Testing for Instrument Independence in the Selection Model
2010/01/05
研討會日期 : 2010-01-05
時間 : 15:15
主講人 : Professor Toru Kitagawa
地點 : B110
演講者簡介 : Professor Toru Kitagawa received his Ph.D. in Economics from Brown University in 2009. He is currently lecture of economics at London University College. His research interest is in econometrics, casual inference, and applied econometrics.
演講摘要 : We develop a specification test for the independent instrumentassumption in the sample selection model. We test the emptiness of theidentification region of Manski (2003): the set of outcomedistributions that are compatible with data and the restriction ofstatistical independence between the instrument and outcome. The sizeof the identification region is characterized by a scalar parameter,the integrated envelope, and in particular the identification region isempty if and only if the integrated envelope exceeds one. Since theempty identification region implies a violation of the exclusionrestriction, we obtain a nonparametric specification test for theinstrument exclusion restriction by developing a testing procedure forwhether the integrated envelope exceeds one. This test procedure has anon-pivotal asymptotic distribution and it is well-known that in thiscase the standard nonparametric bootstrap is not valid to obtain thecritical values. We therefore develop a modified bootstrap procedureand show its validity. Monte Carlo simulations examine the finitesample performance of this bootstrap procedure. We use the procedureto test the independence of the instrument used by Blundell et al.(2003).
