External Shocks and Policy Alternatives in Small Open Economies

Transcription

1 IFPRI Disussion Pper November 2011 Externl Shoks nd Poliy Alterntives in Smll Open Eonomies The Cse of El Slvdor Smuel Morley Vleri Piñeiro Shermn Robinson Mrkets, Trde nd Institutions Division

2 INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE The Interntionl Food Poliy Reserh Institute (IFPRI) ws estblished in IFPRI is one of 15 griulturl reserh enters tht reeive prinipl funding from governments, privte foundtions, nd interntionl nd regionl orgniztions, most of whih re members of the Consulttive Group on Interntionl Agriulturl Reserh (CGIAR). PARTNERS AND CONTRIBUTORS IFPRI grtefully knowledges the generous unrestrited funding from Austrli, Cnd, Chin, Denmrk, Finlnd, Frne, Germny, Indi, Irelnd, Itly, Jpn, the Netherlnds, Norwy, the Philippines, South Afri, Sweden, Switzerlnd, the United Kingdom, the United Sttes, nd the World Bnk. AUTHORS Smuel Morley, Interntionl Food Poliy Reserh Institute Visiting Senior Reserh Fellow, Mrkets, Trde nd Institutions Division Vleri Piñeiro, Interntionl Food Poliy Reserh Institute Collbortor, Mrkets, Trde nd Institutions Division Shermn Robinson, Interntionl Food Poliy Reserh Institute Senior Reserh Fellow, Environment nd Prodution Tehnology Division Noties IFPRI Disussion Ppers ontin preliminry mteril nd reserh results. They hve been peer reviewed, but hve not been subjet to forml externl review vi IFPRI s Publitions Review Committee. They re irulted in order to stimulte disussion nd ritil omment; ny opinions expressed re those of the uthor(s) nd do not neessrily reflet the poliies or opinions of IFPRI. Copyright 2011 Interntionl Food Poliy Reserh Institute. All rights reserved. Setions of this mteril my be reprodued for personl nd not-for-profit use without the express written permission of but with knowledgment to IFPRI. To reprodue the mteril ontined herein for profit or ommeril use requires express written permission. To obtin permission, ontt the Communitions Division t ifpri-opyright@gir.org.

3 Contents Abstrt v Aknowledgments vi 1. Introdution 1 2. The Updted Soil Aounting Mtrix (SAM) 2 3. The Computble Generl Equilibrium (CGE) Model 5 4. Blne of Pyments Shoks 8 5. Dynmis: Simulting the Effet of Inresing Produtivity Conlusions 17 Appendix: Supplementry Tbles 18 Referenes 32 iii

4 List of Tbles 2.2 Imports nd totl supply Shre of exports nd imports by setor Mro results for three negtive blne of pyments shoks Setorl growth rtes of vlue dded Unemployment nd the rel wge Mro impt under three lterntive lbor mrket losures Unemployment nd rel wges under lterntive losures Growth rte of mro vribles with inresed produtivity nd investment Setorl growth rtes Additionl mro results Regionl employment indexes by region nd skill 16 A.1 Setor disggregtion 18 A.2 Household disggregtion 19 A.3 A forml sttement of the dynmi omputble generl equilibrium model 19 List of Figures 3.1 Flow of goods from produers to the ntionl omposite ommodity Perent hnge in GDP in response to remittne shoks The effet of lterntive lbor mrket losures on gross domesti produt (GDP), onsumption, nd bsorption 12 iv

5 ABSTRACT In this pper we used dynmi, regionlized omputble generl equilibrium (CGE) model to nlyze the effet of vrious negtive blne of pyments shoks on output nd employment nd the effet of different lterntive investment strtegies on growth. The model shows lerly how sensitive El Slvdor is to remittne or terms of trde shoks. Eh 10 perent redution in remittnes lowers gross domesti produt (GDP) by 0.2 perent nd household onsumption by 1.4 perent, with the ost rising s the shok intensifies. Any negtive blne of pyments shok fores redution in bsorption, prodution, nd employment nd rel devlution. Beuse El Slvdor s eonomy is dollrized, tht rel devlution n only ome bout through fll in domesti pries brought bout by reession. We show tht the impt of the shok on output depends on how flexible wges re the impt is smller when rel wges re flexible nd gretest when they re fixed in dollrs. We used the CGE model to nlyze lterntive investment strtegies for inresing the growth rte. The investment shre of GDP is low, nd the model mkes it ler tht without some strtegy for inresing investment, the eonomy s overll growth rte is likely to remin low. We hypothesized two lterntive growth rtes for investment, both ssoited with n inrese in exogenous tehnil hnge. Both strtegies require mrginl inrese in the shre of output devoted to investment. We lso showed tht if El Slvdor n inrese the investment shre from 15.5 perent to just 16 perent over five yers by produing growth rte in investment of 8 perent per yer, nd if tht inrese produes 1 perent inrese in the rte of tehnil hnge in ll setors, then the growth rte of the eonomy will prtilly double, rising from 2.85 perent to 4.95 perent per yer. There re eqully fvorble effets on employment for unskilled lbor nd on wges for skilled lbor. Keywords: generl equilibrium models, El Slvdor, development strtegies, regionl development v

6 ACKNOWLEDGMENTS This projet ws finned by Desrrollo Rurl Sostenible de Centromeri (RUTA) under United Ntions ontrt (UNOPS #59526). We wish to thnk the stff of RUTA nd the Centrl Bnk of El Slvdor for their dministrtive support nd for their omments t seminrs where these results were presented. vi

7 1. INTRODUCTION El Slvdor is smll open eonomy tht is inresingly subjet to externl shoks over whih it hs little ontrol, t lest in the short run. It imports most of its fuel, is highly dependent on remittnes, nd runs very lrge trde defiits. Poliymkers re fored to operte in n inresingly turbulent nd diffiult externl environment, mde more diffiult by the onentrtion of exports in very few setors nd by the lrge frtion of neessry intermedite inputs tht re imported. In reent projet, we updted the El Slvdor soil ounting mtrix (SAM) nd built dynmi regionlized omputble generl equilibrium (CGE) model to help in the nlysis of externl shoks nd to explore some poliy lterntives. This pper briefly desribes the updted SAM nd the CGE model nd presents simultion results tht show the severe impt of either the redution of remittne inflows or the deteriortion in terms of trde through inresed oil pries or fll in the prie of offee, one of El Slvdor s min exports. Setion 2 desribes the updted SAM, nd Setion 3, the regionlized CGE model. In Setion 4, we disply the omprtive stti results for three different negtive externl shoks fll in remittnes, rise in the prie of oil, nd fll in the prie of offee. In Setion 5, we use the dynmi version of the model to show the positive effet of inresing investment nd produtivity on the growth rte. Setion 6 onludes. 1

8 2. THE UPDATED SOCIAL ACCOUNTING MATRIX (SAM) Prior to this projet, El Slvdor hd ntionl SAM bsed on dt from We updted the ntionl SAM to 2005 nd extended it in three diretions. First, we regionlized the SAM, using informtion from reent household surveys nd the reent griulturl ensus. 1 Seond, we put lnd into the mtrix, whih permits us to tie prodution in griulture to eh of the four regions in the SAM. Third, we regionlized household inome so tht we ould show the regionl impt of eh externl shok or the inreses in produtivity tht we simulted. The SAM is disggregted into the 46 setors shown in Tble A.1 (in the Appendix) nd into the four regions defined in footnote 1. We seprtely report four ftors of prodution, skilled nd unskilled lbor, lnd, nd pitl. Tble 2.1 displys the mro SAM tht results from ggregting ll the olumns nd rows of the full regionlized SAM. 2 Severl importnt hrteristis of the eonomy n be seen in Tble 2.1. In the first ple, there is very lrge imblne between imports nd exports, thnks to remittnes. Imports in 2005 n be seen in the Commodities olumn nd in the Rest of the World row. There were 7,660 imports, omprising 20 perent of the vlue of totl supply (nd 44 perent of gross domesti produt [GDP]). Exports were only 4,574. Most of the resulting ommeril defiit of 3,086 ws finned by remittnes, whih were estimted t 2,437 in 2005 nd hve grown rpidly sine. 3 Another importnt hrteristi of the Slvdorin eonomy is the high shre of pitl in GDP. Totl GDP t ftor ost is the sum of pyments in the Ativity olumn to lbor pitl (inluding lnd), or 16,035. Of tht mount, the lbor shre is only 37 perent, whih is one of the lowest in the region. Tble Mro SAM for El Slvdor (Millions of U$S dollrs) Commodities Mrgin Costs Ativities Lbor Cpitl Lnd Household Commodities Mrgin Costs Ativities Lbor Cpitl Lnd Household Government Indiret Tx Diret Tx Sving-Investment Chnge of Stoks Rest of the World Totl We ggregted the dt by deprtmentos into four regions, s follows: region oidentl (deprtmentos Ahuhpn, Snt An, nd Sonsonnte), Centrl 1 (Chltenngo, L Libertd, Sn Slvdor, nd Custln), Centrl 2 (L Pz, Cbns, nd Sn Viente), nd Orientl (Usultn, Sn Miguel, Morzn, nd L Union). 2 The full SAM n be found on the IFPRI website. 3 Totl remittnes re found in the Rest of World olumn nd Household row. 2

9 Tble 2.1 Continued Government Indiret Tx Diret Tx Sving- Investment Chnge of Stoks Rest of the World Commodities Mrgin Costs Ativities Lbor Cpitl Lnd Household Government Indiret Tx Diret Tx Sving-Investment Chnge of Stoks Rest of the World Totl Soure: 2005 soil ounting mtrix. Totl The disggregted SAM shows why nd how El Slvdor is so vulnerble to externl shoks. On the one hnd, the import shre of totl ommodity demnd is very high in mny key setors. Tble 2.2 shows imports nd totl supply in nongriulturl setors 42 perent of petroleum is imported, s is 49 perent of metls, mhinery, nd trnsporttion equipment. Perhps most importnt of ll is mquil, treted here s n industril servie of ssembly (sind, in the lst row of Tble 2.2). Mquil is one of El Slvdor s most importnt exports, s disussed below. However, we see here tht of the totl supply of mquil (1,793), 1,360 (or 76 perent) is imported. All of these dt imply tht ny growth strtegy, prtiulrly one relying on export promotion, is going to require gret del of imports. For the sme reson, ny negtive externl shok is likely to hve big impt on domesti prodution. Tble 2.2 Imports nd totl supply Imports Totl Supply Chemils Oil Rubber Non metl Metls Mhinery Trnsport equipment Eletriity Commere Resturnts Trnsporttion Communitions Finnil Servies Rel Stte Soil Servies Mquil Soure: 2005 soil ounting mtrix. 3

10 The SAM not only indites the lrge shre of totl supply in mny key setors tht is imported, but lso the high onentrtion of exports in smll number of setors, ll of whih lso inrese the eonomy s potentil vulnerbility to blne of pyments shoks. Tble 2.3 shows the shre of exports in totl prodution by mjor setor. Forty-one perent of exports ome from mquil (sind); 17 perent from the two tourism setors, resturnts nd hotels, nd trvel; nd n dditionl 5 perent from offee. In other words, lmost two-thirds of totl export revenue omes from just four setors. Other thn offee, griulture mkes only limited ontribution to exports. Other thn mquil, ll the setors of mnufturing mke up no more thn 22 perent of export reeipts, nd s we hve seen, mquil itself uses gret del of imports. At present, wht ll of this is likely to men is tht both the potentil to expnd exports seems to be quite limited, nd the import shre in key setors is high, whih implies limited pity to respond to blne of pyments shoks either by expnding exports or by signifint import substitution. Tble 2.3 Shre of exports nd imports by setor %Totl Exports % Totl Imports Coffee Grins Other griulturl produts Fishery Minnery Milling Sugr Other groindustry Beverges Textiles Pper Chemils Oil Plsti nd rubber Metls Mhinery Trnsporttion equipment Resturnts nd hotels Trnsporttion nd storge Communition Rel stte Soil servies Mquil Soure: 2005 soil ounting mtrix. 4

11 3. THE COMPUTABLE GENERAL EQUILIBRIUM (CGE) MODEL The regionl CGE model used in this prt of the reserh ws bsed on the stndrd model used by IFPRI (see Lofgren, Hrris, nd Robinson 2001), whih follows the neolssil struturlist trdition originlly presented in Dervis, de Melo, nd Robinson (1982), but with some neessry modifitions to pture the multimrket spet of the griulturl setor inluded in this work. The dvntge of the regionlized model is tht it links deisions mde t the ntionl level with outomes on poverty, employment, nd prodution ross regions. This, in turn, permits us to exmine poliies intended to improve the regionl distribution of tivities nd inome in onsistent generl equilibrium fshion, whih inorportes ntionl mrofisl nd monetry onstrints. This dvne ws mde possible by the vilbility of regionl informtion from the reent griulturl ensus. The CGE model hs three omponents. The first shows the pyments tht re registered in the SAM, following the sme disggregtion of ftors, tivities, ommodities, nd institutions shown in the mtrix. The seond inludes equtions tht represent the behvior of the different institutions present. The third desribes the system of onstrints tht must be stisfied by the whole system overing the ftor nd goods mrkets, the blnes for svings, investments, the government, nd the urrent ount of the rest of the world. Eh produer mximizes profits under onstnt returns to sle nd perfet ompetition. There re three ftors of prodution: lbor (differentited by skill nd region), lnd (differentited by region), nd pitl. Prodution is relted to ftor inputs in onstnt elstiity of substitution (CES) prodution funtion, whih llows the produers to substitute these three inputs until they reh the point t whih the mrginl revenue of eh ftor equls the ftor prie (wge or rent). The seond hoie tht the produers mke is the mount of intermedite inputs they will use. This speifition is mde ssuming fixed shres tht speify the pproprite mount of intermedite inputs per unit of output nd lbor/pitl (vlue dded). Finlly, output pries depend on the vlue dded (ost of lbor nd pitl, intermedite inputs, nd ny relevnt txes nd subsidies. Figure 3.1 shows the flow of single ommodity from produers to finl demnd. First, there is the ombintion of goods from ll produers into n ggregte ommodity output. This is hieved using CES produt demnd system, with the intention of leving it up to the buyers to determine how muh to buy of eh produt (mximizing their onsumption). The ggregte output is sold domestilly or interntionlly. The produers llotion between domesti sles nd exports is speified vi onstnt elstiity of trnsformtion (CET) funtion, ssuming imperfet trnsformbility between exports nd domesti sles. The produers will sell their produts to the mrket with the highest profitbility. The domesti prie is the interntionl prie times the exhnge rte, plus ny possible export txes or export subsidies. The domesti good is ombined with imports to produe the omposite ommodity. For this, the Armington (1969) speifition is used, whih mens tht the domestilly produed nd imported goods re imperfet substitutes. 5

12 Figure 3.1 Flow of goods from produers to the ntionl omposite ommodity Imports Consumption Commodity output tivity Commodity output tivity CES Aggregte output CET Domesti sles Exports CES Composite good Government onsumption Investment Intermedite use Soure: Authors retion. Note: CES: onstnt elstiity of substitution; CET: onstnt elstiity of termintion. This model hs three institutions households, government, nd the rest of the world whih do three things (1) produe, (2) onsume, nd (3) umulte pitl. Households sve onstnt frtion of their disposble inome nd buy onsumption goods with the reminder. Household inome is the sum of slries, profits, nd government nd rest of the world trnsfers. Household onsumption of goods nd servies is determined by liner expenditure system (LES). The government reeives txes, onsumes goods nd servies, nd mkes trnsfers to households. The pitl ount ollets the svings from the households, government, nd rest of the world nd buys pitl goods (investment). Our CGE model ontins detiled informtion on the demnd nd supply of 46 eonomi setors/ommodities, with their 67 orresponding tivities (7 griulturl tivities, further disggregted by 4 regions). Lbor is disggregted by qulifition (skilled nd nonskilled) nd by region. Workers within eh region n migrte between setors nd ross regions ording to lbor demnd; however, for skilled lbor, totl lbor supply stys onstnt. Lnd is disggregted by region nd is region speifi, whih is one element tht drives the regionl prodution results. The other feture of our tretment of lbor is the supply urve for unskilled lbor. We ssume tht there is n exess supply of unskilled lbor, t lest over the rnge of solutions tht we nlyze. In essene, this mens tht the supply urve of lbor is flt, or tht the wge is fixed nd employment is endogenous. However, beuse the entire model is rel model or is expressed in terms of the numerire the wge of unskilled lbor is fixed either in rel terms or, in the se of El Slvdor, in dollrs. Household inome nd expenditure ptterns vry ross regions. This is importnt beuse the inomes erned by workers in different setors will benefit different households, depending on their lotion nd ftor endowments. These representtive households reeive ftor inomes nd per pit trnsfers from the ntionl government. Households sve some of their inomes nd use their remining inome to onsume goods under n LES of demnd. All ommodity mrkets re ntionl, so tht pries in ll ommodity mrkets differ only in trnsporttion osts. The Dynmi Version of the Model For the totl ftor produtivity (TFP) simultions, we used reursive dynmi CGE model, whih is solved in two stges. The first stge ims to find solution for one-yer equilibrium using stti CGE model. In the seond stge, model between periods is used to hndle the dynmi linkges tht updte the vribles tht drive growth. The intertemporl equtions provide vlues for ll exogenous vribles tht re needed by the stti CGE model for the next period, whih is then solved for new equilibrium. The model is solved forwrd in dynmilly reursive fshion, with eh stti solution depending only on urrent nd pst vribles. The model does not inorporte future expettions; insted, the behvior of 6

13 its gents is bsed on dptive expettions, beuse the model is solved one period t time. The vribles nd prmeters used s linkges between periods re the ggregte pitl stok (whih is updted endogenously, given previous investment nd depreition), the popultion, the domesti lbor fore, ftor produtivity, export nd import pries, export demnd, triff rtes, nd trnsfers to nd from the rest of the world (ll of whih re modified exogenously). 4 The llotion of new pitl ross setors is done by djusting the proportion of eh setor s shre in ggregte investment s funtion of the reltive profit rte of eh setor ompred with the verge profit rte of the eonomy s whole. Setors with higher (or lower) verge profit rtes will get higher (or lower) shres of the vilble investment. Over time, setor profit rtes should onverge. For the dynmi version of the model, we ssume the sme losures used in the omprtive stti version; we lso ssume tht the lbor fore grows by 0.5 perent per yer. We dd those dditionl workers to the supply of skilled lbor or to the surplus of unskilled lbor. 5 The growth of pitl is determined by the mount of investment, net of depreition. We lso updte the rte of unembodied tehnil hnge by 1 perent per yer or by different mount when we do simulted tehnil hnge experiments. We n then vry over time the exogenous rtes of sving, txes, nd eh of the other poliy prmeters in the model to determine the effet of these hnges on the eonomy s growth rte. As mentioned erlier, growth in the lbor fore by skill lss is exogenous nd relted to popultion growth, whih, in turn, is bsed on lulted growth projetions tken from ntionl dt. For unskilled lbor, the totl size of the vilble lbor fore does not ffet the solution in ny period, beuse in the simultions, we ssume n exess or bklog of unemployed lbor tht is not bsorbed before the end of our simultions. With this reursive CGE model, we re ble to use these senrios for the eonomy, projet the growth pths of the endogenous vribles, nd ompre the bse yer pth (in whih there re no hnges in poliy vribles) with the pths obtined with the proposed poliy hnges. The simultions run with the Slvdorin model give us the growth pth for the Slvdorin eonomy for under number of different poliy lterntives. These pths re ompred with the one obtined from the bse simultion (in whih no exogenous poliy hnges were inluded) to see the impts of implementing vrious TFP senrios ombined with new investment. To summrize, the dynmi umultion proess is updted by the following: Exogenous trends, suh s lbor fore growth, produtivity hnges, pitl stok growth, nd popultion growth Eonomi behvior, suh s distribution of investment by setor, distribution of lbor fore by setor, nd tegory) Implemented poliies, suh s hnges in remittnes, interntionl pries, nd hnges in TFP ompnied with investment) 4 The dynmi model used in this reserh follows the models developed by the IFPRI (see Lofgren, Hrris, nd Robinson 2001; Thurlow 2003). 5 There hs been stedy deline in the popultion growth rte, from more thn 1 perent per yer in the 1990s to less thn 0.5 perent per yer sine

14 4. BALANCE OF PAYMENTS SHOCKS In reent yers, El Slvdor hs been ffeted by t lest three seprte blne of pyments shoks. After period of very rpid growth prior to 2008, remittnes either leveled off or delined in During 2008, the prie of offee, n importnt export, fell by 15 perent, while the prie of oil rose by 85 perent between 2007 nd In our first set of experiments, we simulte the impt of 50 perent redution in remittnes (Remitt3), 20 perent redution in offee pries (PWEoff), nd 20 perent inrese in the interntionl prie of oil (PWMfuel). Tble 4.1 displys the results of these three shoks. Before exmining the results, however, we wnt to point out severl hrteristis of the model tht ber on the results. As mentioned previously, El Slvdor is dollrized. In our model, we hve two lterntive losures: We ould fix either the rel exhnge rte or the level of foreign sving. We hve fixed the level of foreign sving, whih mens tht the rel exhnge rte is endogenous. However, beuse the nominl exhnge rte is fixed (t one dollr), then the internl ost of living index hs to hnge to get the hnge in the rel exhnge rte determined by the model. This is ritil feture of the Slvdorin eonomy. Tble 4.1 Mro results for three negtive blne of pyments shoks Bse Remitt3 PWE off PWMfuel Absorption Consumption Investment Stoks Government Exports Imports GDP t mrket pries Rel Exhnge Rte CPI Soure: Author worksheets. Note: We ssume tht rel wges re onstnt in ll three simultions. The seond losure issue is how the eonomy djusts to hnges in domesti sving. A negtive blne of pyments shok requires some ombintion of n inrese in domesti sving or redution in investment, given tht foreign sving is fixed. We ssume wht is lled blned losure, in whih the investment shre of totl bsorption is fixed nd in whih ll households nd the government hve the sme perentge derese in their nominl spending. Rell tht we re ssuming flt supply urve for unskilled lbor. We n fix the minimum wge either in dollrs or reltive to the ost of living. In this simultion, we hve hosen the ltter, whih will turn out to be signifint distintion, s explined lter. A finl point to ber in mind is tht for these blne of pyments shoks, we re using the omprtive sttis model, whih does not tell us how long it tkes for these full impts to be felt. Insted, the solution gives us good indition of the size nd diretion of the hnge in the eonomy in retion to negtive hnge in externl onditions. Beuse we do not permit the ountry to inrese borrowing when it suffers negtive blne of pyments shok, there must be some ombintion of redued imports or inresed exports to ounterblne the shok. In the lnguge of trde theory, there hs to be n inrese in the prodution of trdble goods nd redution in nontrdbles. In eh se, totl bsorption must deline. A flt unskilled lbor supply urve ends up produing redution in totl prodution, even though there is n 8

15 inrese in exports (exept in the se of flling offee pries). The hnge tht uses this to hppen is the rel devlution. With the 50 perent deline in remittnes, the rel exhnge rte depreites by 8.9 perent. However, beuse the nominl exhnge rte is onstnt, it is brought bout by redution of 6.7 perent in the onsumer prie index (CPI). In other words, the eonomy ontrts, whih in itself tends to drive down the demnd for imports. More to the point, the ontrtion redues the demnd for nontrdbles nd drives down their pries. Beuse pries of trdbles re onstnt, prodution of trdbles beomes reltively more profitble, whih uses prodution to shift from nontrdbles to trdbles. Beuse El Slvdor is smll, open eonomy, it is likely to be prtiulrly sensitive to dverse blne of pyments shoks. To see just how sensitive it is, we rern the omprtive sttis experiments, but using the fixed rel wge losure for unskilled lbor nd vrying the redution in remittnes from 10 perent to 70 perent. The results for GDP re displyed in Figure 4.1. Losing 20 perent of remittnes osts 0.5 perent of GDP, wheres redution of 50 perent osts 1.2 perent. Unfortuntely, the reltionship ppers to be nonliner; more severe shoks hve reltively lrger effet on the growth rte thn do smller ones. Figure 4.1 Perent hnge in GDP in response to remittne shoks Soure: Authors worksheets. The Impt of Blne of Pyments Shoks on Setorl Output In Tble 4.2, we hve ggregted the hnge in vlue dded by mjor setors to show more lerly the shift towrd the prodution of trdbles. The hnge in vlue dded for mquil is shown seprtely beuse of its entrl importne to exports nd the eonomy. The remittne experiment shows the proess most lerly. All the trded goods setors, griulture, proessed food, industry, nd mquil hve positive growth rtes, wheres servies (whih here inlude onstrution) hve negtive growth rte. The results for the two prie experiments re more mbiguous. In the offee se, not surprisingly, griulture ontrts beuse of the impt of the prie shok on the offee setor itself. Other thn griulture, the reltive growth rtes fvor the rest of the trded goods setor reltive to servies. A rise in the prie of oil is lmost universlly ontrtionry, exept for smll inrese in industril prodution. Oil nd fertilizer re key inputs for lmost every setor, so the rise in oil pries rises input osts ross the bord. Beuse there is not muh internl prodution of import substitutes in this key setor, osts rise, infltion inreses, nd output goes down in virtully every setor. 9

16 Tble 4.2 Setorl growth rtes of vlue dded Bse Remitt3 PWEoff PMWfuel Agriulture Proessed food Other industry Servies Mquil Totl Soure: Authors worksheets. Unemployment nd Rel Wges Our simultions ssume tht the rel wge for unskilled lbor is onstnt, whih implies tht ny vrition in demnd will be refleted in hnge in unemployment. For skilled lbor, pitl, nd lnd, the impt of ny exogenous hnge will be on the rel wge. In Tble 4.3, we show how the three negtive blne of pyments shoks ffet wges nd unemployment. In the bse, onstrution unemployment rtes re zero or onstnt. Note tht in the unemployment prt of the tble, the numbers shown re the equilibrium unemployment rte fter the shok. In the rel wge portion of the tble, the numbers shown re the perentge hnge in the rel wge from its bse level. Tble 4.3 Unemployment nd the rel wge Unemployment Rte (UNEMPRX) Bse Remitt3 PWEoff PWMfuel flb-nk-or flb-nk flb-nk flb-nk-o Averge Rel Wge (WREALXP) (%hnge) Bse Remitt3 PWEoff PWMfuel flnd-or flnd flnd flnd-o flb-sk-or flb-sk flb-sk flb-sk-o unskilled lbor fp Soure: Authors worksheets. 10

17 Severl importnt onlusions n be drwn from the tble. The first is tht eh negtive shok uses rise in unemployment bout 1.8 perent on verge in the remittne shok nd good del less for the two prie shoks. For the other ftors of prodution, eh shok leds to redution in rel wges, or rtes of return, whih mens tht these shoks led to nrrowing of wge differentils. Those unskilled workers who keep their jobs gin reltive to skilled workers or profits; however, offsetting tht, fewer of them hve jobs. It my seem obvious tht these negtive shoks would led to rising unemployment, but, in ft, it is not so obvious. All we know is tht the djustment requires fll in bsorption nd shift of prodution in fvor of trded goods. Why should tht djustment use n inrese in unemployment? The reson is tht there is differene in pitl or lbor intensity between the trded nd the nontrded goods setors. The trded goods setors tht expnd must use or require more pitl nd less unskilled lbor thn the ontrting nontrded goods setors. Beuse the totl supply of pitl is fixed in these omprtive sttis experiments, the dditionl mount of unskilled lbor needed in the expnding setors is less thn wht is relesed in the nontrded ontrting setors. The result is rise in unemployment. Hd we ssumed full employment, the rel wge of unskilled lbor would hve ontrted by enough to mke it profitble to substitute unskilled lbor for the other ftors nd hire the otherwise unemployed. The Effet of Rel Wges Demnds on the Eonomy s Retion to Negtive Blne of Pyments Shoks In the simultions nlyzed so fr, we hve ssumed onstnt nominl wge or flt supply urve for unskilled lbor. Beuse this is rel model, however, the onstnt nominl wge trnsltes into two different rel supply urves. In the simultion used so fr, tht nominl wge is expressed reltive to the ost of living index. But in El Slvdor, where wge demnds (nd pries) re expressed in dollrs, n lterntive losure fixes the wge in dollrs, whih mkes firly big differene in how the eonomy rets to shoks. The reson for this differene is tht if workers demnd fixed wge in dollrs, then the fll in domesti pries during the djustment will not redue the ost of lbor in the export setor. In other words, rel devlution will not lower the ost of unskilled lbor. Insted, it will dmpen the expnsion of the exports, whih, s Setion 3 showed, re n importnt prt of the djustment to remittne shok. Of ourse, the wges of skilled lbor, pitl, nd lnd will fll, but the overll hnge in the struture of prodution towrd trdbles will be smller. As result, more of the totl djustment to the shok will ome through domesti reession nd less through djustment in the prodution struture. To see how importnt the tretment of rel wges is for El Slvdor, we rern the remittne simultion using three different losures for unskilled lbor. Remitt1 mkes the unskilled lbor mrket neolssil tht is, the supply of unskilled lbor is fixed, nd the rel wge is endogenous. In Remitt2, the wge is fixed in dollrs. Remitt3 fixes the rel wge reltive to the CPI, whih is the losure used in Setion 3. Tble 4.4 nd Figure 4.2 show the mro results of these three lterntives. The min point of the tble nd figure is the sensitivity of the rel eonomy to how wges re determined. If workers insist on wges fixed in dollrs (Remitt2), then GDP flls by 2.5 perent, ompred with only 0.6 perent with full wge flexibility or 1.2 perent with fixed rel wges. The sme differene is seen in onsumption, investment, nd totl bsorption. We know tht bsorption hs to fll with this negtive blne of pyments shok, but the point is tht it flls further the more inflexible the wges re in dollrs. 11

18 Tble 4.4 Mro impt under three lterntive lbor mrket losures Bse Remitt1 Remitt2 Remitt3 Absorption Consumption Investment Stoks Government Exports Imports GDP t mrket prie CPI Rel xrt Soure: Authors worksheets. Figure 4.2 The effet of lterntive lbor mrket losures on gross domesti produt (GDP), onsumption, nd bsorption Remitt3 Remitt2 Globl mrket prie Consumption Absorption Remitt Soure: Authors worksheets. Consider now the effet of these lterntive losures on rel wges nd unemployment. Unemployment of unskilled lbor is defined to be zero in both the bse run nd the vertil supply urve simultion (Remitt1). However, dollr wge inflexibility pushes the unskilled unemployment rte up to 5.2 perent, whih is signifintly higher thn if workers demnd fixed rel wge in purhsing power. The reson is tht beuse the overll prie index flls (see Tble 4.4), nominl wge demnds in Remitt3 lso fll, whih is prtiulrly importnt in very open eonomy suh s El Slvdor, where mny onsumer goods re imported. Beuse the CPI omprises both domestilly produed nd imported goods, it does not fll s fr s domesti pries do. Thus, if the nominl wge is fixed in terms of the CPI, its hnge must lie between the fll in domesti pries nd the onstnt prie of imports, nd it must be reltively lose to the ltter. The implition of this ft on reltive wges is quite striking: If unskilled 12

19 lbor fixes its wge in dollrs, s it does in Remitt2, then there is lrge inrese in the rel wge beuse of the fll in the CPI. As n be seen, the rel wge for unskilled lbor in eh of the four regions rises by bout 7 perent, wheres the rel wges or rtes of return of ll the other ftors of prodution fll shrply. For unskilled lbor, when wges re fixed in either dollrs or the ost of living, the impt of negtive blne of pyments shok omes through unemployment. Unfortuntely, tht rise in unemployment is one reson GDP hs to ontrt. With wge inflexibility, the djustment to shoks is less in pries nd wges nd more in output nd unemployment. Tble 4.5 Unemployment nd rel wges under lterntive losures Unemployment Rte (UNEMPPRX) Bse Remitt1 Remitt2 Remitt3 flb-nk-or flb-nk flb-nk flb-nk-o verge Rel Wge (WREALXP) Bse Remitt1 Remitt2 Remitt3 flnd-or flnd flnd flnd-o flb-sk-or flb-nk-or flb-sk flb-nk flb-sk flb-nk flb-sk-o flb-nk-o fp Soure: Authors worksheets. Note: Unemployment is for unskilled lbor only. 13

20 5. DYNAMICS: SIMULATING THE EFFECT OF INCREASING PRODUCTIVITY From poliy perspetive, it is useful to know the effets of exogenous hnges in onditions, whih is wht we derived in Setion 4. However, it is lso importnt to know how the eonomy rets over time to poliy hnges. Suppose the government, in response to negtive blne of pyments shok, inreses expenditure on produtivity-enhning investments. Cn tht, over time, offset the negtive omprtive sttis effet of the shok itself? More generlly, wht is the likely effet on the eonomi growth rte of inresing the growth rte of produtivity through dditionl investment? To ddress questions like these, we hve built dynmi version of the CGE model (see Setion 2). In this setion, the model is run over five-yer time horizon, with vrious inreses in setorl or ntionl produtivity with nd without simultneous blne of pyments shok. The tbles tht follow report only the nnul growth rtes over the five-yer time horizon of eh eonomi vrible of interest. We performed two simple produtivity experiments. In one, we inresed the nnul growth rte of investment by 4 perent per yer, whih mde the investment shre of GDP slightly higher thn the bse in yer five. We ssumed tht this ws suffiient to rise the growth rte of produtivity in ll griulturl setors by 1 perent per yer. In the seond experiment, we rised the nnul growth rte of investment by 8 perent per yer, whih rised the investment shre of GDP from 15.5 perent to 16.0 perent by yer five. We ssumed tht this mount of dditionl investment ws suffiient to rise the produtivity in ll setors by 1 perent per yer. (Note: We hve no empiril evidene tht this mount of inresed investment will in ft bring so lrge n inrese in produtivity.) The third experiment ombined the ntionl inrese in produtivity nd ssoited investment with the negtive blne of pyments shok. To mke the experiment roughly onsistent with the omprtive sttis experiment on remittnes, we ssumed 10 perent nnul deline in remittnes suh tht over the five yers, the totl deline ws round 50 perent. In ll these experiments, we ssumed tht the nominl wge ws fixed in dollrs (losure#2) The first thing ws to estblish the eonomy s bseline growth rte, where we get the growth pth of the eonomy in the senrio where nothing hppens (no exogenous hnges re introdued). As noted erlier, El Slvdor hs reltively low investment rte, equl to 15.5 perent in We ssumed tht the growth rte of the lbor fore ws equl to tht of the popultion, or 0.5 perent per yer. Exogenous tehnil hnge ws fixed t 1 perent per yer in ll setors. Given these three ssumptions, the eonomi growth rte ws 2.8 perent per yer (see Tble 5.1). Tble 5.1 Growth rte of mro vribles with inresed produtivity nd investment Demnd-side GDP tble Growth rte per yer INITIAL Bse TFPg TFP-ntl TFP-ntl-remit Absorption REAL Consumption REAL Investment REAL Stoks REAL Government REAL Exports REAL Imports REAL GDP t mrket pries REAL Soure: Authors worksheets. 14

21 In the first experiment, we ssumed tht the ountry sueeds in inresing investment by 4 perent per yer, so tht the investment shre of GDP in the finl yer grew slightly from 15.5 perent. We then ssumed tht ll of the dditionl investment is devoted to griulture. In tht se, not only does the pitl stok grow more rpidly, but lso, it is ssumed, the rte of tehnil hnge in griulture jumps by 1 perent per yer. These hnges inrese the griulturl growth rte by 1.5 perent, the export growth rte by just more thn 1 perent, nd the entire eonomy s growth rte by 0.6 perent. In the seond experiment, we upped the growth rte of investment to 8 perent per yer nd extended the inrese in tehnil hnge to the entire eonomy. Not surprisingly, this hs very lrge impt on the growth rte in every setor. GDP now grows by 5 perent per yer, whih is more thn 2 perentge points higher thn the bseline growth rte. In effet, this is n export- nd investment-led growth strtegy, but even so, privte onsumption rises by more thn 4 perent per yer, ompred with 2.65 perent in the bse run. The importnt point is tht more investment inreses the growth rte of onsumption, even though it rowds out onsumption in the short run. In the third experiment, we posited the sme rpid growth of investment nd tehnil hnge s in the seond experiment, but with remittnes flling by bout 10 perent per yer. As the reder n see, this derese redues the overll growth rte of the eonomy, but the big impt is on onsumption. In this lst experiment, onsumption is rowded out by investment nd exports, nd lthough its growth rte still exeeds the bse, it is lmost 2 perentge points lower thn the totl ftor produtivity (TFP) ntionl experiment. Tble 5.2 shows the nnul growth rtes of setorl vlue dded into our three simultions. Fousing investment on griulture not only rises the growth rte of the eonomy, it lso mkes griulture leding setor, whih hs positive distributionl implitions s well. If we rise the investment rte to 8 perent per yer nd ssume tht this rises produtivity in every setor, we get big inreses in the growth rte of ll setors. Although griulturl growth is surpssed by ll the other setors, its growth rte rises by hlf perentge point reltive to previous experiments. Tble 5.2 Setorl growth rtes Growth rte (% per yer) Bse TFP-g TFPNtl TFPNtl-r Agriulture Pro gri Industry Servies Mquil Totl Soure: Authors worksheets. Some dditionl hrteristis of the growth pth in these three experiments n be seen in Tble 5.3. First, we see tht in eh experiment, the onsumer prie index (CPI) flls nd the rel exhnge rte depreites espeilly in the third experiment, where rising produtivity is ompnied by flling remittnes. This rel depreition is the mehnism by whih the eonomy frees enough resoures to generte the 9.3 perent growth rte in exports neessry to stisfy the foreign sving onstrint. We lso see tht sine investment is growing by 8 perent in both the seond nd third experiments, the investment shre rises from 15.5 perent to round 16 perent. (In the bse run, the investment shre flls to 15.2 perent by the end of the simultion.) The growth of the shre is bit higher in the experiment with remittnes, but only beuse the GDP s growth rte is somewht lower. Both the trde defiit nd the foreign svings shre fll in every one of these experiments. By onstrution, we re foring the eonomy to grow with onstnt mount of foreign svings in dollrs. In effet, we re foring the ountry to grow its wy out of so gret dependene on either remittnes or externl borrowing. Inresingly, it finnes its imports through export ernings, whih is espeilly true when we tke wy hlf of its remittne 15

22 revenue. The experiment shows tht lthough it is possible to do this, household onsumption suffers. Note however tht even with this onstrint, the growth rte of onsumption is higher thn it is in the bse run, in whih there is no hnge in poliy. Tble 5.3 Additionl mro results Prie, rel exhnge rte nd shres Initil vlue or shre Chnge from initil vlue (% hnge) Initil Bse TFPg TFP-ntl TFP-ntl-remitt Rel exhnge rte Domesti prie index Consumer prie index Terms-of-trde Investment/GDP Privte svings/gdp Foreign svings/gdp Trde defiit/gdp Soure: Author s worksheets. The Lbor Mrket The key to understnding the impt of these vrious growth experiments on the eonomy is the lbor mrket (see Tble 5.4). Rell tht for skilled lbor, we ssume tht supply is exogenous nd growing by 0.5 perent per yer. For unskilled lbor, we ssume tht the wge is fixed in dollrs. In other words, if there is ny sort of exogenous shok, the djustment for skilled lbor will be in the wge, wheres for unskilled lbor, it will be in the level of employment. Tble 5.4 shows employment levels by region nd skill for the bse yer nd the finl yer of the simultion. Note tht for skilled lbor, there is no differene between the finl employment levels in ny of the experiments, beuse the lbor supply is exogenous. In the bse run for unskilled lbor, we ssume tht unemployment is zero or onstnt, so the level of employment in the finl yer is simply 2.5 perent higher thn the bse. The tble demonstrtes the impt of investment on the demnd for unskilled lbor. If we invest only in griulture, the rte of inrese in employment for unskilled lbor jumps from 0.5 perent per yer to 0.8 perent. With the higher investment rte in the seond experiment, employment growth rises to 0.9 perent per yer. In other words, if one thinks there is exess unskilled lbor or wht mounts to the sme thing s underemployed lbor ny investment-led growth strtegy will hve big impt on employment growth for the unskilled. Tble 5.4 Regionl employment indexes by region nd skill Unskilled lbor Skilled lbor Region Bse yer Yer five Bse yer Yer five Bse TFPg TFP-ntl TFP-ntl-remit Est Centrl Centrl West Totl Soure: Authors worksheets. 16

23 6. CONCLUSIONS In this pper, we desribed the dynmi CGE model tht we reently built for El Slvdor nd then onduted number of simultions with the model, in prt to show wht the model n do nd in prt to give some guidne to poliymkers on importnt urrent poliy issues. We onsidered two poliy onerns. The first ws the sensitivity of the eonomy to externl shoks, nd the seond is the need to inrese the growth rte. The sensitivity of the eonomy is in prt due to eonomi struture nd to the ft tht the eonomy is dollrized nd wge demnd re mde in dollrs. We nlyzed three negtive blne of pyments shoks. First we nlyzed 50 perent redution in remittnes, nd then we nlyzed two prie shoks one 20 perent rise in the prie of petroleum nd the other 20 perent redution in the prie of offee, n importnt Slvdorin export. The model mde ler tht eh shok fores redution in bsorption, prodution, nd employment nd rel devlution. Beuse the eonomy is dollrized, tht rel devlution n only ome bout through fll in domesti pries brought bout by reession. We showed tht the impts of the shok re muh smller when rel wges re flexible nd gretest when they re fixed in dollrs, beuse the eonomy hs to shift the struture of prodution nd employment towrd the trdble goods setor. However, this shift is diffiult to omplish when wges re fixed in dollrs, beuse in tht se, even domesti reession does not drive down the ost of unskilled lbor. The djustment to shoks is omplited by two struturl hrteristis of the Slvdorn eonomy. The first hrteristi is the big proportion of imports in mny key setors, espeilly petroleum, mquil, nd mhinery. The seond is the high onentrtion of exports in very smll number of setors, some of whih re unlikely to be ble to expnd signifintly in response to hnge in reltive pries. This implies tht ny long-run growth strtegy or ny short-run shok djustment strtegy is likely to inrese import demnd more thn export supply. This inrese, in turn, mkes it more likely tht in response to blne of pyments shok, the eonomy will be fored to redue the demnd for imports through reession rther thn through struturl shift in prodution towrd trded goods. The dynmi CGE model is good tool for nlyzing lterntive investment strtegies to inrese the overll growth rte. The investment shre of GDP is low, nd the model mkes it ler tht without some strtegy for inresing investment, the overll growth rte of the eonomy is likely to remin low. We hypothesized two lterntive growth rtes for investment, both ssoited with n inrese in exogenous tehnil hnge. Both strtegies require mrginl inrese in the shre of output devoted to investment. We showed tht if El Slvdor n inrese the investment shre from 15.5 perent to just 16 perent over five yers by produing growth rte in investment of 8 perent per yer, nd if tht produes in turn 1 perent inrese in the rte of tehnil hnge in ll setors, then the eonomi growth rte will prtilly double, rising from 2.85 perent to 4.95 perent per yer. There re eqully fvorble effets on employment for unskilled lbor nd on wges for skilled lbor. The development lessons re ler, nd the stkes re high. If investment beomes more profitble so tht investment inreses, nd if wy n be found to devote more of tht investment to rising produtivity prtiulrly in the trdble goods setors the impt on growth nd employment will be positive nd lrge. 17

24 APPENDIX: SUPPLEMENTARY TABLES Tble A.1 Setor disggregtion fé offee grn grins zu sugr ne frut fruits vege vegetbles otg other griulturl produts gn livestok vi poultry silv forestry pes fishery mine minnery rn met lt diry moli mining zum sugr ne otl other groindustry bebi beverges tb tobo text textiles vest lothing uer lether mde wood ppe pper impr printing quim hemils petr oil u plsti nd rubber nome minerl produts met metls mqu mhinery mtr trnsporttion equipment ele eletriity gu wter ons onstrution ome ommere rest resturnts nd hotels trn trnsporttion equipment omu ommunition bn bnks nd finnil institutions inmu rel stte lqu property rentls soi soil servies dome domesti servies sgob government sind industril servies oser other servies Soure: Survey dt. 18

25 Tble A.2 Household disggregtion Households hhd-or region orientl hhd-1 region entrl 1 hhd-2 region entrl 1 hhd-o Soure: Survey dt. region oidentl Tble A.3 A forml sttement of the dynmi omputble generl equilibrium model Symbol Explntion Symbol Explntion A Ativities CMN( C) Commodities not in CM ACES( A) ALEO( A) CX ( C) Ativities with onstnt elstiity of substitution (CES) funtion t the top of the tehnology nest CT ( C) Ativities with Leontief funtion f F Ftors t the top of the tehnology nest Trnstion servie ommodities Commodities with domesti prodution fsub F C Commodities f2s F Composite ftors Ftors used in omposite ftors Commodities with domesti sles CD( C) of domesti output i INS CDN( C) Commodities not in CD i INSD( INS) Domesti institutions CE( C) Exported ommodities i INSDNG( INSD) CEN( C) Commodities not in CE h H ( INSDNG) Households CM ( C) Imported ommodities fls F PARAMETERS wts dwts i id ' ie ' im ' int iv Weight of ommodity in the onsumer prie index (CPI) Weight of ommodity in the produer prie index Quntity of s intermedite input per unit of tivity Quntity of ommodity s trde input per unit of produed nd sold domestilly Quntity of ommodity s trde input per exported unit of Quntity of ommodity s trde input per imported unit of Quntity of ggregte intermedite input per tivity unit Quntity of ggregte intermedite input per tivity unit qg qinv shif if shii ii' t te tf f tins i Institutions (domesti nd rest of world) Domesti nongovernmentl institutions Ftors with supply urve Bse-yer quntity of government demnd Bse-yer quntity of privte investment demnd Shre for domesti institution I in inome of ftor f Shre of net inome of i to I (i INSDNG ; I INSDNG) Tx rte for tivity Export tx rte Diret tx rte for ftor f Exogenous diret tx rte for domesti institution i 19