具有独立贡献的相依回归模型

SEEMINGLY UNRELATED REGRESSION MODEL WITH INDEPENDENT CONTRIBUTION

考虑相依回归（ＳＵＲ）模型ｙｉ＝Ｘｉβｉ＿ｅｉ，ｉ＝１，２，…，ｍ，Ｅｅｉ＝０，ｉ，ｊ＝１；２，…ｍ，其中ｙｉ和ｅｉ是ｎ×１维随机向量，Ｘｉ是ｎ×ｐｉ已知矩阵，βｉ是ｐｉ×１维参数向量，∑＝（σｉｊ）ｍ×ｍ＞０．文中给出了两个概念：独立贡献和简洁估计．主要结果是如下五种叙述等价：（１）ＳＵＲ模型具有独立贡献；（２）βｉ的ＢＬＵＥ是简洁估计；（３）协方差改进估计是ＢＬＵＥＺ（４）βｉ的ＢＬＵＥ具有形式其中，ｊ＝１，２，…，ｍ；（５）ＰｋＮｉＮｊ＝０，ｉ≠ｊ，ｋ，Ｉ，ｊ＝１，２，…。

Consider the seemingly unrelated ragression (SUR) model where yi, ei: n×1 are random vectors, Xi: n×pi is a known matrix, βi: pi×1 is a parameter vector, ∑= (σij): m×m > 0. We give two notions, Independent Contribution and Simple Estimator, in this paper. The main result is that the following five statements are equivalent:(1) SUR model has independent contribution.(2) BLUE of βi is a simple estimator.(3) The covariance improved estimator is BLUE.(4) BLUE of βi has the form