[4B3-GS-5] Agents:

[4B3-GS-5-03] Misspecifed Permanent Income Model

〇Kodai Adachi¹, Kei Ikegami², Atsushi Iwasaki¹ (1. The University of Electro-Communications, 2. New York University)

Keywords:Permanent Income Model, Macroeconomics, Robust Control, Riccati Equation, Linear Quadratic Regulator

In economic systems, accurately modeling the dynamics and predicting outcomes pose significant challenges not only for data analysts but also for individual households. Ahn et al.(2014)and Bhandari et al.(2019)argue that individual households who fears the data generated by an unknown distortion of his approximating model dislike uncertainty. This paper addresses these challenges by extending the permanent income model to incorporate endowment misspecification, so that it represents nonlinear and time-dependent feedback of the past information. This misspecification is quantified through a restriction on the Kullback–Leibler divergence between an approximate model and a distorted model. This paper solves problems of worst-case robust performance by transforming a constrained minimax problem into the corresponding multiplier problem using Lagrange duality, and finally applying it to linear quadratic regulator defined by discrete-time Riccati Equation. We confirm that consumption in each period remained constant and savings tends to increase without being influenced by endowment, as the magnitude of misspecification increases.