Kengoay. Pdf

4 

Results 

 

Table 1 presents the results for the simple gravity equation.  Regional dummy 

variables are included all together and then separately and jointly for the case of 

ASEAN and APEC.  The signs of the coefficients on the variables in the first half of 

Table 1 are as expected and are generally highly significant.  The coefficient on GDP 

for example, is generally between 0.7 and 0.8 suggesting that trade increases as 

economic capacity grows.

22

  We also find the traditional negative sign on distance 

and positive sign on our common border variable.  The complementarity index also 

records a positive and significant coefficient and seems to support the hypothesis that 

trade is greater when two countries endowments (reflected in the structure of the 

commodities traded) are complementary.  Only per capita GDP differences records a 

decline in significance (and even changes sign) over time.  One possible explanation 

                                                 

21

 Ideally we would have liked to include all possible RTA’s to avoid bias in the results.  

By including the main groupings in terms of trade volumes however, the bias is likely to be 

minimal.  Se e Endoh (1999) for further discussion.   

22

 Frankel (1993) notes that a coefficient of less than one reflects the phenomenon that the 

smaller the country the greater the trade/production ratio. 

---

 

17

is the Linder hypothesis that emphasises income similarity as the driver of trade 

instead of income differences. 

 

The intra-regional dummies for regressions (1) and (2) are positive for ASEAN and 

APEC implying that countries located within these regions do trade more with each 

other over and above the levels predicted by the basic explanatory variables.  In 

regression (3) we observe that the EU and NAFTA dummies are also generally 

positive and significant.

23

  The inclusion of all four regional groupings in regression 

(4) reveals however, that the NAFTA coefficient was negative and significant for the 

first two periods (prior to the setting up of NAFTA) before becoming positive.  Our 

results, showing a positive and significant ASEAN effect, differ from a number of 

previous studies such as Sharma and Chua (2000) and Soloaga and Winters (2001) 

who both observe a negative relationship albeit for a different estimating equation and 

country coverage but are similar to Frankel  et al. (1995) and Endoh (2000) who 

recorded positive and significant coefficients for APEC.  Note that the ASEAN 

coefficient falls between 1993-1997 after the AFTA process started and it was not 

until after the Asian economic crisis that the trend reverted to an upward one.  One 

possible suggestion is that it took a regional economic shock of the form of the 

currency crisis to trigger the latent forces of ASEAN regional integration that could 

not be stimulated by mere political rhetoric. 

 

Observe that when we include both ASEAN and APEC dummies the ASEAN 

coefficient is significantly lower.  This is consistent with Frankel (1993) who 

                                                 

23

 Previous studies have included a variety of European regional  dummies that have 

generally been found to be insignificant.  The results tend to differ depending on the 

country coverage and method of estimation.  When we included EU12 or EU15 dummies 

the variable was also insignificant. 

---

 

18

observed that in 1980 and 1990 the ASEAN dummy was only significant when no 

other Asian bloc dummies were included and concluded that ASEAN did not seem to 

be an appropriate bloc around which to draw a border. 

 

This leads us to enquire whether the AFTA process has been trade creating or trade 

diverting and whether ASEAN is a discriminating bloc or exhibits “open 

regionalism”.

24

  The lack of a consistent upward trend in the ASEAN dummy over 

our period of analysis deserves closer examination.  In Table 2 we include our two 

additional dummies to represent the case where only the import or export country is a 

member of the RTA.  As previously discussed, AFTA is trade creating if the 

ASEAN coefficient increases and that of the others do not change after the AFTA 

process started and the AFTA is trade diverting in two cases; (i) ASEAN member’s 

welfare is reduced if the ASEAN coefficient increases and that of  imASEAN 

decreases, (ii) non- member’s welfare is reduced if the ASEAN coefficient increases 

and that of  exASEAN decreases. 

 

In Table 2 the three regional dummies are generally significant and justifies their 

inclusion.  A comparison of the RTA coefficients with Table 1 demonstrates that the 

largest differences are for the ASEAN dummy (the EU and NAFTA coefficients 

remain relatively stable). 

 

Concentrating on ASEAN, observe that  ASEAN,  imASEAN and  exASEAN all record 

positive and significant coefficients with the former the largest in all periods.  The 

fact that all three dummies are positive and significant means that members and 

                                                 

24

 For  a  discussion  of  the  concept  of  “open  regionalism”  see  Bergsten  (1997)  and 

Yamazowa (1992). 

---

 

19

non- members have traded with each other more than the hypothetical trade level.  

Examining coefficient changes over time we observe again that ASEAN falls between 

1993 and 1997 and then rises while  imASEAN increases until 1997 (despite the 

ongoing AFTA process) and then decreases while exASEAN demonstrates a consistent 

rise (with a significant jump after the crisis).  Considered jointly, the  imASEAN and 

exASEAN coefficients reveal the extent of the extra-ASEAN trade bias over the 

period.   The generally increasing trend means that there has been a  negative trade 

diversion effect.  More specifically, the upward trend in  exASEAN indicates that