統計学輪講(第3回)

統計学輪講(第03回)
日時      2012年04月24日(火)    14時50分~16時30分
場所      経済学部新棟3階第3教室
講演者    市村英彦 (経済)
演題      Estimation of Models with Incomplete Data by Combining Two Data Sets

概要
We study an incomplete data problem in which one data set does not
contain all variables relevant for an empirical analysis or some
variables are reported to be only in some ranges. The proposed method
presumes the existence of another data set which contains a subset of
variables in the original data
set and the missing variables in the original data set. This
assumption combined with a parametric structural assumption and some
joint variation assumption on the variables in the auxiliary data set
are shown to be sufficient to identify the effects of missing
variables as well as those of the non-missing variables in the
parametric structural relationship for a wide variety of problems.
Our framework extends the existing method for
the linear-in-parameters models in the incomplete data literature to
non-linear models. General conditions under which the proposed
estimators for the
identified parameters exhibit consistency and asymptotic normality are
developed.