The Welfare Effects of Restrictions on U.S. Trade

STEPHEN P. MAGEE* Universityof Chicago

RECENT ECONOMIC AND POLICY DEVELOPMENTS in the areas of stabiliza- tion, allocation,and distribution'help explain the currentpolitical mood in some parts of the United Statesin favor of greaterprotectionism. In the area of macroeconomicstabilization, anti-inflationary policies in 1968-71 led to an undesirablyhigh rate of unemployment.For a number of reasons,these had only limitedsuccess in reducingthe rate of inflation at the hoped-for speed; the result was the price controls imposed on August 15, 1971. The inflation, an increasinglyovervalued dollar, and businesscycle developmentshere and abroadplaced pressure on both our

* I am grateful to Robert Z. Aliber, Arturo Brillembourg,Stephen Easton, Kevin Ferrell,Stanley Fischer,Frank Flatters, William Hart, Dan Heath, ChristineHeckman, Arthur B. Laffer, Carol Nackenoff, Carlos Rodriguez, Osvaldo Schenone, Andrew Schmitz,members of and senior advisersto the BrookingsPanel on Economic Activity, the staff of the Council on InternationalEconomic Policy and other U.S. government agencies,and membersof the Workshopin InternationalEconomics at the Universityof Chicago for assistanceand comments on this paper.I am also indebtedto the National Science Foundation for financialsupport. The views expressedhere are those of the author and do not necessarilyreflect the views of the aforementionedagencies or indi- viduals.The author is responsiblefor errorsthat remain.This paper drawsheavily from three unpublishedworking papers by the author: No. 14, "Tariffsand U.S. Trade,"A Study for the Council of Economic Advisers (June 1972); No. 15, "Trade Liberaliza- tion," A Study for the Council on InternationalEconomic Policy (July 1972); and No. 16, "The WelfareEffects of TariffReductions in the United States"(August 1972). 1. In The Theoryof PublicFinance: A Studyin PublicEconomy (McGraw-Hill, 1959), Richard A. Musgravestresses these areas as a logical division of fiscal policy. 645 646 BrookingsPapers on EconomicActivity, 3:1972 abilityto export and on U.S. industriesthat compete with imports, de- pressingthe U.S. merchandisetrade balancefrom a $3.8 billion surplus in 1967to a $2.7 billion deficitin 1971. Second, over the past five years, allocationpolicy in the international area has changed.The KennedyRound tariffcuts were stagedto reduce U.S. and foreignnonagricultural tariff rates by 35 percentover the period 1968-72,and in a periodof increasingovervaluation of the dollar,helped ease the pressureon the U.S. export sector. However, in the import- competingsector, the U.S. tariffcuts servedonly to createadditional strain. While they were rational on allocationalgrounds, and relativelysmall (probably involving only a 3- to 5-percentage-pointeffect on import prices),their timing unfortunately coincided with an increasein the aggre- gate unemploymentrate from 3.6 percentin January1968 to 5.9 percent in January1972. Third, distributionpolicy failed to neutralizethe shifts occurringin the internationalsector. The Trade Expansion Act of 1962provided that parties injuredby trade liberalizationcould receiveeither adjustmentassistance (technical,financial, tax) or importrelief through an escapeclause, or both. Until recently,because of the wordingof the act, the U.S. TariffCommis- sion has found it difficultto determinethat injuryto domesticparties was caused "in majorpart" by tariffconcessions.2 The resulthas been an in- creasing use of "voluntary"restrictive agreements between the United Statesand.countries exporting some sensitiveproducts to us. Nevertheless, if the increasedpolitical activities of protectionistsare any gaugeof the income losses of import-competingindustries, the TradeExpansion Act of 1962 and subsequentpolicy have failed to compensate adequatelyfor

2. See, for example,Tracy W. Murrayand Michael R. Egmand,"Full Employment, Trade Expansion and Adjustment Assistance," Southern EconomicJournal, Vol. 36 (April 1970),pp. 404-24, for a discussionof these difficulties.In "Policy Problemsin the AdjustmentProcess (U.S.)" (paper deliveredto the Seminar on Industrializationand Trade Policies in the 1970s, InternationalBank for Reconstructionand Development, October 1972; processed),Robert E. Baldwin and John H. Mutti note that from Octo- ber 1962 to November 1969, all twenty-sixpetitions to the U.S. Tariff Commissionby industries,firms, and workers were denied. From November 1969 to March 1972, 16 out of 108 cases involvingworkers' adjustment assistance heard by the commissionwere decided in favor of the workerswhile in another28 cases the votes were evenly divided; 15 out of 23 cases involvingfirms were even or decidedin favor of the firmwhile 7 out of 10 industry-widecases were decided in favor of the industry.(In case of a tied vote, the Presidentdecides whetherto accept or rejectthe petition.) StephenP. Magee 647 shiftsin the distributionof incomeoccasioned by tradeliberalization, and intensifiedby high domesticunemployment.3 Thus, stabilizationpolicy in recentyears has fallen short of the goals of slowing inflation, avoiding high unemployment,and attaining external equilibrium.On the other hand, a rationalallocation policy of reciprocal tariffreduction was successfullyimplemented. Both of these policies,however, led to a redistributionof income away from resourcesin import- competingindustries. Leaving aside the questionof whetherthese resources shouldbe compensated,the existingpolicy mechanisms for handlingredis- tributionproblems in the internationalsector probably did not compensate theseresources sufficiently to eliminatetheir losses relative to the rest of the economy. Regardlessof the cause, certaingroups continue to urge increasedprotection.Should U.S. importsbe restrictedto obtaina higherlevel of aggre- gate employmentor a distributionof incomemore favorable to the factors of productionin import-competingindustries? More important,can welfare be increasedin the United Statesas a whole with greaterprotection? The purposeof this paperis to providesome roughestimates of the welfare effectsin the United Statesof both existingprotection and greaterprotection. To use a concreteexample of the possibilitiesfor increasedprotection, I shall examineTitle III of a widelydiscussed bill introducedby Senator Vance Hartke of Indianaand RepresentativeJames A. Burkeof Massa- chusetts,in the Ninety-secondCongress.4 The first sectionpresents a generaldiscussion of free trade and the ap- propriatenessof trade restrictionsas tools of governmentpolicy. The second section considersthe welfare costs of existing trade restrictions and providesa frameworkfor the analysisof the Burke-Hartkeproposal5

3. The act does not providefor assistanceto export-or import-competingsectors due to overvaluationof the dollar or the recent cyclical developmentsthat have affectedthe U.S. trade position. Also, it is possible that the problems of import-competingareas are attributablelargely to the aggregatelevel of unemploymentand that this sector is attemptingto use trade policy to improve its position, an option not open to the non- traded goods sector. 4. S. 2592 and H.R. 10914, 92 Cong. 1 sess., both introducedSeptember 28, 1971. 5. Severalimportant works on trade restrictionsare used frequentlyin this analysis: Baldwin and Mutti, "Policy Problems";Giorgio Basevi, "The RestrictiveEffect of the U.S. Tariff and Its Welfare Value," AmericanEconomic Review, Vol. 58 (September 1968), pp. 840-52; W. M. Corden, The Theoryof Protection(Oxford: ClarendonPress, 1971); Harry G. Johnson, Aspectsof the Theoryof Tariffs(London: George Allen and 648 BrookingsPapers on EconomicActivity, 3:1972 in the thirdsection. The final sectionsprovide first some caveatsand then a summary.

Free Tradevs. Restrictions

MOTIVATIONS FOR FREE TRADE The primarymotivation for free tradeis that it permitsa countryto im- portproducts that can be producedcheaply abroad in exchangefor products that can be producedcheaply at home.Two primarygains arise from internationaltrade, the consumptiongain and the productiongain.6 The first arisesbecause consumers can purchasegoods at a list of worldprices that are belowthe internalprices that wouldprevail in the absenceof trade, the secondfrom the abilityof producersto shift theiractivity toward products in whichthey are relativelyefficient. Both gainsincrease consumer welfare since trade increasesincome and providesconsumers with an alternative set of marketprices. In principle,a country'stradeable products can be ranked from those of greatestcomparative advantage to those of least. The country obtains both the productionand consumptiongains from trade by increasingproduction and exportingthose high on the list and reducingproduction and importingthose low on the list. If foreigners place tariffson the exportsof a countryand that countryplaces tariffs on importedgoods, the gainsfrom freetrade and specializationin production are reduced.If each tarifflevel is sufficientlyhigh, trade will cease alto- getherand the countrywill revertto the welfareposition it wouldhave had in the absenceof trade. The propositionthat tariffscan offsetthe benefitsof comparativeadvantage in a multiple-productcontext has been illustratedin a study of 1937

Unwin, 1971); Ilse Mintz, U.S. ImportQuotas: Costs and Consequences,AEI Domestic Affairs Study 10 (AmericanEnterprise Institute, forthcoming); Robert M. Stern, "The U.S. Tariff and the Efficiencyof the U.S. Economy," in AmericanEconomic Associa- tion, Papers and Proceedingsof the Seventy-sixthAnnual Meeting, 1963 (AmericanEco- nomic Review, Vol. 54, May 1964), pp. 459-70; Robert M. Stern, "Tariffs and Other Measuresof Trade Control: A Survey of Recent Developments,"Journal of Economic Literature,forthcoming; and U.S. Department of Agriculture, Economic Research Service,Measures of the Degreeand Costof EconomicProtection ofAgriculture in Selected Countries,Technical Bulletin 1384 (1967). 6. See Harry G. Johnson, "The Cost of Protectionand the ScientificTariff," Journal of Political Economy,Vol. 68 (August 1960), pp. 327-45, reprintedin Johnson, Aspects of the Theoryof Tariffs. StephenP. Magee 649 tradeby MacDougall.He showedthat in productareas in whichthe United Stateshad a comparativeadvantage, U.K. tariffrates were high and that U.S. tariffrates were high in areaswhere the United Kingdomhad a com- parative advantage.7Relatively high averagetariff rates in the United States held the U.S. share of U.K. exports of manufacturesto only 4.5 percent, while 11 percent of U.S. exports of manufactureswent to the United Kingdomin 1937.Thus, tariffslargely dissipated the bilateralgain to tradebetween the two countries. If the world moved to free trade in manufacturedproducts, what re- sourceswould the United Statesbe ableto utilizemore efficiently? Since the United Stateshas large endowmentsof skilledlabor and of researchand developmentfacilities relative to the rest of the world, restrictionson the country'sexports reduce its abilityto exploitits comparativeadvantage by exportingproducts using such resources.8On the import side, since the United Stateshas little semiskilledand unskilledlabor relativeto the rest of the world,import-competing industries presumably will tend to release relativelymore of this type of labor. In 1967,manufacturing wage ratesin import-competingindustries were only $2.66 per hour (excludingquota items) comparedwith $2.84 in all manufacturingand $3.16 in the export industries.9Also, Ball has offered1962 cross-sectional evidence that effective rates of protectionare high in low-wageareas and vice versa.10Thus,

7. G. D. A. MacDougall, "British and American Exports: A Study Suggested by the Theory of ComparativeCosts," EconomicJournal, Vol. 61 (December 1951), pp. 697-724. The unweightedaverage for U.S. tariffswas 61 percentad valorem where the United Kingdom had a comparativeadvantage and 28 percentwhere she did not. 8. See, for example, William Gruber, Dileep Mehta, and Raymond Vernon, "The R&D Factor in InternationalTrade and InternationalInvestment of United States Industries,"Journal of Political Economy, Vol. 75 (February 1967), pp. 20-37; and Donald B. Keesing, "The Impact of Research and Development on United States Trade,"in the same issue, pp. 38-48. 9. All manufacturingis from Trade Relations Council of the United States, Inc., Employment,Output, and ForeignTrade of U.S. ManufacturingIndustries, 1958-69/70 (4th ed., Washington:TRC, 1972), Vol. 2, p. 793. Other wage rates were calculatedby Kevin Ferrell.Actually, to prove that import-competingindustries are relativelyinten- sive in unskilledlabor, the shareof factorrewards going to unskilledlabor in such industries should exceedtheir sharein the rest of the economy.The evidencethat this mightbe the case is Stage III (the standardizedproduct) of Raymond Vernon's product cycle ("InternationalInvestment and InternationalTrade in the Product Cycle," Quarterly Journalof Economics,Vol. 80, May 1966, pp. 190-207). 10. David S. Ball, "United States Effective Tariffs and Labor's Share,"Journal of Political Economy,Vol. 75 (April 1967), pp. 183-87. The figure on p. 185 shows the relationshipdramatically. 650 Brookings Papers on Economic Activity, 3:1972 eliminationof U.S. and foreigntariffs would shift relativedemand in the United States from low-wage to high-wage labor and permit greater utilizationof U.S. technology.

REASONS FOR TRADE RESTRICTIONS Althoughtariffs and quotaseliminate the welfaregains from trade,they existbecause they servegoals in at least six areas:domestic-political income redistribution,employment, the balanceof payments,international income redistribution,infant industries,and other social goals. We shall consider each in turn. Domesticdistribution of income.In 1941,Stolper and Samuelsonshowed that if wages and pricesare flexibleso that full employmentcan be main- tained,an increasein tarifflevels would decrease the realincome of the factor used relativelyintensely in exportproduction and increasethe real incomeof the factorused relatively intensely in import-competingindustries.11 Since,by the Heckscher-Ohlintheorem, a countrytends to exportproducts that use relativelyintensely its abundantfactor, higher U.S. tariffswould tend to decreasethe incomeof the abundantfactor-skilled labor-and increase the income of the scarce factor-semiskilled and unskilledlabor. Thiseffect would be accentuatedby any foreignretaliation on U.S. exports. The impact of restrictedtrade on the returnsto both capitaland land are unclear;12but to the extentthat foreignbarriers to U.S. exportsof agricultural goods could be reduced,the returnto land should increase.In the short run, beforerelative factor prices can adjustfully, tradeliberalization by the United Stateswould tend to increaseutilization of skilledlabor and land whilereducing the utilizationof unskilledlabor. Assumingthat the nation wishes to increasethe income of factors in import-competingindustries, the use of trade restrictionsdepends on the welfarecosts they impose on the United Statesrelative to those of alternative programssuch as subsidiesor excise taxes. In a welfaresense, trade

11. Wolfgang F. Stolper and Paul A. Samuelson, "Protection and Real Wages," Reviewof EconomicStudies, Vol. 9 (November 1941), pp. 58-73. 12. See, for example, Wassily Leontief, "Domestic Productionand Foreign Trade; the AmericanCapital Position Re-examined,"Economia Internazionale, Vol. 11 (Febru- ary 1954),pp. 3-32, reprintedin RichardE. Cavesand HarryG. Johnson(eds.), Readings in InternationalEconomics (Richard D. Irwin, 1968), Vol. 11, Chap. 30. See also the discussion by WilliamH. Bransonand Helen B. Junz, "Trendsin U.S. Tradeand Compara- tive Advantage,"Brookings Papers on EconomicActivity (2:1971), pp. 285-33& StephenP. Magee 651 restrictionsare generallyless efficient than alternativemechanisms of redistributingincome. Aggregateemployment. Tariffs can add to total employmentby stimulat- ing job creationin import-competingindustries if the economy is at less than full employment.At full employment,the relativeprice change for importablessimply alters the compositionof employmentand not the aggregatelevel if stabilizationpolicy keeps the economyat the samepoint on a fixed Phillips curve. However,trade restrictionsare an inappropriate meansof stimulatingaggregate employment. First and foremost,they cannot be as efficientas the macroeconomictools of monetaryand fiscal policy. Second, the aggregateimpact of trade restrictionsmay be offset by foreignretaliation. For example,Congress was influencedby the unem- ploymentproblem when it passedthe TariffAct of 1930,which increased dutieson morethan one thousandarticles of trade.'3Within a yeartwenty- five foreigncountries had raisedtheir tariffsagainst American goods in a tariffwar. Between 1929 and 1933 internationaltrade declinedfrom $65 billionto less than $25 billiona year,reflecting both decliningincomes and increasedrestrictions.'4 Trade balance and balance of payments. An increase in import prices causedby a tariffdirectly reduces the value of importsand improvesthe tradebalance. Several factors modify this effect,however. First, retaliation againstthe country'sexports may result.Second, there are indirectreduc- tions in the value of exportsthrough several mechanisms: The rise in the priceof importsdiscourages production of goods that use theseimports as inputs, encouragesthe movementof factors into the import-competing sector,and raisesthe pricesof exportsand nontradedgoods. Finally, the reductionin the value of imports into a countryreduces the foreignex- changeearnings of the rest of the world and thus the country'sexports. The use of the tariffas a means of improvingthe tradebalance and the balanceof paymentsis symptomaticof the failureof other,more general, mechanismsto achieveinternational adjustment of paymentsimbalances.15 Thereis good reasonto be skepticalabout the uses of tariffsfor providing

13. Robert M. Norris, "U.S. Foreign Economic Policy-Progress or Regression?" (speechdelivered at the annualspring meeting of the New HampshireBankers Associa- tion, June 7, 1972; processed). 14. N. H. Engle, "Reciprocityin Foreign Trade Policy," HarvardBusiness Review, Vol. 16 (Autumn 1937), p. 42. 15. See Stern, "Tariffsand Other Measuresof Trade Control." 652 Brookings Papers on Economic Activity, 3:1972 long-runsolutions to countries'balance-of-payments disequilibria. Since thesedisequilibria may be simplemonetary phenomena, tariffs may not be the most appropriateway to approachthe problem.If, as monetaristsargue, a balance-of-paymentsdeficit is evidencethat domesticcredit creation exceedsthe growthin the domesticdemand for cash balances,the properre- sponseto the deficitwould be to slow the growthof domesticcredit creation.'6 Of course, there are difficultiesin applyingto a reserve-currency countrysuch as the United Statesthe monetaryapproach to the balanceof payments.But similararguments hold for otherapproaches to the balance of payments.Policies such as greaterflexibility in foreignexchange rates withinthe band or fully floatingrates would obviatethe need for tariffsas policy tools for balance-of-paymentspurposes. Theoptimum tariff. If a countryis ableto influenceworld prices through the volumeof its trade,it can increaseits welfarein somecases by levyinga tariff,bringing welfare to a maximumby a rate that is calledan "optimum tariff."At its optimumlevel the tariffcan assurewelfare gains from improvingthe termsof tradethat more than offset the losses causedby the distortionin domesticproduction and consumption.In a sense,it operates like a tax on foreigners,lowering the net price they receiveon items they sell. HarryG. Johnsonhas shownthat the welfaregains from an optimum tariffare smaller,the smallerthe shareof importsin free tradein an economy and the moreelastic the foreignoffer curve facing the country.'7Since the United Stateshas a relativelysmall ratio of importsto domesticproduction and foreigndemand is somewhatelastic, the gains to the United Statesfrom an optimaltariff would be relativelysmall. Table 1, reproduced from Johnson'sarticle, shows, for example,that if the elasticityof foreign demandfor U.S. productswere 2.0 and the free tradeimport share were 0.10, the optimal tariff for the United States would be 100 percentbut would yield a percentagegain in U.S. welfareof only 1.4 percent.Such a policy is infeasiblefor severalreasons, such as internationalpolitical costs and the likelihoodof foreignretaliation, which are ignoredin Table 1. In practice,the attemptof any one countryto improveits welfareat the ex-

16. Harry G. Johnson, "The MonetaryApproach to Balance of PaymentsTheory," forthcomingin Michael B. Connolly and Alexander K. Swoboda, InternationalEco- nomics: The GenevaEssays (London: Allen and Unwin, 1972). 17. Harry G. Johnson, "The Gain from ExploitingMonopoly or Monopsony Power in InternationalTrade," Economica,Vol. 35, New Series (May 1968), pp. 151-56, reprintedin Johnson, Aspectsof the Theoryof Tariffs. SteDhenP. Magaee 653 Table 1. Relative Welfare Gain from OptimumTariff Policies Percent Optimumtariff Elasticity 100 Import share underfree trade offoreign 77 - 1 demand (percent) 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 1.10 1000 3.4 6.7 9.9 12.8 15.3 18.1 20.3 22.8 23.7 24.8 1.20 500 2.6 5.1 7.4 9.6 11.7 13.5 15.0 16.3 17.3 18.0 1.30 3331/3 2.1 4.1 6.0 7.6 9.2 10.6 11.7 12.6 13.4 13.8 1.40 250 1.8 3.4 4.9 6.3 7.5 8.6 9.5 10.2 10.7 11.1 1.50 200 1.5 2.8 3.8 5.2 6.2 7.1 7.8 8.4 8.9 9.1 1.75 1331/3 0.9 1.8 2.5 3.4 4.1 4.7 5.2 5.6 5.8 6.1 2.00 100 0.7 1.4 2.0 2.6 3.1 3.4 3.8 4.1 4.3 4.4 2.25 80 0.6 1.1 1.5 2.0 2.3 2.6 2.9 3.1 3.2 3.3 2.50 66% 0.5 0.9 1.2 1.6 1.8 2.1 2.3 2.4 2.5 2.6 2.75 57,j 0.4 0.7 1.0 1.3 1.5 1.7 1.8 1.9 2.0 2.1 3.00 50 0.3 0.6 0.8 1.0 1.2 1.4 1.5 1.6 1.7 1.7 3.50 40 0.2 0.4 0.6 0.7 0.9 1.0 1.1 1.1 1.2 1.2 4.00 331/3 0.2 0.3 0.4 0.6 0.7 0.7 0.8 0.9 0.9 0.9 4.50 28Wi 0.1 0.2 0.3 0.4 0.5 0.6 0.6 0.7 0.7 0.7 5.00 20 0.1 0.2 0.3 0.4 0.4 0.5 0.5 0.5 0.6 0.5 Source: Reproduced from Harry G. Johnson, "The Gain from Exploiting Monopoly or Monopsony Power in International Trade," Economica, Vol. 35, New Series (May 1968), p. 156, reprinted in Harry G. Johnson, Aspects of the Theory of Tariffs (London: George Allen and Unwin, 1971). pense of anotheris likely to resultmerely in a deteriorationin the welfare of both throughescalating restrictions. "Infantindustry." According to the infantindustry argument, tariffs can be levied on manufacturedand otherproducts to encourageindustrializa- tion of the country.'8Once the "infant"has maturedinto a strong economic entity, the traderestrictions are removed.A long-rungain can ac- crue from the traderestriction only if the tariff stimulatesa changein the country'sendowment or qualityof factors,increases its technicalcapabil- ity throughlearning by doing or some otherprocess, or allowsit to exploit potentialeconomies of scale.Even if economicgains are achievedthrough tariffs,but are delayed,their present discounted value may be less than the presentvalue of the costs imposedby the tariff.19 18. It is questionablewhether the "infant industry"argument applies to the United Statesin the currentprotectionist debate. As one wit has put it, most of the U.S. industries arguingfor greaterprotection are in their dotage ratherthan their infancy. 19. Nevertheless,a numberof countrieshave established"tariff factories" by placing high tariffs on finishedproducts and low tariffs on inputs. The "effectiverate of protection" indicatesthe protectionprovided to the manufactureof the finishedproduct. For example, in order to provide protection for an industrialprocess such as automobile assembly, the tariff on imports of autos must be comparedwith the tariffs on all components of automobile production. If there are high tariffs on imports of automobile components,there will be no encouragementof automobileassembly in the country. The "effectiverate of protection"is definedas the percentagechange in value added in some activitythat is attributableto the entiretariff structure.In the introductionto a paper by Humphrey,there is an excellentreview of the literatureon the effectiverate of protection 654 Brookings Papers on Economic Activity, 3:1972 Othersocial goals. Many other reasonsare given for traderestrictions, suchas maintainingdomestic production at highlevels in certainsectors for national defense purposes,protecting the safety and health of domestic consumersand the environment,and generallycontrolling the flows of individualproducts. Bhagwati has shownthat wheneverthe governmentis determinedto achievea social goal, the optimalstrategy is to use policy variablesthat affectthe targetvariable most directly.20For example,trade restrictionthrough a tariffor quotais inefficientsince it is equivalentto the simultaneousintroduction of a productionsubsidy and a consumptiontax. Traderestrictions, therefore, would be optimalonly wherethese two things are desirablesimultaneously and, by coincidence,are desirableat the same ad valoremrate. Even apart from this bizarrecase, it is difficultto con- structcases in whichsociety wishes both to tax consumptionand to stimulate productionof a givenproduct. In the case of the infantindustry that the countrywishes to encourage,a productionsubsidy is demonstrablysuperior to a tariffin that it does not artificiallyraise the priceto consumers.At times, the most efficientuse of governmentresources would be to affectdirectly the country'sfactor en- dowment.In short, if the social goal is to reduceconsumption of a given product,a consumptiontax is superiorto tariffs;if it is to stimulatepro- duction,a productionsubsidy is superiorto the tariff. In summary,each of the six objectivesof traderestrictions could poten- tiallybe achievedmore efficiently through other policy tools. Althoughthe argumentsfor trade restrictions are economically weak, the politicalreasons for theirexistence are more persuasive.First, import-competingindustries

(see David B. Humphrey,"Demand Inflationand EffectiveProtection," Southern Eco- nomicJournal, Vol. 37, October1970, pp. 144-50).The effectiverate of protectionis mea- sured as the ad valoremtariff on the final productless the tariff on the materialinputs into the product weighted by their importance; both terms are then adjustedfor the importanceof all materialinputs into the final product.The generaltendency of countries to place highertariffs on manufacturesof finishedgoods as opposed to raw materials is reflectedin recent data publishedby GATT. See The ContractingParties to the GeneralAgreement on Tariffsand Trade,Basic Documentationof TariffStudy, Summary Tables(Geneva: The ContractingParties, July 1970). The GATT study shows that the averagead valoremtariff rates in the majorindustrial countries (calculated by weighting each country'stariffs by its trade) is 6.2 percentfor raw materials,9.0 percentfor semi- manufactures,and 10.4 percentfor finishedmanufactures. 20. For the most comprehensivesurvey of the optimal policy tools to use in a given situation, see Jagdish N. Bhagwati,"The GeneralizedTheory of Distortions and Wel- fare,"in JagdishN. Bhagwatiand others(eds.), Trade,Balance of Paymentsand Growth: Papers in InternationalEconomics in Honor of CharlesP. Kindleberger(Amsterdam: North-Holland, 1971). StephenP. Magee 655 preferthem to the more directform becauseof the anonymityand lower visibilityof subsidizationthey provide. Second,trade restrictionsare po- liticallyattractive since they simultaneouslygenerate government revenue, subsidizeproducer interests that are well-organizedpolitically, and tax consumerinterests that, until lately, have not been effectivelyorganized. Third, the optimumtariff argumentmay be in the back of politicians' mindswhen a tariffis levied: They hope that foreignersand not domestic residentswill bear the "incidence"of the tax. It is reassuringthat substitutabilitywithin the economicsystem provides some offsetto the welfarelosses imposedby traderestrictions. Mundell has shown that even if internationalcommodity movements are completely stoppedby tariffsor quotas,a smallcountry may avoid any loss of welfare if thereis freemovement of at least one factorof production.21 Thus,even if protectionistsin the United Statessucceeded in stoppingall international trade,under some fairlyrestrictive assumptions, U.S. residentsmight still be able to attainthe free tradewelfare level if (contraryto fact) either(1) free immigrationor (2) free movementsof capital were permitted.Thus, some of the welfarelosses imposedby traderestrictions may be undoneby factormobility. Giventhese arguments on both sidesof the tariffquestion, what has been the historicalpattern of tarifflevels in the United States?Figure 1 shows that from earlyin the centuryuntil the earlytwenties, tariff rates were falling. The rates rose in the 1920s with the Fordney-McCumbertariff and againin the 1930sunder Smoot-Hawley. Since then, therehas been a grad- ual movementto freertrade.

The WelfareEffects of Movingto Free Trade

A FRAMEWORK FOR ANALYZING THE WELFARE EFFECTS As noted at the outset,international trade confers on a countrytwo ad- vantages,both of which increaseits real income: (1) an expansionof its consumptionopportunities, and (2) a shift of its resourcesinto more pro- ductiveareas. One way of quantifyingthese gains is throughconsumers' surplusand producers'surplus, as illustratedin Figure 2. The panel on consumers'surplus reveals that, if the productwere sold on a unit-by-unit

21. Robert A. Mundell, "InternationalTrade and Factor Mobility," American EconomicReview, Vol. 47 (June 1957), pp. 321-35. 00 0~

C.0

02 0 CN0-

0~~~~~~0 CO4 X o~~~~~--4- 3t A l______Z T-- 0>s

0 0

_.t X^~~~~~~IC- 0_) ~~~~~~~~~~~~~~~~~~~~~~

0 ~ ~ ~~~0

0O-

R o o = ~~~~~~~~~~~~~~~~t 0ot

E ___ iz StephenP. Magee 657 Figure2. Measurementof Consumers'and Producers'Surpluses

Price

M \

t CosmesPrdces urp Demandlus A42

P.eR

Si 4i ul Q6 Quantity Price ~~~~~~~~~~~Supply

B, ~~~~~~~~~~Producers'surplus

N ~~~~~~B2

0 ih 1i ki Qe Quantity 658 Brookings Papers on Economic Activity, 3:1972 basis,and the priceexceeded OM, no unitswould be sold. If one unit were sold, someonewould be willingto pay ss1; the secondunit could be sold for ttl, the third unit for uul, and so on. If the price is OP",the amount people wouldbe willingto pay on a unit-by-unitbasis is equalto the area OQQeRM.However, since everyonepays the same pricefor each unit sold, the actualpayment for OQe is equal to the area OQeRP,,or area A2. The residualregion, A1, is called the "consumers'surplus." As the price falls, this areabecomes larger, and consumersare better off. The conceptof producers'surplus is similar.At a marketprice below ON, suppliersare not willingto provideany output.If the priceis equalto ii1, one unit wouldbe forthcoming,a secondif the pricewere raised tojjl, and so forth.Thus, on a unit-by-unitbasis, supplierswould be willingto supply OQeif they re- ceivedarea B2 as theirtotal receipts.However, if OPeis the marketprice, the actualreceipts of suppliersequal the areasB1 + B2. Thus, producers receivea "producers'surplus" equal to area B1. This partialequilibrium approachcan be extendedto a generalequilibrium context in the two- sectormodel by usingcompensated demand curves and generalequilibrium supplycurves (that is, the verticalaxis expressesa price ratio). The use of consumers'surplus and producers'surplus as measuresof the welfarecosts of traderestrictions is illustratedin Figure3.22 Assumethat the United States is importinga product whose world price is equal to OP, The U.S. domesticdemand and supply curvesfor the product are shownin the figure.With no restrictionson U.S. imports,domestic demand would equal OQ2, with OQ2 producedby U.S. producersand Q2Q2 the quantityimported. The consumers'surplus would then be the triangular areaMSPW and the producers'surplus would be areaH1. If a tariffis now imposedon importsof the productbut the worldprice stays fixed, the U.S. price rises to OP.. The ad valoremtariff is equal to the ratio of PIj,, to

22. One analytical problem with these welfare measuresis that the imports under consideration include both consumer goods and intermediategoods while the tradi- tional theory deals only with consumer goods. However, Richard Schmalensee, in "Consumer'sSurplus and Producer's Goods," AmericanEconomic Review, Vol. 61 (September 1971), pp. 682-87, finds that in the constant costs-perfectlycompetitive case, if the ratio betweenintermediate and final goods is fixed, the traditionalconsumers' surplusapproach yields an accuratemeasure of the social welfare effects. If the ratio is not fixed, so that the lower pricecaused by a tariffdecrease causes the ratio to change,the traditionalmeasure overstates the true welfare gain in the competitivecase and under- statesthe social gain in the monopolycase. Sinceimport markets probably fall somewhere betweenthe perfectlycompetitive and the monopoly cases, there is no reason to expect the estimatehere to be too high or too low because of the intermediategoods problem. The one caveat is that Schmalenseedeals only with the constant costs case. Stephen P. Magee 659 Figure3. The WelfareCosts of Tariffs

Price

Supply

Pis R E PIwP H2 I /^. ~L/ DX"

DnDemand N Hr2 G, F G2

Q2 Ql Qj Q2 Quantity D1, D2 = Productionand consumptiondeadweight losses, respectively E = Tariff revenue G1 + F + G2 = Quantityimported in absence of tariff H1 = Producers'surplus without tariff H1 + H2 = Level of domestic productionunder free trade I + H1 = Additionalproducers' surplus with tariff MSP. = Consumers'surplus without tariff MRPUS = Redudtionin consumers'surplus with tariff OP.. Notice that the amount of consumersurplus is reducedby the area PwSRPu8.What happens to this decreasein the value of consumerwelfare? First, the trapezoidalarea I is transferredto producers'surplus, so that import-competingproducers earn windfallgains. Second,area E becomes tariff revenue and is transferredfrom consumersof import-competing productsto the government(area E equalsthe tariff rate P.oP.8times the quantityof importsQi Q' underthe tariff).Thus, two of the main components of the reductionin the welfareof consumersare simplya transferto other U.S. citizens:I is the redistributioneffect to producersand E the 660 Brookings Papers on Economic Activity, 3:1972 revenueeffect. However, two portionsof the reductionin consumers'surplus are capturedby no one. These are the triangularareas D1 and D2, and they are known as the "deadweightlosses" causedby the tariff.Area D1 correspondsto the productiondeadweight loss, whichresults from an artificialmovement of resourcesfrom higher-to lower-productivitypur- suits; and D2 is the consumptiondeadweight loss, whichreflects substitu- tion of less satisfactoryproducts by consumers.The subsequentinvestiga- tion concentrateson areasD1 and D2, the sum of whichmeasures the total social costs to U.S. consumersof U.S. traderestrictions.23 The deadweightloss trianglesin Figure3 are annualflows,that is, these losses are incurredevery year. In analyzingthe welfareeffects of moving towardfree trade,the gainsthrough the reductionin the inefficiencies(D1 and D2) must be adjustedfor the costs of movingresources from import- competingindustries into otherareas. This involvescalculating the present discountedvalue of both the perpetualannual deadweight losses of welfare and the costs of movingresources out of import-competingproduction and into other areas.24The initial assumptionis that neitherthe demandnor the supplyof the productin questionis growing;this assumptionis later relaxed.The long-runsupply elasticity will, however,exceed the short-run supplyelasticity. Thus, the productiondeadweight loss, D1, becomeslarger

23. See W. M. Corden, "The Calculation of the Cost of Protection," Economic Record,Vol. 33 (April 1957), pp. 29-51; Basevi, "RestrictiveEffect of the U.S. Tariff"; Stern, "U.S. Tariff and the Efficiencyof the U.S. Economy"; and FranklinV. Walker, "The RestrictiveEffect of the U.S. Tariff: Comment,"American Econlomic Review, Vol. 59 (December 1969), pp. 963-66. The use of these trianglesas measuresof the welfare costs of trade restrictionshas been discussedextensively by Johnson in "Cost of Protec- tion and the ScientificTariff." The theoreticalunderpinnings of these measuresof welfare have been discussedextensively in Arnold C. Harberger'swork; see, for example, "The Measurementof Waste,"in AmericanEconomic Association, Papers ... 1963, pp. 58-76. The ratherstrong assumptions necessary for the use of this techniqueand all of its theoreticaldifficulties will not be discussedhere. The discussionsby Dale W. Jorgenson, William Vickrey,Tjalling C. Koopmans, and Paul A. Samuelson,in response to Har- berger'spaper and appearingin the samePapers (pp. 86-96), referto a numberof these difficulties. One set of sufficient conditions for consistent additivity of individual preference maps into communitypreference maps is that all individualshave identical and homo- thetic preferences.The measuresused here also imply that the marginalutility of income is the same across individualsand at all income levels. 24. For intertemporalapproaches to deadweightlosses, see D. Levhariand E. She- shinski, "Lifetime Excess Burden of a Tax," Journal of Political Economy, Vol. 80 (January/February1972), pp. 139-47, and Baldwin and Mutti, "Policy Problems." I am especiallyindebted to the Baldwinand Mutti paperfor suggestingthe discountingof both benefitsand adjustmentcosts. StephenP. Magee 661 throughtime. In Figure3, this meansthat the point L, whichis the inter- section of the supplycurve and the worldprice line, will move to the left as the supplycurve rotates through T and adjuststo its long-runposition. For most consumerproducts, the long-rundemand elasticity will exceed the short-rundemand elasticity.25 Thus, the long-runconsumption deadweight loss, D2, will be largerthan the short-rundeadweight loss. The sizes of the two deadweightloss flows increaseuntil the new long-run equilibriumis reached,and thereafterthe two trianglesremain at a constant size in the absenceof growth.Since the losses are changingthrough time, discountingthem is instructivesince it reducesthe changingflows to a singlepresent value; thus, the costs of tariffscan be comparedwith the presentvalues of alternativepolicies. The welfaregains from tariffelimination must be adjustedfor the costs of moving factors from the import-competingsector into other areas. The largestcomponent of this cost is the unemploymentof factorsof pro- ductionduring a transitionperiod from the time the factorsleave the import-competingindustry until they are reemployed.These costs are in- curredonly until a new long-runsupply equilibriumis achieved;at that point resourceflows cease and henceno furthercosts arisefrom the movementof resources.In this paper,the relocationcosts of laborwill be usedas a proxyfor the generalcosts of relocation.There are seriousdifficulties in calculatingthe costs of moving capital that is quasi-specificto import- competingproduction. As a result,the "lost labor time" measurewill understatethe total social cost of the resourcemovement, just as the deadweightloss trianglesunderstate the costs of traderestrictions (some of the nonquantifiablelosses are discussedbelow).

COSTS OF EXISTING RESTRICTIONS ON U.S. IMPORTS Whatare the annualcosts of existingrestrictions on U.S. trade?This section analyzesthe impact on U.S. imports,considered in three categories: those that competedirectly with U.S. productionand are subjectto tariffs;

25. However, H. S. Houthakkerand Lester D. Taylor, in ConsumerDemand in the UnitedStates, 1929-1970: Analysesand Projections(2d ed., HarvardUniversity Press, 1970), show that in cases whereinventory adjustments are present,the long-rundemand elasticityfor productscan be less than the short-runelasticity. But on balancethey find that the "habitformation" phenomenon generally dominates the "inventoryadjustment" phenomenon,so that the long-rundemand elasticities exceed the short-runelasticities. I assume that producersand consumersare in long-runequilibrium at points T and R in Figure 3, so that the supply and demandcurves rotate through these points. 662 Brookings Papers on Economic Activity, 3:1972 Table2. Valueof U.S. Importsand Duties Paid on MajorProducts Subject to QuantitativeRestrictions, 1971 Millions of dollars Product Imports Duties Petroleum 3,278 100 Steela 2,009 126 Man-madeand woolen textiles 1,840 ... Sugar 813 48 Meat 598 34 Cotton textiles 590 Dairy 70 .. Stainlesssteel flatware 28 ... Cotton and cotton waste 8 ... Wheat 0.4 ... Peanuts 0.4 ... Total 9,235 308

Sources: All data except for steel were supplied by Mary Jane Wignot, Office of the Special Representative for Trade Negotiations. Steel data, which are from the Trade Relations Council of the United States, Inc., are for 1969 and are for standard industrial classification (SIC) 3312 only (blast furnaces and steel mills). James Ozzello, U.S. Department of State, Trade Agreements Division, has noted that SIC categories 3315, 3316, and 3317 (steel wiredrawing, cold finishing of steel shapes, and steel pipe and tube, respectively) are also included in the voluntary agreement, but data problems excluded them from this exercise. SIC 3312 is the largest of the four categories. a. 1969 data.

those that competeonly partially,both dutiableand otherwise;and those subjectto quotasor government-to-governmentagreements. I assumethat decreasesin the respectivetariffs lead to equal reductionsin the prices of competingU.S. goods in the first group,but not in the second. Table2 showsmajor import items subject to quotasor otherquantitative restrictions,which in 1971accounted for $9.2 billion in U.S. imports.The remainingtask is to dividethe $36.4billion in nonquotaimports into those that are directlycompetitive and those that compete only partiallywith domesticproduction. The only informationreadily available on imports that competeclosely with domesticproduction has been providedby the U.S. Bureauof Labor Statistics.62Since I rely on these data, which cover only manufacturedimports, costs of the tariffon directlycompetitive non- manufacturedimports will be understated.As the data shown in Table 3

26. U.S. Bureauof Labor Statistics,"Productivity and Unit Labor Costs in Export and Import-CompetingIndustries, 1958-68," in UnitedStates InternationalEconomic Policy in an InterdependentWorld, Papers Submitted to the Commissionon International Trade and Investment Policy (U.S. Government Printing Office, 1971), Vol. 1, pp. 507-33. Stephen P. Magee 663 Table 3. Calculationof CompetitiveU.S. Importsand Duties for 1971 from 1969 Data Dollar amountsin millions

Percent Line Description Amount of total

1969

1 Total imports (c.i.f.) $38,314 100.0% 2 Manufactured 33,115 ... 3 Directly competitivewith U.S. productions 10,281 26.8 4 Import-competingproduction in United States 41,500 ... 5 Customs duties 2,340 100.0 6 On directlycompetitive imports5 814 34.8 7 On remainderof imports 1,526 65.2

1971

8 Total imports (f.o.b.) 45,602 100.0 9 Directly competitivewith U.S. productions 12,220 26.8 10 Quota items 9,235 20.3 11 Other 24,147 53.0 12 Import-competingproduction in United States 44,862 ... 13 Customs duties 2,768 100.0 14 On directlycompetitive imports" 963 34.8 15 On quota items 308 11.1 16 On other imports 1,497 54.1

Sources: Line 1-International Monetary Fund, InternationalFinancial Statistics, Vol. 25 (June 1972), p. 37. Line 2-Trade Relations Council of the United States, Inc., Employment, Output, and Foreign Trade of U.S. ManufacturingIndustries, 1958-69/70 (4th ed.; Washington: TRC, 1972), Vol. 2, p. 793. Lines 3, 4-Derived by applying data in ibid., Vols. 1 and 2, to four-digit standard industrial classification industries in which imports have very close domestic substitutes, using classification from U.S. Bureau of Labor Statistics, "Productivity and Unit Labor Costs in Export and Import-Competing Industries, 1958- 68," in United States internationalEconomic Policy in an Interdependent World, Papers Submitted to the Commission on International Trade and Investment Policy (U.S. Government Printing Office, July 1971), Vol. 1, App. A. Line 5-U.S. Tariff Commission, tabulation, "Value of U.S. Imports for Consumption, Duties Collected, and Ratio of Duties to Values, under the Tariff Act of 1930, 1930-71" (March 1972; processed). Lines 6, 7-Same as Lines 3, 4. Line 8-International Financial Statistics, Vol. 25 (October 1972), p. 366. Note that the data are on an f.o.b. basis. Line 9-26.8 percent (from line 3) of line 8. Lines 10-Table 2. Line 12-Since data were not yet available on U.S. production competing directly with imports. it is assumed to grow at the same rate as total goods output in the United States. The Economic Report of the President, January 1972, p. 200, reports that output of total goods increased from $457.3 billion in 1969 to $494.2 billion in 1971, or an increase of 8.1 percent. Thus line 12 equals 1.081 tinmesline 4. Line 13-Same as line 5. Line 14-34.8 percent (from line 6) of line 13. Line 15-Table 2. Lines 11, 16-Residuals. Calculations are based on data before rounding. a. Directly competitive means those manufactured imports that are dutiable and not subject to quotas, and that compete closely with U.S. production. 664 Brookings Papers on Economic Activity, 3:1972 indicate,directly competitive imports in 1971 were calculatedto be $12.2 billion and to competewith $44.9 billion in U.S. production.Other imports equaled$24.1 billion. Thus, quota items are 20 percentof total imports,directly competitive imports are 27 percent,and the remainderequals 53 percent. Accordingto the allocationof customsrevenue collected in 1971,shown in Table 3, the impliedad valoremtariff rate on all U.S. importswas 6.1 percent:On quotaitems it was 3.3 percent;on directlycompetitive imports, 7.9 percent;and on other imports,6.2 percent.

DIRECTLY COMPETITIVE U.S. IMPORTS

In orderto calculatethe welfareeffects of the existingtariff structure, valuesmust be assignedto the elasticitiesof domesticdemand, supply, and imports.In Figure 3, these correspondto the elasticitiesof the domestic demandand supplycurves in the regionof points T for productionand R for consumption.However, equation (1) impliesthat only two of thesethree elasticitiesare independentin the case of importsthat are perfectlysubsti- tutablewith domesticproduction:

(1) em = (ed - e8) + e, where em = the elasticity of demand for imports ed = the domesticdemand elasticity e8 = the domestic supply elasticity D = the total quantitydemanded in the United States(the sum of domesticproduction plus imports) M = the quantityof importsconsumed.

From Table 3 the ratio of D to M equals (12.2 + 44.9)/12.2, or 4.67,27 assuminga domesticprice of unity. Since more econometricevidence is availableon importprice elasticitiesthan on the relevantdomestic supply and demandelasticities, I have assumeda valuefor emand derivedvalues of ed and e8that satisfyequation (1). The importelasticity, em, is assumedto be

27. For simplicity,in this and similarcalculations in this paper,the roundednumbers are given, althoughthe calculationsare made from unroundeddata. StephenP. Magee 665 -3 in the shortrun and -8 in the long run.28The short-runelasticity con-

strains e8 to 0.82 if ed = 0, and ed to-0.64 if e8 = 0. There is evidence thatthe supplyelasticities exceed the demandelasticities (in absolutevalue) in both the short and the long run.29I assumethat the demandelasticities equalminus one-half the supplyelasticities. These assumptions, with equation (1), yield the following pattern of domestic demand and supply elasticities: Elasticity Short run Long run Domesticdemand, ed -0.25 -0.75 Domesticsupply, e8 0.50 1.50 These elasticitiesseem low and will mean some underestimateof the welfare costs of U.S. tariffs.For example,on the supplyside in the long run, constantcosts may be a reasonableassumption, implying that the long-run elasticityof supplywould be infinite.30But assuminghigher domestic demand and supply elasticitieswould producean implausiblyhigh import price elasticity.For simplicity,the discussionin this section assumesthat eliminationof tariffson U.S. imports does not raise the world price of U.S. importables.Thus the understatementof the welfarecosts using low domesticelasticities is more or less offset by ignoringthe terms-of-trade effecton importables.31

28. The short-runelasticities are consistent with those found in H. S. Houthakker and Stephen P. Magee, "Income and Price Elasticities in World Trade," Review Of Economicsand Statistics, Vol. 51 (May 1969),pp. 111-25, whileboth are somewhathigher than those in John E. Floyd, "The Overvaluationof the Dollar: A Note on the Inter- national Price Mechanism," AmericanEconomic Review, Vol. 55 (March 1965), pp. 95-107. 29. For example, J. Wemelsfelder,in "The Short-TermEffect of the Lowering of Import Duties in Germany," EconomicJournal, Vol. 70 (March 1960), pp. 94-104, found that cuts in West German tariffs in the late 1950s stimulatedimports primarily through contractionof productionrather than through increasesin consumption. 30. See Harry G. Johnson, "Factor MarketDistortions and the Shape of the Trans- formation Curve,"Econometrica, Vol. 34 (July 1966), pp. 686-98. Johnson performed simulations on the degree of curvatureof the productionpossibility curve and found that an economy'sability to transformone productinto anotheris extremelyhigh, even under fairly large variations in the parametersof the production functions. A. A. Walters,in "Productionand Cost Functions; an EconometricSurvey," Econometrica, Vol. 31 (January-April1963), pp. 1-66, also finds that the evidencefor constant returns to scale is very strong. Thus, the assumptionof a long-runsupply elasticityof 1.5 is too low. The domestic demand elasticities here are also somewhat lower than those estimated by Houthakkerand Taylor, ConsumerDemand, p. 175. 31. Both estimatedand assumedvalues of the foreign supply elasticity for U.S. imports are high enough in the short run to make this a fairly innocuous assumption.In 666 Brookings Papers on Economic Activity. 3:1972 Table4. Short-runand Long-run Static Effects, from 1971 Base, of Eliminationof Tariffson U.S. Importswith Close Domestic Substitutes Millions of dollarsin annualflows Changein variable due to price chanige Short Long Analogous Variableand formula run run area in Figure3 Deadweightloss (D WL)elimination 97 291 D = D1 + D2 Consumption,0.5t2Ded 38 114 D2 Production, -0.5t2Se8 59 177 D Decreasein domesticproduction, -tS(1 + es) 4,912 8,187 1 + D1 + G Redistribution:consumer gain 4,275 4,350 I + D + E Producers'loss, tS(1 - 0.5te8) 3,215 3,096 1 Revenue and DWL to consumers, 1,060 1,254 D + E 0.5t2(-Ded + Se,) + t(D - S) Increasedvalue of imports 2,680 8,040 G1 + G2 Increasedconsumption, tDed 1 042 3,126 G2 Reduced production,tSe8 1,638 4,914 G Revenueloss, t(D - S) 963 963 E Sources: Derived from indicated formulas, where in the mathematical expressions t = change in the U.S. price if the quota were eliminated (the ad valorem tariff is 7.9 percent, from text; thus t = 0.079/1.079 = 0.073) D ='total demand in the United States (domestic production plus imports) = $57,082 million (from Table 3) (note that the D in Figure 3 referredto in the last column stands for deadweightloss, and not total demand.) S = total supply in the United States = $44,862 million (from Table 3) ei = domestic demand elasticity (short run = -0.25; long run = -0.75) (from text) e, = domestic supply elasticity (short run = 0.5; long run = 1.50) (from text) The calculations were made from data before rounding. a. The t in this formula differs from the others in the table; it refers to the actual ad valorem tariffrate (7.9 percent), while they refer to the percentage decrease in price (7.3 percent).

The formulasfor calculatingthe welfareand structuraleffects of reducing the 7.9 percentad valoremtariff on directlycompetitive imports to zero are shownin Table4, and describedmore fully in Johnson'spaper.32 The table gives the short- and long-runeffects of the tariff elimination;these data are shown geometricallyin Figure3, althoughstrictly speaking, the functions shouldbe constantelasticity nonlinear curves in the relevantregions ratherthan the linearfunctions of that figure. Table 4 indicatesthat the deadweightloss attributableto the current tariffsystem equals $97 million a year in the shortrun and $291 million a

the long run, a foreign supply elasticity of infinityis plausible so that the adverseeffect on U.S. welfare of a rise in world prices when U.S. tariffs are cut is only a short-run phenomenon. 32. Johnson, "Costs of Protection." StephenP. Magee 667 yearin the long run. Most of this distortionis due to misallocationof production. These numbersare underestimatesof the social cost of these tariffs,since tariffrates vary aroundthe mean of 7.9 percent,and the welfare loss trianglesgrow through time; however, they are overestimates in the shortrun becauseI have ignoredthe cost of labor movements.Ad- justmentsfor all three of these factorswill be performedin a subsequent section.Tariff elimination reduces the value of domesticproduction by approximately$5 billion in the short run and $8 billion in the long run, assumingthe simple static case of no growth.Elimination benefits consumersof import-competingproducts by $4.3 billion, both in the short and in the long run. Most of this is at the expenseof producersof import- competingproduction. The value of imports increasesby $2.7 billion a yearin the shortrun and $8.0billion a yearin the long run,mostly through reducedU.S. productionrather than increasedU.S. consumption.Nearly $1 billionin tariffrevenue is eliminated,in both the shortand the long run.

OTHER U.S. IMPORTS To assessthe effectsof eliminatingthe 6.2 percenttariff on U.S. imports that are not directlycompetitive with U.S. production,and not subjectto quotas, the analysisassumes an elasticityof importdemand of -2 in the shortrun and of -5 in the long run.With theseelasticities, the estimateof the deadweightlosses (DWLs) attributableto the tariffis

(2) DWL = 0.5t2emV DWLsr = 0.5 (0.058)2 (2.0) (24,100) = $81 million DWLL, = 0.5 (0.058)2 (5.0) (24,100) = $202 million, whereV is the valueof importsbefore the tariffcut. The gainto the United Statesfrom eliminatingthis tariffis $81 millionper year in the short run and $202 millionper year in the long run. Eliminationof this tariffresults in a 5.8 percentdrop in the price of importedgoods; t = (0.062/1.062) = 0.058.Assuming that this reducesthe pricedomestically by 1 percent,33and that domesticsupply elasticities are 1.0 in the shortrun and 3.0 in the long

33. Use of differentimport price changes and domesticprices changes follows Floyd's techniqueof handlingimports that are imperfectlysubstitutable with domestic production (see "Overvaluationof the Dollar"). 668 Brookings Papers on Economic Activity, 3:1972 Figure 4. Long-runStatic Effects, from 1971 Base, of Tariff Eliminationon NoncompetitiveU.S. Imports

Price

PUS ~~~R

P. _ q - - Supplyof imports

M' Demand|forLiports

0 Qi Qo Quianitityimnported Value (billionsof dollars Areaa in annualflows) Definition it 1.5 Revenue redistributedto consumers M' 24.1 Old value of imports L' 7.0 Increasein value of imports K' 0.2 Deadweightloss Source: Derived by method discussed in text. a. Not drawn to scale.

run, domesticproduction declines by $4.4 billion and $8.7 billion, respectively. The redistributionfrom producersto consumersis roughly $2.2 billion in the short run and $2.0 billion in the long run.34This redistribution, plus the revenueand deadweightloss effects,increases the welfareof consumersof theseimportable products by $3.8billion in the shortrun and by $3.7 billion in the long run. The changesin trade in the long run are shown in Figure4. The curvesare the U.S. importdemand curve and the foreignsupply curve of goods to the United States.Domestic production and consumptionare not shownin Figure4. AreaK' is the deadweightloss of the tariff.

34. The domesticproduction data for this exerciseare based on an estimateof $217 billion from 1969 data. Stephen P. Magee 669

U.S. IMPORTS UNDER QUOTAS This section considersthe welfare costs associatedwith quantitative limitationson six majorproducts imported into the United States.Imports of thesegoods equaled$9.2 billionin 1971(see Table2). Table5 showsthe valuesof domesticsupply and demandand imports for the six majorquota itemsin 1969.The last columnwill be combinedwith equation(1) in some cases to approximatethe importprice elasticitiesof demand. I shallrely in this sectionon otherstudies of U.S. importquotas. In some casesthese estimate only the consumercost. This exceedsthe social cost to the United States, however, since part of it is a redistributionto U.S. import-competingproducers and part of the "tariff-equivalentrevenue" is capturedby U.S. importers.The actualsocial cost to the United States equalsthe deadweightlosses plus that part of the tariff-equivalentrevenue that is capturedby foreignsuppliers. The latterportion reflectsthe possibilitythat quotasmay allow foreignsuppliers licensed to sell in the United States to charge more than the world price. The quota operateslike a tariff, but the "revenue"is shared by importersand foreign exporters ratherthan flowinginto the U.S. Treasury.Figure 4 can be used to illus-

Table5. U.S. DomesticSupply, Demand, and Imports of MajorProducts Subjectto Quotas,1969 Dollar amountsin millions Domestic Ratio of supplyless Domestic demandto Product exports Imports demand imports Petroleum $12,460 $2,480 $14,940 6.02 Steel 24,300 2,009 26,310 13.10 Textiles, total 17,597 1,632 19,229 11.78 Man-made 12,418 695 13,113 ... Woolen 1,203 410 1,613 ... Cotton 3,976 527 4,503 ... Sugar 2,420 677 3,097 4.57 Meat 9,523 558 10,081 18.07 Dairy 10,817 100 10,917 109.17 All quota items 77,117 7,456 84,574 11.34

Source: All data except those for steel and textiles were supplied by Mary Jane Wignot, Office of the Special Representativefor Trade Negotiations. The textile data were provided by Emanuel Lipscomb of the U.S. Department of Commerce. The steel data are from the Trade Relations Council of the United States, Inc., and include only SIC 3312 (see the source to Table 2). 670 BrookingsPapers on EconomicActivity, 3:1972 tratethese points. If, insteadof the tariff,a quotahad been used to cut imports from OQOto OQi, then the quota would have been equivalentto a tariff equal to PUSPW/OPwin that it cut trade by an equivalent amount.35 The tariff-equivalentrevenue is equal to areaJ'. Since there is a gap between internalU.S. prices and world prices, this area will be captured eitherby U.S. importersor foreignsuppliers, or partlyby each,depending on the competitivesituation in the importmarket. If it is capturedentirely by U.S. importers,the social cost of the quota to the United Statesequals the deadweightloss triangle,K'. If it is capturedentirely by foreignsuppliers,the cost to the United Statesequals K' + J'. In the thirdcase, where the revenueis shared,the cost is K' + fJ'; f is the fractionof the tariff- equivalentrevenue captured by foreigners.The first two situations are simply specialcases of the third in whichf = 0 and f = 1, respectively. Anotherpoint is that the observedvalue of U.S. imports,M, will be equal to the areaM' + fJ'. In order to convert the publishednumbers on the consumercost of quotasinto the socialcosts, I have usedthe followingprocedure. As above, t is the change in the U.S. price if the quota were eliminated;the price changecan be approximatedas

(3) t = CS/D, whereCS is the consumersurplus lost due to the quota and D the value of demand(or consumption),inclusive of CS.36 Variablet is then used in the formulasin Table 4 to calculatethe deadweightlosses. Calculationof the portion of the tariff-equivalentrevenue going to for- eignersinvolves the followingvariables: M' is the value of importsexclu- sive of the tariffrevenue they capture,ftM' is the tariffrevenue they capture, and the sum of these two items equalsobserved imports, M: (4) M =M' + ftM'. Thus, M' is computedby (5) M' = M/(1 + ft);

35. Rachel McCulloch,in "ImportQuotas and ResourceAllocation" (PhD. disserta- tion, Universityof Chicago, 1973),has definedother equivalencesin monopolisticcases. 36. This is an exact measureif the domestic demandelasticity equals zero. This calculation of t places no restrictionon the foreign supply elasticity of the good to the United States. It is alreadyimplicit in the estimatesgiven in the studies surveyed. Stephen P. Magee 671 and the tariff-equivalentrevenue captured by foreigners,RF, in terms of observedimports, M, is

(6) RF=( ft)M. Equation(6) will be used in estimatingthe tariff-equivalentrevenue lost to foreignersunder the steel,meat, and dairyquotas. Petroleum.Bergsten notes that oil quotas raise the domesticprice of petroleumby 60 percent.37The CabinetTask Force on Oil ImportControl estimatedthat the cost to consumersof the oil importquota was $5 billion and wouldincrease to $8 billionper yearby 1980,involving efficiency costs of $1.5 billion to $2.0 billion.38Mintz, in commentingon the administra- tion of the oil quota, arguesthat the loss of tariff-equivalentrevenue to foreignersmay be small.39Thus, the annualsocial cost of the oil import quota is assumedhere to be $1.5 billion. This numberis a bit high in the short run, but low in the long run. Steel. We consider here only standardindustrial classification (SIC) 3312, blast furnacesand steel mills, althoughSIC 3315, 3316, and 331740 are also subjectto the voluntarylimitations. According to Mintz, in De- cember 1968, Japaneseand some Europeansteel industries,which accountedfor 82 percentof U.S. steelimports, agreed to limittheir exports of steel mill productsto the United States.James Ozzello41 estimates that the 37. See C. Fred Bergsten,"The Cost of ImportRestrictions to AmericanConsumers" (AmericanImporters Association, no date; processed),p. 3. 38. Cabinet Task Force on Oil Import Control, The Oil ImportQuestion: A Report on the Relationshipof Oil Importsto the National Security(U.S. Government Printing Office, 1970), p. 124. These long-run efficiencycosts include only the productiondead- weightloss; they do not includeloss of tariff-equivalentrevenue to foreignersor the consumption deadweight loss. See also the study by Kaj Areskoug, "U.S. Oil Import Quotas and National Income," SouthernEconomic Journal, Vol. 37 (January1971), pp. 307-17. 39. Mintz, U.S. ImportQuotas. 40. Steel wiredrawing,cold finishingof steel shapes, and steel pipe and tube, respectively. 41. The estimatesmade by Ozzello, who is in the TradeAgreements'Division of the U.S. Departmentof State, incorporatethe quota, restrictivegovernment procurement policies, and other nontariff barriers.His estimates are not inconsistentwith those of C. R. MacPheein "NontariffBarriers in InternationalTrade in Steel" (Ph.D. disserta- tion, MichiganState University,1970). Restrictive government procurement policies are an important nontariff barrier, both here and abroad; for example, see Norman S. Fieleke, "The Buy-AmericanPolicy of the United States Government:Its Balance-of- Paymentsand WelfareEffects," New EnglandEconomic Review (July/August 1969),pp. 2-18. 672 Brookings Papers on Economic Activity, 3:1972 ad valoremtariff equivalent of this and otherlimitations on U.S. imports of SIC 3312 is approximately17 to 35 percent,and that foreignsuppliers usually quote their prices to importersat roughly 5 percent below the U.S. price.With 17 percenttaken as a conservativeestimate of the cost of the steel limitations,foreigners capture 66 percentof the tariff-equivalent revenue,so thatf = 0.66. Ozzello also believesthat the world supply of steel to the United Statesis highlyelastic, so that the fall in the U.S. price of steel with quota elimination would be t = (0.17/1.17) = 0.145. Equation(2) can be used to calculatethe deadweightloss effectsof the steel quota.Table 5 and equation(1) implythat the importprice elasticity of demand for steel should be high since DIM = 13.1. If em - 10,42 the deadweightloss is as follows: DWL = 0.5 (0.145)2 (10)(2009) = $211 million. In additionto this loss, the tariff-equivalentrevenue captured by foreigners, from equation(6), is RF = (O.66)(O.145) 2009 (1 + 0.66(0.145)) $175 million. Thus, the total annualcost to the United Statesof the steel quota on SIC 3312 is $386 million. Textiles. At present,there are two sets of restrictionson U.S. imports of textiles:the long-termagreement on cottontextiles (LTA) and the voluntary agreementswith Japan,Taiwan, Korea, and Hong Kong to limittheir exportsof wool and man-madetextiles to the United States.43Because of the nature of these agreements,most of the tariff-equivalentrevenue is probablycaptured by foreigners.Mintz found that the biggestsocial cost would be in the area of woolen and man-madetextiles. She hesitatesto place a precisenational cost on the textilelimitations because the data are inadequate,but takes $1.25 billion per year as a best guess (roughlyhalf of her estimateof the consumercost). She estimatesthat this numberwill doublebetween 1972 and 1976.To be conservative,I shalluse $1.25billion in this exercise. 42. This value equals the "high" value used in Baldwin and Mutti, "Policy Prob- lems," App. 1. 43. See Mintz, U.S. ImportQuotas, for a discussionof the historyand the administra- tion of these agreements. Stephen P. Magee 673 Sugar. Mintz has calculatedthe consumptiondeadweight loss of the sugarquota at $10million to $25million per year and the productiondead- weightloss at $79 millionto $110 millionper year,yielding a total annual deadweightloss of $89 million to $135 million. She estimatesthe tariff- equivalentrevenue paid to foreignersat $265million to $317million. Sum- ming the means of these rangesgives an annual averageloss due to the sugarquota of $403 million. Meat and dairyproducts. Bergsten has noted that the annualconsump- tion costs of the meat importquotas is $350million.44 From equation(3), the ad valoremtariff equivalent is 3.5 percent,and the pricedecrease with the eliminationof the quota is 3.38 percent.45I assume arbitrarilythat foreignerscapture half the tariff-equivalentrevenue, and that the price elasticityof demandfor meatimports is -10. Thisyields a deadweightloss of $3.2 million and tariff-equivalentrevenue lost to foreignersof $9.3 million.The total annualsocial cost is $12.5 million.46 Bergstenestimates that approximately$500 millionis paid annuallyby consumersof dairy productsbecause of the quota.47From equation(3), the tariff equivalentis 4.6 percent,implying that t = 0.044. Assuming, arbitrarily,that the foreignsupply is infinitelyelastic and that the U.S. import priceelasticity is -20 (see the largenumber for D/M in Table5), the deadweightloss is $1.9 million and the revenuecaptured by foreigners equals $2.2 million, or a total social cost of $4.1 million a year. Totalquotas. The sum of the annuallong-run costs to the United States of the six quantitativeimport limitationsconsidered in this section is $3,555 million. Assumingthat the textile loss is equallydivided between deadweightloss and tariff-equivalentrevenue lost to foreignersimplies that $2.4 billion of the $3.6 billion annualcost of all quotasis deadweight loss and $1.2 billion is tariff-equivalentrevenue lost to foreigners.While

44. Bergsten,"Cost of Restrictions,"p. 4. 45. GeoffreyH. Jackson,"The Impact of Eliminatingthe Quota on U.S. Importsof Beef," AgriculturalEconomics Research Paper 338 (CornellUniversity, Department of AgriculturalEconomics, January 1972; processed),estimates that eliminatingthe beef quota would result in a 2.6 percent decline in the U.S. price in 1975. 46. Bergsten notes that the voluntary restraintsby the major foreign suppliers of fresh and frozen meat hit low-income families relativelyseverely since most meat imports are used in the manufactureof low-costitems such as frankfurtersand hamburgers ("Cost of Restrictions,"p. 4). PresidentNixon liberalizedthese controls somewhatin 1970 because of these adverseprice effects. This exerciseignores the recent suspension of the beef quota. 47. Ibid. 674 Brookings Papers on Economic Activity, 3:1972 the costs of quotasseem high relativeto the previoustariff estimates, two factsshould be keptin mind.First, the deadweightlosses of the restrictions increasewith the square of the ad valoremequivalent tariff rate. The relatively high rates on petroleum,steel, and textiles generatelarge losses relativeto the size of imports. Second, loss of tariff-equivalentrevenue accountedfor almosta thirdof the U.S. social loss due to quotas.Tariffs, of course,do not impose this cost.

COSTS OF EXISTING RESTRICTIONS ON U.S. EXPORTS What would be the benefitsto the United States of removalof foreign barriersto its exports?In the import section we assumedfor simplicity that the United Stateswas a price taker in world markets;that is, tariff reductionsin the United States would not increaseworld prices notice- ably. However,since countriestend to specializemore in productionthan consumption,foreign demandfor U.S. exports cannot be taken as per- fectly elastic. Thus I assumethat the short-runelasticity of demandfor U.S. manufacturedexports is -1.0 and that the long-run elasticityis _4.5.48 U.S. exportscan be brokeninto three groups:manufactured goods in which exportsare an importantpart of domesticproduction, agricultural exports,and all otherexports. In the last categoryexports are quite small relativeto total productionand the exportsupply elasticity is likely to be extremelyhigh, so that foreigntariff cuts will not significantlyaffect the price of these goods to producersor consumersin the United States. Hence, the welfareeffects are negligibleand the categorycan be ignored here.In the firstcase, U.S. industriesin whichmanufactured exports were a substantialportion of sales exportedan estimated$13.5 billion worth of goods and sold $77.3 billion in the United States in 1971.49 This leaves

48. In Magee, "Tariffsand U.S. Trade,"the direct price elasticitiesof demandused for U.S. exports in a fifteen-regionmodel were -2.7, while the cross-priceelasticities with respect to other suppliers to each market averaged 1.7. Therefore, the elasticity appropriatefor foreign tariffelimination on both U.S. and foreign exportsis the differ- ence betweenthese two elasticities, or -1.0 in the short run. The assumptionthat the long-rundemand elasticity equals -4.5 is arbitrary. 49. BLS data indicatethat in 1967, U.S. manufacturingindustries whose exports accounted for a substantialportion of total sales exported$9.49 billion; the value of total shipments in these industriesin 1967 was $71.84 billion. The included industries are those in which exports "represented10 percent or more of the value of domestic output ... [or]represented at least 6 percentof domesticoutput and had a value [of exports] StephenP. Magee 675 $22.3 billion in "other"U.S. exports(which are not examinedhere), and $7.8 billion in agriculturalexports, giving a total export level of $43.6 billion.50 Manufacturedexports. Considering first the $13.5 billion in 1971 U.S. manufacturedexports that are an importantpart of total productionin theirrespective industries, I assumedomestic supply and demandelasticities for these goods that give plausiblevalues of the U.S. export supply elasticities.It turnsout that the domesticelasticities given in the informal table on page 665 for directlycompetitive U.S. importsgive reasonable valuesfor the U.S. exportsupply elasticities, s,, using formula(7):

(7) sx = S (e- - ed) + ed,

short-runs, = 6.7 [0.5 - (-0.25)] + (-0.25) = 4.8, long-runs- = 6.7 [1.5 - (-0.75)] + (-0.75) = 14.4, whereS is the value of the supplyof exportableproducts produced in the United Statesin 1971and equalsthe sum of the amountsold in the domestic market, D, plus the amount exported, X. In this case, S = (D + X) = (77.3 + 13.5) = $90.8 billion in 1971, and S/X = 6.7, giving a short-run supplyelasticity of 4.8 and a long-runelasticity of 14.451The supplyand demandelasticities for U.S. exportsare as follows: Elasticity Short-run Long-run Foreigndemand, d. -1.0 -4.5 U.S. exportsupply, s. 4.8 14.4

of at least $150 million" (BLS, "Productivityand Unit Labor Costs," pp. 507-08). From subtraction,$62.35 billion is the value of output in these industriesthat was sold in the United States. The $9.49 billion in exports was 31 percentof total nonmilitary merchandiseexports of $30.6 billion in 1967.The growthfactor for 1967-71 of 24 percent on domestic goods output (EconomicReport of the President,January 1972, p. 200) implies that domesticsales would have been $77.3 billion in 1971.Taking 31 percentof total merchandiseexports in 1971 ($43.6 billion) equals $13.5 billion for the exports in question. 50. Jack J. Bame, "Balanceof PaymentsDevelopments: Fourth Quarterand Year 1971," Surveyof CurrentBusiness, Vol. 52 (March 1972), p. 39. 51. In StephenP. Magee, "A Theoreticaland EmpiricalExamination of Supplyand Demand Relationshipsin U.S. InternationalTrade," A Study for the Council of Eco- nomic Advisers(October 1970), the short-run(annual data, unlagged)U.S. exportsupply elasticitywas estimatedto be in the region of 10. However, this is for total exports and includes the high-elasticityitems in which exports are a small proportion of domestic sales. Thus, the short-runelasticity of 4.8 used here is not unreasQnablefor the exports underconsideration. 676 Brookings Papers on Economic Activity, 3:1972 Theseelasticities can be used with the foreigntariff rates to calculatethe changein the priceof exportablesin the UnitedStates if foreigntariffs were eliminated.The averagead valoremtariff rate on dutiableU.S. exportsto the EuropeanEconomic Community,the five largest countries of the EuropeanFree Trade Association (EFTA), Canada, and Japanis 11.3per- cent, using 1968weights.52 The effectivetariff on the priceincluding tariff is thus 0.113/1.113,or 10.1 percent,which is denotedas %AT. The percentage changein the dollarprice of U.S. exports,%zAPX, is shownin equation (8):

(8) %zAPX= (d -" ) %AT. Substitutingthe valuesof the supplyand demiandelasticities into equation (8) yieldsthe followingpercentage changes in the priceof exportablesin the United States: short-run %OAPX= 1.0- 8)10. 1 1.75

long-run %APX= (45-144) 10.1 2.40.

In the short run, foreigntariff elimination raises U.S. home prices of ex- portablesby 1.75 percent,while the long-runprices of U.S. exportables rise by 2.4 percent.Foreign tariff reductions lower the deliveredprices of U.S. goods abroadbut raise the pricesreceived by U.S. exportersas the increasein the foreigndemand causes U.S. producersto move up theirexport supplyfunction. The percentagechanges in the priceof U.S. exportsjust calculatedcan be treatedas the tariffchanges were in the section on imports.Thus, t = 0.0175in the shortrun and t = 0.024in the long run.The short-and long- run effectsof foreigntariff elimination on U.S. exportablesare shown in Table6. The long-runresults in that table are shownin Figure5. I assume, as before,that OP8-= 1. The total gainin welfarein the UnitedStates as a resultof foreigntariff elimination is equalto the sum of the deadweightloss trianglesplus the amount of tariffrevenue that was formerlycollected in foreigncustoms but is now capturedby U.S. exporters.These gains are the sum of areas D and E and equal $246 million in the short run and $380 million per year in the long run. 52. These numbers are derivable from GATT, Basic Docuimentation.See Magee, "Tariffsand U.S. Trade,"for these calculations. StephenP. Magee 677 Table6 also showsthat the value of the domesticproduction of export- ablesincreases by $2.4 billionper yearin the shortrun and $5.6 billion in the long run.Because the U.S. priceof exportablesrises, U.S. consumersof exportablesare worse off: In the short run the consumerloss is $1.4 billion and in the long run it equals $1.8 billion. The increasein producers' surplusequals the sumof the redistributioneffect from U.S. consumersplus eliminationof the deadweightlosses plus area E, which is a portion of revenueformerly collected abroad on U.S. exports.The value of U.S. exports increasesby $1.4 billion in the shortrun and $5.1 billion in the long run, owingto threefactors: the greatersupply of exportablesin the United States,the reduceddemand for exportableproducts in the United Statesas a resultof theirincrease in price,and the revenueformerly paid to foreign customsplus the deadweightlosses. Agriculturalexports. The gains from removal of restrictionson U.S. agriculturalproducts are potentiallylarge. Even though only about 20 percentof U.S. exportsare agriculturalproducts, the nation does have a comparativeadvantage in a numberof productsthat are subjectto heavy protectionabroad and restrictedproduction at home. In 1969, exports provided14 percentof farm cash receipts;however, agricultural production was 7.3 percentbelow potentialand exportscould be increasedby 77 percent,according to Upchurch,if idle resourceswere utilized.53In con- trast with the usual case, an expansionof U.S. agriculturalexports, even without an increasein their price, would still mean a U.S. welfaregain since domesticprograms designed to keep farmresources idle in orderto maintainfarm income could be dismantled.The opportunitycosts are near zero for the resources-namely,land-that could be used to expand U.S. agriculturalexports; every dollar of additionalexports thus represents an additionto U.S. welfare.Lawrence B. Krause,in a studyunder way at this writing,and D. Gale Johnsonagree that the eliminationof barriers againstU.S. agriculturalexports in WesternEurope and Japancould lead to an annualexpansion of at least $3 billion to $5 billion in U.S. exports and welfare.54In whatfollows, I assumea short-rungain of $3 billionand a

53. M. L. Upchurch, "CompetitivePosition of U.S. Agriculture,"in UnitedStates InternationalEconomic Policy in an InterdependentWorld, Vol. 1, pp. 836-38. 54. D. Gale Johnson, "The Impact of Freer Trade on North American Agricul- ture"(paper delivered at the joint session of the AmericanEconomic Association and the AmericanAgricultural Economics Association, Toronto, December29, 1972). 678 Brookings Papers on Economic Activity, 3:1972 Table6. Short-runand Long-run Static Effects, from 1971 Base, of Eliminationof Tariffson MajorU.S. ManufacturedExports Millions of dollars, annual flows Changein variable due to price change Short Long Analogous Variableand formula run run area in Figure5 Deadweightloss (D WL)elimination 10 56 D = D1 + D2 Consumption,-0.5t2Ded 3 17 D2 Production, 0.5t2Se8 7 39 D Increasedvalue of domesticproduction 2,404 5,599 I+ 2D2 + E + G1 + 2D1 Redistribution:consumer loss, 1,350 1,872 1 tD - 0.5t2Ded Redistribution:producer gain, 1,596 2,252 I + D + E tS + 0.5t2 Se, Increasedvalue of exports 1,388 5,096 G2 + 2D2 + E + 2D1 + G1 Increasedsupply, tSe,(1 + t) 808 3,347 G1 + 2D1 Reduced demand,tDed(I + t) 344 1,425 G2 + 2D2 Revenue, t(S - D) 236 324 E Total U.S. welfaregain 246 380 D + E

Sources: Derived from the indicated formulas, where all the symbols and the values of ed and es are as defined in Table 4, and (from text) D = $77.3 billion, S = $90.8 billion, and t = 0.0175 and 0.024 for the short and long run, respectively. The calculations were made from data before rounding. long-rungain of $5 billionper year. Thus, the biggestgains to freetrade for U.S. exportsare in the agriculturalarea.

ADJUSTMENT FOR TRANSITION COSTS, AGGREGATION BIAS, AND GROWTH Transitioncosts. Previoussections have establishedthat the gains from eliminationof traderestrictions are smallin the short run but build up as the economyadjusts to free trade.However, the costs involvedin moving resourcesfrom one sectorof the economyto anothermust be considered. Trade liberalizationmoves resourcesout of import-competingactivities and into nontradableand export activities.Since this analysis takes a long-runview of the efficiencyquestion, I measurethese resourceshifts at full employment.The nation has been both above and below full employment in the past decade and, in any case, these calculationsshould not includethe cost of failureof aggregatedemand policy, in eitherdirection. Stephen P. Magee 679 Figure5. Long-runStatic Effects from 1971 Base of Eliminationof Tariffson U.S. MajorManufactured Exports Price

Supply

\ R R S' S PI D2 E P~~~~

HI Pu,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Demnand

0 U U' T' T Quanitity Value (billionsof dollars, Area, annualflows) Definition I 1.9 Redistributionfrom consumersto producers H1 + G2 77.3 Old level of domesticconsumption D = D1 + D2 0.1 Deadweightloss E 0.3 Old value of foreign tariffrevenue G2 + 2D1 + E + 2D2 + G1 5.1 Change in the value of exports F 13.5 Old value of exports RSTU 18.6 New value of exports Gi 3.3 Increasedsupply of exports (one of the componentsin changein value of exports) G2 1.4 Reduced demand for exports (one of the components in change in value of exports) Sources: Same as Table 6. a. Not drawn to scale. 680 Brookings Papers on Economic Activity, 3:1972 Movingresources is not frictionless,and the need here is to measurethe cost of movingbelow potentialgross nationalproduct (or insidethe economy's productionpossibility curve) in the transitionperiod after trade liberalization.There is no clean measureof this social cost: As a proxy, I follow Baldwinand Mutti55and use the income(and presumablythe production)lost by labor as it moves from one sector to another.For simplicity,I ignorethe short-runoutput response, calculate the long-runoutput changeand the impliedchange in employment,multiply this by a wage rate and an assumed duration of unemployment,and spread this loss equallyover the five-yearperiod that I assumeindustries require to adjust to changesin tradebarriers and reacha new long-runequilibrium. The resultsfor both U.S. importsand U.S. exportsare shown in Table7. The notes to the table explainthe mechanicsof the calculations.The pur- pose of the tableis not a precisegauge but a crudemeasure of the economic cost of movingresources; among its flawsare the failureto examineforeign nontariffbarriers and the guessworkthat underliesmany of the estimates. Becausepeople must move from onejob to another,expanding exports en- tails costsjust as contractingimports does, but the costs arelower because less time is lost, among other things.To the extentthat resourcesleaving import-competingindustries are absorbed into export industries, the transitioncosts are overstated.However, the skill requirementsin the two generallydiffer sufficiently that this overstatementwill be small. The last threerows of the tablegive the presentvalue of thejob-change costs for the assumedfive-year period over which they take place. Aggregationbias. In additionto the costs of moving resources,correc- tion must be made for certainaggregation biases in the estimatesof the deadweightlosses imposedby U.S. and foreigntariffs. Tables 4 and 6 indicate that, for fixedvalues of the relevantelasticity and the base, the social cost of a tariffis a functionof the squareof the tariffrate. The deadweight loss calculationswould not be subjectto aggregationbias if each of the componenttariff rates were equal. However, if they arenot (as is the case), aggregationbias exists,and it increaseswith the varianceof the tariffrates aroundthe mean. To correctthis situation,I have constructedthe "root mean-squaredtariff," tr, which, at the aggregatelevel, yields a bias-free estimateof the welfareloss of the tariffstructure: (9) tr aiti,

55. "Policy Problems." StephenP. Magee 681 whereti is the middleof tariffrange i (for example,ti = 0.015 in the range between t = 0.01 and t = 0.02) and ai is the share of total trade with tariff rates falling in range i. The averagetariff rate used in the earlier calculations, ta, is (10) ta = aiti The proposition that tr > ta can be illustrated in the following two cases. Both deal with two products,with equal weightsin imports(al = a2 = 0.5) and the same mean total tariffrate; in case A, t1 = 10 and t2 = 10, while in case B, t1 = 0 and t2 = 20. The subscriptsindicate the product number.According to equations(2) and (9), the true welfarecosts of case B are greaterthan those of case A:

Case A DWL = 0.5emV(0.5 X 0.12 + 0.5 X 0.12) = 0.01(0.5emV)

Case B DWL = 0.5emV(0.5X 02 + 0.5 X 0.22) = 0.02(0.5emV). The deadweightloss estimatesfor importsin Table 4 ($97 million and $291 million)and equation(2) ($81 million and $202 million)and for ex- portsin Table6 ($10 millionand $56 million)must be adjustedupward by a correctionfactor, C: 2

(11) C = tr a Using tariff dispersiondata from the GATT study,56I estimate that tr = 0.100 and ta = 0.059 for total (not just dutiable)U.S. imports, so that C = 2.87. Multiplyingthis factor times the deadweightlosses of Table 4 gives the followingestimates, which are free of aggregationbias: Short-run Long-run deadweightlosses deadweightlosses (millionsof dollars) (millionsof dollars) Directly competitive 278 835 Other 232 580 Total annuallosses 510 1,415 As a proxyfor the calculationof t, for U.S. exports,I use data on dispersion for "worldtariffs," which, in additionto importsinto the United States, includesthose of the EEC and nine othermajor developed countries. The results yield t, = 0.083 and ta = 0.051, so that C = 2.65. Based on this 56. Basic Documentation. 2g .g8u > St*t: tn t_ o^o^t?~~~~~~~enC1

N00 0 ~~ I,-)

O~~~~~~~~~~~~0 ONr-

S R N~ o O n ~~~~~~~~~~~~~~e *@ON O

0~~~~~~~0

s I i;: : O t 00 0 - 00 o -~~~~~~~~~~~~~~~~~l - Y

B r- l" t 00

0~~~~~~~~~~

00~~~~

{ ~~~~~~~~~o=

6 >o I~~~~~~~~~~~~T4>PC 0 4-4~.~o - q ~ C,, ~ O - 0 >~~~~~~~~W An Y t In W F EECltzkE,E

"O 01%C4, - C 4

I; clie V. j'0i 6 - - 0 uuo 0 C's ,00~~~~~~~2~~~~~~~*~~~~~~~~0 C2 ec

o'0 00 ,

.0 .0C' C,3 40 -,% -. - 13 '04 3 r- .0 C1 O404c)44ctO 0 0 (I 4) 440 ;

~~~~~~~~~4)0~~~~~~~~~C ) 0 4 ~~~~~~~~~~~~~~~~~~~~~~~~~ 0 E too 00 )~ 0 ~ ~ ~ ~ ~ ~ ~ * Z0 0 0 ~~j 0 O O3~~ ~0 % O 0.04) 0l 400 )

0- 4444. > 4)

cd '004 0U4. 44 O

v >44en w o~~~~~~~

0 .0 00 ~ ~ ~ ~ ~ ~ 0 0 . ~' 0~ 0 ~~4~U4444.e4 0~4 C, cl 4) 4) C)~~~~~~~~~~~4 >0\,O to ~ ~~~~~~~ 0~~~~S0cr 1 0) ~ X

c 0U. 0 440to > 00004 a 44~~~~~~~~~~~~l. t, cd - n .. 0~~~~~~~~~~42 44444)4 c 44 44 4 E U2 A OO4 40Cu 0 U Ch 0~~44~)4 ~ 0 684 Brookings Papers on Economic Activity, 3:1972 factortimes the deadweightlosses in Table6, and the revenuecomponents, $262 millionis the short-runand $472 millionthe actuallong-run annual costs of foreign tariffson U.S. exports of manufactures.The following discussionrelies on these correctedfigures. Three aspects of these adjust- mentsshould be noted. First, the largesizes of the correctionfactors indicate that the tariff distributionsare considerablyskewed. Second, the largeradjustment factor for U.S. importsreflects the "tariffdisparities" issue for whichthe United Stateshad to makeconcessions in the Kennedy Round.Third, the tariffdistribution data were not availableseparately for the two categoriesof importsused here. Thus, a single correctionfactor calculatedfrom all manufacturedtrade was appliedto both importcate- gories. Growth.Thus far, the analysishas been static.The only changesin the flows throughtime involvedmovement from short-runto long-run posi- tions. If in fact domesticsupply and demandgrow at an annualrate, g, the estimateof deadweightloss (see Table4) for consumptionin yeari can be written (12) DWLi = 0.5t2edDo (1 + g)i, whereg is the proportionalannual growth rate and Do is the initialquan- tity of the importabledemanded. The presentvalue of the DWL in year i can be written

(13) .5t2edDo(1 + g)i (l+ r)t = 0.5t2edDo(l + d)-i, whered is the discountfactor, incorporating both growthand the rate of capitalization,r, and approximatelyequal to r - g. I assumethat the ap- propriaterate of social capitalization,r, is somewherebetween 6 and 12 percent.The real growthrate for the economyas a whole and for the con- sumptionof items directlycompetitive with importscan be approximated at 4 percent. Thus, assuming r = 0.08 and g = 0.04 gives d = 0.04, that is, a discountfactor of 4 percent. The presentvalue of the consumptiondeadweight loss, includinggrowth, is the sum of all futuredeadweight losses discountedat 4 percent.Table 8 gives the costs of U.S. tariffsand quotasat alternativediscount rates of 4 percent,7 percent,and 10 percent.While 4 percentis probablythe most plausibleestimate, the 7 and 10 percentrates were includedto permitthe ci ci 0 entOr vt Woci 0 A ' oe"o c o 0 CS4 tr- 8O o 00 W o N 0 s ci~c4-) Of) Cf4 OiN o0 N

Cu t t-0fN _i 00 e 4 f .= o No XCt m m N 0 0 t 00 '--4

N o - 0-- No eo ? - N .OtO gr -oNc(N 0n t c4-)q 0 --) Oo)(7%0 00 of&0 ONci W0 0)0

x CN tNct- 000 00 tO -o tn ? (oN N 0 ) 000 cX Ci M ho 0 en c t) 4 to0 0 N e- 00- o

0 0 '-~~ cicicicici - 00~~~~00 O\ ei,; q I m c 00 0c C t'- 00 eo --04 0t r s ?tci ?I 00 C00 . O'O m m C1 Oo N ~o C14to *

J O r- t N 0 n 00 ?? ?0 0 1 M I . 5 C-q-,-rtr) cc~0 -14 C>- &)0 o 5 -4C. C)4 OfON.- f) t)0 C) N W) - q CN cs q I oot . C) NO t C0 t ON '-4 00 00 0 C) C b. 0io ' t-0t)) 0o oO 0 tN c c c c i~l ci~cl~ -4 -z 006 00, '00 , - -- '--4 tno 00 00 o G) C)i O COt 00 00 00 t ci 0 ^ oo c o~ u:~~~~~~~~eC. M ol C, V-1 c) N (= o, D Cu ciciz'-)C~)C)00fNN e) '. tf) 0N

<~~~~~~~~~c3 sv~cN t-W c f 'f'- o N 000o '.co 00 0o N 403 xo o t- O t O - .X C4 C' en N N cc 4 OC) 0 t- o - e 00O N 0 ?o ?00)oo FF ho cs O O)Itt n t 000o C)o N 0 C)r-en C O - O'4i O - "?0 t c o Q~~~~~~~~~~ en 13 ^It 00 oo4 C1 r tr M .o ~~~~~~~~~~~~~~~~~4~ cl *E oi 4 l r -.~~~~~~~~~~~~~~~~~~~~~~~~~~-

000 o N ci00 ooN tt _ - N ci =4 C) Q~~~~~~~~~-r~ g bONoNN'.0ci0 c O0X W) )N cm N 0 0 ene lwr N m 00 0

00ci)tOf) 0 ) ci- Ct- C ,'4 0~ 0 i 0t oR bb~~c X XIt mtn 1.0 en tr tn oo . ~~ 000000 ~~~~~~00N ~~~~t 0~~~ N '.0 0' '~~~~~~.0 o N _ ) -N C) ON 0 c4ON 00 '0

o E cic>)~tOf)O 8co 0O- 00 '0 t1 ?.0

4o i-4 C 0

0 ~~~~~~~~~~~~~~~~)ci - - 0 0 ' 000 C) CA *W W j 2 . - 0)O m O C)O 0 0 Cuan 00?? ?? C 0~~~~~00 0~' C, C (I) 0~~~~~~~~~~~~~~~C 0cC)0 . 686 Brookings Papers on Economic Activity, 3:1972 discountingin the absenceof growthor alternativegaps betweenr and g. Thenext to last line of Table8 givesthe total presentcost of eachrestriction and the previousthree lines provide some feel for the timepath of the total presentvalue. For example,59 percent(12.1/20.4) of the gain from eliminatingthe tariffon directlycompetitive imports comes afteryear 15 when a 4 percentdiscount is used. The componentsof the presentvalues are brokeninto four time periods.If one believedthat Mundellian-typefactor movements(see note 21 and the relateddiscussion) would eliminatetrade afterfifteen years, the fourthperiod should be ignored. The incomelosses due to job changesin the firstfive years are subtracted from the gains, and the resultsshown in the line labeled, "total, 1 to 5, net of job-changecosts." For nonquotaimports in the firstfive years,the costs ofjob changesabsorb nearly a thirdof the short-rungain from reduc- tion in the tradebarrier. In the long run,the gainsfrom free trade are more important.Since most politicalhorizons are rather short, this helpsexplain part of the politicalresistance to freertrade. Table 9 gives the same data for U.S. exports,and shows that the gainsto the United Statesfrom free worldtrade for agriculturalproducts dominate. Remember, however, that no estimatesare made in this paper of the costs to the United States of foreignnonagricultural quotas and nontariffbarriers. Also, since there is no reason to believe that trade liberalizationwill cause changes in the U.S. trade balance,no calculationsare made here for the effectsof cur- rencyrealignments induced by tradeliberalization.

Movingto More RestrictedTrade

CUTTING TRADE TO 1965-69 LEVELS An extremeprotectionist proposal is containedin Title III of the Burke- Hartkebill to amendthe tariffand tradelaws, introducedin the Congress in September1971. They provided that the quantityof U.S. importsin the first year (1972), both by productcategories and by country,should not exceedthe averagequantities imported during the calendaryears 1965-69. The productcategories were to be definedby a proposedUnited States ForeignTrade and InvestmentCommission (FTIC) on the basisof the five- digit and seven-digitclassifications used in the Tariff Schedulesof the United States,Annotated. Products were to be groupedso as not to "ad- 1.0 N ON tn O t- - 0 -o vt 6 o l t-Joo - 1 J en en en e cn t t- Cl00 t 0 t0

~~~~ro - r " c -4 - oo en oo en e r b -b ~~~~~~~~oCW en 00o N ~ o oo 0o r- r- o *i >~~~~~~~~~~~~~~~~~~~~~~~~* U) R s O~0~r . oo 00 a -q arO 0. 0 0

C) "o en V-4 t, C. X N ?~~~~~~~~~~0oo C 0 t o C _ C e _ %C_0N W) t- _ 0 (ON In o o. - enen _?t C _. _0 O0 -I' e(ON- Cl o

0 X~~~~~0tOt0 ee .0 o o

- - - ~~~~~~~~-4 It) I) 00) 00 > 0 t)000' Ot) 00 0 0 0 0 00 o m OCl~t000 C ' 0 ' 00 00 O

~~-4 - - - ~~~~Clt .-. 0

o 0 O)OO) )oVI )W J 0 0 0) I) -,t o)0 O 00 0 C) It It) 0 0 e c k 0 -C 00)Cl 0 %C - C . C, zOntjz00 O N 00 t) Cl I) t) - .-4 b N c

0 .0 -~~~~~~~~~~1ffi~~~~~~~~~~C ne ntn "4C4 t-Clt C ( 14 1mo >N 00000^O001o tla -4It)Cl) oW oo 0oN0.0 Cl e It)t n C )

C-4-4 e ~c - - U). | 0 00 -4'0000I C 00 W ene 0 .w 0- QQ enWC) tl 0b 0 ON 0C ,t (N C0 *i E GB W~~- 000 -r- en enene %C e t- Oo r- -, b

gO N00 e?oh.Clo~0N000 0 '0 O 0 00 o ~~--z -; -u -

0~~~~~~~~~~~~~~~~~~~

'4-~ ~ ~~~~~~~~~~~~~-

o > 0~~~~~~~~~ 0; 'o0 -6 0 01 0 W~~~~~~~~~~~~~~~~~0~ 0 -W

4->IC 04 0o -4- j Ct) II),- 0 0 688 Brookings Papers on Economic Activity, 3:1972 verselyaffect the assemblyor productionof any item or componentin the United States."57 The bills providedthat after 1972the quotascould be increasedbut the ratio of importsto domesticproduction was to remain at the 1965-69 levels.The FTIC could decreasethe level of any quota if it foundthat imports were "inhibitingthe productionof any manufacturedproduct,"58 a particularlystriking provision since the wholepurpose of competition,including foreign competition,is to disrupt inefficientproduction. Con- versely, the FTIC could relax the quotas on inputs and intermediate productsthat would"inhibit the production"of U.S. goods. If in any year, the FTIC determinedthat a foreigncountry was unlikelyto make full use of its quota, the bill providedthat "it shall so informthe President,and shall distributethe quota amongnew or existingsuppliers as the President may direct."59 The only importsexempt under Title III are goods in whichthe FTIC decidesthat (1) quantitativelimitations on U.S. importsare alreadyeffec- tive (throughvoluntary bilateral or multilateralgovernment-to-government agreementor legal controls);(2) "failureto importthe goods would cause long-termdisruption of United States markets";and (3) "the domestic industry producing competing goods has consistently failed to make technologicalinnovations required to remaincompetitive with foreignpro- ducers."60 What would be the effects of the quantitativerollbacks in trade from 1971to the 1965-69levels (1972 data werenot availableat the time of this writing)?Table 2 listed those items that would be exceptedfrom the controls becauseof existingquotas or government-to-governmentagreements. Excludingsteel, these imports amount to approximately$7.2 billion(based on 1971 data).61If the FTIC grantedexemptions to avoid long-termdisruption of U.S. marketsfor goods totally without domesticsubstitutes, another$2.8 billionwould be exempted(out of the $24.1billion of imports in the "other"category), making total exemptionsapproximately $10 bil-

57. H.R. 10914, Title III, Sec. 303(e); S. 2592, Title III, Sec. 303(e). 58. H.R. 10914, Title III, Sec. 301(b)(2)(B). 59. Ibid., Sec. 301(f). 60. Ibid., Sec. 301(d)(1)-(4). These exemptions would not apply, presumably,to automobilesimported into the United States under the Canadianauto agreementsince no restrictionsare involved. 61. Inadequatedata on steel dictatedexcluding it from the exemptionsconsidered in this analysis, and it is consideredsubject to rollback here. Stephen P. Magee 689 lion. This would reduce 1971 importssubject to rollbackfrom the total importlevel of $45.5 billion to $35.6 billion. Columns(5) and (6) of Table 10 show the average1965-69 and 1971 quantitiesof U.S. importsin millionsof 1958 dollars,and column(7) the ratio betweenthem. However,trade will not be cut back for each country in proportionto the amountsindicated in column(7) becauseof the $10 billion in exemptions.Columns (1), (3), and (4) give the value of U.S. imports in 1971dollars, importsexcluding exemptions, and the exemptions themselves.Columns (8) and (9) give the portions of trade affectedand unaffectedby the cutbacks. The ratio of the proposedto existing1971 total U.S. importsby country equalsthe cutbackratio in column(7) timesthe ratio of applicableto total trade in column(8) plus column(9). This is shown in column (10). The percentagedecreases in the quantitiesof total importsare in column(11). For the EEC countries,the importcutbacks range from 20 to 31 percent; for EFTA countries,around 10 percent;for Canada,29 percent;for Japan, 44 percent;and for the less developedcountries (LDCs) (the bulk of the rest-of-the-worldregion), 13 percent. The priceelasticities of demandfor U.S. importsby country,estimated using 1951-69 annualdata, are shown in column(12). Column(13) gives the percentageincrease in the price of U.S. importsby country,using the percentagedecreases in the quantitiesin column(11) dividedby the price elasticities. Two types of price increasesare calculated:those based on total trade (column 13) and those based on trade that is subject to the rollbacks (column14). In the lattercase, the pricesof U.S. importsfrom twelve of the fifteenregions experience price increases of less than 20 percent.Low price elasticitiesof demandfor importsfrom Canadaand the LDCs cause the prices of their productsto rise by 71 and 38 percent,respectively. The averageimport price increasefor all productssubject to the quantitative cutbackis 36 percent.For simplicity,I assumethat the supplyelasticities of U.S. importsby countryare high, so that the proportionatecutback in the quantitiesand the valuesof trade are the same. Thus,a decreasein the valueof total tradeshown in column(15) is simply the quantityrollback in column (11) times the value of 1971 importsin column (1). The hardest-hitcountries are Germany,with declinesin exportsto the United Statesof $1.1 billion;Canada, with $3.7billion; Japan, with $3.2 billion; and the LDCs, with $1.7 billion. The total value of the 690 Brookings Papers on Economic Activity, 3:1972 Table10. Effecton 1971 ImportLevels of QuantitativeCutback to 1965-69 Levels Dollar amountsin millions (1) (2) (3) (4) (5) (6) (7) (8) U.S. imports Share of U.S. imports (1958 dollars) trade Ratio of affected by Total, Imports Average 1965-69 Burke- 1971 Share of excluding Exceptions quantity 1971 to 1971 Hartke Country imports total exceptions (1) - (3) 1965-69 quantity (5)/(6) (3)/(1) United Kingdom $2,461 0.054 $2,193 $268 $1,647 $1,912 0.861 0.89 Belgium 848 0.019 773 75 621 802 0.774 0.91 Denmark 286 0.006 270 16 189 242 0.781 0.94 France 1,080 0.024 1,010 70 703 936 0.751 0.94 Germany 3,665 0.080 3,476 189 1,925 2,846 0.676 0.95 Italy 1,405 0.031 1,211 194 967 1,334 0.725 0.86 Netherlands 532 0.012 458 74 346 448 0.772 0.86 Norway 172 0.004 168 4 137 151 0.908 0.98 Sweden 455 0.010 448 7 315 359 0.877 0.98 Switzerland 492 0.011 458 34 329 337 0.977 0.93 Canada 12,723 0.279 11,737 986 7,027 10,309 0.681 0.92 Japan 7,244 0.159 6,577 667 3,389 6,530 0.518 0.91 Australia 621 0.013 301 320 389 567 0.686 0.48 South Africa 285 0.006 158 127 236 301 0.784 0.55 Rest of world 13,277 0.292 6,321 6,956 9,193 12,586 0.730 0.48 All countries 45,546 1.000 35,559 9,987 ...... 0.78 Sources: The 1971 trade data in columns (1)-(4) were provided by Mary Jane Wignot, Office of the Special Representa- tive for Trade Negotiations. The import data for 1965-69 came from U.S. Department of Agriculture,Economic Research Service, Foreign AgriculturalTrade of the United States, annual issues. The total country export price indexes used to deflate the value data, with 1958 = 100, were computed from various issues of International Monetary Fund, Inter- national Financial Statistics. Price elasticities (column 12), based on annual data for 1951-69, are from Stephen P. decreasein U.S. importsis $11.2 billion, about 25 percentof total trade and 32 percenton tradeaffected by the Burke-Hartkerollbacks. One caveatabout Table 10: At the time of writing,data for the 1965-69 periodwere not availableby countryon the valueof U.S. importsregulated by quotas.Also, the priceindexes used to deflatethe seriesincluded both quota and nonquota items. Includingquota items probablyleads to an underestimateof the actual Burke-Hartkecutback, since the quota price indexespresumably grow faster than the average.Using an index including these items to deflatethe values of tradeleads to an underestimateof the growthin the quantitiesof U.S. importsfrom 1965-69to 1971,and hence an underestimateof the cutback. In the tariffexercise performed in an earliersection, the only variationin the price declinesin each categorywas due to differencesin tariff rates. Thus,if the assumptionof a high foreignsupply elasticity for U.S. imports holds, the percentagedeclines in the import prices of dutiablesdepends only on the tariffsand is independentof the demandelasticities. Use of the StephenP. Magee 691 as Proposedin Burke-Hartke Legislation, by Country

(9) (10) (11) (12) (13) (14) (15) (16) Price increase in U.S. Ratio of (percent) proposed to Percentage Share of actual 1971 decrease in Absolute On On Value of New value unaffected trade total trade value of price total noflexempt decrease in of 1971 trade (7) X (8) [1.0 - (10)] elasticity of imports imports 1971 imports imports 1.0 -(8) + (9) X 100 demanda (11)1(12) [(I) - (7)1(12)] (1) X (11) (1) - (15) 0.11 0.876 12.4 1.55 8.0 9.0 $305 $2,154 0.09 0.794 20.6 4.09 5.0 5.6 175 671 0.06 0.793 20.7 1.11 18.6 19.8 59 226 0.06 0.765 23.5 5.06 4.6 4.9 254 826 0.05 0.692 30.8 2.50 12.3 12.8 1,129 2,536 0.14 0.764 23.6 3.36 7.0 8.3 332 1,073 0.14 0.803 19.7 2.98 6.6 7.7 105 429 0.02 0.909 9.1 1.58 5.8 5.7 16 156 0.02 0.879 12.1 3.35 3.6 3.5 55 400 0.07 0.978 2.2 1.76 1.3 1.3 11 482 0.08 0.706 29.4 0.45 65.3 71.1 3,741 8,982 0.09 0.561 43.9 3.52 12.5 13.6 3,180 4,064 0.52 0.848 15.2 l.Ob 15.2 31.0 94 527 0.45 0.881 11.9 1.20 9.9 18.3 34 253 0.52 0.870 13.0 0.71 18.3 38.0 1,726 11,551 0.22 ...... 1.53 ... 35.9 11,216 34,330 Magee, "Tariffs and U.S. Trade," A Study for the Council of Economic Advisers, Working Paper 14 (June 1972; processed). The calculations are based on data before rounding. a. All elasticities in the column are negative. b. This elasticity was statistically insignificant.However, I assume for simplicity that it equals 1.

import elasticitiesreported on page 665 to calculatethe percentagein- creasein the prices of importablesin the United Statesas a resultof the Burke-Hartkecutback would understatethe actual price increase.The reasonis that the distributionelasticities are less importantwhen the price changesare more or less exogenouslyset (by tariffchanges) whereas they mustbe incorporatedin calculatingan aggregateprice elasticity if the com- ponentsare endogenouslydetermined.62 62. Sufficientconditions for the consistencyof disaggregatedprice elasticitiesand an estimatedaggregate price elasticity have been discussedin T. S. Barker,"Aggregation Error and Estimates of the U.K. Import Demand Function," in Kenneth Hilton and David F. Heathfield(eds.), The EconometricStudy of the UnitedKingdom (Macmillan, 1970), pp. 115-43. An elaboration of this techniqueis discussed and applied to U.S. trade in Magee, "Tariffsand U.S. Trade."This problemwas first broughtto my atten- tion in a paper by WilliamH. White, "Bias in Export SubstitutionElasticities Derived throughUse of Cross SectionSub-Market Data" (InternationalMonetary Fund, March 1970; processed).The total elasticity that is consistent with disaggregateddata can be writtenas follows: e = E e;(m;/m)vi, 692 Brookings Papers on Economic Activity, 3:1972 The datain Table 10can be used to calculatethe implicitaggregate uni- form price elasticity of demand associatedwith the Burke-Hartkeroll- back. Sincethe calculationsare done at a disaggregated(country) level, the averageimport price increase incorporates the appropriatecountry distributionelasticities into the aggregateelasticity. Since the priceof nonexempt importsrises by 36 percentand the quantitydeclines by 32 percent(11.22/ 35.56),the implicittotal priceelasticity is -0.89. However,this numberis lower than desirablefor presentpurposes since it is based on price elasticities (column 12 of Table 10) that were estimatedon total trade (includingquota items). Assuming that the effectiveprice elasticity of demand for quota items is 0 and using 0.16 for the 1971 weightof quota items in U.S. imports,the short-runelasticity for this exerciseon nonquotaimports is -1.06, whichfor simplicityis takenas unity(-1.0). Thelong-run import demandelasticities used above wereapproximately 2.5 times the short-run elasticities.Thus, the assumptionin thissection is thatthe long-rundemand elasticityfor importsis -2.5. Since the Burke-Hartkeproposal reduces the quantityof dutiableim- ports by 32 percent(from $35.6 billion in 1971to $24.3 billion),the price risesby 32 percentin the shortrun and by 12 percentin the long run. These are equivalentto 32 percentand 12 percentad valoremtariff levels on top of existingtariffs. Since tariffs are eliminatedunder the Burke-Hartkepro- posals, the net welfaredistortion of Burke-Hartkemust be measuredby calculatingthe total welfarecost fromfreetrade levels and then subtracting the costs of existingtariffs. The ad valoremaverage tariff rate on dutiabletrade is now 6.8 percent (2.5/36.4). 100. This implies that the long-runand short-runad valorem equivalentsof the Burke-Hartkeproposal above worldprices are 41.0 percent and 19.6 percent,respectively.63 Equation (2) servesto calculatethe where e = total import price elasticity ei = disaggregatedelasticities (mi/m) = share of categoryi in imports vi = distributionelasticity = (dpi/pA)/(dp/p) The variablevi is called the distributionelasticity. It indicatesthe averagepercent varia- tion in the price of category i relative to the total price index, p. Since prices of items with high elasticitiesof demandvary much less than the aggregateindex, vi will be lower for high-elasticityitems than for low-elasticityitems. This negativecorrelation between ei and vi lowers e. 63. One plus the ad valorem equivalentsabove world prices equals (1.32)(1.068)- 1.410, and (1.120)(1.068)= 1.196. Stephen P. Magee 693 deadweightloss portionof the Burke-Hartkecutback. The freetrade value of importsto be used in that formulaequals $40.9 billion in the shortrun and $50.1 billion in the long run.64Thus, the deadweightlosses are as follows: Shortrun Long run DWL = 0.5 (0.410)2(1) (40.9) DWL = 0.5 (0.196)2(2.5) (50.1) = $3.4 billion = $2.4 billion. The long-runannual loss from the quota is less than the short-runloss: The higherelasticity of demandfor importsplus the largerlong-run free trade base is not sufficientto offset the smallerad valorem equivalent tariff. The second part of the loss to the United Statesfrom the quota is the tariff-equivalentrevenue captured by foreign suppliers.It equals the ad valoremequivalent tariffs (0.410 and 0.196, respectively) times the restricted levels of importsunder the cutback($24.3 billion), or $10.0 billion in the short run and $4.8 billion in the long run. Assumingthat half of these amountsare captured by foreignsuppliers, the additionalsocial loss is $5.0 billion in the short run and $2.4 billion in the long run, for a total annual cost of the cutbackof $8.4 billion in the short run and $4.8 billion in the long run. In Table 11, the short-and long-runeffects of the Burke-Hartkecutback are convertedto presentvalues at 4 percent,7 percent,and 10 percent. Rememberthat the discountrates are roughly equal to the rateof capitalization less the growthrates of quota importsin the absenceof restrictions. Sincethis rate is likelyto be high for a numberof years,4 percentis probably the most reasonablerate. It should be emphasizedthat the calculations performedhere assumeno retaliationagainst U.S. exports,and thus minimizethe potentialcosts. As in the tariffelimination case, the shiftof factorsinto import-competing industriesalso createscosts. Assumingthat the entranceinto industries under Burke-Hartketakes only half as long as the departurefrom them underfreer trade, considered in Table7, and with the directlycompetitive and other import categoriesaveraged, present values of the costs of job

64. Directly competitive dutiable imports increase by $2.7 billion in the short run and $8.0 billion in the long run with tariff elimination(see Table 4). Other imports increase by roughly $2.8 billion in the short run and $7.0 billion in the long run. Thus, nonquota importsincrease by 15.1 percentin the short run and 41.2 percentin the long run so that free trade nonquota importsto which Burke-Hartkewould apply are $40.9 billion in the short run and $50.1 billion in the long run. 694 Brookings Papers on Economic Activity, 3:1972 Table 11. Annual Costs to the United States, from 1971 Base; of the Burke-Hartke Cutback on Imports, and Present Values, by Alternative Discount Rates Billions of dollars Annual Present valueof the cutback cost of the Year cutback 4% 7% 10% 1 8.4 8.4 8.4 8.4 2 7.5 7.2 7.0 6.8 3 6.6 6.1 5.7 5.5 4 5.7 5.1 4.7 4.3 5 4.8 4.1 3.6 3.3 Total, I to 5 ... 30.9 29.4 28.3 Total, I to 5, includingjob-change costs ... 31.4 29.9 28.8 6 tol1 ... 18.2 14.9 12.5 11 to 15 ... 14.9 10.6 7.7 After 15 ... 69.6 26.9 12.5 All years ... 133.6 81.8 61.0 All years,including job-change costs ... 134.1 82.3 61.5 Source: The job-change costs and the annual cost for year one are developed in the text accompanying this section. The annual cost is discounted at 6 percent for future years. changesper $1 billion changein output are $45 million at 4 percent,$43 millionat 7 percent,and $40 millionat 10 percent.If the expansionof domestic productionroughly equals the reductionin imports of $11.2 billion,65 the present values of the job-change costs associatedwith the Burke-Hartkeincrease in domesticproduction are $509 million at 4 percent, $486 million at 7 percent, and $452 million at 10 percent. These changesare smallrelative to the costs noted in Table 11. At 4 percent,the present cost of the Burke-Hartkecutback, includingthese job-change costs, is $134 billion.

PROPORTIONAL QUOTAS A second feature of Title III of the Burke-Hartketrade proposal re- quiredthat after 1972trade had been cut back to the 1965-69levels, the ratio of imports to domestic productioncould not exceed the 1965-69

65. This is an extremelyrough estimate,since it assumesthat the reductionin imports due to reduced demand equals the redistributioneffect plus the effect of production deadweightlosses. Stephen P. Magee 695 ratio.This means that importscould growno fasterthan domesticproduction. Startingfrom a 1958-61 base, importsin directlycompetitive areas grew by approximately10 percent per year until 1969, while domestic productionof these goods grew by just over 5 percent.However, import growthwas acceleratedover this periodby a numberof transitoryfactors, so that the calculationof the restrictiveeffect of the proportionalquota assumesthat importsgrow at only 1.5 timesdomestic production. This, in fact, equalsthe incomeelasticity of demandfor importsestimated over the period 1951-69.66This analysisprovides a conservativeestimate of the costs of the proportionalquota by assumingthat the long-rungrowth of import-competingproduction equals real GNP growth and real imports growat 1.5 timesthat rate.From 1950to 1971,U.S. realGNP grewat 3.55 percentper annum.The incomeelasticity factor implies import growth of 5.33 percentper annum. The techniqueused for measuringthe effectsof the proportionalquota is to assumethat importswould grow at 5.33 percentin the absenceof the quota and 3.55 percentwith it. The gap betweenthe actualimport level in year i and the restrictedquota level, Ri, is givenby

(14) Ri = [(1.0533)i - (1.0355)i] Vb, whereVb is the valueof the importsunder consideration in the base period. To providean estimateof the cost of proportionalquotas that is indepen- dent of the Burke-Hartkecutback, calculations here assumethat the non- exempt1971 level of $35.6billion in importsis the base. Columns(1) and (2) of Table 12 show the actualand limitedimports starting from this base. Column(3) is the percentagedecrease in the quantityof importsdue to the proportionalquota. Since the equivalenttariff changes each year, the shortrun elasticityof -1.0 used in the previoussection is applied.This means that column(3) is also the ad valoremequivalent tariff implied by the quota. Column (4) gives the annual deadweightloss and column (5) the tariff- equivalentrevenue lost to foreigners,who are assumedto capturehalf of it. The total costs of the quotaare in column(6), and theirpresent values using 4, 7, and 10percent discount factors are shown in columns(7)-(9). Since growthis alreadyaccounted for, 7 percentis probablya reasonabledis- count rate.The calculationswere not carriedpast the fifteenthyear because

66. See Houthakkerand Magee, "Income and Price Elasticities in World Trade," and Magee, "Theoreticaland EmpiricalExamination of Supply and Demand Relation- ships." oo t- t- (5 C) F A tr- o - t C. 00 m to) oo Il ctn t_ ON CNt o O ._

o o~~~~~Cen OC e N :j C) WI 00 mbo~cr en N

O W t r- tn C) 0O Ct c CN 00 M: 0 tn -.4 c ":t _ 0 _ ~Ct? E Co et oo -h ? ooe 4 taO- _I oo O0

d 0 0 t3 0~~~~~~~~~~~~~~~

21, 00 ~c00 00 00en c i-. 0 0t 00 00t o- 0000 00 o~~~~~~~~~~~~~~~~~~~ en

~~~~~~~~ ~ ~~~O 0 0

w oo O i C3 o- ~~~~~~~~~~0o C) O i oo - ooc -V oo

_/EgN? oY N 4m QO O0C-,t\ " 00 o Q -:

O ~ ~~~ ~ ~~~~~~~~~~~~~~~~~~~__ _ _ C)_ C) _ O _

S @ e < t. 8W~~~~~~, o o t o m tn t o C) eq ?cer o > : -~~~~~~~1 g~~~~~~~~~~~~~~~~~~~~~~~~~Wtn W-!?N n t tmB?bXX

0000~~~~~~~

~s* c u^

0 m tn w0eM t 9'S s 4 r- ".. r-a Ed)c O ooC?o .0~~~~~~~~~~~~~~~~~~

- c~~~~~~~~~~~~------W) .) Z oo C 0~H00 00 M O l 0 0 StephenP. Magee 697 losses grow more rapidlythan the discountrate. Even then, at 7 percent, the presentcost of the proportionalquota is $28.3 billion; it has small short-runcosts but the long-runcosts are high becauseof the cascading restrictiveeffect.

Qualifications

The calculationsin this papershould be viewedas ballparkestimates of the welfareeffects of trade restrictions-they are subjectto great uncer- tainty. At least four nonquantifiablefactors indicatethat the numerical estimatespresented here understatethe true costs of trade restrictions. First, except for some mechanicalexercises, this paperhas ignoredsome importantdynamic gains from free trade.Both economistsand the public at large broadlyagree that economicgrowth leads to an increasein welfare.However, Johnson has shownthat if economicgrowth shifts resources towardprotected industries, it can actuallyserve to worsenwelfare rather than augmentit.67 This occursbecause the benefitsof growthare offsetby the waste of additionalproduction in protectedindustries. A secondqualitative factor is economiesof scale.In areaswhere economies of scale are important,the productiongains are greaterthe largerthe size of the market.Clearly, these production gains will be largerin an open world economythan in one of many small submarketsseparated by high tariffsor otherrestrictions. Furthermore, the frameworkused here cannot serve to quantifythe welfaregains from expandedmarkets in areas of decliningcosts. Third,the effect of trade on the antitrustproblem should also be considered.If U.S. car manufacturersand automobileworkers faced no competition from importedvehicles, the resultsof wage negotiationsand the increasesin both the size and the priceof automobilesin the United States would worsen.Thus, free tradeis a stimulusfor greatercompetition and industrialefficiency in the United States. Quotas are an especiallybad form of restrictionbecause they may convert a potentialinto an actual monopoly.

67. Harry G. Johnson, "The Possibilityof Income Losses from IncreasedEfficiency or Factor Accumulationin the Presenceof Tariffs,"Economic Journal, Vol. 77 (March 1967), pp. 151-54. 698 BrookingsPapers on EconomicActivity, 3:1972 Finally, Tullock has enumeratedother costs related to protection.68 First, the systemimposes administrative costs. Second,if publicfunds are less efficientlyengaged than privatefunds, the transferof fundsfrom indi- vidualsto the governmentwill causelosses not measuredin the deadweight loss calculationsdiscussed earlier. Third, the redistributionfrom consumers to producersmay not increasethe welfareof the lattersince they will be willing to invest resourcesin lobbying for protectionuntil the marginal returnon the last dollar so investedis equal to the likely returnof the transfer.Consumers, on the otherhand, interestedin protectingtheir own surplus,may be willingto invest similarly.While these expendituresmay be rationalfrom the point of view of the economicagents involved, they are wastefulfor societyas a whole. This is particularlytrue since a highly efficientindustrial organization and successfullobbying for protectionism may be very close substitutes.To the extent that firmsbelieve that these two thingsare close substitutes,industrial efficiency is impaired. This paperhas not considereda full generalequilibrium approach to trade restrictions,69which includes the effects of monetarypolicy and financialassets. It has examinedonly the cost of trade restrictions.But since policy analysis involves relative comparisons,in order to decide whetheran alternativepolicy should be used to achievethe goal of trade restrictions,the presentvalue of its costs shouldbe comparedwith the costs calculatedhere. The high costs of one policyhave no meaningin isolation fromcost comparisonswith others. This paperhas alreadyaddressed the argumentthat tradepolicy should be used to correctthe aggregateunemployment problem. A word is in orderon the effectsof tradebarriers on the U.S. distributionof income. Tradeliberalization would improvethe U.S. job mix, with a shift toward a largerproportion of higher-wagejobs.70 But because it wouldalter the job

68. Gordon Tullock, "The Welfare Costs of Tariffs, Monopolies, and Theft," WesternEconomic Journal, Vol. 5 (June 1967), pp. 224-32. 69. See, for example, Michael Mussa, "Tariffs,the Distributionof Income, and the Balance of Payments"(paper deliveredto the Workshopin InternationalEconomics, Universityof Chicago, July 18, 1972; processed),and the studies on real balancesand currencydevaluations in the extensivework by RudigerDornbusch, especially "Deval- uation, RelativePrices and the Real Value of Money,"Report 7130 (University of Chi- cago, Center for MathematicalStudies in Business and Economics, June 1971; processed). See also H. David Evans, A GeneralEquilibrium Analysis of Protection: Thle Effects of Protectionin Australia(Amsterdam: North-Holland, 1972). 70. As noted earlier,production wage rates per hour in 1967 were $3.16 in the export industriesand $2.66 in import-competingindustries. Stephen P. Magee 699 mix, a reductionin tradebarriers would probablylead to a redistribution of incomeaway from low-wageearners. In this context trade restrictions are probablyprogressive taxes. There are severalexceptions to this rule. First, very high wage rates are paid in some quota areas: Data from the Trade RelationsCouncil show that in 1967 wages in SIC 29, petroleum and coal, were $3.89 per hour, and in SIC 3312, blast furnacesand steel mills, they were $3.98 per hour. There are also some low wage rates in areaswhere restrictions on U.S. exports(particularly agricultural) are high. Second, Norman S. Fieleke has found that the U.S. tariff structureis slightlyregressive on the consumptionside.7' The regressivityis probably not greatenough to offset the progressivityon the income side, however. Thus,the gainsfrom freertrade are partialiyoffset, in the absenceof fiscal mechanismsto compensatethose who lose fromit, by a less equitabledistri- bution of income.

Summary The quantitativefindings in this studyare summarizedin Table13. They indicatethe followingcosts to the United States of existingand proposed traderestrictions: First,the averageannual cost of U.S. importrestrictions is $3.3billion to $5.0 billion.The bulk of this is due to quotasand quantitativerestrictions. In the short run (one to five years),the gains from eliminatingtariffs on directlycompetitive and other nonquota imports is partially offset by $330 million per year in labor transitioncosts. After transitioncosts, the United Statescould gain $2.7billion to $3.6 billionper year by eliminating quotas. Second,the presentdiscounted value of the costs of restrictionson nonquota U.S. importsis $33.1 billion. If these restrictionswere eliminated, over a third of the gain would be obtainedbefore the fifteenthyear. The presentvalue of the cost of U.S. quota restrictionsis $88 billion; if they were eliminated,U.S. welfarewould increasein presentvalue by nearly $37 billion beforethe fifteenthyear. Third,foreign tariffs on exportsof manufacturedgoods cost the United Statesrelatively little-$300 millionto $500 millionper year, or a present value of $11.4 billion. However,foreign restrictions on U.S. agricultural 71. "The Incidenceof the U.S. TariffStructure on Consumption,"Public Policy, Vol. 19 (Fall 1971), pp. 639-52. a a .~~~~(.4) F '0 0 .~~0 0 C) ~ t ~~ t ~ ~ ~ ~ ~ e 4 e ~~~~~~~~~q = o R~~~~~~~~~~~~C08 ct6 lf 0 4 0 o0 0c Q~~~~~~~~~~ts o-- .. w oo4ooN N en N enCS Cot Ey to8

Go 00'j X t ? o i P .1 Wbmo0~ ~ ~~ ---4)j4) ' '00

('1 ON 0 ; X $ m t>1000' @ o

U~~~~~ftrr |~~~~~~~~~~~~~Cr-- cl 08 r4 (3 I ?6 (>Iir- atci0 cq cl t 0 =e o "~~ 000'-4'4~~~~r-O . .0 '..4 Ci 'r)c-i--r--~~~~~~~~~-44r)c-i ~~~~~ ~ ~ /N ~~' i 4ft c-S c-i-.0 r- ONc-S 0 z 0 4..~~~~~~~~~~~~~~~~

>tIvrN UvsAo , N I 9 t N cr t oo N c3 - 0v) } o 0ss ^ ?0 I .g:Crz a) . oo m I n ? Y U -0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ci 0 0 0 ,, - PIZ ~~00 CN t'0-40N040 0f

0O C . en 00- cq

4-. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0

o >4 >0 01 ; 4) '0 a

4 O) f ,Ct0O-.- ~ 009~ C!00 .

V0 ,C Nt 0 0x

-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~o o s4~- .v . ? m 8 8 o 3

t 00'4t00tOr-.c--..Sf4 > U "N 0 o - 00S f " O or c00i t) 'C CS'00 t-it'4 :0 4) 0 t :3 "4 ~ ~ ~ ~~ ~ ~ ~ O N .'-~~~~~~~~~~~~~.~~~ U. 0 > - q0 0 f 0 0

tn ~ ~ ~ ~ ~ ~ ~ 1 0 06~ . C) 4-. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~)~~~MI l 0

0 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~00 C4 >4.

C1E *w* 0 0j U;'*2e? n~ ~ '' ON . * . "- 0

4.4 0~~~~~~~~~~c

Po ~ ~ ~ ~3ooo

. -~~~~~~~~o2~~~~~~~~~~ ~~~~~o~0.0u00 StephenP. Magee 701 exportscost the nation $4 billion to $5 billion per year; this presentdis- countedcost is $125 billion and if it were eliminated,U.S. welfarewould increaseby $53 billion withinfifteen years. Fourth, the total costs to the United States of existingtariff and quota barriersto tradeaverage $7.5 billion to $10.5billion per year.The present value of thesecosts is $258billion, $107billion of whichis imposedbefore the fifteenthyear. Fifth, the total welfarecost to the UnitedStates of the Burke-Hartkecut- back of U.S. importsfrom 1971 to 1965-69 levels would be $3.4 billion to $6.0 billion per year.72These imply a presentvalue of $101 billion, $52 billion of which accruesin the firstfifteen years afterenactment. Sixth, the proportionalquota provisionof the Burke-Hartkeproposal would cost the United States $1.1 billion per year in the first five years, $3.5 billionper year in the secondfive years, and $7.0billion per year in the thirdfive years after enactment. The presentdiscounted value of thesecosts is $28 billionwithin the firstfifteen years. The calculationswere not carried beyondthat point becausethe losses weregrowing faster than the discount rate. This makesthe proportionalquota an especiallydangerous form of restrictionbecause of the cascadingloss effectand the politicaltendency to keep quotas on long afterthey have achievedeven short-runobjectives. Seventh,after subtracting the costs of the tariffseliminated with Burke- Hartkeenactment, the net additionalwelfare cost of the proposal is $6.9 billion to $10.4 billion per year, or a presentdiscounted value of $129 billion, $80 biliion of whichcomes withinthe firstfifteen years after enactment. Finally,if Burke-Hartkewere enacted, the total annualmeasured costs of restrictionson U.S. tradewould be $15 billion to $21 billionper year;this has a presentvalue of $387billion, $187billion of whichwould come in the next fifteenyears. These calculations emphasize that quotasimpose greater costs than tariffsbecause (1) foreignerscapture part of the revenuethat would have been generatedby an equivalenttariff; (2) quotas stimulate monopoly behavior; (3) they hide the actual rate of protection given, which tends to be high; and (4) this implicit rate of protectionusually growsthrough time. The last two points are importantsince the efficiency losses of all restrictionsare proportionalto the squareof the rate of protection: Doublingrestrictions quadruples the social costs.

72. These numbersequal the "cutback"line in Table 13 minusthe "tariffseliminated." Commentsand Discussion

C. FredBergsten: I would like to stressand discussfurther Magee's own warningthat the gains of free tradecalculated in the papergrossly under- estimatethe actualgains to the United States.As Mageementioned, three majorelements are omittedin his analysis-dynamiceffects, economies of scale, and monopolyeffects. These may providevery large additional benefits that shouldbe addedto the overallestimate of the potentialgains from free trade. Preservationof competitionis particularlyimportant for the United States becausesome of the quota items, especiallysteel, are the productsof highlyoligopolistic industries in whichforeign supplies may be the only sourceof realcompetitive pressure. The sectoraleffects implied by Magee also merit specialattention: The existingrestrictions were costing consumersabout $18 billion per year, a bit more than the estimatein my own earlierpaper cited by Mageebecause of his higherestimate of the con- sumercosts of presentU.S. tariffs,while their benefits accrued to a limited number of special interests,before the administrationwisely began to liberalizethe oil and meat quotasto help fightinflation. Magee's handling of the "tariff-equivalent"revenue issue should be emphasized.When tariffsare increased,part of the reductionin consumer surplusis capturedby the additionaltariff revenue the governmentcollects; the result is "only" a transferwithin the economy.But with quotas, the scarcitypremiums that driveup pricesmust be dividedbetween U.S. importersand foreign exporters, and the divisionis indeterminate.The foreign supplieralways gets some shareof the higherprice triggered by the quota, and may even get all of it. Foreignsuppliers are particularlylikely to cap- turethe largerpart of the reducedU.S. consumersurplus under the favored new approachof "voluntary"export restraints, where the foreigncountry ratherthan the importingcountry administers the controls.In the case of 702 Stephen P. Magee 703 the textilerestraints, for example,Japanese firms with quota allocations are sellingexport tickets to one another(just as oil importerssell them to one anotherin this country).The price of the ticketsderives from the scarcity premiums,suggesting that most if not all of the price increasesfrom the U.S. quotas are being capturedby Japaneseproducers rather than being transferredwithin the U.S. economy.This transfer from U.S. consumersto foreignexporters must be addedto the deadweightloss in calculatingthe nationalloss to the United Statesfrom the quotas. Perhapsthe most importantreason that the gainsfrom free trade may be understatedby Mageeis that tradepolicy is unstable.The United Statesis virtuallycertain to get increasedprotectionism unless trade is liberalized. Thus any calculationof the benefitsfrom reducingtariffs and othertrade barriersmust include the avoidanceof the costs of increasingtariffs or quotassuch as those proposedin the Burke-Hartkebill. Indeed,when the administrationasks for new tradenegotiating authority, its majormotive is to avoid the Burke-Hartkebill. If it were necessaryto try to eliminate tariffscompletely in orderto head off the Burke-Hartkequotas, as some administrationofficials have implied, the gainsfrom free trade would equal, at a minimum,the sum of Magee'scalculations of the costs of presenttariffs plus the cost of Burke-Hartkeitself. Magee'scalculations also have interestingimplications for the feasibility of a new reciprocaltrade negotiation. They suggestthat the eliminationof dutiesby the United Stateswould raiseU.S. importsby about $15 billion annuallyover the longerrun ($8 billion of "competitive"imports and $7 billion of "noncompetitive"imports). The eliminationof foreign duties wouldraise U.S. exportsby roughlythe sameamount ($6 billionfor manufacturedexports for industrieswhere exports were a substantialportion of sales, $5 billion for agriculturalexports, and some additionalamount for the remainingone-half of all manufacturedexports). Thus a move to eliminate all dutieswould appear to comereasonably close to meetingthe test of reciprocity,at least fromthe standpointof the United States.

LawrenceKrause: Magee's effort to calculatethe incalculableis the first seriousattempt to measurethe effectsof traderestrictions and it deservesto be encouraged.Of course,a lot of judgmentsare called for alongthe way in making these calculations.Although I trust Magee to make these judg- ments,my guessis that the losses due to existingrestrictions are somewhat overstatedin his paper. 704 Brookings Papers on Economic Activity, 3:1972 Magee's discussionand estimationof welfareconcentrate on the economic inefficiencyside of the story and virtuallyignore the incomeredis- tributionside. The gainersand losersare not the samepeople, so thereare genuineredistributive effects. One can arguethat anotherway to offsetthe redistributioncan alwaysbe found,but that is a cop-out.The politicalsys- tem will not provide full compensationfor the losers, and the problem cannot be assumedaway. My guess is that the incomeredistribution producedby the tariffsystem yields some positive welfare gain. In any case,the actualdistributive effects need to be investigatedand takeninto account. No matterhow hard one triesto give the devil his due, I thinkMagee is absolutelycorrect in rejectingthe unemploymentargument in favorof trade restrictions.Fundamentally, trade restrictions are alwaysinflationary and the only reasonwe have excessiveoverall unemployment is becauseof the inflation problem. Furthermore,the unemploymentargument for trade restrictionssupposes that job opportunitiesmust be created in specific products(generally manufactured ones) whereexcess capacity exists. That is not the way cyclicalrecoveries in fact createjobs. Duringthe past twelve months, manufacturingoutput has grown more rapidlythan overallout- put, but the increase in employmentin manufacturinghas been only 500,000,or less than 3 percent,compared with an overallincrease of 2.4 million, or fully 3 percent. A wordis in orderregarding the use of traderestrictions as meansof ac- complishingsocial goals. For example,many argue that oil quotasmay be justifiedfor nationaldefense. In fact, they have workedin the wrongdirec- tion. Whatcould be a strangerpolicy than using more domestic oil in peace- ful times instead of saving it for national defense?On the other hand, Magee's infant industry argumentcannot be dismissed simply on the groundsthat a subsidyis a bettermechanism; one must recognizethe dis- tortingeffect of the extrataxes that wouldbe requiredto producethe reve- nues to pay the subsidy.More generally,any move towardfree trade would requirecustoms revenue to be replacedby some other taxes, which may createconsumption distortions that in part offsetthe gain. I think the studiesof the impactof existingquotas that Mageerelies on grosslyexaggerate the costs. The key assumptionin these studiesrelates to the foreignsupply elasticity to the United Statesand is weakestwhen per- fect elasticityis assumed(as is oftenthe case). Theseare managedmarkets that cannot be describedadequately by elasticityparameters. Quota-free oil importsinto the United Stateswould mean a majorinstitutional change StephenP. Magew 705 in the worldoil market,and one cannotreasonably assume that the price the UnitedStates would pay for importedoil would be unaffected.Magee findsthe greatestcosts by far on the quota side, but I seriouslydoubt the qualityof the calculationshe has to rely on. The estimatesMagee uses for the benefitsto the United States from increasedagricultural exports if barrierswere removed are probablyjust as shakyas those for the costs of import quotas. But I believethey convey the right message:Here is the mostsignificant potential welfare gain to the UnitedStates from freer trade. When one looks at the side of increasedrestrictions such as the Burke- Hartkebill, the quantitativeeconomic costs are not as importantas the politicalimpact. Burke-Hartke would make foreign trade policy a strictly domesticissue, unilaterally set by the UnitedStates wlth no furthernegotia- tions with othercountries. This arrangementwould have tremendousim- plicationsfor internationaleconomic relations that cannotbe capturedin any estimateof the welfareloss. I concludethat moving to freertrade may be worthsomething, but not a greatdeal; however,more restrictions may cost a very great deal indeed. Therefore,we shouldwant to stayjust aboutwhere we are.But, as Bergsten pointedout, we cannotstay where we arebecause the politicalsituation on tradepolicy is basicallyunstable. We have to negotiatein orderto avoid going backward.But we shouldnot wantto achievecompletely free trade, because we could only backslidethereafter. We should really want to negotiate forever, reachingagreements before presidentialelections to demonstratethat we havenot merelywasted time and effort.

GeneralDiscussion

Paul Wonnacottshed light on why the welfarecosts of existingtariffs tendedto appearrelatively small. He emphasizedthe point thatthe welfare trianglesdepend on the squareof the tariffrate and thus when tariffs are cut by 50 percent,the welfarecosts are reducedby 75 percent.In contrastto quotas,which have grown in importance,tariffs have been cut substantially sincethe 1930s;and consequentlytheir remaining welfare costs are under- standablysmall. A numberof participantscommented on the importanceof variousfac- tors that are omittedin Magee'sanalytical framework. Wonnacott felt that economiesof scale wereparticularly important benefits of worldmarkets, 706 BrookingsPapers on EconomicActivity, 3:1972 especiallyfor certainU.S. exportslike aircraft.Lawrence Klein mentioned that othertypes of U.S. traderestrictions can, like importsand quotas,have significantwelfare costs. Amongthese he pointedto restrictionson the use of foreign-flagvessels and export promotion subsidies such as DISC. On the other hand, Klein questionedthe assumptionthat foreign supply to the UnitedStates is highlyelastic; that did not squarewith the evidencefollow- ing our devaluation,when many foreign producers simply cut theirexport prices.Alfred Reifman was concernedthat the estimatedcost of the Burke- Hartke proportionalquota provisionunderestimated the impact of this seeminglyreasonable provision. A ceilingthat preventedeach type of import from increasingits shareof the U.S. marketwould soon restrictim- portsas sharplyas the originalcutback to 1965-69average levels, since, for manygoods, imports could be expectedto continueto risemuch faster than domesticproduction. Reifman also suggestedthat the CommonMarket ex- periencemight be studiedto appraisethe magnitudeof dislocationcosts involvedin movingtoward free trade. It was his impressionthat the evidence was reassuring. Magee's treatmentof job dislocationcosts evoked a numberof comments.Wonnacott stressed that increasedimports can competeeither with industriesthat are experiencingexcess demand or with those that have idle resources.Increased imports that alleviateshortages in bottleneckareas can ease overallinflationary pressures and therebymake it possibleto have less restrictivefiscal and monetarypolicies, and thereforehigher levels of economic activity.In that sense, such additionalimports can actuallycreate jobs. On the otherhand, if pocketsof excessiveunemployment exist in cer- tain industries,additional imports that competewith these industries would clearlysubtract from the aggregateemployment total. WilliamNordhaus complimentedMagee for makinga genuineeffort to examinethe distribution problemand to identifythe industriesthat would have increasedun- employment.On the one hand, Nordhausfelt that Magee might be over- statingthe increasein unemploymentdue to dislocationby ignoringthe normalprocess of attritionof workersthrough voluntary quits and retire- ment;to the extentthat attritionbrought down the employmenttotal over time, the reducedemployment needs of the industrywould not increase turnoveror layoffs.On the otherhand, Nordhauswas concernedthat the costs were measuredentirely in terms of unemploymentand ignoredthe possibilitythat people who were laid off might on balancehave to take lower-paidjobs as a substitute. Stephen P. Magee 707 The discussionalso focused on Magee'sprocedures of discountingand capitalizingthe streamsof benefitsand costs overtime. FrancoModigliani suggestedthat the properinterest rate should be a social measureof im- patiencesince the measurementsdealt with welfareeffects. He felt that 6 percentwas roughly the rightreal rate of interestto discountfor impatience, but he questionedwhether the processof takingpresent values was the best way to make the calculations.Magee felt that this techniquewas a useful way of lookingat the problembecause of the sharplycontrasting time pro- files of costs and benefitsof traderestriction and tradeexpansion.