Ledermna 2010
Ledermna 2010
IN SEARCH OF
EMPIRICAL GUIDANCE
FOR INDUSTRIAL POLICIES
Public Disclosure Authorized
Daniel Lederman
William F. Maloney
blic Disclosure Authorized
Does What You Export Matter?
Does What You
Export Matter?
IN SEARCH OF EMPIRICAL
GUIDANCE FOR INDUSTRIAL
POLICIES
Daniel Lederman
William F. Maloney
© 2012 International Bank for Reconstruction and Development / The World Bank
1818 H Street NW,
Washington DC 20433
Telephone: 202-473-1000;
Internet: www.worldbank.org
1 2 3 4 15 14 13 12
This work is a product of the staff of The World Bank with external contributions. Note that The
World Bank does not necessarily own each component of the content included in the work. The
World Bank therefore does not warrant that the use of the content contained in the work will not
infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely
with you.
The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the
views of The World Bank, its Board of Executive Directors, or the governments they represent. The
World Bank does not guarantee the accuracy of the data included in this work. The boundaries,
colors, denominations, and other information shown on any map in this work do not imply any
judgment on the part of The World Bank concerning the legal status of any territory or the endorse-
ment or acceptance of such boundaries.
Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges
and immunities of The World Bank, all of which are specifically reserved.
This work is available under the Creative Commons Attribution 3.0 Unported license (CC BY 3.0)
http://creativecommons.org/licenses/by/3.0. Under the Creative Commons Attribution license, you
are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under
the following conditions:
Attribution—Please cite the work as follows: Lederman, Daniel, William F. Maloney. 2012. Does
What You Export Matter? In Search of Empirical Guidance for Industrial Policies. Latin American
Devolopment series. Washington DC: World Bank DOI: 10.1596/978-0-8213-8491-6. License:
Creative Commons Attribution CC BY 3.0
Translations—If you create a translation of this work, please add the following disclaimer along with
the attribution: This translation was not created by The World Bank and should not be considered
an official World Bank translation. The World Bank shall not be liable for any content or error in
this translation.
All queries on rights and licenses should be addressed to the Office of the Publisher, The World
Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@
worldbank.org.
Acknowledgments ix
Abbreviations xi
1 Introduction 1
Conceptual Issues 2
What Makes a Good Good? 5
Beyond Goods 7
Notes 10
4 Smart Goods 35
Wage Premiums and Educational Endowments in
Latin America 38
Wage Premiums across Industries 40
Country and Industry Effects on Skill Premiums 46
Exports and Industry Skill Premiums 48
Notes 53
v
vi contents
Bibliography 109
Index 121
Boxes
4.1 Estimating Skill Wage Premiums 37
5.1 Caribbean Super Stars? 62
6.1 Who Makes the iPod? 85
contents vii
Figures
2.1 Net Exporters of Natural Resources during
1980–2005 and Real GDP per Capita in 2005 15
2.2 Net Importers of Natural Resources during
1980–2005 and Real GDP per Capita in 2005 16
3.1 The Distribution of Exporters’ Incomes:
PRODY and 1 Standard Deviation 27
3.2 Regional and Country EXPYs 28
4.1 Skill Premiums and Skill Endowments in
Latin America 42
5.1a Relative Quality by Region, 1990–2001 60
5.1b Region and Country Relative Quality 61
5.1c Relative Quality Ladders for Top Latin American
Exports and Representative Products 63
5.2a Quality Growth by Region, 1990–2001 67
5.2b Region and Country Quality Growth 68
5.3 Quality Growth by Region, 1990–2001, Product
Fixed Effects Included 69
5.4 Growth and Standard Deviation of
Quality Growth, 1990–2001 71
5.5 Unit Value Ratio of New to Incumbent Goods
(25, 50, 75th percentiles) 74
6.1 Brazil and Mexico: Has Their Export Prowess
Resulted in Revealed Comparative Advantage (RCA)
in Patents for Aircraft, Office Computing, and
Accounting Equipment? 79
6.2a–c Revealed Comparative Advantage (RCA) in Exports
versus RCA in Innovation (1980–2005) 80
6.3 PRODY versus Value Added in China 87
7.1 Export-Revenue Concentration and Terms-of-
Trade Volatility 97
Tables
3.1 Influence of EXPY on Growth: Revisiting the
Evidence from Hausmann, Hwang, and Rodrik 29
4.1 Education and Skill Endowments in Latin America
and the Caribbean 39
4.2 Returns to Schooling and Skill Premiums in
Latin America and the Caribbean 41
4.3 Sectors Ranked by Skill Premium in Latin America 43
4.4 Explanatory Power of Industry and Country Effects
on Latin American and Caribbean Returns to
Schooling and Skill Premiums 47
viii contents
This study was undertaken under the Regional Studies Program of the
Office of the Chief Economist for Latin America and the Caribbean. The
authors are grateful to the Office of the Chief Economist for Latin Amer-
ica and the Caribbean, the Development Economics Research Group,
and the International Trade Department of the World Bank for their
support during the preparation of the manuscript. They also gratefully
acknowledge funding from the Multi-Donor Trust Fund on Trade exe-
cuted by the World Bank, International Trade and Integration Unit of the
Development Economics Research Group. Augusto de la Torre, Norman
Loayza, Otaviano Canuto, Jose Guilherme Reis, Bernard Hoekman,
Pravin Krishna, Andrés Rodríguez-Clare, Guido Porto, Irene Brambilla,
Ana Paula Cusolito, Francisco Rodríguez, Vittorio Corbo, and Roberto
Álvarez, among others, provided insightful comments on our previous
work, presentations, and papers that are now part of this manuscript.
Shifra Katz, a graduate student at the University of Chicago, provided
stellar research assistance for chapter 6 when she was an undergraduate at
the University of Maryland. Joana Naritomi, a PhD candidate at Harvard
University’s Political Economy Group, provided research assistance for
chapter 7 when she was a consultant for the World Bank. Finally, the
authors acknowledge the invaluable comments from two anonymous ref-
erees, Ricardo Hausmann and Cesar Hidalgo. All remaining errors are the
authors’ responsibility.
ix
Abbreviations
xi
1
Introduction
Does the content of what economies export matter for development? And,
if it does, can governments improve on the export basket that the market
generates through the shaping of industrial policy? This book considers
these questions by reviewing relevant literature and taking stock of what
is known from conceptual, empirical, and policy viewpoints.
A large literature answers affirmatively to the first question and sug-
gests the characteristics that distinguish desirable exports. Some schools of
thought are best known by their colorful metaphors: For example, natu-
ral resources are a “curse”; “high-tech” goods promote the “knowledge
economy”; a “product space” made up of “trees” (goods) from which
“monkeys” (entrepreneurs) can more easily jump to other trees fosters
growth. More prosaically, but no less controversially, goods which are
intensive in unskilled labor are thought to promote “pro-poor” or “shared
growth,” whereas those which are skilled-labor intensive are thought to
generate positive externalities for society as a whole. Concerns about
macroeconomic stability have led to a focus on the overall composition of
the export basket.
This book revisits many of these arguments conceptually and, wherever
possible, imports heuristic approaches into frameworks where, as more
familiar arguments, they can be held up to the light, rotated, and their
facets examined for brilliance or flaws. Second, the book examines what
emerges empirically as a basis for policy design. Specifically, given certain
conceptual arguments in favor of public sector intervention, do avail-
able data and empirical methods allow for actually doing so with a high
degree of confidence? In asking this question, the book assumes that policy
makers are competent and seek to raise the welfare of their citizens. This
assumption permits sidestepping the debate about whether government
failures trump market failures generically: In this sense, the book attempts
to “give industrial policy a chance.”
1
2 does what you export matter?
Conceptual Issues
The three chapters composing Part I explore the good as a unit of analy-
sis. Each chapter examines a literature that has argued in favor of certain
goods as growth promoting (or inhibiting, as the case may be) and dis-
cusses the conceptual arguments and empirical evidence supporting the
respective viewpoint.
Chapter 2 on Cursed Goods revisits in some detail the ubiquitous litera-
ture on the natural resource curse which has offered numerous arguments
over the course of 200 years, for example, the absence of inter-industry
spillovers, toxic political economy effects, and so on. Although at both
the conceptual and anecdotal levels these arguments are compelling, in the
end there is surprisingly little evidence for these particular effects. Indeed,
the empirical arguments recur to the aggregate level where it can be argued
that the majority of the evidence is, in fact, in favor of a resource blessing.
This is emphatically not to deny that in many countries natural resources
have been associated with negative consequences and these experiences
should be understood and contrasted with the successful growth stories.
However, at present the average effect of natural resource endowments
(and even mining output as a share of GDP or net exports) on growth
appears to be positive, although we remain concerned about the chal-
lenges posed by export concentration.
Chapter 3 on High Productivity Goods and Monkeys examines recent
literature produced by researchers mostly associated with the Kennedy
School of Government at Harvard University. This literature argues that
exporting products currently produced by rich countries yields spillovers
that lead to faster growth. Further, it postulates a learning externality
arising from the production of these goods. The evidence in favor of
this view relies on cross-country growth regressions to demonstrate the
growth-enhancing effects of high-productivity goods. However, these
results are relatively fragile, even after exploring some alternative growth-
model specifications, which arguably provide a fairer (less strict) test of
the underlying theoretical arguments put forth by Hausmann, Hwang,
and Rodrik (2007). This may imply that the high-productivity effect may
6 does what you export matter?
be overstated, or, following our focus on the demand side, it may also
be because goods exported by high-income countries, by definition, are
already generously supplied by competitive economies and, hence, there is
an offsetting low rents effect to any possible productivity externality.
The chapter also revisits the monkey-tree argument of Hidalgo et al.
(2007), and suggests that it can be cast as an externality with attendant
price offset effects. An irony emerges: Goods that are “close” to other
goods in the product space and therefore easy for monkeys (entrepreneurs)
to jump to, by definition, enjoy low barriers to entry. Further, the potential
for rents from Marshallian externalities is likely to have been dissipated.
In addition, historical correlations of indexes of comparative advantage
among industries (the underlying workhorse of the empirical product
space espoused by these authors) are unlikely to be useful predictors of
where the next high-rent, high-productivity goods are to come from. On
the other hand, populating a high-density segment of the product space
could result in export diversification, which in turn could help reduce
macroeconomic volatility, although this dimension was not explored by
the original authors.
Chapter 4 on Smart Goods extends the scope for industrial policy by
exploring labor market data from 16 Latin American and Caribbean econ-
omies. It assesses whether certain types of industries offer human capital
externalities. Although concerns about income distribution and pro-poor
growth would lead to the conclusion that subsidizing agriculture, for
example, could yield growth with higher demand for unskilled labor, it is
difficult to think of the market failure (besides capital market imperfec-
tions) that would justify such a policy. Specifically, there are many other
alternative policies that could redistribute income across the population,
such as taxes and transfers. In contrast, it is known from the empirical
literature on schooling that the aggregate (social) returns to schooling tend
to be higher than the micro-econometric estimates of the returns to school-
ing for individual workers (Krueger and Lindahl 2001). Consequently,
this chapter examines patterns of the “returns to schooling” by looking
at whether there is consistent evidence that certain sectors provide higher
“skill premiums” or returns to schooling than other sectors. Are there
some goods that are “smart” and should be encouraged due to market
failures in the accumulation of education?
Following other empirical analyses in this book, this chapter examines
whether country effects are more or less important than those related to
particular industries, and whether the latter should be looked to to pro-
vide the incentives to invest in education. In addition, the chapter assesses
the role of exports and export-product differentiation as determinants of
industry skill wage premiums. The preponderance of the evidence suggests
that country and industry characteristics help explain national differences
in the skill premium, but exports in general appear to be an important fac-
tor. This could imply that, at most, a combination of orthodox pro-trade
introduction 7
Beyond Goods
The two chapters in the second part of the book raise several issues related
to the previous analyses of the good or industry as the units of analysis.
It argues that, for a variety of reasons, the assumption of the good as a
homogeneous unit of analysis, produced in uniform ways across countries,
is incorrect in important ways, thus shedding further doubt on the wisdom
of pursing product- or industry-centered industrial policies.
Chapter 5 on Export Heterogeneity along the Quality Dimension intro-
duces a newer literature on export quality measured by unit values. It
goes to the other extreme by arguing that the important variance across
countries is differences of quality within narrowly defined product catego-
ries, rather than the products themselves. In sum, the issue for develop-
ment policy is not whether an economy exports wine or microchips; it is
about whether the economy produces Chateau Margaux for US$ 2,000 or
Charles Shaw’s Two-buck Chuck. Without full knowledge of the industry
structure, it is difficult to say anything about the welfare implications of
specializing in one product over another. However, since average quality
rises with level of development, the dynamics of quality (measured by the
growth of export unit values) potentially offers insights into the drivers of
economic growth by acting as a proxy for the accumulation of underlying
factors of production that yield high-quality goods and perhaps greater
productivity.
The findings support the argument that certain goods have greater
potential for quality growth due to longer “quality ladders” that offer
stronger convergence effects toward high unit values. This in itself is a
weak argument for industrial policies since there is no obvious market fail-
ure that suggests that countries are incorrectly specialized should they find
themselves in goods, such as commodities, with shorter ladders. Further,
critical factors affecting unit value growth appear to be country specific.
In particular, there are factors, perhaps deficient credit markets, poorly
articulated national innovation systems, or poor institutions that appear
to inhibit growth of unit values even within the same products where
advanced country export unit values appear to grow more robustly than
those of developing countries.
Chapter 6 on Heterogeneity in the Production of Goods argues that
seemingly identical goods appear to be produced with different tech-
nologies of production in different countries, thereby implying differing
potential for externalities. Looking at historical cases and at patenting
8 does what you export matter?
than a challenge for the foresight of central planners with good intentions;
it becomes a potentially harmful approach that could increase rather than
decrease export concentration.
Chapter 8 concludes with a brief discussion of the main findings and
policy implications. The focus is not a comprehensive list of specific pro-
grams, and “toolkits” are not provided for designing appropriate indus-
trial policies or examples of international “best practices.” Our aim is to
draw links between the basic notions of positive externalities, the best
available empirical evidence, and the challenges policy makers might face
in advocating for different types of industrial policies.
In the end, theory, intuition, and empirical evidence all suggest poten-
tially desirable public policies that go beyond the noninterventionist
orthodoxy. For example, findings indicate that there are arguments for
supporting efforts at diversification in natural resource exporters and
subsidizing exports that raise a country’s returns to schooling. More gen-
erally, a strong case can be made for “horizontalish” (neutral, on average,
across sectors) policies supporting the productivity and quality growth of
existing industries and the emergence of the ever-unpredictable new ones,
for example, in resolving market failures in the development of trade net-
works, improvements in quality, investment in research and development,
and so on.
However, to conclude, the literature to date offers few reliable empiri-
cal guides to the superiority of one type of good over another and hence
to the selection of products or industries for special treatment. Further,
what emerges consistently is an extraordinary heterogeneity of country
experiences within product categories. This ranges from identical goods
being produced with very different levels of productivity, quality, and tech-
nological sophistication, to the fact that, in an evolving global production
system, countries increasingly trade in tasks—fragments in the production
of a good—rendering the concept of a good increasingly anachronistic.
The concern with how countries produce what is currently exported argu-
ably merits more attention than what is produced. Understanding the
roots of the observed differential performance, in turn, feeds back into
the question of what is exported through conventional considerations of
comparative advantage. Throughout this book, simple empirical exercises
suggest that country-specific characteristics, rather than goods character-
istics, go a long way in explaining the incidence of potentially desirable
industrial structures.
With regard to crafting optimal baskets of goods, specific programs and
policies that could be part of such a policy stance remain unexplored here.
However, this is partly because more analytical work is needed to under-
stand how products within countries’ export bundles are correlated in terms
of quantities, prices, and factor demands. Without such knowledge, the
design of pro-diversification industrial policies must remain the subject of
modest policy experimentation with rigorous monitoring and evaluation.
10 does what you export matter?
Notes
1. However, even in this case, retaliation can reverse the original corporate
profit transfer across the competing economies (or the terms-of-trade improvement
in the case when one of the firms is an import-competing domestic monopoly).
2. Even if a country holds market power in the aggregate, a high level of
domestic competition could drive the export price to marginal cost and pass along
all potential rents to foreign importers. Government imposition of an optimal tariff
is effectively an internal coordination mechanism for restricting output of all domes-
tic agents so that the country enjoys the rents itself. The Organization of Petroleum
Exporting Countries (OPEC) performs the same service internationally.
3. See also Basu and Fernald (1995) for an example of how difficult it is to
econometrically identify spillovers.
4. Goldstein and Khan’s (1985) survey estimates price demand elasticities of
aggregate exports for several countries, and finds these centered around 1, imply-
ing substantial market power across many industries. However, Panagariya, Shah,
and Mishra (2001) argue that this contradicts a more mainstream assumption that
most countries face very high, even infinite elasticities, for their goods. They find
that, at greater levels of disaggregation and with an improved estimation approach,
estimates of a set of textile-related products lie between 60 and 136 for Bangladesh.
Estimates across a broader range of goods and countries are not available.
Part I
Perhaps the class of goods that have been considered to have an impact
on growth, in this case negative, are those based on natural resources;
indeed, a vibrant literature persists on the “resource curse.”1 Adam Smith
(1776) was perhaps the first to articulate a concern that mining was a bad
use of labor and capital and should be discouraged.2 The idea reappeared
in the mid-1950s in Latin America when Raúl Prebisch (1959), observ-
ing slowing regional growth, argued that natural resource industries had
fewer possibilities for technological progress. Further, Latin American
countries were condemned to decreasing relative prices on their exports.
These stylized facts helped to justify the subsequent import substitution
industrialization (ISI) experiment in modifying national productive struc-
tures. Subsequently, disenchantment with the inefficiencies of protection-
ism and the consequences of populist macroeconomic policies led to more
open trade regimes and less intrusive microeconomic policies, partly with
the example of East Asia’s rapid export-led growth in mind.
13
14 does what you export matter?
4
log natural resources net exports/labor force
SAU NOR
GAB KWTIS
2 CAN
NZL AUS
TT FIN
COG VEN MYS
DZA SUR CHL ARE DNK
ECU KAZ NLD IRL
0 LBR PNG IRCRI ARG RUS
CIV BOL NAM
NGA IDN COL URY SWE
1980–2005
CM AZE PERMEX
ZMBYEMMRTMNG R VNM PRYBRAZAF
SLB
HND
ZWE
TJ
–2 MWI SYR K BTN LVA
TGO GH GIN GT THA
MDGKEN MOZ FJI
NER UGA
–4 MLI
GNB
CAFTZA
MDA
–6 ETH y = 1.2584x – 12.102
BDI R2 = 0.5593
–8
5 6 7 8 9 10 11
log GDP per capita 2005
about average impact.4 Arguably, while colorful, continued use of the term
distracts from understanding why some countries have done well with
natural resources while others have not.
Numerous channels through which the curse might operate have been
offered. First, Prebisch (1959), among others, popularized the idea that
terms of trade of natural resource exporters would experience a secu-
lar decline over time (meaning without interruption) relative to those of
exporters of manufactures. Prebisch is thus perhaps the exception in being
preoccupied with the demand side of the quality of export debate, although
without identifying any compelling market failure to be addressed. How-
ever, even the stylized fact is somewhat in doubt. Cuddington, Ludema,
and Jayasuriya (2007) find that they cannot reject that relative commodity
prices follow a random walk across the 20th century with a single break in
1929. There is no intrinsic force driving the observed decline, and prices
could just as easily rise tomorrow as fall further.
Although commodity by commodity, important mean-reverting com-
ponents are evident and are, in fact, necessary for stabilization funds to
16 does what you export matter?
KORISR BEL
PRT AUT
JOR LCA GRC ESP FRA
0 MDVLBN GBR USA
DJI VCT SVK BHR
CPV LTUBLR SVN
RO BW HRV
TONWSM MKD CZE
–1 PAN
LSO BIHJAM TUR
VUT ALB TUN
GMB MAR UKR HUN
1980–2005
–2 ERI COM
SEN PHL EGYARMDOM MUS
PAK BLZ BGR
SLV
GEO
–3 BGD NIC LKA POL
SWZ
KHM IND
RWA NPL
–4 BFASDN
SLE CHN
BEN y = 1.0885x – 11.227
–5 R2 = 0.4602
–6 EST
–7
6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11
log GDP per capita 2005
be viable, the notion that long-run prices have a strong unpredictable and
permanent component appears more relevant today than at any time dur-
ing the last half century. Paul Krugman (2008) taking exactly the opposite
position from Prebisch, argues that continued growth by China and India,
combined with simply “running out of planet,” will lead to continued
strong excess demand such that “rich countries will face steady pressure
on their economies from rising resource prices, making it harder to raise
their standard of living.”5
Second, beginning with Smith, observers have argued that natural
resources are associated with lower human and physical capital accu-
mulation, productivity growth, and spillovers. The spillovers argument
is closest to the classic externalities argument, but neither is accepted as
conclusive in the literature. Even in Prebisch’s era, future Nobel Prize win-
ner Douglass North (1955, 252) argued that “the contention that regions
must industrialize in order to continue to grow . . . [is] based on some fun-
damental misconceptions.” Pioneer trade economist Jacob Viner argued,
“There are no inherent advantages of manufacturing over agriculture”
cursed goods: natural resources 17
(Viner 1952, 72). Consistent with Viner’s (1952) early assertion, Martin
and Mitra (2001) find total factor productivity growth to be higher in
agriculture than in manufacturing in a large sample of advanced and
developing countries.
Taking a broader view encompassing interindustry spillovers, Wright
and Czelusta (2007) and Irwin (2000) have argued that, contrary to
Smith’s prejudice, mining is a dynamic and knowledge-intensive industry
in many countries and was critical to U.S. development. Arguably, the
single most important general purpose technology of the 19th century,
the steam engine, arose as a learning spillover from the mining industry.
Blomström and Kokko (2007) have argued the same for forestry in Scan-
dinavia where Saab, a car and aircraft producer, and Volvo emerged from
truck producers serving the forestry industry. So too, Nokia, the telecom
giant, arose from what was originally a forestry company.
The question of why such miracles did not appear in Latin America and
the Caribbean points toward Baldwin’s argument that such spillovers are
not automatic, but depend on how goods are produced. Several authors
stress the economic complementarity of essential factors, particularly
human capital (see Gylfason 2001 and Bravo-Ortega and de Gregorio,
2007). In a related manner, Maloney (2007) argues that Latin America
missed opportunities for rapid resource-based growth due to deficient
technological adoption driven by two factors. First, deficient national
“learning” or “innovative” capacity, arising from low investment in
human capital and scientific infrastructure, led to weak capacity to inno-
vate or even take advantage of technological advances abroad. Second,
the period of inward-looking industrialization discouraged innovation
and created a sector whose growth depended on artificial monopoly rents
rather than the quasi-rents arising from technological adoption. At the
same time it undermined natural resource-intensive sectors that had the
potential for dynamic growth. Hence, natural resources were produced
in a “low-tech” way. The alternate path chosen by the United States,
Finland, and Sweden was and is, to an important degree, still open. Røed
Larsen (2004) argues that Norway’s surge from Scandinavian laggard in
the 1960s to regional leader in per capita income was based largely on the
opposite strategy to that chosen by Latin America and concludes: “Nor-
wegian Oil is a high technology sector which we may assume has much
the same positive spillover effects as manufacturing is supposed to have”
(Røed Larsen 2004, 17).
These arguments are central to the discussion surrounding the “Dutch
Disease” aspect of the resource curse emphasized by, among others,
Gylfason, Herbertsson, and Zoega (1999) and Sachs and Warner
(2001a, 2001b), where perhaps through an appreciated exchange rate or
classic Rybczinski effects, resource booms depress manufacturing activ-
ity.6 However, if the natural resource sector is not inferior in terms of
18 does what you export matter?
the Sachs and Warner results using either a net measure of resource
exports or the gross export measure without the adjustments for the two
countries, they find that the negative impact of natural resource abun-
dance on growth disappears.
In fact, even accepting the modified data, the interpretation of the
Sachs-Warner results is not entirely clear. Sala-i-Martin, Doppelhofer, and
Miller (2004), in their Bayesian search for robust explanatory variables
across millions of growth regressions, find a persistent negative sign when
the proxy enters. However, it is not robust enough to be considered a
core explanatory variable for growth as other variables appear to absorb
its influence. In contrast, domestic oil production as a share of national
income turns out to be a core explanatory variable and with a posi-
tive effect. In a similar vein, Lederman and Maloney (2007) show that,
controlling for fixed effects in a panel context, the negative impact of
resources also disappears, suggesting that it may not be their particular
proxy, but its correlation with unobserved country characteristics that is
driving the appearance of a resource curse. Manzano and Rigobon (2007)
concur and argue that the cross-sectional result arises from the accumula-
tion of foreign debt during periods when commodity prices were high,
especially during the 1970s, that led to a stifling debt overhang when
prices fell. These results, and the analogy to other bubbles, are impor-
tant, not only because they cast further doubt on the resource curse, but
especially because the policy implication is that the right levers to deal
with the lackluster performance of resource-rich developing countries in
recent decades lie in the realm of macroeconomic policy instead of trade
or industrial policies.
Bravo-Ortega and de Gregorio (2007), using the same proxy (as well
as resource exports over total exports), also find a negative cross-sectional
impact, but trace its origin to a Dutch Disease effect working through
human capital. Adding an interactive human capital term suggests that as
the stock of human capital rises, the marginal effect of the exports of natu-
ral resources on income growth rises and becomes positive. This is broadly
consistent with Gylfason, Herbertsson, and Zoega’s (1999) argument that
a national effort in education is especially necessary in resource-rich coun-
tries, although without their hypothesis that resource-rich sectors intrinsi-
cally require, and hence induce, less education. However, Bravo-Ortega
and de Gregorio find that the point at which exports of natural resources
begin to contribute positively to growth occurs at around three years of
education, a level achieved by all but the poorest countries in the world.
Sachs and Vial (2001) and Sachs and Warner (1995b) confirm a negative
and robust relationship using a second, related proxy—the share of natural
resources exports in total exports, and this proved somewhat more robust.
However, it again does not make Sala-i-Martin, Doppelhofer, and Miller’s
(2004) core list of robust regressors. Further, when Lederman and Maloney
(2007) include a generic measure of concentration, the Herfindahl Index,
cursed goods: natural resources 21
Notes
1. This chapter borrows heavily from Lederman and Maloney (2007 and
2008) and Maloney (2007). See also van der Ploeg (2011) for a recent summary of
some aspects of the literature.
2. “Projects of mining, instead of replacing capital employed in them, together
with ordinary profits of stock, commonly absorb both capital and stock. They are
the projects, therefore, to which of all others a prudent law-giver, who desired to
increase the capital of his nation, would least choose to give any extraordinary
encouragement . . . .”
3. See Irwin (2000) for the United States; Innis (1933) and Watkins (1963) for
Canada; Wright (2001) and Czelusta (2007) for Australia; Blomström and Kokko
(2007) and Blomström and Meller (1991) for Scandinavia. Latin America also
offers its success stories: Monterrey, Mexico; Medellin, Colombia; and São Paolo,
Brazil all grew to become dynamic industrial centers based on mining, and in the
latter two cases, coffee. Copper-rich Chile has been the region’s model economy
since the late 1980s.
4. A “venture capital” curse is not discussed because 19 out of 20 venture
capital- financed firms go bankrupt. If the central tendency is that natural resources
have a positive effect, then they remain a blessing, albeit a conditional one. There is
a need to understand the complementary factors necessary to maximize it. This is
cursed goods: natural resources 23
not different than understanding why Taiwan, China, did better with its electron-
ics industry than Mexico, or that Italy did better with its fashion industry than the
Republic of Korea did with “Project Milan.”
5. Krugman, Paul, “Running Out of Planet to Exploit,” The New York Times,
March 21, 2008.
6. These arguments are fundamentally modifications of the Rybczynski theo-
rem of the Heckscher-Ohlin-Vanek framework in which it can be shown, using a
2x2 Edgeworth Box, that an increased endowment of one good necessarily implies
an absolute fall in the production of the good that is not intensive in that particular
factor.
7. Sachs and Warner (1995b) argue that the Dutch Disease leads to concen-
tration in resource exports, which they assume to have fewer possibilities for pro-
ductivity growth. Evidence shows that net exports of energy and mining products
per worker are associated with concentration of export revenues, which in turn are
linked to terms-of-trade volatility. This material is discussed in chapter 7.
8. During 1998–2007, microchips accounted for over 25 percent of Costa
Rica’s total merchandise exports. See Lederman, Rodríguez-Clare, and Xu (2011).
9. The other papers by Sachs and Warner (1995b, 1997, 1999, 2001a, 2001b)
contain the basic results of 1995b, at times using a slightly longer time span
(1965–1990 instead of 1970–1989), and often including additional time-invariant
explanatory variables such as dummy variables identifying tropical and landlocked
countries, plus some additional social variables.
10. Numerous countries in Asia and Latin America have a large presence of
export processing zones that would, using the gross measure, overstate their true
abundance in manufacturing-related factors. The variable also shows substantial
volatility over time, reflecting terms-of-trade movements. Hence, the average for
the period is probably a better measure than the initial period value that was used
by Sachs and Warner in several of their papers.
11. Assuming identical preferences, a country will show positive net exports of
resource-intensive goods if its share of productivity-adjusted world endowments
exceeds its share of world consumption. Usually, the net exports are then mea-
sured with respect to the quantity of other factors of production, such as the labor
force.
12. It should be noted that the cited references show that the Heckscher-Ohlin
model of factor endowments performs relatively well for natural resource net
exports, but performs less well for manufactured goods. The current debate in the
trade literature revolves around the question of how the Heckscher-Ohlin model
might be amended (by considering, for example, technological differences across
countries, or economies of scale) to help better predict the observed patterns of net
exports across countries. However, there is no debate about the use of net exports
as a proxy for revealed comparative advantage in this literature.
13. It is tempting to view the Bayesian approach to testing for robust regressors
as the final word, as in Sachs and Warner (2001b), where the authors (mistakenly)
argue that the curse is robust based on the 2000 working paper version of Sala-i-
Martin et al (2004). As mentioned, Sala-i-Martin, Doppelhofer, and Miller find a
positive effect of the fraction of mining in GDP on growth. However, this approach
is not well suited for dealing with biased coefficients or robustness of coefficients
over time. That is, the results might change with changes in the time period covered
by the data, or by including only exogenous explanatory variables. Furthermore,
the Bayesian approach does not yield robust results even in the presence of mea-
surement errors in the GDP data, and therefore yield different results when using
slightly different versions of the Penn World Tables purchasing-power-adjusted
data. See Ciccone and Jarocinski (2010).
24 does what you export matter?
14. Lederman and Maloney (2008) use quantile regressions to test for inter-
national heterogeneity in the natural resource variable coefficients. They find that
in the long run, net exports of natural resources are positively correlated, with
no statistically significant international heterogeneity, with the level of GDP per
capita. This implies that countries would be poorer if they did not have natural
resources. Still, there is no evidence of a curse, even if it is true that the long-term
gains from natural resources had already been absorbed prior to 1980. There is
heterogeneity in the growth regressions.
3
25
26 does what you export matter?
What are these high productivity goods? Figure 3.1 presents the PRODY
values for relatively aggregated categories. Two findings are important.
rich country “high productivity” goods 27
30,000
25,000
20,000
PRODY
15,000
10,000
5,000
0
r
es
d s
s
ac ls
d ca ry
e
ns uct d
ea
e
ct
ile
an al
as
od
ct
al
an mis
tio
ic
tra rod an
a
al hem ine
bb
bl
po s
du
er
du
et
rv
w
xt
gl
rta
ta
p ls
ru
m
tri ani
se
ot
on min
h
te
ro
ro
ge
fo
d
rp
dp
lie i
ve
m
oo
he
c
tic
w
at
c
as
ec
st
le
pl
el
exports
16,000
14,000
12,000
10,000
EXPY
8,000
6,000
4,000
2,000
0
D
C C
na
ua m ia
lo a
Ve Mex pe
zu ra o
Ar ela, zil
ge RB
N hamna
nt as
C s
d ba e
C ba s
itt Su om o
an na a
U Nev e
Pa gu s
C n ay
D a ma
. ica
ua p.
Ja Pe r
m ru
G B ca
El ate lize
F. lva la
ui r
H Bol na
du ia
Pa uya as
N rag na
ar ay
H a
C iti
a
do
G do
ille
To do
ru i
de ud
ic
anBar hil
ol g
s ri bi
d m
ub
EC
LA
a
G Ber Ind
Sa ma
on iv
Ec Re
ai
G r
ic u
hi
ag
om R
Ba nt
os a
u e
B
O
.A
t
ne
ad
.K
id
St
in
Tr
Note: EXPY = Average Productivity/Income level (PRODY) of Export basket; GDP = gross domestic product; GMM = Generalized Method of
Moments; IV = Instrumental Variable.
Regressions include decade dummies.
Robust t statistics in parenthesis, * significant at 10 percent, ** significant at 5 percent, *** significant at 1 percent.
30 does what you export matter?
Hausmann and Klinger (2006, 2007) and Hidalgo et al. (2007), among
others, have also been pursuing another type of externality as a justifica-
tion for ranking goods by desirability, customarily termed the “Monkey
and Tree” analogy. In this context, a country’s product space is likened
to a forest with monkeys climbing trees as a metaphor for productivity
growth. To capture the evolution of economies, certain goods allow an
easier transition to other goods, and, hence, a continuing dynamic growth
rich country “high productivity” goods 31
process. Monkeys climb up trees but at a certain point would need to jump
to other trees (new goods), which they can then climb again.
The imagery is attractive, although the ambiguous relationship to
standard economic models makes it somewhat difficult to dissect as an
argument. It is not conceptually obvious why jumping from tree to tree
is preferred to being in one very tall tree, although, in practice, most
countries have graduated through industries. Therefore, it is fair to ask
what facilitates jumping into new areas. Again, although the link between
diversification and productivity is not tight (see Harrison and Rodríguez-
Clare 2010), from a volatility point of view, diversification may matter to
growth. However, even here, it is not clear that the answer to having one
hugely productive, rent-generating tree (perhaps looking suspiciously like
an oil derrick) is to diversify production, rather than to financially smooth
and hedge across time.
Whatever the benefits of having other trees in proximity, Harrison and
Rodríguez-Clare (2010) have come the closest to approximating a main-
stream argument by highlighting the analogy from a tree in close proxim-
ity to others where jumps are easy to standard Marshalling externalities
with the same caveats discussed earlier. In particular, if a good provides
easy jumping in one country, then that must be the case in all countries,
and international prices must reflect this. Unless a poor country gets asym-
metric benefits from such agglomeration, which is plausible as it fills out
its industrial structure, proximity effects may be offset. Further, trees in
the dense forest are not only easy to jump from, but they may also be easy
to jump to since there are so many potential trees of origin. The barriers
to entering industries in dense parts of the forest must, by definition, be
lower, and competition must be higher and rents few.
Other issues arise when reflecting on the relationship between the exter-
nalities surrounding proximity and those being captured by PRODY. For
example, is there any guarantee that rich country goods are also those in
the thick part of the forest? Expectations that frontier goods produced by
rich countries, that is, those with the highest quasi-rents from innovation,
would, by definition, be at the edge of the forest with the next obvious
place for monkeys to jump yet to be invented.
This raises an issue that is present in the previous arguments but espe-
cially germane here. In the case of Scotland in the 19th century, mining led
to the invention of the steam engine, a transformative technology. Like-
wise, Saab and Volvo began as trucking companies for the Swedish forestry
industry. Scottish and Swedish monkeys jumped to major new sources
of economic dynamism. As Blomström and Kokko (2007) stress, Nokia
emerged from a forestry company. However, do these trees remain close to
each other today, particularly when, as discussed in chapter 6, production is
so globally fragmented? Both forestry and mining used to have transporta-
tion industries located very close to them in the forest, and had the quality
of “transport industry proximity.” Does that mean that Chile is likely to
32 does what you export matter?
develop its own Volvo? Hausmann and Klinger (2007) find that the product
space they estimate in 1975 does almost as good a job at predicting jumps
between 1995 and 2000 as a product space estimated in 1995, indicating
that it is a relevant analytical structure to study the dynamics in a 20-year
horizon. However, it is unlikely that the next Nokia will emerge from a
forest company again. In fact, it would seem better for Chile to explore
whether the genetic modifications it is undertaking to make its salmon
more disease-resistant might lead to a cure for cancer, for example—a rela-
tionship that could potentially make the country billions. Such a jump, of
course, could not be inferred from past relational patterns any more than
the steam engine, the cell phone, Google, or the iPhone could be.
Finally, as discussed in chapter 6, it may not be just the distance among
trees, but how good the monkeys are at jumping between them. It may not
be the nature of the forest, but the level of the Simian capital that is most
relevant. Hausmann and Hildalgo (2009, 2010) have, in fact, recently
shifted the discussion toward understanding the underlying capabilities
required to produce goods and, presumably, to jump among trees. In the
case of EXPY, they have moved away from productivity per se. Hausmann
and Hidalgo (2009) use information on the structure of the network con-
necting countries to the products they export to create estimates of the
capabilities required to make products. Hidalgo (2010) shows that, in
fact, EXPY can be disaggregated into a component capturing income (pro-
ductivity) and one capturing estimates of the number of capabilities (as
derived from measures of relative comparative advantage). These capabili-
ties can be thought of as very specific factors of production ranging from
infrastructure to norms, institutions, and social networks. The sophistica-
tion of a product relates to the number of capabilities that the product
requires, and the complexity of a country’s economy is related to the set of
available capabilities (see Hidalgo 2010). Certain measures of capabilities,
rather than productivity, are correlated with growth and income. Further,
some measures are shown to be robust regarding the inclusion of concen-
tration proxies, although the intuition behind these particular measures is
not immediately clear.
These discussions reflect a fascinating application of recent advances
in studies of networks and merit further discussion. Several issues, how-
ever, remain on the table. First, the link between capabilities, or factors of
production more generally, and income and growth is expected. What is
not yet clear is the link between what is produced and the accumulation
of these capabilities. Hidalgo, for example, argues that “countries become
what they make.” As an argument in favor of supporting certain goods,
this can be seen as a restatement of the externalities argument: Production
of certain goods will lead to knowledge spillovers beyond that industry.
This is not yet supported by the work in the field, and the automaticity
of a country accumulating this capacity, as discussed in chapters 5 and 6,
remains in doubt.
rich country “high productivity” goods 33
Notes
Smart Goods
35
36 does what you export matter?
where the subscript i denotes individuals, j denotes the industry that the
individual is affiliated with, and t denotes years. The hourly wage is given
by w. It is computed as the reported weekly wage divided by the number
of hours worked per week (in several surveys these answers refer to the
total wages received and the number of hours worked during the week
prior to the survey).
The variable used to construct the skill premium is education, denoted
by Ed. In one approach, skilled workers are those with at least a high
school diploma. Thus, the function f(Edijt) becomes a binary variable (Sk)
that is equal to one if the individual has at least a high school diploma.
Thus, the wage model is as follows:
where YEd are years of education. The coefficient a measures the per-
centage point increase in wages due to an additional year of education.
This model controls for individual characteristics in the vector x, and
for industry and year effects in the indicator variables d. The controls
included in x are gender, age and age squared, marital status, whether the
individual works full time or part time, a dummy variable for individuals
in rural areas, and regional dummy variables. It should be noted that
estimates from these equations provide correlations in a cross-section of
workers, and therefore the estimates should be interpreted as reduced
form coefficients measuring the average difference in wages between
actual skilled and unskilled workers. These are not to be taken as predic-
tions for specific individuals should they move into the skilled group.
(continued on next page)
38 does what you export matter?
The authors estimate average skill premiums for the national economy
as well as for the numerous industries within countries. After summa-
rizing the authors’ estimates of sectoral skill premiums for 61 tradable
and nontradable sectors in each economy, including 23 manufacturing
sectors, the chapter reviews empirical analyses of industry and country
effects on industry-specific wage premiums, and provides a preliminary
assessment of export-related determinants of the skill premium in manu-
facturing industries.
Table 4.1 contains the basic descriptive statistics of the education and
skill variables. The first two columns show sharp differences in average
years of education and in the ratios of skilled to unskilled workers across
countries (skilled workers are defined as individuals who hold a high
school diploma). Average years of education are comparatively high in
Argentina (10.73), Uruguay (9.68), Chile (9.1), Panama (8.97), Colombia
(8.55), and Ecuador, the Dominican Republic, and Mexico (above 7.9).
These countries also show the highest share of skilled workers, ranging
from 27 percent in Mexico to 52 percent in Argentina (in Colombia, by
contrast, the share is lower). Years of education are lowest in Nicaragua,
Guatemala, and Honduras (5.31, 5.70, and 5.92, respectively), but the
share of skilled workers is lowest in Nicaragua and Brazil (9 percent
and 15 percent, respectively). In the cases of Argentina and Uruguay, the
comparatively high observed levels of education and shares of skilled
workers are partly explained by survey design. In these two countries, the
household surveys covered only urban households. In the remaining 14
countries, the surveys are representative of the rural population as well
as the urban one.
Table 4.1 Education and Skill Endowments in Latin America and the Caribbean
Average Share of skilled workersa Share of highly skilled workersb
years of education All Male Female All Male Female
Country (1) (2) (3) (4) (5) (6) (7)
Argentina 10.73 0.52 0.49 0.54 0.24 0.24 0.24
Brazil 7.37 0.15 0.13 0.17 0.25 0.29 0.22
Chile 9.10 0.40 0.39 0.41 0.24 0.25 0.24
Colombia 8.55 0.20 0.21 0.19 0.55 0.55 0.54
Costa Rica 7.68 0.18 0.18 0.18 0.34 0.35 0.32
Dominican Rep. 8.02 0.30 0.28 0.33 0.34 0.34 0.34
Ecuador 8.06 0.32 0.32 0.32 0.33 0.33 0.32
El Salvador 6.20 0.23 0.24 0.22 0.22 0.21 0.17
Guatemala 5.70 0.19 0.22 0.16 0.27 0.32 0.22
Honduras 5.92 0.19 0.19 0.20 0.30 0.37 0.24
Mexico 7.94 0.27 0.28 0.26 0.41 0.45 0.37
Nicaragua 5.31 0.09 0.09 0.09 0.46 0.49 0.42
Panama 8.97 0.37 0.34 0.40 0.31 0.28 0.34
Paraguay 7.45 0.25 0.25 0.26 0.23 0.23 0.23
Peru 7.98 0.23 0.24 0.21 0.45 0.45 0.46
Uruguay 9.68 0.33 0.30 0.35 0.35 0.32 0.37
Source: Brambilla et al. (2011).
Note: (a) The share of workers with a high school diploma or more over the total number of workers. (b) The share of workers with more
39
than a high school diploma over workers with a high school diploma.
40 does what you export matter?
Column 5 in table 4.1 presents the share of workers with more than
a high school diploma (individuals with tertiary education, some college
experience, college degree, and graduate degrees), over the total of work-
ers with at least a high school diploma. The differences across countries
are obvious, demonstrating that the composition of the skilled labor force
varies across countries. Countries with high shares of highly skilled work-
ers in the skilled group (41 to 55 percent) are Colombia, Peru, Mexico, and
Nicaragua. It is important to note that since Nicaragua is the country with
the lowest share of skilled workers (with high school), the relatively few
workers with advanced degrees tend to reach a high level of educational
attainment. Countries with low shares of highly skilled workers are El
Salvador, Paraguay, Argentina, and Chile (19 to 23 percent). The partici-
pation of highly skilled workers in the total labor force can be obtained by
multiplying column 5 by column 2.
Table 4.2 presents the results for the returns to schooling, as well as
for the skill premium for the sample of Latin American and Caribbean
countries. The first three columns under the heading “Years of educa-
tion” show the estimates of the returns to schooling for the baseline
estimation, from an estimation that utilizes only data from full-time
workers (a sub-sample of all workers), and results from a median regres-
sion (which is less sensitive to outliers than Ordinary Least Squares [OLS]
estimates). The results under the “Skill premium” heading follow the
same pattern, but focus on the relative-wage effects of a high school
diploma. Both sets of results come from the online appendix published
by Brambilla et al. (2011).
As with the skill endowments, there is a notable heterogeneity across
countries in the estimated returns to years of schooling and in the skill
premiums. An immediate question that emerges is whether the skill premi-
ums (and the returns to schooling) are correlated with the observed share
of skilled workers. The evidence appears in figure 4.1: The correlation is
negative, suggesting that countries with a higher supply of skilled work-
ers (relative to unskilled workers) also tend to have lower skill premiums.
Hence, country characteristics do matter as determinants of the skill pre-
mium. What about industries?
0.60
0.50
share of skilled workers
0.40
0.30
y = 0.718e–1.80x
R2 = 0.386
0.20
0.10
0.00
0.300 0.400 0.500 0.600 0.700 0.800 0.900 1.000
skill premium
15 out of 58, listed above the first dashed line in the table) includes
numerous nontradable sectors, a few natural-resource intensive indus-
tries, and surprisingly few manufacturing industries. Educational services
(“Education” and “Research and Development”) activities appear in
this group, as expected. However, mining activities and agricultural ser-
vice activities are also at the top of the list. In contrast, manufacture of
seemingly “high-tech” goods such as “Manufacture of radio, television,
and communication equipment and apparatus” appears at the bottom
of the list, among the group of activities in the lowest 25th percentile of
the distribution. This list also includes employment in households and
leisure and tourism services (such as “Hotels and Restaurants”). Thus,
this quick look at the ranking of sectors in terms of their skill premiums
46 does what you export matter?
In order to study the role of exports in shaping the wages of skilled work-
ers in Latin America, Brambilla et al. (2011) provide two sets of explor-
atory evidence. One examines the role of industry exports as a share of
national GDP by industry; the other studies the role of export unit values
(at the 2-digit level) in determining industry wage premiums.
Table 4.5 reports the results concerning the role of the incidence of
industry exports in GDP. The correlation between exports and industry
returns to schooling and skill premiums is positive, and this correlation
coefficient rises after controlling for industry effects (column 2), but its
statistical significance disappears with the inclusion of country effects
(columns 3 and 4). However, the positive effect of exports on the returns
to schooling reappears after controlling for the level of development
of each country (proxied by the log of GDP per capita) and national
skill endowments (the log of the ratio of skilled—with completed high
school—to unskilled workers). The results for the skill premiums are
somewhat weaker, but the partial correlation is also positive after includ-
ing these sets of controls. Moreover, the evidence suggests that richer
countries pay higher skill premiums, and, as expected, countries with
relatively more workers who completed high school tend to pay lower
skill premiums.
Table 4.6 shows the results concerning the role of export prices (unit
values) on skill premiums in Latin America. Neither unit values nor the
dispersion of unit values explains the industry skill premium. It is possible
that this is a result of the noise in the unit values data. Therefore, Brambilla
et al. (2011) reported results, reproduced in table 4.6, that measure the
variance of unit values in different ways in attempts to reduce the noise
due to measurement errors. For instance, in specification C, where the top
and bottom 5 percent of the unit values are trimmed, the dispersion in unit
values becomes significant in some regressions. This hints that the scope
Table 4.5 Exports, Countries, and Industries as Determinants of Industry-Skill Wage Premiums
(1) (2) (3) (4) (5) (6) (7) (8) (9)
A) Return to Schooling
Log exports/GDP 0.00250*** 0.00342*** 0.00025 –0.00027 0.00252*** 0.00309*** 0.00261** 0.00249** 0.00238**
[0.00091] [0.00096] [0.00099] [0.00120] [0.00093] [0.00102] [0.00104] [0.00113] [0.00113]
Log GDP per 0.01943*** 0.01340*** 0.01439*** 0.02057*** 0.01968***
capita [0.00344] [0.00512] [0.00509] [0.00592] [0.00601]
Log skilled/ –0.01131*** –0.00923*** –0.01247*** –0.00710** –0.00899**
unskilled [0.00320] [0.00336] [0.00363] [0.00351] [0.00411]
Enrollment rate 0.00032* 0.00021 0.00021 0.00018
[0.00018] [0.00019] [0.00020] [0.00020]
Export constraints 0.00401 0.00228 0.00628 0.00501
[0.00425] [0.00429] [0.00454] [0.00476]
Doing business –0.00010** –0.00005
index [0.00005] [0.00005]
Average firm size –0.09317** –0.07769*
[0.03803] [0.04188]
Country dummies No No Yes Yes No No No No No
Industry dummies No Yes No Yes Yes Yes Yes Yes Yes
Observations 287 287 287 287 287 287 287 261 261
R-squared 0.026 0.419 0.273 0.608 0.485 0.493 0.502 0.506 0.507
Table 4.5 Exports, Countries, and Industries as Determinants of Industry-Skill Wage Premiums (continued)
(1) (2) (3) (4) (5) (6) (7) (8) (9)
B) Skill Premium
Log exports/GDP 0.03292*** 0.03330*** 0.02187*** 0.00375 0.00997 0.01707** 0.01817** 0.01997** 0.02098**
[0.00798] [0.00968] [0.00687] [0.00913] [0.00714] [0.00788] [0.00808] [0.00883] [0.00891]
Log GDP per 0.17993*** 0.12134*** 0.11874*** 0.13265*** 0.13921***
capita [0.02762] [0.03959] [0.03985] [0.04683] [0.04741]
Log skilled/ –0.42187*** –0.41095*** –0.40506*** –0.40962*** –0.39591***
unskilled [0.02687] [0.02791] [0.02948] [0.02947] [0.03314]
Enrollment rate 0.00290** 0.00317** 0.00262 0.00289*
[0.00145] [0.00151] [0.00159] [0.00162]
Export constraints –0.02069 –0.01805 –0.01319 –0.00466
[0.03309] [0.03340] [0.03606] [0.03727]
Doing business 0.00022 0.00038
index [0.00035] [0.00042]
Average firm size 0.02015 -0.10155
[0.27825] [0.30904]
Country dummies No No Yes Yes No No No No No
Industry dummies No Yes No Yes Yes Yes Yes Yes Yes
Observations 285 285 285 285 285 285 285 259 259
R-squared 0.057 0.256 0.562 0.722 0.619 0.626 0.626 0.624 0.625
Source: Brambilla et al. (2011, table 5). Standard errors in parentheses. Significance at 1, 5, and 10 percent denoted by ***, **, and *.
Note: GDP= gross domestic product.
Table 4.6 Export-Product Differentiation and Industry-Skill Wage Premiums
Years of education Skill premium
(1) (2) (3) (4) (5) (6)
PANEL A
Log unit value 0.0006 –0.0001 0.0037 –0.009
[0.0009] [0.001] [0.006] [0.01]
Log var(Unit_value) 0.00031 0.0003 0.003 0.006
[0.0003] [0.0007] [0.003] [0.006]
Log exports/GDP 0.0024** 0.0024** 0.0024** 0.017** 0.016* 0.016*
[0.001] [0.001] [0.001] [0.008] [0.008] [0.008]
Observations 287 287 287 285 285 285
R-squared 0.5 0.5 0.5 0.63 0.63 0.63
PANEL B
Log unit value –0.0002 –0.0006 0.024 0.021
[0.002] [0.003] [0.023] [0.023]
Log var(Unit_value) 0.0003 0.0003 0.0029 0.0025
[0.0003] [0.0003] [0.0029] [0.0029]
Log exports/GDP 0.0026** 0.0024** 0.0024** 0.017** 0.016* 0.015*
[0.001] [0.001] [0.001] [0.008] [0.008] [0.008]
Observations 287 287 287 285 285 285
R-squared 0.5 0.5 0.5 0.628 0.628 0.629
51
Notes
1. See, for example, Krueger and Lindahl (2001) on the theoretical arguments
and empirical evidence concerning externalities from education. However, it is
plausible that in some countries the private returns to education can be close to the
social returns. For skeptical views with applications to the cases of Italy and the
United States, see Ciccone, Cingano, and Cipollone (2004) and Ciccone and Peri
(2006), respectively.
2. See Griliches (1977), Card (1999), and Krueger and Lindahl (2001) on the
econometric issues that plague estimates of the returns to schooling or the skill
premium. In their review of the literature as of 2000, Krueger and Lindahl conclude
that there is surprisingly little evidence of ability bias in the literature.
3. To be sure, there are likely to be many other plausible explanations for inter-
industry wage differentials other than the export-driven efficiency wages argument.
For example, unionization and policy distortions might make some sectors pay
higher wages to workers with a given skill level. However, if exports are associated
with higher wage premiums, it is unlikely to be driven by unionization, as it is well
known that export-oriented industries are less unionized than import-competing
and public sector industries.
4. In technical terms, to ascertain the extent to which the variance of country
or industry characteristics explains the variance of skill premiums, one would need
to know how much of the variance of national characteristics is explained by indus-
try characteristics and vice versa. To draw such estimates, one would need to make
assumptions about the relative exogeneity of country and industry characteristics.
Part II
Beyond Goods
5
57
58 does what you export matter?
Figure 5.1a compares the median level of export unit values across regions
for exports to the United States. Since unit values come in their “units”—
US$ per bushel, ton, car, bottle, and so on—the quality leaders (90th
percentile) are standardized within each product category to generate a
measure of “relative quality.” Consistent with Schott’s finding, the rich
countries of the OECD have the highest median relative quality level.
The Latin America and the Caribbean region, the Middle East and North
Africa region, and the Sub-Saharan Africa region follow. Eastern Europe
and Asia, including the rich economies of Asia (EAP [High]: Republic of
Korea, Taiwan, China, Singapore, Hong Kong SAR, China) and emerg-
ing Asia (EAP [Low]: China, Philippines, Malaysia, and so on) follow.
The Latin America and Caribbean–East Asia counterintuitive finding was
discovered independently by Schott (2003). Figure 5.1b disaggregates the
Latin America and Caribbean region for reference. The peculiarly high
values found for relative quality (and PRODY) in the Caribbean are dis-
cussed in box 5.1.
At a product level, the graphs in figure 5.1c offer a hyper-disaggregated
view of 12 goods at the Harmonized System 10 level, the finest disag-
gregation available. The categories were selected using a combination
of importance in the export basket of the Latin America and Caribbean
region, as well as the representativeness of certain types of goods. Turbofan
airplanes are not common in the region, but the success of the Brazilian
manufacturer, EMBRAER, does have importance as a potential sector
that merits benchmarking.
60 does what you export matter?
0.5
0.4
0.3
0.2
0.1
0
m -
D -
m -
m -
R
co gh
EC n
co gh
co ow
EN
EC
AF
LA
SA
O (no
e)
EA )
e)
in (hi
in (hi
e
in (l
M
P
pe
D
P
EA
EC
ro
Eu
O
region
The first noteworthy point is that the implicit length of the quality
ladder as measured along the vertical axis varies substantially by prod-
uct. Gold and silver bullion, for example, is concentrated between .8
and 1, with very little vertical differentiation. Footwear, men’s shirts,
and even microprocessors show values from .1 to substantially over
1. In general, we may expect that commodities, almost by definition,
would have less room for vertical differentiation, and this is largely
true. Gold, silver, bananas, and fuel oil have relatively short quality
ladders. Therefore, Latin America’s concentration in commodities may
explain its relatively high overall quality. However, this does not extend
to all resource-based goods: For instance, the variance in peeled, frozen
shrimp and prawns, wine, and coffee is similar to passenger vehicles and
aircraft. Overall, though, once the goods produced are controlled for,
the region’s ranking falls to third-lowest, suggesting that the commodity
impact is important.
Even a casual perusal of these figures brings some anomalies to light
that point to the difficulty of interpreting unit values. First, the data
record imports from countries regardless of whether or not the goods
are produced there. Hence, Singapore is a high-quality exporter of coffee
(and, as in chapter 3, at times “asses, mules and hinnies”) when, in fact,
these are reexports through an entrepôt. Likewise, more than one level of
export heterogeneity along the quality dimension 61
1.0
median relative quality
0.8
0.6
0.4
0.2
0
Ba ent a.
Ba am a
G e ay
te ,RB
ba a
D ndu vis
SuAnt pe
D
N elo da
Bo ana
U Ch o
C lo ica
Ve Pa a R ia
ne ra ica
B la
. R as
ex .
C C
na
ua m ia
P ia
Ja ad s
ad N Be eru
ru ile
Ar na ay
C ma or
itt l S ya il
an va a
H Ne or
a
C iti
H o
rb as
F. ina es
G me
M ep
an icar lize
h in
. K E Gu raz
To u
u o
g
s al n
ub
g m
ic
EC
LA
a
t b
a
G Ber Ind
liv
d d
zu gu
d ag
Pa gu
Ec ad
om r
. u
hi
d u
r ill
os m
m
ua la
ui
O
o
id
St
in
Tr
the production process may be included. Sweden, in fact, has the highest
unit value for coffee (truncated), but this appears to be due to the fact that
a particular Swedish company selects and brands the beans it reexports.
Second, despite the very high level of disaggregation, there may still be
heterogeneity of production along several dimensions other than the most
obvious (see Khandelwal 2008 and Hallak and Schott 2008). A high price
unaccompanied by substantial sales may be due to other factors that may
not actually reflect quality. Third, within a category, countries may export
a variety of qualities of wine, exploiting different submarkets, and this
may drop the average value far below the “peak” value for country. For
example, New Zealand chose to enter the market at a high price point
and does not export the cheaper varieties that Chile, Argentina, or, for
that matter, France do. Finally, it worth highlighting that it is not clear
that being in “high price” goods is obviously better. As Mukerji and Pana-
gariya (2009) note, the United States produces goods at a huge variety
of quality levels, suggesting that exporting low-quality goods to certain
markets is profitable as well. Acknowledging all of these caveats, the fact
that, on average, rich countries produce higher quality goods than poor
countries does suggest that, in the aggregate, there is a link.
62 does what you export matter?
1.00
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
0
1
Hondura
Slovenia
Bulgaria
Brazil
2
Romania
Macedoni Ecuador
Sier_In
Algeria
Hungary
10
Indones
3
Ethiopia
Jamaica Panama
Moldova
Morocco
China
4
Yugoslav
Chile
Dom_Rep Nicaraga
Turkey
20
Armenia
India
5
Argent
Croatia Mexico
Ukraine
Greece
Uruguay
6
Portugal
Italy
Tunisia Cos_Rica
RED WINE
30
Cyprus
BANANAS
Peru
German
7
Israel
Georgia Guatamala
Spain
Denmark
Poland
S_africa
8
Austral
40
Japan Colombia
Sweden
Nethlds Russia
9
Canada Austria
Switzld Venez, RB
Mali
Sd_arab
50
10
France Mexico
Paragua Bel_lux Peru
New_zeal
Dom_Rep
11
Lebanon Hongkong, SR
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1
1
Burma Guyana
Vietnam Pakistan
Thailand India
6 India Nethlds
Ecuador Nigeria
Guinea Macau
Brazil Oman
Venez,RB Arab_Em
Japan
11
Rwanda China
11
New_Guin Hongkong,SAR
Burundi Mozambq
Salvadr Brazil
Uganda Bngldsh
Peru New_zeal
16
Nicaraga Surinam
Mexico Senegal
Bolivia Malaysia
Dom_Rep Panama
21
Hondura Norway
Iceland
21
China Salvadr
Cameroon Denmark
Guatmala Burma
Nigeria Ukingdom
S_Africa Trinidad
26
Panama Estonia
Italy Guatmala
Colombia Greenld
31
Cos_Rica Canada
Malawi Chile
COFFEE, ARABICA
31
Congo Singapr
SHRIMP AND PRAWNS
Zaire Hondura
Tanzania Nicaraga
St_k_nev Bel_Lux
Ethiopia Colombia
Ecuador
36
Nethlds Phil
Canada Peru
Zimbabwe
41
Venez
Indones Cos_Rica
Kenya Belize
41
Djibouti Austral
German Indones
Singapr Sri_Lka
Switzld Gabon Vietnam
Ukingdom
46
Mritius Madagas
Zambia Thailand
51
0.07
0.75
0.80
0.85
0.90
0.95
1.00
0.60
0.65
0.07
0.75
0.80
0.85
0.90
0.95
1.00
1
1
Panama S_Africa
Mexico
Gabon Lithuani
Ukingdom Venez,RB
3
Ecuador
Colombia N_antil Colombia
11
Peru Bahamas
Poland
5
Trinidad Zaire
Bel_lux Sd_arab
Panama Korea,Rep.
Italy
Russia Canada Arab_em
7
Chile
21
Switzld Denmark
Argent Eq_gnea
Dom_rep Estonia
Russia
9
Spain
Peru Congo
Oman
Brazil
SILVER
FUEL OIL
Cos_rica Ukraine
31
Nethlds
Turkey
11
Mexico Bel_lux
Singapr Portugal
German Yemen_N Egypt,Arab Rep
German
13
Canada France
41
Norway Ghana
Chile Ukingdom
Ivy_cst
Nigeria
15
Sweden
Brazil Finland
Senegal
Ireland
China Tunisia
Indones Cameroon
51
17
Nicaraga Angola Algeria
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1
1
11
Mexico Switzld
11
Denmark
France
26
German Morocco
Israel Hungary Austral
Cyprus Canada German
Ireland Nethlds Austria
Austria
61
Bel_lux France
Figure 5.1c Relative Quality Ladders for Top Latin American Exports and Representative Products
(Quality is measured by export unit values standardized by the 90th percentile of each product) (continued)
Russia
St_k_nev Arab_em
Ireland
MEN'S COTTON SHIRTS MICROPROCESSORS
Seychel
1.0 1.0
Denmark 0 0
0 Japan
Iceland
0.9
Nethlds
0 France
0.9
Austral
Switzld
New_zeal 00
0.8
Portugal
Czechre_P
Slovakia
0.8
Taiwan
Sri_Lka
0.7
0
German 00
New_zeal
Israel
0.7
Canada 0 0
Spain
0.6
Colombia
Canada
Yemen_n
Cos_rica
0 Peru
0 Argent
0.6
Austria
Macau 00
0 S_africa
Macau
Italy
Hongkong,SAR
0.5
0 Nicaraga
Czechre
Malaysia
Ukingdom
China 0 0
Brazil
Zimbabwe
0 Colombia
00 German_E
0 Ecuador
Argent
Moldova
Moroco
0 Slovakia
0.5
Mexico
0 India
Austral 00
Sweden
0.4
Chile
0 Russia
German
0 Vietnam
Israel 00
Kenya
P
Japan
00 Finland
Ivy_cst
0.4
Korea,Rep. Taiwan
Chile 00
0.3
Cos_rica
Mexico 00
Malta
Phil
0
Bolivia 00
Indone
Finland
Cambod 00
Ireland
Guatmala00
France
Slovenia
Brazil 00
0
0.3
Jordon 00
0.2
Pakistan 00
India
Egypt, Arab Rep 0
Bel_lux
Romania 0
Nepal 0
Mritiu_S
Burma
Switzld
Singapr
0.2 0.1
Denmark
Syria 0
Mali
China
0.1 0
Thailand
1 11 21 31 41 51 61 71 81 91 1 11 21 31 41 51
Ukingdom
PASSENGER MOTOR VEHICLES NEW TURBOFAN PLANES
France
German
1.0 1.0
Canada
Sweden
Sweden
0.9
0.9
Brazil
0.8
0.8
Nethlds
0.7
Italy
0.6 0.7
Japan
France
German
Canada
Mexico
0.5
Korea,Rep.
0.6
Ukingdom
0.4
0.5
Israel
0.3
0.2 0.4
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7
VL OV $2.50/PR WOMEN), Men’s Cotton Shirts (Mens’ Cotton Shirts, Knit), Microprocessors (MICROPROCESSORS MONO IC, DIG,
SIL, MOS (ASIC) & (PLA) 32BTS&>), Passenger Motor Vehicles (PASSENGER MOTOR VEHICLE, NESOI, SPARK IGN, 4 CYL, 1500-
3000CC), New Turbofan Planes (NEW TURBOFAN PLANES, NON-MILITARY, >4536 & <=15000 KG).
66 does what you export matter?
For this study, Krishna and Maloney (2011) examine the dynamics of export
unit values, that is, the process of change in quality. This is useful for three rea-
sons. First, their work can be seen as the dynamic analogue to Hummels and
Klenow (2005) and Schott (2004). It is known that the unit values of exports
of rich countries are higher than those from developing economies, but what
are the forces driving this pattern? Second, the analysis of the dynamics of
export unit values permits examining whether some products offer better
prospects for development via improvements in unit values. Third, one can
identify what fraction of aggregate unit value growth is due to goods compo-
sition, and what fraction to country-specific characteristics.
Figure 5.2a shows the rate of growth of unit values and figure 5.2b
disaggregates these values by countries within Latin America. The OECD
shows the highest rate of quality growth, something that, given the higher
export heterogeneity along the quality dimension 67
level of relative quality, indicates that quality is diverging over time. How-
ever, seemingly paradoxically, there is evidence of convergence within
products. That is, the export unit values of countries further from the
quality frontier grow faster than those closer to the frontier (the highest
observed unit value). Thus, the differential position of Latin America and
the Middle East and North Africa regions, relative to the high-income
economies of East Asia (Hong Kong SAR, China; Taiwan, China, Korea,
and Singapore) might explain why they grow more slowly as Hwang sug-
gests. To the degree that these regions are near the frontier in their basket
of largely, natural resource–driven goods, they get much less of a catch-up
convergence “kick.”
However, two other factors of importance emerge from the regression
analysis. First, controlling for products preserves divergence but greatly
reduces the gap in growth rates between the OECD and other regions.
This suggests that what goods countries produce do matter to unit value
growth. Second, controlling for the basket of goods, there is a large posi-
tive, free-standing OECD effect unrelated to position along the quality
ladder. In effect, even if the OECD and developing countries produced
the same goods, the countries of the OECD would grow substantially
faster, even given their closer proximity to the frontier. This is clear from
7%
6%
5%
4%
3%
2%
1%
0%
EC pe
A
A
m D
m P
R
m P
R
EN
EC
co E A
co C
co E A
LA
AF
SA
-O uro
)
e)
e)
e)
D
in OE
on E
in
-in
h-
h-
w
ig
ig
(n
(lo
(h
(h
region
6%
5%
median quality growth
4%
3%
2%
1%
0%
L D
C AC
Be Inina
Ba rmudia
ha da
om ex s
N . R ico
nt p.
itt G Brlles
an y zil
Pa Ne na
Paagu is
Ja nam y
m a
S o a
Ve alv livia
zu C or
C ela ile
N olo ,RB
a ia
ua a a
H temdor
Ar nd ala
Ba entras
C rb ina
R s
Su Hica
amiti
e
d gu ru
ba y
Be go
e
D M ma
ta do
To a
El B aic
G Ecu gu
liz
EC
r v
rin a
ar b
.A e
ne ad
an Uru Pe
s u a
h
d a
h
ic m
g u
i
os a
O
ad
.K
id
St
in
Tr
figure 5.3, which controls for product composition. The only coefficient
that is above average (zero) is that of the high-income OECD countries.
Aside from the Latin America and Caribbean region, the coefficients
for the remaining regions are lower than zero, indicating that, on aver-
age, their growth rate is below average. Clearly, the large fall in the gap
between OECD and lesser developed countries’ growth rates suggests
that the composition of the basket matters to overall growth rates. How-
ever, the fact that there is still a divergence after controlling for them
suggests that country characteristics remain very important, confirming
again that how each good is produced matters greatly.
What factors could influence the rate of growth in unit values? Consistent
with findings from the productivity literature, exposure to international
competition appears to stimulate quality upgrading manifested in rising
export heterogeneity along the quality dimension 69
0.002
0
median quality growth
−0.02
−0.04
−0.06
−0.08
−0.1
−0.12
−0.14
A
EC pe
A
C
m P
R
m D
m P
EN
EC
co E A
LA
AF
co E A
co C
SA
-O uro
)
e)
e)
e)
in O E
D
M
E
in
-in
on
h-
h-
w
ig
ig
(n
(lo
(h
(h
region
unit values. Fernandes and Paunov (2009), using Chilean data, confirm
that firms more exposed to trade have higher product quality. The export
demand effects are similar. Iacovone and Javorcik (2008) find that Mexi-
can plants invest in product quality upgrading before they export.2
The destination market also seems to influence the level of quality. For
the United States, in the aggregate, Waugh (2008) found that export unit
values rise with the income level of the destination market, and Bastos
and Silva (2008) find the same for Portuguese exports.3 These findings are
consistent with different qualities being targeted to distinct submarkets. As
Waugh argues, higher levels of quality allow access to more submarkets.
Overall, the traditional prescriptions of increasing competition and
opportunities to export, especially to wealthier markets, would work in
favor of raising quality.
Krishna and Maloney (2010) attempt to unpack the puzzle of the previ-
ous section: Although within products there is convergence across coun-
tries, without product-specific effects, there is unit value divergence. In
70 does what you export matter?
12.5
Ukingdom
Switzld
Kiribati German
Japan
7.5 Guinea Italy Nethlds
Gilbralt Denmark
France
median quality growth
Sweden Zaire
Surinam Kazakhst N_antil Togo
Canada AUstria
Ireland
SingaprHongkong
Czecho
Guyana Nigeria Israel
AUstral
ALgeria Norway
BahamasSt_k_nevPortugalMexico
Fiji Taiwan Bel_lux
Mritius Bulgaria
Poland
EcuadorDom_rep Brazil Lebanon Finland
EstoniaA Malawi
Cambod
Sia_nesParagua
Ghana Cameroon
UgandaJamaica Sri_lka
Chile
Hungary
Phil Romania China
India Spain
Russia Sudan
Cyprus Croatia Panama
Guatmala
Venez
Haiti Liberia
Colombia
Kenya
Greece New_zeal
Salvadr Korea_s Samoa
2.5 Moldova Tanzania
BngldshLithuani
ARgent Czechrep
Iceland
Yugoslav
Benin GreenldAZerbaij Cos_rica
Indones
Hondura
BoliviaThailandKuwait Zambia
Uruguay
Peru
Malta
Us_nes
Slovakia Malaysia
Bosnia-h
Morocco
LaoBarbado
Egypt
Belize
Oman
Trinidad Macau
Macedoni Zimbabwe
Nicaraga S_africa
Slovenia Sd_arabMongola
Yemen_n Nepal ARab_em Turkey
Bahrain Armenia Sier_ln
Madagas
New_guin
Pakistan
Syria Ivy_cst
Ukraine
Jordon Burma
QatarSenegalTunisia
Ethiopia
UzbekistNew_cale
Congo Mozambq
Turkmeni Vietnam
Mauritn ALbania
Belarus
Burundi
Latvia
−2.5 Tajikist GuadlpeRwanda
USSR
Gabon
Mali Georgia
Niger
−7.5
0.05 0.15 0.25 0.35 0.45 0.55
The preceding analysis has not taken into account the dynamic nature
of the composition of regional export baskets. Quality may also increase
with the introduction of new goods of higher relative quality. This section
includes an analysis of the entry and exit patterns of goods in regional
export baskets. “Entries” are composed of goods not traded from 1990
to 1995, and traded at least three times from 1996 to 2001. “Exits” are
composed of goods traded at least three times from 1990 to 1995, and not
traded from 1996 to 2001. “Incumbent” goods are goods traded at least
three times in the 1990 to 95 period.
The median ratio of incoming to incumbent goods for the non-OECD
regions is 1.03, and the median ratio for the OECD (high-income) coun-
tries is equal to 1.06. This implies that new goods enter at approximately
the same level as existing goods within the region, but also that new
goods enter at higher quality levels in richer countries than in poorer
ones. At the upper end of the distribution, the quality ratio of entering
to incumbent goods is larger for the non-OECD regions, suggesting a
degree of convergence at the top end of the quality distribution between
OECD and non-OECD countries.
Figure 5.5 disaggregates this by region, presenting the ratio at the
25th, 50th (median), and 75th percentiles. As compared with the previous
results where there are roughly equal ratios of OECD (high-income) and
non-OECD at the 50th percentile, important regional differences emerge.
Compared to the OECD, Central Asia and the East Asia Pacific regions
are 10 to 20 percent lower at the median, while Eastern Europe and South
Asia are roughly 10 percent higher. The Latin America and the Caribbean
region, the Middle East and North Africa region, and Sub-Saharan Africa
are all roughly similar to the OECD at the 50th percentile level.
These counterintuitive findings are perhaps somewhat allayed by what
is happening at the upper end of the distribution. Although both East
Asia categories have lower median ratios, at the 75th quantile East Asia
(high-income) equals that of the top performers. East Asia (low-income) is
respectable as well. Arguably, the faster growing areas may have a broader
distribution of exporters coming on line, many reflecting their relatively
low average level of “caliber,” but some being global superstars what are
“leapfrogging.” Eastern Europe stands out as having its entire distribution
shifted right, with quantiles showing ratios 30–50 percent higher than the
OECD. Since the period covered begins after the fall of the Berlin Wall
in 1989, this may suggest that the pre-liberalization level of “caliber” or
general technological sophistication could very broadly support goods of
higher quality and some superstars, and that liberalization made this pos-
sible. Central Asia is largely shifted, or at least compressed left, and Africa
is of a similar character, with modest median growth and no superstars.
74 does what you export matter?
1.75
1.50
incoming ratio
1.25
1.00
0.75
0.50
0.25
A
EC pe
A
m D
m P
R
m P
EN
EC
LA
co E A
co C
co E A
SA
AF
-O uro
e)
e)
e)
D
in E
M
h- O
on E
-in
in
h-
w
ig
ig
(n
(lo
(h
(h
region
show a higher price in the year of entry. It also takes longer for them to
converge to the incumbent prices, whereas in the case of homogeneous
goods, the new exporters enter with a higher price but rapidly converge to
the price of the incumbents.
Conclusion
This chapter has examined export quality through the lens of the unit
value. The literature documents a high degree of heterogeneity in prices
even within very finely disaggregated goods. On average, this measure of
“quality” rises with level of development. Therefore, the dynamics of unit
values offers a window on broader development issues.
The chapter finds that there is a convergence dynamic, that is, within a
good, countries further from the frontier will, all things being equal, expe-
rience faster growth rates of their export unit values. In this sense, Latin
America, for instance, is at a bit of a disadvantage in its concentration
in commodities, which tend to have shorter quality (unit value) ladders.
Further, there is evidence that goods matter. Many manufactures appear
to offer greater opportunities for investments that will yield more rapid
growth in quality.
In this context, two observations can be made. First, what is not clear
is that there is a market failure that would dictate that the Latin American
and Caribbean economies should be specialized in goods that defy their
comparative advantage in commodities. Lower possibilities for unit value
growth may translate into lower profitability, but there is no obvious exter-
nality that the market cannot see and which must be corrected. Moving into
other noncommodity goods against a country’s comparative advantage will
likely involve welfare losses.
Second, and critically related to the previous point, the findings indi-
cate that the convergence effects are swamped by idiosyncratic regional
factors which make the OECD (high-income) continue to grow rela-
tively faster, even controlling for goods. Even if the Latin America and
the Caribbean region had the OECD basket of goods, it would perform
much worse. This again points to the “how” and not just the “what”
of export decisions, although as of now one can only speculate on
the factors undermining the region’s performance. However, numerous
behind-the-border factors, such as the resolution of market failures in
technology, the depth of financial markets, and the quality of institu-
tions, appear to be relevant. Further, the literature has put an emphasis
on human capital accumulation generally as a key promoter of quality.6
These factors not only affect the quality growth within a good category,
but also are likly to affect what goods are produced, and, in particular,
whether a country produces those with the greatest potential for quality
growth. Hence, it seems that national sector-neutral policies can help
76 does what you export matter?
Notes
1. See Brooks (2006), Hallak and Schott (2008), Kugler and Verhoogen (2008)
and Khandelwal (2010) have argued that additional information on the relative
demand for products needs to be incorporated to make true quality comparisons.
For purposes here, the assumption is that, on average, the raw unit values capture
differences in quality, albeit with measurement error.
2. This is consistent with Bustos (2010) who added a measure of technologi-
cal choice to the Melitz framework. Bustos found that, for Argentina, reduction of
import tariffs by its Mercado Común del Sur (MERCOSUR) or Common Southern
Market partners increased both the probability of firm entry into these export
markets and spending on technology.
3. Relatedly, Brambilla, Lederman, and Porto (2012) find that Argentine man-
ufacturing firms paid higher average wages and hired more skilled workers upon
shifting their exports from Brazil to high-income markets (the United States and
Europe) during the Brazilian devaluation of 1999. These changes in skill utilization
by firms were associated with exports with higher unit value variances than other
exports.
4. Do and Levchenko (2007) also postulate a model in which financial ser-
vices are endogenous and countries producing low-finance-intensive goods will
have financial markets that cannot support taking on more risky goods. A related
literature is reviewed, and the empirical validity of the Acemoglu and Zilibotti
(1997) theory is in chapter 7 of this book.
5. The association of increasingly complex or involved products suggests that
the diversification channel need not be the only financial barrier, and that barriers
need not, in fact, originate in the financial sector. Bardan and Kletzer (1987) argue
that more sophisticated manufactured finished products require more credit to
cover selling and distribution costs than primary or intermediate products. There-
fore, imperfections in credit markets, even where technology and endowments are
identical, can lead to specialization of countries with higher levels of sovereign risk
or imperfect domestic credit markets in less sophisticated products. Beck (2002)
builds a model in which manufacturing, due to exhibiting increasing returns to
scale, is more finance intensive due to increasing returns to scale.
6. In a paper commissioned for this study, Waugh (2008) offers a general
equilibrium theory of the supply and demand for product quality, and international
trade that sees quality as an important feature to understanding bilateral trade
volumes. He argues that intermediate goods are available in different quality and
this quality is complementary to domestic human capital: Skilled workers are better
able to use higher quality intermediate goods. Higher human capital countries are
able to produce all levels of quality for export, whereas poor countries can produce
only those at the lower end of the spectrum. Introducing quality in his simulations,
he is able to replicate up to 75 percent of the observed variation in the volume of
bilateral trade compared to the model, with no quality considerations.
6
Heterogeneity in the
Production of Goods
In some sense, much of the economic literature sees the process of devel-
opment as the progressively more efficient production of the archetypal
widget. Hence, the notion that a good can be produced in many different
ways is not at all alien. But its importance for the present discussion can-
not be overstated.
The Chilean historical experience with exporting copper is illustrative.
In 1870, Chile was the world’s largest producer. Yet by 1904, output had
fallen in absolute terms to the point where there was a question about
whether the industry could survive. As occurred in Mexico, the mines were
then largely bought by foreigners, particularly from the United States, and
Chilean participation became negligible. Today, fortunes have reversed
and the Chilean national company, the National Copper Corporation of
Chile (CODELCO), is a major global player. This reversal cannot be a
function of the good because copper has not changed that much. However,
it has everything to do with the Chilean ability to bring new technologies
77
78 does what you export matter?
into production at different historical periods. At the turn of the 20th cen-
tury, Chile lacked the innovative capacity to employ new technologies and,
in particular, the Bessemer process and new applications of electrolysis.
By contrast, Wright uses the same U.S. experience with copper as an
example of how nations learn: It developed a knowledge industry with
a network of expertise which, in addition to allowing the Americans to
profitably take over Chilean copper production, also laid the foundation
for expansion into other fields of engineering and manufacturing. It was
the same product but produced with a completely different outcome (see
Maloney 2007). Hence, in Natural Resources: Neither Curse nor Destiny,
Lederman and Maloney (2007) argue that the huge historical variation
in development experience based on natural resources suggests that it
is more fruitful to investigate the variation than the “average” tendency
and notions of a “conditional curse” move in this direction. How coun-
tries leverage their resources is as important as the resource endowments
themselves.
This lesson extends far beyond the natural resource issue to all types of
goods. The American and Chilean examples with copper suggest that
virtually identical goods can be produced at very different levels of sophis-
tication and with very distinct long-run impacts on growth. There is sug-
gestive evidence that this is also true today within goods often thought
to be “high-tech” manufactured goods. These are often thought to be
knowledge industries with the potential for spillovers. However, again,
whether or not such spillovers occur depends importantly on how these
goods are produced.
As a crude measure of whether such high-tech goods actually lead to
knowledge generation and learning, is revealing to consider the degree of
patenting. Figure 6.1 shows the Index of Revealed Comparative Advan-
tage in Innovation (IRCAI) in the production of Mexican computers and
Brazilian aircraft (Lederman and Maloney 2006). This is the number of
Mexican patents given in a particular sector over the total number of
Mexican patents, divided by the global analogue. If Mexico, with a large
computer sector, is producing relatively more patents in this sector than
is the case globally, the IRCAI >1, then it has an innovation comparative
advantage in the sector.
However, this is not the case for Mexico. For a 20-year period, Mexico
has shown a comparative disadvantage (less than 1) in innovation (as
measured by patents) in the sector. Where this country trades more is not
where it innovates more. A similar finding emerges for aircraft in Brazil.
In neither case is it easy to argue that any knowledge cluster in these
heterogeneity in the production of goods 79
1.2
RCA index in SIC 372 (18-yr MA)
1.0
0.8
0.6
0.4
0.2
0
94
95
96
97
98
99
00
80
81
82
83
84
85
86
87
88
89
90
91
92
93
19
19
19
19
20
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
Source: Lederman and Maloney (2005, 2006), based on data from the U.S.
Patent and Trademark Office and its Standard Industrial Classification (SIC) of
patents of 2000.
80 does what you export matter?
USA
UKR
FRA
ISR BRABHS RUS
BOLBEL CAN GBR ESP
CHL
MEX ITA SWE AUT DEU
PRT POL CZE
NLD ROM
URYMYS
TUR DNK CHE
HUN GRC PER SVK
GIN BGR ISL AUS CYP
ZWE NZL
KOR
IRL THA ARG
ZAF JPN
RCA in trade
PHL
LUX
FIN HRV KWT
PRYHKG
IDN CHN ECU
IND VEN
TWN
MAR COL
SAU
VNM
PAN
−6.906877
IRN
0 1 4.902237
RCA in patents
0 1.44528
FIN
ISR KOR
SWE JPN
ATG
GHA HUN
MEX MYS
DOM
DEU CAN GBR
FRA
THA CHN TWN
LUX USA IRL PHL
BMU HRVDNK BEL
SVN THA HKG NLD
CHE CYP
BRA
LBN ROMGRC CZEITA
UKR
NOR CRI
AUS JOR
SVK
BHS PRT
IDN
NZL
BRB BLR KNA
POL
ZAF LTU GEO KGZ
RCA in trade
BGRTUR MDAARM
IND UGA MAC
TTO PAK LKA
CUB HNDSLV RUS
MMR
KEN
GTM ARG BOL EGY
GHA
MAR SAU
CHL PERPAN
IRN
VEN ECU KWT
COL
ISL
NGA
−7.339043
DZA SYR
0 1 2.612473
RCA in patents
world LAC
Vollrath 1991) and patents. The latter varies between zero and infinity.
The vertical and horizontal lines cutting through these graphs indicate
the level of comparative advantage. For example, a value greater than
zero in trade indicates comparative advantage in trade, and a value
greater than one in patents suggests comparative advantage in patenting.
heterogeneity in the production of goods 81
CRI
IDN TWN
PHLMYSCHN THAKOR
HUNMEX JPNIRL
CZEHKG GBR
LKA USA
0
SVNHRVNZL
POLZAF MDA
BOL
RUS TUR
ARG MAR
JAM
CHL UGA
GTM
COL PER EGY PAK
SAUISL
KWT
VEN
IRN
DZA
−7.950758
SYR
NGA
0 1 3.605023
RCA in patents
world LAC
The How
In the cases of both Chilean copper and the aircraft, electronics, and
computer industries, there is evidence that goods can be produced in very
different ways and with very different results. Indeed, the literature sug-
gests that the phenomenon is more general. Blomström and Kokko (2007)
demonstrate the sophistication of the Swedish forestry industry to be far
higher than that of Chile or Brazil. Wright and Czelusta (2007) discuss
“mineral underperformers” to describe Latin America’s low-productivity,
low-exploration mining enterprises. At a very aggregate level, Martin and
Mitra (2001) also find by using simple total factor productivity regres-
sions that differential rates of productivity growth exist within manufac-
tures and agriculture by country. Developing countries, on average, show
lower rates of growth in both manufactures and agriculture.
Combined with the findings of dramatic variance in the quality from
the previous chapter, two important conclusions can be drawn and one
open policy question can be raised. First, both the quality of very finely
disaggregated goods and the way in which even very homogeneous goods
are produced can vary greatly according to country context. Second, the
fact that a country produces a particular good does not guarantee that
whatever positive things may be associated with that good will appear
in that context. Countries do not automatically become the best of what
they produce.
The open policy question, then, is how can the development impact
of whatever basket is produced be maximized (including leveraging it
into new products)? What are the complementary actions that countries
need to take, and might these be more important than what is actually
produced?
As noted previously, Wright documents how the United States lever-
aged copper extraction into a major knowledge network by incorporating
higher level human capital, universities, and private sector firms that both
led to the invention of new techniques for copper processing and laid the
foundation for moving into new industries. Blomström and Kokko (2007)
argue the same for the Nordic forestry industries. Table 6.1 shows just
heterogeneity in the production of goods 83
one set of institutions involved in the pulp and paper cluster in Sweden.
The agglomeration of high-level research in states and institutions for the
development of human capital is far beyond what would be found in Chile
or Brazil today.
These findings also explain how a forestry company in Finland with
a cellulose mill at the town of Nokia would become a major telecom-
munications giant, or how a Finnish copper company would become
Outokumpu, one of the world’s largest producers of fine tolerance stain-
less steel products—while neither of these transformations occurred in
Latin America. Forestry and copper were cultivated as very knowledge-
intensive industries such that, to use Wright’s phrase, the companies
Goods or Tasks?
Each import of a finished iPod into the United States contributes roughly
US$150 to the China–U.S. bilateral trade deficit. Frequently, this kind of
product is considered a desirable “high-tech” product with likely high
knowledge spillovers.
Yet in this era of extraordinarily fragmented production processes
and extended value chains, how much really accrues to China? Following
the value added requires information not easily available, but Linden,
Kraemer, and Dedrick (2009) conclude that it is not very much.
Of the US$299 retail price in the United States, US$163 was captured
by U.S. companies and workers; US$75 went for distribution and retail;
US$80 went to Apple for its invention and overall coordination of pro-
duction; and US$8 went to various component makers. Japan earned
about US$26 through the Toshiba disk drive. In the end, only a handful
of dollars or just above 1 percent of the value added of the iPod accrued
to China’s labor for what is largely assembly work.
Source: Linden, Kraemer, and Dedrick (2009). See also Kraemer, Linden, and
Dedrick (2011) on the iPad and iPhone.
even among Chinese firms drives the margins down to the barest minimum.
As noted by Steinfeld (2004), the emphasis on standardized, non differenti-
ated products offers little alternative but to compete on the basis of low cost
and high volume. Moreover, firms continually run the risk of being unseated
by the next low-cost entrant, locking firms into a mutually destructive price
competition.
Clearly, high-tech exports in this context are not “high value added”
in any sense of the word. Figure 6.3, at a very aggregate level, shows that
electronic products have only 22 percent value added while some other
low PRODY goods, such as coking, furniture, or chemical products, have
much higher. (Interestingly, any positive slope is driven by natural resource
products exported by rich countries, such as nonmetallic mineral prod-
ucts.) Disaggregating further, table 6.2 shows that electronics and other
“high-tech” products are at the bottom of the domestic value-added list.
Computers contribute only 4.6 percent. At the high end are a variety of
natural resource and industrial “basics” such as chemical fertilizers, metal
processing, and the like. The development impact of a unit gain in pro-
ductivity is going to be larger in these goods. Hence, when the literature
talks about “moving into high value added products,” it needs to be more
precise. For China this would imply moving out of computers and into
hemp textiles.
The findings indicate that the focus should be on moving up to higher
value-added tasks. But again, there is little evidence that there is an auto-
matic progression up the value chain or toward greater value added along
other dimensions. As Breznitz and Murphree (2011) argue, even for domes-
tically completed goods for domestic consumption, after licenses and fees
for the use for foreign technology, China can feel trapped at the lowest value
added of final state assemblers. Much of the present discussion around
the Chinese drive for indigenous innovation and indifferent enforcement
of intellectual property rights relates precisely to China’s goal of moving
from a model of “Made in China” to “Innovated in China” (Segal 2010;
Breznitz and Murphrees 2011).
Thus, from a conceptual point of view, the final export product becomes
very misleading as a measure of desirability of the activity actually being
undertaken. As Steinfeld (2004, 1972–1973) notes in his work on China’s
“shallow” integration into the global production chain:
5,000
20 30 40 50 60 70 80 90 100
value added
Source: Authors’ calculations based on domestic value-added estimates from Koopmans, Wang, and Wei (2008).
87
88 does what you export matter?
Notes
91
92 does what you export matter?
Development as Diversification
The seminal article by Imbs and Wacziarg (2003) analyzes the process of
diversification across income levels. The data on production and employ-
ment concentration across countries gathered by these authors suggest a
robust pattern whereby economic diversification increases with the level
of development, until reaching a relatively high level of GDP per capita,
after which time economies become increasingly specialized. This finding
is provocative as it contradicts a basic tenet of neoclassical trade models,
which predict that specialization produces improvements in economic
efficiency and ultimately development.
Klinger and Lederman (2004, 2006, 2011) were among the first to
study the empirical relationship between diversification, export-product
discoveries, and the level of development. Unlike the exponentially posi-
tive trajectory of patenting activity, export discoveries tend to peak at a
low level of income per capita, and fall monotonically with development
thereafter. The Imbs-Wacziarg U-shape function of concentration is appar-
ent in the trade data; diversification peaks around $20,000 purchasing
power parity (PPP) and declines thereafter.1
Numerous models in the literature suggest that market failures inhibit the
discovery process, thereby constraining diversification and possibly devel-
opment. In the words of Harrison and Rodríguez-Clare (2010, 4041),
“… just as research and development subsidies are appropriate responses
to innovation spillovers, policies to promote entry into new industries are
appropriate to deal with information spillovers associated with the discov-
ery of new profitable activities.”
trade quality as portfolio diversification 93
Lederman and Maloney (2007) point out that the curse of natural resources
could be a myth. Of particular relevance is their finding that the most
robustly negatively correlated indicator of natural resource exports with
trade quality as portfolio diversification 95
A first look at the data is not conclusive, especially for the sample
of Latin American and Caribbean economies. Table 7.1 contains the
necessary descriptive statistics. It shows volatility indicators, namely,
the standard deviation of the annual growth of each variable from 1980
to 2005. It also shows the average external prices (export, import, and
terms-of-trade indexes, which are weighted averages of unit values of
exports and imports). In addition, it contains volatility indicators for real
(PPP adjusted) GDP and consumption per capita, as well as for various
potential covariates of macroeconomic volatility, including trade open-
ness, financial development, and rule of law (as a proxy of the quality of
public institutions).
Table 7.1 Natural Resources, Macroeconomic Volatility, and Trade Concentration, 1980–2005
96
Latin American
and Caribbean net Other net
Latin America and exporters of energy exporters of
Global sample Caribbean (LAC) and mining energy and mining
Obs. Mean Obs. Mean Obs. Mean Obs. Mean
Volatility Indicators
Export price index 139 0.132 19 0.133 15 0.122 59 0.151
Import price index 139 0.113 19 0.112 15 0.098 59 0.117
Terms of trade 139 0.091 19 0.094 15 0.088 59 0.110
GDP per capita 155 0.059 21 0.043 14 0.040 70 0.071
Other Variables
Net exports of mining per worker 150 0.375 21 0.091 15 0.171 63 1.200
Net exports of agriculture per worker 151 0.087 22 0.172 15 0.263 63 0.412
Export concentration 153 0.321 22 0.313 15 0.297 65 0.405
Import concentration 155 0.137 22 0.128 15 0.117 67 0.134
Trade over GDP in 1980 109 63.634 20 52.066 15 48.797 50 68.611
Private credit over GDP in 1980 107 30.979 21 27.528 15 24.733 47 25.720
Rule of law index in 1982–84 118 3.072 22 2.359 15 2.493 54 2.994
Source: Authors’ calculations based on data from the World Bank, International Monetary Fund, and United Nations Commercial and Trade
Statistics (UNCOMTRADE).
Notes: GDP = gross domestic product; Obs = number of observations. Export and import concentration are measured by the root of the
Herfindahl index.
trade quality as portfolio diversification 97
The first two columns contain the data from the global sample, followed
by the data from Latin America and the Caribbean. The last six columns
show the corresponding statistics for the sample of Latin American and
Caribbean and non-Latin-American net exporters of mining and energy
commodities. The data show that all Latin American and Caribbean net
exporters of mining and energy were also net exporters of agricultural
commodities during 1980–2005.
The data do not support the view that the Latin America and the
Caribbean region or its exporters of natural resources suffered from
unusually high external volatility. In contrast, non-Latin-American net
exporters of mining and energy did, in fact, face higher export and
import price volatility than the global and Latin American samples. This
group of countries also had export structures that were significantly
more concentrated. Partly because the descriptive statistics are incon-
clusive, it is necessary to assess the validity of the hypotheses linking
natural resource dependence and volatility with multivariate economet-
ric estimations.
Figure 7.1 illustrates the relationship between merchandise export-
revenue concentration and terms-of-trade volatility. The positive
IRN
0.3
RWA
terms-of-trade volatility
NIC AZE
SDN ZMB
0.2 GNB
BIH CMR
UGA COM
GEO TJK
CIV
BGD KAZ
SYR
YUG RUS
TZA MWI GHA MRT
UKR CPV
KGZ ETH
IND PAK VNMMDGTGOECUGMB MOZ BEN
0.1 PER SLV
EGY ARM
NAM BFA
JORIDNKEN
BLRLVA BOL
DOM
PNG CAF
BRAARG LKA SEN
ZWE CHL
MEX ERI
CZE JPN COL
ALBCRI LSO
GUY MLI
BGR GTM HND BWA
ESP
THA URYPHL
ZAF GIN
TUR
ROM SVK
MKDMAR
MYS MDA PRY
CHN
ITA KOR
LTUNZLPOL MUS
DEU
CHE
USA
FRA PRT LBN
GRCFIN
SVN
SWE
GBRESTIRL TUNISR PAN KHM
HUN SWZ
DNK
AUTHKG BEL
HRV
NLD
0 BTNDJI TKM YEM
above, would this policy also succeed in diversifying the export portfolio
of a small, underdeveloped economy with abundant natural resources?
There is no existing evidence or theoretical framework that could explain
how the emergence of one big hit affects other existing exports and the
prices of their required factors of production.
The following chapter returns to the policy challenges and trade-offs
that perhaps should be kept in mind in the pursuit of pro-diversification
policies.
Notes
Conclusion and
Policy Reflections
Does what a country exports matter? The answer can be broken down
into several parts, although the bottom line is that “how you export mat-
ters more.”
First, what a country exports probably does matter. Externalities and
rents exist, and there is no reason to believe that they are associated with all
goods equally. In the former case, there is clearly an argument for interven-
tions to encourage such goods more than the market would naturally do.
Second, the literature still offers us no confident policy guidance on
what those goods might be. Measurement of externalities is notoriously
difficult, and this study has argued that the shortcuts offered to get around
this measurement issue—showing that goods thought a priori to have
externalities that positively affect growth—have proven weak pillars
for policy makers to rely on. The advice to stay out of natural resource
industries and to get into high-tech industries, those which rich countries
already produce, or those offering potential to enter new industries, either
does not prove robust empirically or, where no empirical test is offered,
raises substantial conceptual concerns. Perhaps the market is missing good
opportunities, but it is not apparent at this point that government (or we)
can see them any more clearly.
Third, the policy debate needs to focus far more on the vast hetero-
geneity of experiences within any given sector. Although the study has
shown that there is no robust evidence of a resource curse, there defi-
nitely are countries that appear to have suffered as a result of having
resources, just as there are tremendous success stories, including much
of the industrialized world today. The production technologies used to
produce a good, and the “knowledge intensity” of that process, can vary
greatly across countries. The returns to skilled labor vary as much across
countries as across goods. The range of quality within much disaggregated
categories of goods is so large that some have argued that understanding
105
106 does what you export matter?
109
110 bibliography
Brambilla, Irene, Daniel Lederman and Guido Porto. 2012. “Exports, Export Des-
tinations and Skills.” American Economic Review, forthcoming. Draft version
published as National Bureau for Economic Research Working Paper 15995,
Cambridge, Massachusetts.
Brambilla, Irene, Rafael Dix-Carneiro, Daniel Lederman, and Guido Porto. 2011.
“Exports, Skills and the Wages of Seven Million Latin American Workers.”
The World Bank Economic Review. Draft version published as “Exports, Skills
and the Wages of Five Million Latin American Workers.” World Bank Policy
Research Working Paper 5246. World Bank, Washington, DC.
Bravo-Ortega, Claudio, and José de Gregorio. 2007. “The Relative Richness of the
Poor: Natural Resources, Human Capital, and Economic Growth.” In Natural
Resources: Neither Curse nor Destiny, edited by Daniel Lederman and William
F. Maloney, chapter 6. Stanford University Press.
Breznitz, Dan and Michael Murphree. 2011. The Run of the Red Queen Yale
University Press.
Bravo-Ortega, Claudio, and José de Gregorio. 2007. “The Relative Richness of the
Poor? Natural Resources, Human Capital, and Economic Growth.” In Natural
Resources: Neither Curse nor Destiny, edited by Daniel Lederman and William
F. Maloney, chapter 4. Stanford University Press.
Broda, C., and J. Romalis. 2003. “Identifying the Relationship between Trade
and Exchange Rate Volatility.” Technical Report, Chicago Graduate School of
Business, Chicago, IL
Brooks, E. 2006. Why don’t firms export more? Product Quality and Colombian
Plants” Journal of Development Economics 80:160–178.
Brunnschweiler, Christina N. 2008. “Cursing the Blessings? Natural Resource
Abundance, Institutions, and Economic Growth.” World Development 36(3):
399–419.
Bustos, Paula. 2010. “Trade Liberalization, Exports and Technology Upgrading:
Evidence on the Impact of MERCOSUR on Argentinean Firms.” American
Economic Review, forthcoming.
Caballero, R.J. and G. Lorenzoni. 2007. “Persistent Appreciations and Overshoot-
ing: A Normative Analysis.” Technical Report, National Bureau for Economic
Research Working Paper 2007, Cambridge, MA.
Cadot, Olivier, Celine Carrere, and Vanessa Strauss-Kahn. 2011. “Export Diver-
sification: What’s Behind the Hump?” The Review of Economics and Statistics
93(2): 590–605.
Card, David. 1999. “The Causal Effect of Education on Earnings,’’ in Ashen-
felter Orley and David Card, editors, Handbook of Labor Economics, Vol. 3:
1801–1863. Elsevier Science B.V.
Cashin, P., L.F. Céspedes, and R. Sahay, 2004. “Commodity Currencies and the
Real Exchange Rate.” Journal of Development Economics 75 (1): 239–268.
Chen, Y. and K. Rogoff. 2003. “Commodity Currencies.” Journal of International
Economics 60 (1): 133–160.
Ciccone, Antonio, and Giovanni Peri. 2006. “Identifying Human Capital External-
ities: Theory with Applications.” Review of Economic Studies 73: 381–412.
Ciccone, Antonio, and Marek Jarocinski. 2010. “Determinants of Economic
Growth: Will Data Tell?” American Economic Journal Macroeconomics,
forthcoming.
112 bibliography
Ciccone, Antonio, Federico Cingano, and Piero Cipollone. 2004. “The Private
and Social Return to Schooling in Italy.” Giornale degli Economisti e Annali di
Economia 63(3-4): 413–444. December.
Clark, P., N. Tamirisa, S.J. Wei, A. Sadikov, and L. Zeng. 2004. “Exchange rate
volatility and trade flows-some new evidence.” Technical Report,Washington,
DC: International Monetary Fund.
Clark, P.B. 1973. “Uncertainty, Exchange Rate Risk, and the Level of International
Trade.” Economic Inquiry 11 (3): 302–313.
Corden, W. Max, and J. Peter Neary. 1982. “Booming Sector and De-Industrialization
in a Small Open Economy.” The Economic Journal 92(368): 825–48.
Cuddington, John T., Rodney Ludema, and Shamila A. Jayasuriya. 2007.
“Prebisch-Singer Redux.” In: Natural Resources: Neither Curse nor Destiny,
edited by Daniel Lederman and William F. Maloney, chapter 5. Stanford
University Press.
Czelusta, Jesse W. 2001. “Natural Resources, Economic Growth, and Technical
Change: Lessons from Australia and the United States.” Stanford University.
Davis, Graham. 1995. “Learning to Love the Dutch Disease: Evidence from the
Mineral Economies.” World Development 23(10): 1765–79.
De Long, Bradford, and Lawrence Summers. 1992. “Equipment Investment and
Economic Growth.” Quarterly Journal of Economics 106(2): 455–502.
Do, Q. and A. Levchenko. 2007. “Comparative Advantage, Demand for external
Finance, and Financial Development.” Journal of Financial Economics 86:
796–834.
Dollar, David, and Aart Kraay. 2003. “Institutions, Trade, and Growth.” Journal
of Monetary Economics 50(1): 133–62.
Dunning, Thad. 2008a. Crude Democracy: Natural Resource Wealth and Political
Regimes. Cambridge Studies in Comparative Politics. Cambridge University
Press.
———. 2008b. “Model Specification in Instrumental-Variables Regression.” Political
Analysis 16(3): 290–302.
Dunning, Thad. 2005. “Resource Dependence, Economic Performance, and Politi-
cal Stability.” Journal of Conflict Resolution 49(4): 451–82.
Durlauf, Steven N., Andros Kourtellos, and Artur Minkin. 2001. “The Local
Solow Growth Model.” European Economic Review 45(4–6): 928–40.
Easterly, W., A. Reshef, and J. Schwenkenberg. 2009. “The Power of Exports.” World
Bank Policy Research Working Paper 5081. World Bank, Washington, DC.
Easterly, William, and Ross Levine. 2002 “Tropics, Germs, and Crops: How
Endowments Influence Economic Development.” University of Minnesota.
Eaton, Jonathan, Marcela Eslava, Maurice Kugler, and James Tybout. 2007.
“Export Dynamics in Colombia: Firm-Level Evidence.” NBER Working Paper
13351, National Bureau of Economic Research, Inc., Cambridge, MA.
Estevadeordal, Antoni, and Alan M. Taylor. 2002. “A Century of Missing Trade?”
American Economic Review 92(1): 383–93.
Feenstra, Robert C. “US Imports, 1972-1994: Data and Concordances.” NBER
Working Paper Series. Working Paper 5515. Cambridge, MA. 1996.
Fernandes, A. M and C. Paunov. 2009. “Does Tougher Import Competition foster
Product Quality Upgrading? PRWP 4894, World Bank, Washington, DC.
bibliography 113
Lederman, Daniel, and William F. Maloney, eds. 2007a. Natural Resources: Nei-
ther Curse nor Destiny. Stanford University Press.
Lederman, Daniel, and William F. Maloney. 2007b. “Trade Structure and Growth.”
In: Natural Resources: Neither Curse nor Destiny. Stanford University Press.
———. 2007c. “Trade Structure and Growth.” Chapter 2 in Natural Resources,
Neither Curse nor Destiny, edited by D. Lederman and W. Maloney. Washing-
ton, DC: World Bank and Stanford University Press.
———. 2006. “Innovation in Mexico: NAFTA Is Not Enough.” Chapter 13 in
Global Integration and Technology Transfer, edited by B. Hoekman and B.
Smarzynska Javorcik. World Bank, Washington, DC and Palgrave Macmillan.
Lederman, Daniel, Andrés Rodríguez-Clare, and Daniel Xu. 2011. “Entrepreneur-
ship and the Extensive Margin in Export Growth: A Microeconomic Account-
ing of Costa Rica’s Export Growth, 1997–2007.” The World Bank Economic
Review 25(3): 543–61.
Lee, Donghoon, and Kenneth I. Wolpin. 2006. “Intersectoral Labor Mobility and
the Growth of the Service Sector.” Econometrica 74(1): 1–46.
Levchenko, A. A. 2007. Institutional quality and international Trade Review of
Economic Studies 74:791–819.
Levine, Ross, and David Renelt. 1992. “A Sensitivity Analysis of Cross-Country
Growth Regressions.” American Economic Review 82(4): 942–63.
Linden, Greg, Kenneth L. Kraemer, and Jason Dedrick. 2009. “Who Captures
Value in a Global Innovation Network? The Case of Apple’s iPod.” Communi-
cations of the ACM 52(3): 140–144.
Maddison, Angus. 1994. “Explaining the Economic Performance of Nations,
1820–1989.” In Convergence of Productivity, edited by William J. Baumol,
Richard R. Nelson, and Edward N. Wolff. Oxford University Press.
Maloney, William F. 2007. “Missed Opportunities: Innovation and Resource-
Based Growth in Latin America.” In Natural Resources: Neither Curse nor
Destiny, edited by Daniel Lederman and William F. Maloney. Stanford Univer-
sity Press.
Mankiw, Gregory, David Romer, and David Weil. 1992. “A Contribution to the
Empirics of Economic Growth.” Quarterly Journal of Economics 107(2):
407–37.
Manzano, Osmel, and Roberto Rigobon. 2007. “Resource Curse or Debt Over-
hang?” In Natural Resources: Neither Curse nor Destiny, edited by Daniel
Lederman and William F. Maloney. Stanford University Press.
Marshall, Monty G., and Keith Jaggers. 2002. “Polity IV Project: Political Regime
Characteristics and Transitions, 1800–2002.” University of Maryland, College
Park, Center for International Development and Conflict Management.
Martin, Will, and Devashish Mitra. 2001. “Productivity Growth and Convergence
in Agriculture and Manufacturing.” Economic Development and Cultural
Change 49(2): 403–22.
Mayer, W. 1984. “The infant-export industry argument.” 17(2): 249–269.
McEvedy, Colin, and Richard Jones. 1978. Atlas of World Population History.
New York: Facts on File.
Mehlum, Halvor, Karl Moene, and Ragnar Torvik. 2006. “Institutions and the
Resource Curse.” Economic Journal 116(508): 1–20.
Moreno, María Antonio, and Cameron A. Shelton. forthcoming. “Sleeping in the
Bed One Makes: The Venezuelan Fiscal Policy Response to the Oil Boom.” In
bibliography 117
Boxes, figures, notes, and tables are indicated by b, f, n, and t following the page
number.
121
122 index
P Q
Pack, H., 2, 4, 26 quality. See heterogeneity and
Panagariya, A., 10n4, 58, 61 quality dimensions; trade
Panama quality and portfolio
education in, 38 diversification
wage premiums in, 36, 38b quality ladders, 7, 58–60, 67, 75
paper industries, 83, 83t
Paraguay
R
skilled workers in, 40
wage premiums and, 36, 38b Raddatz, Claudio, 100–102
patents, 78–82, 89n2, 106 Rajan, Raghuram, 18
Paunov, C., 69 Razin, A., 72
Pavcnik, N., 36 rent-seeking activities, 2, 3–4, 6,
Penn World Tables, 23n13 10n2, 18, 105–6
Peru Reshef, A., 8, 102
clothing exports in, 66 resource curse, 1, 13–22, 15–16f,
EXPYs in, 28 105. See also natural
patents in, 89n2 resources
skilled workers in, 40 revealed comparative advantage
wage premiums in, 36, 38b (RCA), 78–82, 79f,
Philippines, manufacturing exports 80–81f
in, 103 Rigobon, Roberto, 20
“picking winners” slogan, 8–9, 91 Robinson, James, 72
index 129
Tornell, Aaron, 18 V
Torvik, Ragnar, 14, 18
value added goods, 86, 87f, 88t, 89
Toshiba disk drives, 85b
Venezuela
total factor productivity, 17, 82
economic growth in, 18
trade concentration. See export
EXPYs in, 27
concentration
venture capital, 22–23n4
trade liberalization, 94
Verhoogen, E., 53, 76n1
trade quality and portfolio
Vettas, Nikolaos, 93
diversification, 8–9,
Vial, Joaquín, 20
91–104, 106
Viner, Jacob, 14, 16–17
development and, 92
volatility in exports, 8, 92, 94–102,
export structure and financial
96t, 97f, 99t, 106
development and, 99–102
Volkswagen, 66
manufactured exports,
Volvos, 17, 31, 84
distribution of, 102–4
“voracity effect,” 18, 95
market failures and, 91, 92–94
natural resources, export
concentration, and W
volatility and, 94–99, 96t,
97f, 99t Wacziarg, R., 92
transformative technologies, 31 wage premiums, 35–46, 37–38b,
Trinidad and Tobago 39t. See also skill
export quality levels in, 62b premiums
natural resources in, 19 Warner, M., 14, 17, 19–20, 23n7,
turbofan airplanes, 59, 66 23nn9–10, 23n13
Tyson, Laura D’Andrea, 3 Waugh, Michael, 69, 76n6
Wealth of Nations (Smith), 14
Weil, David, 19
U Weiss, John, 25
unions and unionization, 36, 53n3 “What You Export Matters”
United States (Hausmann, Hwang,
automobile imports, 66 & Rodrik), 25
copper exports and, 77–78, 82 Who’s Bashing Whom (Tyson), 3
economic diversification in, 21 wine exports, 7, 61, 66, 106
exports from, 58 Wolpin, Kenneth I., 36
export unit values and, 59, 69 Wright, Gavin, 17, 78, 82–83
externalities and, 3
iPod exports to, 84, 85b X
manufacturing exports in, 103
mining in, 17 Xu, Colin, 21, 74, 91, 100
mobility costs in, 36
quality of goods from, 61
Z
technology industries in, 3
urban areas, education levels in, 38 Zhang, Jinkang, 25
Uruguay Zilibotti, F., 70, 76n4
education in, 38 Zingales, Luigi, 18
wage premiums in, 36, 38b Zoega, Gylfi, 14, 17, 20
ECO-AUDIT
Environmental Benefits Statement
The World Bank is committed to preserving Saved:
endangered forests and natural resources. • 2 trees
The Office of the Publisher has chosen • 1 million BTUs of total
to print Does What You Export Matter? on energy
recycled paper with 50 percent postcon- • 271 pounds of net
sumer fiber, in accordance with the rec- greenhouse gases
ommended standards for paper usage set • 1,220 gallons of
by the Green Press Initiative, a nonprofit waste water
program supporting publishers in using • 77 pounds of solid
fiber that is not sourced from endangered waste
forests. For more information, visit www
.greenpressinitiative.org.
LAT IN AM ERICA N D EVELO P ME NT FORUM S E RI E S
Does what economies export matter for development? If so, can industrial policies
improve on the export basket generated by the market? This book approaches these
questions from a variety of conceptual and policy viewpoints. Reviewing the theoretical
arguments in favor of industrial policies, the authors first ask whether existing indicators
allow policy makers to identify growth-promoting sectors with confidence. To this end,
they assess, and ultimately cast doubt upon, the reliability of many popular indicators
advocated by proponents of industrial policy.
Second, and central to their critique, the authors document extraordinary differences in
the performance of countries exporting seemingly identical products, be they natural
resources or “high-tech” goods. Further, they argue that globalization has so fragmented
the production process that even talking about exported goods as opposed to tasks may
be misleading. Reviewing evidence from history and from around the world, the authors
conclude that policy makers should focus less on what is produced, and more on how
it is produced. They analyze alternative approaches to picking winners but conclude by
favoring “horizontal-ish” policies—for instance, those that build human capital or foment
innovation in existing and future products—that only incidentally favor some sectors over
others.
ISBN 978-0-8213-8491-6
SKU 18491