The Wise Use of Dummies in Gravity Models: Export Potentials in the Euromed Region

Abstract

In this paper, we estimate a gravity equation properly accounting for omitted exporter and importer's overall trade resistance, through country-yearly dummies for exporter and importer countries. We find that the omission of time-varying multilateral trade resistance terms in the estimation of a gravity equation introduces important biases in the results, although correcting them means we can only compute differences between actual and predicted export shares, instead of levels, as usually done. An application to the calculation of trade potentials in the Euromed region (Southern and Eastern Mediterranean countries)shows that the omission of time-varying multilateral trade resistance terms greatly influences the computation of export potentials as well as the estimated effect of signing a free trade agreement. Overall, we find that, except for Algeria, Jordan and Lebanon, Euromed countries' share of exports to the EU as a whole is at, or slightly above, those predicted by a correctly-specified gravity model, although the share of exports to some individual EU countries is significantly below the predictions of the gravity model. Except for those three countries, we find significant opportunities for export growth to the US, instead.

Information barriers, export promotion institutions, and the extensive margin of trade

“…Hence, there is potential for omitted variable biases originated in the time dimension of the data. In order to assess whether this is indeed an issue in our case, we check the robustness of the estimates to replacing time-varying exporter and importer fixed effects for their time-constant versions (see, e.g., Baldwin and Taglioini, 2006;and Ruiz and Vilarrubia, 2007).…”

Determinants of trade with solar energy technology components: evidence on the porter hypothesis?

“…Furthermore, the panel context allows for controlling the heterogeneity among countries and temporal effects by means of country and year fixed effects. In this context, the role of bilateral trade costs is addressed repeatedly (Anderson and van Wincoop, 2003;McCallum, 1995 (Egger, 2000;Egger and Pfaffermayr, 2003;Matyas, 1998;Ruiz and Vilarrubia, 2008). Under the assumption of homoscedasticity, applied empirical trade studies transform the model into logs establishing a linear relationship estimated with simple fixed effect ordinary least squares (OLS) that conveniently allows for interpretation in percentage changes.…”

Economic Integration in the Euro-Mediterranean Region

“…These last results are consistent with Söderling's (2005) estimates. On the other hand the results for Morocco show some inconsistency, with Ruiz and Vilarrubia (2007) indicating that it is over-exporting to France, while the opposite was the case in Söderling's (2005) study.…”

“…Finally, a recent study by Ruiz and Vilarrubia (2007) finds that previous studies, which did not properly take into account the overall multilateral trade resistance, have tended to overestimate the trade potentials for the region. The authors employ country-year fixed effects in the estimation to solve the problem of a potential bias stemming from omitted variables.…”

“…This section reviews the following studies: Péridy (2005a and2005b), Ferragina, Giovannetti and Pascore (2005), Nugent and Yousef (2005), Al-Atrash and Yousef (2000), Söderling (2005) and Ruiz and Vilarrubia (2007). Péridy (2005a) aims at evaluating the intra-MENA trade flows for 5 MENA countries.…”