Australia and Japan – a remarkable commercial relationship
Appendix
(Note that this is an edited version of the report. It includes the key tables only and only the relevant data. A full version of the report will be published by the authors.)
DFAT–ANU Services Trade Project
A note on results
Jenny Corbett, Fuku Kimura, Kazu Hayakawa and Arata Kuno
Introduction
This research examines the bilateral services trade between Australia and Japan to assess whether that trade is larger or smaller than would be expected in the context of global flows of services trade.
The method used is a standard gravity model which covers as large a sample of countries as possible to establish a benchmark (average) bilateral trade flow. Dummy variables for trade by Australia and Japan with all trade partners, and for the bilateral flow between the two, are introduced to capture the additional effect, after accounting for the standard explanatory variables. This uncovers whether the bilateral flow is larger or smaller than that between an average pair of countries.
Data for services trade is notoriously poor so the results must be interpreted with some caution and, additionally, there are several avenues for further research that would allow checks for robustness of the results.
2.1 Gravity Equations in Services
There is a small number of other studies using gravity equations for services. Kimura and Lee (2006) provide a brief summary. The present study differs from the existing literature in the following respects:
i. It uses more recent data and more countries giving significantly larger sample size
ii. It breaks the data down in to three components of services trade (transport, travel and other services)
iii. It includes a goods intensity index in the services equation to capture the complementarity effects between goods and services trade
iv. It includes a preferential trade agreement (PTA) that covers only agreements that explicitly include services (previous studies have included all PTA agreements between partner countries but many agreements did not include services).
v. It includes a country risk variable.
vi. It has an explicit focus on the Australia-Japan bilateral trade patterns.
2.2 Method, Data and Definitions
Years: 2002-2004 (Unbalanced Panel)
All Countries (169):
Albania, Algeria, Angola, Antigua and Barbuda, Argentina, Armenia, Australia, Austria, Azerbaijan, Bahrain, Bangladesh, Belarus, Belgium, Belize, Benin, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon, Canada, Cape Verde, Central African Republic, Chad, Chile, China, Colombia, Comoros, Congo, Costa Rica, Cote d'Ivoire, Croatia, Czech Republic, Denmark, Djibouti, Dominica, Dominican Republic, Ecuador, Egypt, El Salvador, Equatorial Guinea, Eritrea, Estonia, Ethiopia, Fiji, Finland, France, Gabon, Gambia, Georgia, Germany, Ghana, Greece, Grenada, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, Honduras, Hong Kong, Hungary, Iceland, India, Indonesia, Iran, Ireland, Israel, Italy, Jamaica, Japan, Jordan, Kazakhstan, Kenya, Kiribati, Korea, Rep., Kuwait, Kyrgyz Republic, Lao PDR, Latvia, Lebanon, Lesotho, Liberia, Libya, Lithuania, Luxembourg, Macao, Macedonia, Madagascar, Malawi, Malaysia, Maldives, Mali, Malta, Marshall Islands, Mauritania, Mauritius, Mexico, Micronesia, Moldova, Mongolia, Morocco, Mozambique, Namibia, Nepal, Netherlands, New Zealand, Nicaragua, Niger, Nigeria, Norway, Pakistan, Palau, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Poland, Portugal, Russian Federation, Rwanda, Samoa, Saudi Arabia, Senegal, Seychelles, Sierra Leone, Singapore, Slovak Republic, Slovenia, Solomon Islands, South Africa, Spain, Sri Lanka, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Sudan, Suriname, Swaziland, Sweden, Switzerland, Syrian Arab Republic, Tajikistan, Tanzania, Thailand, Togo, Tonga, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, United Arab Emirates, United Kingdom, United States, Uruguay, Uzbekistan, Vanuatu, Venezuela, Vietnam, Yemen, Zambia, Zimbabwe
OECD Countries (30):
Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Korea, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, United Kingdom, United States
Sample PTAs (entered into force before 2004 and notified under the GATS Article V)
EC (Treaty of Rome), CER, EEA, NAFTA, Costa Rica-Mexico, EC accession of Austria, Finland and Sweden, CARICOM, Canada-Chile, Mexico-Nicaragua, Chile-Mexico, EC-Mexico, New Zealand-Singapore, Guatemala-Mexico, El Salvador-Mexico, Honduras-Mexico, EFTA-Mexico, United States-Jordan, Chile-Costa Rica, Chile-El Salvador, EFTA, Japan-Singapore, EFTA-Singapore, Panama-El Salvador, Singapore-Australia
(FTAs solely notified under the GATT Article XXIV or Enabling Clause are not incorporated into our dataset.)
Gravity Equation:
ln (1 + Tij) = β0 +β1 ln GDPi + β2 ln GDPj + β3 ln distanceij +β4 ln Riski + β5 ln Riskj
+ β6 ln Remotei + β7 ln Remotej + β8 languageij + β9 colonyij + β10 FTAij
+ β11 ln Goods Intensityij + β12 Ex AUSi + β13 Im AUSj + β14 Ex JPNi + β15 Im JPNj
+ β16 AUS to JPNij + β17 JPN to AUSij + εij.
A time subscript is omitted. i and j are exporter’s country and importer’s country, respectively. Year dummies are also introduced.
Dependent variable (Tij):
Description: Total services trade, Travel service trade, Transportation service trade, Commercial services trade (Communications services, Construction services, Insurance services, Financial services, Computer and information services, Royalties and license fees, Other business services, Personal, cultural and recreational services) (on the balance-of-payments basis, which primarily covers modes 1 and 2)
Source: OECD Statistics on International Trade in Services, Detailed Tables by Partner Country 2001-2004, 2006 Edition.
Description: Total goods trade
Source: UN Comtrade
Note: Dependent variables are deflated by GDP deflator in reporting country (2002 = 100).
Independent variables:
(1) GDP
Description: Gross Domestic Product deflated by GDP deflator in each country (2002 = 100)
Source: World Development Indicator
(2) distance
Description: Geodesic distance calculated following the great circle formula, which uses latitudes and longitudes of the most important cities/agglomerations (in terms of population)
Source: CEPII website
(3) Risk
Description: Country risk index, which is the aggregates of bankers’ evaluation on the risk of default in each country (the larger index indicates that the risk of default in the country is smaller.)
Source: Institutional Investor, various issues
(4) Remote
Description: a measure of the economic remoteness of alternative trading partners: an inverse of the sum of distance-weighted real GDP,
Sources: World Development Indicator, CEPII website
(5) language
Description: Dummy variable taking unity if a language is spoken by at least 9 per cent of the population in both countries and zero otherwise.
Source: CEPII website
(6) colony
Description: Dummy variable taking unity if two countries have ever had a colonial link and zero otherwise.
Source: CEPII website
(7) FTA
Description: Dummy variable taking unity if two countries are members of a common FTA notified under the GATS Article V as of the end of year 2001, 2002, and 2003, one year prior to our trade data respectively.
Source: WTO website
(8) Goods Intensity
Description: The trade intensity index is a measure of whether the value of trade between two countries is greater or smaller than would be expected on the basis of their importance in world trade. It is defined as the share of one country’s exports going to a partner divided by the share of world exports going to the partner:
,
where xij is country i’s exports to country j. Xiw and Xwj are country i’s total exports to the world and country j’s total imports from the world, respectively. Xw is total world exports. An index of more (less) than unity indicates a bilateral trade flow that is larger (smaller) than expected, given the partner country’s importance in world trade.
Source: WDI, UN Comtrade
(9) Ex AUS
Description: Dummy variable taking unity if exporting country is Australia and zero otherwise
(10) Im AUS
Description: Dummy variable taking unity if importing country is Australia and zero otherwise
(11) Ex JPN
Description: Dummy variable taking unity if exporting country is Japan and zero otherwise
(12) Im JPN
Description: Dummy variable taking unity if importing country is Japan and zero otherwise
(13) AUS to JPN
Description: Dummy variable taking unity if exporting country and importing country are Australia and Japan, respectively, and zero other otherwise
(14) JPN to AUS
Description: Dummy variable taking unity if importing country and exporting country are Australia and Japan, respectively, and zero other otherwise
2.2 Results
The basic trade intensities are shown in Figure 1. Services trade intensity for Australia’s exports of services to Japan and Japan’s exports of services to Australia are above 1 (so greater than proportional to the role of each in world trade) but lower than the intensities for the comparable goods trade in both directions. Intensity indexes are useful as a general measure of the extent to which trade is higher or lower than proportional to the share of the exporting and the importing countries in world trade but they do not fully capture whether a bilateral trade flow is greater or smaller than would be expected given the characteristics of each country relative to the characteristics that, on average, determine bilateral trade flows.
Figure 1. Trade Intensity Index
The tables that follow describe a number of regression analyses on the gravity models which do capture those country characteristic effects.
Key Results
Total Services
- The gravity model works well for services. As in other recent work (Kimura and Lee, 2006) it works at least as well, and in some circumstances better than, the model for goods. (The explanatory power of the equations here is better than those published in Kimura and Lee, probably because of the larger sample size).
- The data set with only OECD countries presents slightly more stable results.
- The model with more control variables is preferred to the simple gravity model. Therefore, for interpreting the effects in services trade alone, the results in Tables 5 and 6 are preferred (including the trade intensity index as an explanatory variable). For comparing services and goods effects Tables 2 and 3 give the results.
- Distance has a slightly smaller negative effect on services than on goods trade in contrast to the result in Kimura and Lee (we have not tested the significance of the difference in coefficients – possibly they are not significantly different)
- Remoteness effects are unstable relative to the goods trade results but in the stronger results of tables 5 and 6 remoteness of both exporter and importer both have a negative effect.
- Membership of a PTA that includes services has an uncertain effect on total services (negative in the whole sample, insignificant in the OECD sample) but has a consistently negative effect on transport services. It is not clear what the interpretation for a negative effect would be.
- Goods Intensity has a consistently positive effect on services trade suggesting a complementarity between the two types of trade (alternative tests of complementarity in the literature have found mixed results). In the OECD sample this effect is largest in transportation. These results are not, however, consistently reflected in the detailed data for Australian and Japanese trade in tables 7-10.
- Country risk effects are larger than for goods and consistently positive (i.e. lower risk of both exporters and importers increase trade) across all model specifications.
Australia-Japan Effects
- As noted in the introduction Australia has a higher than average export and import of services and this is maintained across the sectors except in the export of transport services. In the detailed analysis of Tables 8 the above- average effect in Australia’s exports of services is maintained except in communications, royalties and government. Transport services show a positive effect but only significant at the 10% level. The most significant effects are in travel, insurance, financial and computer services
- For Japan the below-average export and below or at average effect in imports is sustained across the models but the pattern is more mixed in the detailed data of Table 10. There the below average export of travel is preserved and below average results recorded for insurance, computer services, business and personal.
Australia-Japan bilateral effects
- Both the bilateral flows from Australia to Japan and from Japan to Australia are below the average against the larger sample of countries but these equations have low explanatory power. (Table 2 Columns 3 and 4 and Table 5)
- Against the OECD average (where equations have higher explanatory power) the Australia to Japan flow does not differ from the average while the Japan to Australia flow appears slightly above average (Table 6) but the results are not strong so it is more accurate to say we do not have strong evidence for performance that differs from the average, at least amongst OECD countries.
-
Tables 7-10, based on better quality data from both Australia
and Japan (but smaller sample sizes), show that
- In Australia’s services exports to Japan there is a consistent below-average effect in travel, communication, royalties, other, personal and government. The effect in transport, insurance, finance and computer services is not consistent across the two estimation methods. Using the PPML estimates of Table 8 (probably more reliable) insurance and finance also have below-average results while transport have weakly
- In Australia’s services import from Japan there is a below-average effect in transport, travel, communications, other, personal and government. Results are inconsistent in insurance, finance and royalties but insurance and royalties are below-average in the PPML estimates.
- In Japan’s exports to Australia there are no consistently significant patterns in the data (i.e. no coefficients are of the same sign and significantly different from zero across both estimation methods in Tables 9 and 10). Accepting the PPML estimates of Table 10 as the best specification gives below-average effects (negative and significant) in construction, computers, royalties, personal and government. An above-average effect is shown in air transport. The important sectors of transportation, travel, insurance and financial show zero effects.
- In Japan’s imports from Australia there are also no consistently significant results. Using Table 10, significant below-average effects are found in other services, construction, computers, royalties and personal. All other sectors show zero effects.
Tables
Table 2. Gravity Results: Goods vs. Services: All Sample
Trade |
Services |
Goods |
Ex GDP |
1.496*** |
1.097*** |
[0.036] |
[0.048] |
|
Im GDP |
1.655*** |
1.169*** |
[0.036] |
[0.048] |
|
Distance |
-1.193*** |
-1.642*** |
[0.056] |
[0.079] |
|
Ex Risk |
1.658*** |
4.319*** |
[0.170] |
[0.223] |
|
Im Risk |
1.364*** |
5.273*** |
[0.179] |
[0.225] |
|
Ex Remote |
-0.745*** |
0.690*** |
[0.116] |
[0.153] |
|
Im Remote |
-1.102*** |
0.262* |
[0.117] |
[0.152] |
|
Language |
1.093*** |
0.067 |
[0.169] |
[0.277] |
|
Colony |
2.104*** |
-0.048 |
[0.256] |
[0.394] |
|
Ex AUS |
2.067*** |
0.012 |
[0.280] |
[0.506] |
|
Im AUS |
2.578*** |
1.146** |
[0.269] |
[0.496] |
|
Ex JPN |
-1.583*** |
0.585 |
[0.220] |
[0.540] |
|
Im JPN |
-1.856*** |
0.174 |
[0.201] |
[0.477] |
|
AUS to JPN |
-0.935*** |
0.488 |
[0.341] |
[0.655] |
|
JPN to AUS |
-1.338*** |
-0.592 |
[0.318] |
[0.690] |
|
constant |
-111.320*** |
-47.085*** |
[4.214] |
[5.581] |
|
Year dum. |
no |
no |
Obs. |
9,469 |
9,469 |
R-sq |
0.5630 |
0.4452 |
Notes: Heteroskedasticity-consistent standard errors (White) are in parentheses. ***, **, and * show 1%, 5%, and 10% significant, respectively. The data is averaged over the three years.
Table 3. Gravity Results: Goods vs. Services: OECD Sample
Trade |
Services |
Goods |
Ex GDP |
0.836*** |
0.909*** |
[0.024] |
[0.025] |
|
Im GDP |
0.816*** |
0.953*** |
[0.023] |
[0.017] |
|
Distance |
-0.934*** |
-1.037*** |
[0.034] |
[0.032] |
|
Ex Risk |
1.652*** |
0.122 |
[0.183] |
[0.128] |
|
Im Risk |
2.140*** |
-0.293*** |
[0.186] |
[0.111] |
|
Ex Remote |
-0.129 |
0.136** |
[0.082] |
[0.069] |
|
Im Remote |
-0.083 |
0.386*** |
[0.063] |
[0.059] |
|
Language |
0.638*** |
0.692*** |
[0.076] |
[0.057] |
|
Colony |
0.500*** |
-0.018 |
[0.087] |
[0.094] |
|
Ex AUS |
0.646*** |
-0.018 |
[0.184] |
[0.150] |
|
Im AUS |
0.681*** |
0.054 |
[0.166] |
[0.279] |
|
Ex JPN |
-0.271** |
0.520*** |
[0.123] |
[0.092] |
|
Im JPN |
0.199 |
-0.182 |
[0.143] |
[0.119] |
|
AUS to JPN |
0.498*** |
1.508*** |
[0.192] |
[0.165] |
|
JPN to AUS |
0.098 |
-0.126 |
[0.192] |
[0.250] |
|
constant |
-39.121*** |
-7.828*** |
[2.569] |
[2.042] |
|
Year dum. |
no |
no |
Obs. |
2,271 |
2,271 |
R-sq |
0.7379 |
0.7814 |
Notes: Heteroskedasticity-consistent standard errors (White) are in parentheses. ***, **, and * show 1%, 5%, and 10% significant, respectively. The data is averaged over the three years.
Table 5. Gravity Results for Services Trade: All Sample
Trade |
Total |
Travel |
Transport |
Other |
Ex GDP |
1.519*** |
1.616*** |
1.509*** |
1.382*** |
[0.038] |
[0.064] |
[0.055] |
[0.053] |
|
Im GDP |
1.669*** |
1.637*** |
1.641*** |
1.242*** |
[0.038] |
[0.066] |
[0.056] |
[0.049] |
|
Distance |
-1.133*** |
-1.659*** |
-1.442*** |
-0.744*** |
[0.065] |
[0.113] |
[0.097] |
[0.092] |
|
Ex Risk |
1.825*** |
0.517 |
1.883*** |
2.041*** |
[0.180] |
[0.379] |
[0.336] |
[0.343] |
|
Im Risk |
1.695*** |
1.249*** |
2.426*** |
1.053*** |
[0.189] |
[0.401] |
[0.336] |
[0.339] |
|
Ex Remote |
-0.771*** |
0.019 |
-0.052 |
-1.062*** |
[0.121] |
[0.212] |
[0.176] |
[0.178] |
|
Im Remote |
-1.061*** |
-1.045*** |
0.21 |
-1.078*** |
[0.122] |
[0.212] |
[0.178] |
[0.176] |
|
Language |
1.221*** |
1.109*** |
-2.112*** |
1.500*** |
[0.181] |
[0.290] |
[0.329] |
[0.151] |
|
Colony |
1.399*** |
1.782*** |
2.443*** |
0.490** |
[0.279] |
[0.339] |
[0.284] |
[0.210] |
|
FTA |
-1.132*** |
0.845*** |
-0.835*** |
-0.003 |
[0.113] |
[0.241] |
[0.195] |
[0.199] |
|
Goods Intensity |
0.118*** |
0.135*** |
0.142*** |
0.141*** |
[0.013] |
[0.021] |
[0.038] |
[0.031] |
|
Ex AUS |
1.519*** |
4.251*** |
0.064 |
1.373*** |
[0.279] |
[0.532] |
[0.711] |
[0.480] |
|
Im AUS |
2.016*** |
5.509*** |
1.561*** |
2.002*** |
[0.272] |
[0.502] |
[0.523] |
[0.423] |
|
Ex JPN |
-2.066*** |
-0.415 |
-0.198 |
-0.629** |
[0.243] |
[0.444] |
[0.249] |
[0.248] |
|
Im JPN |
-2.216*** |
1.081*** |
-0.389 |
-0.122 |
[0.220] |
[0.262] |
[0.256] |
[0.248] |
|
AUS to JPN |
-0.843** |
-3.571*** |
-0.061 |
-1.023** |
[0.353] |
[0.485] |
[0.716] |
[0.504] |
|
JPN to AUS |
-0.953*** |
-2.863*** |
-1.848*** |
-0.909** |
[0.331] |
[0.536] |
[0.512] |
[0.413] |
|
constant |
-114.360*** |
-90.001*** |
-70.391*** |
-110.365*** |
[4.479] |
[7.031] |
[6.016] |
[6.136] |
|
Year dum. |
no |
no |
no |
no |
Obs. |
8,411 |
4,365 |
4,264 |
3,401 |
R-sq |
0.5703 |
0.3790 |
0.4191 |
0.4733 |
Notes: Heteroskedasticity-consistent standard errors (White) are in parentheses. ***, **, and * show 1%, 5%, and 10% significant, respectively. The data is averaged over the three years.
Gravity for Disaggregated Services
The tables below present results for the same model as in the tables above but using detailed data available from Australian and Japanese sources to give a picture of the pattern for specific services sectors.
In addition to OLS, we perform Poisson pseudo maximum likelihood (PPML) estimation technique since there are many zero values of our dependent variable. Besides being consistent in the presence of heteroskedasticity, this method also provides a natural way to deal with zero values of the dependent variable (Silva and Tenreyro, 2006, p 641)[i].
Australian Services Trade
Year: 2002-2004 (Unbalanced panel)
Partner countries (31): Canada, Chile, China, Fiji, France, Germany, Greece, Hong Kong, India, Indonesia, Ireland, Italy, Japan, Korea, Malaysia, Mexico, Netherlands, New Zealand, Norway, Papua New Guinea, Peru, Philippines, Russian Federation, Singapore, South Africa, Sweden, Switzerland, Thailand, United Kingdom, United States, Vietnam
Data source: Australian Bureau of Statistics
Services: Transport; Travel; Communication; Construction; Insurance; Financial; Computer; Royalties; Other; Personal; Government
Note: Since 95% of non-missing data in construction services trade report zero trade, we drop construction services.
Japanese Services Trade
Year: 2002-2004 (Unbalanced panel)
Partner countries (31): Australia, Belgium, Brazil, Canada, China, France, Germany, Hong Kong, India, Indonesia, Iran, Italy, Korea, Luxembourg, Malaysia, Mexico, Netherlands, New Zealand, Philippines, Russian Federation, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland, Thailand, United Arab Emirates, United Kingdom, United States, Vietnam
Data source: Bank of Japan
Services: Transportation (Sea Transport; Air Transport), Travel, Other Services (Communications Services; Construction Services; Insurance Services; Financial Services; Computer and Information Services; Royalties and License Fees; Other Business Services; Personal, Cultural, and Recreational Services; Government Services, n.i.e.)
Note: We dropped observations with negative trade values (1% of all observations).
Table 8. Australian Services Trade by Sector: PPML
Notes: Dependent variable is “Tij”. Heteroskedasticity-consistent standard errors (White) are in parentheses. ***, **, and * show 1%, 5%, and 10% significant, respectively. Since data of Australian imports from Japan in computer services are missing, dummy variable “JPN to AUS” is dropped in computer services.
Table 10. Japanese Services Trade by Sector: PPML
Notes: Dependent variable is “Tij”. Heteroskedasticity-consistent standard errors (White) are in parentheses. ***, **, and * show 1%, 5%, and 10% significant, respectively.
References
F Kimura and H H Lee (2006), “The Gravity Equation in International Trade in Services”, Review of World Economics, 142 (1): 93 – 121.
J. M. C. Santos Silva and Silvana Tenreyro (2006), “The Log of Gravity”, The Review of Economics and Statistics, 88(4): 641-658.