Robustness of the Extensive Margin in the Helpman, Melitz and Rubinstein (HMR) Model

The HMR model extends the classical gravity model of trade to correct for the large number of zeros in the world trade matrix (export selection) and for the unobservable fraction of exporting fi�rms (extensive margin). They �find that, while omission of both of these corrections result in the biased estimates of the gravity model, the extensive margin correction is the most signi�ficant of the two when estimating the trade flows. I test the robustness of this conclusion by splitting the world trade data into OECD and non-OECD countries. The extensive margin should be both economically and statistically more signi�ficant for the OECD exporters, while export selection should play a larger in the trade flows for the non-OECD exporters. I �find that the extensive margin is not signi�ficant for the OECD trade flows, but the export selection is important regardless of the exporter location. These �ndings call into question the conclusions of the HMR model. I posit and test possible hypothesis to explain them.