Improving Hedonic Estimation with an Inequality Restricted Estimator
提出一种不等式约束贝叶斯估计量，利用子市场先验信息解决特征定价模型中的多重共线性问题，蒙特卡洛实验和交叉验证表明其优于普通最小二乘法。
Economists commonly estimate the value of characteristics not traded in explicit markets by hedonic pricing. Unfortunately, these nonexplicitly traded characteristics often display a lack of independent variation or multicollinearity. Often some prior information on the value of these characteristics is available from submarkets. This paper utilizes this type of prior information to circumvent multicollinearity problems in hedonic pricing models using an inequality restricted Bayesian estimator. The authors perform a Monte Carlo experiment and cross-validation analysis to demonstrate the superiority of inequality restricted Bayesian over ordinary least squares at many margins in a variety of situations typically faced in hedonic estimation. Copyright 1995 by MIT Press.