World Bank Reprint Series: Number Ninety Jagdish N. Bhagwati and T. N. Srinivasan Trade Policy and Development Reprinted from Iuteriiatio,ial Ecotivw.ic Policit: Thieory an)d Evidentce, Rudiger Dornbusch and Jacob A. Frenkel, eds. (Baltimore: Johns Hopkins University Press, 1978), pp. 1-38 CHAPTER ONE Trade Policy and Development JAG.DISH N. BHAGWATI T. N. SRINIVASAN The interaction between international trade and development is a subject of such complexity and importance that it has rarely ceased to attract the attention of economic theorists, analysts of the world economy, and de- signers of the international economic system. Inevitably, therefore, it has drawn into its fold and its many controversies some of the best minds of each generation of economists: dating from Adam Smith, David Ricardo, and John Stuart Mill, down to Alfred Marshall and, in our own times, to Dennis Robertson, Ragnar Nurkse, Jacob Viner, Gottfried Haberler, and Arthur Lewis.' There are far too many questions that the topic raises: witness, for example, the elegant recent review by Carlos Diaz-Alejandro.2 We propose rather to concentrate on two sets of analyses that have currently been the focal point of theoretical, empirical, and policy discussions. In Section I, we review the evidence that is currently available on the question that Nurkse had raised in the early 1950s regarding the optimal trade and developmental strategy for a postwar LDC (less developed country) planning to accelerate its economic growth. Arguing that the nineteenth-century mechanism of trade as "an engine of growvth" (in Dennis Robertson's catching phrase) was not available to present-day LDCs for a number of reasons, he noted that a policy of "balanced growth," reflecting essentially domestic demands, was inevitable. Remarkably, he did contrast this, what we would today describe as an IS (import substitu- tion) strategy, with the policy alternative of promoting new, manufactured exports, a la what we would today call the EP (export promoting) strategy: I Cf., in particular, Jacob Viner (1953), Ragnar Nurkse (1959), Gottfried Haberler (1959), and W. Arthur Lewis (1969). 2 Carlos Diaz-Alejandro (1975). The views expressed in this paper are personal and do not reflect those of the institu- tions to which the authors are affiliate.', Thanks are due to the National Science Foundation (Grant No. SOC77-07188 for partially supporting the research under- lying this paper. Section I draws on earlier work for UNCTAD. 2 Jagdish N. hlheiianui and T. N. KSrinhisan but felt that the latter offered little promise, as it was likely to run into DC marlcet disruption-related trade restraints, as with textiles. As it happens, the postwar period did witness both sets of policies, starting in the early 1960s, and we have the evidence of two major projects on these issues so that we can, with hindsight, see which strategpy was the more successful ex post. Our analysis wiii not merely review tlhese empirical results, it will also indicate the unsettled questions on wlhich only future research can generate persuasive evidence. Therefore, while Section I focuses principally on the trade policies of LDCs, in regard to the optimlal methods of utilizing t'le available trade opportunities, we turn in Section II to the complementary subject of how those trade opportunities ought to be defined. in particular, we will con- sider two subjects of recent policy interest, namely, (i) the tlheorctical and policy issues raised by the problem of market disruption-related threats of trade restrictions on imports of manufactures by DCs; and (ii) the recent demands by LDCs, as part of the New International Economic Order (NIEO), for commodity agreements. The reader should be forewarned that this chapter is therefore a selec- tive review of the major trade-and-dcvelopmental policy issues; it is cer- tainly not intended to be an exhaustive guide to the voluminous litcrature on the subject.3 I. PROTECTION, INDUSTRIALIZATION, EXPORT PERFORMANCE, AND ECONONIIC DEVELOPMIENT We turn now to the "foreign trade strategy" issues that were admirably, and with. much prescience, raised by Ragnar Nurkse.4 Cairncross, in an insightful review of Nurkse and HIaberler, having reviewed thc general argumentation couched in terms of trends in world trade and wihether these justified elasticity pessimism or optimism and whether these in turn re- quired balanced growth or not, summed up as follows:5 At the end of it all, the reader may still feel that neither Nurkse nor Haberler has settled the primary isstue: how far a shortagc of foreign exchange (con- trasted with capital. skilled lahbor, land, etc.) is a limiting factor in economic development. The majority of the un(der-developecd couintries are m0onocultutres, dependent for their camnilngs of foreign exchange on a single coimimlodity (or at most two or three). These earnings are highly inelastic excelpt whlien exports of the princilpal commodity form a small fraction of the world's conunimption. At the same time, nearl) all the plant and machinery that they rcclqire has to be imiiportedl, so that the scale of industrial investmcnt is limited by the foreign 3 The m:iny distingigishiedl resewarccirs %h%hoe conlributions are not noted explicitly should equally take note of this facic! 4 This sectioO draws heavily on J. llhagwi ai i ( 1 976), S Cf. A. K. Cairncross (1960, chap. 12, p. 208). Trade Polici, Vind Dei-elopment 3 cxchaini_e availablc to pay for it. In those circumstances, what should be the policy of a country seeking to accelerate its development? We know what most countries have done; it would be interesting if we could be told, by an economist of the standing or Nurkse or Haberler, what the results have been and what thde shotuld have done. Modesty should prevent us from laying claim to the "standing of Nurkse or HThaberler." However, we are certainly now in a positioln to respond to C(airncrosss's query, tlhanks principally (thiough not exclusively) to two major projects on foreign trade regimes and their effects on economic development: the O.C[) project, directed by Ian Little, Tibor Scitovsky, an(d Maurice Scott, wlhose results have been k;nown since the early 1970s; and the N13BER project, directed1 by Jagdish Bhagwati and Anne Krueger, whiose reslults have nosw become generally available.", In particular, we now have statistical evi.ence and ecoinomilic argu- nicnzittioin oIn the followNing, related issues: (i) The degree and structur, of protection that have been practiced in the developing countries; (ii) The analvtic.al rationale for relating this to the pattern of industrialization and C%port performance of these developing countries via t?te effect on the relative incentives for import su1bstilution and export pronlotion; (iii) The statistical evidence for the argument that such incentives affect the pattern of induistriali/ation and export perorrnmance in the developing countries; and (iv) The qiestion whether, and if so why, better export performance is related to better economic performance. A. 71i/u th,Qreo atnd structure of pro,,eti,n: Concepts In ;analyzing protection, one needs to distinguish among three sets of concepts: (i) Trade policy protection versus domestic policy protection: The Organization for Economic C'ooplel.tion ani D)evelopment (OECD) prt)ject (oruani.ned by the OLCD) Developmenit Center) covered Bradil, India. Mfc\icoi, Paki- stan, the Philippines, and Taiwan; whereas the NBER project covered ten countries: I.Al, Chile, Colombia, FL! pt. Ghana, India, Israel. the Philippines, South Korea, and Tuirkey. T-he NBiR project (National Bureau of Economic Research [1975, 1976]) essentiallv takes off from the OECDI) proiect. in extending the analysis to much more systematic attention to the exchange-eontrol aspects of the foreign trade regimes in the k1cvt:opinp COLIIrie'C: it also considers dyn:nui; aspects of the trade regimes and the problems of trade libei ;ili,ation. The OE'D studies (Organization for Economic (Cio,lceation aind Ievelopment f1'9701 have been published in five country volumes an1di otne overall votume: ( 1) I ittle. Sciiovsks. and Scott (1970), overall; (2) leigs- man 1dt-10). I;i/d; (3) Bhagwati and Desai (1970), India; (4) Lewis (1970), Pak.istan; 15) Hsing. Pow\er, and Sicat (1970), 'ITaiwan and the Philippines: and (6) jin!! t 1970)). Mlexico. The NBER stildies are being publishedI in ten-country volumes and wt) sm nii hevi l triitmwv, the following are :itrecly published: (1) Krueger (1970), 'Turlkev: '2) \Mic ick I (1975). Isriael; (3 B Baldwin (1975). the Plhilippines: (4) I.eith 11 i',7, (ihana:; 5) r ink, W'.tphl.il, and Kim (1975). Soutlh Korea, (6) Bhaigwati antdtl Srinivastin (1975), Indiai; (7) Iiansen and N.-lnhshibi (1975), Fgypt; (8) Diaz- AXci.indit to 1)71), C olombia: and (9) B3ehrman (11)766), Chile. Bh:1mgw:ti (1978) and Kint,.rr I )Is) h1a ve wsritten two separate synthes,is vtolulmiices, focusing on (lifTerent partk of the pio.c.t results. 4 Jsg.'lisIh N. Bheigwati tt;id T. N. Sriniveisan an activity may be protccted through tariffs and quotas (QRs), on the one hand, or through domestic subsidies, etc., on the other hand;7 (ii) Tariffs versus quota protection, or alternatively, explicit versus implicit protection: within trade policy, we can distinguish between protection furnisshed by tariffs or by QRs; V turn, QRs may be specifically designed for protecting the activitv in (quc a . or they may be a result of an ovCLrvalueUd exchange rate that results in the use of QRs as a teclhniquc for balanicing international accounts- and (iii) nominal versus effective protection: the protectioni mlay be measured in the conventional way as on goods and services (i.e., as nominal rates) or on l .;lue added (i.e., as effective rates). It is clear from thCese Conceptual distinctions that, in examiniing protec- tion, the international economist aims at coniparinig the total structure of incentives (to import-competing and other activities) as contrasted with those that would be provided under a regime of lhissez, faire, or whalt has been more aptly described as a regime of unified exclhanige rates." Thus, the incentives for domestic import substitutes that would follow from over- valued exchange rates, and the attendant implicit tariffs implied by QRs, must be allowed for; and so must the use of dlomestic subsidies, in several forms, to d]omestic productioni. The early studies of protection in the LDCs allowed for neither the use of ORs nor the presence of dlomiiestic taxes and subsidies." However, the well known IBRD (International Bank for Recoln- struction and Developmentt) studies,1" as also the NBER stullies. typically attempt to allow for implicit tariffs (i.e.. ORs) ancd, occasionally and partially, for indirect taxes iwnofar as thev affect domestic prices of inu)Lts or differently affect import sUbstitutes and imports. The use of implicit tariffs involves, typically, the conversion of import premium data or, zflternati%ely, data on differentials between domestic and c.i.f. prices of comparable items into equivalent lariffs, This procedure is subject to both empirical and conceptual difTicuiltics, a few of which nmay be mentioned here:" (i) quality differenices exist betwveen imllorts and import substitutes, which imply that some of the differential in prices, when used for estimation, is attributable to this factor; (ii) frequently the OR regime may be so restrictive that imports are prolhibited and there is, in consequence, often Ino easy aind reliable way to get comparable c.i.f. prices; The choice between these alternative instriunments of protection has, of cou rse bIcen 'h. subject nmaitcr of contribut ions lby Mfeadle. B3hlatmta. Ramaswami, Srin:iv;isati C'orlen. Johnson, and other theorists of tradc polic%. We do not dkciishe lile issuies here. This phi sine was uised in 1Bhacv.tli ( 1968). 9 This wam true of the early eslirn:lts for Pakistan, for ev\linlple. by Soligo anid. Stern (i 965). '1 Cf. Balassa and Associates ( 1971). '' Cf. the treatment in Balassa (1971 i.hap. 3): also consult Bhatgwati (1978, chap. 5) for a niore detaiiled dicSuIssionI. Trade Policy (an eltL 'pnu'ti /ent 5 (iii) if perfect competition in quota allocation and use, and in foreign nmd domestic supply and demand, cannot be assumed, the import prerni iii cannot be meaningfully converted in general into an equivalent iniplicit tariff; (iv) where domestic licensing conitributes to the generation of monopoly profits, the inmport prenmium will reflect this factor as well and hence is not interlrrctahle as protectiorn from tlle viewpoint of inferring resource allocational shifts; (v) in the nature of the case, QRs will be, and are, associated with fluctuating premiums, so that it is extremely dillicuilt to arrive at one set of lprenliums to convert into implicit tariffs, and totally misleading to use one suclh set to indicate the tariff structure (which is to be taken, in turn, to indicate the struicturi-e of price incentives to domestic protection). Given these, and other, serious shortcomings, it is best to treat thie re- sulting estimates of the implicit tariff structure as desL_riptions, in varying degrees of loose approximation, of the pattern of incentives that may be appearing in the developing countries in question, thanks to QRs.'2 Similarly, the description of the tariff structure in effective tariff, as distinct from nominal tariff, terms raises both conceptual and empirical questions.l" In p)articular, it is not po;sible to utilize the colpUted effective tariff rates to indicate in an unambiguotus Faslhion the direction of change in resource allocation that is resulting from the set of nominal tlriffs thlat we use to conmpute the effective tariffs.1-4 In liglht of these problems, it is best perhaps to regard tlle effective tariff structures that lhave beenl estimated inl the OECD, NBER, and IBRD studies, amnong many others, as also essentially descriptions that, in a very loose way, indicate the difTerential nature of incentives that the combination of tariffs, QRs and (in some instances) dom-iestic stibsi(lies and taxes seem to throw up in the ecoiiono)m being studied. While the tariff st rttcturies are deteine(l an(l estimated in the mnilllner incdi- cated above, and mlust be intcrprete(d with great caution, the concept of '2 Tn some of the studies. the proiection granted is broken down into that re'.ilting from explicit tariffs and the idditional element dtue to QRs, when the implicit tariffs exceed the e'.plicit tariffs. Cf. the 13haign.wati-I)esaii-Panchimtukhi estimates in the OECD Tndiia volume ( 1970). and the Leith estimates in the NBER Ghana vohLinc (1975). I: For a detailed consideration of the empirical qtuestions. see Balassa ( 197 1, chaps. 3 and 4); for conceptual problems, see in p;trHiCilI:II the couiit ibutions by Bruno (1973) and BIh.gwvtli and Srinivasan (1973), to the "Journ;lI of Tnternational Economics Symposium on the 'Iheory of IFTeclive Protection in General FLuilibrium" (1973). 4 "This ptoint has been established, and sufficient conditions undler which the directioni of change in resouirce allocation mu.uv nonetheless be inferred, investigated. by inter- nationol trade theorists roceLitlk.. A good starting point for readhing this literatuLite is in the Jourrn. I of International Fconomiics Svinposiuni on Ihe 'theorv of F'ffective Pro- tection in 'Jeneral Equilibrium- 1973). The uiisi,tic.il evidlence on this question, discussed in the teXt al0ove, also corroborates this iheoret(icl skepticism, while indicait- ing a few of !he reasons for -t. For fuiller dliscuission, ee Bhllhag%,ti (1978. chap. 5). 6 JaIgdish N. a/mgiraii lndel T. N. Srilirasain the degrce of protection rcflects essentially a weighted avelage of suchi tariffs.'15 In addition to such averages, sone economists have also attempted to adjjust the average degree of Ipimotectiori O^w ard by ar,guing that the removal of the tariffs would generally gen-ate a balance-of-payments deficit that would halve to be eliminated by devaluing the exchange rate. The devalation, in tturn, N\otld impiil that thc (lomestic price of the im- ported coniiimodities would fall by less than the tariff remnov'al would inmplv.'"; W'hile this is a theoreticailly correct thling to dio, if onc is interested in what hiappens (net) to the nlominal domiiestic pirice of importables as a result of the tariff inposition. T thle practicail e.stittioti of thi.s adjustmelnlit factor, as attenmllled in several of the 13RD studies, relies oTI prO.C'(IcId'eri that can be Ldi'ende(dLI on ly by making highly rcstrictive assumptions. ' Finallv, in anticipation of the anall is in Subsection D on the inter- aictionl between protectioni and export perfOrmIan;11ce, it nt;v be InedII thaIt the degree of priotectioni is often taken as a reasonable explanatory variable for exlport performance. Additionall\ , three other Concepts are u5Cd, fte- quently as c.xplanaitory \ a riables in analyzing export perforninice. two relating in somne falshion to protection in the broad sense deIined above. First, the ratio of the effective eclian-c rate oni exporls (ElKER,) to that on imports ( EER ) is taken as an ind(lex of iow, far the average exlports are profitLblc relative to average import-conlpething p)rodutction. t' Second, ',The nominall tat illis may bre weighted bv shares in imports or1 in (idnoestic pio- duction: Cileit iL; inirills may be v%eiehliid bh shares in nominall value added of the activities in qIuestion. 1'; Thius a removal of an average tariff of 50 percent, resulting in a devaluation of 20 percent, woultld imnpiv a net, idiutkstd average tariff of 30 percenit: the dlontic,i, nominal price of the imr.orted items sNould fall only hy 30 percenlt when the tariffs were removed and the b alancc-o'f-p:nic tnertt position left unchanged. Note that it would reqtuire, even in theorn, speCial jestrictions to infet fronm suich a (net) effect on the averar,e domestic nominal;) price of inmportables tlhat. r0, example, the share of trade in national incotmie is redtuced by suclh tt t.fl Ill'. '9 Cf. Balassa ( 1971, Appendi. 3 ) for the specification of the procedures 1used, and an excellent theoreticail surveN of them in D)ornbusch ( 1974). Akside fronm the theo- retical objections, spelledi ouit by Dornbiusch, one might note also the general incon- sistency between uising less than infinitely elastic foreign el:t.ticities of denm.and for exports to compute e.xchange riate change and constant international price; for comn- puting effective protection (as requliredI by the fact that the general equilibrilinm analyses of effeCtive prOttion in the :fi il;ible literiture universally make the ws'tnnip- tion of coistant international prices). VIThe effective e-chliatge rate on es poit is defirined as the units of doniestic ciurrenicy tihat can be obtained for a dollar's worth of e\por ik. taking into account export duties, s%1`;id ie' anId sutrch,irLvc. speCitl : nec rates, itiptit Subsidies telated to c \ports. etc. The effective exchange rate otn imports (ER., ) is corieItpondlitnvLk defined as the utnits of 0oncICtl eiL currency that wouild be paid( for a dollat worth of imlports, taking into aiccoti tariTs. sutrJmrarr-es. interest on advance dteposits, etc. In principle the t EER, shoUild incl(de premia on import lienttCe: however, in the N I11R stuoies, the lJ'fR. was dIeltiiedi exclusive of Oheni, for the simIple reason [hat fiItr many' cotintries no relialble dait on implitii I premnia cou]ld be obl.titled citliel directily or via suitable snurvevs of c.i.f. and retail prices. As stated later, the ratio fER/1EER,, as an index of eNport bias dates back to before even the OlFC( I) pr ieci bi tdies and was use(d. % ilhoutit detailed qu.intifict;tion, in Bh:ugwati ( 1968). Tr-ade Policy tnel DIevelolmnent 7 for any one activity, the efTective tariff rate as applicable to production for domestic sales may be compared with the cffective tariff rate as applicable to exports .and the ratio thereof, when exceeding unity, would be described as the "export bias" characterizing that activity."11 But, if the EER,. and EFR7, are defined (as they vere traditionally in the 1960s in India) as including the incentives and disincentives on OlltpLutS as also those rel:ted to inlputs,21 theni the definiitionl of cxlport bias as the ratio of elfTective tariffs in export and doniiestic markets is identical witlh the better-k nown and earlier definition of export bias in terlis of the ratio IIR,F'EAR AR,,,.R'. Third, we may note the concept of real effte, ve exchange rates, or what the NBER project calls the price-level d11efltedL efTectixe cexchlange rates (PLDEERs). In contrast to the EER !EI+R_,, ratio, the PLDEER. would show the relative price of the ex\portables to home ooods (as (listinet from importables) and hence capture a dlifferent CielIemelt of thle total picturle regarding incentives to produce for exports. Furthernmore, the N1I3R project utili,ted, in some studies, the concept of puLrcChasiing-power parity effective e\clhnoe rates, PPPE'E,Rs, WhiCh aldjLust also for changes in the foreign price level. We shall return to these conceptis wlhen we examnine the relationship of protection with export perforimance. For the presnt, it is iniportant to distinguish broadly betwcen two basic implications of any observed tariff structure: (i) the import-conipeting activities are beinci.i broadly speaking, encouraged relative to whiat tihe absence of protection would imply: this is what might be called the '-deoree of inmport sushtitution" aspect of the protective struictuLre; and (ii) there are (usually) difTerenitiail lariffs on, and therefore differential incentives to, different activities withlin the imnport- competing sectors: this is what couIld be cailled the "pattern of import substitution" aspect of the protective structure.:; I an approximate fashion, we can then argue that the degree- of-protection concept Corlrsp)onLds to 2t1 This concept was used in the International Batik for Reconsu ilicr ini and I)evelop- ment (IBRD) studies and is used in the South Korean stuidy of the NBFR proiect: it was not used in the OEC'D project at atl. However. as noted below, it redtuces in effect to the tproperly defined) ratio of EER ER., in any case. 21 Thus, for e\ample. exporters in India typically receive imported materials at international priLes. so 11h;t FER, is definedl as in1lns;iVe of the implied( subsidv from this seheme. See Blhagw:mui and Srinivasan 11975), for e\aimple. 2-'T'hiS is seen ieadil b ti:niing that. fo;r the ustial notation, export bias unider thle former concept amounts to: t . ~ .'2 \lhere thle Superscripts x andl relate to) export and doiestic markets i k,pcct 1' el.\. and the latter amouLits to: 1 z lz3.. t~ (2 .: 'i - t: , ' I. 24Ihe terminology of deglence and pattern of imnport stibNmitituion was introduced in Bhagwati t 1972). 8 Jhigdi.shi N. Bheigitvii antd I' N. Srinivason the degree of iMport-SUbstitutlion aspcct, and the struct me-of-protection concept corresponds to the pattern of import-substiLution aspect, of the process of economic expansion and, in effect, of indUstrialization in the LDCs. In the rest of this section we will essentially deal with both these aspects: Subsection C will consider the pattcrn of import SLIulftiltlltt,io whereas Subsection D, in considering export pcrformnance. wil simllUl tancously imiply consideration of the question of the degree of import substitution. Prior to these analyses, hioever, a brief review of the empiricall stuLIliCs on the degree and structure of protection is presented in the next SuhsecCion 13. B. The degree and structure of protec(t'n: Evidence Although botil the OECD and NBER proijects contain, within their more ambitious and wide-ranging frannxcork, estimiates of the protective strUctture, the best-known and standardiied estimates for a set of deCeClop- ing coLntries are to be found in the six IBRD studies for Brazil, C';hile. Malaya. Mexico, Pakistan, and the Pbilippines.-t Essentially, these estimates relate to specitic dater, typicallv deploy the effective tariff concept, and tiiliic impilicit tarill estinmlates (basekl largely on estimated differenitials b.otweert foreign anid domestic prices (f inmports). Net protection estimates (adjusting for exchange rate chatige) are also included. The studies also procedl to present effective protection by export and domestic nmarkets, so that export bias, so (efined, is also tyl)iCally estimated by the authors. The IBRD studies indicated that the manufacturing sector was pro- tected, relative to the primary sector in nearly all the countries in question and, in the case of Chile and Philippines, the averatec tariff rate for mIanu- facturing was fairly sizable. The OECD sVniheCis VOlunme by Little, Scitovsky, and3 Scott also con- tained estimates of avcraLe tariff levels for manufactures that inidiciated] again that the dcgree of protection used for manufacturcs by de- !loping countries was extr enmel)y high: protection being dlefincc as nominal, explicit 'tThe OF:CD and NBFR sud(ies offered much more comprehensivc and (dcl.iilekl analyses of the countries being studied than the 1BRI) Itcdv. In pcrticti.lr, miost of the NBER volumes examined e\port performn.nce in tlepth, systenma!tie.lllc .iy lyce(l the eNolutioi of the c\ehange-control regime over tinme, exa,mined ftllyv the conditions deternmining the ouitcomes of tihcratli/nion attempts (inliditinig pnlincal factors plui the role of foreign aid. etc.) and ilteniple(l (in some easCS) more syslenmlic examina- tion of dynamic argtimentl relating to investment, innovation, savings fornmaion. etc. ancd their interaction with the foreign trale reginme. In all these rcspects, the N}31PR studies were, for the most part, more compreherisi\e and ambitious than other- etlom t. such a.1, for es:cmple. the IB1RD prroject. thomi!h the latter did touch marginally on some of the issues (e.g.. Balassa's brief treatment of 'dynainic" effects in his intro- ductory essays,. relaling however inainly to static, scale economics and co.nmlpetition aspects). The rel.ttionship between the NIBR and tie OECI) projcets. which were both ambritious, has been spelled out above. Trade Policy andl Deveulminc'nt 9 tariffs alone. By contrast, tfley arguied that most of the present DCs had used substantially lower tariffs in the course of their developnment.25` This contrast was sustained by examination of the effective tariffs as well. While the OECD project did note the "';ariahility of protection" to different manufacturing activities, the main focus of lhe NBJF.R stuiLles has been precisely onl this aspect of the overall foreitn trade re-gime. Thus, while stressing the many ditlicul;lies in initerpretinig tlle strulctUre of protectioni, the estimates vere used to u uderli ne the difrent al-i neen tiMes- generating natture of the regime, whilIlc stressing equally the :dnistratie- cum-allocational procedLurcs that led to auLtonIa,ticity of protection, fluctilati nt incelntives [t rolugh 'varying import preiniia reflecting chnaoingo allocationsi and rules, anid numerouis otlher facets withoLut whose adequale understanding the analyst of the cliTects of tariff sr1ICtictres Nouldc he making, at best, misleading ineerences"' While using the estimated tariffs on several manufacturing activities (and, in the case of Egypt, for agri- cultural crops as well-)27 to show the wide dispersioin in the implZliCe incentives, the NBER project also utilized concepts and meCIsuLes StuCh1 as domestic resource costs (DRCs) to indicate rather the varying social rates of return to pro(luction in alternative nmanufacturing actiities. The major conclusion of the NBER stud(lies is that the protective struic- ture, when inclusive of the implicit tariffs (implicd by QRs) undeler exchlange controls, is characteri,ed by considcrable dispersion and unpredictability, and that the effects are to create resource nisiallocation whlosc ilnCidlence is indicated by the DRC-disper.sion obscrved in the lipirical studies.-"' C. Protectioni and pattern of indu.strializati(J) The effect of the structure of protectiorn on the patterni of indius rialila- tion may first be noted, before procceding to consider the effect of the degree and pattern of suclh protection on export perforniance. 2- Cf. Little, Scitovsky, an(d Scott (1970, chap. 5). It should be added, ohos ever. ,hat an unpublished NBER project-commissioned examination of Japanese tariff protection dturing the period of eairl indtustriali/aition, by Tppei Yamazawa, slggests a substantially greater role of t iif, and other forms of protection, than the Little- Scitovsky.-Scott figure indicates for Japan. 2' For dctails. see Bhagwati ( 1978. chap. 5). - Cf. Hansen and Nashashibi t 1975) on FL,% pt. 2'( Of course, if shadow prices for (lomestic factors, and marginal revenuies for com- modities facing dleclining prices as exports increase. are not used to calculate DRC's, they can be red-uced by a simple transformation to effective rates of protection. 29 The stress on dtispersion of incentives is to be found partieularly in I eihli WRliF1 Ghana voltme (197.5) and i n Blha-imlti and Srinivasan's NBER India volume ( 175 ), and is spelled out overall in l1h;uwvali\s N13FR sindlieNis volume (1978, chalp. 5). Balassa ( 1971 ) also stresses the variahility of incetmtives in the tariff structuire hut needs to pay the same degree of attentioti to interpreting it in light of the unprcmliA.:hility, built-in aumtonamticity of future pr'otectioni, and other integral aspects of QR regimes that define the corptext, and hence the trtie nmcaning. of theme tairifT cstinbites. lore is said on this in the next subsection. 10 Jilgelish N. 1J/higivaii caid T. N. Srinivasan 'While it is true in reality thiat protectio7n of the mllanulfalcturing sector, in toto, is supportive of in(lustrialization in the LDCs, it should be nioted that it does not followv that the pattern of nmnulfactUrint produLIction, or import substitution, also is explainable by the pattern of protection that is being measured for the LDCs. Thus, it is tenClpting to argue thlat industries when arrayed in ascending order by their (eflTCtiV'e or nrmilinal) protective rates should also be raneCdL in ascending or(ler by their urowtil rates, or their import-substitution ratios. But there is neitheI analytical, nor0 clear empirical, support for suchl a hypothesis; and, in fact, it iS CenioUraiing that wvhere theory suggests thlere should be no such corrclttioni, thlere is mixed cvi(lence to be fou.lndl in practice as well. Among tlle favorable results for thie hypothesis stated is that for Coloml- bia. Thus, in the NBER study Of Colom1bia, Carlos Diaz-Alcjandro cites the earlier work of HutCheson oni Colombian protection that regresses growth rates successfully on effectivc protective rates."" Similarly, Frank et al., in tIe NBER study on South Korea, repoort on rankl corielation coenicients between various nicasures of effective protec- tion, and of effectiv,e incentives (defined so as to include the effects of tax rebates, credit preferences, and suclh incentives) an(d rCsource-allocational indices such as import-substitution ratios (or export shares for e.xlort induKtrics) and growth contribution."' Thleir resmil ts, however, are generally poor on the import side: tei correlation let ween import ratios anLl leffCtiVe incentives is sienilicant and positiv-e. sugpesting that imnport . suibtit ution had progre;sed the least [ratler than the most] in those sectors that liadi a high level of effective incentivus on donmestic sales, and the correlations between efTective incentives to domestic sales and groxwth contribUtions are not si-nificant, though they are ncgative.32 Additional cross-sectional analssis of ihis variety was also coinductcd by Jere Behrman in hlis NBFR studly of Chile, to determiniile wheiter the price structures created bv the international econiiomiic rcginies were asso- ciated with growth across sectors.:s' He foLun(d a positive relation between orowth in value-added and in fiorsepov.er capacity bctween 1961 and 1967 and the iniplicit taTiff rates (ITRs) for 1967 and also for the incre- mental ITRs betweenl 1961 and 1967. But this relationshipl has little plausibility, as Bchirmiin notes, and may be rationaliicd only by arguimcnt, such as that the ITRs "perhaps . . servcd as signals, however, of tlhe t' The Huichemon estimates of etTective protection use the early Balassa method for treating nontraded goods as enjoying icro protection ia tlicr than as value added ti la Corden. Note, however, that in the case of Chile. at least, the distinction lbetween thle two rne.suIres is not empirically important. (f. Behrman (1976) on Chile. 1 Fraink, Wes(phial. and Kim (1975, chap. 10) on Sotlih Korea. Ibid. ( 1975, chap. 10, p. 36). Cf. Befihrnmai ( 1976, chap. 12) or ('hile. for ftull deciils of this ;inatvsk. Tnrade Policy anid De'elhpmntw 11 government's intentions to favor particular sub-sectors."', Interestingly, Behrman found no evidence for a link between effective rates of protection and growth: in fact, the only significant nonzero correlation coefficient, using alternative estimates, was a negative one betwecni effective rates and growth in productioni from 1953 to 1961.'5 Going beyond the NBER studies, fuLrtlhc nrmore, wve nmay note two suiccess- ful sets of regressiomls: for Pakistan by Guitinhcrfl: an1d for Niger-ia by Oyejide.37 The Pakistani analysis was unsuccessful for import-substitlltion ratios, but successful for growth rates for a 23-industry study. Thle Ni-criian analysis, for 42 industries, resulted in succeCssfulI recressions of inillort- substituition ratios oni effective rates of protection anid changes thierein. While, therefore, the results for the differeint countries are fairly mixed, we also need to note that the construictioni of a thcoretical rationale for a successful regression of import-sLubstitutionl ratios or growthi rates in cross- section analysis is difficult, and one may reasonably expect to find no relationships of the kinds postulated. It should be useful to spell out why this is so, taking the import-substitution ratio as the depend(lent variable and effective tariffs as the indcpendent variable. i) To begin with, etfectixe tariffs heing tile independent variible, a basic difficulty arises. The effect on the imrport-sUblstituttion (pIrod ucti In- to-total-supply) ratio is not uniquelN determnined by the efTectixc tariff: for the same effective tariff is conipatible witll different comnbinations of nominal tariffs on output and iniputs and helce with (lifferet effects on production and conisumplrtion of the output. Therefore, even if the pairtial- equilibrium, supply-and-demnand curves were identical across the indLustlries the relationship postulated wvould not follow unless the input-output struc- ture and the structure of nominal tariffs on each inridutw'.s outputs alnd inputs were identical. ii) Once we go beyond partial into general-equilibrium analysis, further- more, the hypothesis refuses to hold up for the furtlher reason that tlle theory of general eqLuililbrium tells us unlhappily that, in an n-output (n>2) economy, if more than one price changes, the directionl of indi"idual output changes cannot be pIredictedl from this fact alone: one really has to work out the full general-equlilibriuim soltition.*" This nihilistic conclusion carries : Ibid. (1975, chap. 12). Nor. for that nallter. (lid Behrman managce to find any simnitic:int association be- tween DRCs andl growth indicates. 73.GL]isineer (1971). 3 T. A. ()cjide (1971). XC One may further he temlpted to infer (as we (lid in the ('onfrellmce paper) that. if there are it ('>2) dilTemcnti tariffs. e .illing in it prices changing, there is no theo- retical presimption at all for- asseiitnug ihat the cihanigcs in the n ;activities' omltptit, will be correlated with the i talril's. Ho%ke'er, as noted hy Alan D)eardorfT in his ('Comment, this would be an invalid inference. 12 Jagdishi N. Bhaigwatlti and T. N. Srinivasan over, of course, to a general equilibrium model with imported inputs as well.39 iii) Finally, while the analytical points made above relate to the effects of the tariffs vis-a-vis the free trade situation, with given resources, the exercises testing the postulated hypothesis relate often to a situation of growing resources. But, in this event, there is even less presuitmiptioin theoretically in support of the hypothesis.4"1 Thus, even within the confines of conventional economic theory, one would have serious difficulties with the hypothesis that higher efTective tariffs lead to higher import-substitutioll ratios on a cross-sectional basis. In the context of actual developing countries, these difficulties are accentu- ated indeed, For example, the growth of industries is likely to reflect in- dustrial licensing and targeting; and, as noted below in the context of QRs, anticipation of tariff protection, as distinct from initial protection, once the industry has built up to size leading to an effective political pressure group,1' may be quite important in determining growth incentives. In fact, we must recognize many additional difficulties, specific to ex- change control regimes, (where QRs typically may dominiiiate tariffs), witl the notion that observed protective structures will tell the analyst anytilinri terribly conclusive about growth incentives. In particular-i L, an imporerltant 39 In fact, for predicting outpuL changes (as one must be, if one's interest is in the import substitution ratio), as distinct from "value-added" changes, in moodels with im- ported inputs, the effective protection measures run into troulble even if we confine ourselves to twvo goods. This problem was first raised by V. K. Ramaswami and T. N. Srinivasan (1971) and is extensively analyzed in the contributions of Bruno (1973) and Bhagwati and Srinivasan (1973) to the "Journal of International Economics Symposium" (1973). 40 Thus, take a simple two-sector example, using the standard tv.o-by-two model of trade theory. We know from Rybczynslki'\ 1heorem that the supply curves of the tvo commodities will shift differentially rather than identically, so that even if the supply curves were identical in the initial situation across activities, they would cease to be so with economic expansion (unless all factors expanded uniformly), And hence any effect of the tariff structure on the import-stubstitution ratio would be "mnllddlied' by this additional growth effect. This is clearly a pertinent point when one is relating the import ratios for 1967, for example, to effective protection in 1962 (as in the Nigerian exercise reported above): a period over which the capital stock may have increas;ed by nearly 30 percent (assuming a capital-output ;atio of 3:1 and an average savings rate of 15 percent of GNP), and hence certainly in excess of the labor force. " Thus, the causal relationship may well run from the growth and size of an industry to the magnitude of its tariff protectioni. In fact, it is only recently that economists have begun to concern thenisclves with the ciquestionl of why tarilf structtures are what they are, as distinct from what they should be. At an institutional-analytical level, the work of Padma Desai on the criteria tised by the Indian Tariff Commission in granting tariff protection represents one approach of interest and importanice (ef. Desai 1970). At a statistical-econometric level, the work of Basevi (1966) on exanmin- ing the factor intensity of protected indtistries in the UInited States represents a different, and equally usefUl. :tpproachh; for an intc-e'ting analysis of the relationship between the labor force characteristics of an industry and the degree of e\emption secured by it from the across-the-board 50 percent tariff cut in the Kenniedy Round, see Cheh ( 1974). Trade Policy and Development 13 fact is that many developing countries have operated with rules of "automa- ticity" in protection: QRs were used to grant protection as soon as domestic production was started. Once this "institutior, al" feature of the system is taken into account, it is easy to see that any observed (implicit) tariff structure fails to incorporate the incentive effects of guaranteed, "potential" tariff protection, which is clearly a significanit factor on the Scene. More precisely, we should not expect the resource-allocationial effects of i1 prespecified tariffs to be identical with the effects of a process of tariff- imposition that is characterized by automatic protection to any potential activity, the degree of protection, in turn, being expected by potential in- vestors with uncertainty regarding its precise extent (this, in turn, being dependent largely on the restrictiveness of the forcign exchange situation), and which process winds up with the n observed tariffs in place. It is for this set of reasons that the notion of relating tariffs, effective or nominal, to the pattern of industrial expansion-no matter lbow measured -seems to be lacking in sufficient rationale, especially for countries with restrictive exchange-control regimes: as, indeed, several developing coun- tries have been for the bulk of the postwar period. This may well account for the mixed nature of the statisticatl results reported in this section. On balance, therefore, we should be content to take tlle view, admittedlY less ambitious, that the differential tariff structure among dill'crent activitics merely incdicates, very broadly indeed, the differential nature of the in- centives that exchange-control regimes in developing countries tend to generate: a conclusion that, in itself, is sufnicicntlv interesting and inmportant. Next, we may note, in this context, that the differential nature of tlle incentives, as indicated by the differential rates of protection to different manufacturing activities, can be shown rather to result sinwltancously in differential social returns from the allocation of rcsources in producing these alternative items, This can be done quLlitaftively by slhowing how the actual allocational criteria used for making production and investmenit decisions, whether through the use of ORs or throu;gh the use of domestic licensing or via both sets of instruments (as in India and Pakistan), can hardly be expected to yield anythinp like an cconomicall rational alloca- tion of resources.42 Ouantitativel, it can be done by doing sophisticated cost-benefit analvsis on a number of lilferent activities, tlhereby showing the differential social returns -esulltimig fronm different aclivities eicouratged (or enabled to exist) by the entire franic%%ormk of protection. It can aIlso be done by using somewhat roug,h-andi-ready calculations, suclh as those implied by DRC estimates, of the kind deployed in the NBER studies, which essentially use 'fllustrative" slhadow prices and arrive at notions of 42This, in fact, was done in the India study of the 0FCD) (13hmg%vaii and Desai 1970) and was also thie reason for the :mnatymicyi focuis on methods of echmlimge control in the NBER studies. 14 Jagtisls N. BlJag;i-ati aitd T. N. Siitirasaii differential returns produced by different activities by estimating the foreign exchange that the same value of domestic, primary factors is producing in alternative activities. Primarily using the DRC method there- fore as a rough device for estimating social retur-ns, the NBER project does show the wide variations that obtain in the restrictive foreign trade r-eginmes that have been deployed in the LDCs studied under the project.'" Finally, note that the NBER studies explicitly extend the analysis of the economic consequences of protection, as generatel by restrictive foreign trade regimes, to issues such as underutilization of capacity, exccssive in- ventory holdings, etc., with findings g-,enerally adverse to the case of those who favor the use of restrictive trade regimes. They also investigate the dynamic aspects of foreign trade regimes quite explicitly, examining the effects on domestic savings formation, foreign capital inflows and eflicienlc) thereof, quality of entrepreneurship, technical clhange and innovation, etc. The general conclusion from such analyses is that there is little empirical support for those who would argue that restrictive regimes generate dy- namic gains that offset the static inefficiencies that are documented in the NBER studies and that, in fact, were spelled out also in the earlier OECD studies at some length."- D. Protectioni and export per! oriancuce We turn now to the issue of the degree of import sLibstituLtionl that corre- sponds, as noted earlier, to the question of the degree of protection. Two points need to be noted at the outset. First, recalling that protection is defined here, as in the NBER studies, as inclusive of the effects of the exchange control regime via import premia etc., the analysis in this sub- section will extend to the issue of whether restrictive foreigni trade regillmes, associated with high import premia, lead to deteriorating or inferior export performance, whereas liberalized foreign trade regimes tcn(d to have im- proved expor. performance. Second, we should also note that the common distinction drawn between import-subsiitufting strategy and export- promoting strategy may be made, in sharper analytical terms, by observing that the former group essentially works with a degree of protection that implies that the ratio of EERs for exports is less than unity, whlercas the latter group of counitries essentially has this ratio closer to uinity (as export 13 For detailed analyiical and empirical discussion. see lBh;igwaiti (1978, chap. 5). 44 Many of these dynarnic questions were dealt with e\plicifly for India in Bhagwati- Srinivasan (1975), in particular; they have been considered morc generally, with an eye on the entire set of countries in the NBER nproiect. in Bhlagwaii (1978. ch;ip . 6- 8). Note equally that, contrary to the enthusiasm of many proponents of liberalijied re- gimes, there is no systeniatic evidence on their side either of dynamic cieicicncies. The facts, and for that matter, the theoreticnl argimients. in these dNynamic areas go in both directions and no general concluwions seenm arranted. Cf. 13hagw:iti (1978). Trade PolicY andt Dei'elopmnemnt 15 subsidies of various types bring the EER for exports much closer to that for imports).45 There are several different types of evidence available in the NBER studies, to suggest that restrictive foreign trade regimes, with high explicit or implicit tariffs and lower-than-unity EER JEERm, ratios, are associated with lower export performance and that changing the overall foreign trade regime successfully in the direction of reduced reliance on exchange conitrol and increased liberalization pays handsome dividends in terms of higher exports. First, there is the usual type of evidence that, after successful liberaliza- tion (normally accompanied by devaluation), exports having generally declined tend to sh1ow responsiveness. This phenomenon, known in the literature on devaluation as the J-curve behavior (with initial decline and later rise), has been documented for several (though by no means all) of the liberalization episodes that the NBER countries experienced and that have been studied in depth. Thus, for example, the June 1966 Indian de- valuation and liberalization policy package, once adjustment was made for the exogenous decline in exports brought about by two serious agricultural droughts, showed this type of pattern of export behavior,4" Occasionally, attention has been focused on the short-run export response, so that the medium and long-run response, which was more favorable, has been missed by earlier anal\sts.'7 Second, there is a considerahlc amount of statistical analysis, in the NBER studies, of the responsiveness of minor exports in particular and manufactured exports in general, which (on the basis of regression analysis using mainly time series estimates) suggcests strongly that the exports of these developing countries are, in general, responsive to price changes. This evidence is at the microlevel for specific commodities (in- cluding sometimes even primary products) and also for broad aggregates by sectors.4` It should be noted that the studies do deploy different indices for their price variable; and there is, indeed, here some of the tendency 4'' In practice, the e\pori-promoting countries do not seem to make the export EER identical to that for imports; buit thc do make it substantially closer. In theory, one should want to define export-promoting strategy as making the EERJIEERI, ratio exceed unity, so that there is a net incentive to export rather than serve the domestic market! These issuies are (liscUwsed in BhagwaLti (1978. chap. 8). ,il For a full discussion of the cross-country evidence, see Krrueger (1978). 4' For a notable ecweption, see Cooper (1971). 4'This evidence would ,tiveest that while the 2 X 2 trA.de-theorelic model, where both goods are traded, is uinrealistic, the atugnmentation of this model with a preassigned nontraded good is also incapable of captluring reality adequately. What one needs is a model where, dlepcn(ling or the policy equilibritum, a good may be traded or may cease to be traded. Stuch a model, on Ricardian lines, was considered by Samuelson (1964) many years ago and has been recently e%plorcd by him in a joint study: Dornbusch, Fischer, and Samuielson (1977), anid inlLICpendIelilly in Saimelson (1979). 16 JagdishA N. Blhigaiu.i and T. N. Srinivasan among econometricians to keep shifting among alternative price variables until something works. Buit, with this customary caveat in mind, we should note that the evidence broadly supports those who contend that prices do matter. Third, Krueger's cross-sectional analysis of the ten NBER countries in her synthesis volume also seems to underline the significance of prices in improving or inhibiting the growth of exports. In her regressions, she uses dummies to represent Phases I, II, IV, and V: these refer to diiTcrciit de- grees of restrictiveness of the trade regime (as spelled out in the NBER studies), where Phase I primarily initiates in a simple way the QR regime, Phase II represents proliferation of QRs and increased restrictiveness, Phase III is attempted liberalization, Phase IV reprcsents successful move- ment toward liberalization, and Phase V is a full shift to a liberal-paymelnts regime.49 The Krueger regressions indicate that PLDEERs on exports seem to affect both traditional and nontraditional (otherwise described as minor in many of the studies) export values, and that Phases IV and V do seem to affect export performance favorably.a() In this regard, note furthernmore that there seems to be a general case, underlined by ti detailed analysis in the NBER stidies, for arguing that it is really a shift to successful liberalization ancd tllereforc contintiiiin liberalization that is critical to improved export performance on a sustained basis: i.e., a shift to Phase IV from Phase 1I will show such an improved performanlce, but not really occasional jabs at libcralization, each resulting in eventual relapse into Phase II (from Phase III liberalization attempts). Thus, it is the sustained transition during the 1960s and later from Phase II to Phase IV by South Korea, Taiwan. and Brazil that has been attended by high rates of export growth.5' Additionally, it is also worth notinig that it is not just the price aspects of the restrictive Phase II regimes that inhibit export performance. As has been documented in the Indian case, for exaample, and by contrast in the South Korean case, in the NBER studies, the whole franmeworl; of exchange controls in a Phase II situation militates against export performance. Thus, for exanmlple. the ability to expand production to fill export orders requires access to import licenses for raw materials and capacity expansion requires import (and industrial) licenses: in each case, red tape and uncertainty cloud the sceneo.2 Fourth, statistical analysis of the usual decomposition variety, where the 49 For more ysytcmaLic and careful definition, of the Phases, see any of the NBER voluimes: for conv,cniciicc. they are state(d fully in the Appendix. Cf. Kriuever ( 1978). PI.I)El&R ;tands for price-levcl deflated EERs. 5.1 'he role of expect:ations ensuring that export incentives are seen to have been made favorable over continued periods is ohb%ioisly cr-itical to this result, for that is when entrepreneurs will wish to make investments in export markets, 52 Again, this is the kind of effect on exports that only Phase-change analysis can piclk up statistically, if at all. Trade PolicY antd D'i'elopmnwni 17 export performance of several LDCs is decomposed into that attribUtable to overall growth of demand, regional composition, commodity composi- tion, and a residual "competitive" factor effect, contrasting the 1950s when most LDCs were in Phase 11 and the 1960s when some lhad successfully shifted to Phase IV, shows that the latter group of Phase IV countries had dramatically improved export performance and that a sizable share of it could be assiggned to the residual, "competitive" factorA} Such analysis of the "competitive" factor is not generally consider-ed to be as persuasive as the time-series analysis deployed in many of the NBER studies. However, it has considerable suggestive value and is corroborative of the conclusions arrived at through use of other analytical approaches. E. Protection anid econiomic perforinmacc We. may finally address the central question of whetlher LDCs with superior export performance also have superior econiomic performance and, if so, why? There is little douit .'-at, in the NBER studies for example, the countries that have managed to shift to improved export performnance by reducing export bias have also managed to rcgistcr acceler-ation in their growvth rates, whereas countries that have not done so (and lhave remained in Phase It regimes) have had poorer gro%%th rates. The contrast between the success of South Korea and the failure of India, in this regard, is cross-sectionazilly the most telling. A recent statistical analysis of Irving Kravis also supports this conclu- sion.5 Using decomposition anal\sis to differentiate LDCs with highllexport performance based on domestic policies, and taking a 39-country samples Kravis has noted a 0.51 Spearman coefficient between ranks with respect to the index of such export performanllce and ranks reoardino the growuhl rate of real national product. That the superior-export-performance countries do better compared to both their own earlier growth performance under restrictive trade regimes and other countries with inferior export performance seems therefore to be, generally speaking, a valid assertion. The interesting question is: why? Here, we have a few answers and many que,stions. 1. First, it would appear that the pattern of incentives, and hence of expiort promotion, is less skewed in practice than the chaotic pattern of import-substituting incentives under the restrictive trade regimes. The statistical quantifications of EER,.s for several activities in Southl Korea, for the mid-1960s, for example, sugcest that the variability (including the extremes) of incentives is significlantly lower than the EER,,s for several 5Cf. Askari and Corbo (1975). This sradiss icai Ntiidy also distinguishes between "minor" and other exports, defining the "minor" as all those exports that were below 10 percent of the total value in the initial year. 5- Cf. Kravis (1970; pp. 868-69. in particular). 18 Jagdishl N. B/agw afi and T. N. Srinihasan activities in the restrictive Phase H1-type regimes in other countries, such as India.55 Similarly, the average ratio EERJIEER,,, also seems much closer to unity (at times even exceeding unity, but remaining closer to it) under the liberalized Phase IV- or V-type regimes than unider the restrici;ivc Phase II- type regimes."6 Thus, it wvould appear that, on both the degree and the pattern questionis, distinguished earlier, the export-p-'omoting countries with liberalized re- gimes seem to do better. For both types of allocative reasons, therefore, one could argue that the resulting reduction in allocative inefficicncy must provide some of the explanation of the improved export performance that is observed for the liberalized-regine countries. But, in turn, one rnmust ask the question as to why these incentives are less chaotic and more "neutral," by and large, under the liberalized trade regimes. The reasons would seem to consist in the fact that the successfuil shift to export-promoting strategy (or Phase IV) generally takes place within the o.verall context of continuing exchlmnge controls. and that the OR-caused bias against exports is offset by giving the import premia to exporters through schemes such as supply of iriportecd materials at internationlal prices, etc.,5, and by using exchange rate adjustment more freely and thereby directly reducing import premia and hence the bias against exports. The result is generally (not always) to eliminate or reduce the bias against exports rather than to create excessive bias for exports. BecauLsc of budgetary considerations, cash subsidies that could. conceivably create massive bias for exports are usually not substa-ial (thloughli not unk-nowv.n). On the other hand, the import-substituting strategy, especially via the mechanisms of import premia from ORs, can and has typicalIlN CaIsed EER,,, to get way out of line with EER, (which was then dleterineid(l almost exclusively by the exchange rate): and the costs of suclh a subtantiail rise in EER, /EER , above unity are generallv not undersood and, in any case, do not fall directly on the budget. 2. Next, the sheer improvement in export performance, followviinm from the elimination of the bias aeainst e.xports, must surely play the major role in the full explanation. The link-s here are possibly diverse. i) The NBER studies suggest that there is little evidence that the export- promotring couintries are technically more ITrogressive or that tlley have higher savings rates because of a larger export sector.`5 The asyimetrv in For a more detailed analysis, including statis,tical and analytical reasons for possible skepticism regarding this observation, consuilt Bhalng%hli (1978. chaip. 8), I§ Tn fact, 13h:ewati ( 1978) therefore defincs the cport-pronmoting straeLvw as one where EFSR:EER?,, is brotught fairly close to unitv. There is much documentanion of thesc schemes of expart promotion in the N lBE-R studies. Cf., in particular. Bhiagvati-Srinivasan ( 1975) on India and Frank, Westphal, and Kim ( 1975) on South Korea. These questions have been examined in detail, analytically and empirica.lly, in Bhagwati (1978, chaps. 6 and 7). Trade Policy anid Developnment 19 the export-promoting and import-substituting countries' economic per- formance cannot thus be traced, at least on current evidence, to superiority of the one strategy over the other on these dynamic grounds, even though the proponents of each strategy often indulge in assertions to that effect. ii) Part of the answer rather appears to be in the fact that a more comfortable balance-of-payments position, resulting from imiipro\ed export incentives and earnings, generally eases up the excesses of the import- substituting strategy. This should be obvious from the well-kniown demon- stration that, under a foreign exchange bottleneck (in the sense of Chenery), additional foreign exchange is more productive than under a savings bottle- neck. But it is also apparent from the fact that it eases excess capacity (generated largely by the OR regime in the first place5"), may reduce the need to hold excess inventories, and leads often to elimination of critical bottlenecks, etc. It is perhaps remarkable that these kinds of problems, attendant on economies in the restrictive Phase II, are rarely to be found in the liberalized Phase TV and V economies that have successfully transited to export-promoting strategy on a continuing basis. iii) In regard to the general easing of the balance of payments (and hence of the losses that attend restricti\e paymenits policies) under the export- promoting strategy, it is also worth noting that this effect is reiniforced by the substantial inflow of foreign capital that can attend such a strateg., While political factors help to explain the substantial inflowvs of foreign private investment in Soutlh Korea, these are undoubtedly to be supple- mented by economic factors. And here one probably ought to attribtute to the export-promoting strategy itself the sizable magnitude of the inflow of nonaid foreign funds and its efficacy in promoting economic growth. By contrast, under import-substituting stratey., both the matnllitnldc of the inflow and its social returns are likely to be lower. 'Tlhis conti-ast nmay be explained as follows. Regardinig magnitude, an export-prontolimig strategy, with its lack of discrimination against foreign markets, is likely to attract foreign firms essentially on the nineteenth-century pattern of factor-endo\mnent ad- vantages. \Vhercas in the ninctecntlh-century. this meant natural resouLrcs, today it means exploiting Heckscher-Olhlin style low wages. On the otlher hand, by creating artificial indUcecmlent to invest via tariffs an-d/or ORs, so that one gets "tariff-jurmping" investmnents oriented to the domestic market alone, the import snbstituting strategy provides an artificially limiiited in- centive to invest in the LDC. FuLtlhermllore, even the suibstantial official borroxin-, by Soutlh Korea and Brazil in the inteni-aitional capital miarik-ets surely must lhave been facilitatecl by the dcionAr ation of a superior export perfornmance (for, that would assuage fears of e.xccssive borrowing and inability to repay). Then again. in regard to efliciency, it is easy to show that "tariff-jumnlling" ' On this point, see the arguments in PBhawati-Srinivasan (1975, chap. 13). 20 Jagdislh N. Bhagwvaii and T. N. Srit1nivasan investments, induced under the import-substituting strategy, are more likely to imply social losses or (at minimum) reduced gains than invest- ments attracted by Heckscher-Ohlinesque factors. That foreign capital in- flow can be not merely less productive wben inspired by QRs and/or tariffs, but actually immiserizing, has been shown elegantly by Brecher and Diaz-Alejandro in a recent paper.," For the traditional 2 X 2 model of trade theory, they show that social utility for the small couuntry, lhavilng declined with the tariff, will decline further with the initial inflow of foreign capital when the importable good is capital-intensive. It will con- tinue to decline with additional inflows of foreign capital until autirlk-y is reached, then rise gradually to the level under free trade (a situation dis- cussed by Mundell earlier),',' remain at that level for further inflows and, finally, start rising after complete specialization in production is reached (a situation discussed by MacDougall earlier)."- While the factors noted in the preceding paragraphs would seem to be critical in defining the asymmetrical outcomes under the import-substituting and the export-promoting strategies, some additional factors may be cited that might contribute to the asymmetry, but for which no systematic evi- dence is yet available. iv) Thus, one could argue that the export-promoting strategy may lead to a generally redutced reliance on direct or physical, as distinct from price, measures.03 Direct controls have been argued wvith plauisibility, in botlh tlle OECD and NBER studies, to be very costly in practice. It is possihle that the general incidence of such direct controls may be sigiiificaitlv less uinlder export promotion, because price, distribution, and other controls mav make little sense to bureaucrats when firms' outputs are mainly addressed to overseas, rather than domestic, miarl;ets. A different, andi perhaps more perceptive, formulation of this kind of conitrast was xvell put by ani econo- mist familiar with both the Indiani (Phase II) and the Souith Korcan (Plhase IV) trade regimes: the Indiani regime consists mainly of "don'ts" wilereas tihe Korean regime consists mainly of "do's." Whether thiese contrasts are, in a basic political sense, endemic to the two strategies being contrasted is not clear; but the NBER studies do suggest that they exist currently. v) In the still more grey area, one may further argue that the export- pronmotinig strategy mu