Interpretation of Graphs that Compare the Distribution Functions of Estimators
展示了如何将估计量累积分布函数曲线下的面积解读为有界绝对误差损失下的风险，并通过重设坐标轴将其转化为有界平方误差损失下的风险，用于比较有限信息最大似然和两阶段最小二乘估计量的精确分布风险。
In this paper I examine graphical comparisons of one-dimensional (or marginal) distribution functions of alternative estimators. It is shown that areas under the c.d.f. (cumulative distribution function) curve can be given a decision-theoretic interpretation as risk under a bounded absolute-error loss function. I also show that by a simple rescaling of the graph's axes, graphical areas are created which can be interpreted as risk under bounded squared-error loss. The bounded loss functions are applied to compare graphically and numerically the risk of exact distributions of the limited-information maximum likelihood and two-stage least-squares estimators in a simultaneous equations model.