0 Internatonal Conference on Socal Scences and Socety Impact of U.S. Cotton Subsdes: A Computable General Equlbrum Model Houtan Ge,a, Crstna Echevarra,b,James Nolan,c, Rchard Gray,d Department of BPBE, Unversty of Saskatchewan, 5 Campus Drve, Saskatoon, SK, S7N 5A8, Canada Department of Economcs, Unversty of Saskatchewan, 9 Campus Drve, Saskatoon, SK, S7N 5A5, Canada a hpg338@mal.usask.ca, b c.echevarra@usask.ca, c james.nolan@usask.ca, d rchard.gray@usask.ca Keywords: Computable General Equlbrum modelng, Smulaton, Cotton Subsdy Abstract. The cotton subsdy ssue became one of the most dffcult negotatng topcs at the WTO once t was rased. Afrcan countres clamed that cotton subsdes granted by the U.S. led to artfcally depressed world prces and thus negatvely mpacted both export earnngs and producton levels n non-subsdzed countres. The purpose of ths study s to quantfy the mpact of U.S. cotton subsdes. A computable general equlbrum model (CGEM) s used to examne the effects of removng U.S. cotton subsdy programs on world producton, exports and prces of cotton. The model smulates mpacts for recent crop years, 996 through 009. Smulaton results show that there s a lnk between U.S. subsdes and the collapse n world cotton prces.. Background Global cotton producton doubled from 0 mllon tons n 960 to 5 mllon tons n 00. More than three-quarters of cotton output s accounted for by developng countres. Cotton s one of the most wdely produced agrcultural crops and cotton producton s a vtal source of foregn exchange, nvestment, and economc growth for some of the world s poorest countres. World cotton prces have fallen by half snce the md-990s. All the cotton producers are threatened by the fall n world prces whch reached ther lowest level n thrty years by 00/0. Although such a sharp declne s partly due to competton from synthetc fbers and the slow growth of the world economy, t s manly attrbuted to cotton subsdes. That s, the prce fall was aggravated by cotton subsdes. In 003, some Afrcan countres made a jont submsson to the WTO. They clamed that the cotton subsdes granted by the U.S. and the EU artfcally rased the level of cotton producton and stmulated exports n those countres and, by extenson, led to depressed market prces of cotton. The submsson asked for an accelerated procedure to reduce cotton subsdes, nducng the cotton ssue to be debated ntensvely at the WTO. In fact, the cotton ssue became one of the most dffcult negotatng ssues at the WTO once t was rased. Support to cotton producers has been greatest n the U.S., followed by Chna and the EU. Over the last fve years, the U.S., the EU and Chna accounted for 95 of world cotton subsdes []. In 009, subsdes to cotton producers n the U.S., Chna, and the EU reached 3.5 bllon dollars whch represented 70 of the value of cotton exports worldwde. Domestc subsdes n those countres dstort cotton producton and trade []. 978-0-983693-4-5/0/$5.00 0 IERI ICSSS0 86
Chna s a cotton mportng country. Although EU subsdes reman the hghest per producton unt, they are appled to much less cotton. Comparatvely, the world pays more attenton to U.S. subsdes. The U.S. s the thrd-largest producer of cotton n the world and the world s largest cotton exporter. The U.S. accounts for 0 percent of world output and 35 percent of global cotton exports from 005 to 009. The U.S. shelds cotton farmers by offerng them hgh subsdes, encouragng hgher producton and exports []. The trends of global cotton prces and U.S. subsdes over the perod 996-009, shown n Fgure, seem to support the clam that these hgh subsdes depress global prces. Subsdy [mllon $] Prce [$] Source: Author s Calculaton Based on Data from Envronmental Workng Group: Farm Subsdy Database, and Index Mund: Cotton Prce. Fgure : U.S. Cotton Subsdes and World Cotton Prce. Lterature Revew In the past a few years, the ssue of cotton subsdes and ther lkely negatve effects has attracted consderable attenton and controversy. There are some model-based assessments of the mpact of the removal of subsdes, and ther results dffer markedly. Whle the models utlzed are smlar n structure, the dfferences are n the assumptons about the levels of subsdes, as well as market parameters. ICAC [] estmates that 73 percent of the world producton of cotton was under some knd of drect assstance and the removal of U.S. subsdes alone would have ncreased world prces n 008 and 009 by around 0 percent. Goreux [3], n a report used as background for the Afrcan Countres submsson to the WTO, uses a smple model to analyze the njury to Afrcan producers by subsdes n developed countres, and concludes that world cotton prces would ncrease by 3-8 percent n the absence of these subsdes. Qurke [4] estmates that removal of producton and export subsdes by the U.S. and the EU would have ncreased world prces n 00 by 0.7 percent. Ge [5] examnes the cotton subsdy effects by usng a partal equlbrum model and concludes that the world cotton prce would have ncrease 8.40 n 003 f subsdes n the U.S., the EU and Chna had been elmnated. Sumner [6] uses a log-lnear equlbrum dsplacement model to smulate the effects of removng U.S. cotton subsdy programs. He estmates that removng U.S. subsdes would have decreased U.S. producton and export by 5.76 and.86 respectvely and ncreased the world prce by 8.66 on average durng the 00-005 perod. All above model-based analyses of the cotton subsdy ssues are based on the famlar mult-regon, partal equlbrum, statc world trade models, wthout descrbng the dynamcs of resources movng from one ndustry to the other. 3. Modelng the Impact of Cotton Subsdes: a CGEM Introducton. Whle a partal equlbrum model gnores the nteractons across commodtes and also the ncome effect, a computable general equlbrum model can capture the feedback effect across sectors and countres. In recent years, mprovements n model specfcaton, data avalablty, and computer technology have mproved the payoffs and reduced the costs of quanttatve polcy analyss based on CGEMs, pavng the way for ther wdespread use by polcy analysts throughout 87
the world. The CGEM s attractve because t can yeld precse quanttatve predctons. It allows for more flexblty and s more consstent wth neoclasscal economc theory and thus, may generate less based estmates when compared wth other modelng technques [7]. Unlke partal equlbrum models, a general equlbrum model accounts for the nter-ndustry lnkages n an economy and s vewed as a more approprate framework for conductng economc mpact analyss. In order to assess the effects of U.S. cotton subsdes quanttatvely, a computable general equlbrum model s used n ths study. The use of computers allows the models to be mplemented on a larger scale, wth potentally many dfferent sectors and types of nputs. Constructon of the model requres the followng steps: descrbng the model n detal and collecton and reconclaton of data to ft the model and completng the analyss of smulaton. A CGEM. There are three groups of countres n the model (,,3). Group s formed by cotton-subsdzng countres. Group s comprsed of countres where cotton s not subsdzed. The last group, the rest of the world, s comprsed of countres where no cotton s produced. There are two goods n the model: cotton and non-cotton (j,). We refer to cotton producton as the frst sector and non-cotton producton as the second sector. There exst two non-sector-specfc factors of producton, captal and labour, and one factor specfc to the frst sector: land. For smplcty, we assume that the producton functon exhbts constant returns to scale. The followng s the notaton of parameters and varables n ths model: s cotton subsdy n country q j producton of cotton or non-cotton n country θ share of the capttal on value added n non-cotton sector φ share of the factor return on value added n cotton sector c j consumpton of cotton or non-cotton n country α prce elastcty of demand β ncome elastcty of demand r nterest rate n country w salary pad for labour n country Hcks-neutral productvty term or total factor productvty of cotton or A j non-cotton n country j K j captal n sector j of country L j labour n sector j of country N land used n cotton producton n country land rental of country R τ tax collected by country Ths model s based on two knds of optmzng behavor, utlty maxmzaton and proft maxmzaton. The representatve consumer of each country group maxmzes the utlty functon U[c, c ] subject to the budget constrant, p.c + c r.k + w.l + R.N τ, where p s the prce of cotton. The prce of non-cotton s normalzed to. In each country, the producton functon for the cotton sector equals ϕ ϕ ϕ ϕ q A K L N () whle the one for the non-cotton sector equals q A K L () θ θ Frms maxmze profts: the dfference between expected market revenue and varable costs of producton, П (p j +s ).q j r.k j w.l j, where p p and p. The frst order condtons for the representatve frm n the cotton sector n country are ϕ ( ) ϕ + ϕ ϕ r p s A ϕ K L N 0 (3) j 88
and ϕ ( ) ϕ + ϕ ϕ w p s A ϕ K L N 0 (4) j ϕ ϕ ϕ ϕ R ( p + s ) A ( ϕ ϕ ) K L N 0 (5) j whle n the non-cotton sector the frst order condtons are and θ θ r A θ K L 0 (6) θ θ w A ( θ ) K L 0 (7) A cotton consumpton functon emphaszes the relatonshp between consumpton, ncome and the relatve prce of goods: α β B p Y c (8) where Y refers to ncome as defned n the budget constrant. Each country keeps ts budget balanced. It should be true that s q τ (9) Fnally, the market clearng condtons for the factor of productons n each country are and K j j j j K L L (0) () The global market clearng condtons for each product j are 3 3 c j q () j Equatons to yeld a system of 34 equatons and 34 unknowns. Ths system s smplfed, by substtuton, to a new system of 8 equatons wth 8 unknowns. The 8 equatons are solved n the Secant and the Newton-Raphson methods by usng Fortran 90. Calbraton. Wth respect to the demand sde, there are three parameters: α, β and B. We estmate ther values by usng equaton 9. Based on ICAC producton and prce data, we assume prce elastcty of demand α equals to 0.. Wth the relatve term data for 007 and 008 avalable, we can fnd the values of β and B, whch are equal to 0.38 and 47.6 respectvely. For the non-cotton producton functon, we use estmates of the aggregate producton functon. Thus, φ s approxmately /3 (n between 0.30 and 0.35 accordng to the lterature snce Solow [8]). The parameters for cotton producton are approxmated by the agrcultural producton functon n Echevarra [9]. Accordng to her, captal ntensty n agrculture equals 0.43, labour ntensty equals 0.4 and land ntensty equals 0.6. We take the U.S. as representatve of group snce our focus s on the effects of removng U.S. subsdes whle holdng subsdes n other countres constant. We take mportant cotton exportng countres as representatves of group : Brazl, Mexco, Pakstan Uzbekstan, and West and Central Afrca (WCA) countres. Fnally, we take Indonesa, Thaland, Japan and South Korea, countres that are not cotton-producers and are among the bggest cotton-mporters n the world, as representatves of group 3. 89
We assume both cotton and non-cotton sector n the U.S. enjoy the same total factor productvty (TFP), A j. By usng the Solow model [8], we can calculate the relatve (wth respect to the U.S.) TFP. We take ths aggregate TFP as the TFP of the non-cotton sector. For group, the weghted average of ths TFP equals A 0., relatve to A for the U.S. The weghted average of the TFP for the four countres representatve of group 3 equals A 3 0.33 relatve to the U.S. The relatve TFP for the cotton sector n group s a free parameter. Snce we have no way of estmatng t, t s fxed for these smulatons at 0.3. Notce that although, accordng to ths assumpton, the TFP n cotton s lower n group than n the U.S., the cotton TFP rato, 0.3, s larger than the non-cotton TFP rato, 0.. Wth respect to endowments, labour s approxmated by 65 (employment-populaton rato) of populaton. The populaton s estmated at 30 mllon for the U.S., 47 mllon for group and 463 mllon for group 3 based on the UN database 008. Land used n cotton producton s 5590 thousand hectares n the U.S. and 6380 thousand hectares n group based on the ICAC statstcs 008. Captal s approxmated by 30 (nvestment-to-gdp rato) of GDP. GDPs for the U.S., group and group 3 are 4640 bllon, 8630 bllon and 34400 bllon dollars respectvely based on the OECD statstcs 008. Results. The man smulaton results are provded n Table, ncludng the percentage change n prce, producton, and export consstent wth the removal of cotton subsdes. These results ndcate that removng U.S. subsdes would reduce U.S producton and exports substantally and thereby allow the world prce to rse by a large and economcally sgnfcant amount. If subsdes had been removed, the world prce of cotton would have been 4 to percent hgher over the fourteen-year perod 996-009 and, on average, percent hgher durng those years. U.S. producton would have been 6 to 9 percent lower over 996-009, wth an average of 3.35 percent lower. U.S. exports would have been to 33 percent lower and, on average, 4 percent lower over ths perod. On the contrary, supply and export n the non-subsdzed countres would have ncreased. As reported n Table, producton would have been to 7 percent hgher and, on average, 38 percent hgher over the perod 996 to 009. Exports would have been 0 to 8 percent hgher, wth an average of 65 percent hgher, over those years. At a global level, world producton and exports would not have changed sgnfcantly: there would have been a slght ncrease n both world producton and exports over the perod 996-009. 4. Conclusons Although cotton s not an excepton as a commodty that receves hgh levels of subsdes n some countres at the cost of non-subsdzed countres, the cotton ssue became a controversal subject n the WTO complance panel. Wth the WTO negotatons expected to resume n the future, there s a need for more analyss of the ssue n order for these negotatons to proceed on the bass of facts and fgures rather than conjectures. The purpose of ths paper was to contrbute some analyss on cotton wthn the above context, namely lkely effects of dstortons n the world cotton market on non-subsdzed countres. Overall, the results support the concluson that U.S. cotton programs have a serous and sgnfcant mpact on world producton, exports and global prce of cotton. The very hgh levels of support gven to cotton producers n the U.S., by vrtue of ts absolute magntude, are one of the drect and major causes of the problems faced by global cotton producton. They artfcally ncrease U.S. cotton supples on nternatonal markets, brng down export prces and reduce supples and exports of non-subsdzed countres, partcularly those WCA countres most dependent on cotton for ther ncome and foregn exchange earnngs. It has been establshed that there s a lnk between U.S. large-scale subsdes and the collapse n global cotton prces over the past decade. Reducng or elmnatng these subsdes would have sgnfcant benefts for poor countres for whch cotton producton s an mportant contrbutor to exports and GDP. Therefore, non-subsdzed countres have a powerful moral and economc case for ther demands for reform of the cotton subsdy system. 90
References [] ICAC: Government Support to the Cotton Industry. Washngton, DC: ICAC (00). [] J. Baffes: Cotton Subsdes, the WTO, and the Cotton Problems, Workng Paper, the World Bank (0). [3] L. Goreux: Cotton after Cancun, Dscusson Paper to nform debate, OECD (004). [4] C.C. Lu, and C.Y. Chen: A Computable General Equlbrum Model of Southern Regon n Tawan, Amercan Journal of Appled Scences (3), 0-4 (004). [5] H. Ge: Impact of Cotton subsdes n Developed Countres on Developng Countres, Master Project, Department of Economcs, Unversty of Saskatchewan (005). [6] D. A. Sumner: Effects of U.S. Upland Cotton Subsdes on Upland Cotton Prces and Quanttes, Workng Paper, Unversty of Calforna Agrcultural Issues Center (006). [7] D. Qurke: Trade Dstortons and Cotton Markets: Implcatons for Australan Cotton Producers, Centre for Internatonal Economcs, Australa (00). [8] R. M. Solow: A Contrbuton to the Theory of Economc Growth, Quarterly Journal of Economcs, February (956), 70, 65 94. [9] C. Echevarra: A Three-factor Agrcultural Producton Functon: the Case of Canada, Volume, Number 3, Autumn (998), Internatonal Economc Journal. 9
Table : The Effect of Removal of U.S. Cotton Subsdes on Cotton Prce, Supply and Export over 996-009 year 996 997 998 999 000 00 00 003 004 005 006 007 008 009 Average U.S. subsdy Total amount [mllon $] 807 745 38 945 068 3333 950 550 9 3686 980 54 65 70 46 Amount per kg [$] 0.0 0.9 0.44 0.55 0.57 0.78 0.54 0.66 0.45 0.73 0.69 0.56 0.36 0.9 0.57 Effect on World prce 4.40 4.4 9.9.36.85 7.83.5 4.97 0.06 6.6 5.68.64 7.99 0.93.3 U.S.quantty suppled -6.00-5.68 -.6 3-3.64-4.0-7.38-3.4 8-5.54 -.75-6.6-6.0-3.8 6-9.85-9.3-3.35 U.S.quantty exported Non-subsdzed quantty suppled Non-subsdzed quantty exported -.7 7 3.05.50 -. 8.3 0.05 -.5 30.89 56.88-4.73 39.3 76.70-5.3 4.06 8.09-30.39 59.50 35.9-4.4 8 38.58 74.87-7.66 48.74 0.8 -.7 3.36 57.9-9.8 54.85 0.3-8.38 5.37 09.46-5.0 7 40.3 79.9-8.5 6 4.46 43.0-3.90 7.76 8.45-4.6 37.89 64.64 World quantty suppled 0.43 0.40 0.94.6.0.63.4.38 0.95.5.45.8 0.76.88.5 World quantty exported 0.3 0.30 0.70 0.87 0.90. 0.86.04 0.7.5.09 0.89 Note: Data are from 0 Farm Subsdy Database, Envronmental Workng Group. The calculatons of subsdy per kg. of cotton are based on annual cotton producton data from the Natonal Cotton Councl of Amerca. 0.57.4 0.86 9