Accounting for pairwise heterogeneity in bilateral trade flows: a stochastic varying coefficient gravity model

Abstract

This paper suggests an alternative way for estimating the gravity equation that takes into consideration country-pair heterogeneity in bilateral trade flows. Specifically, a stochastic varying coefficient gravity model based on Hildreth and Houck's (1968) random coefficient regression is proposed, that eliminates heterogeneity bias inherent in standard econometric methods. The results indicate that the standard gravity estimates can differ substantially from what is obtained when heterogeneity is accounted for.

Notes

¹ Recently, Sanso et al. (1993) concluded that the log-linear form, while not optimal, is a fair and ready approximation of the optimal form.

² This may be an oversimplification of the behaviour of β-coefficients if β's vary in a systematic way with these quantifiable variables. However, given that such information is not always available, the present assumption seems reasonable.

³ Endoh (1999) also found similar results in analysing trade relationships between EU, LAFTA and CMEA countries during the 1960 to 1994 period. In addition, Sharma and Chua (2000) reveal that the trade in ASEAN countries is also highly affected by the rekative size of the economies, with economic integration playing a minor role. Finally, Arnon and Weinblatt (1998) found that the divergence of the GDP per capita between exporting and importing countries affect negatively bilateral trade flows.