Public Disclosure Authorized

WPS5371 Policy Research Working Paper 5371

Public Disclosure Authorized

Placing the 2006/08 Commodity Price Boom into Perspective

John Baffes Tassos Haniotis

Public Disclosure Authorized

Public Disclosure Authorized

The World Bank Development Prospects Group July 2010

Policy Research Working Paper 5371

Abstract
The 2006-08 commodity price boom was one of the longest and broadest of the post-World War II period. Apart from strong and sustained economic growth, the recent boom was fueled by numerous factors, including low past investment in extractive commodities, weak dollar, fiscal expansion, and lax monetary policy in many countries, and investment fund activity. At the same time, the combination of adverse weather conditions, the diversion of some food commodities to the production of biofuels, and government policies (including export bans and prohibitive taxes) brought global stocks of many food commodities down to levels not seen since the early 1970s. This in turn accelerated the price increases that eventually led to the 2008 rally. The weakening and/or reversal of these factors coupled with the financial crisis that erupted in September 2008 and the subsequent global economic downturn, induced sharp price declines across most commodity sectors. Yet, the main price indices are still twice as high compared to their 2000 real levels, begging once more the question about the real factors affecting them. This paper concludes that a stronger link between energy and nonenergy commodity prices is likely to be the dominant influence on developments in commodity, and especially food, markets. Demand by emerging economies is unlikely to put additional pressure on the prices of food commodities. The paper also argues that the effect of biofuels on food prices has not been as large as originally thought, but that the use of commodities by financial investors (the so-called financialization of commodities) may have been partly responsible for the 2007/08 spike. Finally, econometric analysis of the long-term evolution of commodity prices supports the thesis that price variability overwhelms price trends.

This papera product of the Development Prospects Groupis part of a larger effort in the department to gain a better understanding of the causes and consequences of the 200608 commodity price boom. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at jbaffes@worldbank.org.

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Produced by the Research Support Team

Placingthe2006/08CommodityPriceBoom intoPerspective
JohnBaffes

THEWORLDBANK
jbaffes@worldbank.org TassosHaniotis EUROPEANCOMISSION anastassios.haniotis@ec.europa.eu

Theviewsexpressedinthispaperarethoseoftheauthorsandshouldnotbeattributedtotheir affiliatedinstitutions.WewouldliketothankAtamanAksoy,JulieDana,BettyDow,ChrisGil bert, Louis Goreux, Bernard Hoekman, and Gauresh Rajadhyaksha for comments and sugges tionsonpreliminarydrafts.RachelWeavingprovidedexcellentediting.Thepaperisacontribu tiontotheWorldBankDFIDGlobalTradeandFinancialArchitectureproject.

1.Introduction
The 200608 commodityboomwasoneofthe longestand broadest ofthepost WWII period. The boomand especially the 2008 rally, when crude oil prices peaked at US$ 133/barrel (up 94 percent from a year earlier) and rice prices doubledwithinjustfivemonthshasrenewedinterestinthelongtermbehavior and determinants of commodity prices, and raised questions about whether commodity prices have reversed the downward course that most of them fol lowedduringmostofthepastcentury.1Ithasalsoproducednumerouscallsfor coordinatedpolicyactionsatthenationalandinternationalleveltoaddressfood availabilityandfoodsecurityconcerns.2 To put the recent commodity boom into perspective calls for a good un derstandingofthekeycharacteristicsanddeterminantsoflongtermcommodity pricemovementsandanappreciationofhowlimitedthisunderstandingis,es pecially with respect to the conditions under which the recent boom unfolded. Suchaperspectiveisimportantinordertoavoidpolicypitfallsthatinthename ofmitigatingfoodsecurityconcernsorimprovingthefunctioningofthemarkets may,infact,exacerbateexistingproblems. Thispaperhastwoobjectives.Thefirstistoanalyzethenatureofthere centboom,especiallyinfoodcommodities,byexaminingwhich keyfactorsfu eled it and whether such factors are likely to remain in place in the long term. The second objective is to place the boom into perspective by examining long term trends and characteristics of commodity prices. The next section begins withadiscussionofrecentpricetrends,includingthecausesoftheboomaswell asacomparisonwithearlierepisodesofhighprices.Particularattentionispaid tothreekey(realorperceived)causesoftheboom:excessliquidityandspecula tion,fooddemandgrowthbyemergingeconomies,anduseofsomefoodcom moditiestoproducebiofuels.Section3analyzesthelongtermbehaviorofcom modity prices, including stationarity, comovement among prices of food com modities, and the price link between energy and nonenergy commodities. The final section summarizes and discusses some policy issues, including the ratio nalityandviabilityofproposalsfordealingwithpricespikes. We conclude that a stronger link between energy and nonenergy com moditypricesislikelytohavebeenthedominantinfluenceondevelopmentsin commodity, and especially food, markets. Demand by developing countries is unlikely to have put additional pressure on the prices of food commodities, al thoughitmayhavecreatedsuchpressureindirectlythroughenergyprices.We also conclude that the effect of biofuels on food prices has not been as large as originally thought, but that the use of commodities by investment funds may havebeenpartlyresponsibleforthe2007/08spike.Finally,econometricanalysis
Page2

ofthelongtermevolutionofcommoditypricessupportsthethesisthatpriceva riabilityoverwhelmspricetrends.

2.TheNatureandCausesoftheRecentCommodityBoom
The recent commodity boom emerged in the mid2000s after nearly three dec adesoflowanddecliningcommodityprices(Figure1).Thelongtermdeclinein realpriceshadbeenespeciallymarkedinfoodandagriculture.Between197576 and 200001, world food prices declined by 53 percent in real USdollar terms. Suchpricedeclinesraisedconcerns,especiallywithregardtothewelfareofpoor agriculturalproducers.Infact,oneoftheDohaRoundschiefmotives(andalso one of its perceived main obstacles) was the reduction of agricultural support andtradebarriersinhighincomecountriesasetofreformsthatwasexpected toinduceincreasesincommoditypricesandhenceimprovethewelfareoflow income commodity producers (Aksoy and Beghin 2005). Starting in the mid 2000s,however,mostcommoditypricesreversedtheirdownwardcourse,even tuallyleadingtoanunprecedentedcommoditypriceboom. Between2003and2008,nominalpricesofenergyandmetalsincreasedby 230percent,those of food andpreciousmetalsdoubled, and thoseoffertilizers increased fourfold. The boom reached its zenith in July 2008, when crude oil prices averaged US$ 133/barrel, up 94 percent from a year earlier. Rice prices doubledwithinjustfivemonthsof2008,fromUS$375/toninJanuaryto$757/ton inJune. The recent boom shares two similarities with the two earlier major com modity booms of the postWWII period, during the Korean War and the early 1970senergycrisis(seeRadetzki(2006)foradiscussionofthethreebooms).Each ofthethreeboomstookplaceagainstabackdropofhighandsustainedeconomic growth as well as an expansionary macroeconomic environment, and each was followedbyasevereslowdownofeconomicactivity.Andallthreetriggereddis cussionsoncoordinatedpolicyactionstoaddressfoodandenergysecuritycon cerns. Yettherecentboomalsoshowssomeimportantdifferencesfromthepre viousones.Bymostaccounts,itwasthelongestlastingandthebroadestinthe numbers of commodities involved. It was the only one that simultaneously in volvedallthreemaincommoditygroupsenergy,metals,andagriculturewith itspeakshowingfoodandagriculturepricesincreasinglessthanallothercom modity prices(World Bank 2009). It was not associatedwithhighinflation,un liketheboomofthe1970s(althoughtheincreaseinfoodpriceshadsomenota ble, albeit shortlived, impact on inflation). Finally, it unfolded simultaneously with the development of two other boomsin real estate and in equity mar
Page3

ketswhoseend led most developedcountriestotheirmostseverepostWWII recession. The recent boom took place in a period when most countries, especially developingones,sustainedstrongeconomicgrowth.During200307,growthin developing countries averaged 6.9 percent, the highest fiveyear average in re cent history (Figure 2). Yet apart from broad and prolonged economic growth, thecausesoftherecentboomwerenumerous,includingmacroandlongtermas wellassectorspecificandshorttermfactors. Fiscal expansion in many countries and lax monetary policy created an environment that favored high commodity prices.3 The depreciation of the US dollarthe currency of choice for most international commodity transactions strengtheneddemand(andlimitedsupply)fromnonUS$commodityconsumers (and producers). Other important contributing factors include low past invest ment,especiallyinextractivecommodities;4investmentfundactivitybyfinancial institutionsthatchosetoincludecommoditiesintheirportfolios;andgeopolitical concerns,especiallyinenergymarkets. Inthecaseofagriculturalcommodities,priceswereaffectedbythecom binationofadverseweatherconditionsandthediversionofsomefoodcommod itiestotheproductionofbiofuels(notablymaizeintheUSandedibleoilsinEu rope). That led to global stocktouse ratios of several agricultural commodities down to levels not seen since the early 1970s, further accelerating the price in creases(Figure3).Policyresponsesincludingexportbansandprohibitivetaxes thatwereintroducedin2008tooffsettheimpactofincreasingworldfoodprices contributedtocreatingtheconditionsfortheperfectstorm. Theweakeningand/orreversalofthesefactors,coupledwiththefinancial crisisthateruptedinSeptember2008andthesubsequentglobaleconomicdown turn, induced sharp price declines across most commodity sectors. But though commoditypriceshavedeclinedsharplysincetheirmid2008peak,theypicked upagainrecentlyandthekeycommoditypriceindicesarestilltwiceashighas theirearly2000slevels(Figure4). Thus the key question is whether at least some of the factors behind the recent boom are more permanent in nature, and likely to remain in place. Past experiencerevealsthatfoodcommoditypricespikesweremainlydrivenbyneg ativesupplyshocks,withhighpricesoftenactingasthebestincentiveformiti gatingtheshocksthatgeneratedthem.Yetthepertinenceofsuchexperiencefor futuredevelopmentshasbeenquestioned,andinattemptstoexplainthecurrent boom,somefactorshavereceivedconsiderablymoreattention(orevensubjected toaconsiderableamountofmisinformation)thanothers.Withthisinmind,the restofthissectionexaminesthecontributionsmadebythreesuchfactors,name ly, excess liquidity and speculation, income growth and dietary changes in
Page4

emerging economies, and the diversion of some food commodities to biofuel production. Excessliquidityandspeculation Duringthecourseoftherecentboom,ithasbeenarguedoftenthatfundamentals donottellthewholestoryandcommoditypriceshavebeendriven,inpart,by factors that go beyond demand and supply considerations. Excess liquidity, whichisoftenplacedwithinthebroadercontextofspeculation,isoftenargued tohavefueledacommoditybubble. Togaugetheimportanceoftheissue,considerthefollowing.Betweenthe secondhalfof2007andthefirsthalfof2008productionofpetroleumincreased from 85.8 million barrels per day (mb/d) to 86.8 mb/d. Consumption fell from 86.5 mb/d to 86.3 mb/d. Prices should have fallen. In December 2007, crude oil averaged US$ 90/barrel while in June 2008 it averaged US$ 132/barrel, almost 50% up. Recent figures on spare capacity give an equally perplexing picture. During2009,OPECsparecapacitystoodat6.3mb/dwhilepetroleumpricesav eraged$62/barrel.However,similarcapacitylevelsduringtheearly2000swere associated with $20/barrel. Stocks of key food commodities are 20% higher in 2009/10 compared to 2007/08; yet the nominal food price index averaged 23% higherinDecember2009comparedtoayearago,rathersurprisinggiventhatan often cited reason for the food price spike of 2008 was low inventories. Admit tedly, the apparent divergence between commodity prices and fundamentals deserves(andhasreceived)attention. While fundamentals have played a key role in commodity price move ments,observationssuchastheabovehaveledmanyresearchersandanalyststo argue that the commodity price boom reflected, in part, excess liquidity and speculation. In other words, a lot of new money (a result of excess liquidity) chasedtoofewassetsandeventuallyfounditswayintocommoditymarkets,in turncausingaspeculativebubble. Various concepts related to speculation have been discussed often inter changeablyandhavebeenanalyzedinisolationfromeachother,ortheyarepre sented in an oversimplified manner, especially in various editorials and blogs. Theseconceptsinclude:excessliquidity(oneofthethreesourcesofnewmon ey); index fund activity (the chief investment vehicle for the new money); speculation(anactivitynotwelldefinedbutnecessaryforthefunctioningoffu tures exchanges and sometimes the source of market manipulation); the role of commodity futures exchanges (the nerve centers of commodity price discovery mechanisms); inventories (which matter most for nonperishable agricultural commodities,mostlygrains);andspeculativebubble(averybroadanddifficult
Page5

to quantify issue). Not surprisingly, such complexity has led to different views andmixedempiricalevidence.Theremainingofthissectionelaboratesonthese concepts (see Appendix A for a discussion on speculation in commodity mar kets). Broadlyspeaking,thenewmoneywhichisnotassociatedwithphysi calcommoditytransactionscanbelinkedtothreesources: Diversificationofinvestmentvehicles.Duringthepastdecadeorso,invest mentfundmanagersnoticedthatexistingassetclasseswerebecomingincrea singlycorrelated.Intheirsearchfornew(uncorrelated)assetstheybroadened theirportfoliosbyincludingassetsfromemergingeconomies.Whentheseas setsbecamethemselvescorrelated(becauseofthesystemiceffectofallfunds investedinthesameassets)fundmanagersbeganinvestingincommodities, thus setting the stage for the socalled financialization of commoditiesa roletypicallyreservedforgold,after1973.Totheextentthatcommoditiesare viewedasanotherassetclass,diversificationmayhaveapermanentcharac ter. Rebalancingofinvestmentportfolios.Therebalancingofinvestmentportfo lios by shifting funds from US$denominated (and other) holdings to com moditiesaddedfurtherinflowsintocommoditymarkets.Thestronginverse relationship between the US$ and commodity pricesespecially crude oil has been invoked often as evidence for such rebalancing (see Medlock and Jaffe 2009).5 The effect of rebalancing is less permanent than diversification and depends largely on how long investors risk attitudes will favor com moditiesordisfavorotherassets. Excessliquidity.Thelowinterestrateenvironmentsupportedbymanycen tral banks resulted in excess liquidity, part of which found its way to com modity markets (on top of the price increase due to expansion of physical demandforcommodities).Itisbelievedthatexcessliquidityhasbeenthekey reasonbehindtheboominrealestatemarkets.Theeffectofexcessliquidityis likelytolastforaslongasinterestratesremainlow.

The key channel through which the new money found its way into commoditymarkets isindexfunds.Themost widelyused andcloselywatched indicesaretheDowJonesAIGandS&PGoldmanSachsCommodityIndex(also known as DJAIG and S&PGSCI). About 95 percent of funds indexed to com modities are replicated by these two indices. The funds take long positions in commodityfuturesexchangesbybuyingcontractsand,priortoexpiration,roll ing them over. While there are no precise estimates on their size, a broadly ac ceptedrangeasofmid2008was$250to$300billion(Masters2008).Areportby amajorcommercialbankestimatedthatanadditional$60billonwentintocom
Page6

moditiesduring2009,placingthe2009totalestimateto$230240billion,margi nallylowerfrom2008duetoweakercommodityprices.Althoughtheseamounts of money represent about one percent of the global value of pension and sove reignwealthfundholdings(bothkeycontributorstoindexfunds,withtheglob alvalue ofthese two groups estimated at$20and $4trillion,respectively)they arelargecomparedtothesizeofcommoditymarkets.6 Theeffectofthenewmoneyoncommoditypriceshasbeenassociated withspeculationthatmighthaveledtoapricebubble.Theviewsonthesubject, however,havebeen,forthemostpart,extreme.Forexample,Krugmaninase riesofNewYorkTimesblogsandeditorials,notonlyrejectedtheviewthatspecu lationfueledtheboombutalsodismissedtheideathatcommoditytradingactiv ityinfuturesexchangesmayhaveaffectedcommoditypricesatall,arguingthat afuturescontractisabetaboutthefutureprice.Ithasno,zero,nadadirectef fect onthe spotprice(NewYork Times,June23,2008). Otherstoo haveargued thatspeculationplayednorole.Wolfisoftheopinionthatifspeculationwere raisingpricesabovethewarrantedlevel,onewouldexpecttoseeinventoriespil inguprapidly,assupplyexceedstherateatwhichoilisburned.Yetthereisno evidenceofsuchaspikeininventories(FinancialTimes,May13,2008).Frankel citedtheCongressionaltestimonybythechiefeconomistoftheCommoditiesFu turesTradingCommission(CFTC)onApril3,2008tosupportinhisweblog(Ju ly 25, 2008) that The evidencedoes not support the claim that speculation has beenthesourceof,orhasexacerbatedthepriceincreases.Verleger(2009),based on IEA (2009) analysis, concluded at his CFTC testimony that the increase in crudeoilpricebetween2007and2008wascausedbytheincompatibilityofenvi ronmentalregulationsandspeculationhadnothingtodowiththepricerise.7The IOSCO Task Force (2009), formed at the request of G8 to examine the issue of speculationincommoditymarkets,foundthateconomicfundamentals,notspe culative activity, are the possible explanation for recent price changes in com modities.8Wright(2009)echoedsimilarviewsbynotingthatiflongfuturesposi tionswerebehindthegrainpricespikeof2008,stockswouldhaveincreased. At the other end of the spectrum, Soros (2008) called commodity index tradingathisUScongressionaltestimonyintellectuallyunsound,potentiallydes tabilizing, and distinctly harmful in its economic consequences. Eckaus (2008) andKhan(2009)authoredpapersentitledTheOilPriceIsReallyaSpeculative Bubble and The 2008 Oil Price Bubble, respectively. Calvo (2008) noted that speculationandlowinventoriesarenotnecessarilyinconsistentwitheachother andconcludedthat[increasesin]commoditypricesaretheresultofportfolio shiftagainstliquidassetsbysovereigninvestors,sovereignwealthfunds,partly triggeredbylaxmonetarypolicy,especiallyintheUS.Roubini(2009)saidthis ontheearly2009crudeoilpriceincrease:improvingfundamentalsjustify
Page7

oilgoingfrom$30tomaybe$50.Ithinktheother$30isallspeculativedemand feedingonitspeculatorsandherdingbehavior.Similarviewswereechoedby othersaswell,includingMedlockandJaffe(2009)andWray(2008). The empirical evidence on the subject has been mixed. Two IMF studies (2006, 2008) found no evidence that speculation had systematically influenced commodityprices.Asimilarconclusionwasreachedbyaseriesofstudiesunder takenbytheUSCommoditiesFuturesTradingCommission,theagencythatre gulatesUSfuturesexchanges.9Sanders,Irwin,andMerrin(2008)expressedskep ticismabouttheassertionthatspeculationhasledtobubblesinagriculturalfu turesprices. Other authors share somewhat different views. Robles and others (2009) identified speculative activity in the futures market as a source of the 2007/08 agricultural commodity price increases. Plastina (2008) concluded that between January 2006 and February 2008, investment fund activity might have pushed cottonprices14percenthigherthantheywouldotherwisehavebeen.Inthenon ferrous metals market, Gilbert (2007) found no direct evidence of the impact of investor activity on the prices ofmetals, but found strong evidence that the fu tures positions of index providers had affected the prices of soybeans (though not of maize) in the US futures exchanges. Perhaps, the strongest evidence is a subsequentstudybyGilbert(2010:420)whoconcludedthatByinvestingacross theentirerangeofcommodityfutures,indexbasedinvestorsappeartohavein flatedfoodcommodityprices. Whyistheempiricalevidencemixed?Oftenitisarguedthatdespitethe fact that investment fund activity remained steady or even increased between midandend2008,pricesdeclinedsharplyduringthatperiod.Hence,theargu mentgoes,investmentfundsdidnotaffectcommodityprices.However,thatis based on the logic that if investment fund activity continued to increase, then prices would increase forever (or, at least they would not decline), an unlikely outcome.Supplyanddemandfundamentalswillprevail,eventually.Thus,from an empirical perspective, the key question should be whether investment fund activity contributed to the recent boom (or any boom, for that matter), not whetheritaffectscommoditypricesingeneral.Consequently,examiningtherole ofinvestmentfundsincommoditymarketsrequiresfirst,theidentificationofthe boomperiodandsecond,eitheranalyzethatperiodinisolationperhapswithin aunivariatemodelingframeworkoruseofearlierperiods(orsomeotherprior) ascounterfactuals. Identifyingthesuspectperiodwillcontainalargeelementofsubjectivi ty and hence will inevitably lead to differing opinions. However, there are at leastthreereasonswhyinvestmentfundactivitymayhaveinfluencedcommodi typrices.First,investmentincommoditiesisarelativelynewphenomenon,and
Page8

fundshavetodateflowedmostlyin,notout,implyingthatsomemarketsmay havebeensubjecttoextrapolativepricebehavior;thatis,highpricesleadingto morebuyingbyinvestmentfunds,inturnleadingtoevenhigherprices,andso on.Infact,thatmayhavebeenthecasewiththesovereignwealthfunds,whose revenue comes from commodities, which in turn it is invested in commodities. Second,indexfundsinvestonthebasisoffixedweightsorpastperformancecri teria, and hence investment often behavesdifferently from whatmarket funda mentalswoulddictate.Third,thelargesizeofthesefundscomparedtocommod itymarketsmayexacerbatepricemovements.Or,asSoros(2008:3)characteristi callyputittheinstitutionsarepilinginononesideofthemarketandthey havesufficientweighttounbalanceit. Whatconclusioncanbederivedfromofallthisinthecontextofexcessli quidityandspeculation?Wasinvestmentfundactivityatleastpartlyresponsible fortherecentcommodityboom?Anycommodityrelatedactivityonthefinancial side is unlikely to alter longterm price trends, which will ultimately be deter minedbymarketfundamentals.But,suchactivitiescaninducehigherpriceva riabilityinthesenseofexacerbatingthelengthandtheamplitudeofpricecycles, astheymostlikelydidduringtheperfectstormof2007/08. Dietarychangesandincomegrowthinmiddleincomecountries Typicallyagriculturalpriceboomsarelinkedtosupplyshockssuchasweather events or animal diseases that disturb normal production patterns. Supply shocks were no exception in the recent boom. Droughts played a major role in the reduction of dairy exports from New Zealand. Australias grain production wasseverelyaffectedbythethreedroughtsexperiencedduring2002/08(ahighly unusualweather pattern,oftenlinkedtoglobal warming).The 2009 droughtin SouthAmericaaffectedtheoilseedsmarketformorethanayear.Butthemagni tude of these events, although cumulative and coming after a long period of normal weather patterns, falls short of explaining the extent of the food price spike. Could a demand shock offer a more plausible explanation, as has often beensuggested?Forthistohavebeenthecase,suchashockshouldhavetaken placeeitherunexpectedlyandsuddenlyorthrougharapidshiftinlongtermex pectationsaboutfooddemandpatterns.Ithasoftenbeenarguedthatastructural shift has taken place in the demand for grain by emerging countries, including ChinaandIndia,andespeciallyduringthepastdecadewhenthesetwocountries experiencedhighincomegrowth.TheJune2009issueofNationalGeographic,for example, noted that as countries like China and India prosper and their people moveup thefood ladder,demand for grains has increased.Similarar
Page9

guments have been advanced by noted scholars as well. Krugman argued that theresthemarchofthemeateatingChinesethatis,thegrowingnumberof people in emerging economies who are, for the first time, rich enough to start eatinglikeWesterners(NewYorkTimeseditorial,April7,2008).Likewise,Wolf asked So why have prices of food risen so strongly? and then answered strong rises in incomes per head in China, India, and other emerging countries haveraiseddemandforfood,notablymeatandtherelatedanimalfeeds(Finan cialTimes,April29,2008).Indeed,thesizeofChinaandIndia,whichtogetherac countfor27percentoftheworldspopulation,impliesthatevenaminorchange intheirpatternofdemandgrowthhasamajoreffectonworldmarketprices. Butacloserlookatthegrowthtrendsofpopulationandincomeoverthe past decades, coupled with those of demand for food commodities, shows no evidence that food demand growth accelerated either in China and India or in theworldasawhole.Table1summarizesdemandgrowthpatternsforanumber ofkeyfoodcommoditiessince1961forfour12yearperiodsroughlycorrespond ingtofourpricecycles:theperiodofthegreenrevolution(196172);theafter mathofthetwoenergyshocks(197384),therecoveryofagriculturalpricesuntil their mid1990s price spike (198596), and the last period until the recent price peak(199708).Thedataclearlyshowthatdemandgrowthhasslowedformost grainsincluding those used for feed, reflecting a slowdown in the growth of demandformeat. Duringthemostrecentdecade,despiteaclearaccelerationofGDPgrowth since2003,strongerdemandforagriculturalproductsbothatworldlevelandin China and India has been the exceptionit occurred in maize and in soybeans (drivenbydemandforedibleoils),andwasrathermixedingrains,butcertainly did not occur in meats or dairy products (see Table 2 with a breakdown of the 19972008 period for the world, China, and India). Similar findings on the role (or,thenonrole)ofChinaandIndiahavebeendiscussedinAlexandratos(2008: 673)whoemphaticallystatedthattheir[ChinasandIndias]combinedaver age annual increment in consumption (both growth rates and absolute incre ments)waslowerintheyearsofthepricesurges,200208,thaninthepreceding period19952001.FAO(2009)arrivedatnearlyidenticalconclusions. These developments reflect the huge gap that existed during the price boombetweenthefundamentalsofagriculturalmarketsandthecorresponding price levels. No other example demonstrates this better than the fact that the highestprice increasestookplacein two commoditieswheatandricewhere fooddemandwasstagnatingandyetwerewidelyexplainedasbeingdrivenby strongfooddemand.Thus, whilesupply shocksmayexplain someofthe price pressures in certain food commodities, by contrast demand growth accelerated inrecentyearsincommoditiesusedforbiofuels,suchasmaizeandedibleoils.
Page10

Biofuels The increasing interaction between the price movements of energy and non energycommoditiesduringtheboomfocusedattentionontheimpactofgrow ing demand for biofuels, including for maizebased ethanol (mainly in the US) and oilseedbased biodiesel production (mainly in Europe). During the boom, maizeandcrudeoilpricesmovedintandem,pointingtoanemergingnewand fixedrelationshipbetweenthem.Obviously,maizeanditsuseforethanolmoved into the picture as significant factors affecting price developments. But how muchimpactwasthere,andwasthereasimilaroneinoilseeds, resultingfrom theiruseforbiodiesel? The contribution of biofuels to the recent price boom, and especially the price spike of 2007/08, has been hotly debated. Mitchell (2009) argued that bio fuelproductionfromgrainsandoilseedsintheUSandtheEUwasthemostim portantfactorbehindthefoodpriceincreasebetween2002and2008,accounting, perhaps, for as much as two thirds of the price increase. Gilbert (2010), on the other hand, found little direct evidence that demand for grains and oilseeds as biofuelfeedstockswasacauseofthepricespike. FAO (2008) compared a baseline scenario, which assumes that biofuel production will double by 2018, to an assumption that biofuel production will remainatits2007levels;itconcludedthatinthelattercasegrainpriceswouldbe 12percentlower,wheatprices7percentlower,andvegetableoilprices15per centlowerthaninthebaselinescenario.OECD(2008)arrivedatsimilarconclu sionsforvegetableoils,findingthattheirpriceswouldbe16percentlowerthan thebaselineifbiofuelsupportpolicieswereabolished;eliminatingbiofuelsubsi dieswouldhavesmallerimpactsonthepricesofcoarsegrains(7percent)and wheat (5 percent). Rosegrant (2008), who simulated market developments be tween2000and2007(excludingthesurgeinbiofuelproduction),concludedthat biofuelgrowthaccountedfor30percentofthefoodpriceincreasesseeninthat period,withthecontributionvaryingfrom39percentformaizeto21percentfor rice.Lookingahead, Rosegrantfoundthat ifbiofuelproductionweretoremain atits2007levels,ratherthanreachingitsmandatedlevel,maizepriceswouldbe lowerby14percentin2015andby6percentin2020.10 Banseandothers(2008)comparedtheimpactoftheEUscurrentmandate to(i)anomandatescenarioand(ii)amandatewherebytheUS,Japan,Brazilal soadopttargetsforbiofuelconsumption.Theyestimatethatby2020,inthebase line scenario (no mandate), cereal and oilseed prices will have decreased by 12 and7percent,respectively.IntheEUonlyscenario,thecomparablechangesare 7percentforcerealand+2percentforoilseeds.Bycontrast,undertheglobal scenario(addingbiofueltargetsinUS,Japan,andBrazil)oilseedpriceswillhave
Page11

risenby19percentandcerealpricesbyabout5percent.TheEuropeanCommis sionsownassessmentofthelongterm(2020)impactsofthe10percenttargetfor biofuels(i.e.thatrenewableenergyfortransport,includingbiofuels,willsupply 10 percent of all EU fuel consumption by 2020) predicts fairly minor impacts from ethanol production, which would raise cereals prices 36 percent by 2020, butlargerimpactsfrombiodieselproductiononoilseedprices;thegreatestpro jectedimpactisonsunflower(+15percent),whoseglobalproductionpotentialis quitelimited.Taheripourandothers(2008)simulatethebiofueleconomyduring 200106.Byisolatingtheeconomicimpactofbiofueldrivers(suchasthecrudeoil price andtheUSand EUbiofuel subsidies) fromother factors ata global scale, theyestimatetheimpactofthesefactorsoncoarsegrainpricesintheUS,EU,and Brazilat14percent,16percent,and9.6percent,respectively. AjointUSDepartmentofAgricultureandDepartmentofEnergyassess ment (USDA/USDE 2008) concluded that the recent increase in maize and soy bean prices appears to have little to do with the runup in prices of wheat and rice.Itfoundthatiftheamountsofcornusedforethanolandedibleoilusedfor biodiesel in the US had remained unchanged at their 2005/06 levels, prices in 2007/2008wouldhavebeen15percentlowerformaize,18percentforsoybean, and13percentforsoybeanoil.Theassessmentalsoconcludedthattheimpactof biofuelsproductionin2007wasa34percentincreaseinretailfoodpricesanda 0.10.15percentincreaseintheallfoodCPI. Clearly US maizebased ethanol production, and (to a lesser extent) EU biodiesel production) affected the corresponding market balances and land use in both US maize and EU oilseeds. Yet, worldwide, biofuels account for only about1.5percentoftheareaundergrains/oilseeds(Table3).Thisraisesserious doubtsaboutclaimsthatbiofuelsaccountforabigshiftinglobaldemand.Even though widespread perceptions about such a shift played a big role during the recentcommoditypriceboom,itisstrikingthatmaizepriceshardlymoveddur ing the first period of increase in US ethanol production, and oilseed prices dropped when the EU increased impressively its use of biodiesel. On the other hand,pricesspikedwhileethanolusewasslowingdownintheUSandbiodiesel usewasstabilizingintheEU. Yetwhilethedebatehasfocusedmostlyontheamountoffoodcropsthat havebeendivertedtotheproductionofbiofuels,andtheresultingeffectonpric es,lessattentionhasbeenpaidtoamoreimportantissuelinkedtothisdevelop mentthe level at which energy prices provide a floor to agricultural prices. Analytically,thisisaverycomplexissue;inadditiontothepricesoftherespec tivecommodities(energyandfeedstockforbiofuels),itinvolvesnumerousother elements,includingsubsidies,mandates,traderestrictions,andsunkcostsofthe biofuelindustry.Therefore,analystsoftenusevariousrulesofthumbtoexpress
Page12

aperceivednewrelationshipbetweenagriculturalandcrudeoilprices.Onesuch ruleisthatthepriceofmaizeexpressedinUS$/tonisroughlydoublethepriceof crude oil in US$/barrel (thus a US$ 75/barrel price for crude oil would corres pond to US$ 150/ton for maize). Other commentators (in the US) have argued thatapriceofUS$3/gallonofgasolineatthepumpisthelevelatwhichthema ize price is determined by the crude oil price. The World Bank (2009) reported that crude oil prices above US$ 50/barrel effectively dictate maize prices; this conclusion was based on the strong correlation between the maize price and crude oil prices above US$ 50/barrel and the absence of correlation below that level. The US Government Accountability Office (2009: 101) while acknowledg ingthateconomistshavedisagreedaboutthecircumstancesthatwouldmakethe 2009USbiofuelmandatesnonbinding(i.e.biofuelsbecomeprofitableatcurrent energy prices), it gave a range between $80 and $120 per barrel (the range was based on anecdotal evidence based on interviews). The empirical basis of such rulesislinkedtotheissuediscussedinthenextsection.

3.CommodityPrices:LongertermTrends
This section focuses on three key characteristics of commodity price behavior: lack of trends, comovement among prices, and a special case of the latter, i.e., thelinkbetweenenergyandnonenergycommodityprices. Trends,cycles,andeverythinginbetween The longterm behavior of commodity prices was first examined systematically by Prebisch (1950) and Singer (1950), who noted that since the late 19th century the prices of primary commodities had been declining relative to the prices of manufacturedgoods(oftenreferredtoasthebartertermsoftrade).Theywarned ofpotentialproblemsforproducersofprimarycommodities,andinfacttheno tion of declining terms of trade formed the cornerstone of the industrialization policiesthatmanydevelopingcountriespursuedduringthe1960sand1970s. ThesocalledPrebischSinger(PSH)hypothesishasbeen,perhaps,oneof the most researched topics in commodity price behavior. Early research (e.g., Spraos 1980; Sapsford 1985; Grilli and Yang 1988), which focused mainly on identifying trends, supplied broad support for PSH. However, later authors foundthatpricesdidnotsimplymovealongalineartrendbutinsteadcontained strong stochastic elements, i.e., long and irregular cycles, thus producing more mixedresults(e.g.,CuddingtonandUrzua1989;Cuddington1992).Studiesus ing better econometric techniques and longer time series allowed for structural breaks (e.g., Leon and Soto 1997; Zanias 2005; Kellard and Wohar 2006). And
Page13

very recent literature, focusing on nonlinear or timevarying alternatives (e.g., BalagtasandHolt2009),findsevenlesssupportforPSH. All this research is perhaps best summarized by Cashin and McDermott (2002) who concluded that the downward trend in real commodity prices is of little policy relevance because it is small when compared to the variability of prices.OrasDeaton(1999:27)succinctlyputit,whatcommoditypriceslackin trend,theymakeupforinvariance. Commodity price variability is at the core of the current policy debate. The difficulty associated with describing past price behavior, and hence with making inferences regarding future trends, can be inferred from Figure 1; the conclusionsreacheddependonwhattimeperiodischosenforanalysis.Statisti cally, this difficulty reflects the problem of nonstationarity, i.e. the fact that the averagepricedoesnotexistinthestatisticalsense.Table4showstheresultsof ananalysisofstationarityforpricesofsixfoodcommodities(wheat,maize,rice, soybeans,soybeanoil,andpalmoil).Forsensitivitypurposes,wereportresults fromtwotests,withandwithouttrend,bothinnominalandrealterms(wealso used US CPI in addition to MUV and the results were remarkably similar). All lendstrongsupporttononstationarity,thusreaffirmingtheconclusionsreached byCashinandMcDermott(2002)andDeaton(1999). Thefactthatcommoditypricevariabilityoverwhelmstrendshasanum ber of key implications. On the methodological side, analysis involving prices needstorecognizethatcorrelationsmaynotbemeaningfulunlesscertaincondi tions are met (see next section), and also that because a mean or a trend of the priceseriescannotbeproperlydefined,thevariabilityinpricesisdifficulttocal culate. Onthepolicyside,attemptstointroducemechanismswithpricetriggers (as has often been proposed recently) are likely to fail. In fact, the absence of trends (or simply put, the nonexistence of an average price) may be the key reasonwhyearlierpricestabilization(orother)mechanismsfailed.11Whenprices stay low for long periods, stabilization funds run out of resources, and when pricesstayhighforlongperiods,stabilizationfundstendtobemisused.Consid er,forexample,thattheagriculturalcommoditypriceindex(showninFigure1) exceededitsperiodaverage(equalto173)inallyearsduring194871andfellbe lowitinallyearsduring19812007. Comovement Because some agricultural commodities can be substituted for one another (e.g. variousedibleoils),whileresourcesontheinputside(e.g.,land,labor,andma chinery)canbeshiftedfromonecroptoanother,thechangesinfundamentalsor
Page14

policy actions in one market will eventually be transmitted to other markets as well.Thus,assessinghowthepricesofvariousfoodcommoditiesmovewithre spect to each other is paramount in understanding the way and the degree to whichmarketconditionsandpoliciesaffectprices.Examiningsuchrelationships ultimately comes down to estimating the degree of price comovement among variouscommodities. Whilethegeneralsubjectofpricecomovementhasbeenextensivelystu died in the literature, analysis of the comovement among prices of different commodities is scarce. (For a brief literature review of price comovement and thereasonswhytheissuehasnotbeenadequatelyresearchedseetheAppendix B.) Here we analyze the comovement of prices using a simple econometric model. The degree of comovement was analyzed among six food commodity prices,usingordinaryleastsquareswithannualdatafrom1960to2008:Pti=+ 1Ptj+2MUVt+3t+t,wherePtiandPtjdenotethelogarithmofcommodityprice iandjinyeart(expressedinnominaldollarterms),MUVtdenotesthedeflator,t isthetimetrend,andtdenotestheerrorterm;,1,2,and3areparametersto beestimated. TheresultsarereportedinTable5.Becausepricesarenonstationary(see previoussection)examiningthestationaritypropertiesoftheerrortermisacru cialstepinestablishingthevalidityofthemodel.Alltheregressionsshowstrong performance,withanaverageR2of0.84andwithunitrootstatisticsthatstrongly confirm the stationarity of the error term. Moreover, in all cases the slope esti mateofthepricevariableissignificantatthe1percentlevel. Theresultsimplythatit isimportantnotto analyzecommoditymarkets inisolationfromoneanother,becausetheimpactofeventsthatseeminglyaffect one market will eventually be equalized among most commodity sectors. Con sider,forexample,thepalmoil/soybeanoilparameterestimateof0.97andanR2 of0.93(Table5,bottomrow).Thissuggestsanalmostsynchronousmovementof palmandsoybeanoilprices,despitethefactthatsoybeanoilisanannualcrop produced chiefly in North and South America and palm oil is a tree crop pro duced almost exclusively in East Asia. The implication is that, whether biofuel mandates are applied to one or the other edible oil market, the effect will be eventually diffused among all edible oil markets. Not surprisingly, policies fa voringbiofuelproductioninthenameofenvironmentalbenefitsmayinfactlead to less desirable outcomes. That is, the environmental benefits from switching from fossil fuel use to, say, rapeseedbased biodiesel in Europe or soybean oil basedbiodieselintheUSmaybelessthantheenvironmentalcostsofexpanding palm oil production in East Asia.12 Similarly, prices of wheat, maize, and soy beanskeyfoodcrops,producedprimarilyintheUS,EU,andSouthAmerica
Page15

showanequallylargecomovement,astheirR2averaged0.93,muchlikethatof palmandsoybeanoil. Forinflation,bycontrast,theestimatedcoefficientiseithernotsignificant lydifferentfromzeroor,inthefewcaseswhereitissignificant,itissmall.And thetimetrendparameterestimateisalmostalwayszeroimplyingthatthereis eithernotrendorthesametrendforallprices. Theenergy/nonenergypricelink It has become increasingly clear that the energy price increases of the last few years have a permanent character. In the 20 years between 1984 and 2004, the priceofcrudeoilaveragedalittlemorethanUS$20/barrelinreal2000terms.13 Now mostanalysts andresearchersbelievethatthe new equilibrium priceof oilwillbethreetofourtimeshigherthanthis,withproportionalchangestaking placeinallothertypesofenergy,atleastinthelongterm.Ifsuchassessmentis correct,then high energy pricescoupled with thehighenergyintensityofagri cultural commodities imply that developments in nonenergy (especially food) markets will depend strongly on the nature and degree of the price links be tweenenergyandnonenergycommodities. The channels through which energy prices affect other commodities are numerous(seeforexampleFAO2002;Baffes2007;WorldBank2009).Onthecost side,energyenterstheaggregateproductionfunctionofmostprimarycommodi ties through the use of various energyintensive inputs and, often, transport of outputs over long distances. Some commodities have to go through an energy intensiveprimaryprocessingstage.Inothercases,themaininputmaybeaclose substitutetocrudeoil,aswhennitrogenfertilizerismadedirectlyfromnatural gas.And,totheextentthatsomecommoditiesareusedtoproducebiofuels(to some degree aresponseto highenergy prices),another importantdimension is addedtotheenergy/nonenergypricelink(seeearlierdiscussiononbiofuels). Weexaminedtheenergy/nonenergypricelinkbyestimatingaregression similar to the one used for the comovement estimates above.14 The results for elevencommoditypriceindicesarepresentedinTable6.Theyshowthatenergy pricesexplainaconsiderablepartofcommoditypricevariability;theadjustedR2 ofallregressionsaveraged0.85.Specifically,theparameterestimateofthenon energyindex(toprowofTable6)is0.28,implyingthata10percentincreasein energypricesisassociatedwitha2.8percentincreaseinnonenergycommodity prices,inthelongrun. ThreeearlierstudiesGilbert(1989),BorenszteinandReinhart(1994),and Baffes (2007)that estimated the elasticities of nonenergy commodity prices with respect toenergypricesreportedtheseas0.12,0.11,and 0.16,respectively
Page16

(Table7,toprow).15Whenthesampleunderlyingthecurrentanalysisisadjusted tomatchthesamplesusedinthesestudies,thepassthroughcoefficientbecomes remarkablysimilar,at0.13,0.12,and0.18,respectively. Underlying these aggregate passthrough coefficients for nonenergy commoditypricesarevariationswithinsubindices.Amongthesubindices,the highestpassthroughelasticityisinfertilizer,at0.55notsurprisingly,sinceni trogenbasedfertilizersaremadedirectlyfromnaturalgas.Interestingly,thefer tilizerandenergypriceincreasesduringtherecentboomwereinlinewiththose experienced during the first oil shock: from 1973 to 1974 phosphate rock and ureapricesincreasedfourfoldandthreefold,whilethecrudeoilpriceincreased fromUS$2.81/barreltoUS$10.97/barrel. The passthrough elasticity for agriculture, estimated at 0.27, reflects a wide range among the components of the agriculture index: beverages (0.38), food(0.27),andrawmaterials(0.11).Forthecomponentsofthefoodpriceindex, bycontrast,theelasticityestimatesfallwithinaverynarrowrange:cereals(0.28), edibleoils(0.29),andotherfood(0.22).Basedonthesameregression,Table8re portsparameterestimatesforthesixfoodcommoditiesunderconsideration.The estimatesforallsixfallwithinanarrowrange,fromalowof0.27inmaizetoa highof0.36insoybeanoil.Thisresultcontrastssharplywithestimatesformet als,whichshowahighdegreeofdiversity(seeChaudhri2001;Baffes2007). Anumberofkeyconclusionsemergefromtheseresults.First,thepricesof most commoditiesrespond strongly to energyprices, withtheresponse further strengthening in periods of high prices (the values of the estimated elasticities increase considerably when the recent boom is included in the analysis). More importantly, the difference between the last two estimates (last two columns of Table7)indicatesthattheeffectofenergypricesonthepricesofallcommodities has increased considerably when the recent boom is taken into consideration suggesting that the energy/nonenergy price link has strengthened (see earlier discussion). Many observers have attributed such strengthening of the relation shiptotheuseofbiofuels,whichalsocoincides(roughly)withtheboom.Yet,it isimportanttonotethatthestrengtheningoftheeffectofenergyonnonenergy prices is more pronounced in nonfood commodities (e.g. raw materials and metals)thaninfood.Thus,commonfactorsappeartohaveplayedamoreprom inentroleintherecentboom(seeVansteenkiste2009;Gilbert2010). Second,foodcommoditypricesrespondtoenergypricesbymovingina very synchronous manner since the elasticities fall within a very narrow range (from0.25to0.36).Suchresultnotonlyemphasizestheinterdependenceofagri cultural markets (as discussed earlier) but also indicates that since a key deter minant of food commodity prices is energy prices, analyzing food markets re quiresanunderstandingofenergymarketsaswell.
Page17

Third,thoughthetransmissionelasticitiesofenergypricestononenergy prices are broadly similar to one another, this is not the case with the inflation coefficient,estimatesofwhichvaryconsiderablyinsign,magnitude,andlevelof significance. The inflation coefficient is positive and significantly different from zero only for agriculture and some of its subindices, and effectively zero for metals and fertilizers. This implies that the relationship between inflation and nominalcommoditypricesismuchmorecomplexand,perhaps,changingover time. Thismay not be surprising if one considers that during 197280 (a period thatincludedbothoilshocks)theMUVincreasedby45percent,andthatduring 200008itincreasedbyonlyhalfasmuch.Theincreasesintheindexofnominal nonenergypricesduringthesetwoeightyearperiodswereidentical,at170per cent. Lastly, the estimates of trend parameters are spread over a wider range than the estimates of energy price passthrough and inflation. For example, the aggregate index of nonenergy prices shows no trend at all, while the index of metal prices shows an almost 2 percent positive annual trend and the index of agriculturepricesshowsa1percentnegativeannualtrend.Further,thetrendpa rameterestimatesoftheagriculturesubindicesvaryconsiderably,from0.08for raw materials to 3.12 for beveragesconfirming the point made earlier that commodity prices do not exhibit welldefined trends, even when one accounts fortheeffectofenergyprices.

4.ConcludingRemarks
Numerous factors have contributed to the recent commodity boom, and have beenanalyzedextensivelyintheliterature.Yettheirrelativeweightcontinuesto beanareaofcontention.Inthispaperweexaminedthreekeyfactorswhoserole has been somewhat controversial: speculation, the growth of demand for food commoditiesbyemergingeconomiesandtheroleofbiofuels.Weconjecturethat indexfundactivity(onetypeofspeculativeactivityamongthemanythatthe literaturerefersto)playedakeyroleduringthe2008pricespike.Biofuelsplayed someroletoo,butmuchlessthaninitiallythought.Andwefindnoevidencethat allegedstrongerdemandbyemergingeconomieshadanyeffectonworldprices. Although tentative, these conclusions provide insights into the determinants of thefuturepathofcommodityprices,whichisstilluncertain. Centralamong the uncertaintiesistherelationshipbetweenthepricesof energy and of food commodities. Our examination of the key characteristics of longerterm commodity price behavior revealed a strong link between energy andnonenergyprices,whichincreasedconsiderablyduringtherecentboom;it also revealed that comovement among the prices of food commodities is very
Page18

strong. The latter implies that events taking place in one sector (e.g., increased demandformaizefortheproductionofethanol)willaffectothermarkets(e.g., forwheat)throughreallocationofresources,especiallyland.Italsoimpliesthat policy changes in one market may affect other markets. For example, expecta tionsabouttheuseofcornforbiofuelscouldresultinhighwheatpricesevenin thepresenceofrecordlevelsofwheatstocks.Ourresultsalsoshowthatagricul tural commodity market fundamentals appear, in the short term, to be playing somewhat less of a role than in the past, tending to be overshadowed by the muchstrongerpullofenergyprices. Ourconclusionaboutthelongtermevolutionofcommoditypricesiscon sistentwithearlierliterature,andsupportsthethesisthatpricevariabilityover whelmspricetrends.Variability issuchthatthe averagepricedoesnotexistin thestatisticalsense(i.e.,pricesexhibitnonstationarybehavior),andtheconclu sionsreachedabouttrendsdependonwhattimeperiodischosenfortheanaly sis. Despiteitssimplicity,thisconclusionhasimportantimplications.Follow ingtherecentfoodpricespike,therehavebeencallsforpolicyactions,essential ly aiming to alleviate the impacts of price spikes on developing countries, throughrelianceonsomelevelofbufferstocks(whetherphysicalorvirtual).His toryhasnotbeenkindtocollectivemeasuresdesignedtopreventthedeclineor reducethevariabilityofprices.Whattypeofmeasureswouldbemorepertinent tomitigateanyundesiredeffectsofpricevariabilitywoulddependonthebetter understandingofthefactorsthatnotonlyaffect,butalsopotentiallyalter,long termpricetrends.

Page19

Endnotes

Numerousauthorshaveanalyzedtherecentcommodityboom.See,forexample,Abbott,Hurt, andTyner(2008),Timmer(2008),Gilbert(2009),Mitchell(2009),PiesseandThirtle(2009),Sarris (2009),Trostle(2008),andCoady,Dorosh,andMinten(2009).AgriculturalEconomicsdevotedan entireissuetothesubject(MastersandShiverly2008).

1

See,forexample,VonBraunandTorero(2009)onvirtualreservesandMendoza(2009)onrice insurance mechanisms. Todays discussions call to mind those in earlier booms (for example, Meadowsandothers1972).
2

Calvo (2008) and Frankel (2007) have argued that interest rates played a key role during the boom.
3

Althoughunderinvestmenthasbeencitedveryoftenasthekeyfactorintheboom,thisassess mentisessentiallyderivedexpost.Certainly,anylevelofpastinvestmentwillbeconsideredlow athighpricesandhighatlowprices.Yet,researchreportedinWorldBank(2009)showsthatthe levelofinvestmentwasrightatthetimeitwasmade.Forexample,during19802007,R&Dand investmentexpendituresbymajormultinationaloilandgascompaniestrackverycloselyoutput prices (as evidenced by their strong correlation with energy prices, R2 = 0.95). Similarly, public R&DagriculturalexpendituresfollowagriculturalGDP.
4 5

ThisinverserelationshipisinadditiontotheeffectoftheUS$exchangerateagainstcurrencies ofcommodityproducingandconsumingcountriesmentionedearlier.

Informationonpensionand,especially,sovereignwealthfundsisnotwidelyavailable.There centlypublishedfirstannualreportonsovereignwealthfundsisanattempttofillthisgap(SWF 2009).

6

Yet, Verleger (2009: 2) argued that The collapse in oil prices from July 2008 to December 31, 2008canbetiedtothepossibleliquidationoffuturespositions.Whichbegsthequestion:If theliquidationoffuturespositionswerepartlyresponsibleforthecollapseinoilprices,should notthetakingofsuchpositionsberesponsibleforthecorrespondingincreaseinoilprices?
7

TheIOSCO(2009:3)reportsrecommendations,whichwerebasedonreviewingotherstudies, were:(i)understandwithgreaterclaritytheroleofspeculativeandcommercialactivityincom modity futures markets; (ii) gain a more comprehensive view of trading activities in, and the structureof,theunderlyingmarketsthatmayaffectpriceformationoncommodityfuturesmar kets; and (iii) detect, deter, and prosecute manipulation and other trading abuses involving commodityfutures,andrelatedcommoditymarkets.
8

See,forexample,Bykahin,Haigh,andRobe(2008)andCommodityFuturesTradingCom mission(2008).
9 10

The models used in the studies discussed in this section are the following: FAO (2008) and OECD (2008) used AGLINK; Rosegrant (2008) used IMPACT; Banse and others (2008) used GTAPE; EU (2008) used ESIMPE; Mitchell (2009) used simple statistical analysis; and Gilbert (2009)usedaCAPMtypeeconometricmodel.

Suchfailedarrangementsincludethe1962InternationalCoffeeAgreement(andasubsequent series of agreements) to restrict exports and boost coffee prices, the 1972 International Cocoa Agreement, and similar efforts by producers of cotton and grains; the International Tin Agree ment;andtheInternationalNaturalRubberOrganization.Foradiscussionofsuchagreementsee Gilbert(1996)andRadetzki(2009).
11

Page20

Alargebodyofliteraturediscussesthisissue;seeforexample,Searchingerandothers(2008) andFargioneandothers(2008).
12

The low energy prices between mid1980s and early 2000s prompted most analysts to argue thatthehighpricesofthe1970swereanaberrationandthatthepre1973levelswerethenorm. ForexampleinitsMarch6,1999edition,theEconomistsleaderarticleentitledDrowninginOil concluded that (p. 19): $10 might actually be too optimistic. We may be heading for US$ 5. Thankstonewtechnologyandproductivitygains,youmightexpectthepriceofoil,likethatof mostothercommodities,tofallslowlyovertheyears.JudgingbytheoilmarketinthepreOPEC era, a normal market price might now be in the US$ 510 range. Factor in the current slow growthoftheworldeconomyandthenormalpricedropstothebottomofthatrange.Indeed, most energy analysts were forecasting real prices to average between US$ 15/barrel and US$ 20/barrelinthelongrun.Forexample,theWorldBanksnominalcrudeoilpriceforecastin1999 wasUS$18/barrelfor2005andUS$19/barrelfor2010.TheDecember2008WTIfuturescontract opened at US$ 18.88 in January 15, 2002, when it was first introduced. During 2008, crude oil pricesaveragedUS$97/barrel,almostfivetimeshigherthanthehighestforecasts.
13

Herethemodelisthesameastheoneusedaboveforcomovement,exceptthatPtidenotesthe pricesoffoodcommoditiesandPtjdenotestheenergypriceindex.Therestofthecoefficientsand variables have the same interpretation. However, in contrast to the price comovement regres sions, regressing food prices on energy prices has a welldefined endogeneity pattern; energy affects food prices but not vice versa. Thus, we estimated only the regressions with the energy priceindex(alongwithdeflatorandtimetrend)astheexplanatoryvariables.Theestimatescan beviewedasenergypricetransmissionelasticitiesratherthanjustcointegrationparameters.See Baffes(2009)forthestructureoftheindices.
14

Table7indicatesthattheelasticitiesforfoodcommoditiesarehigherthanthoseforrawmate rials and metals. This is consistent with the inputoutput table of the GTAP database, which shows that the direct energy component in the US agriculture and manufacturing sectors is 12 percentand3percentrespectively.
15

Page21

TABLE1 ANNUALGROWTH(PERCENT)INGLOBALGDP,POPULATION, ANDCONSUMPTIONFORSELECTEDCOMMODITIES

MACROVARIABLES Population GDP,real CONSUMPTION Rice Wheat Maize Soybeans PalmOil Beef Pork Poultry 196172 2.0 5.5 3.3 3.9 3.7 4.8 8.4 3.2 3.7 12.1 197384 1.8 3.4 2.7 2.9 2.5 2.6 10.2 1.8 4.9 6.9 198596 1.6 2.8 1.9 1.4 2.7 5.5 7.7 1.1 2.7 6.8 19972008 1.3 3.0 1.2 0.9 2.8 4.0 8.8 1.0 2.2 4.0

Source:AuthorscalculationsbasedonFAO,FAPRI,WorldBank,andUNdata.

TABLE2 ANNUALGROWTH(PERCENT)INCONSUMPTIONOFSELECTEDCOMMODITIES
GDP Wheat Rice Maize Soybeans PalmOil Beef Pork Poultry WORLD 19972002 2.9 1.3 1.6 1.8 5.8 10.5 0.8 2.7 4.7 200308 3.3 2.7 1.5 3.6 3.3 7.5 1.6 1.3 3.8 CHINA INDIA 19972002 8.4 0.7 0.5 2.8 16.1 22.1 3.8 2.8 4.9 200308 10.4 0.2 0.6 3.4 8.7 8.2 2.8 1.2 4.6 19972002 5.1 2.7 1.0 1.9 1.4 21.4 0.5 1.0 17.6 200308 8.5 1.8 1.9 5.5 7.1 7.2 2.2 0.4 7.7

Source:AuthorscalculationsbasedonFAO,FAPRI,WorldBank,andUNdata.

Page22

TABLE3 KEYBIOFUELSTATISTICS
200001 200203 200405 200607 200809 Biofuelsasashareofglobalgrainandoilseedarea(percent) EUoilseeds USmaize USMaizearea USGrainarea Worldgrainarea 0.00 0.13 3.63 0.99 0.16 0.06 0.27 7.32 2.00 0.32 0.15 0.37 9.45 2.79 0.43 0.24 0.76 18.03 5.68 0.85 0.34 1.11 27.54 8.44 1.26

LandusedforUSethanolfrommaizeasashareof(percent)

Notes:Theshareshavebeencalculatedbasedonaverageworldyields.

TABLE4 STATIONARITYSTATISTICSFORKEYFOODCOMMODITYPRICES,19602008
Wheat Maize Rice Soybeans Soybeanoil Palmoil NOMINALSERIES Wheat Maize Rice Soybeans Soybeanoil Palmoil 1.63 1.69 2.53 1.32 1.08 1.41 WithoutTrend ADF 2.32 1.59 1.59 1.72 1.34 1.41 PP 1.70 1.49 1.71 1.59 2.06 2.14 0.84 1.22 1.62 1.23 1.17 2.14 ADF 1.83 1.84 3.08 2.09 0.96 1.31 3.10 2.76 3.40
*

WithTrend PP 1.47 1.83 1.92 1.96 2.28 2.47 2.02 2.21 2.27 2.14 2.56 2.47

REAL(MUVDeflated)SERIES

2.11 1.99 1.31

Notes:ADFandPPdenotetheAugmentedDickeyFuller(DickeyandFuller1979)andPhillips Perron (Phillips and Perron, 1988) statistics for unit roots. Asterisks in this and the following tablesdenotesignificanceat10%(*),5%(**),and1%(***)levels.Thecorrespondingtstatisticsare 2.60,2.93,and3.58forthetestswithouttrendand3.18,3.50,and4.16forthetestswithtrend. The ADF statistic corresponds to the MacKinnon onesided pvalue. The lag length of the ADF equationswasdeterminedbyminimizingtheSchwarzlossfunctionwhilethebandwidthofthe PPstatisticwasbasedontheNeweyWestmethod.

Page23

TABLE5 PARAMETERESTIMATES:COMOVEMENTREGRESSIONS
MaizeWheat SoybeansWheat WheatRice SoyoilWheat WheatPalmoil MaizeSoybeans RiceMaize PalmoilMaize SoyoilMaize RiceSoybeans SoybeansPalmoil SoyoilSoybeans PalmoilRice SoyoilRice PalmoilSoyoil 0.29 (1.37) 0.90@ (3.20) 0.34 (1.01) 1.75@ (3.90) 0.02 (0.02) 0.16 (0.54) 0.48 (0.99) 1.27@ (2.69) 1.45@ (3.64) 0.00 (0.01) 0.54 (1.48) 0.89@ (2.09) 1.70@ (1.48) 1.94@ (3.71) 0.13 (0.44)

1
0.85 (14.40)
@

2
0.06 (0.77) 0.19@ (1.99) 0.26@ (2.36) 0.08 (0.51) 0.39@ (3.26) 0.04 (0.39) 0.13 (0.81) 0.04 (0.27) 0.13 (0.98) 0.06 (0.32) 0.41 (3.96) 0.23 (1.61) 0.18 (1.01) 0.12 (0.53) 0.08 (1.03)

100*3 0.33 (1.35) 0.45 (1.40) 0.41 (1.02) 0.03 (0.05) 0.04 (0.10) 0.01 (0.92) 0.86 (1.54) 0.17 (0.31) 0.39 (0.86) 0.79 (1.41) 0.35 (0.91) 0.47 (1.04) 0.01 (0.02) 0.71 (0.86) 0.54@ (1.79)

AdjR2 0.94 0.92 0.90 0.81 0.87 0.91 0.77 0.79 0.86 0.76 0.89 0.86 0.69 0.74 0.93

ADF 4.86*** 5.20*** 4.54*** 6.54*** 5.30*** 6.18*** 5.60*** 5.64*** 6.90*** 5.19*** 4.54*** 7.73*** 5.15*** 5.71*** 4.32***

0.78@ (10.21) 0.60@ (8.31) 0.97@ (7.83) 0.63@ (6.64) 0.85@ (10.82) 1.02@ (7.24) 1.09@ (8.00) 1.12@ (9.72) 1.01@ (7.13) 0.53@ (7.75) 1.12@ (9.82) 0.64@ (5.40) 0.64@ (5.80) 0.97@ (16.83)

Notes:Allregressionswereruninbothdirections.(@)denotesparameterestimatesignificantat the 5 percent level. We report the direction with the largest ADF statistic. The variances have beenestimatedusingWhitesmethodforheteroskedasticityconsistentstandarderrors.Forother notesseetable4.

Page24

TABLE6 PARAMETERESTIMATES:PRICEINDICESREGRESSEDONENERGYPRICEINDEX
NonEnergy Metals Fertilizers Agriculture Beverages Rawmaterials Food Cereals Edibleoils Otherfood Preciousmetals 3.03 (6.54)
@

1
0.28 (5.24)
@

2
0.12 (0.68) 0.17 (0.60) 0.30 (0.95) 0.33@ (2.43) 0.55@ (2.63) 0.51@ (3.15) 0.21 (1.39) 0.17 (0.89) 0.12 (0.58) 0.45@ (4.44) 1.05@ (7.61)

100*3 0.01 (0.02) 1.93@ (2.31) 0.39 (0.48) 0.99@ (2.73) 3.12@ (5.22) 0.08 (0.19) 0.71 (1.80) 0.87 (1.76) 0.80 (1.50) 0.42 (1.18) 1.75@ (3.68)

AdjR2 0.90 0.82 0.81 0.90 0.76 0.91 0.85 0.78 0.80 0.89 0.98

ADF 3.35** 3.30** 3.97*** 3.81*** 4.95*** 3.15** 3.85*** 3.83*** 2.82* 3.60*** 3.91***

3.77@ (4.80) 3.58@ (4.12) 2.51@ (6.90) 1.83@ (3.10) 1.85@ (4.16) 2.91@ (7.11) 3.13@ (5.94) 3.33@ (6.16) 1.86@ (6.28) 1.40@ (3.58)

0.25@ (3.14) 0.55@ (4.79) 0.26@ (5.54) 0.38@ (4.87) 0.11@ (2.15) 0.27@ (4.93) 0.28@ (4.23) 0.29@ (4.51) 0.22@ (3.81) 0.46@ (9.40)

Notes:SeeTables4and5. Source:Baffes(2009).

Page25

TABLE7 COMPARINGLONGRUNTRANSMISSIONELASTICITIES
Holtham (1988) 1967:S1 1984:S2 Nonenergy Food Rawmaterials Metals 0.08 0.17 Gilbert(1989) 1965:Q1 1986:Q2 0.12 0.25 0.11 Borensztein& Reinhart(1994) 1970:Q1 1992:Q3 0.11 0.16 0.18 0.04 0.11 0.28 0.27 0.11 0.25 Baffes(2007) 19602005 Baffes(2009) 19602008

Notes:Holthamusessemiannualdata,GilbertandBorensztein&Reinhartquarterly,andBaffes alongwiththepresentstudyannual.Gilbertselasticitiesdenoteaveragesbasedoffourspecifica tions. Holthams raw materials elasticity is an average of two elasticities based on two sets of weights.indicatesthattheestimateisnotavailable. Source:Holtham(1988),Gilbert(1989),BorenszteinandReinhart(1994),Baffes(2007,2009).

TABLE8 PARAMETERESTIMATES:INDIVIDUALCOMMODITIESONENERGYPRICEINDEX
Wheat Maize Soybeans Rice Palmoil Soybeanoil 3.27@ (6.50) 3.15@ (6.23) 3.58@ (8.11) 3.57@ (5.14) 4.94@ (6.44) 5.25@ (7.83)

1
0.30@ (5.02) 0.27@ (4.66) 0.26@ (4.92) 0.25@ (2.67) 0.35@ (3.72) 0.36@ (4.13)

2
0.12 (1.49) 0.13 (0.70) 0.25 (1.51) 0.32 (0.26) 0.01 (0.02) 0.09 (0.39)

100*3 0.49 (1.07) 0.74 (1.58) 0.82 (1.83) 1.62@ (2.78) 0.95 (1.38) 0.42 (0.53)

AdjR2 0.84 0.80 0.82 0.58 0.63 0.70

ADF 4.35** 3.49** 3.85*** 4.05*** 3.16** 2.56

Notes:Seetables4and5.

Page26

Figure1 CommodityPriceIndices(Real,MUVdeflated,2000=100)
350

Korean war

Oil shocks

Recent boom

300

250

Agriculture

200

150

100

50

0 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003 2008
Source:WorldBank

Figure2 RealGDPGrowth(AnnualPercentChange)
8

Developingcountries

4 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

Source: World Bank

Page27

Figure3 GlobalGrainStockstoUseRation(Percent)
40%

35%

IncludingChina
30%

25%

20%

15%

10% 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

Source:USDepartmentofAgriculture

Figure4 CommodityPriceIndices(Nominal,2000=100)
500

400

300

200

Agriculture
100 Jan04 Jan05 Jan06 Jan07 Jan08 Jan09 Jan10

Source:WorldBank

Page28

REFERENCES
Abbott,PhilipC.,ChristopherHurt,WallaceE.Tyner(2008).WhatsDrivingFood Prices?FarmFoundation,OakBrookIl. Agbeyebbe,TerenceD.,1992.CommonStochasticTrends:EvidencefromtheLon donMetalExchange,BulletinofEconomicResearch44:141151. Ai, Chunrong, Arjun Chatrath, and Frank Song, 2006. On the Comovement of CommodityPrices,AmericanJournalofAgriculturalEconomics88:574588. Aksoy,Ataman,andJohnC.Beghin(eds.),2005.GlobalAgriculturalTradeandDevel opingCountries.Washington,DC:WorldBank. Alexandratos, Nikos, 2008. Food Price Surges: Possible Causes, Past Experience, andLongtermRelevance,PopulationandDevelopmentReview34:599629. Anderson, Kym, Marianne Kurzweil, William Martin, Damiano Sandri, and Ernesto Valenzuela, 2008. Measuring Distortions to Agricultural Incentives, Revisited,WorldTradeReview7:675704. Banse,M.,vanMeijl,HandG.Woltjer,2008.TheImpactofFirstandSecondGen eration Biofuels on Global Agricultural Production, Trade and Land Use. GTAP Conference Paper, June 2008. https://www.gtap.agecon.purdue. edu/resources/download/3693.pdf Baffes,John,1993.DoesComovementAmongExchangeRatesImplyMarketInef ficiency?EconomicsLetters,44:273280. ________,2007.OilSpillsonOtherCommodities,ResourcesPolicy32:126134. ________,2009.MoreontheEnergy/NonEnergyCommodityPriceLink,Applied EconomicsLetters,DOI:10.1080/13504850903120683.AlsopublishedasWorldBank WorkingPaperNo.4982. Baffes,John,andIoannisKaltsas,2004.CottonFuturesExchanges:TheirPast,their Present,andtheirFuture,QuarterlyJournalofInternationalAgriculture,43:153 176. Baffes, John, and Bruce Gardner, 2003. The Transmission of World Commodity Prices to Domestic Markets under Policy Reforms in Developing Countries, JournalofPolicyReform6:159180. Baillie,RichardT.,andTimBollerslev,1989.CommonStochasticTrendsinaSys temofExchangeRates,JournalofFinance44:167181. Balagtas,JosephV.,andMatthewT.Holt,2009.TheCommodityTermsofTrade, Unit Roots, and Nonlinear Alternatives: A Smooth Transition Approach, AmericanJournalofAgriculturalEconomics91:87105. Borensztein, Eduardo, and Carmen M. Reinhart, 1994. The Macroeconomic De terminantsofCommodityPrices,IMFStaffPapers41:236261.
Page29

Bykahin,Bahattin,MichaelS.Haigh,andMichelA.Robe,2008.Commodities and Equities: A Market of One? Washington, DC: US Commodity Futures TradingCommission. Calvo,Guillermo, 2008.ExplodingCommodityPrices,LaxMonetaryPolicy, and SovereignWealthFunds.http://www.voxeu.org/index.php?q=node/1244. Cashin,Paul,C.JohnMcDermott,andAlasdairScott,1999.TheMythofComoving CommodityPrices.IMFWorkingPaperNo.169.Washington,DC:International MonetaryFund. Cashin,Paul,andC.JohnMcDermott,2002.TheLongrunBehaviorofCommodi tyPrices:SmallTrendsandBigVariability,IMFStaffPapers49:17599. CommoditiesFuturesTradingCommission(CFTC),2008.InteragencyTaskForce on Commodity Markets Releases Interim Report on Crude Oil. Washington, DC,July22. Chaudhri,Kausik,2001.LongrunPricesofPrimaryCommoditiesandOilPrices, AppliedEconomics33:531538. Coady,David,PaulDorosh,andBartMinten,2009.EvaluatingAlternativePolicy ResponsestoHigherWorldFoodPrices:TheCaseofIncreasingRicePricesin Madagascar,AmericanJournalofAgriculturalEconomics91:711722. Cowing,CedricP.,1965.Populists,Plungers,andProgressives:ASocialHistoryofStock andCommoditySpeculation,18901936.Princeton,NewJersey:PrincetonUniver sityPress. Cuddington, John T., 1992. Longrun Trends in 26 Primary Commodity Prices, JournalofDevelopmentEconomics39:20727. Cuddington, John T., and Carlos M. Urza, 1989. Trends and Cycles in the Net BarterTermsofTrade:ANewApproach,EconomicJournal99:42642. Deaton,Angus,1999.CommodityPricesandGrowthinAfrica,JournalofEconom icPerspectives13:2340. Deb, Partha, Pravin K. Trivedi, and Panayotis Varangis, 1996. The Excess Co movementofCommodityPricesReconsidered,JournalofAppliedEconometrics 11:27591. Dickey,David,andWayneA.Fuller,1979.DistributionoftheEstimatorsforTime Series Regressions with Unit Roots, Journal of the American Statistical Associa tion74:427431. Dwyer,GeraldP.,andMylesS.Wallace,1992.CointegrationandMarketEfficien cy,JournalofInternationalMoneyandFinance11:318327. Eckaus,RichardS.,2008.TheOilPriceIsaSpeculativeBubble.CenterforEnergyand EnvironmentalPolicyResearchWorkingPaper08007. Fackler,PaulL.,andBarryK.Goodwin,2001.SpatialPriceAnalysis.InB.Gard
Page30

nerandG.Rausser(eds.),HandbookofAgriculturalEconomics,vol.I.NewYork, NY:ElsevierScience. FoodandAgricultureOrganizationoftheUnitedNations(FAO),2002.Commodity MarketReview200102.Rome:FAO. ________,2008.TheStateofFoodandAgricultureBiofuels:Prospects,Risks,andOp portunities.Rome:FAO. ________,2009.TheStateofAgriculturalCommodityMarkets:HighFoodPricesandthe FoodCrisisExperiencesandLessonsLearned.Rome:FAO. Fargione, Joseph, Jason Hill, David Tilman, Stephen Polasky, and Peter Haw thorne, 2008. Land Clearing and the Biofuel Carbon Debt, Science, 319: 123538. Frankel, Jeffrey, 2008. Commodity Prices, Again: Are Speculators to Blame? http://content.ksg.harvard.edu/blog/jeff_frankels_weblog/2008/07/25/commo ditypricesagainarespeculatorstoblame Frankel, Jeffrey, 2007. The Effect of Monetary Policy on Real Commodity Pric es. In John Campbell (ed.), Asset Prices and Monetary Policy. Chicago, IL: UniversityofChicagoPress. Gilbert, Christopher L., 1989. The Impact of Exchange Rates and Developing CountryDebtonCommodityPrices,EconomicJournal99:773783. ________, 1996. International Commodity Agreements: An Obituary Notice, WorldDevelopment24:119. ________, 2007. Commodity Speculation and Commodity Investments. Revised version of the paper presented at conference on The Globalization of Primary CommodityMarkets,Stockholm,October2223. ________,2010.HowtoUnderstandHighFoodPrices.JournalofAgriculturalEco nomics61:398425. Granger,CliveW.J.,1986.DevelopmentsintheStudyofCointegratedEconomic Variables,OxfordBulletinofEconomicsandStatistics48:213228. Grilli,EnzoR.,andMawChengYang,1988.PrimaryCommodityPrices,Manufac turedGoodsPrices,andtheTermsofTradeofDevelopingCountries:Whatthe LongRunShows,WorldBankEconomicReview1:147. Hakkio,GraigS.,andMarkRush,1989.MarketEfficiencyandCointegration:An ApplicationtotheSterlingandDeutschmarkExchangeMarkets,JournalofIn ternationalMoneyandFinance8:7588. Holtham,GeraldH.,1988.ModelingCommodityPricesinaWorldMacroeconom icModel.InOrhanGuvenen(ed.),InternationalCommodityMarketModelsand PolicyAnalysis.Boston,MA:KluwerAcademicPublishers. International Monetary Fund (IMF), 2006. World Economic Outlook 2006. Washing
Page31

ton,DC:IMF. ________,2008.WorldEconomicOutlook2008.Washington,DC:IMF. International Organization of Securities Commissions (IOSCO), 2009. Task Force onCommodityFuturesMarkets:FinalReport.Madrid. Kellard,Neil,andMarkE.Wohar,2006.OnthePrevalenceofTrendsinCommodi tyPrices,JournalofDevelopmentEconomics79:14667. Khan,MohsinS.2009.The2008OilPriceBubble.PolicyBriefPB0919.Washington DC:PetersonInstituteforInternationalEconomics. Len,Javier,andRaimundoSoto,1997.StructuralBreaksandLongrunTrendsin CommodityPrices,JournalofInternationalDevelopment3:4457. LeyBourne,StephenJ.,TimA.Lloyd,andGeoffreyV.Reed,1994.TheExcessCo movementofCommodityPricesRevisited,WorldDevelopment22:17471758. MacDonald,Ronald,andMarkTaylor,1988.MetalsPrices,Efficiency,andCointe gration:SomeEvidencefromtheLondonMetalExchange,BulletinofEconomic Research40:235239. Markham,JerryW.,1987.TheHistoryofCommodityFuturesTradinganditsRegulation. NewYork,NY:PraegerPublishers. Masters,MichaelW.,2008.TestimonybeforetheCommitteeonHomelandSecuri tyandGovernmentAffairs.UnitedStatesSenate,Washington,D.C.,May20. Masters,WilliamA.,andGeraldE.Shiverly,2008.IntroductiontotheSpecialIssue ontheWorldFoodCrisis,AgriculturalEconomics39:373374. Meadows, Donella H., Dennis L. Meadows, Jrgen Randers, and William W. Be hrensIII,1972.TheLimitstoGrowth.Rome:ClubofRome. Medlock, Kenneth B. III, andAmy Myers Jaffe, 2009. Who Is in the Oil Futures MarketandHowHasItChanged?Mimeo.JamesBakerIIIInstituteforPublic Policy,RiceUniversity. Mendoza,RonaldU.,2009.AProposalforanAsianRiceInsuranceMechanism, GlobalEconomyJournal9:131. Mitchell,Donald,2009.ANoteonRisingFoodPrices.PolicyResearchWorkingPaper 4682.Washington,DC:WorldBank. Mundlak,Yair,andDonaldL.Larson,1992.OntheTransmissionofWorldAgri culturalPrices,TheWorldBankEconomicReview6:399422. OrganizationforEconomicCooperationandDevelopment(OECD),2008.Economic AssessmentofBiofuelSupportPolicies.DirectorateforTradeandAgriculture.Par is:OECD.http://www.oecd.org/dataoecd/19/62/41007840.pdf Piesse,Jenifer,andColinThirtle,2009.ThreeBubblesandaPanic:AnExplanatory ReviewoftheRecentFoodCommodityPriceEvents,FoodPolicy34:119129.
Page32

Phillips,PeterC.B.,andPerron,Pierre,1988.TestingforaUnitRootinTimeSeries Regression,Biometrika75:335346. Pindyck, Robert S., and Julio J. Rotemberg, 1990. The Excess Comovement of CommodityPrices,EconomicJournal100:11731189. Plastina, Alejandro, 2008. Speculation and Cotton Prices, Cotton: Review of the WorldSituation61:812.InternationalCottonAdvisoryCommittee. Prebisch, Ral, 1950. The Economic Development of Latin America and its Principal Problems.NewYork,NY:UnitedNationsPublications. Radetzki,Marian,2006.TheAnatomyofThreeCommodityBooms,ResourcesPol icy31:5664. Robles,Miguel,MaximoTorero,andJoachimvonBraun,2009.WhenSpeculation Matters. IFPRI Policy Brief 57. International Food Policy Research Institute. WashingtonDC. Rosegrant,Mark,2008.InternationalFoodPolicyResearchInstituteBiofuelsand Grain Prices. http://beta.irri.org/solutions/images/publications/papers/ if pri_biofuels_grain_prices.pdf Roubini,Nouriel(2009).BigCrushComing.http://www.indexuniverse.com/sec tions/features/6777nourielroubinibigcrashcoming.html Sanders, Dwight R., Scott H. Irwin, and Robert P. Merrin, 2008. The Adequacy of SpeculationinAgriculturalFuturesMarkets:TooMuchofaGoodThing?Marketing and Outlook Research Report 200802, Department of Agricultural and Con sumerEconomics,UniversityofIllinoisatUrbanaChampaign. Sapsford,David,1985.TheStatisticalDebateontheNetBarterTermsofTradebe tweenPrimaryCommoditiesandManufacturers:ACommentandSomeAddi tionalEvidence,EconomicJournal95:78188. Sarris,Alexander, 2009. Evolving Structure of WorldAgricultural Trade and Re quirementsforNewWorldTradeRules.Paperpresentedatthe12thAnnual ConferenceonGlobalEconomicAnalysis,Santiago,Chile. Schaede, Ulrike, 1989. Forward and Futures in TokugawaPeriod Japan: A New PerspectiveontheDjimaRiceMarket,JournalofBankingandFinance45:487 513. Searchinger, Timothy, Ralph Heimlich, R. A. Houghton, Fengxia Dong, Amani Elobeid, Jacinto Fabiosa, Simla Tokgoz, Dermot Hayes, TunHsiang Yu, 2008. UseofUSCroplandsforBiofuelsIncreasesGreenhouseGasesthroughEmis sionsfromLandUseChange,Science319:123840. Sephton,PeterS.,andHansK.Larsen,1991.TestsofExchangeMarketEfficiency: Fragile Evidence from Cointegration Tests, Journal of International Money and Finance10:561570.
Page33

Singer,HansW.,1950.TheDistributionsofGainsbetweenInvestingandBorrow ingCountries,AmericanEconomicReview40:47385. Soros, George, 2008. Testimony before the U.S. Senate Commerce Committee Oversight: Hearing on FTCAdvanced Rulemaking on Oil Market Manipula tion.Washington,D.C.,June3.http://www.georgesoros.com/files/SorosFinal Testimony.pdf Spraos, John, 1980. The Statistical Debate on the Net Barter Terms of Trade be tweenPrimaryCommoditiesandManufactures,EconomicJournal90:107128. SWF,SovereignWealthFundAnnualReport,2009.WeatheringtheStorm:Sovereign Wealth Fundsin the Global Economic Crisis of 2008. Monitor Group and Fonda zioneEniEnricoMattei,April. Taheripour,F.,D.K.Birur,T.W.Hertel,andW.E.Tyner,2008.Biofuelsandtheir Byproducts:GlobalEconomicandEnvironmentalImplications.GTAPConfe rencepaper,Helsinki,June2008. Timmer,C.Peter,2008.CausesofHighFoodPrices.WorkingPaperNo.128.Manila: AsianDevelopmentBank. Trostle,Ronald,2008.GlobalAgriculturalSupplyandDemand:FactorsContribut ing to the Recent Increase in Food Commodity Prices. Economic Research Service,USDepartmentofAgriculture,WRS0801(Junerevision). US Government Accountability Office, 2009. Biofuels: Potential Effects and Chal lenges of Required Increasesin Production andUse. Report GAO09446. Wash ington,D.C. USDepartmentofEnergy/DepartmentofTransportation,2008.ImpactofBiofuels onFoodandGasolinePrices. Vansteenkiste,Isabel,2009.HowImportantAreCommonFactorsinDrivingNonFuel Commodity Prices? A Dynamic Factor Analysis. Working Paper 1072. European CentralBank. Verleger, Phillip, 2009. Testimony to the U.S. Commodity Futures Trading Com mission on the Role of Speculators in Setting the Price of Oil. Washington, D.C.,August5. Von Braun, Joachim, and Maximo Torero, 2009. Implementing Physical and Virtual Food Reserves to Protect the Poor and Prevent Market Failure. IFPRI Policy Brief. Washington,DC:InternationalFoodPolicyResearchInstitute. WorldBank,2009.GlobalEconomicProspects:CommoditiesattheCrossroads.Washing ton,DC:WorldBank. ________, various issues. Commodity Price Data. Development Prospects Group. Washington,DC:WorldBank. Wray, Randall L., 2008. The Commodities Market Bubble: Money Manager Capitalism
Page34

andtheFinancializationofCommodities.PublicPolicyBrief96.TheLevyEconom icsInstituteofBardCollege. Wright, Brian, 2009. A Note on International Grain Reserves and Other Instru mentstoAddressVolatilityinGrainMarkets.Technicalbackgroundpaper presentedattheWorldGrainForum2009,St.Petersburg,Russia,June67. Zanias,GeorgeP.,2005.TestingforTrendsintheTermsofTradebetweenPrimary CommoditiesandManufacturedGoods,JournalofDevelopmentEconomics78: 4959.

Page35

APPENDIXA.WHATISSPECULATION?
Blaming speculative activity in periods of sharp price increases is not uncom mon.Pushedbysuchabelief,numerousattemptshavebeenmadetorestrictor evenclosetheUSfuturesmarkets.Markham(1987)reportsthatduring190709 (60thUSCongress),25billswereintroduced,designedtoprohibitfuturestrad ingfor a comprehensive treatment of all attempts to regulate and/or prohibit futures trading see Cowing (1965). Baffes and Kaltsas (2004) describe how nu merouscottonfuturesexchangeswereshutdownduringthefirsthalfofthe20th century as the result of government intervention, again because of perceived speculative activity. Schaede (1989) documents similar attempts for the Djima ricemarketinJapan,oftencitedastheworldsoldestfuturesexchangeitbegan operationin1730. Whatmakestherecentdebateonspeculationinterestingisthatithasdi vided even noted scholars and analysts. Apart from the insufficient empirical evidenceonthesubject,suchdivisionpartlyreflectsthedifferenttypesofspe culativeactivitythatanalystsandeconomistsreferto.Indeed,thelinesbetween hedgersandspeculators,betweenphysicalandfinancialtransactions,aswellas between legal and illegal trading activities are complex, blurry, and go beyond textbookdefinitions.Togaugesuchcomplexities,considerthetraditionalsepa rationoftheplaceinwhichtransactionstakeplace(physicalversusfinancial)and theactorsinvolved(hedgersversusspeculators),asdepictedintableA1.Thefirst columnshowshedgingtransactionsbyproducers,consumers,andtraders(with the banks as intermediaries) that take place either in physical or financial mar kets (the latter in commodity futures exchanges). This is the typical textbook case. The picture becomes complicated, however, when speculators engage in physicaltransactionsbyholdinginventories,keepingresourcesintheground,or engaging in various types of market manipulation. A more complex picture emergeswhenspeculatorsengageinfinancialtransactions(oftencombinedwith physicaltransactions). Hence, understanding the complexities and the controversial nature of speculationultimatelycomesdowntounderstandingtherightbottomcellofthe TableA1.Onewaytoanalyzespeculationistomapitssourcesanditseffecton commodity markets to the place where transactions take place, the actors in volved,andtheirmotivation. Placeoftransaction Commoditytransactionstakeplaceeitherinfuturesexchangesorphysicalmar kets. Speculation taking place in commodity exchanges forms the backbone of the functioning of the futures markets by injecting the necessary liquidity to
Page36

complete the transactions.This isthetype of speculativeactivitythat was typi cally given as a reason for the closure of commodity futures exchanges men tionedabove. Inthephysicalmarket,ontheotherhand,tradersmaybuyandholdlarge quantities of commodities with the expectation that an upward movement in priceswillgenerateprofits(oftencalledhoarding).Thisisthetypeofspeculative activity that Paul Krugman and Martin Wolf (among others) have referred to. Unless such activity entails market manipulation (in which case it would bean illegal activity), it is the intertemporal equivalent of Adam Smiths invisible hand:tradersbuyatcurrentpricestoselllaterwhen(intheiropinion)themar ketwillbetight,therebybalancingthemarketandhencereducingpricevariabil ity. There is no evidence that hoarding took place during the recent boom, as knowninventoriesinalmostallcommoditiesreachedhistoricallows.However, inthecaseofextractivecommoditiesespeciallycrudeoilonemaywellargue thattargetingoutputtolevelsbelowwhatthemarketfundamentalsdictate(as is the case with OPEC quotas) is a form of hoarding. The difference is that the commodityiskeptintheground,ratherthaninabovegroundandthereforeoff themarketstoragefacilities. Actorsinvolved Apart from the hedgers (e.g., producers and consumers) with interest in the physicaltransactionofcommodities,twootheractorshavebeenoperatinginthe marketduringthelasttwoorsodecadeswithpurelyfinancialincentivesandno transactionsinthephysicalmarkets.Theyarehedgefundsandcommoditytrad ing advisors (CTAs). During the past few years, investment funds (mostly pensionfundsandsovereignwealthfunds)alsoenteredthefinancialmarkets.It has been argued that these groups (mostly the latter) may have affected com modityprices.
Hedge funds. These undertake investment and trading activities in a broad range of assets, including commodity markets. A hedge fund may trade and invest in com modityassetclassesinordertohedgethediverserisksinherentintheirportfolios. Insuchacase,takingapositioninthefuturesmarketforaparticularcommodityor commodityclasscanrepresentaninvestmentinanoncorrelatedassetthatprovides diversificationbenefitstotheoverallportfolio.Hedgefundshaveexistedfordecades andtheireffectoncommoditymarketsistypicallyofshorttermnature(i.e.,theyaf fectshorttermpricemovements). Commoditytradingadvisors(CTAs)areassetmanagersthatoperatealmostexclusively incommoditymarkets.Theyinvestforportfoliosundermanagementandforclients with the objective of earning profits from market volatility. Perhaps, CTA activity reduces price volatility since they trade on the basis of market fundamentals and
Page37

technicalanalysis. Investmentfunds.Theyincludesovereignwealthandpensionfundswhichduringthe pastfewyearsbeganincludingcommoditiesintheirportfoliomixasanotherasset class. Their chief motivation has been asset diversification. In addition to the way thesefundsinvestinfuturesmarkets(i.e.,fixedweightsandpastperformancecrite ria),itistheirsheersizethatmattersmost(seediscussioninmaintext).

Motivation Veryoftenspeculationtakesplaceintheformofmarketmanipulation.Thisre ferstoillegalactivitytypicallyisolatedinoneorafewcommoditymarkets.Itcan takeplaceinthephysicalorfinancialmarkets(oftenitinvolvesboth).Wellpub licized cases are the US onion market in the 1950s, where onion producers ar guedthattradersintheChicagoMercantileExchangecorneredthemarket(this resulted in the passage of the Onions Futures Act which prohibited futures con tractsononions);theHuntbrotherswhoattemptedtocornerthesilvermarketin thelate1970sandearly1980s;Sumitomoschiefcoppertrader,YasuoHamana ka, who cornered the copper market in the 1990s; and the BP cornering of the propanemarketin2006,whichresultedinaUS$300millionfine.Suchactivityis notknowntohaveprevailedduringtherecentboom.

TABLEA1:SPECULATIONINCOMMODITYMARKETS
ACTOR PLACE

HEDGERS
Producers/consumers Traders Banks Producers/consumers Traders Banks

SPECULATORS
Holdinginventories(e.g.,hoarding) Keepingresourcesintheground(e.g.,OPEC) Marketmanipulation(e.g.,corneringthe market) Investmentfunds(e.g.,pensionfunds,sove reignwealthfunds) Investmentanddiversificationinstruments (e.g.,CTAs,hedgefunds) Marketmanipulation(e.g.,corneringthe market)

PHYSICAL

FINANCIAL

Page38

APPENDIXB.COMOVEMENTOFCOMMODITYPRICES
Thegeneralsubjectofcommoditypricecomovementhasbeenexaminedexten sivelyandinvariouscontexts.Yetsometopicshavereceivedmuchmoreatten tionthanothers.Thetopicsthathavereceivedlessattentionmattermostinthe currentpolicydebate. Overall,theresearchoncommoditypricecomovementfallslargelywith intwostrands.Thefirstexaminescomovementamongpricesofthesamecom modityindifferentlocationswithinthemarketefficiencycontext,aphenomenon alsoknownasspatialmarketintegrationorthelawofoneprice.Whilemostof thesestudiesexaminecomovementinabivariatecontext,someusemodelsca pable of examining it within a multivariate setting. This topic has been studied extensively (see Fackler and Goodwin 2001 for a literature review). A less re searched subject (though one more useful for policy analysis purposes) within thatstrandhasbeenthecomovementbetweenworldanddomesticcommodity prices,arelationshipthatincludesapolicydimension.Specifically,thesestudies examine whether world price signals have been fully transmitted to domestic markets or instead have been subjected to policy distortions (see, for example, Baffes and Gardner 2003; Mundlak and Larson 1992). Now, for agricultural products, a research project led by Kym Anderson has produced a consistent globaldatabasethatincludespricesreceivedbyfarmersandpaidbyconsumers in75countries(Andersonandothers2008).Thatdatabaseisexpectedtogenerate moreresearchonthesubject. Thesecondstrandofliteratureexaminespricecomovement(orlackthe reof) among different commodities. The genesis of this literature goes back to Granger(1986:218),whowrote:Ifxtandytareapairofpricesfromajointlyef ficient,speculativemarket,theycannotbecointegratedifthetwopriceswere cointegrated,onecanbeusedtohelpforecasttheotherandthiswouldcontra dicttheefficientmarketassumption.Thus,forexample,goldandsilverprices,if generatedbyanefficientmarket,cannotmovecloselytogetherinthelongrun. Grangersassertionledtoresearchincommoditymarkets(e.g.,MacDonaldand Taylor1988)andothermarketsaswell,notablyexchangerates(see,amongoth ers,Baillieand Bollerslev1989; Hakkio and Rush1989). This researchwaslater questionedonseveralgroundsincludingthefactthatcomovementreflectsres ponses to common fundamentals rather than market inefficiencies. (See, for ex ample,Agbeyebbe1992;Baffes1993;DwyerandWallace1992;andSephtonand Larsen1991.) A similarly controversialsubjecthas beentheexcess comovementhy pothesis first discussed by Pindyck and Rotemberg (1990) who, after analyzing pricemovementsofsevenseeminglyunrelatedcommodities(cocoa,copper,cot
Page39

ton,crudeoil,gold,lumber,andwheat),concludedthatthesepricescomovedin excess of what the macroeconomic variables could explain. A number of likely explanationsweregiven,includinganincompletemodel,endogeneityofthema croeconomicvariables,rejectionofnormalityassumption,andbubblesormarket psychology.Subsequentresearch,however,challengedtheexcesscomovement hypothesis on data and methodological grounds (see Ai, Chatrath, and Song 2006;Cashin,McDermott,andScott1999;Deb,Trivedi,andVarangis1996;and LeyBourne,Lloyd,andReed1994).Inamorerecentpaper,Vansteenkiste(2009) concluded that commodity price comovement is driven mainly by exchange rates,interestrates,andthepriceofcrudeoil. Onemayarguethattherejectionoftheefficientmarkethypothesisinthe presenceofcomovementarguedbyGranger(1986)correspondstoPindyckand Rotembergs (1990) bubbles or market psychology explanation for excess co movementprovidedthatpricesusedinGrangerssensehavebeenadjustedac cordingly by the fundamentals. Perhaps, the fact that two leading articles on price comovement among different commodities have been somewhat contro versialmayhaveledtoaslowingdownofresearchinthatareaaswell.

Page40