THE IMPACT OF COMMODITY DERIVATIVES IN AGRICULTURAL FUTURES MARKETS

Transcription

1 Alghero, June 20 Feedng he Plane and Greenng Agrculure: Challenges and opporunes for he bo-econom THE IMPACT OF COMMODITY DERIVATIVES IN AGRICULTURAL FUTURES MARKETS Zupprol M., Dona M., Verga G., Ran M. Unvers of Parma, Ial AIEAA 20

2 Bacground In recen ears, agrculural mares have experenced a hgh volal wh prces ha boos n a ver quc and srong wa and drop wh he same rapd and nens. The boom and bus of 2008 agrculural commodes prces has been nvesgaed b researchers, mare analss and polc maers whou reachng a common nerpreaon of such a dnamc and a unque explanaon of he man deermnan. Despe he researches on demand and suppl deermnans of he spes, oher auhors dspued he role of mare speculaon as he cause of mare bubbles. The analss has been addressed o undersand he relaon beween he prce dnamcs and he rend of he fnancal acv ha much ncreases s parcpaon n commod fuure mares, he ndex funds. The mpac of commod dervaves

3 Objecves Tes wheher he nvesng acv n he fuure mare of dfferen raders caegores can be denfed as a source of he ncreasng agrculural commod prces. The analss s carred ou b adopng he approach proposed b Irwn and Sanders (200) addng some relevan nnovaons and exenson: A more exended prce me seres; A procedure o chec oulers nsde he daase; The sandardzaon of each varable ncluded n he model; A es of he nverse relaonshps, consderng he mare facors as causal varables; Esmaon of he causal wh and whou he ndependen varable a conemporaneous me. The mpac of commod dervaves

4 Daa The prce behavor s descrbed hrough hree major ndcaors, he mare facors: ) Reurns of nearb fuure prces The reurns are calculaed on he seres of he nearb fuure prces as he weel average of he fve dal reurns, ransformed wh naural logarhms 2) Realzed volal The realzed volal s compued as he sandard devaon of he las 20 frs dfferences of naural logarhm on he nearb me seres, annualzed (facor 20 ). 3) Impled volal The mpled volal s provded b Thompson-Reuers onlne servce whch calculaes he mpled volal a he mone nerpolaed, dsngushng beween pu and call opons. The mpac of commod dervaves

5 Daa The weel daase are organzed wh respec o hree dmensons: mare facors, raders group and radng acv (TA) ndcaors. Daase srucure for Ne Long posons Daase srucure for Ne Shor posons The mpac of commod dervaves

6 Commodes The commodes consdered n he analss are he same of hose proposed b Irwn and Sanders (200). More specfcall: No. Abbrevaon Commod Mare Weel daa Dal daa AN Cocoa NYBT CC Corn CBOT DK HRW Whea KCBT EC Feeder Cale CME GC Lean Hogs CME HN Coffee NYBT LC Lve Cale CME NN Coon NYBT OC Sobean Ol CBOT SC Sobeans CBOT WC SRW Whea CBOT ZN Sugar NYBT The mpac of commod dervaves

7 Daa pre-reaemen Each me seres has been submed o an ouler chec usng an nnovave robus mehod proposed b Ran e al. (2007) amng o denf oulers usng a Forward Search (FS) based on a mulvarae esng. The procedure has he objecve o avod masng (false negave) and swampng (false posve) problems n he ouler deecon. The FS echnque denfes he oulers accordng o a procedure based on he esmaon mare b mare of a mulple regresson model and deecng hem hrough he deleon resdual analss Buld subses of he basc nformaon Esmae mulple regresson model b LS Analss of resduals Addng uns wh mnmum squared resduals o subses Sor of resduals (Malhanobs Dsance All he varables has been sandardzed o have he coeffcens no affeced b he daa level and comparable among commodes. The mpac of commod dervaves

8 Causal Evaluaon Sraeg We have esed he causal of he radng acv ndcaors n respec o mare facors usng he Granger s approach. Each mare was analsed consderng wo me seres as he lnear causal echnque requres. For he me and he commod : Mare facor Mare facor coeffcen Causal coeffcen Resduals,,,, x,, Inercep Lagged Mare facor Causal varable Followng he Irwn and Sanders proposal, n order o ncrease he power of he causal es, we esmaed he mares as a ssem of Seemngl Unrelaed Regresson (SUR). The mpac of commod dervaves

9 Causal Evaluaon Sraeg The mpac of commod dervaves x x x x,2,2,2,2 2,2,3,3,3,3 3,3,2,2,2,2 2,2,,,,, SUR ssem for evaluang causal We assume ha n he mare here s an errac force ha affecs all he commodes prces n he same wa (correlaon beween mare equaon resduals)

10 Causal Evaluaon Sraeg SUR models are esed consderng: Wh and whou Conemporane varables Non conemporaneous equaon,,,,,,x,, x, Conemporaneous equaon,,,,,,x,,x,, x, The coeffcens of he lagged causal varables are cleaned from he nfluence of he ndependen seres a me. Ths las model provdes an ex-pos analss of he degree of causal beween he seres. The mpac of commod dervaves

11 Causal Evaluaon Sraeg SUR models are esed consderng: Causal varables as ndependen and dependen varables Mare facors = f(lagged Mare facors, lagged Causal varables) Causal varables= f(lagged Causal varables, lagged Mare facors) Two pes of models have been esmaed: he prmar model wh he mare facor as dependen varable and he reverse model wh he causal varable as dependen varable. The analss s conssen wh he Granger s heor ha does no resrc he causal evaluaon o onl one drecon of he relaonshp. The mpac of commod dervaves

12 Man resuls The resuls acheved for all he weel seres wdel confrm hose acheved b Irwn and Sanders: a subsanal absence of causal nerference of raders behavour on he dnamcs of reurns, realzed and mpled volales. Despe hs resuls, he Lagrange mulpler sascs developed b Breusch and Pagan (980) confrms ha he SUR approach o model esmaon s alwas he mos adequae. Ths hghlghs a sgnfcance correlaon of he resduals along all he mares. All he mares seem, hus, o reac o exogenous nformaon wh a clear homogene. However, s possble ha he weel daa provde a dsored nformaon abou he real radng acv ha s carred ou on a dal bass. Ths s he reason for we have esed he model usng he onl avalable dal nformaon, ha s he VOIR. Ths analss opon does no allow o dsngush he daa accordng o he radng groups. The mpac of commod dervaves

13 Daa The relaonshps beween mare facors and he Volume o Open Ineres (VOIR) has been esmaed usng dal daa. The mpac of commod dervaves

14 Esmaes Esmaes Man resuls Dal VOIR analss 0,7 0,6 0,5 0, 0,3 0,2 0, 0-0, -0,2-0,3 RV = f(tradng Acv) Tradng Acv = f(rv) Conemporaneous.Wald << 0.05.Wald << ,7 0,6 0,5 0, 0,3 0,2 0, 0-0, -0,2-0,3 No Conemporaneous.Wald > 0.05.Wald < The mpac of commod dervaves

15 Man resuls Dal VOIR analss Realzed Volal (+)? Tradng Acv The mpac of commod dervaves

16 Conclusons The oucomes of he model based on weel daa reproduces he concluson acheved b Sanders and Irwn, confrmng ha here are an sgnal of causal nsde he nvesgaed mares. The SUR model s sascall adequae o analse he causal effec beween radng acv and mare facor dnamcs. The dal daa nclude an mporan par of he nformaon ha mgh be no capured b he weel daa. The use of conemporaneous ndependen varable allows o beer nerpre he relaonshps n an ex-pos perspecve. The homogene beween coeffcens due o prelmnar sandardsaon shows a common behavour among he mares. Fnall, we canno asser ha s lel a non-causal relaonshp beween speculave acv and commod prces, bu we can jus sugges an ndeph analss of he nfluence of he nformaon frequenc for evaluang causal n commod fuures mares. The mpac of commod dervaves