Liberalization of Trade in Services Under ASEAN n FTAs: A Mapping Exercise

This study maps out the degree of liberalization of trade in services under four ASEAN n frameworks. After constructing a database showing the existence of limitations on market access and/or national treatment by each service sector, the study finds that the commitment level differs greatly between sensitive and less sensitive sectors, and that the commitment level under the ASEAN Framework Agreement (AFAS) is the highest among the four FTAs studied. It also finds that there are cross-country and sector-wide similarities in the pattern of service sector commitment under and across each of the FTAs; this implies that the shared domestic sensitivities can be overcome by a shared economic cooperation scheme for enhancing competitiveness in the ASEAN n region.


I. Introduction
While there has been a delay in the WTO-based liberalization of trade in services 1 , East Asian countries are in the process of establishing preferential pluri-lateral free trade agreements (FTAs) with a wide coverage fit for regional community building. 2 They have the potential of merging into a consolidated region-wide free trade framework. This study undertakes a mapping exercise

II. An Overview of WTO/GATS Commitment Tables
Whereas WTO's General Agreement on Trade in Services (GATS) is still ongoing under the current Doha Development Agenda for further multilateral liberalization, its basic framework of negotiation is fully taken into consideration and implemented under the four FTAs in the Asia Pacific region. It is therefore necessary first to give an overview of the framework of GATS. The most recent updated version of the GATS Commitment Tables available on-line is dated January 2003. In the case of "Revised Offer 2006", only a limited number of countries have submitted their revised offers. 3 Therefore the former tables are used in this study.
In a commitment table under GATS, four Modes 4 i.e. Mode 1 up to Mode 4, and two aspects of liberalization, i.e. market access (MA) and national treatment (NT), are listed in tabular formats. In each service sector (see APPENDIX for the the GATS-based classification of service sectors), the four modes and two aspects of liberalization make eight "cells", for each of which the existence of limitations is indicated in text. Such indication is created by ⓒ 2012 Journal of East Asian Economic Integration filling in one of the following three indications: (1) "none" (in the case of no limitation), or (2) "unbound" (in the case where there is no legally binding commitment made), or (3) description of the limitation.
For the sake of analytical tractability, this study adopts the level of 55 sub-sectors. The further disaggregated 155 sectors have been considered at the database construction stage. 5 Also, this study considers specific-commitments only. "Horizontal commitments", or commitments applied to all the GATS service sectors are not considered in this study. This is because the way horizontal commitments are described is oftentimes rather complicated, making a clear-cut and consistent database construction extremely difficult.
The following three-fold symbolic classification is used for constructing a database for the commitment by each sub-sector, by mode and by aspect of liberalization, in each FTA N: No limitation (and bound); L: Limited (or restricted) but bound; U: Unbound.
Since there are sub-categories with slightly different patterns of commitments in each of the most disaggregated 155 service categories, one "conservative" (i.e. most restrictive) pattern is listed in the database 6 constructed. In the case where the word "Unbound", or "None" is followed by such phrases as "except...", the label "U" or "N," respectively, is simply applied. The situation of no description exists is considered as "U." This simplified categorization allows for a "bird's-eye view" analysis of an otherwise analytically intractable style of reporting observed in the original GATS commitment tables. The database has been constructed for the four East Asian free trade agreements, i.e. (1) the ASEAN Framework Agreement on Services (AFAS) package 7, (2) the ASEAN-Australia-New Zealand FTA, (3) the ASEAN-China FTA, and (4) the ASEAN-Korea FTA. 5 At the stage of reporting the Hoekman Index (mentioned in the next section), aggregation up to the 55 sectors is used. While each of the 155 sub-sectors has further sub-divisions, the way each commitment table is described is not comparable with others due to idiosyncrasy in actual offer documents at the most detailed level (e.g. branching out with incomplete indications, incomplete listings, partial merging of different sub-divisions and the like). 6 The data will be published as part of ERIA FTA database at ERIA's website (www.eria.org).
ⓒ Korea Institute for International Economic Policy

III. Indexation of Service Trade Liberalization from the Database
Indexation of service trade liberalization is a new research area, primarily because the trade in services has long been considered as "non-tradable" (which is currently not the case), and also because the modalities of trade in services differs greatly across different sub-sectors (as indicated by Adlung and Martin. 2005). Indeed, so much mention is made of the restricted status of trade in services (see, e.g. Fink and Molinuevo 2008;Gootiiz and Mattoo 2009;Hoekman, Martin and Mattoo 2009;and Urata, Ogawa and Sawada 2011).
Measurement of the degree of service trade liberalization, therefore, naturally faces methodological difficulty, hence the paucity of literature to empirically address indexation methods. Among the limited number of existing work, OECD (2003OECD ( , 2009) offers the results of measuring service trade restrictiveness through undertaking subjective interviews to relevant business experts. Similarly, Ochiai, Dee and Findlay (2007) and Dee (2009) also make subjective evaluation of service trade restrictiveness using basically the same method. While the subjective way of evaluating service trade restrictiveness facing business sectors is a useful method in that attempts to capture the actual trade barriers are made, it also seems to face a difficulty especially in terms of constructing non-biased and comparable indices. Hoekman (1995), on the other hand, proposes an objective indexation method for measuring the GATS-style degree of commitment in the service sector. The present paper draws on Hoekman's approach since it remains the only indexation medhod that is objective in nature. This method assigns values to each of 8 cells (4 modes and 2 aspects-market access (MA) or National Treatment (NT)-), as follows: N=1, L=0.5, U=0; then calculates the average value by service sector and by country. Using the database constructed, the "Hoekman Index" has been calculated for each 155 sub-sectors. Then the simple average at the level of the 55 sectors is calculated. Tables 1-4 report the results by FTA.  Vietnam: The sectors 02B (Courier Services), 02C (Telecommunication Services), 04D (Franchising), 09A (Hotels and Restaurants) and 09B (Travel Agencies and Tour Operators Services) have the largest degree of commitment at 0.75. The average level of commitment is 0.33.
As for ASEAN-wide integration of trade in services, it has "Declaration on the ASEAN Economic Community Blueprint 8 ," in which targeting of some specific service sub-sectors and some aspects (including logistics services, market access limitations for Mode 3 and foreign equity participation for some sub-sectors) is made. It is expected that the use of Hoekman Index provides at least partial, but tangible information in this context.

The ASEAN-Australia-New Zealand Free Trade Agreement (AANZFTA) by country and by sector
The sector 01B (Computer and Related Services) has the highest average commitment by participating countries, at 0.70. The ASEAN average (total) is 0.20. The total average of commitment by country under AANZFTA is 0.23. Brunei: The sector 01B (Computer and Related Services) has the largest degree of commitment at 0.75. The average level of commitment is 0.07.
Cambodia: The sector 01B (Computer and Related Services) has the largest degree of commitment at 1.0 (full score). The average level of commitment is 0.38.
Indonesia: The sectors 08A (Hospital Services) and 09A (Hotels and Restaurants) have the largest degree of commitment at 0.63. The average level of commitment is 0.16.
Laos: The sector 01B (Computer and Related Services) has the largest degree of commitment at 0.80. The average level of commitment is 0.12.
Malaysia: The sector 01B (Computer and Related Services) has the largest degree of commitment at 0.80. The average level of commitment is 0.16.
Myanmar: The sector 01B (Computer and Related Services) has the largest degree of commitment at 0.88. The average level of commitment is 0.11. New Zealand: The sector 01B (Computer and Related Services) has the largest degree of commitment at 1.0 (full score). The average level of commitment is 0.39.
Philippines: The sector 09B (Travel Agencies and Tour Operators Services) has the largest degree of commitment at 0.75. The average level of commitment is 0.11. Singapore: The sector 01B (Computer and Related Services) has the largest degree of commitment at 1.0 (full score). The average level of commitment is 0.32.
Thailand: The sector 01B (Computer and Related Services) has the largest degree of commitment at 1.0 (full score). The average level of commitment is 0.22.
Vietnam: The sectors 02B (Courier Services), 02C (Telecommunication Services), 04D (Franchising), 07A (All Insurance and Insurance-related Services), 09A (Hotels and Restaurants), 09B (Travel Agencies and Tour Operators Services) have the largest degree of commitment at 0.75. The average level of commitment is 0.32.
ⓒ Korea Institute for International Economic Policy

ASEAN-China Free Trade Agreement (ACFTA) by country and by sector
The sector 09B (Travel Agencies and Tour Operators Services) has the highest average commitment by participating countries, at 0.34. The ASEAN average is 0.12. The total average of commitment by country under ACFTA is 0.12.

ASEAN-Korea Free Trade Agreement (AKFTA) by country and by sector
The sector 09B (Travel Agencies and Tour Operators Services) has the highest average commitment by participating countries, at 0.50. The ASEAN average is 0.19. The total average of commitment by country under AKFTA is 0.20. Thailand: NA (due to lack of publicly available data) Vietnam: The sectors 01B (Computer and Related Services), 02B (Courier Services), 02C (Telecommunication Services), 04D (Franchising), 07A (All Insurance and Insurance-related Services), 09A (Hotels and Restaurants), 09B (Travel Agencies and Tour Operators Services) have the largest degree of commitment at 0.75. The average level of commitment is 0.31.

IV. Correlation of commitments among the participating countries
After calculating the Hoekman Index, similarities among participating countries have been measured in the form of correlation coefficients. This has been done by comparing the calculated Hoekman Indices by country and by sector (as in Tables 1-4). The results are presented in Table 5-8. Under AFAS (as shown in Table 5), high correlations can be observed between ⓒ 2012 Journal of East Asian Economic Integration (1) Malaysia and Vietnam (correlation coefficient=0.609); (2) Laos and Vietnam (correlation coefficient=0.608). There is no negative correlation 10 observed among the ten ASEAN countries, indicating that they all have concern for common sensitive sectors as well as less-sensitive ones. Malaysia has the strongest positive correlation with the ASEAN average (correlation coefficient of 0.791). The simple average of all of the coefficients between different countries listed in the Table is calculated as 0.341 (not shown in the Table). This is the second highest among the four FTAs under coverage in this study, as seen below.
Under the ASEAN-Australia-New Zealand FTA (results are shown in Table  6), there is no correlation coefficient higher than 0.700, showing that under this FTA, each country has its own individual sensitivities. 11 All the correlation coefficients are positive (with the highest one being 0.688 between Australia and New Zealand), with just one exception (between Myanmar and the Philippines, yet the coefficient, -0.053 is low in magnitude). Malaysia has the strongest positive correlation with the ASEAN average (correlation coefficient of 0.805). The simple average of all of the coefficients between different countries listed in the Table is calculated as 0.349 (not shown in the Table). This average is the highest, and a little higher than that for AFAS (i.e. 0.341), indicating that, relatively speaking, the member countries are similar in their service sector commitments. 10 A low positive value does not indicate sector-by-sector precise correlations, it only shows the overall tendency. In this sense, the result has a limited ramification. In general, the correlation value between -0.3 and -0.7 generally indicate a weak negative correlation, and if the value is more than -0.3, it is usually judged that there is no correlation. With this caveat, the use of correlation coefficient seems to be valid. Rather than making artificial distinction between sensitive sectors and less-sensitive ones, I have included all the sectors in the calculation. 11 The correlation value between 0.3 and 0.7 generally indicate a weak positive correlation, and if the value is less than 0.3, it is usually judged that there is no correlation. In view of this, if the correlation is between 0.3 and 0.7, there is some commonality among the participating countries with the commitment pattern. And if the correlation is less than 0.3, then each of the countries has its own "unique commitment pattern" determined mainly by its unique domestic sensitivities. I have check a source and found that the following relations hold: for r (correlation coefficient), |r|=0.3 ⇔ r2≒0.1 (that is, one variable has the power to explain 10% of the other variable's variation); |r|=0.7 ⇔ r2≒0.5 (that is, one variable has the power to explain 50% of the other variable's variation).
Of course there is no definite reason why 0.1 and 0.5 should be used as thresholds, but this seems to be the established interpretation.  Table 7), there is no correlation coefficient higher than 0.700, just as in the case of the ASEAN-Australia-New Zealand FTA. The highest coefficient is 0.588 (between Vietnam and Cambodia). Vietnam has the strongest positive correlation with the ASEAN average (correlation coefficient of 0.789). The simple average of all of the coefficients between different countries listed in the Table is calculated as 0.059 (not shown in the Table). This is the lowest among the four FTAs investigated in this study. This seems to signify that the participation by China as a big supplier and market for trade in services, is rather "sensitive" and therefore the commitments by individual countries are diverse, reflecting intensified sensitivities.
Under the ASEAN-Korea FTA (results are shown in Table 8), there is no correlation coefficient higher than 0.700, as in the ASEAN-Australia-New Zealand FTA and the ASEAN-China FTA. The highest coefficient is 0.572 (between Brunei and Indonesia). Vietnam has the strongest positive correlation with the ASEAN average (correlation coefficient of 0.780). The simple average of all of the coefficients between different countries listed in the Table is calculated as 0.241 (not shown in the Table). This is the second lowest correlation among the four FTAs at issue in this study.
Correlation among the ASEAN+n FTAs has also been measured, using the sector-average value of Hoekman Index in Tables 1-4. The result is shown in Table 9. The highest positive correlation of 0.870 is observed between the ASEAN-Australia-New Zealand FTA and the ASEAN-Korea FTA. The lowest correlation of 0.615 is observed between the ASEAN Framework Agreement on Services and the ASEAN-China FTA. This, though, is also a positive value. There is no negative correlation observed among the four FTAs. Since country-difference is not considered in this analysis (due to differing membership across different FTAs), sector-specific factors are relevant here: Sectors with open orientation and those with domestic sensitivities are more or less shared across all the four FTAs.
Overall, strong correlations (coefficients of over 0.8) are observed among the following three FTAs, i.e. among (1) the ASEAN-Australia-New Zealand FTA, (2) the ASEAN-China FTA, and (3) the ASEAN-Korea FTA. In other words, the ASEAN Framework Agreement on Services has an unusual commitment pattern, reflecting some degree of a unified ASEAN membership.

V. Cluster Analysis
The next attempt is to highlight similarities in commitments among individual participating members by FTA. The standard pair-wise clustering method 12 has been applied to the calculated Hoekman Indices (as in Tables 1-4). Figures 1-4 show the results of pair-wise clustering. Figure 1 shows the clustering of countries under AFAS in the form of a "dendrogram" (tree-shaped categorization). As shown, Malaysia is closest to the simple-average of commitments by all the signatory countries (labeled as "ASEAN Ave." in the Figure). The commitment patterns do not seem to be categorized perfectly according to the level of economic development (in terms of per-capita GDP). Also, Cambodia, Laos, Myanmar and Vietnam (so-called "CLMV" countries as latecomer members of ASEAN) are not clustered close to one another, reflecting individual commitment patterns for each of them. Judging from the "distance" (measured by the horizontal axis in the Figure), the distances between ASEAN countries are closest under AFAS among the four FTAs studied, since all the ASEAN countries are clustered together within the distance of 2, whereas in the other Figures, the final clustering is done beyond the distance of 2. Figure  2 reveals that Australia and New Zealand are closest to the "Average", which indicates that their commitment patterns are, interestingly, "typical" of ASEAN members. Figure 3 for the ASEAN-China FTA shows that China is clustered rather far away from the "Average" commitment pattern. 13 Vietnam is closest to the "Average" just as in the case of Figure 1 (for the AFAS). Figure 4 for the ASEAN-Korea FTA shows that Korea is categorized rather close to the "Average" commitment pattern (although Vietnam is closest to the "Average"). 14 Clustering by sector of the country-average commitment under each FTA is shown in Figures 5-8. The upper part of the Figures show a group (or "cluster") of rather highly committed sectors, while the bottom part groups those sectors less committed. Overall, idiosyncratic clustering of the neatly categorized 55 12 Cluster analysis is a method of grouping observations into subgroups (called clusters) so that observations in the same cluster are similar in terms of "distance", which is Euclidean distance. 13 It seems that China has a rather closed trade policy, which is comparable to Brunei (which has the leading oil-related sector, hence less need to open up its market). 14 Cambodia, being aware of its rather less advanced domestic economic condition, has a quite open trade policy, which is comparable with Korea, in order to boost its economy through trade (including trade in services). In essence, both Korea and Cambodia are similar to each other in their open-trade inclination.
service sectors is observed, indicating that sensitivities differ even among similar service sectors. Since the more left-hand side of the Figures indicate shorter "distance" among the clustered pairs), so-called "cluster meeting" as seen in the GATS-based negotiations at the WTO, could also take place under these FTAs with a view to achieving cross-sector convergence in the future.

Figure 5. Clustering of sectors under AFAS (in the form of a dendrogram)
ⓒ Korea Institute for International Economic Policy  ⓒ Korea Institute for International Economic Policy      There are several caveats to be made in interpreting the mapped data. Most notably, there should be a distinction drawn between actual policy provisions and the noted commitments: the former might be well above the latter, indicating that in the actual business setting, a particular country's openness is more than the way the country makes its commitment under certain FTAs.
In addition, "enforcement" of the bound commitments is quite another issue: however deeply committed one country may be at the level of an FTA, such commitment might not be actually realized (enforced). Further, there is also a need to compare each country's commitment under GATS with that under each of the FTAs. This comparison of GATS-based commitments and the FTA-based commitments would reveal whether the so-called "WTO-plus" feature exists or not. 17 And finally, this study exclusively focuses on the "outline description", in the sense that the "Limitation" of individual service sectors is not quantified but simply denoted (in the database) as "L". Measuring the contents of limitations out of the commitment tables (characterized by "positive lists" rather than negative ones) requires an overall picture of each sector's legal framework. In this study, these aspects have not been considered, posing a limitation and at the same time providing an agenda for further study. 18 The mapping exercise in this study has overall revealed that: (1) The commitment level differs greatly between "sensitive" sectors and "less sensitive" sectors; this means that there is much scope for further enhancing international division of labor in terms of trade in services, through utilizing FTAs; 17 While all the pluri-lateral FTAs are expected to have the WTO-plus feature, China's commitment under the ASEAN-China FTA omits its commitment under the GATS, thus leading to the under-estimation of China's bilateral commitment. There are, however, incidences in which China reports in its bilateral FTA the same commitment made under the GATS. A preliminary investigation has revealed this sort of "discrepancy" being observed with several other countries including Thailand. There is thus a need to make some "reconciliation work" between the GATS commitment and FTA commitment overall, as part of the sequel research project. 18 As a separate undertaking, the often used "coverage index" has been calculated (for the use of this index, see, e.g. Adlung and Roy 2005). This index measures "the ratio of countries committed in particular sectors (as N or L) to the total number of countries". After calculating this index for each sector under each of the four FTAs at issue in this study, correlation coefficients between the Hoekman Index and the coverage index under each of the FTAs has been calculated. As a result, it is found that there is a high correlation of a little over 0.90 between these two indices, which implies that the Hoekman Index can serve as a representative index for measuring the commitment level of trade in services.
(2) The commitment level under the ASEAN Framework Agreement (AFAS) is the highest among the four FTAs studied; this means that the ASEAN member countries are rather highly consolidated among themselves, leading up to the formation of an ASEAN Economic Community (AEC); (3) There are cross-country similarities in the pattern of service sector commitment under each of the FTAs; this implies that the shared domestic sensitivities can be overcome by a shared economic cooperation scheme for enhancing competitiveness (through FTA provisions); (4) There are sector-specific similarities (high correlations) among the three FTAs, i.e. the ASEAN-Australia-New Zealand FTA, the ASEAN-China FTA and the ASEAN-Korea FTA; this signifies that in the face of extra-ASEAN market opening, the ASEAN members become more consolidated in terms of the pattern of service commitment; (5) Overall, Mode 4 (movement of people) is least committed, whereas Mode 2 (consumption abroad) is most committed under all the four FTAs studied.
There are two possibilities on the sequence of further streamlining the four FTAs: (1) Start within the same "clusters" among similarly committed countries under a particular FTA; then harmonize the level of commitments across all the signatory countries to the FTA; or (2) Start with harmonizing rather dissimilar countries from different "clusters" of commitments under a particular FTA, which provides small-scale "social experimenting"; then scale up this line of effort at an acceptably later stage to the level of the whole FTA, then eventually attempt to harmonize across all the FTAs centering on ASEAN, if the region covered by ASEAN+n FTAs is to become a more seamless market in terms of trade in services.
Either avenue would generate some degree of domestic concern. Overall, though, the absolute degree of commitment in service sectors remains rather low, even under the FTAs with a preferential nature. Given that there are more benefits than costs arising from deepening trade in services, further harmonization of the service chapters under the four FTAs studied is economically valid for bringing about more benefit to the ASEAN members, as well as all the other participating countries in the Asia Pacific region. As