Across-Product versus Within-Product Specialization in ...€¦ · value isoquants). Relatively...

Across-Product versus Within-Product Specialization inInternational Trade∗

Peter K. Schott†

Yale School of Management & NBER

November 21, 2002

Abstract

The unit values of US manufacturing imports vary widely within very narrowlydefined products. In cross-section, unit values are higher for varieties exported bycapital and skill abundant countries, and they increase with the capital intensityof exporters’ production techniques. Over time, the same products increasinglyare sourced from more disparate countries. These facts reject ‘old’ trade theoryspecialization across products but are consistent with such specialization withinproducts: capital abundant countries use their endowment advantage to manufac-ture varieties that are superior in terms of quality or attributes to those producedby labor abundant countries. The facts are inconsistent with ‘new’ trade theorymodels that have producer price varying inversely with producer productivity be-cause unit values are higher for the set of countries commonly thought to be moreproductive.

Keywords: Heckscher-Ohlin Model; New Trade Theory; Unit Value; Varieties Trade

JEL classification: F11; F14; F2; C21

∗Special thanks to Richard Baldwin, Andrew Bernard and Jonathan Feinstein. I alsothank Alberto Alesina, Gordon Hanson, James Harrigan, Marc Melitz, Stephen Redding,TN Srinivasan and anonymous referees for helpful comments. An earlier version of thispaper was entitled “Do Rich and Poor Countries Specialize in a Different Mix of Goods?Evidence from Product-Level US Trade Data” (NBER 8492).

†135 Prospect Street, New Haven, CT 06520, tel : (203) 436-4260, fax : (203) 432-6974,email : [email protected]

Product Specialization in International Trade 2

1. Introduction

The unit values of US manufacturing imports vary widely within thou-sands of finely detailed product categories. To take one striking example,men’s cotton shirts from Japan are roughly thirty times as expensive asthe identically classified variety1 originating in the Philippines. Across allUS manufacturing imports, the mean high to low unit value ratio in 1994was 24. To put these difference in US prices in perspective, note that theprice of a Big Mac in 1999 varied by a factor of just 3 — across countries —according to the Economist.

In addition to being large, differences in US import unit values aresystematic. Three patterns emerge from the data. First, unit values arehigher for varieties originating in capital and skill abundant countries thanthey are for varieties sourced from labor abundant countries. Second, unitvalues are positively associated with the capital intensity of the productiontechnique exporters use to produce them. Third, over time the unit valuesof skill and capital abundant countries increase relative to the unit valuesof labor abundant countries.

These data provide a completely new dimension for testing the impli-cations of ‘old’ and ‘new’ trade theory. In ‘old’ trade theory, compara-tive advantage drives countries to specialize in unique subsets of goods.Within the Heckscher-Ohlin factor proportions framework, for example, la-bor abundant Philippines ought to export labor intensive apparel and toyswhile capital abundant Japan should focus on capital intensive machineryand chemicals. Existing tests of this framework find scant evidence in favorof endowment-driven trade at the industry level (e.g. Bowen et al 1987,Trefler 1995).

Surprisingly, I find no evidence of specialization across products hereeither. Over time, the US sources the same products from an increas-ingly diverse set of high and low wage countries. However, the facts areconsistent with old trade theory specialization within products. The pos-

1Throughout the paper, imports from different countries within a product categoryare referred to as varieties. It is useful to think of varieties as being both horizontal (e.g.red versus blue telephones made with identical input intensities) and vertical (e.g. hightech versus low tech phones made with different input intensities). My use of horizontaland vertical here is not meant to refer to issues associated with multinational enterprises.


itive relationship between unit values, exporter endowments and exporterproduction techniques supports the view that capital abundant countriesuse their endowment advantage to produce vertically superior varieties, i.e.varieties that are relatively capital intense and possess added features orhigher quality, thereby commanding a relatively high price. If this interpre-tation is correct, conventional tests of the framework using industry-leveldata are problematic because much of the factor proportion action occursat a level that is hidden from the researcher.2 This interpretation of theevidence is similar in spirit to the quality ladder product cycle model ofGrossman and Helpman (1991). In that framework, high wage leaderswith a comparative advantage in innovation (which can be motivated byendowments) continually develop new and improved varieties to replacethose copied by low wage followers.3

The positive relationship between unit values and input intensities issignificant because it indicates that differences in exporter production tech-niques may not be due solely to variation in factor efficiency, an explanationthat has been popular since Leontief (1953).4 The observed capital inten-sity of the Japanese electronics industry may exceed that of the Philippineelectronics industry, not just because their labor is less efficient but becausethe Japanese are manufacturing fundamentally different products.

The influence of aggregation and potentially arbitrary product classifi-cation in seeking evidence of product mix specialization has been a concernsince Balassa (1966). Obviously, an infinitely disaggregate classificationwould reveal complete specialization. Thus, the association between va-riety unit value and production technique found here is quite important,

2Two recent tests of the framework support this view. Davis and Weinstein (2001)allow country input intensities to vary with country capital abundance in their test ofwhether the factors embodied in trade equal countries’ relative factor abundance. Thisassumption, which is equivalent to assuming all countries produce a unique set of goods,provides a closer match between factor content and factor supplies than previous tests.Schott (2001) finds strong support for the implication that a country’s level of industryparticipation varies with its relative endowments when industries are adjusted to allowfor product heterogeneity.

3Feenstra and Rose (2000) show that the order in which countries first begin exportingindustries to the US market is positively associated with estimates of their innovativepotential.

4Trefler (1993), for example, recovers estimates of country productivity by maximizingthe fit of the factor proportions framework.


allowing a direct link between within-product varieties and the manner inwhich goods are conceptualized in the factor proportions framework. In-stead of focusing on machinery versus apparel, we should be thinking abouthigh-definition versus analog televisions. Given the potential for misclas-sification that exists in a product-level trade dataset that is constructedfrom literally millions of US Customs declaration forms, the strength ofthe evidence found here is remarkable.

In new trade theory, international trade patterns are driven by con-sumers’ love for variety, imperfect competition and productivity differencesamong producers within industries. Varieties of new trade models differdepending upon whether the focus is on homogeneous (e.g. Krugman 1979,1980) or heterogeneous (e.g. Bernard et al 2000, Melitz 2002) firms withincountries.5 In both cases a variety’s price varies inversely with its pro-ducer’s productivity. For these models to be consistent with the importunit values of US trading partners, skill and capital abundant countriesmust have relatively low productivity, an assumption most trade econo-mists would find counter-intuitive.6

This evidence against new trade theory should not be too startlinggiven that these models are designed to explain the relatively high vol-ume of intra-industry trade occurring between countries with very similarendowments. The poor fit found here merely highlights an insight — allintra-industry trade is not equal — that is often forgotten given the disap-pointing empirical performance of old trade theory in industry-level factorcontent tests. The consistency of factor proportions and ‘old’ trade flowsat the product level does not mean that new trade theory is irrelevant instudies that focus more squarely on ‘new’ trade flows.7

More broadly, the evidence presented here highlights the need for anew set of firm-based trade models which can simultaneously capture the

5This paper is related to existing tests of homogenous firm new trade models thatexamine intra-industry trade among a group of countries (e.g. Helpman 1987, Hummelsand Levinsohn 1995, Debaere 2002). Unlike those studies, it examines the imports of asingle country, considers product prices rather than intra-industry trade, and focuses ona larger and more diverse set of trading partners.

6Though it is possible for high-productivity firms to compete in terms of qualityrather than price (Melitz 2000), the empirical link between exporter unit values and inputintensities found here suggests that such competition relies upon relative endowments.

7Bernard et al (2002), for example, find support for key implications of heterogeneousfirm models among US manufacturing plants.


richness of demand suggested by the proliferation of US product imports aswell as the important association between factor endowments, factor inputintensities and product prices.

The remainder of this paper proceeds as follows: Section 2 describesold and new trade theory and their implications for product-level tradedata; Section 3 describes the empirical results; Section 4 discusses alternateexplanations; and Section 5 concludes.

2. Old and New Trade Theory

In the Heckscher-Ohlin model of old trade theory, a country’s prod-uct mix varies with its relative factor endowments.8 The multiple coneequilibrium of this model is displayed in the Lerner diagram in Figure 1.It features a world of two factors and four industries, Apparel, Textiles,Machinery and Chemicals, which differ in terms of their capital intensity(production technique). Apparel is the most labor intensive industry whileChemicals is the most capital intensive. Under standard assumptions (seeDixit and Norman 1980), the four industries’ unit-value isoquants delin-eate three cones of diversification, where cone refers to the set of vectorsselecting a product mix.

Because production of an industry outside of the cone in which a countryresides results in negative profit, GDP-maximizing countries specialize inonly the two industries anchoring their cones, i.e. the two industries whoseinput intensities are most closely related to their endowments.9 The neg-ative profits capital abundant Japan would earn in labor intensive Appareland Textiles, for example, can be seen by comparing the amount of capitaland labor that can be bought for one dollar in Japan (via the downwardsloping isocost curve defined by rJPN and wJPN) with the amount of cap-ital and labor needed to produce one dollar’s worth of output (via the unitvalue isoquants). Relatively high costs drive countries out of industries atodds with their comparative advantage.

8The single factor Ricardian model of ‘old’ trade theory also implies product special-ization as a result of international productivity differences (e.g. Dornbusch et al 1977and Eaton and Kortum 2002). I focus here on the Heckscher-Ohlin model because ofevidence with respect to multiple factors of production presented below.

9Leamer (1987) provides generalizations of these implications for higher dimensionalsettings.


Existing trade datasets are almost exclusively collected at the industrylevel. As a result, testing whether countries with disparate relative factorendowments export distinct sets of goods within industries to the US isquite difficult; countries rarely specialize at the industry level.

This paper exploits the richness of finely detailed product-level tradedata to look for specialization both across and within thousands of prod-ucts. To test for specialization across products, I investigate the extentto which capital abundant (high wage) countries ship the same productsto the US as labor abundant (low wage) countries. To test for specializa-tion within products, I examine whether exporter varieties within productsare related to exporter endowments and exporter production techniques.Within-product specialization assumes that the isoquants of Figure 1 rep-resent product varieties rather than industries. Because data on productiontechniques at the product level are unavailable, I link product unit valuesto industry input intensities.10

‘New’ trade theory is based upon consumers’ love of variety (via Dixit-Stiglitz (1977) preferences), monopolistic competition and intra-industryvariation in productivity. Production encompasses a single factor, labor.As is well known given these assumptions, a variety’s price is a constantmarkup over productivity-adjusted marginal cost. The price ratio of anytwo varieties is

pmpn

=w/ϕmw/ϕn

, (1)

where pm and pn are the prices of varieties m and n,w represents thereturn to (homogeneous) labor, and ϕ indexes productivity.11 In earlymodels (e.g. Krugman 1980), countries are assumed to be a collection ofhomogeneous firms, each possessing the same productivity and producinga distinct variety. More recent models (e.g. Bernard et al 2000, Melitz2002) stress firm heterogeneity in an effort to explain issues beyond tradepatterns, in particular why some firms become exporters and others donot. Because the data I examine encompass countries and products, I pro-ceed under the assumption that countries produce unique varieties within

10Lack of product-level input intensity data also precludes performance of a product-level factor content test in the spirit of Bowen et al (1987).11Because varieties face the same elasticity of substitution, the constant markup term

cancels out of equation (1).


product categories.Equation (1) implies that countries with higher productivity will have

lower priced exports.12 This implication is contradicted by the data: unitvalues are higher for skill and capital abundant countries, i.e. the verycountries which enjoy relatively high productivity.

It is possible that firms with relatively high productivity choose to com-pete on quality, using their productivity advantage to produce high-quality,high-price varieties rather than lower-priced versions of goods from lowproductivity firms (Melitz 2000). Like the quality ladder model of Gross-man and Helpman (1991), however, a new trade model incorporating thisactivity begs an old trade theory interpretation. This interpretation issupported by the empirical link between exporter unit values and inputintensities.

The next section examines specialization both across and within prod-ucts. The lack of specialization across products is taken as evidence againsta standard interpretation of old trade theory. Specialization within prod-ucts is used to test old versus new trade theory directly. In new tradetheory, within-product specialization is horizontal and variety price variesinversely with producer productivity. In old trade theory, within-productspecialization is vertical: varieties are related both to exporter endowmentsand to exporter production techniques.

3. Empirical Results

3.1. Data Description

Product-level US import data compiled by Feenstra (1996) record thecustoms value of all US imports by exporting country from 1972 to 1994.

12Wages in equation (1) can also be interpreted as efficiency-adjusted. In that case,wm = ewm/ηm, where ewm and ηm are the raw wage and effieciency of labor in country m,respectively. If workers receive their marginal product, wm = wn = w and cross-countryvariation in wages and worker efficiency have no impact on relative prices. If workersin more productive countries receive exogenously high relative wages that are unrelatedto efficiency, variety prices will diverge less than implied by differences in ϕ. However,such differences would have to be implausibly high to rescue new trade theory: evenif Japan and the Philippines are equally productive (ϕm = ϕn), the unit value ratio inmen’s cotton shirts requires Japanese apparel workers to earn thirty times the wage ofPhilippine apparel workers after adjusting for worker efficiency (η).


Customs (i.e. free on board or fob) value is exclusive of any duties orshipping charges. An extremely useful feature of these trade data is theinclusion of both quantity and value information for a large number of goodsand countries, rendering possible the calculation of unit values. I computethe unit value of product p from country c, upc, by dividing import value(Vpc) by import quantity (Qpc), upc = Vpc/Qpc.13 Examples of the unitsemployed to classify products include dozens of shirts in apparel, squaremeters of carpet in textiles and pounds of folic acid in chemicals. Becauseunits vary by products within industries, unit values cannot be computedat the industry level.

It is important to note that the unit values in this dataset are not per-fect. A 1995 study by the US General Accounting Office identified under-lying product heterogeneity and classification error as two major sources ofunit value error in an in-depth analysis of eight products. Within-productheterogeneity is a focus of this paper, and I explore how unit values varywith respect to exporter endowments and exporter production technique.Classification error involves inaccurate recording of units and misclassifica-tion of goods.

Imports are recorded according to thousands of finely detailed cate-gories, which I refer to as ‘products’ or ‘goods’.14 Imports at higher levelsof aggregation, such as the one digit Standard International Trade Clas-sification (SITC1) system, are referred to as ‘industries’. Table 1 listsexamples of products by industry in 1994 along with the number of prod-ucts in each industry in that year. Manufactured Materials, with overfour thousand products, has the most categories. The analysis below is re-stricted to manufacturing imports (SITC1=5,6,7,8), which are more likelyto be motivated by exporter skill and capital abundance.

A snapshot of across-industry specialization at various levels of aggre-

13For some years and products, there are multiple country observations of value andquantity. In those cases, I define the unit value to be a value-weighted average of the ob-servations. Availability of unit values ranges from 77% of product-country observationsin 1972 to 84% of observations in 1994.14 Imports are classified according to seven digit Tariff Schedule of the US (TS7) codes

from 1972 through 1988 and according to the ten digit Harmonized System (HS10) codesfrom 1989 through 1994. The most salient difference between the two systems is a reduc-tion in the number of Manufactured Materials categories at the expense of Machinery,Chemical and Food categories.


gation is provided in Figure 2. Each line in the figure traces the share ofnon-zero country-industry observations in the dataset for a different levelof aggregation. While 93% of SITC1 cells exhibit positive imports in 1994,the share is just 10% for products. This discrepancy highlights the difficul-ties of using industry-level data to test for old trade theory specialization.15

3.2. US Trading Partners Do Not Specialize Across Products

In old trade theory, countries with different relative endowments exportdistinct sets of products to the US. Testing this hypothesis requires consid-eration of the broadest possible sample of countries because specializationis more apt to appear among more dissimilar trading partners. As a re-sult, I increase sample size in this section by grouping countries accordingto per capita GDP (PCGDP) rather than capital abundance; GDP datais available for roughly three times as many countries as endowment dataover the sample period.

I classify countries as low, middle and high wage if their World BankPCGDP is in the 0 to 30th percentile, the 30th to 70th percentile, or the70th to 100th percentile of the world distribution, respectively. (Below, Ishow that results are not sensitive to the use of alternate cutoffs.) Thesecohorts are meant to correspond to the three cones of diversification inFigure 1. Countries are re-assigned to cohorts each year to control forpotential movement through cones of diversification. Though the numberof US trading partners increases over time, across all years there are anaverage of 40 countries in the low and high wage cohorts, and 55 in themiddle wage cohort. Countries classified as low wage throughout the sam-ple period include China, India, Pakistan and most African countries.16

Turkey and Chile are persistent middle wage countries.Products are classified according to the exporter PCGDP cohort from

which they originate. Low (L), Middle (M) and High (H) products origi-nate solely in low, solely in middle, or solely in high wage countries, respec-

15Each years’ shares in Figure 2 are conservative in the sense that they are computedwith respect to the set of countries exporting any product to the US in that year, andnot with respect to the set of countries in existence in that year.16The set of countries permanently transitioning out of the low income group during

the sample period, and the years of transition, are Thailand (1979), Cameroon (1981),Egypt (1985), the Philippines, (1990), Senegal (1990) and Indonesia (1991).


tively. Products are Low & Middle (LM) or Middle & High (MH) if theyare sourced simultaneously from at least one country of each type. Finally,a product is Low, Middle & High (LMH) if it originates in at least one lowwage country and one high wage country, simultaneously (e.g. both Japanand India export the good to the US). The six product cohorts — L, M, H,LM, MH and LMH — are mutually exclusive.

Figure 3 plots a breakdown of the number of products by type. Thekey message of this figure is remarkable: even when trade is divided intothousands of products, there is little evidence of endowment-related special-ization across products. In 1972, 38% of import products originate solelyin high wage countries (H) and 31% in middle-high (MH) wage countries.By 1994, these share had fallen by roughly half, to 21% and 16%, respec-tively. At the same time, the share of LMH products — those importedsimultaneously from high and low wage countries — rose steadily from 30%in 1972 to 62% in 1994.17

Though the Heckscher-Ohlin model does not indicate what level of LMHtrade constitutes a rejection of the framework, these facts are clearly at oddswith the spirit of the model. They reject old trade theory specializationdue to comparative advantage across products.

This evidence is robust to a number of sensitivity analyses. This robust-ness is summarized in Table 2, which reports the share of LMH productsin 1972 and 1994 according to alternate methods of categorizing products.The second row of the table is the base case definition from Figure 3. Thenext two rows use the 20th and 80th and the 40th and 60th per capitaGDP percentiles, respectively, to classify countries. Though levels change,the upward trend is preserved, though muted for the 20th-80th percentilesplit because of the large expansion of middle countries (e.g. under thisdefinition, China moves from low to middle and Korea from high to middlein 1994). Use of the asymmetric 30th-90th split, reported in the fifth row,is similar to that of the base case: low wage countries are entering theproduct markets of even the fifteen or so highest wage countries.

Between 1972 and 1994 an average of roughly one fifth of LMH prod-ucts receive this classification because of the presence of a single low wagecountry. The countries responsible for these products are predominantly

17Similar evidence, available from the author, is found with respect to the value ofimports.


rapidly growing emerging economies (e.g. China and the Philippines). Thesixth row of Table 2 reports results after excluding LMH products that aredefined in this manner: an upward trend, from 17% to 44%, remainseven after they are removed from the sample. Note that the influence ofthese countries is not inconsistent with the factor proportions framework.Indeed, it may be a manifestation of their movement into the cones ofdiversification occupied by higher wage countries.18

To determine the overall sensitivity of results to China, the seventhrow of the table reports LMH shares using the 30th-70th percentile splitbut excluding China from the sample. The increase in shares from 28% to49% indicates that low wage countries in addition to China are increasinglyexporting the same goods to the US as high wage countries.

The final row of Table 2 reveals that increase in LMH exports is notdriven by very small export flows. The upward LMH trend remains evenif exporter-product observations of less than $10,000 are excluded from thesample.

3.3. Unit Values, Endowments and Production Techniques

Unit values are positively associated with exporter PCGDP, exporterrelative endowments and exporter production techniques. Unit valuesincrease with PCGDP both in cross section and within product-exporterpairs over time. In addition, the ratio of high wage country unit valuesrelative to low wage country unit values increases with time.

Striking examples of the first relationship are provided in Figure 4,which plots exporter unit value versus exporter per capita GDP for fourproducts in 1994. The first three scatters in the figure are plots of man-ufacturing products: dyed woven fabrics, men’s cotton shirts and CRTmonitors. The fourth scatter, of fuel oil, is a natural resource commodity.The manufactured goods exhibit a positive (and significant) relationshipbetween unit value and PCGDP. The natural resource good does not.

18The relatively high increase in LMH goods due to fast-growing low wage countriesis also consistent with the macroeconomic literature on convergence (see, for example,Imbs and Wacziarg 2002). On the other hand, the data indicate that non-converginglow wage countries are also entering products held by high wage countries. As seen inthe next section, the across-product evidence here obscures within-product specializationthat is missed by traditional, industry-level analysis.


Across all US manufacturing imports, the median ratio of high to low unitvalues is 24.

A more formal estimate of the relationship between product unit valuesand exporter income is obtained by estimating

log(upct) = αpt + βpt log(pcgdpct) + εpct (2)

via OLS separately by LMH product and year, where upct is country c’sunit value of product p in year t and pcgdpct is country c’s per capita GDPin year t. The sample is restricted to LMH products because these arethe products originating simultaneously in high and low wage countries.This estimation yields approximately 70,000 βpt’s. The percent of thesecoefficients that are positive and significant (at the 10% level) are reported,by year, in the second column of Table 3. Results show that the share ofproducts exhibiting a positive relationship between unit value and exporterincome grows from 40% in 1972 to roughly 50% by 1994. This evidencerejects a null hypothesis of no association between unit value and PCGDPacross products.

More broadly sourced products are more likely to have a positive as-sociation with exporter PCGDP. The third and fourth columns of Table3 report the share of positive and significant slopes when equation (2) isrestricted to goods exported by at least 20 and 40 countries, respectively.Shares jump by ten to fifteen percentage points with each sub-sample, sothat by 1994, 60% of products sourced from at least 20 countries, and 75%of products sourced from at least 40 countries, show unit values increasingwith PCGDP.19 This pattern of results is reassuring because it reveals thatdifferences in unit values are not driven by relatively few countries partic-ipating in relatively inactive product markets. Indeed, the more varietiesimported, the more likely their price rises with exporter PCGDP.

Exporter-product unit values also increase with exporters’ per capitaGDP percentiles within products over time. This result cannot be pushedtoo hard because it is based upon the severely restricted sample of LMHproduct codes that are valid in both 1972 and 1988.20 Nevertheless, table

19This winnowing of the sample results in a sharp decline in the number of LMHproducts analyzed in each column. In 1994, there are 5528, 2229 and 447 LMH productsin the full, n>20 and n>40 samples, respectively.20Significant product code revisions occur between 1972 and 1988. The regression


4 reports OLS results from estimating

∆upc = α+ β∆ptilec + εpc, (3)

where ∆upc is the log difference in exporter-product unit value across timeless the mean difference for the product, and ∆ptilepc is change in exporterc’s PCGDP percentile across time less the mean difference for countriesexporting that product. Changes are computed over 1972 to 1988. Thecoefficient estimate implies that a ten percentage point (.10) jump in acountry’s relative position in the world PCGDP distribution is associatedwith a 4.8% increase in relative unit value. To the extent that PCGDPpercentile movement is correlated with factor accumulation, this trend isconsistent with factor proportions and product cycle theory: the morelikely countries are to move into more capital intensive cones, the morethey appear to produce higher priced, more capital intensive varieties.

More direct measures of exporter endowments are also positively relatedto unit values. Table 5 reports the results of regressing log unit valueson log exporter capital and skill abundance and product fixed effects in1990.21 Analysis is restricted to a single year due to data constraints.The table reveals that unit values are positively and significantly relatedto exporter skill and capital abundance. The coefficient estimate in thesecond column implies that a 10% increase in country capital abundanceis associated with a 4% percent increase in unit value. Coefficients inthird column, on the other hand, show that a 10% increase in primary andsecondary education attainment are each associated with a 7% increase inunit values. If measures of capital and skill abundance are both included inthe regression, as reported in column 3, the coefficient for capital abundancedeclines and the significance of skill abundance falls as a result of the highcorrelation between capital and skill.

Unit values are also positively related to exporter production techniquein 1990. Though production techniques are unavailable at the product level

sample contains 2105 exporter-product observations across 222 LMH products and 91countries; not all countries export each product. 1972 and 1988 are used as endpointsbecause of the change in product classification systems in 1989.21Data on country capital per labor and educational attainment are from Maskus

(1991) and Barro and Lee (1994), respectively. For these regressions, goods are re-classified as LMH according to the country capital abundance cohort from which theyoriginate, using the same 30-70 percentile cutoffs as above. Countries where capitalabundance is not observed are excluded from these cohorts.


they are available at the industry level from another source.22 As a result,I concord products to industries and regress via OLS product-exporter unitvalues in industry i on industry-exporter techniques,

log(upic) = αp + βikic + εpic, (4)

where αp is a product fixed effect and kic is the capital per labor ratioof three digit ISIC industry i in country c. This estimation yields oneslope per industry. Slopes, R2’s and the number of product-country ob-servations for each industry are reported in Table 6. Across exporters, apositive and statistically significant relationship between product price andindustry capital intensity is evident in 26 of 28 manufacturing industries.The magnitude of the regression slope is highest for Machinery (ISIC 322),where the point estimate implies that a 10% increase in capital intensity isassociated with an 8.5% increase in unit value.

Relating unit values to production technique is very helpful becauseit bypasses potentially arbitrary product classification. Demonstrating apositive correlation between goods prices and their capital intensity relatesdirectly to a key implication of the Lerner diagram in Figure 1, which isthat goods in old trade theory are defined by their input intensities. Thus,this link gets to the heart of specialization irrespective of whether it occursacross or within products.

Finally, the relative unit value of high wage country varieties with re-spect to per capita GDP increases with time. I estimate this relationshipby SITC1 industry to highlight differences across industries. For each yearand across all LMH products within each SITC1 industry, I estimate viaOLS

log(upict) = αp+βi1 log(pcgdpct)+βi2 log(pcgdpct) ∗Timet+ εpict,(5)

where αp represents a dummy for product p and Timet is a time trendranging from 1 to 23 (1972 to 1994). Results are reported by industryin Table 7. All coefficients are positive and statistically significant at

22Country-industry capital intensity estimates are computed using data from the IND-STAT3 database from UNIDO (see Schott 2001 for details on their construction). Forthe regression, HS10 products are concorded to three digit ISIC industries using a com-bination of Maskus’ (1991) SITC to ISIC industry concordance and Feenstra’s (1996)product to SITC concordance.


the 1% level. They are also economically significant: in Machinery in1972, for example, a 10% increase in per capita GDP is associated witha 12% (11.76% + 0.8%) increase in unit value. The coefficient on thetime trend indicates that the magnitude of the association increases overtime in all industries, and relatively more so in Manufactured Materialsand Miscellaneous Manufacturing.23

This last result may reflect a reaction by high wage countries to compe-tition from low wage countries: as low wage countries gain greater access tothe world market due to reductions in trade barriers, high wage countriesreact by moving out of low quality goods and into high quality goods.24

Though the factor proportions framework and product cycle theory maybe used as a rough guide in thinking about this issue, there is a need forfurther theoretical work to derive more specific implications regarding theevolution of relative quality differentials between high and low wage coun-tries over time.

The evidence linking unit values, exporter endowments and exporterproduction techniques presented in this section supports an old trade the-ory interpretation of US trade because it is consistent with skill and capitalabundant countries using their relative endowments to manufacture supe-rior, or vertically distinct, varieties that incorporate higher levels of capitalper worker. In the context of Figure 1, this interpretation has high andlow wage country varieties being represented by distinct isoquants, witheach country producing the set of varieties whose input intensities are mostsimilar to its relative endowments: Italy exports sportswear that is capitalor skill intensive (high quality, fashionable) while China exports sportswearthat is labor intensive (low quality, drab).

The evidence in this section is less consistent with new trade theorybecause varieties from countries with high productivity have a higher pricethan varieties from countries with low productivity.

23This trend is not due to a change in the composition of products within industriesover time. A similar pattern emerges when the sample is restricted to the (constant) setof products classified as LMH in 1972.24The reduction of quantitative trade restrictions such as the global Multifiber Arrange-

ment governing apparel and textile trade, for example, may reduce the demand for highwage country exports. As these barriers fall, low wage countries can satisfy a greatershare of demand.


4. Additional Interpretations

The analysis in the previous section supports the idea that interna-tional product trade proceeds according to comparative advantage. Thissection discusses additional interpretations of the evidence, and argues,that they, unlike comparative advantage, are not consistent with both thecross-sectional and time series evidence.

Specialization may appear stronger within rather than across productsbecause of transfer pricing. To the extent that US-based multinationalssource inputs from developing countries with lower labor costs, and seek tominimize tax liability in those locations, ‘true’ unit values may be higherthan those reported on customs documents. Such behavior could increasethe likelihood of finding evidence of specialization via unit value differences.On the other hand, if US tax rates are higher than those of low wagecountries, multinationals would have an incentive to over-report the valueof exports from low wage countries. (To match the time series evidence,this tax disparity must be increasing with time.) If that is the case, theresults of the previous section are likely to be conservative. Unfortunately,controlling for transfer pricing is not possible with existing datasets. Asnoted above, I am also unable to control more generally for product orvalue misclassification that occurs when customs declarations forms arefilled out.

It is also possible that the unit values of low wage countries are lowerthan those from high wage countries because of the relatively strong bar-gaining power of US firms. If the US is able to obtain lower prices from pro-ducers in low wage countries than from producers in high wage countries,perhaps due to imperfect information or other distortions, the relativelyhigh unit values of high wage country varieties will be biased upwards.Though it is hard to believe such asymmetry would endure (let alone in-crease over time), gathering data to test this hypothesis would be useful.

The evidence is also consistent with demand-side explanations of inter-national trade. To the extent that countries with similar incomes have ataste for similar goods (e.g. Linder 1961), US consumers may be willingto pay relatively high prices for high wage country varieties. However, forthis explanation to render new trade theory consistent with the unit valuepatterns reported above, this taste must be implausibly strong. Indeed,it must overcome both the relatively high productivity of high wage coun-


tries and generate very large relative unit value differences. On the otherhand, to the extent that these taste differences are a manifestation of higherquality or added attributes, they are consistent with old trade theory.

Finally, the within-product evidence for comparative advantage maybe driven in part by firm outsourcing of the type discussed by Feenstraand Hanson (1996). If goods from various stages of a production processare included in a single product category, countries with very differentendowments may export very different intermediate inputs within the sameproduct category. The increase in unit value differences over time is alsoconsistent with the idea that outsourcing has increased since the 1970s.To determine the extent to which identifiable outsourcing influences thetrends in this paper, I have re-run the estimations above on a sub-sampleof products that excluded products containing the word “part” in theirdescription (Ng and Yeats 1999). Results were essentially the same.

5. Conclusion

Product-level US trade data provide a completely new dimension fortesting and thinking about old versus new trade theory. Surprisingly, thedata rule out endowment-driven specialization across thousands of finelydetailed products. On the other hand, the positive association betweenwithin-product unit values, exporter endowments and exporter productiontechniques is consistent with within-product factor-proportions specializa-tion. These relationships suggest high wage countries use their endowmentadvantage to add features or quality to their varieties that is not presentamong varieties emanating from low wage countries.

Unit value patterns appear inconsistent with new trade theory modelsthat have producer price varying inversely with producer productivity. Tothe extent that skill and capital abundant countries enjoy relatively highproductivity, their varieties should sell at a discount relative to the varietiesfrom labor abundant countries. They do not. It is of course possible thata closer look at unit value variation within a subset of high wage countries,or within a subset of low wage countries, will provide support for new tradetheory in the arena it was designed to model. In line with recent theoreticalprogress in the literature, such an examination is likely to be much morefruitful and interesting if undertaken at the firm level.


Another interesting feature of the data is that high wage country unitvalues increase relative to low wage country unit values over time. Thatthis trend occurred during a period when trade barriers fell substantiallymay reflect efforts by developed countries to avoid direct competition withlow wage countries by upgrading their product mix, as in the quality ladderproduct cycle model of Grossman and Helpman (1991). Another issue mer-iting scrutiny is the applicability of Stolper-Samuelson price-wage linkagesto a world containing so many imperfectly substitutable varieties. In thestandard Heckscher-Ohlin model, specialization across relatively few indus-tries can insulate workers in high wage countries from their counterpartsin low wage countries. To the extent that the separate goods producedby high and low wage countries are substitutes, this insulation may bedampened.

The evidence presented in this paper highlights the need for a newround of firm-based trade models that encompass key elements of old andnew trade theory. In particular, future models must capture the richness ofdemand suggested by the proliferation of US product imports as well as theimportant association between factor endowments, factor input intensitiesand product prices.


References

Balassa, Bela. 1966. Tariff Reductions and Trade in Manufacturers Amongthe Industrial Countries. American Economic Review 56:466-473.

Barro, R.J. and J. Lee. 1994. International Comparisons of EducationalAttainment. NBER Dataset.

Bernard, Andrew B., Jonathan Eaton, J. Bradford Jensen, and Samuel S.Kortum. 2000. Plants and Productivity in International Trade. NBERWorking Paper 7688.

Bernard, Andrew, Jensen J. Bradford and Peter K. Schott. 2002. FallingTrade Costs, heterogeneous Firms and Industry Dynamics. Dartmouthmimeo.

Bowen, Harry P, Edward E. Leamer and Leo Sveikauskas. 1987. Multi-country, Multifactor Tests of the Factor Abundance Theory. AmericanEconomic Review 77(5):791-809.

Davis, Donald R. and David E. Weinstein. 2001. An Account of GlobalFactor Trade. American Economic Review 91:1423-1453.

Debeare, Peter. 2002. Testing ‘New’ Trade Theory without Testing forGravity: Re-interpreting the Evidence. University of Texas mimeo.

Dornbusch, R., Fischer, S and Samuelson, P.A. 1977. Comparative Ad-vantage, Trade and Payments in a Ricardian Model with a Continuumof Goods. American Economic Review 47:823-839.

Dixit, Avinash and Joseph Stiglitz. 1977. Monopolistic Competition andOptimum Product Diversity. American Economc Review 67:297-308.

Dixit, Avinash and Victor Norman. 1980. Theory of International Trade.New York: Cambridge University Press.

Eaton, Jonathan and Samuel Kortum. 2002. Technology, Geography, andTrade. Econometrica, 70:1741-1779.


Feenstra, Robert C. 1996. U.S. Imports, 1972-1994: Data and Concor-dances. NBER Working Paper 5515.

Feenstra, Robert C and Gordon Hanson. 1996. Globalization, Outsourcing,and Wage Inequality. NBER Working Paper 5424.

Feenstra, Robert C and Andrew K. Rose. 2000. Putting Things in Order:Trade Dynamics and Product Cycles. Review of Economics and Statis-tics (forthcoming).

General Accounting Office. 1995. US Imports: Unit Values Vary Widelyfor Identically Classified Commodities. Report GAO/GGD-95-90.

Grossman, Gene M. and Elhanan Helpman. 1991. Quality Ladders andProduct Cycles. Quarterly Journal of Economics 106:557-586.

Helpman, Elhanan. 1987. Imperfect competition and International Trade:Evidence from Fourteen Industrial Countries. Journal of the Japaneseand International Economies 1:62-81.

Hummels, David and James Levinsohn. 1995. Monopolistic Competitionand International Trade: Reconsidering the Evidence. Quarterly Jour-nal of Economics 110:799-836.

Imbs, Jean and Romain Wacziarg. 2002. Stages of Diversification. Amer-ican Economic Review (forthcoming).

Krugman, Paul R. 1979. Increasing Returns, Monopolistic Competition,and International Trade. Journal of International Economics 9:469-479.

Krugman, Paul R. 1980. Scale Economies, Product Differentiation, andthe Pattern of Trade. American Economic Review 70:950-959.

Leamer, Edward E. 1987. Paths of Development in the Three-Factor, n-Good General Equilibrium Model. Journal of Political Economy 95:961-999.

Linder, Stefan. 1961. An Essay on Trade and Transformation. New York:Wiley.


Leontief, Wassily. 1953. Domestic Production and Foreign Trade: TheAmerican Capital Position Re-examined. Proceedings of the AmericanPhilosophical Society 97:332-349.

Maskus, Keith E. 1991. Comparing International Trade Data and Productand National Characteristics Data for the Analysis of Trade Models, inP. Hooper and J.D. Richardson, eds., International Economic Transac-tions, University of Chicago, Chicago.

Melitz, Marc. 2000. Estimating Firm-Level Productivity in DifferentiatedProduct Industries. Harvard mimeo.

Melitz, Marc. 2002. The Impact of Trade on Intra-Industry Reallocationsand Aggregate Industry Productivity. NBER Working Paper 8881.

Ng, Francis and Alexander Yeats. 1999. Production Sharing in East Asia:Who Does What for Whom and Why? World Bank Policy ResearchWorking Paper Series.

Schott, Peter K. 2001. One Size Fits All? Specialization, Trade and In-come Inequality. American Economic Review (forthcoming) and NBERWorking Paper 8244.

Trefler, Daniel. 1995. The Case of the Missing Trade and Other Mysteries.American Economic Review 85(5):1029-1046.

Trefler, Daniel. 1993. International Factor Price Differences: Leontief wasRight! Journal of Political Economy 101:961-987.


K

L

ApparelTextiles

Machinery

Chemicals

Japan

Philippines

k2

k1

1/wJPN 1/wPHL

1/rPHL

1/rJPN

Figure 1: Multiple Cone Heckscher-Ohlin Specialization


Notes: Graph displays the percent of non-zero country-industry cells across all US manufacturing imports for the noted level of aggregation. SITC is the Standard International Trade Classification; four digit industries are more finely defined than one digit industries. Product refers to the 7 digit TS or 10 digit HS product categories, which are the most detailed description of US trade available. Product classification switched from TS to HS in 1989.

0

20

40

60

80

100

1972 1977 1982 1987 1992

Non

-Zer

o C

ount

ry-I

ndus

try O

bser

vatio

ns (%

)

1 Digit SITC2 Digit SITC4 Digit SITCProduct

Figure 2: Percent of Non-Zero Country-Industry and Country-Product Ob-servations in a Panel of US Manufacturing Imports


Notes: US import products are classified according to the income level of the exporting country. Exporters are classified as low (L), middle (M) or high (H) wage if their per capita GDP is below the 30th, between the 30th and 70th or above the 70th percentiles of world PCGDP, respectively. Low wage products originate solely in low wage countries (e.g. China), while LMH products originate simultaneously in at least one low and high wage country. The six classifications in the figure are mutually exclusive. Countries are re-ranked each year.

0

10

20

30

40

50

60

70

1972 1977 1982 1987 1992

Perc

ent o

f Im

port

Prod

ucts

by

Sour

ce

Low WageMiddle WageHigh WageLMMHLMH

Exporter

Figure 3: Breakdown of US Imports By Exporter PCGDP Level, 1972-1994


CRT Monitors (8471923200)

Men's Cotton Shirts NES (6205202065)Dyed Woven Fabrics (5208398090)

Number 6-Type Fuel Oil (2710000530)

(mean) gdp646 24,547

9

21

AGO

ARG

BEL

BHS

BRA

CAN

CHE

CMR

COG

COL

DEUDNKDZA

ECU

EGY

ESP

FRA

GAB

GBRGHA

IDNIND

IRL

ITAKOR

MEX

NGANLD

NOR

NZL

PAN

PER

PRTSAU

SEN

SGP

SWE

TTOTUN

VEN

YEM

ZAF

(mean) gdp468 24,547

21

670

AREARG

AUS

AUT

BEL

BGD BGR BHRBLZ

BOL BRA

CAN

CHE

CHLCHN

CIVCOLCRI

CYPCZE

DEUDNK

DOMECUEGY

ESP

ETH

FIN

FJI

FRA

GBR

GHA

GIN

GMB

GRC

GTMGUY

HKGHND

HUNIDNIND

IRL

ISR

ITA

JAMJOR

JPN

KEN KNAKOR

LAOLBNLKA

MAR

MEXMLIMNG

MOZ

MUSMYS

NGANIC NLD NORNPL

NZL

PAKPANPER

PHLPOL

PRT

ROMSAU

SEN

SGP

SLVSWESYRTGO

THATTOTUR

TZA

VEN

VNMYEM ZAFZWE

(mean) gdp776 24,547

0

10

AUS

AUTBEL

BFA

CHE

CIVCMR CZE

DEU

DNK

ESP

FRAGBR

GHAGTM

IND

ITA

JPN

MEX

PRT

SWE

(mean) gdp955 24,547

162

8,143

AUS

AUT

BEL

BMUBRA

CAN

CHECHLCHN

DEU

DNKECU

ESP

FIN

FRA

GBRHKGHUNIDNIND

IRLISL

ISR

ITAJORJPN

KEN KORMEXMUSMYSNLD

NOR

PHL

PRT

SGP

SWE

THATUN

VEN

$ Pe

r Sq

Met

er.

$ Pe

r Mon

itor.

$ Pe

r Doz

en S

hirts

. $

Per B

arre

l. Per Capita GDP ($)

Per Capita GDP ($) Per Capita GDP ($)

Per Capita GDP ($)

Figure 4: Unit Values vs Exporter Per Capita GDP for Four US ImportProducts in 1994


SITC1 Industry Example of SITC2 Industries Example of HS ProductNumber of HS10 Products (1994)

0 Food Meat, Dairy Products, Fruit Sheep, live 1823

1 Beverages/Tobacco Wine, Cigarettes Carbonated soft drinks 163

2 Crude Materials Rubber, Cork, Wood, Textile Fibers

Silkworm cocoons suitable for reeling

833

3 Mineral Fuels Coal, Coke, Petroleum, Natural Gas, Electric Current

Unleaded gasoline 101

4 Animal and Vegetable Oils Lard, Soybean Oil, Linseed Oil Tallow, edible 82

5 Chemicals Organic Chemicals, Dyes, Medicines, Fertilizers, Plastics

Chloroform 1930

6 Manufactured Materials Leather, Textile Yarn, Paper, Steel, Cork Products

Diaries and address books, of paper or paperboard

4219

7 Machinery Power Generating Machinery, Computers, Autos

Ultrasonic scanning apparatus

2898

8 Miscellaneous Manunfacturing Apparel, Footwear, Plumbing, Scientific Equipment, Cameras

Boys' shorts cotton playsuit parts, not knit

3866

9 Not Elsewhere Classified Special Transactions, Coins, Gold Sound recordings for State Dept use

65

Table 1: Mapping Products to Industries


Percentiles Used to Define LMH Products Additional Restrictions 1972 1994

30th - 70th (Figure 3) None 30 62

40th - 60th None 45 65

20th - 80th None 25 34

30th - 90th None 27 61

30th - 70thExclude LMH Products Exported by Just One Low Wage Country

17 44

30th - 70th Exclude China From Sample 28 49

30th - 70thExclude Exporter-Product Observations where Exports Values is Less than $10,000

15 54

Percent of LMH Products in US Imports

Notes: This table reports the share of US import products sourced simultaneously from low, middle and high wage countries in 1972 and 1994 according to different robustness tests. Per capita GDP percentile splits refer to the breakpoints used to classify countries as low, middle and high wage. Products are classified according to the set of countries in which they originate.

Table 2: Robustness of Across-Product Specialization to Alternate Defini-tions of Low, Medium and High Wage Countries


All LMH Products

LMH Products With More Than 20

Exporters

LMH Products With More Than 40

Exporters1972 43 55 611973 45 57 691974 40 53 691975 41 57 681976 43 54 551977 42 55 531978 43 53 551979 45 56 601980 46 56 501981 42 53 591982 43 50 611983 43 49 551984 43 50 571985 45 56 611986 47 56 681987 49 58 721988 52 62 711989 49 60 741990 52 63 751991 51 63 741992 52 63 761993 50 61 741994 49 60 73

Notes: This table reports the number of manufacturing products exhibiting a positive and significant correlation (at the 10% level) between product unit value and exporter PCGDP. Manufacturing products are those in SITC1 industries 5 through 8. The first column reports significance across the full sample of LMH products. Subsequent columns restrict the sample to LMH products imported from the indicated number of countries. Sample size declines across columns: in 1994 there are 5528, 2229 and 447 LMH products in the full, n>20 and n>40 samples,

Percent of LMH Manufacturing Products With Significant Positive Relationship Between Unit Value and Exporter PCGDP

Year

Table 3: Summary of Product-Level Regressions of Unit Value on ExporterPCGDP


Regressor Relative Unit Value ChangeRelative Exporter PCGDP Percentile Change 0.486

(0.221)

Observations 2105

Products 222

Countries 91

R2 0.01

Notes: Table reports OLS estimation results on the sample of LMH manufacturing products used in both 1972 and 1988. Robust standard errors adjusted for country clustering are in parantheses. Relative unit value change is log difference of exporter-product less the mean change for the product. Relative PCGDP percentile change is difference in exporter PCGDP percentiles less the mean change for countries exporting the product. Changes are for 1972 to 1988. Countries do not export all products.

Table 4: Change in Exporter Unit Value versus Change in ExporterPCGDP Percentile, 1972 to 1988


RegressorLog Capital per Labor ($000) 0.419 *** 0.387 ***

0.059 0.065Log Primary Education Attainment 0.688 *** 0.330 *

0.173 0.181Log Secondary Education Attainment 0.686 *** 0.192

0.159 0.159Log Tertiary Education Attainment 0.187 -0.032

0.134 0.102Product-Country Observations 54,425 54,425 54,425Countries 44 44 44Product Fixed Effects Yes Yes YesAdjusted R2 0.81 0.80 0.81Notes: This table reports results of cross-country regressions of exporter unit value on exporter relative endowments. Capital per labor ratios are in $000 and are from Maskus (1991). Education attainment data are from Barro and Lee (1994). All data are for 1990. Robust standard errors based on country clustering are noted below coefficients. ***, **, and * refer to statistical significance at the 1%, 5% and 10% levels.

Log Unit Value

Log Unit Value

Log Unit Value

Table 5: Product-Exporter Unit Values vs Industry-Exporter Capital andSkill Abundance, 1990


ISIC Industry (Rev 2) R2Product-Country

Observations311 Food products 0.35 *** 0.49 8505313 Beverages 0.24 *** 0.46 546314 Tobacco 0.05 0.30 176321 Textiles 0.62 *** 0.43 14440322 Wearing apparel, except footwear 0.81 *** 0.49 8276323 Leather products 0.37 *** 0.44 863324 Footwear, except rubber or plastic 0.58 *** 0.45 2177331 Wood products, except furniture 0.29 *** 0.55 1053332 Furniture, except metal 0.33 *** 0.33 547341 Paper and products 0.03 0.47 1036342 Printing and publishing 0.18 *** 0.64 1393351 Industrial chemicals 0.30 *** 0.49 4192352 Other chemicals 0.45 *** 0.53 2123353 Petroleum refineries 0.18 ** 0.58 310354 Miscellaneous petroleum and coal 0.70 *** 0.66 196355 Rubber products 0.39 *** 0.48 1108356 Plastic products 0.43 *** 0.56 2173361 Pottery, china, earthenware 0.35 *** 0.56 1070362 Glass and products 0.50 *** 0.51 1296369 Other non-metallic mineral products 0.27 *** 0.48 1406371 Iron and steel 0.66 *** 0.64 2140372 Non-ferrous metals 0.67 *** 0.73 852381 Fabricated metal products 0.47 *** 0.54 5736382 Machinery, except electrical 0.46 *** 0.49 7541383 Machinery, electric 0.85 *** 0.45 3134384 Transport equipment 0.33 *** 0.50 3864385 Professional and scientific equipment 0.52 *** 0.37 2813390 Other manufactured products 0.30 *** 0.56 3392

Slope

Notes: This table reports results of cross-country regressions of log country-product unit value on log country-ISIC industry capital intensity, by ISIC industry. All data are for 1990. Regressions include product fixed effects. Industry-country capital intensities are from Schott (2001). Products are concorded to industries using Maskus' (1991) SITC to ISIC industry concordance and the product to SITC concordance provided by Feenstra (1996). ***, **, and * refer to statistical significance at the 1%, 5% and 10% levels.

Table 6: Product-Exporter Unit Value vs Capital Intensity of Industry-Exporter Production, 1990


RegressorLog(PCGDP) 0.539 *** 0.661 *** 1.176 *** 0.989 ***

0.013 0.005 0.012 0.003 x Time Trend 0.010 *** 0.020 *** 0.008 *** 0.030 ***

0.002 0.001 0.002 0.001Country-Product Observations 86,221 392,673 185,392 572,481Products 1,881 5,735 2,786 7,329Product Fixed Effects Yes Yes Yes YesR2 (Within) 0.02 0.06 0.05 0.14R2 (OLS) 0.56 0.68 0.76 0.74

Log(Unit Value) Log(Unit Value) Log(Unit Value) Log(Unit Value)

Notes: Table displays coefficients from a panel regression of country-product unit value on exporter per capita GDP for LMH products by SITC1 industry. Standard errors are listed below each coefficient; results for the constant are suppressed. *** represents statistical significance at the 1% level. Time trend ranges from 1 (1972) to 23 (1994)

ChemicalsManufactured

Materials MachineryMiscellaneous Manufactures

SITC 5 SITC 6 SITC 7 SITC 8

Table 7: Panel Regression of Product Unit Value Versus Exporter PCGDPand Time, by SITC1 Industry

Across-Product versus Within-Product Specialization in ...€¦ · value isoquants). Relatively...

Documents

Transcript of Across-Product versus Within-Product Specialization in ...€¦ · value isoquants). Relatively...