Trade in Services and Human Development: A First Look at The Links

ADBI Working Paper Series
Trade in Services and Human

Development: A First Look at the
Links
Ben Shepherd and Gloria Pasadilla
No. 268
March 2011
Asian Development Bank Institute

Ben Shepherd is the principal of Developing Trade Consultants Ltd. Gloria Pasadilla is a
research fellow at the Asian Development Bank Institute. For helpful discussions, the
authors are grateful to Bernard Hoekman, Mia Mikic, Deunden Nikomborirak, and
participants at the ADBI-ARTNeT conference on Regulatory Reforms and Liberalization
in Services: Examining Impacts on Inclusive and Sustainable Development, held in Bali,
Indonesia on 11–12 October 2010.
The views expressed in this paper are the views of the authors and do not necessarily
reflect the views or policies of ADBI, the Asian Development Bank (ADB), its Board of
Directors, or the governments they represent. ADBI does not guarantee the accuracy of
the data included in this paper and accepts no responsibility for any consequences of
their use. Terminology used may not necessarily be consistent with ADB official terms.
The Working Paper series is a continuation of the formerly named Discussion Paper series;
the numbering of the papers continued without interruption or change. ADBI’s working
papers reflect initial ideas on a topic and are posted online for discussion. ADBI encourages
readers to post their comments on the main page for each working paper (given in the
citation below). Some working papers may develop into other forms of publication.
Suggested citation:
Shepherd, B., and G. Pasadilla. 2011. Trade in Services and Human Development: A First
Look at the Links. ADBI Working Paper 268. Tokyo: Asian Development Bank Institute.
Available: http://www.adbi.org/working-paper/2011/03/10/4485.trade.services.human.dev/
Please contact the author(s) for information about this paper.
Asian Development Bank Institute

Kasumigaseki Building 8F
3-2-5 Kasumigaseki, Chiyoda-ku
Tokyo 100-6008, Japan
Tel: +81-3-3593-5500
Fax: +81-3-3593-5571
URL: www.adbi.org
E-mail: [email protected]
© 2011 Asian Development Bank Institute

ADBI Working Paper 268 Shepherd and Pasadilla
Abstract
Some services directly produce outputs that are important for human development, such as
basic human services. Many other services are important inputs into the production and
distribution of goods that are necessary for human development purposes. A more efficient
services sector should mean that such goods and services can be made available to poor
people more cost effectively and more broadly. In line with this reasoning, we find in the data
that less restrictive services trade policies are associated with better human development
outcomes across a range of sectors. Appropriate services trade liberalization can therefore
promote human development directly through improved outcomes, in addition to indirectly
effects through the income channel.
JEL Classification: F13, O15, O24

Contents
1. Introduction .................................................................................................................... 1
2. Methodology and Data ................................................................................................... 2
3. Results and Discussion .................................................................................................. 3
3.1 Education Services ........................................................................................... 3

3.2 Distribution Services ....................................................................................... 11
3.3 Engineering Services ...................................................................................... 16
3.4 Telecommunications Services......................................................................... 20
3.5 Pooled Estimation Results .............................................................................. 24
4. Conclusion ................................................................................................................... 26
References ......................................................................................................................... 28
Appendix: Trade in Health Services and Human Development ........................................... 29

1. INTRODUCTION
The services sector is becoming increasingly important in modern economies: In many of the
most developed, it can represent two thirds or even three quarters of all economic activity.
International trade in services is also increasing in importance, and has been growing more
rapidly than goods trade over recent years (World Trade Organization (WTO), 2008). It has
also proved more resilient to the global financial crisis and resulting trade collapse (Borchert
and Mattoo, 2009).
Existing work on services trade has largely concentrated on two questions that do not
directly relate to human development. The first is the determinants of international trade
flows. Using methodologies such as the gravity model, researchers have investigated the
range of factors that can promote or inhibit trade in services. In general, they have found that
many of the factors known to impact goods trade—geographical distance, market size, and
cultural or historical factors—also exert a strong influence on services trade (Kimura and
Lee, 2006).
The second strand of research on services trade has examined its implications for economic
efficiency. These studies tend to have shown that a more restrictive services environment is
associated with less efficient and lower-quality service provision, inefficient resource
allocation, and slower economic growth (Arvis et al., 2010; Eschenbach and Hoekman,
2006; and Findlay and Warren, 2000). Because services are often important inputs in the
production of manufactured goods, an inefficient services sector can have economy-wide
implications—including a loss of competitiveness for manufacturers and exporters (Arnold,
Mattoo, and Narciso, 2008).
It has long been recognized that services trade can influence economic and social outcomes
through a variety of mechanisms. On the one hand, the economic gains from reform are
significant (Organisation for Economic Co-operation and Development. (OECD), 2003;
World Bank, 2002), and thus have the potential to promote human development by
increasing per capita incomes. 1 We refer to this linkage as the income channel. However,
there has been considerable controversy as to possible incompatibilities between openness
to services trade and the provision of human development-related services such as
electricity, water, and telecommunications. More broadly, a tension has emerged in the
human development literature between the economic case for liberalizing services markets,
and a perceived social case for maintaining stricter regulations in order to promote human
development objectives (United Nations Conference on Trade and Development (UNCTAD),
2005; United Nations Development Programme (UNDP), 2006).
This paper will bring some of the first empirical evidence to bear on the direct links between
services sector regulation and human development, going beyond the income channel. From
an economic point of view, there are good reasons to believe that services liberalization
might be positively—not negatively—associated with at least some human development
outcomes. The reason is that more efficient provision of public and private services that are
important for development can lead to lower prices for consumers, and more widespread
availability of human development-related goods and services. For instance, Chile’s
liberalization and privatization of its telecommunications sector, along with its use of “smart”
subsidies, led to a strong increase in availability of telecommunications services, and thus an
increase in the general population’s ability to access information and participate in political
and social life (Wellenius, 2002).
A second example of the links between openness to services trade and human development
outcomes is the role that the distribution sector plays in reducing the cost of moving vital
goods to the hinterland of poor countries (Sarley, Allain, and Akkihal, 2005). Examples
include basic foodstuffs, medicines, and mosquito nets. Only with a relatively well-developed
and efficient distribution sector is it possible to ensure that these types of products reach
1
That per capita income is an important determinant of overall human development outcomes is reflected in the
fact that the Human Development Index includes income as one of its components.
those who need them at lowest possible private and public cost. For instance, (Sarley, Allain,
and Akkihal, 2005) found that the logistics cost of moving bed nets from port to hinterland in
Liberia amounts to nearly half the cost of the product. Reducing the logistics cost wedge
clearly has great potential to help bring more bed nets to those who need them.
In this paper, we examine the association between human development and services trade
using simple nonparametric and parametric regression techniques. Our approach is to use
an indicator of human development as the dependent variable, and (at least) per capita
income and a measure of services sector policy restrictions as the independent variables.
Our measures of policy restrictiveness vary at the sector level, so we run separate
regressions using sector-specific measures of human development. Controlling for gross
domestic product (GDP) per capita means that we take account of the fact that it is an
important determinant of human development, and that it tends to be strongly inversely
correlated with service sector restrictiveness. We face formidable data constraints, however,
and are generally unable to include a wide range of other control variables due to small
sample size. The best we can do is to pool observations from three sectors, and use country
and sector fixed effects to control for other influences. Our results should therefore be
interpreted as providing a first indication of some important correlations in the data. In
particular, they should be interpreted in terms of associations between variables rather than
as evidence of causal links.
The paper proceeds as follows. In the next section, we discuss our methodology and data in
more detail. The section after that presents and discusses our results. We cover education,
distribution, engineering, and telecommunications services, as well as pooled results across
all sectors. The final section concludes with some policy implications and suggestions for
further research.
2. METHODOLOGY AND DATA

Our objective in this paper is to provide some first evidence on the direct links between
services trade policies and human development. By “direct”, we mean independent of
income effects. For example, we expect that a less restrictive policy environment in the
distribution sector should lead to more efficient and less costly service provision, and thus to
wider availability of important human development products such as vaccines. More
generally, we will test the following hypotheses using nonparametric and parametric
regression techniques: 2
1. Is openness to trade in education services correlated with human development
performance, as measured by indicators such as: the human development index,
poverty rates, per capita income, and the democracy index?
2. Is openness to trade in distribution services correlated with human development
performance in the form of higher immunization rates?
3. Is openness to trade in engineering services correlated with human development
performance, as measured by indicators such as: road network density, and the
availability of basic telecommunications services?
4. Is openness to trade in telecommunications services correlated with human
development performance, as measured by indicators such as: the HDI knowledge
and education index, a political voice and accountability indicator, and the
availability of basic telecommunications services?
2
Ideally, we would also like to include health services in the above list. However, data constraints mean that it is
currently impossible to do so. The appendix, instead, discusses the human development dilemma involved in
health services, and gives a somewhat more descriptive analysis of possible correlation of health services with
human development outcomes such as equity and access.
2
For each of these hypotheses, our dependent variable is a particular measure of human
development performance. Most data are sourced from the World Bank’s World
Development Indicators and the Human Development Report. For data on democracy and
political voice and accountability, we use Freedom House and the World Governance
Indicators respectively. As our main independent variable, we use sector-specific measures
of policy restrictiveness sourced from the Australian Productivity Commission (APC). 3 These
data are currently available for a single year only (2000), which we take as our base year for
all empirical work. The database includes two main measures per sector, each of which is a
numerical summary of a wide range of underlying policy information. The first (“domestic”)
index represents the entry barriers and ongoing cost burdens to which domestic firms in a
particular sector are subject. The second (“foreign”) index contains the same information in
respect of foreign firms. The difference between the two represents the extent to which trade
policy is discriminatory vis-à-vis overseas operators.
Our methodology proceeds in two steps. First, we use a common nonparametric technique,
the multivariate Locally Weighted Scatterplot Smoother, to examine graphically the
correlations among the variables of interest. The advantage of this method is that it allows us
to analyze the relationship between the dependent and independent variables without
imposing any particular functional form. It proceeds by running an ordinary least squares
(OLS) regression separately for each data point, using a centered 80% sample of the
original data as an estimating window. In the second step of our methodology, we run
standard OLS regressions using the same independent and dependent variables to confirm
the impressions given by the nonparametric regressions.
3. RESULTS AND DISCUSSION

3.1 Education Services
Education services have an obvious link with human development. Liberal policies towards
education can increase availability of education services and improve student access to
them. Greater access boosts adult literacy and enrolment rates. As more people have
access to education, this can also indirectly improve life expectancy, not only as a result of
better knowledge of hygiene but also, possibly, because of greater life satisfaction and
empowerment derived from being able to harness individual talents and capacities.
Education also helps increase income. As increased human capital leads to growth in
productivity (Benhabib and Spiegel, 1994), income growth follows. Education is also seen as
a great ‘equalizer’ because it makes social and income mobility possible. Schooling also has
an impact on the quality of public governance and democracy because it makes people
more informed about society and enables them to better understand and assess potential
risks and opportunities. Hence, education is an important foundation for democratic
societies.
We test the links between various human development outcomes and education services by
regressing restrictions in education services 4 against various human development indicators.
We use the Human Development Index (HDI), the education index of the HDI, per capita
GDP, and a democracy index. Our hypothesis is that more open trade in education services
will help improve the overall HDI and the HDI-education index (literacy and enrolment).
Since education helps people find jobs and earn their living, it should also contribute to
overall development and increase per capita income—an example of the income channel,
3
The Restrictions on Trade in Services Database is available online at http://www.pc.gov.au/research/
researchmemorandum/servicesrestriction
4
We use restrictions on tertiary education services. The reason for this choice is that the collected restrictions on
tertiary education across countries in the study by the Australian Productivity Commission are more
comprehensive, while those for elementary and secondary education services are incomplete. The sample
consists of 20 countries.
3
discussed above. Finally, because education increases social and political awareness, it
should help develop a more robust democracy. The converse of the above is that restrictions
in services trade, therefore, restrict democracy, restrain income growth, and dampen the
overall achievement of better human development.
Unlike the other service sectors used in the rest of this paper, the APC study provides a
breakdown of restrictions in education services trade by mode of supply. To take advantage
of this wealth of information, we regress each of the modal restrictions on the different
human development outcomes to understand which of the mode of supply restrictions
exercise greater influence on human development. We also made a simple average of the
different modal restriction indices to come up with a foreign restriction index, 5 which we also
regress against each of the other chosen dependent variables to make the results for
education services comparable with those of other service sectors below. But unlike the rest
of the paper, we do not examine the domestic restriction index’s effect on human
development outcomes because data on domestic modal restrictions in the original APC
study on education services are highly incomplete. 6
The results show strong links between selected human development indicators and
restrictions in education services. Significantly, of the four modes of supply, restrictions on
commercial presence appear to have greater influence in affecting human development
outcomes than restrictions in cross-border trade, consumption abroad, or movements of
natural persons.
Nonparametric estimates using the HDI education index 7 show that foreign restrictions on
commercial presence in education services have a more pronounced negative effect than
restrictions on other modes of supply. Figure 1 shows that while restriction indices on
consumption abroad, cross-border supply, and movement of natural persons have
ambiguous relationships with education outcomes, the relationship in the case of commercial
presence restrictions is clearly negative. Using the overall HDI index yields similar results as
shown in Figure 1.
5
The limitation with taking the simple average of the modes of supply restrictions is that each of the modes is
assumed to have equal weight or importance in the overall restrictions to trade in education services.
Nonetheless, we try this simple method to be consistent with the rest of the paper, which uses overall foreign
restriction indices.
6
For details of the restriction index construction for education services, see Nguyen-Hong and Wells (2003).
7
The HDI education index is comprised of literacy rates and school enrolment rates.
4
Figure 1: Non-parametric Regressions of Human Development Index (HDI) Education

Index Variable on Indices of Education Services Restrictiveness
Data Source: UNDP (2005); Nguyen-Hong and Wells (2003).
Table 1 corroborates the statistically strong negative correlation between foreign restrictions
on commercial presence and the HDI. 8 Per capita GDP (in log form) strongly and positively
correlates with the HDI indices, while restrictions on the commercial presence mode of
supplying education services correlate negatively. The correlation is slightly weaker with the
HDI than with the HDI education index, which is to be expected in light of the broader range
of influences for which the former accounts. We also regress the simple average of the
modal restrictions, the foreign restriction index, on the HDI and the HDI education index, but
this yields no statistically significant results. This perhaps reflects the fact that the three other
modes of supply—consumption abroad, cross-border, and movement of natural persons—
have been found to have insignificant relationships with the HDI indices.
8
The HDI is comprised of both education and health indicators. Education indicators include enrolment and
literacy rates, while health indicators include life expectancy and income.
5
Table 1: Regressions on Human Development Indicators (HDI)

(1) (2) (3) (4)
HDI Education HDI Education
Dependent Variable: HDI HDI
Index Index
Log (gross domestic
0.81*** 0.087*** 0.048*** 0.056**
product per capita)
(0.01) (0.012) (0.15) (0.022)
Commercial
-0.012* -0.024**
Presence
(0.01) (0.011)
Foreign Index
-0.01 -0.027
(Combined)
(0.02) (0.026)
Constant 0.113 0.045 0.495*** 0.402*
(0.088) (0.13) (0.149) (0.228)
Observations 19 19 19 19
2
R 0.92 0.91 0.69 0.64
Notes: All regressions are estimated using ordinary least squares. Standard errors are reported in parentheses.
Separate regressions using other modes of supply yielded no significant results. Statistical significance is indicated
by: * (10%), ** (5%), and *** (1%).
Data Source: UNDP (2005) for human development index (HDI) and HDI education index; World Development
Indicators (WDI) online database of the World Bank for GDP per capita; and Nguyen-Hong and Wells (2003) for
foreign restrictiveness indices in higher education.
We next tried other human development indicators like per capita income and a democracy
index as dependent variables. Non-parametric estimates with the democracy index and log
per capita GDP as dependent variables yield a different result from that of HDI indices. In
these non-parametric regressions, all modes of supply restrictions show a significantly
negative correlation with per capita income (Figure 2) and a positive correlation with the
democracy index. The democracy index represents a worsening of democratic conditions as
the index rises, hence the positive relationship with education services restrictions (Figure
3).
6
Figure 2: Non-parametric Regressions of GDP per capita on Indices of Education

Services Restrictiveness
Data Source: UNDP (2005); World Bank (2002); Nguyen-Hong and Wells (2003).
7
Figure 3: Non-parametric Regressions of the Democracy Index on Indices of

Education Services Restrictiveness
Data Source: UNDP (2005); World Bank (2002); Nguyen-Hong and Wells (2003).
Table 2 validates the results from the non-parametric estimates. 9 Least squares regression
results show that all modes of supply restrictiveness indices have negative signs and strong
statistical significance. Regressions with the democracy index as the dependent variable
likewise yields a significantly positive correlation. The result shows that greater restriction on
education services, in all modes of supply, is correlated with a worsening of democracy.
We also tried regressing poverty rates, the Gini inequality index, and the cost of tertiary
9
We are conscious that the per capita GDP regressions exclude a number of variables usually included in
income and growth regressions. The reason is our very small sample size. These results should be taken as
indicative only.
8
education on the restriction indices. The poverty rates regressions yielded no statistically
significant correlation with any of the restriction indices for education services. This is
surprising as one would immediately associate greater provision and efficiency of education
services with opening opportunities to a wider population and thus reducing poverty, but the
result is not sufficiently conclusive. One reason might be that we are using restrictions in the
tertiary education sector, due to unavailability of data for the more poverty-relevant primary
and secondary sectors. Using the Gini coefficient as the dependent variable yielded
significant results for restrictions in commercial presence, but with a positive sign. This
means that, instead of education supplied via commercial presence reducing inequality, it
instead worsens it. Again, this might be linked to our use of tertiary education data. The
regression of cost of education, 10 however, yielded a significant negative correlation with
restrictions on commercial presence. That is, education services that are open to foreign
commercial presence are associated with a lower cost of education.
Table 2: Regressions on Income and Democracy
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Dependent
Log of gross domestic product (GDP) per capita Democracy Index
Variable:
Population Growth -0.768*** -0.382* -0.756*** -0.723*** -0.534***
(0.16) (0.206) (0.155) (0.208) (0.174)

Log (GDP per
-0.244 -0.457 -0.368 -0.9056* 0.038
capita)
(0.34) (0.48) (0.35) (0.42) (0.39)
Consumption
-0.749*** 1.4921***
Abroad
(0.17) (0.4)
Commercial -
0.5891*
Presence 0.461***
(0.16) (0.29)
Cross-Border
-1.22*** 2.0013***
Supply
(0.26) (0.41)
Presence of Natural
-0.846*** 1.2669***
Persons
(0.16) (0.41)
Foreign Index
-0.985*** 2.0615***
(Combined)
(0.21) (0.42)
Constant 10.75*** 10.54*** 11.078*** 10.77*** 11.003*** 3.441 5.874 4.402 10.1571** -0.145
(0.23) (0.31) (0.32) (0.22) (0.31) (3.45) (4.81) (3.51) (4.23) (4)
Observations 18 19 18 15 19 18 19 18 15 19
2
R 0.65 0.54 0.6 0.46 0.67 0.64 0.44 0.59 0.65 0.67
10
The cost of education is represented by expenditure per student in tertiary education as a percentage of per
capita GDP. Poverty rates are the percentage of the population living below $1.25/day. For the poverty
equation, we used tobit regression due to the many zeros in the dependent variable representing developed
country situations.
9
Notes: All regressions are estimated using ordinary least squares. Standard errors are reported in parentheses.
Statistical significance is indicated by: * (10%), ** (5%), and *** (1%).
Data Source: Freedom House (www.freedomhouse.org) for the Democracy Index; World Development Indicators
(WDI) online database of the World Bank for GDP per capita and population growth; and Nguyen-Hong and Wells
(2003) for the foreign restrictiveness indices in higher education.
Using individual components of the HDI as dependent variables, however, surprisingly did
not yield the predicted results. For example, regressions with life expectancy, adult literacy
rates, and gross enrolment in tertiary education as dependent variables did not give
significant correlations with the restrictiveness index. Neither did restrictions in education
services show a correlation with the size of the college-educated labor force. However,
restrictions in consumption abroad (mode 2), cross-border supply (mode 1), and movement
of natural persons (mode 4) are negatively correlated with the life satisfaction index (Table
3). 11
Table 3: Results of Regressions Using Different Dependent Variables

Dependent Variable Result
No significance observed for any restrictiveness
Life Expectancy at Birth
index.
Adult Literacy
index.
Gross Enrollment in Tertiary Education
index.
Labor Force with Tertiary Education No significance observed for any restrictiveness
(% of Total Labor Force) index.
Foreign restrictiveness indices for consumption
Predicted Life Satisfaction Index abroad, cross border supply, and movement of
(Value Range 0-10 (High Satisfaction)) natural persons are significant at 1%, 5%, and
10% respectively.
Foreign restrictiveness index for commercial
Gini Index presence is significant at 10%, but the sign is
negative.
Foreign restrictiveness index for consumption
Expenditure per Student, Tertiary abroad and overall foreign index are significant
(% of gross domestic product per capita) at 1% and 10% respectively with positive
coefficients.
Poverty Headcount Ratio at $1.25 a Day No significance observed for any restrictiveness
(purchasing power parity) (% of Population) index.
Source: Authors’ own estimates.
In summary, our results show that restrictions on education services are not only associated
with poorer human development results, but also that some restrictions on modes of supply
have a greater impact than others. In particular, it appears that restrictions on commercial
presence take on greater importance for human development than restrictions on other
11
Detailed regression results used in Table 3 are available from the authors upon request.
10
modes of supply. But for democracy and per capita income, all modes of supply restrictions
are significantly associated with these outcomes. The life satisfaction index is negatively
correlated with restrictions on all modes of supply except commercial presence, while the
Gini coefficient regression seems to suggest that restrictions in commercial presence,
surprisingly, lessen inequality.
3.2 Distribution Services

There are many potential links between distribution services and human development. A
variety of human development outcomes rely on the ability to move important goods
efficiently and cost-effectively from one point to another. (Sarley, Allain, and Akkihal, 2005),
for instance, analyze the cost of upgrading supply chains as part of meeting the increased
movements of certain goods inherent in achieving the Millennium Development Goals. One
example is agricultural staples: an effective distribution network enables poor households to
access markets for basic produce, either as buyers or sellers. Economic and nutritional
wellbeing are therefore both linked to the quality of distribution services.
Vaccines are another important example. Most developing countries lack the means to
manufacture basic vaccines locally. Even in those with domestic capacity, economies of
scale mean that production is usually concentrated in just a few central locations. To
maximize the human development impact of the availability of vaccines, it is of course vital
that they be distributed to those in need. An efficient, high quality, and cost-effective
distribution network represents a necessary intermediate link in the chain between vaccine
producers and consumers. 12 To the extent that regulation of the sector influences prices,
costs, and quality of service provision, it is therefore possible that trade and regulatory
policies in distribution directly affect human development outcomes such as vaccine
availability.
Non-parametric regression analysis provides some initial support for this hypothesis of a link
between distribution sector regulation and vaccine availability. Figure 4 shows that even
after controlling for GDP per capita, there is a noticeable negative association between the
restrictiveness of domestic regulation in the distribution sector, and the rate of the diphtheria,
pertussis, and tetanus (DPT) vaccination. A very similar relationship is observed using the
measles immunization rate as the dependent variable (Figure 5).
12
Public procurement policies are also likely to affect the vaccination outcomes we are interested in here.
However, we do not have data with which to measure such policies in the same way as we can measure
restrictions on trade in distribution services.
11
Figure 4: Non-parametric Regression of the Diphtheria, Pertussis, and Tetanus (DPT)

Immunization Rate on per capita Gross Domestic Product (GDP) and an Index of
Distribution Services Restrictiveness. One Outlier Excluded
Data source: World Development Indicators online database of the World Bank; Australian Productivity Commission’s
database of restrictions to services trade for the restrictiveness indices.
Figure 5: Non-parametric Regression of the Measles Immunization Rate on per capita

Gross Domestic Product (GDP) and an Index of Distribution Services Restrictiveness
One Outlier Excluded
Data source: World Development Indicators online database of the World Bank; Australian Productivity Commission’s
database of restrictions to services trade for the restrictiveness indices.
12
Evidence from standard OLS regressions confirms these insights (Table 4). 13 Of course,
results need to be interpreted cautiously in light of the simplicity of the model and the very
small number of observations. But the first indications are that a more restrictive set of
distribution sector regulations is associated with lower immunization rates: both regressions
using the domestic restrictiveness index show negative and statistically significant
coefficients. This result is robust to the inclusion of per capita GDP as a control variable.
This last point is important because of the strong role played by income in determining
human development outcomes such as vaccination rates. Our results show that even after
controlling for income, more liberal distribution sector regulations are linked to stronger
immunization outcomes.
Interestingly, it is only the regressions using the domestic restrictiveness index that produce
significant results (columns 1 and 3). Neither regression using the foreign restrictiveness
index has a statistically significant coefficient (columns 2 and 4). This result suggests that it
is the overall quality of regulation that matters for distribution sector performance, not just the
degree of discrimination against foreign service providers.
13
Results are qualitatively identical if we use a fractional logit model to take account of the fact that the
dependent variable is bound between zero and unity (Papke and Wooldridge, 1996).
13
Table 4: Regression Results Using Distribution Restrictiveness Indices

(1) (2) (3) (4)
Immunization Rate
Dependent for diphtheria, Immunization Rate Immunization Rate Immunization Rate
Variable: pertussis, and (DPT) (Measles) (Measles)
tetanus (DPT)
Log (GDP per
6.697*** 6.981*** 5.096** 4.880**
capita)
(0.000) (0.001) (0.014) (0.043)
Domestic
Restrictiveness -40.677* -59.944**
Index
(0.082) (0.049)
Foreign
Restrictiveness 10.509 -10.358
Index
(0.425) (0.495)
Constant 27.800* 19.971 42.503** 42.681*
(0.086) (0.287) (0.031) (0.076)
2
R 0.46 0.42 0.32 0.22
Notes: All regressions are estimated using OLS with robust standard errors. Prob. values are reported in
parentheses. Statistical significance is indicated by: * (10%), ** (5%), and *** (1%). One outlier is excluded in each
regression.
Data Source: World Development Indicators online database of the World Bank for immunization rates and GDP per
capita; and the Australian Productivity Commission’s database of restrictions to services trade for the restrictiveness
indices.
For this sector, we can also adopt an alternative empirical strategy that allows us to greatly
increase the number of observations in our sample. To do so, we use one component of the
World Bank’s Logistics Performance Index (LPI 2007) as a measure of sectoral performance
in place of the Australian Productivity Commission regulatory indicators. The LPI is based on
a survey administered to around 1,000 logistics professionals around the world. By asking
them to rate performance in a number of countries with which they trade, the overall sample
size is increased to around 5,000 observations. The LPI itself is a composite of six indices
based on average responses to survey questions. The component we use here asks
respondents to rate on a one to five scale the quality and competence of logistics services in
a given country.
Although a valid measure of the performance of service providers in this area, the LPI data
clearly differ from the policy restrictiveness measures used elsewhere in this paper in that
they measure private sector development rather than public sector regulation. Nonetheless,
using the LPI data makes it possible to greatly expand the sample, and thus to include
additional explanatory variables that help demonstrate the robustness of our results. In
particular, we include controls for the total spend on health in GDP—to account for the fact
that a higher level of health spending should produce higher vaccination rates—and the
14
overall level of government effectiveness, i.e. the quality of public services, the quality of the
civil service and the degree of its independence from political pressures, the quality of policy
formulation and implementation, and the credibility of the government's commitment to such
policies. We also include an interaction term between the LPI services component and per
capita GDP, to account for the possibility that logistics affects vaccination rates differently in
developed versus developing countries.
Results using OLS are in Table 5. 14 The signs of all control variables are as expected:
countries at higher incomes, those that spend more on health, and those with more effective
governments tend to have higher levels of immunization for DPT and measles. All control
variables have coefficients that are statistically significant at the 1% level. In addition, the LPI
services component has a positive and 1% significant coefficient: in line with the smaller
sample results presented above, better logistics performance is associated with higher
immunization rates. Interestingly, the interaction term is negative and 1% statistically
significant, which indicates that the link between performance in logistics services and
vaccination rates becomes weaker as countries get richer. This result highlights the main
argument of this paper, namely that getting service delivery right is particularly important for
poor people in developing countries.
14
Again, we obtain qualitatively identical results using the fractional logit model.
15
Table 5: Regression Results Using the Logistics Performance Index (LPI)

(1) (2)
Immunization Rate for
diphtheria, pertussis, and Immunization Rate (Measles)
tetanus (DPT)
LPI (services) 0.475*** 0.448***
(0.006) (0.004)
LPI*Log (GDP per capita) -0.056*** -0.054***
(0.002) (0.001)
Log (GDP Per Capita) 0.171*** 0.166***
(0.000) (0.000)
Total Health Spending % GDP 1.156*** 0.842*
(0.006) (0.062)
Government Effectiveness
0.072*** 0.071**
Index
(0.007) (0.014)
Constant -0.611 -0.546
(0.139) (0.146)
Observations 142 142
2
R 0.355 0.305
Notes: All regressions are estimated using ordinary least squares with robust standard errors. Prob. values are
reported in parentheses. Statistical significance is indicated by: * (10%), ** (5%), and *** (1%).
Data Source: World Development Indicators (WDI) online database of the World Bank for immunization rates, GDP
per capita, and total spending on health as a percentage of GDP; the World Governance Indicators for the index of
government effectiveness; and the World Bank’s Logistics Performance Index for the LPI services component.
3.3 Engineering Services

We also expect regulation in the engineering services sector to be correlated with a number
of important human development outcomes. Engineering services are an important input in
many development areas. Construction projects that improve the quality of life in rural areas
are an example. So too is the extension of basic services such as telephony. To the extent
that engineering services are available on a high-quality and cost-effective basis, these kinds
of projects become easier and less expensive to implement. We therefore expect that a
regulatory stance that tends to reduce cost in the sector should be associated with better
development outcomes in these areas.
In this case, the data provide only mixed evidence in relation to this hypothesis. Non-
parametric regressions using telecommunications outcomes as the dependent variables
tend to support it. This is true for the number of telephone users and Internet users per 100
population (Figures 6–7): in both cases, there is an apparent negative relationship between
engineering sector restrictiveness and development outcomes after controlling for per capita
income. Surprisingly, the same result does not hold for road density: there is no obvious
relationship between this outcome variable and engineering sector restrictiveness (Figure 8).
16
Figure 6: Non-parametric Regression of the Number of Telephone Users per 100

Population on per capita Gross Domestic Product (GDP) and an Index of Engineering
Data Source: World Development Indicators online database of the World Bank; Australian Productivity Commission
database of restrictions to services trade.
17
Figure 7: Non-parametric Regression of the Number of Internet Users per 100

Population on per capita Gross Domestic Product (GDP) and an Index of Engineering
18
Figure 8: Non-parametric Regression of Road Density on per capita Gross Domestic

Product (GDP) and an Index of Engineering Services Restrictiveness
Results from OLS regressions are similar to those from the nonparametric exercise (Table
6). We find that after controlling for per capita income, the restrictiveness of regulation in the
engineering sector is associated with weaker outcomes in terms of telephone users (foreign
index, column 4) and Internet users (domestic index, column 1). The same is true for road
network density using the foreign index (column 6). Whereas in distribution services it is only
the domestic index that matters for performance, here we find evidence that general
regulation and the degree of discrimination vis-à-vis foreign service providers are both
important.
19
Table 6: Regression Results Using Engineering Restrictiveness Indices

(1) (2) (3) (4) (5) (6)
Internet Internet
Phone Users Phone Users
Users per Users per
per 100 per 100 Road Density Road Density
100 100
Population Population
Population Population
Log (GDP per
15.347*** 13.889*** 23.644*** 22.807*** 0.703* 0.601**
capita )
(0.000) (0.000) (0.000) (0.000) (0.022) (0.006)
Domestic
Restrictiveness -47.962** -15.535 -0.870
Index
(0.047) (0.442) (0.817)
Foreign
Restrictiveness -22.015 -26.271* -3.952*
Index
(0.177) (0.074) (0.096)
Constant -121.343*** -106.656*** -185.161*** -173.644*** -5.439* -3.864*
(0.000) (0.000) (0.000) (0.000) (0.044) (0.050)
Observations 34 34 34 34 28 28
2
R 0.619 0.602 0.840 0.851 0.216 0.295
Data Source: World Development Indicators online database of the World Bank for Internet and phone users per 100
population, road density, and GDP per capita; and the Australian Productivity Commission database of restrictions to
services trade for the restrictiveness indices.
3.4 Telecommunications Services

In the information age, access to basic telecommunication services is an important part of
civic life. It enables citizens to participate in economic, social, and political life. It enables
them to access basic information and to communicate more easily with others. Extension of
basic telecommunication services to the hinterland has long been an important development
objective. Regulation has traditionally played a strong role in shaping the sector. In recent
decades, however, there has been a strong move towards liberalization, which has been
accompanied by a general extension of the availability of basic services.
Indeed, the telecommunications sector provides some of the strongest evidence in favor of
our core hypothesis that less restrictive regulation is linked to better human development
outcomes. In terms of access to basic services, nonparametric regressions strongly suggest
an inverse relationship between regulatory restrictiveness and access to telephony and the
Internet (Figures 9–12). The same is true for the HDI education index and the World
Governance Indicators (WGI) voice and accountability indicator. In line with the discussion in
the previous paragraph, these results suggest that better regulation not only increases
access to basic services, but can also help improve a country’s general development and
20
governance environment by improving internal communication links.
Figure 9: Non-parametric Regression of the Number of Internet Users per 100

Population on per capita Gross Domestic Product (GDP) and an Index of
Telecommunication Services Restrictiveness
Data Source: World Development Indicators online database of the World Bank; Australian Productivity
Commission’s database of restrictions to services trade.
21
Figure 10: Non-parametric Regression of the Number of Telephone Users per 100
Population on per capita Gross Domestic Product (GDP) and an Index of
Data Source: World Development Indicators database of the World Bank; Australian Productivity Commission
Figure 11: Non-parametric Regression of the Human Development Index (HDI)

Education Index on per capita Gross Domestic Product (GDP) and an Index of
Data Source: Human development index website; Australian Productivity Commission database of restrictions to
services trade.
22
Figure 12: Non-parametric Regression of the World Governance Indicators (WGI)

Voice and Accountability Index on per capita Gross Domestic Product (GDP) and an
Index of Telecommunication Services Restrictiveness
Data Source: World Development Indicators online database of the World Bank; World Governance Indicators;
Australian Productivity Commission database of restrictions to services trade for the restrictiveness indices.
Parametric OLS regressions confirm these insights (Table 7). As in Table 5 above (LPI
regressions), the expanded sample size available with the telecommunications data makes it
possible to include some additional control variables, namely: overall government
effectiveness (as defined above); and total government spending as a percentage of GDP,
as an indicator of the extent to which the state is involved in service provision. The control
variables generally have the expected signs, but they are not always statistically significant.
In all eight regressions, however, the indices of regulatory restrictiveness are negatively and
1% statistically significantly associated with our various development outcome measures, i.e.
access to the Internet and telephony, education, and voice and accountability in government.
Together, these results provide strong evidence in favor of our core hypothesis.
23
Table 7: Regression Results Using Telecommunications Restrictiveness Indices

(1) (2) (3) (4) (5) (6) (7) (8)
Internet Internet Phone Phone World governance WGI Voice
HDI HDI
Users per Users per Users per Users per indicators (WGI) &
Education Education
100 100 100 100 Voice & Accountabil
Index Index
Population Population Population Population Accountability Index ity Index
Log (per capita
0.004 0.003 0.056*** 0.055*** 0.120*** 0.118*** -0.026 -0.029
GDP)
(0.545) (0.666) (0.000) (0.000) (0.000) (0.000) (0.710) (0.665)
Domestic
Restrictiveness -0.231*** -0.220*** -0.383*** -0.918*
Index
(0.000) (0.005) (0.001) (0.065)
Foreign
Restrictiveness -0.126*** -0.120*** -0.187*** -0.623**
Index
(0.000) (0.005) (0.002) (0.022)
Government
Effectiveness 0.093*** 0.092*** 0.108*** 0.108*** -0.056** -0.051* 0.729*** 0.702***
Index
(0.000) (0.000) (0.000) (0.000) (0.037) (0.056) (0.000) (0.000)
Government
Spending % -0.038 -0.037 0.105 0.107 0.278 0.252 -0.066 0.021
GDP
(0.747) (0.749) (0.504) (0.500) (0.326) (0.387) (0.952) (0.985)
Constant 0.123** 0.131** -0.228** -0.222** -0.153 -0.151 0.447 0.529
(0.029) (0.021) (0.012) (0.013) (0.375) (0.382) (0.464) (0.375)
Observations 132 132 133 133 113 113 133 133
2
R 0.768 0.768 0.838 0.838 0.595 0.587 0.670 0.676
Data Source: World Development Indicators online database of the World Bank for Internet and phone users per 100
population, GDP per capita, and total government spending as a percentage of GDP; the human development index
website for the education index; the World Governance Indicators for the voice and accountability and government
effectiveness indices; and the Australian Productivity Commission’s database of restrictions to services trade for the
restrictiveness indices.
3.5 Pooled Estimation Results

The most obvious shortcoming of the above sectoral results is that they only control for per
capita income. There are many other factors that could also potentially influence human
development outcomes, in addition to income and sector-level regulation. Examples include
the general level of development of governance institutions, and the overall efficiency of
regulation. However, small sample sizes make it impractical to account directly for these
influences through the inclusion of additional control variables.
To try and deal with this problem, we also estimate models that pool data across all sectors.
24
This approach allows us to include fixed effects by country and by sector. The fixed effects
clean out all external influences that vary in the country or sector dimensions, thereby
relieving to a large extent the omitted variable bias that might be suspected in our previous
results. For instance, the fixed effects control for influences such as government
effectiveness, or the composition of government spending. In terms of the data, we proceed
by selecting one human development indicator per sector: the DPT immunization rate for
distribution; Internet users per 100 population for engineering; and phone users per 100
population for telecoms. At this stage, we do not include education in the panel estimates
because of the different structure of the regulatory indicators, which measure restrictiveness
by mode of supply rather than on an aggregate domestic or foreign basis, as in the other
sectors.
The first two columns of Table 8 present estimation results for the pooled model using data
for all three sectors. Results strongly support our contention: the domestic and foreign
restrictiveness indices both have negative and statistically significant coefficients (1% and
5% respectively). The difference in magnitude between the two coefficients suggests, as
noted above, that it is usually the restrictiveness of domestic regulation that makes most
difference in terms of human development outcomes.
In the last two columns of Table 8, we exclude the telecom sector from our dataset. The
reason for doing so is that it strongly dominates the other sectors in terms of the number of
data points available. Although the domestic and foreign restrictiveness indices both have
the expected negative coefficients, they are no longer statistically significant once we
exclude the telecom sector. Our results in columns 1–2 are therefore being driven to a
significant degree by a close link between regulation and human development outcomes in
that sector. In part, this is a consequence of data limitations: our human development data
are much more closely related to sectoral economic performance in telecom than in the
other sectors. The greatly reduced sample size is also a constraint: our regressions in
columns 3–4 have only 71 observations but a total of 39 dummy variables to account for
country- and sector-level influences. The last two regressions are therefore pushing the data
to their limits, and it is perhaps not surprising that our results lose precision.
25
Table 8: Regression Results Using Pooled Data

(1) (2) (3) (4)
All Except All Except
All Sectors All Sectors
Telecoms Telecoms
Domestic
Restrictiveness -48.676*** -15.732
Index
(0.001) (0.586)
Foreign
Restrictiveness -26.702** -4.951
Index
(0.007) (0.816)
2
R 0.968 0.967 0.959 0.959
Country Fixed
Yes Yes Yes Yes
Effects
Sector Fixed
Yes Yes Yes Yes
Effects
Data Source: World Development Indicators online database of the World Bank for phone users per 100 population,
Internet users per 100 population, and the rate of diphtheria, pertussis, and tetanus immunization; and the Australian
Productivity Commission database of restrictions to services trade for the restrictiveness indices.
4. CONCLUSION
This paper has presented some first empirical evidence on the links between services trade
and human development. The data generally show that a more restrictive services trade
policy environment is correlated with worse human development outcomes. This finding is
consistent with a mechanism in which more restrictive services policies result in higher
prices of basic goods and services for consumers. It has long been argued that trade policy
in services sectors can lead to higher national income, which in turn can promote human
development. Our results go one step further in isolating a direct connection between service
sector restrictiveness and development, which acts independently of the income channel.
One implication of our results is that the tension between service sector openness and
human development outcomes, which is apparent in some of the development policy
literature, has perhaps been overstated. At the very least, our results suggest that there is no
systematic association between greater policy restrictiveness and better outcomes. Rather,
open and efficient services sectors can help promote human development. Reducing the
restrictiveness of service sector policies through well-designed liberalization programs can
be one element of a successful approach to promoting economic and human development
together.
There is considerable room for future research to expand on our approach and results. The
main difficulty we have confronted relates to the availability of data on applied policy settings
in services sectors. As data for more countries and years become available, it will be
possible to expand the sample size used here, and perhaps even move to a genuine panel
26
data framework. Both steps are important in ensuring that our results are robust to the
exclusion of additional country- and sector-specific factors from the regressions.
27
REFERENCES
Arnold, J., A. Mattoo, and G. Narciso. 2008. Services Inputs and Firm Productivity in Sub-
Saharan Africa: Evidence from Firm-Level Data. Journal of African Economies 17(4):
578–599.
Arvis, J., M. Mustra, L. Ojala, D. Saslavsky, and B. Shepherd. 2010. Connecting to Compete
2010: Trade Logistics in the Global Economy. Washington, DC: World Bank.
Benhabib, J., and M. Spiegel. 1994. The Role of Human Capital in Economic Development:
Evidence from Aggregate Cross-Country Data. Journal of Monetary Economics
34(2): 143–173.
Bjørnskov, C., A. Dreher, and J. Fischer. 2008. Cross-Country Determinants of Life
Satisfaction: Exploring Different Determinants Across Groups in Society. Social
Choice and Welfare. Springer 30(1): 119–173
Blouin, C., N. Drager, and R. Smith. 2006. International Trade in Health Services and the
GATS: Current Issues and Debates. Washington, DC: World Bank.
Borchert, M., and A. Mattoo. 2009. The Crisis Resilience of Services Trade, Policy Research
Working Paper No. 4917. Washington, DC: World Bank.
Eschenbach, F., and B. Hoekman. 2006. Services Policy Reform and Economic Growth in
Transition Economies. Review of World Economics 142(4): 746–764.
Findlay, C., and T. Warren, eds. 2000. Impediments to Trade in Services: Measurement and
Policy Implications. London: Routledge.
Kimura, F., and H. Lee. 2006. The Gravity Equation in International Trade in Services.
Review of World Economics 142(1): 92–121.
Nguyen-Hong, D., and R. Wells. 2003. Restrictions on Trade in Education Services: Some
Basic Indexes. Staff Working Paper. Melbourne, Australia: Australian Productivity
Commission.
Organisation for Economic Co-operation and Development (OECD). 2003. Quantifying the
Benefits of Liberalizing Trade in Services. Paris: OECD.
Papke, L., and J. Wooldridge. 1996. Econometric Methods for Fractional Response
Variables with an Application to 401(k) Plan Participation Rates. Journal of Applied
Econometrics 11: 619–632.
Sarley, D., L. Allain, and A. Akkihal. 2005. Estimating the Global In-Country Supply Chain
Costs of Meeting the MDGs by 2015. Arlington, VA: United States Agency for
International Development.
United Nations Conference on Trade and Development (UNCTAD). 2005. Trade in Services
and Development Implications Document No. TD/B/COM.1/71. Geneva, Switzerland:
UNCTAD.
United Nations Development Programme (UNDP). 2006. International Trade and Human
Development: Lao PDR. Vientiane: UNDP National Statistics Center.
Wellenius, B. 2002. Closing the Gap in Access to Rural Communications: Chile 1995–2002.
Discussion Paper No. 430. Washington, DC: World Bank.
World Bank. 2002. Global Economic Prospects. Washington, DC: World Bank.
World Trade Organization (WTO). 2008. World Trade Report. Geneva, Switzerland: WTO.
28
APPENDIX: TRADE IN HEALTH SERVICES AND HUMAN

DEVELOPMENT
Liberalizing trade in health services has both potential negative and positive results. For
example, by liberalizing cross-border trade, telemedicine can help in upgrading knowledge of
local medical professionals, and make available expert medical advice even in remote areas.
Increased foreign investment in hospitals and clinic facilities can improve supply of medical
services, both in terms of quality and quantity. Greater access to state-of-the art medical
technology becomes possible. Improving facilities can help attract more foreign health
consumers who are shopping around for cheaper services for simple medical procedures
and can help increase foreign exchange receipts from medical tourists.
However, these positive outcomes have to be balanced with the potential downside. For
example, in telemedicine, how does one control for potential medical malpractice? Who is
supposed to shoulder the liabilities? Increased foreign-funded hospital facilities can generate
an internal brain drain effect, whereby medical professionals move from the public hospitals
or clinics to higher-paying private facilities. Liberalization’s effect on quality medical access
for the broader population is a serious offset to the potential benefits from greater access to
advanced medical services that those in middle- to high-income groups enjoy. Freer
movements of medical professionals could also lead to a brain drain situation, at the
expense of huge public subsidy for educating them.
Because data in health services is difficult to obtain, it is hard to make correlations between
liberalization of health services and development outcomes, particularly with regards to
equity and access to medical services. On the one hand, there are not many instances of
liberalization in health services, if we go by the number of General Agreement on Trade in
Services (GATS) commitments in the sectoronly about 32 out of 134 members have made
commitments in health services. On the other, some of the potential negative effects have
taken place with or without liberalization: for example, the outward migration of medical
professionals, which has been observed in developing countries regardless of liberalization
or GATS commitments in developed economies. Likewise, the increase in cross-border
trade in medical transcription services has taken place without mode 1 commitments in
many outsourcing destination countries.
As with other service sectors, it is expedient for governments to fix their national regulatory
framework and understand national policies and priorities before opening up an important
sector like health services, where many public services and subsidies may be affected, and
in which the objective for policy goes far beyond mere economics. For many countries, their
first priority is wide, high-quality health care coverage at an affordable cost. Equity and
access considerations are important considerations that should be balanced out against the
potential increase in foreign exchange receipts through health services exports.
29

Trade in Services and Human Development: A First Look at The Links

Uploaded by

Copyright:

Available Formats

Trade in Services and Human Development: A First Look at The Links

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Trade in Services and Human Development: A First Look at The Links

Uploaded by

Copyright:

Available Formats

ADBI Working Paper Series

Trade in Services and Human

Ben Shepherd and Gloria Pasadilla

Asian Development Bank Institute

Please contact the author(s) for information about this paper.

Asian Development Bank Institute

© 2011 Asian Development Bank Institute

JEL Classification: F13, O15, O24

2. Methodology and Data ................................................................................................... 2

3. Results and Discussion .................................................................................................. 3

3.1 Education Services ........................................................................................... 3

Appendix: Trade in Health Services and Human Development ........................................... 29

2. METHODOLOGY AND DATA

3. RESULTS AND DISCUSSION

Figure 1: Non-parametric Regressions of Human Development Index (HDI) Education

Data Source: UNDP (2005); Nguyen-Hong and Wells (2003).

Table 1: Regressions on Human Development Indicators (HDI)

Figure 2: Non-parametric Regressions of GDP per capita on Indices of Education

Figure 3: Non-parametric Regressions of the Democracy Index on Indices of

Population Growth -0.768*** -0.382* -0.756*** -0.723*** -0.534***

(0.16) (0.206) (0.155) (0.208) (0.174)

Table 3: Results of Regressions Using Different Dependent Variables

3.2 Distribution Services

Figure 4: Non-parametric Regression of the Diphtheria, Pertussis, and Tetanus (DPT)

Figure 5: Non-parametric Regression of the Measles Immunization Rate on per capita

Table 4: Regression Results Using Distribution Restrictiveness Indices

Table 5: Regression Results Using the Logistics Performance Index (LPI)

Total Health Spending % GDP 1.156*** 0.842*

3.3 Engineering Services

Figure 6: Non-parametric Regression of the Number of Telephone Users per 100

Figure 7: Non-parametric Regression of the Number of Internet Users per 100

Figure 8: Non-parametric Regression of Road Density on per capita Gross Domestic

Table 6: Regression Results Using Engineering Restrictiveness Indices

3.4 Telecommunications Services

governance environment by improving internal communication links.

Figure 9: Non-parametric Regression of the Number of Internet Users per 100

Figure 11: Non-parametric Regression of the Human Development Index (HDI)

Figure 12: Non-parametric Regression of the World Governance Indicators (WGI)

Table 7: Regression Results Using Telecommunications Restrictiveness Indices

3.5 Pooled Estimation Results

Table 8: Regression Results Using Pooled Data

APPENDIX: TRADE IN HEALTH SERVICES AND HUMAN

You might also like

Population Growth -0.768*** -0.382* -0.756* -0.723* -0.534***