TY - JOUR
T1 - Hahn-Hausman test as a specification test
AU - Lee, Yoonseok
AU - Okui, Ryo
N1 - Funding Information:
Previous versions are circulated under the title “A Specification Test for Instrumental Variables Regression with Many Instruments.” The authors acknowledge the valuable comments from Takeshi Amemiya, the Associate Editor, two anonymous referees, Donald Andrews, Mehmet Caner, Juan Carlos Escanciano, Nikolay Gospodinov, Jinyong Hahn, Han Hong, Simon Lee, Yukitoshi Matsushita, Whitney Newey, Taisuke Otsu, Peter Phillips, Mototsugu Shintani and Katsumi Simotsu, along with seminar participants at Columbia, Yale, the Kobe meeting of the Kansai Econometric Society, the 2009 SETA meeting, and the 2009 Far East and South Asia Meeting of the Econometric Society. Lee thanks the Cowles Foundation for Research in Economics at Yale University, where he was a visiting fellow while writing this paper. Okui appreciates the financial support from Japan Society of the Promotion of Science under KAKENHI22730176 and KAKENHI22330067 . The usual disclaimer applies.
PY - 2012/3
Y1 - 2012/3
N2 - This paper develops a modified version of the Sargan [Sargan, J.D., 1958. The estimation of economic relationships using instrumental variables. Econometrica 26 (3), 393415] restrictions, and shows that it is numerically equivalent to the test statistic of Hahn and Hausman [Hahn, J., Hausman, J., 2002. A new specification test for the validity of instrumental variables. Econometrica 70 (1), 163189] up to a sign. The modified Sargan test is constructed such that its asymptotic distribution under the null hypothesis of correct specification is standard normal when the number of instruments increases with the sample size. The equivalence result is useful in understanding what the HahnHausman test detects and its power properties.
AB - This paper develops a modified version of the Sargan [Sargan, J.D., 1958. The estimation of economic relationships using instrumental variables. Econometrica 26 (3), 393415] restrictions, and shows that it is numerically equivalent to the test statistic of Hahn and Hausman [Hahn, J., Hausman, J., 2002. A new specification test for the validity of instrumental variables. Econometrica 70 (1), 163189] up to a sign. The modified Sargan test is constructed such that its asymptotic distribution under the null hypothesis of correct specification is standard normal when the number of instruments increases with the sample size. The equivalence result is useful in understanding what the HahnHausman test detects and its power properties.
KW - HahnHausman test
KW - Many instruments
KW - Overidentifying restrictions test
KW - Sargan test
KW - Specification test
UR - http://www.scopus.com/inward/record.url?scp=84856337259&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84856337259&partnerID=8YFLogxK
U2 - 10.1016/j.jeconom.2011.10.005
DO - 10.1016/j.jeconom.2011.10.005
M3 - Article
AN - SCOPUS:84856337259
SN - 0304-4076
VL - 167
SP - 133
EP - 139
JO - Journal of Econometrics
JF - Journal of Econometrics
IS - 1
ER -