Gravity channels in trade

Abstract

Gravity variables such as distance, adjacency, colony, free trade agreements or language are used to capture the effects of trade costs in empirical studies. By using actual data on trade costs, this paper decomposes the overall effects of such variables on trade into those through three gravity channels: duties/tariffs (DC), transportation-costs (TC), and dyadic-preferences (PC). As opposed to the existing literature where gravity variables act like supply shifters (through DC and TC), this paper empirically shows that they act like demand shifters (through PC). Regarding policy, it is implied that welfare-improving globalization cannot be achieved only through reductions in direct costs such as duties/tariffs or transportation costs; it is rather the globalization itself that should be promoted in order to shift the preferences of destination countries toward international products and thus reduce indirect trade costs. The results are further connected to several existing discussions in the literature, such as welfare gains from trade and the distance puzzle.

Keywords:

• Gravity variables
• dyadic preferences
• U.S. imports

JEL Classifications:

• F12
• F14

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Also see Head and Mayer (2014) for an excellent survey based on other recent studies.

2 Such an empirical approach is also in line with studies such as by Hillberry et al. (2005) who find that taste parameters are correlated with bilateral trade costs or by Head and Mayer (2013) who suggest that preference differences have the potential to explain some of the large distance and border effects that have been found in the literature. For instance, based on the estimate of indirect trade costs in Head and Mayer (2013) shows that between 50% to 85% of the distance effects on trade flows are due to indirect trade costs (that are called dark trade costs). In addition to these studies, this paper can directly identify the role of direct versus indirect trade costs through each and every gravity variable (rather than only distance) using actual data on trade costs.

3 Studies such as by Giri, Yi, and Yilmazkuday (2021) show that having a common trade elasticity (that can be represented by ${\eta }_t-1$ in this paper) across goods would result in similar (welfare) implications compared to having good-specific trade elasticity as long as these elasticities are estimated in a model-consistent way, which is the methodology followed in this paper.

4 In earlier studies, after using the connection between the elasticity of substitution η and the trade elasticity $\eta -1$, Hummels' (2001) trade-elasticity estimates range between 3.79 and 7.26, the estimates of Head and Ries (2001) range between 6.9 and 10.4, the estimate of Baier and Bergstrand (2001) is about 5.4, Harrigan's (1996) estimates range from 4 to 9, Feenstra's (1994) estimates range from 2 to 7.4, the estimate by Eaton and Kortum (2002) is about 8.28, the estimates by Romalis (2007) range between 5.2 and 9.9, the (mean) estimates of Broda and Weinstein (2006) range between 3 and 16.3.

5 https://dataweb.usitc.gov/

6 Since we estimate the implications of our model for these unit prices in a separate equation, potential measurement errors are captured by the corresponding residuals.