Trade barriers, multinational involvement and intra-industry trade: panel data evidence from India

Abstract

The article analyses the effects of trade barriers and multinationals on the intensity of intra-industry trade (IIT) in a panel of Indian manufacturing industries from 1988 to 1999. We find that the intensity of IIT increases with the reduction of trade barriers. This is expected as greater competition from imports leads individual plants in the domestic industry to specialize in the manufacturing of unique varieties. The analysis suggests that horizontal (market seeking) multinational activities in the domestic industries exert a negative influence on IIT. This is consistent with the view that horizontal multinationals displace exports to the host country. At the same time, our results indicate that IIT will be stimulated to the extent that the entry of multinationals induces intra-industry specialization. We also analyse the role of product differentiation and plant level scale economies in determining IIT.

Acknowledgements

I thank an anonymous referee for valuable comments. Deb Kusum Das kindly provided the estimates of trade barriers in India, which is gratefully acknowledged. Thanks are also due to D. Narayana, K. J. Joseph, P. K. M. Tharakan, K. L. Krishna, Arvind Virmani, B. N. Goldar and participants of ETSG conference (University of Nottingham, September 2004) for comments on an earlier version.

Notes

¹ Some of the previous studies (Greenaway et al., 1995; and Sharma, 2004) have used relative unit values of exports and imports for disentangling total IIT into horizontal IIT (where goods are differentiated by attributes) and vertical IIT (where goods are differentiated by quality). Aiginger (1997) discusses the various problems in using unit value as an indicator of quality. In any case, we do not attempt this due to data constraints.

² Grubel–Lloyd index is a biased downward measure of IIT if the country's total commodity trade is imbalanced (Grubel and Lloyd, 1975).

³ Values of the indices at the disaggregated level are not reported. Interested readers may obtain the table from the author.

⁴ An analysis of country-specific determinants in India's bilateral IIT can be seen in Veeramani (2002).

⁵ Wei (2005) shows that the FDI inflows to India have been mostly horizontal in nature. Horizontal FDI may displace exports to the host country not only from the home country of the multinationals but also from other countries producing similar products.

⁶ See Owusu-Gyapong (1986) for a detailed discussion and application of these specification tests.

⁷ The selected country groups and countries together account for nearly 90% of India's total trade.

⁸ Cross-reference ‘a’ in Tables 2 and 3 indicates that one-year lagged value of the particular variable is used instead of the contemporaneous value. The result does not become better if we include the lagged value of only TAR _it (or QR _it ) while retaining the contemporaneous value of all other variables.

⁹ The rationale for considering this variable (defined as the number of finely disaggregated products comprising of each 4-digit ISIC code) is that IIT may arise from the aggregation of heterogeneous commodities into common industry categories (Tharakan, 1984).

¹⁰ Note that the variable TLD is time-invariant. An advantage of the RE model relative to the FE model is that it is possible to include the time-invariant regressors in the former but not in the latter. Hausman statistics are not reported whenever we include TLD because, in any case, FE model can not be used in such cases.

¹¹ The R ² values for ‘All Industries’ are small. Previous studies have also reported such low R ² values. Examples are, Loertscher Wolter (1980), Greenaway et al. (1995) and Sharma (2004).

¹² Estimations have been done both including and excluding TLD. The results reported in the table are those obtained when TLD is included because these regressions show higher R ² values. Hausman statistic (when TLD is excluded) supported the use of RE model. Experiments have also been made by excluding ADV _it and ADV _it ² and by using the lagged values of other variables as done above. But, the results are not affected.

¹³ These results are not reported to save space. It might be noted that trade liberalization has been undertaken in a gradual and phased manner in India. Therefore, to a certain extent, the year dummies might pick up the effect of trade liberalization, rendering TAR _it and QR _it statistically insignificant.

¹⁴ Further, it is generally held that the rising share of IIT under trade liberalization is associated with a relatively low adjustment cost (Brülhart and Thorpe, 2000).