Financialisation of Commodity Markets
Co-movement Behind-the-Scenes
Abstract
In the early 2000s institutional investors entered the commodity futures markets en masse with passive, long only, index type positions in sharp contrast with those typically assumed by traditional expert participants. A heated public debate soon erupted over the perceived consequences of the phenomenon—commonly referred to as “financialisation”—and, in response to immediate policy concerns, the matter was thrust onto the academic sphere as a burning issue. With the benefit of hindsight it now seems that the academic debate was framed rather narrowly, used contentious research methods, and eventually led to regulatory changes that were therefore perhaps unwarranted. In contrast, we take a broader approach where we consider a large cross-section of liquid commodities as suggested by the nature of financialisation that comprehends commodity futures as an asset class. We examine the cross-sectional interconnectedness of these assets and uses a bespoke asset pricing factors based framework to study the issue through the lens of co-movement. We find that the phenomenon had ontological consequences for the commodity complex with its impact extending beyond the mechanical effects induced by indexation. The onset of the financial crisis and the monetary policy regimes that followed, on the other hand, seem to have set off a motion of reversion to legacy pre-financialisation fundamentals.
Keywords
Commodity markets, Financialisation, Co-movement, Asset pricing
Introduction
The last twenty years have seen major upheavals in the commodity markets with the onset of financialisation1 as well as the period of the 2008 financial crisis and its aftermath. Financialisation appears to have affected multiple economic sectors, including agriculture and energy, and its impact has been hotly debated in the policy, legislative, regulatory, and academic spheres. A rigorous legislative scrutiny accompanied by a vigorous policy and academic debate across a number of disciplines about the perceived effects on commodity prices of the unprecedented inflow of institutional funds brought into commodity futures markets by new index type investors2 eventually led to regulatory changes in several individual markets.
In response to immediate policy concerns, the academic debate was initially framed rather narrowly around adequacy of speculation—the burning issue of the day—at the individual commodity level and the ensuing academic analysis focused on the more mechanical effects of financialisation. With the benefit of hindsight, this approach seems to lack a fundamental understanding of the phenomenon it endeavoured to study and which consequences it attempted to address.
Financialisation was a phenomenon global in nature that, through the channel of index investment, affected the whole cross-section of liquid commodities at once, eventually spurring its transition away from a collection of idiosyncratic markets to its emergence as an asset class. Focusing on individual markets in this context thus seems limited. Most of the studies underpinning the regulatory response to financialisation took the form of pairwise correlation, causation and to some extent co-integration analysis, all of which are idiosyncratic in nature for they study single pairs of individual assets. Besides, some of these techniques since proved controversial and seem to produce results the quality of which may fall short of the standards that may reasonably be expected to lay the ground for legislative action. Although appealing, for it comes with the promise of prompt “significant” results, this approach misses a critical corollary of the substantial advent of a new category of players in a (set of) market(s): the alteration of fundamentals it potentially induces, best observed in this context through the lens of commercial hedging pressure (Keynes 1930; Hicks 1939).
While essential to the understanding of the phenomenon, the global nature of financialisation and its impact on market fundamentals therefore appear to have been originally overlooked and another approach, perhaps more refined, seems warranted.
We strive to provide one in this study by taking a fundamentally different empirical perspective that is broader in scope. We examine the interconnectedness of the commodity complex at the cross-sectional level, considering the entire cross-section of actively traded commodities on futures markets and use a bespoke futures-based asset pricing framework that includes factors constructed using both theoretical fundamentals and liquidity considerations.
The global nature of financialisation makes cross-sectional co-movement a cornerstone issue that the direct impact of increased market participation alone might not be able to fully explain for it potentially had deeper effects, for example on market fundamentals as our results suggest. Commodity futures markets have a history of being the classic physical commodity price risk shifting conduit for commodity specialists. Over the financialisation period we observe a decrease in effectiveness of commercial hedging pressure (CHP) at explaining commodity futures returns at the individual level accompanied by a more pronounced increase at the aggregate level. We further note an overall shift in the nature of hedger’s behaviour away from the traditional Keynesian view of risk transfer with higher individual risk premiums over phases of cross-sectional contango as opposed to backwardation. Both patterns suggest that the phenomenon may have altered this long-standing paradigm of commercial risk transfer with the entry of long only commodity index traders (CITs) with very different demands from traditional participants. This change could have altered the hedging ontology in these markets and our asset pricing framework is tailored to detect these more subtle effects driven by changes in the nature of market participation.
Factor model techniques are well suited to isolating common driving factors and detecting co-movement (Fama and French 1993, 2015; Carhart 1997; C. Asness and Frazzini 2013; Hou, Xue, and Zhang 2015; C. Asness et al. 2015; Frazzini and Pedersen 2014; C. S. Asness, Frazzini, and Pedersen 2019) at a broader level, beyond individual pairs of assets, and are commonly used to study commodity market fundamentals (E. Schwartz and Smith 2000; Miffre and Rallis 2007; Gorton, Hayashi, and Rouwenhorst 2012; Cortazar, Kovacevic, and Schwartz 2013; Yang 2013; Daskalaki, Kostakis, and Skiadopoulos 2014; Szymanowska et al. 2014; Fernandez-Perez et al. 2018; Bakshi, Gao, and Rossi 2019; Boons and Prado 2019; Sakkas and Tessaromatis 2020); they consequently seem well adapted to study these two issues simultaneously. Building on theory we therefore develop a bespoke commodity-based asset pricing framework comprised of factor models based both on fundamentals as well as liquidity considerations.
One of these is based on the Keynes-Hicks (Keynes 1923, 1930; Hicks 1939, 1946) risk transfer approach to the term structure; the corresponding “theory of normal backwardation”, complemented for the possibility of contango (H. Houthakker 1957; Cootner 1960), postulates that futures prices for a given commodity are inversely related to the extent that commercial hedgers are short or long and using commercial hedging pressure (CHP) as a proxy for hedgers net market position we construct factor mimicking portfolios that capture the impact of hedging pressure as a systemic factor (Basu and Miffre 2013). The cross-sectional nature of financialisation (Cheng and Xiong 2014; Basak and Pavlova 2016) further leads us to extend this paradigm by incorporating aggregate CHP (CHP), a market wide measure of hedging pressure first considered in Hong and Yogo (2012). In this context, returns for individual commodities should be high in periods where CHP is low (backwardation) and low in periods where CHP is high (contango). Working’s contending approach that relates futures price formation to storage costs in the theory of “the price of storage” (Working 1949) forms the basis of another model that we consider where, following Szymanowska et al. (2014) and Fuertes, Miffre, and Fernandez-Perez (2015) who demonstrate that the term structure factor has explanatory power for the cross-section of commodity returns, we construct factor mimicking portfolios using roll-yield as a proxy for the front end shape of individual commodity term structures. We further consider a liquidity-based model following Hong and Yogo (2012) who find that open interest growth has predictive power for individual commodity returns and again, using the latter measure as a proxy for individual commodity liquidity, we construct long-short factor mimicking portfolios.
We study the performance of these models on a set of twenty-four US traded commodities and six GB traded (London Metal Exchange) base metals over the 1997/2018 period that we divide into four distinct sub-periods: pre-financialisation (1997/2003); financialisation (2004/20083); financial crisis and aftermath, including loose monetary policy regimes that followed (2008/2013); and post-crisis (2014-2018).
Consistent with earlier studies, we observe that mean returns for the US traded commodities are sharply higher over the financialisation period compared to pre-financialisation with the pattern extending to GB traded metals.
Over the first period the cross-sectional average returns during phases of backwardation for both the US and GB commodities are positive and overall higher than during phases of contango, providing broad support for an aggregate version of the Keynesian hedging pressure hypothesis. Over the second period this pattern reverses, with both the US and GB commodities having higher returns during phases of contango as opposed to backwardation. This change in pattern suggests the onset of financialisation altered the traditional risk shifting nature of hedger’s behaviour, implied by the Keynesian hypothesis.
This change of paradigm due to the arrival of financial investors, on the other hand, raises the question of whether they functioned as liquidity providers or demanders. Arguments have been put forward in favour of both possibilities with Moskowitz, Ooi, and Pedersen (2012) arguing that financial investors provide liquidity and Kang, Rouwenhorst, and Tang (2020) taking the opposite stance. Our analysis of individual commodity returns’ relation with both CHP and CHP before and during financialisation sheds a new light on this issue with our results suggesting that financial investors may have been liquidity demanders leading hedgers to become liquidity providers.
These entered the commodity spectrum en masse with across-the-board index type positions prompting the transition of the complex from a collection of idiosyncratic markets towards a more cohesive ensemble akin to an asset class. In this context of heightened cross-asset market participation the issue of interconnectedness stands out as paramount and we strive to investigate it by first examining the correlation structure of our sample set of assets at the global, country, sector and sub-sector levels. Our results indicate higher interdependence within the full cross-section of commodities over the financialisation period, consistent with both the empirical and theoretical literature. The increase appears to be driven by the US metals sector, a new observation to the best of our knowledge, and, unlike individual commodity returns, does not seem to be regime-driven, as a similar pattern of results is observed across phases of backwardation and contango.
Goldstein, Li, and Yang (2014) and Goldstein and Yang (2022) argue that the arrival of financial investors had an impact on the extent of normal backwardation in commodity futures markets. Based on their analysis we would expect our average CHP factor, based on backwardation as measured by hedger’s positions (Keynesian perspective, theory of normal backwardation: Keynes (1930), Hicks (1939), H. S. Houthakker (1957), Telser (1958)), as well our term structure factor, based on backwardation as measured by roll yield (Working’s perspective, theory of storage: Working (1933), Kaldor (1939), Working (1948), Brennan (1958), Cootner (1960), Weymar (1966), Danthine (1978), Turnovsky (1983), E. S. Schwartz (1997)), to be best suited to measure the impact of financialisation.
We analyse the pricing dynamics over the different periods by first studying the time series explanatory performance of the various factors on our commodity assets sample. The market-based factor outperforms the other three in the first period and, although its performance increases over the financialisation period, that of the CHP factor shows a greater improvement in a relative sense. The performance of the market factor is driven by agricultural and energy commodities in the first period while metals strongly contribute to the performance improvement observed over the second period. The sector is also the driving force behind the dramatic improvement of the CHP factor during financialisation, led by its long leg on which metals predominantly load. The sector is similarly responsible for the improved performance of the short leg of the term structure factor of the period, on which the metals also predominantly load. Taken together, these results provide evidence of convergence of Keynesian and Working’s theories of the term structure in detecting price co-movement during financialisation and strongly suggest that the term structure related attributes (hedging pressure/roll) co-movement and price co-movement link was concentrated in the metals sub-sector.
The GB traded base metals provide a useful set of assets to test how widespread were the changes in pricing dynamics brought about by the onset of financialisation. To that end, we examine the performance of the four factors on the six GB metals; the results show that there was co-movement in global metals returns in the second period and that this co-movement could be detected by our term structure related models, particularly so by our hedging pressure-based model.
The onset of the financial crisis and the monetary policy regimes that followed, on the other hand, also appear to have induced significant changes. Commodity futures returns fell dramatically while the symmetry between the dynamics of CHP and CHP’s’ explanatory power on individual commodity returns breaks with both now seemingly converging. The correlation structure is also affected with evidence of an intensification of the higher interdependence within the full cross-section of commodities prompted by financialisation, consistent with the greater interdependence between equity and commodity markets over the period documented in the literature. The pricing results further indicate the emergence of a systematic factor across the entire cross-section of the US commodities as well as evidence of cross-market linkages with the GB traded commodity subset. The onset of the financial crisis thus seems to have had a global impact on the commodity complex in contrast with financialisation, much more closely related to the metals sector.
The returns pull-back continues into the post-crisis period, interestingly amidst a lower volatility environment, suggesting that the underlying contributing factors–including investment outflows from the commodity futures complex as well as perhaps the evolving nature of global oil demand over the period–have had a lower impact on volatility than witnessed during financialisation. The pattern of reversion toward the pre-financialisation Keynesian paradigm continues stronger with mean returns higher during phases of backwardation relative to contango and the difference more pronounced than over the crisis period. In contrast to the crisis period when they seemed to be converging, the pricing performance of CHP and CHP now part ways, seemingly reverting to the Keynesian paradigm observed pre-financialisation, further suggesting a reversion to legacy commodity specific dynamics. The pull back extends to the correlation structure with an across-the-board decrease in average pairwise correlations down to levels below those observed during financialisation. The average market factor beta for the entire cross-section shows a similar pattern, dropping down below the level observed during financialization with the decline seemingly driven by the metals sector, particularly the GB traded segment. The pattern extends to the pricing performance of all factors that drops below levels observed during financialisation.
Moreover, both the average correlations and the pricing performance of the market factor are considerably stronger during contango phases, unlike previous periods, which showed no substantial difference between regimes with the pattern particularly pronounced for metal assets globally, both US and GB traded. These results document the emerging role/dominance of the contango regime post-crisis and could provide a valuable basis for further research.
Financialisation was an issue of such policy importance that it triggered legislative action. The debate was initially framed around adequacy of speculation, the burning issue of the day, and the ensuing academic analysis focused on the more mechanical effects of financialisation.
With the benefit of hindsight, it appears this approach was perhaps too narrow, and it now seems necessary to address the phenomenon from a broader perspective. Commodity price dynamics appear to have altered substantially in quite different ways over the financialisation and the financial crisis periods and both our cross-sectional approach to examining interconnectedness in the complex as well as our commodity futures asset pricing framework together seem able to provide new insights into the nature of these changes: financialisation was a phenomenon endogenous to the commodity markets transmitted via the commodity futures markets and which effects were particularly strong for the metals sector while the crisis and its aftermath seem to have delivered an exogenous shock across the entire cross-section of liquid commodities, further altering the cross-sectional dynamics, and potentially triggering a reversion to pre-financialisation dynamics in the most recent period.
The initial view of financialisation was that it consisted of speculative flows which had the effect of driving up and creating bubble like conditions in commodity futures prices4. However, the fundamental question of the nature of the impact of financialisation across the entire cross-section of commodity futures markets has not yet been completely answered. We provide an empirical complement to a new stream of theoretical studies that try to model the impact of financialisation on various aspects of commodity futures markets5 and demonstrate that the effects of financialisation extend beyond the mechanical effects induced by indexation as outlined in Tang and Xiong (2012) and Basak and Pavlova (2016).
Data & methods
Background
The period from 1998 to 2008 saw an influx of investment into commodity index linked products most of which made its way into the commodity futures market. Popular commodity indexes whose goal was to track the broad movement of commodity prices became accessible by means of swaps, exchange traded funds and exchange traded notes and attracted at least $100 billion of net super-hoc investment over the 1998-2008 period.
Concerns over the consequences of this financialisation phenomenon eventually led to legislative changes including the approval of Rule 76 FR 4752 issued by the US Commodity Futures Trading Commission (CFTC) on January 26th, 2011. This provision emanates from the Dodd-Frank Wall Street and Consumer Protection Act of 2010 (Title VII, Section 737) that mandates the CFTC to use position limits to restrict the flow of speculative capital into a number of commodity markets. The Rule was approved in a close 3-2 vote and the ensuing rule-making process was extremely contentious with several commissioners (Michael Dunn and Scott D. O’Malia in particular) expressing reservations about the lack of supporting evidence and the Rule also triggering thousands of comment letters as well as a lawsuit against the CFTC spearheaded by two Wall Street trade groups, the International Swaps & Derivatives Association (ISDA) and the Securities Industries & Financial Markets Association (SIFMA)6.
A world-wide debate ensued about the role of index funds in commodity markets. The first responses to the 2007/2008 crisis of escalating food and energy prices took the form of policy reports, many of which reasoned that the growth of commodity index funds came along with an influx of largely speculative capital that was responsible for driving commodity prices beyond their historic highs (De Schutter 2010; Gilbert 2010a; Herman, Kelly, and Nash 2011; Schumann 2011; UNCTAD and Cooperation 2009).
Early US Senate investigations on the matter drew on pricing and trading data supplied by the CFTC as well as interviews with numerous experts including hedge fund manager Michael Masters who linked the growing presence of index investors in US agricultural and energy markets to the observed surges in both futures and cash prices (Michael W. Masters 2008; M. W. Masters and White 2008). This contention has since come to be known as the “Masters’ hypothesis” and was largely endorsed by the final US Senate report on the issue (Senate 2009).
At this stage, the question thrust on the academic sphere was whether “excessive speculation”7 was linked to escalating energy and agricultural commodity prices. A number of ensuing academic studies, most of which used commodity specific correlation and causality based analysis have since disproved this contention: Irwin, Sanders, and Merrin (2009), Sanders, Irwin, and Merrin (2010), Sanders and Irwin (2011a), Sanders and Irwin (2011b), Irwin (2013), Brunetti and Reiffen (2014), Hamilton and Wu (2015) and Bruno, Büyükşahin, and Robe (2017) investigated the issue in the context of the agricultural markets; meanwhile, Büyükşahin and Harris (2011), Tokic (2012), Fattouh, Kilian, and Mahadeva (2013), Kilian and Murphy (2014), Knittel and Pindyck (2016) and Manera, Nicolini, and Vignati (2016) examined the energy markets while Bohl and Stephan (2013), Kim (2015) and Boyd et al. (2016) studied both energy and agricultural markets and Irwin and Sanders (2011), Irwin and Sanders (2012) as well as Stoll and Whaley (2011) examined the commodity markets in general. Other studies have underlined various alterations in commodity pricing dynamics over the period. Gilbert and Pfuderer (2014) reject the view that financialisation has not had any effects in the grains markets and demonstrate that trades originated by financial market participants, and specifically index investors, can move prices but tend to be typically volatility-reducing. Juvenal and Petrella (2015) contend that the oil price increase between 2004 and 2008 was mainly driven by the strength of global demand but that the financialisation process of commodity markets also played a role. Likewise, Henderson, Pearson, and Wang (2015) show that non-information-based financial investments have important impacts on commodity prices. Cheng and Xiong (2014), on the other hand, show that in the case of index commodities, financialisation has transformed the risk sharing, and information discovery functions of commodity futures markets.
Data
Commodity index swaps, commodity index funds (CIFs), commodity-based exchange traded funds (ETFs) and notes (ETNs) or commodity linked notes (CLNs) were amongst the main investment vehicles that allowed institutional and retail investors to build up commodity exposure in the early 2000s (Boons, De Roon, and Szymanowska 2012; Henderson, Pearson, and Wang 2015; Irwin and Sanders 2011; UNCTAD and Cooperation 2009; Schumann 2011). Most of these products were primarily designed to track prominent commodity indexes among which the S&P-GSCI, the most popular.
Most of its constituents at the time of writing form the basis of the broad cross-section of commodities that we consider in this study: corn-#2 yellow (XCBT), oats (XCBT), soybean meal (XCBT), soybean oil (XCBT), soybeans (XCBT), wheat-SRW (XCBT), cattle-feeder (XCME), cattle-live (XCME), lean hogs (XCME), cocoa (IFUS), coffee-C (IFUS), cotton-#2 (IFUS), lumber (XCME), orange juice (IFUS), sugar-#11 (IFUS), natural gas (XNYM), crude oil-WTI (XNYM), gasoline (XNYM), heating oil (XNYM), copper (XCEC), gold (XCEC), palladium (XNYM), platinum (XNYM) and silver (XCEC). Unleaded gasoline, the NYMEX legacy contract for gasoline, stopped trading in 2006 when it was replaced by Reformulated Gasoline Blendstock for Oxygen Blending (RBOB) with the two futures series trading alongside for most of the year. For our NYMEX gasoline data series, we consider unleaded gasoline up to September 1st, 2006 and RBOB gasoline thereafter, date at which liquidity for the latter overtook that for the former.
We also consider the GB metal complex with aluminium (XLME), copper (XLME), lead (XLME), nickel-primary (XLME), tin-refined (XLME) and zinc (XLME). This sample cross-section of assets is further divided into groups for the purpose of the analysis by country (US, GB), sector (agriculturals, energy, metals) and by sub-sector (agriculturals: grains, livestock, softs; energy: gas, petroleum; metals: base, precious)8.
The commodity futures trading commission (CFTC) operates a comprehensive system of collecting information on market participants known as the large trader reporting program (LTRP) where it records and reports to the public position data on market participants who have position levels in excess of a particular market specific threshold9. The commission collects market data and position information daily from clearing members, futures commission merchants (FCMs) and foreign brokers and publishes corresponding summaries weekly in a series of market specific reports. In these reports, individual traders are categorised according to the nature of their trading activity on the basis of self-reported information10 that is subject to review by CFTC staff for reasonableness. Position information is then aggregated by category and asset type with each report providing an aggregation for a set of report specific categories and, for each category, a breakdown between futures only positions and positions in futures and options combined.
In its legacy report, the commitment of trader report (COT) with data dating back to 1962, the CFTC distinguishes “commercial” and “non-commercial” market participants where a “commercial” participant is defined as one “[…] engaged in business activities hedged by the use of futures and option markets”11. A third category, “non-reportable”, aggregates positions for participants not meeting the reporting threshold. In response to concerns related to category accuracy12 the CFTC now refines its classification in the disaggregated commitment of trader report (DCOT) where participant categories include “Producer/Merchant/Processor/User”, “Swap dealer”, “Money manager” and “Other reportable” as well as in the supplemental commitment of trader report (SCOT) that details commodity index trader aggregate positions for thirteen agricultural commodity futures markets. Data for DCOT and SCOT date back to 2006 and 2007 respectively.
We examine the period from 1997 to 2018, during which we utilise futures-only position data from the COT report—the only CFTC report providing data for the entire duration of interest—as well as futures term structure price and open interest data sourced from Bloomberg. The sample period is further divided into four distinct intervals. The pre-financialisation era spans the 1997-2003 period and is naturally followed by the financialisation phase running from 2004 to 2008 which in turn leads into the financial crisis and its aftermath, covering 2008 to 2013. The final interval starts in 2013 with the onset of tapering and the conclusion of the US Quantitative Easing (QE) program.
Methods
For each commodity asset we compute the daily and weekly front futures return as well as weekly commercial hedging pressure (CHP) time series with the CHP calculated as the ratio of the number of long commercial positions to the total number of commercial positions for the considered asset-week combination. For each asset-period combination we compute the average annual daily return and corresponding volatility over the entire period as well as over phases of cross-sectional low and high CHP13 and further test for significance in difference between the two regimes using t-tests for mean returns and F-test for corresponding variances. These statistics are also computed, and their difference between regimes tested, for equally weighted portfolios of the US and GB traded commodity assets independently.
For each period, the weekly return series for the individual US-traded assets are regressed on both the relative change on their respective CHP series and that on the cross-sectional CHP independently. The resulting \(R^{2}\) values are then averaged by sector and sub-sector. Additionally, pairwise daily return correlations are computed for each combination of assets across the periods of interest and for each individual year within the full sample period. These correlation coefficients are then averaged across the entire cross-section of assets, as well as by country, sector, and sub-sector. For each period, the coefficients are also calculated and averaged independently over phases of low and high cross-sectional CHP. We further form equally weighted portfolios from the US traded commodity assets grouped by sub-sector. For each period we compute the average pairwise return correlation coefficients between these portfolios over the entire period as well as over phases of low and high cross-sectional CHP independently.
This approach sheds light on the evolving interconnectedness among commodity assets as the complex transitions from traditional market dynamics—where commodity producers and speculators had distinct roles—to a more intricate landscape shaped by the influx of financial investors.
On the one hand, futures markets have a long history of serving commodity producers to alleviate their commodity price and output risks. Keynes (1930) and Hicks (1939) emphasise the “normal” behaviour of naturally short hedgers outsourcing their commodity risk to naturally long commodity specialist speculators. This market configuration is referred to as “backwardation” where futures are expected to trade at a discount to spot prices, providing speculators with incentives to take the long side. The opposite market paradigm is referred to as “contango”. Working (Working 1933, 1948, 1953) and Kaldor (Kaldor 1939) somehow relax this “supply-of-speculative-services” (Till 2007) approach by introducing processors and merchants, also commodity experts, on the hedgers side and relating the notions of backwardation and contango more directly to the term structure via the theory of storage where the shape of the front end of the term structure (basis) is directly related to the market price of storage.
Financialisation, on the other hand, refers to the entry of financial investors into the commodity futures markets who, for the most part, are not commodity experts. Besides, these new market participants tend to exhibit herding like behaviour, often taking massive long-only positions on the whole cross-section of indexed commodities in an attempt to enhance returns while hedging against inflation and reaping further diversification benefits for portfolios with, in most cases, existing large positions across various asset classes (Brunetti, Büyükşahin, and Harris 2016; Boyd et al. 2016; Cheng, Kirilenko, and Xiong 2014; Juvenal and Petrella 2015; Singleton 2013; Tang and Xiong 2012). This contrasts with the traditional approaches of Keynes and Working and in this context the issues of co-movement and aggregate market participants behaviour seem particularly relevant.
By grouping together assets by shared attribute dynamics in common portfolios, asset pricing techniques—and factor models in particular—are well suited to study co-movement (Fama and French 1993, 2015; Carhart 1997; C. Asness and Frazzini 2013; Hou, Xue, and Zhang 2015; C. Asness et al. 2015; Frazzini and Pedersen 2014; C. S. Asness, Frazzini, and Pedersen 2019). Careful, theory-based, attribute selection further enables a refined analysis of market fundamentals (E. Schwartz and Smith 2000; Miffre and Rallis 2007; Gorton, Hayashi, and Rouwenhorst 2012; Cortazar, Kovacevic, and Schwartz 2013; Yang 2013; Daskalaki, Kostakis, and Skiadopoulos 2014; Szymanowska et al. 2014; Fernandez-Perez et al. 2018; Bakshi, Gao, and Rossi 2019; Boons and Prado 2019; Sakkas and Tessaromatis 2020). We therefore develop a futures-based asset pricing framework that includes factors constructed using both theoretical and liquidity considerations. Using the weekly returns, open interest and market position data described above we construct market portfolios of individual commodity nearby futures as well as mimicking portfolios for market, commercial hedging pressure, term structure and liquidity risk factors in returns. While our market portfolios are long only combinations of particular commodity sets, our other mimicking portfolios for risk factors are constructed as combinations of two sub-portfolios (legs), one held long, one held short, which respective constituents are pooled on a weekly basis according to risk factor specific criteria.
All portfolios are equally weighted and returns for the two-legged factors are calculated as the difference between the average return for constituents of the long leg and that for the constituents of the short. i.e, return on the mimicking portfolio for risk factor \(i\) (\(f_{i}\)) for period \(t\):
\[r_{t}^{f_{i}} = \frac{1}{x} \left ( \sum_{j=1}^{x}r_{j,t} - \sum_{j=n-x}^{n} r_{j,t} \right )\] \(n\equiv\) number of commodities in the set considered for mimicking portfolio construction.
\(x = n \cdot s\).
\(s\equiv\) selection threshold (\(\frac{1}{3}\) here).
\(r_{j,t}\equiv\) period \(t\) front contract return for commodity \(j\).
\(j\equiv\) commodity rank in the ordered set \(Y_{t}^{f_{i}}\).
\(Y_{t}^{f_{i}}\equiv\) period \(t\) ordered set of the commodities considered for mimicking portfolio construction where the ordering rule is specific to \(f_{i}\).
For our portfolio CHP (commercial hedging pressure), we sort the set of commodities considered on past twenty-six-week average CHP from lowest to highest. The bottom third constituents (lowest \(\overline{CHP}\)s) of the corresponding ordered set form the long leg of the portfolio while the top third form the short leg:
\[Y_{t}^{CHP}\equiv\left \{ \left \{ \overline{CHP_{1}}, \overline{CHP_{2}}, ..., \overline{CHP_{n}} \right \}, \leq \right \}\]
\(\overline{CHP_{j}}=\frac{1}{26}\sum_{k=0}^{25}\frac{L_{j,t-k}}{L_{j,t-k}+S_{j,t-k}}\)
\(L_{j,t}\equiv\) number of long positions held by commercial hedgers for period \(t\) on the futures series of commodity \(j\).
\(S_{j,t}\equiv\) number of short positions held by commercial hedgers for period \(t\) on the futures series of commodity \(j\).
For our portfolio TS (term structure), we sort the set of commodities considered on past twenty-six-week average roll-yield from highest to lowest. The bottom third constituents (highest \(\overline{RY}\)s) of the corresponding ordered set form the long leg of the portfolio while the top third form the short leg:
\[Y_{t}^{TS}\equiv\left \{ \left \{ \overline{RY_{1}}, \overline{RY_{2}}, ..., \overline{RY_{n}} \right \}, \geq \right \}\]
\(\overline{RY_{j}}=\frac{1}{26}\sum_{k=0}^{25}(\frac{F_{j,t-k}^{1}}{F_{j,t-k}^{2}} - 1)\)
\(F_{j,t}^{1}\equiv\) period \(t\) close price for commodity \(j\) first nearby futures contract.
\(F_{j,t}^{2}\equiv\) period \(t\) close price for commodity \(j\) second nearby futures contract.
For our portfolio OI (open interest), we sort the set of commodities considered on past twenty-six-week average front contract open interest growth from highest to lowest. The bottom third constituents (highest \(\overline{OI^{\Delta \%}}\)s) of the corresponding ordered set form the long leg of the portfolio while the top third form the short leg.
\[Y_{t}^{OI}\equiv\left \{ \left \{ \overline{OI_{1}^{\Delta \%}}, \overline{OI_{2}^{\Delta \%}}, ..., \overline{OI_{n}^{\Delta \%}} \right \}, \geq \right \}\]
\(\overline{OI_{j}^{\Delta \%}}=\frac{1}{26}\sum_{k=0}^{25}(\frac{OI_{j,t-k}}{OI_{j,t-k-1}} - 1)\)
\(OI_{j,t}\equiv\) open interest at close of period \(t\) on the front (nearby) contract for commodity \(j\).
We use a time series regression approach where individual commodity front futures contract daily returns are regressed on returns to the mimicking portfolio for a particular risk factor with the corresponding \(\beta\)s and \(R^{2}\)s averaged. For the long-short factors the \(\beta\)s and \(R^{2}\)s are computed for models where the regressor is the factor itself as well as both its long and short legs independently14.
For each time period considered in the study, being one of the four multi-year periods described above or any individual year over the entire sample time period, we carry out our analysis over the entire period as well as over phases of low and high aggregate CHP (CHP) over the period; a notion that we believe is particularly relevant in the context of financialisation where issues of co-movement and aggregate market behaviour play a central role. We define CHP as the cross-sectional average of individual CHPs for a particular pool of assets. i.e., period \(t\) CHP for asset pool \(x\): \[\mathbf{CHP}_{t}^{x}=\frac{1}{n}\sum_{j=1}^{n}CHP_{j,t}\] \(n\equiv\) number of commodities in set \(x\).
\(CHP_{i,t}\equiv\) period \(t\) CHP for commodity \(i\) as defined above.
We further define aggregate backwardation (backwardation) and contango (contango) CHP regimes as, for a given period, sub-periods with CHP levels below and above the period’s median respectively and, in contrast with (Hong and Yogo 2012) who aggregate across sub-sectors, our CHP construction draws on the full cross-section of US-traded assets considered in the study.
By relating returns on attribute based risk factor mimicking portfolios to returns on individual as well as group of assets, our time series approach therefore draws the link between attribute and pricing dynamics over different, theory-based market regimes, and thereby enables us to study market fundamentals and co-movement simultaneously, two key issues in the financialisation phenomenon.
Results & discussion
The results show a clear pattern of increase in average return during financialisation as illustrated by Figure 1, with most individual commodities showing a higher mean return, and Table 1 with an equally weighted portfolio of the US traded assets showing a mean returns of 15.9% significant at the 5% level vs. 6.9% with a 10% significance in the pre-financialisation period. The increase is even more pronounced for an equally weighted portfolio of GB traded assets (metals), 2.47% (not significant) to 21.8% (significant at the 10% level), which sets out the global nature of the phenomenon. The pattern of results for volatilities is similar although the increase is less pronounced with a weaker majority of individual commodities showing heightened volatility and both equally weighted portfolios showing an increase, more pronounced for the GB assets. Financialisation seems however to have impacted the commodity complex beyond these mechanical effects of increased long participation ensuing from the entry of index type investments as our results suggest next.
The cross-sectional nature of financialisation naturally leads us to extend the individual hedging pressure paradigm by incorporating aggregate CHP (CHP), a market wide measure of hedging pressure. The traditional Keynesian hedging pressure paradigm postulates a negative relationship between the return on an individual commodity futures and its hedging pressure. The extended Keynesian hedging pressure paradigm, in the context of financialisation, should also imply that returns for individual commodities should be high in periods where CHP is low and vice et versa.
Over the first period twenty three of the thirty commodity assets considered have a higher mean returns during periods of backwardation as compared to contango (Figure 2) including eighteen US and five GB traded assets. For the US commodities the difference is statistically significant at the 10% level for two assets, at the 5% level for one and at the 1% level for another; for the remaining six commodities the higher mean return during contango is not statistically significant for any of them. For the GB commodities, the difference is significant at the 10% level for one asset while for the remaining asset the higher mean return during contango is not statistically significant. An equally weighted portfolio of the twenty-four US commodities had a mean return of 15.4% during backwardation phases, statistically significant at the 1% level, and a mean return of -1.1% during contango, both in the first period, with the difference between them statistically significant at the 5% level. The corresponding figures for an equally weighted portfolio of the GB metals were 4.4% and 0.3% respectively. These results provide broad support for an aggregate version of the Keynesian hedging pressure hypothesis and suggest that hedgers in the aggregate were engaged in risk transfer during this period.
Over the second period the pattern essentially reverses with seventeen of the thirty commodities having higher mean returns during phases of contango over backwardation including fourteen US and four GB traded assets with the difference being statistically significant at the 5% level for one commodity. The equally weighted US portfolio had a mean return of 15.7% over backwardation against 16.2% during phases of contango, both statistically significant at the 10% level. The difference was considerably more pronounced for the equally weighted GB metals portfolio with a mean return of 8% during backwardation rising to 35.6% during contango.
The onset of financialisation thus seems to have engendered a change in the nature of hedger’s behaviour, at the aggregate level in particular, taking it away from the traditional Keynesian view of risk shifting. This phenomenon has been noted in earlier theoretical and empirical work (Danthine 1978; Stout 1998) in different contexts from which two competing models of behaviour have emerged.
The information arbitrage model (Grossman and Stiglitz 1980; Danthine 1978; Kyle 1989) implies that hedgers may often be seeking out counterparties to trade with rather than being purely passive. This issue is also raised in both Cheng and Xiong (2014) and Stulz (1996) who point out that hedgers may be taking a view on prices just as speculators do. In fact, by hedging away some of their risk, hedgers are able to speculate more heavily based on their disagreement against speculators regarding futures price movement (Simsek 2013), which fits into a heterogeneous expectation theory of speculation15. As Stout (1998) points out, this disagreement-based theory of trading is one of the main reason the public and the law disapproves speculation as this form of trading is regarded as non-productive16.
In the second model hedgers are seen as liquidity providers as in Kang, Rouwenhorst, and Tang (2020) with this explanation also consistent with the nature of financialisation which led to the arrival of long term, long only investors who require substantial liquidity to roll-over their positions. Table 2 supports this view. We regress commodity futures front returns on individual commercial hedging pressure (CHP) as well as aggregate CHP (CHP), our market wide measure of CHP, in separate models and find that the explanatory power of CHP at the individual commodity level decreases over the financialisation period with the US cross-sectional average \(R^{2}\) falling down to 20.8% from 25.8% in the previous period while that of CHP increases (7.1% vs 5.4%). Financialisation therefore seems to have had the effect of loosening the extent to which futures risk premiums relate to commodity specific commercial hedging motives while strengthening that to which they relate to broader dynamics such as cross-sectional liquidity provision. The pattern is strongest for the metals sector where the average \(R^{2}\) drops down to 19.1% from 28.1% between the two periods at the individual level while it rises from 8.4% to 15.7% at the aggregate level.
Financialisation was a global phenomenon, through the channel of index investment it simultaneously influenced the entire spectrum of liquid commodities, eventually spurring the transformation of commodity markets from a collection of idiosyncratic markets into a more cohesive collection akin to an asset class. In this context, interconnectedness emerges as a crucial aspect that warrants thorough examination.
We strive to address this issue by first examining the average returns pairwise correlations between commodities across the entire cross-section of the commodity assets considered in the study as well as by country, sector and sub-sector with the results shown in Table 3. For the entire sample the average pairwise correlation jumps from 0.07 in the first period to 0.16 over financialisation. This finding indicates higher interdependence within the full cross-section of commodities starting with the financialisation period, consistent with the empirical findings in Fry-McKibbin and McKinnon (2023) and Mayer, Rathgeber, and Wanner (2017) as well as with the theoretical model of Basak and Pavlova (2016). The sector analysis sheds light on the driving force behind these changes. The US metals see the sharpest increase in pairwise correlations going from 0.22 over the past period to 0.56 over financialisation while the three other sectors see a much smaller percentage increase during financialisation relative to the past period. These results thus suggest that the US metals sector was the driving force behind increased average pairwise correlations during financialisation, a new observation to the best of our knowledge. The pattern of correlations is very similar across phases of backwardation and contango for all the sectors and sub-sectors over the past and financialisation periods17. In this context, the ontological consequences of financialisation are better observed from an investment perspective as illustrated in Table 4 where we observe average pairwise correlation coefficients between equally weighted portfolios of US traded commodity assets grouped by sub-sector18. Although modest in magnitude, these exhibit a telling pattern, mirroring that shown by the individual commodity returns in Table 1, with a substantially higher average coefficient over backwardation phases in the first period (0.12 vs. 0.03) and the pattern reversed over financilisation with an average coefficient substantially higher in contango (0.12 vs. 0.02).
We break down Table 3’s average correlations by year in Figure 3. For the full sample of assets, the average pairwise correlations remain in the 0.06–0.08 range from 1998 through 2003, before suddenly increasing to 0.12 from 2003 to 2004. This suggests 2004 as a distinct break point for the onset of financialisation, as noted in several studies (Baker 2021; Tang and Xiong 2012). We further observe that the US metals appear to have played a major role in this as the increase seems driven by the sector which average pairwise correlation more than doubles (0.49 vs 0.22) over the period.
We investigate further the issue of co-movement from an asset pricing perspective using a bespoke asset pricing framework starting with a market factor mimicking portfolio constituted of the twenty-four US traded commodities with equal weighting. This portfolio enables a refined examination of cross-correlations across sectors as the numerator of the beta of the regression of the returns for an individual commodity against those for the market factor is driven by covariances while the denominator is driven by the individual commodity variance. Thus, an increase in a commodity’s correlation with other commodities should lead to an increase in its beta relative to the market factor portfolio. We accordingly regress the individual commodity front futures returns against those for our market factor portfolio and average the resulting betas across the entire set of commodity assets considered as well as by country, sector and sub-sector for each of the four periods of interest with the results shown in Table 5. The average beta for the US metals sector jumps sharply to 1.12 over financialisation from 0.63 over the previous period; the increase is even more dramatic for the base metals sub-sector (GB traded assets) in a relative sense. The group had a much lower average beta of 0.27 over the past period, suggesting a low level of correlation with US commodities; this figure increases to 0.85 over the financialisation period, even beyond that for the US agricultural sector of 0.80. In contrast, the corresponding figure for the energy sector shows a substantial decrease, down to 1.61 during financialisation from 2.33 over the past period.
The corresponding pricing performance displays a similar pattern as shown in Table 6. The average \(R^{2}\) for the market factor across the entire cross-section of commodity assets, including the GB traded base metals, increases to around 20% during financialisation compared to around 10% over the previous period with the rise driven by the metals sector at large. The sector average for the US traded assets jumps to 31% during financialisation from 7.5% over the past period while for the GB traded assets the corresponding figure increases from less than 2%, or virtually no explanatory power for the (US based) market factor, to almost 12% over the financialisation period. The results for our other risk factor mimicking portfolios show a similar pattern and provide a deeper insights into co-movement dynamics in this context. All three factors—CHP, open interest and term structure—performed poorly in the first period, indicating that the commodities with extreme factor attributes (high or low CHP for example) were uncorrelated with the rest of the commodity cross-section, and the overall performance consistent across periods of contango and backwardation19. The onset of financialisation leads to a substantial improvement in the performance of the CHP factor with its average \(R^{2}\), albeit still small at 5.4%, almost tripling from the past to the financialisation period in contrast to the other factors that show weaker dynamics with a slight decrease and a much smaller increase for the open interest and term structure factors respectively. Here again the increase in pricing performance for the CHP factor is driven by the metals sector, particularly the precious sub-sector with the average \(R^{2}\) for the group jumping to almost 26% during financialisation from 1.2% over the past period. The corresponding figures of 22% and 1.1% for the entire US metals sector stand in sharp contrast with the agricultural sector, which exhibits no notable change, and the energy sector, where pricing performance drops substantially from 9.2% down to 1.6%. The improvement in the CHP factor’s pricing performance remarkably comes predominantly from the long leg (low CHP assets: Keynesian “normal backwardation”) of the factor20 with the corresponding mimicking portfolio’s performance itself driven by the metals, the precious sub-sector in particular. Interestingly, although the term structure factor itself only sees a very modest increase, its short leg (low roll-yield assets: Working’s backwardation) shows a sizeable rise in pricing performance, yet again driven the metals sector. The pattern of results for both factors extends to the GB traded assets and suggest a convergence of Keneysian and Working’s term structure perspective over the financialisation period.
Taken together these results emphasise the global nature of the impact of financialisation on the metals complex suggest that there is information in the factor constructed from hedging pressure based on the Keynesian paradigm relevant to the term structure factor built on the Working paradigm. These findings therefore appear to complement the results of Kang, Rouwenhorst, and Tang (2020) who find that hedging pressure both conveys information about liquidity provision in the short term (Working paradigm) and risk transfer in the longer term (Keynesian paradigm).
The crisis period (late 2008 to mid-2013) shows a contrasting pattern of results. Returns fall sharply relative to the financialisation period (Figure 1) with twenty-four commodities showing lower mean returns, including eighteen US and six GB traded assets. Returns for equally weighted portfolios of both sets of assets (Table 1) fall down to 9.4% and 5.3% from 15.9% and 21.8% respectively. Returns are mostly higher during phases of backwardation relative to contango (Figure 2) with twenty commodities showing higher mean returns in backwardation, including fifteen US and five of the GB traded assets, and the pattern extending to equally weighted portfolios of the two sets of assets. Volatility is moderately higher with eighteen assets showing higher figures, fifteen US commodities and three GB metals; and interestingly eminently higher over phases of contango; out of the US complex only Sugar-#11 (IFUS) and none of the GB traded metals show higher volatility figures over backwardation phases. The symmetry between the dynamics of CHP and CHP’s explanatory power on individual commodity returns breaks (Table 2) with both now seemingly converging; the US cross-sectional average \(R^{2}\) at the individual level bouncing back up (22.5% vs 20.8%) while that at the aggregate level rises further (14.7% vs 7.1%).
The average pairwise return correlation for the entire set of sample assets increases further (up to 0.28: Table 3) and so does the average pairwise sub-sector portfolio correlation, albeit to a smaller level and in a smaller proportion (up to 0.08: Table 4). This indicates an intensification of the higher interdependence within the full cross-section of commodities prompted by financialisation consistent with the greater interdependence between equity and commodity markets over this period documented in Silvennoinen and Thorp (2013). The figure for the US traded section of the metals sector only progresses modestly to 0.59 after peaking at 0.63 in 2006 (Figure 3). In contrast, the GB traded assets see a sharp increase to over 0.65 starting in 2007 up to a peak of 0.69 in 2010, and so does the agricultural sector with the sharpest increase in 2008 when the average pairwise correlation for the sector jumps to 0.30. The energy sector also shows a substantial progression starting in 2007 and peaking at 0.63 in 2008 before leveling back down over the rest of the period. The onset of the financial crisis seems therefore to have had an across the board impact on the interdependence of commodities in contrast with financialisation, much more closely related to the metals sector.
The entire cross-section sees an increase in average regression coefficient relative to the market factor with the strongest progression for the GB traded metals followed by the agricultural sector (Table 5). The pricing performance of the factor dominates the US commodity complex (Table 6), showing a 50% increase in mean average \(R^{2}\) to over 31%, while that for both the CHP and open interest factor decreases and, although increasing, remains modest for the term structure factors; the increase is strongest for the GB traded metal assets with an average \(R^{2}\) of almost 34%. Taken together, these results indicate the presence of a systematic factor across the entire cross-section of the US commodities.
The returns pull-back continues into the post-crisis period with twenty six commodities (Figure 1), including twenty-two US and four GB traded assets, showing lower mean returns relative to the crisis while the same number of US commodities and all GB metals show lower figures relative to the financialisation period; mean returns for equally weighted portfolios of US and GB assets (Table 1) drop down to -0.2% and 2.7% respectively. Interestingly volatility also recedes substantially with twenty six commodities, including twenty US assets and all GB metals, showing lower volatility figures relative to both the crisis and financialisation periods suggesting that investment outflows from the commodity futures complex have had a lower impact on volatility than the corresponding inflows witnessed over the financialisation period. The figures for the equally weighted portfolios of US and GB assets drop to 9.9% and 16.2% respectively, very close to pre-financialisation levels.
The pattern of reversion toward the pre-financialisation Keynesian paradigm initiated during the crisis continues stronger into the post-crisis period with mean returns higher during phases of backwardation relative to contango (Figure 2), more so than over the crisis period with twenty three commodities showing higher mean returns over backwardation phases. Seventeen of those are US traded assets with the difference significant at the 5% level for Lumber (XCME) and at the 10% level for Palladium (XNYM) while no difference is significant for the assets showing higher figures over phases of contango. The pattern is more pronounced for the GB traded subset with all six assets showing higher mean returns in backwardation with the difference significant at the 1% level for Nickel-primary (XLME), 5% level for Aluminium (XLME) and 10% level for Zinc (XLME). It also extends to equally weighted portfolios of US and GB traded assets that both show higher mean returns during backwardation with the difference significant at the 5% level for the GB portfolio and the figure close to the 10% level significance threshold for the portfolio of US assets. Volatilities are overall higher in contango although less so than over the previous period with twenty one commodities showing higher figures over contango against twenty-eight in the crisis period, including seventeen US traded assets. The pattern extends to three of the six GB traded metals with the difference significant at the 1% level for Copper (XLME) and Tin (XLME) and at the 5% level for Zinc (XLME) while differences are not significant for Aluminium (XLME), Lead (XLME) and Nickel-primary (XLME) that exhibit higher volatility over backwardation phases. Equally weighted portfolios of US and GB traded assets on the other hand both show higher volatility in contango with the difference significant at the 1% level for both. Over the financialisation period in contrast volatilities were predominantly higher in backwardation for US traded assets with fourteen commodities showing higher volatility figures in these phases; the post-crisis pattern of results for volatilities thus likewise shows signs of reversion to pre-financialisation standards. After showing signs of convergence in the crisis period the dynamics of the explanatory power of CHP and CHP part ways (Table 2), seemingly reverting to the Keynesian paradigm observed pre-financialisation, further suggesting a return to legacy commodity specific dynamics. The recovery initiated in the crisis period for that of CHP is confirmed with a stronger increase, 27.2% vs. 22.5%, strongest for the metals sector, 37.3% vs. 23.7%, while the energy sector shows the opposite pattern: 11% vs. 16%. In contrast, after increasing stronger in the crisis, that for CHP falls overall post-crisis, 7.5% vs 14.7%, with the decrease strongest for the energy sector in a relative sense: 2.5% vs. 14.8%.
While the average sub-sector portfolio return correlation remain stable (Table 4), the average individual asset pairwise correlations fall across the board, below the levels observed during financialisation, with the figure for the entire cross-section (Table 3) dropping down to 0.12 from 0.28 over the crisis. This substantial decline, that does not seem to have been widely noted, appears to begin in the latter part of the crisis period, around 2012 (Figure 3), before leveling down over the whole post-crisis period.
Similarly, the average market factor beta for the entire cross-section drops down to 0.94 (Table 5), below the level observed during financialization. The decline seems driven by the metals sector, particularly the GB traded segment, with the figures for the US and GB traded assets at 0.91 and 0.72, down from 1.12 and 1.18, respectively. Meanwhile, the pricing performance of the market factor also recedes to below financialisation levels with the average \(R^{2}\) for the entire cross-section dropping down to 14.7% (Table 6). All sectors seem affected with agriculturals, energy and metals all seeing their average \(R^{2}\) decline below the levels observed during the financialisation period albeit remaining above 10% for the GB traded metals against less than 2% pre-financialisation. Similarly, the performance for both the CHP and term structure factors falls back to levels below those attained over financialisation. These results, along with those from the crisis period, suggest that the exogenous shock of the financial crisis, combined with the subsequent accommodative monetary policy, may have triggered a reversion toward pre-financialization fundamentals in the commodity futures complex; an issue that, to our knowledge, has not been widely discussed in the existing literature.
The phases analysis21 further documents the emerging role/dominance of the contango regime post-crisis with average correlations higher during those phases and the pattern stronger for the metal assets both US and GB traded. Similarly, for the latter the pricing performance of the market factor is considerably better during phase of contango in contrast with the previous periods that showed no substantial difference between phase of contango and backwardation. The figures for the US and GB traded segments of the sector are 23% (contango) vs 11.5% (backwardation) against 36.2% vs 43% during financialisation and 14.4% (contango) vs 6.6% (backwardation) against 33.5% vs 34.4% during financialisation respectively. Taken together, these results indicate some recent changes in the correlation structure within the sector and provide additional evidence that the interdependence effects of financialization seems to have been strongest for the entire cross-section of the metals sector with a lesser impact for the other sectors. This observation, along with the decline in interdependence in the commodity sector at large documented above, could form a valuable basis for future research.
Conclusion
In the early 2000s, against the backdrop of a low yield environment and poor stock market performance, a combination of financial innovations22 and regulatory changes23 led to large inflows of institutional capital into the commodity futures markets. This process known as “financialisation” spurred a heated public policy debate about whether the ensuing increase in open interest and trading volume in commodity futures exerted upward pressure on prices.
The debate spread to the legislative sphere, with the US senate launching formal investigations, and was thrust onto the academic community as a matter of urgency. Perhaps in response to this, most of the early studies focused on the more mechanical effects of financialisation in individual markets, relying on commodity specific causality and correlation-based analysis. The contention eventually triggered legislative action and new position limits were introduced in a number of grains and energy futures markets.
With the benefit of hindsight, this focus appears to have been too narrow. As our results suggests, financialisation was a phenomenon global in nature and studying the direct effects of increased market participation in single markets thus seems limited in this context.
The results of our analysis further suggest that financialisation had cross-sectional effects on market fundamentals across the commodity complex beyond energy and agricultural commodities and in fact strongest for the metals sector at the global level; an aspect largely overlooked in the original approach that we believe the existing literature has not fully analysed.
The onset of the financial crisis and the monetary policy regimes that followed, on the other hand, also appear to have induced significant changes in commodity pricing dynamics. In contrast with financialisation, our results suggest that the crisis and its aftermath have delivered an exogenous shock across the entire cross-section of liquid commodities which appears to have further altered the cross-sectional dynamics and eventually triggered a motion of reversion to legacy, pre-financialisation, fundamentals.
Figures
Tables
| asset | estimate | regime | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|---|---|
| US commodities | mean | whole period | *6.89% | **15.89% | 9.39% | -0.18% |
| backwardation | ***15.44% | *15.66% | 13.4% | 6.82% | ||
| contango | -1.09% | *16.16% | 4.9% | -6.33% | ||
| backwardation vs. contango | ** | |||||
| volatility | whole period | 10% | 13.62% | 17.58% | 9.86% | |
| backwardation | 9.88% | 14.21% | 14.83% | 8.59% | ||
| contango | 10.09% | 13.05% | 19.98% | 10.95% | ||
| backwardation vs. contango | ** | *** | *** | |||
| GB commodities | mean | whole period | 2.47% | *21.77% | 5.25% | 2.72% |
| backwardation | 4.41% | 8.01% | 15.06% | **19.95% | ||
| contango | 0.34% | *35.56% | -5.24% | -14.05% | ||
| backwardation vs. contango | ** | |||||
| volatility | whole period | 15.66% | 26.91% | 30.24% | 16.19% | |
| backwardation | 14.39% | 25.92% | 25.94% | 15.13% | ||
| contango | 16.87% | 27.88% | 34.06% | 17.1% | ||
| backwardation vs. contango | *** | * | *** | *** |
| regressor | sector | sub-sector | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|---|---|
| Δ% commodity CHP | all | all | 25.79% | 20.72% | 22.46% | 27.17% |
| agriculturals | all | 25.19% | 21.6% | 23.91% | 28.12% | |
| grains | 33.18% | 27.86% | 29.15% | 36.24% | ||
| livestock | 6.23% | 10.53% | 6.89% | 4.34% | ||
| softs | 26.68% | 20.89% | 27.19% | 31.88% | ||
| energy | all | 25.12% | 19.45% | 16% | 10.96% | |
| gas | 19.69% | 15.65% | 7.36% | 5.29% | ||
| petroleum | 26.93% | 20.71% | 18.87% | 12.85% | ||
| metals | all | 28.12% | 19.1% | 23.27% | 37.32% | |
| base | 47.38% | 25.86% | 18.07% | 34.9% | ||
| precious | 23.3% | 17.41% | 24.57% | 37.93% | ||
| Δ% aggregate CHP | all | all | 5.37% | 7.12% | 14.72% | 7.49% |
| agriculturals | all | 4.91% | 4.5% | 12.14% | 6.23% | |
| grains | 7.25% | 5.4% | 19.75% | 8.45% | ||
| livestock | 1.35% | 0.41% | 0.52% | 0.26% | ||
| softs | 4.34% | 5.64% | 10.35% | 6.98% | ||
| energy | all | 3.35% | 6.27% | 14.83% | 2.45% | |
| gas | 2.65% | 2.58% | 2.6% | 0.27% | ||
| petroleum | 3.59% | 7.49% | 18.9% | 3.17% | ||
| metals | all | 8.36% | 15.67% | 22.37% | 15.33% | |
| base | 10.12% | 12.74% | 23.94% | 7.67% | ||
| precious | 7.92% | 16.4% | 21.98% | 17.25% |
| country | sector | sub-sector | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|---|---|
| all | all | all | 0.0737 | 0.1638 | 0.2849 | 0.1282 |
| US | all | all | 0.0684 | 0.1545 | 0.2526 | 0.1096 |
| agriculturals | all | 0.0910 | 0.1305 | 0.2322 | 0.1075 | |
| grains | 0.4421 | 0.4898 | 0.5673 | 0.3728 | ||
| livestock | 0.3340 | 0.2868 | 0.3707 | 0.3333 | ||
| softs | 0.0391 | 0.0853 | 0.1584 | 0.0764 | ||
| energy | all | 0.4908 | 0.5892 | 0.4687 | 0.4203 | |
| petroleum | 0.7278 | 0.7846 | 0.7692 | 0.7175 | ||
| metals | all | 0.2167 | 0.5553 | 0.5877 | 0.4721 | |
| precious | 0.2883 | 0.6305 | 0.6704 | 0.5762 | ||
| GB | all | all | 0.4047 | 0.4691 | 0.6536 | 0.4412 |
| regime | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|
| whole period | 0.0421 | 0.0631 | 0.0799 | 0.0778 |
| backwardation | 0.1166 | 0.0201 | 0.0633 | 0.0727 |
| contango | 0.0297 | 0.1164 | 0.0889 | 0.0812 |
| country | sector | sub-sector | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|---|---|
| all | all | all | 0.8556 | 0.9709 | 1.0352 | 0.9438 |
| US | agriculturals | all | 0.7706 | 0.7965 | 0.8535 | 0.8252 |
| grains | 0.9891 | 1.1982 | 1.1727 | 1.0301 | ||
| livestock | 0.4320 | 0.2059 | 0.3317 | 0.5484 | ||
| softs | 0.7213 | 0.6900 | 0.7953 | 0.7589 | ||
| energy | all | 2.3308 | 1.6106 | 1.3977 | 1.7639 | |
| gas | 2.6992 | 1.7090 | 1.0044 | 1.2585 | ||
| petroleum | 2.2080 | 1.5778 | 1.5288 | 1.9324 | ||
| metals | all | 0.6320 | 1.1252 | 1.1212 | 0.9131 | |
| base | 0.4902 | 1.1178 | 1.4140 | 0.8780 | ||
| precious | 0.6674 | 1.1271 | 1.0480 | 0.9219 | ||
| GB | all | all | 0.2713 | 0.8520 | 1.1761 | 0.7192 |
| country | sector | sub-sector | factor | past | financialisation | crisis | post-crisis |
|---|---|---|---|---|---|---|---|
| all | all | all | market | 9.69% | 19.7% | 31.3% | 14.73% |
| CHP | 1.92% | 5.39% | 4.43% | 3.78% | |||
| open interest | 5.98% | 5.63% | 4.09% | 4.32% | |||
| term structure | 1.52% | 2.95% | 5.2% | 1.39% | |||
| US | agriculturals | all | market | 9.08% | 16.04% | 25.99% | 12.07% |
| CHP | 1.02% | 1.6% | 1.47% | 1.61% | |||
| open interest | 1.09% | 1.9% | 0.45% | 1% | |||
| term structure | 1.28% | 1.36% | 1.73% | 2.16% | |||
| grains | market | 16.44% | 29.28% | 42.09% | 19.27% | ||
| CHP | 0.6% | 1.37% | 0.78% | 2.98% | |||
| open interest | 2.22% | 4.18% | 0.67% | 2.04% | |||
| term structure | 1.78% | 1.37% | 2.04% | 2.08% | |||
| livestock | market | 2.96% | 1.83% | 8.99% | 4.84% | ||
| CHP | 1.63% | 2.06% | 3.36% | 1.23% | |||
| open interest | 0.43% | 0.55% | 0.18% | 0.34% | |||
| term structure | 0.38% | 1.25% | 1.15% | 1.79% | |||
| softs | market | 4.78% | 9.9% | 18.39% | 8.48% | ||
| CHP | 1.13% | 1.59% | 1.22% | 0.44% | |||
| open interest | 0.29% | 0.28% | 0.36% | 0.28% | |||
| term structure | 1.24% | 1.41% | 1.71% | 2.42% | |||
| all | all | market | 11.67% | 21.7% | 30.69% | 15.73% | |
| CHP | 2.38% | 5.8% | 5.39% | 4.47% | |||
| open interest | 7.45% | 6.95% | 4.57% | 5.29% | |||
| term structure | 1.89% | 2.87% | 5.12% | 1.71% | |||
| energy | all | market | 26.58% | 31.57% | 38.94% | 25.9% | |
| CHP | 9.13% | 1.61% | 3.5% | 4.79% | |||
| open interest | 40% | 33.46% | 21.16% | 27.67% | |||
| term structure | 3.41% | 2.99% | 5.5% | 1.86% | |||
| gas | market | 19.37% | 18.38% | 10.88% | 7.86% | ||
| CHP | 5.29% | 3.62% | 12.6% | 16.2% | |||
| open interest | 22.08% | 27.47% | 15.46% | 16.03% | |||
| term structure | 0.41% | 4.24% | 7.16% | 5.22% | |||
| petroleum | market | 28.98% | 35.97% | 48.3% | 31.91% | ||
| CHP | 10.41% | 0.94% | 0.47% | 0.99% | |||
| open interest | 45.97% | 35.46% | 23.06% | 31.56% | |||
| term structure | 4.41% | 2.57% | 4.95% | 0.74% | |||
| metals | all | market | 7.54% | 30.79% | 38.2% | 18.58% | |
| CHP | 1.07% | 21.76% | 18.67% | 12.79% | |||
| open interest | 0.5% | 0.91% | 3.68% | 0.27% | |||
| term structure | 2.48% | 7.31% | 14.99% | 0.24% | |||
| base | market | 5.39% | 23.13% | 50.77% | 19.74% | ||
| CHP | 0.57% | 5.78% | 0.57% | 0.6% | |||
| open interest | 0.18% | 1.01% | 2.78% | 0.46% | |||
| term structure | 0.06% | 9.72% | 9.63% | 0.24% | |||
| precious | market | 8.08% | 32.7% | 35.06% | 18.3% | ||
| CHP | 1.2% | 25.76% | 23.19% | 15.84% | |||
| open interest | 0.57% | 0.89% | 3.9% | 0.22% | |||
| term structure | 3.08% | 6.71% | 16.33% | 0.24% | |||
| GB | all | all | market | 1.74% | 11.71% | 33.71% | 10.75% |
| CHP | 0.08% | 3.74% | 0.57% | 1.02% | |||
| open interest | 0.07% | 0.34% | 2.17% | 0.42% | |||
| term structure | 0.04% | 3.28% | 5.5% | 0.09% |
Appendix
Commodity assets taxonomy
| asset | country | sector | sub-sector |
|---|---|---|---|
| Corn-#2 yellow (XCBT) | US | agriculturals | grains |
| Oats (XCBT) | |||
| Soybean meal (XCBT) | |||
| Soybean oil (XCBT) | |||
| Soybeans (XCBT) | |||
| Wheat-SRW (XCBT) | |||
| Cattle-feeder (XCME) | livestock | ||
| Cattle-live (XCME) | |||
| Lean hogs (XCME) | |||
| Cocoa (IFUS) | softs | ||
| Coffee-C (IFUS) | |||
| Cotton-#2 (IFUS) | |||
| Orange juice-frozen concentrated (IFUS) | |||
| Sugar-#11 (IFUS) | |||
| Lumber-random length (XCME) | |||
| Natural gas (XNYM) | energy | gas | |
| Crude oil-WTI (XNYM) | petroleum | ||
| Gasoline (XNYM) | |||
| Heating oil (XNYM) | |||
| Copper (XCEC) | metals | base | |
| Gold (XCEC) | precious | ||
| Silver (XCEC) | |||
| Palladium (XNYM) | |||
| Platinum (XNYM) | |||
| Aluminium-primary (XLME) | GB | metals | base |
| Copper (XLME) | |||
| Lead-refined pig (XLME) | |||
| Nickel-primary (XLME) | |||
| Tin-refined (XLME) | |||
| Zinc (XLME) |
References
Acharya, Viral V., Lars A. Lochstoer, and Tarun Ramadorai. 2013. “Limits to Arbitrage and Hedging: Evidence from Commodity Markets.” Journal of Financial Economics 109 (2): 441–65. https://doi.org/10.1016/j.jfineco.2013.03.003.
Asness, Clifford S, Andrea Frazzini, and Lasse Heje Pedersen. 2019. “Quality Minus Junk.” Review of Accounting Studies 24 (1): 34–112. https://doi.org/10.1007/s11142-018-9470-2.
Asness, Clifford, and Andrea Frazzini. 2013. “The Devil in HML’s Details.” The Journal of Portfolio Management 39 (4): 49–68. https://doi.org/10.3905/jpm.2013.39.4.049.
Asness, Clifford, Andrea Frazzini, Ronen Israel, and Tobias Moskowitz. 2015. “Fact, Fiction, and Value Investing.” The Journal of Portfolio Management 42 (1): 34–52. https://doi.org/10.3905/jpm.2015.42.1.034.
Baker, Steven D. 2021. “The Financialization of Storable Commodities.” Management Science 67 (1): 471–99. https://doi.org/10.1287/mnsc.2019.3445.
Bakshi, Gurdip, Xiaohui Gao, and Alberto G. Rossi. 2019. “Understanding the Sources of Risk Underlying the Cross Section of Commodity Returns.” Management Science 65 (2): 619–41. https://doi.org/10.1287/mnsc.2017.2840.
Basak, Suleyman, and Anna Pavlova. 2016. “A Model of Financialization of Commodities.” The Journal of Finance 71 (4): 1511–56. https://doi.org/10.1111/jofi.12408.
Basu, Devraj, and Joëlle Miffre. 2013. “Capturing the Risk Premium of Commodity Futures: The Role of Hedging Pressure.” Journal of Banking & Finance 37 (7): 2652–64. https://doi.org/10.1016/j.jbankfin.2013.02.031.
Bohl, Martin T., and Patrick M. Stephan. 2013. “Does Futures Speculation Destabilize Spot Prices? New Evidence for Commodity Markets.” Journal of Agricultural & Applied Economics 45 (4): 595–616. https://doi.org/10.1017/s1074070800005150.
Boons, Martijn, Frans De Roon, and Marta Szymanowska. 2012. “The Stock Market Price of Commodity Risk.” In AFA 2012 Chicago Meetings Paper. https://doi.org/10.2139/ssrn.1785728.
Boons, Martijn, and Melissa Porras Prado. 2019. “Basis-Momentum.” The Journal of Finance 74 (1): 239–79. https://doi.org/10.1111/jofi.12738.
Boyd, Naomi E., Bahattin Büyükşahin, Michael S. Haigh, and Jeffrey H. Harris. 2016. “The Prevalence, Sources, and Effects of Herding.” Journal of Futures Markets 36 (7): 671–94. https://doi.org/10.1002/fut.21756.
Brennan, Michael J. 1958. “The Supply of Storage.” The American Economic Review 48 (1): 50–72. https://doi.org/10.1007/978-1-349-02693-7.
Brunetti, Celso, Bahattin Büyükşahin, and Jeffrey H. Harris. 2016. “Speculators, Prices, and Market Volatility.” Journal of Financial & Quantitative Analysis 51 (5): 1545–74. https://doi.org/10.1017/s0022109016000569.
Brunetti, Celso, and David Reiffen. 2014. “Commodity Index Trading and Hedging Costs.” Journal of Financial Markets 21: 153–80. https://doi.org/10.1016/j.finmar.2014.08.001.
Bruno, Valentina G., Bahattin Büyükşahin, and Michel A. Robe. 2017. “The Financialization of Food?” American Journal of Agricultural Economics 99 (1): 243–64. https://doi.org/10.1093/ajae/aaw059.
Büyükşahin, Bahattin, and Jeffrey H. Harris. 2011. “Do Speculators Drive Crude Oil Futures Prices?” The Energy Journal, 167–202. https://doi.org/10.5547/ISSN0195-6574-EJ-Vol32-No.
Büyükşahin, Bahattin, and Michel A. Robe. 2014. “Speculators, Commodities and Cross-Market Linkages.” Journal of International Money & Finance, Understanding international commodity price fluctuations, 42 (April): 38–70. https://doi.org/10.1016/j.jimonfin.2013.08.004.
Carhart, Mark M. 1997. “On Persistence in Mutual Fund Performance.” The Journal of Finance 52 (1): 57–82. https://doi.org/10.2307/2329556.
Cheng, Ing-Haw, Andrei Kirilenko, and Wei Xiong. 2014. “Convective Risk Flows in Commodity Futures Markets.” Review of Finance 19 (5): 1733–81. https://doi.org/10.1093/rof/rfu043.
Cheng, Ing-Haw, and Wei Xiong. 2014. “Financialization of Commodity Markets.” Annual Review of Financial Economics 6 (1): 419–41. https://doi.org/10.1146/annurev-financial-110613-034432.
Christoffersen, Peter, Asger Lunde, and Kasper V. Olesen. 2014. “Factor Structure in Commodity Futures Return and Volatility.” Journal of Financial & Quantitative Analysis, 1–74. https://doi.org/10.1017/s0022109018000765.
Cootner, Paul H. 1960. “Returns to Speculators: Telser Versus Keynes.” Journal of Political Economy 68 (4): 396–404. https://doi.org/10.1086/258347.
Cortazar, Gonzalo, Ivo Kovacevic, and Eduardo S. Schwartz. 2013. “Commodity and Asset Pricing Models: An Integration.” NBER Working Papers 19167. National Bureau of Economic Research, Inc. https://ssrn.com/abstract=2287027.
Danthine, Jean-Pierre. 1978. “Information, Futures Prices, and Stabilizing Speculation.” Journal of Economic Theory 17 (1): 79–98. https://doi.org/10.1016/0022-0531(78)90124-2.
Daskalaki, Charoula, Alexandros Kostakis, and George Skiadopoulos. 2014. “Are There Common Factors in Individual Commodity Futures Returns?” Journal of Banking & Finance 40 (March): 346–63. https://doi.org/10.1016/j.jbankfin.2013.11.034.
De Schutter, Olivier. 2010. “Food Commodities Speculation and Food Price Crises: Regulation to Reduce the Risks of Price Volatility.” United Nations Special Rapporteur on the Right to Food Briefing Note 2: 1–14. https://www2.ohchr.org/english/issues/food/docs/briefing_note_02_september_2010_en.pdf.
Domanski, Dietrich, and Alexandra Heath. 2007. “Financial Investors and Commodity Markets.” Working Paper. https://www.bis.org/publ/qtrpdf/r_qt0703g.pdf.
Duffie, Darrell. 2014. “Challenges to a Policy Treatment of sSpeculative Trading Motivated by Differences in Beliefs.” The Journal of Legal Studies 43 (S2): S173–s182. https://doi.org/10.1086/677836.
Ederington, Louis, and Jae Ha Lee. 2002. “Who Trades Futures and How: Evidence from the Heating Oil Futures Market.” The Journal of Business 75 (2): 353–73. https://doi.org/10.1086/338706.
Ekeland, Ivar, Delphine Lautier, and Bertrand Villeneuve. 2019. “Hedging Pressure and Speculation in Commodity Markets.” Economic Theory 68: 83–123. https://doi.org/10.1007/s00199-018-1115-y.
Etula, Erkko. 2013. “Broker-Dealer Risk Appetite and Commodity Returns.” Journal of Financial Econometrics 11 (3): 486–521. https://doi.org/10.1093/jjfinec/nbs024.
Fama, Eugene F., and Kenneth R. French. 1993. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of Financial Economics 33 (1): 3–56. https://doi.org/10.1016/0304-405x(93)90023-5.
———. 2015. “A Five-Factor Asset Pricing Model.” Journal of Financial Economics 116 (1): 1–22. https://doi.org/10.1016/j.jfineco.2014.10.010.
Fattouh, Bassam, Lutz Kilian, and Lavan Mahadeva. 2013. “The Role of Speculation in Oil Markets: What Have We Learnt so Far?” The Energy Journal 34 (3): 7–33. https://doi.org/10.5547/01956574.34.3.2.
Fernandez-Perez, Adrian, Bart Frijns, Ana-Maria Fuertes, and Joelle Miffre. 2018. “The Skewness of Commodity Futures Returns.” Journal of Banking & Finance 86: 143–58. https://doi.org/10.1016/j.jbankfin.2017.06.015.
Frazzini, Andrea, and Lasse Heje Pedersen. 2014. “Betting Against Beta.” Journal of Financial Economics 111 (1): 1–25. https://doi.org/10.1016/j.jfineco.2013.10.005.
Fry-McKibbin, Renée, and Kate McKinnon. 2023. “The Evolution of Commodity Market Financialization: Implications for Portfolio Diversification.” Journal of Commodity Markets 32: 100360. https://doi.org/10.1016/j.jcomm.2023.100360.
Fuertes, Ana-Maria, Joëlle Miffre, and Adrian Fernandez-Perez. 2015. “Commodity Strategies Based on Momentum, Term Structure, and Idiosyncratic Volatility.” Journal of Futures Markets 35 (3): 274–97. https://doi.org/10.1002/fut.21656.
Gilbert, Christopher L. 2010a. “Speculative Influences on Commodity Futures Prices 2006-2008.” UNCTAD, Geneva. https://www.cftc.gov/sites/default/files/idc/groups/public/@swaps/documents/file/plstudy_14_cifrem.pdf.
———. 2010b. “How to Understand High Food Prices.” Journal of Agricultural Economics 61 (2): 398–425. https://doi.org/10.1111/j.1477-9552.2010.00248.x.
Gilbert, Christopher L., and Simone Pfuderer. 2014. “The Role of Index Trading in Price Formation in the Grains and Oilseeds Markets.” Journal of Agricultural Economics 65 (2): 303–22. https://doi.org/10.1111/1477-9552.12068.
Goldstein, Itay, Yan Li, and Liyan Yang. 2014. “Speculation and Hedging in Segmented Markets.” The Review of Financial Studies 27 (3): 881–922. https://doi.org/10.1093/rfs/hht059.
Goldstein, Itay, and Liyan Yang. 2022. “Commodity Financialization and Information Transmission.” The Journal of Finance 77 (5): 2613–67. https://doi.org/10.1111/jofi.13165.
Gorton, Gary B, Fumio Hayashi, and K Geert Rouwenhorst. 2012. “The Fundamentals of Commodity Futures Returns.” Review of Finance 17 (1): 35–105. https://doi.org/10.1093/rof/rfs019.
Grossman, Sanford J., and Joseph E. Stiglitz. 1980. “On the Impossibility of Informationally Efficient Markets.” The American Economic Review 70 (3): 393–408. https://doi.org/10.7916/d8765r99.
Hamilton, James D., and Jing Cynthia Wu. 2015. “Effects of Index-Fund Investing on Commodity Futures Prices.” International Economic Review 56 (1): 187–205. https://doi.org/10.1111/iere.12099.
Henderson, Brian J., Neil D. Pearson, and Li Wang. 2015. “New Evidence on the Financialization of Commodity Markets.” The Review of Financial Studies 28 (5): 1285–1311. https://doi.org/10.1093/rfs/hhu091.
Herman, Marc-Olivier, Ruth Kelly, and Robert Nash. 2011. “Not a Game, Speculation v. Food Security: Regulating Financial Markets to Grow a Better Future.” Oxfam Policy & Practice: Agriculture, Food & Land 11 (7): 127–38. https://www.ingentaconnect.com/content/oxpp/oppafl/2011/00000011/00000007/art00005.
Hicks, John Richard. 1939. Value and Capital. 1st ed. Oxford University Press, Cambridge.
———. 1946. Value and Capital. 2nd ed. Oxford University Press, Cambridge.
Hirshleifer, J. 1975. “Speculation and Equilibrium: Information, Risk, and Markets.” The Quarterly Journal of Economics 89 (4): 519–42. https://doi.org/10.2307/1884690.
Hirshleifer, Jack. 1976. “Reply to Comments on" Speculation and Equilibrium: Information, Risk, and Markets".” The Quarterly Journal of Economics, 689–96. https://doi.org/10.2307/1885330.
———. 1977. “The Theory of Speculation Under Alternative Regimes of Markets.” The Journal of Finance 32 (4): 975–99. https://doi.org/10.2307/2326507.
Hong, Harrison, and Motohiro Yogo. 2012. “What Does Futures Market Interest Tell Us about the Macroeconomy and Asset Prices?” Journal of Financial Economics 105 (3): 473–90. https://doi.org/10.1016/j.jfineco.2012.04.005.
Hou, Kewei, Chen Xue, and Lu Zhang. 2015. “Digesting Anomalies: An Investment Approach.” The Review of Financial Studies 28 (3): 650–705. https://doi.org/10.1093/rfs/hhu068.
Houthakker, H. S. 1957. “Can Speculators Forecast Prices?” The Review of Economics & Statistics 39 (2): 143–51. https://doi.org/10.2307/1928531.
Houthakker, Hendrik. 1957. “Restatement of the Theory of Normal Backwardation.” Cowles Foundation Discussion Papers 44. Cowles Foundation for Research in Economics, Yale University. https://EconPapers.repec.org/RePEc:cwl:cwldpp:44.
Irwin, Scott H. 2013. “Commodity Index Investment and Food Prices: Does the "Masters Hypothesis" Explain Recent Price Spikes?” Agricultural Economics 44 (s1): 29–41. https://doi.org/10.1111/agec.12048.
Irwin, Scott H., and Dwight R. Sanders. 2011. “Index Funds, Financialization, and Commodity Futures Markets.” Applied Economic Perspectives & Policy 33 (1): 1–31. https://doi.org/10.1093/aepp/ppq032.
———. 2012. “Financialization and Structural Change in Commodity Futures Markets.” Journal of Agricultural & Applied Economics 44 (3): 371–96. https://doi.org/10.1017/s1074070800000481.
Irwin, Scott H., Dwight R. Sanders, and Robert P. Merrin. 2009. “Devil or Angel? The Role of Speculation in the Recent Commodity Price Boom (and Bust).” Journal of Agricultural & Applied Economics 41 (2): 377–91. https://doi.org/10.1017/s1074070800002856.
Isleimeyyeh, Mohammad. 2020. “The Role of Financial Investors in Determining the Commodity Futures Risk Premium.” Journal of Futures Markets 40 (9): 1375–97. https://doi.org/10.1002/fut.22122.
Juvenal, Luciana, and Ivan Petrella. 2015. “Speculation in the Oil Market.” Journal of Applied Econometrics 30 (4): 621–49. https://doi.org/10.1002/jae.2388.
Kaldor, Nicholas. 1939. “Speculation and Economic Stability.” The Review of Economic Studies 7 (1): 1–27. https://doi.org/10.2307/2967593.
Kang, Wenjin, K Geert Rouwenhorst, and Ke Tang. 2020. “A Tale of Two Premiums: The Role of Hedgers and Speculators in Commodity Futures Markets.” The Journal of Finance 75 (1): 377–417. https://doi.org/10.1111/jofi.12845.
Keynes, John Maynard. 1923. “Some Aspects of Commodity Markets.” Manchester Guardian Commercial: European Reconstruction Series 13: 784–86.
———. 1930. Treatise on Money. Macmillan, London.
Kilian, Lutz, and Daniel P. Murphy. 2014. “The Role of Inventories and Speculative Trading in the Global Market for Crude Oil.” Journal of Applied Econometrics 29 (3): 454–78. https://doi.org/10.1002/jae.2322.
Kim, Abby. 2015. “Does Futures Speculation Destabilize Commodity Markets?” Journal of Futures Markets 35 (8): 696–714. https://doi.org/10.1002/fut.21716.
Knittel, Christopher R., and Robert S. Pindyck. 2016. “The Simple Economics of Commodity Price Speculation.” American Economic Journal: Macroeconomics 8 (2): 85–110. https://doi.org/10.1257/mac.20140033.
Kyle, Albert S. 1989. “Informed Speculation with Imperfect Competition.” The Review of Economic Studies 56 (3): 317–55. https://doi.org/10.2307/2297551.
Leclercq, Emmanuel, and Remy Praz. 2014. “Equilibrium Commodity Trading.” SSRN Scholarly Paper Id 2464400. Rochester, NY: Social Science Research Network. https://doi.org/10.2139/ssrn.2464400.
Manera, Matteo, Marcella Nicolini, and Ilaria Vignati. 2016. “Modelling Futures Price Volatility in Energy Markets: Is There a Role for Financial Speculation?” Energy Economics, Energy markets, 53 (January): 220–29. https://doi.org/10.1016/j.eneco.2014.07.001.
Masters, M. W., and A. K. White. 2008. “The Accidental Hunt Brothers: How Institutional Investors Are Driving up Food and Energy Prices.” Working Paper.
Masters, Michael W. 2008. “Testimony Before the Committee on Homeland Security and Governmental Affairs.” U.S. Senate 20 (May).
Mayer, Herbert, Andreas Rathgeber, and Markus Wanner. 2017. “Financialization of Metal Markets: Does Futures Trading Influence Spot Prices and Volatility?” Resources Policy 53: 300–316. https://doi.org/10.1016/j.resourpol.2017.06.011.
Miffre, Joelle, and Georgios Rallis. 2007. “Momentum Strategies in Commodity Futures Markets.” Journal of Banking & Finance 31 (6): 1863–86. https://doi.org/10.1016/j.jbankfin.2006.12.005.
Moskowitz, Tobias J., Yao Hua Ooi, and Lasse Heje Pedersen. 2012. “Time Series Momentum.” Journal of Financial Economics, Special issue on investor sentiment, 104 (2): 228–50. https://doi.org/10.1016/j.jfineco.2011.11.003.
Sakkas, Athanasios, and Nikolaos Tessaromatis. 2020. “Factor Based Commodity Investing.” Journal of Banking & Finance 115: 105807. https://doi.org/10.1016/j.jbankfin.2020.105807.
Sanders, Dwight R., and Scott H. Irwin. 2011a. “The Impact of Index Funds in Commodity Futures Markets: A Systems Approach.” The Journal of Alternative Investments 14 (1): 40–49. https://doi.org/10.3905/jai.2011.14.1.040.
———. 2011b. “New Evidence on the Impact of Index Funds in u.s. Grain Futures Markets.” Canadian Journal of Agricultural Economics/Revue Canadienne d’agroeconomie 59 (4): 519–32. https://doi.org/10.1111/j.1744-7976.2011.01226.x.
Sanders, Dwight R., Scott H. Irwin, and Robert P. Merrin. 2010. “The Adequacy of Speculation in Agricultural Futures Markets: Too Much of a Good Thing?” Applied Economic Perspectives and Policy 32 (1): 77–94. https://doi.org/10.1093/aepp/ppp006.
Schumann, Harald. 2011. “The Hunger-Makers: How Deutsche Bank, Goldman Sachs and Other Financial Institutions Are Speculating with Food at the Expense of the Poorest.” Foodwatch, Berlin. https://www.foodwatch.org/uploads/media/foodwatchreport_TheHungerMakers_observationsandcallsforaction_ger_03.pdf.
Schwartz, Eduardo S. 1997. “The Stochastic Behavior of Commodity Prices: Implications for Valuation and Hedging.” The Journal of Finance 52 (3): 923–73. https://doi.org/10.1111/j.1540-6261.1997.tb02721.x.
Schwartz, Eduardo, and James E Smith. 2000. “Short-Term Variations and Long-Term Dynamics in Commodity Prices.” Management Science 46 (7): 893–911. https://doi.org/10.1287/mnsc.46.7.893.12034.
Senate, U. S. 2009. “Excessive Speculation in the Wheat Market.” Majority & Minority Staff Report. Permanent Subcommittee on Investigations 24: 107–8. https://www.hsgac.senate.gov/imo/media/doc/REPORTExcessiveSpecullationintheWheatMarketwoexhibitschartsJune2409.pdf?attempt=2.
Silvennoinen, Annastiina, and Susan Thorp. 2013. “Financialization, Crisis and Commodity Correlation Dynamics.” Journal of International Financial Markets, Institutions & Money 24: 42--65. https://doi.org/10.1016/j.intfin.2012.11.007.
Simsek, Alp. 2013. “Speculation and Risk Sharing with New Financial Assets.” The Quarterly Journal of Economics 128 (3): 1365–96. https://doi.org/10.1093/qje/qjt007.
Singleton, Kenneth J. 2013. “Investor Flows and the 2008 Boom/Bust in Oil Prices.” Management Science 60 (2): 300–318. https://doi.org/10.1287/mnsc.2013.1756.
Sockin, Michael, and Wei Xiong. 2015. “Informational Frictions and Commodity Markets.” The Journal of Finance 70 (5): 2063–98. https://doi.org/10.1111/jofi.12261.
Stoll, Hans R., and Robert E. Whaley. 2011. “Commodity Index Investing: Speculation or Diversification?” The Journal of Alternative Investments 14 (1): 50–60. https://doi.org/10.2139/ssrn.1633908.
Stout, Lynn A. 1998. “Why the Law Hates Speculators: Regulation and Private Ordering in the Market for OTC Derivatives.” Duke Law Journal 48: 701–86. https://doi.org/10.2307/1373070.
Stulz, René M. 1996. “Rethinking Risk Management.” Journal of Applied Corporate Finance 9 (3): 8–25. https://doi.org/10.1111/j.1745-6622.1996.tb00295.x.
Szymanowska, Marta, Frans De Roon, Theo Nijman, and Rob Van Den Goorbergh. 2014. “An Anatomy of Commodity Futures Risk Premia.” The Journal of Finance 69 (1): 453–82. https://doi.org/10.1111/jofi.12096.
Tang, Ke, and Wei Xiong. 2012. “Index Investment and the Financialization of Commodities.” Financial Analysts Journal 68 (5): 54–74. https://doi.org/10.2469/faj.v68.n6.5.
Telser, Lester G. 1958. “Futures Trading and the Storage of Cotton and Wheat.” Journal of Political Economy 66 (3): 233–55. https://doi.org/10.1086/258036.
Till, Hilary. 2007. “A Long-Term Perspective on Commodity Futures Returns.” In Intelligent Commodity Investing. London: Risk books.
Tokic, Damir. 2012. “Speculation and the 2008 Oil Bubble: The DCOT Report Analysis.” Energy Policy 45 (June): 541–50. https://doi.org/10.1016/j.enpol.2012.02.069.
Turnovsky, Stephen J. 1983. “The Determination of Spot and Futures Prices with Storable Commodities.” Econometrica 51 (5): 1363–87. https://doi.org/10.2307/1912279.
UNCTAD, Secretariat Task Force on Systemic Issues, and Economic Cooperation. 2009. “The Global Economic Crisis: Systemic Failures and Multilateral Remedies.” UNCTAD, Geneva. https://unctad.org/system/files/official-document/gds20091_en.pdf.
Villeneuve, Bertrand, Delphine Lautier, and Ivar Ekeland. 2014. “Speculation in Commodity Futures Markets: A Simple Equilibrium Model.” In Séminaire Hotelling (RITM – ENS CACHAN), 37. Cachan, France. https://hal.science/hal-01655848.
Weymar, F. Helmut. 1966. “The Supply of Storage Revisited.” The American Economic Review 56 (5): 1226–34. http://www.jstor.org/stable/1815306.
Working, Holbrook. 1933. “Price Relations Between July and September Wheat Futures at Chicago Since 1885.” Wheat Studies 9 (1388-2016-116727): 187. https://doi.org/10.22004/ag.econ.142876.
———. 1948. “Theory of the Inverse Carrying Charge in Futures Markets.” Journal of Farm Economics 30 (1): 1–28. https://doi.org/10.2307/1232678.
———. 1949. “The Theory of Price of Storage.” The American Economic Review, 1254–62. https://www.jstor.org/stable/1816601.
———. 1953. “Hedging Reconsidered.” Journal of Farm Economics 35 (4): 544–61. https://doi.org/10.2307/1233368.
Yang, Fan. 2013. “Investment Shocks and the Commodity Basis Spread.” Journal of Financial Economics 110 (1): 164–84. https://doi.org/10.1016/j.jfineco.2013.04.012.
Footnotes
Starting around 2004, with a view of commodity markets as an emerging asset class and in an effort to hedge against inflation and seek diversification benefits (Büyükşahin and Robe 2014; Singleton 2013), institutional investors sent forth an unprecedented flow of capital into commodity futures markets. This phenomenon is commonly referred to as the financialisation of commodities (Domanski and Heath 2007).↩︎
See section Section 2.1 for a detailed discussion.↩︎
Starting point based on earlier studies (Baker 2021; Christoffersen, Lunde, and Olesen 2014).↩︎
Michael W. Masters (2008); M. W. Masters and White (2008); UNCTAD and Cooperation (2009); De Schutter (2010); Gilbert (2010b); Gilbert (2010a); Herman, Kelly, and Nash (2011); Schumann (2011); Singleton (2013).↩︎
Etula (2013), Acharya, Lochstoer, and Ramadorai (2013), Cheng, Kirilenko, and Xiong (2014), Leclercq and Praz (2014), Sockin and Xiong (2015), Goldstein, Li, and Yang (2014), Villeneuve, Lautier, and Ekeland (2014), Goldstein and Yang (2022), Ekeland, Lautier, and Villeneuve (2019), Isleimeyyeh (2020).↩︎
ISDA & SIFMA v US CFTC; Complaint, 1:11-cv-02146; December 2nd, 2011.↩︎
Understood as the market activity peculiar to index type investors.↩︎
See Table 7 for a detailed taxonomy of the commodity assets considered in the study.↩︎
Current reporting level thresholds available at: www.ecfr.gov.↩︎
CFTC form 40. Available at: www.cftc.gov.↩︎
See CFTC regulation 1.3, 17 CFR 1.3(z) for details.↩︎
Aggregate CHP or CHP: see below for a formal definition and description of the phase construction process.↩︎
A detailed breakdown of these results is provided in the attached Internet Appendix.↩︎
J. Hirshleifer (1975); Jack Hirshleifer (1976); Jack Hirshleifer (1977)↩︎
Duffie (2014) also discusses some of the challenges faced by a policy treatment of speculative trading motivated by differences in beliefs.↩︎
A breakdown of the average pairwise correlations by regime is provided in the attached Internet Appendix.↩︎
A more detailed breakdown of country, sector and subsector average pairwise correlations is provided in the attached Internet Appendix.↩︎
A breakdown of the performance by regime is provided in the attached Internet Appendix.↩︎
A breakdown of the performance by factor leg is provided in the attached Internet Appendix.↩︎
Results available in the attached Internet Appendix.↩︎
Commodity price indexes and the ad-hoc financial instruments enabling investment in them were the main financial innovations which allowed large global banks to offer commodity investment products to institutional and retail investors. In 1991 Goldman Sachs created the S&P-GSCI which provides investors with buy-side exposure to commodities via the OTC swap market and thus without having to participate in the formal futures markets with their position limit restrictions. At this stage, these restrictions still applied to the issuing institutions though, as they hedged the corresponding commodity swap exposure in the futures markets. The first commodity-based ETFs were created through buying physical precious metals with gold and silver ETFs offered as early as 2002/2003. The regulatory hurdle here related to the licensing of commodity trading professionals. Typically, investors had to sign a statement with their broker stating that they understood the risks of commodity investments; a rather inconvenient paperwork for a product designed to trade like a stock, as set forth by a number of industry players at the time.↩︎
In 2000, the Commodity Futures Modernization Act (CFMA) granted non-agricultural commodity futures statutory exemption from regulation (“Enron loophole”). It still required agricultural commodity derivatives be traded on a CFTC-regulated exchange however. Eventually, the CFTC classified swap dealers as “bona fide” hedgers, granting them position limits exemption (“swap dealer loophole”). In 2005, the CFTC waived the rule that required commodity investors to sign a statement saying they understood the risks, letting the funds replace it with their prospectus. The regulatory bottleneck that prevented the large-scale expansion of commodity index investment was no more.↩︎