当前位置: X-MOL 学术SIAM J. Financ, Math. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Recover Dynamic Utility from Observable Process: Application to the Economic Equilibrium
SIAM Journal on Financial Mathematics ( IF 1 ) Pub Date : 2021-02-22 , DOI: 10.1137/18m1235843
Nicole El Karoui , Mohamed Mrad

SIAM Journal on Financial Mathematics, Volume 12, Issue 1, Page 189-225, January 2021.
Decision making under uncertainty is generally considered as the selection of an optimal sequence of actions in an uncertain environment. Its calibration raises the “inverse” problem to recover the criterion from the data. A classical example in economy is the theory of “revealed preference” introduced by Samuelson [Economica, 5 (1938), pp. 61--71]. The observable at a given date $t$, $\mathscr X_t(x)$, is an increasing function of a real parameter $x$ (the wealth in economy). The process $\{\mathscr X_t(x)\}$ is called the characteristic process. The objective is to recover a dynamic stochastic utility $\{U(t,z)\}$, “revealed” in the sense where its performance is without bias, more formally when “$\{U(t,\mathscr X_t(x))\}$ is a martingale.” The increasing of $\mathscr X_t(x)$ in $x$ and the concavity of the utility lead to privilege the so-called adjoint process $Y_t(u_z(x)):=U_z(t,\mathscr X_t(x))$ in the linearization of the problem; the one to one correspondence between $U_z(t,z)$ and $Y_t(u_z(x))$ for a given characteristic process is used intensively. We focus on the $(u,\mathscr X,Y)$ triplets, bringing great attention to their initial conditions. We establish a necessary and sufficient condition for the existence of at least one solution to the “revealed" utility problem. An operational version is as follows: “$\{\mathscr X_{t}(x)Y_t(y)\}$ is a supermartingale for any $(x,y)$ and $\{\mathscr X_{t}(x)Y_t(u_{z}(x))\}$ is a martingale for $y=u_z(x)$." Moreover, there is an equivalent intrinsic framework, where in addition the processes “$\{\mathscr X_{t}(x)\}$,$\{Y_t(y)\}$,$\{U(t,z)\}$" are supermartingales. Itô's semimartingale framework is used to illustrate this characterization. The operational version ensures that the revealed utility is a solution of a nonlinear stochastic PDE. Less obvious is its interpretation as a stochastic value function of some optimization problem. The financial markets framework appears as a special case, under stronger assumptions. Then, we revisit the dynamic equilibrium problem as in He and Leland [Rev. Financial Stud., 6 (1993), pp. 593--617], by considering it as a revealed utility problem. We solve the problem in a random environment, by characterizing all the equilibria, in showing that the only possible conjugate dynamic utilities are the mixture of stochastic dual power utilities.


中文翻译:

从可观察过程中恢复动态效用:在经济均衡中的应用

SIAM 金融数学杂志,第 12 卷,第 1 期,第 189-225 页,2021 年 1 月。
不确定性下的决策通常被认为是在不确定环境中选择最佳行动序列。它的校准提出了“逆”问题,以从数据中恢复标准。经济学中的一个经典例子是 Samuelson 提出的“显示偏好”理论 [Economica, 5 (1938), pp. 61--71]。在给定日期 $t$ 的可观察量 $\mathscr X_t(x)$,是实参数 $x$(经济财富)的增函数。过程$\{\mathscr X_t(x)\}$称为特征过程。目标是恢复动态随机效用 $\{U(t,z)\}$,在其性能没有偏差的意义上“揭示”,更正式地为“$\{U(t,\mathscr X_t( x))\}$ 是鞅。” $\mathscr X_t(x)$ 在 $x$ 中的增加和效用的凹陷导致了所谓的伴随过程 $Y_t(u_z(x)):=U_z(t,\mathscr X_t(x) 的特权))$ 在问题的线性化中;密集使用了给定特征过程的 $U_z(t,z)$ 和 $Y_t(u_z(x))$ 之间的一一对应关系。我们专注于 $(u,\mathscr X,Y)$ 三元组,非常关注它们的初始条件。我们为“揭示”效用问题的至少一个解的存在建立充分必要条件。一个操作版本如下:“$\{\mathscr X_{t}(x)Y_t(y)\}$是任何 $(x,y)$ 的超鞅,$\{\mathscr X_{t}(x)Y_t(u_{z}(x))\}$ 是 $y=u_z(x)$ 的鞅.” 此外,还有一个等效的内在框架,其中除了进程“$\{\mathscr X_{t}(x)\}$, s semimartingale 框架用于说明这种表征。操作版本确保揭示的效用是非线性随机偏微分方程的解。不太明显的是,它被解释为某个优化问题的随机值函数。在更强的假设下,金融市场框架似乎是一个特例。然后,我们重新审视 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示的效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。s semimartingale 框架用于说明这种表征。操作版本确保揭示的效用是非线性随机偏微分方程的解。不太明显的是,它被解释为某个优化问题的随机值函数。在更强的假设下,金融市场框架似乎是一个特例。然后,我们重新审视 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。操作版本确保揭示的效用是非线性随机偏微分方程的解。不太明显的是,它被解释为某个优化问题的随机值函数。在更强的假设下,金融市场框架似乎是一个特例。然后,我们重新审视 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示的效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。操作版本确保揭示的效用是非线性随机偏微分方程的解。不太明显的是,它被解释为某个优化问题的随机值函数。在更强的假设下,金融市场框架似乎是一个特例。然后,我们重新审视 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示的效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。我们重新审视了 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示的效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。我们重新审视了 He 和 Leland [Rev. Financial Stud., 6 (1993), pp. 593--617],将其视为揭示的效用问题。我们在随机环境中解决了这个问题,通过表征所有均衡,表明唯一可能的共轭动态效用是随机双功率效用的混合。
更新日期:2021-02-22
down
wechat
bug