Project

# Heteroskedasticity Robust Intensive and Extensive Margin Decomposition

Goal: In this paper, we fix the problem of quantile crossing, and therefore the lack of coefficient additivity, by using a original contribution by He (1997) known as restricted regression quantile (RRQ). The method avoids quantile crossing by relating all quantile functions through the median. Importantly, He (1997) shows that the RRQ inherits the finite sample properties of the standard quantile regression estimator developed by Koenker and Basset (1978) as well as the property of consistency. To accommodate the presence of trade flows equal to zero, we apply the 3-step estimator developed by Galvao, Lamarche
and Lima (2013), replacing the standard QR with RRQ (named modified 3-step estimator).

0 new
2
Recommendations
0 new
0
Followers
0 new
13
0 new
53