The World Economy (2009)
doi: 10.1111/j.1467-9701.2008.01150.x
The Impacts of Alignment with
Global Product Standards on Exports
of Firms in Developing Countries
STANDARDS,
World
TWEC
Oxford,
Original
XXX
0378-5920
1467-9701
©GALINA
2008Economy
The
UK
Article
AN
Author
AND
MRAS
KEITH
Journal
ANDE.EXPORTS
compilation
MASKUS © Blackwell Publishers Ltd. 2008
Blackwell
Publishing
Ltd
Galina An1 and Keith E. Maskus2
1
Kenyon College and 2University of Colorado
1. INTRODUCTION
S traditional trade barriers, such as tariffs and quotas, have diminished
in importance, increasing attention has been paid to the trade impacts of
regulatory policies, such as technical regulations and performance requirements.
There can be considerable benefits from more harmonised global standards,
including economies of scale, product compatibility and consumer acceptance of
imports. At the same time, it is costly to comply with international technical
regulations, which therefore might be impediments to trade.1
The compliance challenges are amplified for exporters in developing countries
(DCs) that try to penetrate markets in industrial nations. Compliance costs may
be significantly higher for such firms due to lower initial domestic standards.
Nonetheless, compliance with industrial nations’ standards may provide export
gains to firms in developing countries due to larger demand for their products
and greater incentives to upgrade quality and production processes (Ganslandt
and Markusen, 2001; Jaffee and Henson, 2004).
In this paper we offer the first firm-level study of this empirical question,
using data acquired from a detailed World Bank survey of exporters in several
industries located in developing countries. We examine two types of government
initiatives – mutual recognition agreements (MRAs) and alignment of domestic
standards with international norms. An MRA is a pact between two or more
countries that commits each party to recognise the results of product testing
or certifications of the other in specified sectors. Thus, it does not harmonise
A
The authors are grateful to two anonymous referees for their comments and insights.
1
Surveys may be found in Sykes (1995), Baldwin (2000) and Maskus and Wilson (2001).
552
© 2009 The Authors
Journal compilation © 2009 Blackwell Publishing Ltd, 9600 Garsington Road,
Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
STANDARDS, MRAs AND EXPORTS
553
regulations, but rather recognises that each member’s testing procedures are
adequate to guarantee product safety. It should liberalise trade by reducing the
cost of duplicative testing, inspection and certification.
In contrast, alignment means harmonising national regulations with international standards, such as the ISO 9000 system. This approach may increase
market access by signalling to consumers in all foreign markets that certified
products achieve reliable quality and safety. However, harmonisation imposes
higher compliance costs than MRAs. It becomes an empirical question as to
which approach is likely to raise exports by more (or reduce them by less).
Our empirical contributions are two-fold. First, we employ a firm-level dataset
(from the World Bank Technical Barriers to Trade Survey, described below) of
exporters in developing countries. All previous studies used aggregate trade
flows by industry. Second, we distinguish between two different types of policies:
negotiating MRAs and aligning home measures with international standards.
More specifically, we examine how MRAs and harmonisation of domestic
regulations with international standards influence firm-level decisions on how
much to export to five industrial regions (the United States, the European
Union, Canada, Japan and Australia). We consider exports to developed
countries because the issue of quality standards will most sharply arise there.
Our econometric analysis finds that where poor countries negotiated MRAs
there is a significant export-promoting effect for their firms, a finding that survives
extensive robustness checks. However, alignment with international standards,
though positively associated with such exports, does not have a significant impact.
2. RELATION TO PRIOR LITERATURE
The empirical literature on the trade impacts of standards harmonisation is
scarce, mostly due to data limitations and measurement problems. An early study
was by Swann et al. (1996), who used counts of voluntary national and international standards recognised by the United Kingdom and Germany from the PERINORM database at the three-digit industry level.2 They found that increases in
the number of British national standards raised both manufacturing imports and
exports, while the number of international standards to which Britain was a party
had a significantly positive effect on exports but not imports.
Moenius (2004) extended that analysis by constructing a comprehensive dataset
on industry-specific standards that existed in common between bilateral
trading partners. He found that standards had large trade-promoting effects among
the nations sharing them, while country-specific standards promoted trade in
2
PERINORM is a proprietary database listing detailed product regulations, primarily in developed
countries.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
554
GALINA AN AND KEITH E. MASKUS
manufacturing sectors but hindered trade in agriculture. He attributed the latter
effect to information costs. Manufactured goods must be adapted to foreign
markets, and country-specific standards offer information on particular requirements.
In contrast, agricultural commodities may not require such adaptation and specific
standards are exclusionary.
Chen and Mattoo (2008) examined the effects of regional harmonisation
within the European Union. In their empirical analysis they employed a data
panel covering imports from 28 OECD and 14 developing countries at the threedigit level of manufacturing industries between 1986 and 2001. Using figures on
the counts of harmonisation directives and MRAs at the industry level, the authors
found that harmonisation increased intra-EU trade, while raising exports of
industrialised non-member countries and reducing exports of most developing
countries. However, a higher number of MRAs between the EU and non-member
countries promoted trade among all parties, including developing nations.
A few recent studies specifically address the effects of MRAs on trade. Baller
(2007) studied standards policies among OECD countries between 1986 and
2003 in two sectors, telecommunication equipment and medical devices, in
which MRAs and harmonisation coverage are widespread. He found that MRAs
had a strong positive influence on both export probabilities and trade volumes
for partner countries. Harmonisation, however, had a negligible effect on parties
to the agreement. Vancauteren and Weiserbs (2003) used panel data for 15 EU
countries between 1990 and 1998 and found that sectors with MRAs exhibited
the smallest home bias among all industries subject to some kind of standards
harmonisation. Amurgo-Pacheco (2007) examined the effects of MRAs between the
EU and the United States on third countries in a set of pharmaceutical products.
He found that MRAs are trade-promoting instruments among members, but diminish
exports from other nations. Finally, a report to the European Commission (2003)
that focused on the MRA between the EU and Australia found a positive but
insignificant effect of the MRA in creating trade in four industries.
Case studies of the effects of MRAs on trade are almost non-existent, perhaps
because such agreements are relatively recent, especially with developing countries. Two surveys were undertaken by the European Commission (1998, 2002),
though it is likely that trade impacts were biased downward by the industry selection. The results suggested that for straightforward products that pose few safety
concerns, such as bicycles, tanks and containers, the MRA generally raised trade
within the EU. On the other hand, the application of mutual recognition principles
to technically complex products, such as buses, trucks and road safety equipment,
or products with safety concerns, such as food supplements, had little impact.
There are a few case studies of alignment with international standards by
developing-country governments, though all focus on agricultural commodities.
Stanford (2002) described efforts by the government and avocado growers in
Mexico to impose international standards on producers in that industry, which
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
STANDARDS, MRAs AND EXPORTS
555
resulted in lifting of the USDA ban on avocado imports to the United States in
1997. Henson and Mutillah (2004) discussed Kenyan exports of Nile perch. The
government of Kenya has invested in administrative structures, inspection and
certification procedures, upgrades of laboratory facilities and training. While
these costs have been considerable, there are significant benefits in terms of more
effective quality control. Today most perch exporters have diversified their
export base beyond the EU to Australia, Japan, the United States and other
locations. Jaffee and Henson (2004) examined three case studies of exports of
Guatemalan raspberries, Kenyan fresh produce and Peruvian asparagus, all of
which showed that quality upgrading to international and industrial-countries
standards contributed to export growth.
The theoretical literature on standards and trade is more plentiful, though most
of it focuses on the potential cost-raising impacts of alignment and harmonisation
(Maskus and Wilson, 2001). Borrowing from elements of new trade theory, it is
not difficult to understand why MRAs may be more export-promoting than international standards alignment. As emphasised by Melitz (2003), highly productive
firms are more likely to enter international markets because they can overcome
the fixed costs of exporting more readily than can inefficient firms. Taking this
as his point of departure, Baller (2007) sets out a model in which MRAs reduce
fixed exporting costs because they do not force local firms to invest in testing
and certification procedures that are equivalent to those abroad, which is required
under harmonisation. As a result, his model predicts a larger export-promoting
impact from MRAs than alignment, a finding that is widely anticipated in the
policy literature, at least as a matter of degree (Baldwin, 2000; Maskus and Wilson,
2001; Chen and Mattoo, 2008).3
3. DATA AND ECONOMETRIC SPECIFICATION
Our econometric approach is driven in part by the nature of the available data.
Thus, we first describe the standards data before discussing specification issues.
a. The World Bank Survey
In 2001–02 researchers in the Development Economics Group at the World
Bank commissioned a major survey in order to address how familiar firms in
developing countries were with international product standards and how they
responded to them.4 That effort generated the World Bank Technical Barriers to
3
In his review of EU experience, Pelkmans (2003) adds a further advantage of MRAs, in that they
are less subject to capture by those who might engage in costly overregulation.
4
See Maskus et al. (2001) for an extensive discussion of these issues.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
556
GALINA AN AND KEITH E. MASKUS
Trade Survey, described in detail by Wilson and Otsuki (2004). The survey covers
689 firms in 25 industries in 17 developing countries (DCs) and transition
economies for either 2000 or 2001. The primary goal of that project was to
evaluate the impact of foreign standards and technical barriers to trade on the
operations of firms in DCs. Firms were asked extensive questions about their
knowledge of such regulations in five main developed regions (United States,
EU, Canada, Japan and Australia) and their reactions to them. Questions also
were posed about domestic government policies and whether they hindered or
enhanced the respondent firm’s ability to export.
The data we employ form a subset of the overall survey. Appendix Table A1
presents the detailed industry and country composition of our dataset after cleaning and omitting observations with missing values for the variables we analyse.
Our firms cover both manufacturing and agricultural products. As can be seen in
Appendix Table A1, 40.9 per cent of the observations in our sample come from
India, while the representation of other countries in the sample varies from 0.7
per cent (Senegal) to 7.6 per cent (South Africa). In terms of industry composition,
textiles and apparel constitute 29 per cent of the sample, followed by processed
food and tobacco (12 per cent) and raw agricultural products (10 per cent). The
percentage of other industries varies from 0.5 per cent (materials) to 7.1 per cent
(miscellaneous manufacturing commodities).
Our sample contains 421 firms, each of which reported their exports, if any,
to the five developed regions mentioned above and total world exports. Values
were converted into thousands of US dollars at contemporaneous exchange rates.
Our two policy variables are taken from the survey as well. The first, which
we call STANDARD, is the total number of international standards with which
the country’s government has aligned its domestic regulations, as reported by
firm managers. Specifically, firms were asked the following question: ‘With which
international standards did your government choose to align their national regulations?’ The question listed as choices the International Standards Organization
(ISO), International Electrotechnical Commission (IEC), International Telecommunications Union (ITU), Codex Alimentarius (CODEX), Office International
des Epizooties (OIE), and the Food and Agricultural Organization’s International
Plant Protection Convention (IPPC). For each choice the response was either yes
or left blank.
We constructed STANDARD as the sum of all positive responses. Thus, if a
firm reported that its government had aligned domestic regulations with ISO only,
STANDARD would equal one; if it had also adopted CODEX, STANDARD would
be equal to two, and so on. Detailed information is available in a supplementary
data appendix on request. In summary, approximately 70 per cent of the firms
did not report any known alignment with international standards, while 26 per
cent reported alignment with only one standard and only four firms mentioned
three. Some firms in all countries reported at least one such alignment. At the
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
STANDARDS, MRAs AND EXPORTS
557
industry level, there was at least one firm in all sectors except petroleum, leather
products and paper products that listed positive alignment with international
standards. The largest absolute numbers of responses arose in textiles and
apparel, followed by processed food and tobacco, and raw agricultural products.
As a proportion of total firms in each industry claiming alignments, the highest
were in instruments and photographic and optical goods (66 per cent) and primary
metals (60 per cent).
The second policy variable is MRA, which was constructed from the question:
‘Are any of your top five revenue-generating products subject to mutual recognition agreements in any of the following countries/regions: US, EU, Canada,
Japan, Australia?’ Similar to STANDARD, variable MRA is the sum of positive
responses, with the highest reported value being the maximum of five. To summarise, most firms did not report any MRAs (87.9 per cent). For example, no
firms from India, Senegal and Iran operated under any MRA with the five developed countries. These governments have not placed emphasis on negotiating
such agreements on behalf of exporting firms. Among those firms reporting
agreements, 35 mentioned MRAs with one developed country and 16 reported
MRAs with multiple regions.
Overall, adherence with international standards is more prevalent in the data
than membership in MRAs, perhaps because the former may be less strict than
what developed countries require in the latter. Further, alignment with international standards is a unilateral decision, which is easier than concluding an MRA.
Summary statistics for exports and the policy variables are in Table 2.
b. Econometric Specification
In our dataset 18 per cent of all firms do not export to any of our five
developed regions. Thus, our main approach is to use Tobit estimation of the
following equation:
EXPijk = β′ Xijk + λzijk + ajDj + δkDk + εijk.
This equation determines the amount of firm i’s exports (in log form) in industry
j from country k, EXPijk, summed across five developed markets (United States,
EU, Canada, Australia and Japan) as a function of firm characteristics, Xijk, and
government policies toward compliance with the international standards, zijk.5 The
equation also includes fixed effects for industry and exporting country.
The Xijk vectors include several firm characteristics reported in the survey. The
first is years since the enterprise was founded (AGE), since it is likely that older
firms have established domestic and international markets. Second is fixed assets
5
Because exports were zero for some firms, we actually take logs of one plus exports.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
558
GALINA AN AND KEITH E. MASKUS
(FIX_A) to control for the firm’s size.6 A third factor is the extent of foreign
ownership (F_OWN ), which might enhance the ability to penetrate foreign
markets. A fourth characteristic is the average growth rate of sales in the past
three years (GR_PAST ). Finally, we control for the type of the firm through the
use of dummy variables that indicate whether the firm is a state enterprise,
private company, subsidiary of a multinational firm, joint venture, or other type.
Variables AGE and FIX_A enter the estimation in natural logs.
The industry dummies control for unobserved industry-specific characteristics,
such as technological intensity, trade barriers in importing nations and market
structure. The exporter country dummies reflect geographical characteristics,
aggregate trade policies of local governments, unobserved treatment by industrial
countries of a particular developing country, such as preferential trade agreements, and any other international trade considerations that differ by exporting
country. We assume that εijk is a random disturbance term symmetrically distributed about zero and uncorrelated with the explanatory variables.
c. Blank Entries
A primary problem with the construction of our policy variables was how to
treat blank entries on firms’ responses about MRAs and international standards.
Those entries could either indicate that there were no such measures or that the
firm managers were unaware of them. In principle, we would expect most such
answers to mean the absence of measures since the firms in our sample are
exporters or indicated an intention to export, suggesting that managers would be
aware of product regulations in foreign markets. Nonetheless, we developed
several versions of the policy variables based on different assumptions.
First, we assumed that blank entries truly signify no MRAs (MRA = 0) or no
incorporation of international standards (STANDARD = 0).7 Second, we assumed
that the number of MRAs and standards should be similar by country and industry,
so we assigned average values of MRA or STANDARD, calculated by country and
industry code, to missing answers. Third, we used the maximum value of
STANDARD and MRA by industry and country instead of average value. Finally,
we experimented with assigning averages and maximum values to all firms, not
just to those with blank answers, in our fourth and fifth versions, respectively.
6
We also tried both sales and employment to control for size and the results were qualitatively the
same. However, due to concern about contemporaneous endogeneity of those variables we chose
the size of fixed assets since investment decisions generating capital stocks are made in previous
periods, making fixed assets predetermined. In our sensitivity section we treat fixed assets as
endogenous and use instrumental variables to purge correlation with unobserved firm effects. As
shown below, the main regression results are very similar.
7
Even if managers were unaware of existing measures, a blank entry would suggest they do not
take such measures into account in their decision-making.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
STANDARDS, MRAs AND EXPORTS
559
Our primary reported regressions employ assumption two, assigning average
values for MRA and STANDARD to the blank entries, because it seems to be most
realistic. However, we show results for all versions for comparison purposes.
d. Endogeneity and Instrumental Variables
The most significant problem we face is that our sample is a cross-section,
rather than a panel.8 This issue makes identification of the causal impacts of our
policy variables on exports difficult. The reason is that STANDARD and MRA
may be endogenous, since the government could decide to negotiate more MRAs
and align with international standards in sectors of particular comparative advantage or where exports are growing rapidly. Note that MRA negotiations and
harmonisation with international standards take years, suggesting that such
policies might be considered predetermined. However, firms and sectors with
significant exports in the cross-section may have been high-growth entities in the
past, generating a channel from exports to policy. Alternatively, both the decisions
to export by firms and government involvement with trade issues might be
influenced by an unobserved third factor.
We address endogeneity concerns by incorporating, as instrumental variables,
several standard political-economy measures at the country–industry level.
Specifically, the decision to negotiate MRAs and align with international
standards may be influenced by a variety of economic and political factors. Thus,
for economic instruments in the manufacturing sectors we extracted several series
from the World Bank’s ‘Trade, Production and Protection 1976–2004’ dataset.9
That database contains exports, employment, investment and other measures by
country and industry at the three-digit International Standard Industrial Classification (ISIC) level. We developed a concordance between the ISIC and our industry
breakdown, which is available in a supplemental appendix. For the agricultural
sector we compiled similar figures from the World Development Indicators.
We use data from these sources to construct the following variables, which are
sensible instruments for interests that could affect MRA and standards policy,
while plausibly uncorrelated with the error terms in export regressions using data
for individual firms, given that our sample covers only a small number of enterprises in each sector–country pair.
First, IND_EXP_GR is the growth of industry exports for each country in our
sample between 1989 and 1996. Since we do not have the information on when
MRAs and international standards were enacted, we chose this period to make
8
Unfortunately, the World Bank decided not to replicate the survey in later years.
These data were compiled by the Trade Research Group in Development Economics and are
available at the World Bank website. The most recent figures contain data for all of the countries in
our database. Some data are missing for particular years and industries, which is why we use averages.
9
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
560
GALINA AN AND KEITH E. MASKUS
sure that we are looking at the industry’s export growth before those initiatives
came into effect.
Second, COMP_ADV is the average industry’s revealed comparative advantage
between 1989 and 1996, calculated as:
⎛
⎜ EXPjk
⎝
⎛
⎜ ∑ EXPjk
⎝ j
⎞
∑ EXPjk ⎟⎠
k
⎞
EXP
∑ ∑ jk ⎟⎠
j k
.
This measure captures country k’s share of industry j’s exports in total world
exports of industry j, divided by country k’s share of total world exports. A higher
ratio ‘reveals’ higher comparative advantage, though there are problems with that
interpretation.10 Calculated at the industry level it is a sensible instrument for
trade-policy interests of individual firms.
Third, EMPL is the average industry employment share by country between
1989 and 1996, calculated as the ratio of industry employment over the country’s
labour force.11 Finally, HERF is the Herfindahl index of industry concentration
in the United States for 1992, acquired from the US Census Bureau.
In addition to economic factors, domestic political structure can also influence
the amount of government involvement in trade policy. For this reason we add
to the pool of instruments two government structure variables. These include a
democratisation index, derived from Gastil’s political rights index (Freedom
House, various years), averaged over 1989–96, and the relative size of the
government, defined as the share of government spending as a percentage of
GDP between 1989 and 1996. We also incorporate interaction terms between our
economic instruments and government structure variables.
In the estimation, we choose the instruments used for each specification from
the pool of constructed IVs using the following criteria. First, they have significant
correlation with the policy variables. Second, they satisfy the overidentification
tests of instrument exogeneity (Hansen’s J-test). Third, we consider whether the
exclusion restrictions are appropriate by testing whether the instruments have no
independent influence on the dependent variable, beyond that exercised through
the policy variables. We check that by entering each instrument in the second
stage and testing its significance.
Precise definitions of all variables and their units are presented in Table 1,
while summary statistics are in Table 2.12
10
See Ballance et al. (1987) for an extensive discussion and Balassa and Bauwens (1988) for an
application.
11
Labour force figures were taken from the World Bank, World Development Indicators.
12
A correlation matrix is available in the supplementary data appendix.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
STANDARDS, MRAs AND EXPORTS
561
TABLE 1
Variable Definitions
Description
LHS variables
EXP
EXP_all
RHS variables
STANDARD
MRA
GR_PAST
AGE
FIX_A
F_OWN
Log of Exports to developed countries – EU, USA, Canada, Japan, Australia
(exports are in 2001 US$1,000)
Log of total exports (exports are in 2001 US$1,000)
The number of international standards with which the country’s government
chose to align their national regulations (ISO, IEC, ITU, CODEX, OIE, IPPC)*
The sum of positive responses to the questions whether any of the top five
products are subject to MRA with US, EU, Canada, Japan or Australia
Average growth rate of sales in the past three years (%)
Log of firm’s age = Log(2001 – year of establishment)
Log of value of plant’s fixed assets (in 2001 US$1,000)
Share of foreign ownership:
1 – no foreign ownership, 2 – less (or equal) than 50%, 3 – more than 50%
foreign ownership
Type of ownership (dummies)**
PRIVN
Headquarter local of a privately held, non-listed company
PRIVL
Headquarter local of a publicly traded or listed company
SUBD
Subsidiary/division of a domestic enterprise
SUBM
Subsidiary/division of a multinational firm
JOINTD
Joint venture of a domestic enterprise
JOINTM
Joint venture of a multinational firm
STATE
Completely or partially state-owned company
Instruments
IND_EXP_GR
COMP_ADV
Average growth of industry exports in each country between 1989 and 1996
Average industry j’s revealed comparative advantage for country k between
1989 and 1996, calculated as
⎛
⎜ EXPjk
⎝
⎛
⎜ ∑ EXPjk
⎝ j
EMPL
HERF
GOVT
DEM
⎞
∑ EXPjk ⎟⎠
k
⎞
∑ ∑ EXPjk ⎟⎠
j
k
10
Average industry employment share by country between 1989 and 1996,
calculated as the ratio of industry employment over country’s labour force × 1,000
(number of employees per 1,000 of labour force)
Herfindahl index of industry concentration in the US for 1992 (divided by 1,000)
Average government expenditures as a share of GDP (in %) between
1989 and 1996
Average democracy index (= 8 – Gastil’s political rights index) between 1989 and
1996
Notes:
* ISO – International Standards Organization.
IEC – International Electrotechnical Commission.
ITU – International Telecommunications Union.
CODEX – Codex Alimentarius Commission.
OIE – Office International des Epizooties.
IPPC – FAO International Plant Protection Convention.
** The benchmark type of ownership is cooperative/collective.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
562
GALINA AN AND KEITH E. MASKUS
TABLE 2
Summary Statistics
Variable
Obs.
Mean
Std. Dev.
Min.
Max.
EXP
EXP_all
STANDARD
MRA
GR_PAST
AGE
FIX_A
F_OWN
PRIVN
PRIVL
SUBD
SUBM
JOINTD
JOINTM
STATE
IND_EXP_GR
COMP_ADV
EMPL
HERF
GOVT
DEM
421
412
421
421
421
421
421
421
421
421
421
421
421
421
421
421
421
421
421
421
421
5.096
6.613
0.356
0.185
0.194
2.794
6.398
1.235
0.565
0.164
0.007
0.036
0.026
0.019
0.014
0.309
0.034
0.152
0.948
14.071
4.657
3.226
2.312
0.595
0.593
0.325
0.832
2.285
0.593
0.496
0.371
0.084
0.186
0.160
0.137
0.119
0.362
0.119
0.129
0.397
4.160
1.201
0
0
0
0
−0.311
0.693
0.722
1
0
0
0
0
0
0
0
0.002
0.000
−0.531
0.255
8.854
2.125
11.758
12.476
3
5
3.000
4.956
13.201
3
1
1
1
1
1
1
1
2.158
0.845
0.727
1.873
21.755
6.500
4. ECONOMETRIC RESULTS
We turn next to the presentation and discussion of the statistical analysis,
beginning with the base case and working through endogeneity analysis and
robustness checks.
a. Basic Specification
First, we estimate our regression equation using Tobit in the cross-section
of firms summarised in the data section, but without instrumental variables.
Table 3 presents our initial estimation results using version two (average values
of MRA and STANDARD are assigned to the blank entries) of our policy
variables. Here we include firm characteristics and country and industry dummies.
Table 3 contains three specifications, with the policy variables entering separately
and together.
The estimation results in columns (1)–(3) show that both alignment with international standards (STANDARD) and negotiating mutual recognition agreements
(MRA) are positively and significantly (at 5 per cent and 1 per cent significance
level, respectively) associated with exports to the five developed regions. The
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
STANDARDS, MRAs AND EXPORTS
563
TABLE 3
Tobit Estimation of Export Equations
(1)
MRA
STANDARD
GR_PAST
AGE
FIX_A
F_OWN
PRIVN
PRIVL
SUBD
SUBM
JOINTD
JOINTM
STATE
Log-likelihood
Pseudo-R2
Observations
(2)
(3)
0.818*** (0.304)
0.736 (0.502)
0.184 (0.218)
0.610*** (0.088)
0.572 (0.371)
−0.244 (0.520)
−0.197 (0.636)
−8.113*** (2.428)
1.664 (1.236)
−1.223 (1.350)
−0.837 (1.325)
−1.360 (1.503)
0.816*** (0.231)
0.715** (0.300)
0.687 (0.494)
0.208 (0.214)
0.581*** (0.087)
0.552 (0.365)
−0.218 (0.511)
−0.283 (0.626)
−7.306*** (2.392)
1.684 (1.216)
−0.857 (1.332)
−0.606 (1.306)
−1.243 (1.485)
0.870*** (0.232)
0.752 (0.497)
0.211 (0.216)
0.594*** (0.088)
0.531 (0.368)
−0.143 (0.514)
−0.202 (0.630)
−6.866*** (2.383)
1.652 (1.226)
−0.855 (1.341)
−0.195 (1.304)
−1.489 (1.486)
−952.57
0.0859
421
−955.91
0.0827
421
−949.76
0.0886
421
Notes:
All equations include industry and country fixed effects. Standard errors are in parentheses. *** Significant at 1%,
** significant at 5%, * significant at 10%.
magnitude of the coefficients is similar for both policy variables. In terms of economic effects, the coefficients in this regression imply that, on average, having an
MRA with an additional developed country (or region) in our sample, or aligning
with one additional international system of standards, is associated with approximately 122 per cent more exports of individual firms to these developed markets.13
The magnitudes of these coefficients, though large, seem plausible. Most of the
firms in the sample list no or one MRAs with developed countries, so negotiating
the first such agreement might have a large impact on the exports to industrial
nations for a number of reasons. It could raise demand for a firm’s products by
signalling to importers abroad that those products may be brought in without
further certification. Further, entry into an MRA could substantially decrease the
cost of exporting and there may be learning effects that can encourage exporting
to other rich nations. Similarly, almost 70 per cent of the firms reported no
adoption of international standards. Thus, a harmonisation policy could require
them to undertake substantial investments in production processes and quality
upgrading. Alternatively, since the direction of causality is not clear in this basic
Tobit case, the government may have adopted international standards in sectors
that were already exporting and complying with those norms.
13
We calculate percentage change in EXP as 100(e0.8 − 1) = 122.5 per cent.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
564
GALINA AN AND KEITH E. MASKUS
Concerning other control variables, the size of fixed assets is positively associated with exports to developed countries. It also appears that subsidiaries of
domestic companies tend to export less to developed regions.
b. Endogeneity and Instrumental Variables Estimation
Next we address the endogeneity concerns regarding our policy variables,
while below we consider endogeneity of fixed assets. As discussed earlier, we
develop a series of potential IVs from independent data sources at the World
Bank. The second-stage IV–Tobit regression results are shown in Table 4.14 We
note from the first-stage regressions (not shown here) that MRA is a function of
several political-economic factors, including the Herfindahl index, democracy
and the interaction between them, and interactions between industry export
growth and the size of the government, and industry export growth and democracy. Specifically, MRAs are more likely to be negotiated in more concentrated
industries and less democratic societies. In addition, for an average firm in our
sample industry export growth is associated with a higher number of MRAs.15
Column (1) shows the second-stage regression for the case where only MRA
is potentially endogenous. Of primary interest is that the coefficient for MRA
remains positive and significant at the 10 per cent level. Its magnitude rises from
0.87 without instrumentation in Table 3 to 0.92 in this case. Thus, once we purge
MRA of its political and economic determinants, its effect on exports to developed
economies becomes even greater.
The second column shows the corresponding IV–Tobit estimation accounting
for endogeneity of STANDARD. First-stage results suggested that alignment with
international standards is more prevalent in less democratic societies and for an
average firm it is negatively associated with industry export growth. In terms of
our main question, the second-stage results find that, unlike MRA, the magnitude
of the coefficient on STANDARD falls with IV estimation. Further, the coefficient
is not significant, suggesting that there is little evidence of a causal relationship
from standards alignment to exports to developed countries. In this context,
MRAs appear to be more determinative means of expanding exports than are
international standards for firms in poor countries.
Finally, the third column shows the results when both policy variables enter
the IV–Tobit estimation. Here the coefficient for MRA is still positive and
significant at the 10 per cent level, while that for STANDARD is not. Overall,
our evidence suggests that alignment with international standards may not
14
These regressions were run with the ivtobit routine in Stata. In order to conserve space we do
not show first-stage results here but they are available on request.
15
By average we mean a firm that comes from a country with the average share of government
expenditure in relation to GDP (14 per cent) and the average democracy index (4.65).
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
MRA
STANDARD
FIXED_A
MRA_ALL
GR_PAST
AGE
FIX_A
F_OWN
PRIVN
PRIVL
SUBD
SUBM
JOINTD
JOINTM
STATE
Observations
Wald χ2(43)
Hansen J-test (p-value)
MRA
STANDARD
(1)
(2)
0.924* (0.534)
0.606 (1.380)
MRA and
STANDARD
(3)
0.969* (0.594)
−0.231 (1.513)
MRA and
FIXED_ A
(4)
MRA_ALL
(5)
1.001* (0.589)
0.717* (0.403)
0.748 (0.497)
0.212 (0.216)
0.592*** (0.089)
0.529 (0.368)
−0.141 (0.514)
−0.209 (0.632)
−6.827*** (2.406)
1.654 (1.225)
−0.836 (1.351)
−0.184 (1.307)
−1.474 (1.490)
421
284,045.42
0.90767
0.756 (0.519)
0.184 (0.217)
0.614*** (0.092)
0.568 (0.371)
−0.221 (0.537)
−0.171 (0.654)
−8.046*** (2.457)
1.648 (1.237)
−1.235 (1.350)
−0.721 (1.512)
−1.443 (1.598)
421
36,567.00
0.34229
0.770 (0.518)
0.213 (0.217)
0.593*** (0.092)
0.527 (0.372)
−0.116 (0.542)
−0.192 (0.652)
−6.915*** (2.512)
1.622 (1.234)
−0.856 (1.358)
−0.059 (1.582)
−1.586 (1.602)
421
2,046.69
0.82703
0.709 (0.514)
0.131 (0.336)
0.450 (0.446)
−0.170 (0.523)
−0.489 (1.088)
−7.064*** (2.525)
1.491 (1.333)
−0.835 (1.355)
−0.344 (1.406)
−1.611 (1.557)
421
7,627.15
0.81440
0.612 (0.643)
0.598** (0.272)
0.136 (0.126)
0.636*** (0.055)
0.249 (0.198)
0.128 (0.284)
0.454 (0.344)
0.183 (1.104)
1.117* (0.660)
−0.756 (0.780)
0.153 (0.710)
−0.441 (0.772)
412
416.92
0.67357
STANDARDS, MRAs AND EXPORTS
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
TABLE 4
IV–Tobit Estimation of Second-Stage Export Equations
Notes:
Column headings indicate variables subject to IV estimation in first stage. All equations include industry and country fixed effects. Standard errors are in parentheses.
*** Significant at 1%, ** significant at 5%, * significant at 10%.
565
566
GALINA AN AND KEITH E. MASKUS
promote exports to developed countries as much as negotiation of MRAs. For
further perspective, we present in Appendix Table A2 descriptive statistics for
two groups of firms: those covered by MRAs and those not covered. As the
figures reveal, firms that report the existence of MRAs are younger and larger on
average and have a larger share of foreign ownership. None is state owned or a
subsidiary of a domestic enterprise. However, a larger portion is either a subsidiary of a multinational firm or part of a joint venture. It seems that, at least in
this sample, having foreign ownership or a JV partnership is correlated with the
ability to use an MRA for export growth. This outcome may be due to local
subsidiaries having easier access to higher-quality standards from their investment partners. There also may be greater trust in product quality if a firm has
foreign business partners.
c. Sensitivity Checks
Given that our policy variables are measured somewhat indirectly, it is important
to subject the basic results to sensitivity and robustness checks. As mentioned
above, we also ran the same types of regressions using the other four versions of
our policy variables, with zero entries treated in a different fashion. In Table 5
we summarise the coefficients only for the policy variables, since the results for
other regressors were almost identical.16 Coefficient estimates of the policy
variables are similar for versions (2)–(5). In version one the coefficients for MRA
in the export equation are almost double those in the other versions and suggest
unrealistically high trade elasticities. The main conclusions are consistent across
the other treatments of zero entries in computing the policy variables. In the
remaining analysis we only employ version two, which, as we argued above, seems
the most natural construct and has the advantage over versions four and five of
preserving variation across firms, rather than just across industries and countries.
As a second sensitivity check, we introduced country–industry interaction
dummies in addition to country and industry dummies, a rigorous specification
of controls. Since our data have a limited set of observations, for this purpose we
aggregated countries into five regions (Eastern Europe, Latin America and
Caribbean, Middle East, South Asia and Sub-Saharan Africa) and industries into
five groups (raw food, processed food and related goods, equipment, materials,
and textiles). This resulted in 32 dummy variables (24 interaction, four industry
and four country dummies) in our regression equations. We simply summarise
the results here. The MRA coefficients remain positive and significant in all
equations, though again only marginally so in the IV–Tobit case with both policy
variables included. Thus, our earlier conclusion holds up: MRAs are effective
16
Full results are available on request.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
EXP
Tobit
(1)
Version 1: Blank entries treated as zero
MRA
1.229***
(0.290)
STANDARD
Pseudo-R2
Wald χ 2
0.0878
IV–Tobit
(2)
(3)
(1)
2.550*
(1.401)
0.802***
(0.290)
0.0829
1.174***
(0.289)
0.717***
(0.285)
0.0908
1,155.20
Version 2: Blank entries assigned the mean value for MRA and STANDARD by industry and country
MRA
0.870***
0.816***
0.924*
(0.232)
(0.231)
(0.534)
STANDARD
0.818***
0.715**
(0.304)
(0.300)
Pseudo-R2
0.0859
0.0827
0.0886
Wald χ 2
284,045.42
Version 3: Blank entries assigned the maximum value for MRA and STANDARD by industry and country
MRA
0.681***
0.623***
0.931*
(0.200)
(0.200)
(0.513)
STANDARD
0.784***
0.676**
(0.287)
(0.285)
Pseudo-R2
0.0848
0.0828
0.0875
10,571.31
Wald χ 2
(2)
(3)
1.515
(2.328)
2.683*
(1.594)
−0.395
(2.663)
3,375.83
702.90
0.606
(1.380)
0.969*
(0.594)
−0.231
(1.513)
36,567.00
2,046.69
0.585
(1.277)
0.979*
(0.601)
−0.201
(1.460)
37,199.82
1,798.74
STANDARDS, MRAs AND EXPORTS
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
TABLE 5
Summary of Coefficients for Different Versions of STANDARD and MRA in Export Equations
567
568
TABLE 5 Continued
Tobit
(1)
IV–Tobit
(2)
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
(3)
(1)
Version 4: STANDARD and MRA are country–industry means
MRA
0.896***
(0.234)
STANDARD
0.855***
(0.317)
Pseudo-R2
0.0862
0.0828
Wald χ 2
0.833***
(0.234)
0.724**
(0.314)
0.0888
0.911*
(0.525)
Version 4: STANDARD and MRA are country–industry maxima
MRA
0.788***
(0.203)
STANDARD
0.854***
(0.278)
Pseudo-R2
0.0864
0.0838
Wald χ 2
0.694***
(0.205)
0.674**
(0.279)
0.0892
2.93e+06
Notes:
All equations include industry and country fixed effects. Standard errors are in parentheses.
*** Significant at 1%, ** significant at 5%, * significant at 10%.
(2)
(3)
0.618
(1.351)
0.968*
(0.595)
−0.275
(1.510)
31,876.26
2,014.93
0.591
(1.253)
0.923
(0.583)
−0.118
(1.438)
20,082.73
2,351.96
0.893*
(0.485)
56,985.58
GALINA AN AND KEITH E. MASKUS
EXP
STANDARDS, MRAs AND EXPORTS
569
policy initiatives for export promotion, while alignment with international
standards does not seem to have much impact on firm-level manufacturing
exports to developed countries.
Next, we address another endogeneity concern that relates to the size of a firm,
measured here by the value of fixed assets. Though the size of fixed assets is in
principle predetermined from prior investments, there might still be unobserved
firm characteristics that influence both the size of the firm and the amount of
exports. One example is having ties with the government, which might be helpful
in both export growth and asset expansion. Furthermore, a firm might make
export decisions based on the discounted present value of future assets. Thus, we
treat the fixed assets variable also as endogenous. The results for the case where
we treat MRA and fixed assets as endogenous are in the fourth column of Table 4.
In the second stage we find that the coefficient on MRA is quite similar to that
in the first column. To conserve space we do not show the results for the cases
with STANDARD included, for those coefficients are consistent with earlier
findings.
As a fourth robustness exercise we consider potential heteroscedasticity in
the second-stage IV–Tobit regression residuals. When such residuals have
unequal error variances, the consistency of Tobit estimates is in question
(Wooldridge, 2003). The most likely source of heteroscedasticity is variation in
firm size, as captured by fixed assets. Thus, to address this problem we first
regress the squared residuals from the IV–Tobit specification of the export
equation with MRA as the only policy variable, taken from Table 4. This regression yields:
2
=
eijk
0.0896
0.5539
+
* FIX _ Aijk + uijk .
(0.1430) (0.0211)
This result suggests there is positive correlation between squared residuals and
the size of fixed assets. Accordingly, we scale all variables by the square root of
(0.5539 + 0.0896 × FIX_A) and run the new regression equation, verifying that
the squared residuals in the new specification are not statistically significantly
related to asset size. The coefficient estimate on MRA in the scaled regression is
1.062, similar to those in Table 4, and remains significant at the 10 per cent level.
This finding confirms that our main results still hold and the size of the
coefficient for MRA is close to that in the basic specification.
A further question to check is whether our results are dominated by the high
number of Indian firms in the data sample. As noted earlier, such firms comprise
nearly 41 per cent of the firms overall, yet report no MRAs in their survey
responses. Running the IV–Tobit estimation after removing Indian enterprises,
we find that the responsiveness of exports to the number of MRAs is somewhat
larger than that in the base case in the first column of Table 4 and remains
significant.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
570
GALINA AN AND KEITH E. MASKUS
As a final check, we also consider the relationship between our policy variables and total exports, rather than just exports to the five developed regions. In
these specifications, MRA includes all mutual recognition agreements, not just
those with the developed countries. As shown in the final column of Table 4, the
IV coefficient on MRA becomes insignificant. That the responsiveness of exports
to all MRAs is lower when we include all export destinations accords with
intuition, because standards for achieving mutual recognition are generally less
rigorous in other developing countries than in developed markets. In turn, the
impact on consumer acceptance in export destinations may be less.
5. CONCLUDING REMARKS
This paper offers the first firm-level econometric evidence on the effects of
aligning domestic product standards with international norms and of negotiating
mutual recognition agreements on firm-level manufacturing exports of developing nations. Without considering endogeneity of the policy variables, we found
that both measures are associated with more exports to developed countries.
However, when we apply instrumental variables to deal with the problem of
endogeneity we found that the influence of MRAs remains significantly positive,
but that of standards alignment becomes insignificant. Note that there is no evidence from this analysis that linking with international standards actually causes
exports from developing-country firms to be reduced through higher costs. These
results are consistent with the findings in Baller (2007).
On this evidence, it seems that MRAs are particularly stimulating for exports,
presumably because they reduce the direct costs of exporting associated with
certification and testing. Alignment with international standards, such as ISO and
CODEX, has lesser ability to expand exports from poor countries, perhaps
because those norms may be insufficient to meet the requirements of developedcountry markets.
These results suggest that if governments of developing countries are interested in increasing exports to developed countries’ markets, resources could be
spent on negotiating MRAs. Of course, in making such choices governments are
constrained by their available resources and often it is costly to negotiate comprehensive MRAs. Further, it is not evident from these results that this approach
is the first-best means of expanding exports. Overall, however, the findings in
this paper support the notion that some forms of linkage to international product
norms can enhance export performance of individual firms, rather than serve as
export barriers.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
TABLE A1
Industry and Country Composition
Code Industry
Number of Observations
ARG
1
2
3
4
5
6
7
8
9
11
12
13
14
15
16
18
19
20
22
BLG
CHL
CZE
Raw agricultural products
1
1
7
1
Meat and fish products, livestock
1
1
0
1
Electrical and electrical equipment
1
2
0
6
Fabricated metal
1
1
0
0
Industrial machinery and equipment 1
1
0
0
Industrial or agricultural chemicals
2
3
2
5
Instruments and photographic and
0
0
0
1
optical goods, watches and clocks
Leather and leather products
0
1
0
0
Paper and allied products
2
0
0
0
Processed food and tobacco
6
2
2
3
Rubber and plastic products
4
0
2
0
Telecommunications and
0
0
0
3
terminal equipment
Textiles and apparel
1
5
1
3
Transportation equipment and
4
1
0
5
automotive parts, and dealers
Lumber, wood and furniture
0
0
1
3
Primary metal and metallic ores
0
0
0
0
Petroleum and other
0
0
0
0
non-metallic minerals
Miscellaneous manufactured
0
0
0
0
commodities
Material
0
2
0
0
Total
24
20
15
31
Per cent
5.70 4.75 3.56 7.36
HND
IND
IRN
JOR
KEN
NGA
PAK
PAN
POL
SEN
ZAF
UGA
Total
Per
Cent
STANDARDS, MRAs AND EXPORTS
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
APPENDIX
1
1
0
0
0
0
0
4
0
9
6
7
1
1
0
0
1
0
0
4
0
2
0
0
1
0
1
0
12
0
0
0
0
0
0
8
0
0
0
0
0
0
0
1
0
0
0
0
1
0
3
0
0
0
1
0
2
0
5
0
0
3
0
0
0
0
0
0
2
0
0
0
1
6
1
2
0
3
2
1
0
0
0
0
42
10
26
15
10
26
3
9.98
2.38
6.18
3.56
2.38
6.18
0.71
0
0
2
0
0
15
0
11
1
1
1
0
1
0
0
0
0
7
2
0
0
0
3
0
0
0
1
3
1
0
2
0
0
1
0
0
0
2
2
0
0
0
2
0
0
0
0
0
0
0
0
0
4
3
1
1
0
3
0
0
20
3
51
16
5
4.75
0.71
12.11
3.80
1.19
1
0
87
4
2
0
3
0
0
0
1
0
6
0
1
0
6
3
0
0
5
4
0
0
122
21
28.98
4.99
1
0
0
0
0
0
1
0
2
0
1
3
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
2
2
0
1
1
6
5
8
1.43
1.19
1.90
0
25
1
0
1
0
1
0
0
1
1
0
30
7.13
0
0
0
0
0
0
0
6
172
13
20
16
14
12
1.43 40.86 3.09 4.75 3.80 3.33 2.85
0
0
9
22
2.14 5.23
0
0
0
2
0.48
3
32
12
421
100.00
0.71 7.60 2.85 100.00
571
Note:
ARG: Argentina; BLG: Bulgaria; CHL: Chile; CZE: Czech Republic; HND: Honduras; IND: India; IRN: Iran; JOR: Jordan; KEN: Kenya; NGA: Nigeria; PAK: Pakistan; PAN: Panama; POL: Poland;
SEN: Senegal; ZAF: South Africa; UGA: Uganda.
572
Variable
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
EXP
EXP_all
STANDARD
GR_PAST
AGE
FIX_A
F_OWN
PRIVN
PRIVL
SUBD
SUBM
JOINTD
JOINTM
STATE
Reported MRAs
Did Not Report MRAs
Obs.
Mean
Std. Dev.
Min.
Max.
Obs.
Mean
Std. Dev.
Min.
Max.
51
51
51
51
51
51
51
51
51
51
51
51
51
51
6.821
7.46
0.588
0.23
2.602
7.179
1.51
0.608
0.176
0
0.078
0.059
0.039
0
2.248
2.135
0.726
0.31
0.738
2.056
0.809
0.493
0.385
0
0.272
0.238
0.196
0
0.403
1.785
0
−0.31
0.693
3.398
1
0
0
0
0
0
0
0
11.76
12.48
3
1.5
4.644
13.2
3
1
1
0
1
1
1
0
370
361
370
370
370
370
370
370
370
370
370
370
370
370
4.858
6.494
0.324
0.189
2.821
6.291
1.197
0.559
0.162
0.008
0.030
0.022
0.016
0.016
3.270
2.314
0.568
0.328
0.842
2.297
0.547
0.497
0.369
0.090
0.170
0.146
0.126
0.126
0
0
0
−0.200
0.693
0.722
1
0
0
0
0
0
0
0
11.724
11.724
3
3.000
4.956
12.235
3
1
1
1
1
1
1
1
GALINA AN AND KEITH E. MASKUS
TABLE A2
Descriptive Statistics for Firms According to MRA Experience
STANDARDS, MRAs AND EXPORTS
573
REFERENCES
Amurgo-Pacheco, A. (2007), ‘Mutual Recognition Agreements and Trade Diversion: Consequences
for Developing Nations’, HEI Working Paper No. 20-2006 (Economics Section, The Graduate
Institute of International Studies).
Balassa, B. and L. Bauwens (1998), Changing Trade Patterns in Manufactured Goods: An Econometric Investigation (Amsterdam: North-Holland).
Baldwin, R. (2000), ‘Regulatory Protectionism, Developing Nations and a Two-Tier World System’,
Centre for Economic Policy and Research Discussion Paper No. 2574 (London).
Ballance, R. H., H. Forstner and T. Murray (1987), ‘Consisency Tests of Alternative Measures of
Comparative Advantage’, Review of Economics and Statistics, 69, 1, 157–61.
Baller, S. (2007), ‘Trade Effects of Regional Standards Liberalization. A Heterogeneous Firm’s
Approach’, World Bank Policy Research Working Paper No. 4124 (World Bank).
Chen, M. X. and A. Mattoo (2008), ‘Regionalism in Standards: Good or Bad for Trade?’, Canadian
Journal of Economics, 41, 3, 838–63.
European Commission (1998), Technical Barriers to Trade, Volume 1 of Sub-Series III:
Dismantling of Barriers of the Single Market Review (Luxembourg: Office for Official
Publications).
European Commission (2002), Report from the Commission to the Council, the European
Parliament and the Economic and Social Committee. Second Biennial Report on the
Application of the Principle of Mutual Recognition in the Single Market (Brussels: European
Commission).
European Commission (2003), The Economic Impact of Mutual Recognition Agreements on
Conformity Assessment: A Review of the Costs, Benefits, and Trade Effects Resulting from the
European Community MRAs Negotiated with Australia and New Zealand, Final Report
(Brussels: European Commission).
Freedom House (various years), Freedom in the World: Political Rights and Civil Liberties (New
York: Freedom House).
Ganslandt, M. and J. R. Markusen (2001), ‘Standards and Related Regulations in International
Trade: A Modeling Approach’, in K. E. Maskus and J. S. Wilson (eds.), Quantifying the
Impact of Technical Barriers to Trade: Can It Be Done? (Ann Arbor: University of Michigan
Press).
Henson, S. and W. Mutillah (2004), ‘Kenyan Exports of Nile Perch: Impact of Food Safety
Standards on an Export-Oriented Supply Chain’, World Bank Policy Research Working Paper
No. 3349 (World Bank).
Jaffee, S. and S. Henson (2004), ‘Standards and Agro-Food Exports from Developing Countries:
Rebalancing the Debate’, The World Bank Policy Research Working Paper No. 3348
(World Bank).
Maskus, K. E. and J. S. Wilson (2001), ‘A Review of Past Attempts and the New Policy Context’,
in K. E. Maskus and J. S. Wilson (eds.), Quantifying the Impact of Technical Barriers to Trade:
Can It Be Done? (Ann Arbor: University of Michigan Press).
Maskus, K. E., T. Otsuki and J. S. Wilson (2001), ‘An Empirical Framework for Analysing
Technical Regulations and Trade’, in K. E. Maskus and J. S. Wilson (eds.), Quantifying the
Impact of Technical Barriers to Trade: Can It Be Done? (Ann Arbor: University of Michigan Press).
Melitz, M. (2003), ‘The Impact of Trade on Intra-Industry Reallocations and Aggregate Productivity Growth’, Econometrica, 71, 6, 1695–725.
Moenius, J. (2004), ‘Information versus Product Adaptation: The Role of Standards in Trade’,
Kellogg School of Management Working Paper (Northwestern University).
Pelkmans, J. (2003), ‘Mutual Recognition in Goods and Services: An Economic Perspective’, European Network of Economic Policy Research Institutes, Working Paper No. 16.
Stanford, L. (2002), ‘Constructing “Quality”: The Political Economy of Standards in Mexico’s
Avocado Industry’, Agriculture and Human Values, 19, 4, 293–310.
Swann, P., P. Temple and M. Shurmer (1996), ‘Standards and Trade Performance: The UK
Experience’, Economic Journal, 106, 1297–313.
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009
574
GALINA AN AND KEITH E. MASKUS
Sykes, A. O. (1995), Product Standards for Internationally Integrated Goods Markets (Washington,
DC: Brookings Institution).
Vancauteren, M. and D. Weiserbs (2003), ‘The Impact of Removal of Technical Barriers to Trade
on Border Effects and Intra-Trade in the European Union’, Working Paper (Université
Catholique de Louvain).
Wilson, J. S. and T. Otsuki (2004), ‘Standards and Technical Regulations and Firms in Developing
Countries: New Evidence from a World Bank Technical Barriers to Trade Survey’, World Bank
Working Paper (World Bank).
Wooldridge, J. M. (2003), Introductory Econometrics: A Modern Approach, 2nd edn (Mason, OH:
Thomson South-Western).
© 2009 The Authors
Journal compilation © Blackwell Publishing Ltd. 2009