The commercial benefits for Japan of its aid to Asia

1
The commercial benefits for Japan of its aid to
Asia Sabit Amum Otor ANU graduate and Research
Associate at the Development Policy Centre
2
1. Introduction

• The core objective of official development
  assistance (ODA) is to promote sustainable
  economic development of the recipient country.
• But the strategic and commercial goals of donors
  are also recognized to be important for aid.
• Commercial impact of aid is possible through
  tying of aid (including informal tying) and
  through creation of goodwill and habit
  formation
• But most studies focus on the development impact
  of aid. There are only a few on the commercial
  impact of aid.

3
1. Introduction (cont.)

• There are several studies on the trade impact of
  European aid, but only one that includes Asia.
• They have mixed results, but several studies
  find
• positive impact of own-country aid on its trade.
• negative impact of other-countries aid on the
  countrys trade.
• The study by Nowak-Lehmann and his colleagues
  (N-L, 2009) on the impact of German aid on its
  exports is the model for this paper.
• N-L find
• US1 of German aid has an average return of US
  1.04-1.50 in exports.
• aid from other European countries crowds out
  German exports
• ODA causes exports not vice versa

4
1. Introduction (cont.)

• I use the same methodologies as N-L to ask the
  same questions of Japanese exports and aid to
  Asia.
• Japan is of interest because
• It is a major donor.
• In 2010, Japan disbursed about US11 billion of
  foreign aid to developing countries. About 67
  percent of this amount was allocated as bilateral
  aid.
• Japan's bilateral ODA has been concentrated in
  Asia.
• Wagner (2003) the only study to examine Japan
  (and other major donors)
• Finds an average return for Japan from aid on
  trade of US 1.20
• Only looks at short-run, and study now out of
  date.

5
2. Research Questions

• This study investigates
• the short-and long-run effects of Japans ODA on
  Japans exports to the recipient countries
• the short-and long-run effects of DAC ODA
  (excluding Japan ODA) on Japans exports to the
  recipient countries
• the causal relationship between Japans ODA and
  Japans exports to the recipients

6
3. Methodology

• Like N-L, this study uses a gravity model of
  international
• trade, and applies two different econometric
  techniques
• Dynamic Ordinary Least Squares to get long-run
  estimates
• Error Correction Model to get short-run and
  long-run estimates, to provide a robustness test,
  and to test for Granger causality
• Data from Japan and15 recipient countries of
  Japans ODA in Asia (including West Asia), during
  the period between 1972 and 2008
• The sample includes Bangladesh, Bhutan, India,
  Indonesia, Lao, Lebanon, Malaysia, Maldives,
  Myanmar, Nepal, Pakistan,, Philippine, Thailand,
  Sri Lanka and Syria.
• Focus on Asia because it is a major recipient of
  Japanese aid and because of data limitations

7
3. Methodology (cont.)

• The gravity model states that the trade between
  two countries is explained by their gross
  domestic products and populations, by the
  distance between their two economic centres, and
  by country-pair fixed factors that impede or
  facilitate trade such as whether two trading
  partners have trade agreements, common language,
  and common border and whether one or both of
  them have had a colonial history.
• For my research purposes I also include
  own-country and other-country aid to the
  recipient as an explanatory variable for
  own-country exports to the recipient.

8
4. Results

• Panel unit root and co-integration tests
• The unit root tests show that the variables in
  levels are not stationary, but the
  first-differenced of the variables are
  stationary. The test for co-integration shows
  that there is convincing evidence of
  co-integration relationship among the data
  series. This means that we can use the DOLS
  (Dynamic Ordinary Least Squares) and the ECM
  (Error Correction Model).
• Weak exogeneity test (to address the issue of
  endogeneity)
• The test shows that the dependent variables
  (except for exports) are weakly exogenous. This
  means we can use ECM (Error Correction Model) to
  explain exports.
• Note the DOLS method produces good results even
  some or all of regressors are endogenous.
• 
  

9
4. Results (cont.)
Summary results The return of Japanese and other
DAC aid on Japanese exports (USD)
Long-run Long-run Short-run
DOLS ECM ECM
Average return on bilateral aid (Japan) 1.2 1 1
Average return on other DAC bilateral aid (Japan excluded) 2.2 1.4 0.2
Total average return on bilateral aid (Japan other DAC countries) 3.4 2.4   1.2
Summary results reported using favoured model
version controlling for heteroscedasticity and
serial correlation, but all versions for each
technique give similar results. All aid
coefficients significant at 10 level, except for
short-run other DAC.
10
4. Results (cont.)
Tests for Granger causality show that in both the
short and long-run, Japanese aid causes exports
but not vice versa.
Dependent variable Source of causation (independent variable) Source of causation (independent variable) Source of causation (independent variable) Source of causation (independent variable) Source of causation (independent variable) Source of causation (independent variable)
Short run Short run Long run
LEXPJAP LAIDJAP ECT Joint (ECT and LEXPJAP) Joint (ECT and LAIDJAP)
LEXPJAP - 4.41 -0.26 - 18.69
LAIDJAP 0.13 - 0.04 0.23 -
11
5. Conclusion

• Impact of Japanese ODA on Japanese exports
• Japans ODA has positive and significant impact
  on Japans exports to Asian countries
• These impacts are not only limited to the
  short-run, but are larger in the long-run.
• These results are similar to Wagner for Japan
  for the short run (Wagner US 1.20 Otor 1.00)
• These results are also similar to those of N-L
  for Germany (cf N-L, Germany US 1.04-1.50 v.
  Japan US 1.0-1.80 for long-run)
• We also find that an increase in Japanese ODA
  causes an increase in Japanese exports, but not
  vice versa.

12
5. Conclusion (cont.)

• Impact of other-DAC country ODA on Japanese
  exports
• ODA from other DAC donors also has positive and
  sometimes significant impact on Japans exports
  to Asian countries
• These impacts are not evident in the short-run,
  but are very large in the long-run.
• These results are different to those of N-L for
  Germany and also studies of Switzerland
  (Zarin-Nejadan 2008), which mainly find either
  complete or partial crowding out of other-country
  ODA on own-country exports.
• Unclear why we get a different result. May rest
  on large commercial benefits for Japan from
  aid-induced Asian development

13
5. Conclusion (cont.)

• Summary
• This research supports other findings that
  own-country aid does increase own-country
  exports.
• It also suggests, though less clearly, that
  other-country aid increases own-country exports,
  but this seems to vary from country to country.
• Given the increasing interest in aid for trade in
  Australia, it would be useful to conduct a
  similar analysis for Australian aid and trade.

14
Thank you

• And happy to take your comments and questions

15
Appendix 1

• The gravity model of international trade
• Following the recent literature, I included the
  exchange rate and ODA for both Japan and other
  major donors (Japan excluded) variables into the
  equation (1), and then transformed into
  log-linear form. After restricting
• and The equation (1) can be written as
  
• 
  

16
Appendix 2

• Dynamic Ordinary Least Squares (DOLS)
• This model was proposed by Kao and Chiang (
  2000). They propose regressing the dependent
  variable onto contemporaneous level regressors,
  lags and leads of the first differences, and a
  constant using ordinary least squares

This estimation technique produces unbiased
estimates even when some or all regressors are
endogenous
17
Appendix 2 (Con.)

• Dynamic Ordinary Least Squares (DOLS)
• This model was proposed by Kao and Chiang (
  2000). They propose regressing the dependent
  variable onto contemporaneous level regressors,
  lags and leads of the first differences, and a
  constant using ordinary least squares.
• This estimation technique produces unbiased
  estimates even when some or all regressors are
  endogenous

18

• Appendix 2 (Con.)
• Weak Exogeneity and Causality Tests
• 
  
• 
  (4)
• The test for the null hypothesis (in each
  equation) that against the alternative that
  using t-test. If the estimated coefficient of the
  lag of equilibrium residual variable is
  insignificant (i.e. fail to reject the hull
  hypothesis), then the dependent variable of that
  equation is weakly exogenous.

19

• Appendix 2 (Con.)
• ECM
• 
  
  
• 
  
• Note By comparing equations (5) and (6) it is
  easy to derive estimated coefficients of the
  variables in equation (5) from estimated
  coefficients of equation (6).
• The equation (6) is estimated with 3 lags. And
  after applying the General-to-Specific technique
  we reported the estimated results of this
  equation in Table (6).

20

• Appendix 3
• Results
• Unit Root Test
• Two tests the first test was proposed by
  Breitung (2000), and the second was proposed by
  Choi (2001)

Breitung Breitung Fisher-ADF Fisher-ADF
Statistic Prob Statistic Prob
Level
LEXP 1.21 0.89 -3.56 0.00
LTGDP 2.19 0.20 -0.60 0.28
LTPOP -10.86 0.00 -0.70 0.24
LEXCH 1.59 0.94 2.70 1.00
LAIDJAP -2.66 0.00 0.71 0.76
LAIDDAC 1.80 0.96 0.21 0.58
First-difference
?LEXP -8.22 0.00 -16.34 0.00
?LTGDP -2.43 0.00 -10.69 0.00
?LTPOP 4.32 1.00 -1.95 0.03
?LEXCH -9.87 0.00 -10.33 0.00
?LAIDJAP -11.93 0.00 -21.40 0.00
?LAIDDAC -9.98 0.00 -16.37 0.00
21
Appendix 2 (Con.) Results

• Panel Co-integration test, Pedroni (1999, 2004)
• Estimated results
• Note indicates statical significant at 1
  level. Probabilities

unweighted unweighted weighted weighted
Statistic Prob Statistic Prob
Within-dimention
Panel v-Statistic -0.35 0.64 -3.20 1.00
Panel rho-Statistic 1.37 0.91 -0.35 0.36
Panel PP-Statistic -2.18 0.01 5.66 0.00
Panel ADF-Statistic -2.05 0.02 -5.77 0.00
between-dimension
Group rho-Statistic 0.51 0.70
Group PP-Statistic -5.86 0.00
Group ADF-Statistic -6.11 0.00
22
Appendix 3 (Con) Results
Dynamic Ordinary Least Squares (DOLS)
Technique (1) (1) (1) (2) (2) (2) (3) (3) (3)
Variable Estimates Estimates Stats Estimates Estimates Stats Estimates Estimates Stats
LTGDP 0.25 1.98 1.98 0.25 0.97 0.97 0.53 5.06 5.06
LTPOP -2.35 -4.57 -4.57 -2.35 -2.69 -2.69 -1.44 -2.26 -2.26
LEXCH -0.21 -4.51 -4.51 -0.21 -2.83 -2.83 -0.18 -2.56 -2.56
LAIDJAP 0.19 3.53 3.53 0.19 2.14 2.14 0.13 1.67 1.67
LAIDDAC 0.28 4.04 4.04 0.28 1.85 1.85 0.29 2.98 2.98
Long-run return on bilateral aid (Japan) US1.7 US1.7 US1.7 US1.7 US1.7 US1.7 US1.2 US1.2 US1.2
Long-run return on other DAC bilateral aid (Japan excluded) US2.2 US2.2 US2.2 US2.2 US2.2 US2.2 US2.2 US2.2 US2.2
Total long-run return on bilateral aid (Japan other DAC countries) US3.9 US3.9 US3.9 US3.9 US3.9 US3.9 US3.4 US3.4 US3.4
Dummy for country fixed effects yes yes yes yes yes yes Yes Yes Yes
Dummy for year fixed effects yes yes yes yes yes yes No No No
Adj R2 0.96 0.96 0.96 - - - - - -
R2 0.96 0.96 0.96 - - - 0.95 0.95 0.95
Obs 480 480 480 480 480 480 480 480 480
Model 1 doesnt control for heteroscedasticity
and serial correlation Model 2 and 3 do. ,
and indicate statistical significance at the 1
5 and 10 respectively.
23
Appendix 3 (Con.) Results
Weakly Exogeneity test results
Dependent Variable Number of lags
1 lags 2 lags 3 lags
Estimates t-stats Estimates t-stats Estimates t-stats
?LEXP ?LTGDP -0.47 -17.64 0.00 0.98 -0.81 -35.17 0.00 0.35 -0.26 -7.77 -1.12 0.74
?LTPOP -0.00 -1.35 -0.00 -1.30 -0.00 -1.91
?LEXCH -0.02 -0.24 -0.00 -0.09 -0.01 -0.37
?LAIDJAP 0.00 0.15 -0.00 -0.08 0.04 0.71
?LAIDDAC -0.00 -0.31 -0.01 -0.45 -0.02 -0.54
24
Appendix 3 (Con) Results
Error Correction Model (ECM) long-run results
Technique (1) (1) (2) (2) (3) (3)
Variable Estimates Stats Estimates Stats Estimates Stats
Long run estimates Long run estimates Long run estimates Long run estimates Long run estimates
LTGDP 0.38 3.3 0.38 1.83 0.62 7.66
LTPOP -1.82 -3.7 -1.82 -2.26 -1.52 -3.4
LEXCH -0.20 -4.63 -0.20 -3.08 -0.15 -2.86
LAIDJAP 0.20 4.37 0.20 2.97 0.11 2.34
LAIDDAC 0.22 3.64 0.22 1.76 0.18 2.81
ECTt-1 -0.82 -37.43 -0.82 -6.63 -0.95 -60.68
Long-run return on bilateral aid (Japan) US1.8 US1.8 US1.8 US1.8 US1.0 US1.0
Long-run return on other DAC bilateral aid (Japan excluded) US1.7 US1.7 US1.7 US1.7 US1.4 US1.4
Total long-run return on bilateral aid (Japan other DAC countries) US3.5 US3.5 US3.5 US3.5 US2.4 US2.4
Model 1 doesnt control for heteroscedasticity
and serial correlation Model 2 and 3 do. ,
and indicate statistical significance at the 1
5 and 10 respectively.
25
Appendix 3 (Con)Results
Error Correction Model (ECM) short-run results
Short run estimates Short run estimates Short run estimates Short run estimates Short run estimates Short run estimates Short run estimates Short run estimates
LEXPJAPt-1 -0.03 - 1.51 - 1.51 -0.03 -1.31 -1.31 -0.06 -3.56
LTGDP 0.74 3.60 3.60 0.74 1.91 1.91 0.62 5.92
LTPOP -9.49 -2.28 -2.28 -9.49 -1.61 -1.61 -11.84 -1.38
LEXCHt-2 0.10 1.00 1.00 0.10 1.4 1.4 0.07 1.06
LAIDJAP 0.15 4.17 4.17 0.15 3.78 3.78 0.11 3.33
LAIDDAC 0.02 0.35 0.35 0.02 0.31 0.31 0.03 0.6
Short-run return on bilateral aid (Japan) US1.4 US1.4 US1.4 US1.4 US1.4 US1.0 US1.0 US1.0
Short-run return on other DAC bilateral aid (Japan excluded) US0.2 US0.2 US0.2 US0.2 US0.2 US0.2 US0.2 US0.2
Total short-run return on bilateral aid (Japan other DAC countries) US1.6 US1.6 US1.6 US1.6 US1.6 US1.2 US1.2 US1.2
Model 1 doesnt control for heteroscedasticity
and serial correlation Model 2 and 3 do. ,
and indicate statistical significance at the 1
5 and 10 respectively.