Professional Documents
Culture Documents
1990 - DeMelo Tarr - WelfareCostsOfUSQuotasInTextilesSteelAndAutos
1990 - DeMelo Tarr - WelfareCostsOfUSQuotasInTextilesSteelAndAutos
1990 - DeMelo Tarr - WelfareCostsOfUSQuotasInTextilesSteelAndAutos
OUOTAS IN TEXTILES,
STEEL AND AUTOS
Jaime de Melo and David Tarr*
AbsIruct—this paper quantifies the welfare effects and re-
source shifts that would occur if U.S. quantitative restrictions for so that, unlike in the PE analysis, capturing
(QRs) in textiles, steel and autos were removed. Estimates are quota rents affects resource allocation. Third,
derived from a static ten-sector general equilibrium model of
the U.S. economy. The welfare loss from the ORs is estimated
economy-wide resource constraints and in-
at approximately U.S. $20 billion (1984 dollars) Due to the terindustry linkages provide a more accurate esti-
high rent transfer component of ORs (about 75Wo), the aver- mate of welfare and sectoral employment effects.
age across-the-board tariff equivalent of QRs (in terms of
welfare costs) is estimated at about 209c, a rate which pre-
This paper deals with the problems of PE analysis
date,s the early days of multilateral tariff reduction. by presenting estimates from a static ten-sector
computable general equilibrium (CGE) model of
I. Introduction the U.S. economy calibrated to the year 1984 when
quantitative restrictions (QRs) in tex- tiles and
TABLE 1. —E s ici
S
rECI FICATION
Sector (CENTRAL CASE)
QRs in textiles and autos. We discuss these briefly that our welfare estimates are quite insensitive to
below. (Tarr (1989, chaps. 5 and 6) is devoted to a considerable changes in their values.7
detailed discussion of how these estimates were For the premia rate estimates for apparel (ex-
derived.) pressed as a percentage of landed U.S. import
Table 1 presents all elasticities and our con- price inclusive of tariffs), we have relied on
structed estimates of pre-existing premia. These Hamilton (1988). Hamilton’s data are based on
elasticities correspond to the “central” elastici- the sale of rights to export apparel products to
ties. (When we report in table 3 the likely range the United States. His data give an estimate of
of the welfare effects of removing all trade re- the quota rents captured by Hong Kong. In order
strictions, we are reporting results from the “low” to understand the problem of premia rate estima-
and “high” elasticities that are derived from the tion, it is crucial to recognize that the existing
“central” elasticities in table 1 by subtracting and Multifiber Arrangement (MFA) allows a number
adding, respectively, a standard deviation. 6 ) Be- of marginally inefficient foreign suppliers to sell
cause much econometric work has been done to in the United States. If the MFA were abolished,
estimate the capital/labor substitution elastici- many of them would be squeezed out of the U.S.
ties, we are fairly confident about the accuracy of market by competition. The quota premium rate
these elasticities. We are also fairly confident of earned by these inefhcient suppliers is less than
the import demand elasticities, for which a num- the quota premium rate paid by U.S. consumers
ber of estimates are available. Less confidence as a result of the MFA. Thus, we had to deter-
can be placed on export supply elasticities. How- mine the marginal supplier to the United States if
ever, because these latter elasticities enter only the MFA were abolished. Data in Hamilton
indirectly into our estimations, we have found (1988) allow us to determine that Hong Kong,
which had a quota premium rate of 479c for
apparel sold to the United States in 1984, or a
As the sources for table 1 detail, the estimates were
obtained from many studies. These studies generally provide supplier more
standard errors of estimates. For those elasticities taken from
Stern, Francis and Schumacher, the high and low estimates
are generally the high and low estimates from their survey,
7
which are not based on standard deviations. In a few cases the For example, doubling (halving) the elasticity estimates in
high and low estimates are obtained by doubling and halving column (3) increases (decreases) our estimates of the welfare
the central estimate. See Tarr f1988, ch. 5) for details. gains with the central elasticity estimates from US$22.0 billion
(with infinite foreign trade elasticities) to $22.1 ($22.9) billion,
respectively, or + 0.5 to.
492 THE REVIEW OF ECONOMICS AND STATISTICS
efficient than Hong Kong, would be the marginal sumption of imported steel products by 159c from
supplier if quotas were removed.' the observed levels in 1984. This restriction re-
The United States is much more competitive in sults in a 7Wo premium on imported intermediate
textiles than in apparel. Consequently, for textiles steel products. The employment hgures in col-
we take 5% as the premium rate, which is much umn 2 are only for the industry subject to quota
more conservative than the 159c proposed by removal; the total economy-wide relocation of
Cline (1987, p. 167). A positive rate is indicated, workers is reported in column 3. This latter mea-
however, based on data in a study by the U.S. sure is a summary of interindustry effects (see
International Trade Commission (1987). For au- Deardorff and Stern (1986)).
tos we relied on the quality-adjusted premia esti- The figures in table 2 reveal that the largest
mates of Feenstra (1988) for Japanese car im- welfare costs are due to the ORs in textiles and
ports by the United States. For European car apparel. This may seem surprising since that sec-
imports, we relied on the quality-adjusted price tor is smaller than autos, and imports are three-
increase of European cars sold in the United fourths of auto imports in value (including pre-
States estimated by Dinopoulos and Kreinin mia). Also, the proportion of the total welfare
(1988). Our resulting premia estimates, which are costs due to distortionary costs are much higher
weighted averages, are 41% for textiles and ap- in textiles and apparel than in autos even in the
parel and 32% for autos.9 case of an infinite elasticity of import supply in
autos. This is so because the premium rate is
N. Welfare Cost Estimates higher and because the price elasticity of demand
The costs of QRs reported in tables 2 and 3 for textiles and apparel is almost four times higher
below are based on estimates of the premia that than the corresponding elasticity for autos. Fur-
accrue to the exporting countries rather than to thermore, the relative homogeneity of domestic
the United States. One can therefore distinguish and imported steel reduces the production costs
two components of the costs of the QRs: (1) the of distortion.''
income or rent transfer to foreigners and (2) the Because elasticity estimates are not precise, we
distortionary cost due to the usual consumption report in table 3 a range of welfare gains from
and production costs of protection.' 0 Table 2 removing all QRs simultaneously for the cases of
summarizes these costs for each of the three low, central, and high elasticities. Because the
industries: textiles and apparel, autos and steel. marginal benefit to the economy of an income
Since the restrictions on steel—which resulted in transfer is a decreasing function of demand and
approximately a 15% reduction in imports of supply elasticities, the welfare gains from captur-
steel starting in early 1985—were not imple- ing rents from foreigners is higher in the low
mented until 1985, we obtain the estimates for elasticity case than in the high elasticity case. Of
steel by reducing intermediate and final con- course, these estimated welfare gains due to the
8
We note that estimates of quota premia, in both textiles The United States imposes ORs in other sectors in addi-
and automobiles, vary from year to year. We obtained esti- tion to textiles, autos and steel. For example, Hufbauer,
mates of quota premia for 1984, the year for which we Berliner and Elliott (1986) reported QRs in book manufactur-
benchmarked the model. Thus, our estimates apply to the ing, motorcycles, shipbuilding and maritime industries, sugar
conditions prevailing in that year. For other years, the premia and food products containing sugar, cheese and other dairy
rates and our cost estimates may differ. products, pean uts and meat. We have simulated the effects of
If the marginal supplier is more efficient than Hong Kong, removing all the QRs in the economy by assuming a high
our premium rate estimate is conservative and the associated (low) premia rate (tariff equivalent) of the ORs (with rents
welfare gain estimate is downward-biased. There is evidence going to the foreigners) of 5to (2.5Wo) in agriculture; 3Wo
that, during the quota era, U.S. dealers of Japanese autos (1.5 to) in other manufactures, other consumer goods and
captured about $550 per vehicle in rents above those rents food; and zero in the remaining sectors. Given these premia
captured by foreigners. This would correspond to a higher rates, the welfare gain from removing all QRs, including those
premium rate on autos of 43to, of which foreigners capture in textiles, autos and steel, is $27.6 billion in the high and
80Wo of the rents. We shall provide separate estimates for this $24.0 billion in the low premia rate cases. These estimates of
scenario. the benefits of removing all QRs, however, can only be
’0 In the cases where there are terms of trade effects, the considered suggestive because, unlike the case of our esti-
estimates of the distortionary costs of protection are reduced mates of the premia rates in textiles, autos and steel, we have
because the terms of trade effects partially offset the distor- not done a careful estimate of the premia rates in the remain-
tionary cost component of the ORs. ing sectors of our model.
Employment
Change in the Economy- Vide
Welfare“ Industry Losing Gain Employment
Its OR Relocation'
- 4, e/„„, ' ñ Small country a'•8umptlon for all ct he:r export demand and lmpcrt supply e I a ,tdeities
Results in pa rent heses are the welfare estimates with infinite toreign trade e lasticit res
e'' for cxpcir ts = 3 and e,‘ of supply of imports — 4 whe re i ranges over all traded sectors.
capturing of rents from foreigners are overesti- Insofar as removing QRs leads to increases in mates to the extent that
rent-seeking activities agricultural exports and auto imports, the U.S. dissipate them. For this reason, an auction quota terms
of trade will decline and hence the welfare mechanism is superior to a direct allocation of gain will be less. For the
central elasticity case, quota rights to imports. Thus our estimate of the the welfare gain is $20.3 billion, of which $5.4
annual cost of ORs in these three industries for billion is the distortionary cost component of 1955 protection levels
is between $20 billion and QRs. Obtaining the quota rents is a pure income
$22 billion. transfer from the rest of the world to the United
494 THE REVIEW OF ECONOMICS AND STATISTICS
as when they are removed in their sector alone, currently with the VER on autos. This is so because
the auto sector gains 1,700 jobs when ORs on steel, an important input into auto pro- duction,
steel and textiles and apparel are removed con- becomes cheaper and because the in- come
Copyright O 2001 All Rights
elasticity of demand for autos is high. Both effects TABLe 4. —Eurcovunsi Errors or Reuovi o
benefit the domestic auto industry which expands ORs
even when QR protection is removed. in Autos, STEEL AND Tex+ires
(CENTRAL ELASTI CITv CASE)‘
Given that QR protection is obtained through (thousand jobs)
the political process, one can argue that the U.S.
Agriculture 6.4
Congress has decided to value a job in the pro- Food 0.8
tected sectors more highly than a job in other Mining 4.5
sectors. In our experiment, 172,000 jobs in tex- Textiles and Apparel — 156.4
Autos 1.7
tiles and steel were protected by the QRs on the Steel — 13.5
three sectors at a cost of $20.3 billion. Hence, the Consumer Goods 17.5
annual cost per protected job is about $118,000, Other Manufacturing 87. I
Traded Services 35.4
approximately 8 (3) times the average annual Non-Traded Services 18.fi
total compensation of workers in the textile (steel) Foreign trade e lasticitieS defined in table 3
industry.
If QRs are removed, displaced workers will
incur search, relocation, and retraining costs (see
Mussa (1978)). Net benefits from QR removal are That is, for every dollar of earnings losses saved,
obtained by subtracting these costs. A proxy for the economy loses $65.
these costs is the discounted value of a displaced
worker’s earnings losses over a lifetime." This
measure allows us to estimate how much gainers VI. CORCIUSIO¥IS
will have left after compensating displaced work- Perhaps the most striking result is the high cost
ers for their earnings losses. Earnings losses for of protection from NTBs relative to that from
displaced workers last approximately six years." tariffs. While this is repeatedly mentioned in pol-
A conservative estimate of net benefits, NB, is icy discussions, few relative estimates are avail-
obtained as able. The figures in table 3 suggest that the
welfare cost of tariff protection is between 2%
(low elasticity case) and 4% (high elasticity case)
of the welfare cost of QRs in textiles, autos and
steel. Since there are other QRs in the United
States beyond those examined here, this estimate
is a lower bound.
An alternative way to evaluate the costs of
ORs is to ask what tariff structure would give the
5 EV — C, 1 same welfare loss as existing QRs in the three
NB —— sectors. In the central elasticity estimates (with
(1 + r)’
infinite foreign trade elasticities), the total wel-
fare cost is estimated at $22.0 billion (1984 dol-
where ñK is the equivalent variation measure, lars), of which $7.2 billion is the distortionary
C,
component. For that set of elasticities, moving
is estimated earnings losses in year t, and r —— 7%
from the actual tariff structure to a uniform tariff
is the discount rate. (This estimate is conservative
structure yielding the same (import weighted) av-
since earnings losses are zero after six years,
erage protection would represent a welfare gain
while yearly benefits do not decay.) NB is $104
of $0.70 billion. (Removing tariffs altogether
billion, with an associated benefit/cost of
would give a welfare gain of $0.94 billion.) Start-
65.
ing from the existing tariff dispersion, to get the
distortionary cost element of ORs would require
'‘ We call our estimate a proxy for these adjustment costs multiplying each tariff by 3.6 times its 1984
because we have not incorporated into our model an endoge-
nous sector that moves resources, as suggested by the theoret- value, which would amount to an average (import
ical work by Mussa. weighted) nominal protection of 12.59r. Adding
" See Jacobson (1978). To be conservative, we measure the loss due to rent transfers would require multi-
total compensation losses, which exceed earnings losses by the
amount of fringe benefits. Morkre and Tarr (1980, chapter 3) plying tariffs by 7.4 times their 1984 value,
discuss the merits of this measure and the alternative unem-
ployment cost measure.
Tarr, 13., A General Equilibrium Analysts of the Welfare and 1984 Farm Legislation (Washington, D.C., 1984).
Employment Ejects of US Quotas on Textiles, Autos, U.S. International Trade Commission, US Global Competitive-
and Steel (Washington, D.C.: U.S. Federal Trade Com- ness: the US Textile Infill Industry, Publication No. 2048,
mission, 1989). Dec. 1987.
Tarr, D., and M. Morkre, Aggregate Costs to the Uniied States Whalley, J., Trade Liberalization Among Major World I"rading
of Tariffs and Quotas on Imports: General Tariff Cuts Areas (Cambridge, MA: MIT Press, 1985).
and Remot’al of Quotas on Aiuomobtles, Steel, Sugar, Winston, C., and Associates, Blind Intersection* Policy and the
und Textiles (Washington, D.C.: U.S. Federal Trade Automobile Industry (Washington, D.C.: Brookings In-
Commission, 1984). stitution, 1987).
U.S. Department of Agriculture, Dairy: Background for the