Jacobian-free Efficient Pseudo-Likelihood (EPL) algorithm
提出一种无需计算均衡约束雅可比矩阵的简单方法，用于估计动态离散博弈中的高效伪似然估计量，利用数值导数和Krylov方法，数值实验表现良好。
This study proposes a simple procedure for computing Efficient Pseudo Likelihood (EPL) estimator (Dearing and Blevins, 2024) for estimating dynamic discrete games, without computing Jacobians of equilibrium constraints. It utilizes the ideas of numerical derivatives and the Krylov method. • Propose a simple method to compute EPL estimator without computing Jacobians. • Computing Jacobians of equilibrium constraints is prone to coding mistakes. • The proposed method utilizes the ideas of numerical derivatives and the Krylov method. • Show good performance by numerical experiments.