Does the GATT/WTO promote trade? After all, Rose was right

Abstract

This paper re-examines the effect of the GATT/WTO on trade using recent econometric developments that allow us estimating structural gravity equations with the Poisson pseudo-maximum likelihood (PPML) estimator on a large dataset that requires computing high-dimensional fixed effects. By doing so, we overcome computational limitations that are present in previous studies. In line with Rose's (Am Econ Rev 94: [98][99][100][101][102][103][104][105][106][107][108][109][110][111][112][113][114] 2004) seminal work, we find that, unlike regional trade agreements and currency unions, the GATT/WTO accession has not generated positive trade effects. This result is robust to the use of alternative measures of trade flows, across periods and country groups, to changes in the periodicity of the data, when taking into account the GATT/WTO accession dynamics, to controlling for the participation of only one country of the dyad in GATT/WTO, to the consideration of non-member participants, and to the use of alternative datasets. Notwithstanding, we also find that PPML results are sensitive to the definition of the dependent variable.

The impacts of the World Trade Organization on new members

“…By controlling these aspects, the author suggests that the WTO did not promote trade. Esteve‐Pérez et al (2020) reached the same conclusion using the PPML estimator with three types of fixed effects, simultaneously accounting for unobserved bilateral heterogeneity and endogeneity (with country‐pair fixed effects) and multilateral resistance terms (with exporter‐time and importer‐time fixed effects) in a large dataset with more than 200 countries over the period 1948–2013.…”

“…Dutt et al (2013) included zeros when adopting the two‐stage estimator proposed by Helpman et al (2008), (henceforth HMR), but they did not take into account existing WTO asymmetries. Esteve‐Pérez et al (2020) used the PPML estimator and three types of fixed effects, but only on aggregate trade flows. Therefore, none of the studies cited used the PPML estimator to analyse the effects of the WTO on disaggregated trade flows, to the best of our knowledge.…”

“…When this ‘mystery’ seemed to have been unravelled, Esteve‐Pérez et al (2020) again suggested that the GATT/WTO did not promote world trade after correcting the omitted variable bias in previous studies and computing high‐dimensional fixed effects on a large dataset, respectively. Although the gravitational equation was the tool of analysis for the aforementioned authors, the way to specify the fixed effects, the estimator, and the periodicity of the database were different, which possibly explains the lack of consensus on the subject and renders this debate still open.…”

Political economy of aid allocation: The case of Arab donors

“…An alternative and more robust approach to the ad‐hoc shifted log specification is the Poisson pseudo maximum likelihood (PPML) method suggested by Silva and Tenreyro (2006, 2011). This approach has become a standard tool in the trade literature when a data set has many zero values in the dependent variable (Esteve‐Pérez, 2020). Silva and Tenreyro (2006, 2011) demonstrated that PPML models perform better than OLS in the presence of heteroscedasticity and many zero observations.…”

Has the euro been fattening the European pig meat trade?

“…Most studies have shown that the introduction of the euro had a positive and significant effect on trade among eurozone members (Cieślik et al 2012;Camarero et al 2014;Glick and Rose 2016;Kunroo et al 2016;Larch et al 2018;Esteve-Pérez et al 2020). Meanwhile, Santana-Gallego et al (2016) assess the positive impact of the euro on tourism flows.…”