Estimating the effect of the Internet on international trade

Abstract

This paper estimates the effect of the Internet on promoting international trade. The Internet can reduce the information cost for traders. We study the effect of the Internet on trade by augmenting the gravity equation with the Internet. The empirical results show that a 10% increase in the Internet users increases international trade by 0.2%–0.4%. The results are not sensitive to circumstances such as adding time-varying country fixed effects into the gravity and controlling for infrastructure measures. The results do not seem to be resulting from some small subset of the data and the results are robust to considering zero trade flows. Furthermore, our results are not driven by the time series variation of the panel data either. To address the endogeneity issue, we use the civil liberty index from the Freedom House as an instrument, system-GMM (generalized method of moments) approach and heteroskedasticity identification strategy, and the results still show strong effect of the Internet on trade improvement.

Keywords:

• Internet
• international trade
• gravity equation

JEL Classifications:

• F10
• O30

Acknowledgements

I am grateful to the editor, professor David Giles and the anonymous referees for their invaluable comments that helped to improve this paper. I also thank Kofi Otumawu-Apreku for helping to proof-read parts of the paper.

Notes

1. Fixed cost of trade is an important assumption in current trade theory with firm heterogeneity, see e.g. Melitz (2003), Chaney (2008) and Helpman, Melitz, and Rubinstein (2008).

2. For instance, Baldwin (1988) shows that to the extent that such costs are sunk they may be responsible for explaining observed hysteresis in trade flows. Roberts and Tybout (1997) find evidence that sunk entry costs help explain why Colombian firms export in a particular year. Freund (2000) finds that trade links between original members of the European Union are highly persistent, and hypothesizes that entry costs are likely to blame.

3. See e.g. Rauch (1996), Belderbos and Sleuwaegen (1998) and Rauch and Casella (2003).

4. Though using Anderson and van Wincoop's approach the bias lessens, it still exists quantitatively.

5. During the earlier years before the late 1970s, such a model was not grounded in formal theoretical foundations and it appealed to informal economic foundations and a physical science analogy. Since 1979, formal theoretical economic foundations for a gravity equation have come up and a notable feature common to all these studies is an explicit role for prices or specific form of multilateral price indexes. See Anderson (1979), Bergstrand (1985), Eaton and Kortum (2002) and Anderson and van Wincoop (2003) (among many others) for examples. Feenstra (2004) notes that we can simply use country-specific fixed effects to capture such price effects.

6. For example, a country's business cycle, its cultural, political or institutional characteristics, as well as unobserved factor endowments. Baier and Bergstrand (2007), Stack (2009) and Rose and Spiegel (2011) also incorporate country–time-variant effects in gravities.

7. Some of these variables (usually dummies) are time-invariant, and these dummies will be excluded once we control for the time-invariant country-pair fixed effects. Therefore, we obtain the estimates of these time-invariant dummies without controlling country-pair fixed effects.

8. Though there are cases on restricting exporting or promoting importing like raw materials and high-tech goods. The figure is calculated based on the data in UNCTADstat. See http://www.unctad.org/Templates/Page.asp?intItemID=1584&lang=1.

9. See e.g. the subsidy in agriculture goods in South Korea, Japan, Europe and the United states and the duty drawback scheme in China and more examples in WTO website.AU:Please spell out “WTO” in full at first mention.

10. Railways are the measures of infrastructure commonly used in the literature, see e.g. Donaldson (2010).

11. Ferrantino, Liu, and Wang (2008) find strong statistical evidence of under-reporting exports at Chinese border to avoid paying value-added tax.

12. For example, it would not be surprising to find that Tajikistan and Togo did not trade in a certain year. The existence of zero trade between many pairs of countries is directly addressed by Hallak (2006) and Helpman, Melitz, and Rubinstein (2008).

13. Here, we do not report the first stage results. Economic scale, the Internet, having the same language, with the same border, colony relations and RTA will positively affect the probability of doing bilateral trade. Landlocked and island characteristics seem to negatively affect the probability of doing bilateral trade.

14. Including truncation of the sample (that is, elimination of zero-trade pairs) and further nonlinear transformations of the dependent variable.

15. Lin (2013) uses the PPML method to solve the distance puzzle with zeros.

16. Zhang and Zhu (2011) use Chinese Wikipedia block by government as a natural experiment to study the relationship between group size and incentives to contribute.

17. The website is http://www.freedomhouse.org/.

18. Arellano and Bond (1991) use the difference-GMM method first proposed by Holtz-Eakin, Newey, and Rosen (1988).

19. For example, an AR(1) regression using mean log value of exports from country i to country j yields an autoregression coefficient of 1.01 when controlling for a linear trend. Moreover, the augmented Dickey–Fuller test does not reject the hypothesis of a unit root in trade levels at the 90% confidence level.

20. For instance, in our case, skeptical readers may argue that CL is not purely exogenous assigned. We refer the reader to these literatures for details of the methodology.