The WTO Trade Effect

Pao Li Chang, Singapore Management University

Abstract

Rose (2004) showed that the GATT/WTO membership does not enhance trade between countries, once other conditioning factors are taken into account. Rose's study is subject to three potential shortcomings. First, the approach is parametric and thus is prone to biases due to misspecifications. Second, the countries are likely to be spatially correlated, and this problem gets aggravated by taking `country-couples' as the observation units. Third, added to this spatial dependence is the usual temporal dependence over time in the country panel data. This paper proposes nonparametric methods to estimate the treatment effect of the GATT/WTO membership, which circumvent the complicated spatial and temporal dependence in the data. In particular, we adopt matching methods to obtain point estimates of treatment effect, and permutation test and signed-rank test to assess the statistical significance of the treatment effects. We examine the effect of the GATT/WTO membership, in comparison with the effect of the Generalized System of Preferences (GSP), on bilateral trade flows. Various matching criteria are experimented: matching without restriction, matching restricted to observations of the same country-couple, matching restricted to observations of the same year, matching restricted to observations of the same GATT/WTO period, and matching restricted to observations of the same income class. Contrary to Rose (2004), the finding is an overwhelming positive effect of GATT/WTO membership, not less than that of the GSP.