What is the evidence on aggregate food price behavior?
Aggregate food price has been less variable over recent years than many of its components, particularly food grains.Footnote 1 In 2005 the FAO food price index (Fig. 1) showed a modestly rising trend 20% higher than the 1998–2000 average. Prices accelerated in 2006, and by October were climbing sharply until the summer of 2008, when the index exceeded twice its 2005 level; by late summer, it had fallen far from the peak, but remained much higher than in 2000. However, this aggregate food price index understates the price fluctuations of the major food grains that attracted most attention.
For a longer view, the real price of wheat in the USA trended downward for decades (Fig. 2), reflecting yields that generally outpaced demand growth. Along their downward paths, prices generally fluctuate moderately within a reasonably defined range. However, episodes of steeply rising prices, followed by precipitous falls, are prominent features. The price series are asymmetric, there being no equally prominent troughs to match spikes, and when the price is relatively low the probability of a sudden fall is negligible. These features are characteristic of commodities generally and, relative to other spikes in the figures, those of recent years are not particularly high, when properly deflated.
The overall downward trend in prices is principally due to the remarkable success of plant breeders and farmers in continually developing and adopting new crop varieties with increased yields, and to the development of cheap and plentiful supplies of fertilizers and other inputs.
Increases in world consumption of the major grains have occurred even as the scope for expanding cultivated area has greatly diminished in most countries (Fig. 3). Note also the recent surge in diversion of maize to biofuel uses.
These aggregate figures mask great regional variation in prices and consumption, but the globalization of markets and reduction in shipping costs offer great opportunities for smoothing local fluctuations. Figure 4 shows wheat production in individual countries and in the whole MENA region. Pooling the entire regional output variation and sharing it proportionally would considerably stabilize wheat supplies, especially for Morocco, Iraq, Saudi Arabia and Tunisia. However, they would still need to import large portions of their grain supplies to feed their populations.
The increased demand for grain for human consumption has been driven mainly by increased global population, although the rate of increase has begun to slow. Only in poorer countries is increased income an important driver of grain consumption per capita, which is naturally limited by human stomach capacity. For grains used for animal feed, the trend towards increased consumption has been greater, because human consumption of animal products continues to rise with income long after minimum calorie requirements are satisfied. Use of maize as an animal feed has boosted its demand beyond the expectations based on its use as a human staple. Animal feed accounts for a smaller but significant share of wheat production, notably in Europe. Rice is used predominantly as a food. There is substantial agreement about the drivers of these long-term trends in grain consumption and prices; however, opinions differ on the causes of recent grain price volatility.
What caused recent grain price fluctuations?
A consensus is emerging among economists concerning the contributions of different factors to the recent market volatility for grains. Predictable disturbances can cause price trends, but not spikes, unless normal market responses are somehow constrainedFootnote 2—spikes are generated by surprises. Thus, the recent rapid increases in income in China and India have increased global demand for food and feed grains, but these increases, sustained for several years, were only surprising in their continuation into 2008. Similarly, reductions in the rate of yield increases in rice and wheat could have contributed to a tighter market, but as medium-term phenomena related to global neglect of crop research, they were hardly surprises. For detailed discussions see Abbott et al. (2008, 2009), Mitchell (2008), Timmer (2008) and Gilbert (2008). Factors such as the unprecedented extension of the Australian drought, other regional production problems, the possible effects of global warming, and exchange rate movements were much less predictable. However, their influence was not sufficient to explain most price spikes of 2007–2008.
Other market disturbances, which it was not possible to predict before 2007, were global in influence and deserve particular attention: the changes in biofuel policies and demand, and price spikes of fertilizers and fuel. As all relate directly to recent price spikes in the petroleum market, they merit special consideration by a regional group which includes many of the world’s major petroleum exporters.
The conversion of oilseeds into biodiesel in Europe, the USA and elsewhere; of sugar into ethanol in Brazil; and of maize into ethanol in the USA, link events in the world energy markets to food market behavior in a qualitatively new way.Footnote 3
In the USA, the diversion of maize and soybean to biofuel approaches 30% and 20%, respectively. This will further increase under current policies using subsidies and mandates, as well as protection from competition from more efficient Brazilian sugar-derived ethanol. The southern corn leaf blight infestation of 1971 cut maize supply in the USA by only half the percentage currently diverted to biofuels; however, it was a very serious shock that directed attention to the USA’s food security. Furthermore, the mandates for diversion of US maize for biofuel are expected to increase and have much more serious implications for supplies of maize for feed and food than an equivalent yield drop due to blight which was controlled relatively quickly by substituting genotypes that were resistant.
As existing government mandates for the use of grains, sugar and oilseeds were viewed as solid policy commitments, strong biofuel demand was clearly foreseeable by 2006—before prices took off. However, unanticipated oil price jumps must have encouraged upward revisions in the expected growth of biofuel-related demand for grains and oilseeds, as did upward revisions in the mandates in the USA. Furthermore, the diversions were too great to be made up in the short run by increased yields. They must have had large effects on the decreases in grain stocks, and the steady price increases in the years immediately preceding 2007–2008.
Wheat and other food grains were diverted to animal feed to substitute for maize diverted to ethanol and oilseeds diverted to biodiesel. Consumers increased rice demand, to replace the wheat used for feed. Biofuel demands and surges in meat demand caused by rising incomes also affected food grain markets less directly by diverting inputs from food grains toward feed and biofuel. Some rice land may have been diverted to produce maize or soybean, but it is unlikely this greatly affected overall rice production. The best rice land is ill-suited to corn or soybean in the temperate zones where much of the world’s maize and soybean are grown. However, on Asian croplands where two or three crops are possible annually, wheat can replace rice as a dry-season, irrigated crop when its relative price increases.
Prices of fertilizers and fuels
Worldwide adoption of modern high-yielding plant varieties and lower opportunities for expansion of cultivated area has increased fertilizer demand. Prices of some fertilizers rose faster than any other agricultural commodity price in 2007–2008, reflecting short-term supply constraints, energy costs and transport costs. Although some farmers and ethanol producers have blamed fertilizer and oil prices for grain price spikes, the evidence is not convincing. Grain price rises associated with previous harvests preceded fertilizer price movements, not vice versa.
Crude oil is an important input into modern agriculture, both directly and indirectly. Its price is virtually independent of disturbances in the grain markets. Crude oil prices have been very high recently, but there was no large effect on acreage or yield even in countries that use petroleum intensively in production. Farm prices in the USA rose dramatically as grain, fuel and fertilizer prices rose, indicating the net effect on farmers’ profits and incentives was positive and large. In contrast, unpredictable changes in petroleum prices affected grain demand, since jumps in petroleum prices now not only affect farmers’ costs, but also shift demand for the grain they produce via increased biofuel demand. This is a new phenomenon. When ethanol production exceeds mandated levels, marginal fuel price changes increase total demand for grains even as they increase input costs.
Pursuing this line of argument, a reasonable expectation might be that income growth and biofuel demand should have less influence on the price volatility of rice relative to maize and wheat. Yet the price spike was the highest for rice in 2008, indicating another significant contributor to chaos in the world grain markets—panic in the rice trade.
Panic in vulnerable markets
On 9 October 2007, the Indian Government banned exports of rice other than basmati, and increased rice availability to its consumers to reduce their concerns about inflation, and the adequacy of staple food supplies after a poor wheat harvest. The rice price outside of India began to rise (Fig. 5)—a problem with wheat supply thus triggered a sequence of decisions by other major exporters leading to the crisis in rice prices and market access (Slayton 2009).
As reports of production problems in other countries proliferated, governments of grain exporting countries were pressured by their urban consumers to reduce grain prices. These pressures temporarily outweighed the interests of producers and traders in selling to the highest bidder. One by one, rice exporters imposed their own export restrictions, including in March 2008, Viet Nam, an important supplier.Footnote 4 China, itself adequately supplied, did not make its substantial grain stocks internationally available as a supplier of last resort, as it had done earlier in the decade. Instead it taxed wheat and rice exports. Other countries also imposed export controls or taxes on wheat or maize exports; e.g. Argentina, South Africa, and the MENA countries, Egypt, Syria, Iraq and Yemen (World Bank 2009; Yemtsov 2008).
Countries dependent on imports of staple foods became increasingly anxious to secure supplies. Ultimately, only Thailand and the USA remained in the market as exporters. Many countries, including Morocco, Saudi Arabia, India and China, reduced their tariffs or taxes on imports, which reduced domestic prices relative to world prices, but also contributed to increasing world prices. One discouraging example of inadequate international cooperation was the failure to negotiate timely sale of Japanese rice stocks (imported to comply with WTO mandates, and never destined for domestic consumption) to desperate international importers (Timmer 2008).
This episode highlighted the strong substitution, at the margin, between the three major grains. Indeed, the market for the major grains can be considered as a market for grain calories (Bobenrieth and Wright 2009),Footnote 5 and so reliance of MENA countries on wheat is no insulation from disturbances in other grain markets. They must recognize that shocks in demand for maize for biofuels, for example, will have a relative impact on international wheat prices and availability comparable to the impacts on the maize markets. Following the causal chain further, OPEC-induced shocks to oil supply will likely affect the wheat import markets of OPEC countries.
Reviews of grain price volatility during recent years have allocated percentage shares of responsibility to a set of factors—a sensible approach if these factors had a linear cumulative effect on food price volatility; however, the effect is highly nonlinear. When supplies are already tight, another unexpected small supply reduction or new market order can become the ‘straw that breaks the camel’s back’ and cause a sharp price spike. This nonlinearity merits attention, as it is a key to understanding recent market events and constructing appropriate policy responses.
The economics of storage arbitrage determines the highly nonlinear relationship between grain prices and available supply, and aids evaluation of claims that other factors are the key drivers of market volatility. To understand the nonlinearity of the price–supply relationship, it is necessary to grasp some fundamental features of grain storage as an economic activity. The theoretical models separate demand for consumption from implicit demand for storage, under specific assumptions on the competitive structure of the storage industry. Once the inter-temporal price smoothing potential ensured by storage management is understood, policy analysts may be better able to value the alternative policy interventions to increase the effectiveness of storage management in enhancing food security.
The economics of storage
The effects of storage on the consumption and price of grains are illustrated in Fig. 6. The annual harvest in year t, h
, is random, reflecting the influences of weather and other disturbances on production. Total demand is the horizontal sum of the demand for consumption in the current period, c
; and the demand for grain stocks, x
, to carry forward for later consumption.Footnote 6 Consumption responds to price according to the downward-sloped market demand function P(c
). Stocks x
cannot be negative.
In storage arbitrage, regardless of the economic setting (monopoly, competition, state control of resource allocations), two accounting relations hold: available supply, A
, is the sum of the harvest, h
, and stocks carried in from the previous year, x
; and consumption is the difference between available supply and the stocks carried out. Profit-maximizing storage units hold positive stocks only if they expect returns to cover costs. If storage units are competitive, the current price of a unit stored must be expected to rise at a rate that covers the cost of storage and the interest charge on the value of the unit stored.
Given available supply, A
, storage units carry stocks x
from year t to year t + 1 following a version of ‘buy low, sell high’. Where the consumption demand is a downward sloping straight line (e.g. Fig. 6), if the price is sufficiently high, carried-over stocks are zero. When the price is high and no stocks remain, those who consume grains (e.g. rice, wheat or maize) as staples are willing to forego other expenditures (including health and education) to continue to eat their grain, so the consumption demand is very steep and price inelastic. Large changes in price result as consumption adjusts to the full effects of a supply shock; e.g. in 1972–1973, a reduction in world wheat production of <2%, when stocks were almost negligible, caused the annual price to more than double (Fig. 2). When stocks are high, a similar supply shock would have a far smaller effect on price (Fig. 6).
By acquiring stocks when the price is low, storage units can reduce the rise in consumption and thus cushion the associated fall in price. The disposal of stocks when supplies become scarcer reduces the severity of price spikes. If the supply of speculative capital is sufficient, storage can eliminate negative price spikes, but can smooth positive spikes only as long as stocks are available. When stocks run out, aggregate use must match a virtually fixed supply in the short term. Less grain goes to feed animals and the poorest consumers reduce their calorie consumption, resulting in malnutrition, hunger or even death.
If producers can respond to incentives with a one-year lag, that response is very stabilizing for consumption and price. If, for example, an irrigation system which has been shut down to save scarce ‘fossil’ water can be maintained in usable condition, it could be an emergency production reserve to stabilize consumption during severe shortage, without the high capital cost of holding emergency stocks to ensure a similar level of security.
In such markets, measuring both consumption and stocks (including stocks held by consumers) is very difficult (thus grain statistics refer to ‘disappearance’ rather than consumption), and complicates food policy directed at ensuring minimum consumption for all during shortages. Also, a common feature of all such physical storage activities is that, from a global viewpoint, aggregate stocks are constrained to be non-negative (even if there are conditions by which further release of stock would be profitable, because if current stocks are zero, it is impossible to ‘borrow from the future’). This fact makes modeling storage behavior particularly challenging, and thus estimating market behavior based on available data is very difficult.Footnote 7
How global storage affects world grain markets
To interpret the behavior of grain market prices, and identify the causes of high volatility, it is crucial to understand how storage behavior affects the relation between prices and available supplies of grain. The series of annual stocks-to-use ratios and annual real price indexes for wheat, reveals that the wheat price spikes in the 1970s, 1995–1996, and in 2007–2008 occurred when world stock-to-use ratios were low (Fig. 7). For the market to function effectively, a virtually irreducible minimum amount of grain must be held in the system, to transport, market and process grains. For example, no matter how urgent the present demand for grain, some must be held on docks during ship loading/unloading, and elsewhere in the supply chain. Stocks data are notoriously imprecise, and minimum working stocks are about 20% of use.Footnote 8 Comparison of two series (Fig. 7) shows that stocks are very unresponsive to price at these minimum levels. A similar comparison for maize would reveal the same phenomenon—spikes in price when stock-to-use ratios were low.
Another important feature of these grains is that the marginal cost of storage per period, including physical protection, insurance and spoilage, is usually modest. The assumption of constant unit costs is reasonable and a good approximation in regions where humidity is low, modern infrastructure is available, and deterioration unimportant, but not in many hot and humid environments (Paul 1970). The main cost of storing grains is usually the cost of capital invested in accumulating stocks. Increases in grain stocks are not generally limited by storage capacity. The storage of grain is profitable only if the value of the grain when released exceeds the sum of the storage cost and interest on the capital.