Affine Incentive Schemes for Stochastic Systems with Dynamic Information
Affine Incentive Schemes for Stochastic Systems with Dynamic Information
复制标题
具有动态信息的随机系统的仿射激励方案
DOI：
--
复制
发表时间：
1982
期刊：
影响因子：
--
通讯作者：
T. Başar
中科院分区：
文献类型：
--
作者：
T. Başar
在本文中，我们研究了在一般希尔伯特空间设置中具有分层决策结构的二主体随机决策问题中最优激励方案的推导。假设位于层次结构顶部的代理可以访问其他代理的决策变量的值以及一些公共和私有信息，并且第二个代理的损失函数被认为是严格凸的。在这个设置中，结果表明，在一些相当温和的结构限制下，存在第一代理的最优激励策略，该策略在动态信息中是仿射的，在静态（公共和私有）信息中通常是非线性的。还讨论了某些特殊情况并解决了数值示例。
In this paper we study the derivation of optimal incentive schemes in two-agent stochastic decision problems with a hierarchical decision structure, in a general Hilbert space setting. The agent at the top of the hierarchy is assumed to have access to the value of other agent's decision variable as well as to some common and private information, and the second agent's loss function is taken to be strictly convex. In this set-up, it is shown that there exists, under some fairly mild structural restrictions, an optimal incentive policy for the first agent, which is affine in the dynamic information and generally nonlinear in the static (common and private) information. Certain special cases are also discussed and a numerical example is solved.