KCG Seminar on Extensive Margin of Trade by Julian Hinz

With intensifying global trends towards protectionism, it becomes even more important than before to provide reliable research evidence on determinants of international trade and on their economic and social consequences. This requires the application of adequate (and creative) research methods for data analysis, taking limitations and difficulties of different potential research methods best into account.

Focusing on one of the major methodological challenges for trade-related analysis, namely the difficulty in considering bilateral unobserved heterogeneity in estimating the extensive margin of trade, Dr. Julian Hinz, KCG Research Fellow and Postdoctoral Researcher at the Kiel Institute for the World Economy, will give a presentation “Zeroing in on zeros: Bias-Corrected Fixed Effects Panel Estimation of the Extensive Margin of Trade” to show their methodological progress made in dealing with this challenge. Abstract of the paper, co-authored with Joschka Wanner from the University of Bayreuth, can be found below.

Abstract: Estimates of the determinants of the extensive margin of trade do not take into account bilateral unobserved heterogeneity for two reasons: (i) the large number of required fixed effects precludes the practical implementation of appropriate specifications and (ii) even if estimates could be obtained, they would suffer from an incidental parameter problem. We show how a fixed effects logit estimator with exporter-time, importer-time, and country pair fixed effects can be implemented for large panel data sets via an iteratively re-weighted least squares procedure for generalized linear models. Additionally, we demonstrate how a jackknife bias correction for the incidental parameter problem associated with these three-way high-dimensional fixed effects can be applied. We use this procedure to reassess the effect of joint WTO membership on the probability that a country pair engages in (aggregate and sectoral) bilateral trade and demonstrate the behaviour of the estimates in a calibrated Monte Carlo study.

The seminar will take place in the Medienraum (A-211) at the Kiel Institute for the World Economy (Kiellinie 66, 24105 Kiel, Germany) at 12.00 – 1.00pm on August 24, 2018 (Friday).