Ranking Performance Measures in Multi-Task Agencies
研究了多任务委托代理模型中，如何基于似然比的二阶随机占优条件对绩效评价系统进行排序，并指出在特定条件下该条件可简化为方差协方差矩阵条件。
ABSTRACT: We derive sufficient conditions for ranking performance evaluation systems in multi-task agency models (using both optimal and linear contracts) in terms of a second-order stochastic dominance (SSD) condition on the likelihood ratios. The SSD condition can be replaced by a variance-covariance matrix of likelihood ratios (VCM) condition when the utility function is square-root, the performance measures are normally distributed, and for LEN models. We identify existing results derived under the LEN assumptions that rely on the VCM condition and, thus, also hold for optimal contracts.