Efficient Estimation with Panel Data: Instrumental Estimators Variables with Bootstrap and Jackknife Method

Authors : Mohamed BIJOU, Said LAARIBYA

Pages : 58-66

View : 16 | Download : 6

Publication Date : 2020-12-31

Article Type : Research Paper

Abstract :Hausman and Taylor insert ignore into journalissuearticles values(1981); have proposed an effective instrumental variable estimator for panel data regression models, where individual effects can be correlated with some of the regressors. Amemiya and MacCurdy insert ignore into journalissuearticles values(1986); and Breusch, Mizon and Schmidt insert ignore into journalissuearticles values(1986); suggested instrumental variables estimators potentially more efficient than the estimator of Hausman and Taylor. In our empirical exercise on educational returns inspired by the pioneering empirical work of Cornwell and Rupert 1988, we propose a new empirical covariance variance matrix revisited by bootstrap and jackknife methods to estimate robust standard deviations. The results show that three variables `smsa, union and black` become insignificant with robust standard deviation estimation via both Bootstrap and Jackknife methods. This modest contribution somehow qualifies the results revealed by BADI H. Baltagi and Sophon Khanti -Akom 1990 and Cornwell and Rupert 1988
Keywords : Instrumental variable, panel data, covariance variance matrix, jakknife