index funds, financialization, and commodity futures markets

31
Featured Article Index Funds, Financialization, and Commodity Futures Markets Scott H. Irwin* and Dwight R. Sanders Scott H. Irwin is the Lawrence J. Norton Chair of Agricultural Marketing, Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign; Dwight R. Sanders is a professor in the Department of Agribusiness Economics, Southern Illinois University Carbondale. *Correspondence may be sent to: E-mail: [email protected]. Submitted March 2009; accepted December 2010. Abstract Some market participants and policy-makers believe that index fund investment was a major driver of the 2007-2008 spike in commodity futures prices. One group of empirical studies does find evidence that commodity index investment had an impact on the level of futures prices. However, the data and methods used in these studies are subject to criticisms that limit the confidence one can place in their results. Moreover, another group of studies provides no sys- tematic evidence of a relationship between positions of index funds and the level of commodity futures prices. The lack of a direct empirical link between index fund trading and commodity futures prices casts considerable doubt on the belief that index funds fueled a price bubble. Key words: Index funds, commodity, futures markets, prices, specu- lation, bubble. JEL Codes: D84, G12, G13, G14, Q13, Q41. Introduction The financial industry has developed new products that allow insti- tutions and individuals to invest in commodities through long-only index funds, over-the-counter (OTC) swap agreements, exchange traded funds, and other structured products. Regardless of form, these instruments have a common goal – to provide investors with buy-side exposure to returns from a particular index of commodity prices. 1 Several influential academic studies concluded that investors can capture substantial risk premiums and reduce portfolio risk through rela- tively modest investments in long-only commodity index funds (e.g., # The Author(s) 2011. Published by Oxford University Press, on behalf of Agricultural and Applied Economics Association. All rights reserved. For permissions, please email: [email protected]. 1 In the remainder of this article, the term “commodity index fund” or “index fund” is used generically to refer to all of the varied long-only commodity investment instruments. See the discussion in the fol- lowing section for further details. Applied Economic Perspectives and Policy (2011) volume 33, number 1, pp. 1–31. doi:10.1093/aepp/ppq032 1 at Pennsylvania State University on February 27, 2013 http://aepp.oxfordjournals.org/ Downloaded from

Upload: d-r

Post on 04-Dec-2016

221 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Index Funds, Financialization, and Commodity Futures Markets

Featured Article

Index Funds, Financialization, and CommodityFutures Markets

Scott H. Irwin* and Dwight R. Sanders

Scott H. Irwin is the Lawrence J. Norton Chair of Agricultural Marketing,Department of Agricultural and Consumer Economics, University of Illinois atUrbana-Champaign; Dwight R. Sanders is a professor in the Department ofAgribusiness Economics, Southern Illinois University Carbondale.

*Correspondence may be sent to: E-mail: [email protected].

Submitted March 2009; accepted December 2010.

Abstract Some market participants and policy-makers believe that index fundinvestment was a major driver of the 2007-2008 spike in commodity futuresprices. One group of empirical studies does find evidence that commodity indexinvestment had an impact on the level of futures prices. However, the data andmethods used in these studies are subject to criticisms that limit the confidenceone can place in their results. Moreover, another group of studies provides no sys-tematic evidence of a relationship between positions of index funds and the levelof commodity futures prices. The lack of a direct empirical link between indexfund trading and commodity futures prices casts considerable doubt on the beliefthat index funds fueled a price bubble.

Key words: Index funds, commodity, futures markets, prices, specu-lation, bubble.

JEL Codes: D84, G12, G13, G14, Q13, Q41.

Introduction

The financial industry has developed new products that allow insti-tutions and individuals to invest in commodities through long-only indexfunds, over-the-counter (OTC) swap agreements, exchange traded funds,and other structured products. Regardless of form, these instruments havea common goal – to provide investors with buy-side exposure to returnsfrom a particular index of commodity prices.1

Several influential academic studies concluded that investors cancapture substantial risk premiums and reduce portfolio risk through rela-tively modest investments in long-only commodity index funds (e.g.,

# The Author(s) 2011. Published by Oxford University Press, on behalf of Agricultural and AppliedEconomics Association. All rights reserved. For permissions, please email:[email protected].

1In the remainder of this article, the term “commodity index fund” or “index fund” is used genericallyto refer to all of the varied long-only commodity investment instruments. See the discussion in the fol-lowing section for further details.

Applied Economic Perspectives and Policy (2011) volume 33, number 1, pp. 1–31.doi:10.1093/aepp/ppq032

1

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 2: Index Funds, Financialization, and Commodity Futures Markets

Gorton and Rouwenhorst 2006; Erb and Harvey 2006). Combined with theavailability of deep and liquid exchange-traded futures contracts, theseconclusions set off a dramatic surge in commodity index fund invest-ments. While there is uncertainty about the best way to measure the mag-nitude of this surge, it is safe to say that at least $100 billion of newinvestment moved into commodity futures markets between 2004 and2008. Domanski and Heath (2007) term this the “financialization” of com-modity futures markets. Given the size and scope of the index fund boom,it should probably not come as a surprise that a worldwide debate hasensued over the role of index funds in commodity markets. This debatehas important ramifications from a policy and regulatory perspective, aswell as practical implications for the efficient pricing of commodityproducts.

On one side, a number of hedge fund managers, commodity end-users,policy-makers, and some economists contend that commodity indexinvestment was a major driver of the 2007-2008 spike in commodityfutures prices (e.g., Masters 2008; Masters and White 20082; USS/PSI 2009;De Schutter 2010; Baffes and Haniotis 2010). The massive wave of indexfund buying allegedly created a “bubble” that forced commodity futuresprices well above “fundamental values.” The following statement by FadelGheit, Managing Director and Senior Oil Analyst for Oppenheimer &Company, Inc., before the Subcommittee on Oversight and Investigationsof the Committee on Energy and Commerce, U.S. House ofRepresentatives, in June 2008 illustrates this perspective:

“I firmly believe that the current record oil price in excess of $135 per barrel isinflated. I believe, based on supply and demand fundamentals, crude oil pricesshould not be above $60 per barrel . . . There were no unexpected changes in industryfundamentals in the last 12 months, when crude oil prices were below $65 per barrel.I cannot think of any reason that explains the run-up in crude oil price, beside exces-sive speculation,” (Gheit 2008, p.2).

This line of reasoning fueled political pressure to limit speculative pos-itions in commodity futures markets, particularly in energy futuresmarkets. The recently-passed 2010 Dodd-Frank Wall Street Reform andConsumer Protection Act provided regulators with the authority needed toimplement such limits. The Commodity Futures Trading Commission(CFTC) now has broad authority to set aggregate speculative positionlimits on futures and swap positions in all non-exempt “physical com-modity markets” in the United States.

On the other side, a number of market analysts and economists haveexpressed skepticism about the bubble argument, citing logical inconsis-tencies and contrary facts (e.g., Krugman 2008; Pirrong 2008; Sanders andIrwin 2008; Smith 2009). This group has argued that commodity marketsin 2007-2008 were driven by fundamental supply and demand factors thatpushed prices higher not “excessive speculation.” Factors cited as drivingthe price of crude oil include strong demand from China, India, and otherdeveloping nations, a leveling out of crude oil production, a decrease inthe responsiveness of consumers to price increases, and U.S. monetarypolicy (e.g., Hamilton 2009a; Kilian and Murphy 2010). In the grainmarkets, the diversion of row crops to bio-fuel production and weather-

2Masters, Michael W., and Adam K. White. 2008. The Accidental Hunt Brothers: How InstitutionalInvestors are Driving up Food and Energy Prices. http://accidentalhuntbrothers.com/.

Applied Economic Perspectives and Policy

2

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 3: Index Funds, Financialization, and Commodity Futures Markets

related production shortfalls are cited, as well as demand growth fromdeveloping nations, U.S. monetary policy, and a lack of investment inbasic production infrastructure (e.g., Trostle 2008; Wright 2009).

Even though over two years have passed since the 2008 peak in commodityprices, the controversy surrounding index funds continues unabated. Forinstance, Michael Masters, Portfolio Manager for Masters Capital Management,LLC, made the following statement at a March 2010 CFTC hearing:

“Passive speculators are an invasive species that will continue to damage the marketsuntil they are eradicated. The CFTC must address the issue of passive speculation; itwill not go away on its own. When passive speculators are eliminated from the markets,then most consumable commodities derivatives markets will no longer be excessivelyspeculative, and their intended functions will be restored,” (Masters 2010, p.5).

Joachim von Braun, director of Germany’s Center for DevelopmentResearch, made these comments at a March 2010 conference linking highfood prices and world hunger:

“We have good analysis that speculation played a role in 2007 and 2008 . . .Speculation did matter and it did amplify, that debate can be put to rest. Thesespikes are not a nuisance, they kill. They’ve killed thousands of people,”3.

These statements illustrate the acrimonious and heated nature of thepublic policy debate surrounding the role of index funds in commodityfutures markets.

The purpose of this article is to provide a thorough review and synthesisof the arguments on both sides of the debate about the impact of index fundsin commodity futures markets, as well as a careful assessment of the latestempirical research on the subject. We consider empirical studies that test forgeneral bubble-like price activity over the 2007-2008 period, as well as thosethat specifically link market behavior to index fund positions. This “takingstock” is important given the stakes involved in the outcome of the debate. Ifindex fund investment was responsible for a large bubble in commodityfutures prices, difficult questions are raised about the basic price discoveryand risk-shifting functions of these markets. New regulatory limits on specu-lation would likely be justified even if costly to some market participants. Ifsupply and demand fundamentals rather than index investment wereresponsible for the commodity price spike, then new limits on speculationwould not be justified and could do substantial harm to the efficient func-tioning of these markets. Before delving into the arguments and evidence,we provide an overview of commodity index funds in the next section.

Commodity Index Funds

Commodity index investments share the common goal of tracking thebroad movement of commodity prices. The Standard and Poor’s-GoldmanSachs Commodity IndexTM (S&P-GSCI), is one of the most widely trackedindices and is generally considered an industry benchmark; it is computedas a (quantity) production-weighted average of the prices from 24 com-modity futures markets. While the index is well-diversified in terms ofnumber of markets and sectors, the production-weighting results in a

3As quoted in: Ruitenberg, Rudy. 2010. Global Food Reserve Needed to Stabilize Prices, ResearchersSay. Bloomberg.com. March 29. http://www.bloomberg.com/apps/news?pid=newsarchive&sid=au9X.0u6VpF0.

Index Funds and Futures

3

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 4: Index Funds, Financialization, and Commodity Futures Markets

relatively large 67% weight towards the energy sector, and a fairly small21% in traditional livestock and agricultural markets.4 The other industrybenchmark is the Dow Jones-UBS Commodity IndexTM (DJ-UBS). TheDJ-UBS market weights are based on a combination of economic signifi-cance and market liquidity, with a maximum weight of 33% in any sector.Energy markets receive a weight of 28%, and the livestock and agriculturalsectors combine to 40% of the DJ-UBS index.5 In both of these popularindices, the market weights, contract switching or rollover conventions,and contract months traded are well-publicized, resulting in a transparentand well-defined index.

Investors can gain exposure to commodity indices through a number ofinvestment vehicles, but the positions often find their way to the commodityfutures markets.6 A minority of institutions and individuals may gain com-modity exposure by directly purchasing futures contracts in a manner thatmimics a popular index. However, this number is relatively small, as mostinstitutions are barred from directly trading futures (Stoll and Whaley 2010),and most individuals would have difficulty replicating a broad-based index.As an alternative to directly purchasing futures, institutions often invest in afund that promises to mimic a popular commodity index. The fund managerwill then either directly invest in futures or gain the promised marketexposure by entering an over-the-counter (OTC) swap contract with a swapdealer. Swaps and other OTC derivatives are popular because they can be tai-lored by a swap dealer to meet the specific needs of a client. An index swap isa derivative contract structured to provide a payoff that is indexed to one ofthe popular commodity indices (see Chapter 5, Hull (2000), for examples ofswap contracts). The swap dealer will in turn enter the futures market andtake long positions in the corresponding futures contracts to offset the riskassociated with their side of the OTC derivative. As an alternative, insti-tutional investors may choose to bypass the fund and enter directly into acommodity return swap with a swap dealer. Again, the swap dealer will bethe agent who actually takes the long positions in the futures markets.

For individual investors, investment firms offer funds whose returns aretied to a commodity index. Both exchange-traded funds (ETFs) and struc-tured notes (ETNs) have been developed that track commodity indices.ETFs are essentially mutual fund shares that trade on a stock exchangeand are designed such that the share price tracks a designated commodityindex. ETN’s are actually debt securities where the issuer promises tomake pay-outs based on the value of the underlying commodity index.Both ETFs and ETNs trade on exchanges in the same manner as stocks onindividual companies. The management company that initially offers andmanages the fund or note collects a fee for their services. To gain com-modity exposure, ETF and ETN managers can either buy futures contractsdirectly, or more likely, utilize OTC commodity return swaps. The swapdealer will subsequently purchase commodity futures contracts to hedgetheir commodity exposure related to the swap transactions.

4As of October 15, 2010. (http://www2.goldmansachs.com/services/securities/products/sp-gsci-commodity-index/index.html).5As of September 30, 2010. (http://www.djindices.com/mdsidx/downloads/fact_info/Dow_Jones-UBS_Commodity_Index_Fact_Sheet.pdf).6See Engelke and Yuen (2008) and Stoll and Whaley (2010) for further details on the various commod-ity index investment instruments.

Applied Economic Perspectives and Policy

4

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 5: Index Funds, Financialization, and Commodity Futures Markets

Sanders, Irwin, and Merrin (2010) show that roughly 85% of index-related positions in agricultural futures markets are held by swap dealers.In the non-agricultural markets this percentage is likely to be even higher.However, swap dealers run a much larger “book” of OTC transactions thanjust index investments. A swap dealer may have a customer who is a tra-ditional short hedger, whose position offsets the long position desired byan ETF. In this case, the swap dealer “nets” the two positions internallyand may not have to go to the futures market to place a long hedge if thepositions offset one another. If long hedging in futures is required, it maybe in a much smaller quantity than the original index swap made with theETF due to internal netting. Swap dealer netting has been shown to be rela-tively small in agricultural futures markets, where swap dealers do rela-tively little non-index business; however, it can be quite large in the energyand metals markets (CFTC 2008). Figure 1 demonstrates the alternativeflow of total commodity index investments into net futures positions.

Measuring the size of commodity index investment is no simple matter. Inaddition to the netting issue discussed above, there are issues related to count-ing U.S. versus non-U.S. investments, and how to determine what qualifies asan “index” product. The CFTC has developed three different series related tocommodity index investment in U.S. futures markets; this reflects the variousassumptions about measuring index investment (see Irwin and Sanders (2010)for a discussion of the three series). Figure 2 presents combined U.S. andnon-U.S. asset totals in commodity index products collected by BarclaysCapital – one of the longest and oldest series available – from the fourthquarter of 2004 through the third quarter of 2010. This series indicates thatassets surged by over $200 billion starting in late 2004, reaching a peak of over$250 billion in mid-2008. Following a sharp decline during the recession, assetsresumed their upward climb, reaching a new peak of about $300 billion at theend of the third quarter of 2010.7 The plot also reveals that nearly all of the

Figure 1 Flow of index investments into commodity futures markets

7Index investment data collected by the CFTC provides smaller estimates of the assets in index funds.At the end of the second quarter of 2008, CFTC data indicated the total was $201 billion compared to$269 billion for the Barclays Capital data. The gap between the two series expanded to $132 billion atthe end of the second quarter of 2010.

Index Funds and Futures

5

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 6: Index Funds, Financialization, and Commodity Futures Markets

recent growth in assets has been in exchange-traded products and commoditynotes. Finally, it is important to remember when viewing such data that itmeasures total assets under management, which is affected by inflows and out-flows of investment funds and commodity price changes.

Further perspective on the size of commodity index investment is pro-vided in Figures 3–6; these figures show the net long (long minus shortcontracts), open interest of commodity index funds, and the percentage oftotal long open interest (futures only) represented by these positions infour grain futures markets: Chicago Board of Trade (CBOT) corn, soy-beans, and wheat, and Kansas City Board of Trade (KCBOT) wheat. Thedata are observed at weekly intervals from January 7, 2004 to September7, 2010, and are reported in the CFTC’s Supplemental Commitments ofTraders report (widely referred to as the commodity index trader report

Figure 2 U.S. and non-U.S. assets in commodity index products, first quarter, 2004-thirdquarter, 2010

Figure 3 Weekly net long open interest of commodity index traders (CIT) and CIT percentageof market (long) open interest for Chicago Board of Trade (CBOT) corn futures market,January 6, 2004 - September 7, 2010

Applied Economic Perspectives and Policy

6

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 7: Index Funds, Financialization, and Commodity Futures Markets

(CIT).8 Trends in the net long positions of CITs in these four marketsroughly parallel the overall trends in commodity index investment shownin Figure 2. There are some differences across markets, however. CIT netlong open interest in CBOT corn and wheat first peaked in early 2006, anddid not exceed this peak until very recently. CBOT soybeans and KCBOTwheat did not reach their first peak until later, in mid- to late-2007. CITpercentage of total market open interest probably best illustrates the sheersize of index fund positions, which has ranged recently from approxi-mately 20-30% of the long side of the market in CBOT corn and soybeans

Figure 4 Weekly net long open interest of commodity index traders (CIT) and CIT percentageof market (long) open interest for Chicago Board of Trade (CBOT) soybean futures market,January 6, 2004 - September 7, 2010

Figure 5 Weekly net long open interest of commodity index traders (CIT) and CIT percentageof market (long) open interest for Chicago Board of Trade (CBOT) wheat futures market,January 6, 2004 - September 7, 2010

8Position data for 2004-2005 were prepared by the CFTC at the request of the U.S. Senate PermanentSubcommittee on Investigations (USS/PSI 2009). We thank the staff of the subcommittee for allowingus to use this data.

Index Funds and Futures

7

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 8: Index Funds, Financialization, and Commodity Futures Markets

and KCBOT wheat, and 30-40% in CBOT wheat. By any reasonable stan-dard, index investment positions, at least in these grain futures markets,are indeed very large.

A relevant question in light of the large positions of commodity indexfunds is whether the positions should be considered speculation, hedging,or something altogether new. Stoll and Whaley (2010) argue that commod-ity index investments are not speculative because the investments arelong-only, passive, fully-collateralized, and motivated by portfolio diversi-fication benefits. This is an accurate description of the way commodityindex investments are developed and marketed to investors. Most publicstatements by institutional investors emphasize the primary advantage ofcommodity investments as diversification – providing a return that isuncorrelated with core equity and bond holdings. However, what is notknown with any degree of certainty is the actual motivation of insti-tutional or individual investors in commodity index funds. While thefunds are strictly long-only, they can still be used by investors as a tradinginstrument; money can be invested in index funds when expected returnsare believed to be high and withdrawn when returns are expected to below or negative. Intriguing evidence on the motivation of commodityinvestors is provided by a recent Barclays Capital survey (Norrish 2010).In response to the question, “Why do you invest in commodities,” 43% ofrespondents said portfolio diversification, 31% said absolute returns, 9%said inflation hedge, and 17% said emerging market growth. Theresponses indicate that there is a speculative component to at least somecommodity index fund positions.

What is a Bubble?

Brunnermeier (2008, p.578) defines a bubble as, “ . . . asset prices thatexceed an asset’s fundamental value because current owners believe thatthey can resell the asset at an even higher price in the future.” While thisdefinition is slightly different from some of the more classical references

Figure 6 Weekly net long open interest of commodity index traders (CIT) and CIT percentageof market (long) open interest for Kansas City Board of Trade (KCBOT) wheat futuresmarket, January 6, 2004 - September 7, 2010

Applied Economic Perspectives and Policy

8

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 9: Index Funds, Financialization, and Commodity Futures Markets

(O’Hara 2008), it is largely consistent with popular notions. That is, abubble is characterized by two elements. First, prices are “inexplicablebased on fundamentals” (Garber 2000, p. 4). Second, the trading motiveitself is unrelated to fundamental values, as market participants believethey can always sell to a greater fool. The end result is the classic marketaction identified (ex post) as, “ . . . an upward price movement over anextended range that then implodes,” (Kindleberger 1996, p. 13). The econ-omic damage from an asset pricing bubble may be extensive since themarket provides false signals about value. In particular, an inflated pricemay induce the over-production of an asset and misallocation of pro-ductive resources.

To understand the conditions for an asset pricing bubble to occur, it isuseful to first review the assumptions that ensure rational asset pricing—usually defined as a situation where market price equals the presentvalue of the future income stream associated with an asset (fundamentalvalue). Theoretical modeling (e.g., Tirole 1982) has deduced several keyassumptions necessary for rational asset pricing: i) an infinite number ofpotential traders; ii) traders believe other traders are rational; iii) tradersstart with the same prior beliefs about the future income stream; iv) norestrictions on short-selling or arbitrage; and v) the economy has an effi-cient allocation of resources before trading begins. If any of these assump-tions are violated, then a pricing bubble can form. For example, bubblescan occur with the large-scale entry of unsophisticated “noise” traderswho have mistaken beliefs about their prospects for future incomestreams. Rational traders would normally arbitrage any mispricing thatresults, but they may also restrain arbitrage positions due to noise traderrisk (De Long et al. 1990). Noise trader risk stems from the unpredictabil-ity of noise trader positions. However, if noise traders behave in a pre-dictable fashion, there is no noise trader risk, and asset pricing bubblescannot occur.9

From the perspective of economic theory, commodity futures contractsare not the type of market one would predict as being likely susceptible tobubble-like activity. Specifically, futures contracts are finite-horizon instru-ments with virtually no constraints on short-sales, which eliminate thepotential for a bubble in theoretical models (if the other assumptionslisted above hold). In addition, long-lasting bubbles are less likely inmarkets where deviations from fundamental value can be readily arbi-traged away (easily “poached” in the terminology of Patel, Zeckhauser,and Hendricks (1991)). There are few limitations to arbitrage in commod-ity futures markets, because the cost of trading is relatively low, trades canbe executed literally by the second, and prices are highly transparent. Thisstands in contrast to markets where arbitrage is more difficult, such asresidential housing. The low likelihood of bubbles is also supported bymany studies on the efficiency of price discovery in commodity futuresmarkets (see Zulauf and Irwin 1998; Carter 1999; Garcia and Leuthold2004). Finally, experimental evidence – which largely supports the exist-ence of bubbles – shows that the introduction of a futures market lessensthe incidence of bubbles (Forsythe, Palfrey, and Plott 1984; Noussair andTucker 2006).

9See Barlevy (2007) for a readable and non-technical discussion of the conditions in which assetpricing bubbles can occur.

Index Funds and Futures

9

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 10: Index Funds, Financialization, and Commodity Futures Markets

While it is arguable that bubbles have actually occurred with any fre-quency in financial history, at least in the sense defined by economists, theterm has definitely been more in vogue over the last decade, with pro-claimed “bubbles” in information technology stocks and residentialhousing, as well as suspected bubbles in U.S. Treasury securities and com-modity futures prices. Indeed, while economists are typically careful intheir use of the term “bubble,” the popular press has seemingly adoptedthe label for any market that makes a large advance quickly and thenmakes an equally hasty retreat. As Barlevy (2007) points out, the problemwith this popular usage is that large price swings sometimes occur natu-rally and in response to shifts in supply and demand. This highlights themain difficulty of identifying an asset pricing bubble: disentangling move-ments in an asset’s price due to changing supply and demand fundamen-tals from those due to bubble behavior. This is especially hard becausefundamental value depends not only on expectations about futuresincome streams, but also upon discount (interest) rates.

It Was a Bubble

The Arguments

Masters (2008) has interwoven investment and price data to create themost widely-cited bubble argument, painting the activity of index fundsas being akin to the infamous Hunt brothers’ cornering of the silvermarket in the early 1980s. Masters blames the rapid increase in overallcommodity prices from 2007-2008 on institutional investors’ embrace ofcommodities as an investable asset class. As discussed in the previoussection, it is clear that considerable investment dollars flowed into com-modity index funds over this time period. However, the evidence pro-vided by Masters is limited to anecdotes and the temporal correlationbetween money flows and prices. Masters and White (2008) recommendspecific regulatory steps to address the alleged problems created by indexfund investment in commodity futures markets, including there-establishment of speculative position limits for all speculators in allcommodity futures markets, as well as the elimination or severe restrictionof index speculation.

A similar position was taken by the U.S. Senate PermanentSubcommittee on Investigations in its examination of the performance ofthe CBOT wheat futures contract:

“This Report finds that there is significant and persuasive evidence to conclude thatthese commodity index traders, in the aggregate, were one of the major causes of“unwarranted changes” – here, increases – in the price of wheat futures contractsrelative to the price of wheat in the cash market. The resulting unusual, persistentand large disparities between wheat futures and cash prices impaired the ability ofparticipants in the grain market to use the futures market to price their crops andhedge their price risks over time, and therefore constituted an undue burden oninterstate commerce. Accordingly, the Report finds that the activities of commodityindex traders, in the aggregate, constituted “excessive speculation” in the wheatmarket under the Commodity Exchange Act,” (USS/PSI 2009, p. 2).

Based on these findings, the Subcommittee recommended the: 1) phaseout of existing position limit waivers for index traders in wheat; 2) ifnecessary, the imposition of additional restrictions on index traders, such

Applied Economic Perspectives and Policy

10

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 11: Index Funds, Financialization, and Commodity Futures Markets

as a position limit of 5,000 contracts per trader; 3) investigation of indextrading in other agricultural markets; and 4) the strengthening of data col-lection on index trading in non-agricultural markets.

One of the limitations of the bubble argument made by Masters is thatthe link between money inflows from index funds and commodity futuresprices is not well developed. This allows critics to assert that bubble pro-ponents make the classical statistical mistake of confusing correlation withcausation. In other words, simply observing that large investment hasflowed into the long side of commodity futures markets at the same timethat prices have risen substantially does not necessarily prove anythingwithout a logical and causal link between the two. One attempt to estab-lish this linkage was delivered in testimony by Todd Petzel, ChiefInvestment Officer for Offit Capital Advisors, at a July 2009 CFTC hearingon position limits in energy futures markets:

“Seasoned observers of commodity markets know that as non-commercial partici-pants enter a market, the opposite side is usually taken by a short-term liquidity pro-vider, but the ultimate counterparty is likely to be a commercial. In the case ofcommodity index buyers, evidence suggests that the sellers are not typically otherinvestors or leveraged speculators. Instead, they are owners of the physical commod-ity who are willing to sell into the futures market and either deliver at expiration orroll their hedge forward if the spread allows them to profit from continued storage.This activity is effectively creating “synthetic” long positions in the commodity forthe index investor, matched against real inventories held by the shorts. We have seenhigh spot prices along with large inventories and strong positive carry relationshipsas a result of the expanded index activity over the last few years,” (Petzel 2009,pp. 8-9).

In essence, Petzel (2009) argues that unleveraged futures positions ofindex funds are effectively synthetic long positions in physical commod-ities, and hence represent new demand. If the magnitude of index funddemand is large enough relative to physically-constrained supplies in theshort-run, prices and price volatility can increase sharply. The bottom lineis that the size of index fund investment is “too big” for the size of com-modity futures markets.

Hamilton (2009b) develops a formal model of price determination in thecrude oil market that generates specific theoretical conditions for speculat-ive impact; he begins by noting that the key challenge is reconciling aspeculative bubble in crude oil prices with changes in the physical quan-tities of crude oil. A standard argument is that a price bubble will inevita-bly lead to a rise in inventories as the quantity supplied at the “bubbleprice” exceeds the quantity demanded (Krugman 2008). Hamilton’s theor-etical model is based on a profit-maximizing representative refiner ofcrude oil into gasoline, an exogenous supply of crude oil, and a demandfunction for gasoline. Solution of the model for the optimal inventorymanagement condition of the refiner leads to important predictions aboutprice behavior and inventories. In particular, if the demand for gasoline isperfectly inelastic and speculators are able to force up the forward(futures) price of crude oil for non-fundamental reasons, this will forcethe current (spot) price of crude oil to rise in order to maintain equili-brium in the storage market. However, if the elasticity of gasoline demandis less than perfectly inelastic, the standard result follows that inventoriesmust also increase.

Caballero, Farhi, and Gourinchas (2008) develop a theoretical modelthat emphasizes the underlying “macro” forces that may be driving a

Index Funds and Futures

11

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 12: Index Funds, Financialization, and Commodity Futures Markets

bubble in commodity prices. The model divides the world into tworegions; the first has developed financial markets, and the other does not.Since commodity producers are assumed to be located in the region withless developed financial markets, producers are forced to hold financialassets from the developed region as a store of wealth, which leads to adecline in real interest rates in this region. Bubbles form in asset markets,including commodity markets, in response to the decline in real interestrates. Once again, a key prediction of this model is that commodity inven-tories accumulate during the building phase of the price bubble. An inter-esting feature of this model is that equilibrium is invariant to the presenceof a futures market. The bubble is ultimately fed only by global imbal-ances in the supply of liquid and safe financial assets.

Empirical Evidence

We review here five studies that provide empirical evidence of a bubbleand/or a statistical linkage between index fund activity and commodityfutures price movements. These studies are selected because they containstatistical analysis that directly tests whether the recent commodity pricespike contained a bubble component and/or whether index fund tradingcontributed to the price spike.10

Gilbert (2009) examines price behavior in crude oil futures, three metalsfutures, and three agricultural futures over 2006-2008. He first testsdirectly for bubbles using a new time-series test developed by Phillips,Wu, and Yu (2009). Tests using daily data indicate bubbles in seven of thenine markets, but for a surprisingly small percentage of the days in thesample. For instance, statistically significant bubbles are present in only 21out of the 753 days for crude oil futures, or less than 3% of the entire2006-2008 sample period. Not surprisingly, the “bubble days” are concen-trated entirely near the peak of prices in the summer of 2008. Gilbert alsoconstructs a quantum index of commodity index fund investment infutures markets using positions of index traders in the 12 agriculturalmarkets reported in the CFTC’s weekly CIT report. Granger causality testsare used to establish whether lagged changes in the quantum index helpto forecast returns (price changes) in each of the seven markets includedin the analysis. The results indicate a significant relationship betweenindex fund trading activity and returns in three of the seven markets:crude oil, aluminum, and copper. Gilbert estimates the maximum impactof index funds in these markets to be a price increase of 15%.

Gilbert’s (2009) results point towards a non-negligible impact of indexfunds on price in at least some commodity futures markets. There arereasons for caution, however. The first is that rejection of the null hypoth-esis in a Granger causality test should be viewed as only a preliminaryindication of a true casual relationship. Newbold (1982) provides a classicdiscussion of the problems that can arise in rejecting the null in such tests,and these include: i) economic agents’ expectations about the future canmake it appear as if there is a relationship between two variables when infact there is none; ii) the apparent causal relationship between two vari-ables may actually be the result of a third variable omitted from the

10We make no claim to having uncovered any and all studies that may contain such evidence. This iscertainly an active area of current economic research around the world and additional studies willundoubtedly emerge.

Applied Economic Perspectives and Policy

12

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 13: Index Funds, Financialization, and Commodity Futures Markets

analysis; and iii) without an explicit theory that establishes a link betweentwo variables, the meaning of a non-zero correlation is unclear.

The second reason is related to the index of commodity index fundactivity used by Gilbert (2009); his quantum index is a weighted averageof index trader positions in agricultural futures markets (as reported inthe CFTC’s CIT report). The weights are nearest-to-expiration futuresprices on the first date of his sample. Gilbert (2009) then applies this indexnot only to three of the agricultural markets included in the index, butalso to the energy and metal markets not included in the index.Application of an index constructed from the 12 CIT agricultural marketsto the energy and metals markets is justified only if one believes thepattern of index trader positions in agricultural markets closely mirrorsthe pattern in energy and metals markets. Gilbert (2009) presents no evi-dence whether this is a reasonable assumption or not. This point isreinforced by noting that he did not find evidence of a relationshipbetween index fund activity in agricultural markets, upon which the indexis based, but did find evidence in energy and metals markets, which arenot included in the index.

Gilbert (2010) investigates the factors driving the 2007-2008 increase infood prices; he argues that agricultural price booms are better explainedby common demand-side factors than individual market supply-sidefactors. He goes on to specify a multi-equation model where changes inaggregate food commodity prices are a function of changes in the price ofcrude oil, commodity index investment, and a U.S. dollar exchange rateindex. Changes in the price of oil and commodity index investment areendogenous variables, and the change in the U.S. dollar exchange rate isexogenous. Commodity index investment is measured using the samequantum index as in his 2009 study, and it is specified as a function ofchanges in the U.S. dollar index, Chinese industrial production, and U.S.and Chinese stock market indices. This specification is based on the argu-ment that index investment is not driven by portfolio diversification, butrather by investors’ view that the demand for raw material from Chinawill push up commodity prices.

Gilbert (2010) estimates model parameters using monthly data fromMarch 2006 to June 2009, which indicate a large and statistically significantimpact of commodity index investment on the food price index. However,Gilbert does not interpret index investment as a fundamental cause of thefood price boom, but instead it is viewed as the channel by which themain drivers – rapid Chinese economic growth and dollar depreciation –affected the price of food. This is another form of the argument thatunderlying “macro” forces created a bubble in commodity prices ratherthan index funds per se (Caballero, Farhi, and Gourinchas 2008). Onceagain, there are good reasons for caution when interpreting these results.For example, the sample period is very short for estimating a structural-type econometric model (which the author acknowledges). Further, themodel implicitly assumes that index investment is entirely speculative,certainly a controversial assumption. Finally, there is little informationabout the sensitivity of estimation results to alternative specifications ofthe variables included in the model. It would not be surprising to findthat results are quite sensitive to model specification given the short timeperiod used to estimate the model.

Index Funds and Futures

13

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 14: Index Funds, Financialization, and Commodity Futures Markets

Einloth (2009) devises a test of bubbles based on the joint behavior ofconvenience yields and prices in a storable commodity futures market.Here, convenience yield is the flow of benefits that accrue to inventoryholders from having stocks of the commodity on hand. For example, thecosts of shutting down an oil refinery due to the unavailability of crudeoil may be very high; thus, the operator may hold some stocks as insur-ance against this possibility. The basic theory of storage predicts that lowinventories lead to a rising commodity price and an increasing marginalconvenience yield (Working 1948; 1949). Thus, the combination of a risingfutures price and a falling marginal convenience yield is a violation of thistheory. Einloth uses this condition as a test for speculative impact in thecrude oil futures market, and finds that marginal convenience yield(derived from differences between futures prices) rose for most of theincrease in crude oil futures prices up to $100/barrel, but fell as crude oilincreased from about $100 to $140/barrel. While this indicates that specu-lation may have played a role near the peak in crude oil prices, the tech-nique cannot attribute this to a particular group of market participants,such as index funds.

There is no doubt that Einloth is correct in arguing that the basic theoryof storable commodity markets rules out the combination of falling mar-ginal convenience yield and rising futures prices. However, more sophisti-cated versions of the theory show that this combination can occur, not dueto speculative impacts, but because market participants are reacting ration-ally to a particular set of incentives. For example, Routledge, Seppi, andSpatt (2000) develop an equilibrium model of price behavior for a storablecommodity where the spot commodity has an embedded timing optionthat is a function of endogenous inventory and exogenous supply anddemand shocks. In this model, falling marginal convenience yield can beobserved at the same time that futures prices are rising if the value of thetiming option is large enough. This could occur if the probability of asequence of large positive demand shocks or negative supply shocksincreases substantially. The implication is that Einloth’s test may not ident-ify bubbles, but instead unusual periods where the timing option derivedfrom holding inventories increases at the same time that price also rises.

Tang and Xiong (2010) do not test for bubbles in commodity futuresprices. Instead, they argue that commodity markets were not fully inte-grated with financial markets prior to the development of commodityindex investments and, “The increasing presence of index investors incommodities markets precipitated a fundamental process of financializa-tion amongst the commodities markets, through which commodity pricesnow become more correlated with the prices of financial assets and eachother,” (p. 2). Statistical tests confirm that the correlation of commodityfutures returns with stock, bond, U.S. dollar, and crude oil returnsincreased significantly starting in 2004. Some of the increases are remark-able. For example, the correlation of daily soybean and crude oil futuresreturns before 2004 moved in a narrow range between .10 and 0.20, butafter 2004 increased steadily to a peak of about 0.60 in 2009.

Tang and Xiong recognize that increasing correlations could be due to avariety of factors, such as the financial shocks associated with the recentrecession and the increasing impact of crude oil prices on many commod-ities (e.g., de Gorter and Just 2010), and address this “identification” issuewith a difference-in-difference approach. Specifically, they test whether the

Applied Economic Perspectives and Policy

14

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 15: Index Funds, Financialization, and Commodity Futures Markets

increase in correlation of commodity futures returns with stock, bond, U.S.dollar, and crude oil returns post-2004 is greater for commodities includedin major indices (such as the-S&P-GSCI) compared to commodities notincluded in the indices. A statistically significant increase for in- versusoff-index commodity markets is estimated for five non-energy sectors,leading the authors to conclude that index investment has increasinglyimpacted commodity futures prices (although in a non-bubble manner).

Tang and Xiong’s study appears to provide concrete evidence of alinkage between commodity index investment and futures price move-ments. The results are also consistent with other types of evidence aboutthe price impact of investment flows in financial markets (Barberis,Shleifer, and Wurgler 2005). Nonetheless, there are several reasons forviewing Tang and Xiong’s findings with caution. The first reason is thatthe average magnitude of correlation increases is small after 2004 for in-versus off-index commodities. The simplest way to emphasize this pointis to compare the estimated correlations across in- and off-index commod-ities for the final year of the sample, 2009. Considering the results forregressions in which correlations (exposures) are estimated jointly for allfour financial variables (regression (6) in their table 5), one discovers thatdifferences in correlations for in- versus off-index commodities in 2009average only 0.05, -0.03, -0.06, and 0.14 for equity index, bond index,dollar index, and crude oil returns, respectively, across the five commoditysectors. While these differences generally are statistically significant, theeconomic significance of such small differences is debatable.

The second reason for viewing Tang and Xiong’s findings with cautionis that the difference-in-difference approach used by the authors may notadequately control for fundamental factors that are common to all com-modities in a given sector. Ai, Chatrath, and Song (2006) show how afailure to properly condition on market fundamentals can lead to incorrectinferences in co-movement tests. A particular concern with Tang andXiong’s study is the representativeness of the non-index “control” marketsin some sectors. Two of the four non-index markets in the grain sector arerough rice and oats futures, both of which are thinly-traded and illiquidmarkets. The same can be said of the only two non-index markets in thesoft commodity sector, lumber and orange juice. The most egregiousexample is in the livestock sector, where the sole non-index market is porkbellies. This is, for all intents and purposes, a dead market with close tozero open interest. Comparison to illiquid and unrepresentative marketscould induce a bias in the difference-in-difference regression estimates.

The third reason for viewing Tang and Xiong’s findings cautiously is thatseveral other studies indicate that commodity futures markets have beenintegrated with financial markets for decades, contradicting Tang andXiong’s basic thesis. For example, Bjornson and Carter (1997) examine sevencommodity futures markets from 1969-1994, and find that expected returnsare lower during times of high interest rates, expected inflation, and econ-omic growth, leading the authors to conclude that commodities provide anatural hedge against business cycles. In a recent paper, Hong and Yogo(2010) document that information about future inflation has been priced intocommodity futures markets since at least the mid-1960s (albeit with a delay).

Phillips and Yu (2010) introduce a new recursive test for bubble charac-teristics in financial time series. Essentially, the sequential procedure usesstandard tests (Dickey-Fuller t tests) for unit root behavior in prices

Index Funds and Futures

15

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 16: Index Funds, Financialization, and Commodity Futures Markets

against “mildly explosive alternatives” that would characterize a bubble.The recursive procedure is powerful in the sense that it allows for the“time stamping” of the start and end date for a bubble, and it can beapplied in real time to identify when a bubble is forming. FollowingCaballero, Farhi, and Gourinchas (2008), the authors hypothesize thatglobal liquidity imbalances drove a series of financial bubbles thatmigrated from technology stocks to housing, on to subprime mortgages,and then into commodity markets and eventually the bond market (afterthe subprime crises broke). Test results are largely consistent with thissequence, identifying a bubble in subprime mortgage derivatives fromAugust 2005 through July 2007, which then migrated to selected commod-ity markets. Specifically, the authors find evidence of bubble price charac-teristics from March 2008 to August 2008 in crude oil and heating oil.Notably, however, there is no evidence of a bubble in prices for coffee,cotton, cocoa, sugar or feeder cattle.

The findings of Phillips and Yu (2010) are certainly interesting and rep-resent some of the more convincing evidence of bubble-like activity inenergy prices. However, the lack of evidence for the other commoditiestested is inconsistent with the notion that a wave of index fund buyingpushed up all commodity prices. Moreover, the authors’ selection ofmarkets (e.g., feeder cattle) and data (e.g., deflated spot prices) may nottranslate to equivalent findings in the grain and oilseed futures markets,where concerns about a bubble are typically focused.

To summarize, the studies reviewed in this section are interesting, rigor-ous, and make a valuable contribution to the debate about the marketimpact of commodity index funds. Results in these studies also negate theargument that no evidence exists of a bubble in commodity futures pricesor a relationship between commodity index investment and movements incommodity futures prices. However, the data and methods used in thesestudies are subject to a number of important criticisms that limit thedegree of confidence one can place in the results.

It Was Not a Bubble

The Arguments

A number of economists have expressed skepticism about the bubbleargument (e.g., Krugman 2008; Pirrong 2008; Sanders and Irwin 2008;Smith 2009). These economists cite several contrary facts and argue thatcommodity markets were driven by fundamental factors that pushedprices higher. Irwin, Sanders, and Merrin (2009) and Pirrong (2010)present useful summaries of the counter-arguments made by economists.We categorize these into three logical inconsistencies in the argumentsmade by bubble proponents and four instances where the bubble story isnot consistent with observed facts.

The first possible logical inconsistency within the bubble argument isequating demand with money inflows to commodity futures markets.With equally informed market participants, there is no limit to thenumber of futures contracts that can be created at a given price level.Index fund buying in this situation is no more “new demand” than thecorresponding selling is “new supply.” Futures contracts are “pure bets”in this context (Black 1976), with a long for every short such that

Applied Economic Perspectives and Policy

16

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 17: Index Funds, Financialization, and Commodity Futures Markets

commodity futures markets are zero-sum games. This implies that moneyflows in and of themselves do not necessarily impact prices. Rather, priceswill only change if new information emerges that causes market partici-pants to revise their estimates of physical supply and/or demand.

What happens when market participants are not equally informed?When this is the case, it is rational for participants to condition demandson both their own information and information about other participants’demands that can be inferred (“inverted”) from the futures price(Grossman 1986). The trades of uninformed participants can impact pricesin this more realistic model if informed traders mistakenly believe thattrades made by uninformed participants (“noise”) reflect valuable infor-mation about supply and demand fundamentals. Hence, it is possible thatother traders in commodity futures markets interpreted the large orderflow of index funds on the long side of the market as a reflection of valu-able private information about commodity price prospects, which wouldhave had the effect of driving prices higher as these traders revised theirown demands upward. Of course, this would have required a largenumber of sophisticated and experienced traders in commodity futuresmarkets to reach the conclusion that index fund investors possessed valu-able information that they themselves did not possess.

The second possible logical inconsistency of the bubble story is in theargument that index fund investors artificially raised both futures andcash commodity prices when they only participated in futures markets.These contracts are financial transactions that only rarely involve theactual delivery of physical commodities. In order to impact the equili-brium price of commodities in the cash market, index investors wouldhave to take delivery and/or buy quantities in the cash market and holdthese inventories off the market. Index investors are purely involved in afinancial transaction using futures markets; they do not engage in the pur-chase or hoarding of the cash commodity and any causal linkagesbetween their futures market activity and cash prices is unclear at best.

A third possible logical inconsistency of the bubble argument is theblanket categorization of speculators, particularly index fund speculators,as wrongdoers, and hedgers as victims of their actions. In reality, the “badguy” is not so easily identified since hedgers sometimes speculate andsome speculators also hedge. For example, large commercial firms mayhave valuable information gleaned from their far-flung cash market oper-ations and then trade based on that information.11 The following passagefrom a recent article on Cargill, Inc. nicely illustrates the point:

“Wearing multiple hats gives Cargill an unusually detailed view of the industries itbets on, as well as the ability to trade on its knowledge in ways few others canmatch. Cargill freely acknowledges it strives to profit from that information. “Whenwe do a good job of assimilating all those seemingly unrelated facts,” says GregPage, Cargill’s chief executive, in a rare interview, “it provides us an opportunity to

11Hieronymus (1977) argued that large commercial firms dominated commodity futures markets, andspeculators tended to be at a disadvantage. Based on his theoretical analysis, Grossman (1986,p. S140) asserted, “ . . . it should come as no surprise if a study of trading profit finds that tradersrepresenting large firms involved in the spot commodity (i.e., commercial traders) make large tradingprofits on futures markets.” In the classic empirical study on this subject, Hartzmark (1987) showedthat large commercial firms in six of seven futures markets make substantial profits on their futurestrades.

Index Funds and Futures

17

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 18: Index Funds, Financialization, and Commodity Futures Markets

make money . . . without necessarily having to make directional trades, i.e., outguessthe weather, outguess individual governments,” (Davis 2009).

The implication is that the interplay between varied market participants ismore complex than a standard textbook description of pure risk-avoidinghedgers and pure risk-seeking speculators. The reality is that marketdynamics are ever–changing, and it can be difficult to understand themotivations and market implications of trading, especially in real-time.

In addition to the logical inconsistencies, there are several ways thebubble story is not consistent with the observed facts. First, theoreticalmodels (Caballero, Farhi, and Gourinchas 2008; Hamilton 2009b) showthat if a bubble raises the market price of a storable commodity above thetrue equilibrium price, then stocks of that commodity should increase(much like a government-imposed price floor can create a surplus). Thefact that stocks were declining, not building, in most commodity marketsfrom 2006-2008 is therefore an important contradiction of the price bubbleargument in these markets (see Korniotis (2009) and Pirrong (2010) fordetailed statistics and analysis). There are two possible counterpoints tothis observation. Caballero, Farhi, and Gourinchas (2008) argue that inven-tory for an exhaustible resource like crude oil includes un-extracted oil inthe ground and it is possible that the below-ground rise in inventory morethan offset the above-ground (measured) decline in inventory. Hamilton(2009b) notes that the buildup of (above-ground) inventories due to aprice bubble could be so small that it is not easily detected if the elasticityof demand for the storable commodity is very small or zero.

The second factual inconsistency of the bubble story is that if indexfund buying drove commodity prices higher, then markets without indexfund investment should not have seen a price advance. Irwin, Sanders,and Merrin (2009) show that markets without index fund participation(fluid milk and rice futures) and commodities without futures markets(apples and edible beans) also showed price increases during the2006-2008 period. Stoll and Whaley (2010) report that returns for CBOTwheat, KCBOT wheat, and Minneapolis Grain Exchange (MGEX) wheatare all highly positively correlated from 2006-2009, yet only CBOT andKCBOT wheat are used heavily by index investors. Stoll and Whaley alsoobserve that Commodity Exchange (COMEX) gold, COMEX silver,New York Mercantile (NYMEX) palladium, and NYMEX platinum futuresprices are highly correlated over the same time period, but only gold andsilver are included in popular commodity indices. Pirrong (2010) reportsthat the price of shipping services increased dramatically between June2007 and July 2008 in concert with other commodity prices, which, heargues, contradicts the bubble story since shipping services are non-storable and therefore should not be susceptible to asset bubble influences.There are two notable counterpoints to these observations. Petzel (2009)argues that many of the comparisons are problematic because a bubble inone market could be transferred to other markets through well-knownmarket linkages (i.e., cross-price elasticities).12 Tang and Xiong (2010)point out that the relative magnitude of price increases for index and non-index (storable) commodities is relevant whether or not non-index com-modities increased in price.

12Sanders, Irwin, and Merrin (2009) address this issue by selecting markets for comparison that havelow-cross price elasticities.

Applied Economic Perspectives and Policy

18

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 19: Index Funds, Financialization, and Commodity Futures Markets

The third factual inconsistency with the bubble story is that speculationwas not excessive when correctly compared to hedging demands. The stat-istics on long-only index fund trading reported in the media and discussedat hearings tend to view speculation in a vacuum focusing on absolute pos-ition size and activity. Working (1960) argued that speculation must begauged relative to hedging needs. In particular, speculation can only be con-sidered “excessive” relative to the level of hedging activity in the market.Utilizing Working’s speculative T-index, Sanders, Irwin, and Merrin (2010)demonstrate that the level of speculation in nine agricultural futuresmarkets from 2006-2008 (adjusting for index fund positions) was not exces-sive. Indeed, the levels of speculation in all markets examined were withinthe realm of historical norms. Across most markets, the rise in index buyingwas more than offset by commercial (hedger) selling. Buyuksahin andHarris (2009) use daily data from the CFTC’s internal large trader databaseto show that Working’s T-index in the crude oil futures market increased inparallel with crude oil prices from 2004-2009, but the peak of the index wasstill well within historical norms. Till (2009) reports similar results for crudeoil, heating oil, and gasoline futures from 2006-2009 using data in theCFTC’s Disaggregated Commitments of Traders (DCOT) report.

The fourth observable fact that contradicts the bubble story relates to theimpact of index funds across markets; a priori, there is no reason to expectindex funds to have a differential impact across markets given similar pos-ition sizes. In other words, one would expect markets with the highestconcentration of index fund positions to show the largest price increases.But Irwin, Sanders, and Merrin (2009) find just the opposite when compar-ing grain and livestock futures markets. The highest concentration ofindex fund positions was often in livestock markets, which had the smal-lest price increases through the spring of 2008. This is difficult to reconcilewith the assertion that index buying was responsible for a price bubble.

Empirical Evidence

We review here nine studies that do not provide empirical evidence of abubble and/or a statistical linkage between index fund activity and com-modity futures price movements.13 As in the previous section, studies areselected because they contain statistical analysis that directly tests whetherthe recent commodity price spike contained a bubble component and/orindex fund trading contributed to the price spike. Most of the studiesfocus on testing for a relationship between index fund trading and com-modity futures price movements since this is the most direct type ofevidence.14

Buyuksahin and Harris (2009) use daily data from the CFTC’s internallarge trader database from July 2000-March 2009 to conduct a variety oftests on the relation between crude oil futures prices and positions ofvarious types of traders. Granger causality tests provide no statistical evi-dence that position changes (number of contracts) by any trader group

13As before, we make no claim to presenting an exhaustive review of all such studies.14Though not reviewed here, Robles, Torero, and von Braun (2009) utilize weekly CIT data to conductGranger causality tests between index positions and futures returns. Despite the overwhelmingly nega-tive results of their statistical analysis, the authors nonetheless assert that there is sufficient evidenceof the damaging role of speculation to warrant the creation of a new international organization tocounteract this market failure. See Wright (2009) for further discussion.

Index Funds and Futures

19

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 20: Index Funds, Financialization, and Commodity Futures Markets

systematically precede price changes. This result holds for the entiresample period, and when the sample is broken into two parts: from July2000-June 2004 and July 2004-March 2009. Buyuksahin and Harris assumethat swap dealer positions are linked to commodity index investing, andhence, the results for swap dealers are interpreted as a failure to reject thenull hypothesis of no relation between crude oil futures prices and indexfund trading.

A key question, then, is whether it is reasonable to assume that swapdealer positions in crude oil futures accurately reflect the total effective“demand” of index investors. As noted earlier, swap dealers internally netlong commodity index positions with other long and shortover-the-counter (OTC) positions. A 2008 CFTC report indicates that swapdealer netting is substantial in crude oil, with only 41% of long swapdealer positions in crude oil futures on three dates in 2007 and 2008 linkedto long-only index fund positions (CFTC 2008). Hence, the measurementerror inherent in using swap dealer positions to represent index fund pos-itions in crude oil futures may be substantial and could limit the con-clusions that can be drawn from studies like Buyuksahin and Harris.

Brunetti and Buyuksahin (2009) also use daily data from the CFTC’sinternal large trader database to analyze whether speculators destabilizeprices in five futures markets: NYMEX crude oil and natural gas, ChicagoMercantile Exchange (CME) Eurodollar and mini-Dow, and CBOT corn.The sample periods begin in either January 2005 or August 2006 (corn) andend in March 2009. A unique feature of the study is that price impacts forthe five largest trader groups are estimated jointly in a vector autoregres-sion (VAR) system, which allows tests not only of the direct effect of eachgroup on returns or volatility, but also the interaction of positions amongthe groups. Three groups of speculators are considered: swap dealers,hedge funds, and floor broker/traders. Like Buyuksahin and Harris (2009),this study assumes that swap dealer positions represent commodity indexfund investments. Granger causality tests based on the estimated VARmodels yield two main conclusions. First, returns are not forecast by pos-itions of any trader group in the five markets, including swap dealers, withthe exception of mini-DOW futures, where there is evidence of a dampen-ing impact of hedge funds. Second, volatility is forecast by swap dealerpositions in crude oil and natural gas and by other speculators in allmarkets except corn. Since the impact is negative, Brunetti and Buyuksahinargue that speculators reduce risk in the futures market and thereforelower the cost of hedging. Once again, the use of swap dealer positions torepresent index fund investment is a potential limitation of this study.

Stoll and Whaley (2010) use data from the CFTC’s weekly CIT report toconduct a variety of tests on the null hypothesis that index fund tradingdoes not cause commodity futures price changes. The authors examine all12 agricultural markets included in the CIT report from January 2006-July2009: corn, soybeans, soybean oil, CBOT wheat, KCBOT wheat, feedercattle, lean hogs, live cattle, cocoa, cotton, coffee, and sugar. Granger caus-ality tests between weekly futures returns and weekly index fund invest-ment flows reject the null of no causality in only 1 of the 12 markets. Theauthors also examine the contemporaneous relationship between returnsand changes in index fund investment flows after controlling for changesin other speculator investment flows. A significant contemporaneousrelationship is found for only 2 of the 12 markets. Two additional “roll”

Applied Economic Perspectives and Policy

20

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 21: Index Funds, Financialization, and Commodity Futures Markets

tests are conducted to determine whether the shifting of index fund pos-itions between contracts impacts the relative price movements of thenearest- and next-nearest futures contracts. This analysis is conducted forsix commodities tracked in the CIT report, as well as crude oil. These testsfind evidence of a significant relative price impact only in crude oil.Overall, Stoll and Whaley conclude that commodity index investing haslittle impact on futures prices.

Stoll and Whaley’s analysis generates virtually no evidence that indextrading had an impact on agricultural futures price movements. However,it is not known whether their results generalize to a broader set of com-modity futures markets, especially energy futures markets that have beenat the center of the current controversy. Another issue is that the power oftheir statistical tests to reject the null hypothesis of no causality may bedecreased by the use of (U.S.) dollar flows to measure index fund partici-pation. Dollar flow measures are computed by multiplying each week’snet position (in contracts) by the price of the futures contract nearest-to-expiration to obtain a total notional value, and then differencing thetotal from week-to-week. Because U.S. dollar flows impound into activitymeasures both the variability of prices and positions, it is a less precisemeasure of actual index fund buying and selling. Finally, there is thequestion of the degree to which the CIT data represents actual indexinvestment in agricultural markets. Sanders, Irwin and Merrin (2010)show that the bulk of CIT positions are in fact swap dealers hedging OTCexposure. However, swap dealers operating in agricultural marketsconduct a limited amount of non-index long or short swap transactions(CFTC 2008), so there is less measurement error in using the net positionof swap dealers in these markets to represent index fund investment.

Sanders and Irwin (2010b) investigate the price impact of index fundtrading in four grain futures markets. Weekly CIT data for CBOT corn,soybeans, wheat, and KCBOT wheat futures contracts from January 2004to August 2009 are used in the analysis.15 The data from 2004-2005 are notpublicly available, and are used for the first time in this study. Extendingthe sample period with this data is advantageous for two reasons. First,previous research suggests that the buildup in index positions was mostrapid from 2004-2005 (Sanders, Irwin, and Merrin 2010); this period maythus be the most likely to show the impact of index traders, if any.Second, the additional two years of data increases the sample size ofweekly observations considerably compared to the publicly available CITdata, which should improve the power of time-series statistical tests.

The data confirm that there was a fairly dramatic and large build-up inindex fund positions in the four grain markets examined by Sanders andIrwin (2010b). For instance, the number of wheat contracts held by indexfunds at the CBOT increased nearly fourfold from 2004 to 2006. However,the buildup in index contracts and the peak level of index holdings pre-dates the 2007-2008 increase in grain futures prices for which they areblamed. This observation casts doubt on the hypothesis that index specu-lation drove the 2007-2008 increase in grain futures prices. In 15 out of 16tests, both Granger causality and long-horizon regressions fail to reject thenull hypothesis that commodity index positions (change in net long CIT

15The same position data, updated through early September 2010, are used in figures 3-6 of thisarticle.

Index Funds and Futures

21

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 22: Index Funds, Financialization, and Commodity Futures Markets

contracts or net long CIT contracts/total long market open interest) haveno impact on futures prices. Overall, the authors conclude that data trendsand statistical test results are not consistent with the widely-touted bubbletheory. A limitation of this study is that tests are restricted to four grainfutures markets.

Sanders and Irwin (2010c) investigate the price impact of index fundtrading in 12 agricultural markets and 2 energy futures markets (crude oiland natural gas). Data compiled by the CFTC in the DisaggregatedCommitments of Traders (DCOT) report on positions held by swap dealersfrom June 2006 to December 2009 is assumed to reflect index-type invest-ments. Bivariate Granger causality tests are used to investigate lead-lagdynamics between index fund positions and futures returns or price vola-tility in each commodity futures market. In addition, a new systemsapproach to testing lead-lag dynamics is introduced and applied. Thesystems approach improves the power of statistical tests by taking intoaccount the contemporaneous correlation of model residuals across marketsand allows a test of the overall impact of index funds across markets.

The system of Granger-style causality tests fails to reject the nullhypothesis that trader positions do not lead returns in the 14 markets ana-lyzed by Sanders and Irwin (2010c). Hence, there is no evidence of alinkage between index trader positions (as represented by swap dealers)in commodity futures markets and price levels. There is a tendency toreject the null hypothesis that index trader positions do not lead to marketvolatility. Similar to the results in Brunetti and Buyuksahin (2009), thedirection of the impact is consistently negative. While index positions leadto lower volatility in a statistical sense, Sanders and Irwin (2010c) note thatit is possible that trader positions coincide with some other fundamentalvariable that is actually causing the lower market volatility (see the discus-sion of Newbold (1982) above). Sanders and Irwin (2010c) are also awarethat swap dealer positions may be an imperfect proxy for overall indexinvestments due to internal netting of OTC positions by swap dealers, par-ticularly in the energy futures markets. However, they argue that the netposition may still be relevant with respect to price impacts since this is theposition actually taken to the commodity futures market. Whether nettedor non-netted positions are more relevant is an empirical question.

Irwin and Sanders (2010) conduct a test of the “Masters Hypothesis” incommodity futures markets using the newest of the CFTC’s data series onindex fund investment – the quarterly Index Investment Data (IID) report.The IID data is not only a better measure of index investment because pos-itions are measured before internal netting by swap dealers, it also coversthe complete spectrum of agricultural, energy, metal and soft commodityfutures markets. A cross-sectional regression test is applied to the IID dataover the first quarter of 2008 through the first quarter of 2010, which cap-tures the rapid increase, peak, and subsequent decline in commodityprices. A cross-sectional approach is likely to improve statistical powercompared to time-series Granger-type tests (Sanders and Irwin 2010a) anda quarterly horizon should better capture a slowly accumulating pricepressure effect if it is present in the data. Both lagged and contempora-neous effects are also considered.

Irwin and Sanders’ regression estimates provide very little evidence thatindex positions influence the cross-section of 19 commodity futuresmarket returns or volatility. Only 3 of the 18 average estimated slope

Applied Economic Perspectives and Policy

22

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 23: Index Funds, Financialization, and Commodity Futures Markets

coefficients are statistically significant, and 2 of the 3 significant cases havenegative signs that contradict the claim that index investment increasesreturns or volatility. The results are robust regarding whether lagged orcontemporaneous effects are considered. The cross-section tests are sup-plemented with time-series tests using daily positions held by the U.S. Oiland Natural Gas Funds, two of the largest exchange-traded index funds(ETFs), in the crude oil and natural gas futures markets, respectively.Granger-causality tests show no causal links between daily returns orvolatility in crude oil and natural gas futures markets and the positions ofthe two large ETFs. Long-horizon regression tests likewise fail to reject thenull hypothesis of no market impact for the ETFs. Overall, the empiricalresults of the study fail to support the hypothesis that index funds were amajor driver of the recent spike in commodity prices.

Irwin et al. (2011) examine whether index funds were responsible forrecent episodes of non-convergence between CBOT corn, soybean, andwheat futures prices and cash prices during the delivery period of thecontracts. Some argue that index funds were a major driver of the epi-sodes of non-convergence because index fund investment put upwardpressure on the spreads between futures prices for the same commoditydue to the “rolling” of positions or the initiation of positions in a“crowded market space” (e.g., USS/PSS 2009; Petzel 2009). If spreadsapproach full storage and interest opportunity costs, an incentive iscreated for those taking delivery that effectively de-couples cash andfutures prices. The authors do not find evidence of a statistically signifi-cant increase in spreads during the roll window of index funds. Whetherlagged or contemporaneous effects are considered, Granger causality testsfail to find a statistically significant link between weekly positions of indexfunds and movements in spreads. Irwin et al. conclude that other factorsmust be responsible for the non-convergence between cash and futuresprices.

Buyuksahin and Robe (2010) investigate the source of the recent increasein correlation between stock and commodity returns. The authors firstdocument that the correlation between commodity (S&P-GSCI) and stockindex returns (S&P 500 IndexTM) from January 1991 to August 2008 gener-ally fluctuated between -0.40 and 0.40. Almost immediately after the col-lapse of Lehman Brothers in September 2008, the correlation spiked ashigh as 0.70 and remained at historically high levels until March 2010.Buyuksahin and Robe utilize disaggregated data from the CFTC’s largetrader reporting system to test whether the market activity of differentgroups of traders in commodity futures markets is helpful in explainingthe increasing correlation. After accounting for business cycle and othereconomic factors, only hedge fund activity is a significant factor. In con-trast to Tang and Xiong (2010), no evidence is found that index fundactivity (as measured by swap dealer positions) is associated with theincreasing correlation between commodity and stock returns.16 The differ-ence in results is likely due to the detailed information on trading activi-ties of all major participants in commodity futures markets used byBuyuksahin and Robe.

16Silvennoinen and Thorp (2010) also investigate the source of increasing correlation between commod-ity and financial returns and reach similar conclusions to Buyuksahin and Robe (2010).

Index Funds and Futures

23

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 24: Index Funds, Financialization, and Commodity Futures Markets

Kilian and Murphy (2010) estimate the price elasticity of crude oil (andgasoline) demand because theory indicates a zero or near zero price elas-ticity of demand is required for speculation to impact spot prices in a stor-able commodity market (see Hamilton 2009b). The authors develop astructural VAR (vector autoregression) model of the global market forcrude oil that explicitly allows for shocks to the speculative demand forcrude oil, as well as shocks to the flow demand and flow supply. Theirmedian estimate of the short-run price elasticity of demand for crude oil is-0.41, which is about seven times higher than typical conjectures in therecent literature. In a related fashion, the median estimate of the short-runprice elasticity of demand for gasoline is -0.20, several times higher thanprevious estimates. Since the short-run elasticity estimates are substan-tially larger than zero (in absolute terms), Kilian and Murphy concludethat speculation did not play a major role in the run-up in crude oilprices. The model estimates imply that the surge in crude oil prices from2003-2008 was caused by fluctuations in the flow demand for oil driven bythe global business cycle. As with all structural modeling exercises, Kilianand Murphy’s structural VAR model depends on a fairly lengthy list ofidentifying restrictions. The impact of these restrictions on elasticity esti-mates will undoubtedly be the subject of further research.

Overall, the results of the studies reviewed in this section provide nosystematic evidence of a relationship (statistically or economically)between positions of index funds and the level of commodity futuresprices or spreads between commodity futures prices. This finding is con-sistent across a variety of commodity futures markets, measures of indexfund activity, statistical tests, sample periods, and time horizons. Of par-ticular note is the robustness of the results across combined on- andoff-exchange index fund positions, netted or non-netted swap dealer pos-itions, and individual ETF positions. It should also be noted that the lackof a relationship is consistent with the majority of academic evidence per-taining to speculation and price behavior (e.g., Petzel 1981; Sanders, Boris,and Manfredo 2004; Bryant, Bessler, and Haigh 2006; Sanders, Irwin, andMerrin 2009). Finally, there is mild evidence of a negative relationshipbetween index fund positions and the volatility of commodity futuresprices, consistent with the traditional view that speculators reduce risk inthe futures market and therefore lower the cost of hedging.

Summary and Conclusions

Investment in long-only commodity index funds soared over the lastdecade. This surge and its attendant impacts have been labeled the “finan-cialization” of commodity futures markets. In view of the scale of thisinvestment – at least $100 billion – it is not surprising that a worldwidedebate has ensued about the role of index funds in commodity futuresmarkets. Those who believe index funds were responsible for a bubble incommodity futures prices (e.g., Masters 2008) make what seems to be anobvious argument – that the sheer size of index investment overwhelmedthe normal functioning of these markets. Those who do not believe indexfunds were behind the run-up in commodity futures prices in recent yearspoint out logical inconsistencies in the bubble argument, as well as severalcontradictory facts (e.g., Irwin, Sanders, and Merrin 2009; Pirrong 2010).

Applied Economic Perspectives and Policy

24

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 25: Index Funds, Financialization, and Commodity Futures Markets

Not surprisingly, a flurry of studies has been completed recently in anattempt to sort out which side of the debate is correct. One group ofstudies finds evidence that commodity index investment directly orindirectly had an impact on commodity futures prices (Gilbert 2009, 2010;Einloth 2009; Tang and Xiong 2010). Results in these studies negate theargument that no evidence exists of a relationship between commodityindex investment and movements in commodity futures prices. However,the data and methods used in these studies are subject to a number ofimportant criticisms that limit the degree of confidence one can place intheir results. Another group of studies fails to find systematic evidence ofa relationship (statistically or economically) between positions of indexfunds and the level of commodity futures prices (Buyuksahin and Harris2009; Brunetti and Buyuksahin 2009; Stoll and Whaley 2010; Sanders andIrwin 2010a, 2010b, 2010c; Irwin and Sanders 2010; Buyuksahin and Robe2010). The results are robust across combined on- and off-exchange indexfund positions, netted or non-netted swap dealer positions, and individualexchange-traded fund (ETF) positions, as well as a variety of statisticaltests, sample periods, and time horizons.

Even if one ignores the data and methodological concerns in the firstgroup of empirical studies, the weight of the evidence is not consistentwith the argument that index funds created a bubble in commodity futuresprices. Whether the wave of index fund investment simply overwhelmednormal supply and demand functions (Masters 2008), channeled investors’views about commodity price directions (Gilbert 2010), or integrated finan-cial and commodity markets (Tang and Xiong 2010), the linkage betweenthe level of commodity futures prices and market positions of index fundsshould be clearly detectable in the data. Very limited traces of this linkageare visible, however. To date, no “smoking gun” has been found.Moreover, results of studies that test for a bubble component in commodityfutures prices – regardless of the cause – are decidedly mixed (Gilbert2009; Phillips and Yu 2010; Kilian and Murphy 2010). Therefore, it isunclear if there was a bubble in commodity futures price from 2007-2008,and even less clear whether one was caused by index funds.

From this perspective, the current controversy surrounding index fundscan be viewed as simply the latest in a long history of attacks upon specu-lation in commodity futures markets (Jacks 2007). For example, Labys andThomas (1975, p. 287) investigated the 1973-75 commodity boom anddescribed the situation in terms remarkably similar to the present contro-versy, “This paper analyses the instability of primary commodity pricesduring the recent period of economic upheaval, and determines the extentto which this instability was amplified by the substantial increase infutures speculation which also occurred. Of particular interest is thedegree to which this speculation rose and fell with the switch of speculat-ive funds away from traditional asset placements and towards commodityfutures contracts.” While index funds have become sizable participants incommodity futures markets, there is once again precious little evidencethat these “new speculators” drove price movements.

Important implications for public policy follow from the conclusion thatindex funds were not a primary driver of the 2007-08 commodity priceboom. New limits on speculation are not grounded in well-establishedempirical findings and could impede the price discovery and risk-shiftingfunctions of these markets. In particular, limiting the participation of

Index Funds and Futures

25

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 26: Index Funds, Financialization, and Commodity Futures Markets

index fund investors would diminish an important source of risk-bearingcapacity at a time when such capacity is in high demand. Commodityfutures markets would become less efficient mechanisms for transferringrisk from parties who don’t want to bear it to those that do, creatingadded costs that ultimately get passed back to producers in the form oflower prices and back to consumers as higher prices. Moreover, newspeculative position limits in futures may simply push index fund inves-tors into physical commodity markets (Kaminska 2010), where indexinvestments would not be subject to speculative futures position limits.Index fund positions in physical (cash) markets would provide a muchmore direct means by which financial investors could impact price discov-ery and stockholding in commodity markets; seemingly not the desiredoutcome of the regulatory drive to limit futures speculation.

There are several issues related to the 2007-2008 spike in commodityprices that merit further investigation. First, a number of potential expla-nations for the spike have been offered beyond speculation – includingdemand growth from China and other developing economies, biofuel pol-icies, monetary policy, trade restrictions, and supply shortfalls – but thereis no consensus about their relative impacts. While some researchers havebegun to tackle this question (e.g., Frankel and Rose 2009; Kilian andMurphy 2010; Martin and Anderson 2010; Heady and Fan 2010), muchmore effort is needed before a clear consensus can be reached. Second,even if index funds did not affect the level of commodity futures prices,this does not preclude other impacts. A logical place to investigate is themarket for storage. Classical economic writers such as Cootner (1961) andWeymar (1968) argued that the introduction of futures trading in a com-modity market flattens the supply of storage curve because the activity offutures speculators increases risk-carrying capacity. It is not unreasonableto posit that the large-scale addition of index investors increased stock-holding capacity in a similar fashion, a point made by Verleger (2009)with respect to crude oil. The available evidence on index funds’ impacton the market for storage has so far been limited to time-series analysis ofthe behavior of futures spreads (Stoll and Whaley 2010; Irwin et al. 2011).Research is needed that investigates possible structural impacts on thesupply and demand for storage in commodity markets. Third, the “finan-cialization” of commodity futures markets generally has been defined interms of the rise of commodity index funds. The emerging evidence(Etula 2009; Buyuksahin and Robe 2010) indicates that other types oftraders, such as broker-dealers and hedge funds, play key roles in trans-mitting shocks to commodity futures markets from other sectors. This is apromising avenue for increasing our understanding of the pricingdynamics in commodity futures markets.

Acknowledgements

The authors thank Kevin Norrish of Barclays Capital for providing data on assetsin commodity index funds. Nicole Aulerich, Jean Cordier, Linda Fulponi, JimHamilton, Todd Petzel, Aaron Smith, Carl Zulauf and participants in the OECDWorking Party on Agricultural Policies and Markets provided helpful commentson earlier versions of this article.

Applied Economic Perspectives and Policy

26

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 27: Index Funds, Financialization, and Commodity Futures Markets

References

Ai, Chunrong, Arjun Chatrath, and Frank Song. 2006. On the Comovement ofCommodity Prices. American Journal of Agricultural Economics 88(3): 574–88.

Baffes, John, and Tassos Haniotos. 2010. Placing the 2006/08 Commodity Boominto Perspective. Policy Research Working Paper 5371, The World Bank,Washington, D.C.

Barberis, Nicholas, Andrei Shleifer, and Jeffrey Wurgler. 2005. Comovement.Journal of Financial Economics 75(2): 283–317.

Barlvey, Gadi. 2007. Economic Theory and Asset Bubbles. Economic Perspectives.Federal Reserve Bank of Chicago, third quarter, p. 44–59.

Bjornson, Bruce, and Colin A. Carter. 1997. New Evidence on AgriculturalCommodity Return Performance under Time-Varying Risk. American Journal ofAgricultural Economics 79(3): 918–30.

Black, Fisher. 1976. The Pricing of Commodity Contracts. Journal of FinancialEconomics 3(1-2): 167–69.

Brunetti, Celso, and Bahattin Buyuksahin. 2009. Is Speculation Destabilizing?Working Paper, U.S. Commodity Futures Trading Commission, Washington,DC.

Brunnermeier, Markus K. 2008. Bubbles. In The New Palgrave Dictionary ofEconomics, Second Edition, Steven N. Durlauf, and Lawrence E. Blume eds., pp.578–82. New York, NY, Palgrave Macmillan.

Bryant, Henry, David A. Bessler, and Michael S. Haigh. 2006. Causality in FuturesMarkets. Journal of Futures Markets 26(11): 1039–57.

Buyuksahin, Bahattin, and Jeffrey H. Harris. 2009. The Role of Speculators in theCrude Oil Futures Markets. Working Paper, U.S. Commodity Futures TradingCommission, Washington, D.C.

Buyuksahin, Bahattin, and Michael A. Robe. 2010. Speculators, Commodities, andCross-Market Linkages. Working Paper, U.S. Commodity Futures TradingCommission, Washington, D.C.

Caballero, Ricardo J., Emmanuel Farhi, and Pierre-Olivier Gourinchas. 2008.Financial Crash, Commodity Prices, and Global Imbalances. Brookings Papers onEconomic Activity (Fall): p. 1–55.

Carter, Colin A. 1999. Commodity Futures Markets: A Survey. Australian Journal ofAgricultural and Resource Economics 43(2): 209–47.

Commodity Futures Trading Commission (CFTC). 2008. Staff Report onCommodity Swap Dealers and Index Traders with CommissionRecommendations. http://www.cftc.gov/stellent/groups/public/@newsroom/documents/file/cftcstaffreportonswapdealers09.pdf.

Cootner, Paul H. 1961. Common Elements in Futures Markets for Commoditiesand Bonds. American Economic Review 51(2): 173–183.

Davis, Ann. 2009. Cargill’s Inside View Helps It Buck Downturn. The Wall StreetJournal. January 14. http://online.wsj.com/article/SB123189501407679581.html?mod=rss_whats_news_us.

de Gorter, Harry, and David R. Just. 2010. The Social Costs and Benefits ofBiofuels: The Intersection of Environmental, Energy and Agricultural Policy.Applied Economics Perspectives and Policy 32(1): 4–32.

De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, and Robert J.Waldmann. 1990. Noise Trader Risk in Financial Markets. Journal of PoliticalEconomy 98(4): 703–38.

De Schutter, Olivier. 2010. Food Commodities Speculation and Food Price Crises:Regulation to Reduce the Risks of Price Volatility. Briefing Note 02 by theUnited Nations Special Rapporteur on the Right to Food.

Domanski, Dietrich, and Alexandra Heath. 2007. Financial Investors andCommodity Markets. Bank for International Settlements Quarterly Review (March),p: 53–67.

Index Funds and Futures

27

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 28: Index Funds, Financialization, and Commodity Futures Markets

Einloth, James T. 2009. Speculation and Recent Volatility in the Price of Oil.Working Paper, Division of Insurance and Research, Federal Deposit InsuranceCorporation, Washington, D.C.

Engelke, Lynn, and Jack C. Yuen. 2008. Types of Commodity Investments. In TheHandbook of Commodity Investing, Frank J. Fabozzi, Roland Fuss, and DieterKaiser eds., pp. 549–69. Hoboken, NJ: John Wiley and Sons.

Erb, Claude B., and Campbell R. Harvey. 2006. The Strategic and Tactical Value ofCommodity Futures. Financial Analysts Journal 62(2): 69–97.

Etula, Erkko. 2009. Broker-Dealer Risk Appetite and Commodity Returns. StaffReport No. 406, Federal Reserve Bank of New York.

Forsythe, Robert, Thomas Palfrey, and Charles Plott. 1984. Futures Markets andInformational Efficiency: A Laboratory Examination. Journal of Finance 39(4):955–81.

Frankel, Jeffrey A., and Andrew K. Rose. 2009. Determinants of Agricultural andMineral Commodity Prices. Working Paper, Kennedy School of Government,Harvard University.

Garber, Peter M. 2000. Famous First Bubbles. Cambridge, MA: MIT Press.Garcia, Phillip, and Raymond M. Leuthold. 2004. A Selected Review of

Agricultural Commodity Futures and Options Markets. European Review ofAgricultural Economics 31(3): 235–72.

Gheit, Fadel. 2008. Testimony before the Subcommittee on Oversight andInvestigations of the Committee on Energy and Commerce, U.S. House ofRepresentatives, June 23. http://energycommerce.house.gov/images/stories/Documents/Hearings/PDF/Testimony/OI/110-oi-hrg.062308.Gheit-testimony.pdf.

Gilbert, Christopher L. 2009. Speculative Influences on Commodity Futures Prices,2006-2008. Working Paper, Department of Economics, University of Trento,Trento, Italy.

———. 2010. How to Understand High Food Prices. Journal of Agricultural Economics61(2): 398–425.

Gorton, Gary B., and K. Geert Rouwenhorst. 2006. Facts and Fantasies aboutCommodity Futures. Financial Analysts Journal 62(2): 47–68.

Grossman, Sanford J. 1986. An Analysis of “Insider Trading” on Futures Markets.Journal of Business 59 (2, part 2): S129–46.

Hamilton, James D. 2009a. Understanding Crude Oil Prices. Energy Journal 20(2):179–206.

———. 2009b. Causes and Consequences of the Oil Shock of 2007-08. BrookingsPapers on Economic Activity (Spring): p. 215–61.

Hartzmark, Michael L. 1987. Returns to Individual Traders of Futures: AggregateResults. Journal of Political Economy 95(6): 1292–1306.

Heady, Derek, and Shenggen Fan. 2010. Reflections on the Global Food Crisis:How Did It Happen? How Has It Hurt? And How Can We Prevent the NextOne? Research Monograph 165, International Food Policy Research Institute,Washington, D.C.

Hieronymus, Thomas A. 1977. Economics of Futures Trading for Commercial andPersonal Profit, Second Edition. New York, NY: Commodity Research Bureau.

Hong, Harrison, and Motohiro Yogo. 2010. Commodity Market Capital Flow andAsset Return Predictability. Working Paper, Department of Economics,Princeton University.

Hull, John C. 2000. Options, Futures, and Other Derivatives. Fourth Edition. UpperSaddle River, NJ: Prentice Hall.

Irwin, Scott H., Phillip Garcia, Darrel L. Good, and Eugene L. Kunda. Forthcoming.Spreads and Non-Convergence in CBOT Corn, Soybean, and Wheat Futures: AreIndex Funds to Blame? Applied Economics Perspectives and Policy 22.

Irwin, Scott H., and Dwight R. Sanders. 2010. Testing the Masters Hypothesis inCommodity Futures Markets. Working Paper, Department of Agricultural andConsumer Economics, University of Illinois at Urbana-Champaign.

Applied Economic Perspectives and Policy

28

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 29: Index Funds, Financialization, and Commodity Futures Markets

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 and Applied Economics 41(2): 393–402.

Jacks, David S. 2007. Populists versus Theorists: Futures Markets and the Volatilityof Prices. Explorations in Economic History 44(2): 342–62.

Kaminska, Izabella. 2010. Why Base-Metal Physical ETFs Could Be a Bad Deal.Financial Times, October 27. http://ftalphaville.ft.com/blog/2010/10/27/384851/why-base-metal-physical-etfs-could-be-a-bad-deal/.

Kilian, Lutz, and Dan Murphy. 2010. The Role of Inventories and SpeculativeTrading in the Global Market for Crude Oil. Working Paper, Department ofEconomics, University of Michigan.

Kindleberger, Charles P. 1996. Manias, Panics, and Crashes: A History of FinancialCrises, Third Edition. New York, NY: John Wiley and Sons.

Korniotis, George M. 2009. Does Speculation Affect Spot Price Levels? The Case ofMetals with and without Futures Markets. Finance and Economics DiscussionSeries 2009-29, Divisions of Research & Statistics and Monetary Affairs, FederalReserve Board, Washington, D.C.

Krugman, Paul. 2008. More on Oil and Speculation. New York Times, May 13.http://krugman.blogs.nytimes.com/2008/05/13/more-on-oil-and-speculation/.

Labys, Walter C., and Harmon C. Thomas. 1975. Speculation, Hedging andCommodity Price Behavior: An International Comparison. Applied Economics7(4): 287–301.

Martin, Will, and Kym Anderson. 2010. Trade Distortions and Food Price Surges.Paper presented at the World Bank-UC Berkeley Conference on Agriculture forDevelopment – Revisited, Berkeley, CA, October 1-2.

Masters, Michael W. 2008. Testimony before the Committee on Homeland Securityand Government Affairs, U.S. Senate. May 20. http://hsgac.senate.gov/public/_files/052008Masters.pdf.

———. 2010. Testimony before the U.S. Commodity Futures Trading Commission.March 25. http://cftc.gov/ucm/groups/public/@newsroom/documents/file/metalmarkets032510_masters.pdf .

Newbold, Paul. 1982. Causality Testing in Economics. In Time Series Analysis:Theory and Practice I, O.D. Anderson ed., pp. 701–16. Amsterdam, Netherlands:North Holland Publishing Company.

Norrish, Kevin. 2010. Commodity Investor Audience Survey Results. 6th AnnualCommodity Investor Annual Conference, Barcelona, Spain, March 4-5.

Noussair, Charles, and Steven Tucker. 2006. Futures Markets and BubbleFormation in Experimental Asset Markets. Pacific Economic Review 11(2): 167–84.

O’Hara, Maureen. 2008. Bubbles: Some Perspectives (and Loose Talk) fromHistory. Review of Financial Studies 21(1): 11–17.

Patel, Jayendu, Richard Zeckhauser, and Darryll Hendricks. 1991. The RationalityStruggle: Illustrations from Financial Markets. American Economic Review 81(2):232–36.

Petzel, Todd E. 1981. A New Look at Some Old Evidence: The Wheat MarketScandal of 1925. Food Research Institute Studies 18(1): 117–28.

———. 2009. Testimony before the Commodity Futures Trading Commission.July 28. http://www.cftc.gov/ucm/groups/public/@newsroom/documents/file/hearing072809_petzel2.pdf.

Phillips, Peter C.B., Yangru Wu, and Jun Yu. 2009. Explosive Behavior in the 1990sNasdaq: When did Exuberance Escalate Asset Values? Cowles FoundationDiscussion Paper No. 1699, Yale University.

Phillips, Peter C.B., and Jun Yu. 2010. Dating the Timeline of Financial BubblesDuring the Subprime Crisis. Cowles Foundation Discussion Paper No. 1770,Yale University.

Pirrong, Craig. 2008. Restricting Speculation Will not Reduce Oil Prices. The WallStreet Journal. June 11. http://online.wsj.com/article/SB121573804530744613.html.

Index Funds and Futures

29

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 30: Index Funds, Financialization, and Commodity Futures Markets

———. 2010. No Theory? No Evidence? No Problem! Regulation. (Summer): 38–44.Robles, Miguel, Maximo Torero, and Joachim von Braun. 2009. When Speculation

Matters. Issue Brief 57, International Food Policy Research Institute, Washington,D.C.

Routledge, Bryan R., Duane J. Seppi, and Chester S. Spatt. 2000. EquilibriumForward Curves for Commodities. Journal of Finance 55(3): 1297–1328.

Sanders, Dwight R., Keith Boris, and Mark Manfredo. 2004. Hedgers, Funds, andSmall Speculators in the Energy Futures Markets: An Analysis of the CFTC’sCommitments of Traders Reports. Energy Economics 26(3): 425–45.

Sanders, Dwight R., and Scott H. Irwin. 2008. Futures Imperfect. New York Times.July 20. http://www.nytimes.com/2008/07/20/opinion/20irwinsanders.html?_r=1&oref=slogin.

———. 2010a. A Speculative Bubble in Commodity Futures Prices? Cross-SectionalEvidence. Agricultural Economics 41(1): 25–32.

———. 2010b. The Impact of Index Funds on U.S. Grain Futures Prices. WorkingPaper, Department of Agricultural and Consumer Economics, University ofIllinois at Urbana-Champaign.

———. 2010c. The Impact of Index Funds in Commodity Futures Markets: ASystems Approach. Working Paper, Department of Agricultural and ConsumerEconomics, University of Illinois at Urbana-Champaign.

Sanders, Dwight R., Scott H. Irwin, and Robert P. Merrin. 2009. Smart Money? TheForecasting Ability of CFTC Large Traders. Journal of Agricultural and ResourceEconomics 34(2): 276–96.

———. 2010. The Adequacy of Speculation in Agricultural Futures Markets: TooMuch of a Good Thing? Applied Economics Perspectives and Policy 32(1): 77–94.

Silvennoinen, Annastiina, and Susan Thorp. 2010. Financialization, Crisis andCommodity Correlation Dynamics. Research Paper 267, Quantitative FinanceResearch Centre, University of Technology Sydney.

Smith, James L. 2009. World Oil: Market or Mayhem? Journal of EconomicPerspectives 23(3): 145–64.

Stoll, Hans R., and Robert E. Whaley. 2010. Commodity Index Investing andCommodity Futures Prices. Journal of Applied Finance 20(1): 7–46.

Tang, Ke, and Wei Xiong. 2010. Index Investing and the Financialization ofCommodities. Working Paper, Department of Economics, Princeton University.

Till, Hilary. 2009. Has There Been Excessive Speculation in the US Oil FuturesMarkets? What Can We (Carefully) Conclude from New CFTC Data? RiskInstitute Position Paper, Ecole des Hautes Etudes Commerciales du Nord(EDHEC).

Tirole, Jean. 1982. On the Possibility of Speculation under Rational Expectations.Econometrica 53(6): 1499–1528.

Trostle, Ronald. 2008. Global Agricultural Supply and Demand: FactorsContributing to the Recent Increase in Food Commodity Prices. Outlook ReportNo. WRS-0801, Economic Research Service, U.S. Department of Agriculture.

United States Senate, Permanent Subcommittee on Investigations (USS/PSI). 2009.Excessive Speculation in the Wheat Market. Washington, DC: U.S. GovernmentPrinting Office.

Jr.Verleger, Philip K. 2009. The Great Glut: The Influence of Passive Investors.Petroleum Economics Monthly 26(3): 1–37.

Working, Holbrook. 1948. Theory of Inverse Carrying Charge in Futures Markets.Journal of Farm Economics 30(1): 1–28.

———. 1949. The Theory of Price of Storage. American Economic Review 39(6):1254–62.

———. 1960. Speculation on Hedging Markets. Food Research Institute Studies 1(2):185–220.

Weymar, F. Helmut. 1968. Dynamics of the World Cocoa Market. Cambridge, MA:The M.I.T. Press.

Applied Economic Perspectives and Policy

30

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from

Page 31: Index Funds, Financialization, and Commodity Futures Markets

Wright, Brian. 2009. International Grain Reserves and Other Instruments toAddress Volatility in Grain Markets. Policy Research Working Paper 5028, TheWorld Bank, Washington, D.C.

Zulauf, Carl R., and Scott H. Irwin. 1998. Market Efficiency and Marketing toEnhance Income of Crop Producers. Review of Agricultural Economics 20(2):308–31.

Index Funds and Futures

31

at Pennsylvania State University on February 27, 2013

http://aepp.oxfordjournals.org/D

ownloaded from