Higher-order asymptotic expansions of the least-squares estimation bias in first-order dynamic regression models

详细信息    查看全文

• 作者：Jan F. Kiviet^a ; ¹ ; ^{j.f.kiviet@uva.nl} ; Garry D.A. Phillips^b ; ^{phillipsgd1@cardiff.ac.uk} ; ^{G.D.A.Phillips@ex.ac.uk}
• 关键词：ARX-model ; Asymptotic expansion ; Bias approximation ; Lagged dependent variable ; Monte Carlo simulation
• 刊名：Computational Statistics and Data Analysis
• 出版年：2012
• 出版时间：November, 2012
• 年：2012
• 卷：56
• 期：11
• 页码：3705-3729
• 全文大小：347 K

文摘

An approximation to order is obtained for the bias of the full vector of least-squares estimates obtained from a sample of size in general stable but not necessarily stationary ARX(1) models with normal disturbances. This yields generalizations, allowing for various forms of initial conditions, of Kendall¡¯s and White¡¯s classic results for stationary AR(1) models. The accuracy of various alternative approximations is examined and compared by simulation for particular parameterizations of AR(1) and ARX(1) models. The results show that often the second-order approximation is considerably better than its first-order counterpart and hence opens up perspectives for improved bias correction. However, order approximations are also found to be more vulnerable in the near unit root case than the much simpler order approximations.