Do Deep Trade Agreements with Intellectual Property Provisions Actually Increase International Trade?

This paper examines the impact of preferential trade agreements (PTAs) and deep trade agreements (DTAs) containing intellectual property rights (IPRs) provisions on export flows, for a panel of 87 countries spanning 1990-2017. We estimate a structural gravity model using the Poisson Pseudo-Maximum Likelihood (PPML) framework that simultaneously accounts for heteroskedasticity and a preponderance of zeros in trade flows, and include a rich set of fixed effects to account for multilateral resistance factors and endogeneity in binary trade agreement indicators. We introduce treatment leads and lags to account for anticipatory and phased-in effects of our key trade policy instruments. Our estimation results reveal, first, that there are no significant contemporaneous effects of PTAs or DTAs with IPRs-provisions on export flows. Any discernible effects manifest over time, as businesses gradually adapt to changes in tariff and non-tariff barriers. Second, the heterogeneous impact of deep versus shallow agreements does not appear to matter in our sample. Third, the depth of PTAs though positive significant, is economically small in magnitude. However, the depth of IPRs-related provisions and alternative policy areas within IPRs reveals a significantly negative impact on export flows. This indicates that the inclusion of additional provisions actually diminishes trade volumes, one conceivable explanation for which might be that the increased complexity of IPRs-related provisions in the DTAs renders compliance with the agreements more challenging. Fourth, our results indicate a lagged response of both PTAs and IPRs on high-IP and low-IP export flows. The patent-related, copyright-related and trademark-related provisions have a lagged impact on trade flows in the patent-intensive, copyright-intensive and trademark-intensive industries, respectively. Our findings suggest that IPRs promote trade contemporaneously in specific industrial clusters, demonstrating increased sensitivity to intellectual property. Our results are robust to various checks.