Trade Liberalization and Labor Markets: the case of Indonesia

Transcription

1 Trade Lberalzaton and Labor Markets: the case of Indonesa M. Thea Snclar, Guntur Sugyarto and Adam Blake Abstract Trade lberalzaton has long been advocated for standard reasons of ncreasng competton and welfare. Yet, there s stll consderable debate, partcularly n developng countres, about the range of effects that result from lberalzaton and the relatve magntudes of dfferent effects. Such debate stems, n part, from the fact that trade reform takes place n a dstortonary context. The effects of trade reform n the presence of other dstortons fall nto two strands: trade lberalzaton and dstortonary domestc taxes, and trade lberalzaton and labor market dstortons. Whereas the former has receved sgnfcant attenton, the analyss has generally been undertaken n the context of an unchangng labor market regme. However, labor market reforms are commonly under consderaton, so that t s mportant to take account of ther possble effects on trade lberalzaton and welfare of dfferent sectors of the populaton. Ths paper examnes trade lberalzaton n the presence of alternatve labor market regmes n urban and rural sectors. The analyss frst consders a form of labor market dstorton that s relatvely common n developng countres. The presence of wage rgdtes n the form of mnmum wages, unonzed labor or government controls n formal or urban sectors contrasts wth the absence of such rgdtes n nformal or rural sectors. Ths paper examnes the consequences of trade lberalzaton n such a settng, showng that, under certan condtons, labor market rgdtes can mean that trade lberalzaton reduces welfare. The paper also examnes the effects of trade lberalzaton n the context of hgher or lower rgdtes n labor markets. The alternatve combnatons of labor market rgdty/lberalzaton and trade lberalzaton are tested n a computable general equlbrum model of Indonesa that ncorporates 18 producton sectors, 8 types of labor, 5 types of captal and 8 representatve household groups. Whle full trade lberalzaton s found to be welfare mprovng, lberalzng only part of the tarff schedule can lead to welfare losses.

2 Trade Lberalzaton and Labor Markets: the case of Indonesa M. Thea Snclar, Guntur Sugyarto and Adam Blake 1. Introducton Trade lberalzaton has long been advocated for standard reasons of ncreasng welfare. Yet, there s stll consderable debate, partcularly n developng countres, about the range of effects that result from lberalzaton and the relatve magntudes of dfferent effects. Such debate stems, n part, from the fact that trade reform takes place n a second-best world where a multtude of other dstortons exst. The effects of trade reform n the presence of other dstortons have only recently been addressed, and fall nto two strands: trade lberalzaton and dstortonary domestc taxes, and trade lberalzaton and labor market dstortons. The lterature that concentrates on trade lberalzaton and dstortonary domestc taxes admts that there are two reasons why lberalzaton may be harmful to welfare. Frstly, through a revenue replacement effect when the forgone tarff revenues are rased through domestc taxes whch may be more dstortonary than the orgnal tarff levels (Harrson et al., 1993). Secondly, f the structure of domestc taxes mposes a hgher tax burden on ndustres that are protected by the hghest tarffs, then the dstortonary nature of the domestc tax system may be offset by dstortons n the opposte drecton n the tarff structure (Konan and Maskus, 2000). Lberalzaton of tarffs may remove these offsettng dstortons and could lead to a more dstorted post-lberalzaton economy. The case of jont reform of tarffs and taxes was consdered by, for example, Clarete and Whalley (1987, 1988), and Ylmaz (1999) examned the welfare effects of optmal export taxes. Konan and Maskus (2000) extended the analyss to encompass the decomposton of the welfare effects nto the effects of trade reform, tax reform and ther nteracton. However, lttle attenton has been pad to the second strand of lterature examnng how labor market dstortons affect the welfare effects of trade lberalzaton. Notable exceptons are Devarajan et al. s (1997) study of the effects of unons on the outcomes of economc reform and Bussolo et al. s (2002) work on the effects of trade and labor market dstortons on trade volumes and the wage gap between the sklled and unsklled. The frst of these papers found that the presence of actve labor unons ncreases the gans from trade lberalzaton as the power of such unons to extract rents s dmnshed n a lberalzed economy. Despte ths sgnfcant fndng, there has been lttle attempt to examne how other types of labor market dstortons alter the welfare effects of trade lberalzaton. Ths paper attempts to fll a small part of ths gap by examnng the effects of tarff lberalzaton n the presence of wage rgdtes. The ssue of the welfare effects of trade lberalzaton s further complcated by the fact that dfferent types of trade and labor market reforms have dfferent effects on dfferent groups wthn the populaton. Increasng concern about the dstrbutonal repercussons of lberalzaton (Harrson et al., 2002) rases questons about whch groups gan the most from jont polcy reform and whch, f any, lose. Ths paper wll therefore also examne the dstrbutonal effects of lberalzaton n the presence of wage rgdtes. 2

3 2. The theoretcal framework The ntuton behnd the framework presented below s relatvely straghtforward. Consder an economy that protects mport-competng sectors through tarffs and n whch there are real wage rgdtes n those mport-competng sectors but not n other sectors. The rgdtes could be mposed by the government (mnmum wages or ncomes polces) or by unons (settng fxed real wages). Typcally, t s useful n the developng country context, to consder mnmum wages that are () not appled n agrculture; () below current wage levels n servces; and () at a bndng wage level n (mport-competng) manufacturng sectors. Labor s not fully moble between the mport-competng sectors and other sectors. In ths economy, tarff removal would result n less protecton of the mport-competng sectors, lower levels of output and employment n those sectors, and therefore lower levels of employment of labor n those sectors and n the economy as a whole. There are therefore two effects on welfare: frstly, a removal of dstortons that ncreases welfare by allowng producers and consumers to react to world prces, and secondly a resource effect that reduces welfare because some of the economy s resources (labor) are unemployed. Standard arguments for tarff reductons, based on a two good, two factor model, predct that trade wll rase welfare by allowng an economy to consume beyond ts producton possblty fronter, and that tarffs restrct the ablty of countres to consume beyond that fronter. Fgure 1: Tarff lberalzaton wth wage rgdtes X a b Q ** P * Q * P T Q T C T C * C ** c d 3

4 A two-sector specfc-factors model can be used to llustrate the effects of trade lberalzaton, ntally n a frst best context of no labor market dstortons and, subsequently, n the context of rgdtes n labor markets. The model conssts of sectors producng exportables and mportables n a small open economy, n whch both sectors use moble captal and sectorspecfc labor to produce ther respectve goods. Both sectors are perfectly compettve and have constant returns to scale technology. The producton possblty fronter aq T c n Fg. 1 shows the dfferent combnatons of exportables and mportables that can be produced n an economy wthout dstortons. Under free trade at world prces P * = P M /P X, the economy s able to produce at pont Q * and consume at pont C *. A tarff t forces producers and consumers to respond to the dstorted prce rato P T = (1+t)P *, when producton occurs at pont Q T and consumpton at pont C T, on a lower ndfference curve than pont C *. The fronter bq T d shows the output combnatons that are produced when the economy s subject to a fxed real wage n the mportables sector that s equal to the real wage receved by mportable-specfc labor at pont Q T. To the left of Q T, any reducton n mportable producton s accompaned by a reducton n employment of mportable-specfc labor. Consequently, more of mportable producton must be forgone n order to produce an addtonal unt of exportables, compared to the case wthout labor market dstortons on the orgnal fronter. To the rght of Q T, the reverse holds as ncreases n mportables producton are accompaned by ncreases n employment at the fxed real wage. A realstc nterpretaton of a mnmum wage s that the economy s restrcted to producton possbltes gven by bq T c, where (as the mnmum wage clears the mportables-specfc labor market at pont Q T ) the economy follows the orgnal producton possbltes fronter to the rght of pont Q T, but s restrcted to the left of Q T. Followng tarff lberalzaton (here, tarff removal), producton wll move around the producton possbltes curve to a pont such as Q **, at whch the levels of consumpton at C ** gve a lower level of utlty than the tarff-dstorted levels at C T. The slope of the producton possbltes curve at pont Q ** s crucally not tangental to the world prce rato, and t s ths fact that generates the fall n utlty. In order to understand why the prce rato s not tangental to the producton possbltes curve at pont Q **, recall that the output of mportables and exportables are gven by producton functons, X (, ) = x L K x x (, ) M = m L K m m The slope of the producton possblty fronter, dm dx, can easly be derved by takng dervatves of these functons, and settng margnal value products equal to wages: dm P w dl + rdk = dx P w dl + rdk X M M M M X X X In a Hecksher-Ohln model both factors are moble so that both sectors face the same wage rate ( wx = wm ) and any change n employment of ether labor or captal n ether sector must be accompaned by an offsettng change n employment n the other sector ( dlx = dlm and dk X = dkm). Eq. 1 then leads to the concluson that the slope of the PPF must equal the negatve of the prce rato, dm dx = PX PM, the standard result where the slope of the producton possbltes fronter depends only on the relatve prces. 4

5 If labor s mmoble then nstead of facng the same wage rate, employment n each ndustry must equal labor supply, f wages are flexble and clear labor markets ( dl X = 0 and dl M = 0 ). Ths agan leads to the same concluson n Eq. 1 that dm dx = PX PM. If, however, wages n the mportables sector are fxed relatve to the captal rental rate 1, wm = γ r then whle the exportable-specfc labor market must stll clear ( dl X = 0 ), the mportablesspecfc labor market need not do so ( dlm 0 ). Eq. 1 then leads to: dm P X γ dl = 1+ dx PM dk M M Because dlm dl X < 0, the slope of the producton possbltes curve at pont s less steep than the world prce rato whch leads to a fall n utlty when the tarff s removed. Note that hgher real wages (hgher γ ) wll lead to an even larger dfference between the prce rato and the slope of the producton possbltes curve, at a pont to the left of Q**, whch would lead to an even larger fall n utlty. The basc logc behnd ths concluson generalzes to models wth more complex structures, although there are other factors that need to be taken nto consderaton. In mult-sector appled models, such as the one employed below, there are usually several sectors that can be classfed as mport-competng; the tarff levels appled to mports may dffer n these sectors. There are often more than one factor of producton and t may be more realstc to consder wage rgdtes as applyng to a certan type or types of labor than to labor n specfc ndustres. The CGE model detaled below ncludes assumptons of labor mmoblty, for example, but also apples wage rgdtes to moble labor. In ths framework the share of mport-competng sectors n total employment of these labor types becomes mportant. If mport-competng sectors domnate n the employment of one or more of these labor types, results smlar to those derved above may be found; f other sectors also employ sgnfcant levels of these labor types then any fall n employment n the mport-competng sectors may be offset by ncreases n employment n other (for example, exportng) sectors. Dstortonary domestc taxes also exst n real-world economes, and t s mportant that appled models take these nto account. It s equally mportant n the present context to take care that welfare losses caused by the presence of a dstortonary domestc tax system, or to the dstortonary effects of revenue replacement, are not wrongly attrbuted to the effects of wage rgdtes. For ths reason, the effects of tarff reform wll be examned both wth and wthout wage rgdtes, and both wth and wthout revenue replacement through dstortonary domestc taxes. 3. Tarffs and labor markets n Indonesa Trade lberalzaton has a checkered hstory n Indonesa, as governments have contnued to rely on nternatonal trade taxes as a source of revenue. Major changes n trade polcy have been trggered by poltcal and economc crses, generally lnked to unfavorable external 1 Note that the results are not determned by the prce that real wages are set aganst. The wage could also be fxed relatve to a consumer prce ndex wth the same concluson. 5

6 condtons such as falls n the prces of prmary commodtes. The sequencng of trade lberalzaton n Indonesa followed substantal lberalzaton of captal and the fnancal market, as the man trade reforms were mplemented only after 1985, as ndcated n Table 1, manly n response to balance of payments problems. Government polces have tended to revert towards protectonsm n the absence of such problems. Table 1: Indcators of Reform (%) Measure Average tarff: Unweghted Producton weghted Import Lcensng: Import weghted Producton weghted Index of Dsperson 1) ) Measured by the coeffcent of varaton. Source: World Bank (1992), Indonesa Growth, Infrastructure and Human Resources, Report No IND. 6

7 Table 2: Trade lberalzaton epsodes n Indonesa, 1945 onwards The Chaotc Years ( ) Hgh nflaton and frequent economc polcy/government changes. Domnant role of taxaton on trade. Imposng multple exchange rates, export surcharges, quanttatve restrctons on mports and tarffs Stablzaton and Rehabltaton ( ) New nvestment law, development plan and balance budget. Abolsh multple exchange rates and peg to US $. Adopt an open captal account. Trade Lberalzaton Measures Adopted n Each Perod The Ol Wndfalls and Boom Years ( ) Domnant role of ol. Non-tradable and mport substtutng ndustry. Dutch Dsease. Ignore trade lberalzaton. More protectve. Adjustment to External Shocks ( ) Internatonal Debt and Mexco crses. Tax and fnancal reforms. Export promotng measure (TRIMs). Approved mporter system. Promotng Use of Domestc Products. Ambvalent towards trade lberalzaton. Further Trade Lberalzaton (1985-md 1997) Regonalsm of AFTA & APEC. Sgn GATT-Code (on subsdes & countervalng dutes). Ratonalze tarffs. Deregulate shppng & custom unons. Duty exempt and duty drawback. Removes export lcenses and convert QRs wth tarffs. The Economc Crss and Afterwards (1997-to date) The Asan Crss and IMF package. Reduce tarff more. Abolsh export taxes and mport restrctons. Lberalze domestc market.

8 The trade lberalzaton measures adopted by the Indonesan government snce ndependence n 1945 are summarzed n Table 2. The measures are classfed nto sx man stages, to reflect the nature of government polces at each stage Further trade lberalzaton seems nevtable, gven the Indonesan government s commtments to the World Trade Organzaton, Asa-Pacfc Economc Co-operaton and Assocaton of South East Asan Natons agreements to lberalze nternatonal trade. Moreover, a reducton of tarffs, n conjuncton wth other lberalzaton measures, has been part of the polcy package of condtonal loans to Indonesa by the IMF and World Bank. Labor markets n Indonesa are also experencng sgnfcant changes. Controls have been a prncpal feature of labor markets n the manufacturng sector, ntally by means of a sngle, government-controlled unon and, from 1989 onwards, by means of the mplementaton of a mnmum wage polcy for the manufacturng and clercal sectors. The government s tax and subsdy polcy towards agrculture has also contrbuted towards wage fxng wthn ths sector. Real wages n the manufacturng sector have tended to rse over tme, owng to labor unrest n the post-suharto era, ncreases n the mnmum wage and, more recently, to labor sheddng n response to the Asan fnancal crss. Wages n the small busness, agrculture and nformal sectors decreased durng the crss and have generally remaned low, as the relatvely hgh mnmum wage n the export-orented manufacturng sector has lmted employment n manufacturng, causng surplus labor to move nto the other sectors of the economy. Although the share of output provded by the manufacturng sector has ncreased consderably over the medum term, the share of labor wthn the agrculture and servce sectors of the economy remans hgh relatve to countres such as Korea and Malaysa. Ths employment structure rases nterestng dstrbutonal ssues for trade reform, the effects of whch wll be consdered wthn the framework of the CGE model of the Indonesan economy. 4. The Model The model ncorporates all producton, trade and consumpton n the Indonesan economy, and s calbrated to a socal accountng matrx (SAM) for 1993 (Central Bureau of Statstcs 1996). Producton s aggregated to eghteen sectors or ndustres, wth eghteen commodtes correspondng to these ndustres. Each ndustry produces only ts correspondng product. Table 3 shows the eghteen sectors n the SAM and model, and shows the relatve szes of the sectors, n terms of ther gross value added. Industres are characterzed by constant returns to scale producton technologes and act n a perfectly compettve manner, maxmzng profts gven exstng prces; and freedom of entry and ext ensures no profts are made above captal rental rates. Output s specfed as a fxed coeffcent or Leontef functon of dfferent ntermedate nputs and value added. Value added s a Cobb Douglas functon of sxteen dfferent types of labor (four categores, agrcultural, producton, clercal and professonal, each havng four components, waged rural, waged urban, non-wage rural and non-wage urban) and seven dfferent types of captal. Table 4 shows each of the labor and captal types, wth data for ther total GVA earnngs. Ths table also shows the szes of ndrect tax payments n the SAM. 8

9 Table 3: The Eghteen Sectors n the Model: Gross Value Added, GVA (Bn Rupah) Percent of total GVA ag-fc Agrculture-food crops 32, ag-oth Agrculture-others 29, mnng Mnng and quarryng 31, foodpro Food processng 24, textle Textle ndustry 27, constrc Constructon 7, pap-met Paper and metal products 15, chemc Chemcals 24, electgw Electrcty, gas and water 3, trades Trade servces 41, restaur Restaurants 8, hotels Hotels 2, landtra Land transports 11, otracom Other transport and communcaton 7, bankns Bankng and nsurance 14, realest Real estate 14, ser-pub Publc servces 24, ser-prv Other prvate servces 8, The model contans dfferent labor market formulatons that wll be used n dfferent smulatons. When labor s moble between sectors and market clearng condtons are nvoked the wage rate w adjusts to mantan equlbrum: j N j L, j = j = { agrcultural, producton, clercal, profesonal} (1) The combnaton of full moblty of labor and market clearng wages wll be termed the compettve case. In a second case, termed moble, mnmum real wages, agrcultural and professonal labor are moble wth market clearng wages whle producton and clercal labor, of whch there are four types each, wll have fxed wages,.e. N j = L j, j = { agrcultural, profesonal} (2) N j L j, j = { producton, clercal} (3) w j cp w j j = { producton, clercal} (4) Where labor s sector specfc n mport competng ndustres wth mnmum real wages, moblty and market clearng stll takes place n other ndustres,.e. N j = L j, j = { agrcultural, profesonal} (5) 9

10 Table 4: Factors of Producton: Gross Value Added, GVA (Bn Rupah) Percent of total GVA lbagwg-rl Labor agrculture wages-rural 9, lbagwg-ur Labor agrculture wages-urban 1, lbagnw-rl Labor agrculture non-wages-rural 27, lbagnw-ur Labor agrculture non-wages-urban 1, lbpnwg-rl Labor producton wages-rural 16, lbpnwg-ur Labor producton wages-urban 18, lbpnnw-rl Labor producton non-wages-rural 4, lbpnnw-ur Labor producton non-wages-urban 2, lbclwg-rl Labor clercal wages-rural 6, lbclwg-ur Labor clercal wages-urban 27, lbclnw-rl Labor clercal non-wages-rural 10, lbclnw-ur Labor clercal non-wages-urban 11, lbplwg-rl Labor professonal wages-rural 3, lbplwg-ur Labor professonal wages-urban 7, lbplnw-rl Labor professonal non-wages-rural lbplnw-ur Labor professonal non-wages-urban cp-land-ag Captal land and other agrculture 16, cp-own-hs Captal own occuped house 8, cp-oth-rl Captal others-rural 25, cp-oth-ur Captal others-urban 27, cp-pdom Captal prvate domestc 38, cp-gov Captal government-owned 3, cp-foregn Captal foregn-owned 35, Import tarffs 6, Other ndrect taxaton 15, Total 329, N N j = L, j = { producton, clercal} (6) N j where N s the set of non-mport competng sectors and ndustres, whch s fxed. In mport competng ndustres, N N j s total employment n these, j = { producton, clercal} (7) L j L j,, j = { producton, clercal} (8) w j cp w j, Equatons 5 to 8 descrbe a set of smulatons termed sector specfc (pn,cl) n the results secton. A fourth set of smulatons wll be performed where all labor types are sector specfc n mport competng ndustres s termed sector specfc (all). Wth ths labor market specfcaton, equaton 5 s dropped and equatons 6 to 8 apply to all labor types. Intermedate nput consumpton s set as a constant elastcty of substtuton (CES) aggregaton of domestcally produced and mported commodtes, allowng mperfect substtuton between the two commodtes, wth a dfferent degree of substtuton for each type of commodty, as reflected by the value of elastcty used. The producton functon also allows for substtuton between smlar types of labor and captal as well as between labor and captal n general. Thus: 10

11 ( [ α 1 )/ ( )/ ( 1 α ) 1 ] σ σ σ σ σ/( σ 1) d d INT = A D + M (9) where A = scale parameter, α d = share parameter for domestcally produced commodtes as a share of total commodtes avalable n the domestc economy (0<α d <1), and D and M are domestcally produced and mported commodtes, respectvely. The elastcty of substtuton between domestcally produced and mported commodtes s gven by σ. The domestc prce of each composte commodty (P ) s a CES functon of the domestc prces of mported (PM ) and domestcally produced goods (PD ): ( [ α 1 )/ ( )/ ( 1 α ) 1 ] σ σ σ σ σ/( σ 1) d d P = PD + PM (10) Trade s governed by the small country assumpton, mplyng that the domestc economy s a prce taker and there s unlmted supply from the rest of the world (ROW) for mports at the gven world prce. The domestc prce of mports s gven by: M PM = PW ( 1+ tm ) ER (11) M where PW s the world prce, ER s the exchange rate, tm s the tarff rate on mported commodtes and the bar sgn ndcates that the varable s fxed. Export prces are determned n a smlar manner: X PX = PW /(1 + te ) ER (12) where te s the export subsdy rate. Table 5 shows exports, mports and mport tarff levels by commodty. Two products (paper and metals, and chemcals) clearly domnate the structure of Indonesa s mports, and both these products have a reasonably hgh tarff appled to them. The largest tarff s appled on textles (16.0%) but mports of textles are neglgble. Indonesa s exports are domnated by four products at ths level of commodty detal: mnng and quarryng, textles, constructon and chemcals. 11

12 Table 5: Trade Data n the SAM (Bn Rupah, 1993). Exports Imports Import Tarff ag-fc Agrculture-food crops 172 1, ag-oth Agrculture-others 1, mnng Mnng and quarryng 13,252 2, foodpro Food processng 6,558 2, textle Textles 13, constrc Constructon 14,667 4, pap-met Paper and metal products 8,491 34, chemc Chemcals 18,357 18, electgw Electrcty, gas and water trades Trade servces 463 restaur Restaurants 651 hotels Hotels 1,049 landtra Land transports 214 otracom Other transport and communcaton 1,789 bankns Bankng and nsurance 1,736 realest Real estate 2,080 ser-pub Publc servces 2, ser-prv Other prvate servces 1, Household ncomes (Y h ) consst of factor ncomes (wages and rent payments for captal used domestcally and abroad) and transfer ncomes from the government (TGH) gh, domestc frms (TFH) fh, other households (THH) hh and the ROW (TWH) wh, gven by: Y h WL k k + ( PNX WL k k) h = k k + ( TGH) gh + ( TFH) fh + ( THH) hh + ( TWH) wh ER (13) Frms ncomes (Y f ) nclude payments for captal used n producton, transfers from other frms (TFF) ff and transfers from the ROW (TWF) wf, set as a resdual: Yf = ( PNX WkLk) f + ( TFF) ff + ( TWF) wf ER k (14) Government ncome (Y g ) s categorzed nto payments for captal used n producton actvtes, ncome taxes from domestc nsttutons (households, domestc frms and government-owned companes), ncome from ndrect taxes leved on commodtes and transfers from the ROW (TWG) wg, set as a resdual: Y g ( PNX WkLk ) g + thyh + t fyf + k h f = + tdx S PD + ( TWG ) wger (15) 12

13 Household expendture (E h ) conssts of consumpton of composte commodtes, drect tax payments to government, transfers to other household groups and savngs: E = ( C ) + ( t Y ) + ( THH) + S h h h h h g hh h (16) Expendture by frms (E f ) conssts of transfers to households, drect tax payments to the government, transfers to other frms (retaned proft), transfers to the ROW (TFW) fw and savng: E = ( TFH) + ( t Y ) + ( TFF) + ( TFW) + S f fh f f g f ff fw f (17) Government expendture (E g ) conssts of consumpton of composte commodtes, transfers to households (TGW) gh, transfers to the government (TGW) gg, transfers to the ROW (TGW) gw and savng.: E = ( C ) + ( TGH) + ( TGG) + ( TGW) + S g g gh gg gw g (18) Transfer payments from the ROW to households are set exogenously (shown by a bar sgn on the varables n the equatons), the transfers to government and frms are set endogenously (as resduals). Ths s consstent wth the behavor of domestc frms as well as the fscal polcy of the government; both rely on foregn sources for fundng ther defcts. Transfer payments consst of foregn loans, grants and other transfers. Total savng n domestc economy conssts of household savngs (S h ), frms savng (S f ), government savng (S g ) and captal njectons from the ROW (S w ). In equlbrum, total savng equals total nvestment, whch s dstrbuted to each sector based on fxed shares. Aggregate fnal demand (total fnal consumpton of composte commodtes) s gven by: C = Ch + Cg + I (19) where C j = δ j (1 MPS j )Y j, j = h, g For non-agrcultural and non-producton workers n Indonesa, wages are set n compettve markets and reflect the margnal product of labor: n D k) = Wk wth L = =1 PN ( X / L L and L = L D S (20) For labor n the agrcultural sector and producton [manufacturng] workers, wages are fxed and the last part of equaton (20) becomes L D L S where L S L S k = k k < * k. Thus, allowng for unemployment n the agrcultural sector and among producton workers, D and S n the equatons above refer to demand and supply whle W k s the equlbrum wage. The balance of payments equlbrum equaton s gven by: 13

14 PW M + ( TSW) + ( TGW) + ( TFW) + ( RMTW) sw gw fw kw PWE E + ( RMFW) + ( TWH) + ( TWF) + ( TWG) wk wh wf wg = (21) The left hand sde of the equaton above s the ROW revenue, consstng of mports, captal outflows, transfers from government and frms and captal payments from foregn captal used n domestc producton to the ROW (remttances). On the rght hand sde s the ROW total expendture, coverng exports, captal payments and transfers to domestc households, frms and government. The balance of payments can be fxed or resdual, dependng on the assumpton about the exchange rate. In the fxed exchange rate case, the balance of payments defct s resdual to clear the market, whle n the flexble exchange rate case, the defct s fxed. Table 6 shows the part of the SAM where factor of producton rows ntersect wth sectors. It therefore shows factor employment by sector, wth payments made by sectors n the columns to factors n the rows. For example, the frst four rows relate to agrcultural labor (waged rural, waged urban, non-waged rural, non-waged urban). These labor types are employed only n the frst two sectors (agrculture-food crops and agrculture-other), and the non-waged rural component s clearly larger than the other components of agrcultural labor. Table 7 shows aggregated employment of factors of producton for sectors grouped nto net mportng and other sectors. The net mportng sectors can be seen n Table 5 to be: agrculture-food crops, paper and metal products, chemcals, and all servce sectors wth the excepton of the non-traded utltes sector. As Table 7 shows, these net mportng sectors are partcularly large users of clercal and professonal labor and are more ntensve users of agrcultural labor than other sectors. The other sectors are more ntensve users of producton labor and captal than the net mportng sectors. Table 8 shows a further analyss of factor payments by ndustry, wth employment of factors (n columns) at the aggregated level gven for each ndustry (row). From ths table t be dscerned that the more ntensve use of producton labor by the other sectors n Table 7 s largely due to the textle sector, whch uses 30% of all producton labor n Indonesa. Ths sector s notably an exportng sector. The more ntensve use of clercal labor n mport competng sectors can be seen to be due to most of the employment of ths type of labor occurrng n servce sectors. The two mport competng sectors that are protected by hgh tarffs (paper and metals, and chemcals) use relatvely small quanttes of clercal labor. Table 9 shows the dstrbuton of factor payments (columns) to households (rows) n the SAM, wth factors of producton aggregated as n prevous tables. In general, hgher ncome households (.e. groups 7 and 10) receve a hgher proporton of ther ncomes from captal, professonal labor and clercal labor than other households, and lower ncome households (.e. groups 5 and 8) receve larger proportons of ther ncome from producton labor, although these generalzatons have ther exceptons; for example, the hgh ncome rural group (7) receves a hgher proporton of ts ncome from producton labor than any other group. 14

15 Table 6: Factor Payments n the SAM (Bn Rupah, 1993). Ag-fc ag-oth mnng foodpro textle constrc pap-met chemc electgw trades restaur hotels landtra otracom bankns realest ser-pub ser-prv lbagwg-rl 4,061 4,981 lbagwg-ur 894 1,046 lbagnw-rl 21,878 6,013 lbagnw-ur 1, lbpnwg-rl ,685 2,814 5, , , ,084 lbpnwg-ur ,025 6, ,415 1, , ,122 lbpnnw-rl ,529 lbpnnw-ur lbclwg-rl , , lbclwg-ur , ,799 1, , , lbclnw-rl , lbclnw-ur , lbplwg-rl , lbplwg-ur , lbplnw-rl lbplnw-ur cp-land-ag 3,823 13,023 cp-own-hs 8,953 cp-oth-rl 1,827 5,722 1, , , ,886 1, cp-oth-ur 736 2, , ,696 2, ,627 1, ,258 cp-pdom 5 2, ,807 9,450 1,917 4, , ,685 4,125 1,876 1, cp-gov cp-foregn ,798 1, , , ,961 15

16 Table 7: Total Factor payments by net mportng and other sectors (Bn Rupah, 1993). Net mportng Net mportng Other sectors sectors sectors Other sectors (Bn Rupah) (Bn Rupah) (%) (%) Agrcultural 27,947 12, Producton 18,277 24, Clercal 53,730 2, Professonal 10,756 1, Captal 82,087 74, Total 192, , Table 8: Factor payments by Sector n the SAM (Bn Rupah, 1993). Agrculturatoonal exports mports Produc- Profess- % of % of Clercal Captal Total ag-fc 27, ,840 31, ag-oth 12, ,236 29, mnng 2, ,990 31, foodpro 5, ,028 17, textle 12, ,448 26, constrc 2, ,345 6, pap-met 3, ,500 10, chemc 3,059 1, ,147 23, electgw ,323 3,247 trades , ,674 37, restaur 55 2, ,167 7, hotels ,374 1, landtra 4, ,576 11, otracom ,082 7, bankns 95 4, ,250 13, realest ,760 13, ser-pub 1,385 12,018 8,394 2,524 24, ser-prv 4,431 1, ,194 8, Total 42,952 55,834 11, , ,

17 Table 9: Factor Endowments n the SAM (Bn Rupah, 1993). Produc-ton Labor types Clercal Agrcultural Professonal Captal 1. Agrcultural employee household 3, ,893 8, Agrcultural small farmer household 13,441 4,071 1, ,609 38, Agrcultural medum farmer household 5, ,020 10, Agrcultural large farmer household 9,060 1, ,757 15, Non-agrcultural rurallow ncome group 1,454 2,264 2, ,539 10,178 household 6. Non-agrcultural ruraldependent ncome group ,171 3,285 household 7. Non-agrcultural ruralhgh ncome group 6,481 13,392 12,124 3,011 8,772 43,779 household 8. Non-agrcultural urbanlow ncome group 474 3,509 6, ,071 22,384 household 9. Non-agrcultural urbandependent ncome group 25 1,451 1, ,214 5,821 household 10. Non-agrcultural urbanhgh ncome group ,328 30,077 7,057 15,907 67,761 household Frms ,020 66,020 Government ,250 4,250 Total 40,337 42,953 55,834 11, ,223 Total 17

18 5. Results The model descrbed above s subjected to a number of smulatons under dfferent labor market assumptons and macroeconomc closure condtons. In every smulaton reported here the results show the effect of complete removal of all mport tarffs on all commodtes. The smulatons are desgned to show the relatve magntude of factors that nfluence the sze of the net gans from trade lberalzaton. Table 10 shows the equvalent varaton of tarff removal under the dfferent labor market condtons and closures. The dfferent labor markets and macroeconomc closures are as follows. Four types of labor markets are used, from compettve to sector specfc (all) wth ncreasng levels of labor mmoblty and wage rgdty. In the compettve case for each of the sxteen segmented labor markets, labor can move freely between ndustres and real wages adjust to clear markets. There are no opportuntes for ncreases or decreases n unemployment overall for any labor category, but ndvdual sectors can change ther employment levels at the market clearng real wage. No movement s allowed between labor market categores n ths or any other type of labor market. In the moble, mnmum real wages case the markets for the four types of producton labor and four types of clercal labor are changed from the compettve case so that mnmum real wages exst n these eght categores of labor. A fall n demand therefore leads to unemployment. In the sector specfc (pn,cl) case the markets for the four types of producton labor and four types of clercal labor are further segmented. Employment of these labor types n mportcompetng sectors s sector-specfc wth a mnmum real wage for each combnaton of labor type and sector. Full moblty exsts wthn each of the sxteen labor market segments n other sectors, and for agrcultural and professonal labor n the mport competng sectors. In the sector specfc (all) case the markets for all sxteen labor markets are further segmented, wth employment n mport-competng sectors beng sector-specfc wth a mnmum real wage for each combnaton of labor type and sector. Full moblty exsts wthn each of the sxteen labor market segments n other sectors. The welfare effect of trade lberalzaton wth these four labor markets s shown n the frst row of Table 10. In ths row, smulatons are conducted wth any reducton n government revenues leadng to reductons n government consumpton, whch s ncluded n the EV fgures; wthn these smulatons there s therefore consderable change n the composton of consumpton, wth falls n publc consumpton and ncreases n prvate consumpton. Wth compettve labor markets, tarff removal leads to welfare benefts; wth mnmum real wages but no mmoblty of labor, ths beneft s reduced slghtly. Wth the two types of sector-specfc labor there are large reductons n welfare from trade lberalzaton. The second row of Table 10 shows the welfare effects of the same smulaton under the same labor market condtons where the government changes ncome tax to ensure revenue neutralty. Government consumpton does not change n these smulatons. In the compettve labor markets case, revenue neutralty reduces the welfare benefts of trade lberalzaton but the benefts are stll postve. Under the sector specfc labor markets revenue neutralty reduces the magntude of the welfare loss, although these losses are stll 18

19 much larger n magntude than the effects of movng to revenue neutralty wth compettve labor markets. Table 10: Equvalent Varaton (Bn Rupah) Labor markets Compettve closure Moble, mnmum real wages Sector specfc (pn,cl) Sector specfc (all) Government consumpton ,010-1,852 Revenue neutral ,004 Domestc dstortons removed Table 11: Employment (Bn Rupah constant value) Labor markets Compettve Moble, mnmum real wages Sector specfc (pn,cl) Sector specfc (all) closure Government consumpton ,772 Revenue neutral Domestc dstortons removed Conclusons Ths paper has examned mport tarff lberalzaton n the presence of labor market dstortons (real mnmum wage restrctons) and moblty restrctons. Theory suggests that n ths case there s the possblty of trade lberalzaton beng welfare worsenng, but prevous emprcal work on labor markets and trade lberalzaton has concentrated more on the presence of market power and unonzaton. The lterature also suggests that the presence of domestc dstortons and the need for revenue replacement as reasons why tarff lberalzaton mght be welfare worsenng. The emprcal nvestgaton has been through a computable general equlbrum model of Indonesa, wth dfferent labor market assumptons. The results have shown that the presence of sector specfc wage rgdtes has a sgnfcant effect on the net gans from tarff lberalzaton. Revenue neutralty has a small and negatve effect on the net gans from tarff lberalzaton. The presence of domestc dstortons has a larger effect on the net gans from tarff lberalzaton than revenue neutralty, but s much smaller than the effects of sector specfc wage rgdtes. Mnmum real wages have lttle effect on the net gans from tarff lberalzaton where labor s moble between sectors. The magntude of welfare losses when labor markets are rgd (sector specfc) and where reductons n employment n mport competng sectors feeds through to ncreases n unemployment are much greater than these other sources of welfare loss. Overall, these results provde support for the need for measures to ensure greater labor market flexblty when undertakng trade lberalzaton. They also mply that where labor market flexblty cannot be acheved trade lberalzaton may be detrmental to welfare. 19

20 References Bussolo M., Mzala A., and Romaguera P. (2002), Beyond Heckscher-Ohln: trade and labour market nteractons n a case study for Chle. Journal of Polcy Modelng, 1-29 Central Bureau of Statstcs (1996), The Indonesan Socal Economc Accountng System (Sstem Neraca Sosal Ekonom Indonesa) 1993, Volume: I and II, Jakarta, Indonesa. Clarete, R.L., Whalley, J. (1988), Interactons between trade polces and domestc dstorton n a small open economy. Journal of Internatonal Economcs 24, Dervs, K., de Melo, J., Robnson, S., (1982), General Equlbrum Models for Development Polcy. Cambrdge Unversty Press, New York. Deverajan, S., Ghanem H., and Therfelder K.(1997), Economc Reform and Labor Unons: A General-Equlbrum Analyss Appled to Bangladesh and Indonesa. The World Bank Economc Revew, Vol. 11(1) Harrson, W. Glenn, Rutherford, F. Thomas and Tarr, G. D. (1993), Trade Reform n the Partally Lberalzed Economy of Turkey, The World Bank Economc Revew, Vol. 7(2) Konan, E D., Maskus E. K. (2000), Jont trade lberalzaton and tax reform n small open economy: the case of Egypt. Journal of Development Economcs, Vol 61, Ylmaz, K., (1999), Optmal export taxes n a multcountry framework. Journal of Development Economcs, Vol