Policy Research WORKING PAPERS Trade Polloy Policy Research Department The World Bank September 1993 WPS 1173 Morocco's Free Trade Agreement with the European Community A Quantitative Assessment Thomas F. Rutherford E. E. Rutstr6m and David Tarr Welfare benefits to Morocco from a free trade agreement with the European Community would be about 1.5 percent of GDP. But welfare benefits would be 2.5 percent of GDP if Morocco liberalized trade with the whole world - and with only slightly higher adjustment costs. Policy RtcshWoiklngP md_natet effindings of woi in pogreca and encouage the exchange of ideu among Bank affand dlothe ioigWn davelopmnemintisewlpap, dimubutedbytheReserchAdvis(nyStaff, curythenames oftheauw,n clect ody thdrviw,andahouldbeuedandcitedaccordingly.Thefindin,gsbntertautions.andconclusionsareth1authweown.bysThCyould no be attiaed to th Wald Bank, its Board of Diracto, its managncnt, or any of its member countrieL Policy Researchj Trade Policy WPS 1173 Thispaper-a productoftheTrade Policy Division, PolicyResearch Department-is part ofa largereffort in the department to investigate the consequences of regional integration. The study was funded by the Bank's Research Support Budget under the research project "Impact of EC '92 and Trade Integrationr on Selected Mediterranean Countries" (RPO 675-64). Copies of this paper are avaihble free from the Wor.d Bank, 1818 H Street NW, Washington, DC 20433. Please contact Nellie Artis, room NIO-013, extension 38010 (September 1993, 27 pages, plus 18 pages of appendices). Morocco is interested in developing a reciprocal the rest of the world - with only slightly higher free trade agreement with the European Commu- adjustment costs. Liberalizing trade with the nity (EC), although it already enjoys free access world would provide great.-r benefits because it to EC markets in industrial products and is not would eliminate the trade diversion costs associ- obligated to give EC exporters reciprocal access. ated with discriminatory trade liberalization. But Moroccan agricultural exports are impeded (Although the fact that significant benefits would by agricultural protecdon in the European accrue from discriminatory liberalization against Community. imports from either the European Community or the rest of the world indicates that trade diver- A free trade agreement would require that sion is not dominant.) Morocco lower . s moderately high tariffs against its most important trading partner. Tariff * As a result of improved access to the reductions against the European Community but European Community, employmcnt and output not against the rest of the world may provide in the vegetable and citrus fruit sectors would benefits provided the trade diversion costs of expand. But the phosphate sector stands to gain preferential tariff reduction do not dominate. most from the free trade agreement because liberalization would induce a depreciation in the Rutherford, Rutstr6m, and Tarr apply a 39 real exchange rate. sector general equilibrium model of the Moroc- can economy which includes the sectors most * Morocco's cereal, meat, dairy, and sugar likely to be affected by such an agreement. They sectors would lose most in terms of employment, investigate the economic effects of the prospec- because of significantly lower import prices from tive free trade agreement as well as five other the European Community. The nontraded goods trade liberalization scenarios for Morocco. sector would also contract slightly. Among their most important findings: *The value-added tax would have to be * The welfare benefits to Morocco from a free increased to compensate for the loss in tariff trade agreement with the European Community revenues, on which Morocco depends. would be about 1.5 percent of GDP. Such substantial welfare gains partly reflect the Estimates are provided as ranges, with benefits of reducing dispersion in the tariff probability assessments, because of the element regime. of uncertainty. * Welfare benefits of about 2.5 percent of GDP would accrue from liberalizing trade with The Policy Research Working Ppe Series disseminates the fuidings of work under way in theBank Anobjectiveof the series is to get these findings out quickly, even if presentz dons are less than fully polished. The findings, interpretatims, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the Policy Research Dissemination Center .~~~~~~~~~~ MOROCCO'S FREE TRADE AGREEMENT ViTr THE EC: A QUANTITATIVE ASSESSMENT by Thomas F. Rutherford E.E. Rutstr&-m and David Tarr * University of Colorado, University of South Carolina, and Wodd Bank, respectively. Table of Contents Page No. 1. Introduction 1 2. A Small Open Economy Model 3 2.1 Trade Protection in Morocco and the Free Trade Agreement Shocks 3 2.2 General Model Structure 7 2.3 Empirical Implementation of the Morocco Model 9 3. Results 12 3.1 Welfare, Revenue and Aggregate Primary Factor Movement 13 3.2 Employment Impact by Sector 16 4. Impact of Key Parameters and Sensitivity Analysis 19 4.1 Impact of Key Parameters 19 4.2 Sensitivity Analysi- 22 References 27 APPENDICES Appendix A. Elasticity Parameters A-2 A. 1 Choice of Point-Estimates A-2 A.2 Effects of Choice of Trade Elasticities A4 Appendix B. Adjustments to the Social Accounting Matrix A-6 Appendix C. The Relationship Between the Elasticity of Supply and the Sector-specific Capital Share A-9 Appendix D. Algebraic Formulation of the Model A-10 1. NRODUCTION Since 1986 Morocco has shown an interest in developing a closer trade association with the European Community (EC). Thli interest culminated in March 1992 with concretes discussions for a reciprocal Froa Trade Agreement with the EC. To some the Moroccan interest may be surprising, since Morocco already enjoys privileged relations with the EC. It has freo access to markets in industrial products (there are some minor exceptions such as trousers and canned sardLnes), and is not obligated to provide reciprocal access to Lts market to producers of the EC. However, EC agricultural protection impedes Moroccan exports (notably in the areas of vegetables and citrus fruits), so that improved access to EC markets is an important issue. Moreover, the reciprocal obligations of a Free Trade Agreement (FTA) with the SC will require that Morocco lower its moderately high tariffs against its most important trade partner. These tariff reductions against the EC may provide additional trade liberalization benefits to Morocco, provided the trade diversion costs of preferential tariff reduction do not dominate. This paper reports on an applied general equilibrium modelling exercise that investigates the economic eQ fects of the Moroccan proposal. Given the questions of improved market access of Moroccan fruits and vegetables and of trade diversion and trade creation (which arises in any preferential trade area), we decompose the effects on the Moroccan economy from this proposal into the following six policy scenarios: a) improved access for Moroccan fruits and vegetables in the EC ("ACCESS"); b) unilateral tariff reductions in Morocco against the EC alone ("ECLIB"), against imports from the rest of the world alone ("LIBROW"), and against all trading partners ("LIBALL"), without improved access to EC agricultural markets; c) cooperative tariff reductions with the EC, where on the EC side this implies extended market access for Moroccan fruit and vegetables ("FTA"); and d) full free trade agreement with the EC augmented by unilateral liberalization of tariffs against rest of world imports ("FTAALL"). Our analysis provides a quantitative indication of the income gain to Morocco of these strategies. Among our most important results, we find that the welfare benefits to Morocco from the free trade agreement are about 1.5 of percent CDP, and are about 2.5 percent of GDP if Morocco adds trade liberalization with the rest of the world to the free trade agreement. These welfare gains are quite substantial ln the context of other model estimates with constant returns to scale such as ours, and partly reflect benefits from eliminating dispersion in the tariff regime, since dispersion is eliminated in the process of liberalization. The larger welfare gains from Morocco adding elimination of protection against the rest of the world to a free trade agreement with the EC, reflects the trade diversion costs associated with discriminatory trade liberalizaticon. On the other hand, the significant benefits that accrue from discriminatory liberalization agail.at either EC or rest of world imports indicates that trade diversion is not dominant. One interesting conclusion is that broader trade liberalization yields greater welfare gain than the PTA, but wlth only alightly hLgher adjustment Coste. Thls can be seen as an lmportant argument ln favor of the efforts towards lowering tariffs against non-EC sources subsequent to achieving a FTA. We evaluate the overall welfare sffects on the Moroccan economy from the above scenarios, as well as the sectoral impacts. The sectoral impacts are particularly Lmportant glven the emphasis that has been placed in Morocco on dLversLfication of production and exports, ..e well as concern over the costs of adjustment. The model provides a quantltative indication of the extent of output (not reported) and employment (report. 1 below) adjustment that will be required by industry. We find that although citrus frults and vegetables expand as expected from their improved access to the NC market from the PTA, by far the sector that gains the most from the trade liberalizatlon scenarios is the phosphate sector. Phosphate exports expand sLgniflcantly after the PTA or other trade liberalization scenarios due to the deprecLatlon of the real exchange rato induced by the liberalization. Conversaly, non-traded goods sectors slightly contract after the PTA or other liberalizing scenarios. The largest losers of employment, however, are the cereals, meat and dairy and sugar sectors. These 2 sectors face significantly lower import prices after trade liberalization. Estimates are provided within a range (wLth probability assessments over the range), as there is an element of uncertainty in the estimate. which we also quantify. Given the importance of the tariff au an instrumekt of revenue generation in Morocco, we exploit the ability of a "simulation laboratory" to zontrol for this effect by adopting the value added t&c (VAT) as an explicit replacement tax such that government revenue rem tns con'tant. The VAT induces distortion coats (marginal excess burden). We indicate the extent the VAT would have to be changed .in order to avoid a reduction in government revenue. The w'arginal excess burden of raising government revenue from the VAT is incorporated in the analysis. The model we uoe is a 39 sector computable general equilibrium model of the Moroccan economy. This level of disaggregation captures most of the important sectors of the Moroccan economy that would be affected by the Free Trade Agreement. In particular, citrus fruits, vegetables, cereals, sugar and meat and dairy products, textiles, apparel, fishing and phosphates are included as separate sectors. The model that we use in deliberately very simple, to facilitate the confrontation of policy-makers' intuition with easily interpreted simulations. The model assumea no terms-of-trade eofocts, a single household, no capital accumulation, and constant returns to scale production with competitLve pricing. In addition, the model is a "comparative statics" model which ignores the costs of adjustment of factors. Consequently, the benafits of the integration-liberalization scenarios will be les than our estimates to the extent of costs of adjustment. 2 * A SMULL OPEN ECONOMY MODEL 2.1 Trade Protection in Morocco and the Free Trade Agreement shocks Structure of Protection. Since 1983 Morocc has dramatically liberalized its foreign trade regime. In 1983 import licenses were required in all sectors, tariff rates were high and dispersed (some over 1CA% in a4- -cn to the special import tax of 15%), and 3 there were export licensing requirements and a state marketing board monopoly on exports of processed food products.' In 198S, the maximum rate for customs duties was lowered to 45%. Mout impressively, ? -A'occo 'as progressively reduced its import licensing requLrements so that by 1993 no imports will require a license (other than for health and safety reasons). On the export side only minor restraints remain since special customs regimes for exporters were extended and improved, export licensing was removed with only a few exceptions, and the monopoly of the state marketing board was abolished (see Mateus et al. (1988, p.ll)J. We therefore take as our point of departure, a trade regime that is free of non-tariff barriers. Tariff rates as of 1991 by sector are presented in column 10 of table 1. The structure of the rates is taken from legal applicable rates at the tariff line level that are aggregated, based on an unweighited average, to the 39 sectors of our model. Legal tariff rates, however, are not Indicative of actual tariff collections, because there are exemptions to the tariff (as well as exemptions to the fiscal import tax and the value-added tax) for a number of purposes, most importantly for inputs into products that will be exported. In view of the importance of the revenue implications of the proposed tariff changes, we therefore proportionally adjusted all the legal rates. The rates that appear in table 10 have been proportionately adjusted for all sectors, such that the total tariff collections in the economy, based on 1991 imports, equals actual aggregate tariff collections from budget data for 1991.2 The average import tariff based on collections is 19%, which includes tariff surcharge (fiscal import duty).3 The most important sectors regarding import revenues are Industrial Machinery, and Coal and Crude Oil, which together account for over 30% of revenues from trade taxes. Among the most important non-tariff barriers scheduled to be removed in early 1993 are those in several agriculture sectors: sugar, cereals, meat, dairy 't8. Mateus at al. 19881 and UNDP-Wordd Bank (19921 for detols. 'Appendix B dicus thds adjument 'As memben of the Msghreb pmferential tmding ae, inoas fiom Algea an Tunisa are exempted frnm the fial import duty. 4 Table 1: Morocco - Basic Data for the Socild Accounting Matrix 8ECTOR (Abbrov.) SECTOR SHARB OP TOTAL: TRADE BY SECTOR: TARWPRA (In percent) (In percent) (1) (2) (3% (4) (5) '6) (7) (8) (9) (10) Output Labor Capital Imports % fromEC El. wur: %to eC lmoogir Er ______________________________ .________________ ___________ _ Consumption Output kORICULTURB 1. Cereals (CER) R 5.44 3.04 10.32 7.16 53.8 0.85 54.2 13.20 150 45* 2. Sugar (SUG) R 0.26 0.12 0.41 0.00 0.0 0.00 0.0 0.00 0.00 45* 3. Citrus FruIts (Cfl) R 2.18 1.50 2.84 0.03 64.8 9.26 69.8 0.40 40.20 31.56 4, Vegetables (VEO) R 1.84 1.14 2.15 0.26 100,0 2.75 dP.6 2.40 14.00 1135 5. Meat and Dairy (MAD) R 4.90 1.62 9.88 0.23 50.2 0.02 100.0 0.70 0.00 45S* 6. Eshln (FSN) R 0.38 0.08 0.25 0.00 64.8 0.66 53.6 0.00 16.30 7. Forestry and Other Agri- cultue (FOR) 1.02 0.51 1.74 2.44 59.4 0.17 86.6 32.80 7.00 29.54 MINIG AND RELATED _ 8. Phosphates (PHS) 2.52 2.5 4.04 0.00 0.0 22.45 51.1 0.00 89.20 9. Other Non-Metallic Mining (NDM) R 0.40 0.46 0.30 1.99 7.7 0.73 41.2 41.10 15.40 16.48 10. Metals Mining (MIN) R 0.59 0.58 0.91 0.14 86.4 4.57 80.6 18.30 81.40 14.74 11. Coal and Crude Oil (CAO) R 0.16 0.47 0.08 18.82 1.6 0.14 84.2 96.20 22.90 11.35 12. Refined Oil (OIL) 4.14 031 0.20 1.89 51.6 2.19 73.9 19.50 7.50 23.80 13. Electricity and Water J(ELB) 1.80 1.54 1.77 0.00 0.0 0.00 0.0 0.00 0.00 piANUAClU RJINO_ 14. Food Products (POO) 4.10 5.38 0.95 3.38 53.6 0.44 95.7 10.90 1.00 35.23 15. Other Food Products (OFP) 5.40 4.24 2.16 3.98 47.7 7.40 633 12.90 13.30 27.91 16. Bverages and Tobaeco (DEV) 1.01 1.84 0.51 0.34 76.8 153 94.2 5.30 15.00 30.02 17. Texties (T) 4.10 6.07 2.28 5.3C 71.0 625 61.0 21.10 14.40 25.30 18 Clothing (CLO) 2.31 2.27 1.81 0.02 86.7 4.02 94.1 0.20 1550 34.13 19. Leather and Shoes (LEA) 1.24 1.88 0.68 0.15 89.5 4.78 71.5 2.40 3850 24.01 20. Wooden ProducU (WDS) 1.9S 2.93 0.64 2.36 31.4 C.26 83.7 17.60 3.00 19.98 21. Pper and Printing (PAP) 1.11 1.92 029 2.27 47.1 0.92 69.4 28.20 8.20 27.65 22. Cement (CEM) 1.66 358 0.41 0.95 38.1 0.11 682 11.00 0.70 21.86 23. Iron and Steel (LAS) 0.46 0.44 0.12 657 63.3 1.87 79.2 75.60 31.90 16.84 24. Electro-Mchncal Ind. ustry (E ai) 1.79 3.93 1.12 2.29 88.9 0.11 87.3 22.60 0.60 21.45 25. Industrial Machinery (IND) 0.98 0.74 0.62 17.83 45.3 0.06 68.2 74.10 0.60 20.44 26. Transport Equipment (TEQ) 1.14 153 0.46 4S6 62.4 0.25 66.1 46.90 2.10 15.04 27. Electrical Equipment (E!O) 0.78 1.40 0.17 3.84 57.4 0.13 99.4 49.80 1.60 2.I1 28. Offlce Machinery (MAC) 0.07 0.04 0.02 1.06 73.9 0.02 9c2 73.70 1.90 30.54 29. Chenicals (CUM) 3.41 3.80 2.45 .15 67.1 7.41 29.4 34.90 21.20 19.40 30. Rubber and Plastics (RBR) 0.75 1.22 0.24 L03 74.5 0.15 20.5 21.50 0.80 23.16 31. Other Industries (OTH 0.33 0.44 0.18 0.19 77.2 .05 44.7 11.00 1.50 26.26 ERVICES 32. Construction (CON) 9.51 3.83 6.44 0.00 0.0 0.00 0.0 0.00 0.00 33. Trade (TRD) 12.57 2.99 21.61 0.00 0.0 0.00 0.0 0.00 0.00 34. Transport (TRN) 5.49 3.87 4.19 1.41 100.0 9.39 57.2 4.80 20.90 10.00 35. Cormunicatons (COM) 0.50 0.46 0.70 0.01 100.0 0.02 57.2 0.40 0.60 10.00 36. Banking (BNK) 1.63 1.73 2.10 0.00 100.0 0.02 57.2 3.00 9.30 37. Insurmce (INS) 0.34 038 0.00 0.00 100.0 1.24 57.2 3.20 10.70 10.00 38. Other Services (SRV) 11.82 6.99 14.87 0.86 100.0 7.70 57.2 1.20 7.60 10.00 39. Administration (ADM) 12.86 2.85 0.00 0.00 0.0 0.00 . 0.00 0.00 _ = * Sectors marked R (for resource) have sector-sopccfic capital. 4Nontariff barrier estimated at 45 pmecnt legal tariff rates after adjustment for collections are 10 percent. 16 percent and 28 perent in cereals. sugar and meat and dairy respectively. and edible oils. These non-tariff barriers are believed to be quite binding, and Morocco intends to increase tariffs in the meat ana dairy industries, and impose variable levies in the others to cushion the adjustment costs. We therefore assume in our benchmark that the tariff levels in the meat and dairy, sugar and cereals sectors are 45 p6rcent. Shockg of the Free Trdad AqSegmet An a reault of the decade long liberalization of trade, some dlversificatlon in exports has been achieved, mainly in textiles and phosphate derivatives, but the development of export markets in agriculture has been inhibited by protectionist policies in the EC. Given the generally free access to EC markets by Moroccan producers, SC restrictions of importance on imports from Morocco only remain in fruit and vegetables, trousers, and canned sardines. At the level of aggregation of our model, increased access to the EC markets for Moroccan products will primarily Lnfluence the export price of the fruit and vegetable sectors.4 The Moroccan Free Trade Agreement proposal would involve a lor aring and eventual elLmination of all remaining trade barriers on all imports from the EC (i.e., lowering of the tariff), with correspondingly increased market access for Moroccan products into the EC, most notably in agriculture. We estimate the price distortion in agriculture due to border barriers in the EC to be about 8%, following the EC model developed by Harrison, Rutherford, and Wooton [1989). An upperbound" scenario would assume that the EC demand schedule for Moroccan fruits and vegetables is infinitely elastic, so that a removal of tariffs and other barriers will be entirely passed on to Moroccan producers as an increase in the export price. If the EC demand schedule were less than infinitely elastic, the price increase passed on to Moroccan exporters would be less than 8. Moreover, if Moroccan exporters are currently capturing some of the rents from the EC trade barriers, then the export price increase would result in less than the full 8% of benefits to Moroccan exporters. Since Moroccan 4Mormco's exporu in fruit and vegetables are dominated by oaga aed aother ciua fruits, which are both the major conomponts of our citma fruits, and by tomatoer, poatoes, and presetred vegeables, which ae the major conmontas of our vegetables sector. production is small relative to EC demand (suggesting highly elastic demand in the EC), and since the variable levy is assumed to capture rents in the EC, we take as our base case scenario the full 0% increaue in the price of fruits and vegetables from improved access to the SC market. Domestic taxee consist of the value added tax (VAT), employment and corporation taxes, and production taxes and subsidies. Sinco we do not have good updated data on collections by sector7 all of these rate-i have been set uniformly. The most important '-ax in eur model is the value-added tax, and import taxation is the next most important. The legal VAT rate is 19%, applied to both importe and domestic production. As witl tmport taxation, however, there are exemptions to the VAT. In order to be consistent with aggregate VAT revenues collected on domestic production and imports, the domestic VAT rate hao been set to 3% and the rate applied to imports to 11%.5 The VAT taxation on imports introduces further distortion in the trade regime.6 The other tax rates are calibrated based on aggregate tax collections as recorded in the SAK. They ares 1% production tax (net of subsidieo), an 8% labor tax (net of subsidies), and a St corporation tax (net of subsidies). 'In the appwdix we provide a detailed documentationof he x rae Wpled In our mol. Bely, however, tal Moroccan valadded in 1991 was 189.4 billion di;a2, wile total impots wer 59.7 bilion dirhm. Valusimdded tax an domesti snler foport) was 5359 (7853) mDllion dirhams, which represent. 3 percent of value-dded uad 11 pereeot of iports. We applied tim rat to dte uctue of value-added and impots in our 1980 SAM, which yields an amount of value-added Sovernme rev-'10 equal to 32 percat of total govenme evne. Import tae (cluding the P.F.I.) equal 11,465 million didbms In 1991, or 19 perce of the valuo of lmpots. We cale all cual taiff rates in 1991 so that the weighted aveage lat equals ths 19 percent. i yields that 30 perct of governmt revenue I our mDode Is from taiff collections. Actu valuedded ad tadff collectios i 1991 were 24 and 21 pec of ovem n rewve, repetivey, but these ae Many xes employed in Morocco that ar not presnt in our model (ach n a pesnal w , exci taxes, liceng feew and a corpora tax. With our mapping we have asured that the atmacure of the VAT and iipost ta fomn 199& gs impimed in our model. Moteove, rogarding what is importat for the revenue implcatiorn of the model, the relative in_on of toh VAT to impot te u a percet of toWl govemet revenu is presved approximately (3230 verms 24121). 'Given that the VAT is appUied on al production and in our model it i soo rebated on expor, iie entire VAT appled on impot im a diseriminatory tax on imponrs. TeMr are two methods of value-added taxatio that do ot dcdminate agalt imports: (1) apply the VAT on all domestic production, including that destined for exports, but do not apply the VAT on Lmpors (the origin principle); or (2) apply the VAT on imporls and domestic production for the domestic market only (the destination pdrciple). If all don_stc production is subject to the VAT and imports re abo taxed the VAT on import Is a discrminatorq teaiff. In Morcco, an effort is made to rebate the VAT on exports. Then, in principlo, Morocco implements the VAT according to the desinadon principle, and provided the rebte of VAT on eoRt is complete, the VAT is not dIsrmbItory against Imports. 6 2.2 general Model Structure Our Small Open Economy (8OE) model is doslgnad for trade policy analjuiL with a large number of sectors. The model io a "generic" general equilibrium - model of a alngle economy along the llnes of de Molo and Tarr (19921 and HarrLson, Rutherford and Tarr '19931. Goods are produced using primary factors and intermediate Lnputs. Prlmary factors Lnclude labor and capital. Land is not included expllcitly, but we nonetheless have a sector-opecLfLi factor by varylng the share of capital that lo sector-specific in "resource" sectors ('he nine ,esource sectors are denoted by an R following their names in table 1). Labor is assumed fully mobile acroso sectors. Production exhibLts constant returns to scale, and producers behave competitively, selecting output levels such that marginal cost at th.-% output levels equals the glven market price. In export sectors, output is differentiated between goods destined for the domestLc, EC and all other export markets. This relatLonship is characterized by a two-level constant elastLcity of transformation frontier. ComposLt* output is an aggregate of domestic output and composite exportel composite exports are an aggregate of exports for the SC and non-EC markets. Final demand by prlvate households arlses from nested constant elasticity of iubstitution utility functlons. Thioa allows consumer decLison-making to occur in multi-stage budgeting. At the top level, goods from dLfferent sectors compete subject to the budget constraint of the consumer, where all income elasticities are unity. In tbh second stage, the consumer decides how much to spend on domestic versus aggregate imports, subject to income allocated to spending in the sector from the first stage, with possibly different elasticities of substitution by commodity. Finally, having decided how much to spend on imports, the consumer allocates this expendlture on SC versus non-EC imports. In two sectors, meat and dairy, and sugar, we depart from the "Armington" assumption and assume that imports and domestic production are perfect substitutes. This is becauso there are no (or negligible) imports in the initial 7 equilibrium due to the non-tariff barriers discussed above- the Armington assumption, without very high elasticitLes of substitution, will imply (contrary to expectations) that trade liberalization yields very little increase in imports. In principle, the appropriate model is the one that is bencnmarked to an econometrically estimated elasticity of supply in the sugar, and meat and dairy Lndustries. Absent explicit estimates, we use the model that in closest to o-r assessment of the supply elasticity. In our sensitivity analysis, we adjust che share of sector-specific capital and thereby the supply elasticity, yielding alternate estimates of import penetration after liberalization. As discussed above, the only Moroccan trade distortions currently included in the model are ad valorem tariffs (or subsidies) on imports and a value added tax that is applied at different rates on imports and domestLc products. The model allows tariff rates on imports to differ depending on whether the import is from the EC or the rest of the world (ROW); and we allow exports to have different prices dependtng on whether they are sold in the SC or ROW. These distinctions allow us to study policies such as accession to a free trade area. The Free Trade Association with the SC also involves changes in Morocco's access to EC markets. The main effect of increased access would be an increase in the Moroccan export price, and these are therefore included as policy instruments that can be varied in counterfactual simulations. Government expenditures and Investment demand are exogenous. Funding of government expenditures is provided by net tax revenues. There are three other components of government income in addition to import tariffs. These are (i) value-added taxes on factor inputs to production and on imports, (Li) employment and corporation taxes on factor employment, and (iLi) ad valorem production subsidies net of excise taxes on production output. In a counter-factual scenario the value-added tax adjusts endogenously to balance government (net) tax revenues with expenditures. Thus the welfare effects of changes in trade policy explicitly incorporate the appropriate marginal excess burden of raising government revenue from other sources. 8 Since private consumption equals the income from primary factors plus net transfere to the consumer by the government (from domestic and foreign trade taxes), Walras law is satisfied. Public consumption is balanced with the value of public endowmeato and tax revenue. World market impurt and export prices are fixed, so there are no endogenous changes In the terms of trade. In other words, import supplies and export demands are infinitely elastic. The current account balances the value of exports and imports taking into account exogenously-fixed capital inflows. This guarantees no "free-lunch" either taken from or given to foreigners.7 2.3 Eupirical implementation of the Morocco model We employ a 1980 Social Accounting Matrix (SAM) for Morocco which distinguishes 39 production sectors. This provides a consistent set of input- output relationships showing intermediate, final demand and value added transactions. Table 1 summarizes some of the most important industry data employed in our model, and displays the names of each of our sectors along with a 3-letter acronym for later reference.$ Columns 1-3 show the share of Moroccan output, employment and capital by sector, derived from the 1980 8AM. Although a full update of the 1980 SAM is unavailable, appendix B shows that, at the 9 sector level of aggregation, output shares did not significantly change between 1980 and 1991. Over 40% of both output and factor employment originates in the service sectors, about 30% of the economy is in manufacturing (food and textiles and apparel comprise about half the manufacturing sector) and the remaining 30% is in the agricultural and mining sectors. Columns 4 and 6 display the sectoral decomposition of imports and exports, where the share accounted for by EC imports and exports is displayed in columns 5 and 7. These shares are updated 1991 data, aggregated from tariff line data provided by the government of Morocco. Phosphates are the most important export 7A morm formal decription of the model is given ia Appendix D. 7bis SAM was constmoted by Mstm et Ai. (1988). 9 sector, and they encounter no trade barriers in the ZC. Fruit and vegetables exports together make up 14% of all exports to the PC. Columns 8 and 9 show the importance of trade for each sector. Clearly the mining sectors are very dependent on exports, as is the citrus fruit sector to a lesser degree. We therefore expect some benefits to this latter sector from increased access to EC markets. The benchmark values of all elasticities in the model are reported in Appendix A. Estimates of elasticitie must be assembled for primary factor .substitution, import demand, import source, domestic demand, and the transformation of domestic supply into domestic and exported products.9 Despite our literature search, there are many elasticities about which there is considerable uncertainty. our "remedy" for this problem, which is endemic to any large-scale model of this kind, is to undertake systematic sensitivity analyses of our major results with respect to plausible bounds on these elasticities. Even if we are unable to specify a point estimate with any precision, our priors over the likely bounds that these elasticities could take are quite strong. To the extent that our major conclusions are robust to perturbations over these bounds, we do not see our uncertainty over specific values of these elasticities as a weakness of the model.10 We report the results of these sensitivity analyses, which involve a minimum of 1000 simulations for each counter-factual policy in Section 4. They will allow us to conclude whether or not our main results are robust, at least with respect to plausible uncertainty over elasticities. We numerically elaborate in appendix C, the model parameters that define the gross substitute-complement relationship between domestic and export production. Although this relationship is important for sectors such as citrus '11in detail, thene elasticities refer to the elasticy of ubstitution betwen primay fuctor of production in each ector; the elasticity of substitution between domestic production ad an impoes composite in each ector, the ebaicity of ubstitution between imports dinguished by source, also by sector, the elasticity of substitution between domestic consmption of each good (the compones of which are, in uns, compositus of domestic and imporled production); and the elsicity of transformadon of domoesic production into does tic uss and export. 'sMes remarks should not bo interprcted as denying the value of any new empirical work on gnerting such elasdcities. On the contrary, any effort that could generate better boundson the e point estimats i eful in generating policy concluson that carry greater credibility, even if those conclusions will sill be probabilistic i nature. 10 fruits and vegetables, whlch are expected to experience export price lncreasoe, lt is typleally not transparent in models of thli type for the followlng reason. Let a denote the supply elasticity of the composite output ln a sector and j denote the elastlclty of transformatlon between domestlc and exported output ln a sector. Abstractlng from general equlilbrlum effects from other markets, de Halo and Tarr 119921 show that lf and only if a > ja, an lncrease ln the export prlce wlll lncrease output of the domestic variety and ralse the domentlc prlce, l.e., the lmport and domestlc varletles are gross complements 3.n production." Although a slmilar condition exlits in consumption, all elasticlties in consumption are entered parametrically, and lt li straightforward to examlne whether the import and domestlc varletles are gross substltutes.12 Although jo li entered parametrically, a (the industrywlde elasticity of supply) is only defined lmplicitly and, ln a model with constant returns to scale such as ours, could potentially assume extremely large values, especially for small sectors where output expansion will not significantly alter the relative costs of its inputs. we report our basic results for the cases of low, medium and high industrywide elasticities of supply. We implement a change in the elasticity of supply, by assumlng three different shares of sector speciflc capital ln all the resource (R) sectors: 50, 75 and 90 percent. Cotarua parlbus, the larger the share of capital that is sector specific, the lower the industrywide elasticity of supply. Appendix C numerically elaborates the relationship between the assumed share of sector-specific capltal ln the cltrus anO vegetable sectors, and the Itibe intuion for ths result as foliow. When th export pric increa_s, fim tovouer wil incae if output levels a held con Frms will therfor puhase more Iu In order to produce mom. DEigarding th relativo pice chag betwen domec and export makt for a moment, this wald res In an inoae In conpoolt pducton, I.e., of goods desind for both tbe domesic and expost murke. li increae In comoste uput we label the output effect. The extent of ths effet depend on e. It i equivalt to the Income effect in consumer theory. Thero is also a tansformaton effect, however, my from produciWg domestie goods in favor of export production, due to the change in the relative price of exports to domesi vadeties. The extet of this trnsfomat effect depends on , the tansformation elastiity, and it is equivalent to the subsiution effect in consumer theory. When e > p, the output effect domin tho trformtion effect, and the goods are gros complements. '2da Melo and Tarr (1992 show that a necess ad sufficient codition for the price of the dometic import competing good in a sectr to be a gross substitute with the import good is that the price elstieity of demand for the composite Armin good is less than the Anmington substitution elatieity. This condition euns thtth ubtitutio effect wil dominate the icom effect in demand. A similar rsult is dscued in Rutmt6m (9M2. 11 supply elasticity of the composite output for domestic firm. in these sectors. In the case of citrus fruits, the Industrywide elasticity of supply varies from about unity to 3.5, but takes slightly lower values in the vegetable sector. Given that the elasticity of transformation (go) is S in our benchmark, this implies that exports and domestic outputs in these sectors are gross substitutes. In the present version of the model we only have one private household in Morocco. It is important to note, however, that there are several powerful theorems in international trade theory to show that one can effect Pareto- efficient reforms for multiple households providing there are aggregate (real) income gains and one accepts some weak conditions on patterns of demand and ownership.13 These results do nct rely on the availability of lump-sum redistributive taxes, nor do they address the issue of an optimal reform package. What they do show is that one can focus initially on aggregate gains in income and welfare, knowing that the redistributive aspects of the problem do have a solution that leaves each household at least as well off as before the reform. This is not a complete substitute for actually solvlng for the equity effects of a reform package, but it is a partial substitute. The SOB model is generated with the GAMS software developed by Brooke, Kendrick and Meeraus [19881 and solved with the MPS/GE software developed by Rutherford (19891. The systematic sensitivity analyses are undertaken with the ZPSS software developed by Harrison (19901 and using the procedures developed by Harrison and Vinod 119923. 3. RBSULTS The policy simulations that we consider and the aggregate results on welfare and taxes are summarized in table 2, and the employment effects by sector are summarized in table 3. In the following section, we present the results of systematic sensitivity analysis to determine the robustness of the results to a Sc Dixit and Norman (1980; pp.79/801119861. The conditons on demad d factor ownerahip pa a pdmaily to nde opure exchange' economies. 7Tea conditios are tivialy met in our model. 12 parameter specificption. One parameter which stands out in importance regarding some of the results in the industry-wide elasticity of supply assumed in the resource sectors. Consequently, ln this section, we present our "best guess" estimates under three different assumptions regarding the lndustry-wlde elasticity of supply in resource sectors. 3.1 Welfare, Revenue and Aggregate Primary Factor Movement The first three columns of table 2 show the welfare gain measured as the Equivalent Variation as a percent of benchmark GDP. Columns 7-9 show the percentage labor adjustment measured by the necessary reallocation of labor across sectors as a percentage of total labor supply. Columns 10-12 likewise measure the necessary reallocation of capital across sectors as a percentage of total capital supply. The results depend on the industrywide elasticity of supply assumed in the resource sectors. The greater the elastlcity of supply, the more resource movement across sectors (more labor has to change jobs), but the more welfare gain as well. For example, under ACCESS, the cLtrus fruits and vegetables sectors obtain higher EC prices. With larger elasticity of supply, they expand output more. ThLs results in greater welfare gain to the economy, but also more movement of capital and labor between sectors. Conversely, the meat and dairy, sugar, and cereal producers will suffer a decline in demand as a result of lowerlng tariffs under all scenarios except ACCESS. The greater the elasticlty of supply, the more output, labor and capital reduction there will be in these sectors, but the larger the welfare gain as a result of shlftlng these resources into more efficient sectors.4 All the results of columns 1-3, .7-9 and 10-12 follow thls pattern. '"Th different elsticities of supply re implemented through varying the are of ector specific capiwl. Given ector ecific capital, in reponse to a change in demand, the rent on capital in the sector will cbanp which implies there will be less movement of tsource in or out of the sctor. For example, specific capital owners in citrus fruits and vegetables ear greater ren under ACCESS, but the increase in their rens increaes their pnces and dininishe the expansionofoutput. Convesey, in contmcting ectors, ector pecific capital rsults in a eduction of ren, coss and prices, and a diminished reductioa in output. 13 Table 2: Free Trade Agreement with the EC and Related Trade LLberalizations Welfare, Tax and Factor Adjustment Effects on Morocco' ________ % Chango in Welfare %Cbango in VAT e X of bbor that hange jobs * of capital t adjua H M L H M L H M L M MN H FrA 2.2S 1.52 1.20 54.0 583 60.3 3.2 2.S 2.2 5.1 3.3 2.7 EC81 2.05 1.29 0.97 54.6 5dS9 60.9 33 2.6 2.3 5.4 3.6 3.0 ACCESS 0.31 0.27 0.25 -1.8 -1.2 -1.0 0.4 03 0.3 0.6 0.4 0.3 LIBROW 1.86 1.10 0.78 55.7 60.1 62.0 3.2 2.5 2.2 5.3 3.5 2.9 LIBALL 3.12 2.37 2.06 80.7 8S.5 87.7 3.7 3.0 2.8 6.0 4.3 3.6 FIAALL 336 2.60 2.29 80.0 84.9 87.1 3.6 3.0 2.7 5.6 3.9 3.3 a. All aslmlations use tho Value-Added Tax as replcement tx. Reult amt for high (H), medium @4), and low (L) elatcity of supply in rceource ecton. DESCRTION OF POLICIES: FrA ... FuU free trade agreement with the EC. Inmreaed expost price for chtra fmuit ad vegtables for SC deintion by 8 percent, dimination of inpost proetion from EC sources. ECLIB ... Eimination of impont protection againt C import. ACCESS ... Incraed expont pries for citms fiuits nd vegetable to EC deations by 8 porco. LIBROW ... Emiation of impost protection against non-BC impots. LIALL ... Eliminaon of impost protcwtion against all imports, BC and n-BC. FTIAALL ... Ful free trade agreemt with the BC, augmened by elImiation of impost proteon from non-SC su. as weU. Flrst conelder the pollcy scenario ACCESS. Improved access to the EC for citrus fruits and vegetables will lmprove Moroccan welfare by slightly more than one-fourth of a percent of GDP, due to improved resource allocation and better terms-of-trade in cltrus fruits and vegetables. Since columns seven through nine show that factor movement is small under ACCESS, It is primarily the terms-of- trade improvement that is provldlng the benefits from improved access. Removing tariffs against EC imports (ECLIB) results ln an improvement in Moroccan welfare of between one and two percent, which is about 4 to 7 times the benefits of lmproved access alone. The free trade agreement (FTA), which combines the polLcies of ACCESS and ECLIB, results in galns ln Moroccan welfare that are 14 roughly additive in the separate policies. Removing tariffs against the non-EC rest of the world (LIBROW) results in !jains of about 80 percent of those from liberalizing trade with the EC alone, reflective of the fact that the EC is the larger trade partner. If tariffs are lowered against all imports (LIBJLL) another substantial increase in Moroccan welfare (of about one percent of GDP) is obtained (compared with ECLIB), interestingly, without significantly additional shifting of labor and capital among sectors. The reason that the additional welfare is obtained with little additional resource movement is that lowering tariffs against only the EC induces resource movement, but that resource moveme"t is not necessarily toward the most efficient sectors by world standards, i.e., there is trade diversion from the Morocco-EC Free Trade Agreement.0 Resource movement that is induced by trade diversion will not occur when tariffs are lowered to all supplying countries. The significant benefits that accrue from discrim&.natory liberalization against either EC or rest of world imports indicates, however, that trade diversion is not dominant. Liberalizing tariffs to the rest of the world in combination with a free trade agreement (FTAALL) with the BC results in benefits that are roughly additive in the separate poll3ies that make up FTAALL, i.e., LIBALL plus ACCESS. All simulations are performed assuming that the rate of VAT taxation would be altered so that revenue to the government is unchanged. For scenarios involving reduced tariffs against the EC, columns 4-6 of table 2 show that the VAT would have to be increased by about 55-60 percent. This means that the VAT collection rate on imports would rise to about 16-17 percent (from the collected 11 percent) and on domestic products to about 4-5 percent (from the collected 3 percent). For scenarios involving full tariff liberalization against all imports, the VAT rate would have to rise by about 80-90 percent. "Trade diverson would occur In a Moroccan Fre Trade Ageentni with the BC, when a cupper outside the SC would spply the product to Morocco at a cheaper price than the EC suppier, but the baff incluJvN price of the DC supplier is chae. Ttade diversion coss ae more likely to be high relative to trade ceation benefits: (1) the higher the tariff rcte ginst and M) the lager the are of trde with the countries that are excluded from the integation agreement. We alo show in ection 4.1 tt (3) the lower elascticity of mbution of composite impott ad (4) the higher the elsicity of substitution for imports fron differeot counties of origin, the grea the relte cost of trade diversion. 15 To sum up the aggregate effects, there are significant trade diversion costs when only partially liberalizing import protection, implying that a complete elimination of the protective system would result in higher welfare gains than a free-trade agreement with the EC. horeover, tnmere is a clear correlation between the welfare effects and the necessary factor adjustments. The higher the welfare gain, the higher is generally the adjustment needed. One interesting conclusion is that broader trade liberalization yields greater welfare gain than the FTA, but with only slightly higher adjustment costs. This can be seen as an important argument in favor of the efforts towards lowering tariffs against non-EC sources subsequent to achieving a PTA. Finally, welfare benefits of 1.5 to 2.5 of GDP, from the free trade agreement or broader liberalization, is quite substantial for models with constant returns to scale. We have shown, however, that a considerable portion of the benefits derives from eliminating dispersion in the tariff regime, since dispersion is eliminated in the process of liberallsation.16 3.3 Employment Impact by Sector The sectoral employment adjustments occasioned by the policies are depicted in table 3. By far the sector that gains the most from the trade liberalization scenarios is the phosphate sector, where employment increases by over 60 percent. Despite the fact that phosphate exporters do not obtain improved terms of trade on world markets from the free trade agreement, the reduction of tariffs depreciates the real exchange rate in Morocco, and allows them to obtain more in domestic currency for their exports even if the price of their exports in foreign currency is unchanged. 17 Citrus fruits, vegetables and leather goods (all significant exporters) are the other industries that expand the most. 'lia hct, when we fis humonied aiffs for al uen to dir weightd avenge evel in the bmmnuk C1IJ pero) ad basquweny simuted dh vaious poliies of shown in tble 2 (in the maum ldticty sco) we obtined considerably maer benfits. m potular, tse welfae benefits a a peecenge of GDP ae as follows: FA, 0549; BCLE, 0325; ACCESS, 0.243; UBROW, 0290; U3BALI, 1.233; FrAALL, 1.456. Thi shows that about two4-id of the benofits in many of the sdos deives from ha_monzation of the tuaf regime. "Formlly speaking (since there is no money in the model, the price of expots buys more in terms of domest nontraded goods. 16 Conversely, since the cost of the imported goods rises in tnrms of non- traded goods, sectors which are primarily non-traded generally lose employment after the PTA or other liberalizing scenarios. The largest losers of employment, however, are the agricultural sectors that lower tariffs significantly (cereals, meat and dairy and sugar). For reasons mentioned above, we treat two sectors in the model as homogeneoust meat and dairy, and sugar. In these sectors the industry-wide elasticity of supply play. an especially important role in determining the decline in employment. The elasticity of supply in these sectors is controlled by the share of sector specific capital. More precise estimates for these sectors could be obtained if econometrically estimated supply elasticities were available. The estimates for these sectors are illustrative, given our best guess of the supply elasticities. 17 Table 3, Percent Employment Change by Sector and Scenario * I FTA EI.3 A5SU 3.331W LIALL fIALL SEOR 1(A_v.) N U L H 3d L H 3 L H U L a U L N t L 1. C.mh (CE) lw .13 94 46 4la .9.5 .9 . .i -e1.1 -9.1 .6.3 .13A -333 .30.9 -333 -llf .101 2. &w OM f -32 4.4 41.5 23 41.1 4.2 -2.4 43 -1.0 42. -.5 1.6 -,p.5 o41. -56. -7L7 - 4" 3. Cb FDm (CU) a 2A4 I73 15.9 23 I .J IA 17I2 I" 333 2.2 W . I3 2. 2.2 2. 23. 182 3S4 aL V _,IabI(e0) a 9.1 6.4 Li -3.1 .34 43 89 .7 a.7 -33 .20 .2. 4.4 -..7 -3.9 93 ae . 7.1 5. %di a Wq (MA 1 47.3 4.e 43 47.1 .53.1 453 - IJ 4.3 4.6 67.7 42 .4.7 64 4.2 413 43-as 4" . IL Fhb, MO f .2* -2*, 43 .22 .33 42 .3 e02 62 0.1 0.2 .1 3 0.4 3J 13 es 7. )'muuy*ceuAviv. (FOR) -.4.7 9 s e 0 43 .4.5 .4.7 -43 43 .03 3.3 -3.3 .. -1.3 -724 -7. -73 -73 AD .ATED a ftmobm(MM5 924m .5 53.I 3.7 7P9 67.3 46 -7.1 -62 102 733 1.1 333 n3.5 719 3t. 3)* *10.7 9. Ost Nmua MhftS 6 .7 sA 6.1 53 .3 47 43 21 -245 44 133 -3J .3 -J.1 -IA 43 3 mm" au (We a 3J 2.7 2A 4* 2.9 2. 4.7 4A 4.3 29 2.1 33 4.2 3* 2.7 3 2* 2.5 It.cu3ad Ca& on CA) 1 93 3.5 8.1 9gJ OA a 40 4.7 47 4- 43 41 2A 23 2.7 21 2.3 22 t. oboO -2*w -3A -3 *I 33 0.3 0.1 0J e 50 42 4A 2.3 2.a 4.2 7 2 13.32sdaidumd watow.D) -L.l 4A 4 -*1.0 42 0.3 4L 43 43 -*I 4.2 6. 4 0.31 OA -ID 43 03 MANUFAC IUDO 14. Food P (/FM) 4.0 452 J9 4 4 43 0.1 0.1 0.31 - 2 4.* A 43 4J5 -1A 47 -4A I5. Od Food to06w (OF3? -3.9 -3I .2.3 4* .3.3 .2 43 46 4.3 .1 -3J .4.3 43 4* 41 7 4 14L $moqsw d T_o OM" -2.2 .23 .23 .13 .23 -4 43 42 43 4* 14J9 .13 .3* -3. .3.1 -J -3 17. T. MMXT -44 .7 .3 4* 39 -33 43 42 4 -2.7 -2.7 4 43 4 3 4.2 IL ae () -* 4.1 4.1. -,1 *.1,1 43 42 4.2 43 -3.7 -IJ -13 -IJ -J -41 -3. .*I -1.9 it. Laiie Sad e (IA 33.2 30.9 9.7 14.3 31. 303 .I3 433 .3.1 3a 5.7 4.5 l.7 Ms5 9 I 9.9 I 23 Wood bII (WDH) 03 Os 0.5 0 4 0O5 4L 40 0* -. 7 a * - -2 -2 -2* .2A .13 -IJ 21. r_red tar (PAIP) 2.a -27 2.7 .2A -2.7 -2 43 4.2 42 -3.3 -3.2 -32 -5* 4. .4* .-5 .4 As 2. C E) .13 .3.1 13* -33 -3.1 49 4.3 43 43 1. 4 .33 .33 -22 -2* -1 -2.2 -2* -s 21kmadS Sala) 2.5 Is 1.5 3a 23 2w -43 - * 4 332 2.7 2A4 [A. 4A 0. ID 4. 43 24. lJ.a&Ueim3 _Oy .42 .37 34A 43 -.3 -. Qa O D 0 0 e9 33 42 1 1 4J 4* 25. b9W Ma y OMND -3.7 4J 4 S -24 4J22 44 44 4OA .7D 4. 1 4 - 42 .33.3 -9.9 -13 -1. -33. 2L T _mu Eq _ (m) .1. 43 4.1 -1. -1.2 49 43 43 0.2 0.7 Is -2 -2a -3. - -2. 4- 21. EB IE01 4.m1 4. 43 -. 4* 4 4J AA 43 43 .7 .4.5 -4A .300 43 .932 *10A 4J .03 21. oUbf ! y AOUX) -72 -171* -36. -363 -3" -165 43 47 4A .32 .-n 439.2 -49.1 -190 -B3. -9A 2. bt7db 3O 03 43 IJ 0.7 03 49 47 7 5.1 4J 4A Si 22 Is 2.5 La I 30.3d6o1.a.r.(d31O 47 .13 -I3 4A4 -2 -IJ 43 402 42 2A 1. 3.3 4.1 49 43 4A -. -43 31. O&w &"M 61n) -33 4J .13 42 41 -I3 3.1 0.3 0 -IA -3j 4S -32 -J3 .3 -1S -3 -43 m. . 3.C4MmbU (CON) 3 03 -345 0.2 ID0 0 4.1 42 4J 032 IJ -3A 03 w -1A 0 33. Tl (1`D) 4. 4S 4.9 4.7 -a .3J 13 0. 7 0A 43 -2.5 -1. 43 -33 -2 45 -24A 4A 3L T _mp. (ntN) 23 22 2l 23 Li7 2 40 43 45 53 52 S 5 5A 5.A 5. 4.9 49 35. CoMmIlw (00U) -1.0 4L 03 -3 6 O 3O 4. 1 4. 43 4S e 5 09 450 4.5 IL 0.1 36X '.¶a m 0NI 0.4A OA 13 3. I3 3I 43 43 4-A " IJ I. 0.7 U 33 u. la I2 Yr _aNm m 0A4 0.9 3.3 O.4 1 la 4.4 43 43 1.7 2.2 2A 12 3. 2I la 1.7 Is 3S. oeb ItSl(3XV) 4 03 03 43 0.7 3J e.1 4.2 -03 43 03 I.3 0.3 3* IA 0.3 l0 39. Aihmb&Ubm (ADkd 4.3 42 432 4 4.2 43 * 0* 0D 4.1 42 42 4 41. 41 . 41 4J 4. 4J * H, M, L reer to bigh. medium ad low _ngywidy elascity of ippy In hb. mesco () se oma. ** Sectrsm denoied with R (mause) hv sect ific cap . Souce: Moded esitiutes 18 4. INACT OP KU PAPAIW_U AND SUSITIVITY ANALYSIS 4.1 Impact of Key Parameters Sensitivity analysis over the parameters of our model has revealed the parameters which are most important regarding the welfare, revenue and factor adjustment estimates. One which we discussed above is the industrywide elasticity of supply. We have also found that the results are sensitive to the elasticity between imports and domestic consumption (the Armington elautLeity), and to a looser extent the elasticLty of substitution in consumption between imports from the EC and the ROW. in this section we discuss the impact of these parameters as well as the Armington assumption in sectors with small inLtial shares of imports. Elasticity between Domestic Consumotion and Comoosite Imoorts In the results repirted in table 2, the Armington elasticity is equal to 2 for all sectors other than sugar and meat and dairy (imports and domestic production are assumed to be perfect substitutes in these latter two sectors). Increasing the Armington elasticity for all Armington sectors increases the welfare benefits, as shown in table A2. At a value of 10 for the Armington elasticity, the welfare benefits of the integration-lLberalization strategies increase more than 3 times in all scenario. except ACCBSS (where trade diversion is not an issue). At a value of 1 for the Armington elasticity the welfare benefits are reduced. In figure 1 we provide an interpretation of the welfare economics of why an increase in the Armington elasticity increases the welfare benefits of trade integration. To simplify, and to isolate the impact of the Armington elasticity, we assume that imports from EC and ROw sources are homogeneous in the preferences of consumers. (In figure 2 we show how to generalize the graphical interpretation to where imports from different sources are imperfect substitutes, as in our model.) The case of trade diversion in a given sector is depicted. Tariffs are lowered preferentially against imports from the SC, but imports from the ROW are the cheapest. The cost advantage of ROW suppliers is not large enough to overcome the tariff preference toward the EC, so that Moroccan importers shift from all ROW imports to all EC imports. Initially equilibrium is at El and shifts to 19 1: of i in Increase with the Elasticity of Demand for Composite Imports SUPPLY AND DEMAND FOR HOMOGENEOUS IMPORTS Price of (1 +t)PEc Imports ~~~~EC~~~E LAOW +t)FROW (Price from ROW) I IE I' I~~~~~~D I I - _ ', QO Qi, QlE Quantity of Imports The trade diversion case, where consumers regard imports from different sources as homogenenous, is pictured. If tariffs are eliminated preferentially against the EC, the welfare change is equal to: A - T with D, (inelastic demand for imports), or A + B - T with D. (elastic demand for imports). In the trade creation case there is no area T to subtract, and the triangles A and B extend down to the price of the low cost supplier. either El or ER dependlng on the elasticity of demand for composite imports. Ceterus parLbus, the laxger the Amington lasti.city, the larger the elasticity of demand for composite imports. Consumers, surplus analysis (our general equilibrium model uses 'UcksLan equlvalent variation) implies that the net change in welfare is equal to A - T in the inelastic case, or A + B - T in the elastic demand case, i.e., the triangle B representu the net difference in the welfare between the elastic and inelastic demand cases. In the oauo of trade creation, there is no rectangle T to subtract and the triangles A and B extend down to the delivered price of the low cost supplier; but an enlarged triangle B remains the net difference in welfare between the elastic and inelastic demand cases. ElastLcity of Substitution betweo i Imports fro- Different sources Our benchmark value of the elasticity of demand between imports from the EC and imports from the ROW is S. In table A2, we show that increase.ng this value reduces the welfare benefits of preferential tariff reduction, either against the SC or against the ROW. This parameter has no effect on the welfare effects when the tariff changes are not preferential. Figure 2 depicts how an increase in the elasticity of substitution among imports impacts on the estimates of the change in welfare. The demand curves of Moroccan consumers for imports from the SC and from the ROW are drawn to show that they depend on their own price and among other prices, most notably the price of the import substitute (in all cases it is the tariff inclusive price that is relevant). Preferential tariff reduction against EC imports will reduce the tariff inclusive price of EC imports, and therefore induce an inward shift in the demand curve for imports from the ROW. The inward shift in the demand curve for imports from the ROW will be larger the larger the elasticity of substitution among imports from different sources. In the market for EC imports, triangle A is the gain in consumers surplus that is not offset by lost tariff revenue. In the market for ROW imports, there is a loss of tariff revenue equal to T, in the low cross-elasticity case or equal to Ta + T2 in the high cross- 20 FIGURE 2: Welfare Benefits of Trade Integration Decrease with Greater Substitutability among Imports IMPORTS FROM EC IMPORTS FROM REST OF WORLD 0) 0~~~~~~~ 0~~~~~~~ 0. 1 +t ) P 0~~~~~~~ 0) Din1 [(1+ opR; PIC DECI [(I + PI; ( + tP*J 0. DR, [(1+ tP; PE, D(U1+tPR; (1 + OPEC] Quantity of EC Imports Quantity of Imports from Rest of World 0 If tariffs are eliminated preferentially against the EC, the welfare change is equal to: A - T, with D13, (small cross-elasticity of demand between EC and Rest of World imports), or A - T, - T2 with DO. (large cross-elasticity of demand between EC and Rest of World imports). elasticity case, that iL not offset by a consumers, surplus change. Thus, the nat change in welfare is equal to A - T, or A - T, - T1 Homogeneous or Armington Sectors Since both the sugar and the meat and dairy sectors have little or no imports, we have, assumed that they are homogeneous sectors in our model. An alternative modelling procedure would be to assign a very small amount of imports to a sector that has no imports and treat the sector am an Armington sector. In view of the above discussion on the impact of the Armington elasticity, it should be apparent that treating the sector an a homogeneous sector (equivalent to an .infinite Armington elasticity) will increase the welfare benefits of our trade integration scenarios, i.e., the Armington assumption mutes resource movement and reduces the welfare impact.19 What may be less apparent is that under the Armington assumption, in response to a change in trade policy in a given sector, resource movement and the welfare impact will be quite small in sectors with a small import share compared with sectors with a large share of imports. For example, with our point estimate Armington elasticity (elasticity of substitution) of 2, a fifty percent decline in the relative price of imports will induce a 100 percent increase in the ratio of imports to domestic sales iA consumption. But if imports were le than 0.5 percent of consumption, they will remain under one percent after their relative price reduction, i.e., the absolute increase in the import share of consumption is lese than 0.5 percent. Wlth the same elasticity of substitution, if a sector has a significant initial share, the same relative price reduction will result in a much larger absolute increase in imports as a percent of total consumption in the sector. That is, ceterus 'OTe jurtification for the welfar nalysi of fi6g 2 b equato(8) [or its ecW cu, equio (S".). in Harberger (1971). Habrge considers the cme where thene is a change in the tax on good I (in ourcae leng the tariff Udst C impns) in the presce of tues oan other goods in the economy my LJods 2,...,n. In our cm, the most notable oher tax b te ardff on copting impot from the rt of the wodd ia the same ector. Then, the chnge in welfare b the chnge in scuplus oan good 1, plus th chag in .uphu on goods 2,...,o, wheo the latter is equal to the tax In the other setor tim es the change in quantity In thonectows, anmed over al mcectost. To simlify figue 2, and becan the crosubasion effect in deanod will be smaller and of either daa in odr sectors, we have ignond sector otuade of the directly competing import sector. Our quantitative alysis, however, whch b baed aon Hicksa equWalt variaton, icorp the welfae changes fiom all goods. "See de Melo and Tarr (1992, chapter 2) for an elborion. 21 paribus, there is dramatlcally lese resource movement (and consequently welfare impact) in the case of a small initial import share. In the case of sugar and meat and dairy, it appears inappropriate to model these sectors as Armington sectors after quantitative restraints are removed, where the removal of tariff equivalents of even as high as 100 percent would result in an import share of less than 2 percent. These sectors have been protected by quantitative restraints because of the fear that they would contract precipitously. On the other hand, the homogenous product assumption will tend to result in excessive resource movement without the presence of specLfic factors of production. Thus, as discussed in section 3, we have employed specific factors in the homogeneous sectors, to approximate the appropriate supply response. 4.2 Sensitivity Analysis How robust are our major policy conclusions to the many assumptions of our numerical model? We answer this question partially by considering a systematic sensitivity analysis of the main results with respect to all of the elasticitLes of the model.' our sensitivity analysis employs the procedures developed by Harrison and Vinod (1992). Essentially these procedures amount to a Monte Carlo simulation exercise in which a wide range of elastLcities are independently and simultaneously perturbed from their benchmark values. These perturbations follow prescribed distributions, such as a t dLitribution with a specified standard deviation and degrees of freedom, or a uniform dLstributLon over a specified range.21 For each Monte Carlo run we solve the counter-factual policy with the selected sit of elasticitLes. This process is repeated until we arrive at the desired sahple size, in our case 1000. The results are then tabulated as a distribution, with equal weight being given (by construction) to each Monte Carlo run. The upshot is a probability distribution defined over the ondogenous W. pprciaatha the aro my otheraetlom thtat moafixd weutvay easci, butgardthor exmoemu beyoand the mcope of the prewnt study. 2 lbe exact diotibutional asumption used ae documewed in Appendix A. 22 variables of interest. In our case we focus solely on the welfare impacts of each policy. TABLE 4: Results from sensitivity analysis _________ FTA ACCESS ECLIB LIBROW LIBALL FTAALL sample 1005 2284 1096 1002 1002 1078 size . . . ._._ I WELFARE 1.52 0.27 1.29 1.10 2.37 2.60 PE , . , I Median 1.72 _ 0.29 1.51 1.39 2.73 2.96 Mean 1.74 0.29 1.51 1.39 2.72 2.96 St. d. 0.39 0.02 0.31 0.32 0.35 0.35 Prob>O 1.0 1.0 1.0 1.0 1.0 1.0 Prob>PE 0.705 0.789 0.721 0.870 0.809 0.826 50% LB 1.49 0.28 1.27 1.11 2.43 2.69 50% UB 2.00 0.31 1.77 1.64 2.96 3.23 75% LB 1.38 0.27 1.14 1.00 2.26 2.52 75% UB 2.14 0.32 1.90 1.79 3.12 3.40 VAT rate 58.3 -1.20 58.90 60.10 85.50 84.90 PE Median 56.33 -1.55 57.08 57.11 81.69 80.99 Mean 56.39 -1.58 57.11 57.25 81.75 81.01 St.d. 2.95 0.03 3.07 3.24 3.13 3.07 Prob>0 1.0 0.0 1.0 1.0 1.0 1.0 Prob>PE 0.262 0.086 0.297 0.197 0.133 0.129 50% LB 54.42 -1.81 54.96 55.06 79.52 78.75 50% UB 58.49 -1.39 59.35 59.63 84.19 83.34 75% LB 53.02 -1.99 53.53 53.72 78.11 77.32 75% UB 59.90 -1.29 60.60 61.55 85.70 85.00 LABOR 2.4 0.3 2.5 2.5 3.0 2.9 ADJ PE I a_ _ Median 2.65 0.32 2.75 2.77 3.48 3.38 Mean 2.67 0.32 2.78 2.80 3.49 3.39 St.d. 0.36 0.04 0.38 0.39 0.48 0.45 PE - point estimate, st.d. - standard deviation, LB - lower bound, US - upper bound 23 The results of the sensitivity analysis are reported in Table 4. In the interests of reporting all of the pertinent data in a compact manner, some of the column and row headings are necessarily uome4hat cryptic at first glance. The acronyms for each simulation (column headlng) are defined in Figure 1. The "Sample Size" row refers to the number of Monte Carlo runs that were actually completed. In each case we have at least 1000 runs, whlch should be enough to obtain a reliable picture of the dlstribution of results. The "Point Estimate* (PR) row shows the effect of the policy when all elasticities are set equal to their benchmark, or point estimate (PS), values. These are the results reported .and discussed earlier. As before, we report the change in welfare due to the policy as a percent of GDP, the revenue as the change in the VAT rate required, and the labor adjustment as the percent of the labor force that is reallocated. The remaining rows report the results of the sensitivity analysis proper. We list the median, the mean, and the standard deviation, so as to provide simple indicators of the location and dispersion of the distribution of results. We do not report here the skewness and kurtosis statistics that are necessary to gain a more complete impression of the distribution. In all cases we find that both the skewness and kurtosis are insignificant. In order to obtain an indication of the qualitative policy results we report the "Prob. 2 0" row for the welfare and the VAT rate results, which shows the probability from the empirical distribution that welfare increased in the counter-factual policy. This gives us a measure of the confldence that we have the sign right when we look at the Point ZstLmate effect or the Mean or Median. Similarly, we report a row showing the probability that an effect greater than or equal to the PE effect was obtained. If the PR result is perfectly representative of the location of the dLstribution of results we should see this value around one-half; this would be the case if the PE result exactly equalled the reported Median result. A value lower (higher) than one-half indicates that the distribution generally lies below (above) the PE result.A bWdh the excepdon of the itenediate input subibdzion, where the point eanmte is 0. 24 Flnally, to galn a better sense of the confldence to be attached to the PR or Mean welfare and VAT rate result, we report lower and upper bounds from 50% and 75% symmetrlc confldence lntervals around the Median result. Theme confldence lntervals simply show the smallest and largest values that 1Le wlthin 50% or 75% of the dLitribution centered on the Medlan. Thus a 50% confidence interval between 1.1 and 2.3 can be Laterpreted as saying that 50% of the Monte Carlo runs resulted ln welfare results between these values. What, then, do we learn from these sensLtLvity analyse. regardlng our pollcy conclusLons? Most mean and median welfare effects are above the point estimates reported earlier. These hLgher welfare effects are also coupled wlth higher adjustment costs. We can conflrm our conclusLon above that PTA is a preferred policy package to BCLIB ln the sense of provldlng hlgher median welfare galne at lower median adjustment costs. Note, however, that the mean welfare effect for ECLIB lies wlthln one standard deviatlon of the mean for FTA. There ie therefore considerable overlap between the two distributions so the concluslon regardLng whlch pollcy package is preferred might stlll not be robust. We flnd, however, that the welfare effect ln PTA is greater then the medlan welfare effect in ECLID with a probabillty of 0.711. The reverse case that the welfare effect ln ECLIB Li larger than the medlan welfare effect in PTA only occurs with a probabillty of 0.282, however. Similarly, we flnd the second-best argument ln favor of elimLnatLng all import protection and not just protectLon from EC competitlon to be robust to varLatLons ln the value of key parameters. The welfare effect in LIBALL is much larger than the welfare effect of ECLIB, but wlth not much additlonal labor adjustment. InterestLngly we flnd that the median revenue effects are smaller than the poLnt estimates. In no case would the VAT rate have to increase to more than about 5 5 and 20 percent from a benchmark value of 3 and 11 percent for domestically produced and lmported goods, respectively. In summary, we flnd that our general conclusions are quite robust with respect to any uncertainty ln key parameters. Welfare and labor adjustments tend 25 to be higher, due to the inclusion of higher Armington elasticities, but the revenue effect is smaller. 26 Brooke, Anthony, KendrLck, Davld, and Meeraus, Alexander, (198831 GAMS: A User's auide, Redwood Clty, CAh The Scientlflc Press. Dixit, AvLnash K., and Norman, VLctor, (19801; Theory of Intornational Trade, Welwyn, UK: NLibet. Dixon, Peter B., Parmenter, Brlan R., Sutton, John, and Vlncent, Davld. P., (198231 ORANI: A Multisectoral Model of the Australian Economy, Amsterdam: North-Holland. Harberger, Arnold, [1971]; "Three BasLc Postulates for Applied Welfare Economicst An InterpretLve Essay", Journal of Economic Literature v.9, 785-797. Harrison, Glenn W., 119901; "The Sensitivity AnalysLi of Applied General EquilibrLum Models wlth MP8S: Users GuLde", Unpublished Manuscript, Department of Economics, UnLversLty of South Carollna, June. HarrLson, Glenn W., Jones, Rlchard, Kimbell, Larry J., and Wlgle, Randall, [1993); "How Robust Is Applled General Equillbrlum Modelling?", Journal of Policy Modelling, v. 15, forthcoming. Harrison, Glenn W., Rutherford, Thomas F., and Tarr, David 119933, "PLcemeal Trade Reform ln Partially LLberalized EconomLes", World Bank Economic Reoiw,o v.7, No 2, May. HarrLson, Glenn W., Rutherford, Thomas F., and Wooton, Ian (19891, " ", American Economic Roview, Proceadlngs. HarrLson, Glenn W., and Vlnod, H.D., (199231 "The Sensitivity AnalysLs of Applied General EquLilbrium Models: Completely Randomized Factorial Sampllng Deslgns", The Review of Economics and Statistics, v. 74, No 2, May. Mateous, et.al. (19883; "A MultLiector Framework for Analysi of Stabilization and Structural Adjustment Pollcles - The Case of Morocco", World Bank discuasion papers, no 29. Melo, Jaime de, and Tarr, Davld, (19923; General Equilibrium Analysis of U.S. Foreign Trade Policy, CambrLdge Mass.: MIT Press. Rutherford, Thomas F., (19893, wGeneral Equilibrium Modelling With MPS/GE", Unpublished Monograph, Department of Economics, University of Western Ontario, April. Rutstr8m, E *E, [19913; The Political Economy of Protectlonism In Indonesia, Economic Research Institute, Stockholm School of Economics, Stockholm. Rutstr8m, E.E., [19923; "Friends, Enemies and Rentseeking: Who Protects Whom in Indonesia?", Working Paper 1-92-04, College of Business Administration, University of South Carolina, April. UNDP-World Bank, [19923; "Morocco 2000 - An Open and Competitive Economy", UNDP- World Bank Trade Expansion Program, Country Report 7. 27 APPENDICES TO MOROCCO'S FREE TRADE AGREEMENT WITE THE EC: A QUANTITATIVB ASSESSNENT by Thomas F. Rutherford E.E. Rutstrdm and David Tarr APPENDIX A Easddty Paramete A.l. Choice of Point-Estimates In all sectors, the elasticity of transformation (between output for domestic and export markets) i set at 5 and the transformation elasticity (between exports for the EC and non-SC markets) is met at 8. The elasticity of substitution between domestic and imported goods (the Armington elasticity) is set equal to 2 across all sectors, and the substitution elasticity between BC and non-SC imports is set at 5.' Effects due to the choice of values of these parameters are traced out in the next section. Detailed estimates of the Armington elasticities for the ORANI model of Australia are reported in DLxon, Parmenter, Sutton, and Vincent (19821, ranging from 0.34 for oil and coal products to 6.8 for footwear, however the majorLty of sectors receive a value of 2. LackLng more specific informatLon about substitutablilty Ln Morocco, we not thli elasticity to 2 for all sectors, but trace out the effects from variations in its value in the next section. The consumer's *top level" elasticity of demand for composite output of the sector is taken to be one for all sectors. Given that this value, whlch does not vary in the sensitivity analysis, is le than the Armington elasticity, imports and domestic goods wlthln a sector may be expected to be gross substltutes. The prlmary factor substitutlon elastLeLties are based on the detailed regression estimates of Harrison, Jones, Kimbell, and Wlgle (19911, and are lLsted ln Table Al as they apply here. They range from 0.43 for OIL up to 1.99 for servlces sectors (CON, COX, SRV, and ADM), but the vast majority are close 'the systeat wutvty analysis Investiga varations in de easIciaes bsed On uwIfom dlibuoa wich have the fowinU rang: between 2.5 and 735 for the asfonnon elksticit between dometc ad foreign mzout betwen 6.5 d 9.5 brdb trfonnationelaic between foreign outputs of differerA destinations; between 0.5 ad 3.S for th A DMon elsicity between impots ad dometic ouput; ad between 2.5 and 7.5 for the Annington elsicity for impons from difhren omus. TABLE Al: Capital-Labor Substitution elasticities Scctor In Morocco model Sector in H J K W Points dmait Standard deviaton CER 20 0.95 0.04 suo 20 0.95 0.04 Clr 20 0.95 0.04 VEG 20 0.95 0.04 MAD 20 0.9S 0.04 FSH 20 0.95 0.04 FOR 20 0.95 0.04 FOO 20 0.95 0.04 BEV 20 0.95 0.04 OFP 20 0.95 0.04 PHS 14 0.43 0.11 NMM 24 0.43 0.11 MIN 14 0.43 0.11 CAO 14 0.43 0.11 OIL 13 0.43 0.09 usE B 1.88 0.25 TX_ 22 0.93 0.08 CLO 23 1.19 0.03 LEA 31 0.75 0.16 WDN 24, 25 0.93 0.10 PAP 26,27 1.00 0.13 CEM 32 0.96 0.13 LAS 33 0.91 0.24 T_Q 35 1.20 0.09 EEQ 36 0.98 0.03 IND 35 1.20 0.09 MAC 35 1.20 0.09 EMI 39 1.19 0.05 CHM 28 1.01 0.03 RBR 30 0.97 0.08 OTH 39 1.19 0.05 TRD C, D 1.28 0.53 TRN B 1.88 0.25 BNK E 2.06 0.25 INS B 2.06 0.25 CON Nonitded 1.99 0.48 COM Nonraded 1.99 0. I SRV Nontided 1.99 0.48 ADM Nontaded 1.99 0.48 Source: Harrison, Jones, Kimbell, WVigb 11992M. A-3 to unity. In addition there is an elasticity of substitution between intermediate inputs and value-added in each sector. The tradition, no doubt born of Input- Output modelling habits, is to set this elasticity at zero. We do likewise, but also consider values of 0.25, 0.5, 0.75, and 1 (for each sector) in our censitivity analysis. A.2 Effects of choice of trade elasticities We trace out the effects on welfare in order to assess the model's sensitivity to the choice of value of the Armington elasticity (qA) between goods from domestic and foreign nources and (am) between EC and ROW sources. These results are presen'ed in Tables A2. The welfare effects are- wry ssnsitive to variations in the substitution elasticity between domostic and foreign sources. The welfare effects from the PTA simulation, for examples, goes from 1 to 5 percent as the elasticity increases from 1 to 10. With the exception of ACCESS, the welfare effects are at least quadrupled. The sensitivity with respect to the elasticity of substitution between EC and ROW sources is not that pronounced. The table illustrates, however, how trade diversion costs become more pronounced with the value of this elasticity. Again ACCESS provides a slight exception due to the small effects in general. In summary, the confidence intervals provided by the sensitivity analysis of the main text provide more reliable estimates of welfare and adjustment effects than a point estimate. This is especially true due to the sensitivity of the estimates to the Armington elasticity, and the uncertainty regarding an exact point estimate. A-4 TABLE A2: Welfare Effects of variations i elasdcites of subsitution in demand r FTA ECLIB LEBALL ACCESS LIBROW FrAALL A 1 1.184 0.962 1.814 0.243 0.882 2.035 oA= 2 1.518 1.288 2.369 0.265 1.099 2.598 3 1.859 1.622 2.939 0.285 1.317 3.176 A= 4 2.213 1.970 3.535 0.306 1.541 3.777 A = 5 2.587 2.338 4.164 0.326 1.775 4.412 oA= 6 2.986 2.732 4.837 0.347 2.024 5.089 cA = 7 3.419 3.160 S.557 0.368 2.292 5.814 =A - 8 3.890 3.676 6.323 0.389 2.584 6.584 oA = 9 4.406 4.137 7.127 0.411 2.905 7.393 ^A =10 4.969 4.695 7.956 0.433 3.260 8.225 44 = 1 2.005 1.774 2.369 0.265 1.652 2.598 arm 2 1.857 1.626 2.369 0.265 1.499 2.598 o4m = 3 1.723 1.493 2.369 0.265 1.351 2.598 OM = 4 1.610 1.380 | 2.369 0.265 1.216 | 2.598 aM = 5 1.518 1.288 | 2.369 0.265 1.099 | 2.598 44 = 6 1.448 1.219 2.369 0.265 1.002 2.598 om = 7 1.398 1.169 2.369 0.265 0.927 2.598 aM = 8 1.365 1.136 2.369 0.265 0.871 2.598 OM = 9 1.346 1.117 2.369 0.265 0.832 2.598 OM = 10 1.339 1.110 2.369 0.265 0.809 2.598 Table value repre the change In welfare (equivaen variation) a * peenta of 3DP. See table 2 for an explation of the policies. -A th-e A_nn oluaicity of sbsitution between do_nic out ad compoaite iwpofs. am - the elasiciy of subtton between imports fonm different sourees. Souce: Model enmate. A-5 APPlDZZ 3 adjustments to the Social Accounting Matrix - 5Thi apperndl documents the adjustments to the BAN required to benchmark the Moroccan 503 sodel, and the extraneous data incorporated. The original B1W ix explained in Matous et al. (19881. The 1980 SAM cousnsts of 39 sectors, and all are retained in the SO model. 14 labor accounts are aggregated into one. Throe capital accoua-s are aggregatod into one. One prlvate representatLve household owns both of these aggregated production factors, and any ownership by the governmoet and the enterprise sector has been transferred into tho privato household. Final demad by four "true households" aro aggregated Lnto this private representative household. gove-zatmnt demand is only for sector Administration (the public sector). The vector called "endow" shows consumption of citizens living abroad. We show thLs as exports by the bankLng sector. The tourist account (consumption by tourists in Morocco) is similarly incorporated into exports. Several input taxes are levied in production. The value-added tax is levied on both capital and labor as well as imports of all goods, the social-security tax only on labor, and the corporate tax only on capitatl. These taxes are all calculated not of any subsidies. There is also a production subsidy (not of the excise tax) on output. All these taxes show some variation in collection rates in the SAM. Lacking more detailed information regarding existing exemptions and variations in tax collections across sectors wv assume that collection rates are the uniform across sectors for these taxes, however. Therefore we are able to isolate resource allocation effects that derive from distortions due to trade restrictions (assuming that trade taxes do not offset other sectoral differentials). The income tax is only about 3% and is therefore excluded for simplicity. As the SAM is constructed based on data from 1980, before the major liberalisations were undertaken, we considered it important to establish whether the structure of the economy had changed dramatically between 1980 and 1991. A-6 TABLE B1: Comparison of value-added in 1980 and 1991 Aggregated sector 1980 SAM share 1991 share Agriculture 20.1 24.0 Mining 4.8 2.4 Coal and Oil 0.2 0.2 Refined Oil 0.2 4.6 Electricity and Water 1.7 3.4 Manufacturing 15.9 22.7 Trade 15.2 14.2 Transportation and Communication 5.7 7.8 Other Services 36.2 20.6 ,1_______________________________ 100.0 99.9 Table B1 summarizes a description of value-added shares in some aggregatCd sectors according to the 1980 SAN and offiLial value-added data for 1991. As can be seen no such dramatic structural changes have occurred, so we can retain our confidence in the results based on the 1980 SAM. We decided to use 1991 data on tariff rates rather than tariff revenue collections recorded in the 1980 SAM. The rates we applied are based on the legal rates from World Bank "Stntiaw trade files based on data provided by the Moroccan Ministry of External Commerce, adjusted so revenues match the 1991 collected tariff revenaes. This adjustment to match collections was needed due to the number of exemptions that exist in tariff collections. We encountered som problems in the sectoral mapping from the Siatia data base to the 39 sectoral level of the SAM. We therefore had to turn to auziliary data for rates on sectors CIT, FOO and MAC. For all 39 sectors we applied the simple average of the legal rates reported at the most disaggregated level before adjusting to collection rates. We also calibrated the $iport and export shares by source and destination A-7 country to 1991 according to the sintia files. Again, due to mapping difficulties, we needed to refer to auxiliary data 14!portf-ctors CIT, FOR, PRO, P00, WDN, MAC, anjdexportAuctors CUR, BUG, CAO, POO, CIV, 3ZN, 3BQ, MAC, 333, -as well as exports and imports for all services sectors are therafore disaggregated by source and destination country according to the United Natlons database on Direction of Trade (the CONTRAD3 data base) for 1990. We decided against including more detail in the model by identifying the Maghreb and Other Arab countries as separate trading partners due to their relatively small importance. The Maghreb countries account for 3 and 8% respectively in imports and exports and corresponding figures for Other Arab countries are 10 and 3%. We also decided to adjust the VAT rate to reflect the collections in 1991. We therefore set the domestic rate to 3% and the rate on imports to 11%. When the VAT rates ara adjusted to keep the government budget balanced through the counterfactural simulations the domestic and the imported rates are adjusted by the same proportion. Therefore, as imported rates are higher than domestic rates in the benchmark they will tend to change by a larger number of percentage points. This reflects the differences in exeaptions between imported and domestically produced commodities. Finallv, the corporation tax (levied on capital use) of 5 percent reflects the collected rate in 1991. A-8 AmPmEIX C nh Relationship Betweon the Blastcity of Supply and the Sector-specific Capital Ibare This appendlx numerlcally establishes the relatLonshLp between the share of sector-specific capital in the cLtrus frults (CIT) and vegetables (VKG) sectors and the industry-wide elasticity of supply La these sectors. Our numerical procedure is as follows. We start wlth a value of j, whlch is the elasticity of transformation between output destined for exports or the domestic market. This value is given in the first column of table Cl. We first confirm that the prices of the domestic variants of CIT and VUG decrease in ACCESS (where their export prices increase) for a situatlon where the entire capital st%ck in each sector is mobile. We then increase the share of the sector-specific capital stock in the resource sectors in small steps until we encounter a situation where the domestic price in citrus fruits increases The crltlcal value of the share of sector specifLc capital where the export and domestLc varieties are on the edge of the gross complesent-substLtute relatLonship, glven the elasticity of transformation and other parameters of the model, ls listed in column 2. As discussed ln the text, we may infer from the relationship li de Nelo and Tarr (19921, that at the listed value of the share of sector specific capital, (. the elasticity of transformatLon is approximately equal to , industry elastLclty of supply. With a point estimate of the export transformatLon elasticity of 5.0, we would need e. sector specific capital share of at least 37.5 percent to ensure that a gross-substltute relationshLp exists between export and domestLc varLetles of citrus fruits. We may infer from table Cl, that (given all the parameters of the model) if the share of sector specific capital in all resource sectors is 92.5 (45) percent, then the elasticity of supply in citrus fruits ls about 1 (4). A-9 TABLE Cl: The Relationship between the Supply Elasticity and the Share of Sector-specific Capital Imputed Elasticity of Supply Percentage of sector specific | . . ~~~~~~~~~~~~~capital in resource sectors 1 9. 2 70 3 55 4 45 5 37.5 6 32.5 7 27.5 8 25 9 22.5 10 20 APPENDIX D ALGEBRAIC FORMULATION OF TER MODEL The model is formulated as a system of nonlinear equations corresponding to the three classes of equilibrium conditions associated with an Arrow-Debreu general equilibriums price-cost relations for producers, supply-demand balance for commodity and factor markets (including balance of payments), and income- expenditure balance for domestic consumers and government. In SO these models are generated using the GAMS programming language and solved using the modified Newton (SLCP) algorithm due to Mathiesen (19851. in this framework a central set of variables (prices, activity levels and income levels) characterize the economic equilibrium. All important notation is summarized in Figure Al. Technology, Preferences and itarket Clearance Conditions Domestic production is an aggregate of domestic and exported varieties with A-10 .. . ,Xe .,Ctem.esiate .lnpoztof good:ki c¢:* ~. ......................... ... Valaze added otion tow v. '::''4--'a:tow" Fiur A1Lfr tefo tNotationtv.doeti cnum CapitXl inut (o seto I )/' 2 x~ terediteinp~tsof oo k n A-ll ~ This relationship can be interpreted as inplying differences in the technical processes associated with production for domestic and export markets. The elasticity of transformation defined by el will be lower for goods which are highly differentiated and higher for goods which are relatively homogeneous. The specification of this elasticity may be influenced by the intended time frame of the analysis. In the short-run it is more difficult to transform plants between domestic and export oriented products. Imports from different trading partners trade off with domestic varieties in intermediate demand, investment demand and final demand. For simplicity (and due to limitations of data) we assume that the import composition and import- domestic substitution possibilities in investment, Intermediate and final demand are identical. Under these conditions we can represent inputs as though they were composed of a single import-domestic aggregate for each commodity. The aggregation of domestic and, imported varieties is characterized by a nested constant-elasticity function of domestic and imported goods: Si a= ( 1/4 (aly 4 im, a(V,IO (3) -, *XDgM) (aDl' Di ' ' + cAow Hi') 3 where X, represents a composite import from two or more regions rt ml ( St J64 Mlr1IS The market clearance condition for domestic supply balances output from the Armington aggregation function with intermediate, investment and final demand. This condition is: Si - an2jXYj + GI + X, + *, (4) in which Y, is the activity level of sector J, a, is the input requirements of good I in sector J, and G,, I, and C, are components of final demand associated with government, investment and final consumption. Variable inputs to production include primary factors as well as intermediate inputs of commodities. These are comki.ned in a linearly homogeneous nested Leontief-CES form: A-12 Yls ~ 42 mi[r'^ t a VA 1(5 where VI (4) = ( Sk 56 f'( ) In this equation XB represents intermedlate inputs of good k in sector I, f2 ig the variable input of primary factor k in sector i, V5( represent. the value- added function for variable factors, fi represents primary factor inputs to variable cost in sector I, and fr represents the in'ut of factor k to the formation of fixed costs in sector i, due to the possibility of sector-specific capital. Domestic welfare is defined by consumption levels of market goods: Y=U (Cj,..., C.) (6) The current account is balanced at international prices (pf and ), taking into account exogenous capital flows (B): Pi X, + B u- pim', (7) The prices which appear in this equation are exogenous parameters, the international prices of imports and exports. This constraint has an associated variable which is the "real exchange rate". The model, however, contains no monetary instruments and determines only relative prices. The exogenous increase in the export price that represents access to SC markets in our model, is equivalent to an improvement in this "real exchange rate". Factor markets always clear with flexible prices: E fS + 4f Ek Income-Expenditure Balance Consumer income includes primary factor earnings plus foreign capital inflows less transfers. Final demand is modelled by budget-constrained utility A-13 maximization by a representative agent. The budget constraint is written: £ fri C - 4w5EA + a - 'rr () In this equation vk represents the market price of primary factor k, B represents the foreign exchange balance and Tr T represents the level of lumpsum transfer. Unlike private households, government demands are held constant in all simulations. The government budget constraint is accommodated through endogenous scaling of one of the three government tax instruments so that revenue balances with expenditure. Government income consists of five components: (i) lumpsum transfers from households (T), (il) import tariffs (tu), (iii) value-added taxes on factor input. to production and on imports (P,), (iv) employment and corporation taxes on factor employment (ta), (v) less production subsidies net of excise taxes (at), (vi) less export subsidies (sf,). The government budget is: ,r G1 - ?T 2 + T avi tag 4 + T, pv wt f4 + , pi, tfrmi.r (10) sip (pi Di + pfX, ) _ 8X pfX, in the government budget equation parameters which endogenously adjust to bala-e income and expenditure ares Tr for lumpsum transfers, r1 for factor taxes, and i, for value-added taxes. In any given equilibrium only one of these parameters departs from the default value of unity. PrIce-Cost Balance In Competitive Miarkets When technology exhibits constant return. to scale producers price at marginal cost. In production the marginal cost of supply for sector I (ce) is defined by: ciY -, f1jXfl + (1 + `rA) LF,fA (11) The competitive market structure with constant returns to scale technology and no barriers to entry drives excess profits to zero. Producers then equate marginal cost with market price gross of subsidy, providing the following zero profit condition: A-14 (1+48) (PI DI+ PJ XJ) +,PZ XI S - Cl Yi (12) In this equation the first term represents the value of output gross of production subsidy, and the second term captures the effect of the export subsidy. The import aggregation always equates price with marginal cost. This means that the value of domestic supply equals the cost of domestic inputs plus imports gross of tariffs and rentat Iris, . p,D, + ,(1s+rt,t,) py a4 (13) Model solver The model is solved with the MPS/GE software developed by Rutherford [19891, and generated with the GAMS software developed by Brooke, Kendrick, and Meeraus (1988J. For the purposes of this model some modifications of GAKS have been included allowing the specification of the MPS/GE format in an index format. We include an annotated version of this code: * MODEL DEFINITION IN MPS/GE VECTOR SYNTAX $MODEL:MOROCCO The sectors statement lists all the activities of the model: the utility aggregator for the consumer (U), the same for the government (G), all production activities (Y(I)), export activities (X(I)), and import activities (M(I)). $SECTORS: U G NVK Y(I) X(I)$EXPORT'I) M(I)$IMPORT(I) A(I) The commodities on the model include: a utility good for the private consumer (PU), foreign exchange that translates export revenues into purchasing power for imports (PFX), an investment good (PNVK), a government consumption good (PG), factors of production (PF(F)), Armington goods (PA(I)), domestic output goods (PD(I)), export and import goods (PE(I) and PI(I)). $CONMODITIES: PU PFX PNVK PG PF(F)$E(F) PA(I) PD(I) PE(I)$EXPORT(I) PI(I)$IMPORT(I) The auxiliary constraints formulates the equal yield constraint tax for the government. $AUXILIARY: TAU(AUX) A-15 $CONSUMERSt REPAGT GOVT $REPORT: V:DF(F,I)$DFB(F,) I:PF(F) PROD:Y(I) V:DS(I) I:PD(I) PROD:A(I) Producers demand production factors and Armington intermediate good. and produce output for domestic and export destinations. The production subsidy applies independent of destination. Either the VAT or the FACT (taxes on factor employment) can be used an an equal yield constraint. * AGGREGATE PRODUCTIONs $PRODsY(I) ttETRNDX(I) atESUBKL(I) O:PD(I) Q:DB(I) P:PYB(I) AsGOVT T:(-SP(I)) O:PE(I)$EXPORT(I) Q:XD(I) P:PYB(I) A:GOVT Ts(-SP(I)) I:PA(J) QtIOB(J,I) I:PF(F) QsDFB(F,I) PsPFB(F,I) a: + AsGOVT N:TAU( "VAT") MSVATD(I) + NsTAU("F") M:FACT(F,I) The export variety of the good can be exported to different export destinations. The aggregate export good produced above is the input to this activity. Exports generate foreign exchange, which quantity depends not only on the quantity of exports but also on the price received. The transformation effact between export varieties to different destinations caused by a change in the relative price of the varieties in captured in the P: field. The good with the higher relative export price (PX) will tend to become a relatively cheaper contributor to purchases of foreign exchange than other goods. PXB represents the international price of exports in the benchmark. * EXPORT: $PROD:X(I)$EXPORT(Z) t:ETRNXX(I) I:PE(I) Q:XD(I) O:PPX#(XR) Q:(XB(I,XR)*PX(I,XR)) + Ps(PXB(I,XR)/PX(I,XR)) AsGOVT T:(-SX(I,XR)/(l-SX(I,XR))) Imports are aggregated with a CES function in two step. * ARMINGTON IMPORT AGGREGATOR: $PROD:A(I) stESUBDM(I) O:PA(I) QsAB(I) I:PD(I) Q:DB(I) I:PI(I) Qt(AB(I)-DB(I)) $PRODtM(I)$IMPORT(I) s:ESUBMH(I) O:PI(I) Qs(AB(I)-D8(I)) I:PFX#(MR) Qs(MB(I,MR)*PM(I,MR)) Pt(PMB(I,MR)/PM(I,MR)) + A:GOVT TtT(I,MR) NtTAU(nVAT") Ms((l+T(I,MR))*VATX(I)) + A:REPAGT TsNTB(I,MR) The private consumer utility aggregator is a Cobb-Douglas function. * COBB-DOUGLAS UTILITYs $PRODsU s:1 O:PU QsUB I:PA(I) Q:CF(I) A-16 The government consumption aggregator assumes no substitutability between goods demanded * GOVERNMENT CONSUMPTIONs $PROD s G O:PG Q:(SUM(I, GB(I))) I:PA(I) Q:GB(I) Neither does the aggregation of the investment good * CAPITAL FORMATIONt SPROD:NVK OsPNVK Q:(SUM(I, la(,))) I:PA(I) Q:IB(I) Domestic consumer and government endowments (E:) and expenditures (D:). * DOMESTIC CONSUMER: $DEMAND:REPAGT E:PF(F) Q:E(F) E:PA(I) Q:EN(I) E:PNVK Q:(-SUM(I, IB(,))) E:PFX Q:BOPDEF E:PG Q:(-GOVDEF) E:PU Q:-l R:TAU("LS") E:PU Qsl R:TAU("LO") DtPU Q:UB * GOVERNMENT AGENT: $DEMAND:GOVT E:PG Q:GOVDEF E:PU Q:l R:TAU("LS") E:PU Q:-1 R:TAU("LO") D:PG * AUXILIARY CONSTRAINTS DETERMINE LEVELS OF FACTOR OR LUMPSUM * TAXATION, DEPENDING ON WHICH INSTRUMENT IS USED TO ACHIEVE * EQUAL YIELD: $CONSTRAINT:TAU("F")$ENDOG(-FACT-,SC) ZsG K:-l $CONSTRAINT:TAU("F")$(NOT ENDOG(NFACT"ISC)) Z:TAU("F") K:-l SCONSTRAINT:TAU("VAT")$ENDOG("VAT",SC) ZsG K:-l $CONSTRAINT:TAU("VAT")$(NOT ENDOG("VAT",SC)) Z:TAU("VAT") K:-1 $CONSTRAINT:TAU("LS")$ENDOG("LUMPSUM",SC) Z:G K:-1 A-i7 $CONSTRAINT:TAU("LS")$(NOT ZNDOG("LUMPSUX",SC)) K:-1 ZsTAU("LSW) $CONSTRAINT:TAU( "LOW) K:1 Ks-1 ZaG ADDITIONAL REVMUBC1S APPEIWIX D Mathleson, Lars, "Computation of Economic Equilibria by a sequence of Linear Complementarity Problems", Mathematical Prograxmnng Study 23 (^msterdams North-Holland, 1985). A-18 Pollcy Research Working Paper Series Contact Title Author Date for paper WPSI 151 Is Growth Bad for the Environment? Charles van Marrewijk July 1993 J. Verbeek Pollution, Abatement, and Federick van der Ploog 33935 Endogenous Growth Jos Verbeek WPS1152 Population, Health, and Nutrition: Denise Vaillanceurt July 1993 0. Nadora Annual Operational Review for Fiscal Stacye Brown 31091 1992 and Others WPS1 153 North American Free Trade Alberto Musalem July 1993 P. Infante Agreement: Issues on Trade In Dimitri Vhtas , 37664 Financial Services for Mexico Asli DemirgUg-Kunt WPS1 154 Options for Pension Reform in Tunisia Dimitri Vittas July 1993 P. Infante 37664 ,VPS1 155 The Regulation and Structure of Martin F. Grace July 1993 P. Infante Nonlife Insurance in the United Michael M. Barth 37664 States WPS1 156 Tropical Tlmber Trade Policies: What Panayotis N. Varangis July 1993 D. Gustafson Impact Wlll Eco-Labeling Have? Carlos A. Primo Braga 33714 Kenji Takeuchi WPS1 157 Intertemporal and Interspatial Sultan Ahmad July 1993 E. O-Reilly- Comparlsons of Income: The Meaning Campbell of Relative Prices 33707 WPS1 158 Population Growth, Externalities, Nancy Birdsall July 1993 E. Homsby and Poverty Charles Grifflin 35742 WPSI 159 Stock Market Development and Asih DemirgOg-Kunt July 1993 P. Sintim- Financial Intermediary Growth: Ross Levine Aboagye A Research Agenda 38526 WPS1160 Equity and Bond Flows to Asia and Punam Chuhan July 1993 R. Vo and Latin America: The Role of Global Stijn Claessens 31047 and Country Factors Nlandu Mamingi WPS1 161 Increasing Women's Participation in Molly Maguire Teas July 1993 L. Maningas the Primary School Teaching Force and 80380 Teacher Training in Nepal WPS1162 The Slovenian Labor Market in Mllan Vodopivec July 1993 S. Moussa Transition: Issues and Lessons Samo Hribar-Milic 39019 Learned WPS1163 Domestic Distortions and James E. Anderson July 1993 D. Gustafson Intemational Trade J. Peter Neary 33714 Policy Research Working Paper Series Contact Tttle Author Date for paper WPS1 164 Power, Distortions, Revolt, and Hans P. Blnswanger July 1993 H. Binswanger Reform In Agricultural Land Relations Klaus Deininger 31871 Gershon Feder WPS1 165 Social Costs of the Transition to Branko Milavnovic August 1993 R. Martin Capitalism: Poland, 1990-91 39026 WPS1 166 The Behavior of Russian Firms In Simon Commander August 1993 0. del Cid 1992: Evidence from a Survey Leonid Liberman 35195 Cecilia Ugaz Ruslan Yemtsov WPS1 167 Unemployment and Labor Market Simon Commander August 1993 0. del Cid Dynamics in Russia Leonid Liberman 35195 Ruslan Yemtsov WPS1 168 How Macroeconomic Projections Rashid Faruqee Au3ust 1993 N. Tannan in Policy Framework Papers for the 34581 Africa Region Compare with Outcomes WPS1 169 Costs and Benefits of Debt and Eduardo Fernandez-Arias August 1993 R. Vo Debt Service Reduction 33722 WPS1170 Job Search by Employed Workers: Avner Bar-llan August 1993 D. Ballantyne The Effects of Restrictions Anat Levy 37947 WPS1 171 Finance and Its Reform: Beyond Gerard Caprio, Jr. August 1993 P. Sintim- Laissez-Faire Lawrence H. Summers Aboagye 38526 WPS1172 Liberalizing Indian Agricuture: Garry Pursell September 1993 D. Ballantyne An Agenda for Reform Ashok Gulati 37947 WPSI 173 Morocco's Free Trade Agreement with Thomas F. Rutherford September 1993 N. Artis the European Community: E. E. Rutstrom 3e010 A Quantitative Assessment David Tarr