Global Simulation Analysis of Industry-Level Trade Policy

Transcription

1 Global Simulation Analysis of Industy-Level Tade Policy Joseph Fancois, Tinbegen Institute and CEPR H. Keith Hall, U.S. Commece Depatment Vesion 3.0: 21 Apil 2003 Abstact: In this pape, we outline a modeling stategy fo the patial equilibium analysis of global tade policy changes at the industy level. The famewok is scalable, employs national poduct diffeentiation, and allows fo the simultaneous assessment of tade policy changes, at the industy level, on a global, egional, o national level. Results allow the assessment of impote and expote effects elated to taiff evenues, expote (poduce) suplus, and impote (consume) suplus. With additional data, domestic poduction effects can also be fit into the famewok. Keywods: patial equilibium model, tade policy modeling, simulation model, global makets 1. INTRODUCTION In tade negotiations, thee is a need fo capacity within developing counties to assess the impact of taiff changes. This includes not only multilateal libealization, but also egional and unilateal tade libealization. In past GATT ounds, this has often involved the Wold Bank/UNCTAD sponsoed SMART model. This is because while CGE models povide estimates of aggegate effects, national policy is made at the taiff line level.

2 Ultimately, tade ministies need a stuctued way to combine infomation on tade flows and tade policy fo detailed poduct categoies if they ae to weigh the political foces that suound initiatives to libealize tade. In this pape, we outline a global simulation model (GSIM) fo the analysis of global, egional, and unilateal tade policy changes. Ou goal in developing the model is to povide a elatively simple, yet flexible famewok fo detailed analysis of tade policy in combination with the detailed taiff and tade flow data found in datasets like TRAINS and WITS. In this sense, we shae goals with the developes of the GSIM pedecesso, SMART. Whee we depat fom ealie applications in this aea is in taking advantage of available geate computational powe, and in stessing global maket cleaing conditions athe than impot makets. By focusing on global makets, we hope to facilitate the analysis of the value of collective maket access concessions fo expotes, in addition to the impot maket effects stessed by existing tools in this aea. The appoach we develop is patial equilibium, being industy focused but global in scope. By definition, patial equilibium models do not take into account many of the factos emphasized in ou elegant geneal equilibium tade theoy. This implies pactical limitations to the appoach developed hee. It also implies some useful advantages. Because we focus on a vey limited set of factos, the appoach followed allows fo elatively apid and tanspaent analysis of a wide ange of commecial policy issues with a minimum of data and computational equiements. In ou view, as long as the limitations of the patial equilibium appoach ae kept in mind, useful insights can be dawn with egad to elatively complex, multi-county tade policy changes at the industy level. This includes inteaction of multiple maket access concessions acoss vaious tading patnes, expote gains, consume suplus (impote) gains, and changes in taiff evenue. The pape is oganized as follows. Sections 2 and 3 develops the mathematical stuctue of the simulation model. This includes calibation of elevant own- and coss-pice elasticities, as well as global maket cleaing conditions. The definition of evenue and welfae effects is also discussed. Section 4 is focused on mapping GSIM elationships to the SMART concepts of tade ceation and divesion. Section 5 then discusses a simple 4 egion implementation of the model in Excel. This seves to illustate calculation of poduce and consume suplus changes, taiff evenue changes, and the oveall stategy fo solving the model. Section 6 discusses a stand-alone vesion of the model, with additional functionality but equiing subsidy and domestic poduction data. Two Excel files, GSIM4x4.XLS and GSIM25x25.XLS, ae meant to be distibuted with this pape. 2

3 2. BASIC RELATIONSHIPS When modeling tade policy at an industy level, the potential exists fo out model to quickly become unmanageable. Fo example, it is well known that the complexity of global geneal equilibium models tends to incease geometically as we add egions and sectos. A simila poblem exists even when we focus on an individual secto. Fo example, if we ae modeling tade policy fo lefthanded hoseshoe nails acoss 100 counties, thee ae 9,900 potential bilateal tade flows. To avoid this poblem, we educe the solution set of the model to those global pices that clea global makets. Once we have a global set of equilibium pices, we can then backsolve fo national esults. Within this context, we wok with a log-lineaized (pecent-change) epesentation of impot demand, combined with geneic expot-supply equations. (See Fancois and Hall 1997). This educed-fom system, which only includes as many equations as thee ae expotes, is then solved fo the set of wold (expote) pices. A basic assumption is national poduct diffeentiation. 1 As developed hee, this means that impots ae impefect substitutes fo each othe. The elasticity of substitution is held to be equal and constant acoss poducts fom diffeent souces. The elasticity of demand in aggegate is also constant. Finally, impot supply is also chaacteized by constant (supply) elasticities. Such an appoach is consistent with the Amington (1969) appoach to poduct diffeentiation at the national level (See Fancois and Hall 1997, Roningen 1997), o with the Flam-Helpman (1987) model of fimlevel diffeentiation (whee fim-specific capital fixes vaieties). In this section we spell out the basic stuctue of the model. This includes the development of elevant own- and coss-pice elasticities, and the inclusion of these tems in global supply and demand definitions and maket cleaing conditions. 2.1 Elasticities A citical element of the model appoach developed hee is the undelying ownand coss-pice demand elasticities. To aive at these values, we stat by assuming that, within each impoting county v, impot demand within poduct categoy i of goods fom county is a function of industy pices and total expenditue on the categoy: 1 This can esult, in an Ethie-Kugman type model, if poduct vaieties ae fixed. It may also be a esult of national diffeences in poduct chaacteistics (like Fench vs. Austalian wine). 3

4 Table 1 Notation,s v,w i expoting egions impoting egions industy designation Indexes Paametes Q, The composite good in egion v. i v AV g ( i, E s E m,(i,v) An efficiency tem calibated so that the pice of Q, P=1. The CES expenditue weight tem The CES exponent tem, whee the substitution elasticity 1 E S = 1 - elasticity of substitution aggegate impot demand elasticity Defined fo aggegate impots M (i,v) and composite pice P (i,v) = M (i,v) P (i,v) P (i,v ) M (i,v ) E x,(i,) elasticity of expot supply = X (i,) P (i,) * P (i,) * X (i,) N (i,(,) N (i,(,s T (i, q (i, f (i, Calibated coefficients own pice demand elasticity coss-pice elasticity The powe of the taiff, T=(1+t) demand expenditue shae (at intenal pices) q = T M T M /Â s s s expot quantity shaes f (i, = M (i, / Â w M (i,w), Vaiables M impots (quantity) X expots (quantity) P Composite domestic pice P* (i,) Wold pice fo expots fom egion P (I,),v Intenal pices fo goods fom egion impoted into egion v. t (i,),v Impot taiffs fo goods fom egion impoted into egion v. 4

5 (1) M (i, = f (P (i,,p (i,s, y (i,v) ) whee y (i,v) is total expenditue on impots of i in county v, P (i, is the intenal pice fo goods fom egion within county v, and P (i,s is the pice of othe vaieties. In demand theoy, this esults fom the assumption of weakly sepaability. (To avoid confusion on the pat of the eade o the authos, Table 1 summaizes ou notation). By diffeentiating equation (1), applying the Slutsky decomposition of patial demand, and taking advantage of the zeo homogeneity popety of Hicksian demand, we can then deive the following (See Fancois and Hall 1997): (2) N = q ( E + E ) (, s) s m s (3) N = Â ) E (, ) q Em - q s Es = q Em - ( 1-q s s whee q (i,v ),s is and expenditue shae, and E M,v is the composite demand elasticity in impoting egion v. 2.2 National Demand and Supply Equations Having defined own-pice and coss-pice elasticities, we next need to define demand fo national poduct vaieties. In addition, we will need national supply functions if we ae to specify full maket cleaing. Defining P i, * as the expot pice eceived by expote on wold makets, and P (i, as the intenal pice fo the same good, we can link the two pices as follows: (4) P = 1+ t ) P * = T P * ( i, i, 5

6 In equation (4), T =1+ t is the powe of the taiff (the popotional pice makup achieved by the taiff t.) We will define expot supply to wold makets as being a function of the wold pice P*. 2 (5) X i, = f (P i, *) Diffeentiating equations (1), (4) and (5) and manipulating the esults, we can deive the following: (6) P ˆ (i, = P ˆ i, *+ T ˆ (i,v ), (7) X ˆ i, = E ˆ X (i,) P i, * (8) ˆ i v = N i v Pˆ (, ), (, ),(, ) + Â N M (, s) P( i, s s ˆ whee ^ denotes a popotional change, so that dx x ˆ =. x An impotant point to make hee is that while we cente the discussion in the text aound poduction fo expot, one can also include domestic poduction fo domestic consumption within ou famewok. In paticula, we can index home maket demand though equation (11), supplied as is othe demand fo poduction though equation (10). This means that, when data on domestic poduction ae available, we can include domestic industy effects by modeling home maket tade in addition to foeign tade, using a non-nested impot and domestic demand stuctue. 2 While we do not do so hee, it would be staightfowad to intoduce expot subsidies o taxes, in addition to impot taxes. These would ente into equations (5) and (7). We could also intoduce poduction subsidies though the same equations. 6

7 2.3 GLOBAL EQUILIBRIUM CONDITIONS Fom the system of equations above, we need to make futhe substitutions to aive at a wokable model defined in tems of wold pices. In paticula we can substitute equations (6), (2), and (3) into (8), and sum ove impot makets. This yields equation (9). (9) M ˆ i, = Â M ˆ (i,v ), = Â N ˆ (i,v ),(,) P (i, + ÂÂN ˆ (i,(,s) P v v = Â N (i,v ),(,) [P *+ T ˆ (i,v ), ] + ÂÂN (i,(,s) [ P ˆ s *+ T ˆ v v s v s (i,v ),s (i,s ] We can then set equation (9) equal to the modified vesion of equation (7). This yields ou global maket cleaing condition fo each expot vaiety. (10) ˆ M i, = ˆ X i, fi E ˆ X (i,) P i, * = Â N ˆ (i,(,) P (i,v ), + ÂÂN ˆ (i,v ),(,s) P v v s (i,s = Â N (i,v ),(,) [P *+ T ˆ (i,v ), ] + ÂÂN (i,(,s) [ P ˆ s * + T ˆ v v s (i,s ] Equation (10) is the coe equation fo the system implemented in the speadsheet example in Section 4. Fo any set of R tading counties, we can use equation (10) to define S<R global maket cleaing conditions (whee we have R expotes). If we also model domestic poduction, we will have exactly R=S maket cleaing conditions. 3. WELFARE AND REVENUE EFFECTS In this section we wok with the basic solution set of pices to calculate national welfae and evenue effects. Once we solve the system of equations defined by 7

8 (13) fo wold pices, as we do in ou speadsheet example, we can then use equations (7) to backsolve fo expot quantities, and equations (9) to solve fo impot quantities. We can also solve fo the change in composite pices fo consumes. Fom thee, calculations of evenue effects ae also staightfowad, and they involve the application of tade values against taiffs. Pice and quantity effects can be combined with patial equilibium measues of the change in poduce (i.e. expote) suplus DPS and net consume (i.e. impote net of taiff evenue changes) suplus DCS i,v as a cude measue of welfae effects. (See Matin 1997). Conceptually, ou measue of poduce suplus is shown in Figue 1 as the aea of tapezoid hsnz, and appoximates the change in the aea between the expot supply cuve and the pice line. Fomally, this is epesented by equation (11) below. (11) DPS (i,) = R 0 (i,) P ˆ i, *+ 1 2 R 0 (i,) P ˆ i, * ˆ = R 0 (i,) P ˆ Ê *) 1+ E ˆ X,(i,) P Á Ë 2 i, * X i, ˆ 0 In equation (11), R ) epesents benchmak expot evenues valued at wold pices (which is identical to calibated base quantities). Fo consume welfae, we focus on the implicit composite good, assuming an undelying CES aggegato. This composite good theefoe takes the functional fom È (12) Q i,v = A v Í Âg (i, M (i, Î i=1 1/ Because we define the pice of the composite good to be 1 in the benchmak equilibium, the popotional change in the pice of Q (with total quantity then equal to total consume expenditue) will be: (13) P ˆ = dp P = q (i,v ), P ˆ È Ê Â (i,v ), = q (i,v ), (1+ P ˆ * i, ) T ˆ 1,(i,v ), Â Í Á i=1 i=1 Î Í Ë T 0,(i,v ), 8

9 Whee the eade is again efeed fo Table 1 fo help on notation. Equation (13) is the composite pice equation applied in the speadsheet example and in the actual model. It builds on the following elationship: (14) dp (i, P (i, ( ) 1-1= ( P (i,v ), ) 0 = P (i,v ), È Ê Ï Ô ( P * i, ) 0 + dp * i, Ô Ì Ó Ô ( P * i, ) 0 Ô T ˆ Í 1,(i,v ), Á Ë T -1 Î Í 0,(i,v ), The change in consume suplus is also epesented in Figue 1, as the aea of tapezoid abcd. It is defined as the change in the aea between the demand cuve fo the composite good and the composite good pice, as peceived by consumes. This is fomalized in equation (15). (15) Ê 0 0 ˆ DCS i v = Á R i v T ( 1 i v EM i v Pˆ 2 (, ) (, ), (, ),,(, ) v) sign( Pˆ v) ) - Pˆ v) ) whee Pˆ Ë v) Â = Â q Pˆ * + Tˆ 2 In equation (15), consume suplus is measued with espect to the composite impot demand cuve, with P v) epesenting the pice fo composite impots, 0 and R i T 0 (, ) v), epesenting initial expenditue (and identically quantity since the implicit calibated base pice is 1 fo the composite) at intenal pices. To make an appoximation of welfae changes, we can combine the change in poduce suplus, consume suplus, and impot taiff evenues. 4. OWN- and CROSS- TRADE EFFECTS The SMART model employed measues called tade ceation and tade divesion to quantify the effects of tade libealization. Hee, we biefly discuss the compaable measues. It tuns out that, in the case of a single, small county, these ae identical to the SMART equations fo these values. Because these ae 9

10 Figue 1 Poduce and Consume Suplus Measues 10

11 not actually the Vineian tade ceation and divesion measues, we instead will call them own- and coss-tade effects. Within the system developed above, assume that wold pices ae fixed, so that pice changes ae simply diven by taiff changes. In this case, fo a single county we have: (16) M ˆ (i, = N ˆ (i,(,) P = N (i,(,) ˆ T (i,v ), + (i,v ), +  s  s N (i,v ),(,s) ˆ P (i,s N (i,v ),(,s) ˆ T (i,s Whee we can futhe decompose equation (16) into an own-pice and cosspice tade effect: (17) Own-Tade Effect: TC (i, = M (i, [N (i,v ),(,) ˆ T (i, ] (18) Coss-Tade Effect: TD (i, = M (i, N ˆ (i,(,s) T  s (i,v ),s In equations (17) and (18), we have defined own-pice (o tade ceation in SMART) as tade geneated by diect taiff eductions fo the poduct concened, and coss-pice (o tade divesion in SMART) as tade changes geneated by changes in taiffs on impots fom thid counties. These ae eally just a special case of the coss-pice and own-pice effects that make up impot demand in equation (9) and equation (10). 11

12 5. IMPLEMENTATION AN EXAMPLE A 4x4 sample implementation of the model developed above is available as an Excel file. The data input section is illustated in Figue 2, which highlights the basic data equiements. These include tade flows (valued at a common set of wold pices), tade policy wedges, and elevant demand, supply, and substitution elasticities. The same types of data (with geate matix dimensionality) ae also equied fo lage applications. Note that while elasticities ae symmetic fo the pesent example, this is not necessay. On the basis of input data, othe key paametes (as defined in equations (2) and (3) above) ae calculated fo coss-pice and own-pice effects. These ae shown in Figue 3. The Excel solve is then used to solve the excess demand conditions specified in equation (10) above fo equilibium pices in the countefactual. This involves specifying one of the R excess demand functions fo expots as the objective function, with the othe excess demand functions then specified as constaints. The same appoach can be specified fo vesions of the model with highe dimensionality. Such an extension is coveed in Section 6. (Fo moe on the use of the Excel solve fo solving computational models, see Fancois and Hall 1997, and Devaajan et al 1997). The coe solution values, involving pices and excess demands, ae shown in Figue 4. On the basis of equilibium pice values, othe changes in the system can be calculated as well. These include, of couse, poduce and consume suplus measues (equations 14 and 15), changes in taiff evenues, tade quantities, and tade values. These ae illustated in Figues 5 and 6. The speadsheets can be used to exploe the actual calculation of values. The expeiment esults, while based on synthetic data, still illustate the types of effects captued in the model. We have modeled an expeiment whee two egions, the United States and Euopean Union, intoduce ecipocal taiff cuts (as might happen fom a fee tade ageement). What ae the effects of this taiff eduction? As we might expect, thee is an incease in impot demand on the pats of the EU and US, yielding an incease in pices fo both expotes (7.83 pecent fo the U.S., and 4.55pecent fo the EU). This in tun tanslates into gains in poduce suplus: 45.6 fo the U.S. and 37.7 fo EU poduces. Fo poduces outside the egion, the opposite happens. The pefeential libealization eodes demand fo thid county expots, and thei pices fall. The esults is a loss in poduce suplus: fo Japan and -4.8 fo the ROW. 12

13 On the consume side, composite pices fall by oughly 9 pecent fo US and 8 pecent fo EU consumes. The net effects, involving the combination of poduce suplus, consume suplus, and taiff evenue changes, is also summaized in the speadsheet. (See Figue 6). The net effect involves gains fo the EU and U.S., and losses fo Japan and ROW. In the case of both Japan and ROW, poduce losses coespond to a tems-of-tade deteioation. 6. AN EXPANDED STAND-ALONE VERSION The 4x4 vesion implemented above is designed to wok given the limited data envionment (in tems of domestic poduction data) in which lage-scale detailed taiff analysis is often undetaken. This 4x4 example is implemented in WITS. Thee is also a stand-alone vesion of the model, designed to accommodate a mode detailed set of policy and poduction data. This is the GSIM25x25.XLS speadsheet implementation. The GSIM25x25 model has the following addition featues (not implemented in WITS itself, howeve). _ Domestic poduction can be included, whee data ae available. _ Domestic poduction subsidies can be included, whee data ae available. _ Bilateal expot subsidies can be included, whee data ae available. _ Up to 25 counties/egional patnes can be specified. This additional functionality makes necessay the following changes to the basic theoy: (19) X ˆ i, = E P ˆ X (i,) ( *+ G ˆ i, i, ) (20) P ˆ (i, = ( 1+ P ˆ i, *) (( T (i,v ), ) 1 / T (i, ) S (i, ( ) 0 ( ) -1 ( ) 0 /( S (i,v ), ) 1 13

14 whee The subsidy paid fo expot of poduct i fom egion to egion v in ( S time peiod j=0,1 and whee S=1+s and s is the ad valoem subsidy (i,v ), ) j ate (as a shae of wold pice), so that an expote eceives a subsidy of s fo each unit of evenue eaned diectly by expots. G, A poduction subsidy in egion. i While the model is still solved fo wold pices, poduced and consume pices will vay fom wold pices by the combined effects of impot taiffs, poduction subsidies, and expot subsidies. In addition, while taiff evenue is netted against consume suplus to obtain net consumption benefits, poduce and expot subsidies must also be netted against poduce suplus to obtain poduction benefits. The GSIM25x25 implementation also allows fo own-tade (i.e. domestic absoption), such that the impot demand elasticity is eplaced by the aggegate demand elasticity (see Fancois and Hall 1997). All the emaining algeba goes though as specified, with the modified assumption that the CES aggegation function in equation (12) is now an explicit non-nested CES aggegato defined ove impots and the domestic good. 3 Because of the diffeences outlined above, the GSIM25x25 speadsheet involves a geate set of data equiements. These ae outlined in steps on the speadsheet itself, as shown in Figue 7. The epoted esults ae somewhat diffeent as well, including diffeences in consume, maket, and poduce pices, as well as changes in domestic poduction and the contibution of change sin subsidy payments to total welfae. This is illustated in Figue 8. 3 (Note that the 25x25 implementation uses a slightly diffeent appoximation fo equation (6), using the intenal pice change epoted in equation (12), so that thee may be slight diffeences in appoximate esults unde the two speadsheets.) 14

15 REFERENCES Amington, P. (1969), A Theoy of Demand fo Poducts Distinguished by Place of Poduction, IMF Staff Papes 16: Devaajan, S., D.S. Go, J.D. Lewis, S. Robinson, and P. Sinko, Simple Geneal Equilibium Modeling, in J.F. Fancois and K. Reinet, eds., Applied Methods fo Tade Policy Analysis: A Handbook, Cambidge Univesity Pess: Cambidge, Flam, H. and E. Helpman (1987), Industial Policy unde Monopolistic Competition, Jounal of Intenational Economics 22: Fancois, J.F. and H.K. Hall, Patial Equilibium Modeling, in J.F. Fancois and K. Reinet, eds., Applied Methods fo Tade Policy Analysis: A Handbook, Cambidge Univesity Pess: Cambidge, Matin, W., Measuing Welfae Changes with Distotions, in J.F. Fancois and K. Reinet, eds., Applied Methods fo Tade Policy Analysis: A Handbook, Cambidge Univesity Pess: Cambidge, Roningen, V.O., Multi-maket, Multi-Region Patial Equilibium Modeling, in J.F. Fancois and K. Reinet, eds., Applied Methods fo Tade Policy Analysis: A Handbook, Cambidge Univesity Pess: Cambidge, UNCTAD/Wold Bank (1989), A Use s Manual fo SMART (Softwae fo Maket Analysis and Restictions on Tade Vesion 2), Geneva and Washington, DC. 15

16 Figue 2 Excel 4x4 implementation of GSIM -- model inputs oigin oigin oigin INPUTS tade at wold pices: destination Totals USA JAPAN EU ROW USA JAPAN EU ROW Totals initial impot taiffs destination USA JAPAN EU ROW USA JAPAN EU ROW final impot taiffs destination USA JAPAN EU ROW USA JAPAN EU ROW Elasticities: USA JAPAN EU ROW Em Impot Demand Ex Expot Supply Es Substitution

17 Figue 3 Excel 4x4 implementation of GSIM -- Calibated values Calibated values Notation definitions q Impot shaes at intenal pices destination USA JAPAN EU ROW USA JAPAN EU ROW SUM oigin Notation definitions f Expot shaes at wold pices destination USA JAPAN EU ROW SUM USA JAPAN EU ROW oigin Equation (3) Equation (2) N(i,(,) oigin N(i,(,s) oigin Own pice elasticities destination USA JAPAN EU ROW USA JAPAN EU ROW Coss pice elasticities destination USA JAPAN EU ROW USA JAPAN EU ROW Figue 4 Excel 4x4 implementation of GSIM Coe solution values MODEL SOLUTIONS MARKET CLEARING CONDITIONS Relative pice changes benchmak change in change in Excess pices new pices supply demand Demand Equation (10) USA JAPAN EU ROW oigin 17

18 Figue 5 Excel 4x4 implementation of GSIM Tade Effects oigin Tade at wold pices: change in values destination Expot USA JAPAN EU ROW Total USA JAPAN EU ROW Impot Total EXPORT CHANGES (wold pices) USA EU JAPAN ROW USA JAPAN EU ROW

19 Figue 6 Excel 4x4 implementation of GSIM Welfae Effects county Total welfae effects A B C D=A+B Poduce suplus Consume suplus Taiff evenue Net welfae effect USA JAPAN EU ROW USA JAPAN EU ROW Taiff evenue Consume suplus Poduce suplus 19

20 Figue 7 The GSIM25x25 Speadsheet STEP 1 STEP 2 tade at wold pices: Ente Region Names destination name Load the initial bilateal USA JAPAN EU ROW Reg5 Region1 USA tade matix, at wold pices. USA Region2 JAPAN JAPAN Region3 EU note: Domestic absoption is EU Region4 ROW included as tade with self. ROW Region5 Reg5 Reg Region6 Reg6 Reg Region7 Reg7 Reg Region8 Reg8 Reg Region9 Reg9 Reg Region10 Reg10 Reg Region11 Reg11 Reg Region12 Reg12 Reg Region13 Reg13 Reg Region14 Reg14 Reg Region15 Reg15 Reg Region16 Reg16 Reg Region17 Reg17 Reg Region18 Reg18 Reg Region19 Reg19 Reg Region20 Reg20 Reg Region21 Reg21 Reg Region22 Reg22 Reg Region23 Reg23 Reg Region24 Reg24 Reg Region25 Reg25 Reg Totals note: Fo less than 25 Step 3 INPUTS initial bilateal impot taiffs destination egions, leave the est Load the initial matix of USA JAPAN EU ROW Reg5 of the labels and table bilateal impot taiffs USA values empty. in ad valoem fom. JAPAN EU note: taiffs ae enteed ROW as T=1+t, whee t is the ate Reg of the taiff makup elative Reg to wold pice. Reg oigin oigin 20

21 Figue 8 GSIM25x25 Results Step 11 View Summay Results Summay of Effects Poduce suplus Consume suplus welfae Taiff evenue Change in subsidy payments Net welfae effect Change in Oveall Consume Change in Pices Output othe Poduce Pice fo Home Good Maket Pice fo Home Good E= A+B=C+D pecent pecent pecent pecent A B C D USA % -51.5% % -1.46% JAPAN % 10.5% 7.01% 7.01% EU % 13.2% 8.81% 8.81% ROW % 14.7% 9.78% 9.78% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 12.9% 8.57% 8.57% Reg % -5.0% -3.35% -3.35% Reg % 7.5% 5.01% 5.01% Reg % -3.0% -2.00% -2.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% Reg % 0.0% 0.00% 0.00% 21

22