Trade Liberalization and Complementary Domestic Policies: A Rural-Urban General Equilibrium Analysis of Morocco

Transcription

1 TMD DISCUSSION PAPER NO. 41 Trade Liberalization and Complementary Domesti Poliies: A Rural-Urban General Equilibrium Analysis of Moroo Hans Löfgren Moataz El-Said Sherman Robinson International Food Poliy Researh Institute Trade and Maroeonomis Division International Food Poliy Researh Institute 2033 K Street, N.W. Washington, D.C , U.S.A. April 1999 TMD Disussion Papers ontain preliminary material and researh results, and are irulated prior to a full peer review in order to stimulate disussion and ritial omment. It is expeted that most Disussion Papers will eventually be published in some other form, and that their ontent may also be revised.

2 Trade Liberalization and Complementary Domesti Poliies: A Rural-Urban General Equilibrium Analysis of Moroo by Hans Löfgren Moataz El-Said and Sherman Robinson International Food Poliy Researh Institute Washington, D.C., U.S.A. May 19, 1999 Revised version of paper presented at Expert Meeting on "The Dynamis of New Regionalism in MENA: Integration, Euro-Med Partnership Agreements and After," organized by the Eonomi Researh Forum, the OECD Development Center, and the World Bank, held in Cairo, February 6-7, 1999.

3 ABSTRACT In this study, a dynamially reursive general equilibrium model of Moroo is used to examine alternative trade and domesti poliy senarios involving the implementation of the EU Assoiation Agreement for the period The model has a detailed treatment of the agriultural and rural eonomy in Moroo. The results for the trade liberalization senarios indiate that tariff unifiation has small aggregate effets whereas the removal of non-tariff barriers has strong positive aggregate effets: fator inomes and household welfare expand onsiderably more rapidly than for the base. However, trade liberalization disfavors the rural poor, espeially in rainfed areas. We simulate the introdution of omplementary domesti poliies with a non-distorting transfer program that fully ompensates the owners of rainfed resoures and skill upgrading for the rural labor fore. The results indiate that, if ombined with at least one of these omplementary domesti poliies, trade liberalization an lead to a win-win outome: the welfare of all household groups inreases signifiantly more rapidly than if status-quo poliies are followed.

4 Table of Contents INTRODUCTION...1 BACKGROUND...3 MODEL STRUCTURE AND DATA...7 Model disaggregation...7 Prodution ativities...8 Institutions...9 System onstraints...10 Database and solution approah SIMULATIONS CONCLUSION...21 REFERENCES...24 TABLES AND FIGURES...27 APPENDIX...44

5 1. INTRODUCTION Moroo is about to start implementing an Assoiation Agreement with the European Union (AAEU) at the same time as the ountry adjusts its trade poliies to onform to WTO rules. A major poliy question faing Moroo s poliy makers is whether and to what extent they should pursue additional unilateral trade liberalization. The aggregate impat of a arefully formulated program of trade liberalization is likely to be positive. However, like other major hanges in eonomi poliy, it may have very different effets on different segments of the population. Given that Moroo s agriulture urrently enjoys substantial protetion, additional broad-based trade liberalization is likely to have a detrimental impat on rural households, inluding the bulk of the poor population. This means that poliy makers may onsider the introdution of omplementary domesti poliies that ompensate those who lose from hanges in trade poliies. In the absene of suh poliies, the politial feasibility of signifiant agriultural tariff uts is questionable. In this study, a dynamially reursive general equilibrium model of Moroo is used as a laboratory for analyzing alternative poliy senarios for the period In order to fous on agriulture and issues of poverty, the model distinguishes expliitly between rural and urban ativities and households. It has a relatively detailed treatment of agriultural and other rural prodution, the labor market (disaggregated into skilled and unskilled), and households (disaggregated into four types: rural poor, rural non-poor, urban poor, urban non-poor). The basi simulation senario assumes gradual implementation of the EU partnership without additional bilateral hanges in trade poliies. The first set of simulations investigates the impat of alternative senarios for unilateral trade liberalization. In the seond set of simulations, we simulate the impat of a maximum unilateral trade liberalization senario in ombination with two alternative domesti poliy hanges, inreased rural eduation (enhaning the skills of the rural labor fore), and a non-distorting program for ash ompensation to owners of resoures used in rainfed agriulture. In addition to eonomi indiators, the analysis of the different senarios uses indiators of politial feasibility based on the distributional results. Given the likely 1

6 uneven impat on different population groups, deisions about trade liberalization are informed not only by their eonomi onsequenes, but also by the politial power of different groups. Setion 2 provides a brief bakground on the Moroan eonomy and eonomi poliy, with a fous on trade poliy, agriulture, and rural areas. In Setion 3, the CGE model and its database are presented; Setion 4 is devoted to simulations while Setion 5 summarizes the results and disusses poliy impliations. The Appendies of the paper inlude a mathematial statement of the stati module of the model as well as additional bakground data. 2

7 2. BACKGROUND 1 Table 2.1. summarizes the setoral struture of the national eonomy. In terms of the eonomy as a whole, agriulture provides somewhat less than 20% of GDP but around 45% of total employment, attesting to its relatively high labor-intensity. Rural households derives most of their inome from agriulture as 77% of rural labor is employed in agriulture. The servie setors (inluding the government administration) is the major employer in urban areas. If defined narrowly, Moroo's agriulture plays a relatively limited role in the ountry's relatively diversified foreign trade, aounting for around 8% of exports and 6% of imports; however, if agroindustrial trade is inluded, the shares are onsiderably higher. The most important agriultural exports are fish, fruits and vegetables. Wheat and sugar are the major agriultural imports (Royaume du Maro, 1997; EIU, 1997, pp ). <<Table 2.1>> Sine the early 1980s, Moroo has gradually reformed its eonomy in the diretion of trade liberalization and inreased reliane on market fores and the private setor. Moroo's maroeonomi management has sine the mid-1980s been more suessful than in most other ountries in the Middle East and North Afria aording to key indiators suh as rate and volatility of inflation, level of the budget defiit, and stability of the real exhange rate (Page and Underwood, 1997, pp ). In the trade area, the level and dispersion of tariffs have both been redued while quantitative restritions (QRs) have been eliminated (Alonso-Gamo et al., p. 24; IMF 1997, p. 7). Compared to most strutural-adjustment-oriented ountries, Moroo was suessful in ombining positive growth with rapid restoration of internal and external balane (Karshenas, pp ). Nevertheless, ompared to the 1970s, eonomi growth deelerated in the 1980s and even more so during the period In spite of far-reahing trade reforms, Moroo still has signifiant trade barriers with a high degree of dispersion of protetion rates 1 Appendix Table A.1.2 inludes additional information on strutural hange in Moroo's eonomy,

8 aross setors. Table 2.2 shows 1994 data for tariff and non-tariff barriers that are used for the model-based analysis of this paper. Non-tariff barriers are defined broadly to inlude all domesti prie deviations that annot be aounted for by import duties. As shown, the agriultural trade regime was, as of the mid-1990s, partiularly distorted, espeially for ereals and animal produts. <<Table 2.2>> For Moroo s poliymakers, rural development remains a key hallenge. Aording to data from the early 1990s, rural per-apita onsumption is around half of the urban level. While rural areas house less than 50% of the population, they aount for 70% of the poor. As shown in Table 2.3, rural areas are also strongly disfavored aording to other indiators suh as aess to eletriity and safe water, literay, and shool enrollment, with the female population standing out as partiularly disadvantaged. Low eduational ahievement is refleted in a labor fore that for the most part is "unskilled" (in the sense that most jobs require no formal eduation). The skill gap is a major soure of inequality between rural and urban areas; on average skilled workers earn 6-7 times the wage of unskilled workers (Karshenas, 1994). Relatively unfavorable rural onditions have led to rapid rural-urban migration, whih provides an important outlet for the rural labor fore (absorbing the bulk of its natural growth), but exaerbates urban unemployment and puts downward pressure on urban wages. The rural eonomy is dominated by agriulture whih represents lose to 80% of total employment and may aount for some 60% of total rural value-added. While agriultural GDP is highly variable, sine the early 1980s the setor has arrested an earlier seular deline in its share of the eonomy. The agriultural setor is haraterized by onsiderable heterogeneity, perhaps most importantly between relatively prosperous irrigated zones (17% of the ultivated area in the early 1990s) and disfavored rainfed zones that, inter alia, suffer from frequent but irregular droughts. Moreover, the rainfed areas differ greatly in terms of average annual rainfall. <<Table 2.3>> 4

9 Given the high degree of agriultural protetion, the hallenge of rural development is intertwined with trade poliy. Currently, Moroo is in the proess of implementing its Assoiation Agreement with the EU and its GATT/WTO ommitments in the Uruguay Round. The EU is the major trading partner of the ountries in the Middle East - North Afria (MENA) region, reeiving 25% of the region s total exports and providing 44% of regional imports (ERF, 1998, p. 64). For Moroo, EU export and import shares are even higher, in 1994 at 64% and 57%, respetively (Royaume du Maro, 1997, p. 572). Sine 1994, the EU has sought the onlusion of assoiation agreements with most non-eu Mediterranean ountries. The agreements inlude the establishment of a Free Trade Area (FTA) overing the EU and eah non- EU partner. In line with World Trade Organization (WTO) rules, the FTAs are to be implemented 2 gradually over transitional periods lasting a maximum of twelve years. In 1996, Moroo signed suh an agreement with the EU. For industrial imports from the EU, Moroo is ommitted to a gradual elimination of tariff rates, and the abolishment of any quantitative restritions, taxes, and other measures that have the same effet as tariffs. In return, Moroo will reeive aid for eduation and infrastruture projets over a period of five years (Oneworld, 1995). With few exeptions, Moroo s non-agriultural exports will ontinue to enjoy unrestrained, duty-free aess to the EU. In spite of a slightly improved aess, Moroo s agriultural exports to the EU remain stritly regulated, with limited sope for expansion. As a member of the WTO, Moroo is ommitted to respeting the rules of GATT/WTO. On the basis of the agreement of the Uruguay Round, Moroo is replaing non-tariff measures with ordinary tariffs (tariffiation). For all agriultural ommodities, Moroo is ommitted to maximum tariffs defined by the bounds it submitted, and the 24% redution in those bounds in equal annual derements over a period of ten years. The tariff bounds are of no eonomi 2 Aording to Artiles I and XXIV of the GATT, the establishment of Free Trade Areas and Customs Unions are permitted as long as they are based on the Most Favored Nation (MFN) priniple and meet the following four onditions: a. all trade is overed by the agreement; b. tariff rates imposed on imports from a third ountry after forming the FTA are not on the whole higher or more restritive than those before the FTA formation;. the transition period to free trade does not exeed 12 years; and d. the ontrating parties are promptly notified of the arrangement itself, as well as modifiations or enlargement to it (ERF, 1998, p. 65). 5

10 signifiane sine no hange is required as the submitted bounds are well above the atual applied rates, whih is typial of developing ountries. In pratie, for a signifiant period of time neither the GATT/WTO or the AAEU will oblige Moroo to redue signifiantly the high agriultural protetion that is a major soure of ineffiient resoure alloation. Currently, trade and rural development are two major and interrelated issues in Moroo s eonomi poliy debate. In the trade area, the debate revolves around the impat of the implementation of the Assoiation Agreement with the EU and of Moroo s GATT/WTO ommitments in the Uruguay Round. Another major onern is whether Moroo should pursue further general import liberalization, overing agriultural imports irrespetive of soure and industrial imports from regions beyond the EU. Broad unilateral import liberalization may have a positive impat on aggregate eonomi performane. However, given the high degree of agriultural protetion, the impat of broad-based trade liberalization on rural welfare is a partiular onern. Liberalization may also give rise to transitional osts as labor is realloated between different setors. Moroan poliy makers are well aware of the link between rural welfare and agriultural rop pries in Marh 1998, in the very first deree he signed, Moroo's new prime minister Youssoufi imposed a sharp inrease in tariff rates on imported wheat to ounterat a reent drasti fall in world pries (EIU, 1998, p. 20). In fat, it may be more appropriate to onsider agriultural trade liberalization in the ontext of omplementary poliies that ompensate vulnerable rural households. As an example of poliies that an be pursued in the short run, Mexio introdued an inome transfer program (PROCAMPO) where farmers were ompensated for redued protetion of agriultural markets. By making payments proportional to assessments of past earnings in agriulture, the program aimed at being non-distortionary in terms of urrent prodution deisions (World Bank, 1997). In Moroo, baseline payment levels ould be established on the basis of data from the 1997 agriultural ensus. It should be easier to administer than diret payments based on the urrent ropping pattern. Over a longer time horizon, options inlude support for an eduational system that enhanes the skills of the rural labor fore and the development of infrastruture that failitates the development of rural non-agriultural ativities. 6

11 In this paper, we use a rural-urban CGE model of Moroo to explore the impat of different senarios for trade reforms and omplementary poliies on the rural eonomy, the labor market, and the rural poor. For eah senario, we will also assess politial feasibility, using various aggregate indiators. 3. MODEL STRUCTURE AND DATA 3 We will here present the rural-urban CGE model and its data soures. The urrent model, whih draws on existing eonomywide models of Moroo, is distinguished by an expliit separation of ativities and households into rural and urban. The disaggregation aims at identifying the rural poor, as well as the fators and ativities from whih they earn their inomes. Hene, the model has a detailed treatment of aspets that are most losely linked to the rural eonomy and the welfare of the rural poor, inluding agriultural and other rural ativities, and rural fators of prodution. Although the treatment of urban prodution is more aggregated, the model also permits an analysis of the impat on the urban poor of poliies and exogenous shoks. Moreover, the resulting eonomywide perspetive permits us to avoid the fallay of viewing the rural eonomy as an isolated island. This is important sine the rural and urban eonomies and the welfare of their households are interdependent, with numerous linkages in the markets for ommodities and fators. Model disaggregation Table 3.1. displays the disaggregation of ativities, fators, and institutions. Among the 45 ativities, 38 are rural and 7 urban. Most rural setors are part of rop or livestok agriulture. The non-agriultural setors of the eonomy (disaggregated into the major types of industrial and servie setors) are lassified as rural or urban. 3 This presentation uses no mathematis. Table A.1.1 inludes a mathematial model statement. 7

12 <<Table 3.1>> All ativities use apital and labor. Agriultural ativities demand additional fators: livestok makes use of pasture-fallow land; rop ativities rely on rainfed land; irrigated rop ativities also use water. Outside agriulture, the labor fore of eah ativity inludes both skilled and unskilled labor whereas for all agriultural ativities, exept fishing and forestry, the labor fore is made up of a separate ategory of (unskilled) agriultural labor. In rop and livestok agriulture, most ativities produe multiple ommodities and most ommodities are produed by two ativities, one in rainfed and one in irrigated areas. Fodder byproduts are produed by most rop ativities. Livestok ativities produe meat and milk (disaggregated by animal type) and, for the ow ativities, manure. Multiple-output ativities produe their ommodities in fixed physial proportions. Outside rop and livestok agriulture, eah ativity produes only one ommodity. Given that servie ommodities tend to have loation-speifi harateristis, rural and urban servie ativities are viewed as produing distint ommodities. For industrial and agriultural ommodities, markets are treated as integrated aross regions (irrigated and rainfed agriultural zones or rural and urban regions) and with international trade. The model inludes four household types, disaggregated by region (rural and urban) and inome level (poor and non-poor). The other institutions onsist of the government and the rest of the world, divided into the European Union (EU) and non-eu in the area of goods trade. The rest of the world is thus disaggregated given that one purpose of the analysis is to understand the impat on rural development from Moroo s partnership agreement with the EU. Prodution ativities Produers are assumed to maximize profits given their tehnology and the pries of inputs and outputs. As shown in Figure 3.1, the tehnology of the prodution ativities is speified as a Leontief funtion of aggregate value-added and an aggregate intermediate input. Value-added is produed by a CES funtion of primary fators, and a Leontief funtion of intermediate input use. 8

13 In order to permit tehnique hange in response to signifiant prie hanges for inputs, the intermediate oeffiients are flexible inside agriulture but fixed for other setors. For irrigated rop agriulture, an aggregate land-water fator is among the arguments in the CES funtion. This aggregate fator is produed by a set of alternative fator-aggregation ativities based on Leontief tehnology that speifies substitution possibilities between the land and water along a linearized CES isoquant. This Leontief representation is preferred to a ontinuous CES funtion to allow for the possibility of water or land being in exess supply, with a orresponding prie of zero for the non-sare fator. The inome of eah fator is alloated to domesti institutions (households and government) in fixed shares, after adjustments for fator payments to and from the rest of the world (both of whih are fixed in foreign urreny). <<Figure 3.1>> Institutions In the base year, both rural and urban households reeive the bulk of their inomes from fator earnings in their respetive regions. Compared to the non-poor, the poor in both regions depend more heavily on labor inomes in general and unskilled labor inomes in partiular (See Table A.1.3 for 1994 inome shares derived from the SAM.). In addition to fator inome, households reeive transfers from the government (the transfer reeived by eah household is a fixed GDP share) and the rest of the world (fixed in foreign urreny). Total household inome is used to pay diret taxes, save and onsume. Diret taxes and savings are fixed shares of household inome. Consumption demand is determined by the linear expenditure system (LES). Besides fator inomes, government revenue onsists of taxes diret taxes from households, indiret taxes from domesti ativities, domesti sales taxes, and import tariffs (with different rates applying to EU and non-eu goods' imports). All taxes are ad valorem. Apart from the above-mentioned transfers to households, the government uses its inome to buy a fixed quantity of onsumption goods, transfers to the rest of the world (fixed in foreign urreny), and 9

14 onsumer subsidies (a fixed share of the onsumption value for manufatured goods, representing food items). The rest of the world interats with Moroo through ommodity trade and the abovementioned transfers (whih add to or dedut from the inomes of fators and domesti institutions). System onstraints System onstraints, or losure rules are those onstraints that have to be satisfied by the eonomi system, but whih are not onsidered in the deisions of any miro agent (Robinson 1989, pp ). They onsist of the markets for ommodities and fators as well as a set of maro aggregates. Commodity markets Commodities are supplied by domesti prodution ativities and by imports. On the other side of the market, we find domesti demand and exports. Imperfet substitutability is assumed for ommodities from different soures (different domesti ativities, different import regions, or the outside world versus domesti produers). Commodities delivered to different destinations (domesti market vs. aggregated export market or different export markets) are imperfetly transformable. Figure 3.2. summarizes the ommodity flows that underlie the market for a ommodity that is produed by two ativities and is traded in both diretions, both with the EU and the rest of the outside world. A separate prie is assoiated with eah ommodity flow (box). <<Figure 3.2>> In the bottom left, prodution from the two ativities ombine to form aggregate output that, in turn, is transformed to domesti sales and aggregate exports. In the next stage, the latter 10

15 are further transformed into exports to the EU and the rest of the world. On the domesti supply side, imports from the EU and the rest of the world generate aggregate imports that, together with domesti sales, are aggregated to give the domesti omposite ommodity supply. On the other side of the omposite ommodity market, demand is made up of household and government onsumption, investment, and intermediate input use. The above Figure is simplified for ommodities that enter international trade in a less omplete fashion (or not at all for non-traded ommodities) and/or are supplied by a single domesti ativity. Moreover, for imported servie ommodities, the first step in the aggregation is eliminated sine imports are not disaggregated by soure. The funtional forms for transformation and aggregation are, respetively, Constant- Elastiity-of-Substitution (CES) and Constant-Elastiity-of-Transformation (CET) funtions. At eah stage, the shares of ommodities from different soures or to different destinations are sensitive to relative pries. These assumptions embodied in these funtions imperfet substitutability and transformability grant the domesti prie system a ertain degree of independene from international pries and dampen responses of imports, exports and domesti sales to prie hanges. Eah box an be viewed as representing a market that, although linked to other markets, has a separate market-learing mehanism. With the partial exeption of export and import markets, eah prie performs the role of learing the market the quantities supplied and demanded are, respetively, positively and inversely related to the prie. For imports, the supply side lears the market: it is assumed that Moroo is a small-ountry faing infinitely elasti supplies at exogenous world pries. For most exports, it is similarly assumed that Moroo is a small ountry faing infinitely elasti demands at an exogenous world prie: in this setting, the demand side lears the market. The only exeption is for agriultural exports to the EU. A dual-regime formulation is used aording to whih an inrease in Moroo s supply prie will give rise to redued exports along a onstant-elastiity demand urve. However, a derease in the Moroan prie will not give rise to a orresponding inrease in demand. The EU will purhase the base-year quantity at the 11

16 (lower) prie, in the proess apturing the rent produed by the onstraint. As a result, the EU pays exatly the prie needed to indue Moroo to export the fixed quantities. Fator markets Given the medium- to long-run perspetive of the urrent analysis, the dynami model version assumes that eah fator is mobile aross the ativities that use it. A market-learing prie generates demand-supply balane in the ontext of full resoure utilization. The only exeption applies to land and water in irrigated agriulture where the model allows for the fat that flexibility in tehnique hoie may not be suffiient to assure that both fators always are fully utilized. Hene, for eah fator, two regimes are possible: full employment with a market-learing prie or unemployment with the utilization level as the learing variable. Given that the setoral prodution funtion always demands the land-water aggregate, at most one of the two fators is unemployed at any given point in time. Maro onstraints These onstraints determine the manner in whih balane is ahieved for the maro aggregates assoiated with the aounts for the government, the rest of the world, and savingsinvestment. Government savings the differene between the government s urrent revenues and urrent spending is a fixed share of GDP. Proportional adjustments in the rate of valueadded tax (uniform aross all setors) assure that the government savings target is met. Foreign savings are fixed. A flexible real exhange rate (measuring the ratio between pries of traded ommodities and domesti outputs sold domestially) lears the balane of the rest of the world. As noted earlier, for eah household ategory, savings is a fixed share of its disposable inome. Hene, none of the three types of savings government, household, and foreign is free to equilibrate aggregate savings-investment balane. Hene, the model has a savings-driven determination of investment: aggregate investment (for gross fixed apital formation) varies endogenously to ahieve savings-investment equilibrium. 12

17 The Dynami Module The within-period, stati model is solved for 1994 (the base year for the database) and 1998 (to update the model to the base year for the model-based analysis), and every two years thereafter until Between the stati-model solutions, seleted parameters are updated in the dynami (between-period) module, either using lagged endogenous variables (from solutions in previous periods) or exogenous trends. The aggregate apital stok is updated endogenously on the basis of previous investment and depreiation, interpolating for the inter-period years. Total population, supplies of skilled and unskilled labor, foreign savings, institutional payments to and from the rest of the world, and total fator produtivity by ativity are all updated exogenously. Database and solution approah The model data is based on a disaggregated SAM (a 108x108 matrix) for 1994, to whih the model parameters are alibrated. The SAM was onstruted on the basis of data from various data soures, most importantly: (i) disaggregated agriultural information from the Moroan 4 government, the World Bank, and the FAO, primarily for 1990/91; (ii) a disaggregated eonomywide framework represented by Soial Aounting Matries (SAMs) for 1990 and 1994, an input-output table for 1990, as well as data on the 1994 poliy regime taxes, subsidies, and non-tariff barriers (Bussolo and Roland-Holst, 1993; Roland-Holst, 1996a); (iii) 1994 maro and trade data from Royaume du Maro (1997), the RMSM data base (World Bank, 1997a), and United Nations (1998); and (iv) disaggregated population, onsumption, and labor fore data from Royaume du Maro (1993, 1995, 1996, 1997), World Bank (1994, 1995, 1997a, 1997b), International Monetary Fund (1997), and Karshenas (1994). It should be emphasized that in areas where detailed information was laking (for example regarding wage gaps aross different 4 The Moroan government soures inlude MAMVA (DPAE/Division des Statistiques and DPV, AGER, DPA, and ORMVA), Ministère des Initations à l Eonomie (Diretion de la Statistique), Ministère des Finanes, Ministère de l Industrie, Ministère des Travaux Publis, and Caisse de Compensation. 13

18 ativities), some simplifying assumptions had to be imposed. In doing so, we were guided by the underlying premise of the analysis: the impat of trade poliy on the rural eonomy annot be properly assessed without a model struture that aptures the salient harateristis that are related to the urban-rural divide, inluding large skill and wage gaps, and differenes in setoral struture. Available information was brought together in one matrix, the disaggregation of whih parallels the disaggregation of the urrent model. Underlying the onstrution of suh a SAM is an attempt to make the best possible use of available sattered data. Inevitably imbalanes appear when data from different soures and years are integrated in one framework; a ross-entropy method was used to generate a balaned model SAM that uses all the information ontained in the original data set (Thissen and Löfgren, 1998; Robinson et al, 1998). A maro version of the model SAM idential to the disaggregated SAM exept for the aggregated depition of fators, household, ativities, and ommodities is shown in Table 3.2. A variety of other studies of Moroo were onsulted for estimates of elastiities for the Armington, CET, CES (prodution), LES (household onsumption), and export-demand funtions. 5 <<Table 3.2>> The urrent model is solved as a mixed-omplementarity problem (MCP), onsisting of a set of simultaneous equations that are a mix of strit equalities and inequalities but without an objetive funtion. This approah, made feasible by the reent development of solvers, makes it 5 The onsulted studies inlude Aloui et al,. 1989; de Janvry et al., 1992; Goldin and Roland- Holst, 1995; Laraki, 1989, Mateus et al.,1988; Morrisson, 1991; and Rutherford et al., In summary, the values used are: 1. Elastiity of substitution for CES value-added funtions: 0.8 for all ativities exept Publi Administration (0.19); 2. Elastiity of substitution for CES intermediate-input aggregation funtions for agriultural ativities: 0.5 for all ativities exept vegetables (2.0); 3. CES (Armington) funtion elastiities for aggregation of imports from different regions and of imports and domesti output: between 2 and 7 for all ommodities with the higher values for grains; 4. CET funtion elastiities for transformation of domesti output to aggregate exports and domesti sales and of aggregate exports to exports disaggregated by region: between 2 and 5 for all ommodities; 5. Elastiities for onstant-elastiity export demand funtions for agriultural exports to the EU and for servie exports: Household expenditure elastiities were omputed on the basis of Royaume du Maro (1993). 14

19 possible to formulate a model that ombines desired features of mathematial programming models (in partiular by permitting exess supplies of agriultural resoures, suh as water) while allowing the full range of assumptions for onsumer demand, government poliies, and foreign trade that appear in standard CGE models. The GAMS modeling software is used both to generate the database and to implement the model. The model is solved with PATH, a solver for mixed omplementarity problems SIMULATIONS The simulations, based on the CGE model presented in the preeding setion, explore the impat of alternative senarios for trade liberalization and the potential role of omplementary domesti poliy hanges. The first set of simulations are defined in Table 4.1. <<Table 4.1>> The results are summarized in Table 4.2. The simulation AAEU defines the status quo in the sense that poliy hanges are limited to what Moroo unambiguously made a ommitment to implement in assoiation with the EU. On the aggregate level, real GDP at fator ost grows at an annual rate of 3.7%, a rate that hanges very little aross the simulations that are reported in this 7 paper. Growth is biased in favor of urban prodution and non-agriultural setors, in part beause these do not depend on natural resoures, the physial quantity of whih annot be expanded easily. (In the model simulations, the quantities of agriultural resoures land, water, 6 For GAMS, see Brooke et al. (1988). Rutherford (1995) provides more information on PATH. 7 The simulated aggregate growth rate is in line with the expetations of the World Bank, both for Moroo and the region at large (al-hayat, Deember 16, 1998; RMSM database). Growth in aggregate real GDP at fator ost (an index of real prodution) varies little aross the different simulations sine supplies are exogenous for all fators exept apital (for whih supply growth is endogenous but quite similar aross the different simulations given similar levels of real investment while utilization rates are permitted to vary only for irrigated resoures. 15

20 and pasture-fallow areas are fixed at the 1994 level.) Changes in the alloation of labor between agriulture, rural non-agriulture, and urban ativities are minor. <<Table 4.2>> On average, real fator inomes grow at a similar pae to GDP, with the most rapid growth for agriultural resoures, both irrigated and rainfed. Household welfare grows at 2-3% 8 per year, both on the aggregate level and for the different household groups. Welfare growth is slightly biased in favor of the poor and rural areas, a refletion of the fat that growth in fator inomes is faster for agriultural resoures and unskilled labor than for skilled labor and apital. 9 Rapid growth in the rents of agriultural resoures, inluding pastures, may lead to overexploitation and environmental degradation. The result that resoure inomes in irrigated agriulture grow more rapidly than for other fators points to the growing importane of effiient resoure management in this area. Imports and exports grow faster than GDP and inomes; the eonomy is gradually beoming more open. The agriultural trade defiit grows as domesti prodution, hampered by the limited resoure supply, is unable to keep up with growing domesti demand. For industry, export growth is more rapid than import growth; imports from EU grow rapidly while imports from non-eu ountries deline, i.e., the AAEU leads to trade diversion toward the partner ountries in the free-trading area. Compared to 1998, the real exhange rate appreiates. To 8 The welfare index is derived from the ompensating variation (defined as the amount of money whih, if taken away from the household after a prie and/or inome hange, would leave it just as well of as before the hange; i.e., what the household would be willing to pay for enjoying the hange). More speifially, the index was defined as the ratio between the simulated value of household onsumption and the onsumption value that would have left household welfare at the 1994 level (simulated onsumption value minus ompensating variation). In Tables 4.2 and 4.4, the household values in all olumns exept "1998" in Table 4.2 show the perentage annual growth rate in the welfare index between 1998 and The 1998 olumn shows per-apita onsumption in 1998 (at 1994 pries). 9 Note that the household are lassified on the basis of their 1994 harateristis (inluding loation, inome level, and patterns of asset holdings). Labor that migrates between rural and urban employment does not hange its household affiliation. 16

21 maintain savings at a predetermined share of GDP in the fae of redued tariff rates and other trends, the government ollets value-added taxes at a level equivalent to 4.1% of GDP. The other three simulations in the first set assume that Moroo implements poliies that redue the extent of prie distortions aused by trade poliies. For the simulation TARIFF, unifiation of all ommodity tariffs exept industrial imports from the EU at the 1994 average rate redues protetion signifiantly for agriulture but has less impat on the industrial protetion rates faing suppliers from outside the EU. Aordingly, ompared to the AAEU senario, agriultural imports expand (most strongly for ommodities from the EU, inluding beef) while growth in agriultural prodution and resoure inomes slows. Growth deelerates for rural nonagriulture whih, more than the urban setor, is driven by demand from agriultural prodution and rural households. The fat that the rural eonomy primarily relies on unskilled labor redues inome growth for this ategory. Growth in the urban eonomy aelerates slightly, generating higher inomes for apital and skilled labor as well as on the national aggregate fator level. Aggregate household welfare is enhaned, due to a signifiant improvement for urban non-poor households with small delines for other ategories, an outome that stems from the pattern of hange in fator inomes. In general, the reperussions of tariff unifiation, inluding effiieny gains, are quite limited. This is not surprising sine tariff unifiation takes plae in a setting where numerous other distortions are in plae. In the simulation TARIFF+NTB, non-tariff barriers, primarily affeting agriulture, are gradually removed. The impat is muh stronger than for tariff-unifiation alone and eonomi openness is enhaned signifiantly. The driving fore is redued pries for demanders (both onsumers and produers) of agriultural ommodities. Fator inomes grow onsiderably more strongly for apital and skilled labor. In agriulture, resoure inome growth delines, espeially in the rainfed subsetor. A growing part of the labor fore migrates away from agriulture to the urban and rural non-agriultural ativities. All households groups gain exept the rural poor who are unaffeted. Households that rely more heavily on inomes from rainfed resoures than the representative rural households in the model would be likely to see their situation deteriorate under this senario. Rapid agriultural import expansion raises the trade defiit, as a result of whih the real exhange rate depreiates, 17

22 further enhaning exports and dampening imports. Import growth engenders higher tariff revenues, reduing revenue from the value-added replaement tax to 2.2% of GDP. Unifiation of tariffs at 10% (industrial imports from the EU exluded; the simulation TRADE-LIB), a signifiant ut, generates further exhange rate depreiation, inreased openness, redued tariff revenues, inreased government reliane on the value-added tax, and redued growth in agriultural resoure inomes. Other effets are relatively minor. On the aggregate level, fator inomes and household welfare stay virtually unhanged. There are minor uts in agriultural prodution and rural household welfare. The negative impat on agriulture and rural well-being is driven by the fat that, for agriulture, redued protetion is transmitted more strongly into lower domesti pries as imports are relatively highly substitutable with domesti output. 10 Figures ompare the evolution over time of welfare for eah household group under the last senario (TRADE-LIB), whih inorporates major trade liberalization, to the AAEU senario. The rural poor lose while all other households groups are better off under liberalization in all periods. Aording to the ompensating variation measure, by the year 2012, aggregate gains under TRADE-LIB exeed those of AAEU by more than 5% of the GDP of the AAEU simulation in the same year. Hene, there is sope for having the winners ompensate the losers in a way that assures that both groups are better off than for the AAEU run. Figures show the evolution of disaggregated fator inomes (indexed to 100 for 1998) for the same two simulations. The figures show that, at this level, the gains are highly unevenly distributed. Households that do not have signifiant non-agriultural inomes would tend to lose from the reforms during the time frame onsidered. However, as a longer time period passes, the households are more able to develop strategies where they shift away from reliane on agriultural inome. In the seond set of simulations, we juxtapose the poliy hanges under TRADE-LIB with omplementary measures that aim at ompensating rural, vulnerable losers. These ations may be 10 In an additional simulation, TRADE-LIB was rerun but without the gradual elimination of tariffs on industrial imports from the EU. The results were very similar in most respets, inluding welfare effets. The only major differene was that, as expeted, the new simulation showed higher tariff revenues and less reliane on the value-added tax. 18

23 onsidered worthwhile in their own right (sine they aim at improving the welfare of rural poor) and may also serve to mitigate politial resistane to trade reforms. Table 4.3 desribes the ontent of the two simulations. For the first simulation (whih aims at imitating Mexio s PROCAMPO program, f. disussion in Setion 2), part of the program ost overs administration: out of every Dirham spent on the program, 70% is transferred to farmers while 11 30% is spent on administration. The skill enhanement program is motivated by the fat that, as noted in Setion 2, the skill gap between urban and rural areas is a major soure of rural-urban inequality. It assumed that the program an be ahieved without eonomi loss by realloating eduational expenditures, for example from urban-foused higher eduation that produes graduates laking skills in demand in the labor market. It should be noted that neither one of the programs is narrowly targeted to poor households: they respetively benefit all owners of rainfed resoures and all rural households with unskilled labor. <<Table 4.3>> Seleted results for the seond simulation set are shown in Table 4.4. Figures summarize the impat on household welfare. To failitate omparisons, data for AAEU and TRADE-LIB are repeated. Compared to TRADE-LIB, the TRANSFER simulation generates gains for rural households, espeially the poor, while the urban households lose. This result reflets the fat that urban households own little rainfed resoures while they, like the rural households, suffer from delining fator inomes beause of the value-added tax, whih inreases in rough proportion to the value of the transfer in 2012 it is lose to 3.8% of GDP. Figure 4.14 shows the evolution of the transfer over time. It inreases gradually during the implementation of the redution in border protetion. After reahing a peak in 2004, it starts a steady but slow deline. In other respets, the transfer has a limited impat. It primarily funtions as a devie for inome redistribution. As opposed to the TRADE-LIB senario, the rural poor are 11 This harge is highly approximate it is not lear what the ost of suh a program would be in Moroo, inter alia sine it depends on the apaity of the existing administration to manage an additional program. International experiene suggests that 30% is a plausible figure for the administrative ost share in well-managed publi works programs (World Bank 1997, p. 53). 19

24 now better off ompared to the base senario. However, relative to GDP, the total ost of the transfer program is substantial, suggesting the need to target suh programs more narrowly, perhaps to rainfed regions with little rainfall. <<Table 4.4>> In the simulation SKILL-UPGRADE, we explore the impat of raising the skill level of part of the unskilled rural labor fore. In every year starting from 1999, the skilled rural labor fore is augmented by 5%, boosting its annual growth rate for the period from 3.8% to 7.7%. The rural unskilled labor fore is redued so as to leave unhanged the total labor fore, utting its growth rate from 2.4% to 2.0%. Compared to the TRADE-LIB simulation, GDP growth aelerates signifiantly for rural non-agriultural and urban ativities, but not in agriulture sine this setor only uses unskilled labor. Inomes go up for all fators exept skilled labor, an indiation that, in the fae of supply shifts, demand for this labor type is inelasti. Welfare is boosted strongly for all households exept the urban non-poor, who initially depend more heavily than others on skilled labor inomes. They see their wages deline without any hange in their endowment of skilled labor. Figures 4.15 and 4.16 show the results for two indiators of politial support for the senarios TRADE-LIB, TRANSFER, and SKILL-UPGRADE ompared to AAEU. Aording to both indiators, total politial support is a weighted average of the degree of approval or disapproval of eah household group. For eah non-base simulation, the degree of politial approval or disapproval of eah household group is measured, in eah time period, by the perentage deviation of its disposable inome from inome under AAEU in the same period. This assumes that the population has a oneption of what their onditions would have been like under an alternative, status-quo senario. Under dirham power," the weight of eah household (indiating its politial influene) is defined as its share, during eah time period, in total household inome. Under people power," the household weights are population shares (Dervis et 20

25 12 al., pp ; de Janvry and Sadoulet, 1995, p. 24). Aording to both measures, all non- AAEU senarios are supported (sine the support measures invariably have positive values) further liberalization would be supported ompared to the status quo. Support is growing signifiantly as long as the poliy hanges are implemented. SKILL-UPGRADE is most strongly supported aording two both measures sine it on average leads to the strongest inome improvement. However, given its pro-poor harater (the benefiiaries have larger population shares than inome shares), it sores higher for the people-power measure. Dirham power supports TRADE-LIB more strongly than TRANSFER while people-power support for these two senarios is virtually idential, a refletion of the relative pro-poor harater of the agriultural transfer program. These summary measures provide a simplisti view of the determinants of politial support. Among other things, it is possible that household attitudes should be measured at a more disaggregated level where the impat is onsiderably less even and most likely negative for some groups (f. the evolution of fator inomes shown in Figures ). Nevertheless, aording to these measures, the pro-poor trade-domesti poliy pakages ould have broad support, in partiular if votes in eletions play a strong role in the determination of eonomi poliy. 5. CONCLUSION In this paper, we use a rural-urban CGE model of Moroo to simulate the impat of alternative senarios for trade and domesti poliies. In the base senario, the AAEU is implemented without other poliy hanges. For the period , real GDP at fator ost grows at an annual rate slightly below 4%. Rural poor and urban poor households enjoy the most rapid welfare inreases, a refletion of the fat that the pattern of fator inome growth is 12 In 1998, the people-power shares of the different households (in %) are as follows: urban poor 3.6, urban non-poor 47.5, rural poor 8.8, and rural non-poor In the same year, the dirhampower shares are urban poor 1.1, urban non-poor 66.2, rural poor 2.6, and rural non-poor The shares do not hange signifiantly over time. 21

26 pro-poor: agriultural resoure inomes grow most rapidly followed by unskilled labor, with lower growth rates for skilled labor and apital. The results for the trade poliy simulations indiate that, in a world where poliy alternatives are seond-best, tariff unifiation has a relatively limited impat on aggregate fator inomes and household welfare. However, removal of non-tariff barriers (expressed in tariff-equivalent form) has strong positive aggregate effets. Lowering of tariffs and removal of non-tariff barriers lead to depreiation and major expansions in non-agriultural exports and agriultural imports. Growth aelerates for non-agriultural setors but slows down in agriulture. Resoures (labor and apital) move from agriulture to other parts of the eonomy. Aggregate fator inomes and household welfare expand onsiderably more rapidly than for the base. However, trade liberalization redues inome growth for agriultural resoures, espeially in rainfed areas. The owners of these resoures tend to be a relatively poor part of the rural population. On the household level, the trade liberalization senarios disfavor the rural poor, who represent 70% of all poor in Moroo. Two domesti poliy senarios aim at addressing the relatively negative impat of trade liberalization on the owners of rainfed resoures and the rural poor. In one of the senarios, we introdue a non-distorting transfer program that fully ompensates the owners of rainfed resoures for the losses they inur from trade liberalization ompared to the base senario. On the household level, the result is a pro-rural development pattern, with poor and non-poor rural households registering the strongest welfare improvements. Also the urban households are signifiantly better off than under the base senario. However, the tax burden on the government is quite heavy (lose to 3.8% in 2012), suggesting the need for targeting, perhaps by providing transfers to rainfed farmers in low rainfall zones. In the seond domesti poliy senario, we upgrade the skills of the rural labor fore, approximately doubling the rate of growth for rural skilled labor (from a low base) and reduing growth for rural unskilled labor. This leads to a signifiant growth expansion for GDP (driven by non-agriultural expansion), aggregate fator inomes, and aggregate household welfare. In terms of household welfare, the outome is pro-rural and pro-poor: the two rural households reord the 22