RESHAPING INSTITUTIONS: EFFECTS OF ICT ON BILATERAL TRADE
By
Tithi Chattopadhyay

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Media and Information Studies – Doctor of Philosophy
2013	
  

	
  
	
  

ABSTRACT
RESHAPING INSTITUTIONS: EFFECTS OF ICT ON BILATERAL TRADE
By
Tithi Chattopadhyay
Internet and telecommunication technologies have changed the way people communicate and do
business across the globe. This technology revolution has been visible not only in high income
countries, but also in developing and lower income countries. It has impacted business
transactions between these groups of countries. This dissertation studies the effect of the Internet
on bilateral trade. A few studies have looked at the impact of the Internet on bilateral trade. This
study offers a novel analysis of its interactions with the institutional underpinnings of trade by
examining the effects of Internet use on trade in industries that require varying degrees of
relationship-specific investments.
Industries can be categorized by their level of dependence on investments that are
relationship-specific. A higher dependence on relationship-specific investments is generally
associated with higher levels of risk in the transaction process. Countries with more stable and
differentiated institutional arrangements are generally capable of better contract enforcement,
and have often been the preferred trading partner in such industries. This study analyzes whether
Internet use impacts industries with a higher need for institutional arrangements differently from
those that have a smaller need for these institutional arrangements. Internet is able provide access
to different kinds of information (verbal and visual) almost instantly, and this plays a role in
reducing transaction costs (traditionally managed through the use of institutional arrangements
between two countries). The underlying conceptual framework of this study takes into account
the affordances of the technology based on past literature. The effect of these affordances or

	
  
	
  

features on coordination costs, information costs and productivity costs is analyzed to develop
the hypothesis for this study. The adopted approach analyzes the impact of internet use on
institutional intensive sectors by understanding the relationship of these different technological
affordances on transaction and production costs in an industry.
A pooled OLS estimation and panel data estimation using gravity specifications of
international trade was performed on 25 industries and for 72 countries between the time periods
2002 to 2010. The findings indicate that higher Internet use by the exporting country has a
positive impact on trade with the US in sectors that require higher institutional arrangements.
This has important policy implications because countries with relatively poorer institutional
arrangements can now benefit from the use of Internet, as the technology is able to reduce
transaction and information costs to some degree. It also implies that the use of Internet or
broadband could reduce the need for stronger institutional arrangements in industries with
relationship-specific investments.
The findings from this study also suggest that industries with low institutional intensities
are positively impacted by the use of the Internet because of its effect on productivity
particularly in markets that are highly competitive in nature. The underlying cause and reason for
impact on industries with a lower need for institutional arrangements is a little more ambiguous
because of multiple factors that can influence the outcome including the nature of the industry
(competitive or not) and the existence of other potential importers in the market.

	
  
	
  

ACKNOWLEDGEMENTS
The past few years of being a PhD student have taught me life skills beyond the obvious
academic ones. A lot of people have helped me and influenced me during this time and have
made my experience deeper and richer. I have met a lot of inspiring people that have challenged
me to broaden my horizons and change my worldview. This section is dedicated to those
individuals that I owe my deepest gratitude towards.
First of all, I would like to thank Dr. Johannes Bauer for mentoring me and keeping me on track
throughout the course of my PhD program. His patience and guidance helped me through this
experience that can be somewhat challenging at times. I would also like thank the tireless efforts
of my committee members Dr. Steve Lacy, Dr. Steve Wildman and Dr. Kurt DeMaagd. I have
been fortunate to have had worked with all of them on other projects besides this dissertation,
and I have learnt a lot from each one of them. I am also very grateful to my loving family. I
would first of all like to thank my parents for pretty much everything! They have supported me
and inspired me even though they were miles and time zones apart. I would also like to thank my
extended family and friends (the list is long so I will not mention everyone) that have encouraged
me and kept me sane through all this. Finally, I would like to thank Shishir for the million big
and little things he has done for me during this process! I really do not think I could have begun
and finished this without the support of all these wonderful people.

iv	
  
	
  

TABLE OF CONTENTS

LIST OF TABLES

vi

LIST OF FIGURES

vii

1. INTRODUCTION

1

2. LITERATURE REVIEW
2.1 Internet and communication technology and economic productivity
2.2 ICT and its effect on trade
2.3 Importance of Institutions in trade
2.4 Research Questions and Contribution of this Study

8
9
13
17
21

3. CONCEPTUAL FRAMEWORK
3.1 Affordances of the Internet (broadband technology)
3.2 ICT and its impact on trade costs
3.2.1 Effect on Transaction Costs
3.2.2 ICT and its impact on aggregate productivity
3.3 Institutional arrangements in trade and broadband technology
3.4 Impact of Internet technology: Overall conceptual framework
3.5 Hypothesis to be tested

23
24
27
29
33
36
38
44

4. METHODOLOGY

48

5. EMPIRICAL EVIDENCE

55

6. CONCLUSION

71

APPENDICES
Appendix A: Panel Regression Results
Appendix B: OLS Regression Results
Appendix C: Summary Statistics

75
76
98
123

REFERENCES

167

v	
  
	
  

LIST OF TABLES
Table 1 Affordances of broadband and effect on coordination and information costs

39

Table 2 Data Sources

52

Table 3 List of countries

55

Table 4 Industries and their institutional intensities

57

Table 5 Summary Statistics of Internet Hosts by Country

59

Table 6 Results of OLS Estimation

62

Table 7 Results of Panel Data Estimation

65

Table A.1 321 Wood Products

76

Table A.2 336 Transportation Equipment

77

Table A.3 313 Textiles and Fabrics

78

Table A.4 323 Printed Matter and Related Products

79

Table A.5 326 Plastics and Rubber Products

80

Table A.6 324 Petroleum and Coal Products

81

Table A.7 322 Paper

82

Table A.8 211 Oil and Gas

83

Table A.9 511 Newspaper

84
vi	
  

	
  

Table A.10 327 Nonmetallic Minerals

85

Table A.11 333 Machinery

86

Table A.12 112 Livestock and Livestock Products

87

Table A.13 316 Leather & Allied Products

88

Table A.14 337 Furniture and Fixtures

89

Table A.15 311 Food and Kindred Products

90

Table A.16 114 Fish, Fresh/chilled/Frozen and Other Marine Products

91

Table A.17 332 Fabricated Metal Products

92

Table A.18 335 Electrical Equipment, Appliances & Components

93

Table A.19 334 Computer and Electronic Products

94

Table A.20 312 Beverages & Tobacco Products

95

Table A.21 315 Apparel and Accessories

96

Table A.22 111 Agricultural Products

97

Table B.1 324 Petroleum and Coal Industry

98

Table B.2 111 Agriculture Products

99

Table B.3 315 Apparel and Accessories Industry

100

Table B.4 312 Beverage and Tobacco Products

101

vii	
  
	
  

Table B.5 325 Chemical Industry

102

Table B.6 334 Computer and Electronics Industry

103

Table B.7 335 Electrical Equipment, Appliances and Components

104

Table B.8 332 Fabricated Metal Products

105

Table B.9 114 Fish and Marine Industry

106

Table B.10 311 Food and Kindred Industry

107

Table B.11 113 Forestry Products Industry

108

Table B.12 337 Furniture and Fixtures

109

Table B.13 316 Leather and Allied Products Industry

110

Table B.14 112 Livestock Industry

111

Table B.15 333 Machinery Industry

112

Table B.16 327 Nonmetallic Mineral Products

113

Table B.17 511 Newspaper Industry

114

Table B.18 211 Oil and Gas Industry

115

Table B.19 322 Paper Industry

116

Table B.20 326 Plastics Industry

117

Table B.21 323 Printed Matter and Related Products

118

viii	
  
	
  

Table B.22 313 Textile Industry

119

Table B.23 336 Transportation Industry

120

Table B.24 321 Wood Industry

121

Table B.25 212 Minerals and Ore

122

Table C.1 111 Agriculture Basic Statistics

123

Table C.2 111 Agriculture Correlation and Covariances

124

Table C.3 112 Livestock & Livestock Products Basic Statistics

125

Table C.4 112 Livestock & Livestock Products Correlations and Covariance

126

Table C.5 114 Fish, Fresh/Chilled/Frozen and Other Marine Basic Statistics

127

Table C.6 114 Fish, Fresh/Chilled/Frozen and Other Marine Correlations and Covariances

128

Table C.7 211 Oil and Gas Basic Statistics

129

Table C.8 211 Oil and Gas Correlations and Covariance

130

Table C.9 311 Food and Kindred Products Basic Statistics

131

Table C.10 311 Food and Kindred Products Correlations and Covariances

132

Table C.11 312 Beverages and Tobacco Products Basic Statistics

133

Table C.12 312 Beverages and Tobacco Products Correlations and Covariances

134

Table C.13 313 Textiles and Fabrics Basic Statistics

135

ix	
  
	
  

Table C.14 313 Textiles and Fabrics Correlations and Covariances

136

Table C.15 315 Apparel and Accessories Basic Statistics

137

Table C.16 315 Apparel and Accessories Correlations and Covariances

138

Table C.17 316 Leather and Allied Products Basic Statistics

139

Table C.18 316 Leather and Allied Products Correlations and Covariances

140

Table C.19 321 Wood Products Basic Statistics

141

Table C.20 321 Wood Products Correlations and Covariances

142

Table C.21 322 Paper Basic Statistics

143

Table C.22 322 Paper Correlations and Covariances

144

Table C.23 323 Printed Matter and Related Products Basic Statistics

145

Table C.24 323 Printed Matter and Related Products Correlations and Covariances

146

Table C.25 324 Petroleum and Coal Products Basic Statistics

147

Table C.26 324 Petroleum and Coal Products Correlations and Covariances

148

Table C.27 326 Plastic & Rubber Products Basic Statistics

149

Table C.28 326 Plastic & Rubber Products Correlations and Covariances

150

Table C.29 327 Non Metallic Mineral Ores Basic Statistics

151

Table C.30 327 Non Metallic Mineral Ores Correlations and Covariances

152

x	
  
	
  

Table C.31 332 Fabricated Metal Products Basic Statistics

153

Table C.32 332 Fabricated Metal Products Correlations and Covariances

154

Table C.33 333 Machinery, Except Electrical Basic Statistics

155

Table C.34 333 Machinery, Except Electrical Correlations and Covariances

156

Table C.35 334 Computer and Electronic Products Basic Statistics

157

Table C.36 334 Computer and Electronic Products Correlations and Covariances

158

Table C.37 335 Electrical Equipment, Appliances & Components Basic Statistics

159

Table C.38 335 Electrical Equipment, Appliances & Comps Correlations and Covariances

160

Table C.39 336 Transportation Equipment Basic Statistics

161

Table C.40 336 Transportation Equipment Correlations and Covariances

162

Table C.41 337 Furniture and Fixtures Basic Statistics

163

Table C.42 337 Furniture and Fixtures Correlations and Covariances

164

Table C.43 511 Newspapers, Books & Other Published Matter Basic Statistics

165

Table C.44 511 Newspapers, Books & Other Published Matter Correlations and Covariances 166

xi	
  
	
  

LIST OF FIGURES
Figure 1 Impact of Broadband on bilateral trade at an industry level

39

Figure 2 OLS Estimate vs. Industry Institutional Intensities

65

Figure 3 Panel Estimates vs. Industry Institutional Intensities

65

xii	
  
	
  

1. INTRODUCTION
Over the last few decades, Internet and telecommunication technologies have been adopted
across the globe. Unlike other technology revolutions, this growth has been visible not only in
high income countries but also in developing and lower income nations. This large scale growth
in communication across the globe has had multi-faceted impacts. Information and
communication technologies have influenced interpersonal communications, social interactions,
and are transforming economic activities at local, national and international levels.
ICTs affect business transactions in several ways. On an international level, information
and communication technology can possibly have an impact on market-driven activities
(international trade), activities that involve a more hierarchical structure of governance (some
foreign direct investments, multinational corporations etc.) and activities that are more
contractual in nature (licensing agreements etc.). The above mentioned kinds of economic
structures are different in their coordination structures and hence have different costs and risks
involved. The impact of information and communication technology on these cross-border
economic activities varies with the differences in their respective organizational structure, cost
structures and risks.
ICT benefits market-driven activities like international trade that are potentially affected
by problems of asymmetric information by reducing search costs and basic logistics of trade are
easier to monitor with the use of these technologies. More hierarchical organizations like
multinational corporations have a more complex structure and require a higher level of coordination. These hierarchical economic activities typically also exhibit a high degree of asset
specificity that might require large investments. ICTs have multiple affordances that improve
coordination and transfer of information within these organizations. Use of video conferencing,
1	
  
	
  

centralized information networks and other such applications within these organizations have
hugely reduced costs and improved efficiencies. Contractual economic relations, such as
licensing agreements, are typically less complex, but there are certain information asymmetries
that could result in poorer decisions about the licensing agreement. The use of ICT has also led
to a more awareness about cultures around the world. There is more information about different
kinds of markets making the global market more accessible to most countries. These are a few of
the many ways in which ICT can help reduce information asymmetries and improve decision
making. This study however, focuses only on the impact of these information and
communication technologies on the market-driven activity of international trade.
While there are several studies that look at the impact of the Internet on international
trade (Freund and Weinhold 2000, 2002, 2004; Berthelon and Freund 2008), this is a first
attempt to understand the impact of information and communication technologies on
international trade in different industrial sectors with varying levels of relationship-specific
investments. A measure of the level of relationship-specific investments in a particular industry
was constructed by Nunn (2007) and is termed “contract intensity” of that particular sector. In
other studies such as Levchenko (2011), contract intensity is also called “institutional intensity”
of an industrial sector, defined as the dependence of that particular industrial sector on
relationship-specific investments and contracts.
Traditionally, countries with better institutional quality have a comparative advantage
with respect to industries that involve relationship-specific investments and contracts
(Levchenko 2004, Antras 2007, Acemoglu et al. 2005, Costinot 2005). The institutional
dimension that is of interest in this scenario is the quality of contract enforcement in a country. In
the literature, there are several variables that have been used as indicators of the quality of

2	
  
	
  

contract enforcement in a country. These include the “regulatory quality index” calculated by the
World Bank (2004) and the “rule of law” from Kaufmann, Kraay and Mastruzzi (2003).
This study examines the effect of information and communication technologies on
different industries to analyze if these technologies have changed the trends in trade in more
institutionally intensive industries. While it is difficult to dispute that information and
communication technology has had some role to play in the way international trade has changed
over time, this study attempts to analyze whether the Internet influences institutionally intensive
industrial sectors (investments in these industries are relationship specific) more than other
sectors (sectors where investments are not as relationship specific), thereby indicating a
reduction in transaction and information cost associated with institutionally intensive industries.
Countries that have good institutional arrangements (better regulatory quality or better systems to
enforce contractual obligations) had an advantage while trading in such institutionally intensive
industries. Reduction of transaction costs not only benefits countries that have poorer
institutional arrangements but also benefits countries that already have higher institutional
quality by improving coordination and information transfer.
Internet and communication technologies have an impact on most kinds of economic
activities as they usually act as enablers of communication between individuals and
organizations, including those in different countries. Information and communication
technologies enable information acquisition, market access, export promotion and efficient
coordination. Costs associated with these activities have been identified as important
impediments to international trade between two countries in a particular industry (Samiee and
Walters, 1990 and Samiee 1998). Besides being able to reduce information asymmetries and
transaction costs of these economic activities explicitly, Internet and communication

3	
  
	
  

technologies could lead to better awareness of national cultures and this could lead to more
global cultural integration or fragmentation in some cases. In an organizational context, greater
integration has frequently been linked to greater firm performance (Barney 1991, Chalmeta et al..
2001). Hence, we expect that the integration enabled by Internet and communication
technologies would also impact the overall production of the country and hence have a positive
impact on its international trade. This study explores the role of information and communication
technology in trade by reducing information asymmetries and transaction costs especially in
industries that require relationship-specific investments. However, these technologies may also
help highly competitive, but institutionally less intensive industries by reducing search costs and
enabling better coordination.
Since 1980, the world has seen a disproportionate rise in trade volumes as compared to
the increase in world real GDP (Berthelon and Freund 2008). New industries have come into
existence and new trading partnerships are being entered into all over the world. It is possible
that some of the factors that were traditionally responsible for facilitating better trade between
some countries are now affected by the use of information and communication technology. The
deluge of technical features in current-day technology is capable of impacting not only economic
costs, but also cultural factors that traditionally played a role in international trade.
This study explores if broadband access influences bilateral trade flows in more
institutionally intensive sectors more heavily than others. Better institutional quality of a country
has been important for those countries trading in institutionally intensive sectors as these
institutions by themselves are a critical factor in increasing the country’s comparative advantage
in that sector (Levchenko 2007, Nunn 2007). Better institutions enable more reliable transactions
and lower coordination costs.

4	
  
	
  

This study proposes that the availability of broadband technology in the last decade may
have also played an important role in changing trends in bilateral trade by reducing certain costs
associated with it. Costs associated with trade are broadly categorized as production costs,
coordination costs (distance related and others) and information costs. Different industries are
characterized by varying levels of these costs; hence one could hypothesize that the different
affordances of the Internet have varying effects on trade in different industries. It is of interest to
take a closer look and examine some whether certain sectors affected by the use of the
technology more than others. Different sectors also have different levels of institutional intensity,
and some manage to mitigate a significant amount of their production and transaction costs by
adopting information and communication technologies. In other words, this study is interested in
exploring if the adoption of information and communication technologies affects certain industry
sectors more than others?
This study is a first of its kind to analyze the impact of Internet technologies on trade
particularly on industrial sectors that have a high institutional intensity. The impact of
institutions on trade has been studied before (McLaren 2000, Grossman and Helpman 2002,
2003, 2005, Levchenko 2004, Antras 2005, Acemoglu et al. 2005, Costinot 2005), but this is a
first study that looks at how the use of Internet technology affects trade, with a particular
emphasis on institutionally intensive sectors. It is interesting to investigate whether the Internet
enhances the existing institutional arrangements in countries that are known to have good
institutional arrangements, and further if the technology is able to provide some affordances of
good institutions in those countries that lack these arrangements. In other words, it is intriguing
to explore whether Internet technology is able to provide similar institutional affordances to

5	
  
	
  

countries that lack a high quality of contract enforcement, and give them a chance to compete in
trade sectors that are more contractually intensive.
One motivation for this study is to examine whether the use of Internet affects trading
firms in those countries where the government is rather sluggish in improving overall
institutional arrangements along with influencing those countries that already have these
arrangements in place. It must be kept in mind that the Internet benefits trade in multiple ways
and could potentially also benefit sectors that are not institutionally intensive, especially sectors
that are highly competitive.
The main contribution of this study is that it gives us an understanding of how Internet
affects bilateral trade in different sectors by impacting the underlying costs in these sectors. This
industry wise break up according to institutional intensities provides us with some insight into
what sectors that benefit from the use of Internet technology in trade and why. Empirically, the
dissertation analyzes the effect of broadband technology on bilateral trade in the US between
2002 and 2010. The study includes 22 industries characterized by different institutional
intensities Nunn (2007). The study uses the traditional international trade gravity equation which
is essentially designed around the concept that a country is more likely to trade with another
country that is physically closer and larger, than a country that is smaller and further away. In
this study, the gravity equation is modified to account for other factors important to trade
including the size of the information and communication sector of the respective trading
countries.
The study finds evidence that the use of ICT positively affects both sectors with high and
low institutional intensity. The use of the technology plays a role in reducing transaction costs
and production costs. Findings from this study have important policy implications for the use of

6	
  
	
  

Internet technology for firms in countries with poorer institutional arrangements. Broadband
Internet access increases their ability to compete in sectors that they traditionally had a
disadvantage in. This study also has important policy implications for the use of Internet
technology in industries that tend to be highly competitive in nature.
The next section of the dissertation will review the existing literature in this area of
research. The following two sections provide a conceptual framework for the study as well as
define the methodological approach. The final two sections look at the empirical evidence and its
policy implications.

7	
  
	
  

2. LITERATURE REVIEW
This dissertation studies the impact on overall bilateral trade flows as a result of technology
adoption and use. It is interesting to study the effect of information and communication
technologies on specific sectors or economic activities, and evaluate whom these technologies
have benefited. This can also inform us on whether and how countries and industries should use
these technologies to improve coordination, reduce risk and gain information in this highly
interconnected and globalized world. Even in trade, it is possible that some countries benefit
more than some others by using the technology because of the inherent nature of industries that
they invest in or because of the existence of other country level factors (or lack of
complementary resources) that make the adoption and efficient use of the technology difficult. In
this section, we review literature that has investigated the economic impact of ICT. In the first
section, we examine the broader literature that addresses the impact of ICT on overall economic
productivity, which has repercussions for trade. In the next section, we narrow down to and
review the literature that addresses the impact of ICT on one specific economic activity “trade”.
Broader findings that apply to the overall impact on economic productivity also apply to trade.
Besides economic productivity, ICT also additionally impacts another important factor
influencing trade called “transaction costs” (discussed in the third section). Finally, the last
section highlights contributions of this dissertation and the motivations derived from past
literature.

8	
  
	
  

2.1 Internet and Communication Technology and Economic Productivity
Since the coining of the term “Solow’s paradox” in which Robert Solow argued that you can see
the computer everywhere except in productivity statistics, there have been several studies that
have made an attempt to explain this observation. Solow’s paradox was coined because a key
finding was that since 1973 the residual economic growth had slowed down just when IT
investments had risen. Residual growth is the spillover after growth of all inputs is taken into
account (Robert Solow 1957). Several studies including Brynjolfsson and Yang (1996) did not
find a correlation between IT and productivity. They largely attributed this to lack of data and
inability of analytical methodologies to capture the impact quantitatively. A recent study by
Forman et al. (2010) found evidence for increasing wage inequality with the adoption of the
Internet.
However the paradox was largely demystified when scholars like Brynjolfsson looked
into the reasons for the paradox and found them to be associated with measurement errors, lags
in productivity and the lack of accountability of quality and variety changes brought about by
these technologies. Brynjolfsson and Hitt (2000) find that complementary investments in
institutions are required for benefits from IT investments to be fully realized. Other studies have
tried to explain the reason for this paradox by defining conditions under which productivity is
most likely to increase (Hughes and Morton 2005).
Some studies find that the impact of information technology depends on the availability
of complementary resources. Some scholars (Brynjolfsson and Hitt 2000 and Draca et al. 2006)
find that traditional studies confirming Solow’s paradox do not account for the existence of these
complementary resources. Others (Dedrick 2003) refuted the productivity paradox in both firm
level and country level studies and explained that complementary investments accounted for

9	
  
	
  

difference between organizations. This was supported in some studies (Melville 2004) that also
find that the impact of IT investments depends on the level of complementary environment.	
  
Complementary environment refers to other changes in organizational structure that make work
environment more conducive to the use of IT (e.g. adequate worker training in IT). Some studies
(Bresnahan and Trajtenberg 1995) advocated the idea that information and communication
technology had some impact on productivity when they used a transaction cost perspective to
discuss how general purposes technologies can act as ‘engines of growth’. Similarly, some
(Hubbard 2003) explored how Internet and other ICT products help increase productivity by
improving the matching process between people and products. While some other scholars like
Athey and Stern (2000) studied productivity effects in specific industries (in this study the
researchers looked at how information technology impacts the health care industry in specific).
The above studies find that the productivity paradox was largely a result of empirical limitations,
but in some cases the benefits of the technology were not realized because of a lack of
complementary investments that are required to make the most of the technology. Van Aek and
Inklaar (2005) find the impact of ICT on productivity is also time dependent as countries and
economies take time to realize the benefits of the technology and adapt accordingly. Similarly,
they find that different economic sectors benefit from the technology differently because of the
inherent characteristics of the different sectors. Despite the typical endogeneity challenges that
are involved in measuring the impact of technology on productivity, Crandall, Lehr and Litan
(2007) find sufficient evidence that ICTs do impact productivity growth. A literature review of
broadband’ s impact on productivity (Holt and Jamison 2009, Kretschmer 2012) found that the
technology had a positive impact on various productivity indicators, but it was hard to measure

10	
  
	
  

the impact of the technology with precision because of endogeneity issues that studies like these
typically tend to have.
While it is difficult to evaluate the overall welfare gains from the use of information and
communication technology, what we understand from previous literature is that the technology
may help reduce costs in some sectors and increase productivity in some industrial sectors more
than in others. While the use of information and communication technologies may result in net
gains for consumers and some industrial sectors, there may be some sectors that for several
reasons are unable to use the technology effectively for their benefit.
In a literature review about firm-level productivity effects, Pilat (2004) explores the
different factors/ complementary resources that need to be considered while studying the impact
of ICT on industry level impact. The factors that impact ICT at a firm level help understand why
some sectors and some countries perform better with ICT investments. It depends on the level of
ICT related skills that already exist and if they don’t exist the technology takes longer to have an
impact. The impact of ICT also depends on several organizational factors like the amount of
training provided, the firm’s experience with innovation and its use if advanced business
practices and the restricting of the organization that should accompany ICT use for its benefits to
be realized. There is also some correlation between the size and age of a firm on how it utilizes
ICT to improve productivity. Larger and newer firms tend to use the technology better and hence
see greater benefits. This firm level analysis does aggregate to overall impacts on the economy.
Gretton et al. (2004) find considerable impacts of ICT while using firm level data in Australia.
Hampell et al. (2004), Arvanitis (2004), Baldwin (2004) find that labor productivity increases are
correlated with ICT use in German, Dutch, Swiss and Canadian firms. The above studies capture
how these technologies have not only been able to cut down some transaction costs and

11	
  
	
  

information costs at the firm and industry level, but they have also created new markets and
products.

12	
  
	
  

2.2 ICT and its Effect on Trade
One area where information and communication technologies should have significant impact on
productivity is trade. Internet and communication technology can affect trade by acting as an
economic enabler as well as by integrating the country socially, economically and culturally with
the rest of the world. Internet and communication technology can act as economic enabler in two
possible manners. Information and communication technology have the ability to reduce
transaction costs in trade, further Internet technology has also been rather successful in creating
new opportunities that did not exist before. While both trade in goods and services are impacted
by the availability of Internet and communication technologies, one must note that trade in goods
is impacted largely by the reduction of non-distance related co-ordination and information costs.
For example, some products have huge transportation costs that comprise a major chunk of their
overall costs, and those costs probably cannot be substantially mitigated by ICT. ICT could
improve the logistics involved in the transportation process in order to ensure that perishables
and other time sensitive goods are delivered in time, but the actual costs of transportation still
continue to exist. Hence some distance related costs are mitigated because of increased coordination and reduced transaction costs, but the cost of fuel and the actual transportation process
are not impacted by the technology.
Trade in services just like trade in goods benefits from the lower transaction and
information costs that Internet technology enables, but there is a bigger impact of information
and communication technology on the trade of services because it can mitigate distance effects in
some sectors completely and make services more tradable (although legal restrictions also have
an impact). Internet technology adoption results in the creation of new opportunities because
temporal and location boundaries now cease to exist because of the unique properties of the new

13	
  
	
  

technology (Freund and Weinhold 2002). Some studies (Choi 2010) have found that access to
the Internet almost doubles the trade in services. The widespread use of the technology has also
given rise to a new class of jobs requiring the skills to set up, maintain, upgrade, modify and
innovate different uses of the technology itself.
Freund and Weinhold (2000) studied Internet technology specifically and found the
following: first, trade benefits from information and communication technology because the risk
of sunk costs associated with trade can be largely reduced. This is because the use of information
and communication technology results in better information and co-ordination of the trade
process. For products that involve relationship-specific investments, informed decision making
reduces the risk of these investments as the trading parties involved are better informed about
each other’s expectations, and the co-ordination process involved in the supply chain is more
regulated. They also found that for developing and new markets, Internet technology reduces the
effect of historical linkages because better communication enables higher levels of trust and
mutual understanding of the environment of business and partnership. They found that countries
with the fewest past trade linkages (most likely the poor countries) have the most to gain from
the Internet as it gives them the opportunity to build new relationships and develop new
industries. If these countries had competitive advantages in particular industries but were unable
to enter the market because of historical linkages, the information and communication
technology provides them with tools to establish a trade relationship that is characterized by a
higher level of information and co-ordination.
Several scholars have investigated further, and have found unique trends in trade when
access to internet is increased. Clarke and Wallsten (2010) found that increases in internet access
in developing countries increases trade but only exports to other developed countries. The

14	
  
	
  

widespread use of Internet technology does create some winners and some losers in the process.
Countries that do not use the technology effectively to improve their processes are rendered more
vulnerable. Similarly, countries that earlier benefited because of historical linkages may now
have to face more competition. Countries that have some competitive disadvantage but could still
trade with neighboring countries will now find it harder to find a market for their products. There
are several studies that have looked at the effect of ICT in specific countries or regions (Dimelis
and Papaioannou 2011, Martinez 2003, Rana 2012, Wong Pellan 2012). There are several studies
that have studied the impact of ICT on a variety of outcomes including the effect on trade secrets
(Majdali 2000), international conflicts (Mansfield and Pollins 2009), polarization of skills
demand (Van Reenan 2012) and impact on culture (Affortunato et al 2013). Some studies in the
past have looked at the effect of Internet on specific industries like health care, food and
beverage and so on (Fleming 2011, Chung et al 2010, Corso and Gastaldi 2010). Whatever, the
scenario it is likely that information and communication technologies have changed some trading
patterns.
While there have been studies showing impact of trade on particular countries/regions,
particular outcomes and particular industries, this study is the first one to analyze the impact of
internet on bilateral trade on a list of industries specified by their institution intensities. In other
words, we are particularly interested in the effect of Internet technology on trade to analyze if
industries with high relationship-specific needs are impacted in any way. These industries
traditionally tend to be very risky and there is a significant amount of trust involved in these
trade relationships. It is possible that some of these industries are directly related to the ICT
sector and their productivity is directly impacted by ICT investments and use. However, this
study attempts to also include other relationship specific manufacturing and agricultural sectors

15	
  
	
  

in the sample to analyze if there is also a reduction in contractual transaction and information
costs.

16	
  
	
  

2.3 Transaction Costs and the Importance of Institutions
The concept of transaction cost was introduced by Ronald Coase (1937) who discussed why it
was not efficient for all exchanges to be organized as markets and how firms played an important
role in enabling exchanges and improving transactions. The argument was further elaborated by
Williamson (1975, 1985) when he specified why transaction costs exist and what were the risks
involved when transaction costs were high. The overall argument was that transaction costs
existed because human beings were limited with respect to knowledge, foresight, skill and time.
Certain institutional arrangements were required in order to avoid incomplete contracts and
enable exchanges that would have been impossible otherwise. He goes on to state that there are
essentially two types of risks involved in carrying out any exchange. One is relational risk that
arises because of information asymmetries, limited foresight, lack of skills, cognitive abilities
and even purely from coordination problems that arise because of communication issues. The
next type of risk is environmental risk and does not pertain to the relationship between the two
parties but exists because people have very limited knowledge about their environment as well as
the environment of the party they are conducting business with. For example, people may not be
aware of policy changes made by the government in their own country leave alone the country of
the other business party involved.
Institutions are particularly important in industries that require relationship-specific
investments and contracts. Several studies have depicted the importance of contract enforcement
in these industries (Williamson 1985, Grossman and Hart (1986) and Klien, Crawford and
Alchian 1978). The incomplete contracts literature shows that the absence of proper institutional
arrangements could potentially lead to a well-known hold-up problem. Since investments in
relationship specific contracts involve investments that are specific to the contract, there exist
certain inefficiencies because of investment irreversibility. Parties often have to enter into
17	
  
	
  

elaborate contracts that prevent hold up problems or at least distribute the losses due to hold up
problems by assigning appropriate property rights. Hence institutional arrangements become
rather important in such industries and particularly in international trade. Levchenko (2008)
states that institutions are a source of comparative advantage in industries that are more
relationship-specific. Some countries have better institutions and are therefore able to attract
more business in more relationship specific sectors. It is possible that the Internet could play an
important role in reducing the importance of these institutional arrangements in more
relationship specific or institutionally intensive sectors.
Keeping the basics of transaction cost theory, it is evident that ICT can reduce the burden
from traditional institutions in a country that often play a critical role in reducing transaction
costs in an international market exchange. Given the nature of ICT there is no doubt that it
certainly has properties that reduces information asymmetries, improves coordination efforts,
enables the provision of better communication technology at a very low marginal cost and
transcends the boundaries of time and location to reduce transaction costs significantly. Both
environmental and relational risks are significantly lowered with the use of Internet technology,
which therefore should have a positive effect on exchanges that take place across boundaries.
As discussed earlier, there is rich economic literature that has shown that there is a
significant gain in productivity from the use of Internet technology. Hubbard (2003) discusses
how Internet and other ICT products help increase productivity by improving the matching
process between people and products. In today’s world where there is so much variety within and
across nations. Internet and communication technology brings the most closely matching
products to the buyer thereby leading to a productivity gain. The faster and easier search process
results in inventory savings as well. The existence of various directories and search engines and

18	
  
	
  

communication technologies gives opportunities to the less established exporters to gain
information about market conditions.
Hamill and Gregory (1997), Hoffman and Novak (1996), and Athey and Stern (2002)
look at how information technology impacts the health care industry in specific. They look at
lower transaction costs that have been achieved by the use of Internet technology because of an
improvement in the speed of processing information. The speeding up of processes can reduce
opportunity costs of both time and material by reducing market uncertainties. They find that in
the medical industry prevention of lagged information produced economic gains that covered
85% of their costs. Paul David (1990) refers to the same thing in terms of ‘amenities’ provided
by technology. In his paper, he goes on to look at the problem of productivity paradox that might
arise because of the use of a new technology. This is particularly relevant to our study as we do
not know whether Internet technology actually leads to productivity gains in trade in all sectors
equally, or if only some sectors that are inherently capable of restructuring their businesses to use
the technology most efficiently benefit. In this respect Chun et al. (2005) uses Schumpeter’s
creative destruction approach to explain how newer competitive firms use Internet and other
information technologies to make more established firms obsolete. Even a non-exporting firm
can enter the market in some industries if it is able to use the Internet and create a website to
reach out to buyers in different parts of the world thereby creating a competitive advantage itself
(Samiee 1998). It is also possible that some industries might end up exporting lesser because ICT
improves search within a country and an exporting firm might end up supply more domestically
and a non-exporting firm might find more business in its own country.
From the above literature it is clear that Internet and communication technologies may be
useful in reducing transaction costs provided, some other basic economic and technological

19	
  
	
  

factors are taken care off. This is true particularly in the case of services where Internet
technology not only lowers transaction costs but also creates new opportunities by transcending
temporal and location boundaries (Freund and Weinhold 2002). In order for Internet and
communication technologies to have a positive impact on trade, it is essential that the
competitive advantage that ICT technology gives to some countries is larger than the competitive
disadvantage that some trading nations(either because their ICT usage is low or because of
increased competition) may now face (Samiee 1998). Secondly, even though the Internet and
communication technologies might enable easier transactions, trade is possible only if a country
decides to open its legal and regulatory framework to allow for these kinds of exchanges.
Thirdly, technological misuse like intellectual property violation or lack or privacy or security
could further cause impediments to trade and leading to higher transaction costs. Fourthly, it is
important to keep in mind that the proper adoption and use of the technology is possible only if
some basic socio-economic factors are taken into consideration and this may play an important
role in which countries benefit by the use of the technology in trade (Benjamin and Wiegand,
1995; Hoffman and Novak, 1996; Quelch and Klein, 1996).
Under the right set of conditions, information and communication technologies can
benefit both exporters and importers by lowering their firm level co-ordination costs, reducing
risks involved in international trade by enabling better information flows, by helping some
industries be more productive because of overall lower costs and by reducing cultural distance
between countries. It is also possible that some industries that are not capable of restructuring in
order to use the technology more effectively, may not see any changes. Additionally some
industries with higher transportation costs might just end up finding suppliers domestically
thereby reducing demand for exports.

20	
  
	
  

2.4 Research Questions and Contributions of this Study
In this study, an attempt is made to use organizational communication literature as well as
transaction cost theory to understand the mechanism by which information and communication
technology (particularly the Internet) can benefit trade in some industrial sectors more than
others. This study looks at how certain kinds of information and transaction cost in different
industry sectors are reduced with the use of certain information and communication
technologies, whereas certain other kinds of costs are not affected by information and
communication technologies. A de-compositional framework is used that examines how
affordances like audibility, visibility, co-presence, mobility, co-temporality, co-presence,
reviewability and revisability affect the important collaborative tasks of initiating conversation,
establishing common ground, and maintaining awareness of potentially relevant changes in the
collaborative trade environment. These affordances are beneficial in varying degrees to different
industrial sectors. A conceptual framework is developed to understand the contribution of the
Internet in different industrial sectors that are largely classified by the degree relationship
specific investments involved. While the reasons for the impact of Internet on trade in different
sectors maybe difficult to test empirically, the conceptual framework will provide us with some
understanding of the varying trends in different industries.
The main research question explored in this dissertation is how information and
communication technologies have impacted international trade flows in sectors that have varying
levels of relationship-specific investments. Do arguments of increased productivity due to ICT
also hold for international trade in some sectors? Different industries have inherent
characteristics that have varying levels of benefits from using the different of information and
communication technologies. Industries can be classified by the nature of investments made in
them. One particular kind of investment that involves high levels of transaction costs are
21	
  
	
  

relationship specific investments. Industries with higher levels of such investments require better
institutional arrangements to avoid hold up problems (Levchenko 2008). Does the Internet or
broadband technology provide some sort of alternative arrangements that reduce transaction and
information costs? A detailed study of the different industries between 2002 and 2010 can shed
light on which industries have benefited the most by the use of broadband technologies and
hence what the policy implications for different trading nations are.

22	
  
	
  

3. CONCEPTUAL FRAMEWORK
Information and communication technologies influence trade through their impact on multiple
levels of economic activity. These technologies have an effect on productivity on a more
aggregate industry and national level through their impact on the productivity at an individual
firm level. Furthermore, information and communication technologies have an impact on trade
because of its overall impact on transaction costs involved in trade. The following subsections
are set up in the following manner. In the first subsection, technical affordances of the
technology are discussed. Identifying these affordances is crucial in understanding how the
technology impacts trade costs (transaction and production), and this impact on trade costs is
discussed in the next subsection. The following subsection then goes into a discussion of
institutional arrangements and how technology impacts trade in different industries through its
impact on these institutional arrangements. In the final subsection, the overall framework is
developed by understanding the relationship between the technological affordances that
broadband provides and the factors that drive bilateral trade in different industries. Controlling
for the other factors that play a major role in bilateral trade, a conceptual framework is developed
to predict the relationship between broadband technology and trade in specific industrial sectors
characterized by their relationship specific investments.

23	
  
	
  

3.1 Affordances of Internet (Broadband Technology)
The Internet and in particular broadband technology has different technological affordances that
can help mitigate costs involved in bilateral trade in different ways. Based on a study by (Clark
and Brennan 1991 and Kraut el al 2002) the technological affordances of information and
communication technologies are described in Table 1.
The first one is ‘audibility’ where the participants can hear people and the sounds of an
environment. This enables quicker co-ordination as it allows for a simultaneous conversation to
take place, doubts and questions can be immediately addressed and there is a slightly higher
awareness about the general environmental and relational risks involved. The next affordance
that is discussed is ‘visibility’, and this allows participants to view people and objects in the
environment. The Internet is the only information and communications technology with this
affordance as it enables one to see pictures, or chat live via video chatting services online and
review documents and emails. Thirdly, Internet technology enables interacting entities to
experience ‘co-presence’, and this essentially allows participants to be mutually aware that they
share an environment. This affordance is slightly more difficult to comprehend in the context of
trade, but can be interpreted as the ability of both trading partners to understand the physical
environment in which they are carrying out the activity. Broadband Internet is able to offer this
affordance at the highest level by enabling tracking of a product, performing quality check and
so on by allowing for frequent conversations, exchange of documents and instantaneous updates
on the status a delivery or production. The fourth affordance is ‘mobility’, which is a feature that
enables participants to move around while sharing an environment (or exchanging information
about it). Broadband Internet is becoming relatively more mobile these days with the availability

24	
  
	
  

of wireless service everywhere, and the ability of participants to access information from any
computer that is connected to network service provider.
‘Co-temporality’ is an affordance that enables participants to be present simultaneously. This
affordance is offered by broadband Internet to some degree, but requires individuals to use
applications online that are specifically designed for this purpose. ‘Reviewability’ is an
affordance that ensures that messages do not fade over time and can be reviewed at any time.
Internet allows for this affordance at the highest level as documents can be exchanged between
the two trading parties in short amounts of time. The final affordance that we are interested in is
‘revisability’, which essentially allows for messages to be revised or checked before being sent.
Broadband Internet allows for revisability depending on the Internet application used, because
the nature of most voice/video services requires conversations to be instantaneous, hence making
‘revisability’ more difficult while using those mediums of communication that instantaneously
transfer voice and video messages. However, there exist several video and voice applications that
only host visual and verbal messages that are pre-recorded and such applications do allow for
‘revisability’.
Besides having these technological affordances it should be noted that broadband technology
has changed over the past decade contributing to a large extent on the effect that they have on
international trade. According to Clemons et al. (1993), the ways in which ICT usually change
over time is the following:
a. Decreasing cost of the technology: In the past decade markets have witnessed significant
drop in the cost of broadband access. The falling cost of the service will result in these
technologies having a bigger influence on trade over time.

25	
  
	
  

b. Improved quality of the technology: The quality of broadband technology has changed
over time thereby changing their technical affordances. While analyzing data over time it
is important to keep in mind the additional and improved affordances offered by these
technologies. For example, the mobile phone migrated from 2G to 3G and now to 4G and
mobile Internet is widely used all over the world; information technology has seen
increased processing capacity, better speeds of computer hardware, higher bandwidths,
Internet speeds and larger number of applications being developed for the web. These
changes need to be considered while understanding the influence of Internet technology
over time.
c. Better connectivity and standardization: Open systems, standardized hardware etc. enable
better coordination and dissemination of information. These changes have occurred
because of technological improvements as well as policy changes. These changes are
very important in enabling the use of broadband technology efficiently across countries
and hence have had an impact on how they influenced international trade.
d. Growing number of applications: Not only has broadband infrastructure access and
service improved over the last decade, but the number of applications that are developed
and hosted over the network has exponentially grown. The growth of these applications
have created added value for the network itself, and has given rise to huge network
benefits (e.g. more peer to peer collaborations, use in government transactions) that
impact economic activities all around the world.
In the following subsections we will discuss how these technological affordances impact
trade by impacting transaction costs involved in trade and overall aggregate productivity at the
industry level.

26	
  
	
  

3.2 ICT and its Impact on Trade Costs
It is important that before we begin our discussion on ICT and its impact on international trade,
we briefly discuss highlights of existing theories of international trade. The existing trade
theories can be divided into two main groups. The first group assumes perfect competition and
suggests that external economies of scale are the biggest driver of international trade. This group
of trade theories generally makes an assumption that goods and services produced by a country is
dependent on factors that are very country/region specific. Trade theories that fall into this group
are Adam Smith’s model of absolute advantage (Adam Smith 1776), the Ricardian model of
comparative advantage, and finally the Heckscher-Ohlin model that states that international trade
is determined differences in factor endowments. The second group consists of scholars that
looked into the concept of imperfect competition and internal economies of scale. This was
largely known as the new trade theory and demonstrated that countries trade with each other
even when there is no perfect competition or specialization. This is because consumers like
choice and consider varieties of the same product substitutable. Krugman (1980) and Helpman
and Krugman (1985) found that firms produce a variety of goods with increasing returns to scale
technology. This phenomenon was particularly helpful in explaining why two similar developed
countries enter into trade. ICT in particular impacts both internal and external economies of scale
by reducing production and transaction costs. In this study we use the gravity model derived
from the Cobb Douglas expenditure system (derived from the production function) that is
successful in empirically explaining trade patterns under both perfect specialization and
imperfect specialization assumptions. Irrespective of whether we assume perfect competition or
not, the use of the gravity equation helps empirically explain the trends in trade caused by ICT.
There are different ways in which ICTs affect economic activities. ICT is capable of
reducing costs in production and transaction related activities, and this influence of ICT has been
27	
  
	
  

studied widely. Clemon et al. (1993) argue that information and communication technology
because of its ability to lower coordination costs, reduces the need for more hierarchical
organizational systems and one can observe that this gives rise to over-all higher economies of
scale and better long term relationships between internal agents of the firm as well as external
ones. The authors also find that most studies do not account for increase in quality and variety
while studying productivity. Both increased quality and variety are very important to firms that
want to enter the export market. They also find that for ICT to result in firm productivity, some
amount of internal restructuring maybe required for a firm to fully realize all the benefits.
In this section, we will discuss the impact of ICT on the transaction costs involved in an
activity like international trade, as well as impact of the technology on the overall productivity of
industries. Transaction costs were first discussed by Ronald Coase, and are essentially the cost of
coordinating a market exchange in the presence of imperfect information. Two important aspects
of transaction costs are therefore the cost of coordinating the activity and the cost of gathering
information or making decisions in the presence of limited information.

28	
  
	
  

3.2.1 Effect on Transaction Costs
Bilateral trade benefits from information and communication technology because of its effects on
overall transaction costs that include risks involved in the production and distribution process.
Since information asymmetry is a root cause of risk that an agent has to deal with during its
decision making process, this study will refer to the costs involved with better understanding and
mitigating these risks as information costs. The cost of coordinating the market activity will be
referred to as the coordination cost. In order to understand how these different information and
communication technologies help reduce co-ordination/transaction and information costs in
trade, we must understand these kinds of costs in more detail. For any agent involved in bilateral
trade the overall reduction in transaction cost is the following:
1. Coordination costs: This category includes all costs that are involved in the process of carrying
out the trade of the good or service between the buyer and seller or within the supplying firm.
This could include logistical co-ordination costs, transportation costs and other possible costs
involved in the process of trading a particular good or service. Co-ordination costs within a firm
can also be lowered by ICT because use of these technologies lead to restructuring and work
place re-organization that may help the firm focus on its area of core competency, increase
specialization and reduce learning costs involved in the production process (Clemons et al. 1993,
Bresnahan and Brynjolfsson 2002)
According to a detailed study by Malone and Crowston (1991), co-ordination activities can
be categorized into different co-ordination processes. The four broad categories of co-ordination
processes are managing shared resources, managing producer and consumer relationships,
managing simultaneity constraints and managing task/subtask relationship. Managing shared
resources and tasks/subtask relationships are important activities for exporting firms because

29	
  
	
  

they are involved in a production process that possibly uses shared resources and certainly has a
certain distribution of tasks that need to be coordinated. However, these two co-ordination
processes exist only for the exporting firm and are not very important in the actual market
exchange involved in international trade. In the trade of goods and services, it is always the case
that one partner produces a product or delivers a service without the involvement of the trading
partner in the process of production. For the purpose of international trade, the two co-ordination
activities that are important are managing consumer/producer relationships (in this case it is the
relationship between the two trading partners) and managing simultaneity constraints, which
essentially include the co-ordination of the logistics of scheduling, payment processes,
transporting and so on. Malone and Crowston (1991) find that information and communication
technology can reduce these co-ordination costs in the following ways
1. First-order effect of reducing coordination costs with ICT: substitution of the technology
for some human coordination. For instance, many banks and insurance companies have
substituted automated systems for large numbers of human clerks in their back offices.
2. Second-order effect of reducing coordination costs with ICT: Increase of the overall
amount of co-ordination with the use of the technology.
3. Third-order effect of reducing coordination cost with ICT: Coordination intensive
structures are set up or changed by the technology
2. Information Costs: The next category of cost that is involved in the process of carrying out the
production or distribution (in this case the actual trade of a good or service) is information cost.
This kind of cost arises out of information asymmetries that exist when working with someone
based in a different country. These costs could be search costs or costs of finding a suitable
trading partner, risks that producers may have from any asset specificity in that particular

30	
  
	
  

industry (e.g. if a producer invests in specific machineries because they perceive a demand for
the good, but they also face risks of having large sunk costs if for some reason their product goes
out of demand), costs involved in finding information about the product price and quality as well
as the cost of finding if there are any environmental risks involved in trading with a partner in a
particular country or region. Based on the kind of industry the trade is taking place in, there are
varying levels of these information costs. For example, some industries trade in products that are
very country-specific and this entails added information costs about the demand of the product in
the importing country. If sufficient information is not gathered about potential demand the
exporting country could end up making investments on a product that cannot be exported in
another market if demand falls short. Similarly, products in industries with specific
environmental and health standards also require higher amount of information gathering.
Some industries find it beneficial to start the process of transaction and then gather all the
information that they need. This is called the transaction to information model and often takes
place when the transaction costs are low. The other model is the information to transaction model
and this takes place in industries where the transaction costs are significantly high, so it is
beneficial to first gather information about who is the most suitable trading partner, the costs
involved and the risks involved if the agreement is to be entered into. Several of these costs are
reduced by the use of information and communication technologies (Levchenko 2008).
Information costs are related to two categories of information asymmetry issues, namely
uncertainty and equivocality (Daft, Lengel and Trevino 1987). According to these authors,
uncertainty is the difference in the amount of information that is required to perform a task in the
most efficient manner and the amount of information that is actually available. If the amount of
information available about the task, environment and partners is low, then this creates a cost for

31	
  
	
  

the parties involved in the transaction. Equivocality on the other hand means ambiguity of
messages and this is often present when two parties present in different locations carry out a
transaction. Equivocality in trade can be mitigated to a large extent by the use of appropriate
information and communication technologies that have the following features of (1) providing
feedback, (2) providing multiple cues in the form of voice messages, visual gestures, documents
and so on (3) language variety which includes languages, pictures or numbers in conveying a
message and (4) providing personal focus which enables messages to be tailored to the frame of
reference, needs and current situation of the receiver.
Transaction costs as described by Williamson (1979), arise from human cognitive
limitations to receive, transmit and process information about the business transaction.
Williamson discusses different kinds of governance structures and how they relate to transaction
costs. As described earlier, this dissertation focuses on the market driven activity of trade and
some inter firm transactions involved in the production process. Based on Williamson (1979,
1985), contracts are the most suitable governance structure to mitigate transaction costs in this
kind of a market driven environment. This dissertation focuses on how Internet technology helps
reshape the importance of this institution in particular. The affordances discussed earlier impact
the different kinds of transaction costs and productivity costs to reshape the traditional institution
of contracting.
In this section, we have discussed the different kinds of transaction costs involved in
international trade. In the next section we will briefly discuss the impact on aggregate
productivity before going on to a discussion of how affordances of ICT impact overall trade
costs.

32	
  
	
  

3.2.2 ICT and its impact on aggregate productivity
On an aggregate industry level, this means that broadband technology will increase overall firm
productivity. It may even encourage certain new entrants to enter the market because of lower
costs of coordination and information. Meltiz (2002), makes an argument for trade by saying that
trade encourages the most productive firms to enter the market with the lesser productive firms
producing for a domestic market. This may in the long run force less productive firms to leave
the market and hence additional intra-industry reallocation of firms may take place towards more
productive firms. If this argument is extended, it is possible to say that information technology
further increases firm level productivity (through their impact on production, coordination and
information costs as discussed earlier) and thereby results in reallocation of firms in the market
to include firms that are more productive because of the use these information and
communication technologies. This may have an impact at a country level because different
countries have had varying degrees of success in adopting these technologies and this may affect
the overall productivity of some of their industries.
There may also be varying influence on different kinds of industries based on inherent
industry characteristics. This may result in compositional change in bilateral trade over time,
meaning that the very nature of some industries may have changed over time because of
changing technology, globalization, environmental standards or other reasons that have caused
industries to change significantly over time. An example of an industry that has significantly
changed in nature over time is the entertainment industry. ICT has given rise to different
production models for TV shows over several mediums (TV and Internet). The above discussion
of aggregate productivity of an industry should be studied for different industries differently.
Based on inherent industry characteristics, different industries and firms with in them can use

33	
  
	
  

information and communication technologies with varying levels of success. Some industry
characteristics that are important in this respect are the following (Levchenko 2008, Freund and
Weinhold 2000, 2002, 2004).
1. Does the product involve relationship specific investment?
2. Is the product a durable or non-durable good?
3. Is there a significant transportation cost involved? Can this transportation cost be
mitigated by information and communication technology and to what extent?
4. Has the nature of the product itself changed over time because of the influence on ICT?
These characteristics of the industry influence the degree to which information and
communication technology can be successfully used to increase productivity. Different countries
because of differences in factor proportions (theory of comparative advantage) or increasing
returns to scale (new trade theory), choose to invest more in certain industries and this in turn
affects the degree with which they can use the technology successfully. It is possible that better
adoption of information and communication technologies at a national level benefits only those
countries that primarily invest in certain industrial sector like services. For example, service
oriented countries with a lower adoption of these technologies may face a setback, but countries
that export natural resources may not be affected by the adoption of information and
communication technologies to the same extent.
Increased productivity leveraged by the use of information and communication
technology has an overall impact on trade because it brings down costs for the exporting country
making the product cheaper. There are other theories about what impacts industry productivity in
a country and what makes a country more favorable in trade. The most common theory being
countries with a relative comparative advantages (or relative factor differences) in certain

34	
  
	
  

industries are more likely to enter into trade exchanges. However, new trade theory attributes the
phenomenon of trade between countries with similar skill sets without relative factor differences
between them to specialization over time in certain industries due to various reasons like learning
over time and so on. Irrespective of the underlying framework driving trade between the two,
ICT can supplement productivity and reduce transaction costs in both scenarios. Countries that
are slower to adopt and utilize the technology may be significantly impacted in a competitive
global environment. ICT may potentially also create new digital products in certain industries as
a direct consequence of its use in the country. It may also make some products obsolete because
of the availability of new digital counterparts. Hence, ICT impacts trade between two countries
because of its role in reducing transaction costs and increasing productivity. ICT may also
significantly change the industry itself and might have a direct impact on the industry over time.
An example of an industry that has been created by the use of ICT is the digital books industry,
causing reduced demand for the traditional paper book industry.

35	
  
	
  

3.3 Impact of ICT on Trade through its Impact on Institutional Arrangements
The importance of institutions has been widely discussed in trade literature. Change in
institutional arrangements is often cited as one of the most important variables influencing trade
flows (Baier and Bergstrand 2001). Institutions can be described as a set of rules or explicit
arrangements that authorize or prescribe certain business codes of conduct (Levchenko 2007 and
Nunn 2007). They find that institutions play two major roles in trade. Institutions generate rents
for some parties within the economy and are also a source of comparative advantage. Institutions
according to Williamson (1985) also overcome the hold-up problem that could arise in any
market exchange between two parties. Institutions are of high importance in those industrial
sectors where transaction costs are high or there exist certain relationship specific investments.
Transaction costs have been defined in the previous section as co-ordination and information
costs involved in the trading process.
Relationship specific investment is one where the production investment is unique to the
relationship between the two trading parties. These kinds of investments are riskier and could
result in the well-known hold up problem if it is not backed up by strong institutional
arrangements (Levchenko 2008). Countries with weaker institutions could have rather
incomplete contracts making international trade more risky (Caballero and Hammour 1998).
Institutions could therefore end up giving certain countries a comparative advantage in certain
trading sectors (Levchenko 2008). Some traditional institutions used by countries as a source of
comparative advantage is better judicial quality and predictability and the enforcement of
contracts in countries. As discussed earlier, broadband Internet technology has certain
affordances that could reduce transactions and information costs involved in trade. By that logic
the use of broadband technology should give countries trading in institutionally intense sectors

36	
  
	
  

the similar benefits that institutional arrangements provide (E.g. better monitoring, reduced
search costs). Countries that already have good institutions are further strengthened by the use of
ICT, and countries that do not have good institutions now benefit from the use of broadband as it
reduces certain transaction costs and uncertainties involved. As described earlier the use of ICT
improves communication, allows for better and faster coordination and reduces information
costs.
It is possible that with the use of broadband a country that has weak institutions but
significant factor differences now becomes more favorable as a trading partner. In this study,
institutional intensity is sourced from Nunn (2007). Institutional intensity is the quality of a
certain industry to be more dependent on institutions because of higher transaction costs
involved. The higher the institutional intensity of an industry the higher the risk faced by the
producing country if for some reason the exchange does not go through. The risk is higher if an
industry’s inputs are tailored to very specific needs thereby making it more risky for both parties
if trade is not successful. In Nunn (2007), it is defined as the fraction of each industry's inputs
not sold on organized exchanges or reference priced, and is constructed based on US InputOutput Tables. Inputs that cannot be bought this way require relationship-specific investments
and thus rely on good contracting institutions being in place leading to higher institutional
intensity. Broadband technology can to some degree be a substitute to institutional arrangements
in a particular country, though this might be empirically very difficult to show. While it
improves coordination and reduces information asymmetry, it may be difficult to quantify the
costs saved. In this study, we explore the possibility that broadband might have a higher impact
on sectors that have high institutional intensity provided all other factors like transport costs and
durability are similar.

37	
  
	
  

3.4 Impact of Internet technology: Overall conceptual framework
In this section, we will analyze the relationship between the affordances of broadband
technology and their effect on international trade in different industrial sectors. Based on the
affordances of broadband technology and the two kinds of costs involved in trade, the following
table summarizes the effect of broadband on the costs of coordination and information.
With these affordances of broadband technology taken into consideration, it can be
inferred that broadband technology will have an impact on bilateral trade (in this case with the
US). The use of the technology provides some sort of comparative advantage in a particular
industrial sector by impacting coordination costs, information costs and improving productivity.
As discussed in the previous subsections, one possible way of being a source of comparative
advantage (Levchenko 2008), is by reducing transaction costs particularly in industries that are
highly institutionally intensive. Institutionally intensive industries are often ones with higher
information and coordination costs and therefore may benefit the most by use of ICT provided
other factors remain constant (like changing nature of the industry, large transportation costs).
The second way in which broadband technology can impact bilateral trade with the US is
if it helps the overall productivity of a particular sector. There are some sectors that do not
require relationship specific investments but still benefit from ICT because the technology can
make business processes more efficient (better management and logistic processes, better record
keeping, better market information and so on). These industries trade with countries because of
specialization or just differences in factor proportions. These industries benefit from broadband
technology simply because of increased factor productivity and lower co-ordination costs that
ICT may introduce.

38	
  
	
  

Table 1: Affordances of broadband and effect on co-ordination and information costs
Technology

Features/
affordances

Effect on Coordination
Costs

Effect on Information
Costs

Effect on Productivity

Internet

Audibility,
Visibility, Copresence, Cotemporality,
Reviewability and
Revisability

1.
Managing
shared
nd
rd
resources (2
and 3
order effect in reducing
costs)
2.Managing task/subtasks
nd
rd
(2 and 3 order effect
in reducing costs)
3.Managing Producer /
Consumer Relationships
• Improves logistics
st nd
rd
(1 , 2 and 3
order effect in
reducing costs)
4.Managing Simultaneity
Constraints
• Improves logistics
st nd
rd
(1 , 2 and 3
order effect in
reducing costs)
• Progressive
st
arrangements(1 ,

1.Uncertainty (information
costs, monitoring costs and
vulnerability costs)
• High effect on
reducing costs
2. Equivocality (subjective
messages)
• High effect on costs
by reducing the risk
from equivocality
enabled by
technology features
such as
i.
Feedback
ii.
Multiple cues
iii.
Language
iv.
Personal

1.Relationship specific Investment
sector
• High Effect in reducing costs
nd
2. Managing shared resources (2

nd

rd

3 order effect in reducing costs)
4.Managing Producer / Consumer
Relationships
st nd
• Improves logistics (1 , 2
rd

and 3 order effect in
reducing costs)
5.Managing Simultaneity Constraints
st nd
• Improves logistics (1 , 2
rd

and 3 order effect in
reducing costs)
• Progressive arrangements
st nd
rd
(1 , 2 and 3 order effect
in reducing costs)
st
6.Change in nature of industry (1
order effect that may significantly
impact the industry)

rd

2 and 3 order
effect in reducing
costs)

39	
  
	
  

rd

and 3 order effect in reducing
costs)
nd
3.Managing task/subtasks (2 and

Finally broadband can also be instrumental in changing the nature of a particular sector
(first order effect). This means that broadband technology has substituted technology for some
human coordination or has changed the nature of a physical good or process of producing it.
Broadband could have also increased demand of a product because of globalization and for some
industries the technology may serve as a marketing tool (either directly or indirectly). The last
decade has witnessed the nature of certain industries such as the entertainment industry, food and
beverage industry and so on because of global information sharing that broadband technology
has enabled. To summarize: broadband technology firstly reduces transaction costs in
institutionally intensive sectors; secondly it also improves trading opportunities for countries that
could trade with the US because of inherent differences in factor proportions by improving
productivity; and finally broadband impacts trade by changing the very nature of certain
industries (For example: music, books and entertainment), or could make information available
more globally and increases the demand of certain products (for example: the demand for
specialty food and beverage could increase because the world is more connected because of the
technology).

40	
  
	
  

Figure 1: Impact of Broadband on bilateral trade at an industry level

	
  

T&P Costs

+

P Costs

Low II, High RCA or
Specialization

+ or -

P Costs

Internet
Use of
Exporter

High II, High RCA or
Specialization

Low RCA or
Specialization

-

41	
  
	
  

Trade
with US

+ or -

Nature of
Industry

	
  

No
trade
with
US

Other contributions
(income, population,
exchange rate)

The above conceptual framework highlights the effect of Internet on bilateral trade with
the US. Use of broadband technology could reduce both transaction costs and production costs in
sectors that have higher institutional intensity (High II), Ricardian comparative advantage (where
countries have factor differences causing the production of a certain good to be more efficient in
the country with the relative comparative advantage thereby leading to trade between the two
countries) and highly specialized trade sectors because these industries do have additional
production costs because of relationship specificity that can be reduced by the use of the
technology. In cases where the industry is a low institutional intensive industry, broadband
technology can be still used by technologically superior countries to reduce co-ordination costs
or improve search. In cases where there is no Ricardian comparative advantage (RCA) or lack of
specialization, trade does not take place even with the availability of better Internet technology.
The importer country (US) also can reduce its transaction costs in general by the use of this
technology as it provides them with better co-ordination tools and higher information. It helps
them improve their search process and identify countries with factor differences or
specializations so that they can mutually benefit from international trade.
Based on the above framework of reduction in costs, we infer that highly institutionally
intensive industries should benefit more from the use of broadband technology. Industries with
low institutional intensities also benefit from the use of the technology in bilateral trade if there
are factor differences or specialization that makes trade between the two countries beneficial.
Finally, broadband could have an overall impact on the industries by changing the very nature of
the industry. This can happen because of multiple reasons. Firstly, broadband could give rise to a
new generation of products because technology can often replace some traditional goods and
services. Second, broadband is playing a big role in making the world more globalized. People

42	
  
	
  

all over the world now have access to a huge range of information and this has changed people’s
tastes and preferences for certain goods and services.
In this study, we make an assessment of how Internet affects individual industries by
studying how broadband has impacted trade with the largest importer in the world, the US. It
may be difficult to tease out the exact nature of the impact, but the underlying framework will
provide us with information about potential pathways causing these changes. Using the threedigit ISIC (International Standard Industrial Classification) Codes, we consider 25 different
industries for our study. Since this study focusses on relationship-specific investments, the 25
industries were chosen such that it was easy to obtain their institutional intensities from Nunn
(2007). This is discusses in more detail in the empirical evidence section.

43	
  
	
  

3.5 Hypotheses to be Tested
Based on the different possible effects of Internet use on trade costs, the following two
hypotheses are proposed.
H1: Internet use has a positive impact on sectors that have high institutional intensity.
As discussed earlier, there is literature that suggests that institutional arrangements themselves
have become a source of comparative advantage along with being able to generate rents for some
agents in the industry (Levchenko 2008). As discussed earlier, relationship specific investments
require better co-ordination and the minimization of information costs. Since these products are
not sold in the organized market or reference priced, connections between the two trading parties
are made through search processes that are largely driven by pre-existing ties and networks (Fink
et al. 2002). Since these sectors are associated with a well-known risk known as the ‘hold up
cost’ it was common practice to trade with countries that were known to have better institutional
arrangements. The growing popularity of Internet technology especially in business largely
changes this trend because the affordances of the technology can now make it possible to
monitor logistics and co-ordinate more effectively to minimize risk. The presence of multiple
verbal and visual cues also makes it easier to minimize risk because of better information
transfer between the concerned parties. Therefore the use of Internet technology can to some
degree substitute for institutional arrangements in countries where superior technology for
production exists but institutional arrangements are poorer.
H2: Internet use in sectors with low institutional intensity has a smaller positive impact
(compared to high II sectors).
This hypothesis is driven by two primary factors; a) factor differences or specialization that a
country may possess; and b) the low institutional requirement of that particular industrial sector.

44	
  
	
  

The first factor could be positively impacted by the use of broadband technology. This is because
use of ICT may lead to higher overall productivity in the sector (E.g. better coordination and
management, better access to information). The second factor might cause a negative impact as
far as bilateral trade flows with the US is concerned (though the use of broadband might have an
overall positive effect for the exporting country). This is because trade exchanges in this industry
are less relationship specific and are less contractual in nature. In a lot of cases, these industries
have products that are sold in the commodities market, and already have instruments in place to
mitigate any transaction costs that may exist in the process of trading these products. The use of
ICT may actually increase market competition and the exporting country may seek better
suppliers (ones with higher demand and willingness to pay) because ICT use makes search costs
lower.
Let us discuss the first factor driving this hypothesis. If trade is beneficial because of
existing difference in factor proportions, broadband technology can drive production costs down.
Trade is often driven by factor proportion differences or by the specialization of a production
process in a country over time. A country will export a product/service in sectors in which it has
a cost advantage as far as factor proportions or are concerned, and will import products in sectors
where the primary production factor costs are domestically very high. Since it makes sense that
the difference in factor proportions will lead to trade, one must also note that countries can be
rather competitive in their exports prices, and countries around the world benefit by taking
measures to not only increase factor productivity, but also by improving logistics of supply. As
discussed earlier, one such measure is to use Internet technology to improve factor productivity.
As discussed before there are several affordances that the Internet has that make co-ordination
within internal units involved in the production process easier. There is also literature that

45	
  
	
  

supports the link between the use of ICT and improved factor productivity (Stiroh 1999, 2001
and 2002). In cases where trade happens because of specialization or increasing returns to scale,
the use of broadband technology has a similar impact by increasing productivity. Hence, any
exporting country benefits by the use of broadband technology as it increases productivity and
reduces some transaction costs.
Let us now discuss the second factor of low institutional intensity and examine how
broadband technology may potentially impact bilateral trade is such industries. Low institutional
intensity essentially means that countries have rather low transaction costs and products are not
relationship specific. Of course, the degree of relationship specificity in every industry is
different, but industries with a low institutional intensity have fewer transaction costs incurred
because the product is rather generic in nature and can easily be resold in another country if trade
with a particular partner does not successfully work out. Since the exporter’s product is easily
sold in another country, the use of broadband technology gives the exporting country the
flexibility and the information tools to carry out trade with another country that has a demand for
that product. This means that the use of broadband technology by a country has opened the
country up to other potential trade partners, thereby leading to a potential negative impact on
bilateral trade with the US. If the exporting country has high supplies and is capable of handling
multiple trading partners then it is possible that bilateral trade with the US is not negatively
impacted. The extent to which this factor drives down bilateral trade between the two countries
depends on several factors like the overall demand of the product globally, the exchange rates
between countries, the nature of the product (durable or not) and so on. An industry that is in
high demand like Oil & Gas is likely to use the technology to find other trading partners if the

46	
  
	
  

terms of trade with the US are not favorable. The use of broadband gives them more information
about the options available to them.
Finally, besides these two fundamental hypotheses, it must be noted that the use of
broadband technology may have pervasive impacts on some industries by changing the very
nature of the product itself, or by changing tastes and preferences in a country. This study will be
unable to empirically tease out the reason for these changes, but it will be able to capture an
anomaly in results for industries that have been significantly impacted. It is not difficult to then
attribute a cause for the anomaly by observing the general trends in those industries.

47	
  
	
  

4. METHODOLOGY
Existing literature has provided empirical evidence of the positive relationship of the Internet on
international trade. This study explores if the Internet matters more in industries that are
institutionally intensive. In other words, this study explores if the Internet plays a positive role in
bilateral trade in sectors that are more institutionally intensive as opposed to sectors that are less
institutionally intensive. As described earlier, institutions can be described as a set of rules or
explicit arrangements that authorizes or prescribes certain business codes of conduct. Some
industries have historically required the presence of institutions for trade, where as some other
industries have a smaller need for institutional arrangements. As discussed earlier, there are
various factors like relationship specificity and other transaction costs that impact the need for
institutions.
In order to study the effect of technology on bilateral trade, the gravity equation is
employed. The gravity equation has been used for such studies by numerous scholars like Freund
and Weinhold (2000, 2002, 2004, 2008), Deardorff (1995), McCallum (1995), Frankel, Stein and
Wei (1995), Helliwell (1996), and Wei (1996). Anderson (1979) used the Cobb Douglas
expenditure system to develop the gravity equation. This equation was originally designed to
describe bilateral trade patterns based on the distances between countries as well as their relative
economic condition. The standard equation was as follows;
Gdpi Gdpj

(1)  Tradei,j =   α Distance   
ij

The basis of the gravity model is that the volume trade between two countries is directly
proportional to the size of the economy of each country and is inversely proportional to all the
factors that contribute to the cost of trade. The gravity model of trade was used in several

48	
  
	
  

empirical studies with considerable success, but there were several scholars that questioned the
theoretical justification of the model. The gravity equation is theoretically derived from the Cobb
Douglas expenditure system Anderson (1979), and the model has been successful at explaining
variation in bilateral trade under both constant returns to scale and increasing returns to scale.
Several trade theories like the Ricardian, Heckscher-Ohlin (H-O), and increasing returns to scale
(IRS) models (Helpman and Krugman 1985, Deardoff 1998 and Helpman 1998) can be used to
derive the gravity equation. However, the biggest criticism of the gravity equation is that it
assumes perfect specialization. In response to this critique, several scholars including Evenett
and Keller (2001), evaluate the gravity equation with imperfect specialization, and find that
when the model is based on increasing returns to scale, the gravity equation is successful in
accounting for trade variations across country pairs even when there is no difference in factor
proportions (large differences in factor proportions led to perfect specialization in constant
returns to scale world). Further, Hummels and Levinshon (1995) and Helpman (1987) in separate
studies find that the gravity prediction is an empirical success for developed countries if the
model is based on an increasing returns scale. Moreover the gravity equation works for
developing countries too because specialization in these countries is driven by factor proportion
differences.
For the purpose of this dissertation, the gravity equation will be modified to evaluate the
effect of various other variables, which are originally unaccounted for in the basic gravity
equation. This can be simply done by using a first order log-linearization of equation (1) and is a
common practice in studies using the gravity equation. Using this log transformation, the results
are interpreted as percent changes. In other words, the coefficients allow to assess the percent
effect of a one percent change in the independent variable on the dependent variable. In this case,

49	
  
	
  

trade is imports from country i to j in the time period t. Now if we represent trade by total
bilateral trade exports between the two countries i,j as importsij and then taking log on both sides
of the gravity equation we get the following;
(2)  logimportsijt= β1 +  β2   loggdpit  +β3   loggdpjt  +β4   logdistij+  εijt   
As commonly done in literature, we add cross-country specific effects. Since we are also
interested in the effect of ICT we add the variable “hosts” which captures the number of secure
servers in the respective countries. Internet hosts of both countries give us a measure of the
extent businesses in the two countries use the Internet. The country effects are captured in the
variables β2 to β9. This is represented in equation (3).
(3)  
logimportsijt=
β1   +  β2   loggdpit  +β3   loggdpjt+β4   logpopit+β5   logpopjt+β6   logexrateijt+β7   logdistij
+β8   loghostsit+β9   loghostsjt+  εijt   
In the above equation, the economic sizes of the two trading countries are accounted
along with the distance between the two countries. The income of the exporting country gdpi and
the population popi are included to account for supply (larger economies are usually capable of
producing more). The income of the importing country gdpj and the population popj are included
to account for demand. Both income and population data is collected from the World Bank’s
‘world development indicators’ (WDI) database. Exchange rate data is also collected from the
world developmental indicators database as it affects both supply and demand between the two
50	
  
	
  

countries. The other gravity equation variable disti,j is the average trading distance between the
two trading countries and this is defined as the distance in kilometers between capital cities. This
data is standard in the trade literature and is obtained from the IMF Direction of Trade Statistics.
Finally data on the Internet availability is collected from the World Development
Indicators database. Two variables fixed broadband subscribers and number of secure servers
(hosts) are two possible choices for studying the effect of Internet on trade. The use of secure
hosts has some limitations (organization of the ISP and hosting market has an influence on the
number of secure servers in a county) secure servers (hosts) still capture the use of Internet in
businesses better that individual fixed broadband subscriber. The use of encryption technology in
Internet transactions is a good indicator of whether the technology is actually used for business
purposes or not. Several other studies have used number of Internet users per capita of a
country’s population for analysis. This data is based on the number of individual homes that have
direct access to Internet and the average size of a family in the corresponding country. While this
is a good reflection of the penetration of the Internet in developed countries, it may not be an
efficient indicator of penetration in developing countries where Internet resources are heavily
shared between multiple users. Internet hosts on the other hand refers to a machine or an
application that is connected to the Internet and that has an Internet protocol address assigned to
it. Hence, hosts stand to reflect more closely the penetration of Internet in a country. The number
of Internet secure servers per capita of population of a country in particular reflects how
embedded the Internet is in the commercial activities of a country.

51	
  
	
  

Table 2: Data Sources
Variable
Source
Imports by country US Trade Online
to the US
https://usatrade.census.gov/
Accessed through MSU main library
Population
World Development Indicators
http://data.worldbank.org/data-catalog/world-development-indicators
GDP
World Development Indicators
http://data.worldbank.org/data-catalog/world-development-indicators
Exchange Rate
World Development Indicators
http://data.worldbank.org/data-catalog/world-development-indicators
Distance
IMF Direction of Trade Statistics
http://www.imf.org/external/data.htm
Internet Hosts
World Development Indicators
http://data.worldbank.org/data-catalog/world-development-indicators

Data on the dependent variable imports of country i to country j from 2002 onwards was
collected from US Trade online that has monthly and annual data on U.S. imports and exports by
country and U.S. customs port area with up to 10-digit Harmonized System classification detail.
In their study of determinants of bilateral trade flows, Egger (2000) and Matyas (1997) propose
the general gravity model which includes time and exporter and importer effects. In this study, a
three-way gravity model capturing time and exporter and importer effects is used. The two-way
model is a more generalized representation of a three-way model that only includes time
invariant main effects. The two-way model includes all time invariant effects (time-invariant
main effects and time invariant exporter and importer effects), whereas the three way model only
includes the main effect. In addition to time invariant main effects, the two-way model also
includes time-invariant bilateral effects, which the three-way model does not include. In this
dissertation, the three-way model is adequate for purposes of our analysis because the only time
invariant effect of interest is the main effect of ‘distance’ between the two countries. This is

52	
  
	
  

because the impact of ‘Internet’ on ‘distance’ and therefore transportation costs may have a
crucial role to play in determining trade patterns between countries in certain industries.
The statistical software package used to perform the empirical analysis is Stata 11. In this
dissertation, both cross sectional pooled regressions using OLS as well as panel data regression
is used to estimate the above gravity equation. This approach examines the impact of Internet use
on trade assuming that time is constant. The use of two different methods of estimation not only
provides a robustness check on the results, but because the two methods have different pros and
cons they help capture different aspects of the underlying process.
While the OLS estimation is more likely to have better goodness of fit because we are
analyzing a larger amount of data, it also helps us eliminate co-trending variables over time
(Freund and Weinhold 2004). The assumption the time is constant eliminates the bias that could
arise out of two variables having a linear relationship in their trends. Our second method of
estimation uses panel data and will help us avoid any omitted variable bias. The two most
common ways of panel data estimation is a fixed effects model and a random effects model.
According to scholars like Eggar (2002), the model selection depends not only on data
properties, but also depends on the interest of the analysis.
In general it is assumed that the fixed effects model provides more robust results.
However, one major restriction with the fixed effect model is that variables that are collinear
with the fixed effects need to be dropped from the model. In our case, data that varies by
exporter but is constant for the importer cannot be included in the model. Since the US is our
primary importer in our study, using a model with this restriction on variables that are collinear
with fixed effects will not work. Hence, the most obvious panel estimation choice here was the
random effects model. Further, when we tested the three-way model for each set of regressions,

53	
  
	
  

the random effect model was determined to be more appropriate than the fixed effects model
after conducting the Hausman test (Wooldridge 2001), that indicates that both the fixed effects
and random effects model are consistent, but the random effects model is more efficient. The
general rule of thumb as suggested by Judge et al. (1998) was also applied where he states that if
the cross sectional units are regarded as random drawings, the number of cross-sectional units N
is large (72 in our case) and T is small (9 in our case), the random effects model is more efficient
than the fixed effects model. Taylor (1980) also has shown that for T greater than or equal to 3
and N-K greater than or equal to 9, where K is the number of regressors, the random effects
model is more efficient. Hence after weighing both theoretical considerations and testing the
data, the random effects model seemed more appropriate for this study. Additionally, in this
study, regressions where carried out separately for different industries categorized by their
institutional intensities. It would be possible to analyze the interactive effect of institutional
intensity and Internet by introducing Institutional Intensity as a variable in the gravity equation.
This approach was explored but ultimately not chosen because of empirical challenges of using a
time invariant variable interacting with ‘Internet Hosts’ in a panel data estimation.

54	
  
	
  

5. EMPIRICAL EVIDENCE
Data from the years 2002 to 2010 was used for this study based on data availability. After
attempting to ensure that adequate data was available for the main variables in the gravity
equation, as well as for the specific industries considered in this study the resulting sample size
finally used in the analysis were 72 countries. After cleaning up the data for missing values we
end up with a relatively large sample size of 72 countries. Even though data availability was a
factor in selection of the sample, the sample does not only contain countries from developed or
OECD nations. The sample includes a large number of medium and low income countries with
different country characteristics. They also include countries from the American, African, Asian,
Australian and European continents. This was critical since distance is an important determinant
of trade patterns. Descriptive summary statistics of the variables are available in the appendix by
each of the 25 industries.
Table 3: List of countries
Albania
Algeria
Australia
Azerbaijan
Bahamas,	
  The
Bangladesh
Belarus
Belize
Bolivia
Bosnia	
  and	
  Herzegovina
Brazil
Bulgaria
Burkina	
  Faso
Cambodia
Cameroon
Canada
Chad
China

Colombia
Cote	
  d'Ivoire
Croatia
Cyprus
Czech	
  Republic
Denmark
Dominica
Dominican	
  Republic
El	
  Salvador
Estonia
Ethiopia
Gabon
Israel
Japan
Jordan
Kenya
Korea,	
  Rep.
Lao	
  PDR

Lebanon
Macedonia,	
  FYR
Madagascar
Malaysia
Maldives
Mali
Moldova
Morocco
Mozambique
Namibia
New	
  Zealand
Nicaragua
Oman
Panama
Papua	
  New	
  Guinea
Paraguay
Peru
Philippines

55	
  
	
  

Romania
Russian	
  Federation
Saudi	
  Arabia
Senegal
Slovak	
  Republic
Sri	
  Lanka
Sweden
Switzerland
Syrian	
  Arab	
  Republic
Tanzania
Thailand
Togo
Trinidad	
  and	
  Tobago
Ukraine
Uruguay
Yemen,	
  Rep.
Zambia
Zimbabwe

The 25 industries selected are listed below in Table 4, along with their institutional
intensities. The trade code used to categorize these industries is their three-digit ISIC
(International Standard Industrial Classification) Codes. The institutional intensities used in this
study to categorize the industries into low, medium and high institutional intensity sectors are
taken from a study conducted by Nunn (2007). Institutional intensity as defined by Nunn (2007)
as the share of intermediate inputs that cannot be bought on organized exchanges and is not
reference priced. In his study Nathan used the ISIC codes and classified them into the Bureau of
Economic Analysis (BEA) 1997 codes in order to calculate the institutional intensities. The data
used for human capital and physical capital came from Antweiler and Trefler (2002), while the
data for skill and capital intensities came from Bartelsman and Grey (1996). Data about
relationship specific investments comes from the 1997 input-output table and data about input’s
that are relationship specific came from Rauch (1999). Since the data used to calculate
institutional intensities come from specific sources and the calculation of these individual inputs
for different time periods are rather time-consuming and beyond the scope of this dissertation,
we will assume that the institutional intensities of these different industries have remained fairly
constant over the time period of interest in this study.
It is possible that some inputs have changed over time and the institutional intensities
may have altered, but for the purpose of this study we will assume that for the 25 industries
included in the study, the nature of relationship specific investments has not changed drastically
over a decade. Since the data used in this study spans from 2002 to 2010, it is reasonable to
assume that institutional intensities calculated by Nathan Nunn in 2007 are relatively time
appropriate to be used by this study to classify industries. In the 25 industries used for this study
we observe that oil and gas is the least institutionally intensive and transportation equipment is

56	
  
	
  

the most. As suggested in Nathan Nunn’s paper these numbers make sense for the time period of
our study. Petroleum and Coal, and Oil and gas have the least relationship specific investments
and that has not changed between 2002 and 2010, while transportation equipment require higher
amounts of relationship specific investments as the country specific/location specific factors play
a crucial role. It seems that the basic nature of these industries have not changed significantly
even if there are some changes in costs or processes.
Table 4: Industries and their institutional intensities
Industry
324 Petroleum and Coal Products
211 Oil & Gas
325 Chemicals
112 Livestock & Livestock Products
311 Food & Kindred Products
313 Textiles & Fabrics
322 Paper
111 Agricultural Products
327 Nonmetallic Mineral Products
212 Minerals and Ore
326 Plastics & Rubber Products
332 Fabricated Metal Products
113 Forestry Products
312 Beverages & Tobacco Products
321 Wood Products
114 Fish, Fresh/chilled/frozen & Other Marine Products
337 Furniture & Fixtures
316 Leather & Allied Products
323 Printed Matter And Related Products
335 Electrical Equipment, Appliances & Components
315 Apparel & Accessories
333 Machinery, Except Electrical
334 Computer & Electronic Products
511 Newspapers, Books & Other Published Matter
336 Transportation Equipment

57	
  
	
  

Institutional intensity
0.057654288
0.171116009
0.240283594
0.271475737
0.330635756
0.339565582
0.348113626
0.373506114
0.376581877
0.395959088
0.407733738
0.434656501
0.482169494
0.509090653
0.516188145
0.521434605
0.567659855
0.570608377
0.71282208
0.740018547
0.745411098
0.763578296
0.809803036
0.830137992
0.858740389

Other data used in the modified gravity model comes from several sources and data
sources are referenced in table 2. In this study, the dependent variable is US imports and we are
interested in understanding the impact of the number of Internet hosts in trading partner nations
on US imports in these 25 industries that have varying levels of institutional intensities. An OLS
pooled regression model and a modified gravity equation using a random effects model were
selected for this study. The OLS pooled regression was used to eliminate the effect of cotrending variables as well as to distinctly capture the effect of distance on trading patterns. In the
panel regression, a generalized least squares estimator was used to account for the correlation
structure involved in panel data of this nature.
The data is analyzed for 72 countries between 2002 and 2010. Each industry was studied
separately and then the standardized beta coefficients were compared.

Table 5 provides

summary statistics of the number of Internet hosts. It can be observed that the mean number of
Internet hosts has increased by over ten times. The standard deviation over time seems to decline
at first and then exponentially increase. As with most technologies, some early adopters first
adopt the technology and this leads to the discovery glitches and issues with the technology.
After things have been rectified and improved to make adoption easier, the rest of the population
adopts the technology. At this stage, it is possible that different countries follow different trends
in adopting the technology depending on their demographic and socio-economic factors. Urban
populations are more likely to have access to the technology first, and the time it takes for the
rural and less urban area to adopt and get access to the technology might vary depending on
several factors. In this study, we proceed to analyze if this difference in access to the information
and communication technologies affect trade in different industries differently.

58	
  
	
  

Table 5: Summary Statistics of Internet Hosts by Countries in this Study
Year
2003
2004
2005
2006
2007
2008
2009
2010

Mean
641.8361
1784.209
2350.957
2826.333
4214.139
5191.278
6163.324
8199.278

Standard Deviation
4.242641
3.535534
1.414214
1.414214
2.828427
5.656854
9.899495
9.899495

The empirical findings of the OLS estimation are listed below in Table 6. The table
depicts only the results of the internet host variable on imports. For results of other variables
please refer to the appendix. The focus of this section is the use of Internet (secure hosts) by the
exporting country as several studies in the past (Freund and Weinhold 2004) have indicated that
it is technology use by the exporting country that impacts trade patterns. Additionally, since we
are dealing with the same importing country (US), and the exporting country is the one that
benefits by the decreasing production and transaction costs, we will focus on the impact of
broadband use by the exporting country on bilateral trade.
In order to test for the assumption of normality, we checked the standardized normal
probability plot (p-p) and found that the error distributions were not tailed. As discussed above,
we tested for the absence of specification errors due to presence of endogenous variables in the
model by running Hausman’s endogeneity test for all the independent variables. The chi-squared
values for the Hausman tests of all variables indicated that the differences in coefficients are not
systematic, thereby ruling out endogeneity captured in an omitted variable bias or because of
measurement error. Finally, since multi-collinearity may inflate standard errors, the significance
of all the independent variables in our analysis indicates that collinearity between variables is not

59	
  
	
  

a significant threat to our results even if present. The large number of observations, high sample
variance of explanatory variables and low residual variance makes multi-collinearity less of
threat in this analysis. We also checked for heteroscedasticity between variables in the direct
impact model and the Breusch-Pagan / Cook-Weisberg (Heij and Boer 2004 and Wooldridge
2001) test for heteroscedasticity ruled out any concerns of deviation from constant variance.
A first key finding is that hypothesis 1 is supported in most of the highly institutionally
intensive sectors. The hypothesis states that industries that require more intensive institutional
arrangements are likely to be positively impacted by the use of broadband technology. In
general, industries with institutional intensities smaller than 0.4 are considered low, and ones
higher than 0.6 are considered high. However, industries between 0.4 and 0.6 are more
ambiguous and can fall into either categories depending on industry sub-classification. If
industries with an institutional intensity of above 0.5 are considered, we find that with the
exception of the newspaper industry all the industries have a positive standardized beta
coefficient that is significant. The newspaper industry could be an exception because of the
changing nature of the industry especially a migration towards online news sources. However,
the result of the regression in this industry is not statistically significant.
Hypothesis 2 is also supported in most industries with a lower institutional intensity, but
as discussed in the conceptual framework we find that there a few different factors that may
influence the slightly more diverse results. Some of the low institutionally intensive sectors have
a significant negative beta coefficient. This indicates that in these sectors, the exporting country
may benefit from the use of broadband technology when products are not very relationship
specific, because they can export their products to markets that are more favorable with a lot
more ease. Another explanation for these negative values is that Oil and Gas, Petroleum, Coal

60	
  
	
  

and Agricultural products have low institutional intensity but also have very high demand. They
are significantly impacted by trading arrangements/pacts between countries and Internet may
have a small role to play. There is also a high transportation cost involved these industries and
the affordances that Internet technology provides may not do much to lower these costs. These
products are also observed to be ones that are traded in the commodity markets. Commodity
markets are characterized by the use of instruments like futures contracts and other derivatives
that have been used for decades. This existence of several instruments that already mitigate
transaction costs have an impact on how much internet technology can impact these industries in
general. Further, the importing country (US) may have also increased domestic production in
these industries, and this could be a third explanation for these negative values. We also find that
when institutional intensity is closer to 0.5, those industries generally tend to largely benefit from
the use of broadband technology. In other words, even when the institutional intensity is low,
some of these industrial sectors still do have some transaction costs that cause broadband
technology to be of more utility in these sectors. They may also need to use digital processes if
their business partners in the importing country use the technology. Some benefit more than
others because broadband technology not only improves productivity in these sectors, but these
sectors could also benefit from merging tastes and preferences as a result of the technology
making the world a more global place. An example of an industry that benefits from this kind of
globalization is the food and kindred products industry because knowledge of global foods is
increasing demand for specialized food products from certain countries.

61	
  
	
  

Table 6: Results of OLS Estimation: Impact of Internet Hosts of Importing Country on Imports
Industry
Industries with low institutional intensity
324 Petroleum and Coal Products
211 Oil & Gas
325 Chemicals
112 Livestock & Livestock Products
311 Food & Kindred Products
313 Textiles & Fabrics
322 Paper
111 Agricultural Products
327 Nonmetallic Mineral Products
212 Minerals and Ore
Industries with high institutional intensity
326 Plastics & Rubber Products
332 Fabricated Metal Products
113 Forestry Product
312 Beverages & Tobacco Products
321 Wood Products
114 Fish, Fresh/chilled/frozen & Other Marine Products
337 Furniture & Fixtures
316 Leather & Allied Products
323 Printed Matter And Related Products
335 Electrical Equipment, Appliances & Components
315 Apparel & Accessories
333 Machinery, Except Electrical
334 Computer & Electronic Products
511 Newspapers, Books & Other Published Matter
336 Transportation Equipment

Inst. Intensity

SE

Pvalues

Rsquare

0.057654
0.171116
0.240284
0.271476
0.330636
0.339566
0.348114
0.373506
0.376582
0.395959

-0.1921
-0.30136
0.292402
0.717319
0.711624
0.365804
0.430032
-0.18928
0.47875
0.089691

0.099145
0.139822
0.05501
0.116188
0.072082
0.099331
0.097421
0.097067
0.066452
0.161863

0.053
0.032
0
0
0
0
0
0.052
0
0.58

0.2873
0.1623
0.7835
0.1442
0.5784
0.0864
0.2248
0.0464
0.7226
0.004

0.407734
0.434657
0.482169
0.509091
0.516188
0.521435
0.56766
0.570608
0.712822
0.740019
0.745411
0.763578
0.809803
0.830138
0.85874
62	
  

	
  

Beta Host

0.202094
0.598627
0.294794
0.826213
0.526411
0.401206
0.538908
0.498816
0.500388
0.264616
0.267445
0.676744
0.638674
-0.22565
0.18737

0.095358
0.05346
0.143966
0.076872
0.094334
0.130518
0.076039
0.105447
0.057191
0.093287
0.104976
0.047444
0.057663
0.19825
0.095041

0.035
0
0.041
0
0
0.002
0
0
0
0.005
0.011
0
0
0.257
0.049

0.0722
0.8157
0.2631
0.6579
0.065
0.3796
0.6623
0.0507
0.798
0.1511
0.0061
0.8455
0.764
0.5575
0.1252

In order to ensure that our results are robust and any omitted variable bias is captured a
random effects panel model is also applied. The main empirical findings of the panel data
regression study are listed in Table 7. Using the gravity equation, the main control variables used
in this study are the GDPs of the importing and exporting country, the population of the two
countries, the exchange rate between the two countries and the distance between the two trading
countries. As discussed earlier, the two most common ways of panel data estimation is a fixed
effects model and a random effect model. Even though most studies involving panel data use the
fixed effects model because it provides robust results, we have discussed in the methodology
section our reason for using the random effects model. It is often the case that in applied policy
studies with specific subjects of interest the random effects model is a better. The impact of
Internet hosts (secure servers) on bilateral trade (imports to the US) in 25 different industries is
2

displayed below. The standard beta coefficients, standard errors, the goodness of fit R and the
p-values are reported below. The R square value varies between regressions conducted between
the different industries. The gravity model used for this study has been kept as generic as
possible to include several industries, but this means that there is a possibility of some omitted
variable bias or the existence of other endogeneity issues that result in different goodness of fit
for these different industries. If one observes the results tables, a lot of industries that exchange
goods in the commodity markets have a poorer goodness of fit indicating the possible existence
of an omitted variable. Examples of such industries are Minerals and Ore, Wood Products,
Plastic and Rubber Products and Petroleum and Gas. These industries have lower goodness of fit
in both OLS and panel data estimation. However, for comparison sake the equations for all 25
industries are kept generic and industry specific variables have not been accounted for. Some of

63	
  
	
  

these variables have been discussed in the interpretation of the results as it is important to
account for these while analyzing the results of the OLS and panel data estimation.

64	
  
	
  

Table 7: Results of Panel Data Estimation: Impact of Internet Hosts of Importing Country on Imports
Industry
Industries with low institutional intensity
324 Petroleum and Coal Products
211 Oil & Gas
325 Chemicals
112 Livestock & Livestock Products
311 Food & Kindred Products
313 Textiles & Fabrics
322 Paper
111 Agricultural Products
327 Nonmetallic Mineral Products
212 Minerals and Ore
Industries with high institutional intensity
326 Plastics & Rubber Products
332 Fabricated Metal Products
113 Forestry Products
312 Beverages & Tobacco Products
321 Wood Products
114 Fish, Fresh/chilled/frozen & Other Marine Products
337 Furniture & Fixtures
316 Leather & Allied Products
323 Printed Matter And Related Products
335 Electrical Equipment, Appliances & Components
315 Apparel & Accessories
333 Machinery, Except Electrical
334 Computer & Electronic Products
511 Newspapers, Books & Other Published Matter
336 Transportation Equipment

Inst. intensity Beta Host

Pvalues

Rsquare

0.057654288
0.171116009
0.240283594
0.271475737
0.330635756
0.339565582
0.348113626
0.373506114
0.376581877
0.395959087

-0.08605
-0.3514197
0.1035312
0.495167
0.1566343
-0.0028095
0.3122754
-0.1962806
0.2631353
0.0196294

0.145163
0.197219
0.075238
0.157935
0.066187
0.145014
0.140285
0.122963
0.074559
0.411688

0.553
0.075
0.169
0.369
0.018
0.985
0.026
0.11
0
0.0962

0.01
0.125
0.7761
0.1375
0.5318
0.04
0.233
0.0616
0.7199
0.03

0.407733738
0.434656501
0.482169494
0.509090653
0.516188145
0.521434605
0.567659855
0.570608377
0.71282208
0.740018547
0.745411098
0.763578296
0.809803036
0.830137992
0.858740389

0.0049828
0.2472891
-0.0487679
0.1047977
0.4086147
0.1513534
0.3694672
0.0734178
0.2699192
0.346459
0.3351666
0.2852997
0.1079856
-0.1687481
0.176982

0.141387
0.06003
0.112997
0.07147
0.140056
0.105199
0.067559
0.15543
0.074428
0.138668
0.143298
0.060583
0.051087
0.218771
0.118381

0.972
0
0.666
0.143
0.004
0.15
0
0.637
0
0.012
0.019
0
0.035
0.441
0.135

0.0726
0.7967
0.2563
0.5869
0.075
0.3777
0.6568
0.028
0.794
0.1639
0.0215
0.8188
0.723
0.5744
0.1385

65	
  
	
  

SE

Hypothesis 1 is supported in this study, because we see that Internet use has a positive
impact on sectors that have high institutional intensity. Some sectors that have high institutional
intensity and have significant results are 333 Machinery, 334 Computer &Electronic Products,
335 Electrical Equipment, 337 Furniture and Fixtures and 315 Apparel & Accessories. While the
sign of the standardized beta coefficient is positive in all high institutional industries, a few
industries have results that are not statistically significant. This could be because of co-trending
variables or because of the smaller number of data points available for analysis using the panel
regression method. However, the overall positive sign implies that the use of Internet does
benefit industries that have high institutional intensities by reducing transaction costs (both coordination and information costs) and production costs. These industries often need high
institutional arrangements to secure themselves against the well-known hold up problem that
exist in industries with relationship specific investments. Similar to the OLS results, the
newspaper industry is the only industry that is an exception and this is potentially because of the
changing nature of the newspaper industry.
The panel data results of low institutionally intensive sectors support hypothesis 2, but
are slightly different from the OLS estimates. Figure 2 and 3 demonstrates how industries with
different institutional intensities are impacted by the use of Internet. These charts depict
industries institutional intensities against the estimated parameters for both OLS and panel data
estimation. The trend lines on both charts depict an upward trend indicating that Internet has a
bigger impact on industries with high institutional intensities. It is interesting that the panel
estimation have results that indicated that exporting countries in some low institutionally
intensive sector are likely to find other more favorable export markets besides our primary
importer US. This is because the use of Internet technology has affordances to provide higher

66	
  
	
  

amounts of information and better trade coordination. In this case, we also find that even though
the signs of the coefficients support hypothesis 2 they are not statistically significant. Some low
institutionally intensive exporting sectors also benefit from the use of the technology as
discussed in the conceptual framework earlier. Irrespective of the institutional intensity of a
sector, trade takes place when factor proportions are different or there is some specialization.
These sectors do have productivity benefits of using broadband technology. The 311 Food and
Kindred products, 112 Livestock and Livestock Products, 325 Chemicals, 326 Plastics and
Rubber Products, 327 Nonmetallic Mineral Products and 322 Paper are found to be positively
impacted by the use of Internet in the exporting country. Most of the estimates were statistically
significant indicating that the use of Internet benefits overall productivity whether they are
institutionally intensive or not. This indicates that the use of Internet reduces production costs in
sectors that naturally have differences in factor proportions that benefit trade. This could indicate
that co-ordination costs are reduced or search costs are reduced, and exporting countries are
better able to put out information about their products because of the Internet. It could also
indicate that with more countries using the Internet, the use of the technology becomes essential
for any country that was to do business with other countries in this global world. Earlier, these
kind of trading relationships were dictated either by trade agreements or traditional alliances.
Now the presence of Internet in exporting firms around the world gives exporters a chance to
display information about their products and prices very conveniently. This not only reduces
search costs but improves the overall logistics involved in the trading process.

67	
  
	
  

Figure 2: OLS Estimate vs. Industry Institutional Intensities

1	
  
Es#mated	
  Beta	
  

0.8	
  
0.6	
  
0.4	
  
0.2	
  
0	
  
-­‐0.2	
  
-­‐0.4	
  
0	
  

0.3	
  

0.6	
  
Industry	
  II	
  

0.9	
  

1.2	
  

Figure 3: Panel Estimate vs. Industry Institutional Intensities
0.6	
  
0.5	
  

Es#mated	
  Beta	
  

0.4	
  
0.3	
  
0.2	
  
0.1	
  
0	
  
-­‐0.1	
  
-­‐0.2	
  
-­‐0.3	
  
-­‐0.4	
  
0	
  

0.3	
  

0.6	
  
Industry	
  II	
  

68	
  
	
  

0.9	
  

There are certain industries that do not fit the general pattern in the results using both
estimation techniques. In sector 511 Newspaper, Books and Published Matter, we find that the
use of Internet technology over time has a negative impact on trade. This can be explained by
understanding the nature of this industry. The Internet has globalized the availability of news and
similar published matter. We also find similar results in most industrial sectors using both
estimation techniques. The OLS regression has more significant results adding to the robustness
of this analysis. The panel data estimation has a larger number of negative beta coefficients in
the low institutionally intensive sectors. However, since the results are not significant we cannot
state this with statistical certainty. Hypothesis 1 is supported by both estimation techniques, and
hypothesis 2 is supported more clearly in the OLS estimation. The panel data regression has
more statistically insignificant results and the general goodness of fit is lower. This could be
because the number of data points is fewer in the panel data estimation and the methodological
assumption driving the pooled approach makes it easier to detect a pattern.
The other independent variables and their beta coefficients are also reported industry wise
in the appendix of this study. The focus of this study if Internet technology and its effect on
bilateral trade with the US and we largely find that the hypothesis H1 and H2 are supported in
this study. These results indicate that Internet benefits industries that require high institutional
arrangements, but trade between countries is largely benefited from the use of technology
irrespective of the institutional intensity. Even in cases where a natural comparative advantage
exists, Internet improves the total factor productivity because of its affordances that improve coordination and reduce information costs. The sectors in which Internet is negatively correlated
with trade are ones like 324 Petroleum & Coal that have a high demand all around the world and
thus traditional ties and alliances still play an important role. These products are not very

69	
  
	
  

relationship-specific, and the use of the technology allows the exporting country to export its
product on more favorable terms of trade. This changes the patterns of bilateral trade, and gives
rise to different groups of winners and losers. One other characteristic of these industries that
show negative correlation to trade are the fact that they are products that are sold in commodities
markets, and as discussed earlier there are several other instruments used in these market that
already mitigate some transaction costs. The use of Internet may not have a significant impact on
reducing these costs and may changes the patterns of trade by enabling more countries to enter
the market.

70	
  
	
  

6. CONCLUSION
This study looks at the impact of Internet technology on bilateral trade with the US. The focus is
to understand if Internet use impacts industries that require higher institutional arrangements
differently from those that do not. The motivation for this comes from studies in the past that
have emphasized the importance of institutional arrangements in trade (Levchenko 2008,
Kormenos et al. 2001) and from studies that have found that Internet access in a country may
have some effect on trade (Freund and Weinhold 2000, 2002, 2004). Internet because of its
inherent nature is able provide access to different kinds of information (verbal and visual) almost
instantly, and this reduces transaction costs (traditionally managed through the use of
institutional arrangements between two countries). However, no study had really attempted to
look at the impact of Internet use on trade by categorizing industries based on their institutional
intensities. This study is a first attempt to look at how Internet use impacts the importance of
traditional institutional arrangements in trade.
One must be very careful in interpreting the importance of Internet on trade particularly
in highly institutionally intensive sectors because trade is a rather complex phenomenon with
various other important variables. This study has tried to control for important variables like the
size of the economies, exchange rates and the importance of distance. However, there are certain
other variables that might play a role–like cultural factors, languages and similar industry
specific variables–that could not be accounted for due to data availability issues. This more
parsimonious approach also allowed keeping the basic equation as generic as possible to most
industries included. It is very important to acknowledge basic trade theories that state that
comparative advantage makes trade possible and differences in factor proportions lead to
specialization. This study maintains that differences in factor proportions and increasing returns
to scale are the biggest factors leading to trade, and Internet just makes the transaction process
71	
  
	
  

easier. In cases where the institutional intensity is high, Internet may have an additional role to
play because institutional arrangements themselves become a source of comparative advantage
(Levchenko 2008).
An OLS and panel data estimation was performed on 25 different industries, and resulted
in strong support of hypothesis 1 (with the exception of the newspaper industry), and a weaker
support of hypothesis 2 (strongly supported by the OLS estimation, but lack of statistical
significance in the panel data estimation). This indicates support for both hypothesis but a
weaker one for hypothesis 2. Firstly, increase in use of Internet technology by the exporting
country definitely has a positive impact on sectors that require higher institutional arrangements.
This indicates that Internet use in businesses (through the use of secure servers and hosts) has the
ability to change the way trading alliances are entered into. One often finds that institutional
arrangements are of higher quality in more developed nations leading them to be the preferred
choice in industries where institutional intensities are high. Internet technology makes it possible
for firms in countries with poorer institutional arrangements to enter into trading agreements
with firms in the US.
The Internet reshapes the importance of institutions because the technology has certain
affordances that reduce the need for stronger institutional arrangements. In particular the Internet
is able to reduce transaction costs and production costs in these sectors. The second hypothesis
that this study tests is that Internet could continue to have positive effect on low institutionally
intensive sectors because of productivity effects. However, certain sectors may also see a
negative impact on bilateral trade (especially if relationship specificity of investments is very
low or negligible) because they now can find other trading partners that perhaps have more
favorable terms of trade. There are two processes in play here. If factor differences continue to

72	
  
	
  

exist (for example labor being expensive the US will import labor intensive products from
countries where labor is cheaper) or a country is highly specialized in a particular industry the
US is more likely to import those products from that country. Since there could be a large
number of countries with those factor differences, the use of Internet technology could further
help reduce production costs and some transaction costs particularly search costs and coordination costs. Countries that specialize in an industrial sector that is competitive therefore also
benefit from the use of Internet technology. Internet plays a huge role in reducing production
costs giving them an advantage when the market is competitive. However, there is also another
factor that plays a role in bilateral trade patterns. If relationship specificity of an industry is low
or almost negligible then there is a larger potential for trading goods with another country if for
some reason terms of trade with the US is not favorable. This is supported in the OLS estimation
and the panel data estimation also shows similar trends but without statistical significance.
This study not only emphasizes the importance of Internet in markets that are rather
competitive (hypothesis 2), but also suggests that markets that require high institutional
arrangements can positively benefit because of unique affordances of Internet technology that
result in lower production and transaction costs (hypothesis 1 is strongly supported). There are
several ways in which this study can be improved and extended. Firstly, it would be interesting if
we could run the same OLS/panel regression on specialized industry sub-groups (based on trade
alliances) to confirm the hypothesis. It would also be of interest to obtain a breakdown of costs in
every industry to study the impact of broadband of the underlying cost structure of different
industries. The above two questions are not dealt with in this study because of data limitations
but would certainly be interesting extensions. However, this study is still the first one to look at
the impact of Internet on trade by analyzing the how different industries with different

73	
  
	
  

institutional intensities are impacted. The results of this study definitely depicts that broadband
technology does have an impact on bilateral trade. The nature of this impact is positive in sectors
that are very relationship specific, but is more ambiguous in sectors that have low relationship
specificity. There are several studies that have attempted to study impact of the Internet on trade,
on overall economic productivity. However, this study looks at the impact Internet has on the
importance of traditional institutional arrangements during trade. The fact that Internet positively
impacts high institutionally intensive sectors does indicate that technological affordances of
broadband plays an important role in reducing transaction and production costs in trade.

74	
  
	
  

APPENDIX

75	
  
	
  

Appendix A: Panel Regression Results
Table A.1 321 Wood Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

Interval]

loggdp

-0.4945

0.188129 -2.63 0.009

-0.86323

-0.12578

loggdpus

0.214551 0.141992 1.51 0.131

-0.06375

0.49285

logpop

0.337893 0.132072 2.56 0.011

0.079037 0.596749

logpopus

-0.18345

-0.60959

0.242694

logexrate

0.009554 0.063326 0.15 0.880

-0.11456

0.133671

logserv

0.408615 0.140056 2.92 0.004

0.13411

0.68312

logservus

-0.03687

-0.54369

0.469939

logdist

0.015675 0.07665

0.20 0.838

-0.13456

0.165907

_cons

0.025745 0.093763 0.27 0.784

-0.15803

0.209516

0.217424 -0.84 0.399

0.258583 -0.14 0.887

76	
  
	
  

Table A.2 336 Transportation Equipment
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.194526 0.154784 1.26 0.209

-0.10885 0.497896

loggdpus

0.05083

0.149733 0.34 0.734

-0.24264 0.344301

logpop

-0.05474

0.103973 -0.53 0.599

-0.25852 0.149045

logpopus

-0.10625

0.229324 -0.46 0.643

-0.55572 0.343217

logexrate

0.055489 0.052639 1.05 0.292

-0.04768 0.15866

logserv

0.176982 0.118381 1.50 0.135

-0.05504 0.409004

logservus

0.045679 0.27299

-0.48937 0.58073

logdist

-0.13058

_cons

0.148346 0.084547 1.75 0.079

0.17 0.867

0.056815 -2.30 0.022

77	
  
	
  

Interval]

-0.24194 -0.01923
-0.01736 0.314056

Table A.3 313 Textiles and Fabrics
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.278869 0.199212 1.40 0.162

-0.11158 0.669318

loggdpus

0.039675 0.129757 0.31 0.760

-0.21464 0.293994

logpop

-0.02856

0.145004 -0.20 0.844

-0.31276 0.255644

logpopus

0.11211

0.198559 0.56 0.572

-0.27706 0.501279

logexrate

0.079232 0.06533

logserv

-0.00281

0.145014 -0.02 0.985

-0.28703 0.281413

logservus

-0.11268

0.235658 -0.48 0.633

-0.57456 0.349204

logdist

-0.09192

0.087225 -1.05 0.292

-0.26288 0.079039

_cons

0.057293 0.097752 0.59 0.558

1.21 0.225

78	
  
	
  

Interval]

-0.04881 0.207276

-0.1343

0.248883

Table A.4 323 Printed Matter and Related Products
[95%
logimports

Coef.

loggdp

Std. Err.

z P>z

Conf.

Interval]

0.626537 0.113618 5.51 0.000

0.40385

0.849224

loggdpus

0.107232 0.054616 1.96 0.050

0.000187 0.214278

logpop

-0.01179

0.089256 -0.13 0.895

-0.18673

0.16315

logpopus

-0.08231

0.085774 -0.96 0.337

-0.25043

0.085802

logexrate

0.026676 0.033041 0.81 0.419

-0.03808

0.091434

logserv

0.269919 0.074428 3.63 0.000

0.124043 0.415796

logservus

-0.09932

0.101073 -0.98 0.326

-0.29742

0.098776

logdist

-0.11417

0.054908 -2.08 0.038

-0.22179

-0.00656

_cons

-0.12051

0.056613 -2.13 0.033

-0.23147

-0.00955

79	
  
	
  

Table A.5 326 Plastics and Rubber Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.249981 0.191642 1.30 0.192

-0.12563 0.625593

loggdpus

-0.00207

0.136162 -0.02 0.988

-0.26894 0.264803

logpop

-0.08824

0.136566 -0.65 0.518

-0.35591 0.179424

logpopus

0.032521 0.209295 0.16 0.877

-0.37769 0.442732

logexrate

-0.17901

-0.30478 -0.05325

logserv

0.004983 0.141387 0.04 0.972

-0.27213 0.282096

logservus

-0.00453

0.248307 -0.02 0.985

-0.4912

logdist

-0.068

0.080456 -0.85 0.398

-0.22569 0.089692

_cons

0.085373 0.094793 0.90 0.368

0.064168 -2.79 0.005

80	
  
	
  

Interval]

0.482147

-0.10042 0.271165

Table A.6 324 Petroleum and Coal Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.642431 0.202706 3.17 0.002

0.245134 1.039729

loggdpus

0.087269 0.125813 0.69 0.488

-0.15932

0.333859

logpop

-0.17032

0.148219 -1.15 0.251

-0.46083

0.120182

logpopus

-0.0403

0.194548 -0.21 0.836

-0.42161

0.341007

logexrate

-0.04353

0.06849

-0.64 0.525

-0.17777

0.090705

logserv

-0.08605

0.145163 -0.59 0.553

-0.37056

0.198469

logservus

-0.01946

0.230441 -0.08 0.933

-0.47112

0.432196

logdist

-0.24316

0.089762 -2.71 0.007

-0.41909

-0.06723

_cons

-0.01393

0.099493 -0.14 0.889

-0.20893

0.181077

81	
  
	
  

Interval]

Table A.7 322 Paper
[95%
logimports

Coef.

loggdp

Std. Err.

z P>z

Conf.

Interval]

0.245224 0.19068

1.29 0.198

-0.1285

0.61895

loggdpus

0.116775 0.13249

0.88 0.378

-0.1429

0.376451

logpop

-0.03383

0.136441 -0.25 0.804

-0.30125

0.233587

logpopus

-0.12329

0.202706 -0.61 0.543

-0.52058

0.274007

logexrate

0.112441 0.063357 1.77 0.076

-0.01174

0.236618

logserv

0.312275 0.140285 2.23 0.026

0.037323 0.587228

logservus

-0.00723

0.240623 -0.03 0.976

-0.47884

0.464384

logdist

-0.02332

0.080601 -0.29 0.772

-0.1813

0.134657

_cons

0.034244 0.093648 0.37 0.715

-0.1493

0.217791

82	
  
	
  

Table A.8 211 Oil and Gas
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.755367 0.271322 2.78 0.005

0.223585 1.287149

loggdpus

-0.05854

0.169987 -0.34 0.731

-0.39171

0.27463

logpop

-0.09677

0.192159 -0.50 0.615

-0.4734

0.279854

logpopus

-0.0368

0.257485 -0.14 0.886

-0.54146

0.467865

logexrate

0.051831 0.133466 0.39 0.698

-0.20976

0.313419

logserv

-0.35142

-0.73796

0.035123

logservus

0.196439 0.309289 0.64 0.525

-0.40976

0.802635

logdist

-0.09959

0.109002 -0.91 0.361

-0.31323

0.114055

_cons

-0.2232

0.13524

-0.48826

0.041869

0.197219 -1.78 0.075

-1.65 0.099

83	
  
	
  

Interval]

Table A.9 511 Newspapers
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

1.272882 0.301984 4.22 0.000

0.681003 1.86476

loggdpus

-0.22704

0.143747 -1.58 0.114

-0.50878

0.054697

logpop

-0.43077

0.193532 -2.23 0.026

-0.81009

-0.05145

logpopus

-0.86906

0.202434 -4.29 0.000

-1.26583

-0.4723

logexrate

-0.18289

0.132475 -1.38 0.167

-0.44253

0.07676

logserv

-0.16875

0.218771 -0.77 0.441

-0.59753

0.260036

logservus

0.805205 0.240072 3.35 0.001

0.334673 1.275737

logdist

-0.01137

0.092567 -0.12 0.902

-0.19279

0.17006

_cons

-0.83404

0.157226 -5.30 0.000

-1.14219

-0.52588

84	
  
	
  

Interval]

Table A.10 327 Nonmetallic Minerals
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.565154 0.118794 4.76 0.000

0.332322 0.797986

loggdpus

0.09295

0.051068 1.82 0.069

-0.00714

0.193041

logpop

0.061412 0.099026 0.62 0.535

-0.13268

0.255499

logpopus

-0.1521

0.078241 -1.94 0.052

-0.30545

0.001247

logexrate

-0.00021

0.030929 -0.01 0.995

-0.06083

0.060409

logserv

0.263135 0.074559 3.53 0.000

0.117002 0.409268

logservus

-0.03305

0.093106 -0.35 0.723

-0.21553

0.149437

logdist

-0.11337

0.066453 -1.71 0.088

-0.24362

0.016872

_cons

-0.12192

0.065327 -1.87 0.062

-0.24996

0.006116

85	
  
	
  

Interval]

Table A.11 333 Machinery
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.685443 0.094205 7.28 0.000

0.500806 0.870081

loggdpus

0.07915

0.044157 1.79 0.073

-0.0074

0.165696

logpop

-0.0969

0.075726 -1.28 0.201

-0.24532

0.051522

logpopus

-0.27857

0.068471 -4.07 0.000

-0.41277

-0.14437

logexrate

0.032502 0.031097 1.05 0.296

-0.02845

0.09345

logserv

0.2853

0.16656

0.404039

logservus

0.114429 0.080673 1.42 0.156

-0.04369

0.272545

logdist

-0.07111

-0.16625

0.024037

_cons

0.008652 0.048734 0.18 0.859

-0.08686

0.104168

0.060583 4.71 0.000

0.048545 -1.46 0.143

86	
  
	
  

Interval]

Table A.12 112 Livestock and Livestock Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

Interval]

loggdp

-0.48019

0.210772 -2.28 0.023

-0.89329

-0.06708

loggdpus

-0.02325

0.161479 -0.14 0.885

-0.33975

0.293239

logpop

0.482869 0.144847 3.33 0.001

0.198973 0.766765

logpopus

-0.47313

-0.96272

0.016466

logexrate

0.080028 0.069884 1.15 0.252

-0.05694

0.216999

logserv

0.495167 0.157935 3.14 0.002

0.18562

0.804715

logservus

0.287928 0.296572 0.97 0.332

-0.29334

0.869199

logdist

-0.08523

0.082829 -1.03 0.303

-0.24757

0.07711

_cons

0.01511

0.105239 0.14 0.886

-0.19116

0.221374

0.249798 -1.89 0.058

87	
  
	
  

Table A.13 316 Leather & Allied Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.093504 0.220356 0.42 0.671

-0.33839 0.525394

loggdpus

0.117388 0.129962 0.90 0.366

-0.13733 0.372109

logpop

-0.112

0.166455 -0.67 0.501

-0.43825 0.214244

logpopus

-0.30178

0.200221 -1.51 0.132

-0.69421 0.090643

logexrate

-0.0199

0.070865 -0.28 0.779

-0.1588

logserv

0.073418 0.15543

0.47 0.637

-0.23122 0.378056

logservus

0.179958 0.236987 0.76 0.448

-0.28453 0.644443

logdist

0.021453 0.103397 0.21 0.836

-0.1812

_cons

0.061286 0.110074 0.56 0.578

-0.15446 0.277027

88	
  
	
  

Interval]

0.11899

0.224106

Table A.14 337 Furniture and Fixtures
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.463165 0.120105 3.86 0.000

0.227763 0.698566

loggdpus

0.076532 0.045257 1.69 0.091

-0.01217

0.165235

logpop

0.063506 0.108419 0.59 0.558

-0.14899

0.276003

logpopus

-0.18898

0.070927 -2.66 0.008

-0.32799

-0.04996

logexrate

-0.00159

0.028612 -0.06 0.956

-0.05767

0.054487

logserv

0.369467 0.067559 5.47 0.000

0.237054 0.501881

logservus

-0.02264

0.083417 -0.27 0.786

-0.18613

0.140853

logdist

-0.11852

0.073324 -1.62 0.106

-0.26223

0.025194

_cons

-0.11156

0.071105 -1.57 0.117

-0.25092

0.027804

89	
  
	
  

Interval]

Table A.15 311 Food and Kindred Products
[95%
logimports

Coef.

Std. Err.

loggdp

0.465457 0.119616 3.89 0.000

0.231014 0.699899

loggdpus

-0.03906

-0.127

0.048884

logpop

0.077938 0.111137 0.70 0.483

-0.13989

0.295762

logpopus

0.010817 0.069731 0.16 0.877

-0.12585

0.147487

logexrate

-0.07445

-0.13297

-0.01592

logserv

0.156634 0.066187 2.37 0.018

0.02691

0.286359

logservus

0.009429 0.081951 0.12 0.908

-0.15119

0.17005

logdist

-0.11488

-0.26847

0.038718

_cons

0.067067 0.074575 0.90 0.368

-0.0791

0.213231

0.04487

z P>z

Conf.

-0.87 0.384

0.029862 -2.49 0.013

0.078366 -1.47 0.143

90	
  
	
  

Interval]

Table A.16 114 Fish, Fresh/chilled/Frozen and Other Marine Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.234281 0.203559 1.15 0.250

-0.16469 0.633249

loggdpus

0.032212 0.063586 0.51 0.612

-0.09242 0.156839

logpop

0.316601 0.182843 1.73 0.083

-0.04176 0.674966

logpopus

-0.24929

0.097481 -2.56 0.011

-0.44035 -0.05823

logexrate

-0.36891

0.128626 -2.87 0.004

-0.62101 -0.1168

logserv

0.151353 0.105199 1.44 0.150

-0.05483 0.35754

logservus

0.158676 0.11443

-0.0656

logdist

-0.12666

0.119431 -1.06 0.289

-0.36074 0.107419

_cons

-0.27457

0.125176 -2.19 0.028

-0.51991 -0.02923

1.39 0.166

91	
  
	
  

Interval]

0.382954

Table A.17 332 Fabricated Metal Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.693771 0.095192 7.29 0.000

0.507198 0.880344

loggdpus

0.041683 0.043525 0.96 0.338

-0.04362

0.12699

logpop

-0.05946

0.079316 -0.75 0.453

-0.21491

0.095999

logpopus

-0.1548

0.06752

-0.28713

-0.02246

logexrate

0.034692 0.026413 1.31 0.189

-0.01708

0.086461

logserv

0.247289 0.06003

0.129633 0.364945

logservus

0.008725 0.079961 0.11 0.913

-0.148

0.165446

logdist

-0.11022

0.053332 -2.07 0.039

-0.21475

-0.00569

_cons

-0.06328

0.052586 -1.20 0.229

-0.16635

0.039785

-2.29 0.022

4.12 0.000

92	
  
	
  

Interval]

Table A.18 335 Electrical Equipment, Appliances & Components
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

Interval]

loggdp

-0.00324

0.187148 -0.02 0.986

-0.37004

0.363567

loggdpus

0.124741 0.136323 0.92 0.360

-0.14245

0.391928

logpop

0.020105 0.132533 0.15 0.879

-0.23965

0.279865

logpopus

-0.0256

0.209477 -0.12 0.903

-0.43616

0.384971

logexrate

-0.13305

0.062868 -2.12 0.034

-0.25627

-0.00983

logserv

0.346459 0.138668 2.50 0.012

0.074676 0.618242

logservus

-0.11613

0.248598 -0.47 0.640

-0.60337

0.371117

logdist

-0.0319

0.077562 -0.41 0.681

-0.18392

0.120115

_cons

0.038735 0.092856 0.42 0.677

-0.14326

0.22073

93	
  
	
  

Table A.19 334 Computer and Electronic Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.75498

0.093319 8.09 0.000

loggdpus

0.006996 0.034666 0.20 0.840

-0.06095

0.07494

logpop

-0.10255

0.087017 -1.18 0.239

-0.2731

0.068001

logpopus

-0.1824

0.053406 -3.42 0.001

-0.28707

-0.07773

logexrate

-0.08

0.023048 -3.47 0.001

-0.12517

-0.03482

logserv

0.107986 0.051087 2.11 0.035

0.007857 0.208114

logservus

0.099543 0.062971 1.58 0.114

-0.02388

0.222965

logdist

-0.05818

0.061524 -0.95 0.344

-0.17876

0.062405

_cons

0.08208

0.058246 1.41 0.159

-0.03208

0.196241

94	
  
	
  

Interval]

0.572078 0.937883

Table A.20 312 Beverages & Tobacco Products
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

loggdp

0.825902 0.133422 6.19 0.000

0.564399 1.087405

loggdpus

0.030754 0.045396 0.68 0.498

-0.05822

0.119729

logpop

-0.26155

-0.49896

-0.02413

logpopus

0.113172 0.07028

-0.02457

0.250918

logexrate

-0.0112

-0.17101

0.148606

logserv

0.104798 0.07147

-0.03528

0.244876

logservus

-0.13726

0.082644 -1.66 0.097

-0.29923

0.024723

logdist

-0.25006

0.082778 -3.02 0.003

-0.4123

-0.08782

_cons

-0.29835

0.081187 -3.67 0.000

-0.45748

-0.13923

0.121132 -2.16 0.031
1.61 0.107

0.081537 -0.14 0.891
1.47 0.143

95	
  
	
  

Interval]

Table A.21 315 Apparel and Accessories
[95%
logimports

Coef.

Std. Err.

z P>z

Conf.

Interval]

loggdp

-0.2991

0.190393 -1.57 0.116

-0.67227

0.074061

loggdpus

0.114489 0.154924 0.74 0.460

-0.18916

0.418135

logpop

0.153983 0.131611 1.17 0.242

-0.10397

0.411936

logpopus

-0.23219

-0.69816

0.233776

logexrate

0.060178 0.064582 0.93 0.351

-0.0664

0.186757

logserv

0.335167 0.143298 2.34 0.019

0.054307 0.616026

logservus

0.030134 0.2825

-0.52356

0.583825

logdist

-8.1E-05

-0.14683

0.146673

_cons

0.051702 0.09686

-0.13814

0.241544

0.237743 -0.98 0.329

0.11 0.915

0.074876 -0.00 0.999
0.53 0.593

96	
  
	
  

Table A.22 111 Agricultural Products
[95%
logimports

Coef.

Std. Err.

loggdp

0.164918 0.161242 1.02 0.306

-0.15111 0.480947

loggdpus

0.195682 0.149698 1.31 0.191

-0.09772 0.489084

logpop

0.051216 0.108904 0.47 0.638

-0.16223 0.264664

logpopus

0.446262 0.230174 1.94 0.053

-0.00487 0.897395

logexrate

-0.15676

0.054903 -2.86 0.004

-0.26437 -0.04915

logserv

-0.19628

0.122963 -1.60 0.110

-0.43728 0.044723

logservus

-0.56125

0.273946 -2.05 0.040

-1.09818 -0.02433

logdist

0.056817 0.060028 0.95 0.344

-0.06084 0.17447

_cons

-0.04929

-0.21869 0.120114

0.08643

z P>z

Conf.

-0.57 0.569

97	
  
	
  

Interval]

Appendix B: OLS Regression Results
Table B.1 324 Petroleum and Coal Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.803892 0.129284 6.22

0

0.549816 1.057968

loggdpus

0.076624 0.169984 0.45

0.652

-0.25744

0.410685

logpop

-0.29005

0.001

-0.45338

-0.12673

logpopus

0.000778 0.261638 0

0.998

-0.51341

0.514961

logexrate

-0.07745

0.043198 -1.79

0.074

-0.16235

0.007441

logserv

-0.1921

0.099145 -1.94

0.053

-0.38695

0.002745

logservus

-0.02407

0.311853 -0.08

0.939

-0.63694

0.588803

logdist

-0.21434

0.041939 -5.11

0

-0.29676

-0.13192

_cons

-0.01145

0.086693 0.13

0.895

-0.18183

0.158919

0.083107 -3.49

98	
  
	
  

Table B.2 111 Agriculture Products
[95%
logimports Coef.

Std. Err.

t

P>t

Conf. In

terval]

loggdp

0.171429 0.125028 1.37

0.171

-0.07425

0.417113

loggdpus

0.201473 0.166245 1.21

0.226

-0.1252

0.52815

logpop

0.057257 0.080895 0.71

0.479

-0.1017

0.216219

logpopus

0.41747

0.254807 1.64

0.102

-0.08323

0.918173

logexrate

-0.13565

0.041769 -3.25

0.001

-0.21773

-0.05357

logserv

-0.18928

0.097067 -1.95

0.052

-0.38002

0.001455

logservus

-0.54899

0.304112 -1.81

0.072

-1.14658

0.048601

logdist

0.045467 0.041408 1.1

0.273

-0.0359

0.126835

_cons

-0.05744

0.494

-0.22238

0.107493

0.083937 0.68

99	
  
	
  

Table B.3 315 Apparel and Accessories Industry
[95%
logimports Coef.

Std. Err.

P>t

Conf.

Interval]

loggdp

-0.1979

0.135216 -1.46

0.144

-0.46361

0.067798

loggdpus

0.117198 0.180004 0.65

0.515

-0.23651

0.470907

logpop

0.082451 0.087542 0.94

0.347

-0.08957

0.254471

logpopus

-0.17052

0.535

-0.71061

0.369561

logexrate

0.055385 0.045025 1.23

0.219

-0.03309

0.14386

logserv

0.267445 0.104976 2.55

0.011

0.061166 0.473723

logservus

0.001633 0.328113 0

0.996

-0.64311

0.64638

logdist

-0.00118

0.044836 -0.03

0.979

-0.08928

0.086928

_cons

0.05265

0.09069

0.562

-0.12555

0.230863

0.27485

t

-0.62

0.58

100	
  
	
  

Table B.4 312 Beverage and Tobacco Products
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.05092

0.099197 -0.51

0.608

-0.24593

0.144096

loggdpus

0.076026 0.128792 0.59

0.555

-0.17717

0.329224

logpop

0.164436 0.064039 2.57

0.011

0.038538 0.290334

logpopus

0.047406 0.196893 0.24

0.81

-0.33967

0.434486

logexrate

0.053022 0.035583 1.49

0.137

-0.01693

0.122975

logserv

0.826213 0.076872 10.75

0

0.675086 0.977339

logservus

-0.2393

0.235838 -1.01

0.311

-0.70295

0.224342

logdist

-0.20314

0.030314 -6.7

0

-0.26274

-0.14355

_cons

-0.22808

0.065689 -3.47

0.001

-0.35722

-0.09894

101	
  
	
  

Table B.5 325 Chemical Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.679811 0.072038 9.44

0

0.538234 0.821387

loggdpus

0.117185 0.094319 1.24

0.215

-0.06818

0.302551

logpop

-0.12226

0.046503 -2.63

0.009

-0.21365

-0.03087

logpopus

-0.22451

0.143976 -1.56

0.12

-0.50747

0.058449

logexrate

-0.02964

0.023707 -1.25

0.212

-0.07623

0.016954

logdist

-0.12953

0.023001 -5.63

0

-0.17473

-0.08432

logserv

0.292402 0.05501

0

0.184291 0.400513

logservus

-0.01386

0.171732 -0.08

0.936

-0.35137

0.323645

_cons

-0.02716

0.047602 0.57

0.569

-0.12072

0.066386

5.32

102	
  
	
  

Table B.6 334 Computer and Electronics Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.234847 0.07415

3.17

0.002

0.089138 0.380555

loggdpus

0.024643 0.097683 0.25

0.801

-0.16731

logpop

0.130957 0.047566 2.75

0.006

0.037487 0.224426

logpopus

-0.25461

0.088

-0.54681

0.037594

logexrate

0.022221 0.024469 0.91

0.364

-0.02586

0.070304

logserv

0.638674 0.057663 11.08

0

0.525363 0.751986

logservus

0.045732 0.177697 0.26

0.797

-0.30345

0.394917

logdist

-0.01108

0.648

-0.0588

0.036636

_cons

0.110333 0.049108 2.25

0.025

0.013803 0.206803

0.1487

-1.71

0.024283 -0.46

103	
  
	
  

0.216597

Table B.7 335 Electrical Equipment, Appliances and Components
[95%
logimports Coef.

Std. Err.

loggdp

0.01989

t

P>t

Conf.

Interval]

0.120161 0.17

0.869

-0.21623

0.256008

loggdpus

0.103966 0.159961 0.65

0.516

-0.21036

0.418292

logpop

0.010075 0.077794 0.13

0.897

-0.14279

0.162942

logpopus

0.082012 0.244247 0.34

0.737

-0.39794

0.561961

logexrate

-0.1148

0.004

-0.19342

-0.03617

logserv

0.264616 0.093287 2.84

0.005

0.081306 0.447927

logservus

-0.15667

0.29158

-0.54

0.591

-0.72962

0.416293

logdist

-0.02665

0.039844 -0.67

0.504

-0.10494

0.051646

_cons

0.066293 0.080592 0.82

0.411

-0.09207

0.224658

0.040012 -2.87

104	
  
	
  

Table B.8 332 Fabricated Metal Products
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.311619 0.070042 4.45

0

0.173955 0.449283

loggdpus

0.045627 0.089891 0.51

0.612

-0.13105

logpop

0.11748

0.044219 2.66

0.008

0.030569 0.204391

logpopus

-0.19779

0.138025 -1.43

0.153

-0.46907

0.073494

logexrate

0.003398 0.022577 0.15

0.88

-0.04098

0.047773

logserv

0.598627 0.05346

0

0.493554 0.7037

logservus

-0.02521

0.165003 -0.15

0.879

-0.34952

0.299094

logdist

-0.06034

0.021805 -2.77

0.006

-0.1032

-0.01748

_cons

-0.02449

0.045764 0.54

0.593

-0.11444

0.065457

11.2

105	
  
	
  

0.222305

Table B.9 114 Fish and Marine Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.22848

0.162351 -1.41

0.16

-0.5479

0.090936

loggdpus

0.032036 0.188769 0.17

0.865

-0.33936

0.403429

logpop

0.558507 0.098507 5.67

0

0.364699 0.752316

logpopus

-0.26866

0.285126 -0.94

0.347

-0.82964

0.292308

logexrate

-0.30569

0.056487 -5.41

0

-0.41682

-0.19455

logserv

0.401206 0.130518 3.07

0.002

0.144418 0.657994

logservus

0.116803 0.338734 0.34

0.73

-0.54964

0.783247

logdist

-0.11575

0.040033 -2.89

0.004

-0.19452

-0.03699

_cons

-0.13306

0.095294 -1.4

0.164

-0.32054

0.054432

106	
  
	
  

Table B.10 311 Food and Kindred Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.18462

0.092756 -1.99

0.047

-0.36689

-0.00235

loggdpus

0.004231 0.122696 0.03

0.973

-0.23687

0.245338

logpop

0.398755 0.05991

0

0.281028 0.516481

logpopus

0.043662 0.188519 0.23

0.817

-0.32679

0.414115

logexrate

-0.00114

0.971

-0.06176

0.059478

logserv

0.711624 0.072082 9.87

0

0.569978 0.853269

logservus

-0.16072

0.224737 -0.72

0.475

-0.60234

0.280904

logdist

-0.11454

0.030549 -3.75

0

-0.17457

-0.05451

_cons

0.065883 0.061862 1.07

0.287

-0.05568

0.187446

6.66

0.030848 -0.04

107	
  
	
  

Table B.11 113 Forestry Products Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.014146 0.183627 0.08

0.939

-0.34705

0.375345

loggdpus

0.015967 0.200512 0.08

0.937

-0.37845

0.41038

logpop

0.382458 0.108205 3.53

0

0.169617 0.595299

logpopus

-0.26591

-0.85

0.393

-0.87779

0.345976

logexrate

0.049501 0.05748

0.86

0.39

-0.06356

0.162566

logserv

0.294794 0.143966 2.05

0.041

0.011608 0.57798

logservus

0.17207

0.47

0.641

-0.55262

0.896762

logdist

0.029023 0.046061 0.63

0.529

-0.06158

0.119626

_cons

-0.08224

0.425

-0.2847

0.120215

0.31107

0.36842

0.102926 0.8

108	
  
	
  

Table B.12 337 Furniture and Fixtures
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.188415 0.098457 1.91

0.056

-0.00512

0.38195

loggdpus

0.072763 0.120445 0.6

0.546

-0.16399

0.309521

logpop

0.206944 0.061265 3.38

0.001

0.086518 0.327371

logpopus

-0.13779

0.185832 -0.74

0.459

-0.50308

0.227494

logexrate

-0.06829

0.030532 -2.24

0.026

-0.1283

-0.00827

logserv

0.538908 0.076039 7.09

0

0.38944

0.688375

logservus

-0.07264

0.221804 -0.33

0.743

-0.50864

0.363356

logdist

-0.1079

0.029027 -3.72

0

-0.16496

-0.05084

_cons

-0.05363

0.0613

0.382

-0.17412

0.066868

0.87

109	
  
	
  

Table B.13 316 Leather and Allied Products Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.41388

0.135824 -3.05

0.002

-0.68077

-0.14698

loggdpus

0.145447 0.180812 0.8

0.422

-0.20985

0.500746

logpop

0.127875 0.087935 1.45

0.147

-0.04492

0.300668

logpopus

-0.39564

0.153

-0.93815

0.14687

logexrate

0.090315 0.045227 2

0.046

0.001443 0.179188

logserv

0.498816 0.105447 4.73

0

0.291611 0.706021

logservus

0.167301 0.329587 0.51

0.612

-0.48034

0.814945

logdist

0.036487 0.045038 0.81

0.418

-0.05201

0.124987

_cons

0.086993 0.091098 0.95

0.34

-0.09202

0.266001

0.276085 -1.43

110	
  
	
  

Table B.14 112 Livestock Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.702

0.148825 -4.72

0

-0.99448

-0.40951

loggdpus

0.006837 0.192769 0.04

0.972

-0.37201

0.385684

logpop

0.573186 0.094736 6.05

0

0.387002 0.759369

logpopus

-0.5427

0.068

-1.12595

logexrate

0.102001 0.048579 2.1

0.036

0.006529 0.197472

logserv

0.717319 0.116188 6.17

0

0.488977 0.94566

logservus

0.256452 0.354022 0.72

0.469

-0.4393

0.952205

logdist

-0.04243

0.372

-0.13584

0.050981

_cons

0.014706 0.098481 0.15

0.881

-0.17884

0.208249

0.296774 -1.83

0.047531 -0.89

111	
  
	
  

0.040542

Table B.15 333 Machinery Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.264997 0.061999 4.27

0

0.143149 0.386844

loggdpus

0.11163

0.081007 1.38

0.169

-0.04757

logpop

0.086333 0.039949 2.16

0.031

0.007821 0.164846

logpopus

-0.34035

0.12445

-2.73

0.006

-0.58493

-0.09577

logexrate

-0.00326

0.02046

-0.16

0.873

-0.04347

0.036945

logserv

0.676744 0.047444 14.26

0

0.583503 0.769985

logservus

0.06293

0.148325 0.42

0.672

-0.22858

0.354435

logdist

-0.00791

0.019765 -0.4

0.689

-0.04676

0.030931

_cons

0.024047 0.040906 0.59

0.557

-0.05635

0.10444

112	
  
	
  

0.270833

Table B.16 327 Nonmetallic Mineral Products
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.312985 0.085075 3.68

0

0.145766 0.480205

loggdpus

0.064109 0.109894 0.58

0.56

-0.15189

logpop

0.213855 0.053771 3.98

0

0.108164 0.319545

logpopus

-0.1945

0.249

-0.52591

0.13691

logexrate

0.018904 0.027377 0.69

0.49

-0.03491

0.072716

logserv

0.47875

0.066452 7.2

0

0.348136 0.609365

logservus

-0.03587

0.2018

0.859

-0.43252

0.360784

logdist

-0.09386

0.026248 -3.58

0

-0.14545

-0.04227

_cons

-0.05983

0.056302 -1.06

0.289

-0.17049

0.050836

0.168608 -1.15

-0.18

113	
  
	
  

0.280112

Table B.17 511 Newspaper Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

1.426555 0.245628 5.81

0

0.941164 1.911945

loggdpus

-0.2968

0.239003 -1.24

0.216

-0.7691

0.175495

logpop

-0.53065

0.134136 -3.96

0

-0.79572

-0.26558

logpopus

-0.82103

0.337261 -2.43

0.016

-1.4875

-0.15456

logexrate

-0.16856

0.068575 -2.46

0.015

-0.30408

-0.03305

logserv

-0.22565

0.19825

0.257

-0.61742

0.166113

logservus

0.872561 0.399686 2.18

0.031

0.082732 1.66239

logdist

-0.03452

0.042514 -0.81

0.418

-0.11853

0.049493

_cons

-0.79145

0.134849 -5.87

0

-1.05792

-0.52497

-1.14

114	
  
	
  

Table B.18 211 Oil and Gas Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.642726 0.186227 3.45

0.001

0.276131 1.009321

loggdpus

-0.05352

0.242283 -0.22

0.825

-0.53046

0.42342

logpop

-0.01686

0.115898 -0.15

0.884

-0.24501

0.211292

logpopus

-0.21558

0.367715 -0.59

0.558

-0.93944

0.508278

logexrate

0.051765 0.068659 0.75

0.452

-0.08339

0.186922

logserv

-0.30136

0.032

-0.5766

-0.02611

logservus

0.334794 0.441261 0.76

0.449

-0.53384

1.203432

logdist

-0.05167

0.049809 -1.04

0.3

-0.14972

0.046383

_cons

-0.08746

0.125577 0.7

0.487

-0.33467

0.159738

0.139822 -2.16

115	
  
	
  

Table B.19 322 Paper Industry
logimports Coef.

Std. Err.

t

P>t

[95 Conf.

Interval]

loggdp

0.0262451 1.256974 0.21

0.835

-.2207588

0.2732489

loggdpus

0.138371

0.167689 0.83

0.410

-.1911488

0.467892

logpop

0.085947

.0813482 1.06

0.291

-0.739075

0.2458018

logpopus

0.076553

-0.25469

0.764

-.5770381

0.423932

logexrate

0.098299

0.042037 2.34

0.02

0.015694

0.180904

logserv

.4300322

.0974211 4.41

.000

.2385931

.6214713

logservus

-0.07952

0.30429

0.794

-0.677475

0.518426

logdist

-.0250036

.0415246 -.60

0.547

-.1066023

0.0565951

_cons

0.0413514 0.084074 0.49

0.623

-.1238598

0.2065626

-0.3

-0.26

116	
  
	
  

Table B.20 326 Plastics Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.000335 0.122828 0

0.998

-0.24102

0.241693

loggdpus

0.031068 0.163512 0.19

0.849

-0.29023

0.35237

logpop

0.030113 0.079521 0.38

0.705

-0.12615

0.186372

logpopus

0.087903 0.249668 0.35

0.725

-0.4027

0.578504

logexrate

-0.04344

0.289

-0.12381

0.036931

logserv

0.202094 0.095358 2.12

0.035

0.014715 0.389473

logservus

-0.11457

0.298051 -0.38

0.701

-0.70025

0.4711

logdist

-0.0648

0.040728 -1.59

0.112

-0.14483

0.015234

_cons

0.101313 0.082381 1.23

0.219

-0.06057

0.263192

0.0409

-1.06

117	
  
	
  

Table B.21 323 Printed Matter and Related Products
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

0.405238 0.075421 5.37

0

0.256984 0.553493

loggdpus

0.107945 0.094684 1.14

0.255

-0.07818

0.294065

logpopus

-0.07895

0.147435 -0.54

0.593

-0.36876

0.210861

logpop

0.13

0.048079 2.7

0.007

0.035492 0.224509

logexrate

0.038183 0.024112 1.58

0.114

-0.00921

logserv

0.500388 0.057191 8.75

0

0.387969 0.612808

logservus

-0.1588

0.175431 -0.91

0.366

-0.50365

0.186041

logdist

-0.08983

0.022948 -3.91

0

-0.13494

-0.04473

_cons

-0.1173

0.048968 -2.4

0.017

-0.21356

-0.02104

118	
  
	
  

0.08558

Table B.22 313 Textile Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.16621

0.127567 -1.3

0.193

-0.41689

0.084469

loggdpus

0.101689 0.170155 0.6

0.55

-0.23268

0.436056

logpop

0.167939 0.082616 2.03

0.043

0.005592 0.330286

logpopus

0.077162 0.258723 0.3

0.766

-0.43125

logexrate

0.149528 0.042598 3.51

0

0.065821 0.233236

logserv

0.365804 0.099331 3.68

0

0.170611 0.560998

logservus

-0.1806

0.309073 -0.58

0.559

-0.78795

0.426752

logdist

-0.05075

0.042098 -1.21

0.229

-0.13348

0.031974

_cons

0.077049 0.085507 0.9

0.368

-0.09098

0.245077

119	
  
	
  

0.585573

Table B.23 336 Transportation Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

1.1

0.272

-0.10582

0.375293

loggdp

0.134735 0.12242

loggdpus

0.051924 0.162969 0.32

0.75

-0.26831

0.372161

logpop

-0.03108

0.079257 -0.39

0.695

-0.18682

0.124664

logpopus

-0.04215

0.24884

0.866

-0.53113

0.446821

logexrate

0.027111 0.040764 0.67

0.506

-0.05299

0.107213

logserv

0.18737

0.095041 1.97

0.049

0.000612 0.374128

logservus

-0.00266

0.297063 -0.01

0.993

-0.5864

0.581068

logdist

-0.12116

0.040593 -2.98

0.003

-0.20093

-0.04139

_cons

0.157787 0.082108 1.92

0.055

-0.00356

0.31913

-0.17

120	
  
	
  

Table B.24 321 Wood Industry
[95%
logimports Coef.

Std. Err.

t

P>t

Conf.

Interval]

loggdp

-0.61342

0.121527 -5.05

0

-0.85223

-0.37462

loggdpus

0.230696 0.161481 1.43

0.154

-0.08662

0.548015

logpop

0.36201

0.078445 4.61

0

0.207862 0.516158

logpopus

-0.15544

0.246312 -0.63

0.528

-0.63945

0.328579

logexrate

0.059183 0.040342 1.47

0.143

-0.02009

0.138458

logserv

0.526411 0.094334 5.58

0

0.34104

0.711783

logservus

-0.08017

0.785

-0.65808

0.497744

logdist

0.039144 0.04004

0.98

0.329

-0.03954

0.117825

_cons

0.052597 0.081096 0.65

0.517

-0.10676

0.211955

0.294093 -0.27

121	
  
	
  

Table B.25 212 Minerals and Ore
[95%
logimports Coef.

Std. Err.

loggdp

-0.12061

loggdpus

-0.15167

logpop

t

P>t

Conf.

Interval]

0.204529 -0.59

0.556

-0.5232

0.281971

0.260524 -0.58

0.561

-0.66447

0.361131

0.042427 0.131162 0.32

0.747

-0.21575

0.300599

logpopus

-0.12651

0.402629 -0.31

0.754

-0.91903

0.666003

logexrate

-0.0244

0.070516 -0.35

0.73

-0.1632

0.114395

logserv

0.089691 0.161863 0.55

0.58

-0.22891

0.408295

logservus

0.19615

0.477506 0.41

0.682

-0.74375

1.13605

logdist

0.016198 0.062252 0.26

0.795

-0.10634

0.138733

_cons

0.044138 0.132096 0.33

0.739

-0.21587

0.30415

122	
  
	
  

Appendix C: Summary Statistics
Table C.1 111 Agriculture Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

645 5.026783 1.670139

0

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

123	
  
	
  

6.39693

13.0103

Table C.2 111 Agriculture Correlation and Covariances
Variable

logimports loggdp

logimports

1

loggdp

loggdpus logpop

logpopus logexrate logserv

0.0964

1

-0.0627

0.0827

1

0.1532

0.6845

0.0759

1

logpopus

-0.0626

0.1016

0.9009

0.093

1

logexrate

-0.1032 -0.1851

0.0401

0.0716

0.0307

1

logserv

-0.0134

0.7915

0.2268

0.2353

0.2628

-0.3482

1

logservus

-0.0747

0.0932

0.9469

0.0865

0.9813

0.0358

0.2538

0.084

0.04

0.0277

0.3101

0.0293

0.2313 -0.1726

loggdpus
logpop

logdist

124	
  
	
  

Table C.3 112 Livestock & Livestock Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

594

4.43017 2.077538

0 6.216606

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

125	
  
	
  

13.0103

Table C.4 112 Livestock & Livestock Products Correlations and Covariance
Variable

logimports loggdp

logimports

1

loggdp

loggdpus logpop

logpopus logexrate logserv

0.2205

1

-0.0326

0.0778

1

0.2221

0.6868

0.0738

1

-0.0495

0.0963

0.8999

0.0948

1

logexrate

0.0045 -0.1573

0.0345

0.0883

0.0339

1

logserv

0.2214

0.7947

0.2202

0.2381

0.2528

-0.3287

1

logservus

-0.0397

0.0897

0.9466

0.0873

0.9814

0.0341

0.2466

logdist

-0.0057

0.0657

0.0326

0.3302

0.0353

0.2211 -0.1578

loggdpus
logpop
logpopus

126	
  
	
  

Table C.5 114 Fish, Fresh/Chilled/Frozen and Other Marine Basic Statistics
Variable

Obs

Mean

Std.

Min

Max

logimports

425 15.65378 3.117196 7.696213 21.41443

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

127	
  
	
  

13.0103

Table C.6 114 Fish, Fresh/Chilled/Frozen and Other Marine Correlations and Covariances
Variable

logimports loggdp

logimports

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.5049

1

loggdpus

0.0652

0.0954

1

logpop

0.4111

0.6962

0.0951

1

logpopus

0.0676

0.1215

0.9036

0.124

1

-0.3126 -0.1113

-0.0024

0.1828

-0.0033

1

logexrate
logserv

0.4396

0.7912

0.2659

0.2249

0.2985

-0.3482

1

logservus

0.0676

0.1121

0.949

0.114

0.981

-0.0002

0.2902

-0.0833

0.1017

0.0135

0.3347

0.0164

logdist

128	
  
	
  

0.2497 -0.1214

Table C.7 211 Oil and Gas Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

376 2.986719 2.285671

0 5.513429

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

129	
  
	
  

13.0103

Table C.8 211 Oil and Gas Correlations and Covariance
Variable

logimports loggdp

logimports

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.3875

1

loggdpus

0.0562

0.0365

1

logpop

0.3579

0.7424

0.0641

1

logpopus

0.0523

0.0444

0.8994

0.089

1

logexrate

0.0533 -0.0974

-0.0718

0.0571

-0.0569

1

logserv

0.226

0.804

0.2113

0.3409

0.2267

-0.2874

1

logservus

0.0583

0.041

0.947

0.0798

0.9812

-0.0599

0.2264

logdist

0.0485

0.1501

-0.0025

0.302

0.0166

130	
  
	
  

0.1497 -0.0244

Table C.9 311 Food and Kindred Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

629 15.85557 3.360476 7.667158 23.11375

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

131	
  
	
  

13.0103

Table C.10 311 Food and Kindred Products Correlations and Covariances
Variable

logimports loggdp

logimports

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.6799

1

loggdpus

0.0687

0.0871

1

0.408

0.6852

0.0834

1

0.0969

0.1043

0.9002

0.0987

1

-0.2326 -0.1764

0.0326

0.0753

0.0254

1

logpop
logpopus
logexrate
logserv

0.7045

0.7921

0.2257

0.2363

0.2618

-0.3437

1

logservus

0.0849

0.0966

0.9464

0.0933

0.9813

0.0299

0.2519

-0.1365

0.0473

0.0276

0.3138

0.0302

0.2211 -0.1666

logdist

132	
  
	
  

Table C.11 312 Beverages and Tobacco Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

492 14.47918 3.027795

7.72533 20.63566

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

133	
  
	
  

13.0103

Table C.12 312 Beverages and Tobacco Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.6517

1

loggdpus

0.0569

0.0806

1

logpop

0.2775

0.7368

0.0863

1

logpopus

0.0778

0.0978

0.9002

0.1071

1

-0.2093 -0.0837

0.0266

0.0975

0.034

1

logexrate
logserv

0.7778

0.7775

0.2232

0.2844

0.2588

-0.2923

1

logservus

0.0691

0.0915

0.9467

0.1013

0.9813

0.0354

0.2515

-0.2208

0.1289

0.0427

0.3264

0.0455

logdist

134	
  
	
  

0.176 -0.0741

Table C.13 313 Textiles and Fabrics Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

634

4.3212 2.200151

0 6.251904

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

135	
  
	
  

13.0103

Table C.14 313 Textiles and Fabrics Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.2359

1

loggdpus

0.0713

0.0883

1

logpop

0.1444

0.6843

0.0795

1

logpopus

0.0745

0.1028

0.902

0.094

1

logexrate

0.0567 -0.1813

0.0345

0.0701

0.0278

1

logserv

0.2586

0.7893

0.236

0.2319

0.2665

-0.3458

1

logservus

0.0713

0.0959

0.9477

0.0883

0.9814

0.031

0.2594

-0.0436

0.0435

0.0278

0.3109

0.0307

logdist

136	
  
	
  

0.2245 -0.1687

Table C.15 315 Apparel and Accessories Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

647 5.259247

1.37909

0 6.437752

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

137	
  
	
  

13.0103

Table C.16 315 Apparel and Accessories Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.0629

1

loggdpus

0.0168

0.0841

1

0.007

0.6845

0.0762

1

logpopus

-0.002

0.1029

0.9013

0.0933

1

logexrate

0.0011 -0.1863

0.0318

0.0703

0.023

1

logserv

0.1108

0.7911

0.2299

0.2353

0.2658

-0.3523

1

logservus

0.0047

0.0946

0.9473

0.0868

0.9813

0.0276

0.2568

-0.0226

0.0402

0.0283

0.3101

0.0299

logpop

logdist

138	
  
	
  

0.229 -0.1717

Table C.17 316 Leather and Allied Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

647 5.031027 1.669294

0 6.400258

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

139	
  
	
  

13.0103

Table C.18 316 Leather and Allied Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.0559

1

loggdpus

0.0621

0.0841

1

-0.0314

0.6845

0.0762

1

logpopus

0.0317

0.1029

0.9013

0.0933

1

logexrate

0.0056 -0.1863

0.0318

0.0703

0.023

1

logserv

0.1655

0.7911

0.2299

0.2353

0.2658

-0.3523

1

logservus

0.0461

0.0946

0.9473

0.0868

0.9813

0.0276

0.2568

-0.0112

0.0402

0.0283

0.3101

0.0299

logpop

logdist

140	
  
	
  

0.229 -0.1717

Table C.19 321 Wood Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

632 4.887999 1.788474

0 6.356108

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

141	
  
	
  

13.0103

Table C.20 321 Wood Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.0499

1

loggdpus

0.1056

0.0895

1

logpop

0.0853

0.6826

0.0827

1

logpopus

0.0798

0.1087

0.9011

0.0986

1

logexrate

0.0199 -0.1891

0.027

0.0709

0.0197

1

logserv

0.1163

0.7923

0.2287

0.2332

0.2648

-0.3574

1

0.089

0.0998

0.9472

0.0923

0.9812

0.0233

0.255

0.0528

0.0408

0.0301

0.3112

0.0313

0.2303 -0.1711

logservus
logdist

142	
  
	
  

Table C.21 322 Paper Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

630 3.819961 2.409088

0 6.144186

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

143	
  
	
  

13.0103

Table C.22 322 Paper Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.4299

1

loggdpus

0.0971

0.0915

1

logpop

0.2113

0.6888

0.0768

1

logpopus

0.0911

0.1055

0.902

0.0939

1

-0.0568 -0.1698

0.0211

0.0723

0.0201

1

logexrate
logserv
logservus
logdist

0.463

0.7883

0.2418

0.2372

0.2713

-0.3354

1

0.0924

0.0984

0.9477

0.0874

0.9813

0.0225

0.2639

-0.0523

0.0494

0.0227

0.3117

0.028

144	
  
	
  

0.2182 -0.1627

Table C.23 323 Printed Matter and Related Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

533 12.94797 3.321482 7.649693 21.55362

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

145	
  
	
  

13.0103

Table C.24 323 Printed Matter and Related Products Correlations and Covariances
Variables

logimports loggdp

logimports

1

loggdp

loggdpus logpop

logpopus logexrate logserv

0.8539

1

0.005

0.0572

1

logpop

0.4814

0.6939

0.0666

1

logpopus

0.0024

0.0646

0.9001

0.0746

1

-0.2142 -0.1703

0.0391

0.0733

0.0363

1

loggdpus

logexrate
logserv

0.8065

0.7878

0.2049

0.2386

0.2347

-0.3371

1

logservus

0.0024

0.0635

0.9462

0.0722

0.9818

0.0368

0.2311

-0.0859

0.0728

0.0433

0.3072

0.043

logdist

146	
  
	
  

0.2118 -0.1232

Table C.25 324 Petroleum and Coal Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

612 3.015609 2.524168

0 5.924256

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

147	
  
	
  

13.0103

Table C.26 324 Petroleum and Coal Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.4452

1

loggdpus

0.0212

0.0742

1

0.149

0.6955

0.0815

1

0.0132

0.0883

0.901

0.0998

1

-0.1993 -0.1467

0.0353

0.0781

0.0347

1

logpop
logpopus
logexrate
logserv

0.4254

0.7805

0.2249

0.2358

0.2538

-0.3187

1

logservus

0.0147

0.0828

0.9471

0.0929

0.9815

0.0362

0.2488

-0.2413

0.0726

0.0364

0.3197

0.0428

0.2081 -0.1444

logdist

148	
  
	
  

Table C.27 326 Plastic & Rubber Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

648 4.620459 2.035555

0 6.329721

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

149	
  
	
  

13.0103

Table C.28 326 Plastic & Rubber Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.2226

1

loggdpus

0.0394

0.0841

1

logpop

0.0634

0.6845

0.0762

1

logpopus

0.0505

0.1029

0.9013

0.0933

1

-0.1537 -0.1863

0.0318

0.0703

0.023

1

logexrate
logserv

0.2804

0.7911

0.2299

0.2353

0.2658

-0.3523

1

logservus

0.0451

0.0946

0.9473

0.0868

0.9813

0.0276

0.2568

-0.1207

0.0402

0.0283

0.3101

0.0299

logdist

150	
  
	
  

0.229 -0.1717

Table C.29 327 Non Metallic Mineral Ores Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

546 14.53994 3.601517 5.568345 22.36176

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

151	
  
	
  

13.0103

Table C.30 327 Non Metallic Mineral Ores Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.8173

1

loggdpus

0.0051

0.0738

1

logpop

0.4968

0.6905

0.0873

1

logpopus

0.0086

0.0949

0.9002

0.1083

1

-0.2045 -0.1636

0.0298

0.0927

0.023

1

logexrate
logserv

0.7505

0.7912

0.2137

0.2389

0.2475

-0.345

1

logservus

0.0074

0.0876

0.9469

0.1009

0.9814

0.0253

0.2423

-0.0796

0.0714

0.0329

0.3155

0.0348

0.2176 -0.1413

logdist

152	
  
	
  

Table C.31 332 Fabricated Metal Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

566 14.26057 4.093736 7.613325 23.42445

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

153	
  
	
  

13.0103

Table C.32 332 Fabricated Metal Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.8443

1

loggdpus

0.0135

0.0675

1

logpop

0.4397

0.6942

0.0751

1

logpopus

0.0248

0.091

0.9003

0.093

1

logexrate

-0.2672

-0.171

0.0381

0.0827

0.0303

1

logserv

0.8421

0.7919

0.2119

0.245

0.2511

-0.3458

1

logservus

0.0209

0.082

0.9467

0.0856

0.9813

0.0336

0.243

-0.0915

0.0741

0.0368

0.3151

0.0372

logdist

154	
  
	
  

0.216 -0.1367

Table C.33 333 Machinery, Except Electrical Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

588 14.85973 4.166297 7.634337 23.93851

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

155	
  
	
  

13.0103

Table C.34 333 Machinery, Except Electrical Correlations and Covariances
logimports loggdp
logimports
loggdp

loggdpus logpop

logpopus logexrate logserv

logservus

1
0.8475

1

-0.0078

0.032

1

0.4598

0.7268

0.0807

1

logpopus

-0.0154

0.0498

0.8931

0.1055

1

logexrate

-0.2602

-0.1414

0.0465

0.0793

0.0457

0.8006

0.8082

0.2801

0.472

logservus

-0.0172

0.0342

0.9642

0.0885

0.9499

0.0476

0.2903

logdist

-0.0259

0.0974

0.075

0.3057

0.0832

0.2206

-0.0525

loggdpus
logpop

logserv

156	
  
	
  

1

0.298 -0.2441

1

Table C.35 334 Computer and Electronic Products Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

635 14.98132 4.367618 6.309918 25.61189

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

157	
  
	
  

13.0103

Table C.36 334 Computer and Electronic Products Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.8172

1

loggdpus

0.0447

0.0871

1

logpop

0.4254

0.6858

0.0764

1

logpopus

0.0495

0.1038

0.9017

0.0936

1

-0.2425 -0.1794

0.0264

0.0709

0.0221

1

logexrate
logserv
logservus
logdist

0.82

0.7933

0.2344

0.2388

0.267

-0.346

1

0.0476

0.0958

0.9475

0.0869

0.9813

0.0256

0.259

-0.0711

0.0463

0.0264

0.3125

0.0297

0.2233 -0.1689

158	
  
	
  

Table C.37 335 Electrical Equipment, Appliances & Components Basic Statistics
Variable

Obs

Mean

Std. Dev

Min

Max

logimports

648 4.490916

2.11918

0 6.302619

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

159	
  
	
  

13.0103

Table C.38 335 Electrical Equipment, Appliances & Comps Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.3009

1

loggdpus

0.0708

0.0841

1

logpop

0.0808

0.6845

0.0762

1

logpopus

0.0763

0.1029

0.9013

0.0933

1

-0.2542 -0.1863

0.0318

0.0703

0.023

1

logexrate
logserv

0.3836

0.7911

0.2299

0.2353

0.2658

-0.3523

1

logservus

0.0722

0.0946

0.9473

0.0868

0.9813

0.0276

0.2568

logdist

-0.111

0.0402

0.0283

0.3101

0.0299

160	
  
	
  

0.229 -0.1717

Table C.39 336 Transportation Equipment Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

648 4.206718 2.273535

0 6.244167

loggdp

578 23.97549 1.863965

19.4306 29.27214

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

0.04693 29.98035 30.11611

161	
  
	
  

13.0103

Table C.40 336 Transportation Equipment Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.2924

1

loggdpus

0.0648

0.0841

1

logpop

0.0807

0.6845

0.0762

1

logpopus

0.0654

0.1029

0.9013

0.0933

1

logexrate

-0.109 -0.1863

0.0318

0.0703

0.023

1

logserv

0.3459

0.7911

0.2299

0.2353

0.2658

-0.3523

1

logservus

0.0656

0.0946

0.9473

0.0868

0.9813

0.0276

0.2568

-0.1757

0.0402

0.0283

0.3101

0.0299

logdist

162	
  
	
  

0.229 -0.1717

Table C.41 337 Furniture and Fixtures Basic Statistics
Variable

Obs

Mean

Std. Dev

Min

Max

logimports

544 14.19315 3.610159 5.843544 23.44755

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

163	
  
	
  

13.0103

Table C.42 337 Furniture and Fixtures Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.7553

1

loggdpus

0.0212

0.061

1

logpop

0.3922

0.6699

0.0564

1

logpopus

0.0244

0.081

0.899

0.076

1

-0.3179 -0.1905

0.0444

0.0765

0.0436

1

logexrate
logserv

0.7503

0.7977

0.2243

0.2184

0.2549

-0.3534

1

logservus

0.0205

0.0723

0.9459

0.0696

0.9813

0.0463

0.2466

-0.1446

0.053

0.0132

0.3035

0.0159

logdist

164	
  
	
  

0.225 -0.1484

Table C.43 511 Newspapers, Books & Other Published Matter Basic Statistics
Variable

Obs

Mean

Std.Dev

Min

Max

logimports

160 12.14138 2.373729 7.646832 16.83724

loggdp

578 23.97549 1.863965

loggdpus

648 30.06781

logpop

648 16.12662 1.599653 11.12368 21.01467

logpopus

648 19.51439 0.023611 19.47779 19.54902

logexrate

643 3.265752 2.709218

-2.89864 22.62881

logserv

549 4.127796 2.641877

0 11.32446

logservus

576 12.52186 0.360845 11.83873

logdist

648 8.967749 0.541395 6.598359 9.677719

19.4306 29.27214

0.04693 29.98035 30.11611

165	
  
	
  

13.0103

Table C.44 511 Newspapers, Books & Other Published Matter Correlations and Covariances
Variable
logimports

logimports loggdp

loggdpus logpop

logpopus logexrate logserv

1

loggdp

0.6706

1

loggdpus

0.0933

0.2477

1

0.273

0.7102

0.1311

1

logpopus

0.0918

0.2781

0.9153

0.1453

1

logexrate

-0.1017

0.0607

-0.0584

0.1204

-0.0259

1

logserv

0.6199

0.7743

0.4592

0.2178

0.5006

-0.121

1

logservus

0.1027

0.2654

0.9565

0.142

0.98

-0.0307

0.4867

-0.0875

0.024

0.0155

0.1598

0.0144

logpop

logdist

166	
  
	
  

0.0576 -0.0869

REFERENCES

167	
  
	
  

REFERENCES
Acemoglu, D., Antras, P., & Helpman, E. (2007). Contracts and Technology Adoption. The
American Economic Review, 97(3), 916-943.
Acemoglu, D., Johnson, S., & Robinson, J. A. (2005). Institutions as a fundamental cause of
long-run growth. Handbook of economic growth, Ed. Aghion P., Durlauf S.
Amsterdam:Elsevier, 385-472.
Affortunato, F., Ciommi, M., Crociata, A., & Mattoscio, N. The Performances of Italian Vertical
Intra-Industry Trade in Cultural vs ICT Goods. Procedia-Social and Behavioral Sciences, 81,
200-205.
Anderson, J. E. (1979). A Theoretical Foundation for the Gravity Equation. The American
Economic Review, 69(1), 106-116.
Anderson, J. E., & Marcouiller, D. (2002). Insecurity and the Pattern of Trade: An Empirical
Investigation. Review of Economics and Statistics, 84(2), 342-352.
Baier, S. L., & Bergstrand, J. H. (2001). The Growth of World Trade: Tariffs, Transport Costs,
and Income Similarity. Journal of International Economics, 53(1), 1-27.
Baier, S. L., & Bergstrand, J. H. (2009). Bonus vetus OLS: A Simple Method for Approximating
International Trade-Cost Effects using the Gravity Equation. Journal of International Economics,
77(1), 77-85.
Baldwin, J. R., & Gu, W. (2004). Trade liberalization: Export-Market Participation, Productivity
Growth, and Innovation. Oxford Review of Economic Policy, 20(3), 372-392.
Barney, J. (1991). Firm Resources and Sustained Competitive Advantage. Journal of
Management, 17(1), 99-120
Berthelon, M., & Freund, C. (2008). On the Conservation of Distance in International Trade.
Journal of International Economics, 75(2), 310-320.
Blum, B. S., & Goldfarb, A. (2006). Does the Internet Defy the Law of Gravity? Journal of
International Economics, 70(2), 384-405.
Bresnahan, T. F., Brynjolfsson, E., & Hitt, L. M. (2002). Information Technology, Workplace
Organization, and the Demand for Skilled Labor: Firm-Level Evidence. The Quarterly Journal of
Economics, 117(1), 339-376.
Bresnahan, T. F., & Trajtenberg, M. (1995). General Purpose Technologies 'Engines of Growth'?
Journal of Econometrics, 65(1), 83-108.
Brynjolfsson, E. (1993). The Productivity Paradox of Information Technology. Communications
of the ACM, 36(12), 66-77.
Brynjolfsson, E., & Hitt, L. (1996). Paradox Lost? Firm-Level Evidence on the Returns to
Information Systems Spending. Management Science 42(4), 541-558.
168	
  
	
  

Brynjolfsson, E., & Hitt, L. M. (2000). Beyond Computation: Information Technology,
Organizational Transformation and Business Performance. The Journal of Economic
Perspectives, 14(4), 23-48.
Caballero, R. J., & Hammour, M. L. (1998). Jobless Growth: Appropriability, Factor
Substitution, and Unemployment. Paper presented at the Carnegie-Rochester Conference Series
on Public Policy, 48, 51-94
Chalmeta, R., Campos, C., & Grangel, R. (2001). References architectures for enterprise
integration. Journal of Systems and Software, 57(3), 175-191.
Choi, C. (2010). The Effect of the Internet on Service Trade. Economics Letters, 109(2), 102104.
Choi, Y. J., Lee, S. Y., & Wildman, S. S. (2011) Why Do Large Markets Dominate? A New
Look at Sources of the Large Home Market Advantage in the International Film. Paper presented
at the 101st AEA International Conference on Media Studies and Communication.
Clark, H. H., & Brennan, S. E. (1991). Grounding in Communication. Perspectives on Socially
Shared Cognition, 13(1991), 127-149.
Clemons, E. K., Reddi, S. P., & Row, M. C. (1993). The Impact of Information Technology on
the Organization of Economic Activity: The "Move to the Middle" Hypothesis. Journal of
Management Information System, 10(2), 9-35.
Coase, R. H. (1937). The Nature of the Firm. Economica, 4(16), 386-405.
Corso, M., & Gastaldi, L. (2010) Managing ICT-driven Innovations to Overcome the
Exploitation-Exploration Trade-off: A Multiple and Collaborative Research Methodology in the
Italian Health Care Industry. Paper presented at the 11th Int. CINet Conference, Zurich (CH).
Costinot, A. (2005). Contract Enforcement, Division of Labor, and the Pattern of Trade.
Mimeograph, Princeton University.
Crandall, R., Lehr, W., & Litan, R. (2007) The Effects of Broadband Deployment on Output and
Employment: A Cross-sectional Analysis of US Data. Issues in Economic Policy, 6, 1-34
Cuihua Shen, & Williams, D. Unpacking Time Online: Connecting Internet and Massively
Multiplayer Online Game Use with Psychosocial Well-Being. Communication Research, 38(1),
123-149.
Daft, R. L., Lengel, R. H., & Trevino, L. K. (1987). Message Equivocality, Media Selection, and
Manager Performance: Implications for Information Systems. MIS Quarterly, 11(3), 355-366.
David, P. A. (1990). The Dynamo and the Computer: An Historical Perspective on the Modern
Productivity Paradox. The American Economic Review, 80(2), 355-361.
Day, G. S., & Hubbard, K. J. (2003). Customer Relationships go Digital. Business Strategy
Review, 14(1), 17-26.

169	
  
	
  

Deardorff, A. V. (1995). Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical
World? Cambridge, MA: National Bureau of Economic Research.
Dedrick, J., Gurbaxani, V., & Kraemer, K. L. (2003). Information Technology and Economic
Performance: A Critical Review of the Empirical Evidence. ACM Computing Surveys, 35(1), 128.
Dimelis, S. P., & Papaioannou, S. K. ICT Growth Effects at the Industry Level: A Comparison
between the US and the EU. Information Economics and Policy, 23(1), 37-50.
Draca, M., Sadun, R., & Van Reenen, J. (2006). Productivity and ICT: A Review of the
Evidence. Centre for Economic Performance, London School of Economics and Political
Science, London, UK.
Egger, P. (2000). A Note on the Proper Econometric Specification of the Gravity Equation.
Economics Letters, 66(1), 25-31.
Egger, P. (2002). An Econometric View on the Estimation of Gravity Models and the
Calculation of Trade Potentials. The World Economy, 25(2), 297-312.
Eid, R., Trueman, M., & Ahmed, A. (2002). A Cross-Industry Review of B2B Critical Success
Factors. Internet Research, 12(2), 110-123.
Erumban, A. A., & de Jong, S. B. (2006). Cross-Country Differences in ICT Adoption: A
Consequence of Culture? Journal of World Business, 41(4), 302-314.
Evenett, S. (2003). Do All Networks Facilitate International Commerce? The Case of US Law
Firms and the Mergers and Acquisitions Wave of the Late 1990s.	
  Journal of the Japanese and
International Economies, 17(4), 520-537.
Fink, C., Mattoo, A., & Neagu, I. C. (2005). Assessing the Impact of Communication Costs on
International Trade. Journal of International Economics, 67(2), 428-445.
Forman, C., Goldfarb, A., & Greenstein, S. M. (2009). The Internet and Local Wages:
Convergence or Divergence? Cambridge, Mass: National Bureau of Economic Research.
Frankel, J., Stein, E., & Wei, S.-j. (1995). Trading Blocs and the Americas: The Natural, the
Unnatural, and the Super-natural. Journal of Development Economics, 47(1), 61-95.
Freund, C., & Weinhold, D. (2002). The Internet and International Trade in Services. The
American Economic Review, 92(2), 236-240.
Freund, C. L., & Weinhold, D. (2000). An Empirical Investigation of the Internet and
International Trade: The Case of Bolivia. Revista Latinoamericana de Desarrollo Economico, 2,
33-55.
Freund, C. L., & Weinhold, D. (2004). The Effect of the Internet on International Trade. Journal
of International Economics, 62(1), 171-189.

170	
  
	
  

Griffith, R., Redding, S., & Van Reenen, J. (2004). Mapping the Two Faces of R&D:
Productivity Growth in a Panel of OECD Industries. Review of Economics and Statistics, 86(4),
883-895.
Grossman, G. M., & Helpman, E. (2002). Integration versus Outsourcing in Industry
Equilibrium. The Quarterly Journal of Economics, 117(1), 85-120.
Grossman, S. J., & Hart, O. D. (1986). The Costs and Benefits of Ownership: A Theory of
Vertical and Lateral Integration. The Journal of Political Economy, 94(4), 691-719.
Hamill, J., & Gregory, K. (1997). Internet Marketing in the Internationalization of UK SMEs.
Journal of Marketing Management, 13(1-3), 9-28.
Haveman, J., & Hummels, D. (2004). Alternative Hypotheses and the Volume of Trade: The
Gravity Equation and the Extent of Specialization. Canadian Journal of Economics/Revue
canadienne d'économique, 37(1), 199-218.
Helpman, E. (1987). Imperfect Competition and International Trade: Evidence from Fourteen
Industrial Countries. Journal of the Japanese and International Economies, 1(1), 62-81.
Helpman, E. (1998). The Structure of Foreign Trade. Journal of Economic Perspectives, 13(2),
121-144
Helpman, E., & Krugman, P. R. (Eds.). (1985). Market Structure and Foreign Trade. Cambridge:
MIT Press.
Hiscox, M. J. (2001). Class Versus Industry Cleavages: Inter-Industry Factor Mobility and the
Politics of Trade. International Organization, 55(01), 1-46.
Hoffman, D. L., & Novak, T. P. (1996). Marketing in Hypermedia Computer-Mediated
Environments: Conceptual Foundations. The Journal of Marketing,60(3), 50-68.
Holt, L., & Jamison, M. (2009). Broadband and Contributions to Economic Growth: Lessons
from the US experience. Telecommunications Policy, 33(10), 575-581.
Hughes, A., & Scott Morton, M. S. (2005). ICT and Productivity Growth - The Paradox
Resolved? SSRN eLibrary http://ssrn.com/abstract= 881797
Hummels, D. L., & Levinsohn, J. A. (1993). Monopolistic Competition and International Trade:
Reconsidering the Evidence. Quarterly Journal of Economics, 110(3), 799-836.
Kaufmann, D., Kraay, A., & Mastruzzi, M. (2003). Government matters III: governance
indicators for 1996-2002: The World Bank, Policy Research Working Paper 3106.
Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical Integration, Appropriable Rents,
and the Competitive Contracting Process. Journal of Law and Economics, 21(2), 297-326.
Kretschmer, T. Information and Communication Technologies and Productivity growth: A
Survey of the Literature. OECD Digital Economy Papers, 195.

171	
  
	
  

Krugman, P. (1992). Does the New Trade Theory Require a New Trade Policy? World
Economy, 15(4), 423-442.
Krugman, P. R. (1979). Increasing Returns, Monopolistic Competition, and International trade.
Journal of International Economics, 9(4), 469-479.
Levchenko, A. A. (2004). Institutional Quality and International Trade. Review of Economic
Studies, 74(3), 791-819
Hofstede, G. H. (1980). Culture's Consequences: International Differences in Work-Related
Values. Beverly Hills, CA: Sage Publications
Lonsdale, P., Baber, C., Sharples, M., Byrne, W., Arvanitis, T. N., Brundell, P., et al. (2005).
Context Awareness for MOBIlearn: Creating an Engaging Learning Experience in an Art
Museum. Mobile Learning Anytime Everywhere.	
  Proceedings of MLEARN 2004. Bracciano,
Rome: Learning and Skills Development Agency.
Majdali, D. G. (2000). Trade Secrets Versus the Internet: Can Trade Secret Protection Survive in
the Internet Age. Whittier Law Review, 22, 125.
Malone, T. W., & Crowston, K. (1990). What is Coordination Theory and How Can it Help
Design Cooperative Work Systems? Proceedings of the ACM conference on Computersupported cooperative work, 357-370
Malone, T. W., & Crowston, K. (1994). The Interdisciplinary Study of Coordination. ACM
Computer Survey, 26(1), 87-119.
Mansfield, E. D., & Pollins, B. M. (2009). Economic Interdependence and International Conflict:
New perspectives on an Enduring Debate. Ann Arbor: University of Michigan Press.
Martinez-Frias, J. (2003). Comment. The Importance of ICTs for Developing Countries.
Interdisciplinary Science Reviews, 28(1), 10-14.
Matyas, L. (1997). Proper Econometric Specification of the Gravity Model. The World
Economy, 20(3), 363-368.
McCallum, J. (1995). National Borders Matter: Canada-US Regional Trade Patterns. The
American Economic Review, 85(3), 615-623.
McLaren, J. (2000). "Globalization" and Vertical Structure. American Economic Review, 90(5),
1239-1254.
Melitz, M. J. (2003). The Impact of Trade on Intra-Industry Reallocations and Aggregate
Industry Productivity. Econometrica, 71(6), 1695-1725.
Melville, N., Kraemer, K., & Gurbaxani, V. (2004). Review: Information technology and
organizational performance: An integrative model of IT business value. MIS Quarterly, 28(2),
283-322.
Moy, P., Scheufele, D. A., & Holbert, R. L. (1999). Television Use and Social Capital: Testing
Putnam's Time Displacement Hypothesis. Mass Communication & Society, 2(1/2), 27.
172	
  
	
  

Nunn, N. (2007). Relationship-Specificity, Incomplete Contracts, and the Pattern of Trade. The
Quarterly Journal of Economics, 122(2), 569-600.
Pilat, D. (2004). The ICT Productivity Paradox: Insights from Micro Data. OECD Economic
Studies, 38(1), 37-65.
Quelch, J. A., & Klein, L. R. (1996). The Internet and International Marketing. Sloan
Management Review, 37(3), 60-75.
Rana, V. (2012). An Innovative Use of Information & Communication Technology (ICT) in
Trade Facilitation in India. BVICAM's International Journal of Information Technology, BIJIT,
4(2), 492-495
Rauch, J. E., & Evans, P. B. (2000). Bureaucratic Structure and Bureaucratic Performance in
Less Developed Countries. Journal of Public Economics, 75(1), 49-71.
Riordan, M. H., & Williamson, O. E. (1985). Asset Specificity and Economic Organization.
International Journal of Industrial Organization, 3(4), 365-378.
Ryan, C., & Fleming, D. (2004). Digital Eco-Sense: Sustainability and ICT-A new terrain for
innovation. Lab Report 03, ISBN 0-975-67431-5
Samiee, S. (1998). The Internet and International Marketing: Is There a Fit? Journal of
Interactive Marketing, 12(4), 5-21.
Samiee, S., & Walters, P. G. P. (1990). Influence of Firm Size on Export Planning and
Performance. Journal of Business Research, 20(3), 235-248.
Smith, A. (1937). The Wealth of Nations (1776). New York: Modern Library.
Soloaga, I., & Alan Wintersb, L. (2001). Regionalism in the Nineties: What Effect on Trade?
The North American Journal of Economics and Finance, 12(1), 1-29.
Solow, R. M. (1957). Technical Change and the Aggregate Production Function. The Review of
Economics and Statistics, 39(3), 312-320.
Stiroh, K. (1999). Is There a New Economy? Challenge, 42(4), 82-101.
Stiroh, K. (2001). Investing in Information Technology: Productivity Payoffs for US Industries.
Current Issues in Economics and Finance, 7(6), 1-6.
Stiroh, K. J. (2002). Are ICT Spillovers Driving the New Economy? Review of Income and
Wealth, 48(1), 33-57.
van Ark, B., & Inklaar, R. (2005). Catching Up or Getting Stuck? Europe's Troubles to Exploit
ICT's Productivity Potential. Groningen Growth and Development Center, GD 79
van Ark, B., Inklaar, R., & McGuckin, R. H. (2003). ICT and Productivity in Europe and the
United States Where Do the Differences Come From? CESifo Economic Studies, 49(3), 295318.
173	
  
	
  

Wei, S.-J. (1996). Intra-national versus International Trade: How Stubborn are Nations in Global
Integration? : National Bureau of Economic Research, 553.
Wildman, S. S., & Siwek, S. E. (Eds.). (1988). International Trade in Films and Television
Programs. Cambridge: Ballinger Pub Co
Williamson, O. E. (1975). Markets and Hierarchies: Analysis and Antitrust Implications: A
Study in the Economics of Internal Organization. New York: The Free Press.
Williamson, O. E. (1985). The Economic Institutions of Capitalism. New York: The Free Press.
Wong, M.-H., & Pellan, M. I. (2012) Trade Facilitation: The Way Forward for ASEAN and Its
FTA Partners. Policy Brief, 2012-04.
Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge,
Massachusetts: The MIT Press.

174