio is found to be, as expected, negative and highly significant. Baffes and Haniotis (2016) report a panel estimate of -0.37 (remarkably similar to findings reported elsewhere (for example, Bobenrieth et al. 2012, or FAO 2008).
Likewise, the effect of a rise in the real crude oil price is found to be significantly different from zero for all six commodities, with the panel estimate implying a 10 per cent increase in oil prices associated with a 1.5 per cent increase in agricultural prices. The strong relationship between energy and non-energy commodity prices was established in the literature long before the post-2004 price boom (Gilbert 1989; Hanson and others 1993; Borensztein and Reinhart 1994; Chaudhuri 2001), and was confirmed in more recent studies (Baffes 2007 and 2010; Moss and others 2010).
Yet not all studies concur with a strong oil-non-oil price relationship. Saghaian (2010; established a strong correlation among oil and other commodity prices (including food prices) but the evidence for a causal link was mixed. Gilbert (2010) found a correlation between oil and food prices, but noted that this could reflect common causation rather than a causal link. Zhang and others (2010) found no direct long-term relationship between fuel and agricultural commodity prices, and only a limited short-term relationship. Reboredo (2012) concluded that grain prices are not driven by oil price fluctuations.
The mixed evidence on the energy-non-energy price link could reflect the frequency of the data series used in the analysis or the presence of biofuels (Baffes 2013). Zilberman and others (2013) noted that higher-frequency
(and hence ‘noisier’) data are typically associated with weaker correlations. On the other hand, an exogenous shock pushing crude oil prices down under a mandated ethanol-gasoline mixture would increase fuel consumption and push ethanol and maize prices up, and thus lead to a negative relationship between food and oil prices, other things being equal (De Gorter and Just 2009).