130(oś The Effects of Protectionism on a Small Country The Case of Urutguay EDITED BY AND JAIME DE MELO The Effects of Protectonism on a Small Country ne Case of Uruguay WORLD BANK REGIONAL AND SECTORAL STUDIES The Effects of Protectionism on a Small Country The Case of Uruguay EDllED BY MICHAEL CONNOLLY AND JAIME DE MELO The Vorld Bank Was4ingAm, D.C. @ 1994 The Enternational Bank for Reconstruction and Development J The World Bank *X- - 1818 H Street N.W., Washington, D.C. 20433 All rights reserved Manufactured in the United States of America First printing May 1994 The World Bank Regional and Sectoral Studies series provides an outlet for work that is relatively limited in its subject matter or geographical coverage but that contributes to the intellectual foundations of development operations and policy formulation. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this publication are those of the authors and should not be attrbuted in any manner to the World Bank, to its affiliated organizations, or to the members of its Board of Executive Directors or the countries they represenL The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright aearance Center, Suite 910,222 Rosewood Dr., Danvers, Massachusetts 01923, US.A. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C 20433, U.S.A, or from Publications, The World Bank, 66, avenue d'IWna, 75116 Paris, France. At the time of writin& Michael Connolly was professor of economics at the University of Miami H4e is currently visiting professor at Columbia University. Jaime de Melo was clief of the Trade Policy Division in the World Bank's Policy Research Department He is currently a professor at the University of Geneva. Cover design by Sam Ferro. Library of Congress Cataloging-in-Publication Data The effects of protecdonism on a small country . the case of Uruguay I edited by Michael Connolly and Jaime de Melo. p. an- (World 3ank regional and sectoral studies) Includes bibliographical references. -ISBN 0-213-2788-7 1. Protectionism-Uruguay. 2. Uruguay-Commercial policy. L Cormolly, Michael B. (Michael Bahaamonde), 1941- . II De Melo, Jaime. IL Series. HF2016.E36 1994 382t'3'09895-dc2O 94-7420 CIP Contents Preface vii Acknowledgments ix Contributors x 1. The political economLy of protectionism in Uruguay Michae B. Connolly andjfaime die Melo Protectionism and stagnation: an inter :14 Midhael B. Connoliy and faine die Melo 3. Adminiteed protection: reference prices and minimum export prices 45 FMaimo Ckangnna quit and PatTink A. Merssalin 4. Domestic content restrictons and compensatory export requirments in the automobile sector 64 Wendy Takacs 5. Gains and losses from bilateral trading arneet with Argentina and Brazil 81 Jai=e de,melo, CJudio mantene4gro, and Wendy Takacr 6. T~he labor market and trade reforun in mianufacturing 108 Martin Parna 7. Economywide costs of protection and labor market rigidities 124 Jaime de Mdo and David Roland-Hois 8. A long-rum perspective on trade policy, instability1 and growth 150 Jorge F Roklds iM : Preface Why have we put together this volume of essays on protectionism in Uruguay when there are alreadyvexcellent comprehensive stUdies of the broad2aSpects ofprotection in th-at country (Anichini, Caumont,and Sjaastad 1978; Favaro -and Spiller 1 990; and Favaro and Sapelli 1990)? We see at least four reasons for further analysis. Firt, pr-evious work has concentrated mostly on describing the protec- tive measures rather than an-alyzing their- consequences and trying to quan-- tify their effects on resource allocation, growth, antd welfare. The essays in this volume go into more detail on the various regulatory mechanisms, modeling the protective instruments and quantfifjng their effects. For ex- ample, the modeling and analysisof so-called antidumping mechanisms such as refernce prices (chapterS3) is novel, detailing not only the magnification effects of protection but also its. undesirable side effects on the quality of dom estically produced manufactured goods. Since proposals for simil-ar mech anisms are often put forth in developing countries, such in-dept dex- ploration can guide the analysis of their effects wherever such measures might. appear. Similarly, the modeling -and welfare analysis of the automobile as- sembly industry (chap ter 4) provide a unify~ing fr-amework for analyzing the welfare costs of such 'screwdriver" assembly operations anywhere. -Second, economists should be interested in the essays in this -volume because the policies analyzed have been adopted in many other countries. Each chapter develops and uses an analytical -approach to examine the p:ar- ticular measure under study, followed by empirical measuremen't. Third, tiinre is an interest for the policymaker. The essays go beyond quanrilying thfe effects of trade interventions to show the interactions between various forms of intervention (between labor markers and protcecdion in chapter 6, between protection and rent seeking in chapter 2, or between market intervention and resource allocation in chapter 7). Thus, the essays show the harmful effects of protectionism and of excessive intervention gen- viii Eflncr oFPWYrCoNLSM ON A SMAL COUNmr THE CAS OF URUCUAY erally on a small economy, providing vivid examples of policies to avoid as- siduously-notably those associated with antidumping, the promotion of infimt industries, and non-price protective measures in general. Finally, Uruguay offers an excellent laboratory for studying both com- parahivelyandindividually the consequences of highlyrestrictive regulations for a small economy, bringing out elements of relevance to many current economic problems in the world today. A case in point is that of the socialist econonies in transition and of the small states of the former Soviet Union, such as Georgia and Armenia, which share many of the characteristics of Uruguay. This volume is therefore a useful compendium for students, ana- lysts, and policymakers in small economies that stand to reap considerable benefits from a more open tading pattern. Acknowledgments In December 1989 we visted Uruguayunder the auspices of thejoint United Nations Development Program (UNDP)-World BankTrade Expansion Pro- gram (TIEP) and subsequently delivered a report in june 1990 to the Uru- guayan governmen t, Untguay: Trade Reform andEconomicEffidaucp. Often, in policy work, analysis and advice have to be produced on short notice. This ,mission was no exception. However, for most of us, unanswered questions remained& We wanted to probe deeper into the complex Uruguayan trde regime. This volume is the outcome of the follow-up work. Throughout we benefited from support and advice from friends and colleagues in Uruguay-. rrmon Diaz and Elbio Nattino at the Central Bank were most generous with their time. So were Luis Macadar and Martin Ramia at the CGentro de Investigadiones Econ6mnicas-Uruguay and Jorge Roldos, Glaudio S-apeui, -and Luis Vian-a at the Centro de Estudios de la ReaHidad Econ6micay Social, who provideduswith helpful suggestions along the'way- Carls Rodriguez of GEMA. in Argentina also aided the mission. At the World Bank, we were fortunate to have the un6iling support of Eric Abalahin, MariaAmeal,jeff Hayden, and Rebecca Sugui. We also thank Men de Coquereauraont who edited the volume. Two people deserve specia thanks. Tom Cox, the manager of the TEP, was!a constant source of support -and encouragement -at every stage of our. work- In Montevideo, we were fortunate to be able to counxton Alberto Sojit, the Resident UNDP Representative until the time of his ftragic death during our mission. This volume owes much to his encouragement, and we hope that lie would have found it a useful contribution to the study of protrection- ism in Un:guay. .: R : s-~~~~~~~~~~i Coyntributors Federico Chnmganaquf Trade Policy Division, World Ba-nk Mfichael Connofly Departmnent of Economics, University of Miami jaimne de Melo Trade-Policy Division, Worldl Bank, -and University oF C-eneva, Switzerland Patrick Messetlin Intemnational Trade Division, World Bank Claudio Monteneo Trade Policy Division, World Bank M4ardin Ran Trade Policy Division, World Bank, and Centro de Investigaciones Econ6- micas,-Uruguay (CIVE) David Rolnd;Hoist QEGD Development Centre and DepartraLent of Economics, Mills College Jorge Roldos International Monetary Fund and Centro de Estudios de la Realidad Econ6-_ micay Social (CERES) Wendy Takacs Trade Policy Division, World Bank, and Department of Economics, Ucniver skty of Marland, Bfaltimore County The political economyof protectionism in Uruguay M.ichael B. Connolly andJaime de Melo Under the presidencies of Batile y Ordoiiez in 1903-07 and 1911-15 Uru- guay created progressive democratic institutions and awellhire state thatmade it the most advanced, yet regulated, country in Latin America by the middle of this century. With ahomogeneous population and awell-functioning demo- cratic system, this small country of less than 3 million people was know in the early post-World War II period as the "Switzerland of the South.' Four decades later the country had undergone a precipitous decline that put it an xing the least prosperous middle-income counties. Afew comparative figures tell the stozy. In 1950 Uruguay's GDP per capita was $3,784 (in 1985 prices and adjusted to have the same purchasing power as in the United States), dose to that in comparable small and wealthy Euro- pean countrieslike Belgium ($4,151) or Demnark ($4,512). In the Republic ofKorea, a country hailed today as a major development success, CDP per capita inl1953- was $822. In 1960 school enrollment ratios inUruguay reached 11l percent at the primary level and 37 percent at the secondarylevel. The corresponding figures for the European countries are 109 percent and 69 percent foriBelgium and 1 03 percent and 65 percent for Denmark; and for Korea, they are 94 percent and 27 percent) Uruguaywas definitely amaongr the mostdeveloped countries in the world in1950. Measured in purely economic ters Uruguay's GDP per capita was among the highest in a group of thirty-five countries with a GDP per capita near $3,800 in 1950 (in 1985 prices). But Uruguay's relative position dete- rirated so sharply chat by 1988 it was among the last in that grup (figure 1.1). This economic dedine was accompanied by social unrest and political crisis. What happened? The essays in thisivolumenare aboutthe cononeiczaspects of this decline, primarily the economic polides that accompaied it They examine the mechanisms bywhicli these polices particuarlyprotectionism in allitsforms, I~~pa onrelleBlim411 remr |452 ndeRpbi .2 EmCTS OF PtOTECrONISM ON A SMALL CouNTRY: THE CAsE OF URUGUAY Figure 1.1 Uruguay's GDP per capita ranking among thirty-five * semi-industrial economies, 1950-88 RCnk 5 I 0 AL 20 25 30 35. 1 950 1955 1960 1965 1970 1975 1980 1985 Now The foAowing economies are included in the sample; Argentina, Ausbala, Austia, Barbados, Belium, Canada, Chile, Cyprus, Denmudk, Finand, Rance, Germnan Great Britain, Greece, Hong Kong, Iceland. Irelnd, srael, lb1 Japan, lxembour Mexico, the Netherlands, New Zealand, Norway. PRrWgal, Sngapore, South Afiia Spain, Sweden, Swiezaland, Taiwan (China), the UnLted States, Unrguay, and enezuela. contributed to Uruguay's economic deterioration. Policies applied to the manufctuning sectorand their economywide consequences receive the clos- est attention, even though much intervention and regulation also took place in agriculture. While the essays focus on the Uruguayan economy, the regu- lations and distortions examined are pervasive in a large number of coun- tries. In many respects, Uruguay is a vivid example of how excessive regula- tion can bring about economic sclerosis. Because of its small size, Uruguay should have been pursuing outward- oriented policies to ensure its greatest possible integration into the world economy. The economy would have benefited from larger markets and the gains from specialization according to comparative advantage, aswell as from access to worldwide best-practice technologies and the incentives for adopt- ing them. Uruguay chose instead to protect industry and to tax agriculture, the sector inwhich ithad a comparative advantage. Thus, Uruguay is a good example of-a small country that, despite the obvious advantages offreer trade, has pursued industralization through import substitution behind a wal of protection andreguladon. The cost of the strategyhas been astunting of the export sector, notably agriculture, and a loss of general well-being, due to the high inefficiency costs of protection. Ultimately, these inward-looking policies and the rent-seeldng actiities and income redistribudon that surrounded them bred stagnadon and a low Thepoliticleconomy ofproteaionirm ti Uneguay 7 standard of living. That result would be anticipated for any small country that attempts to industrialize behind a barier of tariffs, quotas, and admin- istered protection-especially one like Uruguay, whose obvious long-run comparative advantage lies in agricultural goods. In Uruguay these policies achieved a redistribution of income and production from the country to the city and promoted considerable rent-seeking acdvity around a small indus- trial base in Montevideo. This resource reallocation promoted an inefficient industrialization. . - Mechanisms of protectionism At various times in its history Uruguay has relied on a broad array of instru- ments ofprotecdon. MoststiR existin the earlyl99Os, though some are being reformed. They have proved to be a strong force fornearly two hundred years. Tariffs and reference prices The most widespread and certainly the best known restrictions on trade are tariffs. There are two types of tariffs: an advalorem tariff, which is calculated as a percentage of the c.i.f. import price of a product, and a specific tariff, which is a specified amount levied on each unitimported. Both kinds of tariffs raise the domestic price of the importgood in the marketplace. The two forms are equivalent in the sense that, appropriately defined, they have the same effects. Tariffs have the familiar protective effect on production and a tax effect on consunption.2 In Uruguay official tariffs are often applied to a ref- erence price rather than to the declared ci.f. import price (see below and chapter 3), which is likely to magnify the protective effect of the official tariff. Qzots- Q!uotas specify arnaximum physical quantity of a good that can be imported. The quotais bindingwhen the maximum amount specifiedis less than would otherwise be imported inafree market, givenworldpices, the domestic tariff, and ansportation costs. Under perfect competition quotas and s have a similar effect on the domestic price of an import good, so a quota also has a protective effect on domestic producdon. Its distributional effect is differ- ent, however, because the revenue does not go to the treasury. A volunlt export restriction is a quota on foreign exporters that confers on them the scarcity value of their exports to the domestic market The restriction may be voluntary in name only since there can be an implied threat to impose duties * or quotas ifagreement is not reached on the voluntary levels. . - i 4 - FXmcas 0oF PtoTECnONISl ON A SMALL CoUNmRx. THE CASE or URUGUAY Licenses Alicense transfers an official right to import undera quota or other restric- ton arrangement to a single individual or a few individuals. The license may be given away, sold at a negotiated price, or auctioned off competitively. When the import license is given away, the holder of the license captures the scarcity value amounting to what otherwise wouid be tariff revenue. Under a fiully competitive auction, by contrast, the government captures the implict revenue or rent represented by the auction price of the license. Li- censes can be used in conjunction with a quota or tariff or they can be is- sued for unlimited amounts of imports. For instance, the cupos discussed in chapter 5 are export licenses for goods that enter Argentina duty free un- der bilateral agreements. Domestic content requirements and compensatory exports A less well known form of protection to domestic industry and suppliers is the domestic contentrequirement,which stipulates thatproduction processes use a mninimum percentage of domestic materials and inputs as part of value added. Domestic contentrequirementspermitdomesticsuppliers to sellparts at higher than world prices and provide a nominal rate of protection equal to the percentage difference between the domestic and the world price. In Uruguay this scheme offered protection to the automobile parts industry, which was further protected by the requirement that automobile assembly firms export products in order to romp ensate for their imported parts. The assembly industry, in turn, was protected by a tariff of 40 percent for fSlly assembled automobiles in 1990 and compensatory export requirements for imports of finished vehicles (see chapter 4). Ad hoc antidumpingprocedures While antidumping legislation is commonplace, its effects are seldom ana- lyzed.Antidumping measures can be used to effectivelyraiseprotectionrates on a goodjustas when customs imposes a duty based on a higher reference price instead of the declared ci.f. price of a good. Because reference prices have a stronger effect than an official duty-a reference price twice the de- clared cif. price has an effective duty twice the official duty-domestic pro- ducers are likely to lobby for higher reference prices. A stricter form of ref- erence price known as the minimun export price collects, in addition, a moving surcharge on imports equal to the difference between the reference price and the dedared ci.f price, leaving the importer litde incentive to Thepoiicaltavonmy oftprteaion in Unrguay 5 declare a price less than the reference price. These procedures do not con- form to GATT antidunmping or counterailing duy provisions. (The refer- ence pnce system is analyzed in detail in chapter 3.) Prior deposit reqtirements Requiring advance depo:its for the inportation of goods is equivalent to imposinga tax on these funds equal to the opportunity costofholdingmoney since the deposits typically do notearn interest. Where the nominalinterest rate is established freely in the market, the implicit tax rate on prior deposits equals the nominial interest rate. With interest rate ceilings the implicit tax equals the real rate of return in the economy pius the inflation rate. Natu- rally, zero nominal interest on prior deposits combined with a high rate of inflation mayresultin an extraordinarilyhigh tax rate on prior deposits and, consequently, on importing. Unti 1975 importers were required to make advance deposits for imports which, at a time of high inflation rates, inplied high taritf equivalent rates (see chapter 2). Government procurement practices Government procurement regulations or practices may give partial or com- plete preference to nationallyproduced goods over imports. This policymay be formallyincorporated in a 'buynational"law orinf-ormaly implemented through preferential purchasing practices. The poiicy has an obvious pro- tective effect on domestic production since it allows domestic producers to charge more than world prices. Industrial countries often usify national preference in procurement on strategic or national defense grounds. In Uruguay, govremment preference to national goods in procurement occurs more by custom and practice than for explicitly stated goals. (Government procurement practices are not discussed in this volume.) Export banrs Export bans are usually set to protect a downstream processing industry. An export ban on a specific good has a protective effect because itlowers its do- mestc pnce, which provides asubsidy to producion processes Using te input since they pay less than the world price. Export bans tend to induce smug- gling, however. Uruguay previously had export bans on hides and cattle on the hoot; which protected the meat pacling and leather industries. The ban was easy to evade by walking cattle over unattended borders. (Export bans are not analyzed in this volume.) ;-6 -;EFFECIS or PRDTECIIOMS ON A SMALL CouNTRv THE CASE OF UwRUuAY Mudipe exchngw -rat Charging different exchange rates for different types of foreign exchange transactions, whedter on the import or the export side, amounts to having a different tariff or subsidy rate on each good according to the exchange rate used.When an importer of a luxury good mustpay apremium of dpercent greater than the official foreign exchange rate as well as a tariff of t percent, the effective tariffisraised by dpercent compared to imports enteringat the officialrate. Similarly, a foreign exchange bonus of spercent paid to export- ers for repatriating foreign exchange amounts to an export subsidy ofsper- cent relative to the official rate. Until the 1970s Uruguay had a widespread system offoreign exchange nintegrosor exportbounties. These are discussed briefly in chapter 2. Administered and hidden potection Trade and transport costs thatmust be paid on imports and exports provide extra, often hidden protection to import-competing actvities. Studies have found such costs to be significant in industrial countries. They are likely to presentan even greater baflier to trade in developing countres. In Uruguay portalhclities are veiyineffidentby intemtionalstandards, anda stronglabor movement adds to the impediments to trade through strikes, delays, and featherbedding (see chapters 6 and 7). Another distortion arises from the ineffiiencygenrated bywage premiums tatmightnothave been obtained by labor unions had protection not existed.3 Sources of inefficiencies in highly regulated economies Whatare the mechanisms throughwhich such instruments of protection get translated into economicinefficiency? Trade polices are anatural firstplace to look for the cause of failure in Uruguay because small countries need an open trade regime to overcome the limits imposedbysmall domesticmarkets. Considerwhat happens when there is excessive regulation. Manufactur- ing firms become less efficient as their choice of optimum output level and input mix is restricted. Regulations affecting input markets all have an ef- fect. Fims are impeded by labor legislation from adjusting theirworkforce. Managementofinventoriesis complicatedbyuncertaintyaboutwhetherand when firms wi be -able to import spare parts and raw materials. Enterprises are unsure abouthow much capital theywil have access to-and atwhat cost. In heavily regulated economies finns aIso operate in restricted product markets. There are barriers to the entry offoreign companies. New domestic firms may be barred by a lack of access to capital-the limited pool of capital The poEand wono7y of prmeaio in Uruguay 7 may be parceled out according to long-sanding des between exsting pro- ducers and their creditors. Barrers to entry, of course, benefitexisting firms bygiving them more leeway in their pricing polides. Thus, finns sufferfrom some of the restrictions and benefit from others. Made-to-measure protectionis commonin highlyregulated economies, a situation that guarantees the coexistence within a sector of firmswith large differences in efficiency. Indeed, in such economies most firms are ineffi- dent The level ofprotection is often chosen to make the least efficientfirms profitable, so many fins can operate at a profit with dated machinery and equipment (technical ineffiiency) or too many employees. Too many ofthem operate at too small a scale (scale inefficiency), fail to equate marginal prod- uctswith factorcosts (allocative inefficiency), and price theirproducts above themarginal costofproduction (priceinefficiency). Because consumershave nowhere else to turn, leading firms can charge higher than normal mark- ups, and other finns wi follow. Firms with a wide range of technical effi- ciency thus coexist, leading to high rates of idle capacity and unproductive work forces. In addition, the incentive system is distorted because protection has been tailored for individual products to ensure the survival of firms already in operation. The made-to-measure protection usually indudes quantitative restrictions that give rise to enormous rents and induce economic agents to expendmuch oftheir effort-appropriatingrents rather than engagingin more socally productive activities. Resources are badly allocated and underused, and the resulting market structures are not compeatirve. The efficiency of investmentis low, growth is slow, and the economy stagnates. This is the story of Uruguay, but the messages are clear for other developing and trnsition economies. Political economy of trade liberalization Many readers will view Uruguay's reluctance to liberalize trde as baffling, considering the complexity and obvious inefficiency of measures restricting trade. What lies behind this reluctance to liberalize is the uncertainty and the redistribution of income that typically accompany trade liberalization. A great amount of income redistribution-whose exact outcome is unknown in advance-must take place to achieve a small amount of efficiency gains from expandedtrade. Unless the economyisin dire circumstanceswithhigh inflation and impending economic collapse, the gains from trade lhberalia- tion will not be enough to compensate for the income redistribution in the political calculus of voters and the government This argument, put forth by Rodrik (1992), can be demonstrated by a simple graph illustting the gains and the redistibution of income associ- 8 Etncrs oF PROTEaCONISM ON A SMALL COuNTRY: THE CASE OF URUGUAY ated with. elminating an import quota. In figure 1.2 the distance SO repre- sents a quota that has the effect of raising the domestic price to p, which is above the world price p*. Import license holders can purchase the quota amountat the world price for p*(SD) and sell it at the domestic market price forp(S), realizing rents equivalent to area rc Domestic producers are able to expand their production to the quantity S thanks to the higher domestic price. Consequently, they receive an increase in revenues equal to (p -p*)S but incur increased costs above world levels equal to area b. The producer's surplus, or rent, increases by area a thanks to the quota. Consumers, on the other hand, lose the consumer surplus equal to the sum of areas a, b, c, and d. They lose from the higher cost of a + b + c over the quantity D and from the decline in consumption at the higherprice associatedwith the quota and equal to area d- Removing the quota on inports has the following distributional effects: import-competing producers lose a, import license holders lose c, and the rest of the economy gains a + b + c + 4 for a net gain of b + d- The point of Rodrik's analysis is to illustrate thatincome redistribution is large relative to the net gains to the economy. He proposes ameasure that computes the total redistribution of income divided by the net gains minus one, or the political cost-benefit ratio (PCBR) of trade liberalization: (LI<) PCBRR= (a + b + c + d)-(b + d) a + c b + d b + d jist Figure 1.2 Distributional effects of quota removal Pice a b c d Demnand 5 ~D Quantity The political econoy of protectionism in Uruguay 9 where p. is the share of imports in domestic consumption, Eis the absolute value of the elasticity of import demand, and t is the tariff equivalent of the quota. Rodrik shows that PCBR is likely to be small for most trade liberaliza- tions taken alone. Take the case of the 1974-82 trade liberalization in Uruguay as an example- The share of imports of goods and services in domestic consump- tion private and govenmment, (p.) was 0.184in 1974, and the initial tariffand tariff equivalentwas about 100 percent ( = 1). Assuming an import elastcity (e) of 2 then yields a PCBR of 2.7, which implies that the political costs of redistributon are substantially greater than the benefits of trade liberalization. Combining trade liberalizationwith economic stabilization that benefits all individuals in the economy lowers the PCBR markedly, silice it is now measured by: (1.2) PCBR = where ris the percentage bywhich everyone is made better off (presentvalue of income), w is the reduction in the domestic pnce of importables relative to exportables due to the reform (internal terms of trade), and 0 is the share ofimportables in GNP (Rodrik 1992). Thus, coO represents the improvement in the terms of trade weighted by the share of importables in GNP. This re- vised fornula suggests that the costs of trade liberalization fall dramadcally relative to the benefits when stabilization measures undertaken during dire economic circunstances result in a return to prosperity. Now consider the PCBR for Uruguay in 1974 when stabilization and recovery are taken into accounL Assume an annual increase in the growth rate of 3 percent for 10 years-an approximation of the increase ir. growth during 1974-82 (see chapter 2)-and a discountfactorof5percent. Individuals' wealth (y) would increase by 19 percent. If importables fall 70 percent (co) in tenns of exporcables because of the trade reform, and the share ofimportables in GNP (8) is 17 percent, the PCBR now falls to 0.5. In other words, the PCBR in- dex declines from nearly 3 to less than 1 as a result of the additional ben- efits of stabilization. Or consider the case ofjust one industry in Uruguay, the automobile assembly industry. The elaborate system of protection of the industry results in large efficency losses and tansfers to local assemblers and parts suppli- ers. Wendy Takacs estimates that the bulk of efficiency losses and transfers comes from the assembly industry rather than from the components indus- try (see chapter 4, table 4.1) . She estimates annual effidency losses from the system of incentives benefiting the assembly industry at $26.9 million and annual transfers of$39.8 million. The redistributions thatwould occurwidh reform are thus-nearly one and one-half times the efficiency gains. : 10 EFFECS OF PROTEoNISM ON A SMALL CouNTRY. THE CAsE OF URUGUAY These estimates are also useful n elucidatingwhy the automobile indus- try is so heavily protected in the flrstplace. Uruguay has ten assembly fins, with the two leading firms each accounting for a third of the market Apm proximately 12,000 vehicles are sold a year, on which the average additional cost of protection to the consumer is $6,834 per vehicle. Takacs calculates that the total efficiencyloss (including the componentindustry) pervehicle is $2,427. The annual benefits of protection in the automobile industry are estimated at $14.3 million for each of the two largest firms. It is no wonder that there is strong incentive to lobby for protection on the parts industry andinsuffidentincentive to lobbyagainstprotection on the partofcorisum- ers. The combination of a small number of big winners and a large number of small losers (here, producers relative to consumers) iswhat Olson (1965) argues is a situation conducive to intense lobbying. Plan of the book In chapter 2 Michael Connolly andJaime de Melo give a brief history of protectionism and stgnaton in Uruguay. Tariffschedules in 1815 provided fora general tariffof25 percentwith some exemptions and amaximum tariff of 45 percent. By 1829 the general tariff rate had been cut to 15 percent and the maxinum was 25 percent-the first know.n instance of trade liberaliza- tion in Uruguay. The authors stress the political poWer of the city of Montevideo in distributing income away from the country and toward the ctybyimposinghigh taxeson agricultural exports. Theyalso report evidence of pervasive rent seeking, including a time senres on regulations from 1928 to 1983 that impeded trade, and of resource waste and lower savings and investment rates. In chapter S Federico Changanaqui and Patrick Messerlin focus on cis- guised protectonism in Uruguay through administered reference prices that are used as the base for tarifflevies. Their theoretical analysis shows that this disguised form of protectionism, by providing more protection for goods whose price falls below the reference price, induces Uruguayan producers to specialize in low-price, low-quality products, contrary to the county's comparative advantage in products requiringhigher skilled labor. High-price, high-quality import goods with prices above the reference pnrce escape ad- ministered protection, which fvors their consumption. Thus, an unintended effect of the reference price system is to encourage production of low-qual- ity goods and the consumption of high-quality goods. Changanaqui and Messerlin analyze the protective effects ofadministeredpricesapplied to 500 products and find that protection is strongly "magnified " In chapter 4 Wendy Takacs models the complex system of pro tection in the automobile sector that was instituted in 1971 to encourage leaning by The polticaleconomy of prokc6ionim in Uruguay 11 doing through automobile assembly. The chapter offers a useful lesson on how not to provide for growth and technical change. The combination of high tariffis (40 percent in 1990) on assembled automobiles, low tariffs (10 percent) on automobile assembly kits, domestic content requirements, and compensatory export requirements encourages the local production of au- tomobiles by finns thatassemble the imported kits. The cost to the consumer per car assembled in 1990 of this high effective protection is estimated at more than $6,000-more than half ofittransferred as arent to the automo- bile assembler. A net efficiency loss of over $2,000 per vehicle represents a waste of Uruguayan resources in this "screwd[river" activity. Takacs provides a novel formal framework for analyzing the welfare benefits and costs of domestic content and compensatory export requirements, mechanisns all too often used to promote industrialization. Ihe industries that grow pros- perous tmder this type of protective regime will lobby for contnued protec- tion and fail to introduce innovative technologies or to compete with other producers in the world market Chapter 5 byJaime de Melo, Claudio Montenegro, and Wendy Takacs deals with tdie gains and losses from bilateral trade agreements with Argen- tina (Conveno Argentino-Uruguayo de Cooperacion Econ6micaor CAUCE) and Brazil (Protocolo de Expansion Comercal Uruguay-Brazil orPEC) - The authors deve!op a theoretical model of the effidency and rent transfers impliedby the arrangements. They thenapplyagravitymodel to intaregional trade to estimate the increase in the volume of trade with the region result- ing from these preferential arrangements. The accord widt Argentina seems to have allowed uncompetitive manufacturing exports to that country, while th-at with Brazil concentrates on Uruguay's exports in agricultural and natu- ral resource-based products, more along the lnes of Uruguay's pattem of comparative advantage. Under the agreement with Argentina Uruguayan- exports enter duty free but are subject to a cupo or export coupon that ra- tions the rights to export to Argentina as a share of the Argentine market. Under the agreementwith Brazil the cupos are negotiated quantitiesrather than shares of Brazil's market Both systems could result in rents that allow inefficientfirms to continue to export, though econometric analysis of price data does nor reveal any significant rents. In chapter 6 Martin Ramaanalyzes the labor marketand trade reform in manufacturing and asks whetherunionizaion blocks the effects of trade lib- eralization. The Uruguayan case provides experimental conditions For an- swering this question since some of the trff reforms took place under the penod of authoriarian rule, when unions were forbidden. Barriers to trade give rise to rents to domestic firms, thus providing an incentive for labor to uionize and attempt to negotiate wage increases to capture some of these rents. Ramafinds that sectoral union membership rates in industrial sectors 12 EFmCS OF PROTECTIONISM ON A SMALL CouNTr1y THE CAsE OP URUGUAY are correlated with indicators of market power of domestic finns, including an mdustry concentration index and the effective rate of protection. Analy- ais of forty manufacturing sectors over six years shows that the employment elasticity to changes in nominal protection rates was significantly lower in' the period with active trade unions, and the real wage elasticity was higher. In Chapter 7Jaime de Melo and DavidRoland-Hoistuse ageneral equi- librium model to quantify the economywide welfare losses due to the com- binedinfluence of trade-and factormarketdistortions. Buildingon theanaly- sis in chapter 6, the model incorporates the effects of labor union activity, settingupan interaction between trade policyand the determination ofwages at the sectoral level. The simulations suggest that the system of administered Jimport prices distorts the Uruguayan economy more than the tariffs them- selves. The costs of administered protection-removing administered prices couldadd4percentto GDP-farexceedthose due to tariffs.Addingresources. squandered in rent-seeking activities could raise the total cost of adminis- tered protection to 8 percent of GDP. The simulations also suggest that as a consequence of removing administered protection, nearly 4 percent of the labor force (or 50,000 workers) would have to relocate, explaining some of labor's resistance to trade reform in Uruguay. Resources would flow from formerly protected sectors to agriculture, casing an expansion of up to 15 percent, and light manufactures. The immense difference in trade margins between imports, at 35 percent, and domestic goods, at 9 percent (see chap- ter 3), would also shrink, although the notorious ineffidency of the Port of Montevideo probably contributes substantially to this marked difference. Chapter 8 byJorge Roldos provides a long-run perspective on the rela- tionship between trade policy and ec -omic growth. Based on endogenous growth theory, Roldos tests for increasing returns to scale in Uruguay using neoclassical production functions. While the sum of the capital and labor exponents is greater than unity, it is not possible to reject the null hypothesis of constant returns to scale, so specialization according to comparative ad- vantage does not appear to lead to significant economies of scale, at least at the economywide level. Calculations of the Solow residuals show total factor productivity to be correlated with export growth, suggesting that export externalities help explain the residuals. Roldos finds cross-sectional support for the hypothesis that trade openness is associatedwith a high rate of growth in GDP per capita and that Uruguay's growth was lower than the cross-sec- tion model predicted. Notes 1. All cross-country comparisons are from the Sumrners and Heston (1991) data. We should note, however, the authors assign a C- grade to the reliability of the correc- ThepoEtlitica ecmnmy ofprotectionism in Unrguay 13 Lions for purchasing power for Uruguay. The comparisons of levels of GDP per capita across countries are therefore subject to a wide margin of error. Growth rates compari- sons arc unaffected. 2. Total costs of the import good in the market are increased because of the tariff surchargs An ad valorem duty boosts the cif. pricc to the importer by the percent- agc amount of the tariff. A specific duty boosts price to the importer by the amount I perunitso that the total importprice including the duty isPF + t Itcan easily be shown thatdomestic taxeson consumption combinedwith subsidiestoproduction atthesame ratc replicate the effect of a tariff. 3. Imports offoodstuffs, pharnaceuticals, and other products are sometimes re- stricted or prohibited for health, sanitary, or environmental purposes. Since such re- strictions and prohibitionsmay have astrong protective effect, domestic producers may lobby to enact them. This ype of restraint is not used in Uruguay. References Anichini,JuanJose,Jorge Caumont, and Larry Sjaastad. 1978. La Poittica Comercialy az -+otecci6n en dUruguay. Montevideo: Banco Central de la Repfublica Oriental del Uru- guay. Favaro, Edgardo, and Clauidio Sape1li. 1990. Etp oPrtvnotion andEconomic Growth. San Francisco, Calif.: Institute for Contemporary Studies. Favaro1 Edgardo, and Pablo Spiller. 1990. 'Uruguay. In A. Choksi, M. Michaely, and D.Papageorgiou, eds., The 7imingand Seuencing ojTrade Libera za iom London: Basil Blackwell. Olson, Mancur. 1965. The Logic of Colctdive ACtion: Public Goods and the Thwny of Goods. Camnbridge, Mass: Harvard University Press. Rodrik, Dani. 1992. 'Me Rush to Free Trade in the Developirng World: WhySo Late? Why Now? Will It last?" NBER Working Paper 3947. National Bureau of Economic Research, Cambridge, Mass. Summers, R., andA. Heston. 1991. 'he Penn World Tables (Mark V):An Expanded Set of International Comparisons, 1950-8." Quarterly Joumnal of Econodcs. Mar. 327-68. Protectionism and stagnation: an interpretative history Michael B. Connolly and faime die Melo Uruguay has such a long history of pro tection thatwe could have called this chapter 'Two hundred years ofprotectionismY' High levels of protection of manufacturing were maintained by taxing agriculture, the main activity in which Uruguay has astrong comparative advantage. Whatwere the causes of the early protectionism, and what have been its consequences? This book prvdes some answers by analyzng several f-acets of protectionism in Uruguay's recent history. Most of the chapters deal with the welfaLre costs of the inefficiendies introduced by the various forms ofprotectionismn-tariffs, administered prices, bilateral trading arrangements, and local content re- quirements. Chapter 8 takes a longer-run perspective on the relationship between the trade regimne and growth since the 1950s. This chapter complements the more specific analyses that follow by pro- viding -an overview of protection against which to judge the rel-ative imnpor- tance of the various instruments of protectionism in Uruguay. Italso pl-aces Uruguay's overall performance in a comparative context, -and for the reader unfamili-ar with Uruguay's economic policies, the chapter sketches some of .the milestones in the country's long history of protectiomism. * What is striking in this description of protectionism before World War II is that a large nulmber of the protective instruments :adopted in the nineteenth century are stil around today. The p-anoply off measures that emerged during the deepening of protection from the 1950s.up to.the 1974-82 liberaHization episode is typical of the measures implemented by countries following a development strategy based on industrialization through import substitution. Under the militar rule of 1974-82 and later * after the retum to democracy in 1985, wide-ranging reforms were canTied out A fascmnating aspect of the 1974-82 period is the apparent break in the downward trend in postwar economic performance (see figure 1.1 in chap- te-r 1). It is tempting to concdude that this resumption of growth resuflted from the reforms, whic inlddtaeadfnnillberalization as well '74 . .dProtnism and sttiaonW an interprelaive his" 15 :as an overhaul of the fiscal system and a concurrent effort at stabilizing the economy. To any student of the influence of economic policies on performance, the causes of Uruguay's stagnation are as fascinating as those of the Asian economic ˝miracle." In exploring the contribution of various factors to this stagnaadon, we start with a comparative analysis of Uruguay's growth perfor- mance, noting an improvement in its relative position following the 1974 refonns. We fit a simple cross-country growth model to a sample of thirty-five comparable developing countriesand find thatneitherinvestmentlevels nor initial conditions (GDP per capitaandeducationlevels) satisfactorilyaccount for Uruguay's poor performance. Uruguay's perfonnance is worse than pre- dicted by this simple growth model. So we look for other causes. Drawing on recently collected data by Rama (1993), we note the high incidence ofrcnt-seeldng activities in Uruguay and relate them to the pervasive regulation of the economy. And we consider whether there really was a break in performance after the 1974 reforms, looking at evidence of changes in investment and savings behavior and at the role of external shocks. We conclude that there was probably a slight inprovement in performance attributable to the reforms carried outin 1974, but as the more detailed analyses in the remaining chapters in this voluxme suggest, Uruguaywas stillahighly distorted economy by the end of the 1980s. It is difficult not to conclude that the system of regulations and protec- tion in Uruguay has been made-to-measure. In the end, protection was adopted more in response to lobbying acdvity than as a part of a rational design to promote industrial activities. Ramna (1993), in a detailed examina- tion offoreign trade legislation between 1925 and 1983, counted 1,849 pieces oflegislation that explicidy identified their promoter. Rana translates one of them, dernvingfrom an application in 1937 concerning Sacman-type cIay tiles: "Mr. Horacio Acosta y Lara, proprietor of a nadonal firm manufactur- -ig this product... considers that the low price on which customs tariff is based damages his own interests and puts him at a disadvantage relative to foreign industry.' -What accounted for this made-to-measure trade regime? Industrial con- centration must have piayed an important role. The manufacturing sector. in Uruguay is concentrted in a few hands: as late as 1978 there were only 310 establishments that employed more than 100 workers (up from 210 in 1960). It is therefore not srisig that lobbying activity by manufacturers has been intense. According to Olson (1965) the smaller the number offinis that benefit from a specific action, the larger the benefit share to each firm from that action, and hence the more intense the lobbying activity. Considering this strong lobbying for protection, it is natural to ask how much the reforms of tie 1970s and 1980s actually reduced effective protec- - 16& EMCIS OF PROrEcmToMNsM ON A SMALL COUNTRY: THE CAsE OF URUrUAY tion.Analysis suggests that, atleastuntil the mid-1980s, the reforms progres- sively eliminated primarily redunadantprotection, with tariff reductions hav- ing little effect on the spread between domestic and international prices (Macadar 1988). These estimates, based on compaisons of domestic and foreign comparable products, are approximate -at best. But it is clear that Uruguay has alow trade share in GDP forits size, despite a noticeable increase in the share of exports in GDP following the reforms begun in 1974 (figure 2.1). These reformnswentalongway toward removing the most egregous trade restrictions, such as the prohibitioni of' capital goods imports, but Uruguay's foreign trade regime remained heavily controlled throughout the 1980s. Birth of protectionism: 1815-1939 High and vaniable rates of protection are not new to Uruguay--they were already in place in 1815 (table 2.1). Protectionwas common atthe tme (Great Britain's repeal of the corn laws in 1848 was an exception), holding sway until 1850, when a bilateral trade accord betweenFrance and Great Britain ush- ered in an era of free trade in Europe. Uruguay, npervious to Britsh influ- enceand to commercial interests, continued to pass stiff amrifflaws that dis- criminated increasingly against consumer goods. Poor harvests, diminished livestock outputand accumulated fiscaland merchandise deficits led to the agricultural and finanial crises of 1874, and then to the first attempt to simultaneously protect Uruguay's nascent indus- Figure 2.1 GDP and exports (three-year moving averages), 1935-92 (billions of 1978 pesos) Output Expans 50- 12 40 8 30 Otu 6 20 Fgure 2.1 GDP and fflc~~~~ports 4treya oigaeae) 959 10 2 0 0 1935 1945 1955 1965 1975 1985 SO&CeAuthors! calculations, based on Centbal Bank of Unrguay data. Protectionism and stagnation: an interfrstaidve htstoty 17 Table 2.1 Tariff schedules, 181 5-1930 (percent) Machinery Year ;etneral Maximum Minimum GConswpUn IntermedIate and drLgs 1815 25 45 0 0-45 0-25 0-25 1829 15 25 0 5-25 0-15 0-15 1831 18 25 0 5-25 0-18 0-18 1833 15 25 0 5-25 0-15 0-15 1837 15 31.5 0 0-31.5 0-15 0-15 1850 20 30 0 15-30 0-15 0-15 1856 15 35 0 15-35 0-15 0-15 1861 15 22 0 0-22 0-15 0-IS 1875 15 90 0 0-42 0-90 0-35 1886 30.5 5t 0 6-51 043 -30.5 1888 31.0 SI 0 6-51 0-43 0-31 1912 None 51 0 5-51 0-31 0-25 1927 None 51 0 5-51 0-31 0-25 Source Adapted fro Favaro and Sapelli (I 989, table 2.12). try andraise revenues. Duties of 10 to 20percentwere imposed on mostim- ported goods, but rates reached as high as 90 percent for lithographs and published works.1 In October 1875 duties were lifted on primary materials, industrial and agricultural equipment, and most inputs. The motivation for this pattem of duties was protection and development of the import-substi- tution industry in Uruguay. InJanuary 1888 a 31 percent basic dutywas set on merchandise imports, while imported parts for assembly (in this case, of steamboats), industrial equipment, and fenacng material were exempt or had low duties. Duties on consumption goods such as cheese, butter, brushes, shoes, clothing, hats, matches, and pu:ses were high, ranging from 44 percent to 51 percent. Ac- cording to published report of the Treasury Commission of the House of Representatives, the high duties on consumption goods and low duties on inputs were intended to promote Uruguayan indusy and to raise revenues. The effect on imports of consumer goods in 1890-1900 was devastating. Despite high rates of population growth and immigration the value of im- ports fell 44 percent, from 17 million pesos during 1872-74 to 9.5 million pesos during 1890-1900. As earlyas 1859 there was concern that the high rates of protectionwould lead to smuggling (Favaro andSapelli 1989,42)t Despite talkoflowering tariff rates to discourage smuggling, protection continued to rise. A natural ques- tion to ask is: Why was protection so high in Uruguay while the rest of the world was pursuing free trade policies? There are no obvious polidcal economy reasons for the rise in protec- tion in the nineteenth century. The agroexporters were in power, and they clearly benefited from free trade. The persistence and consolidation of pro- : 718 IEECTS OF PROTECnONISM ON A SMALL CouNwTr: THE CASE OF URUcUAY tectionism in the early parts of this century during the presidencies of Batie y Ordofiez (1903-07 and 1911-15) can be at least partly explained by the shift of political powecr to Montevideo and by the influence of anarchist im- migrants. Batlle y Ordoiiez promoted social legislation and active economic policies. The next watershed was the temporary admission law of 1911 and the raw material and intermediate input law of 1912. The duty drawback and temporary admission law of 1911 allowed the duty-free import of all primary materials and internediate goods used for production, provided that the finished good was exported within a specfied period. The tariff law of 1912 forinally established special duty treatment for imports of primary materials lused in the production of goods sold in the domestic market as well. The legislation of exceptions had begun, introducing new incentives for rent- seeking activities. Once again the motivation was to promote domestic industry-and to createjobs. The log of the House of Representatives (October 4,1912) states that "every nation which protects its industries to create employment and icrease its population thanks to the greater employment, opens its doors to primary materials not available or in short supply at home in order to carry outmanufacturing inits national enterprisesandfactories" (cited inAnichini, Caumont, and Sjaastad 1978,19-20). The policy of high effective rates of protection for manufacturing has endured to this day. Tariff schedules still favor the import of raw materials and equipment (with the exception of automobile parts). The basic draw- back- law remains in existence as well. Supply difficulties during World War I reduced imports and thus tariff revenues. With less competition from abroad, domestic manufacturing grew: the numberofmanufacturingestablishments rose from 1,000 in 1919 to 7,400 by 1930. When the war ended, exports fell and imports rose, prompting legislators in the 1920s to establish higher reference prices for calculating duties on some goods, particularly consumer items facing greater foreign competiton. Duties were levied on either the dedared c.if. value or the ref- erence price, whichever was higher. This mechanism of protection is still in place today in Uruguay, albeit somewhat reforned. Its effects are analyzed in chapter 3. The stock market crashes of 1929 and the subsequent finandal panics inducedanewboutofprotectionism onaworldwide scale.The Smoot-Hawley Tariff in the United States in 1930 was followedwithin twoyears by increased tariffs in sixty other countries (Ethier 1983). Beggar-thy-neighbor policies in the industrial countries sought to protect domestic manufacturing and employment, but instead the contraction in trade led to a reversal of the substantial economic gains that had arisen from rade. On August 6,1931, Protectionism and stqgndon: an intapretative hist 19 Uruguay passed a law authorizing the executive branch to raise duties to as high as 48 percent (up from 31 percent) on goods forwhich therewas "simi- larnationalproduction.' OnAugust20 the presidentwasalso given the right to suspend imports for up to a month on awide range of goods thatwere also produced nationally. On the export side multiple exchange rates were established in 1933 through a compensating system of preferential exchange rates with up to a 50 percent premium for exports. In an attempt to promote domestic indus- tries, subsidies were given through tax exemptions (franquicias) for indus- tries that moved to Uruguay or expanded their local productive capaciy. The antiprotective effect of a revaluation of the peso in December 1937 was un- done by the reimposition of import licensing. What is remarkable in this period of Uruguay's economic history is that the economy grew fairly rapidly despite the relatively high protection. Uru- guay benefited from the boom caused by the meat shortage during World War IL and the "natural" protection afforded by the depression was largely inevitable. But the relatively high rates of protection seemed to have had no perceptible effecton growth. Anichini, Caumont, and Sjaastad (1978) char- acterize theperiodas one of growth through importsubstitution,which ended in the 1950s, when the growth opportunities made possible by protection were exhausted. These authors, however, measuredgrowrh at domestic prices, which overvalues the increase in domestic output due to protection. Mea- sured at constantworld prices, growth up to 1950 is lower. Consolidation of protection: 1941-74 The Korean war provided another, and the Last, positive external shock. On the whole, Uruguay's economic history after World War II is one of increas- ing protectionism and market interventions up to 1974. There were, how- ever, several attempts at liberalization after 1959, including efforts to ratio- nalize the trade regime and attempts to reduce the antiexport bias of the protective regime by extending various subsidies to nontraditional exports. 1941-59 Congress approved abill on export and import control in 1941 that consoli- dated the measures that had been implemented in piecemeal fashion dur- ing the depression. The law also authorized the president to create a comp- troller of exports and imports with authority to prohibit imports that competed with national production. Foreign e-xchange was allocated by the central bank (Banco Reptiblica Oriental del Uruguay) by country and byitem, with pniority going to raw materials, "criticalr consuimer goods, intermediate 20 EFECTS OF PRoTECIoNIsM ON A SMAL.L CouNTR THE CAsE OF URuGUAY and capital goods for the agricultural sector, and raw materials and machin- ery for the manufacturing sectoi. Multiple exchaange rates were introduced for broad categories of goods: necessities, intermediates, and luxuries. While this system made some sense during war years, itmade little sense later on, when it stifled foreign trade. Favaro and Spiller (1990) give a flravor of the discriminatory nature of the bill, describing how the law established norms for the distribution of foreign currency according to the "need each firm showed, the number of employees and their salaries, the volume of activity, and the time that the firm had been established in the market" (p. 347). During a period when world trade was booming, Uruguay was adopt- ing a trading system based on bilateralism, mnuch along the lines pursued in Eastem Europe. Uruguay registered respectable gro*th during this period of rising pro- tectionisrn.Justas it had benefited from not being involvedin World War IIL it now benefited from the Korean War, which led to another export boom. Real GDP growth averaged 3.9 percent a year and real industrial product grew 5.8 percent from 1942 to 1958. And once more, manufacting growth was fueled by import substitution. 1960-73 The period 1960-73 was markedbystalledgrowth and economic stagnation. Some reforms had been introduced in December 1959, induding tempo- rary unification of exchange rates for commercial transactions. The tariff reforms of l959, however, ended up reinforcing the multiple exchange rates for imports, while the free market rate applied to exports and most imports of raw materials and capital goods. The executive branch was authorized to impose surcharges of up to 300 percent, to require prior deposits forimports, and to prohibit 3inports for up to 180 days. Moreover, the new tariffs were levied on administered prices rather than on ci.L prices. Despite lobbying by the Chamber of Commerce for a 150 percent surcharge, the executive branch set the surcharges in September 1960 at 45 percent, 75 percent, and 150 percent (for luxury" goods.) Naturally, with such high surchargesadded to the triffs, demand for the affected goods nearly disappeared, and cus- toms collections declined. lJruguayjoined the Latin American Free Trade Area (LAFTA) in 1960, which was established to provide preferential treatment for regional prod- ucts. Implementationwas patchy, however, and tlhe preferences seem to have resulted in trade diversion rather than trade creation (see chapter 5 and Nogues and QuintaniUla 1993). Protection led to even more intervention in 1964, in an effort to com- pensate exporters through subsidies of up to 20 percent of the f.o.b. value nPrtectionism and slagnation an intypretative histor 21 and by relievng them of import duties on imported inputs used in export production. Discimination againstagriculture intensified, -is taxes were lev- ied on exports of traditional products like wooL beef, and hides. The objec- tive was to promote industrial exports. The result was to superimpose addi- tional layers of taxes and subsidies on the existing structure of tariffschedules, prior-deposit requirements, and multiple exchange rates. D Despite the high levels of protection and the export subsidy schemes, industrial production increased by only 1.1 percentayearin real tenms from 1959 to 1973-slower than the rate of population growth. Quotas imposed on imported materials because of balance of payments problems also con- tributed to the stgnation of manufacturing. Growth possibilities based on import substitution had been exhausted. How high was protection during th}is period? To estimate protection Favaro and Spiller (1990) use a measure-based on the ratio of import to export prices lor a sample of goods-that includes the effects of adminis- tered prices. Average nominal protection, according to this measure, is very high until after the reform of 1974-79: Year Rate 1961 384.4 1964 684.1 1968 578.1 1971 534.5 1974 452.4 1979 69.5 1980 62.3 1982 52.7 While the magnitude of these estimates may be questioned (for a discussion see Favaro and Spiller 1990, 351-2), other evidence supports the claim of a verystrongprotective effectfrom the numerousinterventions, includingother estimates of protection (see Mezzera and de Melo 1985, for example) and measures of the restrictiveness of the trade regime (the share of final goods to total imports fell from 60 percent in 1951 to 31 percent in 1974). In any event, whatever the precise magnitude of protection provided by the system ofinterventions, itis dear thatprotectionwas on the rise until thewide-rang- ing reforms that began in 1974. Economic reforms of 1974 Following an external crisis caused by falling terms of trade and a ban on beef exports to the European Community, Uruguay began a sweeping.rever- 22- Enas1 OF PROTECTIONSM ON A SMALL CouNwra THE CASE OF URUGUAY sal of its proteetionist polides in 1974. Controls on domestic prices and in- terest rates were relaxed, and a floating exchange rate was adopted hIn 1975 qantitative restrictions were eliminated, and in 1979 a program of tariff reductions was adopted. (See appendix for detils on reforrn measures.) Even after these reforms, however,.he range of instruments used to control trade and influence resource allocadon remained large. Import deposit requirements and quotas Some of the trade reforms begun in 1974 have been loosely referred to as the elimination of quotas on imports. Technically, the system in place up to 1975 was not a formal quota system in that there were no set limits on the quantity orvalue of imports overa specified time period. Rather, for imports above specified amounts a set of prior import deposit requirements came into play. In prctice, however, the system operated like a set of value-de- nominated quotas once the import deposit requirements and inflation be- came so high that the cost of the deposits for over-uota imports was almost prohibitive. UntilJune 1975, imports subject to prepayments had to be reg- isteredwith the central bankbefore orderswere placed abroad, and import- ers had to deposit, in Uruguayan pesos, a given percentage (over 100 per- cent) of the cif. value of imports. These deposits were refimdable after six months, upon application, but they accrued no interest Imports by public agences or financed by loans from international agencies or foreign gov- ernmentswere exemptfrom the depositrequirement. Other exemptionsfor particuar types ofproductswere occasionallyintroduced. Importers not only had to forgo interest earings on the deposits, but in an inflationary envi- ronment, they also suffered a loss of real purchasing power. The depositrequirements orprepayments applied only to importsabove a stated percentage of base period imports. Both the percentages and base periods were changed frequently, usually two to five times a year? The de- posit share also varied, by productand over time. To illustrate: in 1968 there were three import deposit-rates: 150 percent, 300 percent, and 400 percent The 150 percentrate applied to goods exemptfrom the taziffsurcharge, goods subject to a 10 percent surcharge, and some goods subject to a 60 percent surcharge. The 300 percent rate applied to other goods subject to a 60 per- cent surcharge and to goods subject to a 90 percent surcharge. The 400 percent deposit applied to goods subject to surcharges higher than 90 per- cent. In November 1971 the deposit requirements were tripled to 450 per- cent, 900 percent, and 1,200 percent respectively) Deposits held by monetary authorities rose after the increase in deposit requirements in 1971 and fell after the 1975 liberalization. The decline in the realvalue ofthe*fmds on depositcanbeused to estimate the tariffequiva- Protcktionism and stagnaton: an intepretaive histoy 23 lentin real tenns of the inport deposit requirements. If prices increase by a factor of rwhile the deposits are being held, this is equivalent to a tax of r/(1 + r) percent of the funds held. Thus, a price increase of 100 percent (which implies r= 1) cuts the purchasing power of deposited funds in half, an increase of 50 percent cuts it by one-third, and so on. Given that the de- posit requirements are multiples of the value of imports, the impact is also multiplied. The real tariffequivalent of the import depositrequirement, T, can be calculated as: (2.1) .T= (1+ r) where D is the percentage of the value of imports that must be deposited. A I100 percent deposit requirement implies D = 1. A 400 percentrequirement (D =4) togetherwith a price increase of 100 percent (r= l) during the de- positperiodimplies that the importergivesup purchasingpower equivalent to 200 percent of the value of the imports requiring deposits. Similarly, a 1,200 percent deposit requrement implies that the importer gives up pur- chasing power equivalent to 600 percent of the value of imports. This approach can be used to calculate the ex post tariff equivalents- thatis, once the inflation rate for the period is known-of the lowest and the highestapplicable import deposit requirements for fnds. From 1968 to the dird quarter of 1971 the lowest and highest deposit requirements were 150 and 400 percent; from the last quarter of 2971 to the second quarter of 1978 theywere 400 percent and 1,200 percent. Estimated tariff equivalents of the import deposit requirements range from a low of about 8 percent for the lowest deposit requiremenit during the period of lowest inflation to ahigh of more than 430 percent for the highest dep osit requirement during the most inflationarypenod. T ese high estimated tariffequivalents are consistentwith contemporary observatons that the cost of the import deposits to importers was so large that itwas virtually prohibitive.5 There are two reasons to suspect that these estimates of tariffequivalents may overstate the restrictiveness of the import deposit scheme. First, if dLe higher depositrequirements were actuallyprohibitive, there would have been at least some amnount of 'waterin the tariff equivalent' for these categories. But since at least some deposits were being held at all dimes during the pe- riod 1968-75, the lowest depositrequirementwas neverprohibitive and can be interpreted as a rough lower-bound estmate of the restrictiveness of the system. Second, firms thatargued that they couldnotmainrain their actvity levels withoutan increase in theirlicense allocations were granted increases. There is also at least one reason to suspect that these estimated tariffs may underestimate the real cost to importers. For commodities for which refer- ence pnces were established, prior deposits were calculated on the basis of 24 Ercrs OF PROnEcroNlsM ON A SMALL CouNRY: THE CASE OF URUGUAY reference prices when they were&higher than the ci.f values. An artificially higher valuation basis would result in larger deposits and a higher ad valo- rem tariff equivalent. A somewhat lower measure of the restrictveness of the import deposit requirements is implied by the outcome of a central bank auction in October 1974 ofUS$625,000 in deposit-exemptforeign exchange forhigeduty luxury imports. Importers paid up to UrS1,500 per U.S. dollar allocated plus the Ur$1,392 costof foreignexchange atthe time ofthe auction. Thatmeans that importers were willing to pay an extra 108 percent rather than pay the prior deposit Thisbidis roughly the same as the estimated triff equivlentfor the lower deposit requirement (96 percent for fimds deposited during the firt quarter ofl974) but is considerably lower than the estimated tariffequivalenLt for the highest deposit requirement of the time, which was over 200 percent. The prepaynent system was liberalized as part of the sweeping reforms of 1974-77. In April 1975 the govermnuent announced that the system of * deposits and quotas would be eliminated as ofJuly 1, 1975. To avoid a large increase in liquidity as import deposits were released, an advance deposit requirement ofS35 percent of the ci.f value, to be held for 150 days, was imposed on all private sector imports as of April 14, 1975. The 35 percent deposit was maintained until December 1975, when it was replaced by a 7 percent surcharge on all imports except some investment and capital goods. The elimination of import deposits and the associated quota-like system lowered costs for-importers who were importing over quota and, more im- portant, allowed substantial changes in the composition of imports and in their allocation among firms. Import quotas below the prior-deposit thresh- old were granted to firms according to the value of their imports in the reS evant base period. The system was inherendy discriminatory, and it severely impeded shifts in the allocation of imports toward more efficent firms or new finms. Maintaining the prior-deposit scheme would have made it diffi- cult for firms to respond to the trade reforms by importing the capital goods, raw materials, or intermediate goods needed to expand export actities or to startup new businesses. Favaro and Spiller (1990) report that eliminating the prior-depositrequirementallowed import-competing sectors to become more effident, so that output increased during the first phase ofliberaliza- tion despite the rise in competing imports. The industrial sector grew at an annual rate of more than 4 percent a year for the three years following the liberalization measures. Tariff reform program of 1 980-89 While the system of prior deposits and quotas was being reformed, the tariff system remained highly protective. In 1976 five different duties were being Pm :ecionism and sagapion: an intetrpraaive history 25 levied on imports: the tariff (applied to the higher of the cif. price or the reference price), aconsularfee of4percent of the fo.b. value, an 18 percent duty (on the higher of the c.i.f. price or the reference price), an exchange rate duty (ranging from 0 percent to 200 percent OFl ae c.if. ralue), and a 7 percent consignment tax forsome inputs andcapital goods. The sum of these duties varied; itwas 3 percent for tractors, 47 percent for electric motors for industrial use, and 197 percent for electric vacuum cleaners. In 1979 Uru- guayjoined the General Agreement of Tariffs andTrade (GAIT), signaling an end to its longstanding practice of providing subsidies to nontraditional exports. The second phase of trade liberalization began in earnest inJanuary 1980 and involved six equal tariffreductions to be taken at the beginning of each year. From twenty-eigh tlevels of duties averaging 45 percent, with apeak rate of 116 percent, tariffs were reduced to eightrates, with asimple average of 36 percent and a maximum rate of 75 percent The crawling peg exchange rate policyintroduced in 1978 was replaced by a floating rate in 1982. The real apprecation that had occurred turned out to be unsustainable. Capital had flowed into the country following the opening of the economy and difficuldties in Argentina, iinancing an invest- ment boom that led. to real growth in GDP averagig mure than 4 percent from 1973 to 1981. Realized real rates ofinterest, however, became unsustain- ably high in 1980 as agents anticipated the collapse of the pre-announced exchange rate (known as the tabita) by reducing capital inflows. As intera- tional rates rose higher in 1981 and commodity prices other than oil fell, Uruguaysufferedadebtcrisis and economic collapse of severeproportions- real GDP fell abruptly by 9 percent Tariff reductions were abandoned, re- placed injanuary 1983 by a new tariffstructure with five levels and a 15 per- cent export tax on traditional exports (table 2.2). Tariffrates continued tofall, butmore graduallythanplanned.The only reversalwas asmall temporaryincrease inJune 1985. Tariffswere reduced 5 percentage pointsin 1989, andinAugust 1990 the highesttariffwas reduced to 30 percent. The reforms also narrowed the range of taff rates, bringing Table 2.2 Chronolog of tariff rates, 1979-87 Efectied te Primryrmazids intermenxategoods Find goods Decenber1979 Oto18 22to94 76to116 January 1982 10and 15 25 3545 55 65 75 January 1983 10 254050 55 June 1985 Is 203545 60 August 1986 10 203545 50 August 1987 .10 203040 45 ŁOurc Mamdar (I1988). 26 E}FEC1s OF PROTECrEOnISM ON A SMALL CoUNTRY TmE CASE OF URUGUAY Uruguay closer to a uniform tariff. However, effectve protection remained muchhigher-andmore dispersed than these tarifffiguressuggestbecause of the operation of the reference price system (see chapter 3) and the special tariff regime for automobiles (chapter 4). Reference and minimum exportprics, 1982-92 The old reference price system was replaced in 1982 by the current, more formalized and fully specified system of administered prices, the change prompted by the difficulty of implementing the antid umpiag and antisubsidy law ofJune 1981. Fifty-four types of products have referenea prices-textiles and apparel alone are covered by 177 separate reference prices. The mecha- nism is simple: when the dedared c.i.f. import price is less than the refer- ence price, the official duty is applied to the reftrence pri_e. Thus, the world price of synthetic tufted carpets maybe US$3.22 ayard, butitsreferenceprice is US$750 a yard. As a percentage of the world price a 40 percent duty rate represents an effective duty of 93 percent A new minimum export price was added to the system of administered prices in 1983. The name is something of a misnomer since itapplies not to Uruguayan exports butto Uruguayanimports-other countries' exports. Be system covered seventy-five types of goods in 1990 (thirty-five of them trans- ferred from the reference price system). The minimum export price system is more protective than the reference price system since it adds a moving surcharge equal to the difference between the declared importprice and the mnimum export price. Consider automobile tires, with a world price of US$3 .80 a pound and a minimum export price of US$4.30 a pound. Import- ers declaring an import price of US$3.80 would pay not only the official duty of4O percent on the $4.30 price, but also the US$0.50 difference between the declared world price and the minimum export prices for a total effective duty of58 percenL Nantrally, there isa strong incentive for importers to avoid the surcharge by declaring an importvalue equal to the minimum export price. The system ofadministeredprices has been reformedinrecentyears by reducing the ratesand the numberofgoodscovered, but the systemremains in place. By boosting the true tariffs above the apparent tariFf, this system undermines the country's trade liberalization efforts. The special regime for automobiles, 1970-92 Another form of import deterrence is achieved by the elaborate system of domestic content and compensatory export requirements established in October 1970 in the automotive industry. In 1990 importers offully assembled automobiles paid a 40 percent tariffandhad to have 'compensating" exports Potectionim and 5stgnation: an interptalive his" 27 of automobile parts equal to 70 percent of the Łob. value of the fully as- sembled automobile. Importers of automobile kits had to pay a tariff of 10 percent, have a minimum Uruguayan value-added content of2O percent, and have compensatory exports ranging from a minimum of 20 percent for a domestic content of 40 percent or more to a maximum of 60 percent for a domestic content of 20 to 25 percent The compensatory export scheme was less restrictive than it appeared, however, since export receipts could be purchased on the secondary market for 8 to 35 percent and used to satisfy the compensatory export requiremenL Export taxes and subsidies In addition to controls on imports Uruguay has a long tradition of taxing its traditional exports of beef, wool, and tanned hides. Furthermore, exports of key raw materials such as cattle on the hoofand untanned leather vere pro- hibited in order to provide cheap inputs to the meat packing and tanning industries. To compensate for this bias against tradable acdvities, export subsidies were introduced for nontaditional exports in 1964, and expanded in 1967and 1970.Nontrditionalexportsalso received exchangeraterebates, while traditional exports were penalized by an unfavorable exchange rate. Clearly this complex structure of import protection and promotion was a strong incentive for lobbying and rent-seeking activties. Bilatercr trade afrnngmntg- - Bilateral trade arrangements with Argentina (CAUCE, negotiated in 1974) and Brazil (PEG, 1975) were an integral part of the reform package launched in 1974. The bilateral arrangements were intended to reduce Uruguay's chronic trade deficit with its neighbors and to secure access to these mar- kets. The anrangements, which are descibed in detail in chapter 5, provide forfree access into Argentina and Brazil for a selected number of lUruguay manufactures. However, free access was limited by quota ceilings for each product Thepreferenceswere regulated through coupons, which controlled access to the Argentine and Brazilian markets. Cross-country evidence on trade, growth, and stagnation What happened to growth ratesas Uruguay's economy became progressively more regulated andpro tected before 1974, when refonns gradually opened the economy to foreign trade? Four patterns stand out in the long-term relation between GDP and exports (see figure 2.1): stagnation of exports between 1953 and 1973;6 two 28 *EFcEsC OF PNorEcnoNisM ON A SMALL CouNTRY: THE CAsE OF URucuAY growth periods, fom the late 1930s to the mid-1950s and after 1973; an as- sociation between exports and GDP after 1974; and an explosion in exports after 1974. The question then is whether the changes can be attributed to policies or to external events. The first place to look for an external explanation for the rise in exports and tie acceleration in economic growth after 1974 is an improvement in Uruguay's commodity terms of trade. Far from improving, the terms of tde fell from 226 in 1973 to 122 in 1974 and stayed at or below 100 (with 1978 = 100) until 1987. Despite this worsening in the tenns of trade, real GDP ex- panded rapidly from 1974 to 1981 (figure 2.2). During a similar worsening in terms of trade from 1950 to 1972, income stagnated. The detenoration in the terms of trade may have contributed to.the slowdown in growth in this earlier p eriod, but not in the later period, an observation confirmed by Rama (1991) using causality tests that reject Granger causality from the terms of trade to outputgrowth. Anotier external factor thatprobably did play a role was the macroeconomic situation in neighboring countries, especially Ar- gentina Uruguay's boom during 1979-80 and the ensuing deep recession in 1982-83, when real GDP fell by 14.7 percent, were strongly influenced by the similar boom-bust cycle in Argentina27 Cross-section analysis using average growth rates for a sample of thirty- five countrieswith GDPpercapitaabove $3,800in 1973 (Surmmersand Heston 1991) shows Uruguay, alongvwith Argentina and Chile, its tvo Southerm Cone neighbors,with the lowvestaverage GDP growth for the tvo periods combined, 1950-74and 1974-88 (figure 2.3). Uruguay's relative position in the sample Figure 2.2 Terms of trade and GDP growth, 194249 Reo GCDP (billions of 1978 pesos) Term of trde (I 978=I 00) 40 150 30 100 0~~~~~~~~~~~~~~~~~~~5 1940 1950 1960 1970 1980 1990 Sowgce GDR. IM Anemcnirnd 9nanckt Stndstkssterirs of trade: Central Bank of Uruguay. JWotectionism and neag7adon: an iniepretative his"oi 29 improved during the second period, however, because while performance deteriorated for most of the sample following a series of price shocks begin- ning in 1973, UJruguay m-anaged to at least stand still. To explore some of the underlying determinants ofgrowth, we fit asimple growth model to the sample countries, using investmnent, initial income lev- els, and education levels as key explanatory variables.8 Since the model ig- nores other factors, such as trade policies, that could also affect growvth, the results are merely suggestive. But they do broadly confirm those of recent. cross.-courntry growth exercises, such a-s Barro (1991) and Levine and Ren elt. (I 992). (For this sample of relatively advainced developing countries, how- ever, catch-up from low income levels is significant only in the first period and the influenceof education variables--proxies for human capita-is only marginal; table 2.3.)Jorge Roldos takes acloser look in chapter 8at the same sample in an extended model thatattempts to control for the effects of trde policies on growth. More interesting here is the difference between actual and predicted growth rates for Uruguay during each subperiod, after the effects of invest- ment and education are taken into account. During 1960-73 actual goth was 2.5 percen thelowthe value predicted by the regression-byfar the larg- est negative residual in the sample. For the later period, 1974-88, actual Figure 2.3 GDP growth for thirty-five countries with GDP per capita above $3,800 in 1973 F&irent 1974-886 8 7 6 ISL f: C : l -PNe O' ~~~~~AUT M e : 0 NZLo -i ~~~~~~ARGO -2 : -2 -I 0 I 2 3 4 S 6 7. 6 PercenL 1 95r-74 Source: rnmers and Hreston (1991), wih amounts adjustedto have the sne purchangpoweras in the United Stints. -= '30 EiFns OF PROTE=TIONISM ON A SMALL CouNTRY: THE CASE OF URUCUAY Table 2.3 Regresion results for cross-country growth detenminants (dependent vcrtabe. oerage GDP per @1itO gowth) -Phlmaryschdoo Secondrscol Red GDP Investment Constant enrallbt ratio enralfient rtio per capita GDP Period term (EDPJ' (EDS) (InGDPCQ (AIW) F 196D-72 27.2 0.014 0.02 -3.4 0.11 172 (4.0) (0.015) (0.01) (0.51) (0.03) 1973-88 -23.5 0.08 -0.02 1.5 0.19 9.1 -(9.2) (0.04) (0.01) (08) (0.03) Note: Sam ple aftiryfive countries shown i jfgure 23. Nunbems in parernheses are standard errors, based on Whies (1981) heteroskedasticity-cnsiswnt coimrmnce matrix. a. Bgining of period values. Sowcc. Authors' dcubtions based on data from World Bank (various years) and Sumnmers and zon (1 991). growth was still some 0.3 percent below its predic:ted value, but the differ- ence was not significanL Thus the regression analysis confirms a relative im- provement in the period af-,er the initiation of the reforns. Rent seeking, reforms, and investment One obvious place to look for reasons for Uruguay's dismal economic per- formance since 1950, especially until 1974, is the distortionary effe cts of trade poicies and othermarketinterventions. The microeconomic effects of such distortions, which are explored in some of the chapters that follow, do not translate directly into effects on growth. And while chapter 8 looks for fur- ther evidence on the relation between trade policies and growth, it does not account for other possible linkages between policy and growth, such as how the policy environment and distnbutional activities affect growth by influ- encing investment and domestic savings. In particulars there is evidence of a broader stuge in Uruguay over the distribution of income between the three main social groups-civil servants, manufacturers, and trade unions- and of its effect on investmentand growth. There is also evidehce suggesting that the reforms initiated in 1974 led to a significant upward shift in private savings and investment during 1974-82. Rent seeking and investment As the discussion of the array of trade control instruments in Uruguay makes clear, discrctionary polices were widely used to influence resource alloca- tion. From the 1930s to the 1960s restrictive regulations were used to close the economy to foreign trade. From the mid-1970s, as the economywas pro- gressively opened to foreign trade, discretionary measures such as export subsidies were used to promote foreign trade. A considerable amount ofrent- seeldng activity would be expected in such an environment. Protectionism and stagntion: an intuvpretative hisutoy 31 Adetailed examination offoreign trade legislation from 1925 to 1983 by Rama (1993) yielded 1,849 pieces oflegislation thatexplicidtlyidentified their promoter-or 3973 if those that concerned only a small number of produc- ers are induded as well. A sense of the importance of these rent-seeking activitiesin relation to the overall size of the economy can be derived by scaling rentseeking (measured as the number of trade-related statuLes benefiting a - - single fi orindustry) by thelevel of economicactivity over the period 1925- 83 (figure 2.4). The scaling shows two peaks, one-during the 1940s and one during the 1970s. The most intense penrod of rent seeking occurred during the import-subsdtution regime of the 1940s. The second peak is almost en- tirely accounted for by rent-seeking acdvities associated with exporting. Using dummy variables to capture changes in policies and institutions, Ramna finds that rent-seeking activities intensified during militar dictator- ships and during periods ofintense imporisubstitution (1931-69) or export promotion (1974-79). The estimated coefficients also suggest that import substitution had a stronger effect on rent-seeking activity than did export promrotdon. Particularly interesting is the relationship Rama finds between rent-seek- ingactivityandl.ong-run growth. He regresses the residuals (or 'innovations") from univariate models of the growth of outputand exports on indicators of rent-seeking intensity (following the approach of Blanchard and Fischer 1989), with time lags of up to twenty years to capture long-run effects. For both output growth and exports the results reveal a positive correlation in the earlyyears, followed bya negative one (figure 2.5). The change could be Figure 2.4. Rent-seeldng intensity, 1925-83 (regulations scaled by economic activity) Scdee (average= 100) 300 ----..--------..'---.--- . -.----'.-.-....- . -..--..-i- ...._.,, aled by exports 200 --. .-.-..-.- *..-............ ..... .-..... -.. lO ---.--.- -....--.- .---. - 50- -_--. -....-_--.--. ---- ---..---. . -..-- ~- Scaled by output 1925 1935 1945 1955 1965 1975 198 S&rrea Rarna (I 993). 32 EncTs OF PRoTnenONISM ON * SMALL CouNmY: THE CASE OF UGRUUAY Figure 2.5 Rent seeking and export growth Carefaton coeffident 0.15 0.10 0.05 0 -0.05 -0.15 -0.20 -0.25 -0.30 -0.35 I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 AU regucdons Iaged t years Cnrrelaon coefficient 0.4 0.3 44~ .1 0.2~U -0.1 -0.2___ E -0.3 -0.4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Regulations on exports, bgged t yeats Soure Rama(1993). explained by an intense bout of lobbying following the trade liberalization of 1974-82 by those eager to preserve their rents. The correlations between the intensity of rent seeking and the growth rate of exports are positive at first but tum negative after three to four years, suggesting that capturing the rent requires investing in the favored sector. By stifling creativity and inno- vation, rent seeking lowers growth. Considering the lagged effects of the correlations, Uruguay's current economic malaise couldbeviewed as the after effects of the 1970s rent-seeking peak. Protectionism and stagnation: an ineerpreta due history 33 Impact of refms on savings and investment In addition to regulations and interventions in the foreign trade sector Uruguay had pervasive regulations affecting product and Factor markets. Before the reform program of 1974 price controls affected more than 90 percent of the consumer basket, the financial system was highly regulated and credit was rationecd, and a systemn of indirect taxation heavily penalized the agricultural sector. The reform package launched in 1974 eliminated most of these regulations, simplified th-e tax system, and reduced taxation on the agricultural sector (for a chronology of the main reforms see the apperdix). If the financial sector deregulation and other liberalization measures significantly changed savings and investmnent behavior, that niight explain some of the apparent relative improvement in perfonnance during 1974-84 (figure 2.3). De Melo and Tybout (1986) examined the pattem of savings and invest- ment during 1962-83, finding several patterns that tend to confirm a change of 'regime" after the 1974 reforms. For one, incremental capital-output ra- tios (ICORs) fell by nearly 40 percent during 1975-S1 relatve to 1967-74. Short-term swings in capacity utilization can, of course, also affect ICOR val- ues, but these averageswere computed over a relatively long period, and there is no other evidence of significant changes in capacity utilization between the two periods. Also, average pnvate sector investment rose markedly after 1974 from a very low base in the preceding period, funded mosdy by foreign savings. Second, estimation of private savings and investmentfunctions reveals a clear shift around 1974.1 The intercept term of the private investment equa- don shifted upward significantly after the 1974 reforms and investment re- sponded more significantly to real interest rates, as one would expect after the changes in the regulatory environment. During the 1960s private savings appeared to respond mainly to polices associated with financial repression and to (mostly) exogenous foreign capital inflows, whereas during the 1970s, when there were no restrictions on foreign borrowing, savings were nega- tively correlated with the real exchange rate. The private savings function also exhibited a significant upward shift after the 1974 reforms when other factors were controlled for. A sttistically significant upward shift in the private investment and sav- ings functions around 1974 tends to suggest a change of regime. However, withi the benefit of hindsight (and in view of the evidence in chapter 8) it does not suggest a permanent acceleration in the rate of growth after the 1974reforms. Itis more likely that the relative imnprovementin perfornance during 1974-82 reflected the one-shot effects of a static improvement in re- source allocaton. . 34 EFmCS OF PROTECTONISM ON , SmALL CouNnTv: THE CASE OF URUGUAY Distributional acivities, investment, and growth Though it is difficult at the microeconomic level to directly trace the effects of resource misallocation on growth, the evidence points toward a negative effect of distortions on growth. The discussion so far suggests that Uruguay continued to show respectable growth until the mid-1950s despite all the market interventions because it had notyet exhausted the opportunities for import substitution. But the interventions were pervasive and the domestic marketwas small, so the scope for domestic-ledgrowthwas small. Thus, otier factors influencing the turning point in the 1950s deserve attention. One of them is the progressive weakening of the Uruguayan state until the military coup of 1973. The loss of autonomy for policy decisions had a strong impact in an economy whose growth had been built largely on the development of a variety of public utilities. Civil servants were a large and influential group in Uruguayan politics-they rose from less than 3 percent of the population to more than 8.5 percent in 1969.'0 Rama (1991) argues thatduringthisperiodofweakening, the pressures exerted by the three dominantsocial groups-civil servants, manufacturers, andwage earners-intensified. He developsasimpleportfolio model inwhich manuflacturers choose betreen owning a finn and investing abroad on the basis of expected profitability, with profitability negatively related to wages and the terms oftrade. He finds that during the period of economic stagna- tion from the mid-i950s and to the early 1970s realwages in all but two years were above the hypothetical real wage that would have resulted in an accu- mulation of capital suffcient to generate zero GDP per capita growth. Prof- itability and investment were restored frllowing the suppression of wage councils and tacde union activities under the military rule. This finding supports other evidence, discussed above, pointing toward a shift in pnrvate investment and savings behavior at the time of the 1974 refonns. During the period when the government was weakening, the three social partners seem to have been strong enough to have an effect on eco- nomic equilibrium, but not broad-based enough to bear the costs of their activities. The manufacturers group was small, and the earnings of civil ser- vants and unionized members amounted to less than a third of GDP. Thus, the organiations representing each group struggled for larger slices of the same pie rather than striving to enlarge the pie, as they mighthave done had they represented a larger share of the population. That a weakening of the state encouraged further exDloitation of the rent-seeking opportunides created by excessive state intervdntion in the first place in no way provides evidence that a strong state will bring economic efficiency. It is probable, however, that the process of economic stagnation ProtctienLm and stagnation: an intsprnoav hiscAry 35 and decline was accelerated by the state's loss of autonomy in decision-mak- ing after the early 1950s and that the arrival of stagnation was delayed by a stronger and more independent state in earlier years. Conclusion Uruguay's history of protectionisrn resulted in economic distortions and an inflexible economy that stagnated over a long period. Rent-seeking activi- ties were rampant, and the labor union movement grew strong behind the high protective walls, since an organization having a monopoly on the la- bor supply could capture some of the rents conferred by protection through higher wages. A substandal opening of the economy followed the liberaliza- tion that started in 1973, but the reforms were suspended in 1982 with the onset of the international debt crisis. Occasional, small reductions in tariff rates were made in later years, but these were temporarily suspended under IMP programs in 1985 and 1990, which raised duties 5 percent for revenue purposes. Significant liberalization of foreign trade is the obvious way of diluting the power of interest groups and of diminishing rent-seeking activities, but ithas proved difficult to achieve in a societywhere the leading political groups benefited from protectiorism. With its decision to enter the MERCOSUR customs union with Argen- tina, Brazil, and Paraguay, Uruguay has committed itself to firther trade lib- eralzation, particularly in the automobile sector. (For a discussion of the merits of this regional approach to trade liberalization, see chapter 5.) The levels of protection will be jointly decided by the member countres, and Argentinaand Brazil, as the two largest, are certain to have the greatest influ- ence on the outcome. During negotiations-in April 1993, Brazil appeared to be proposing high duties to protect some of its manufacturing industres, while Argentina and the smaller members were proposing lower, more even duties to promote freer trade within the region. For a country like Uruguay, with a long history ofrent-seeking interests, participation in MERCOSUR is likely to improve the credibility of its refonns. At the very least, a commit- ment to an across-the-board reduction in tariffs according to a timetable agreed on by the four countries should dilute the influence of domestic lob- bies in policy decisions. Appendix. Chronology of reforms, 1949-89 The firstsection provides a chronology of major reforms during 194949; the second section dassifies reforms during 1974-89 by their key area ofimporL 3 . EFncrs or PRoTcEcIoNIsX ON A SMALL COuNrY: THE CASE OF URUGUAY C-rono-og y, 194949 January 1941 Congress approves export and import control bilL Ex- ecutive branch appoints a comptroller of imports and exports, with authority toprohibitimports thatcompete with national production. December 1959 Multiple exchange rates for imports are eliminated (the price of foreign exchange is to be determined in the market) and replaced by explicit tariffs. The executive branch is authorized to impose surcharges up to 300 percentofci.f prices, require prior deposits foriimports, and prohibit imports for up to 180 days. 1960 - Uruguay joins Latin American Free Trade Area, with status of 'lesser relatively developed" country. September 1960 Presidential decree sets importsurcharges of45 percent, 75 percent, and 150 percent for "luxury" goods. July 1964 Export subsidies available for up to 20 percent of the f.o.b. value, and imported inputs incorporated into exports are exempt from duties. 1965 Currency is devalued (twice). 1967 Currency is devalued. June 1968 As part of a stabilization program, all prices including salaries are fixed; any price increase has to be approved by the price control agency; currency is devalued. 1969 Position of comptroller of imports and exports is abol- ished. May 1971 Capital goods imports become subject to priorapproval, effectively prohibiting imports. March 1972 New government takes office. Beginning of transition period leading to reforms. Recognition that system of price controls established in 1968 must be relaxed. Hundred percent devaluation. Protedtionism andstagnation. aninterpretative history 37 1972 Passive crawlingpegintroduced-pursued untilNovem- ber 1978; creates double foreign exchange market 1973 Militarytakeover. Dissolution of Parliamentandsuspen- sion ofpolidcal actvities.Approvalof developmentplan. September 1974 Capital-market is opened, and Uruguayans are permit- ted to hold foreign exchange for the first time. The exchange rate is determined freely for capital transac- tions, and byacrawlingpegfor trnsactions in thegoods marketL 1974-75 Exportsubsidies are permitted on armassive scale, upon request October 1978 Commercial and capital markets for foreign exchange are formally unified. The exchange rate is announced 90 days in advance. December 1978 Tariff reforms are instituted, with a uniform tariff of35 percent and a uniform export subsidy of 5 percent to be achieved by 1985 through a linear tariff reduction. 1979 Uruguay signs the GATT subsidy code. November 1982 Export subsidies cease, and indirect tax rebates are in- stituted. Areas of imp4 1974-89 FNAcIAIL MARKZr September 1974 Uruguayan residents can buy or sell assets denominated in extemal currencies withoutrestrictions, leading to de factoconverdbilityof the peso. Ceilings on peso-denomi- nated deposits are raised from 15 percent to 30 percent May 1975 Foreign investors are allowed to repatriate earnings and capital. Rules on credit-allocation are established. March-September Domestic interest rates are effectively freed, with the 1976 ceiling raised to 62 percent Rediscount faclities for 38 Emcrs OF PROTErONISM ON A SMALL CouNrRY: TnE CAsE OF URIucuAY commercial banks are eliminated. Currency trde is permitted outside commercial banks. The central bank begins to pay interest on legal reserve requirements. Mid-1977 The 1965 banking law controlling the number of non- bank financial intermediaries is relaxed. (The number - - ofsurh institutions rises from one in 1976 to tventy-three in 1981.) They still cannot offer checking account ser- vice and are chiefly engaged in intermediating foreign currency funds obtained abroad (from nonresidents). November 1977 The dualforeign exchange marketis unified. Commer- cial banks are allowed to pay interest on cash accounts. October 1978 Three-month treasury bills, redeemable in dollars, are sold. In effect the central bank is preannouncing the exchange rate (official unification of dual foreign ex- change market)- Early 1979 Finanicial interest subsidies for exports are removed. May 1979 Subsidized credits from the centralbank to "Frigorificos" are eliminated. The legal reserve requirement is elimi- nated. Commercial bank reserve requirements depos- its in domestic and foreign currency are abolished. The banking tax is abolished (6 percent lender tax and 2.4 percent tax on loans to commercial banks). February- To narrow the persistent cross-currency gap in lending September 1981 rates, thecentra bank offersimplidt exchange guarantees. COMMODnryAND LABOR MARS October 1970 Legislation establishing domestic content requirements and compensatory exports for the automobile industry is established for the purpose of "learning by doing." June 1973 All trade union activities are banished by the new mili- tary governenL July 1974 Personal income and inheri tance taxes are removed. A uniform 25 percent tax on corporate net profits is es- Poetdionism and stagnation: an interpretative histoy 39 tablished, with exports exempted for up to 50 percent of profits reinvested. Export taxes on beef and wool are removed to compensate for the dedine in world prices. July 1974 Internal prices of nonessential goods begin to be liber- alized. At this time 94 percent of the consumption bas- ket (for calculating the CPI) is still controlled, as is 65 percent of agriculture. Some barriers to imports are removed. Compulsory extemal financing for 180 days and prior permission to import capital goods are abol- ished. October 1974 To encourgeforeigndirectinvestments, anewlawguar- antees the repatriation of profits and capital. Jauary 1975 Remaining quantitative restrictions are removed. July-December Thirteeni percent of items in the CPI consumption bas- 1975 ket are deregulated. February 1976 Prices not part of the consumption basket are liberal- ized, except those produced under monopoly. Later in the year. another 25 percent of prices in the CPI con- sumption basket are deregulated. July 1978 Anew agency (DINACOPRIN) takes overprice settang role,with the objective ofincreasingfiexibilityin setting prices and promoding comppetition in the economy. 1978 Temporary admissionregime instituted, provding duty- free access to imports of inputs used in export produc- ton. August 1978 Another 13 percent of CPI consumption basket prices are decontrolled, as is the meat market. December 1978 The maximum surcharge (on c.iŁ prices) is gradually reduced to 90 percent February 1979 Tariff reductions are accelerated to prevent 'exagger- ated" price increases. There is a return to some price controls (dothing, private schooling). : - - 40 EFFECTrs OF PROUCflONaIS OrN A SMALL CouNT.in THE CASE OF URUGUAY March 1979 List of goods and services with prices fixed admiinistra- t-ively is reduced. September 1979 To fightinflation, tariffs are further reduced and prices liberalized. November 1979 Tax reforms are put in place, including a uniform 18 *percentVATfornonessential commodities, reduction of social security charges, elimination of 8.4 percent banking tax, and elimination of subsidized credits and tax exemptions for exporters. June 1980 Car prices are fixed because of the absence of foreign competition. ByJune 1980, 29 percent of prices in the CPI consumption basket are still controlled (16.1 per- cent excluding price of public utilities). Controlled agricultural prices dropped from 65 percentin Decem- ber 1973 to 14 percent in December 1979. 1982 The reference price system is established, with tarifrates based on reference prices rather than c.if. importprices when these are lower than the reference pnce. June 1982 A 10 percent surcharge tax is imposed on all imports. November 1982 Thepreannounced exchange rate experimentcollapses, and a floating exchange rate is adopted. All export sub- sidies and the 10 percent surcharges on imports are dropped, with the subsidies replaced by indirect -tax rebates. A temporary tax of 15 percent is imposed on traditional exports. A five-level tariff system is created, with rates ranging from 10 percent for low materials to 55 percent for final goods. 1983 The minimum export price system is established, pro- viding for a moving surcharge equal to the difference between the declared c.if. import price and the mini- mum exportprice. LE addition, the tariffrate is applied to the minimum export pnce. 1984 The passive crawling peg is reinstituted, based on a bas- ket of currendes but with no government commnitment .Protctionism and stagnation: an nterpretative history 41 to a preannounced exchange rate. Taxes on traditional exports are reduced to 5 percent March 1985 Re turn to democracy. Unions are allowed. Wage bargain- ing mechanisms are set up at the sectoral level, and unions begin to play a significant role in manufactur- ing- June 1985 Additional 5 percent tariff surcharge is reinposed for all tariff levels for fiscal purposes. (Removed in August 1986.) Augus t 1987 Tariffs are cut 5 percent for the-two highest levels. Maxi- mum tariff is set at 45 percent 1991 New tariff reform, with a uniform tariff of 15 percent to be achieved by 1999 through a stepwise tariffreduction. 1991 The minimum export price and reference price systems are reformed. 1992 The compensatory export requirements and domestic content requirements are abolished and the duty on assembled cars is reduced to 25 percent The highest tariff is now 24 percent BmIATERAL TRADE AGREEMENTS May 1975 Under the bilateral agreement with Argentina (CAUCE), no tariffs or quantitadve restrictions apply for Uru- guayan exports to Argentina, except for agricultural and "sensitive" products. For each product covered, there is a ceiling of 5 percent of the previous year's Ar- gentine production. Uruguay allows duty-free imports from Argentina of as much as 60 percent of the previ- ous year's increase in Uruguayan exports to Argentina under CAUCE. January 1976 Trade begins under the bilateral agreement with Brazil (PEC), signed in 1975. No taiffs on a negotiated list of products, with a ceiling for each product of 5 percent of the previous year's production. 42 EFFECrS OF PROTECTIONISM ON A SMALL CouNTRY: THE CASE OF URUCUAY May 1985 C? (JCE is enlarged to cover all products up to the 5 percent ceiling. Nontariffbarriers on the Argentine side are removed. The ceiling is set at 2.5 percent of Argen- tine production for sensitive products and is negotiated on a case-by-case basis for critical" products. August l986 PEC product coverage is significandy increased. Non- tariff barriers on the Brazilian side are removed. An automatic device to adjust the ceilings is set up to allow up to a 30 percentyearlyincrease in some cases, although the 5 percent ceiling remains in force. March 1991 Argentina, Brazil, Paraguay, and Uruguay sign the Treaty of Asuncion creating the MERCOSUR customs union. All tariffs must be removed by December 31, 1994, fol- lowing a stepwise reduction scheme. Each country has an allowed list of "sensitive" products for which the ex- ternal tariffregime remains in force (initially 960 pro d- ucts in the case of Uruguay). The number of products on the list has to be reduced by a fifth each year. There is no agreement yet on the extemal common tariff. Notes l. Themajor studies on protection in Uruguay areAnichini, Caumont, andSjaastad (1978) and Favaro and Spiller (1990). For a description of the early stages of protec- dion, also see Favaro and-Sapelli (1990). For a broad desciption of Uruguayan eco- nomic history, see Finch (1981). 2.Formore detailed accounts of the liberalization measures starring in 1974, see Favaro and Spiller (1990). On the sequencing of liberalization measures, they reach the same conclusions as Hanson and de Melo (1 985), that therc was no evidence that opening the capital account before the current account was a source of Uruguay's difficulties. 3. For example, for the period April 1 to September 30, 1968, the prepayments applied to an importer's imports in excess of 30 percent of the value of his imports in the periodJuly 1, 1966 to Oc tober 31, 1967; for the period October 1 to December 31, 1969, the surcharge applied to imports over 100 percent of imports exempted from pr-epayments in the period October 1 to December 31,1968; and forJanuary 1 to Sep- tember 30, 1972, to imports over 174 percent of imports exempt from prepayIents during the second half of 1971. 4. Lowerdeposit requiremcnts applied to imports from LAFlAcountries, and from time to time exemptions from the deposit requirementswere announced. Forexample, beginningin February 1969, importsof certain capital goodsand transportequipment for export industries, industries producing goods for mass consumption, labor-inten- sive industries, and import substitution industries with low levels of protection were exempted from import deposit requirements. Protectionism and tagnaion: an interpretative hist 43. 5. Observers of the import deposit system often incorrectly referred to it as "pro- hibitive," but te syster was never entirclyprohibitive, as indicated by a relatively large stock of deposits even when the real cost to importers was highest. It seems likely that the system was prohibitive only forover-quota imports in the categorieswith the high- estdeposit requirements. Somc over-quotaimports continued in categorieswith lower deposit requirements. If so, the system operated like a tariff-quota for categories of pro- ducts with lower requirements and likc a quota for categoneswith higher requirements. 6. The turning points in figure 2.1 are similar to those in de Melo and Tybout (1986,figure 1), butdifferentfrom Rama (1991, figure 5.1) who identifies twogrowth periods, 1935-56 and 1968-81, with stagflation during 1957-67. His break points are the outcome of a maximization of the coefficient of determination of a regression of real GDP per capita on an cxponential time-trend equation. 7. De Melo (1987) gives a short-term macro model in which the level ofeconomic activity depends on aggregate demand of Argentine tourists and cxpectations of dc- valuation account for the capital flight of 1982-83. 8. Data on education levels before 1960 are not available for a large number of countries, so the regression data stat with 1960. 9. The shiftwas detected econometricalyby the technique of recursivc residuals, which has the advantage of endogenous detection of functional shifts. 10. The constitutional amendment of 1951 set a rule for sharing public admninis- trationjobs between the two leading parties (three for the wirmer, two for the loser) and prohibited the dismissal of civil servants. Many of the institutional arrangements adopted under this sharing rule greatly reduced public sector efficiency. References Anichini,JuanJose,Jorge Caumont, and Larry Sjaastad. 1978. La Politica Comercialy la Proteccidn n el Uruguay. Montevideo: Banco Central de la Repuiblica Oriental del Uru- guay. Barro, Robert. 1991. "Economic Growth in a Cross-Section of Counties." Quartery Joura ofEconomicg 106 (2): 407-44. Barro, Robert andJ. W. Lee. 1992. "Intemational Comparisons of Educational Attain- ment, 1960-85." Harvard University, Cambridge, Mass. Bl-anchard, Oliver, and Stanley Fischer. 1989. Lectures on Matroeconomics. Cambridge, Mass.: MIT Press. Corbo,Vittorio, andJaime de Melo. 1989. "ExternalShocks and PolicyReformsin the Southern Cone: AReassessment." In G. Calvo, and others, eds., Debt; Stabilation and Development Eays in Memory of Caros5DiazAlandro. London: Basil Blackwell. Ethier, Wilfred. 1993. Modern International Economics. New York- W. W. Norton & Co. Favaro, Edgardo, and Claudio Sapelli. 1990. Export Promotion and Economic Gnrwth. San Francisco, Calif.: Institute for Contemporary Studies. Favaro, Edgardo, and Pablo Spiller. 1990. "Uruguay." In A. Choksi, M. Michaely, and I. Papageorgiou, eds., The Timingand SequancngofT radeLiberzztion London: Basil Blackwell. Finch, H. 1981 -A PoliticalEconomy of Uruguay since 1870. London: Macmillan. Hanson,James, andJaime de Melo. 1985. "External Shocks, Financial Reforms and StabilizationAtternpts in Uruguay during 1974-83. " WorldDevelpmet 13(8): 917-39. IntemationalMonetaryFund (IMP). 1969-79.AnnualRepotonExchangeArrangemnts and Exchange R.stritions and IntanatioFinancial Statistis. Washington, D.C. 44 EFrEcn OF PROTECrIONISM ON A SMALL CouNirx THE CASE OF URUGUAY Macadar, Liis. 1985. "Protccci6n, VentajasComparativas yEficiencia Industrial." SUMA 3(5): 7-59. Melo,Jaime de. 1987. "Fimancial Rceforms, StabilizaLion and Growth under High Capi- tal Mobility Uruguay 1973-83." In M. Connolly and C. Gonzialcz-Vega, cds., Economic Refom and Stabilization in Latin Amcnca New York: Praeger. Mclo,Jaimc de, andjames Tybout 1986. 'The Effect of FinanciaI Libclization on Savings and Investment in Uruguay." Economic Development and Cultural Change. 560-87. M&zzeraj., andJaime de Melo. 1985. "Adjus!imentby Industrial Finnsin Uruguay: 1974- 82." In Vittorio Corbo andjaime de Malo, cds., Scramblingfor Surivmat How Firms Adjusted to the Recent Reform in the Southen Cone. World Bank Staff Working Paper 764. Washington, D.C. Nogues,J., and R Quintanilla. 1993. "Latin America's Integration and the Multilateral Trading System." InJ. de Mclo and A. Panagariya, eds., New Dimensions in Regional :ntegration. Cambridge University Prcss. Rama,Martin. 1991. "Economic GrowthandStagnationinUruguay.'InM. Blomstrom and P. Maller, eds., Diverging Paths: GComaring a Cen"try of Scandinavian and Lati-n American Developmnent Baltimore, Md.:Johns Hopkins University Press. - 1993. "Rent-Seeking Trade Policy: A Time-Series Approach." Policy Research Working Paper 1142. World Bank, Trade Policy Dlivision, Washington, D.C. Summers; R., and A. Hcston. 1991. 'The Penn World Tables (MarkV): An Expanded Set of International Comparisons, 1950-88.' Qyartriyjouimrnal of Economics (May): 327-68. White, H. 1981. 'Consequences and Detection of Misspecified Nonlinear Regression Models."Jowrnal of theAmenican StatwisticaAssociation 77:419-33. World Bank. Various years. WorldDevelopment Report NewYork: Oxford University Press. Administered protection: reference prices and mtntmum export prices Federico Changanaqut and Patick A. Messerlin Because of difficultieswith the formal procedures forimplementing the 1981 antidumpingandantisubsidylaws and the threat of retaliation byother coun- tries, Uruguay sought otherprotectve instruments that would allow a quick response. The reference price system already in place to counter under- invoicing of imports began to be used to take the place ofantidumping and countervailing duty measures. In 1983 a new and more powerful instrument, the minimum export price, was introduced. The two have been applied to some 500 products and are now the major tools of administered protection in Uruguay. Uruguay is not alone in using reference pnces to protect domestic in- dustries and to counter the underinvoicing often triggered by other protec- tionist measures. Peru and Cote d'Ivoire in the mid-1980s did the same. Reference prices have also been used to reduce the shocks associated with the opening of an economy, as Mexico did in 1985.1 The economic consequences of such administered pricing have gener- ally been neglected in the literature. Our analysis of the effects of this form of administrative protection on the Uruguayan economy shows that the eco- nomic costs have two main components. First, reference prices magnify the tariff protection againstimports whose trnsaction prices are below the cor- responding reference pnces, generadng distortions of the sort analyzed in traditional trade theory. Second, for each product covered, reference prices substandally reduce the price differentials between imports with prices lower and those with prices higher than the reference price. By increasing the domestic price of the cheaper variety relative to that of the more expensive variety, reference prices are likely to have an adverse effect on quality. Refer- ence price procedures can induce foreign firms to shift their exports to the country toward expensive varieties of goods, induce domestic producers to shift production toward the cheap varieties, and induce domestic consum- ers to reallocate expenditures toward expensive goods, which have become 45 46 EFFECTS OF PROTECTIONISM ON A SMALL CoUNTrr THE CASE OF URUGUAY relatively less expensive than before. Thus, this system of administered pro- tection has negative distributional effects on the consumption side and pro- motes the production of low-quality domestic goods. Resource and welfare consequences of administered minimum prices This analysis of the resource and welfare consequences of reference prices begins with the simplifying-assumption that consumers are interested in dte goods themselves and not in the services they provide. This approach opens the wayforan analysis of the impact ofadministered protection on the struc- ture of exports and on domestic production. The nextstep is to assume that consumersare interested in the services provided by the goods, that the quality of the goods imported or produced domestically matters to them. This ex- tension permits administered minimum prices to be compared with other instrumnents of protection in their impact on incentives to alter the product mix. How admninistered minimum prces work When the declared import price (c.i.f.) is less than the reference price (a floor price fixed by regulation), the official duty is applied to the reference prce. For transactions with a unit price higher than the official floor price, the published tariff rate is applied to the declared shipment price, so the apparent (or published) tariff and true (or paid) tariffs are the same. Thus, the final domestic price (P) of an imported good equals the import price declared by the importer (p) plus the reference floor price established by the Uruguayan authorities (F) times the nominal tariff rate (t): P-p p+ tF. If F is less than p, ther p is used for calculating import duties. An even more powerful protective measure is the minimum exportprice mechanism, which adds to the simple reference price mechanism a moving charge equal tf the difference between the minimum export price and the declared import price (F- p). Under the minimum export price system the domestic price in the importing county would be: (3.2) P -p + tF (F-p) -F( + 0. One way to administer these reference prices is to fix them in nominal terms, without adjustment for foreign inflation, so that their ad valorem equivalent protection is eroded over time. Morocco uses this approach. A Administered protetion: reference prices and minimnum carn prices 47 second way is to build in a mechanism that allowss frequent revisions in the reference prices. This is the approach taken in Uruguay and in several other developing countries. W4tfare costs of the administered system. We assume for now that goods are homogeneous, rather than differentiated by quality, so that the impact of protection on the quality of imports and on the domestic production mix can be ignored. The administered minimum price systems increase the cost of imports and generate the usual distortions expected from tariff protection: domestic production is diverted from exportables to importables, as shown in figure 3.1 for the reference price system. For the one-quality good foreign supply is pW domestic supply, pZS, and domestic demand, D. In afree trade situation, the marketwould clearat W, with aworld price ofp and quantity sold equal to 0L1. An advalorem tariff (t) shifts the equilibrium to W', where the price isp(l + t) and the quantity sold is 0L1 (ON from domestic suppliers and NI4 from foreign suppliers). Domestic producers receive transfers equal to area a, govenment receives tariffrevenues equal to area c+ c', and consumers experience a netloss equal to areas b + d. Inposition of a reference price (fl on this system generates an effective supply curve ofpZY. The market clears atE, where the price is p + tF, and the quandty sold is 04 (ON2 from domestic suppliers and N2L4 from foreign sup- pliers). Since the domestic price is determined by the reference price and Figure 3.1 Reference price system Pike Z ;eE p+tF Y p(l +t) a 0 N, N1 Lk Li Quandty 48 EFFECIS OF PROTECTONISM ON A SMLL CouNTRr. THE CAsE OF URUcUAY the duty, the system severely restricts competition in the domestic market There is an additional effciency loss equal to areas b' + c '+ d, and domestic producers gain extra transfers equal toarea a'. The governmentcollects extra tariff revenue equal to area c'' but loses tariff revenue equal to areas c'. The analysis for the minimum exportprice system is similar (figure 3.2). With the imposition ofa minimum exportprice, the market dears atL where the price is F(1 + C) and the quantity sold is OL (ONfrom domestic suppliers and hL from foreign suppliers). Domestic producers sell quantity ON, on which they capture rents equal to the sum of the two a areas. The govern- ment collects tariff revenue equal to area c + c', which includes the moving charge equal to area c. These revenues are offset by the welfare loss to con- sumers equal to area pZEWp, which includes the higher prices paid for im- ports (areas c + c-), the higher cost of the domestic source of supply (the two b areas), andthe efficiency loss m consumption (the two d areas). The net effect then is indicated by the sum of the band dareas. In addition, because the Uruguayan price remains unchanged when the world price fialls (curve pWshifts downward), the implicit tariff equivalent of the minimum export prnce increases as world suppliers become more efficienL Impact on quaity and domestic specialization Nextwe assume that goods are differentiated by quality and that consumers buygoods for the services these goods provide. The actual quality of le goods determines the features and longevity of the services provided. In this it is Figure 3.2 Minimum export price system Price p=F(I +t) V F p D .0 N L Quantihy Adm.nistered protection: refrence prices and minimum expoprtdpcts 49 intuitivelyclear thatminimium pnce procedurescan influence qualityin three ways: they can induce foreign firms to shift exports to the country toward the more expensive, higher-quality varieties; they can induce domestic produc- ers to shift production toward chQaper, lower-quality varieties; and because these procedures lower the relativeprice of the expensive varieties, they can induce domestic consumers to shift their spending toward expensive goods. Thus, this system of administered protection has negative distributional ef- fects on the consumption side and promotes domestc production of low- quality goods. Administered protection may tius getin the way of a country's achieving an optimalpattern ofspecialization. Itmayinduce local producers to produce goods of a quality range below their capacities (a situation that fits well the Rodriguez 1979 or Krishna 1987 models) as determined by the country's comparative advantage-or even at too high a quality niche. (See the appen- dix at the end of this chapter fora detailed analysis of the effect on quality.) Estimates of the extra protection provided by administered minimum prices Between 1981 and 1989 Uruguayan administered protection covered roughly 500 products and tariff positions a year (table 3.1). As a rough measure of comparison, the United States and the European Community undertook some 200 antidumping actions during the same period. Even if such special protection as bilaterally negotiated quotas in textiles and steel in industrial countries is also taken into account, administered protection in Uruguay appears intensive by world-standards. While these are dearly rough indica- tors, precise companson that takes into account the size of the import flows involved and the degree of openness of the domestic economies would find Uruguay's adminiistered protection to be high. For nontextile products, the number ofgoods covered rose between 1981 and 1984, and then fell after 1985. At the same time, however, there was a marked shift of goods from the reference price to the minimum exportprice system. For texiles and apparel, coverage remained stable from 1981 to 1987, when it began to fall. Anotherway to look at the coverage of administered protection is through the percentage ofimports affected. Roughly 4 percent of Uruguayan imports, excluding oil, were covered by administered protection in 1987 and 5 per- centin 1989. This low coverage is misleading, however, since the very grant- ing of additional protecton to domestic goods under the administeredprice system introduces a systematic bias that leads to an underesdmation of the importance of these mechanisms. The apparent tariff for the sample of goods under the administered protection examined in this study varied from 50 30 EmaTs OF PROTECTlONISM ON A SMALL CouNTrav THE CASE OF URUCUAY Table 3.1 Scope of administered protection in Uruguay, 1981-89 (number orproducs cowered) Type cfprotection andproduct 1981 1982 1983 1984 1985 1986 1987 1988 1989 Nolsnreiks Reference priCes 8 34 70 95 89 70 62 54 54 Mininum export prims - - - 1 2 1 8 20 25 40 40 Tranrsferred goods" 22 46 46 47 47 46 36 35 35 Subtoal 30 80 116 154 154 136 123 129 129 Textiksb - Yams 177 177 177 177 177 177 177 177 177 FabriCs 105 lO5 105 lOS lOS 10$ 105 105 105 Othertextiles 129 129 129 129 129 129 129 104 82 Subtotal 411 411 411 411 411 411 411 386 364 Al products 441 491 527 565 565 547 534 515 493 - Not applicable. a. Goods initially under refeence pnics and progressively tramslcrred to the minimum export pie system. b. Number of NADI items under the admninistrative pricng system. S-rct: Nattino (I 989). percent to 58 percent on average for 1986, and from 39 percent to 45 per- cent for 1989. The average apparent tariffimposed on all goods imported by Uruguay in 1986 was 29 percent; in April 1989, nominal tariffs ranged from O percent to 45 percent. Thus, using the number of percentage imports af- fected by administered protection underestimates the impact of the refer- ence price systems on Uruguayan trade. A betternotion of the impact comes, for exanple, from learning that the value of textiles and apparel products covered by reference prices equals one-third of Uruguay's total value added in manufictured goods. Effects of admninistered protection The costs of administered protection derive from three sources. Reference prices and minimum export prices magnify tariff protection for selected goods, erode the effects of trade liberalization measures implemented in recent years, and distort the relative prices of cheap and expensive imports of the same item (table 3.2). The prolection-aug7nentation effect For import transactions at prices below the floorprice, administeredminimum prices are likely to substantiallyboost protection since administered prices affect goods that already Eace relatively high nominal tariffs (50 to 58 percent in 1986, 45 percent in 1987, 44 to 45 percent in 1988, and 37 to 39 percent in 1989 for a sample of nontextile goods). The true tariffs fTaced by cheap imports are thus likely to be high. Administered proteedon: referenac pices and minimum export pics 51 To calculate the real tariff under the reference price mechanism, WFin equation 3.1 is divided and multiplied byp the c.i.f. price of imports: (BY'S) P=p(1 +tp), - where , =F/pandmeasures the protection-augmentation effectintroduced -by the reference price mechanism. The augmentation coeffident ,B is calcu- lated by dividing the floor price for each item by the average price of trans- actions cleared at prices lower than the floor price for each item.2 The com- putedaverage true tariffs-are higher than the average apparent tariffsby28.9 percentage points for 1986, 27.6 for 1987, 24.9 for 1988, and 18.1 for 1989. Real tariffs of more than 80 percent are not rare, and some goods have real -triffs of 150 percent or more.5 .For the minimum export price mechanism real tariffs are computed as F(1 + t)/p, where p is the average price (before protection) of transactions cleared at prices lower than the floor price. Real tariffs arising from the mninimumn export price mechanism are lower than those from the reference price system, althoughsdll substantiaL 20.4points highertlhan apparenttariffs for 1986, 4.1 for 1987, 7.6 for 1988, and 7.5 for 1989. These results seem to show that the minimum exportprice system is less protectionist than the reference price system, but itis the differing coverage of the two systems that is responsible for this apparent contradiction. By re- -writing equation 3.2, itis easy to show that shiftingproducts from the reference price to the minimum export price system increases the level of protection: (3.2' P=pą F+ (F-p) = p(l + tp + (F-p)/p). All else being equal, real protection is higher for the minimum export pro- tection system than for the reference price system by the difference between the floor price and the import price as a percentage of the import price, (F - p)/p. As a result, simply transferring a good from the reference pnce system to the minimum export price systm leads to an increase in protec- tion if the import price is lower than the floor price, holding the apparent tariff, the transaction price, and the floor price constant. But transaction prices are not likely to remain constant, because the minimum export price system, far more than the reference price system, reduces the incentives for foreign exporters to keep their prices lower than thefloorprice. Underthereferenceprice system foreign exporters can com- pensate for the protection-augmentation effectwith lower prices. Under the minimum export price system, offering prices lower than the floor price provides no advantage to the foreign exporters because the sale prices in Table 3.2 Effects of administered protection on real tariffs and prices, selected Items, 198649 (percent) Price cristottion effect Protection augmentation (change in rmlatke unit value)' effect (tanscacdons Erosion of Jberalizadon effect betow floo prices) Before After 7Tye of Giobol .Global administered administered protection Apparent Real Apparent weighted Percentage unwe&ghted Percentage price system price system Percentage and year tarif'1 tariff c Difference tarifft t1ariff change d tariff change' d(rato) (ratio) changed Reference price system 1986 48.9 77.8 28.9 57.9 68.5 18.3 69.4 19.9 2.73 1.65 39.6 1987 44.1 71)7 27.6 45.0 54.3 20.7 54.5 21.1 3.18 1.80 43.4 1988 42.1 67f0 24.9 43.8 51.3 17.1 54.8 25.1 2.93 2.03 30)7 1989 38.5 56.6 18,1 39.4 46.9 19.0 44.8 13.7 1.B4 1.44 21.7 Minimum exeport price system 1986 47.9 68.3 20A4 50.0 63.2 26.4 61.6 23.2 2.31 1.98 14.3 1987 45.0 49.1 4.1 45.0 47.4 5.3 47.1 4.7 1.77 .1.59 10.2 1988 38.1 45.7 7.6 45.0 47.3 5.1 47.6 5.8 2.65 2.11 20.4 1989 36.4 43.9 7.5 36.7 38.5 4,9 40.1 9.3 2.02 1.75 13.4 Reference price protection for textiles, $989 Yam 40.0 .47.1 7.1 - - - - - 1.71 1 A3 16.4 Fabric 40.0 55.9 15.9 - ----2.43 1.75 28.0 Others 40.0 64.7 24.7 - -- 3.23 2.13 34.1 All texties 40.0 58.3 18.3 - ---...- -Not applicable. Not av-ailable. Note: For definitions of the price effects, see text. a. Ratio of average prices of transactions Wth prices above the administered minimum price to transactions with prices bel-ow tosne prices, computed for all goods ror which transactions data were available. b. Unweighted averages of the apparent tafrifs imposed on the selected products. c. Unweighted avenages oft.e real tariffs under adminiistered minihmum pricing, d. Percentage increase of global taurifF over the apparent tarff. e. Reducton of the relative unit values due to the RP or MEP systems. Source: Central Bank of Uruguay. Adminisered pwtetion: reference prices and minimumw eport prices 53 : Uruguaywill bedetermined byadministeredfloorpricesand apparent tariffs, two variables overwhich foreign exporters have no control. In these circum- stances the lowest prices that foreign exporters will be induced to offer are close to the floor prices. Thus, they keep for themselves the price difference they would have been ready to offer in the form of a lower import price or to share (illegally) with the Uruguayan importeroruser who maybe a producer. The average augmentation coefficientof all items under reference prices (1.51) is 31 percent higher than the coefficient for all the items under mini- mum export prices (1.15). Moreover, until 1988 the augmentation coefli- cients for items under the minimum export price system actually decreased while the floor prices increased, suggesting that foreign exporters learned quickly to raise their prices. Finally, the results suggest that the ausmentation effect is relativelyevenly distributed among industries, with no obvious pattern of concentration. The iberaation erosion efect. In l986 the Uruguayan authorities lowered nominal tariffs, resuming the process of trde liberalation that had been suspended in 1982. These reductions, however, have probably been under- mined by the application of administered protecton, as the following com- putations indicate (table 3.2). Average apparent tariffs and global real unweighted and weighted (by market share) triffs were computed for items with at least two consecutive years of imports. Global real tariffs are defined as the average of the appar- ent tariff for items on which transactions take place above the floor prices and of the real tariffs on goods whose transactions take place below the floor pnces. Trade liberalization is reflected in the fall in apparent tariffs, from 58 percent in 1986 to 38 percent in 1989 for goods under reference prices, and from 50 percent to 37 percent forgoods under minimum export prices. The global weighted real tariffs were considerably higher than the app'arent tar- iffs, however. That difference-19 percent for reference price goods and 5 percent for minimum export price goods in 1989-is a measure of the ero- sion of the effects of trade liberalization.4 Theprice-distort7ion effect. A rough estimate of the extent of the price-dis- tortion effect can be derived by comparing annual average relative prices (unit values) of cheap and expensive varieties of import items before and after protection (both tariffs and administered prices). (Transactions in an item below the floor price imposed by the authorities are assumed to be sales of a homogeneous cheap variety of the product imported and transactions in an item above the floor price a homogeneous expensive variety of the product.) The changes (annual averages for all goods) in relative prices af- ter protection provide an estimate of the price-distortion effect attributable .54 EFFEcrs OF PROnaCbONISM ON A SMALL CouNTRY: THE CASE OF URUGUAY to administered protection. That effect is substantial (table 3.2). For the period 1986-89, relativepricesfell 34percentforgoodsunderthe reference price system (27 percent for textiles), 15 percent for products under the minimum export price system. This distortion effect is probably underest- mated in the case of the minimum export price mechanism, because it in- duces exporters to stick to the floor price in order to maxinize their rents. Effects on Uruguayan textile and apparelproducers Textiles and apparel under administered protection represent a substantial percentage (roughly 33 percentin 1983) of Uruguayan value added in manu- factured goods. Textiles other than yarn are subject to a particularly com- plex form of administered pncing. Reference prices for fabrics and apparel are based on the reference prices imposed during previous stages ofproduc- tion, resulting in multiplicative protection coefficients. This form of administered pricing has a more severe impact on textiles than on other manufactures. For 1989 the difference between apparent and real tariffs rises from 18.1 percentfor nontextiles to roughly 18.3 percentfor textiles under the reference price system and from 7.1 percent to 7.5 per- centunder the minimum export price system And effective protection rises with the degree of processing (escalation of protection): the protection-aug- mentation effect averages 7 percent foryarns, 16 percent for fabrics, and 25 percent for other textile products considered..-That the apparent tariff is 40 percent for all textile goods studied means that the escalation of protection is entirely the result of administered protection, which constitutes a substan- tial source of uncertainty and confusion for Uruguayan producers of fabric and other texdle products. The price-distortion effect is also more severe for textiles than for nontextile goods: 30.5 percenton average for textiles in 1989 comparedwwith 21.7 percentforallgoods under the reference price system and 13.4percent for goods under the minimum export price system (table 3.2). Escalation is also apparent: the price-distortion effect rises from 16 percent foryarns to 28 percent for fabrics and 34 percent for other textile goods. A more severe distortion effect for textiles than for nontextile goods is a finding of particu- lar relevance for a. developing country, whose people spend a substantial proportion of personal income on textiles and apparel. Protection against macroeconomic disturbances in Argentina and Brazil The reference price and minimum export price systems are often defended as a means of protecting Uruguay against the transfer of macroeconomic Adminiterd prtecdion: reference priea rtnd minimum expwo pticas 55 disturbances from Argentina and Brazil. The evidence does not support that argument, however, vhichwould require finding thatadministered pricing wasappliedprimarily toimports fromArgentdnaand Brazil. Evidenceshows, instead, thatadministeredprotecdon is applied as much againstimports fom the rest of the world (all trade partners of Uruguay except Argentina and Brazil) as itis against imports from Argentina and Brazil.5 F1k. the eighty-eight nontextile products under admiistered pricing, 'forty three have transactions below the floor price only for exports from Argentina and Brazil, thirteen for exports coming only from the rest of the world, and thirty-two for exports coming from both sources. Administered protection is triggered exclusively by exporLs from Argentina and Brazil for only 48.3 percent of products, a proportion that corresponcls roughly to the relative importance of trade flows from these two countries in 1986 (38.6 percent of total Uruguayan imports were from Argentina and Brazil). Also, the economic impact of administered pricing is similar for exports coming from Argentina and Brazil and for those from the rest of the world. The magnification of protecdion achieved by administered pricing is on average similar for exports from Argentina and Brazil and for exports from the rest of the world (tabIe 3.3). When all products with transactions under the floor prices are considered, irrespective of the country of origin, the aug- mentation effect on protection is slightly higher for goods from the rest of the world than for those from Argentina and Brazil, except for goods under the minimum export price system in 1986. But for goods in which only Table 3.3 Uruguay's use of administered prices on selected imports from Argentina and Brazil compared with the rest of the world, 198649 (rnifjcation of protection) AJl prodcts- Common productsa Type of protection Al trade Agenbna Restof Al) trde Argentina Restof and year parteis and Brazl the wśrdb trtnse and Brazil the weddb Referfce price sytem 1986 1.63 1.37 1.84 i.50 1.42 1.60 1987 1.62 1.53 1.57 1.75 1.61 1.57 1988 1.51 1.38 1.68 1.37 1.36 1.60 1989 1.46 1.34 1.40 1.39 1.32 1.44 Average 1.56 1.41 1.62 1.50 i.43 1.55 Minimum export price slstem 1986 1.13 1.15 1.12 1.28 1.21 1.22 t987 1.09 1.09 1.34 1.11 1.09 1.47 1988 1.22 1.19 1.48 1.12 1.15 1.42 1989 1.19 1.20 1.46 1.17 1.19 1.49 Average 1.16 1.16 1.35 1.17 1.16 1.40 Gobalaverage 136 1.28 1 i.49 1.34 1.29 1.48 a. Products with imports corring rom both Argentina and Brazi and the rest orthe world, b. All nrade partners of Uruguay excludingArgenrtina and Brazil. Sowre: Central Bank of Uruguay. 56 EFFECIS OF PROTECrIONISM ON A SMALL CouNTnY: THE CASE OF UR UGUAY Argentina and Brazil or only the rest of the world have transactions below the floor prices, the augmentation effect is systematically higher for the goods from the rest of the world than for the goods from Argentina or Brazil-unambiguously so in the case of the minimum export price system. The explanation may be that Argentine and Brazilian exporters are more familiar than other exporters writh the loopholes of the Uruguayan minimum export price system and its capacity to generate price realignments and rents. Conclusion Byincreasing the costs of imports, administered minimum prices generate the usual distortions associatedwith triffprotection. Butadministered protection also creates other ncgative effects that a pure tariff system does not. Admin- istered minimum prices distort the quality content of imports, domestic production, and exports. Underperfect competition a taniffdoes notchange the optimal quality contentof imports, butan administered minimum price induces foreign firms to upgrade the quality of their exports to the affected market In a noncompetiidve setting the impact of niinimum prices on qual- ity is ambiguous: a minimum price can induce a foreign monopolist to raise orlower quality content While more research is needed on the impact of mini- mum prices on the quality choices of domestic producers facing stronger competition from upgraded foreign goods, it seems likely that administered mnmimum prices impede a developing country's achievement of an optimal pattem of specalization based on compartive advantage. The reference and minimum export pnce system affect goods that ac- count for roughly one-third of Uruguayan value added in manufacturing- a substantial share. These mechanisms undennine the effects of trade liber- alization, boosting published tuiffrates by 7 percentfor the minimum export price systemn (probably an underestimate) and by 18 percent for the refer- ence price system. These systems also create massive distortions (from 15 to 30 percent) between the relative domestic prices of goodsimported-atprices under and above the floor prices. Appendix. Impact of reference and minimum export prices on quality If goods are differentiated by quality and consumers buy goods for the ser- vices they provide, a crucial component of quality is the features and longev- ity of the services provided. Intuidvely, then, minimum price procedures have three quality-related effects: they caninduce foreignfims to shiftexports to the country toward more expensive, higher-quality varieties; they can induce domestic producers to sbiftproduction toward cheaper, lower-qualityvariet- Admiisted pnfkctow reference prices and minimum cp:n prices 57 ies; and they can induce domestic consumers to reallocate expenditures toward expensive goods, whose rlative price is now less expensive. The re- sults are negative distributional effects on the consumption side and the promotion of domestic production of low-quality goods. This appendix refines these intuitive results and extends the literature on-the quality effects of import quotas to cover the quality effects of mini- mum price procedures. Two questions are examined: What is the impact of the minimum price procedure on the quality of the goods imported, par- ticularly compared with the impact of a tanrff or quota? What are the conse- quences of the reference pnce and minimum export price systems on the product mix of domestic producers competing with foreign producers of goods subject to these procedures? The competitive case Foreign suppliers of most Uruguayan imports under reference price ormini- mum export price systems are likely to face competitive markets. A model suggestedbyRodriguez (1979) can be used to compare the impactofa triff and that of a quota on import quality, with the results extended to a mini- mum price system (see also Santoni and Van Cott 1980). The results showv thata tarff does not change the optimal quality content ofa physical unit of imports but that a quota and a minimum prce induce foreign firms to in- crease the quality content of their exports to the country. Under free trade foreign exporters determine the optimal quality con- tent per physical unit they export by maximizing their profits pxq - xi4(q) with respect to quality content, where p is the price per unit of services, x is the numberofphysical unitsproduced, and hkq) is the constant average (and marginal) cost per unit of physical output The optimal quality content qp per physical unit under free trade is thus determined in the long run, where the price per unit of services equals the marginal cost of services: (3A.1) p p= k'(q). Freedom of entry and exit in a competitive market structure implies that in long-run equilibrium firms will make zero economic profits, so that in the aggregate total revenue equals total cost: (3A.2) p(.S)S= A(q), wherep(S) is the inverse of the demand function for imports of services, S =qX the total low of services provided by the goods, and X= l:(x). The long-run equilibrium situation is thus determined by: 59 EFFECTS 0F PRoTECTIONISM oN A SMALL. Couwgrny: THEs CAsE oF URcuGAY th-at is, the traditional equality between marginal and average costs Figure 3A.1 illustrates the model. Figure SA.1a shows the optimal quality contentq with a price per unit of services equial to P , and the total imports of services equal to 5,,. Figure SA.1 b illustrates the equilibriumn situation F for the ser- vices market. Under an ad -valorem tariff (0', foreign firms maximnize their profits pl- tqxq - xls(q), which yields the optimal quality choice: (SAX) p( - = :'(q). Under the zero-profit condition the tariff solution yields: (SA.2') p(l - 0 = h(q)/q, (3X3) pA(q) = (q)/q, showing thatprofit-maximnizing foreign exporters under atariff face the same optimization condition as under free trde wihen they determine thie opti- mal quality content of theirproducts. They thus keep the sme optimal quality content 9,per physical unit of their exports as under free trade. Keeping q, impliesthatthe price per unit ofserviceu nder a tariff P,isesin proporion to the tariff up to 1= b (i) in the figure on the right In other words, -a tariff does not induce foreign finns to change the optmal quality content per physical unit, butitreduces the total amount of services imported from 5, ,(under free trade) toS8T.and the toralv'olume of imported goads fromX2,.to 24.. By contrast, the introducdion of a quota creates an incentive for foreign exporters to increase the quality content per physical unit. The quota is as.- sumed to lead to the same level of physical units imported, a the tariff. This constraint is illustrated by the curve P(qXw, which shows the domnestic price per unit of services as the quality content is changed, 4 being given and fixed. If foreign firmswere to maintain quality ue, they would be able to chiarge a price pFwhile bearing a cost of sr (q)v (at the margin). At that level of imports i,n the domestic price is P, with a rent equal to the difference [-lijdacaruing to quota holders. However, as long as each quota holder re- gards the domestic price for services as given, it ill pay to order imports Of higher quality content whenever the price is higher than the marginal cost. Foreign suppliers are thus induced to increase ateir revenues by choosing a -higher quality content, q such as p = pe(q) for qQ (at the margin), as shown in figure 3A.la. As a result, as shown by the equilibrium point Uin figure 3A.1b, total imports of services thus rise to Se, wvith SoO> 8i.e ihe introducton of an administered minimum price leads to a similar Administered protection: reference picas and minim'umn eA$of pyiers 59 quality upgrading. The minimum export price mechanism is treated as a specific tariffr, thatis, afunction of the quality contentof the imported good: (3A4) r= r(q). For prices up to the floor price, the higher die quality (and therefore unit price)the. smallerthedifference betweenthe arbitrrminimum export price and the declared impbirt price. tis assumed that dr/dlq<0, because the highier the quality, the lower the spedific tariff. In such a cas-e,- foreign.firms deter- mine the optimal quality content by maximizing their profits [xqd- xh(q) -xr(q)] That gives: (3A.1 p= r(q)r illustrated by the curve ((q) in igure 3A.1a. The zero-profit condition cor- responding to free entry imposes: (SA.2-) p(S)=Uh(q) + r(q)]/ q, a condition illustrated by the curve g(q) in flgur q SA.la. As in the quota case foreign firms are induced to increase quality content. They will increase it Figure 3A I Determination of optimal quality content o-seMces -.a (a) P(q (b) - -- (3*1~) pft tq) r q)) _ _ _ _ _ _ _ E u u~~~~~~ - I DXor nt ()(a q) ~~~~~~~~~~~~~~~upy a : ~~~~h(q)lq C . \ emand PF D'=W 0 qF qq qR Quality content 0 ST SQSR SFS=qX -; ~~~~~~~~~~~~~(servcelurit) 60 EFFECs OF PROTECTIONISM ON A SMALL CouNTRY: THE CASE OF URUGUAY up to q,F which is higher than q and even q. The total flow of services in- creases up to SR, with SA> S> SrT. In sum, a minimum price intensifies the incentive to increase the quality content of foreign exports. Tzhe noncornpetitiTe case -An altemative scenario is tat foreign exporters are monopolists, a case that may fit for some Uruguayan imports, although competitive markets are more likely in most cases. The results, derived using a model suggested by Krishna (1987), are quite different froom those achieved under competition. A tariff induces a change in the optimal quality content for the foreign monopolist, and a quotamay lead to alower quality content (see also Das and Donnenfeld 1987). The two following optimization conditions apply to the foreign monopolist's choice between quantity Xand quality q for maximizing prof- its XP94q) -XC(q): XPV(X,q) + P(X,q) - C(q) =0 and (3A.5) 7XPQ(Xq)- XC,(q)= 0. The effect ofta quota on quality is obtained by differentiating totally the sec- ond first-order condition alone. This gives: (3A-6) dq/dX=-P./(P - Cs). The sign of this expression depends on the sign ofP,x (since the denomina- tor is a diagonaI element of the Hessian-that is, is negative in order to en- sure the second-order conditions for a maximum). Nevertheless, Px1 may be positive or negative. As shown by Spence, P is negative when 'the marginal value of quality falls as absolute willingness to pay falls, that is, when the av- erage value attached to quality exceeds the marginal consumer's valuation" (1976, p. 419). As a result, a quota can induce a decrease-notan increase- in the optimal qlualitycontentof thegoods exportedby the foreign monopolist. The minimum price case can be easily introduced to Krishna's model by restating the profit function as XP(Xq) - XC(q) - Xr(q). The new first-order conditions for maximizing profit are: A- x,Xzq) + P(Xq)-C(q) - r(q) = O and (3A.5 ) XPq(Xq) - XC (q) - x.rq) = 0. The foreign monopolist can react by changing the optimal quality content, taking the quantity as constant; by changing the optimal quantityproduced, taking quality as constant; or by changing both the optimal quantity and Administrd protection: reference prices and minimum exaot prices 61 quality produced. For comparing the quality effects of the two trde mea- sures, the bestapproach is to consider the same level of restrictions in termis of physical units. The corresponding comparative statics require differentia- tion of the second first-order condition and give: (3A.62) dq/dX--P / (P -Cq -r) The sign of dq/dXis the same as in equation 3A.6. Nevertheless, the value of dq/dXin equation 3A.6' can be higher or lower than in equation 3A.6 since it depends upon the sign of r As a result, a minimum price can induce the foreign monopolist to amplify or reduce the changes in quality contentwith respect to the changes undertaken under a quota. The welfare effects The Rodriguez (competitive) and the Krishna (noncompetitive) modelsyield different results in terns of optimal quality content. But both models yield similar-and unusual-results on theweltAre impactof the tariffor the quota. In many cases (and in all cases in the Rodriguez model), a quota is shown to be unambiguously superiorto a tariff, contrary to the findings of the traditional trade literature of the -welfare superiority of tariffs over quotas. We find, however, that the two types of models have limitations that diminish the robustness of this unusual result. The discussion here focuses on the com- petitive Rodriguez model because it seems a better fit for the Uruguayan economy (and because it leads to more striking results). The proof of the welfare superiority of a quota over a tariff goes as fol- lows: a quotaincreases the optimal quaity contentperunit (q,> qT = q) and so allows for an overall quality level of imported services higher than that allowed by a tarifffor the same amount ofphysical units. The welfare gain is illus- trated by the area TUBCin figure 3A.lb. Of this gain, TUEaccrues to domes- tic consumers and EUBCto import quota holders. This expansion in service quality comes at a higher unit cost of production (relative to those in the tariffcase); however, as illustrated by area CDPFC in figure 3SA.b. Rodriguez shows that the net welfare gain TUBC- CDPJFC is equal to the area tuf in figure 3A.1a and is thus positive. The same analysis can be applied to the minimum price case. A mini- mum price is said to be welfare improving (that is, less costly) relative to a ariffand a quotain thatit expands the services provided by the same amount of goods (as shown by TRB'C) -and welfare reducing in that it entails addi- tional costs (as illustrated by CDPAC) - How robust are these unusual results? They depend in both models on one crucial assumption; that there is no reaction by domestic firms." When .62 Em,FcTs OF PROTECTIONISMi ON A SMAULL CouNmrv: Tim CA.s oF URuGuAY compa-ring the free trade and tariffcases in the Rodriguez model, forinstance, this assumption iswithout consequence, because foreign firmns keep the samne optimial quality content in both cases (this is not always true for the Krishn-a model). As a result, domiestic firms have no incentives to modifyr the quality content of their own production sold in dom-estic markets. But the same -assumption creates a bias when comparing the free trade and tariff cases on the one hand with the quota and minimum price cases on the other. Domes? tic finns have incentives to modify their quality con tent if foreign firms have decided to change their quality content. That domestic firns will face stron- ger competition in the marke't niches with a high quality content implies adjustment costs for the domestic finns. These adjustnent costs. should be added to the welfiare costs of the quota or reference price, when comipairing these two cases with the free trade and ctariff cases. Notes I. At the beginning of the Mexican tradecliberalization in 1985, following theclim4- nation of most quantitative restrictions, the perc-entage coverage of production of tradables under refe-rence prices increas-ed from 18 percent to 25 percentand remained at this level until the restrictions wvere completely eliminated in 1987. 2. Thec computations are based on data from- Uruguay's central bank, which per- mits separating transactions cleared above and below floor prices. The method of av- erag-ing prices for the two types of transactLions has the advantage of avoiding the risk of overestimaLing protection by focusing on the transactions with the lowest prices. Indeed, the price differentials between the twvo tyes of transactions are a multiple of 2 and 3, a "normxal" figure by any standard. 3. Several reference prices were applied for television sets. The reported results are based on the lowestreference price. Thus, there is a possible underestimnation rather than a'i. overestimation of the augmentation effecL 4. Itis interesting to note that therecis nosign of catching up in the reference price system in the sense that the differentials between the apparent and global tariff-s are stable over time. That is not the case in the minimum export price system. As before, this result may be caused by the pricing behavior the system triggers among foreign suppliers. 5. This discussion deals only with non texti'le goodls. Textiles are subjeCt to 'indi- rect" imports when Uruguayans purchase these goods in Argentina and Brazil. That these indirect imports are generally considered to be significant (although there is no estimate) implies that the computations done in table 3.1 would be upward biased for textiles. 6. As Krishna mentions, another im portan t limitation of these models is that they consider one good, not a product line. For an analysis of a product line, see Krishna (1984). Reactions bydlomestic finns to foreign firms' decisions can be dealtwith through game theory. Das and Donnenfeld (1989) provide such an analysis within a duopolistic market structure, showing that a trade instrument's impact on the firms' decisions depends on the location of the firms in the quality spctr-um. In the c-ase- where foreign firxm. produce the high-quality variety, the domestic firm responds by upgrading qual- Administered protection: ieference prices and minimum expoM prices 63 iy and expanding sales, but the total quantity sold to donmestic consumers declines. The global impact on welfare is thus ambiguous, because the negative impact on tle consumersurplus-dominatedby thelose in tlehigh-qualityniche-iscompensated to somc degree by the increase in the profits of the domestic monopoly. References Das, Satya P., and Shabtai Donnenfeld. 1987. 'Trade Policy and its Impact on Quality of Imports: A Welfare Analysis.fJourndl of lntrnaional Economics 23: 77-95. . 1989. 'Oligopolisdc Competidon and Iternational Trade: Quantity and Quality Restrictions."Jourmna oflnternationalEconoamics 27: 299-318. Krishna, Kala. 1984. Protection and th Product Line. NBER Wori*ingPraper 1535. Cambridge, Mass.: National Bureau for Economic Research. .1987. 'TariffsVersus Quotas With Endogenous Quality. "Joumnazoflnt ional Economics23: 97-122. Mecsserlin, Patrick A. 1990. "Antidumnping Regulations or Procartel Laws? The EC Chemical Cases." Policy Research Working Paper 397. International Trade Division, Iterntational Economics Department, World Bank, Washington, D.C. Nattino, Elbio. 1989. "Instrumentos Antidumping: Recomendaciones para Reducir la Protecci6n Diferencialy DiscrecionaL".Centro de Estudiosdela Realidad Econo6mica y Social (CERES), Infornzsy Propwstas l: 1-4. Rodriguez, Carlos A. 1979. 'The Quality of Imports and the Differential Welfare Ef- fects of Tariffs, Quotas, and Quality Controls as Protective Devices." Canadian Jour- naTl of Econonics12(3): 439-49. Santoni. GJ., and T.N.Van Cott. 1980. "mnportQuota.s TheQualityAdjustmentProb- Tem.' SoutenEconomicJournat46(4): 1206-11. Spence, AM. 1976. "Monopoly, Quality and Regulation." BellJournal ofEconomics (Au- tumn): 417-29. Domestic content restrictions and compensatory export requirements in the automobile sector Wendy Takacs Uruguay's complex system of protection and controls is perhaps best illus- trated by the set ofpolicies that until recently regulated the automobile sec- tor. These controls included tariffs on imported vehides and parts, domes- dc content requirements, and compensatory export requirements. Domestic content requirements mandate a minimum percentage of domesticvalue added or domestic components for products soldwithin the country or provide strong incentives to substtute domestc for imported inputs. These measures are by no means unique to Uruguay. Many Latin American countries, Canada, and Australia have used such regulations as part of an industrialization strategy based on inport substitution to foster the development of domesticmotor vehide and pards industries (see Lloyd 1973; Baranson 1969; andMunk 1969). These protective regimes often resulted in the emergence of domestic assembly operations: firms would import kits or sets of components from abroad and combine them with domesdcally pro- duced components to produce finished vehides. As the limits of import substitution were reached and interest shifted toward export promotion, many countries superimposed export promotion policies on the underlying domestic content requirements. Brazil (Mericle 1984) and for a time Argentina (Jenkins 1985) offered tax exemptions and waivers of inport deposit requirements if certain amounts of output were exported. At one dime Argentina even linked approval for expansion of domestic output to fulfllhent of quantitative export targets. In other coun- tries, such as Mexico and Uruguay, exportpromotion took the form of more explicit compensatory exportrequirements. Finns importing kits or finished vehicles had to have exports equal to a specified percentage of the value of those imports (Bennet and Sharpe 1985; World Bank/UNDP 1990)- In I990 Uruguay's protective regime for the motorvehicle industry con- sisted of all these elemens Fully assembled vehidles were subject to a tariff of 40 percent. Fmns importing assembled vehicles (completely built up or 64 Protedionthe automobil sectr 65 CBU) were also obligated to exportautomnobile industry productsrith avalue added in Uruguay equal to 70 percent of the f.o.b. value of the imported vehicle. Firns assembling automobiles in Uruguay from imported kits (com- pletely knocked down or CKD) were required to pay a tariff of 10 percent on the kits and comply with domestic content requirements and compensatory export provisions. For passengervehicles assembled from kits a minimum of 20 percent of the value of the product had to be of national origin. At the minimum domestic content of 20 percent, compensatory exports of 60 per- cent of the f.o.b. value of the kitwere required. Domestic content above the. 20 percentminimtum could be traded off for compensatory exports ata ratio of I to 2. For example, for a 5 percentage point boost in domestic content to 25 percent, compensatory exports could drop by 10 percentage points to 50 percent.' Amarket in the credits for compensatory exports developed, with vehicle assembly firms in need of compensatory exports paying parts exporters to credit their exports to the assemblers. Observers report that assemblers paid 5 to 15 percent of the value of the invoices for these export credits, with the average being about 8 percent, though payments reached as high as 30 per- cent in times of very high domestic demand for automobiles. A model of domestic content and compensatory export requirements A model was developed to investigate the distortions, costs, and transfers among groups resultingfrom the combination of domestic contentand com- pensatory exportrequirements in the automobile industryand to derive some rough estimates of the costs and transfers arising from this protection regime. The model simplifies byignoring differentiation among types ofcomponents and assembled vehicles, the trdeoffbetween domestic content and compen- satory exports, regulations on levels ofminimum disassembly of components in kits, and prohibitions against importing certain components. It further assumes that domestic content and compensatory export requirements are binding (that is, that less domestic contentwould be used by assembly firms if there were no domestic content requirements and that exports of compo- nentswould be lower in the absence ofcompensatory export requirements). The model also assumes that the country imposing the domestic con- tentand compensatory export requiirements is small, so that the world prices, or import prices, of assembled automobiles (PA*) and of auto components (Pe*) are both given.. Aperfectly competitve domestic components industxy manufactures components, and a perfectly competitive domestic automo- bile industry assembles automobiles from imported and domestically pro- duced components. The country produces a single type of finished or as- sembled automobile, madebyassemblingagiven number W" ofcomponents. 66 E.z-rs OF PROTErCONISM ON A SMALL CouNrvR THE CASE OF UzucuAy Differences-among components are ignored3 Finished automobiles can be imported, subject to a tariff and compensatory export requirements. The domestic assembly industry is subject to domestic content requirements and compensatory export requirements for a package of imported components, called'a kit. Equilibrium prices and quantities in the markets for assembled automobiles and for components are determined jointly because they are tied together by normal input-output relationships and by the domestic con- tent and compernsatory export requirements. Domestic market for assembled automobiles Importation of assembled automobiles requires compensatoiy exports of a certain proportion xA of the value of the imported automobiles. The value of compensatory exports required perautomobile would be xAPf= J%qC, where qc is the quantity of components that must be exported per imported fin- ished automobile and Pc is the domestic price of componens If the compo- nents sell in the world market at PJ*, the cost to the firm of the required compensatory exports would be (P-PCS cx [ -PC )/PC] xAPA7. An ad valorem tariff rate of tA on automobiles increases the cost of the imported vehicle by the amount of the duty. Thus, the cost ofan imported automobile is PI*(' + Q + XAl(4 P? )/&&*. In the long run a perfectly competitive industry would be expected to make zero economic profits, so that in the long run the total cost of the imported assembled vehicle equals its price. Then: (4.1) PA G + tA + XA [(PC PP ) AS where PA is the domestic price of assembled vehicles and P. of components. Given the world prices of assembled automobiles and components and the domestic price of components, equation 4.1 deternines the domestic mar- ket price of assembled automobiles. The quanltity of automobiles sold (do- mestically assembled from kits and domestic components, QA' plus already assembled, MA) is then detennined by the domestc demand for finished automobiles: (4.2) D(PA) = MAą QA If there is an upward-sloping supply function ofvalue added in domestic assembly operations, the quantity of automobiles that firns are willing to assemble increases as the value added per unit (VA) increases: (4.3) VA=V(Q) V' positive Protection in the automobile sector 67 Thevalue ofan automobile assembly kitat world-.n'rketprices is ot(1 -8)Pc* where S is the proportion of total components that must be of domestic ori- gin.3 The ralue of compensatory exports required to import the kitwould be x,cot (i - S)Pc* = PC qc, where xA is the value of the imported kit that must be compensated by exports of automobile industry products. The tariff on ki ts increases the cost of'kits to the domestic assembly industry by the anount of the tariffpaid per kit, or a(l -.) Pc*tK. The cost of domnestic components would equal a8P . The assumption of a perfectly competitive assembly in- dustry implies that unit cost equals price in the long run, so: (4.4) P = a.(l -5)P*[l + t,+ x,;( P0*)/PCI + aSJ%+ V(Q4). Equation 4.4 can be thought of as the long-run inverse supply curve of the assembly industry. Itis constructed by adding vertically the domestcvalue added required to induce fimns to assemble various quantities of vehicles, the cost per vehicle of domestic components used as intermediate inputs (OSP), and the effective cost of the imported kit, ac(1 - 5)PC*[1 + t,+ xJ(PC -P*)/RPC Equations 4.1 through 4.4 determine P, MA, QA, andDA, given Pc, A*, PC' JA' Ix' XA, X,, s, and &- TIhe market for assembled automobiles is depicted graphically in figure 4.1. The price of automobiles to consumers is PA, which lies above the world price PA* by the amount of the extra costs imposed by the tariff tAPA* and the compensatory export requirements on assembled automobiles, X4[ (PC -PC*)/AP4IPA At this price consumers would demand D vehides. Figure 4.1 Domestic mnarket for assembled automobiles Prce s A CC~~~(l4)Pcx K S A*'=aCPC*+V(Q,, as(P Pc*)+a(lAxr *)P O/P~X)jP( P~/P~ P PA X QA ^ , DA ^* Qusnuy~'~A C. ~~A~A A_ DA---------tD 68 EFFCmS OF PRoTECrTONISM ON A SMALL CoUNrRy: THE CASE OF URUGUAY Figure 4.2 Domestic markeat for automobile components Price SC 0 p PC Fur 42Dmsi act fo auoobl D.mDonents x.ces is based on verbal estimates made by individuals familiar with the industry or involved in administering the system, rather than on observed data. Third, the model assumes homogeneous output ('assembled automo- bile") and homogeneous input ("components"). This obvious oversimplifi- cation biases estimates of the protection and transfers to the assembled ve- hicle sector upward and those to the components industry downrward. Overestimation of protection to assembled vehicles arises because fully as- sembledimportedautomobiles tend to be more luxuriousand expensive than the automobiles assembled in Uruguay, yet the loss to consumers is based on percentage price increases using the average import price as the base. Un- .erestimadon of pro tection and oftransfers to the components industry arises because the industry will tend to export components that are the most com- petitve on the world market Thus, the premium on export invoices mea- sures the relatively small deviation between domesdc and world prices for the most competitive components, and not the undoubtedly much larger deviation for the least competitive components. Fourth, the assumption of competiton in the automobile industry may be unrealstic for Uruguay. In 1989 there were about ten motor vehicle as- sembly firms, with two of them accounting for about a fourth each of the 74 EFFECrS OF PROTECTIONISM ON A SMALL CouNrRv THE CASE OF URUCUAi domestic market. With this market structure firm behavior might not corre- spond to that predicted by the competitive model. Yet studies of the U.S. automobile industry indicate thatfinn behavior does not deviate significantly from competitive market behaavior (see Dixit 1988; Krishna, Hogan, and Swagel 1989). The model presented here is intended to clarify the protective effects of and interactions between the domestic content and compensatory exportrequirements and provide rough estimates of the order ofmagnitude of the potential costs of the protective regime. Amodel that captrred strte- gic interactions among firms could provicle a richer analysis and possibly more accurate estimates of the true costs. Fifth, preferential trading arrangements with Argentina and Brazil cre- ate distortions that also cause underestimation of transfers to the domnesutc components industry and of the efficiency losses caused by the domestic content requirements. The impact of the protective regime on the compo- nents industry isvery sensitive to varationsin the deviations of domestic prices of components fromworld prices. These deviations are measured in the model by the premium paid by importers of assembled vehicles and kits for export invoices to credit toward their compensatory export requirements. The in- clusion of exporw shipped to the protected markets ofArgentina and Brazil under preferential trading arrangements pushes the estimated average pre- mium for export invoices below the premium thatwould be necessary to ship only to the world market This underestimation of the premium is not a problem in measuring tie cost imposed upon Uruguayan automobile buy- ars by the compensatory export requirements. In the absence of the prefer- ental trading arLarLgerm.ents, however, de: costs could be much higher be- cause compensatory exports would have to rece-.; an implicit subsidy high enough to make them competitive in the world market.8 Underestimation of the premium paid for export invoices will also result in underestimation of the effect of the domestic content requirements, because the rclevantprice differential is that between world market and domestic prices. Estimates of trnsfers to the components industry and efficiency losses from protected components production are very sensitive to this parameter, making it liklely that the method applied here underestimates the degree of protection to the componentis industry, perhaps significantly. Conclusion The model developed to analyze the impact of the protective regime for automobiles indicates that domestic content requirements, compensatory export requirements, and triffs on finished vehicles and kits kept vehicle prices high, maintained high-cost domesic production of both vehicles and components, and transferred large sums to special interest groups. Protctiorn in the automobik sector 75 Domestic assembly operations were affected by the various elements of the protectiveregime in different, andpotentallycontradictory, ways. Higher prices on finished vehicles encourage greater output from domestic assem- bly operations, but tariffs and domestic content and compensatory export requirements for kit imporLs discourage domestic assembly activity by increas- ing input costs. The net effect could be either to discourage or to encourage domestic assembly operations. In Uruguay the protective regime appears to have encouraged domestic assembly, so part of the consumerloss from higher prices represented a transfer to the assembly industry and part represented an efficiency loss due to increased domestic assembly.ofvehides ata higher cost than the price of assembled vehicles in the world market. Domestic components producers were unambiguously helped by all el- ements of the protective regime, which increased the demand for compo- nents produced within the country and drove up both price and output in the marketfor domestic components. The triffon kits provided component producers with protection from imported components, the tariff on as- sembledvehides helped maintain domesticassembly operations anddemand for components, the domestic content requirements forced domestic assem- bly operations to use domestically produced components, and the compen- satory export requirements for imports of both finished vehicles and kits increased demand for domestically produced components for export, act- ing as an export subsidy to the industry. The protective regime imposed substantial costs on consumers and encour- aged the allocation of resources in relatively high-cost activities. Compensatory export requirements and tariffs drove up the price of finished vehicles to con- sumers. The consumer loss was in part a transfer in the form of higher profits to the domestic producers in the assembly and components industries and in part effidency losses due to the distortion of consumer decisions and produc- don leveLs and to the extra cost of producing components within the country hat could be obtained at lower costin the worldmarket. The protective regime imposed a loss on Uruguayan consumers of automobiles of from $70 mfllion to $80 million a year and transferred from $43 million to $48 million to domestic assembly operations and components manufacturers. The estimated net loss to the country ranges from about $14 million to $28 million a year- These esti- mates are tentative, however, because the model assumes a competitive indus- try, excludes some aspects of the protective regime, ignores the differentiation of automobiles and components, and does not adjust for trade distortions aris- ing from preferential trade anrangements with Brazil and Argentina The govemment of Uruguay dismantled the protective regime for auto- mobiles in 1992. Compensatory export requirements and domestic content requirements were abolished, and the tariff on assembled vehicles was re- duced to 25 percent, 'with a further scheduled reduction to 20 percent in 76 ŁFvsc'r oFtrPOecTIoNJsm oN A SMAL.L COUNTRY. Tuz GAsR or URucuAV January 11993. Temporary provisions through December 1992 allowed com- ponents producers to import assembled automobiles free of duty up to the -value of their exports. This measure was designed to compensate components producers for their losses during the transition period once the protection regime was fuilly dismantled; The industry had been protected for decades, and such transitional niea- sures allow components producers a little tlime to adjust to the new regime. An alternative method would have been. to phase- in c-hanges by gradually reduc- ing taiffs and domestic content and compensatory export requirements ac- cording to a preannounced schedule. Care nee&; to be taken with such an ap- proach to avoid inadvertently increasing the degree of effective protection to the assembly industry by, for examnple, phasing out domestic content and corn- pensatory export requirements on kits fiaster than those on finished -automo- biles. Doing so -could temporarily increase the costs, of protection and provide false signals to domestic industry about the direction of adjustmnenL Appendix. Calculation of trAnsfers and costs of the protective regime for motor vehicles Te consumer loss is area adek in figure 4.1. Let2rt= CPc-Pc*)/IPcbe the pre- mnium paid for invoices to be credited as compensatory exports, and TlDA be the elasticty of demand for assembled vehidles. Then, given that: (4A.1) (PA -?A*)i PA*(,tA + xAlt) and (DA* - DA) =nDA,,(DA/IPA) PA*Q(,+ xA70, Area adek =(PA - PA*) D.A+ 0. 5 (PA -PA*) (DA* - DA) = 'PAQA XAIE[D+ O-5(D* -D0) (4A.2) =P*D(i+ x'rrfl + 0.SniDAlt+ xiifll The deadweight loss in consumption, area deft would be: Area def= 0.5 [PA*(44+ xATt)] (DA* -DA) =0.5 [ PA* (tA + Xjt)]f2,qDA/PA (4A.3) ~~= 0.5PA*DAQA+ XA7t)kjfA4 In calculating the gain to the assembly industry (area nlik) and the dead- weight loss to the economy from excess assembly operations (area lAO) it is useful to note that the height of each of these areas. equals the net impact of the restrictive regime, that is, the -amount, net of cost increases, by which revenue per unit exceeds free trade revenue. Let this distance (nk = hi) be designated N: JProfedton in Me automnobile sector 77 (4A.4) N PA* tA + AcA x( )tP*-a7P C( )xcCc LetoC aPc*/PA* be the share of component production in the final cost of a finished vehicle. Then: (4A.5) N1= Pś*tA+ xA7c r aU -68) Q. xgr7) + &c/ (I -T). Let t-he elasticity of the supply of vehicle asembly with respect to value added be zsA, and note that (QA4- QV) =s&JQ/ (PA*(I -ac)]N. Then, (tAG6) Area lki =0. 5N2eM Q/(PA*(1 -Ca)]. The gain to the assembly'industry, area nlik, can be calcul-ated as area ni/dk less area 1h4 or: (4A.7) Area nlik= N05Ne QA/[A (l-)] Let the elasticity of supply of components be s., and 1{,=P.,Q be the value of domestic components production. The deadweightloss from excess production in the components industry is shown in figure 4.1 as area qvt Are-a qrt = 0.59P%- P*) (Qc- Q2) (4A.8) = 0.5 Vtcc r2 The tra-nsfer to the domestic components industry as a result of the protec- tive regime is area oqu which equals area oqru less the deadweight loss: (4A.9) Area oqtu = (P c) c . VcEx7r2= Vcr -0.5 Vce.j2. Application of the model to the Uruguayan protective regime requires data on quantities assembled, sales, import price of assembled vehicles, deviation of components prices from world prices, the ratio of components cost to the final price of a finished vehide, and the value of components production. Tentative estimates of the magnitude of the welfare effects of the pro- tective regime were calculated using the data presented in table 4A.], as- sumned combinations of elasticities of demand of -0.5 and -1.0 and elas- ticities. of supply of 1.0 and 2.0, and equations 4A.2, 4A.3, 4A.6, 4A.7, 4A8, and 4A.9. T'he results appear in table 4.] in the main text. 78 EFFECmS OF PROTECrIONISM ON A SMALL CoUwRmY. THE CASE OF URUGUAY Table 4A. I Data used in calculations of impact of domestic content and compensatory export requirements Variabe Vdue Mesue it 0.08 Percentage of value of invoice paid to exporter of components for invoices to be caedited for purposes of fulfilling compensatory export requirements. (Source: verbal estimates of individuals intenvewed.) tA 0.561 Price-increasing effect of the 40 percent tariff on assenbled automobiles, taking into accountthatthetariff-indlusive price iste base forthe irternal tax(impuesto especifico interno) of 15 percent and the tax-indusive price is the base for the 22 percent value-added tax tA - .40(I.15)(1.22). 0. 165 Price-increasing effect of the 10 percent tariff on its, which applies only to approximately 14 percent of imported kits (imports from Argenina and Brazil are exempt from duty under bilateral trading arrangements), taking into account that kits are subject to an mntemal tax of 12 percent and a 22 percent vWalue-added tax: tK= 0.14(.1)(1.12)(1.22)+.12(1.22).' -A x,, 0.70 Compensatory export requirement for assembled vehicles. xK 0.53 RNtio of compensatoy exports to kit imports chosen by firms (based on data from two firms representing 27% of the market). - 0.235 Average domestic content ratio chosen by domestic assembly firms (based on data from two firms representing 2796 of the market). v $8,086 Unit value of imported assembled vehides (1989), categores A, C, D. E. and H (Source: Amara de Fabricantes de Automotores). QA 1 1,690 Number of vehicles assembled in Uruguay (1989), categories A, C, 0. E. and H (Source: Cimara de Fabriates de Autornotores). - 2-D 12,237 Number of vehicles sodd in Uruguay (I1989), calcuated as QA plus 529 assembled vehicles imported. a 0.621 Share of components cost in final cost of assembled vehicle (weighted averagc of costs of domestic components and kits as a share of final sales prices far four Uniguayan assembly firms representing 70 to 80 percent ofthe industry, using units assembled in 1989 as weights). Vc Value of component production, calculated within the program as: 0.53(value of kit imports) + 0.7 (value of finished vehide imports) + 0.235 (value of assembled vehides sales = ,'). a. The internal tax on kits is induded in the cakulation of the price-increasing efect of the tarif on kits because it directly affects costs and thus supply in the assemt4 industry. In this model the degree of protection to components manufacturers is rneasured by the prermiurn on export iwokes, which is assurmed to measure the percentage difference betveen domnestic and world narket prices for components. Notes 1. Data on the actual mix of conpensatory export and domestic content require- ments are available for two firms representing approximately 27 percent ofthe market in terms of the number ofvehicles assembled in 1989. Compensatoryexports averaged 53 percent of kit imports and domestic content averaged 23.5 percenL Protection in the automobik sector 79 2. This approach is similar to that of Grossman (1981) in that it assumes that doncstic and imported components are perfect substitutes. Mussa (1984) develops a model in which domestic and imported inputs arc less than perfectly substitutable. 3. Grossman (1981) shows that the effects of domestic content requirements will vary according to whether they are defincd in terms of physical quantities or valuc added. Uruguay's restriction is similar to a restriction in quantity terms because it uses servaluations, by weight, for the components. If P, is the set valuation byweight for the component ofwcightW, P' is the set valuation by weight for an assemblcd vehicle of weight WA, and &' is the required ratio of the valuc of domcstic compo- nents to the value of output, then the demand for components to be incorporated in domestically assemblcd vehicles (cquivalent to aSQA in the text) would be (PU WAI PC W4) S' Q4 4. Domestic content requirements arc calculated using the weightof the compo- nents and officiallyde tcrmnined prices for thc various types of componen ts. This method implies that highcr prices for domestic components will not affcct the quantity of domestic components demanded, so given Q,, t2lae demand curve for components to be combined with kits will be vertical. The use of administratively determined prices avoids the situation in which higherprices fordomestic components increase thcvalue added domestically and thus decrease the quantity of components necessary to fulfill the requirement. 5. Area xwfA = xA[ (PC - P*) /PCPI Ml = D,(P0- Pf) = area pqr 6. Area zyln = [a8(P- Pc*) + a(1 -)xK(Pc - PC*) (Pc"/Pc)] Q = (PC-~PC ) [a6Q+ V (1i - 8)X,QA(PC IPr)] = (Op) 7. It is interesting to note that multinational firms at times explicitly recognized the subsidyelement to components exports. Bennettand Sharpe (1985,p. 186) report that Chrysler arranged for its Mexican assembly operations to transfer funds to its U.S. asscmbly operation to cover the extra cost of Mexican parts. S. Some of the costs of the Uruguayan automotive protective regime are also be- ing paid by users of components ir. Brazil and Argentina, who are paying higher than world market prices for Ulruguayan components, and by the Brazilian and Argentine governments, in the form of forgone tariff revenue. References Azzini, Juan Eduardo y Asociados. 1979. La Indusffa Autootriz en el LUguay. Montevideo: Asociaci6n Nadonal deArmadoresdeVehiculosAutomotores. Cmanra de la Industria Automotriz. Baranson, J. 1969. AutowtiveInustnis inDeudeping Cozunbis. World Bank Staff Occa- sional Papers 8. Washington, D.C. Bennett, Douglas C., and Kenneth E.Sharpe. 1985. TransnatWonalCorporatonsvesus the State: hePuiiiicaE&onmy oftheMxaicanAutomobileIndustry. Princeton, N.J.: Princeton University Press. Camara de Fabricantes de Automotores. 1989? 'La lndustriaAutomotriz en Uruguay: Un ENsfuezo sin Pausas.' Pamphlet Montevidco. CENCI (Centro de Estadisticas Nacionales y Comercio Inteacional del Uruguay). 1989. Manutal rdctico deUImporlador. 13th ed., voL 2. Montevideo. S0 EFmES OF PROTECTIONISM ON A SMMUą CouNrrn THE CASE OF UrUGUAY Corden, W. Max. 1971, The Theory ofProtection. Oxford: Clarendon Press. Dixit, A. 1988. 'Optimal Trade and Industrial Policies for the U.S. Automobile Indus- try." In R.C. Feenstra, ed.,Empi icacMethodsfoIlntmationaI Tradt Cambridgec, Mass.: MIT Press. Grossman, Gene M. 1981. "Te Theory of Domestic Content Protection and Content Prcfcrence." QuartrlyiJournal of Economics 96(4): 583-603. Jenkins, Rhys. 1985. The Rise and Fall of the Argentine Motor Vchicle Industry.' In Rich Kronish and Renneth S. Mericlc, eds., The PoliticalEconomy of de LatinAmerican Motor Vehicle Indushy. Cambridge, Mass.: MIT Press. Krishna, K., K. Hogan, and P. Swagel. 1989. 'Thc Nonoptimality of Optimal Trade Policies: The U.S. Automobilc Industry Revisted, 1979-1985." Paper presented at thc National Burcau of Economic Research and Cen tcr for Econom ic Policy Research Conferencc on Emnpirical Studies of Strategic Trade Policy, Cambridge, Mass., October 13. Lloyd, P.J. 1973. NontarifDistortios ofAustralian Tradc Canberr: Australian National University Press. Mericle, Kenneth S. 1984. 'Te Political Economy of the Brazilian Motor Vchicle In- duty." In Rich Kronish and Kenneth S. Mericle, eds., ThePoarticalEconmy of the La in American Motor VYe/ide Industry. Cambridge, Mass.: MIT Press. Munk, Bernard. 1969. 'he Welfare Costs of Content Protection: The Automotive Industry in Latin America."Journal ofPolitiadEconomy 77: 202-16. Mussa, Michael 1984. 'The Economics of Content Protection. NBERWorking Paper 1457. National Bureau of Economic Research, Cambridge, Mass. Sevel Uruguay, S.A. 1989. "La Industria Automotriz en Uruguay: Lineamientos Gencrales y Principales Indicadores de su Actividad." World Bank and INDP (United Nations Development Program). 1990. "Uruguay. Trade liberalization and Economic Efficiency." ULNDP-World Bank Trade Expan- sion Program Country Report 5. World Bank, Trade Policy Division, Country Eco- nomics Dcpartment, Washington, D.C. Gains and losses from bilateral trading arrang nts wt Argentina and Brazil Jaime de M4elo, Claudio Montenegro, and Wendy Takacs Uruguay is sandwiched between two giant neighbors, Argentina and Brazil. Compared to Uruguay, the two countries are economic giants, and they of- fer the prospect of virtually unlimited outlets for Uruguay's products. Uruguay's trade with them has oscillt,ted widely over the years, but in the last two decades accounted for about a third of exports and 40 percent of im- ports. Communications and transport with the two countries are also excel- lent, making Argentina and Brazil 'natural" trading partners for Uruguay. UJruguay's economic policy toward Argentina and Brazil reflects a strong desire to tap these large mark3ets and reap the benefits of trade with these neighbors. The case for economic integration with Argentina and Brazil extends beyondintegrationofgoods markets. Laborand capitalmove briskly betveen Uruguay and its neighbors as well. Conditions seem right for formnation of a currency area, except for one hitch: the macroeconomic instability plaguing the thiree countries, and the large swingsin real bilateral exchange rates. (The impact of regional instability on growth is explored in chapter 8.) UTruguay has tied to walk a fine line by pursuing dose long-term economic ties with Argentina and Brazil while also isolating its economy from the macro- economic instability of its larger neighbors.' Preferential trade polices have been at the center ofmuch of Uruguay's regional developmentstrategy, beginningwith theLatnAmericaFree Trade Area (LAFTA) accord of 1960 and culminating in the treaty of Asuncion in 1991, an agreement to create a customs union amorigArgentina, Brazil, Para- guay, and-Uruguay (Tratado pan la Constituci6n de un Mercado Comuin, or MERCOSUR) . MERCOSUR is still under development and so may ben- efit from insights gained from an exarninatiorn of the most significant prece- dent to this evolvingarrangement the bilateralpreferendal trdeagreements withArgentina (ConvenioArgendno-Uruguayo de Cooperaci6n,Econ6mica, 81 82 Emecn oFz PRoTECrIONISM oN A, SmuALL Coumrav THiE CAsE. oF URuG:UAY or GAUGCE) and Brazil (Protocolo de Expansi6n Gomercial Uruguay-Brazil, or PEG) negotiated in the mid-I 970s. Whether acounrtty should pursue a preferential approach to trade is hotly debated because the issue of gains and losses is difficult to setde. The choice between a bilateral and a mnultilater-al approach to trade policy is especially imnportant today, with the resurgence of interest in regionalism, of which MERCOSUR is just one example. Looking at how GAUGE and PEG were designed and implemntendz and at whether they benefited the participants can therefore have aipplications beyond Uruguay. A small country entering into a customs unio(n w.ith large neighbors stands to gain significant market access, especially if the partners raise trade barriers against the rest of the world. If the analysis shows no g-ains in the case of GAUGE and PEG, that would provide fuirther evidence of potential problems in using regionalism as a development strategy, especially if the preferential agreements are lim- ited to specific: prducts rather than made across the board. The analysis of bilateralism begins with a primer on the economics of preferential trading arrangements that defines the concepts of "trade cre- ation" and "trade diversion." It moves on to the details of Uruguay's bilat- eral arrangements with Argentina and Brazil and the intraregional trade pattem that has resulted. A model is used to identify the.sources of static gains and losses under GAUGE and PEG, and data on unit values at the product-line level are used to establish wthether these arrange ments resulted in trade creation or trade diversion. For a. broader perspective, a gravity mnodel is used to explore whether. the gamut of regional -arrangements be- tween Uruguay -and its neighbors resulted in a noticeable increase in intaregional trade. While the analysis is crude and does not lead to a direct * identification of welf-are g-ains and losses, it suggests th-at Uruguay gained little fr-om GAUGE and PEG. Under GAUGE Uruguay imported goods from a relatively high-cost supplier and exported goods in which it had, for the mnost part, no comparative advantage. While the -analysis of unit values of export products under PEG showed no evidence of rents to Uruguay, de- spite the presumption in theory for such rents, Uruguay probably gained on balance from PEG because it permitted expansion of exports in which Uru- guay had a comparative adv-antage. The basic economnics of preferential trading arrangements A regional approach to trade liberalization involves preferential treatment of some trading partners, unlike unilateral trade liberalization, which reduces tradebarriers across trading partners in -anondiscriminatory m-ann er. GAUGE and PEG exempted only selected products from tariffs in what was in effect a partial free trade area. Each member maintained its own tariffs against G0M4i1and lOSSes finn bik&lat tradng- a~rwagemenis uith ArgenAi#d dnd Brazl 83 nonmembers. By contrast the treatyoan MERCOSUR calls for fonrmation of a customs union with a common external tariff. The -basic insight on free trade areas goes back to Viner .(1950), who introduced the concepts of urade creation and trade diversion. The simnplest model to Hiustrate these concepts is a three-country, two-good model. Con- sider three countries, A, B, C, potential partners in a free trade area. Let A stand for.Uruguay, B for Brazil, and C for the rest of the world. Figure 5.1I shows the general equilibrium inipoi-t-demand curve for countryA (MA) and the horizontal export-supply curves (PI, and P6) for the two parmner coun- tries. Suppose that two goods are traded and that each,partner imports a different good (a strong but necessary assumption for a free trade area be- tween A and B to have any effect). In the absence of trade intervention, the gain from trade (compared with the no-trade situation) for counta-yA is given by area ag~i. Now if A imposes a nondiscriminatory tridff, *, on all imports, country A wilI continue to import from' country C, the low-cost supplier. Reduced gains from trade are now equal to area abk If Aforms afree trade area with B, the cost to consurners wivll be less for duty-free imports from B than for imports from C, which are subject to the duty, so imports will now come from the partner. There will be trade cre- ation, equal to -area bd4 as the price paid by consumers fals by efand imports expand to point d. But there will also be trade diver-slbn, equal to areafek; as A's terms of trade worsen and tariff r'evenue falls. Area cbefrepresents a gain to consumers but a loss in tariff revenue, which cancel each other. Whether Figure 5.1 Trade creation and trade diversion Rdative price a (I +t)Ph bn As nafter free traade agreement itoueth ocpof eradecetoan rdcivso.Thsipls pb Ass kiss after free trade agreement A MA tries Supose Ft tw good areaadedand hat ech prmerimportsa 84 Emcrrs oF PRoTEcrloNIsM ON A SMALL CouNrRy: THE CASE OP URUGUAY the free trde areawillraisewelfare in A (and in the world asawhole) depends on which of the two effects-trade creation or trade diversion-dominates. Viner's analysis also points out the conditions under which a preferen- tial approach to trade liberalization is more likely to raise welfare: * The larger the initial tariff in the sector. * The lower the tariff on nonpartners, an important factor in the cre- ation of MERCOSUR - The smaller the difference in the cost of imports from the partner and the rest of the world. * The more elastic A's import demand curve. (On this last point: a large proportion of Uruguay's imports are concentrated in internediate and capital goods. forwhich the elasticity ofunport demand is likely to be low.2) The literature on customs unions shows that it is possible to design a welfare-augmenting free trade area and thata preferential trade agreement can be superior to a free trade area. But whether in practice preferential trading arrangements have increasedwelfare is an empirical issue. The con- sensus about integration schemes among developing countries is tat the agreements have generally been only partially implemented (usually on a product-by-product basis rather than across-the-board) and that designing and implementing schemes that provide equitable compensation for part- ners has been difficult? (In the case of CAUCE and PEC, as is demonstated below, the welfare effects of preferential arrangements depend on which country gets the rents when the ceilings on preferences are reached.) With few exceptions the literature on customs unions also indicates that a unilateral approach to trade liberalization is superior to apreferential one. The one important exception is the theoretical case (first discussed by Wonnacott and Wonnacott 1981) of a world divided into a small number of trading blocs with free trade among members and high trade barriers against nonmembers. For a small country, the market access provided by bloc mem- bership may be superior to the nondisciminatory route to trade liberaliza- don. Whether this scenario is relevant for Uruguay is difficult to assess since it depends on tade policies in Argentina and Brazil as well. All that can be said is that it is less relevant now than in the past, since both Argentina and Brazil have liberalized their tade regimes in recent years. Bilateral arrangements and the coupon quota system Uruguay has alonghistory of close economic tieswiti Argentina and Brazil.4 Notal the anrangements have dealtwith trade preferences. Many recent ones involve economic cooperation, mostly with Argendna, to establish common policies forportandtelecommunication acdvties (table 5.1) . Arrangements that establish common standards, regulatory frameworks, and practices re- Gains and losses from bilateral rading airangements uwi Argentina and Brazil 85 duce transaction costs, lead to the exploitadon of economies of scale and scope in transport, and reduce barriers to trade stemming from tansport costs and differences in regulatory frameworks. Clearly, such cooperation is welfare augmendng. Less cleariswhether thatis the case for Uruguay's trade policies extended on a preferential basis. Table 5.1 Bilateral agreements of Uruguay with Latin American countries Mcain eld Nrme of the agreenent Country Date and pbce Trade preferences Tratado de Asunci6n: Argentia. March 26, 1991 Merrado Cornun del Sur Brazil, Asuncid6n (MERCOSUR) Paraguay Protocolo de Exparisi6n Brazil June 1, 1975 Comerdal Uruguay-Brazil (PEC) Cornerdo Argentina-Uruguay Argentina August 12, 1974 de Cooperaci6n tdcn ica (CAUCE) Economic cooperation Acuerdo para la Realizaci6n Argentina February 22, 1985 (poicy ind planning) de Estudios Relativas a la Buenos Aires Construcci6n del Puente Buenos Aires-Colonia Acuerdos sobre Poliica ArgehnTna Febuary 22. 1985 Comnin en Matena Porttnria; BuenosAres Arnpliaci6n de Telecomunicaciones Acta de Colonia sobre Argentina May 19. 1985 Integraci6n Fsisa Colonia Pcuerdos sobre Sistema de Argentina February 2, 1985 Control Unico de Fronteras; Buenos Ares Poritica Cornmn en Materia Portuaria; Anpliaci6n de Telecomunicaciones.; Incremento de Ia Red Comun de Gasoductos, Estudios para la Construcd6n del Puente Buenos Aires-Colonia Mornorttdum sobre el Brazil August 13, 1986 Desarrolo de la Cuenca de Brsilia Ia Laguna Meon y Aprovechamniento de Recursos Hidricos del Ri)o Yaguar6n Acta de Colonia sobre Argentina May 19' 1985 lntegacid6n Fisica y Complementaid6n Agropecuaria Source: Peyes Chavez (I 988) and Ncgues and Quntanilla (1993). 86 EFFEC=s OF PROTECrIONISM ON A SMALL CouNmTv: THE CASE OF URUGUAY MERCOSUR - -Arger.tina and Brazil initiated the MERCOSUR arrangement in 1989; Uru- guayand Paraguayjoined in late 1990. This new approach to integration will substitute automatic across-the-board elimination of tariffs and nontiff barriers for the previous sectoral approach to integration. MERCOSUR is a commitnent to trade goods and services free of trade barriers as well as to ensure factor mobility among member countries. The timetable for tariff reductions sauto-matic, srting '"i'x the stecpest tariff cLits injune 1991 (47 percent) and progressing with gradual reductions every six months, until all tariffs are removed by December 1994. While it is too early to venture an assessment, several elements of the agreement suggest a need for caution. First, though the across-the-board -approach to tariffreduction is preferable to the product-by-productapproach (see discussionbelow), exemptions from the-automatic tariffreductionswere permitted for many products (especially for Uruguay and Paraguay) to ease adjustment.) Second, there has been considerable disagreement among partners about the level of the common external tariff, which as ofApril 1993 had notyet been agreed on. Brazil, the largest partner with the most protec- tive trade regime, seems to want a relatively high common external tariff. Third, the macroeconomic instability in the region has impeded implemen- tation of the agreemenL In September 1992 Argentina adopted a temporary surcharge of 6 percent on all imports, ostensibly as a buffer against Brazil's recent strong real exchange rate depreciation against the Argentine peso. CA UCE and PEG The oil crisis of 1973 and the dramatic slump in foreign demand for beef in 1974 as a result of European agricultural policy led to renewed interest in Uruguay in regional initiatves to encourage trade with its neighbors. The authorities believed that opening Uruguay to the world economy (see chap- ter 2) would not be sufficient to deal with the new adverse external environ- ment. They looked to bilateral trading arrangements with Argentina and Brazil to help reduce Uruguay's chronic trade deficts with its neighbors by reorienting exports to the region and increasing capacity utilization. The CAUCE and PEC agreements of 1974 and 1975 were to become an integral part of Uruguay's shift to an outard-oriented development strategy and reduced antiexport bias. The bilateral agreements lasted throughout the 1980s, until they were superseded by the MERCOSUR agreement at the end of the decade. Under CAUCE (and its extension in the Act of Colonia in 1985), Argen- tina allowed duty-free entry of Uruguayan products of up to 5 percent of Gains andw sesfiom bilatl tratding arangements witls Argenina and Brazil 87 Argentine domestic production of each productin the previous year. Miner- als, agricultural goods, and textiles were excluded. Uruguay was to enjoy unilateral tariff conce.ssions for five years. After 1979 these concessions were to be complemented by Uruguayan tariff reductions.7 The limitations of CAUCE are apparent from a closer inspection of the concessions themselves. Rules of origin under the Act of Colonia classified a goodasaUruguayan productonlyif less than .50 percentofits contents came from third countries. Uruguay's concessions covered capital goods made or assembled in Argentina only if there was no similar production in Uruguay. OtherArgentine goods were exempted fron tariffs if theywere classified as industrial products and had a low duty in Uruguay's import tariff system. Argentine goods whose share had exceeded 30 percent of Uruguay's imports between 1982 and 1984 were not eligible for tariff exemptions. The bilateral agreement with Brazil under PEC involved far more reci- procity. Brazil offered trade concessions on 1,406 items, Uruguay on 1,009 items. Rules of origin also required domestic content of 50 percent of the final value of the item. In an effort to balance bilateral trade, corrective neasures were caled for if the bilateral trade balance for goods included in PEC showed a deficit of more than 10 percent of the total value of wade. PECwas revitalized in 1986 through the treaty of Brasilia (see table 5.1). Uruguayan export products eligible for preferential treatment were placed into seven groups. Five had quota limits in value terms of $0.3, $0.5, $2.5, and $5 million (1986 dollars). One group had limits specified in volume or value terms, and anothergroup was exempt from limitation altogether. Goods could be reclassified into-a higher group-including the no-limitation group-if exports reached 90 percent of the quota limits (see description below) for three consecutve years. Butimportant loopholes remained. Each partner could limit imports if they surpassed 5 percent of usimilar" domestic production in the previous year. The quota limits under CAUCE and PEC were enforced through a cou- pon (cupO) system. In Uruguay the Chamber of Industries administered the system by giving out coupons to meniberswhowantedpreferential treatment for their exports to Brazil orArgentina. The coupons were generally given to the more established and influential firms. Unused coupons were to be re- turned to the Chamber for redistribution, but, in practice, the-y were traded in secondary markets. Trade patterns under CAUCE and PEC Though macroeconomic instability in the region led to considerable fluc- tuation in Uruguay's pattem of tradewithArgentina and Brazil, trade shifted progressively toward Brazil during the 1970s and 1980s (table 5.2). CAUCE. 88 FEmCaS OF PJOTECnoNIsM ON A, Svms COuRrRn THE CAsE oF URGUCJAY and PEC did not eliminate Uruguay's chronic trde defidt with Argentina and Brazil. During the 1980s Uruguay had a cumuladve merchandise trade deficit of $34 million with Argentina and $204 million with BraziL There are some fimdamental differences in Uruguay's pattems of trade with Argentina and Brazil Almost three-quarters of Uruguay's exports to Brazil are agricultural and natural resource-based products, whereas more thant three-quarters of its exports to Argentina consist of manufactures. The pattern of Uruguay's exports to Brazil is much more in line with Uruguay's pattem of commodity exports to the rest of the world and so with its com- parative advantage. That the composition of its exports to Argentina is quite different suggests that CAUCE allowed Uruguay to export prodtvcts to Argentina's protected mark-et thatwere not competitive on theworIdmarket. Data on the percentage of total exports by industry going to Argentina or Brazil provide further evidence of different patterns of exports to Brazil and Argentina Export concentration indices for 1987 based on trade classi- fications at the three-digit industry level show thatArgentina was the major outletfor motor cars (90 percent of total exports), paperproducts (70 per- cent), electrical machinery (48 percent), and "other chemicalproducts (37 percent). Brai was the major outlet for exports of barley (100 percent), fertilizer (89 percent), cerealpreparations (88percent),rubber (85percent), dyeing and tanning (83 percent), organic chemicals (80 percent), plastics (68 percent), and rice (46 percent). Table 5.2 Uruguay's exports under CAUCE and PEC, 1975-88 (mrrdiorsof U.S. do&ars) 4UCE e*om as PEC xportp as a 1 onm -f gfpernae ( Totol expos CAUJCE Eporrs to Total Total e%aors PEC E4xorts Total Year toAgena aoes Argentina epors to frail elfS to Braz eqors 1975 29.6 11.0 37.1 .. _ .. .. 1976 26.8 i3.5 50.3 21 . .. ..1 1977 31.9 15.4 48.4 . 1978 38.2 21.7 56.8 .. .. .. .. _ 1979 97.1 51.9 53.5 .. 1980 1433 88.8 62.4 8.4 1981 114.7 77.1 67.3 63 1982 109.1 66.0 63.5 6.5 145.8 26.7 17.7 2.6 1983 91.1 53.0 58.2 6.1 1213 19.6 162 1.9 1984 88.3 68.9 78.0 7.4 114.8 18.7 16.3 -20 1985 63.1 45.5 72.1 6.3 143.4 46.7 32.6 5.5 1986 88.7 743 83.8 6.9 295.6 82.4 27.9 7.6 1987 1132 97.9 86.4 8.2 204.1 137.4 67.3 11.5 1988 100.2 873 87.1 6.2 229.1 184.7 80.6 13.2 Not aaibble. S3re: Inoati ('990, table 5). CGaro de tddca Nacionales y Comnerc Irternadonal del Uruguay (CENCq. Banco de la RepCbica Orinal del Uruguay (BROU), Asocibcin Latirrnerina de Integrai6n (ALADI), and loint and pemranet olFice of the CAJCE (for 1975-81 ). Gains and blssesfrom likztcn trading a7angwments with Agentina and Brwzl 89 CAUCE and PEC appear to have had little impact on Uruguay's import composition. The commodity composition of Uruguay's imports from Ar- gentina and Brai remained quite similar to its overall import pattern for food, minerals, chemicals, andfinished consumergoods. Marked differences occurred only in fuel imports (extraregional concentration of imports), machinery (concentration on Brazil), and semifinished manufactures (con- centration on Argentina). How important were CAUCE and PEC in Uruguay's trade with Argen- tina and Brazil Historical data show that over time the share of Uruguayan exports toArgentina and Brazil covered by the bilateral agreements increased sharply (table 5.2) - By 1988 more than 80 percent of Uruguay's exports to Argentina and Brazil came under the bilateral agreements.8 These bilateral agreements can thus notbe brushed aside as oflittle importance in Uruguy's overall trade policies. But what about their welfare impact for Uruguay? The welfare inpact of bilateral preferences Apartial equilibrium modelis developed here to study thewelfare effects for Uruguay of CAUCE and PECG To simplfy the analysis, all three countries are assumed to be small in theworld market for the productreceiving preferen- tial treaunent and to supply exports to each other atincreaing marginal costs. Since Uruguay granted preferences onlyTforproducts notproduced lbcally, themodel assumes no import-competing domestic production in the import market. The discussion presents the case ofArgentina and Uruguay, but the analysis would be similar for the Brazilian arrangements, except as noted. Assume thatArgentina is a sufficently small importer of each good that the price in theworld marketis given (Pw; see figures 5.2 and 5.3). Argentina imposes a triff of tpercent (or maintains other trade restrictions thathave the same price-increasing effect). The supply curve of the domestic Argen- tine industry is SA, and Ss,u is the horizontal sum of the Argentine domestic supply curve and the Uruguayan export supply curve to Argentina in the absence of restrictions. With the coupon quota system limiting duty-free Uruguayan exports to Argentina to ppercent of Argentine output, the total supply curve for the Argentine market is shown by S4(1 + p) . The Argentine demand curve for the productisD. Consider first the case in which the quotabinds (figure 5.2). The pUice of the product in the Argentine market would be PA(1 + t ). Domestic pro- duction would be Qt, domestic consumption Q., and total imports Q.- Q, (ac) ofwhich Q2-Q, (ab = X) are from Uruguay and Q,- Q2 (c) are from the rest of the world. The bilateral agreement involves a cost to Argentina in lost tariff revenue on imports that enter free of duty from Uruguay under the arrangements (area abhg).Areafehgis adeadweightloss from inefficient 90 EFFECs OF PROTEnONISM ON A SMALL COUNIRY: THE CASE OF URUGUAY resource allocation because of the higherproduction costs in Uruguay than in the rest of the world. For Uruguay area defrepresents a gain to Uruguayan producers- Area abedrepresents quotarents Uruguayan producersare willing to supply quan- tity XJ atprice P,., but the product will sell in the protected Argentine mar- ketat Pw(1 + t). The distrbution of the quota rents depends on the market structures ofimportinginArgentinaand theexportindustryin Uruguayand on the method of administering the coupons. If importing is relatively com- petitive, administraton ofthe quotas on the exporting side by the Uruguayan Chamber of Industries would probably mean that the rents would go to the exporters in Uruguay, especially if they-are sufficiently concentrated to con- trol the market The BraziJian import markets would be analyzed in the same way, except that under PEC the quotas are specific quantities or values rather than per- centages of domestic production. With a quantityt-denominated quota, the line in figure 5.2 showing the total supply available from Brazilian and Urn- guayan supplierswould be parallel to the Brazilian industry supply curve but displaced to its right by the -amount of the quota. Since Uruguay's exports to Brazil are doser to Uruguay's general export patter than are its ex-ports to Argentina, Uruguay's export supply curve is probably not too fr above the world price, implying larger quota rents relative to deadweight losses. Now consider the case in wshich the quota does not bind (figure 5.3). Again, the equilibrium, price would be PA (1 + tA), imports would be Q - QI (km), ofwhich Q2- Q, (kl = Xu) would be fiom Uruguay. The lost ariff rev- Figure 5.2 Argentine import market: binding quotas Pice SA a i SA ~~~~~~~~SA( I +P) \ PwI +tl4 c,~r~Ip SA+ u d e f PW g h \I DA Q; Q2. Q, ~~~~~~~~Quantity Gains and tossr afinn bilateral trading armngements with Aigendna and Brazl 91 enue would be kir, of which kln represents a gain to Uruguayan export in- dustries and nlsr the deadwveight loss from higher production costs in Uru- guay than in the world market Because the quota does not bind, no rents are generated. Under both GAUCE and PEC agreements Uruguay offers duty-free en- try for specified products that are not produced in Uruguay. In view of the small size of the Uruguayan market, it is reasonable to assume that the world price of the imported product is given at Pw. Suppose that Uruguay imposes a triffat rate tt, (or other form of trde restrction that causes asimilar diver- gence of domestic and import prices). Consider first the case in which imports under the bilateral agreements do not take over the entire Uruguayan market (figure 5.4). Suppose that the bilateral arrangement under consideration is with Argentina and that the Argentine export supply curve is S... Then MA would be imported from Ar- gentna and M, - M4, from the rest of the wvorld. There are no consumer gains (gains from trade creation), just a displacement of imports from the rest of the world by imports from Argentina. The loss of tariff revenue to Uruguay (the loss from trade diversion) is vwzy, of which vwxis a transfer to Argendine exporters and xaty is the deadweight loss due to Argentine pro- duction costs that exceed the world market price. Duty-free access allows Argentine exporters to gain, at the expense of the Uruguayan treasury, and generates efficiency losses. Now consider the case in which there are gains from trade creation, al- though the overall welfare impact is ambiguous (figure 5.5). Suppose the Figure 5.3 Argentine import market: nonbinding quotas PRice PWAI +t, k'A0) Pw r S t Dy Q2 22Q Quandity 92 EFiECrs OF PROTECTIONISM ON A SMALL COUNTRY: THE CASE OF URUGUAY same '7orld market price (P.) and Uruguayan tariff (1.f) and an Argentine exportsupply curve SX If the same tariffwere charged on all imports- from Argentinaas well as from the rest of the world-no importswould come from Argentina. The domestic price to Uruguayan consumers would be P. (1 + tu), and M, imports would come in from the rest of the world. If the taiff were eliminated on imports from Argentina, the equilibrium price would drop to PA, reslting in gains to Uruguayan consumers of abec from the bi- lateral arrangements. In the case of perfect substitutability between Argen- Figure 5.4 Uruguayan import market: diversified sources of imports Pnice SXA x PW Wy z MA mu --\Qantiy Figure 5.5 Uruguayan import market: partner country sole supplier Prce a b _ _ _ _ _ _ _ _ Pnce 2 i u~~~~~~~~ 11 e~1 P w h_ M1 M2 on "t,,L l,b !- ; y Gains and loss fronm bil ate bdingaungmentstwunthAWge7nina andBrazl 93 tine products and products from third countries illustrated here, Argentine exports would take over the markeL However, tariff revenue of abh would be lost, leaving a net gain for Uruguay if area bed exceeds area oid or a net loss if itfalls short It seems liely that Uruguay's losses from trade with Argentina couln be large because Argentn does not generally export the capital goods and industrial products it exports to Uruguay, implying a relatively large differ- ential benveen P,f and Pw. In the case of Brazil relatively greater gains from trade creationand relativelysmalUerlossesfrom trade diversionare likelysince if Brazilian products are competitive on world markets, then the Brazilian supply curve is not likely to be far above the world market price. Analysis of product prices under CAUCE and PEC The analysis presented above suggests that the preferential arrangements would lead to price differentials on the export and import sides. Uruguayan exports would fetch a higher price in the Argentine and Brazilian markets than inworldmarkets (figure 5.2),?while imports (excluding dutypaid) from Argentina orBrazilwould be higher priced than imports from the rest of the world (figure 5.4), particularly if the products were not normally exported to world markets or the exporters had some control over market prices. To see whether the bilateral trade agreements resultedin systematic price differentials, unit values byproduct orin or destination (Argentina, Brazil, and the rest of the world) were collected and analyzed for a sample of prod- ucts covered by CAUCE and PEC. The sample includes f.o.b. unit values for the twenty-nine mostimportantproducts exported by Uruguay to Argentina and Brai during 1986-89 and the c.i.Ł values for the thirteen most impor- tantproducts imported under CAUCE and PEC during 1986-90. (The prod- ucts covered in the sample are listed in appendix table 5Al1.) Two methods of analysis were used. The firstwas to compare unitvalues in eah commodity classification for Uruguayan trade with Argentina, Bra- zil, and the rest of the world. On the basis of the previous analysis of trade patterns under CAUCE or PEC, wewould expect the unitvalue ofexports or imports to be higher on trade with Argentina or Brazil than on trde with the rest of the world. On the export side, the results are consistent with this expectation: for the thirty export products in the data set for 1986-90 the unit values of Uruguay's exports to Argentina or Brazil exceeds those of exports to the rest of the world more frequendy than thereverse (table 5.3). On the import side, the results are contary to expectaadons: for the thirteen most important Uruguayan import products covered by CAUCE and PEC, the unitvalue of imports from Argentina and Brazil was lower than the unit value of imports from the restof the worldmore frequendy than the reverse. Table 5.3 Comparison of unit values of Uruguay's exports to and imports from Argentina, Brazil, and the rest of world (number ofoccurrences) 1986 1987 1988 1989 1990 Higher to or fom Higher to or from Higher to or from Higher o or from Higher to or from Rest or No Rost or No Rest of No Rest or No Rest of No vrld Partner comparison ˝rvd Partner comparison wrfd Partner comparison *vdd Partner comparison Wordd Partner co poadson Exports Argenbtna 5 7 18 5 10 15 8 7 Is 4 9 17 7 5 18 Brazil 7 a 15 5 12 is 7 11 12 5 13 12 6 13 11 Imporn Argentina 12 1 0 12 1 0 10 3 0 10 3 0 8 4 i Brazil 9 3 1 9 3 1 9 2 2 8 3 2 7 3 3 Source. Authoes calcblations. Gains and lossesfAm bilatezml Wadingarmrngmernss withArtina and Bazir 95 Unobserved product differentiation may account for this result, with price differentials perhaps reflecting lower-quality products from Argentina and Brazil than from Uruguay's other trading partners. Also, the analysis does nottakeintoaccountthe dergeofprice difference across origins ordestinations. The second approach was to pool the products and to test for systematic pnce differentials between trade-with Argentna or Brazil and trade with the rest of the world, using the following model = X tYa4DPi*DT, JIPXDA*DP& +XDB*D& (51) + A+D+Eje where Pis the unitvalue for export ito countryjin year t DP. and DT are dummy variablesforproduct and time, DAandDBDare dummyvariables for Argentina and Brazfl, and Ev, is an error term. The first set of interactive variables is designed to capture the effects of differences in productand time period. The second and third sets of interactive variables show the effect of the bilateral agreements, by product. The indiiidual country dummy vari- ables indicate the average effect of the bilateral agreements across products and time- A similar specification would apply for import unit values. This model lends itself to several tests. Omitting the second set of in ter- active variables (those showing the effects of the bilateral agreem ents prod- uct byproduct) permits testing for the presence ofa significant difference in the mean value of the markup on sales to Argentina or Brazil. Omitting the country variables allows testing for significant differences in prices at the product leveL The overall fit is generally very poor for the import unitvalue equation (R2 of around 0.04), except for one specification discussed below, and very high for the export unitvalues (R2 above 0.92). This resultreflects the obser- vation noted above that most imports into Uruguay covered under CAUCE and PEC are manufactured products that tend to be highly differentiated. Tests for differences in prices at the product level show no significance on the export side buta strong significance on the importside. The F-value for the joint significance of the inclusion of the second set of interactive variables for the effectofbilateralagreements byproductis 7.04 (criticalvalue at the 1 percent significance level is 1.64). This result again confims the pres- ence of more producthe terogeneity on the import side. However, the F-value for a significant difference in the mean value of the markup on exports to Ar- gentina and Brazil is only 0.92. So the data do notindicate a significantmarkup (positive or negative) for exportproducts sold to Argentina and Brazil. On the import side the coefficients of the duramy variables captring the presence of a mean markup are positive and highly significant for prod- 96 EFFECIS OF PROTECrlONIS ON A SMALL CoURY: THE CASE OF URUCUAY ucts originatng in both Argentina and Brazil (tvalues above 6 and a com- puted F-value of 23.6). While this resultis along the lines expected from the previous discussion of trade patterns under CAUCE and PEC, it is at odds with the simple comparisons reported in table 5.3. It is difficult to establish with confidence from these tests that Uruguay paid a higher price for im- ports under CAUCE and PEG, especially in view of the strong heterogeneity of products across exporters to Uruguay and the inability to control for dif- ferences in product quality across suppliers. At the product level the coefficients are statistically significant (at the 5 percentlevel or better) on three of the tfirty products one positive and two negative for Argentina and two positive and one negative for Brazil. On the import side none of the thdrteen products imported from Brazil and Argen- tina has a statistically different mean value, after varations in exogenous factors are controlled for through the use of the time dummies. In sum, examination of price differentials across product origins and destinations reveals no clearpatternL On the export side there were no signs that rentswere accruing to Uruguay from export sales to Brazil, which suggests that the coupon quotas were not binding.'0 On the import side the econo- metric evidence of ahighermean price forimports originatinginArgentina andBrazil suggests thatthe preferentialarrangements gave rise to inefficiency. Because differences in qualityacross suppliers are not capturedin the model, however, this result cannot be established with confidence. Have CAUCE and PEC affected the vohlme of trade with Argentina and Brazil? So far the data have indicated that an increasing share of Uruguay's exports was covered by the bilateral arrangements (table 5.2) and that the resulting trade pattern could well have been inefficient.hThe inspection of price dif- ferentials was inconclusive, however, especially on the import side, so the welfare effects of the agreements remain ambiguous. Acomplementaryapproach is to considerwhether there is evidence that the bilateral trade agreements significantly expanded trade between Uru- guayanditsneighbors. One approach is toinvestigatewhetherUruguay trades more orlesswith these countries thanwould be expected on the basis ofsize, income, location, and other characteristics thathave been successfully used in other cases to predict trade patterns across countries. The idea is to find out whether, controlling for other factors, Uruguay's trade with Argentina and Brazil is significantly different from that predicted by the model. A gravity modelwas developed to test for this possibility (see Bergstrand 1989 for the theory ofgravity equations) . The model is similar to models used to examine the effects of preferential trading areas (Aitken 1973) and tO Cmn and bossesfimo bnila ading anrnngents udth Arentirnaand Brzil 97 predict the geographic shift of East European trade from 'east' to "west" followingrecent economic and political changes (Havrylyshyn and Pritchett 1991).The results suggest that thebilateralarrangementshadlitde effecton Uruguay's pattern of trade. The value of trade, T, between two countries iandjis postulated to de- pend on the total trade potential, TP, of each country and on the trade at- traction between them, TA: (5.2) T f(TP', TPi, TA"Y) Separate equadons are estimated for imports and exports of manufactures industry classification codes 5,6,7, and8). Trade potentialdepends on total outputand the trade intensity of output, which depend on economic factors such as level of development and geographic factors such as country size. Although the theoretical underpinnings of the gravity model are not clearly specified, they can be linked to tastes and factor endowments (see, for ea- ample, Bergstrand 1989). Two versions of the gravity model are used to test for the presence of bilateral effects. ModelI] regresses Uruguay's trade (im- ports and exports) with agroup of eighty-four trading parmers and includes separate duramyvariables for Argentina and Brazil. Because there is perfect correlation betwveen the dummyvariables and the usualbordervariable used to measure geographic proximity, a second sample was constructed. Model 2 uses asample of twenty-three semi-industrialized countries selectedfor their comparability with Uruguay. This model, similar to that of Havrylyshyn and Pnitchett (1991), allows separate testing for the effects of the bilateral agree- ments with Argentina and Brazil. The equation for model 1 is: T" = P+P1(DSTANCE )+ P2(GDP) + pS(GDPPC )+p4 (AREA) +f35(lSLAND j )+ (LlDERJ) + XmDL (5.3) +t7DA + P8DB where Th refers to Uruguay's merchandise imports from or exports to a trading partner j, and DA and DB are dummy variables for Argentina and Brazil, designed to capture any influence of CAUCE and PEC on Uruguay g's pattem of imports and exports. The en-or term has been omitted. The other variables are proxies for size (AREA and ISLAAD), geographic proximity (DISTANCE), economic simiarity (L[DER: absolute difference in per capita GDP between Uruguay and country j at purchasing power parity), and cultural factors (DL). (Precise definitions and data sources are given in the appendix.) 98 EFrcr oF PROTECTnoNISM ON A SMALL CouNmv THE CAE OF. URUGUAY 'rhe equation for model 2, which estimates trade flows of twenty-three countries (indexed over t) with their eighty-four trading partners (indexed overj), is: T"i = 0 +1(D1STANGE 1)+2(BORDER) + P.(GDP) +134(cDPPC ) +f35C(AR i)+ P6(ISLAND )+ P70Pi +J35(GDPPCj)+ 39(APEAj) +1j0({ISLANDi)+jj(LJFNDERO)+ XSWZDLY +8,DA+52DB (5.4) + 5LAT Model 2 is similar to model 1, but allows the bilateral arrangements to be isolated by including a separate BORDERvariable to control for geographic proximity. In an alternative formulation, DA and DB are replaced by the dummyvariable DABU, vhich controls for trade between Argentinaand Brazil as well as trade between Uruguay and its neighbors. Both models are esdmated ona cross-section basis usingyearly trade and icome data. The estimation is repeated for severalyears (1970-87 formodel 1, 1980-88 for model 2) because trade between Uruguay and its neighbors has fluctuated significantly from year to year. (Appendix table 5A.2 reports results for 1985; table 5A.3 shows the stability of the results over time.) Since bilateral de flows have a lower bound of zero, censoring bias renders or- dinay leastsquares inconsistent (Greene 1981). Tobitmaximum likelihood estimates are therefore reported instead." The results for model 1 are quite good, as they are with other applica- tions of the gravity model. All the usualvariables have the expected sign and strong sttistical significance. Uruguay's trade decreases with the distance and physical size of the trading partner and increases with the partner's eco- nomic size and income per capita, although the difference in per capita income (a proxy for similarity) is insignificant As in other applications, the island dummy is generally not significant The dummy variables for Argentina and Brazil are of particular inter- est, since they capture thejoint effects of a border dummy and of prefer- ential arrangements Surprisingly, they are negative for imports and exports, and strongly significant for exports (though the significance is not stable over the years-see table 5A.3). The grvity model does not control for the effects of trade policy or of other important determinants of the pattern of trade, such as factor endowments. But insofar as the variables in the model are good proxies for the deterninants of Uruguay's trade, the results indi- cate that Uruguay is exporting significantly less to Argentina and Brazil than would be predicted by size, income, and geographic proximity, casting doubt on the effectiveness of the preferential agreements in increasing intraregional trade. gairn and losssfrom bilateral bading arrangments with Avgentina and Brazil 99 Thiese results do not carr over to model 2, which expands the number of sample countries and includes dummy variables for BORDER and for Imembership in the Latin American Free Trade Area (LAFTA). Now the ef- fectof the BORDERvariable is significantlypositive, as expected. The LAFTA dummy (Argentina, Brazil, andUruguayare LAFTApartners) isalsostrongly significant, suggesting that the preferential arrangements had a positive ef- - fect on the volume ofimports and exports, and thatsignificance is stable over time. After controlling for the influence of bordering countries on trade * patterns, there is no additional effect from CAUGE and PEC. Replacing the dummy variables for Argentina and Brazil with a dummy variable that controls for all patterns of trade among Argentina, Brazil, and Uruguay (DABR) yields a statistically significant negative coefficient (table 5AS3). Because the values of all the other coefficients are stable and close to those reported in table 5A.2, they are not reported for this alternative for- mulation. Within this sample of semi-industrial countries, then, the three countries seem to have rather similar production structures ('competitive" cost structures) so that they trade mostly with the rest of the world rather than among themselves. If that is the case-and the same observation has been made frequently in evaluations of regional arrangements among de- veloping countries (see de Melo, Montenegro, and Panagariya 1992)-inte- gration would have little effect since most trade would continue to be with the outsideworld. This seems to be the resultsuggested by the gravityrmodel, which is not surprising given the incomplete coverage of GAUCE and PEC. Alternatively, the significantly negative coefficient on DABU could be cap- turing the effect of some omitted variable that is an important deterninant for the pattern of trade for the three countries. How robust are these results, and are there any discemible trends dur- ing the 1980s? Despite relatively large changes from year to year in the esti- matedvalues of the coefficients for the dummyvariables capturing the effects of the bilateral accords in model 1, there is a trend in the value and signifi- cance ofthe dummies on the importside toward less negative numbers (table 5A.3). That trend suggests that regional. policies were progressively increas- ing Uruguay's imports from Argentina and Brazil. For exports both dummy variables always have a negative sign though the significance is greater for CAUCE than for PEC. Significance falls only in the later part of the 1 980s, suggesting a progressive effect for GAUCE and PEG and torroboratng the previous observaton of an increasing share ofpreferential exports in the value of toml exports toArgentnaand Brazil (table 5.2). Formodel 2 thevalues of the LA4KJA and DABU dummy variables are quite stable, and so no trend emerges for changes in the values of the coefficents of the dummyvariables. To sum up, the gravity model suggests litde evidence of a changing pat- tern of imports and exports to Argentina and Brazil during the 1980s as a 100 EFFCIS OF PROT M ONISM ON A SMALL CouNwRY: THE CASE OF URUGUAY result of the preferences granted through CAUGE and PEG. The gravity model, though it suffers from omitted variable bias and uncertain dteoreti- cal foundations, corroborates the results of the analysis of Uruguay's trade patterns presented earlier to suggest that the bilateral arrangements had li tde effect on Uruguay's pattern of manufactured goods trade. Conclusion GAUCE and PEC share many of the characteristics often found in regional arrangements among developing countries. Though the arrangements granted duty-free access to a large number of products, numerous restnc- tions limited overall effectiveness (ceilings on the volume of duty-free im- ports, rules of origin, corrective measures if bilateral trade became imbalanced). Analysis ofthe pattern of trade under CAUCE and PEC revealed that an increasing share of regional trade during the 1980s was covered by the arrangements. Estimates of bilateral trade flows from a gravity model suggest that the bilateral agreements had no significant effect on the overall pattem of Uruguay's trade during the 198Os but that LAFTA did have a'sig- nificantly positive effect on Uruguay's manufactures imports and exports. If anything, Uruguay traded less with her neighbors than would be expected from the estimates of the gravity model. -Did Uruguay benefitfrom CAUCE and PEC? Afirm answeris difficult to reach, even in the narrow sense of static gains and losses. But it can be said that under CAUCE Uruguay was importing products from a reIatively high- cost supplier and, to a greater extent than under PEC, exporting products (mostly manufactures) in which it has no comparative advantage. It is ther- fore likely that CAUCE resulted in netlosses for Uruguay. In the case of PEC, theory suggests that Uruguay would gain rents from its access to Brazil's protected agricultural market, but the analysis of export productunitvalues found no significant evidence of rents. Overall, however, Uruguay probably gained from PEC because the agreementallowed Uruguay to expand expors of products in which it is a low-cost producer. Whatabout MERCOSUR? It is too early to tell whether it is likely to have benefits for Uruguay. The across-the-board approach to tariffreductionand the absence of quota limitations should contribute to greater trade expan- sion than under CAUCE and PEC. But credibility problems arise because of the longlistofUruguayan products exempted from the tariffreduction sched- le undl 1994 and Argentina's temporary import surcharge of 1992. Also, implementation maybe difficult ifmacroeconomic instability persists in the region. Overall, MERCOSURwillprobablyincrease Uruguay'swelfare if the common extenal tariff is low. But if the adopted external tariff is high, the preferential approach to trde liberalzation mayreduce Uruguay's welfare. Gains and losseśflom bi&atnil frding arrangmnts with Arentina and Brazil 10J Appendic This appendix lists the products analyzed in the section on prices under CAUCE and PEC (table 5A.1) and describes the samples and variables used in the gravity model. Table SA. I List of sample of products covered under CAUCE and PEC NAVE' code Desaflption Import side 27100919 Other lbdcatingoil of heavygrade 29030000 Sulfonated, nitrted, or nitrosated derivatives of hydroarbons 29300200 Tduene di-4socynate 32070199 Other colorig matter-either of inorganic products, induding those of mineral origin, or of clored-organic material 39011901 liquid polyesters 39020210 Polymerization and copolymnerization in any form 47010520 Pulp derived by mechanical or chemical means frorn any fibrous vegetable material 48010101 Paper or paperboard (including cellulose wadding) in rolb or sheets-for newsprint 51040100 Woven fabrics of synthetic fibers (continuous), induding wvven fabrics of monfil 51040100 Cotton, not carded or conbed 73130300 Sheets or plates of iron or steel, hot rolled or cold rolled 76010200 LUnwrought aluninum; aluminum waste and scrap 87020112 Passenger automobile Its with cylinder engines between 100cc and 1600cc, induding parts used in other motor veh ides (ergine and cab are classified here) EVr !sde 02010600 Meat and edible offals of cows, pigs, sheep, lambs, goats and kids, and horsemneat, fresh, cilled, or frozen 03014505 Fish, fresh live or dead), chilled or frozen 04020201 Mik and cream, preserved, concentrated or sweetened 04030100 Butter 04040209 Cheese and curd 10030199 Barley 10060200 Rice, husked or unhusked but not parboiled 11070100 Malt, roasted or nct 29030101 Sulfonated, nitrated, or nitrosated derivatives of hydrocarbons 29150501 Octyl phthalate 31030302 Mineral or chemical fertilizers, phosphaic 32090300 Vamishes and lacquers; distempers; prepared water pigrments ofthe knd used for finishing leather 32090803 Uquid vamishes containing no soens 381 10500 Disinfectants, insecticides, fungicides, rat poisons, and herbicides 39020204 Chloride polyvinyl in dfferent forms 39020401 Polyester in dferent brmrs 39030200 Regenerated cellulose 40050200 Plates, sheets, and strips of unvulcanized natural or synthetc rubber 40110102 Tires for passenger and commercial automobiles 48010202 Paper used for books 48070101 Paper and paperboard to which superfiaal coating or layers of other materials have been applied to the whole or part of one or both surfaces (ccrtinued on next page) 102 Erncrs OF PROTECTIONISM ON A SMALL Counmur. THE CAsE OF URUGUAY Table SA List ofsample of products covered under CAUCE and PEC .(cant.) NADE' code Descaption 5 poit side (con) 48160101 Boxes, bags, and other paddng containers of paper and paperboard 5101 0302 Yarn of synthetic fibers (conrtnuous), not for*retail sale 51040202 Woven fabrics of synthetic fibers (continuous) 53110101 Woven fbrics of sheep's orlarnbs wool orof fine a irmal hair 55098901 Other woven fabric o(cottDn 56010102 Synthetic fibers (discontinous). not carded. combed or otherwise prepared for spinning 56070400 Woven fabrics of synthetic fiber (discortnous or waste) 61010503 Cotton pants for men and boys a. Unrguays tarilf dassiictiorn system aNomendatura ara L dassifiocion Ce ta merconcias en bs aranceles Ce aduanas). The data set used for model 2 contains twenty-three reporter countries (subscripted by index t). These twenty-three coantries were chosen from among those classified as "middle-income economies" in the World Develop- ment Report 1991. They are: Algeria, Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Greece,Jamaica,Jordan, Korea, Malaysia, Mexico, Para- guay, Peru, Portugal, Thailand, Trinidad and Tobago, Tunisia, Tur- key, Uruguay, and Yugoslavia. Models 1 and 2 were estimated for eighty-four partner countries (sub- scripted by index ). The sample was restricted to partner countries whose imports totaled more than $300 million in 1980 and forwhich no data were missing. The selected partner countries are: Angola, Argentina, Australia, Austria, Bangladesh, Benelux, Benin, Bolivia, Brazil, Burma, Cameroon, Canada, Chile, Colombia, Congo, Costa Rica, Cote d'Ivoire, Denmark, Dominican Republic, Ecuador, Egypt, El Salvador, Ethiopia, Finland, France, Gabon, Ghana, Gua- temala, Hait, Honduras, Hong Kong, India, Indonesia, Iran, Ireland, Israel, Italy,Jamaica,Japan,Jordan, Kenya, Korea, Kuwait, Liberia, Madagascar, Malaysia, Mauritania, Mauritius, Mexico, Morocco, Netherlands, Nicaragu2, Nigeria, Norway, Oman, Pal;istan, Papua N. Guinea, Paraguay, Peru, Philippines, Portugal, Saudi Arabia, SenegaL Singapore, Somalia, Spain, Sri Lanka, Sudan, Sweden, Switzerland, Syria, Thailand, Tobago, Togo, Trinidad and Tunisia, Turkey, United Kingdom, United States, Venezuela, West Germany, Zaire, Zambia, Zimbabwe. Gains and losssfrm biterd tradingarrangments mith AeTnlia and Braud 103 Variables are defined as follows: T*. the dollarvalue of either inports or exports ofmanufacaured goods (the sum of SITC codes 5, 6, 7, and 8). Source: UNSO COMTRADE database. * D1STANC'E'i: logarithm of the straight-line distance between the econo- miccenterofgravityoftherespectvecountries. Source:Linneman (1966). - BORDER': dummy variable, with a value of 1 when the reporter and the parmer share a border, and 0 otherwise. * CDPPC': logarithm of the purchasing powerparity (PPP) income per capitaof the reporter countries Source: Summers and Heston (1991). * AREA 4 logarithm of the area of the reporter country (measured in thou- sands of square kilometers). * ISLAAD ' dummy variable indicating whether the reporter country is an island value of (1) ornot (0). * ISLAADJ: dummy variable indicating whether the partner country is an island (1) or not (0). * COPi.: logaithm of the totl GDP of the reporter county (millions of dollars). Source: World Bank dart = CDi': logarithm of the total GDP of the partner country (millions of dollars). Source: World Bank data. * CDPPCf: logarithm of the PPP income per capita of the partner coun- tries. Source: Summers and Heston (1991). * AREAJ: logarithm of the area of the partner countries (measured in thous-ands of square kilometers). * LAETA#: dummyvariable thattakesavalue of 1 when the reporter and the partner counEres both belong to the LAFTA, -and 0 otherwise. - DA: dumnmyvariable that takes a value of I if the trade is between Ar- gentna and Urugluay, and 0 otherwise. * DR dummyvariable that akes avalue ofl if the trade is between Brazil and Uruguay, and 0 otherwise. * DL/: dummyvariable that takes avalue of l if the reporter countryand the partner country share English as a common language, and 0 otherwise. * DL 7: dunmyvariable that takes the value of i if the reporter county and the partner country share Spanish as a common language, and 0 otherwise. * DL.J dummy variable that takes the value of 1 if the reporter country and the partner countryshare Portuguese as a common language, and O otherwise. LTNDER: logarithm of the absolute value of the difference in per capita GDP between the reporter and the partnar'. Source: Summers and Heston (1991). 104 Emcrs OF PROTECnONISM ON A SMALL CoUmN: THE CASE OF URUcUAY Table SA.2 Gravity model estimates (1985) -Ncn-oi imports Non-dl eaports Voaiable Model I Model 2 Model I Model 2 Constant 34.76 (-3.22)) -46.16 (-15.38) -5.22 (-0.62) 67.14 (-20.27) hI (DISTANCEU) 4.42 (-3.33) -0.48 (4.49) -5.96 (-5.48) -0.67 (-5.90) ' BORDERJ (= 1) 3.89 (6.49)* 4.38 (6.90)" PRorter -k In (GDP') 138 (10.42) 2.70 (19.18) In (GDPC') -0.31 (-0.83) 1.83 (4.57)" In (AREA') -0.55 (-5.96) ' .73 (-7.45)" ISLAND' -0.40(-1.12) -1.25 (-3.25) In (GDPJ) 2.70 (19.18)" 1.43 (14.73)" In (GDPPC') 4.82(7.86) e 0.62(4.05)" 3.89(8.82) 0.02 (). 10) In (AREA') -1.06 (-3.29) -0.43 (-5.74) -0.52 (-2.0) -0.47 (-5.86)H in (POPJ) 3.05 (6.07) 2.24 (5.93)" ISLAND (-j) 0.36 (031) -0.03 (-0.10) -0.15 (-0.16) -0.66 (-2.41)' UNDER (= 1") -021 (0.52) 0.72(7.82) 0.29 (-0.87) 0.86 (.65)" Preeren- I YI) LAFTAI 2.22 (4.07) 2.47 (4.25) Language(= ') ENGLISH -0. r I (-0. 14) 0.65 (0.78) SPANISH 1332(4.85) 1.01 43.47) PORTUGUESE -0.61 (-1.96) *0.25 (-0.80) Bikarel dumrries DA (1) -339 (0.72) 0.12(0.03) -9.34 (-2.38) 1.63 (039) D5 (=1) -2.46 (-0.59) 035(0.09) -7.94(-2.26) 1.71(0.41) Sumrnmary swdstics N (Obs) 84 1932 (23 x 84) 84 1932 x23 84 SignfiFrntat5% 'Signifcantat 1%. Nbt Numbers in parentheses are t-sbtastics. Cans and lossesfinm bibl tradinganMnge unt ih Arntina and Braid 105 Table SA.3 Stability of the bilateral dummy vraables Model I hnpwts Expmr= Year DA DB 1970 - 11.68 (-2.56) -5.62 (-1.47) -13.81 (-3.14) -4.11 (-1.13) 1971 -10.71 (-2.19) -4.93 (-1.19) -14.31 (-3.22) -4.19 (-1.17) 1972 -7.45 (-1.69) -4.93 (-1.34) -13.66 (-2.61) -4.10 (-0.96) 1973 -7.79 (-2.02) -5.71 (-1.72) -11.41 (-2.31) -5.29 (-132) 1974 -9.41 (-1.80) -6. (- 38) -I [.46 (-2.58) -6.20 (-1.6S) 1975 -3.53 (-0.63) -2.93 (-0.60) -J 1.65 (-220) -679 (-1.60) 1976 -4.73 (-0.85) -3.64 (-0.74) -1052 (-1.88) -6.16 (-1.26) 1977 -8.05 (-1.46) -5.96 (-1.4) . -13.41 (-2.84) -8.02 (-20) 1978 -6.10 (-1.1 1) -4.96 (-1.01) -8.40 (-1.76) -5.73 - 135) 1979 -7.25 (-1.24) -5.29 (-1.01) -11.76 (-39) -7.93 (-1.83) 1980 -5.05 (-0.90) -4.12 (-0.82) -1039 (-2.26) -8.21 (-1.97) 1981 -5.46 (-0.95) -3.54 (-0.69) -8.62 (-2.12) -6.76 (-1.87) 1982 -2.33 (-0.46) -2.15 (-0.48) -7.68 (-1.70) 6.84 (-1.69) 1983 -5.64 (-1.09) -3.95 (-0.86) -735 (-1.73) -6.03 (-1.60) 1984 -4.17 (-090) -2.81 (-0.69) -11.63 (-283) -8.31 (-2233) 1985 -3.39 (-0.72) -2.46 (-0.59) -934 (-2.38) -7.94 (-2.26) 1986 -0.48 (-0.12) -0.49 (-0.14) -6.70 (-1.69) -5.78 (-1.62) 1986 -1.30 (-037) -1.49 (-0.48) -8.38 (-2.09) -6.89 (-1.92) 1988 -3.71 (-1.10) -3.23 (-[.06) -5.27 (-1.39) -3.66 (-1.07) Model 2 IMPo EDarts Year A DABU LAFTA DABU 1980 248 (4.42) -4.99 (-337) 2.79 (4.78) -5A46 (-3.55) 1981 1.84 (3.24) -4.51 (-3.02) 239 (4.21) -5.11 (-3.40) 1982 1.90 (334) -3.06 (-2.04) 2.13 (3.70) -3.76 (-2.47) 1983 246 (4.19) -2.76 (-1.77) 2.80 (4.79) -3.61 (-2.31) 1984 248 (439) -2.56 (-1.71) 2.62 (4.40) -374 (-2.37) 1985 2.40 (423) -2.27 (-.5 2.76 (4.68) -3.37 (-2.14) 1986 .248 (4.83) -290 (-213) 3.04 (5.45) -4.08 (-2.76) 1987 267 (5.52) -2.51 (-1.99) 2.75 (5.03) -3.87 (-2.72) 1988 3.64 (6.20) -3.20 (-1.73) 3.86 (5.63) -4.69 (-2.17) A isthe value of the acocient for the dumny vriable forArena mn equation 5.3. DB is the value of the coeffident for durnmy te variable for Brazil in equation 5.3. LARA is the %aue of the coefrident for the LAFTA dummy in equabon 5.4. 04BU is the value of the coellident for the dummy varv ble for trade between Uruguay ard its neighbors or between Asrgfnna and Bazil in an alterative fbrmuhtitn of equabon 5.4. Abte Numrrbers in parentheses are t-stati;cs. 106 EFnErs oF PROTECTONISM ON A SMALL CoUNTRY THE CASE OF URUGuAY Notes We thank Elbio Nattino for helpful conversations and for the product line data, and Lant Pritchett andJames Tybout for comments on an earlier drafL 1. Uruguay's small size, coupled with thecas of communication with its two neigh- bors, has made 5uch isolation difficult (see Hanson and dc Melo 1985 and de Melo 1987). Indeed, a major motivation for the reference price and minimum export price systems examined in chapter 3 was to protect Uruguayan producers from shocks due to the large and sudden exchange rate shifts with Argentina and BraziL 2. In the case of increasing costs for partner B, the greater the complementarity in import demands between A and B, the largcr the gains from a free trade area. 3. For an illuminating discussion of the compensation issue see Foroutan (1993), and for an assessment of implementation see de Melo, Montenegro, and Panagariya (1992). For general propositions about conditions under which the discriminatory approach to trade lbheralization is likly to be trade creating (or trade diverting), see de Melo, Panagariya, and Rodrik (190Z). 4. This section, and the next, draw on Inotai (1990) 5. Some 300 Umxguayan products have been exempted from the automatic time- table for tariff reductions, with the move to free trade for these products to take place in one shot in the last phase of tariff climination in December 1994. 6. MlERCOSUR was preceded by seventeen protocols signed between Argentina and Brazil during 1986-89. It calls for the complete elimination of tariffs on regional trade by 1995 and the creation of a common market by 1998. Tariff reductions are automatic, following a timetable that puts tb.e steepest tariff cuts at the beginng (see Nogues and Quintanilla 1993). 7. Very little reciprocityactually took place on the Uruguayan side. Between 1975 and 1 985 Argentina provided preferences on 1,042 products, Uruguay on 377 (Ferrere 1984). 2. Even though alarge proportion oftrade was under CAUCE and PEC, there has been little change in the pattern of intraindustry trade. Kaplan, Sarachaga, and Vera (1988) calculated indices of intraindustry specialization for Uruguay's trade with Ar- gentina and Brazil and found no significant effect for CAUCE and PEC. 9. This is the market-access argument developed by Wonnacott and Wonnacott (1981) . TIhey show that access to a market that has high barriers with the rest of the world can be superior to unilateral trade hlberalization. 10. In a recent study Correa (1993) examined quota udlization rates for 1987-91 under CAUCE and PEC and conduded that about 60 percent of the quotas were less than half filled- 11. Heteroskedasticity-consistent standard errors are used for the maximum like- lihood estimates (see White 1981). References Aitken, N. D. 1973. 'The Effect of the EEC and EFEA on European Trade: A Temporal Cross-Section Analysis." Amernican Enomric Riew 63(5): 881-92. Bergstrand,J. 1989. "rhe Generalized Gravity Equation, Monopolistic Competition, and the Factor Proportions Theory in International Trade." Review ofEconomics and Statsti cs: 143-53. Gains and hbsrsfiom bi&ateal trading anzngement idih Argentina and Bradl 107 Correra, Femando. 1993. "Los Cupos y Ias Perspectivas del Comercio Regional: La Experiencia del CAUCEy dcl PEC." SU/MA 8(14): 39-64. Ferrcre, D. 1984. 'The Uruguayan-Argentinian Trade C operationAgreement."Jour- 7UZn of World Trade Law 18: 320-34. Forou tan, F. 1993. "Regional Integration in Sub-Saharan Africa: Past lExperience and Future Prospects." InJ. de Melo and A. Panagariya, eds., NewDimensions in RPgional Integration. Cambridge University Press. Greene,W. 1 981. 'On thcAsynptotic Bias of the 015 Estimaborof theTobit Model.* Eeonometrica 49(2): 505-13. Hanson, J., andJ. de Melo. 1985. "External Shocks, Financial Reforns and Stabiliza- tion Attempts in Uruguay: 1974-83." WorldDevekpmnt 13 (8): 917-39. Havzylyshyn, O., and L. PritchetL 1991. 'European Trade Patterns after the Transi- tion. 'PolicyResearch WorkingPaper 748. World Bank, Trade Policy Division, Wash- ington, D.C. Inotai,A. 1990. Unruguay's Bilateral Economic Relations with Argentina and Brazil.' World Bank, Trade Policy Division, Washington, D.C. Xaplan,M., D. Sarachaga, and T- Veia. 1988. ntegracion, politica comercialycomercio mtnindustrial evidcncia para la valoraci6n del comercio intraregional.' Reista de Ecnonzmza 2(3): 89-118. inneman, H. 1966. An Econometric Study of finmrnational Trade Amsterdam: North- Holland. Mclo, J. dec 1987. 'Financial Reforms, Stabilization and Growth under High Capital Mobility: Uruguay 197-83.' In RP Connolly and C. Gonz5lez-Vega, eds., Economic Refom and Stabilzation in Latin America Praeger. Melo,J. de, A Panagariya, and D. Rodrik. 1993. 'he New Regionalism: A Countiy PerspectivcInj. de Melo and A. Panagariya, eds, NewDirsioinRegionalI7tgra- don Cambridge University Fress. Melo, J. de, C Montenegro, and A. Panagariya. 1993. L'integration rigionalc hier et aujourd'huit"RPs dEconomic du dve%eoppmzenz 2: 748. Nogues,J., andR Quintanilla. 1993. LatinAmerica'sIntegration and the Multilateral Trading System." InJ. de Melo and A. Panagauiya, eds., NewDinmnnsions in Rgional Integration. Cambridge UnJrivcrsity Press. Reyes Chavez, Mazio. 1988. La cooperaci6n econ6mica bilateral en America Latina y su relaci6n e influencia en la consecuci6n del proceso de integracion regional. Inzegradan LatimanaMeana-: 136-37. Summers, R., andA. Heston. 1991. 'he Penn World Tables (MarkV):An Expanded Set of Real ProductPrices and Price Levels: Estimates for 130 Countries." Quarterly Journal ofEconomics (May): 327-68. Vimer, J. 1950. The Cusoms Union IAune New York: Camegie Endowment for Intema- tional Peace. White, H. 1981. Consequences and Detection of Misspecified Nonlinear Regression Modcls."JournaleftheAmerican StatistiLclAssodation76: 419-33. Wonnacott, R, and P. Wonnacot. 1981. Is Unilateral Tariff Reduction Preferable to - a Customs Union? The Curious Case of the Missing Foreign Tariffs." Ameican Ec- nomicReview 71: 704-14. The labor market and trade reform in manufacturing Martin Rama The labor market and trade reform are related in two ways. Sluggish adjust- ment of the labor market to changes in product market conditions could slow the movement of labor from import-competing activities to export sec- tors Also, the rents created as a result ofbarriers to foreign competition may provide an incentive forworkers to unionize and push up wages. That means that trade barriers are a potential source of labor market distortions and that these very distortions make it more difficult to dismantle trade barriers. The relationship between trade barriers and the labor market in Jru- guay is an interesting case for many reasons. On the policy side, Uruguay represents one of the most extreme examples of an import-substitution strat- egy. Despite the small size of its domestic market (fewer than 3 million people), Uruguayapplied the whole gamut of protectionistpolices for more than 40 years. Concurrently, a strong labor movement had been developing since the middle of this century. Trade unions were severely repressed dur- ing 1973-84 by a harsh dictatorship. Following the recovery of democracy the labor movementreboundedstrongly in 1985- And concurrentlywith both these periods there has been a gradual dismantling of trade barriers, begin- ning in the mid-1970s. These dramatic changes in economic policies, polidi- cal regime, and the treatment of labor unions provide good experimental conditions for studying the effects of trade reform on labor allocation under both competitive and noncompetitive labor market conditions. The analysis begins with an exploration of the main features of the Uru- guayan labor market in the manufacturing sector, highlighting the prevail- ing institutional arrangements and wage setting procedures in force since democratic recovery. It also measures union strength through sectoral mem- bership rates at two points in time. Examination of the rent-sharing issue shows that Uruguayarz unions played almost no role in the rent-seeking pro- cess leading to the establishment of trade barriers. Yet they are claimants to these rents, as suggested by the significant relationship found between I08 The lainrmar*et and frade refonn tn manufactuing 109 union membership rates and variables reflecting market power at the sec- toral level (the concentration index and the effective protection rate). Mea- surement of the effects on sectoral employment and real wages of the lib- eralization process carried out from 1978 to 1986 shows that sectoral real wages were not sensitive to changes in protection, whether under dictator- ship or democracy. But sectoral employment adjusted significantly when there were no unions. Major features of the Uruguayan labor market Of the 1.3 million people in the Uruguayan labor force, about three-quar- ters work in the private sector, the rest in the public sector. Most public sec- tor employees belong to the centrl administration or to local governments. Only a fourth of them work in one of the dozen smta-owned firms. Many of these firms are commercial banks; others are natural monopolies or have been granted legal monopolies. Although informal sector activities are less important in Uruguay than in other Latin American countries, they expanded greatly during the 1 980s, accounting for 15 to 30 percent ofMontevideo's total employment (Diez de Medina and Gerstenfeld 1986; Aguirre and Mendez 1988, estimated from household survey data). A large share of infonnal activities is related to smug- gling of consumpton goods from Argentina and BraziL About two-thirds of urban jobs are in services. Since these are mainly nontradables, this large share ofservicesjobs reflects the low degree of open- ness oftheUruguayan economy. The manufacwring sector provides approxi- matelyone-fifth ofurbanjobs, mostofthem associatedwith highly protected actiities (figure 6.1). For instance, in 1985 less than ahalfofindustrialjobs were in sectors with an effective protection rate below 40 percent (estimated according to Corden).1 And about 10 percent ofjobs were in sectors with an effective protection rate above 150 percent Patterns were even worse 4 years earlier, when only one-fifth of employment was in sectors with protection rates below 40 percent and almost half of industrial jobs benefited from protection rates above 150 percent. This change suggests that trade reform had significant effects on labor allocation, at least until 1985. Inst iutuionalftramework Almost two-thirds of private sector workers are wage earners, so institutional arrngements affecting wage setting play a crucial role. These arrangements changed dramatically over the last decades. In 1968, within the context of a stabilization program, wage increases for all workers were to be decided by the government. This policy was reinforced in 1973 by the military govern- 110 EFFECrS OF PRomaEcboNLsM ON A SMALL COUNTRY: THE CASE OF URUGUAY Figure 6.1 Protecton and industrial employment, 1981 and 1985 Enployment in mantbcturing (percent) 100 90I 80 70 60 18 50 40 30 20 I0 -40 -20 0 20 40 60 80 100 120 140 Efectdve proeccion rote (percent) Sourr Based on dalafrom Centro de Investigaciones Econ6mzics (CINVE) and Direccidn General de Estadfstia y Censos tDGEC), with efffective protection rates estimarted accordingto Corden (1971). ment's ban on trade unions. By the late 1970s, however, the government had reduced its interference in wage setting to the establishment of minimum wages, and by 1985 the wage-bargaining mechanism in use from 1943 to 1968 (the heyday of the labor movement) was fully restored. Wage councils (cons4os de salarios) are the keystone of this mechanism. Delegatesfrom firms, trade unions, and the governmentsetminimumwages by sector and labor category. Until 19S J wage levels were adjusted every 4 months. From then on, they were adjusted whenever cumulative inflation exceeded some preset threshold. Adjustments are allowed even if the wage councils do not meet, provided that expllcitagreementwas reached in former rounds on the algorithm to be used. Such long agreements usually last 16 to 24 months. In May 1985 forty-eight activity groups were created to negotiate wage agreements One of them has never met, and the others continued to split up into more and more groups. Today, there are more than two hundred wage councils, organized by sectors or by large firms. They cover from less than one hundred to many thousand workers. Government delegates to a wage council are not just referees in the bargaining process. Their consent is required to validate any agreement, thereby extending it to the whole sector covered by the wage council. With- out it, the terms of the contrctare binding onlyfor the firms andtrade unions directly involved in the negotiation. In practice, validation has been condi- tional on wage increases in line with official inflation targets.2 The labor market and trade refom in manufachuing 111 A furious outbreak of strikes in the private sector accompanied the revival ofwage bargaining: newspapers reported some one thousand strikes in 1985- 86 (Filgueira 1990). The intensity of these conflicts is usually explained as a manifestation of the accumulated social demands that went unmet during the military regime. More likely, however, it reflected a learning process by trade unions concerning the true location of the labor-demand schedule. The wages of only about 350,000 workers are set by direct negotiation throughwage councils, so notall activities are covered. Wage bargainingapplies primarily to manufacturing. For labor categorits not covered, such as rral workers orhousekeepers, and for cases in which no agreement is reached by dic wage council; the government-set minimum wag: or wage increase applies.' Labor contracts in the public sector have teatures that distinguish them from others in the economy. Once civii servants have tenure (usually, after 6 months) they cannot be dismissed except for manifest incompetence or the commission of misdemeanors. Even then, the Senate must consent to the dismissal Workers in state-owned enterprises can be dismissed for other reasons aswell, and Senate authorization is not required. In practice, however, dismissing aworker is almost as difficult as it is in the central administration. Trade unions Trade unions or federations are not organized by labor categories or skills, bursimply by sector. Thus, blue-collar workers and foremen (and sometimes even managers) belong to the same guild. Union membership rates remained almost unchanged in the public sector from 1985-87,just after the demo- cratic recovery, to 1990 atLan average level of 58 percent (table 6.1). Nor has there been much change in the number of public sector strikes (or their harshness) since the return of democracy (Filgucira 1990). In the private sector, however, membership rates fell, dropping from an average of 54 percent of wage and salary workers in 1985-87 to 47 percent in 1990 (from 39 percent to 34percentas ashare of all personnel). The decline mayreflect some initial overshooting of labor union affiTiarions following the return of democracy or even the weak auditing of declared membership before 1987. The 1990 figures probably provide a more reliable picture of the actual strength of the labor movement Wage setting in manufacturing -Higher union wages may result from the capturing of rents that arise when foreign competition is blunted by tradebarriers. Did unionswork to gettrade barriers erected or did they simply work to gain the rents after barriers were already up? 112 EFFwErs OF PROTECTIONIS4 ON A SMAU. CouNTRiY THE C4SE OF URUGUAY Table 6.1 Union membership, 198547 and 1990 A a shore ofwge As a share c( and saolry wvkws ag personned kxdtwyckss4ctun Industry 1985-87 1990 J985-87 1990 I Agriculture 5.9 4.1 3.4 2.3 31.11 Meat products 60.3 42.1 38.0 26.5 3112 Dairy products 71.0 55.6 66.7 52.1 3113+3119 Preservesandcandies 35.3 34.6 26.7 28.7 3114 Fish 47.5 49.2 40.7 42.4 3115 Edible oils 75.7 67.9 57.0 50.8 3115+3122 Grain mill products 108.4 112.5 61.8 64.9 3117 Bakeryproducts 14.7 .. 7.4 3118 Sugar 81 c . 69.9 313 Beverages* 787 109.4 57.1 79.3 314 Tobacco manufacturers 70.0 68.9 41.2 40.2 321 Texile products 51.9 49.3 31.9 30.3 322 Apparel 93.6 21.8 45.7 10.7 3231 Leathertanning 473 63.0 46.4 61.8 3233+324 Footwear. leather products 60.4 45.4 35.1 26.7 341 Paper and allied products 56.6 31.4 54.9 30.5 342 Printing and publishing 38.5 32.9 21.8 18.6 3522 Drugs 59.9 55.8 49.1 45.8 351-+352-3522 Otherchiemicals 423 371 34.8 303 355 Rubber products 74.9 53.4 70.4 49.4 362 Gass 108.9 .. 106.4 381+382+383 Metal and machinery 73.4 633 41.0 353 4 ULRdfes (public) 75.4 78.8 75.4 78.8 5 Construction 32.7 232 22.2 15.8 6 Trade 12.4 8.1 -9.2 6.0 7 Transport and conmmunications (public) 70.1 72.1 70.1 72.1 Trasport and commnunications (private) 34.8 343 28.8 29.0 8 Fnance and insurance 43.8 41.9 27.0 25.8 9 Central govnmment 26.6 28.2 26.6 28.2 Local government 59.9 53.2 59.9 53.2 Other services (private) 26.9 26.1 18.0 1 7.2 Notavmalable. ftte: Because the polit weight ofa unicn in the abor movment depends on its number of aRfliates, these membersip figures may be overes6mated-as rates over 1 00 percent strongly suggesL a. IndLides one or rnore tate-owned finns Source: Based on data frm kireci6n General de Esadkdca y Censos (DGEC Oina de Plrearieo y Presupueszo (OPP). and Rienar Inotersindcal de Trabajadores-Convend6n Nacional de Trbaladores (Prr-CNT). Rent seeking by organized labor The first question, then, is whether trade unions participated in the rent- seeking process underlying protective measures. Uruguayan economic his- tory suggests that the answer is no (Rama 1991). Inward orientation intensi- fied in the l930s and early l940s. The general tariffrose from 31 to 48 percent The labor market and trade reform in manufaauning 113 in 1931, and multiple exchange rates were established in 1933-34 and im- portlicensesin 1938. In 1941 a central bureauwasgiven authority to allocate foreign exchange, approve import licenses, and ban imports that competed with domestic production. Organized labor was not a relevant social partner during that period. Its strength had been dramatically reduced by conflicts between Marxists and Anarchists at the beginning of the century and later by the Great Depres- sion. The initiative to create the wage councils in 1943 did not come from trade unions but from a parliamentary committee charged with investigat- ing the living conditions of the working class. More direct evidence of labor's lack of involvement is available from Rana's (1993) surveys of privately promoted foreign trade regulations in Uruguay from 1925 to 198N. Roughly half of the four thousand or so laws, decrees, and resolutions creaung or modifying trade barriers for the benefit of a few firms or sectors explicitly idendfyr their petitioners. Of these, only two include a trade union among their sponsors. In October 1947 a lumber firm and its corresponding trade union got a change in the import tariff on planks and in March 1949 the leather guild and the sectoral trade union got an export subsidy for tanned hides and footwear. In all other cases, rent seeking was carried out by firms only. Rnt shadng by organized labor Although Uruguayan trade unions did not play a direct role in petitioning for trade barriers, they could have claimed some of the rents created by such barriers. When firms face harsh competition, there is little scope for wage increase Trade unions are better placed in situations wbere firms enjoy significant market power. Unions are thus expected to be stronger in sectors in which competitors are few. One way to test the rent-sharing hypothesis is to check whether union membership rates are higher when there are barriers to competition in the corresponding product markets.4 (Table 6.2 reports 1990 rates only, since no significantregression results can be obtained from 1987figures, confirm- ing their unreliability.) Monopoly power in the product market is capmured by the sectoral employment share of the four largest firms in each sector in 1987, measured at the four-digit industrial classification level (table 6.2) . That means that figures reported for aggregated sectors at the three-digit level (apparel, for instance) are in fact employment-weighted averages of the dis- aggregatedshares. In a small country, concentration is agood index ofmarket power only if there is little competition from abroad, and so the concentra- tion index is usedjointly with the corresponding effective protection rate 114 EFFEcrs OF PaOTCON SM ON A SMALL CouNTRr: THE CASE OF URRUGUAY Table 6.2 Union membership and the rent-sharing hypothesis (percent) Shore ofwvge and Share of fotr Effecive sakzywir.ers Abrgestfirms in protection Industry oronized in unions ernpfowrmenr rates b cassifiotvn kidustry (1990) (1987) (1985) 3111 Meatproducts 42.1 295 - 3112 Dairy products 55.6 753 -3.6 3113+3119 Preserves and candies 34.6 49.4 17.9 3114 Fish 49.2 48.9 -1.4 3115 Edible oils 67.9 57.9 1383 3117 Bakery products 14.7c 7.2 3.5 3118 Sugar 81.5c 100.0 3103 314 Tobacco manufcrers 68.9 55.2 4.5 321 Textile products 49.3 295 25.1 322 Apparel 21.8 10.1 0.3 3231 Leathertanning 63.0 46.5 -1.7 3233+324 Footwear, leather products 45.4 30.4 -30.9 341 Paper and allied products 31.4 60.8 27.8 342 Printing and publishing 32.9 20.1 3522 Drugs 55.8 17.5 351+352-3522 Other chemicals 37.2 51.8 513 355 Rubber products 53.4 72.8 53.7 381 +382+383 Metal and machinery 63.3 24.0 167.1 N Notaaitable. Note: Sectorswih memrbership rates that exceed IWU percentare dropped from the analysis. a. Atthe fburd gt level. For aggregte industries, reportd fres are empment-weighted averages. b. Corden method (1971). c Last"baiblefgures are for 1985-87. S'urce Table 6.1 and dta from Centro de Investigaiones Econ6rnicas (CINVE) and Direcr6n General de Estadisticay Cens (GEC). (var-iables are measured at different points in time because of uneven data availability) Regression results are reported for both linear and log-linear speci- iications (table 6.3). The coefficient for the employment share of the four largest firms in -a sector is statistically significant in all cases, with an elas- ticity of union membership rates with respect to this share of approximately 0.4. Higher protection rate-s also lead to higher union membership rates, but elasticity rates are hard to assess because of the high variance of the estimates. Together, the two variables chosen as indicators of imperfect competi- tion account for roughly half the variation in union membership rates. That does not mean that trade liberalization would immediately bring union membership rates down. For one thing, the market power of domestic firms would not vanish immediately.6 For another, once the costs of organizing unions and setting up bargaining procedures have been incurred, such. ar- rangements are likely to remain in force. The labor market and trade rejon in manufacturing 115 Table 6.3 Regression results: explaining union strength Linecr model Lag-&near model Speciricadon Spedficoaon Specification Specification Specification Specifation Variable I 2 3 1 2 3 Constant 0.296 0.438 0.295 -0.368 -0.910 -0.414 (4.217) (8.867) (3.545) (-2.733) (-6.421) (-2.500) Concentration index, 0.426 0.352 0,428 0.451 (3.032) (2.020) (3.915) (3.728) Elfecive protection rate 0.120 0.072 0.437 0.196 (2.520) (1.469) (1.712) (1.038) R2 0.325 0.292 0.436 0.457 0.129 0.580 Note.: Number. in parentheses are tstaistics. a. Ermployment share ofaour largest Oirmns in a sector. Source: Ordnaay least square estimates based on data in table 6.2. The log of one plus the effective protection rate was wed as the exogenous variable. Product markets and labor dermand If, as the preceding discussion suggests, workers share the rents created by protection and the level of protection has no effect on union membership in the short run, what are the likely effects of trade liberalization? A prelimi- nary answer can be provided using a simple analytical model that leads to testable predictions on the relationship between protection and sectoral employment and sectoral real wages. There are n firms in an industry that compete with each other by pro- ducing differentiated goods. The firms or industries also compete with im- ports, whose price expressed in domestic currency is normalized to one. The tariff-inclusive price of the imported good is (1 + T). - From the standard monopolistic competition model, the demand sched- ule faced by a firm or industry i (with i = 1, ..., n) can be written as follows: (6.1) q1 = a, Y(J/V) - O < ai. < 1, cr > I where q1 is output of firm or industry i, P is price, Y is aggregate demand for the nt + I goods in the model and is expected to be positively correlated to economic activity, and a is the price elasticity of demand (the lower a the greater the monopoly power of firm or industry i). The average price of the imported good and its corresponding domestic substitutes is given by V: (6.2) V [ Aaiu. + T)i- Ti + t= 1 i=l - ~~~~~~~~~~i=1 116 EFrECrs OF PROTECTIONISM ON A SMALL COUNTRY: THE CASE OF URUGUAY where .L increases with individuals' preference for imported goods (a, is as- sumed to be small, so that dV/dP = 0). For simplicity, suppose that domestic goods are produced with labor only (L;), under constant returns to scale: (6.3) where [,B is the labor coefficient in firm or industry i. The first-order condi- tion for profit maximization implies: (6.4) Pi = 5U cf W where W1 is the nominal wage paid by the firm. The price of the good pro- duced by firm i is thus set by applying a constant markun ratio over wage costs. The analytical expression of the labor demand schedule can be drawn from equations 6.1, 6.2, and 6.4: (6.5) Lj=A~Y(-7) w-Y, A1=acj317 (--), ,wj=W/P. (6.5) ~ j S a In this expression, P is the general price index for all goods, notjust the nt + 1 goods in the model. Hence, w; is the relevant real wage for individuals' decisions. The V/P ratio is higher (other things being equal) the higher the tariff rate T. The demand for labor in firm or industry i increases with aggregate demand Yand with trade barriers T, and decreases with the real wage wi. The elasticity (a) with respect to the realwage is the price elasticity of demand in the product market. By assumption, this parameter does not depend on the tariff level. In the short run, then, trade liberalization in this sectorwould shift the labor demand schedule to the left, without modifying its shape (figure 6.2). The role of trade unions The effects of trade liberalization depend on whether the labor market is competitive. Assume that individuals have no risk aversion. The utility Uof an individual working in firm or industry i can be written as: (6.6) U= rw, with parameter r depending on preferences. If the individual is laid off, utility becomes: The lahar ma,*et and trade reform in nanufSacing 117 (617) U= r-r(Y, L,, T), dr/dY > 0, dr/dL, > 0, dr-/dTc 0 where rf Y,L, 7) represents either the wage level in the competitive sectors of the economy or the real income level associatedwith domestic activities. This reservation wage is a positive functicri of aggregate demand Y. In addition, itincreaseswith employment in firm or industry ibecause of imperfect sub- stitution between specific labor skills. Finally, it is a negative function of tar- iffs, under the asstumption that trade barriers reduce economic efficiency. In acompetitive labor market, firm orindustry ipays the reservation wage: (6.8) zu. = r. When equation 6.8 holds, trade liberalization leads to a fall in employment, but its effect on the real wage level is ambiguous. A lower demand for labor in firm or industry i reduces the reservation wage for the corresponding skills, while gains from trade shift the r(Y, Lj, 7) schedule up. The marketequilib- rium moves from point C4 to 4 (figure 6.2). That labor demand is not infmitely elastic creates an incentive forwork- ers to raise wages. The simplest analydcal tool for dealing with this issue is the monopolyunion model, inwhich the trade union sets the wage level and the finm or industry the employment level (see Oswald 1985). This condi- dion corresponds to a Stackelberg equilibrium in which the union, as leader, takes into account the reaction function of the firm or industry, represented by its labor demand schedule. Such asymmetric power seems plausible in a country with as strong a labor movement as Uruguay's. Figure 6.2 Trade Uiberalization and the labor market Reet wvige i Union wage A ~~~RI Reservadon wage I ~~~C2 = = 5 = | ~~~~~~~Labordemnd b I~ oo 118 Emas OF PROECIONIS ON A SmAr. CouNTrr THE CAsE OF URUGUAY Assume that the trade union in finn or industry i has M. members and thatM,does notdepend on trade barriers T in the shortrn. The trade union in firr or industry i sets the nominal wage level W (and thus the real wage level w,.) that maximizes the expected utility of its representative member: (6.9) AX Li rw, + (1 - )rr sulbject to equations 6.5 and 6.7. This yields: (6.10) Li where e represents the elasticity of the reservation wage r to employment L. Note that for e = 0, the wage level set by the trade union is a constant markup over r, with the markup ratio being the same as thatused by firm or industry i to set the price P. (see equation 6.4). But in the general case the markup ratio increases with L, and so does the wedge between the union wage and the reservation wage. Thus, in a unionized fiwm or industry trade liberalization shifts the labor market equilibrium from R, to R2 (figure 6.2). As in the competitive labor market case employment falls and the effect on real wages is ambiguous. But the percentage drop in employment is smaller, and the real wage falls more- or rises less-than in the case of competitive labor markets. Two caveats apply to this analysis. First, different assumptions about the wage-bargaining mechanism could have resulted in a different impact of trade reform. For instance, de Melo and Tarr (1990) consider a model in which bargaining involves both wages and employment, so that the outcome is no longer on the labor demand curve. Grossman (1984) assumes a seniority layoff rule, implying that senior workers have a disproportionate influence on union decisions and, therefore, that wages respond sluggishly to trade reform. Second, the results hold only in the short run. In the long run foreign competition in the product market should increase the price elasticity of the demand for goods (4D,+ 2 log (1 + Tg) + 4.D, log (1 + 7) 4 log Y, (6.14) log wi, =Oj +04D+ + 0 log (1 + Tt) + O5D, log (1 + T,) + 03 log Y, where D; is a dummy variable that takes a value of 1 in 1985 and 1986, and 0 otherwise. When workers are organized in unions, the elasticity of employ- ment with respect to the tariff-inclusive price of imports is 02+ 45 (instead of 4; in the competitive labor marker case); the corresponcding elasti-ityfor real wages is 02 + 05 (instead of 0-). According to the model above, 45 should be negative and 65 positive. Estmates for 4o ' 8, and ishow that neither 6(0.048 or -0.069) nor 05 (-0.021 or-0.066) isstatisticallydifferentfromzero,indicatingthattrade liberalization did not affrct real wages whether the labor market was com- petitive or not (table 6.4, columns 2,3, 5, and 6). The real wage levelwas not the same inboth cases, however. Coefficient64 is positive (notshown in table- 6.4) and significant at the 10 percent level when implicit protection rates are used.1 Real wages are higher when workers are organized in unions. The elastidty of employment to the mriff-indusive price of imports is posi- tive in the period with labor market competition (42 of 0.236 or 0.441),12 but ]lUs after workers unionize, as figure 6.2 suggests itwoulci Coefficient 65 is al- ways negative (4.186 or -0.443) and is significantly different from zero when tariffs are measured by implict protection rates. That means that in 1985-86 -trade liberalization became a less effective tool for achieving labor relocation. Is this result due to unionization only, or are there other factors behind it as well? There were no changes in legal dismissal procedures following the democratic recovery, so the negative sign of coefficient O, does not arise from higher fiing costs. Uncertainty in the first years of democracy was high, however, and firms may have been reluctant to adjust to changes in protec- tion if they believed them to be temporaxy. Conclusion Trade barriers may strengthen the market power of domestic firns and cre- ate rents. Since wages can be raised by capturing rents, protection is an in- centive to unionization. There is some evidence of a relationship between 122 EiFte rsO PROTECTIONISM ON A SiuAL Couwrnt THE CAsE oF Uxuouay union membership rates and variables measuring a firm's markCEr power, including a concentration index and the effective protection rate. These correlations are observed in cross-sectional data, however, and dynami'c effects may differ. Although trade liberalization should weaken la- bor unions in the long run, wage-bargaining procedures might remain in force for some time. If labor market imperfections outlast the product mar- ket imper-fections that gave rise to them, there may be problems during the transition if trade liberalization-proceeds too quickly. The short-run effects of trade refor-m should not be exaggerated, though. The scctoral elasticity of employment to protection rates is much lower when workers are organized in unions than when labor markets are competitive. This result suggests that labor market imperfections may reduce the short-mun effectiveness of trade policy as an instrument for improving labor allocation. Notes 1. The effective protection rare measures the gap between value added at domes- tic prices and at international prices. The rate increases with trade barriers on output and decreases with tr-ade barriers on inputs. 2. In fhact, the threat of nonvalidation has operated as an incentive to continue to divide into activity groups until e-ach group includes homogeneous firms, in the sense that they can all afford the- same wage level. 3. Until 1990 all private sector wages were adjusted at the same time. Since then the setting of minimum wages (or wage increases) by the govermecnt has not necessar- ily been synchronized with the adjustments decided by wage councils 4. The rent-sharing hypothesis is tested by Rose (1987) for the United States, by Stewart (1990) for the United Kingdom, and by Christofides and Oswald (1992) for and.The studies use wages as the endogenous variable, taking unionizaton asgiven. 5. No statistically significant coefficients were obtained when using other indica- tors of sectoral openness, such as Krueger's ratio (the ratio betwe-en the foreign-trde deficit of a specific sector and its correspontding domestic demand), the share of ex- ports over total sales, or the share of imports over domestic demand. 6. A CMNVE report (1987) cites the example of firms in Uruiguay that became ini- porters after trade liberalization and kept using the same trade network. 7. Estimates of nominal protection rates are available for thirty-nine sectors for 1978,1980,1981,1982,1985, and 1986. Ikwould have been better to use information on effective protection rates, but estimates are available at the four-digit. level for 1981 and 1985 only. Even for nominal protection rates, not all required information is avail- able for all sectors and years. Thus, sonicobservations have to be set aside in the analysis. 8.. In econometric terms, allowing sector-specific intercepts corresponds to the covariance model specification, which is a simnple way to deal with panel data (cross- section for sectors, time series for years).- 9. Both indicators differed in the first stages of tr-ade liberalization, when redun- dant protection was still high (Rama 1982). 10. The figure r-eported byI Freeman and Katz (1991) is actually 0.545 in the case ofemploymcnL However, this estimate is obtained for a given real wage. Since employ- The labor mar*u and trade refom in manufactuing 123 ment decreases with real wages, and the latter decrease with import penetration, the reported figure overestimates the emnployment impact of foreign competition. 11. It is even signifiCant at the 5 percent level under the restriction 02 = 05= 0. Estimates for 04 are 0.062 or 0.050, depcnding on the chosen protection indiator. 12. This is in accordance with the significant labor relocation depicted in figure 6. 1. References Aguirre, R., and E. MEndez 1988. "El trabajo informal urbano en Uruguayr una aproximaci6n a travEs del analisis de las encuestas de Hogares. SUMA 3(4): 89-103. Christofides, L., and A. Oswald. 1992. 'Real Wage Determination and Rent-Sharing in Collective BargainingAgreements.' OwanerlyJounaiofEconmics (August): 985-1002. CINVE (Centro de Investigaciones Econ6micas). 1987. "La industria fiente a la competencia extranjera CINVE-Ediciones dc la Banda Oriental, Montevideo. Corden, M. 1971. The Thery ofProtection Oxfordc Clarendon Press. Diez de Medina, R., and P. Gerstenfeld. 1986. "Sector informal urbano: marco te6rico, cuantificad6n ypropuesta de medici6n para el caso Uruguayo. Direcci6n General de Estadistica y Censos, Montevideo. Filgueira, C 1990. 'Organizaciones sindicales y cmpresariales ante las politicas de estabilizaci6n: Uruguay 1985-7." In Estabitacidny Respusta So&ac International .Labour Organisation, Santiago de Chile. Freeman, R., and L Katz 1991. Industrial Wage and Employment Determination in an Open Economy." InJ. Abowd and R. Freeman, eds., Immigration, Trade and the LaborMaTiet University of Chicago Press. Grossman, G. 1984. "International Competition and the Unionized Sector.' Canadian Journad of Econmics 17(3): 541-56. Macadar, L. 1988. "Protecci6n,ventajas comparadasy cficiendaindustial."SL hA3(5): 7-59. de Melo,J., and D. Tarr. 1990. 'Industrial Policy in the Presence of Wage Distortions The Case of the U.S Auto and Steel Industries." Center for Economic Policy Re- search Working Paper 435. London. Oswald, A. 1985. "he Economic Theory of Trade Unions: An Introductory Survey." Scandiumianfourrau ofEconomics 87(2): 160-93. Rana, M. 1982. "Protecci6n ycrecimiento industrial.' ClNV-Ediciones de la Banda Oricntal, Montevideo. .1991. "Economic Growth and Stagnation in Uruguay." In AL Blomstrom and P. Meller, eds., DiveingPaths: CompainWga Cenu7ay of&Sandinavian and Latin Ameri- can EonomicDevkopmenL Baltimore, Mdz:Johns Hopkins University Press. 1993. "Rent-Seeking Trade Policy: A Time-Series Approach." Policy Research Working Paper 1142. World Bank, Policy Research Department, Washington, D.C. Revenga, A. 1992 'Exporting Jobs? The Impact of Import Competition on Employ- ment and Wages in U.S. Manufacturing." QyarteriJournal of Economis 107(1): 255-8t Rose, N. 1987. 'Labor Rent Sharing and Regulation: Evidence from the Trucking In- dusty."Journal ofPolitcalEconomy 95(6): 1146-78. Stewart, M. 1990. "Union Wage Differentials, ProductMarketInfluences and the Divi- sion of Rents." EconomicJournal l0f(403): 1122-37. Economyzwide costs ofprotection and labor market rgdities Jaime de Melo and David Roland-Holst Preceding chapters have documented the extensive distortionsin Uruguay's goods and labor markets. Evidence was presented of rent-seeking activities and their associated costs in terms of lower growth rates in sectors that re- ceived disproportionately large amounts of discretionary protection (chap- ter 2).- Reference and minimum export pnce systems were shown to have boosted protection by some 7 to 18 percent (chapter 3). The high costs of protective measures in the automobile industrywere shown to generate sub- stantial rents (chapter 4). And although no direct evidence of trade diver- sion was detected from the operation of the bilateral trading arrangements 'with Argentina and Brazil, the coupon quota system was suspected of lead- ing to inefficient specialization patterns with Brazil and especially with Ar- gentina (chapter 5) . Uruguay's labor market in manufacuingwas shown to be dominated by strong labor unions (chapter 6) Wie these manyinterventions in product and labormarkets are indica- tive of the distortions in resource allocaton, th%re are still others that have not been analyzed. The prohibition against exporting catde on the hoof providedprotection to the meat-processingand leatherindustriesin theform of lower prices. Interest rate ceilings led to extensive credit rationing. And even after the progressive decontrol of interest rates that accompanied the financial sectorreforms begun in 1974 (see the chronology ofreforms in the appendix to chapter 2), remaining con trols permitted preferential access to credit for selected activities, particularly for public sector enterprises. The complexity and pervasiveness of such interventions suggest that it is appropriate to investigate their economywide inplications in a general equi- librium setting by quantifying the likely welfare losses due to the combined influence of interventions in the labor market and in foreign trade. The estimates are derived from a static computabIe general equilibrium (CGE) model, using input-output data for thirty sectors for 1983, the most recent year with reliable data. 124 Bea nosquide casts ofpprfecsion and labor iwn*igidizes 125 Not all the effects of trade and labor market interventions are incorpo- rated in these estimates. Trying to account for all of them would be a Herculean task under any circumstances, especally for an intervention- fiddled economylike Uruguay's. The decision aboutwhat effects to explore was therefore guided by both the availability of data and the desire to illus- tiate how an improved allocation of resources gets tanslated into higher welfare. Thatled to aconcentration on the effects oftarifis andadministered protection in foreign trade and on wage premiums in the labor markeL Also, because the estimates are forarelativelyfine disaggregation of the economy, they capture the effects of differences in price wedges across sectors. In this stepwise approach to estimating the costs of protection and labor marketrigidities, the first step is to estimate the likelycosts ofindividual trade interventions, assuming no distortions in the labor marketand perfect com- petition and constant returns to scale throughout the economy. The next step considers the implications of labor market distortions, taldng existing trade interventions as given. Finally, the welfare costs of protectioia and la- bor market rigidities are estimated, allowing for increasing returns to scale and noncompetitive behavior in manufacturing. The estimatesfind thatthe costs ofadministeredprotectionfaroutweigh the costs of tariffprotection. The costs of tariff protection are small because weighted arrage nominal protection is about 10 percent and the import share in GDP is 12 percent But administered pro tecti n exacts a high price: trade and transport margins are four times higherfor imports than fot 'simi- lar" domestic goods. The general eqailibrium estiaates indicate that elimi- natine this extraprice wedge wouldadd some tc 4percent to base GDP while removing tariff protection alone would add only 0.1 percent to base GDP. The costs of administered protection double if the value of the rents gener- ated is dissipated in unproducive activities. The simulation analysis suggests thatthe resource allocation gainsfrom eliminating wage premiums would be relatively small (tiough twice as large as t he gains from removing tariffpro tection alone) because for tradables the premiums are concentrated in sectors with low employment levels, while for nontradables the scope for labor relocation is quite limited. When econo- mies of scale are introduced for selected manufacturing sectors, the gains from removing tariffs and administered protection are about 50 percent higher than under coastant returns, as resources move from sectors operat- ing under constantretumrns to those operatingunderincreasingreturns to scale. Modeling the effects of Uruguay's market interventions Three sources of foreign trade distortions are explored: the price wedge resulting from import tariffs and the operation of the reference price system :126 Ewzcrs OF PROTECTIONISM ON A SMALL CouNTra THE CASE OF URUGUAY (administeredprotection), thepossibilityofpricingabovedomesticaverage costs when trade barriers limit entuy, and rent-seeking behavior. Consider the wedge between imports and the donmestically produced competing product The value of imported and domestic goodswas decom- nosed into producer value, tax, and margin components (see following sec- tion).1 In line with the evidence provided by Changanaqui and Messerlin in chapter 3, differences in margins are interpreted as reflections of the effects of nontriffbarriers. These effects are decomposedinto two effects: the higher tradeand transportmargin on imports and the resource costof rent-seeldng activity. The wedge resulting from the excess" trade and transport margin paid byimportsovercomparable domesticgoods (ew) forsectoriiscalculated as: (7.1) twJ= frm.-ttrd where tnnr and tnid are the margins on imports and domestically produced goods in sector i These margins appear as a source of final demand for tie r-ade and ransport sectors. A case can be made that the revenues accruing to the trade and transport sectors-notably those arising from the port ac- tivities in Montevideo-are rents. So the second effect of nontariffbanriers arises from the possibility of rent-seeling activities. Following the tradition of the literature on rent-seeking, we assume tat individuals will spend re- sources to obtain these rents and that this rent-seeking activity consttutes an additional cost to society. As an approximation of this resource waste we assume that the value of rents in sector iis given by: (7.2) Df= X(tm - ird)PMmM where PM. is the domestic currency price of imports and M. is the volume of imports. Rents, collected on the tariff-inclusive price of imports, are equal to the value of resources committed to the trade and transport sectors to ob- tain the rents. As in the literature, it is assumed that rent seeking uses re- sources. This is a strong assumption for which there is litde direct evidence. Az exception (coincidentally) is Rama's (1991) examination for Uruguay (discussedin chapter 2), which providesjustification for exploring this path- way of the effects of protection. The formulation follows Grais, de Melo, and Urata (1986) and assumes that the efficiency gain fromremovingnontariffbanriers can be capturedby augmenting the efficiency parameter, A, by the value of rents in the total value of output, X: (7.3) X-=,-(1+R; /PDjXj)Fj(Kj,L,) Eoanomywide cosafofp/tDecton and labor mari* iigidties 127 where PD, is the price of the domestically produced good and 15() is the production functionforsectoriŁAserviceable ifsimplistic formulation, equa- tion 7.3 imputes all costs of rent seeking to the trade and transport sectors and assumes that the 'technology' of rent seeking is similar to that of the regular producdton activity in the trade and tansport sectors. Before turning to labor market distortions, we briefly describe ourmod- eling of oligopolistic behavior. Increasing returns to scale are modeled by assuming constantmarginal costs and positive fixed costs (figure 7.1). Asim- plifying assumption in this modeling of oligopolistic behavior is that the response to a change in policy takes place through changes in the number of products rather than the number of firms.' The various forms of protec- tion are assumed to result in sufficient baniers to entry that domestic pro- ducers collude to obtain an above-normal profit rate, 'fj, on domestic sales. The domestic pricing rule under protection is then: (7.4) PDj= (1 + TW) AC, where A4. is average cost in sector Ł The resulting equilibriumwith protection is shown in figure 7.1. Output drops to X, and profits are equal to the shaded rectangular area It is as- sumed that the above-normal profit rate would not be possible without ad- ministered protection, so '. is set to zero when protection and labormarket interventions are relaxed. The costs to society are equal to areas 1 plus 2. The size of the costs depends on the extent of unexploited economies of Figure 7.1 Increasing returns to scale A*ke L PDj(1 ++) _SiL AC o xI Output .128: Eacrs OF NoNTEcnoNIsm ON A SMALL COUNTRY: THE CASE OF URUCUAY scale. This fornulation does not allow for rationalization effects to occur through changes in scale due to firm entry or exit.4 Distortions and rigidities in the labor market result from labor union activity or other labor market regulations that allow different wages to be paid to the same type of labor in different sectors. Two cases are considered in the modeling: wage differentials due to exogenous distortions and those due to (endogenous) labor union activity.5 The econometric estimates ofthe sectoral wage differentials, 4, (based on household survey data; see below), control for education and seniority and so can be interpreted as premiums rather than payments for differences in worker characteristics. - Th.ewelfare cost ofan exogenously causedwage differentialis illustrated in figure 7.2 for the case of a two-sector price-tking economy. Awage differ- ential in favor of sector R (manufacturing) results in too little employment in tha-. sector and too much in sector U (agriculture). If there are no other market Imperfections, the social costs of the rents accruing to labor in sector Rare equal to the private costs, that is, to area abc. But because the wage premiums are in manufacturing-the most pro- tectedsectorinUmuguay-removingprotectionwill exacerbate the social costs of distortion ir. the urban labor market by shifting the social demand curve for labor in for manufacturingand out foragriculture. This second-bestsitu- ation means that the outcome of removing trade distortions is ambiguous, so the effects of removing trde and labor market distortions are examined separately. - The second specification of wage differentials assumes them to be en- dogenous, following Rama's evidence in chapter 6 that wages in Uruguay Figure 7.2 Labor market distortions from exogenous interindustry wage differentials WFt SP MPI WU .~ ~~ LR Wa=Wo ,_ < 4 ' Lu .-';-' - - ' Ecenomywade costs of pmtcdon and laborma*el rigidities 129 are determinedlargelybylaborunions. Followingde Melo andTarr (1993), the labor union is assumed to have a generalized Stone-Geary linear expen- diture system udlity function:- (7.5) U= f(W4L) with the avenge wage W employment 4 and the wage premium x as argu- ments (sectoralsubscriptshave been omitted). The union eithersets the wage rate unilaterally or negotiates it with the finn, and the firm unilaterally de- terxnines the level ofemployment. The outcome is shown in figure 7.3, where the firmnchooses the profit-maximizing level of employment Lk along its la- bor demand curve, L = L(WO4, where p is the endogenously determined wage premium chosen by the union to maximize its utility at WI. It can be shown that: L -(7.6) *=F(a,8y): F1O where ais the elasticity of substitution between labor and capital, 6 and yare the weights atta&hed to employment and wages in the union's utility func- ton, and the F teons are the partial derivatives of the function determining the endogenous wage premium 4. Thus, the endogenous wage premium depends on the union's objective function and on the relative weights at- tached to wages and employment.6 In this formulation the labor union sets the wage, so the costs of labor market rigidity are endogenous. For example, a reduction in protection in Figure 7.3 Wage detennination with a labor union w ; Wug \ U = U z~~~~~~~~~=U \1 .\ u=~~~~~~u=u - L U- UI Lv= Labordemard LU L : - 130 Em.ars OF PROTECnONISm ON A SMALL COUNTRrY THE CASE OF URUGUAY a sector will result in a lefward shift in the labor demand curve for that sec- tor because of a shift away from the domestic substitute good, causing the union to choose a lower wage premium. (Calibration of the initial value for the wage differential is cxplained in the following section.) The remaining features of the model are standard for a computable general equilibrium model, except for the inclusion of trade margins as a proxy for the distortionary effects of administered protection.7 A represen- tative consumer maximizes utility and a representative produlcer maximizes profits, with atomistic behavior throughout (except for one case that allows for an above-normal profit rate in manufacturing sectors under increasing returns to scale). The economy's terms of trade are fixed by the small-coun try assumption, and capital and labor are in fixed supply. Equilibrium is reached in the usual manner, with a vector of relative product and factor prices that eliminates excess demands in product and factor markets. Assumptions about functional forms are given in figure 7.4, and the resulting social accounting matrix for 1983 to which the model was calibrated is given in appendix table 7A.1. Estimates of price distortions and wage premi Estimates of price distortions in product and factor markets, based on de- tailed input-outputaccounts of 1983 (BCBJ 1989), include the wedges intro- duced by tariffs, by trade and transportation margins, and by wage differen- dials (appendix table, 713.1). The average import-weighted tariff is 10 percent,' but the major wedge between the c.if. import price and the reta pnce is accounted for by landing costs and other import margins, which together avenge 35 percent of the markup on the c.i.f. import value. The 1983 input-output table forUruguay gives unusuallyspecifi.caccount- ing information on the tax and margin components of prices for domestic goods and comparable imports. Trade and transport margins are the usual payments for distribution services on domestic goods. For imports they are likely to include in addition not only landing costs but also any price effects of nontariff barr,iers. If the data are sufflciendly disaggregated to distinguish comparable goods at the sectoral level, the effects of nontariff barriers-the reference and minimumr export price system-will showup as the differences in margins on imports and comparable domestic goods. The microeconomic evidence provided by Changanaqui and Messerlin (chapter 3) supports the use of differences in trade and transport margins (columns 2 and 3 of ap pendix table 7BJ1) as proxies for the tw1 variable defined in equation 7.1. While in principle the transport component of price margins should be similar for imports and comparable domestic goods, that is not the case here. Ewrnwi&d cos fofedion and lkbor ma*vngidit s IS Figum 7 lasEisdky spedfication (a) AJocadon of production Domestic wuply Gross outpt CEr- I E~~~~~~~q~~Boft SUP*l Vai ade Idnterediate I '. CES Leontief cap es Labor CL A~~~~~~ I\ '-ES CES Domestic impoed Dorestic Imported i intern ediate - - - m t ie (b) Comnsuer demand Consumptio LES C1 e con s p : --- -- M2u (a) CES CES Domestic Imported Domestc Impted 5x nde MebadTarw(:992)taibe 3.1. 132 EFFECas OF PzOToCnONISM ON A SMALL COUNTRY: THE CASE OF URUGUAY One reason for the differences is the inefficiency of the port of Montevideo, which harndles most imported commodities. The port is unusually inefficient by intemational standards. According to World Bank estimates the port's cargo handling performance is about 60 percent of the average for a group of 220 international ports, and its hourly productivity is about 62 percent that of a control group of ports in developing countries. The National Port Authority has an antiquated and overgrown administrative structure, and most activities are under the strong control of labor unions. Its financial situation has deteriorated steadily since 1980, with rising costs, shrinking traffic, and declining productivity. At the same time, portuser costs have risen, both nomi- nally and in terms of waiting time, damage, and other losses. In the United States,. where wholesale margins are also found to be systematically higher on imports than on comparable domestic goods (Rousslang and To 1992), delivery lags and additional storage costs for imports account for a 9 to 11 percent price difference between domestic steel and physically identical imports (Jondrow, Chase, and Gamble 1982). But other factors may also account for differences in the value of mar- gins on imports and domestic goods. Distibution channels may be quite differentsince imports tend to go towholesalerswhereas domestic manufac- turers sell directly to retailers, performing the wholesale services themselves. In any event, whatever the other causes of the observed differences, they are unlikely to lead to differences as large as those found for Uruguay: on aver- age imports pay over four times more in trade and transport costs than do domestic sectors (appendix table 7B.1). For reference, consider again the United States, the only country forwhich detailed comparisons are available. The added protection attributable to higher wholesale and freight costs for imports isabout 15 percent (RousslangandTo 1992, table 7.3)-farless than the fourfold difference observed in Uruguay. This large difference between the two countries suggests that the inefficiency of the port of Montevideo cannot alone explain the large margins between importand domestic goods inUruguay and that administrtive importdeterrence, such as the reference price system, must play a substantial role. . Concentration ratios for the thirty manufacturing sectors average 44 percent, indicating that the largest four firms in a sector generally account for nearly half the employment in their sector. Though the figure varies from a low of 17 percent (bakery) to a high of 100 percent (sugar), the concentra- tion measures are fairly uniform and indicate that most sectors could be strongly influenced by the individual or collective decisions of a few fiwms, especially if there are barriers to entry for imports. These high concentra- tion ratios provide indirect support for the assumption that protection may have allowed above-normal profits in manufacturing. Economywide cats ofrpmtection and labor market rigidifies 133 Finally, the estimates of wage premiums, adapted from more disaggre- gated estimates by Bucheli (1992), were obtained econometrically from household survey data (appendix table 7B.1). Following the methodology of Krueger and Summers (1987), the premium estimates control for education and seniority, therebyjustifying the use of these premiums as a proxy for the extent of distortions in the labor marketL Costs of tariffs and administered protection To evaluate the welfare costs of trade restrictions and to get a rough idea of the relative costs of each form of protection, trade restrictions are decom- posed into tariffs and administered protection. To help in intexpreting the results of the experiments, the trade and production structure in the base solution is described in appendix table 7B.2. Uruguay's pattern of trade is typical for an economy with a strong agri- cultural base-agriculture contributes 13 percent to GDP and is the second largest employer. Exports are concentrated in activities that use agricultural inputs and in light manufactures, with imports concentrated in capital and skill-intensive activities. The estimated pattern of effective protection from tariffs alone (appen- dix table 7B.2) is quite different from the pattem of norninal tariff protec- dion (appendix table 71B.1) because of the effects of triffs on intermediate inputs in the determination of a sector's effective rate of protection. For exanple, dairyproductshave the second highestrate of effective protection, but only the fifth highest rate of nominal protection. Both dairy and bakery products sectors have a nominal protection rate of 19 percent, yet the effec- tve rate of protection for bakery products is less than half thatfor dairy prod- ucts, a divergence due to the much highcr valuc-added ratio in bakery prod- ucts. Of the two measures of protection, the pattern of effective rates of protection is the better indicator of the likely pattern of resource pulls that will resultfrom the elimination of tariffprotection since it measures the pro- tection to the factors employed in each activity. The first set of experiments (all experiments are cumulative) compares the welfare costs of tariff protection with those of administered protection: * B-1: Set all tariffs to zero. Measures the standard costs of protection. * E-2: E-1 plus equalize trade and transport margins-remove adminis- tered protection-between domestic and imported goods (let trm. = rd.). Adds the costs of administered protection due to higher trade and transport margins for importsm * E-3: E-2 plus assume that the value of rents imputed to administered protection is dissipated in rent-seeking activities (see equation 7.3). Z34 EFrns OF PROTECEONISM ON A SMALL CouNmrxv THE CASE OF URUGUAY Assumes that the value of rents accruing to the trade and transport sectors is committed to rent seeking and results in resource waste. Given the modeling assumptions about the costs of the various forms of pro- tection, experiment 1 represents a lower bound for the costs of protection and experiment 3 an upper bound. The aggregate effects on welfare are reported using the equivalentvaria- dion measure, which estimates the change in the representative consumer's income as a result of the policy changc (table 7.1).Y To simplify compari- sons, results are evaluated against the calibrated base simulation (appendix table 7B.2), using the pattcrn oftarffsand administered protection described in appendix table 7B.1. Thus, the equivalent variation-welfre measure is expressed as a percentage of income in the calibrated base simulation. To give an idea of the likely resistance of workers to policy changes, an aggre- gate measure of labor relocation is also estimated, expressed as the percent- age of the labor force that must relocate as a result of the policy change. The aggregate results for the welfare costs of tariff protection are very small (less than 0.1 percent of initial income). The cost is low because aver- age (import-weighted) nominal protection is only about 10 percent and the share of trade in GDP is about 12 percent. The estimated costs of adminis- tered protection are much higher, at more than 4 percent of base income, bringing tie total costs of protection to 418 percent of initial income. As noted earlier, imports pay an average offour times as much for distribution as similar domestic products (appendix table 7B3 1). Taking into account the resources dissipated to obtain the rents in sectors engaged in trading and transport (in the model these are the wholesale trade, transport and com- munication, and other services sectors) boosts the potential costs of protec- tion to 8.3 percent of GDP. This figure is a reasonable naximum consider- ing that the costs of administeredprotection are difficult to evaluate and that the total value of rents is probably not dissipated in rent-seeking activities. Estimates of the share of the labor force that would have to relocate in the long run as a result of each policy change help explain whyworkers resist Table 7.1 Aggregate effects of removing tariffs and administered protection aQeentage change frmn base income aid enplyrnent) Eweiment Chaxne in welbrme Labor rebcae E-I: Removetrifrprotecton 0.08 1.6 E-2: Remove tariff and administered protection 4.18 3.8 E-3: Remove tart and adminstered protection and cease rent-seeking acites 8.27 5.1 a. Measured by equ?valent variation. Wefare gin is expressed as a perenage of base GDP of 174.6 billion pesos b. Cornputed as hlf the sum of the absolutevale of ie eployentdiages in thousnds ef work-years and epressed as the percentge of workers that must rebate. Base employment is 1. 159 mDion worker-yean Sorce: Authcrs' calcladans. based on Bcu (1989) datL Ec-nomywdde cost of proecxon and labormarket ngugdes 135 efforts to dismantle protection. Between 1.6 and 5.1 percent of total employ- ment of 1.16 million workers would be affected, so removal of protection would result in as many as 50,000 workers having to relocate. The microeconomic results describe the sectoral resource shifts that would be induced by the removal of protection (appendix table 7B.3). The pattem of resource pullsatthe sectoral level is shown as the percentage change in output and employment from the calibrated base. Removing only ariffs causes the greatest contraction in sectors with the highest rate of protection. For example, footwearshrinks by 22.5 percent.'0 There is also asmall resource shift toward tradable sectors brought about by a real exchange rate depreca- tion of 3.9 percenL This shift toward tradables is more pronounced when thegap in distribution margins is eliminated byremovingadministeredpro- tection as well as tariffs, as the real exchange rate depreciates 13.5 percent The wholesale trade sector contracts by 11.8 percent and the transport and communications sector by8.7 percent. Agriculture expands 7.3 percent, more than twice as much as when only tariffs are eliminated. Finally, ifit is assumed that administered protection was accompanied by rent seeking, the results are magnified because fewer resources need be used-in trade and transport activties. Agriculture expands by 15 percent, and the contraction in the wholesale trade and transport and communication sectors is less than half that resulting from the removal of tariffs and administered protection with- out rent seeking. Te pattern of effective protection rates calculated using the standard pardal equilibriumformulaisnotagood predictorof the pattern ofresource pulls in a model with differentiated products. One reason is the generAl equilibrium effects operating through interindustry linkages. For examplt, the leather sector contracts more than footwear even though footwear has one of the highest rates of effective protection and leather one of the lowest (appendix table 7B.2). The reason, of course, is that the fate of the leather sector is linked to that of the footwear sector, a link that can be properly captured only in a general equilibrium setting. Thus, the estimated sectoral pattem ofcontraction and expansion is llustrative of the likelyresource shifts thatwould occur if protection were rem oved, butnot too much significance can be attached to the magnitudes themselves since the structure of tariffs and administered protection is very roughly estimated. Another reason is that productediffeirentiation confers some amount of autonomy on the domestic price system since itmeans that there is no longer a one-to-one relation between a change in trade policy and the correspond- ing change in the domestic price. Hownmuch a 10 percent tariff reducton onimnported drugs will affect the price of domestically produced drugs de- pends on supply and demand elastiities and on the respective axport and import shares in the sector producing drugs." 136 EFFES OF PRoTECnroNISM oN A SMALL CouTRY THE CASE OF URUcUAY Effects of labormarket rigidities Labor market rigidities also affect the welfare costs of protection. To the extent that the protected manufactuning sectors pay wages that are higher than the opportunity cost of workers, not enough labor is employed in manufacturing. When such distortions exist (a second-best situation), re- moving protecton might lower welfbre. Whether it does in a particular case is an empirical matter since it is not clear beforehand whether wage preni- ums occur in the sectors with high rates of effective protection. And when labor unions are strong, wage premiums are likely to be endogenous, so the adjustment in wage premiums resulting fiom import liberalization is of in- terest in itself. - To assess the -effects of labor market rigidities, the effects of tariff and administered protection are combined and the possibility of rent seeking is ignored. For the following four experiments changes are measured relative to as base solution that incorporates the wage differentials estimated earlier (appendix table 7B.1, last column): e Ł4 Setall exogenouswage differentials equal to unity, keeping tariff and administered protection. Measures the cost of wage differentials under tie assumption that theyare policy imposed and can therefore be removed. * E-5: Remove tariffandadministeredprotection, with exogenous*age differentials at base values. Intended to detect any second-best effects due to the combination of trade and labor market distortions. * E-6: E4 plus E-5. Measures the combined effiects of labor market dis- tortions and protection. * E-7: Same as E-5 but with endogenous wage differentials (see equa- tdon 7.6). Recognizes thatwages are determined through wage bargain- ing, making wage premiums endogenous. - Eliminating wage distortions leads to an estimated welfare gain of 0.14 percent of base income (table 7.2), about twice as large a gain as from elimi- nating tariff protection (table 7.1). The gain is larger, as expected, because labor income is a larger share of GDP than are imports. Eliminating taifh and administered protection (E-5) shows a largergain than in experiment 2, suggesting thathighlyprotected sectors payalowerwagepremium than other sectors, on average 12 Eliminating protection and wage distortions together results in an estimated gain of 5.04 percent of initial income. For the more realistic case in whic h protection is eliminated but labor unions remain ac- tive; the estimatedgain drops to 4.85 percentbecause less labor is reallocated)13 All in all, the estimates suggest that labor market distortions add an addi- tional 25 percent to the costs of tariffs and administered protection- Econmyuide costs of proion and labor maAet rigidties 137 Table 7.2 Aggregate effects of protection and labor market nrgidities (wpentge dwnge ftorn base inmGe and eoramert) &penmiet Char in welfare Labor reiocotbt E4: Erirate wage differtials 0.14 4.6 E-5: Siminre protectn with wage differentals 4.85 4.3 E-6: Bimmnit protctin and wage differentials 5.04 9.3 E-7: Elimiriate protection wih endogenous wage differentials 4.84 :2.3 a. Measured by equivatvaridaon. Welfare gain is expressed as a perentage of base GOP or 174.6 blion pesos. b. Computed as halthe surn o(the absoute value ofthe employmentdwVhanesit thoands owork-yewsarsd eqssed as the pernage otworkers that must rebcatc Base employment is 1.159 mnilion worker-years. Satcre Authors calcuiaior's based on BCU C1989) datb. The detailed sectoral results for employment andwages show that elkn- nating exogenous wage differentials boosts demand for labor and raises the wage-rentl ratio by 1.6 percent (appendix table 7B.4). For sectors with no wage differentials (agriculture, for example) wages rise 10 percent For sec- tors in which wage differentkals are eliminated, wage changes reflect changes in the economywidewageandin each sector'swage differential. Employment shifts toward manufacturing and services as a result of the elimination of the wage differentials (appendix table 713.4, column 1). If tariffs and admiistered protection are eliminated while wage differ- entials are kept unchanged (E-5), the wage-rental ratio increases by only 0.74 percent This increase implies that sectors benefitingfrom turiffs and admin- istered protection are, as expected, capital intensive. The combined effect of eliminating wage differentials plus tariffs and administered protection (E-6) indicates a reversal of resource shifts in many cases since several sec- tors with high protection also pay high wage premiums. Probably the most realistic pattern ofexpected resource shifts from elimi- nating protection is shown for the case in which wages in sectors with wage premums are determined by labor union actiity in each sector, according to the mechanism described in equations 7.5 and 7.6. (In the results reported in appendix table 7B.4, column 8,/6 = 2 in the labor union's objective func- tdon, implying a strong preferencc for wages over employment, a not unre- alistic assumption in the case of Uruguay.) Sectors that experience a large outward shift in their demand-for-abor schedule experience the highest increase in wages. Consider, for example, the grain mill products sector. Removing tariffs and administered protection (E-2) results in a strong re- source shifttoward the sector (appendix table 7B.3), causingarightward shift in its labor demand schedule. If labor unions determine wages, this increase in demand for labor will be reflected in a large increase in wages (81.8 percent). 138 EFEEcrs OF PRorEIloNIsM ON A SMALL CouNTR. THE CASE OF URUGUAY lIncreasaig returns and oligopolistic market structures The high values for sectoral concentration ratios (appendix table 71.1) and the barriers to entry created by administered protection in Uruguay are likely to have contributed to the establishment of oligopolistic markets in manu- facturing. A comprehensive assessment of the importance of these oli- gopolistic structures would require identifying and measuring the excess price-cost margin made possible by the barriers to firm entry created by administered protection. This information is not available, but there is grow- ingcross-country evidence supporting the import-disciplinehypothesis. (For early evidence on Chile, see de Melo and Urata 1986.) Neither are there reliable estimates of the returns to scale at the sectoral level. In place of this informaton the experiments here rely on rough estimates of economies of scale and excess profits that are assumed to be the same across all manufac- wring sectors. The simulations are therefore merely suggestve of the likely effects of increasing reurns to scale and noncompetidve domestic market structures. To explore the influence of oligopolistic markets on the effects of pro- tection, the elimination of tariffs and adniinistered protection is simulated under the following two sets of assumptions: - E-8: Same as E-2, butwith uniform increasing returns to scale inmanu- facturing sectors (CDR, = 0.10). * E-9: Same as E-8 but with an above-normal profit rate of 10 percent (I' = 0.1) in manufacturing sectors that disappear when protection is eliminated. Both experiments are comparable to the elimination of taiffs and adminis- tered protection carried outin experiment 2, except that the calibrated base solution now incorporates unexploited economies of scale in manufactur- ing and, for experiment 9, excess profits under protection. Whether the aggregate benefits of eliminating tariffs and administered protection rise or fall depends on whether unexploited economies of scale are more or less extensive in sectors that attract resources than in those that lose resources. Although tiis is an empirical issue that can be settled only with detailed econometric estimates of returns to scale at the sectoral level, lower gains could be expected because of the resource shift toward agricu- ture, which is assumed to operate under constant returns to scale. But comparing experiment 8 (table 7.3) with experiment 2 (table 7.1) indicates that the gains from eliminating ariffts and administered protec- don are almost 60 percent higher, even though unexploited economies of - scale are assumed to be relatively smalL Eliminating protection reallocates resources from nontradable sectors that operate under constant returns to scale to tradable sectors that in many cases operate under increasing re- Econoemywde costs ofptrotection and labor ma7t rigidities 139 Table 7.3 Welfare gain of removing tariff and administered protection with oligopolistic market structures (pecentace chrn fjbrn bose iromrne and e*p/oymenq &pement Chage n ri wrelre LborrelocOab E-8: Same as experiment 2c but with uniform increasing returns to scale in manufactring 7.61 3.14 E-9: Sare as experinent 8 but with above-norrnal profit rate of 10% in rnanufictng sectors that disappears with loss of protecon 9.62 4.42 a. Measured by equivalent varation. Welfare gain is expesed as a perrentage of base GDP cf 174.6 bEkn pesos. b. Computed as halfte sum of the absolute value ofrte ewiplaymendcanges in thousnds of vart-years and eprssed as the percentage orworkers that must reloatt Base cerplomnt is !.159 milion wonker-years. c. See table 7.1. Sourc Authas calculatios based on BCLJ (I 989) data. scale, the main exceptions being agriculture and mining. Since expansion in agriculture is largely balanced by contraction ir. mLhing, there is a net resource shift toward sectors with increasing returns to scale. Under the as- sumpdon that adjustment occurs through changes in the number of prod- ucts produced by a firm rather than through firm enty or exit, this shift in resources means that firms move down their average cost curves in response to the righcward shift in demand induced by the elimination of protection (see figure 7.1). Of coarse, underan alternative pricing rule such as monopo- listic competition, adjustment woufld occur through the entry of new firms and firms would move up their respective average cost curves. 4 Finally, suppose that the elimination of pro tection also removes barriers to entry so that firms are forced to price more competitively. There is evi- dence that trade liberalization results in lower price-cost margins,"5 which supports the assumpton that above-normal profits (the shaded area in fig- ure 7.1) disappear when tariff and administered protection is eliminated. This move toward more competitive pricing behavior by manufacturingfirms adds another 2 percentage points to GDP. Added together, then, the costs of administered protection would be about 10 percent of GDP, only a small fraction of it due to tariff protection. While the stylized numerical analysis cannot pretend to capture precisely the various components of the costs of protection, it is clear that the costs of nontariff protection in Uruguay are relatively high, especially if some resource waste accompanied rent-seeking activities and if protection allowed firms to engage in more ohgopolistic behavior. Conclusion The numerous and pervasive distortions in Uruguay's economy present a difficult modeling challenge. In the absence of more detailed information, 140 EmasC OF PROTPZnONISM ON A SMALL COUNw: THE CASE oF URUGUAY the general equilibrium simulation' analysis was confined to modeling the effects of tariff protection, administered protection, and intersectoral wage differentdals not accounted for by education and other worker characteris- tics. Though using the wedge between imports and similar domestic products created by different trade and transport service costs as the mechanism through which administered protection affected resource allocation and welfare is only a rough approximation, it permitted exploration of some of the likely effects of administered protection. The simulation estimates of the likely magnitude of the welfare effects and the pattern of resource reallocation that would take place if selected protection measures were eliminated confirm the more detailed analysis in chapter 4: the costs of administered protection are far greater than those due to tariffs. According to the crude estimates from the simulation exer- cises, the statc costs of administered protection could total 4 percent of GDP-even double thatamount ifadmninistered protection gives rise to rents whose full value is dissipated inwasteful rent-seeking activities. The benefits of eliminating differences in sectoral wages not accounted for by education or other characteristics of workers are estimated to be twice those of eliminating tariff protection. These gains total only some 0315 per- cent of GDP, however, because wage premiums are largest in sectors that employ a relatively small share of the labor force engaged in tradable actvi- ties and because the extent of labor reallocation is more limited for non- tradable activities, which have little scope for specalization. The disaggregated model used to estimate the effects of distortions also gives an approximation of the likely resource reallocation thatwould occur if protection were elmin-ated. The simulations indicate that the resources released by the elimination oftarffs andadministeredprotectionwould move toward agriculture and, within manufacturing, toward light manufactures. The patter of trade and labor market distortions does not give rise to sec- ond-best effects because the highestwage premiums are notusually found in sectors with the highest rates of effective protection. What can we say about the robustness of the estimates presented here, considering that not all the protective instruments affecting international trde have been incorporated and that sensitivity analysis of elastcity choice and model structure has not been conducted?) No doubt the least robust estimates are those for the resource allocation costs of administered protec- tion, which were modeled indirectly. While the results are at best suggestire, it is nevertheless likely that the costs of administrative protection are sub- standal larger than the costs of tariff protection. Considering Uruguay's low ratio of trade to GDP for a small country, it is not unreasonable to imag- ine that the costs of administered protection are ten to twenty times those Econon-wide costsof protection and laborkza*et gidWIies 141 of tariff protection. Regardless of assumptions about elasticities or model structure, the exercises here show dearly that the costs of nontariffbani- ers are, under any plausible scenario, many times larger than the costs of tariff protection. The general equilibrium estimates also suggest thatif administered pro- tection is as pervasive asis modeledhere, removingprotectionwould involve a depreciation of the real exchange rate on the order of at least 10 to 20 percent. A move toward free trade would thus entail a substantial contrac- tion of manufacturing and an expansion of the agricultural sector of per- haps 10 percent Such a shift is to be expected in a country that has taxed its agriculture heavily for so many years. Finally, the estimates on labor relocation give some notion of howmuch resistance is likely to greetattempts to substandall5y change trade policy.While granting that the magnitude of the estimates is sensitive to assumed param- eter values governing firms' demand for labor, it is nonetheless instructive to considerwhat the need to relocate some 50,000 workers would mean in a country like Uruguay, with a strong labor union movement Stiff labor resis- tance should certainly be anticipated. Appendix A. The Uruguayan social accounting matrix The social accounting matrix (SAM) underlying the computable general equilibrium model used for the analysiswas constructed with 1983 data from the input-output matrix, national accounts, and balance of payments ac- counts. All data were provided by the Central Bank of Uruguay. A SAM is a tableau form of single entry accounting for income and ex- penditure flows between institutions in the economy. It gives an exhaustive and disaggregated accounting of the composition of national income and product The SAM used here was constructed to give particular attention to production accounting, because of the study's focus on the effects of trade policy on resourcc allocation- The SAM indudes thirty domestic actity sec- tors, detailed by input-output linkages, value added, and final demand. An aggregate version of the SAM with ten producing sectors is pre- sented in table 7A.1 for seventeen institutions. Rows of the matrix indicate income sources for each instimtion, and columns their expenditures. Row and column sums are equal for every agent by individual income-expendi- ture identities. The blocks of the matrix correspond to various accoundng categories. The largest block (first ten rows and columns) details interindus- try flows or production accounts, faniliar from input-output analysis. These entres itemize the receipts for sectoral deliveries (rows) and the expendi- ture requirements (columns) of internediate production. The next block 142 EFFECrs oF PboTECTnoNsM ON A SMALL CouNTRY: THE CAsE oF URUCUAY Table 7Ak1 Social accounting matrix for Uruguay, 1983 (miXions orcuntent new pesos) Othernon- Constwc- Agrictture Minhg Food Apprel durables Durobles Uwrity ton I Agimt,.ure 1,634 0 20.470 3.625 287 116 0 15 2 Mining 2.127 11,777 1,116 295 379 1,007 536 1,282 3 Food 956 0 4,580 1.796 350 7 4 1 4 Apparel 166 5 236 5.090 77 198 2 3 5 Other nondurables 2.630 69 [.296 1.27i 5,768 1,421 76 2,143 6 Durables 1.001 204 1,549 620 876 3.481 173 7,720 7 Utrity 53 51 509 199 284 333 1,187 216 8 Construion 392 0 167 20 37 32 8 0 9 Dtibution 6.820 1,174 6.093 2,666 2,726 2,707 99 118 10 Cther services 483 152 1,614 611 895 509 252 299 II Factors 22,953 896 13,257 7,908 6,472 7.748 3.611 6,832 12 Households 0 0 0 0 0 0 0 0 13 Importtaxes and margns 462 1,675 804 255 3,01B 4,719 0 0 14 Domesfictax and subsidies 847 3,804 5.766 1,007 660 926 2.051 516 15 Govemrment 0 0 0 0 0 0 0) 0 16 Capital account 0 0 0 0 0 0 0 0 17 Rest ofworkd 1,978 10,809 1,288 724 4.896 9,051 0 0 Total 42,504 30,616 58,744 26,088 26,727 32.254 8,001 19,145 SDuwi: BCU (I989). covers the factor accounts for labor and other productive factors. Row 11 details wage and rental payments by sector, and factors in turn remit their income to domestic institutions (column 11), including households (row 12) and government (row 15). Household and enterprise accounts receive income from factors in their rows. Households disburse their income in consumption across sectors and to private savings (column 16). Uruguay has no personal income tax, but there are payments to enterprise taxes, expatriated profits, and savings and to finance inventory changes. The last accounting block consists of "exogenous" accounts for govern- ment, capital and real accumulation accounts, and the restof theworld. The government collects revenue from a variety of direct and indirect fiscal in- struments and disburses its revenue in the form of sectoral subsidies, public expenditure on goods and services, transfers, public savings, debt service, and foreign transfers (column 15). Capital accounts accumulate private and Economywide costs ofproection and labor ma*Aetigdities 143 Import Domestic taxes taxes Distribu. Other House- and and Coven- Capitol Rest of don seMces Factors holds marzuins subsidiies ment account wattd Total 247 21 0 7,616 0 206 479 2,239 5,548 42,504 3,917 590 0 5,057 0 694 1,150 -523 1,213 30,616 3,405 72 0 29,199 0 754 435 704 16,483 58,744 55 290 0 5,427 0 1,904 93 847 11,694 26.088 680 2,366 0 5,557 0 26 670 918 1,836 26,727 1,576 1,868 0 4,183 0 7 747 6,983 1,267 32,254 786 522 0 3,033 0 0 761 0 67 8,001 377 1,521 0 146 0 0 436 16,008 0 19,145 2,208 911 0 12,509 5,265 1,007 640 0 3.831 48,775 3,619 4,464 0 36,446 810 14 1,175 14,174 1,971 67,489 27,551 50.446 0 0 0 0 0 0 0 147,675 0 0 134,806 0 0 0 21,831 0 14,412 171,049 0 0 0 0 0 0 0 0 0 10,933 3,632 3,529 1,083 17.035 0 0 4.612 0 0 45,467 0 0 11,786 3,147 4,858 40,855 754 5,835 0 67,234 0 0 0 38,031 0 0 0 0 10,062 48,093 723 889 0 3,664 0 0 33,453 909 0 68,384 48.775 67,489 147,675 171,049 10,933 45,467 67,234 48,093 68,384 public savings (row 16) and finance sectoral investment demand and gov- ernmentborrowing (coluKnn 16). The rest oftheworld receives incomefrom import sales (row 17), purchases exports, and engages in other remittances and capital account transfers. Appendix B. Sectoral tables (Tables begin on page 144.) 144 EFFECrS OF PROTEmONISM ON A SMALL COUNTRrY THE CASE OF URUGUAY Table 76.1 Estimates of price distortions, domestic concentration, and wage premiums (pecent) IMpot onet goods Tadif Trade and Trade and Ftr-firrn Indusry on ci.f urnspar t mspon concentroaion Wage Indtuy code price mfagin margin ratio preiniumn I Agricuture I 8 54 I 1 2 Mining 2 1 0 7 5 - 3 Meatproducts 3111 2 28 10 30 - 4 Dairy products 3112 1 9 9 1 1 6 75 35.0 S Preserves. cardy 3113.3119 9 92 20 49 35.0 6 Fish 3114 17 68 5 49 35.0 7 Edibleoils 3115 8 32 9 58 35.0 8 Grain mil producaS 3116,3122 1 1 84 7 - 35.0 9 Bakeryproducts 3117 19 a1 10 7 35.0 10 Sugar 3118 2 60 7 100 35.0 11 Beverages 313 12 30 24 - 35.0 12 Tobacco 314 1 94 49 55 35.0 13 Textiles 321 9 55 1 3 30 37.1 t4 Apparer 322 0 72 16 10 37.1 t5 Leather 3231 2 31 9 47 37.1 16 Footwee 3233. 324 45 43 19 30 37.1 17 Woodproducts 33 16 69 7 - 5.9 1 8 Paper products7 341 9 38 13 61 44.7 19 Printg, publications' 342 44 40 20 20 60.5 20 Drugs 3522 11 77 1 0 1 8 60.5 2 1 Other chemicals 351. 352.3522 14 34 II 52 60.5 22 Tires and rubbee 355 16 41 9 73 60.5 23 Glass 362 20 109 22 - 7.4 24 Metal. rnachinery' 381. 382, 383 12 42 8 24 26.6 25 Utilkies 4 0 0 2 - - 26 Construction 5 0 0 I - 20.8 27 Wholesale, retail trade 6 0 0 I - 11.7 28 Transport, comrnunicaton 7 0 0 5 - 16.9 29 Fnance, insurance 8 0 0 1 - 48.9 30 OtherseMices 9 10 35 9 - - Weghted average' 10 35 9 44 - Not applcable. * Indicates sonme forr of administered protecdon (see chapter 4). a. Weights are imports and gros outputs. Sovrce: Avxrs calculirrns. based on BU (1989) data chapter 6 of tis volume for conceradton raios and Bucheri (I 992. tale I) fr wage premiums. Econonywide costs of protwciion and labor mairet vidities 145 Table 7B.2 Trade and production structure in the calibrated base solution (percent) V-lue- Emplor- Capital Efetaive Grss added ment output Ipors E&orz protectbn Industry -ouput ratio shore ratio demand output rate" I Pgriculture 12.2 58.0 16.5 4.4 8.9 13.8 5.1 2 Mining 5.6 4.9 0.5 2.4 49.4 6.7 11.9 3 Meat products 6.1 21.9 2.0 2.6 0.3 46.8 -1.7 4 Dairyproducts 1.3 17.0 0.4 1.5 0.1 31.8 152.8 5 Preserves. candy 1.3 25.7 0.4 2.4 14.3 16.4 99.8 6 Fish 0.5 23.8 0.2 2.0 6.7 83.3 38.7 7 Edible oils 0.3 18.4 0.1 1.8 33.6 15.3 65.5 8 Grainrmilproducs 2,1 21.1 0.7 2.0 7.5 40.5 64.7 9 Bakeryproducts 1.6 32.4 0.5 3.1 0.0 13 -1.3 I10 Sugar 0.7 39.6 0.2 3.9 2.9 11.6 31.4 I I Beverages 2.1 -28.7 0.6 44 1.9 0.5 57.1 12 Tobacco 1.I 12.2 0.1 6.8 -2.0 2.7 -2.1 13 Textese 3.7 326 2.4 3.1 10.0 45.8 27.7 14 Apparel 1.9 37.9 2.2 2.2 2.0 36.4 -9.1 I5 Leather 1.4 17.6 0.6 1.7 6.5 65.7 -11.4 16 Footweae' 0.7 35.6 0.5 2.9 0.9 44.1 146.0 7 Wood products 0.8 39.3 1 A 3.4 11.6 2.9 61.6 18 Paper products' 0.9 34.9 03 35 273 I 13 40.0 19 Printig, publications 0.7 31.1 0.7 23 15 4.5 160.0 20 Drugs 0.3 43.5 0.0 4.5 77.2 14.8 31.0 21 Otherchemicals 3.1 33.0 0.8 3.3 39.9 12.0 39.7 22 Tires and rubbee 1.1 39.2 0.6 3.8 12.3 44 625 23 Glas 1.5 45.1 0.7 4.3 9.8 5.5 66.0 24 Metal, machinery 3.0 42.1 2.6 3.7 627 8.4 334 25 Utilities 2.4 45.5 1.5 6.3 0.0 0.8 -2.1 26 Cosructon 5.8 36.1 5.5 .8 0.0 0.0 -44.1 27 Wholesale, retail trade 9.9 53.1 12.5 5.9 1.1 4.3 -5.2 28 Trasport. communication 7.1 46.1 5.2 5.2 2.7 10.8 -6.1 29 Finance, insurance 13.6 82.6 2.1 8.5 !.3 2.8 -0.4 30 Other services 7.0 61.3 38.1 !.3 1.5 3.1 -7.6 Total 100.0 100.0 Weighted average 45.5 4.3 9.5 13.4 7,3 Indicates sone form of adinistered provtedion. a. The effective rate of prtection orsecors is calcdated as: ERP, = (t,- t, )/C(l - S o,,) from the zariTrats in table 78 1. Source: Aud' olculatiors, based on BCU (1989) data. 146 EFmCI OF PROTECnONISM ON A SMALL CouNTRY: THE CASE OF URUGUAY Table 7B.3 Resource shifts from removing protection (perxntage dcrMge from calbruted base solution) E&errnent I ExperimienE 2 ELperiment 3 Itizsty Output Eaports Output Expots Output Exports I Agriculture 3.5 9.5 7.3 20.9 15.0 32.7 2 Mining -11.1 -18.9 59.0 233.0 -14.6 -31.3 3 Meat products 2.4 8.5 20.9 37.5 30.8 50.8 4 Dairy products -47.8 -70.5 -15.1 -18,6 -20.6 -29.1 5 Preserves, cady 1.6 85 -9.5 -10.0 48.5 126.5 6 Fish 25.1 27-7 -43-4 -45.3 -7.4 -7.3 7 Edible oils -32.2 -53.0 10.3 35.9 15.1 44.8 8 Grain mill products 23.5 40.8 24.7 46.4 37.2 64.6 9 Bakery products 3.2 1 2.7 6.5 26.4 14.5 43.9 I 0 Sugar 0.1 5.4 2.5 16.7 7.8 22.6 11 Beverages -0.5 6.4 -L8 11.5 3.8 27.1 1 2 Tobacco 0.0 6.9 0.4 14.2 4.4 22.1 13 Textilese 20.4 35.2 -10.9 -8.6 -25.8 -33.4 14 Apparel -6.7 -7.5 -24.5 -33.1 -19.6 -26.0 1 5 Leather -49.6 -67.9 -64.1 -85.0 -56.5 -77.4 16 Footweas -22.5 -30.8 -30.4 -40.5 -13.9 -16.6 17 Wood products' -0.3 10.9 0.7 24.0 3.5 30.0 18 Paperp p!uct 1.8 7.8 63.6 175.5 7.6 22.9 1 9 Prinirng. publicatiOnS 2.1 8.8 1 1.4 37.6 7.4 23.6 20 Drugs -2.0 5.4 -4.1 14.4 -11.4 13.4 21 Otherchemicals 16.3 51.2 43.8 139.5 9.3 38.2 22 TreSand rubber 1.2 15.1 2.7 30.4 3.7 28.1 23 Glass 0.4 7.7 2.7 26.1 4.1 24.4 24 Metal, machinery 0.5 11.6 -9.9 -2.5 20.7 73.9 25 Utilities -0.6 5.5 -1.1 10.9 1.2 14-2 26 Construcion 1.7 0.0 S. I 0.0 7.0 0.0 27 Wholesale, retail trade -0.8 5.0 -1 1.8 -1.6 -5.8 20.6 28 Trnsport communication -8.6 -16.4 -8.7 10.8 -2.6 29.6 29 Finance, insurance -0.5 8.2 -6.3 4.2 I. I 24.4 30 Otherservices 0.8 12.2 1.8 24.9 3.7 26.9 Weighted averae -0.3 42 3.1 15.4 3.5 14.8 - Indicates some form or adrninisered protection. Saurac rthors' aonrs, based on BCU(I 1989) data Economwide costs of trotedion and labor maret rigidities 147 Table 7BA Resource shift from eliminating protecdon and labor market rigidities (rcentage chanEe rtmrn the calibrated bose solukcon) E*peermenc 4 Experment 5 bmpefrret 6 Epenrment 7 Wage Wage Wage Wage hdustry Employment rote Employment rmce &rpoyrment rate Employment rrate I Agriculture -125 10.0 10.8 4.0 -10.2 14.2 1 1.5 7.9 2 Mining -16.9 10.0 8.1 4.0 -30.5 14.2 -5.1 7.9 3 Meatproducts -145 10.0 253 4.0 -25.2 14.2 -10.6 7.9 4 Dairy products 22.9 -15.9 -17.9 4.0 29.2 -37.7 -3.1 3.5 5 Preserves. candy 28.0 -15.9 2.7 4.0 60.7 -37.7 -1.5 5.7 6 Fish 215 -15.9 -37.3 4.0 29.4 -37.7 -8.2 -3.6 7 Ediblecils 26.7 -15.9 12.5 4.0 55.3 -37.7 -1.4 5.8 8 Grain mil produces 24.6 -15.9 76.0 4.0 267.8 -37.7 41.4 81.8 9 Bakery products 28.7 -15.9 11.8 4.0 108.1 -37.7 11.8 25.9 10 Sugar 303 -159 -1.0 4.0 59.3 -37.7 -3.0 3.6 11 Beverages 15.9 -15.9 -3.1 4.0 29.3 -37.7 -2.0 2.2 12 Tobacco 17.4 -15.9 -1.6 4.0 33.3 -37.7 -1.5 3.5 13 Textile 29.7 -17.0 -43.0 4.0 20.9 -40.0 -2.2 4.7 14 Apparel 19.8 -17.0 -31.6 4.0 23.9 -40.0 -1.3 6.1 15 Leaher 21.0 -17.0 -17.8 4.0 48.3 -40.0 -1.5 57 16 Footvear 22.6 -17.0 142.6 4.0 234.8 -40.0 0.0 7.9 17 Wood products -3.9 4.4 -0.6 4.0 -6.9 3.0 - I.5 53 18 Paper producte 20.9 -20.9 0.3 4.0 41.7 -47.6 -2.2 1.1 19 Printng. Publicarors 16.3 -27.7 2.1 4.0 44.4 -61.2 -0.3 6.9 20 Drugs 34.5 -27.7 -13.1 4.0 84.3 -51.2 -7.5 -13.6 21 Otherchemical 31.3 -27.7 18.6 4.0 62.1 -612 -0.5 6.1 22 Tire and rubber 24.9 -27.7 0.9 4.0 58.6 -61.2 -0.2 7.2 23 Glass -1 .7 3.1 0.7 4.0 -0.7 0.4 -0.8 5.8 24 Metal, rnachinery 10.9 -11.0 0.5 4.0 15.8 -27.9 -3.0 0.1 25 Ulics -7.7 10.0 -2.3 4.0 -16.4 14.2 -4.4 7.9 26 Construcion 9.8 -7.2 4.0 4.0 25.9 -20.3 0.6 8.8 27 Wholesale, retail trade -0. 1 -0.5 -11.5 4.0 -8.8 -6.8 -5.5 -1.4 28 Trasport communicaton 3.8 -4.4 -128 4.0 -14.1 -14.8 -6.1 -2.5 29 Finance. insurance 42.3 -22.8 -5.7 4.0 76.4 -51.5 -2.9 2.3 30 Otherservices -5,3 iO.0 0.5 4.0 -14.5 14.2 -1.7 5.4 Weighted average 0.0 2.4 0.0 4.0 0.4 -1 .1 0.0 53 Icircates somre form of administerd protecion. Source: AuN' calculations, based on BCU (1989) data. 148 EFFECrS OF PROTECTIONISM ON A SMALL CoUNTRr. THE CASE OF URUGUAY Notes 1. As discussed in the next section, margins are systematically higher for imports than for 'comparable" domestic goods. Caveats in our interpretation of differences in margins reflecting the effects of nontariff barriers are also discussed in the nextsection. 2. 2This distortion-and associated rents-is likely to have a strong distributive effect (see following section), because the model onlyhas one representative consumer, it does not capture the likely distributive implication of this wedge. 3. The alternative is to assume that the zero profit is achieved through firm enty and exit in a monopolistic compedtion framework. See, for example, Harris (1984), Devarajan and Rodrik (1992), and de Melo and Roland-Hoist (1991). 4. Fordetailson parametcrvalues, see the following section. Fora comparison of results derived from this pricing rulc with those for monopolistic pricing allowing for entry and exit, sec de Melo and Tarr (1992, chapter 7). 5. The treatment of wage differentials is the same as that specified in de Melo and Tan (1992, chapter 6), which contains a more detailed explanation of the specifica- tion. - & An alternative is to assume chat the union and the firmnjointlynegotiate wages and employment, as in McDonald and Solow (1981). 7. The model's structure closely resembles that of de Melo and Tarr (1992, chap- ters 3, 6, and 7) except for the treatment of rent seeking, which is inspired by Grais, de Melo, and Urata (1986). Foran introductorypresenration to the structure of the model used here, see de Melo and Tar (1992, chapter 2). & There are also sectoral taxes and subsidies, but to simplify the interpretation of results, they are not included in the model. 9. The equivalent variation measure uses final prices in the evaluation. An alter- native measure, the compensating variation, uses base prices and gives very similar results. 10. Dairy products, a major export sector, also shrinks by 48 percent because of a high effective rate of protection. In this case, the effective rate of protection is prob- ably not a good indicator of the sector's competitiveness. 11. For a formula indicating how the domestic price is related to a change in the import price, in terms of the parameters describing supplyand demand elasticities for the functional forms selected in table 7.1, see de Melo and Tr (1992, section 2.5). 12. In the case of the U.S. auto and steel industries, estimates go in the other direction, withl a slightly smaller welfare gain from eliminating protection in the pres- ence of exogenous wage distortions. 13. For a sensitivity analysis of the results to changes in the weights on wages and employment, see de Melo and Tarr (1992, table 6.8). 14. For a comparison of the two pricing rules and a formula decomposing the welfare effects of a change in trade policy into scale and other efficiency effects, see de Melo and Roland-Holst (1991). 15. For example, de Melo and Uraa (1986) compared price-cost margins for Chilean manufacturing firms in 1967 and 1979. They found that price-cost margins fell drasticall following the trade reforms. They also found econometric support for the import discipline hypothesis, which is captured in an ad hoc way in experiment 9. Economyuwide costs of protedion and l mborket giWities. 149 References Anichini,J.J.,J. Caumont, and L. Sjaastad. 1978. 'La politic comercial y la proteccion en cl Uruguay.' Banco Central del Uruguay, Montevidco. BCU (Banco Central del Uruguay). 1989. "Matriz de insumo-producto del Uruguay." Montevideo. Bucheli, M. 1992. "Diferencias scaoriales de salarios en cl Uruguay." SUAMA 7(12)- 55-80. Devarajan, Shanta, and Dani Rodrik. 1991. 'Pro-competitive Effects ofTrade-Reform: Results from a CGE Model of Cainrcon."Eurpean EconomicReview 35: 1157-84. Grais,WafikJaime de Melo, and Shujiro Urata. 1986. 'A General Equilibrium Estima- tion of the Effects of Reduction in Tariffs and Quantitative Restrictions in Turkey in 1978." In T. N. Srinivasan andJ. Whalley, eds., GeneralEquilibrium TrzdePoliy Mod- eling. Cambridge, Mass.: MIT Press. Harris, R. 1984. 'Applied General Equilibrium Analysis of Small Open Economies with Scale Economies and Imperfect Competition." American Economic RBview 74: 1016-32. Jondrowj., D. Chase, and C. Gamble. 1982. 'The Differential between Domestic and Imported Steel."Journal ofBusiness 55: 383-99. Krueger, A., and L Sumners. 1988. "Efficiency Wages and the Inter-Industry Wage Structure." Economebica 56: 259-93. Macadar, Luis. 1989. "Protecci6n,ventajas comparadas y eficiencia industial." SUM 3(5): 7-59. McDonald, r. M., and R M. Solow. 1981. 'Wage Bargaining and Employment." Ameri- canEconomicRview71:896-908. MeloJ. de. 1987. "Financial Reforms, Stabilization, and Growth uider High Capital Mobility: Uruguay, 1974:83." In M. Connolly and C. Gonzilez-Vega, eds., Economic Reform and Stabilization in. Latin America New York: Praeger. MeloJ. de, and ID. Roland-HoIsL 1991. T'rade Policy and Industrial Organization: Evidence from Korea. In R- Baldwin, ed., EmpiricatStudies of CommercialPolicy. Uni- versity of Chicagc Press. MeloJ. de, and D. Tarr. 1992. A Gneral Equilbrium Analysis of U.S. Tr& Poliy. Can- bridge, Mass.: MIT Press. . 1993. "Industial Policy in the Presence of Wage Distortions The Case of the U.S. Auto and Steel Industries." Inerwional Economnic Review 34: 833-51. Melo,J. de, and S. Urata. 1986. The Influence of Increased Foreign Competition on Profitability." Inn4w.iona1Journai of industrial Orcnization 4: 287-304. Rama, M. 1991. "Endogenous Trade Policy: A Time-Series Approach." World Bank, Policy Research Department, Trade Policy Division, Washington, D.C. Roussiang, D. and T. To. 1992. "Domestic Trade and Transportation Costs as Barriers to International Trade." CanadianJournal of Eonoamics 26: 209-21. A long-run perspective on trade policy, instability, and growth Jorge E. Roldos The impact of trade policy on economic growth has attracted the attention ofboth growth theorists and development economists, and their insights are beginning to converge in thenewhendogenous growth theories.Though some of the forecasts derivedfrom these theoriesmake them observationally equiva- lent to the neoclassical growth model, the new theories have different impli- cations for economic policy. Anumber ofwell-known comparative studieshave presented the empiri- cal evidence suggesting thatdevelopmentstrategiesbasedon exportpromo- tion lead to better economic outcomes than import-substitution strategies. Some endogenous growth models now capture the intuition that export growthleads to higherproductivitygrowth through theexploitation of econo- mies of scale, technology tansfer, or increasing competition (see Nishimizu and Robinson 1984)-whatBhagwati (1986) and others characterize as "grey area" dynamic effects. Within the endogenous growth framework, the static- or level-effects traditionally considered distortions can be reinforced or off- set by dynamic or growth effects (Easterly and others 1990). This chapter draws on some of the insights from endogenous growth theory to examine the role of trade policy in the performance of the Uruguayan economy. After a period of rapid growth in the 1940s based on a strong import substitution policy, Uruguay went through a period of stagnation between 1955 and 1974 (see chapter 2). While Uruguay was stagnating other coun.- trieswere growing fairly rapidly as world trade expanded in the postwar period. In response to its dismal performance, a sharp deterioration in the terms of trade, and a balance of payments crisis, Uruguay shifted away from its im- port-substitution strategy, and at least for the second half of the 1970s, its no-growth recordwas reversed. The debt crisis and the deepening of prefer- ential trade agreements with neighboring Argentina and Brazil have condi- tioned Uruguay's more recent performance in terms of growth, trade, and m3croeconomic stability. This chapter explores these linkages. 150 A long-run paspective on trade poliey, instability, and growth 151 It first addresses the perennial question: 'Does growth drive trade or is there a reverse link from trade to growth?' (Helpmarn 1988, 6). A purely sta- tistical look at the issue-using Granger causaIity tests-uncovers no evidence of temporal causality in either direction for annual or quarterly data, and no indication of cointegration of exports and GDP. Some evidence is found of a positive correlation of export growth and Solow growth residuals, suggest- ing that growth extemalites may have played a role in the relationship be- tween exports and growth in real output. Economies of scale, which play a large role in manyendogenousgrowth models, may be importantin Uruguay, although the evidence is weak. Some formal and informal evidence is also presented on whether increasing returnsdie hallmark of the new growth tbeories-are importanL Finally, an exploration of the effects of preferential trade agreements on exports andgrowthfinds clearevidence thatexports to the region are more unstable than those to the rest of the world and that this instability is assodated with less GDP growth at the sectoral level. Exports and growth: causality tests Mostnormative arguments in favor of an export-promotion trade strategy- defined as one tLat creates no biased incentives toward import substitution (see Bhagwati 1986)-are static. They stress an import-substitution regime's generation of distortions in p: oduction and consumption, the increase in directly unproductive or ldbbying activities, and the welfare-reducing, tariff-jumping invesunent it induces. Bhagwati (1986) adds a final group of arguments he calls "grey area" dynamic effect, which include the impact of the trade regime on savings and innovation. These effects are becoming better understood with the development of new trade theories-based on increas- ing returns to scale and imperfect competition-and endogenous growth theonres. There is ample empirical evidence on the superior performance of economies under export-promotion regimes over those with import-substi- tation reginmes. Most studies involve some cross-section regression of changes in exports oni changes in GDP (see Michaely 1977 and Feder 1982). But as Helpm-an (1988) notes, the crucial question to be answered, in theory and empirically, is whether growth causes trade or trade causes growth. Early two-sector models of growth and trade-generally extensions of Solow'sneoclassicalmodel-explain the dynamics ofboth variablesin terms of capital accumulation and exogenous technological progress. When tech- nological progress occurs in the exportable sector, exports grow faster than GDP-a pattern found for many fast-growing economies. The problem with predictions derived from those models (Helpman 1988) is that they suggest that causality goes from outputgrowth to exportgrowth, whereas many econo- mists believe it goes the ocher way. The new endogenous growth theorists 152 EFFtrs OF PROTEwOnSM ON A SMALL CouNTr. THE CASE OF URUCJAY have started to develop models in which export growth causes CDP growth. This section looks at this causality problem for Uruguayan data for 1935 to 1989 (see figure 8.1). Conventional augniented Dickey-Fuller tests show both variables to have a unitroot or a stochastic trend. Indeed, both are well characterized by inte- grated moving average ([MA) processes, IMA(1,1): (8.1) AY = 0.02 + (1 + 0.285L)e, (3.48) (2.06) (8.2) AA= -(1 -0.476L);r (3.39) where Y, is the logarithm of GDP and X, is the logarithm of export volume. With these results in mind, causality tests were conducted on first differences of both variables (lower-case letters denote rates of growth). A traditional regression of GDP on export volume growth rates yields the following result, after a Cochrane-Orcutt zorrection: (8-3) Y= 0.15 + 0.074X, + 0.095x, I (2.08) (2.64) (3.40) R2 0.25 DW = 1.60 F =5.32 Figure 8.1 Exports and GDP, 1935-89 Index (1961 100) 2.50 200- 100 so ' 0 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 Source: See appendix. A ongrun pedivect Ol on cde po6cy, instabily, and gowth 153 Theresultshowsapositive correlation between outputgrowth and exportgrowth, especially if a lagged export growth vaiable is induded, but this probably re- flects a cyclical correlation and not the kind of trend growdt of interest here. Granger (1969) causality tests and a test suggested by Geweke, Meese, and Dent (1983) were used to test for temporal causality. According to the Weiner-Granger definition of causality (Grangerand Newbold 1986) avari- able xis said to cause another variable yif prediction of the currentvalue of y is improved (in a least-square sense) by the use of past values of x The practical tests are undertaken by running the regression: k (8.4) YJ = .o +YPjyf-j +AYjxt_} $* j=1 j=1 where jis the number of lagged periods, and testing for the null hypothesis that y = 0 (x, does not cause y) by means of the usual F-tests. The order k of the lags is often set at some arbitraily high number, keeping in mind that the power of causality tests is low for heavily parametrized models. Simulation results in Geweke, Meese, and Dent (I 983) suggest an alter- native test After equation 8.4 is estimated using ordinar least squares and denoting by C2 the error-vaiance estimate, the following equation is est- mated, also using ordinary least squares: (8.5) Y, =Po +Y5Jc -iE, 1=' denoting by a the residual variance estimate of this regression. The test sta- tistic is: 2 a2) (8.6) T = * (y~~~2 which has an asymptotic chi-square distribution with k degrees of freedom under the null hypothesis tat x does not cause y. Under both tests the hypothesis that export growth does not cause GDP growth cannot be rejected (table 8.1). There is weak evidence of causality for the case of two lags, but it is not confirmed by the more powerful t-test. Strong evidence that output growth does not cause export growth rules out reverse causality. Splitting the sample and doing the same tests for the sub- period 1965-89-the series seems to exhibit a structural change in the early 1970s coinciding roughly with the change in trade policy-yields similar results: a lack of temporal causality in either direction. One problem with this kind of causality testing is temporal aggregation. If the true lag between cause and effect is one quarter but the stochastic process is observed annually, then causalitywill seem to occurinstantaneously and so will not be captured by these tests- Quarterly data-available for GDP 154 EFFECww OF PROTECrbONISM ON A SMALL CouNmY: THE CASE OF URUGUAY since 1975-were used to overcome this problem. The results again show no causality in either direction (table 8.1). Another problem with these causality tests is philosophical: Cooley and LeRoy (1985) among others have suggested that causation can be accepted only if the empirical evidence is associated with a complete and convincing theory about how the cause produces the effect A rather casual, Bayesian sort of answer to this concern is given by Granger and Newbold (1986,222- 23). A prior belief abouta particular theory could be expressed in probabil- ity terms, so that after the data have been gathered and proper testing has been done, the probability could be raised or lowered, according to.the re- sults. For the case being considered here, this process would probably result in a downward adjustment of the probability since there is little evidence for Uruguay to support the hypothesis that exports cause growth. A final problem with these causality tests is their failure to allow for in- stantaneous causality. While estimation and testing of a structural model of the determination of exports and output is beyond the scope of this study, the possibility of cointegration of the two variables was examined. Since both variables have aunitroot, they couldbe cointegrated, that is, they could share a common stochastic trend (Engel and Granger 1987). If the variables are cointegrated, they can, following the Granger Representation Theorem, be represented as an error-correction model, in which long-run (permanent) components of exports and GDP obey some equilibrium constraints, and short-run (transitory) components of the variables have a flexible dynamic specification. No evidence was found, however, of cointegration between exports and GDP, even after dropping the early sample covering the period of fast growth in tie early stages of the import-substitution strategy. Table 8.1 Causality tests for export and GDP growth Eor grwh cses Output growh cLmu cutpLtt growth exportgmwth Series F-tet T-test F-test T-test Annual (1935-89) 4 lags 1.33 1.5 1 0.35 3 lags 2.04 3.40 0.62 * 2 lags 3.17' 4.60 0.01 Annual (1965-89) 3 lags 1.64 2.25 1.45 4.54 2 lags 1.25 0.56 0.29 Qwrwml (1975-8) 6 lags 1.41 2.80 0.83 0.72 5 lags 0.89 * 0.89 S 4 lags 0.86 * 0.94 0 *'The ciflerential error-variance is negUi&ble *Sinifintatte IOpercentleveL Sorce See apperndc A tong-un paenchiwe on trade poicy, insfabilit4, and gr=k 155 Scale effects, trade policy, and growth If there is no statistical causality between exports and growth, is there evi- dence of other scale or trade policy effects? A brief review of traditional anud endogenous growth theories demonstrates why such effects could be important Traditional and endogenous growth theories Following the neoclassical growth model elaborated by Robert Solow in the 1950s, the growth accounting literature has attempted to quantify the con- tribution to growth of each physical input and technological progress. The model assumes constant returns to scale, diminishing returns for each fac- tor, and exogenous technological progress. Even with adjustments for the quality of inputs (education, equipment vintage) these studies suggest that physical inputs account for only 50 to 70 percent of growth in output (see Middison 1987). In the simple Cobb-Douglas case, after taking log- differentials, a large part of output growth (Y) is explained by the residual A, (where a caret indicates percentage change): A 1% A A (8.7) Y, = Al + aK, + (I - a)LI. This residual is generally attributed to exogenous technoIogical progress, the onlyfactor that could generate the nondeclining rates of productivity growth found for industrial economies. Thus, dissatisfaction with the model mounted, intensified by its inability to explain other important aspects of economic development, such as why productivity levels and growth rates in developed and less developed coun- tries were not converging. Building on the neoclassical growth model, new endogenous growth theories have begun to consider the implications of exteralities and increasing returns and to treat the choice of technology, human capital, and labor supply as endogenous. Especially intriguing is the insight that economic policy can affect not only the level of output, as in the neoclassical model, but also the rate of growth. Following Romer (1987) and Helpman (1988), the models can be clas- sified into two broad groups: those in which economies of scale are extemal and those in which theyare internal. Among the furst group is Romer's model (1986) that makes technological change endogenous by assuming that pri- vate investment in physical capital increases the private stock of productive knowledge, which also becomes available to other firms. The externality associated with this technological spillover could overcome the diminishing returns to investment. It yields a production function of the forn: 156 EFFBis OF PRoTEcraoNIsM ON A SMALL COUNTRY: THE CAsE o; URUCUAY (8.8) Y, =B,K!bKtLl-a =B, KabL-. where B,is basic knowledge. that grows at an exponential rate and a + bis the output-capital elasticity, which differs from the share of capital. When a+ b is greater than or equal to 1, capital stockper capita can grow withoutbounds; with a fixed labor force the borderline case of a + b = 1 behaves like a model with a linear production function. Lucas (1988) presents a model in which private investment in human capital yields a similar externality and increas- ing returns to scale. The second group of models allows for fixed costs in the production of intermediate inputs (Romer 1987,1990), so that Rin equation 8.8 would be replaced by koX(i)Adi Productivity in the final goods sector depends on the range or number of middle products, which, together with the fixed costs of production, isakeyfactorin the appearance of increasing returns. Grossman and Helpman (1990,199laand b) explicitly model the introduction of new varietes of consumer goods as a result of calculations of the fixed costs of research-and development and future monopoly profits. Though the mod- els appear to be more relevant for industrial countries than for developing counties, the same economic principles that guide innovation could be applied to the adoption or adaption of technology by developing countries. These models suggest that some policies can affect growth rates (East- erly and others 1990). For instance, policies thatweaken incentives to invest in physical or human capital or to innovate tend to reduce growth. Rebelo (1990) suggests that polcies that alter the return of factors that can be pro- ducedwithout the use offixed factors affect the growth rate, whereas policies that alter the return of fixed factors can affect only the output leveL Some of these models suggest the possibility of a tradeoff between the negative static effectofsome distortionarypoliciesand theirpositive growth effecLK rugman (1988), for instance, shows how the positive static effect of removing tariffs could be lost if it results in a country specializing in sectors with slow uptake of technology change. The Grossman and Helpman (1990, 1991a and b) models show that trade liberalizadon can accelerate technological progress by increasing the size of the market available to technology producers and can boost the growth rate of countries that imitate the new technology. If trade policy shifts resources toward product innovation, it can accelerate growth (though the welfare resultsmaybe ambiguous). Butitcouldalso divert talented people into rent-seeking endeavors and slow the pace of innovation and imitation. - Though some of these endogenous growth models havetnorinative in- plications that may favor an active role for policy in stmulating growth, Helpman (1989, 183) stresses that 'identifying useful policies requires an understanding of market structure and conduct, entry constraints, inter- A long-run p epctive on trade poliop irlabiitA, 4nd gMWt/a 157 sectoral links, and the ]ike.... Meanwhile, policy should be designed on a case-by-case basis and-because good policies improvewelfare only slightly- no intervention (free trade) remains a good rule of thumb." Crass-section evidence These new growth theories have motivated a series of cross-section studies on the detenninants of growth. Forninety-eightcountries in the peiod 1960- 85, Barro (1991) shows that the growth rate of GDP per capita is positively related to initial human capitl and political stability and negatively related to tde inintal level of GDP, the share of govemment consumption in GDP, and the level of market distortions. Uruguay is an oudier in Barro's regres- sion, with a negative residual of -0.0295, resulting from an actual growth rate of 0.23 percent and a predicted growth rate of 3.18 percent; Rerunning Barro's regression for the sample of thirty-five countries se- lected in chapter 2, using a tade openness measure (the share of exports plus imports in GDP) to check for the impact of trade policy, shows trade openness to be positively related to growth (figure 8.2). Uruguay is still an outlier, but its predicted growth rate has been reduced to about 2 percent because of the low share of trade in GDP. Similar results are obtained when amore sophisticated measure of openness is used, such asPritchett's'(1991) structure-adjusted trade-intensity ratio, demonstrating the robustness of this relationship. Figure L2 Partial assodation between per capita growth and openness, I960-85 Per capito CDP growth rare 0.04 OHKG . 0.08 OscP 0.02 _ _ * wxCAN* 0.01 ECHE-A# SE NO LUE ESP* sU *ND * BEL MEXO 0 ~ZAPs * 0vEN -FIN OIRS E -0.02 0 20 40 so 100 120 140 160 180 Share of trade in GDP Notc Per capit growth is net of the vahue predicted by all explanatoryvariables except openness. Countries are as lied in dcapter 2 S&EJrJ Author's alcuktions, rerunning Bwrm's (1991) regression. 158 Enwwrn OF PROTECToNISM ON A SMALL CouNTnY: THE CASE OF URUGUAY The cross-section evidence covers the entire period 1960-85. To see whether Uruguay's performance improved after 1974, as the results in chap- ter 2 suggest, the following section examines time-series evidence. This ap- proach also has the advantage of allowing for better isolation of the role of policy. Evidence on scale and the role of policy The traditional growth models assume constant returns to scale and consider Uany kind of shift in the production function" (Solow 1957, 312) to be the result of technological change. Thus, in per capita terns (lower case letters are in per worker tenns): (8.9) y,= At + ak. FollowingRomer (1987) and Baldwin (1989), the data are allowed to 'speak for themselves" before the elasticir' of outputwith respect to capital or labor is forced to equal each factor's share in GDP. An Andex of employment was constructedfor Uruguay, using the mostsensible measures-available for each time period (see appendix). A capital stock series was also constructed, us- ing the perpetual inventorymethod (see Harberger and Wisecarver 1978 and CXniara de Comercio 1986). When the index of GDP per worker is plotted against the index of the capital stock perworker, the capital stockindex seems to follow the GDP perworker index much better than when the assumption of constant returns to scale is applied (represented by the lower plotted line in figure 8.3, where 0.46 is the share of capital in national inconme). A regres- sion of equation 8.9 yields a coefficient of 0.80, which is larger than the 0.46 share but not as large as estimates for industrial countries reported in Romer (1987) and Baldwin (1989). A more direct way to test for constant or increasing returns to scale is to run a regression on an equation like 8.7 to see whether the sum of the coef- ficients equals one. Four regressionswere run (table 8.2): regressions RI and R2 use a measure of capital stock adjusted for utlization; regressions RS and R4use the unadjusted index. In the absence of a true measure of utilization, the raw capital stock figures were adjusted by the share of the labor force unemployed in each year (as in Solow 1957), which tends to account for cyclical movements. Except for this adjustment, data limitations precluded making any of the other adjustments usually made in growth accounting (as for education, age, and gender of the labor force, vintages of capital goods, structurl change; see Maddison 1987). For the two regressions with the adjusted index (RI and R2) the estimates of the elasticity of output to capital are dose to the average estimate of the share of capital in national income A long6n pmpective on Irade poEfy, instability, and grww& 159 Figure 8.3 Capital, labor, and productivity, 1955-89 Index(1955=100) ISO 170 160 GDP per worker ISO 140 130 120 ."tstv t ll 100 90 1955 .1960 1965 1970 1975 1980 1985 Source: See appendix. Table 8.2 Estimates of production function aozY,=B1,gP + 21ogL, + 3TWME) RI R2 R3 R4 Di 0.495 CA 16 0.373 0.339 (2.81) (2.22) (2.01) (1.73) B˝ 0.758 0.804 0.976 0.927 (2.33) (2.56) (2.88) (2.79) 13 0.014 0.015 (4.04) (432) AR(I) 1.239 1.117 1316 1.157 (6.82) (6.46) (7.27) (7.01) AP(2) -0.266 -0.407 -0.338 -0.484 (-1.47) (-2.35) (-1.87) (-2-85) R2 (adjusted) 0.97 0.97 0-96 0.97 DW 1.84 1.84 :.75 1.80 f-staistic 322 284 288 273 Esrin'tes of scale ecanoies C:entralestimTate - 1.253 1.21!0 .349 1.266 95% confidnce interval (0.931, 1.576) (0.961. 1.479) (1.004, i.694) (0.988, 1.544) Saurce: Author's calcultions; see append. 160 EFFECas OF PROTECnONISM OrN A SMALL CouVNIY: THE CASE OF URUGUAY (0.46).' Estimates mith the other two regressions are smaller, than that share, but they are more in line with those for industrial countries (see Maddison 1987, table 9). The estimates of the elasiicity ofGDP to laborare much larger than the corresponding share of labor in national income. Time trendswere highly significant (R2 and R4), suggesting technological progress at an ex- ponential rate of 1.4 to 1.5 percent a year. Tests of the null hypothesis of constant returns to scale-coefficients on capital (f1,) and labor ([2) inputs equal to one-tend to suggest the presence of increasing retums to scale (coefficients total 1.22 to 1.35; table 8.2). Nevertheless, the hypothesis of constantreturns to scale could not be rejected at the 95 percent confidence level, except for R3. A key theme in the new theories of trade and growth is that trade in- duces more specialization and so allows for the exploitation of economies of scale. These economies could be due to spillover effects across finns, indus- tries, or countries. Backus, Kehoe, and Kehoe (1990) tiow thatwhen spillovers operate at the industry level with little spillover across industries, aggregate growth is a simple fimction of the product ofa specalization index and GDP. Thatvariable reflects the effect of economy size on growth and the intuition thatfor agiven size a higher level of spe-cialization leads to more learning by doing and so to more growth. Specialization indices were constructed for the Uruguayan manufctur- ing sector based on data from the 1968 and 1978 census and the 1987 Survey on Economic Activity, where the specialization index is defined as, (8.10) S = wher e the term in brackets is the shz e of each industrial sector in total value ad(.ed (at the three-digit industrial classification level). Uruguayan industry ha's increased its degree of diversification with the specalization index fall- ing from 10.1023 in 1968 to 0.0817 in 1978 and 0.0786 in 1987. Since less specialization slows down leaing by doing, the result according to this dass of models, is reduced growth potential in the manufacturing sector. In sunmary, the evidence on increasing returns at the aggregate level is not clearcutL Nor are the implications clear of the new endogenous growth theory for trade policy. Nevertheless, several studies have shown that export growth leads to higher total productivity growth through the exploitation of economies of scale, technology transfer, and increasing competitiveness (see Feder 1983; Nishimizu and Robinson 1984; de Melo and Robinson 1992). In the case of Uruguay, calculation of the Solow residuals (A), or total factor productivity (see details in the appendix), shows dLat changes in productiv- ity are correlated wit export growth but not with changes in capital per worker (see figure 8.4): A lorog-nn perspecliue on trade policy, iistability, and grpvwth 161 At = 0.164k,,+ 0.073+ (0.97) (2.74) R120.13 DW=1.54 F=4.48 The statistical analysis suggests that export externalities help explain Solow residuals whereas increases in the capital stock-as suggested by the Romer (1986) model-have no explanatory power.2 Finally, it is interesting to see what role trade policy played in Uruguay's export growth. In 1974 Uruguay shifted from an import-substitution strat egy to export promotion. progressively lowering tariffs (the average nomi- nal rate of protection was 453 percent in 1974; see Favaro and Spiller 1987) and providing first subsidies and then tax rebates for exports. These policy changes are captured in a trade policy index (TPI), defined as the ratio of subsidies to the average nominal tariff (see the appendix for details). An increase in the value of the index means an increase in the relative price of exportables to importables. Though a crude measure of trade policy (for example, it does not include the effect of reference prices), this measure captures the major changes after 1974. Though the share oftraditional exports (basically meat, wool, andhides) in total exports fell substantially, From more than 75 percent before 1970 to Figure 8.4 Exports and productivity, 1955-89 Export volume index (1981=100) Factorproduciviyindex (1957=100) Capitcl per worker index (1 9SS=100) 140 250 130 130 M~~~~~~ota facto?r t / \> 2D0 . ~~~~~producik 120- 1) 110~~~~~~~~~~~~~~~~~~~5 - 4, ~~~~~~~~~~~~~~~~100 - C0 A. mevoum 100 ~ ~ ~ ~ ~ %.~ Export vlm 90 50 1955 1960 1965 1970 1975 1980 1985 Sorre See appendix. 162 EFECIsa OF PROTECToNISM ON A SMALL CouNwrm THE CASE OF URUGUAY less than 50 percent after 1975, the share remained importantL Because tra- ditidnal exports are especially subject to price fluctuations, the following regression attempts to measure the importance of changes in trade policy against changes in the terms of trade (TOT) in the evolution of the volume of exports: log XA = 5.72 - 0.267 log TOT+ 0.007 J+ 0.084 PTAl + 0.21 PTA2, (7.61)(-1.76) (4.17) (0.82) (2.72) R2 Adj. = 0.90 DW = 1.60 F = 63.7 where X is export volume and PTA1 and PTA2 are dummy variables for the preferential trade agreements with Argentina and Brazil, and for tleir ex- pansion in 1985-86. The results indicate an important role for the trade poliqr variable and for the preferential trade agreements after their expan- sion, but not before. The terms of trade have an unexpected negative sign. The strong deterioration of the terms of trade in 1974 required an impor- tant adjustment in the level of exports, but it also seems to have had an im- pact on the rate of growth of exports (figure 8.5). These may be ansitory growth effects, however, rather than a permanent change in the steady-state rate of growt.h. Figure 8.5 Trade policy and exports, 196049 Exprt volume index (1961 = 100) Trade policy index lrms of trade index (1978=100) 90 250 80 225 Termsof TofedoIjc 70 trade 2ndex 200 60 175 50 40 ume 125 30 20 4311* t l, 10 75 1960 1965 1970 1975 1980 1985 Note: The trde poliy index is the raio of subsidies to the average nomina Wiff. Sow'ce: See apperndkc. A long-runperpeaiveon tradeapoicy, instabity, and grwth 163 These results are consistentwith those ofFavaro and Spiller (1991), whose examination of the impact of trade liberalization at the sectoral level showed increases in producdvity and a sectoral relocation of factors, especially dur- ing 1974-78. These achievements were pardally reversed in 1979-82, how- ever, because of a real appreciaton of the currency induced by increased capital inflows; the exchange rate-based stabilization program, and rising aggregate demand due to Argentine tourists. Viana (1990) decomposes the sources of growth from the demand side and finds that exports are an im- portant determinant of growth between 1974 and 1978, but that internal demandis the tnain determinant of growth from 1979 to 1987. At the sectoral level he finds that between 1984 and 1987 the most dynamic sectors were those that had not previously produced for export but that were exporting to Argentina and Brazil foilowing extension of the preferential tade agree- ments. The increases in productvity induced in other sectors by the growth of intraregional exports are much smaller than those induced by exports to the rest of the world-which Favaro and Sapelli (1989) and Kaplin, SarAchaga, and Vera (1988) attribute to the preferential trade aWeements being merely an extension of the import-substitution strategy to "captive" markets in the neighbor countries. These findings cast doubt on the prospects for contin- ued productivity growth through export growth. Export instability and growth The big export-promotion push ofthe second halfof the 1970s was discrimi- natoiy, granting higher subsidies to the more protected sectors (Favaro and Sapelli 1989,157). Ihis bias was reinforced by the preferential trade agree- ments. The resultwas a substantial increase in trade flows within the region. Driving this policy was a. desire to diversify Uruguayan exports. Critics of outward-oriented policies (see Yotopoulos and Nugent 1978) have suggested that small open economies would end up with exports concentrated geo- graphically and by commodity, thereby increasing the risks of instability and secular decline in the terms of trade. Yet diversification of exports through increased regional exports has involved more instability in export income rather than less (Roldos 1989). Exports to the region are strongly correlated witi exports to the rest of the world, and so expanding regional exports increases the instability of aggre- gate export income. Because of wide fluctuations in the real exchange rate with neighboring countries, swings in export income from the region are more unpredictable and frequent than fluctuations in the prices of its natural-resource-based exports to the rest of the world. The negative effect of increased umcertaintyon capital accumulaton andgrowth wod thus have to be added to the static effects of the preferential trade agreements. .164~ EFnCTs OF PNOTECTIONISM ON A SM~ALL CouNqTgv: THE CAsE OF URUGUAY Following Yotopoulos and Nugent (1978, 331), export instability indi- ces were constructed for export sectors and destinations: (cbogx, - ogkY) (8.11) logX n where X, is the value of exports, X the mean value, and Xthe trend compo- nenL The instability ofexportincome is much higher for the flows to Arn'en- tina and Brazil than for those to the rest of theworld, despite an annual growth rate of exports to the region double that to the rest of the world from 1962- 65 to 1987-90 (table 8.3). Over the more recentpast, performance has dete- riorated with Argentina and has improved substantially with Brazil, so that together, the rate of growth of exports to the region is similar to th2t to the restof dteworld. Considering the expansion of the preferential trade agree- nents, greater growth ofregional exportswould be expected, suggesting that macroeconomic instabilityin the region might have reduced export growth. To analyze the impact of export instability on GDP growth, the stochas- tic trends of each sector's exports were estimated using quarterly data for a sample of the main industrial exports from 1975 to 1990. Adequate datawere available fornineteen sectors. To calculate the instabilityindexfor each sector, log X, was estimated as the permanent component of a Beveridge-Nelson (1981) decomposition. (The methodology aswell as the main properties of the series are reported in the appendix.) Annual growth rates of the output volume index were also calculated for each industry between 1975-77 and 1987-89. Indices and rates were weighted by the share of each sector in in- dustrial value added. A cross-section regression of the output growth rates on the respective export instability indices yields a negative but statistically insignificant coefficient. Sectors that export outside the region havealmnost double the growth rate of output of those thLat export to the region and face half the export instability (table 8.4). Regardless of the crudeness of this measure, it does suggest a negative inpact of regional instability on output growth, the result ultimately of discriminatory export-promotion policies. Table 8.3 Instability and export growth Argentina Brazil Rest of the world Instability index 59.1 49.1 15.4 Expo,t grMlh 1987-90/1962-65 15.7 18.5 7.6 1987-90/1977-80 2.1 8.7 6.4 Note: Instability indices dokted according to equation 8. 1 1, wth a detemiristic trend between 1962 and 1990; exportgrowiL' is the annualized growLh rate between averge of samnple extremes. SowruAuthor's calcubtions. A long-n pmpective on itrae policy, instability, and gnh 165 Table 8.4 Export instability and output growth for main industrial exports Industry code Industry Output gtowth' Export irlobitityb Exports outside the regon 3111 Meatpacking -1.08 3.97 13114 Fishing 1.74' 1.35 3211 Textiles 5.67 131 323[ Leather 1.67 3.12 3240 Footwear -1.18 202 Average 2125 2-48 Exports to the region 3112 Dairy products 4.73 7.56 3115 Oil 0.28 4.46 3116 Rice 2.62 5.33 3133 Breweries 0.17 2.50 3411 Paper 0.95 5.97 3512 Chemicals 0.77 5.44 3513 Plastics. 0.72' 4.63 3521 Paints 3.27 9.52 3551 Tires -0.35 4.33 3620 Glass 0.56 10.53 3710 Basic metaJs -130 19.98 3820 Machinery, norKlectric 5.73 5.43 3830 Machinery, electric 136 1.92 3843 Automobiles and pins 0.18 4.92 Average 1.21 5.57 a. Annuaked growth rates at output (NIF) between the averages for 1987-89 and 1975-77 (or 1978, the base of the index, frthe sectors vwith an asterisk). b. Instabilty index defined in equation 8.11, wIth log equal to the stochastic tend in a Bevericigels-on deconposition. Source: Authar's caulations; see appendix. Conclusion Tests for the effects of trade policies onrgrowth over the period 1960-85 us- ing cross-sectional evidence for a sample of thirty-five countries confinn, as expected, thatopenness is positively correlatedwith growth. When per.capita GDP growth for this sample, net of the value predicted by all other explana- tory variables, is plotted against openness, Uruguay is an outlier, with actual growth about 2 percentage points below predicted growth. These expected findings from the cross-sectional evidence did not carry over as clearly to the evidence based on time-series data for Uruguay. First, no evidence was found of temporal causality from exports to GDP (or vice versa), confirming the results of Jung and Marshall (1985) for a large sample of countries. These results cast some doubt on the direct effects of export-promotion polides on growth, though the results may be influenced by the period of strong but temporary growth due to import-substitution 166 Emen OF PROTEmONISM ON A SMALL CouNTRY: THE CSE OF URUGuAY policies and the long period of stagnation in 1955-74 and the recession of 1982-84. It is likely that the important positive correlation between contem- poraneous exports and GDP growth was due to cyclical rather than to growth factors. Second, the evidence on scale economies was mixed. A key theme of the new growth theories is that trade induces greater specialization, which al- lows for the exploitation of economies of scale. Though the sum of the co- efficients on a Cobb-Douglas production function totaled more than 1 (1.22 to 1.35), the hypothesis of constant returns to scale could not be rejected at a 5 percent level. Further, greater diversification rather than greater spe- cialization -was found for the industrial sector, suggesting a reduction in growth potential since less specialization slows learning by doing. Solow technology residuals were found to be correlated with exports but not with capital per worker, confirming the simulation results of de Melo and Robinson (1992) on the potential for a neoclassical model with export ex- ternalities to account for the stylized. facts of trzde and growth in develop- ing countries. Though the shift to export-promotion policies in 1974 had a big push effect on exports, it is not clear whether the growth in 1974-80 reflected a permanent change in the rate of growth or transitional dynamics of adjust- ment to substantial deterioration in terms of trade. The discriminatory bias ofr de subsidies to exports and the preferential trade agreements thatinduced thatbig push castdoubt on the alternative of apennanent change in growth rate. These policies generated substantial trade flows within the region, yet sectors exporting to the rest of the worldwere found to have higher growth rates and less export instability. Thus, Uruguay's export-promotion strategy- better described as a regional extension of the import-substitution strategy- had not only a static distortionary effect but also a negative effect on growth thiough increased instability. To expand exports widely and boost the long- run GDP growth rate, Uruguay would need Eo reduce the discriminatory bias-by sector and destination- of its export-promotion policies. Appendix. Data sources and assumptions The data on the capital stockwere constructed by Harberger and Wisecarver (1978), using the perpetual inventory methodology, and updated in Cimara de Comercio (1986). We extended the series to 1989. The share of capital in national income was also calculated based on those studies. It includes inter- est income, dividends, and rents from the national income accounts, with a deduction for the imputed income of owners, family members, and inde- A kang'n pempective on trade pdiyq, instabily, and growth 167 pendent workers. It was not possible to construct extended series on those imputedincomesfor the agricultural sector, so the share has an upward bias estimated at 4.8 percent for 1978. The employment index was constructed from the best sources available. For 1965 to 1989, we used the number of workers in Montevideo (source: Insituto de Estadistica, Facultad de Ciencias Economicas y Administraidn, andDireccion General deEstadisticayCensos);we interpolated 1964, using data for 1965 and that from the 1963 census. We projected the series from 1962 to 1955 using the annual rte of growth of the sum ofemploymentin manufacturing (source: Ciman de Industrias, Departamento de Estudios Economicos) and in the public sector (source: constructed from data from the OficinaNacional de Servicio Civil). The data on unemployment, which were used as a proxyfor capital udlizadon (as in Solow 1957), were from the same sources for 1963 to 1989, with the avenge for 1962-68 used for the period 1955-61. Table 8A.1 presents the series on the share of capital, output and capital per worker, as well as the Solow residuals (Solow 1957, table 1). The tade policy index (TPI) was calculated as the ratio of one plus the subsidy rate diided by one plus the tariff rate times 100. The subsidy rate was calculated as the ratio of certificates issued over nontraditional exports. (Sources: C(mama de Comercio 1979; Departamento de Estadios Econ6micos, for 1961-77; GINVE 1983 for 1978-82; CINVE 1986 for 1983-85; for the pe- riod 1986-89, indirect taxes rebates from Direccion General hnpositiva.) The same could not be done with the tariff rate, because the implicit tariff on actual imports was very low due to prohibitive rates. We took estimates on a sample of products presented in Favaro and Sapelli (1989, cuadro 5.5), plus the ones in CINVE (1986) for 198586, and the average of the five rates of the structure for 1986-89, and interpolated linearly the missing years. Though a crude measure of trade policy (it disregards quotas, reference prices, and other nontariff barriers), it seems to capture the main shifts in policy. The instability indices for the sectoral exports were calculated accord- ing to the definition in equation 8.10. Whenever appropriate, a stochastic trend (instead of a deterministic one, see Stock and Watson 1988) was con- structed from the currentvalue of exports. That stochastic trend is the per- manentcomponentofa Beveridge-Nelson decomposition of the series. Miller (1988) shows how that permanent component can be generated from an autoregressive filter obtained from the first difference of the series, then applied to the levels of it. The ARMA processes used to generate that AR filter, and the stochastic trend, are presented in table 8A.2. 168 EmcnS OF PROTEcrIONISM ON A SMALL COUNTRE THE CASE OF URUGUAY Table BA. I Calculation of Solow residuals Year KSHARE YL K2L DLA A 1955 0.469000 27.24155 79.19048 0.000000 1.000000 1956 0.469000 27.37893 80.68803 -0.003687 0.996000 1957 0.475000 27.71300 82.17400 0.003465 1.000000 1958 0.463000 27A2683 82.43847 -0.011919 0.988000 1959 0.515000 27.08654 82.97582 -0.015898 0.972000 1960 0.511000 28.43941 .83.78407 0.042641 1.014000 1961 0.434000 29.95666 84.80900 0.045403 1.060000 1962 0.395000 28.88775 85.76386 -0.041400 1.016000 1963 0.393000 28.68639 85.97275 -0.007974 1.003000 1964 0.401000 29.28125 87.46797 0.013460 1.021000 1965 0.435000 29.71676 89.13491 0.006520 1.028000 1966 0.420000 30.61887 89.64286 0.027083 1.056000 1967 0.397000 2971424 91.12557 -0.036904 1.017000 1968 - -0.419000 30.26591 89.17521 0.027391 1.045000 1969 0.395000 32.10879 89.358 I5 0.056586 1.104000 1970 0.383000 3359739 91.30727 0.036131 1.144000 1971 0.363000 33.45327 91.30646 -0.004305 1. 139000 1972 0.396000 33.73450 93.28669 -6.94e-05 1.139000 1973 0.417000 34.10098 92.07607 0.016230 1.157000 1974 0.434000 35.40464 93.32452 0.031016 1.193000 1975 0.447000 3738078 94.27297 0.048368 1.251000 1976 0.455000 38.46B80 90.94421 0.044937 1.307000 1977 0.485000 38.45769 93.78143 -0.014962 1.288000 1978 0.490000 40.48014 99.15085 0.023426 1.318000 1979 0.543000 43.59496 107.7780 0.027984 1.355000 1980 0530000 46.45474 114.7044 0.029557 1.395000 1981 0.516000 47.88527 121.8709 -0.000469 1.394000 1982 0.493000 46.09008 126.3266 -0.056338 1.316000 1983 0.547000 43.72235 121.4180 -0.032040 1.273000 1984 0.555000 42.54190 121.7462 -0.029244 1.236000 1985 0.506000 41.63513 118.6048 -0.008377 1.226000 1.986 0.493000 44.22036 120.1909 0.051956 1.289000 1987 0.493000 45.69784 118.1523 0.040838 1.342000 1988 0.480000 46.14522 119.5471 0.004095 1.348000 1989 .. 47.19024 120.9489 Not availabit Mote: iHSARE is capil as a share ci income. Y is output per worker, K2L is catal perworker. and DIA ard A are Solow reiduals. SDurc See appendix text A lng-nrunppetdive on irade polcy, instability; and gswtk 169 Table 8A.2 ARMA models for sectoral exports Iduscy cod . Model Qsws}c 3114 (I + 0.462L) Ax, 10.3 (3.97) 3211 (I + 0.447L + 0.476L2 + 0.392L3) Ax, ; 8.3 (3.53) (3.79) (3.12) 3240 (I + 0.555L + 0.252L2 + 0.12L3 - 0.214L") Ax,= 13.1 (4.18) (1.85) (0.91) (1.66) 3112 A=x = (1 -0.692L) i; 15.7 (5.41) 3115 (I + 0.343L + 0.3 LI Ax=s; 15.7 (2.53) (2.28) 3133 (I +0.506L+0.551L2i+0.203L')Ax, = 12.8 (4!56) (5.47) (1.84) 3411 Ax, = (i - 0.533L) ; 16.0 (3.66) 3512 Ax, = (I -0.501L); 7.6 (3.86) 3551 (I + 0.282L + 0.255L2 + 0.235L)Ax, = Et 9.1 (2.14) (1.91) (1.78) 3620 (I + 0.634L + 0.522L2 + 0.222L3)Ax, =; 23.9 (4.77) (3.71) (1.68) 3710 Ax, (I -0.314L-0.387L'); 5.7 (2.37) -91) 3820 Ax, (I -0.52L); 11.4 (4.03) 3830 (I +0395L)Ax,=; 8.9 (3.24) 3843 Ax, = (I - 0465L) 16.1 (3.52) Note:Ax stardsforefirstdifferncecf(ocg)exportsoftherespectivesector.Seors3111.3231.3116.3513. and 352 ( are not ricuded here beouse a determinisc trend vwS more appropriate fan a nodiasc one. SuDe: See appendix txL 170 ErFEcrs OF PROTCrIONISM ON A SMALL COUNTRY: THE CAsE OF URUGUAY Notes I wish to thankLuisViana forinterestingdiscussions,Adriana Cassoni, Michael Connolly, Jaime de Melo,William Easterly, and MarcelVaillant for their comments on a previous draft, and Dolores Benavente and Carlo Graziani for extending and updating the data. All errors remain my own responsibility. 1. This may overcstimate capital's share since we were unable to obtain estimates of income inputed to independent and family workers in the agricultural sector. We estimate this bias to be less than 5 percent of national income (see appendix), 2. Changes in import volumes-the import externality suggested by de Melo and Robinson (1992)-did not have any explanatory power when added as a regressor. 3.Thisisimportantbecause, asshown byStockandWatson (1988) and Cuddington and Urznua (1989), the removal ofa deterministic time trend can alter substantially the nature of the remaining cyclical component of the series. References Anichinij.J.,J. Caumont, and L. Sjaastad. 1978. LaPolitica Comercial y la Protecci6n en el Uruguay." Banco Central del Uruguay. Montevideo. Backus, D., P. Kehoe, and T. Kehoe. 1990. "In Search of Scale Effects in Trade and Growth. Journal of Economic Theory 58: 377-409. Baldwin, R. 1989. 'On the Growth Effects of 1992.7EconomicePolicy9: 247-81. Barro, R. 1991. "Economic Growth in aCrossSection of Countries." QuarlyjounIaof &conomics 106(2): 407-44. Beveridge, S., and C. Nelson. 1981. 'A New Approach to the Decomposition of Eco- noomic Time Series into Permanent and Transitory Components with Particular At- tention to the Measurement of Business Cycles."Journal of Monetary Economics 7: 15r1-74. Bhagwatij. 1986. "ExportPromotingTradeStrategies: Issues andEvidence."Worid Bank Disussion Paper, Development Policy Issues Series no. VPERS7. Washington, D.C. Cdxnara Nacional de Comercio, Departamento de Estudios Econ6niicos. 1979. "Comercio exterior del Uruguay, 1961-1978." Montevideo. . 1986. Publicaci6n no. 17. Montevideo. CINVE (Centro de Investigaciones Econ6micas). 1983. "Industria yprotecci6n en un contexto de apertura externa: Uruguay 1978-1982." Montevideo. 1 1986. "Estudio sobre la reestructura de la protecci6n arancelaria.' Montevideo. Cooley, T., and S. LeRoy. 1985. "Atheoretical Macroeconomics: A Critique.Journal of Monetary Economics 16: 283-308. Cuddington,J., and C, Urzua. 1989. 'Trends and Cycles in the Net Baner Terms of Trade: A New Approach." Eonomcfourna199: 426-42. Easterly, W., Ri King, R. Levine, and S. Rebelo. 1990. "Do National Policies Affect Long-Run Growth?' World Bank, Policy Research Department, Tra-nsition and Macro- Adjustment Division, Washington, D.C. Engel, R., and C W.J. Granger. 1987. "Co-integration and Error Correction: Repre- sentation, Estimation and Testing." Exonometzica 55(2): 251-76. Favaro, E., and P. T. Spiller. 1991. 'Thc Timing and Sequencing of aTrade Lberaliza- tion ELcperiment: Uruguay 1974-1982." In D. Papageorgiou, M Michacly, A Choksi, eds., LiberaizingForeign Trad, vol. 1. Washington, D.C.: World Bank. A long-prn perspective on trade polig, instability, and growth 171 Favaro, E., and C. Sapelli. 1989. 'Promoci6n de e:cportacioncs y crecimiento econ6mico." Centro de Estudios de la Realidad Econ6mica y Social, Centro Internacional pam el Desarrollo Econ6mico. Montevideo. Feder, C. 1983. "On Exports and Economic C rowth. Journal of Dueelopmt Economics 12: 59-7.- Gewcke,J. R, R. Mecse, and W. Dcnt. 1985. ComparingAltcrnative Tests of Causality in Temporal Systems."Journal ofEconometrics 21: 161-94. Granger, C. W. J. 1969. "Investigating Causal Relations by Econometric Models and Cross-spectral Methods." Econometrica 37: 424-38. Granger, C. W.J., and P. Newbold. 1986. ForecstingEconomic Time Series. 2nd. edition. San Diego, Calif;: Academic Press, Inc. Grossman, C. M., and &- Hclpman. 1990. 'Comparative Advantage and Long-Run Growth.' American Economic Revitw 80: 796415. .1 991a. "Endogenous Product Cycics."EconomicJoumnal 101: 1214-29. . 1991b. "Growth and Welfare in a Small Open Economy." In E. Helpman and A. Razin, eds. Inernational Trade and TradtPoliy. Cambridge, Mass.: MIT Press. Harberger, A., and D. Wilsecarver. 1978. 'Tasas de Retorno al Capital en los Ambitos Privado y Social en el Uruguay.' Banco Central del Uruguay. Montevideo. Helpman, E. 1988. "Growth, Technological Progress and Trade.' Austrian Economic Papers 1: 5-25. , 1989. The Noncompetitive Theoryof InternationalTrade andTrade Policy." Proceedings of the World Bank Annual Conference on Delpament Economics, Washing- ton, D.C. Jung, W. S., and P.J. Marshall. 1985. "Eorts, Growth and Causality in Developing Countries."Joumnal of Development Economics 18: 1-12. Kaplin, M., D. Sarachaga, and T. Vera. 1988. "Integraci6n, Politica Comercial y Comercio Intra-industrial: Evidc. a4s pan la Valoraci6n dei Comercio Intrarregional." Reuista deEconomia S%3). Montevideo: Banco Central del Uruguay. Krugn-an, P. 1988. 'he Narrow Moving Band, the Dutch Disease, and the Competi- tive Consequences of Mrs. Thatcher Notes on Trade in the Presence of Dynamic Scale Economies Journal ofDevelopmew Econoomics 27: 41-55. Lucas, R. E. 1988. "On the Mechanics of Economic Development"Journal ofMonetary Economics 22: 3-42. Maddison, A. 1987. 'Growth and Slowdown in Advanced Capitalist Economies: Tech- niques and Quantitative Assessment."Journal ofEconomic Literature 25: 649-98. Melo,J. de, and S. Robinson. 1992. "Productivityand Externalities: Models ofExport-Led Growth."Journal oflntemnational Trade andDevelopment 1: 41-69. Michaely, M. 1977. "Exports and Growth: An Empirical Investigation."Journal ofDevel- opmentEconomics4(1): 49-53. Miller, S. M. 1988. 'The Beveridge-Nelson Decornposition of Economic Time Series." Journal of Monetary Economics 21: 141-42. Nishimnizu, M., and S. Robinson. 1984. 'Trade Policy and Productiviy Change in Semi-Industrialized Countries."Journal ofDevelopmet Ecmomics 16. Pritchett, L. 1991. "Measuring Outward Orientation in Developing Countries: Can It Be Donc?" Policy Research Working Paper 566. World Bank, Trade Policy Division, Washington, D.C. Rebelo, S. 1990. 'Long Run PolicyAnalysis and Long Run Growth."Jornal ofPolitical Economy 99:500-21. Romer, P. 1986. Increasing Returns and Long-Run Growth."JournalofPdliticacEconomy 94: 1002-37. .172 Emrcrs OF PoarploNosw ON A SMALL COUNTRY: THi CASE OF URUGUAY . 1987. "Crazy Explanations for thc Productivity Slowdown." NBER Macro- economics Annual 2: 163-202. ____, 1990. "Endogenous Technical Change." Journal of Political Economy 98(5): S71-102. Roldos, J. 1989. "Export Instability and Growth." Centro dc Estudios de la Realidad Econ6mica y Social. Montevidco. Solow, R. 1957. 'Technical Change and the Aggregate Production Function."Reviewof Economics and Statistics 39: 312-20. Stockj. H.. and M. W. Watson. 1988. "VariableTrends in Economic Time Series."Jour- naI offenrinic Pe-sFkaives. 2:1 47-74. Viana, L. 1990. "Los Efectos dc los Cambios en el Nivel de Protecci6n sobrc las Exportaciones." Paper presented to thcjornadas de Economia, Banco Ccntral dcl Uruguay. Montevideo. Yotopoulos, P., andJ. Nugent. 1978. Economics of Development: Empirical Investigations. New York: Harper and Row. T HE W OR LD B AN K Other World Bank Regional and Sectoral Studics Education in Asia: A Comparative Study of Cost and Financing, Jec-Pcng Tan and Alain Mingac Healt Carc in Asia: A Comparativc Study of Cost and Financing, Charles C. Griffin Bolivia-'s Answer to Poverty, Economic Crisis, and Adjustment: T'he Emer,gency Social Fund, edited by Steen Jorgensen, Margaret Grosh, and Mviarkc Schactcr Commodity Price Stabilization and Policy Reform: An Appropac to the Evaluation of rhe Brazilian Pricc Band Proposals, Avishay Braverman, Ravi Kanbur, Antonio Salazar P. Brandao, Jcffrey Hammcr, Mauro dc Rezendc Lopes, and Alexandra Tan Crop-Livestock Interaction in Sub-Saharan Africa, John McIntire, Danicl Bourzat, and Prabhu Pingali The Transition from Socialism in Eastern Europc: Domestic Restracturing and Forcejn Trade, edited by Arye L. Hillman and Branko Milanovic Women's Employment and Pay in Latin America: Overview and Metlsodolqgy, Georgc Psacharopoulos and Zafiris Tzannatos Case Studies on Women's Employmcnt and Pay in Latin America, edited by George Psacharopoulos and Zafiris Tzannatos Rapid Appraisal Mcthods, edited by Krishna Kumar Administring Targeted Social Programs in Latin America: From Platitudes to Practice, Margaret E.Grosh Russia and the Chbaleng of Fiscal Fedcralism, edited by Christine I. Wallich : *: