Does firm size matter in exporting and using FTA schemes?

Abstract

In this paper, we empirically compare the role of firm size when exporting with that when using free trade agreement (FTA) schemes. We employ a unique survey providing detailed information on FTA use by Japanese affiliates in ASEAN, India, and Oceania. Our findings from the analysis on Japanese affiliates in ASEAN are as follows. First, firm size matters in both decisions on exporting and on using FTA schemes. In particular, firm size is more quantitatively important in decisions on FTA use than on exporting. Second, firms with experience in utilizing FTAs for exporting have an approximately 40% higher probability of using an FTA for exporting to a new country. Third, larger-sized firms use a larger number of FTA schemes.

Keywords:

• FTA
• export
• firm size

JEL Classifications:

• F15
• F53
• O53

Acknowledgements

I am grateful to the Japan External Trade Organization for providing me with the micro data used in this study. I also would like to thank Chih-hai Yang, Toshiyuki Matsuura, and seminar participants in National Central University for their helpful comments.

Notes

1. There are several papers in this literature in addition to Bureau, Chakir, and Gallezot (2007) and Takahashi and Urata (2010). Cadot et al. (2006) focus on EU and US trade with their preferential trading partners. Francois, Hoekman, and Manchin (2006) and Manchin (2006) examine the preferential trade relations of the EU with non-least developed African, Caribbean, and Pacific (ACP) countries under the Cotontou agreement. These three papers focus not on firm/shipment size but other elements including tariff margin or ROO.

2. In this paper, we do not examine the selection effects of FTA use in importing because fixed costs for FTA utilization are trivial for importers. All of the necessary documents concerning the use of FTA arrangements basically deal with exporters. Importers simply need to submit to customs the certificates of origin prepared and presented by exporters. Therefore, unlike the case of FTA usage in exporting or engaging in importing activities, firm size will not play a significant role in FTA scheme utilization.

3. The difference between π_FTA and π_EX eliminates f_EX. For specifics details, see Demidova and Krishna (2008).

4. One weakness of our dataset is that it does not include any other potential measures of ‘firm size’ such as sales, capital, and value-added. In our dataset, the number of employees is the only variable that can show absolute firm size.

5. Our estimation results on employment are qualitatively unchanged even if we introduce only FTA rates or both general and FTA rates into the selection equation.

6. Although the previous section does not consider any components of trade costs other than tariff rates, some components, such as non-tariff barriers, will affect firms’ decisions regarding exporting and FTA use. However, we cannot control directly for those components since data on non-tariff barriers are not available for our sample countries, which include least developed countries (e.g., Myanmar). Nevertheless, if those components are country-specific and do not change significantly during a short period like that of our sample, our inclusion of importing country dummy variables will play a role of controlling for those components.

7. It is obvious that the number of employees per se is not a perfect measure of firm size. However, our inclusion of several dummy variables will reinforce the role of employment as an indicator of firm size. For example, industry fixed effects control for differences in capital–labor ratio across industries. Such differences within an industry will also be controlled, to some extent, by the inclusion of supplier dummy. As a result, we believe that the coefficient for our variable of employment plays a significant role of showing the effects of firm size.

8. We use the share of exports rather than a dummy variable to indicate whether an affiliate has experience in exporting to other countries because almost all sample affiliates have such experience. Therefore, we expect a larger reduction of fixed trade costs in the more export-intensive affiliates.

9. The industry classification in this dataset is rather rough: food industry, textile industry, chemical industry, iron and steel, non-metallic mineral products, metal products, electric machinery industry, transport equipment, and other manufacturing industries.

10. These importing countries also have bilateral FTAs with some ASEAN countries (see Appendix). Therefore, if firms use bilateral FTA schemes and if ROOs are different between bilateral and multilateral FTAs, then ROOs will differ by country-pairs rather than by importing countries. However, the JETRO survey does not allow us to identify which specific FTA scheme is used.

11. The assigned score for each ROO is as follows: (1) for ‘Change-in-Subheading or Regional Value Content,’ (2) for ‘Change-in-Subheading,’ (3) for ‘Change-in-Heading or Regional Value Content,’ (4) for ‘Change-in-Heading,’ (5) for ‘Change-in-Chapter or Regional Value Content,’ (6) for ‘Change-in-Chapter,’ (7) for ‘Regional Value Content,’ (8) for ‘Change-in-Subheading and Regional Value Content,’ (9) for ‘Change-in-Heading and Regional Value Content,’ (10) for ‘Specific Process,’ and (11) for ‘Wholly-obtained.’

12. As mentioned above, this number is not exactly consistent with the number of countries to which they export because we treat ASEAN as a single unit of analysis.

13. Yeaple (2009) found a hierarchical relationship in investment destinations.

14. The insignificant results in FTA equation are unchanged even when excluding the size variable (i.e. in employment). In addition, instead of this continuous variable on the local input share, we tried to include a dummy variable in the local input share. This variable takes the value one if its share is as high as or higher than 40% and zero otherwise. This reflects the fact that the threshold level in the regional value content rules is mostly set to 40%. However, we again obtained insignificant results for the dummy variable in the FTA equation. These estimation results are available upon request.

15. Since Takahashi and Urata (2010) include only firm sizes (e.g. employment) in addition to industry dummy and the dummy variable taking the value one if a firm has overseas affiliates in FTA partner countries and zero otherwise, we cannot compare our results on country-level or industry-level variables with those in Takahashi and Urata (2010).

16. Instead of quartile dummy variables, we also tried the inclusion of a dummy variable that takes the value one if the total number of employees is over 500 and zero otherwise. The results are qualitatively same as those in the case of quartile variables. Namely, the coefficient for this dummy is estimated to be significantly positive in both export and FTA equations and is larger in the FTA equation. These estimation results are available upon request.

17. As in the analysis presented in the previous sections, we cannot identify the name of specific FTAs used. Specifically, even if firms use both multilateral and bilateral FTAs in exporting from ASEAN to Japan, we count the number of FTAs used as one.