Food Prices, Ethics and Forms of Speculation

A regular commentary, during the commodities super cycle of the early to mid 2000s, was the obvious role played by financialization and, by extension, speculators. References to commodities such as oil and in particular the food versus fuel debate were to the fore during this period. High-profile examples include the Time magazine article entitled “Betting on Hunger: Is Financial Speculation to Blame for High Food Prices?” (Paramaguru, 2012), highlighting the influence of financial speculators on food prices. While the article did provide a brief discussion of possible fundamental causes such as the weather, non-OECD demand and biofuel policy, the main focus was on the role of commodity speculation on food prices. Testimony to the US Congress by hedge fund manager (Masters, 2008) also highlighted this issue. In particular, Masters (2008) maintained that growth in index investment created a bubble in commodity future prices which then fed through (via arbitrage) to spot prices. However, compelling the argument by Masters (2008), most academic studies have been unable to find a direct link between index trading and commodity futures price movements or the physical prices for that matter. Irwin and Sanders (2011), Cheng and Xiong (2014) and Haase et al. (2016) all provide excellent reviews of the literature.¹

To explain commodity prices within the food sector, we examine the role of a combined set of actors or investors. We distinguish between long-term (LT) investors, namely those focusing on fundamentals, and short-term (ST) investors. We consider two types of ST investors, Speculators and Manipulators. Our model develops an empirical application of Angel and McCabe (2009) description of speculation and manipulation in commodity markets. We consider the case of ST investors operating on legitimate grounds, who attempt to benefit from price changes (Speculators) and those that actively engage in pushing prices away from their fundamental value (Manipulators).² Both take a short-term horizon, however, the Speculators act rationally (explained in detail in the next paragraph), while the Manipulators take a contrarian focus. We also examine the case of agents that are only interested in long-run fundamentals (Fundamentalists). Rather than examining each case separately, we simultaneously combine all three agents in relation to commodity price movements. Evidence consistent with the popular narrative described above would lead to a definitive increase in the role of Speculators or Manipulators since the period of financialization.

At the heart of our classical fundamental-based model, lies the rational asset pricing model (RAPM). The RAPM indicates that commodities are valued based on their cash flows (convenience yield) that are likely to be generated, see Bredin et al. (2018). We refer to this price as the price determined by the long-run actors, the Fundamentalists. As highlighted above, we will also examine Speculators and Manipulators, using our ethical motivation. As a sensitivity test we compare our results with a purely data driven proxy for the role of Fundamentalists, Speculators and Manipulators using data from the ‘Commitment of Traders’, which is sourced from the Commodities Futures Trading Commission.

Our results provide consistent evidence that the price of commodities within the food sector are certainly influenced by the behaviour of both Speculators and Manipulators. In particular we find that commodity prices are far better explained by simultaneously taking account of fundamentals and both speculation and manipulation. The popular narrative described above might interpret this result as being a recent phenomenon and driven by the process of financialization. However, our findings highlight that this role exists before any financialization process began. While the importance of Speculators/Manipulators are certainly relevant, the vast majority of the post 1990 sample is driven by Fundamentalists. We do, however, find evidence that while the role of Manipulators is small, their importance has grown during the period most associated with financialization.

A commodity can be defined as a raw material or a primary agricultural resource, that can be purchased or sold for production and/or consumption. Most references to commodities over the last decade, however, have in reality viewed them in a different light. In particular, rather than as a consumption asset, commodities have hit the headlines as an investment asset. The flow of investors’ funds into commodities since 2000 has been striking, with the $200 billion in 2008 growing to $380 billion by 2011, see Rouwenhorst and Tang (2012). In 2018, commodity derivatives globally represent close to $2 trillion of outstanding over the counter (OTC) derivatives.³ Investors’ activity sparked considerable interest from the academic literature, with a flood of empirical evaluations on the benefits of commodity investing. The benefits of commodity investing, as discussed by Gorton and Rouwenhorst (2006) and more recently by Bhardwaj et al. (2015), are believed to include consistent risk adjusted returns and low correlations relative to US equities.⁴

The flood of funds into commodities, via commodity futures, has been referred to as the financialization of commodities. This process represents a structural change in commodity market participation and, combined with the price increase during the super cycle, leads to concerns over the potential role of speculation. Parts of the media and, to some extent, regulators have adopted the view that this process of financialization has been feeding through to physical commodity prices, to such an extent that physical prices are no longer closely tied to their fundamental value. Formal evidence of the increased role of financialization since 2005 has been provided by Tang and Xiong (2012) and most recently Bhardwaj et al. (2015).⁵ Further evidence is provided by Büyükşahin and Robe (2011, 2013) and Silvennoinen and Thorp (2013), who show that the return correlation between commodities and conventional financial assets (stocks) has turned positive and statistically significant since 2008, after being negative and significant in previous years. While Cheng and Xiong (2014) examine the implications of financialization in relation to both risk sharing and price discovery, Gilbert (2010) highlights the process of financialization as the principle cause of food commodity price rises during the super cycle.

The natural interpretation is and has been that the process of financialization will lead to greater levels of speculation within commodities. This interpretation very much relies on speculation forcing prices away from their fundamental values and so might be refereed to as manipulation following Angel and McCabe (2009). To date, there has been no formal empirical analysis of the role of both forms of short-run behaviour in relation to commodity price movements. There is, however, considerable evidence in favor of fundamentals driving recent movements in commodity prices and oil prices in particular. For the case of a broad range of commodities, Sanders and Irwin (2011) find evidence against speculative effects and consistent with fundamentals. Andreasson et al. (2016) also provides evidence against the link between speculation and food prices. Smith (2009) finds no empirical evidence to indicate that speculation increased oil prices, while Juvenal and Petrella (2015) find that speculation as well as economic fundamentals played a significant role in the oil price increase during the 2004–08 period. Hamilton (2009a, 2009b) provides a comprehensive overview of the potential causes of oil price changes and he concludes that speculation contributed to the price rise. Finally, Knittel and Pindyck (2016) find that speculation certainly played a role in the 2004-08 oil price increase, but was not the only contributor of the price increases. Our study is novel in the literature, as we examine for the first time the simultaneous role of fundamentals, speculation and manipulation in relation to food based commodities, and with a particular focus on financialization.

Our starting point is the present value model of asset prices, with a specific focus on commodity prices. Commodity prices can be defined as the present value of expected future ’payoffs’ associated with holding the commodity.Following Pindyck (1993), commodity prices (\(P_t\)) will change when there is changes in expected future ’payoffs’ and/or changes to the discount factor (\(\delta \)). The term \(\delta = \frac{1}{1+\mu }\), where \(\mu \) represents the expected return to holding the specific commodity. The normal application in equity markets would adopt the cash flows (dividends) as the payoffs. However, drawing on the theory of storage, these payoffs are defined as the benefits that accrue to the holder of a storable physical commodity, rather than the owner of the futures contract (see Bredin et al., 2018). These benefits are referred to as the convenience yield (\(\psi \)) and represent the benefits associated with holding inventory, e.g. maintaining a smooth production process and so avoiding any disruption to the flow of goods being produced. The market price of all commodities (e.g. agricultural and energy) will be determined by the expectation of market scarcity, reflected in the interaction between current supply and demand. The above present value model can be interpreted as a reduced form dynamic demand-supply model of commodity prices. The implication is that even when the expected capital return is small (relative to the risk adjusted rate) or negative, firms will still maintain positive stocks of inventory. Although formally the convenience yield is unobserved, an indication can be determined via an arbitrage relationship between spot prices and futures prices.⁶ See Bredin et al. (2018) for a detailed discussion.

A simplified version of the model is described below in graphical terms in the upper part of Exhibit 1. Rather than focusing on prices, for modelling reasons we examine the differential form of Eq. 1. The percentage net basis (PNB) is defined as the ratio of the convenience yield to price. This is consistent to the dividend price ratio when examining equities. On the vertical axis the relative value of the commodity, the (normalized) PNB, is reported. The time horizon in years is indicated on the horizontal axis, going from 1 year to 100 years (proxying the infinite investment horizon in the future). The relative value of the commodity today is determined by some combination of expected next period prices (diagonal pattern) and the present value of the convenience yield (grey).

The RAPM reported in Eq. 1, is represented by the commodity price being determined by an infinite stream of future payoffs associated with the physical commodity, reflected on the far right hand side of the figure. As the horizon grows, the importance of the convenience yield (grey) in relation to the relative commodity value increases in importance. Equation 1 reflects the case with infinite horizon and is consistent with the grey bars in the exhibit. This reflects the LT investor or Fundamentalist. Rather than focusing purely on the long-term, we could also examine a short-term perspective, the ST investor. We first consider the ST investor with a rational perspective. In the one period case, today’s relative value are primarily determined by expected one period returns (diagonal pattern). This is a perfectly rational setting, with the only caveat being that the investment horizon is short-term. We refer to this ST investor setting as the Speculator. As the investment horizons increase in time, the relative value of the commodity is increasingly driven by the fundamental element (or the convenience yield) and less by the one period return component. Critically both indicators of relative commodity value are consistent, as we move through time. The key take away from the exhibit is that commodity values will be consistent, irrespective of whether prices are driven by fundamentals (grey) or speculation (diagonal pattern).

Finally, we also consider the case of the agent that actively engage in pushing prices away from their fundamental value (Manipulators). Angel and McCabe (2009) define manipulation ‘as taking actions designed to move prices away from prices justified by economic fundamentals and to profit from the disruption’ (p. 282). This definition places considerable emphasis on the unethical aspect of this behaviour. A consistent interpretation, which also encompasses our model, is that this behaviour is primarily irrational. This may be as a result of herd behaviour following an announcement, e.g. an over-reaction to poorer than expected crop yields. The critical element in relation to manipulation is that prices are being pushed away from their fundamental value. Rather than the combined actions resulting in stable long-run prices, Manipulators push prices higher or lower with the resulting relative price not converging to a stable level. The case of Fundamentalists, Speculators and Manipulators are presented in the lower part of Exhibit 1, with the role of Manipulators reflected in the horizontal patterns. Rather than the one period and the infinite horizon being perfectly consistent, we now see that Manipulators have an adverse impact on relative value, in particular at the short horizon. Our aim is not to formally test for financialization in food commodities. We are, however, interested in examining whether the conventional view of financialization inspired manipulation is present and a driving force for food commodities. Rather than focusing on an individual actor, in our empirical analysis we simultaneously model Fundamentalists, Speculators and Manipulators, with the fraction assigned to each actor being determined by the relative predictive performance.

Following Brock and Hommes (1998) and, more recently, Lof (2015), we aggregate the three agents in a heterogeneous agent model in which the PNB, \(y^{T}_t\), is determined as follows;Here, the term \(G^j\) represents the fraction of the PNB determined at the margin by each agent (Fundamentalists (F), Speculators (S) and Manipulators (M)).⁷ The fraction assigned to each agent increases when the agent’s prediction outperform the others’, and the fractions sum to unity, as described by the following multinomial logit model;The \(\omega ^j\) parameters represent the willingness of agents to switch their strategy. A higher \(\omega ^j\) indicates greater willingness to switch. The term \(U^j_t\) represents a measure of fit, defined as the squared difference between the theoretical and the actual normalized PNB. Following the original recommendation of Hong et al. (2007) and the subsequent application by Lof (2015), we allow each \(\omega ^j\) to vary depending on the agent.

Our second and third hypotheses test examine the extent of systematic manipulation across our commodities and over time. If financialization-inspired manipulation exists, we should see some systematic behaviour for manipulation in the latter half of our sample. To examine the nature of manipulation, we plot the share of manipulation across all commodities. The literature to date has highlighted early to mid 2000s as the likely start of the financialization process within commodities. For example, Hamilton and Wu (2014), Cheng and Xiong (2014) and Chari and Christiano (2017) point to 2005, 20004 and 2002, respectively.⁸ The extent of systematic manipulation at the commodity level and the implications of financialization (if any) are tested using the following two hypotheses.⁹

Drawing on Angel and McCabe (2009), we examine the role of a combined set of actors in relation to food price movements. We specifically examine Fundamentalists, Speculators and Manipulators. We adopt a behavioural model and data from traders to formally examine the role of speculation in food commodities during the period of financialization. Following Angel and McCabe (2009) we quantify the role played by ’good’ speculators (Speculators), those that push prices closer to the fundamental value and ’bad’ speculators, those that push prices away from the fundamentals (Manipulators).

Our results indicate that Speculators and Manipulators play a consistent, yet relatively minor, role throughout our sample. A sample that has become most associated with the process of financialization. Although there is no evidence of a wave of manipulation occurring, we do find results that point to an increased role played by Manipulators during the period most associated with financialization. Rather than relying purely on our behavioural model, we also implement our hypothesis tests using the weights of different agents obtained directly from CFTC ’Commitment of Traders’ data. Consistent results are found when using CFTC data, to those reported using our behavioural model.

Following Angel and McCabe (2009), ours is the first known study to formally quantify the role of Fundamentalists, Speculators and Manipulators for commodities in general and food commodities in particular. Our study clearly highlights the importance of identifying the share associated with each of these agents individually and over time. While manipulation certainly exists in the food commodity sector, there has been no evidence of a wave of manipulation during the recent period of financialization. Rather than rely solely on our behavioural model, we also examine trader behaviour within the industry. The ’Commitment of Traders’ data fully supports our findings. Finally, our results point to two approaches that can be adopted to monitor manipulative behaviour within food prices and can be used to inform the need for regulatory interventions.