关键词: 固定效应, 面板数据, 截面相关性检验, 偏误更正

Abstract: Based on the test statistics of Schott (2004), this paper proposes a bias-corrected cross-sectional correlation test in a fixed effects panel data model. Under big n and big T framework, this paper first studies the asymptotic bias between the Schott statistics using the estimated within residuals and standard normal distribution; then it proposes a bias-corrected new test and derives its limiting distribution jointly; finally, it conducts Monte Carlo simulations to examine the finite properties of the proposed test. The new test converges to standard normal distribution jointly, and Monte Carlo simulation results show that the test has a correct size and relatively better power property. This paper improves and extends the asymptotic theory of the bias-corrected Schott test in the fixed effects panel data model, and provides a new non-normality robust test for application.

Key words: Fixed Effects, Panel Data, Cross-sectional Correlation Test, Bias Correction