since Henry Ford quipped that customers

could have cars in “any color as long as it is

black.”Today’s consumers are able to purchase a wide variety

of goods that were not available in the past. Not only

can they choose their cars from hundreds of makes and

models, but they can also purchase a wealth of technologically

sophisticated new products. One development that

has significantly broadened consumers’ choice of goods in

recent decades is the growth of international trade. As

trade with other nations has expanded, U.S. consumers

have been able to acquire varieties of goods not available

from domestic producers—Japanese cars, for example,

and French wine.

availability of new goods and varieties through international

trade has affected the welfare of U.S. citizens.While

the benefits of free trade have traditionally been associated

with declines in the price of existing products,

recent trade theory suggests that the introduction of new

imported goods constitutes another important gain from

trade. Our task in this article is to provide a measure of this

gain over the past three decades.

To do so,we first estimate the increase in global varieties

from 1972 to 2001.We then estimate how the change in

import prices over this period—a standard gauge of

consumer welfare—would be affected if this increase in

variety were taken into account.Using our results,we determine

what consumers would be willing to pay to access the

wider range of goods available in 2001 than in 1972.

Significantly, we find that global varieties grew more

than threefold over the 1972-2001 period.When we adjust

import price growth for the increased variety, we find that

import prices in this period fell markedly faster—by about

1.2 percentage points per year—than the conventional, or

calculations one step further, we conclude that consumers

would be willing to pay $260 billion, or roughly 3 percent of

GDP in 2001, to avail themselves of the expanded range of

goods on the market. This sizable sum indicates that U.S.

consumers see increased choice in goods as an important

benefit of international trade.

Classical international trade theory postulates that the opening

of an economy to trade improves welfare by allowing

CURRENT I S S U E S I N ECONOMICS AND FINANCE V O L U M E 1 1 , N U M B E R 4

consumers to access cheaper imported goods. In the new

models of international trade, however, countries benefit

from trade not only because the prices of individual goods

change, but also because consumers in open economies have

access to a wider range of goods than consumers in closed

economies. The new models are typically predicated on an

assumption that no one country can produce all of the varieties

available in the world. If consumers value new varieties

and individual countries cannot supply them, then consumers

stand to benefit from the increased choice of goods

that comes with trade.

In these new trade models, the gains from trade hinge

directly on a number of variables. The first is the “elasticity

of substitution” among varieties, or how substitutable consumers

believe varieties to be. If varieties of a particular

good are so alike that consumers will readily substitute one

for another—as in the case of gasoline, for example—then

having two varieties of that good will have little impact on

welfare. However, if the varieties are quite distinct—consider

Irish and American beer, or Italian and American

cheese—then U.S. consumers will benefit from the opportunity

to obtain both varieties.

The gains from trade also depend on differences in quality

across varieties. Consumers will place a higher value

on access to varieties regarded as superior. Thus, most

Americans would presumably prefer the opportunity to buy

French red wine to the opportunity to buy Japanese red

wine. The final variable affecting the gains from trade is

import quantity: All things being equal, consumers will

place a higher value on variety growth in product sectors

that command a large share of spending—for example,

automobiles—than on variety growth in small sectors.

To track the growth of variety in goods entering the United

States over the past three decades, we examine the most disaggregated,

or finely detailed, import data available. For the

1972-88 period, we use the seven-digit import classifitions

of the Tariff Schedule of the United States Annotated

(TSUSA) system; for the 1990-2001 period, we use the

ten-digit import classifications of the Harmonized Tariff

Schedule (HTS), the system that replaced the TSUSA in 1989.

These classifications break down U.S. imports into approximately

15,000 goods, each characterized by a level of detail

on the order of “red wine in bottles of under one liter.”

We define a variety to be a good emanating from a particular

country—for example, French red wine. This definition

suggests both the power and the limitations of working with

trade data. On the one hand, our definition is specific enough

to enable us to examine the interactions of literally hundreds

of thousands of good-country pairs. On the other hand, as

the example of French red wine indicates, our definition is

still very general, collapsing many kinds of French red wine

Looking at the composition of U.S.imports over the 1972-

2001 period (Table 1), we see two unmistakable trends. The

first is a dramatic increase in the number of measured

goods. Between 1972 and 1988, the number of goods

imported by the United States almost doubled, rising from

7,731 to 12,822. Similarly, between 1990 and 2001, the number

of imported goods rose from 14,572 to 16,390. Closer

scrutiny reveals that the increases in these two sample

periods did not consist simply of new imports being added

to the old (Table 1, rows 3 and 4 and rows 9 and 10). Indeed,

it appears that only half to two-thirds of the goods in each

of the two samples were imported both at the start and at

the end of the period.Moreover, approximately one-third of

2

Table 1

Median Total Number

Number of of Varieties

Number Exporting (Good-Country

Year of Goods Countries Pairs)

(1) (2) (3) (4)

U.S. Imports, 1972-88

All 1972 goods 1972 7,731 6 74,667

All 1988 goods 1988 12,822 9 173,937

Goods common to

1972 and 1988 1972 4,171 6 36,191

1972 and 1988 1988 4,171 10 56,183

Goods in 1972

not in 1988 1972 3,560 7 38,476

Goods in 1988

not in 1972 1988 8,651 8 117,754

U.S. Imports, 1990-2001

All 1990 goods 1990 14,572 10 182,375

All 2001 goods 2001 16,390 12 259,215

Goods common to

1990 and 2001 1990 10,636 10 132,417

1990 and 2001 2001 10,636 13 173,776

Goods in 1990

not in 2001 1990 3,936 10 49,958

Goods in 2001

not in 1990 2001 5,754 11 85,439

Sources: NBER Trade Data on CD-ROM; Center for International Data at UC Davis,

<http://data.econ.ucdavis.edu>.

Note: For the 1972-88 period, goods are defined at the seven-digit level of the TSUSA

(Tariff Schedule of the United States Annotated); for the 1990-2001 period, they are

defined at the ten-digit level of the HTS (Harmonized Tariff Schedule).

the goods in each sample had disappeared by the end of the

sample period. It is possible, of course, that the shifting composition

of the imported goods over the sample periods may

owe as much to changes in the way the goods are classified as

to actual changes in the goods themselves. Nevertheless, the

data do suggest that the growth of international trade in

recent decades has led to both the creation and the elimination

of imports—a fact often overlooked in discussions of

globalization.

The second trend that emerges from the import data is a

dramatic increase in the number of countries exporting each

good to the United States (Table 1, column 3). Over the full

1972-2001 period, the median number of countries supplying

each good doubled, rising from six countries at the start

of the period to twelve at the end.

Taken together, the data in Table 1 indicate that the number

of good-country pairs—that is, the number of varieties

(column 4)—rose 133 percent in the first period and 42 percent

in the second period, for a total increase of 231 percent.

This increase constitutes more than a threefold rise in

the number of varieties over the three decades from 1972

to 2001. Roughly half of this increase is attributable to a

doubling in the number of goods and half to a doubling in

the number of countries supplying each good.While we have

not yet explored how consumers value these new varieties,

the dramatic growth in the count of new varieties makes a

prima facie case that understanding variety growth is

important for measuring changes in consumer welfare.

To get a more concrete sense of what is driving the surge

in import variety, consider how specific goods have fared. In

some cases, variety growth stems primarily from an increase

in the number of countries exporting the good. For example,

in 1972, the United States imported roasted or ground coffee

from twenty-five countries. By 2001, however, the nation was

the number of countries supplying beer and wine to the

United States rose by about 195 percent and 50 percent,

respectively, over the period. The number of countries supplying

eyeglasses rose from nine to forty-seven.

In other cases, variety growth stems from both new

goods and new sources for each good. Car audio is a prime

example: In 1972, twenty-one countries exported car radios

of all types to the United States. By 2001, there were nine

different classifications for car audio systems and as many

as twenty-eight countries exporting each of the nine.

Overall, the number of varieties would appear to have risen

from 21 to 174. Clearly some portion of this increase—for

example, the splitting of the single 1972 classification of car

audio into the 2001 classifications of AM radios and AM/FM

radios—does not represent the addition of new goods.

However, other 2001 classifications for car audio systems—

radios with tape and compact disc players, for example—

probably do constitute new goods that have produced a

genuine increase in choice.

The rapid growth of import varieties has been accompanied

by changes in the relative importance of various countries as

exporters to the United States. In the first column of Table 2,

countries are ordered according to the number of goods they

Table 2

Ranking in Year

Country 1972 1988 1990 2001

Japan 1 1 3 7

United Kingdom 2 4 4 3

Germany 3 3 2 2

Canada 4 2 1 1

France 5 6 5 6

Italy 6 5 6 5

Switzerland 7 11 11 11

Hong Kong 8 9 12 16

Netherlands 9 13 13 14

Taiwan 10 7 7 9

Spain 11 14 15 12

Belgium-Luxembourg 12 15 14 15

Mexico 13 12 10 8

Sweden 14 17 16 19

Denmark 15 22 21 23

Austria 16 18 18 21

India 17 19 23 13

South Korea 18 8 9 10

Brazil 19 16 17 18

Australia 20 20 20 20

Israel 21 21 22 22

Portugal 22 26 28 32

Norway 23 31 31 37

Ireland 24 27 26 28

Finland 25 28 30 31

Colombia 26 33 34 35

Philippines 27 25 25 26

China 28 10 8 4

Argentina 29 29 29 39

Greece 30 38 44 47

Sources: NBER Trade Data on CD-ROM; Center for International Data at UC Davis,

<http://data.econ.ucdavis.edu>.

Notes: The table reports rankings for the thirty countries that exported the highest

number of goods to the United States in 1972. For 1972 and 1988, goods are defined

at the seven-digit level of the TSUSA (Tariff Schedule of the United States

Annotated); for 1990 and 2001, they are defined at the ten-digit level of the HTS

(Harmonized Tariff Schedule).

CURRENT I S S U E S I N ECONOMICS AND FINANCE V O L U M E 1 1 , N U M B E R 4

exported to the United States in 1972, with the largest

exporter placed first; the following columns report the rankings

of these same countries in 1988, 1990, and 2001.

Although the countries exporting the most goods to the

United States have tended, throughout the 1972-2001 period,

to be large and high-income economies, the position of

individual countries within the ranking has changed.

Canada moved from fourth to first place, and Mexico moved

from thirteenth to eighth place.The sharp rise in the ranking

of these two countries may reflect their adoption of a free

trade policy: Canada’s jump to first place followed a trade

liberalization in the 1970s and 1980s; Mexico rose from

tenth to eighth place after it liberalized trade in the 1990s.

In the case of other countries, economic growth, perhaps

coupled with liberalization,may help explain the change in

rank. Fast-growing economies such as China and Korea

have advanced rapidly as import sources for the United

States. The increase in the number of U.S. imports from

China has been especially stunning. In 1972,China exported

only 510 goods to the United States; by 2001, that number

had risen to 10,199 (Table 3). In other words, while Chinese

firms competed in only 0.6 percent of the import markets

in existence in 1972, they participated in 62 percent of the

markets in 2001. Thus, although Chinese firms account

for only 9 percent of all imports to the United States, there

is a Chinese firm selling in almost two-thirds of the U.S.

import markets.

The twenty-fold increase in the number of goods

exported by China has elevated China from the twentyeighth

position in the 1972 ranking of exporters to the

fourth position today. Similarly, India has risen in the rankings—

from twenty-third in 1990 to thirteenth in 2001—

with the sharp rise in exports that began with the country’s

liberalization in the last decade. At the other extreme,

countries such as Japan and Argentina, whose economies

grew slowly at the end of the 1990-2001 sample period,

have seen fairly substantial drops in the number of varieties

they export.

Although increased product variety is generally believed to

bring welfare gains, standard national measures of welfare

and prices do not assess how much better off consumers are

when a new variety of an existing good or a new good

becomes available. Both the U.S. import price index and the

consumer price index (CPI) measure the current cost of a

particular basket of goods and services relative to the cost

of those same goods and services in some base period in the

past. Thus, both indexes largely fail to capture the introduction

of new varieties and the increase in the standard of living

the two indexes fall short of being a true cost-of-living

index—one that measures the cost of maintaining a certain

level of welfare without restrictions on what is in the basket

In this section, we report our efforts to recalculate the

U.S. import price index for 1972-2001 taking variety growth

into account. Our object—both here and in the longer, technical

truer measure of the cost, over this thirty-year period, of

maintaining a particular level of satisfaction from the consumption

of imports. If new varieties of goods enhance the

standard of living, then consumers can spend less to achieve

the same level of satisfaction as in the past. By comparing

our variety-adjusted estimate of the rate of change in import

prices over the sample period with the conventional estimate

that does not include variety growth,we obtain a measure of

the nation’s welfare gains from variety growth.

In the box on page 5, we present an equation for calculating

a variety-adjusted price index. Although the equation

describes a simple case in which all varieties are uniform in

quality and priced identically, it nonetheless captures the

most important elements of the more sophisticated calculations

performed in our longer study.

Table 3

Percentage

of Total

Number of U.S. Imports

Year Goods in Year

1972-88

All 1972 goods 1972 510 0.1

All 1988 goods 1988 4,673 1.9

Goods common to

1972 and 1988 1972 215 0.0

1972 and 1988 1988 215 0.2

Goods in 1972 not in 1988 1988 295 0.0

Goods in 1988 not in 1972 1988 4,363 1.7

1989-2001

All 1989 goods 1989 5,587 2.5

All 2001 goods 2001 10,199 9.0

Goods common to

1989 and 2001 1989 3,567 1.3

1989 and 2001 2001 3,567 3.9

Goods in 1989 not in 2001 1989 2,002 1.2

Goods in 2001 not in 1989 2001 6,582 5.0

Sources: NBER Trade Data on CD-ROM; Center for International Data at UC Davis,

<http://data.econ.ucdavis.edu>.

The key variables in our adjusted estimate of the price

index are (1) the extent of the growth in varieties for each

good over the sample period and (2) the degree of similarity,

or elasticity of substitution, among the varieties of each

good. The simplest way to measure variety growth for a

particular good is to compare the number of varieties in

that product category at the beginning of the sample

period with the number of varieties at the end of the

period. For instance, we used this method earlier in the

article to calculate the extent of variety growth in coffee,

beer, and other products. However, a simple count of varieties

ignores the fact that we may import a huge quantity of

one variety and a small quantity of another.While we make

no allowance for differences in quantity in the simple case

we present in the box, we adjust for the share of each variety

in total import expenditures in our longer paper (Broda

and Weinstein 2004).

The second key variable—the elasticity of substitution—

is a measure of the degree to which consumers value new

varieties of a particular good. If U.S. consumers see blueberries

from Chile and Argentina as equivalent or “perfectly

substitutable,” then they will not value access to the Chilean

berries if they already have access to Argentinean blueberries.

In other words, the opportunity to buy both

products will do nothing to enhance consumers’ standard of

living. By contrast, if consumers see Italian and Chinese

shirts as essentially different, then they will regard themselves

as better off if they are able to purchase shirts from

of substitutability between the two varieties of shirts.

For our recalculation of the import price index, we document

the degree of substitutability for varieties of more than

30,000 goods imported by the United States (see Broda and

Weinstein [2004]).We find that for the period between 1972

and 1988, the goods sector with the highest substitutability

among its varieties was crude petroleum and shale oil. The

goods sector in which varieties were the least substitutable

was footwear. In general, the degree of substitutability was

higher for homogeneous products (petroleum is an apt

Our recalculation of the import price index, based on a more

sophisticated version of the methodology outlined in the

box, suggests that the variety-adjusted index fell 28.0 percent

faster than the conventional index between 1972 and 2001,

or about 1.2 percentage points per year. The difference in

the rate of decline was particularly marked for the earlier

sample period: the variety-adjusted price of imports fell

19.7 percent faster than the unadjusted price between 1972

and 1988, or about 1.4 percentage points per year. For the

later sample period, the impact of variety growth was much

smaller, with the adjusted index falling 8.3 percent faster

Our empirical calculation of the import price index

begins with some assumptions about how consumers

think about varieties. Here, we assume that all consumers

evaluate varieties using a constant elasticity of

substitution (CES) utility function that places equal

weights on imports from every country. If the prices

of all varieties of an imported good are identical and

there are no differences in quality across varieties (an

assumption we relax in the longer version of this article),

we can write the price index that takes variety growth

a different good. As the equation makes clear, the effect

of variety growth on the import price index depends on

two factors. First, increasing the number of source countries

for a given good will lower the ratio of old to new

fall in the price index reflects the assumption that consumers

value greater choice. Second, the precise amount

by which new varieties affect the price index depends on

how substitutable varieties of each good are. If varieties

above equation is small, and increases in the number of

varieties will have little effect on the price index. In

sum, our methodology assumes that there are two determinants

underlying how varieties affect the price index:

the magnitude of the increase in varieties and the degree

of similarity among varieties.

of varieties is that it erroneously captures splits or mergers of

product classifications as a change in the number of varieties and

hence could artificially impact the calculation of the varietyadjusted

price index. Fortunately, the methodology we use in

our longer technical paper (Broda and Weinstein 2004) is robust

to a wide variety of data problems arising from the creation and

elimination of product classifications. For example, if goods are

randomly split or merged, then the index remains unchanged.

Moreover, the methodology is robust to new goods being of

higher quality than old goods because superior-quality goods will

claim a larger share of total import expenditures at fixed prices.

1

,

CURRENT I S S U E S I N ECONOMICS AND FINANCE V O L U M E 1 1 , N U M B E R 4

than the official index between 1990 and 2001, or about

0.8 percentage point per year.The lower rate of decline in the

later period may reflect the fact that many of the gains

resulting from the rising importance of East Asian trade may

have been realized before 1990.

To calculate the impact of variety growth on consumer

welfare, we have to make an additional assumption about

how the increased availability of foreign varieties affects

domestic production. For this exercise, we follow Krugman’s

(1980) argument that the number of domestic varieties

remains the same as the number of import varieties

changes. In reality, some imported varieties replace domestic

varieties while others just complement them. However, if

domestic varieties are replaced, the resources used in their

production may be put to work in the production of a new

variety of that product or some other good.

With the assumption that the number of domestic varieties

is unchanged,we can proceed to estimate the impact of

global variety growth on the well-being of U.S. consumers.

We found earlier that the official import price index understates

the rate of decline in import prices by 28.0 percent

over the three decades from 1972 to 2001. If imports

account for about 10 percent of U.S. GDP, then the value

to consumers of the increase in global varieties is about

sum represents what consumers would be willing to pay to

access the expanded set of varieties available in 2001.

Our estimates should be viewed with some caution. As we

noted earlier, we treat the interaction of domestic and

imported varieties very simply, relying on the assumption that

the number of domestic varieties remains the same when

the number of imported varieties increases. In addition,

we choose one way of modeling variety—the CES model

presented in Dixit and Stiglitz (1977)—while recognizing

Although our method has certain limitations, our findings

are striking indeed.A welfare gain of $260 billion from

global variety growth is three to six times larger than traditional

estimates of the gains from eliminating protectionism

(see, for example, Feenstra [1992] and Romer [1994]). Our

finding that the conventional import price index, by failing

to correct for the increase in varieties, understates the

decline in import prices by 1.2 percentage points per year

also underscores the considerable impact of variety growth.

To understand the magnitude of this understatement,

or “bias,” consider how it compares with other estimates of

bias in aggregate prices. In 1996, the Advisory Commission

to Study the Consumer Price Index found that the CPI at

that time overstated the increase in the cost of living by

about 1.1 percentage points per year (Boskin et al. 1996).

Improvements in the quality of consumer goods—which

were not taken into account in the calculation of the CPI—

numbers suggest that the variety-growth bias we find in the

import price index is, in fact, quite large: specifically, it is

twice as large as the quality-change bias in the overall price

This article reports the results from the first attempt to estimate

the impact of new varieties on the U.S. import price

by 1.2 percentage points per year for the 1972-2001

period. The real cost of imports was almost 30 percent lower

at the end of this period than the conventional price index

would suggest. This drop in import prices, we contend,

has raised U.S. welfare by $260 billion, or about 3 percent of

2001 GDP.

The magnitude of this gain from trade suggests that the

effects of variety growth on prices and welfare merit further

exploration. Additional research is needed on the interaction

between imported and domestic varieties. Our own

future work on these issues will consider how the growth

and elimination of domestic varieties affect consumer

prices and welfare.

1. The conventional index used as a benchmark in our analysis mimics the

official U.S.import price index in that it makes no adjustment for the impact of

variety growth. Although it is derived from a different formula, it is closely

related to the official index in practice.

2. Whether consumers assess goods coming from different countries as

different “varieties” is something we will be able to determine by estimating

the degree of substitutability between them. For example, high levels of substitutability

between Brazilian and Colombian coffees would suggest that

consumers do not regard these products as distinct varieties. An alternative

definition of a variety that appears in the trade literature is “goods coming

from a particular firm.” The problem with this definition, however, is that all

goods from a given firm are not the same variety.More practically, it is very

difficult to get firm- or plant-level data for a large set of countries that export

to the United States.

3. The 2001 classification for roasted coffee contains only caffeinated coffee in

packages of under two kilograms. The actual number of suppliers could be

higher.

4. In calculating the CPI, the Bureau of Labor Statistics omits the introduction

of new goods until they are eventually discovered as part of the gradual

rotation of the sample of goods. Even when the no-longer-new good enters the

CPI calculation, no adjustment is made for the consumer gains it provides in

relation to the earlier goods. For a good summary of CPI biases, see Lebow and

Rudd (2003).

5. The import price index and the CPI are both fixed-weight indexes; they

assign weights to the prices of individual goods and services that are fixed over

time.While the assessment of cost-of-living changes is a recognized measurement

goal of the fixed-weight index, it is a theoretical construct that cannot

readily be translated into a single, straightforward index formula. Therefore,

the fixed-weight index cannot easily track all of the factors that affect the cost

of living in today’s dynamic economy.

6. This article summarizes the findings of empirical work described in detail

in Broda and Weinstein (2004).

7. Broda and Weinstein (2004) show that the estimates obtained are reasonable

under a number of criteria.

8. Intuitively, if the price of 10 percent of the goods in the CPI falls by 30 percent,

then the CPI falls by 3 percent. This implies a rise in real GDP of 3 percent.

9. About 2 percentage points of the welfare gain accrue to the earlier sample

period.

10. We choose the Dixit and Stiglitz model because of its prominence, tractability,

and empirical feasibility.

11. While we compare the magnitude of the import index and CPI biases, we

cannot add the variety-growth bias to the existing CPI bias. This is mainly to

avoid a “double counting” of the effect of imported varieties. Many of the

imported goods are intermediate goods that already show up as lower prices in

final consumption goods.

12. We note, however, that the average quality-improvement bias in the CPI

masks a large variance across different CPI expenditure categories. Both the

1996 Boskin et al. study and a more recent paper by Lebow and Rudd (2003)

present estimates of quality bias in many important CPI subindexes—including

medical care, personal computers, televisions, toys, and audio equipment—that

are larger than our estimate of variety bias. Overall, the weighted sum of the

bias found for these categories, which account for 9.7 percent of the CPI, is

2.25 percentage points per year.Another category,“personal computer services

(Internet),” accounts for 0.1 percent of the CPI but has an estimated annual bias

of 19.0 percentage points.

13. Although some researchers have studied the price impact of variety growth

at the micro level (a single good, or at most a few goods), we know of no other

study that considers the effects of variety growth on aggregate prices.

Boskin,Michael J., Ellen R. Dulberger, Robert J. Gordon, Zvi Griliches, and Dale

Broda, Christian, and David Weinstein. 2004.“Globalization and the Gains from

Variety.”NBER Working Paper, no. 10314, February.

Dixit,Avinash K., and Joseph E. Stiglitz. 1977.“Monopolistic Competition and

297-308.

———. 1994. “New Product Varieties and the Measurement of International

Krugman,Paul. 1980.“Scale Economies,Product Differentiation, and the Pattern

Lebow, David E., and Jeremy B. Rudd. 2003. “Measurement Error in the

41, no. 1 (March): 159-201.

Romer, Paul M. 1994.“New Goods, Old Theory, and the Welfare Costs of Trade

Christian Broda is an economist in the International Research Function of the Research and Statistics Group; David Weinstein

is Carl S. Shoup Professor of the Japanese Economy in the Department of Economics at Columbia University.

of New York.Dorothy Meadow Sobol is the editor.

Editorial Staff:Valerie LaPorte, Mike De Mott, Michelle Bailer, Karen Carter

Production: Carol Perlmutter,David Rosenberg, Jane Urry