Instituto Nacional de Investigación y Tecnología Agaia y Alimentaia (INIA) Spanish Jounal of Agicultual Reseach 2010 8(1), 3-17 Available online at www.inia.es/sja ISSN: 1695-971-X EU impot estictions on genetically modified feeds: impacts on Spanish, EU and global livestock sectos G. Philippidis* Aagonese Agency fo Reseach and Development (ARAID). Govenment of Aagón. Avda. Montañana, 930. 50059 Zaagoza. Spain Abstact Ove the last decade, much contovesy has suounded the usage of genetically modified oganism (GMO) technology in commecial agicultue. Moe specifically, it is feaed that GMOs may intoduce new allegens into the food chain o contibute to antibiotic esistance. At the cuent time, the Euopean Union (EU) adopts a zeo toleance policy towad «non-appoved» GMO impots, whilst the appoval pocess has not kept pace with the polifeation of new GMO vaieties. In the EU livestock sectos, this appaent mis-match theatens to inteupt supplies of high potein feed inputs (e.g., soymeal) fom counties with moe elaxed egulations egading GMOs. Employing a well known multi-egion computable geneal equilibium famewok, this study quantitatively assesses the impact of a hypothetical EU impot ban on unappoved GMO vaieties of soybean and maize impots on livestock, meat and daiy sectos. The model code is heavily modified to impove the chaacteisation of the agicultual sectos and land usage, whilst a ealistic baseline is employed to update the global database to 2008, the yea the hypothetical ban is implemented. In the «wost case» scenaio, thee ae significant competitive losses in EU livestock, meat and daiy sectos. In Spain, the negative impacts ae paticulaly ponounced given the impotance of pig poduction in agicultue. In contast, all non-eu egions tade balances impove, with notable tade gains in the USA and Bazil. To conclude, the EU must ugently find a long tem stategy fo GMOs if it is to econcile political expediency with pagmatic economic concens. Additional key wods: computable geneal equilibium, global tade analysis poject. Resumen Las esticciones sobe impotaciones euopeas de los OMGs: impacto sobe los sectoes ganadeos españoles, euopeos y mundiales El uso de oganismos genéticamente modificados (OMGs) en los sectoes ago-ganadeos ha desatado mucha polémica. En paticula, se teme que puedan intoduci nuevos alégenos dento de la cadena agoalimentaia, o subi el nivel de toleancia hacia los antibióticos. Actualmente, la Unión Euopea (UE) aplica toleancia ceo hacia las impotaciones de OMGs «no-apobados», aunque la tasa de apobación no se mantiene en paidad con la polifeación de nuevas vaiantes de OMGs. En la UE, este desajuste podía inteumpi los suministos de piensos con alto contenido poteico desde los países que aceptan el uso de OMGs. En este estudio se emplea un modelo de equilibio geneal computable mundial, paa analiza el impacto de una pohibición hipotética de las impotaciones de soja y maíz tansgénicos no-apobados sobe los sectoes ganadeos con oientación cánica y láctea. Se modifica intensamente el modelo paa efleja con más pecisión el secto agaio y el uso de la tiea. Además, se emplea un «baseline» ealista paa actualiza la economía global hasta 2008, año en que se implanta la pohibición. En el peo escenaio planteado, se dan pédidas gandes en los sectoes ganadeos, tanto de cane como de leche, en la UE, mientas en España el impacto es peo debido a la impotancia del secto de pocino. En contaste, teceos países expeimentan ganancias en sus balanzas comeciales, especialmente EEUU y Basil. En conclusión, la UE debe adopta una estategia sobe los OMGs paa econcilia las amenazas económicas potenciales sobe los sectoes ganadeos con las peocupaciones sanitaias. Palabas claves adicionales: modelos de equilibio geneal computable, poyecto de análisis del comecio global. * Coesponding autho: gphilippidis@aagon.es Received: 18-02-09; Accepted: 12-11-09. Abbeviations used: CGE (computable geneal equilibium), DEFRA (Depatment of Envionment, Food and Rual Affais, UK), EU (Euopean Union), GMO (genetically modified oganism), GTAP (Global Tade Analysis Poject), MTR (mid tem eview), ROW (est of the wold), RPI (etail pice index), CES (constant elasticity of substitution), CET (constant elasticity of tansfomation), SFP (single fam payment), AusNZ (Austalia and New Zealand), RussiaFSB (Russia and Fome Soviet Bloc), EU3 (Austia, Nethelands, Sweden), AC2 (Bulgaia and Romania), EU15 (Euopean Union 15 membes), EU27 (Euopean Union 27 membes), EC (Euopean Commission), MARM (Ministeio de Medio Ambiente y Medio Rual y Maino, Madid), USDA (United States Depatment of Agicultue).
4 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 Intoduction With the advent of biotechnology and its peceived competitive advantages to commecial agicultue, thee has been a apid polifeation and usage of genetically modified oganisms (GMOs) ove the last 10 yeas. In contast, Euopean Union (EU) enthusiasm fo GMOs has been seiously hampeed by scientific concens elating to the possible long tem impacts on the food chain and ultimately consume health and safety issues. At the cuent time, the EU adopts a zeo toleance policy towad non-appoved GMO impots, whee if tace levels of GMO ae found, the whole shipment is efused enty 1. To futhe complicate mattes, the authoisation pocess in the EU has so fa failed to keep pace with the speed with which new stains of GMO cops ae being adopted and accepted in non-eu egions. Thee ae cuently 70 GMOs fo maize (Zea mays L.), ape (Bassica napus L.) and soybean (Glycine max (L.) Me.) in the appoval pipeline, which is expected to incease to 100 in the next two yeas (Cady-Bown, 2008). Moe discouagingly, thee appeas to be little consensus amongst membe states, o a clea long tem EU stategy egading GMO usage (EuActiv.com, 2009b). This uncetainty casts a long shadow ove the EU livestock sectos which heavily depend on feed impots. Fo example, last yea the new stain of RoundUp Ready 2 soybeans which was eady fo planting in the US, theatened EU animal feed secuity until an eleventh hou ageement was appoved by the Commission. Futhemoe, impoted substitutes fo oilseed meal (paticulaly soybean) and othe poteinich feedstuffs at the quantities equied ae only available fom limited souces, with ove 90% of EU soybean impots oiginating fom Agentina, Bazil and the USA. Owing to climatic and agonomic factos, thee is no viable pospect fo developing EU poduction of potein ich plants at shot notice. Even stepping up the poduction of substitute potein cops such as field peas, field beans and sweet lupines as altenatives fo soybean, would still leave a shotfall in meeting EU demand equiements. In the past, the EU s position was potected by its status as a key custome maket, howeve, the emegence of lage impotes such as India and China, both of which employ moe libeal egimes with espect to acceptance of GMOs, theatens to educe the EU s leveage ove supplie counties. The aim of this study is to quantitatively assess the impact on EU livestock, meat and daiy sectos fom a hypothetical EU impot ban on unappoved GMO vaieties of soybean and maize impots fom one o moe of the majo supplies (Agentina, Bazil and the USA). As a basis, the Global Tade Analysis Poject (GTAP) database (vesion 6) (Dimaanan, 2006) is employed, coveing 87 egions and 57 commodities. To achieve this aim, the accompanying computable geneal equilibium (CGE) model has been heavily modified to incopoate explicit modelling of agicultual facto, input and output makets, whilst a ealistic baseline updates the global database to 2008, the yea the hypothetical ban is implemented. Futhe to the above, the impot bans ae modelled using a novel appoach, whilst the impact of global bio-fuels poduction on competing land usage is also chaacteised in an attempt to impove the cedibility of the model estimates. Methods GTAP (Hetel, 1997) is a «demand» led model, based on a system of neoclassical final, intemediate and pimay demand functions. Given the assumption of weak homothetic sepaability, optimisation is boken into nests to allow geate flexibility though the incopoation of diffeing elasticities of substitution, whilst accounting identities and maket cleaing equations ensue a geneal equilibium solution. Once the model stuctue is calibated to the chosen data aggegation, specific exogenous macoeconomic o tade policy «shocks» can be imposed to key policy vaiables (i.e., changes to tax/subsidy ates, facto endowments, technical change vaiables etc.). The model esponds with the inteaction of economic agents within each maket, whee an outcome is chaacteised by a «countefactual» set of equilibium conditions. In this study, the standad famewok is modified in a numbe of ways. All modifications and elevant mathematical deivations ae discussed in the technical appendix. 1 Given the chemical difficulty in peventing the accidental pesence of GMOs in conventional seeds, thee has been some debate within the EU on acceptable levels of toleance. While most membe states apply the zeo toleance pinciple, the levels accepted ae 0.1% in Fance, 0.5% in the UK and 0.9% in Romania (EuActiv.Com, 2009a).
GMO feeds and impacts on livestock sectos 5 Model modifications-agicultual facto, input and output makets Following the wok on GTAP-AGR by Keeney and Hetel (2005), constant elasticity of substitution (CES) possibilities ae modelled between intemediate inputs and pimay facto demands, whilst in livestock sectos, intemediate feed inputs ae also now CES substitutable 2. A constant elasticity of tansfomation (CET) contols the tansfe of labou and capital factos between agicultual/non-agicultual sectos to captue obseved diffeentials in wages and ents in each sub-secto. Othe modifications include the incopoation of a thee-stage weakly sepaable CET nest to captue land heteogeneity acoss diffeent agicultual activities. In addition, following Tabeau et al. (2006), an endogenous non linea land supply function is econometically estimated, whilst in the EU egions, additional model code is inseted to enfoce an uppe limit to the egisteed agicultual land aea upon which the single fam payment (SFP) is based. Employing ecent developments in the liteatue, the study also incopoates an explicit epesentation of the EU s CAP (e.g., set aside, CAP budget, intevention pices, quotas, etc.), which constitutes an impotant component of ou «baseline» scenaio. Modelling an impot ban In this study, a novel method fo modelling an impot ban is poposed. Moe specifically, with educed confidence in feed impots (due to food safety feas), thee is a eduction in associated utility coesponding to that bilateal oute, which in tun motivates impot eductions. Employing cost minimisation and expessing in pecentage changes (denoted by lowecase lettes) gives: q i,,s = u i,s σ i [ p i,,s p i,s ] +σ i z i,,s [1] Lineaised Hicksian impot demands of commodity «i» fom egion to impot egion «s» (q i,,s ) ae a function of commodity pices (p i,,s ), utility (u i,s ), the utility scaling vaiable (z i,,s ) and the elasticity of substitution paamete (σ i ). Implementation of the impot pohibition, chaacteised as a downtun in «confidence» fo EU feed impots, is captued by swapping the scaling vaiable (z) with impot demands (q). The inclusion of bio fuels in the GTAP data and model In the baseline, ecent inceases in bio-fuel poduction ae chaacteised to ecognise its impact as a fom of competing land usage, paticulaly in the USA and Bazil. This study daws on two studies by Taheipou et al. (2008) and Biu et al. (2008) espectively. Taheipou et al. (2008) include additional bio-fuels activities (bio-diesel; gain based bio-ethanol; cane base bio-ethanol) to the GTAP database by splitting them out of existing sectos. A peceived advantage is that the database bette chaacteises changing pattens of land usage (paticulaly in Bazil and the US) fom apid bio-fuels expansion. Since bio-fuels ae diectly substitutable with petol at the pump, following Biu et al. (2008), adjustments ae made to the GTAP pivate demand stuctue to chaacteise «demand diven» inceases in bio-fuels poduction fom inceases in cude oil pices. Futhe discussion of the incopoation of biofuels into the model is given in pat IV of the technical appendix. Data aggegation The choice of model aggegation is detailed in Figue 1. All pimay agicultual sectos ae disaggegated 24 Sectos: ice, wheat, othe gains, vegetables fuits and nuts, oilseeds, aw suga, othe cops, cattle/sheep, pigs/ poulty, aw milk, wool, ed meat, white meat, daiy, othe food pocessing, beveages and tobacco, enegy (gas, coal, electicity), bio diesel, bio ethanol (ceeals and cane based), cude oil, efined petoleum, manufactuing; sevices. 19 Regions: Fance, Gemany, UK, Spain, EU3 (Austia, Nethelands, Sweden); Rest of EU15, Accession 10 1, Accession 2 2, RussiaFSB (Russia and Fome Soviet Bloc), Tukey, USA, Canada, Agentina, Bazil, RoLatAme (Rest of Latin Ameica), Austalia & New Zealand, China, India, ROW 3. Figue 1. GTAP data aggegation. 1 AC10: Cypus, Czech Republic, Estonia, Hungay, Latvia, Lithuania, Malta, Poland, Slovakia, Slovenia. 2 AC2: Bulgaia, Romania. 3 ROW: est of the wold. 2 The standad GTAP employs a Leontief specification. This implies that, fo example, the intensiveness of fetilise application on land cannot alte, o competing feeds ae not substitutable in livestock sectos. Substitution elasticities ae calibated to OECD cental values of Allen patial elasticities (Keeney and Hetel, 2005).
6 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 including the thee livestock sectos of cattle/sheep, pigs/poulty and aw milk. In the food pocessing sectos, ed and white meat sectos and daiy ae disaggegated to captue the impacts of inceasing feed costs in these downsteam sectos. The «new» bio-fuels sectos ae disaggegated along with an enegy composite (gas, coal, electicity), cude oil and petoleum. The emaining sectos ae captued within the composites of manufactuing and sevices. The EU consists of the «big-thee» (Fance, Gemany, UK), Spain (majo EU pok poduce) and fou composite EU egions 3. The non EU egions consist of the main supplies of maize and soybean to EU27 makets (Agentina, Bazil, USA). In addition, «lage» agicultual playes (e.g., AusNZ 4, Canada, China, India) on wold makets as well as othe potentially impotant EU tade patnes (e.g., RussiaFSB 5, Tukey) ae featued. Scenaio design In the fist pat of this expeiment a baseline scenaio is un (see Fig. 2) to captue the main tade policy dives which have occued since the benchmak yea of 2001. Maco pojections data between 2001 and 2020 on GDP, endowments and poductivity ae taken fom Walmsley (2006). The shocks between 2001 and 2008 ae calculated and aggegated to the 19 GTAP egions employed in this study. Impotantly, all taiff shocks account the taiff ovehang between the bound and applied taiff ates, employing the wok of Jean et al. (2005). In addition, the 2003 Mid Tem Review (MTR) CAP efoms ae implemented. Finally, to captue the inceased impotance of bio-fuels in global land usage, a shock to the wold pice of cude oil is implemented which coesponds to the pice ise between 2001 and 2008. The updated 2008 data ae subsequently employed as the benchmak data in the policy scenaios. In the study, thee scenaios ae examined: 1. No impots of GM soybean meal and maize fom the US to the EU. 2. No impots of GM soybean meal and maize fom Agentina and the US to the EU. Baseline assumptions 1. Uuguay Round Commitments (+) Enfoce developed county commitments (expot subsidy limits, applied taiff levels). Complete developing county commitments (expot subsidy limits, applied taiff levels). 2. EU enlagement to 27 membes (+) Remove bode potection between existing and «new» membe states. Impose common extenal taiff fo all new EU membes of the customs union. 3. Additional tade policy shocks (+) Chinese Accession. 4. Agenda 2000 (A2000) commitments and the mid tem eview (MTR) Modelling of CAP mechanisms (CAP budget, modulation, quotas, set-aside, intevention pices). Reduction of intevention pices unde A2000 and MTR efoms. Removal of ALL coupled suppot in the AC12 and MTR ageed components of coupled suppot (#) in the EU15. CAP budget including the implementation of modulation funding and the UK ebate mechanism. Full implementation of the SFP and land idling shocks. 5. Cude oil pice shock of 166% 6. Update shocks (2001-2008) ( ) Shocks to GDP, facto endowments, poductivity. Figue 2. Assumptions shaping the baseline. +: all taiff shocks account fo the binding ovehang. #: data taken fom DEFRA. : data taken fom Walmsley (2006). 3. No impots of GM soybean meal and maize fom Agentina, Bazil and the US to the EU. In the GTAP database, soybean/soymeal and maize ae subsumed within the aggegate sectos «oilseeds», «othe ceeals» and «othe food pocessing» 6. Thus, the elevant impot eductions in these GTAP sectos ae based on the aveage popotion of the affected cops within each secto. To acquie this infomation, data wee solicited fom the UN COMTRADE (2008) database fo a seies of ecent yeas (2001-2007) and aveages wee calculated. Fo example, efeing to Table 1, the equied negative shock to impots to emove non appoved maize and soybean/soymeal fom the «othe food pocessing» secto on expots fom Agentina to Spain is 42.0%. 3 Due to the modelling of the CAP budget, the EU3 (Austia, Nethelands, Sweden) must be sepaated fom othe EU egions. 4 Austalia and New Zealand. 5 Russia and Fome Soviet Bloc. 6 «Pepaed animal feeds» appea in the «othe food pocessing» secto. Fo infomation on the GTAP concodance with specific disaggegate sectos, see Dimaanan (2006).
GMO feeds and impacts on livestock sectos 7 Table 1. EU tade shae data in pecentage (2001-2007 aveage) Shae of «maize» Shae of «soybean» Shae of soyben/maize feed 1 Regions in «othe ceeals» in «oilseeds» impots in «othe food pocessing» Ag 2 Ba 3 USA 4 Ag Ba USA Ag Ba USA UK 98.0 43.4 3.3 6.7 96.5 53.8 17.7 32.7 0.8 EU3 86.1 51.3 17.4 8.8 99.1 88.8 50.5 20.2 1.2 Ge 13.4 23.9 5.5 24.4 99.7 93.5 27.1 18.4 0.2 Fa 70.0 50.0 49.4 2.3 99.3 77.4 32.1 69.5 0.7 Spa 97.6 92.5 23.4 30.2 99.5 85.5 42.0 15.6 1.1 Ro15 92.5 55.4 27.0 64.3 99.5 87.2 59.8 23.8 2.9 AC10 53.2 74.3 67.9 5.1 47.4 15.5 56.3 41.3 4.2 AC2 63.8 74.6 72.3 24.3 50.0 38.6 39.8 12.2 2.0 1 See Figue 1. 2 Agentina. 3 Bazil. 4 United States of Ameica. Souce: UN COMTRADE (2008) and own calculations. In addition to the impot shocks, estimates of inceases in feed costs ae also implemented into the EU livestock sectos. In CGE models, multistage budgeting compatmentalises input and facto demands into nests, each with an individual elasticity of substitution. Consequently, when faced with supply constaints, CGE models have a tendency to «substitute aound» poblems, theeby mitigating the impacts on poduct makets. Fo example, impoted animal feed inputs ae lagely constained to the secto «othe food pocessing», whilst the coesponding cost shae is small 7. Thus, with an elimination of non-appoved maize and soybean elated feed impots and substitution possibilities in favou of cheape «domestic» equivalents, the total cost impact in livestock sectos is unealistically modeate. An immediate esponse would be to assume Leontief (i.e., zeo) substitution technology (o something vey close to zeo) in the livestock sectos. Unfotunately, this assumption would affect the substitutability of all animal feed inputs (domestic and impoted), which is had to justify in policy tems. To einfoce the point futhe, if impoted soybean usage by pigs/poulty in Spain fell by 50%, by vitue of the Leontief assumption, one would be imposing the estiction that all inputs, and theefoe outputs, would also be falling by 50%. Accodingly, it was seen as moe desiable to implement exogenous estimates of aveage feed cost ises fom the loss of (pimaily) impoted soybean deived feeds 8. This appoach captues the «essential» natue of non-substitutable feeds without puging the essential substitutability which chaacteises input decision making in these models. Feed cost estimates will diffe between livestock activities due to diffeing dietay equiements fo soybean based feeds. To povide the animal with geate quantities of enegy and potein as well as moe apid weight gain feed concentates ae needed, of which the most impotant ae gains (maize) and oilseed meal deived fom soybean. Pigs and poulty ae lagely fed on such feed concentates. On the othe hand, uminant animals (cattle and sheep) can digest only cetain quantities of such high concentate feeds, whilst cheap «on-fam» (i.e., pastue based) souces of foage povide impotant souces of fibe. Accoding to Bookes et al. (2005), appoximately 22% of boile feed is soybean elated, whilst Cady- Bown (2008) estimate that 22-25% of high pefomance pig feed is soybean based. In addition, data fom FEDNA (2008) gives tables of limits fo the usage of soybean ingedients in diffeent types of Spanish livestock poduction, which fo pok and poulty ae also aound 20%, whilst fo cattle and sheep and daiy, the values ae close to 9% and 8% espectively. Having appoximated the cost popotion of feeds in the diffeent livestock sectos, it is necessay to employ futhe assumptions to impose plausible cost ises fom a hypothetical GM ban on soybean 9. In EC 7 Between 6-10% shae in the GTAP database. 8 In EC (2007) it is noted that the loss of maize impots fom these thee outes could be eplaced by EU substitutes, by othe domestic ceeals o by impots fom elsewhee. In addition, the loss of only US feed impots could conceivably be compensated by Agentina and Bazil. Fo this eason, the analysis does not associate feed cost ises with losses in maize, whilst in scenaio 1 no feed cost ises ae implemented. 9 To implement feed cost inceases, an exogenous Hicks neutal technical change vaiable is employed. Fo example, a 10% eduction on impoted feeds implies that to attain the same level of feed poductivity, the unit cost of impoted feed inputs is now 10% highe.
8 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 Table 2. Impacts of the genetically modified oganisms (GMO) impot ban in the EU livestock sectos in scenaio 2 (pecentages) Fa 1 Ge Spa UK EU3 R15 AC10 AC2 EU27 2 EU livestock feed demands Cattle/sheep 1.87 2.07 2.23 1.53 3.42 2.45 1.80 5.96 2.83 Pigs/poulty 7.00 7.64 7.89 7.21 7.74 8.35 8.47 10.22 8.16 Raw milk 1.82 1.79 2.42 1.75 3.12 2.14 1.44 5.07 2.35 EU livestock, meat, agicultual and economic poduction Cattle/sheep 1.37 0.92 2.09 0.69 2.24 1.38 1.68 6.45 1.92 Pigs/poulty 8.76 9.41 9.74 5.96 8.28 8.11 7.70 14.10 9.00 Raw milk 1.62 1.21 4.27 1.00 1.35 2.03 1.59 7.83 2.43 Agicultue 2.00 2.43 3.12 1.76 2.49 2.26 2.73 7.36 2.99 Red meat 0.71 0.71 0.53 0.43 1.70 1.27 1.08 4.78 1.28 White meat 2.99 5.49 4.51 2.90 6.08 5.79 8.37 5.31 5.09 Daiy 0.73 0.92 0.71 0.53 1.01 1.00 1.51 4.71 1.04 Real gowth 0.06 0.04 0.08 0.04 0.05 0.03 0.23 1.85 0.05 EU livestock, meat, agicultual and maket pice indices Cattle/sheep 1.18 0.63 1.62 0.92 0.76 1.01 1.44 3.14 1.31 Pigs/poulty 7.23 7.02 7.97 5.07 5.57 5.91 5.70 9.36 6.89 Raw milk 1.00 0.82 2.67 0.78 0.88 1.29 1.11 3.72 1.40 Agicultue 1.38 1.54 2.07 1.42 1.50 1.35 1.64 3.61 1.77 Red meat 0.82 0.89 0.75 0.75 0.36 1.08 0.89 1.82 0.97 White meat 3.67 4.49 4.70 3.57 4.09 4.23 4.94 3.87 4.16 Daiy 0.70 0.78 0.75 0.49 0.66 0.71 0.85 1.28 0.72 RPI 3 0.07 0.05 0.09 0.06 0.06 0.06 0.20 0.63 0.07 1 See Figue 1. 2 EU27: all 27 Euopean Union membes. 3 RPI: etail pice index. (2007) 10, it is estimated that aveage feed costs in the EU livestock sectos, fom the loss of US and Agentinean soybean impots, could ise by 23%. This estimate is employed in scenaio 2 fo pigs/poulty, whilst popotionate aveage feed cost ises of 10% fo cattle/sheep and 9% fo aw milk poduction ae assumed 11. In scenaio 3, EC (2007) estimate feed cost ises of 600% fom the loss of Agentinean, Bazilian and US soybean impots. This pecentage incease is well beyond the thesholds of the model. Consequently, a five-fold incease in feed costs compaed with scenaio 2 is assumed. such that the esults fo scenaio 1 ae not pesented. Indeed, given that the US maket constitutes a mino shae of EU feed impots, impot substitution effects fully mitigate against any impacts in the livestock sectos. The loss of US feed to the EU is picked up (pincipally) by Agentinian and Bazilian expotes. In scenaio 1, slightly moe favouable endogenous cost changes on the pat of Agentinian feed supplies, lead to a lage popotion of EU27 feed impots fom that egion, although these cost diven estimates ae negligible. Results Scenaio 1 The loss of non-appoved US feed impots eveals only negligible impacts in the EU27 livestock sectos, Scenaio 2 The imposition of the GM ban on Agentinean and US impots of maize and soybean has maked epecussions on Spanish livestock sectos (Table 2). In the Spanish pigs/poulty secto, feed demands fall by 10 This study is elaboated futhe in the «discussion» section. 11 These values ae based on the elative limits of soybean in the feed diets of pigs and poulty, cattle and sheep and aw milk poduction.
GMO feeds and impacts on livestock sectos 9 Table 3. Tade balance changes in scenaio 2 ( millions) Cattle/sheep Pigs/poulty Raw milk Red meat White meat Daiy Fa 1 29 1 2 35 29 Ge 0 1 0 17 190 43 Spa 0 8 0 9 109 10 UK 4 6 0 18 21 15 EU3 2 16 0 49 106 11 Ro15 7 3 1 40 553 40 AC10 2 0 0 34 271 37 AC2 14 8 1 23 56 32 EU27 1 8 66 4 173 1,341 217 RusFSB 2 0 3 0 14 76 13 Tukey 0 3 0 0 38 4 USA 6 36 0 38 304 29 Canada 3 8 0 2 92 9 Agentina 1 1 0 24 20 25 Bazil 0 1 0 12 120 1 RoLaAm 3 0 0 0 2 13 2 AusNZ 4 3 8 0 16 40 50 China 0 9 0 0 135 2 India 0 1 1 2 2 1 ROW 5 5 11 1 4 425 71 1 EU27: all 27 Euopean Union membes. 2 RusFSB: Russia and Fome Soviet Bloc. 3 RoLaAm: est of Latin Ameica. 4 AusNZ: Austalia and New Zealand. 5 ROW: est of the wold. 7.89%, whilst in cattle/sheep and aw milk sectos, coesponding falls ae estimated at 2.23% and 2.42% espectively. These estimates compae with EU27 aveage feed demand falls of 8.16% (pigs/poulty), 1.83% (cattle/sheep) and 2.35% (aw milk) 12. Note that the Spanish esults ae elatively close to the EU27 aveage. With inceases in feed costs, Spanish maket pices of pigs/poulty (Table 2) ise 7.97%, whilst moe modeate aveage feed cost ises in cattle/sheep and aw milk lead to maket pice inceases of 1.62% and 2.67% espectively. Consequently, Spanish poduction (Table 2) of cattle/sheep and aw milk declines 2.09% and 4.27% espectively, whilst in pigs/poulty poduction falls 9.74%. Compaing acoss the EU27, the diffeences in maket pice ises fo livestock ae attibuted to the total cost shae of feed costs in poduction. Equally, the tansmission of pices fom upsteam livestock to downsteam meat and daiy sectos eflects the magnitude of the livestock/aw milk cost shae to the total intemediate and value added costs of meat/daiy poduction in the undelying input-output tables. Note that Spanish pice ises in white meat poduction is amongst the highest in the EU27 (4.70%), esulting in a 4.51% eduction in poduction. With lowe feed cost ises in cattle/sheep and aw milk sectos, Spanish (EU27) ed meat and daiy maket pice estimates ae notably smalle. Given the stategic impotance of the EU livestock sectos in agicultue, agicultual output (Table 2) falls by 3.12% in Spain, compaed with 2.00% in Fance, 2.43% in Gemany, 1.76% in the UK and an aveage EU27 fall of 2.99%. In the AC2, falls in agicultual (7.36%) and maco gowth (1.85%) ae consideable. The index of pimay agicultual pices (Table 2) in Spain shows an incease of 2.07%, which esults in a etail pice index (RPI) ise of 0.09%. In the EU27, the coesponding EU27 estimates ae 1.77% (agicultual pice index) and 0.07% (RPI). An examination of the tade balance impacts on EU livestock sectos fom the feed ban in scenaio 2 is pesented in Table 3. In the GTAP database, the vast majoity of «livestock» elated tade occus in the downsteam pocessing sectos, whilst livestock tade is much smalle, especially on exta-eu tade outes. Futhemoe, it is impotant to note that aw milk is lagely non-tadable. With the fall in domestic poduction, 12 The inelastic demand falls in each of the livestock sectos ae detemined by the elasticity of substitution paamete between feed inputs.
10 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 expots of Spanish white meat, ed meat and daiy fall by 12.51%, 1.98% and 1.59% espectively (not shown), whilst cattle/sheep, pigs/poulty and aw milk expots ae estimated to decease by 3.08%, 12.10% and 14.19% espectively (not shown). Compaing with othe Euopean patnes, Spanish expot falls appea to be amongst the highest given the lage impacts on poduction. At the EU27 level, the esults indicate that pigs/poulty and white meat tade could fall by between 8-9% (not shown). With maked deteioations in EU meat poduction, thee is geate consume dependency on non-eu souces of white meat, ed meat and daiy poducts. In white meat, EU impots ise by between 10-18% (not shown), whilst in ed meat and daiy, EU impots ise by magnitudes of appoximately 3% and 2% espectively (not shown). In tems of the EU27 tade balances (Table 3), thee ae deteioations of 4 m (aw milk), 8 m (cattle/sheep) and 66 m (pigs/poulty), whilst lage base tade volumes in downsteam commodities esult in geate deteioations of 217 m (daiy), 173 m (ed meat) and 1,341 m (white meat). Examining the non-eu egions of the aggegation, the main expotes of ed meat to the EU ae Austalia and New Zealand (39%), Bazil (23%), Agentina (6%), USA (6%), and the Rest of Latin Ameica (6%). In white meat tade, Bazil has the lagest tade shae (19%), followed by Tukey (13%), the USA (7%), China (7%) and Austalia and New Zealand (7%) 13. Finally in daiy tade, Austalia and New Zealand have a 35% tade shae, followed by Tukey (16%) and RussiaFSB (9%). Non-EU egions gain at the expense of the EU, whilst lost expot makets to Agentina and the USA depess feed costs, esulting in geate tade competitiveness 14. With a elatively lage tade shae in EU ed meat impots, USA and Agentinean ed meat tade balances impove 38 m and 24 m espectively (Table 3). Fo the same easons, the USA s white meat tade balance impoves by 304 m (followed by China with a tade balance impovement of 135 m) 15. Given the size of thei initial tade shae, Austalia and New Zealand ealise the lagest tade suplus gains in daiy of 50 m espectively. Table 4. Impacts of the genetically modified oganisms (GMO) feed ban on wold pices (pecentages) Wold commodity pices Finally, pe unit wold feed costs (see Table 4) ae estimated to ise by 0.68%, 2.95% and 1.02% fo cattle/ sheep, pigs/poulty and aw milk espectively, due to the weighted incease in aveage EU animal feed costs. Given the tansmission of feed pices into highe livestock (and eventually) meat/daiy pices, the tade weighted index of wold pices in these poducts ae also expected to ise. In scenaio 2, pigs/poulty and white meat wold pice inceases ae estimated at 2.25% and 1.57% espectively. Equally, in emaining livestock, meat and daiy sectos, wold pices incease by 0.59% (cattle/ sheep), 0.49% (ed meat), 0.43% (aw milk) and 0.56% (daiy). Scenaio 3 Pe unit wold feed costs Impacts in scenaio 2 Cattle/sheep 0.59 Cattle/sheep 0.68 Pigs/poulty 2.25 Pigs/poulty 2.95 Raw milk 0.43 Raw milk 1.02 Red meat 0.49 White meat 1.57 Daiy 0.56 Impacts in scenaio 3 Cattle/sheep 3.41 Cattle/sheep 8.93 Pigs/poulty 14.97 Pigs/poulty 25.38 Raw milk 2.19 Raw milk 6.73 Red meat 2.60 White meat 10.05 Daiy 2.92 In scenaio 3, pactically all feed impots ae lost, whilst feed costs ae inceased five-fold acoss all EU membes. As expected, thee ae majo impacts on livestock poduction (see Table 5), paticulaly in pigs/ poulty, which has highe potein feed dependency. In Spain, pigs/poulty declines by 37.20%, compaed with a coesponding contaction in EU27 pigs and poulty activity of 33.95%. As expected, cattle/sheep 13 In the case of Tukey, this is due to poulty tade only, whilst fo China, white meat tade is lagely dominated by pok poduction. 14 Pe unit feed costs in the USA (Agentina) fall by 0.13% ( 1.34) in cattle/sheep, 0.43% ( 1.78%) in pigs/poulty and 0.13% ( 1.46%) in aw milk poduction 15 Examining the oveall pe capita eal income change, it appeas that in Agentina, inceased livestock competitiveness does not compensate fo lost feed sales to the EU27, such that eal pe capita utility falls 0.13%. In the USA, pe capita utility emains static.
GMO feeds and impacts on livestock sectos 11 Table 5. Impacts of the genetically modified oganisms (GMO) impot ban in the EU livestock sectos in scenaio 3 (pecentages) Fa Ge Spa UK EU3 R15 AC10 AC2 EU27 EU livestock feed demands Cattle/sheep 9.84 10.58 10.97 7.85 14.12 12.04 9.57 14.98 12.09 Pigs/poulty 33.16 33.09 35.25 30.56 34.35 33.00 35.26 37.87 34.99 Raw milk 8.50 8.27 10.66 8.08 10.32 9.65 6.93 16.39 10.37 EU livestock, meat, agicultual and economic poduction Cattle/sheep 5.37 3.57 8.09 2.51 8.74 5.39 6.66 23.99 7.47 Pigs/poulty 34.66 37.01 37.20 25.25 32.71 29.90 28.73 45.43 33.95 Raw milk 6.35 4.70 15.56 4.10 5.15 7.80 6.41 28.32 9.32 Agicultue 8.69 10.54 13.22 7.86 10.57 9.12 11.17 27.60 12.29 Red meat 2.69 3.10 2.16 2.35 6.26 4.68 3.98 15.18 4.08 White meat 11.57 21.20 18.06 9.99 22.17 18.75 25.79 11.01 17.60 Daiy 2.69 3.38 2.59 2.00 3.74 3.73 6.10 16.16 3.84 Real Gowth 0.26 0.20 0.37 0.16 0.22 0.29 1.00 7.88 0.38 EU livestock, meat, agicultual and maket pice indices Cattle/sheep 6.53 3.40 8.90 5.24 4.17 5.70 8.29 20.11 7.44 Pigs/poulty 60.21 58.44 66.26 39.89 44.39 45.90 44.45 108.73 56.15 Raw milk 5.47 4.38 14.76 4.36 4.76 7.14 6.35 24.53 7.87 Agicultue 8.94 9.93 13.63 9.42 9.73 8.80 10.91 26.79 11.73 Red meat 4.32 4.64 3.96 4.41 2.16 5.70 4.64 9.26 5.18 White meat 22.92 29.48 32.59 20.41 26.20 26.68 30.26 21.28 26.14 Daiy 3.59 3.97 3.77 2.52 3.45 3.70 4.58 6.46 3.68 RPI 0.36 0.21 0.49 0.29 0.28 0.31 1.14 4.67 0.36 and aw milk poduction falls ae moe modest acoss EU membes, leading to EU27 declines of 7.47% (cattle/sheep) and 9.32% (aw milk). Spanish agicultue contacts by 13.22%, compaed with coesponding falls of 8.69% (Fance), 10.54% (Gemany) and 7.86% (UK), whilst in the «accession 2» (Bulgaia and Romania), agicultue contacts by ove a quate. In downsteam sectos, white meat poduction falls by 18.06% in Spain (close to the EU aveage), with coesponding ed meat and daiy contactions of between 2-3%. In tems of maco gowth, Spanish GDP contacts 0.37%; a lage eduction than othe EU15 egions although consideably less than the accession 12. As expected, EU livestock feed demands fall moe damatically then in scenaio 2 (Table 5). In Spain, feed input pice ises lead to pigs/poulty pice ises of 66.26% (Table 5). In white meat poduction Spanish pices ise 32.59%, eflecting both the high pig/poulty input pice ise and the its cost shae in white meat poduction. The RPI in Spain inceases 0.49% (above the EU27 aveage), whilst in the ecent accession membes, lage RPI ises eflect the geate impotance of ago-food poducts in consume expenditues. With majo contactions in EU27 pigs/poulty and white meat poduction, expots witness eductions of 51.11% and 39.44% espectively (not shown). This compaes with even lage falls in coesponding Spanish sectos of 61.47% and 62.07% espectively (not shown). Examining the EU27 tade balances (Table 6), white meat wosens by 5,991 m, whilst in ed meat and daiy, coesponding tade balance deteioations ae ecoded as 996 m and 1,058 m espectively. Fo Spain the tade balance deteioations ae 42 m (ed meat), 722 m (white meat) and 43 m (daiy). As in scenaio 2, the loss of feed makets in Agentina, Bazil and the USA impoves livestock competitiveness though cheape feed costs 16. The USA and Bazil ealise significant impovements in thei white meat tade balances of 1,135 m and 847 m espectively, whilst China ( 557 m), Canada ( 347 m) and 16 In Agentina and the USA, the aveage feed costs falls ae of a simila magnitude to scenaio 2. In Bazil, pe unit feed costs fall on aveage by 1.64% (cattle/sheep), 1.75% (pigs/poulty) and 1.64% (aw milk).
12 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 Table 6. Tade balance changes in scenaio 3 ( millions) Cattle/sheep Pigs/poulty Raw milk Red meat White meat Daiy Fa 1 151 3 6 179 129 Ge 4 2 2 78 1,320 192 Spa 1 40 1 42 722 43 UK 21 40 1 110 292 81 EU3 11 97 1 267 615 50 Ro15 37 42 6 198 2,279 199 AC10 14 2 1 193 1,108 225 AC2 81 60 2 101 524 139 EU27 38 432 18 996 5,991 1,058 RusFSB 1 28 2 76 276 64 Tukey 2 11 2 1 278 20 USA 21 196 2 124 1,135 130 Canada 17 46 0 17 347 46 Agentina 1 3 0 39 56 41 Bazil 0 7 0 134 847 11 RoLaAm 0 1 0 13 65 13 AusNZ 16 50 0 105 218 257 China 1 61 0 2 557 8 India 0 9 5 10 13 6 ROW 36 95 6 16 1,941 377 Tukey ( 278 m) also see notable tade balance impovements 17. Much of the emaining EU white meat tade deficit is picked up collectively by the est of the wold composite egion. In daiy tade, the lagest positive gains accue to Austalia and New Zealand ( 257 m) on account of its lage EU tade shae, whilst that of the USA also impoves 130 m. Finally, with its lage shae of EU impot makets and impoved tade competitiveness, Bazil ealises a ed meat tade balance impovement of 134 m, followed by the USA ( 124 m) and Austalia and New Zealand ( 105 m), with Agentina s coesponding tade balance impoving 39 m 18. Examining wold pice impacts in Table 4, ising costs in EU27 animal and meat poduction inflate tade weighted wold pices by 3.41% (cattle and sheep), 14.97% (pigs and poulty) and 2.19% (aw milk), whilst in elated downsteam sectos, pices ise by 2.60% (ed meat), 10.05% (white meat) and 2.92% (daiy). Similaly, with steep inceases in EU aveage costs of feeds, pe unit wold feed costs ise by 8.93% fo cattle/ sheep entepises, whilst in pigs/poulty and aw milk coesponding ises ae estimated at 25.38% and 6.73% espectively. Discussion The study employs thee scenaios to quantitatively assess the impacts fom withdawal of «non-appoved» US, Agentinean and Bazilian maize and soybean feed expots to the EU. The esults of the study focus on EU livestock and downsteam meat and daiy sectos, whilst some discussion is eseved fo non-eu makets. Scenaio 1 is consideed a «minimal impact» expeiment, since it is envisaged that Agentina and Bazil would be able to compensate the loss of US makets. Fo this eason, no feed cost inceases wee imposed. In scenaio 2 («medium impact») the combined loss of US and Agentinean supply would only be patly compensated by Bazil, esulting in exogenous feed cost inceases boowed fom the liteatue. In scenaio 3 («wost case»), thee is vey little compensation fom the est of the wold fo the loss of all thee expot makets, such that significant feed cost ises ae implemented into EU livestock sectos. As expected, the esults have alaming impacts on EU poduction, maket pices and tade, paticulaly in the pigs and poulty secto which has a highe dependence of soy deivative feeds. 17 Whilst USA feed pices fall by less than Bazil, the value of thei global expots of white meat is almost thee times the size of Bazil in the data. Agentina, by contast, has a elatively small global expot base of white meat in the GTAP tade data. 18 That this gain is smalle than the USA (despite lage feed pice falls in Agentina) can be attibuted to the fact that USA global expots of ed meat ae ove twelve times the magnitude of Agentina.
GMO feeds and impacts on livestock sectos 13 In a biefing document by EC (2007), the same expeiments wee conducted using the AgLink PE model 19. Whilst AgLink is bette placed to estimate feed costs (which have been used as inputs in the cuent study), unlike GTAP it does not model detailed bilateal tade elationships. Thus, thee is no endogenous teatment of tade divesion between the EU membes and key patne counties. Instead, impot eductions in EC (2007) ae modelled as exogenous eductions in aggegate EU impots. Due to this modelling diffeence, and the employment of the Amington assumption in GTAP 20, changes in EU impot tade epoted in this study ae of a smalle magnitude. In tems of EU poduction, esults ae easonably simila between studies, paticulaly in scenaio 2, although given the 600% incease in feed costs in EC (2007), EU white meat poduction falls in the wost case scenaio epoted in EC (2007) ae of geate magnitude (pigs/pok 34.7%; poulty 43.9%). An inteesting esult fom the cuent study is that esulting eductions in feed costs in the US, Bazil and Agentina impoves tade competitiveness in thei livestock sectos, leading to noticeable tade balance impovements. Fo example, in the white meat secto, the tade balance impoves by to 56 m (Agentina), 847 m (Bazil) and 1,135 m (USA), compaed with an EU27 tade balance deteioation of 5,991 m. Examining the case of Spain, pigs/poulty constitute almost 16% of agicultual output vis-à-vis 13% fo the EU27 (MARM, 2008). This statistic concus with the esult that both agicultual and maco gowth will suffe elatively moe in Spain compaed with othe EU15 membes 21. With apidly inceasing unemployment, futhe edundancies fom contactions in livestock, meat and daiy activities would be untenable in the cuent political climate. Ultimately, the likelihood of such a ban depends on the EU s impotance as a custome fo feed impots. In the case of the USA, the EU constitutes a mino maket which explains why in the past the US has not woied about EU appoval when cultivating new stains of cops (EC, 2007). The case, howeve, is makedly diffeent fo Agentina and Bazil, which depend heavily on EU makets fo expots of feed, paticulaly soybean. At the cuent time, this patially mitigates the likelihood of the scenaios examined hee, although with the ise of China and India as altenative destination makets, the EU cannot affod to be complacent. Acknowledgements The autho would like to thank the Depatment of Envionment, Food and Rual Affais (DEFRA-UK Govenment) fo its financial assistance. The autho would also like to thank two anonymous efeees fo thei comments on an ealie daft. Refeences BIRUR D.K., HERTEL T.W., TYNER W., 2008. Impact of bio fuel poduction on wold agicultual makets: a computable geneal equilibium analysis. GTAP Woking Pape No. 53. BROOKES G., CRADDOCK N., KNIEL B., 2005. The global GM maket: implications fo the euopean food chain. Consultancy epot Bookes West, UK, Neville Caddock Associates, UK, Biotask AG, Gemany. CARDY-BROWN CONSULTANCY, 2008. Impacts of EU Unauthoised GM s on the feed and livestock sectos. Euopean Shepa Goup, 10 th Octobe 2008. DIMARANAN B. (ed), 2006. Global tade assistance and poduction: the GTAP 6 database. Cente fo Global Tade Analysis, Pudue Univesity, West Lafayette, IN, USA. EC, 2007. Economic impact of unappoved GMOs on EU feed impots and livestock poduction. DG-Agi, Biefing pape. EURACTIV.COM, 2009a. EU uged to impose GMO limits on «clean seeds», 3 d Mach 2009. Available in http:// www.euactiv.com/en/cap/eu-uged-impose-gmo-limitsclean-seeds/aticle-179910. [13 July, 2009]. EURACTIV.COM, 2009b. Gemany joins anks of anti- GMO counties, 15 th Apil 2009. Available in http:// www. euactiv.com/en/cap/gemany-joins-anks-anti-gmocounties/aticle-181267. [14 July, 2009]. FEDNA, 2008. Indice tablas FEDNA de ingedientes paa piensos. Fundación Española paa el Desaollo de la Nutición Animal. Available in http://www.etsia.upm.es/ fedna/tablas.htm#con. [6 Dec. 2008]. [In Spanish]. GELLER H.S., 1985. Ethanol fuel fom suga cane in Bazil. Ann Rev Eneg 10, 135-164. 19 Unfotunately, only a vey limited selection of esults ae pesented fom which to compae with. Moeove, estimates ae confined to the EU27 composite egion. 20 Amington diffeentiates between poducts fom diffeent makets, theeby endeing each county with a degee of maket powe (i.e., lowe tade elasticities). 21 Measued in eal income tems, it is estimated that Spanish (EU27) pe capita income falls 0.15% (0.10%) in scenaio 2 and 0.81% (0.59%) in scenaio 3.
14 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 HAAS M.J., MCALOON A.J., YEE W.C., FOGLIA T.F., 2005. A pocess model to estimate bio diesel poduction costs. Bioesouce Technol 97, 671-678. HERTEL T.W. (ed), 1997. Global tade analysis: modelling and applications. Cambidge Univesity Pess, Cambdidge, UK. INTERNATIONAL ENERGY AGENCY, 2004. Bio fuels fo tanspot: an intenational pespective. OECD Publ, Pais, Fance. JEAN S., LABORDE D., MARTIN W., 2005. Consequences of altenative fomulas fo agicultual taiffs. In: Agicultual tade efom and the Doha development agenda (Andeson K., Matin W., eds). Ch. 4. Wold Bank and Palgave MacMillan. KEENEY R., HERTEL T., 2005. GTAP-AGR: assessing the implications of multilateal changes in agicultual policies. GTAP Tech Pap 24, Pudue Univesity. Available in https:// www.gtap.agecon.pudue.edu/esouces/tech_papes.asp. [12 Jan 2009] MARM, 2008. Anuaio de Estadística Agoalimentaia. Ministeio de Medio Ambiente y Medio Rual y Maino, Madid. Available in http://www.mapa.es/es/estadistica/ pags/anuaio/intoduccion.htm. [1 Feb 2009]. [In Spanish]. TABEAU A., EICKHOUT B., VAN MEIJL H., 2006. Endogenous agicultual land supply: estimation and implementation in the GTAP model. 9 th Annual Confeence on Global Economic Analysis, Addis Ababa, Ethiopia. June 15 th -17 th, 2006. TAHERIPOUR F., BIRUR D.K., HERTEL T.W., TYNER W., 2008 Intoducing liquid bio fuels into the GTAP data base. GTAP Reseach Memoandum No. 11 2007, Updated May 1, 2008. TIFANNY D.G., EIDMAN V.R., 2003. Factos associated with success of ethanol poduces. Dept Appl Econ, Univ Minnesota, USA. Staff Pape P03-07. UN COMTRADE, 2008. United Nations commodity tade statistics database. UN Statistics Division. Available in http://comtade.un.og. [1 Nov 2008]. USDA, 2006. The economic feasibility of ethanol poduction fom suga in the United States. US Depatment of Agicultue, Washington, USA. WALMSLEY, T., 2006. A baseline scenaio fo the dynamic GTAP model. GTAP Resouce 2204. Available in https:// www.gtap.agecon.pudue.edu/esouces/es_display.asp? RecodID=2204. [10 Dec 2008]. Land supply estimation Technical appendix In estimating land supply functions fo each of the 87 membe counties/egions of the GTAP database, a non linea functional fom is employed: b0 Accumulated Aea = a [A.1] C 0 + Rent p whee «a» is the asymptote of the function epesenting the maximum potential available land fo agicultual puposes; b 0, C 0 and p ae estimable paametes. Fo the econometic estimation, data on potential agicultual aeas and yields developed by the Intenational Institute fo Applied System Analysis (IIASA, 2007) ae employed. Moe specifically, yields and aea data fo fou diffeent levels of land suitability (4 types) acoss 23 cop types ae available fo each egion (92 obsevations). In an initial step, data obsevations ae soted in descending ode of yields and the coesponding potential aea is accumulated. Given the natue of data available, the «total accumulated potential aea» fo agicultual activity is lage than the conceptual asymptote o «maximum available (agicultual) land aea» (i.e. the same gid-cell can be suitable fo altenative cops). Accodingly, the supply function is e-scaled, assuming that the «total accumulated land aea» coesponds to the actual maximum available land aea (i.e. the distibution of the accumulated aea is popotional to the distibution of the available land along the ange of yields). The «maximum available land aea» (o asymptote) fo each county is calculated as the emaining land excluding bodies of wate, closed foest ecosystems, othe land potection schemes and land employed fo housing and infastuctue. Assuming that the most poductive land is employed initially, the maginal cost of land inceases, which eflects the inceased convesion cost of additional units of maginal land. The ental ate of land is defined as the ecipocal of the potential yield (1/yield). All ents (yields) ae nomalised by dividing by the minimum ent (maximum yield) in each sample, which leads to ents above 1 and yields between 0 and 1. This scaling helps to infe the elative suitability of each county fo each cop, while fom an econometic standpoint it acceleates convegence to a solution. The empiical land supply equation becomes: b R_Aea j = 1 [A.2] C 0 + R _ Rent +ε p j j whee the sub-index j efes to each of the 92 obsevations available fo each county/egion; R_Aea is the elative accumulated aea fo obsevation j; R_Rent is the elative land ent fo obsevation j; b, C 0 and p
GMO feeds and impacts on livestock sectos 15 ae paametes to estimate, with b = b 0 /a; and finally, ε j is the eo tem, which is assumed to be nomally distibuted, N(0,s). Eq. [A.2] is estimated by Weighted Maximum Likelihood (a suitable method fo non-linea models). To impove the fit of the estimated function to the oiginal data, highe weights ae assigned to those obsevations with geate R_Rent j. The location of each county/egion on its land supply cuve is the use atio (R_Aea C ) of agicultual land use in 2000/2001 to maximum available land aea measue discussed above. Substituting calculated land use atio estimates (R_Aea C ) into equation [A.2] and e-aanging, the «cuent elative ent» (R_Rent C ) is obtained. The point elasticity of the land supply function at these coodinates can then be expessed as: = E S = R _ Aea R _ Rent R _ Rent C R _ Aea C = ˆb ˆp R _ Rent Cˆp (Ĉ0 + R _ Rent ˆp ) Ĉ C 0 + R _ Rent Cˆp ˆb * [A.3] whee the cicumflex ove the paametes indicates the estimated coefficients 22. In the model famewok, equation [A.2] is inseted diectly into the model code, whee ents in the 2001 benchmak data can be calibated given knowledge of the emaining paametes and land use atio. To validate the coect implementation of the land supply function, calculated land supply elasticities fom a simple shock must be sufficiently close to the point elasticities calculated in equation [A.3]. Othe CAP modelling issues ( ) Suga and milk quotas ae chaacteised employing complementaity equations in GEMPACK to allow binding/non-binding status of the quota. Estimates of milk and suga quota ents fo the EU15 in the benchmak ae based on an aay of liteatue souces and expet opinion within Defa (UK Govenment). To chaacteise set aside an exogenous hicks neutal poductivity vaiable is employed. A negative shock of 10% implies that of evey hectae used, only 0.9 is poductive. Since the value of land in the GTAP database only eflects «poductive» land, it is assumed that 2001 set aside levels ae implicitly included in the benchmak data (i.e., as pat of the egisteed land aea). Changes in set aside ae based on pojections fom the Euopean Commission. Intevention pices ae explicitly modelled employing complementaity equations. If the suppot pice falls to the exogenous intevention pice (which itself is shocked to simulate the MTR intevention pice falls), stock puchases occu. Since stocks ae not the esult of constained optimisation, but athe ae «tiggeed», they must be subtacted fom the egional income equation such that income emains equal to expenditue. The decoupling of EU agicultual suppot is modelled by the emoval of all output, intemediate input, capital and land subsidies in the GTAP database in 2001 (at diffeent agenda 2000 ates) and eplacing these with a single fam payment (SFP) chaacteised as a homogeneous land payment to all agicultual sectos. As a homogeneous land subsidy ate, the SFP does not favou any poduction activity (i.e., no coss commodity effects) such that the payment is poduction neutal. The calculation of total modulation savings and allocations to each EU27 egion follows the Commission s poposals. Modulation savings ae calculated at a 20% ate of the ceiling SFP ceiling limits. To allocate modulation funds acoss EU membes, egional allocation shaes ae based on the agicultual aea shaes (65% weighting) and agicultual employment shaes (35% weighting). This weighted estimate is subsequently coected employing a elative GDP pe capita weighting. A futhe constaint is imposed within the calculation to ensue that all egions eceive at least 80% (as specified by the Euopean Commission) of thei initial modulation contibutions (except Gemany which should eceive 90%). Modulation flows ae incopoated within the common budget mechanism. In the 2001 benchmak, the CAP budget only applies to the EU15 egions. Thus, each EU egion contibutes to Bussels via 75% of agicultual taiff evenues and modulation, and eceives funding fo domestic suppot policies. The diffeence between total eceipts and total contibutions by each membe gives a net esouce cost of the CAP which is met by unifom pecentage GDP contibutions by each membe state such that the total CAP budget balances at zeo. The analysis also includes the UK ebate mechanism, whee 66% of the UK s net contibution is efunded, whilst the emaining 22 A full list of paamete estimates, standad eos, mean log-likelihood values, land use atios and point elasticities fo each of the 87 egions of the GTAP vesion six data is available fom the authos on equest.
16 G. Philippidis / Span J Agic Res (2010) 8(1), 3-17 EU26 fund the bill based on GDP shaes. In the case of Austia, Gemany, the Nethelands and Sweden, the shae of the efund bill is educed to only one quate of thei GDP shae. In tems of the thee nested CET land allocation stuctue, the top nest CET elasticity is calibated to econometic estimates of land supply to agicultue (Keeney and Hetel, 2005), which is inceased by a facto of two on descending down the nest. Consequently, the mobility of land usage between agicultual sectos is educed in compaison with the standad model (which also educes agicultual supply esponsiveness). In the standad GTAP model, labou and capital ae pefectly mobile, whilst in this model vaiant, the tansfeence of these factos is contolled by a CET elasticity. The CET elasticity of tansfomation is calibated to econometic cental estimates of facto supply elasticities to agicultue in the liteatue (Keeney and Hetel, 2005). Consequently, the supply esponsiveness of agicultual/non-agicultual subsectos in esponse to a emoval of diect suppot in pimay agicultue will be dampened compaed with standad GTAP. Modelling the impot ban Stating with the modified CES function: [A.4] whee U i,s is the level of sub-utility fom the consumption of diffeentiated commodity i in egion s; Q i,,s is consume demand in egion s fo epesentative vaiety i fom egion ; Z i,,s is bilateal utility; A i,s is a scale paamete; δ i,,s is a CES shae paamete; and ρ i is an elasticity paamete. Minimising cost subject to [A.4] gives fist ode conditions: P i,,s =ΛA i,s U i,s = A i,s ρ δ i,,s Q i i,,s Z i,,s ρ δ i,,s Q i i,,s Z i,,s (1+ρ i ) ρ i 1 ρ i δ Q (1+ρ ) i i,,s Z i,,s i,,s [A.5] U i,s = A i,s δ i,,s Q ρ i,,s Z i,,s [A.6] whee P i,,s is the pice of epesentative vaieties. Substituting [A.6] into [A.5]: 1 ρ i ρ P i,,s =ΛA i ( 1+ρ i,s U i ) δi,,s i,s Q (1+ρ) i,,s Z i,,s [A.7] Lineaisation of [A.6] gives: u i,s = S i,,s q i,,s 1 z [A.8] ρ i,s i whee lowe case lettes ae pecentage changes in the coesponding uppe case vaiables, and z i,s is a lineaised expenditue shae weighted aveage of bilateal utilities, with expenditue shaes given as: P S i,,s = i,,s Q i,,s [A.9] P i,,s Q i,,s Lineaisation of [A.7] gives: p i,,s =λ+ ( 1+ρ i )u i,s ( 1+ρ i )q i,,s + z i,,s [A.10] whee λ is a lagangian vaiable. Thus, equations [A.8] and [A.10] ae lineaised fist ode conditions. Reaanging [A.10] in tems of q i,,s gives: q i,,s = σ i p i,,s +σ i λ+u i,s +σ i z i,,s [A.11] whee σ i is the elasticity of substitution between all pai-wise types of epesentative vaieties in the nest: 1 σ i = [A.12] 1+ρ i Substituting [A.11] into [A.8] and eaanging in tems of σ i λ yields: σ i λ=σ i S i,,s p i,,s σ i S i,,s z i,,s + [A.13] + 1 S ρ i,,s z i,,s i Substituting [A.13] into [A.11] eliminates λ. Factoising the esulting expession gives lineaised CES Hicksian pimay facto demands: whee q i,,s = u i,s σ i p i,,s S i,,s p i,,s + +σ i z i,,s z i,s + 1 z ρ i,s i z i,s = Fo consistent aggegation: P i,s U i,s = S i,,s z i,,s P i,,s Q i,,s [A.14] [A.15] [A.16] By lineaising [A.16], substituting [A.8] and eaanging:
GMO feeds and impacts on livestock sectos 17 p i,s = S i,,s p i,,s + 1 z [A.17] ρ i,,s i Using the weighted composite hieachical utility vaiable expession [A.15] and eaanging gives: S i,,s p i,,s = p i,s 1 z [A.18] ρ i,s i Substitution of [A.18] into [A.14], expanding the backets and collecting tems gives: q i,,s = u i,s σ i [ p i,,s p i,s ] + [A.19] 1 σ i [z i,,s z i,s ] σ i z ρ i,s σ i z i ρ i,s i Reaanging [A.12] in tems of ρ i and substituting the esult into [A.19]: q i,,s = u i,s σ i [ p i,,s p i,s ] +σ i z i,,s [A.20] Modified pivate demand stuctue In Taheipou et al. (2008), the authos intoduce thee additional sectos into the standad vesion 6 database to captue the poduction of liquid bio fuels. In boad tems, these thee sectos ae divided into «bio diesel» fom oilseeds cops (lagely based in the EU); «bio ethanol» fom stachy ceeals cops (lagely poduced in the USA and to a lesse extent the EU) and «bio ethanol» based on suga cane (mainly poduced in Bazil). To avoid compomising the undelying equilibium accounting conventions of the standad database, these thee sectos ae split out of existing sectos within the standad GTAP database. Moe specifically, the «vegetable oils and fats» secto (bio diesel), «othe food pocessing» (bio ethanol fom ceeals) and the «chemicals ubbe and plastics» secto (bio ethanol fom cane). A peceived advantage of having thee sepaate sectos, is that the database bette chaacteises the diffeent poduction technologies fo each bio fuel output. To estimate output levels and the intemediate input/pimay facto mix fo these sectos in 2001 (benchmak yea), the authos daw on an aay of liteatue souces. Fo estimates of poduction levels and tade, a epot by the Intenational Enegy Agency (IEA, 2004) is employed. Similaly, assuming zeo pofits the value of poduction is divided between intemediate inputs (i.e., feed stocks, chemicals, enegy, othe) and pimay factos labou and capital employing cost component estimates fom Tiffany and Eidman (2003) (ceeals based ethanol), USDA (2006) and Gelle (1985) (suga cane based ethanol) and Haas et al. (2005) fo bio diesel based on oilseeds. Due to data availability constaints, it is assumed that all inputs ae poduced domestically, except fo the feedstock used in the bio diesel industy in the EU. It is noted that the EU impots an impotant potion of its oilseeds consumption, whee these same tade shaes ae applied to the bio diesel industy impots. Whilst this wok undoubtedly epesents an impotant step into developing the GTAP database in this diection, it is clea that the quality of this type of ventue is typically esticted by both the availability and eliability of the undelying data souces. In thei teatment, the authos had to assume that poduction pocesses fo each of the bio fuels sectos ae homogeneous acoss egions. Moeove, the poduction and tade infomation employed is not exhaustive and some degee of ceative accounting will have been equied to fill in missing gaps in the database. Finally, a lack of data esticted the possibility of epesenting othe possible souces of bio fuels poduction (i.e., fom palm oils, suga beet, wine). Modifications to the pivate household demand nest ae also included to account fo bio fuel demand. Thus, in the top nest, all enegy commodities ae gouped into a single composite commodity within the CDE (pivate) function demands. The enegy composite is divided into coal, oil, gas, electicity and a petoleum and bio fuels composite. Typically, enegy demands ae vey pice inelastic, which is eflected in the elasticity of substitution (ESUBPEN) value of 0.1, based on estimates in Taheipou et al. (2008). In the lowe nest, final demands ae allocated between petoleum and bio fuel poducts, whilst substitution elasticity estimates ae taken fom Taheipou et al. (2008). In the case of Bazil, the EU, and the USA (which dominate bio fuel poduction), substitution elasticities (ESUBPFU) have been calibated to epoduce histoical pecentage inceases in bio fuels poduction between 2001 and 2006 in esponse to inceases in the pice of cude oil. Thus, in Bazil, the EU, and the USA the values ae 1.35, 1.65 and 3.95 espectively. As in Taheipou et al. (2008), in the emaining egions a default elasticity value of 2 is employed.