Skip navigation
Please use this identifier to cite or link to this item:
Title: Macroeconomics and Heterogeneous Reality with Machine Learning
Authors: Yang, Yucheng
Advisors: Violante, Gianluca
E, Weinan
Contributors: Applied and Computational Mathematics Department
Keywords: Deep learning
Heterogenenous agent model
Knowledge graph
Machine learning
Monetary policy
Optimal policy
Subjects: Economics
Artificial intelligence
Applied mathematics
Issue Date: 2023
Publisher: Princeton, NJ : Princeton University
Abstract: The study of macroeconomics is characterized by two primary approaches: the structural approach and the statistical approach. This dissertation aims to explore the impact of machine learning on both approaches in three independent chapters. The first two chapters focus on the potential of machine learning in enhancing structural models, specifically heterogeneous agent models, while the third chapter delves into the statistical approach. In the first chapter, coauthored with Jiequn Han and Weinan E, we propose an efficient, reliable, and interpretable global solution method, the Deep learning-based algorithm for Heterogeneous Agent Models (DeepHAM), for solving heterogeneous agent models with aggregate shocks. The state distribution is represented by a set of generalized moments. Deep neural networks are used to approximate the value and policy functions, and the objective is optimized over directly simulated paths. In addition to being an accurate global solver, DeepHAM boasts three additional benefits: it is computationally efficient and free from the curse of dimensionality, it offers an interpretable representation of the distribution over individual states, and it can solve the constrained efficiency problem as easily as the competitive equilibrium, which opens up new possibilities for normative studies in macroeconomics. In the second chapter, I study optimal monetary policy rules in a quantitative heterogeneous agent New Keynesian (HANK) model where inflation has redistributive effects on households through their different (1) consumption baskets, (2) nominal wealth positions, and (3) earnings elasticities to business cycles. Unlike in representative agent models, a utilitarian central bank should adopt an asymmetric monetary policy rule that is accommodative towards inflation and aggressive towards deflation. Specifically, by accommodating stronger demand and higher inflation, the central bank benefits low-income households through nominal debt devaluation and higher earnings growth. The third chapter, coauthored with Yue Pang, Guanhua Huang, and Weinan E, constructs a knowledge graph (KG) that consists not only of linkages between traditional economic variables but also of new alternative big data variables. This KG offers a systematic approach to incorporating human knowledge when dealing with a large number of variables in macroeconomic models. We propose an active learning natural language processing algorithm to extract these variables and linkages from the vast textual data of academic literature and research reports. By utilizing the KG as prior knowledge to select variables for forecasting, we achieve significantly higher accuracy, particularly for long-term forecasts, compared to statistical variable selection methods.
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Applied and Computational Mathematics

Files in This Item:
File Description SizeFormat 
Yang_princeton_0181D_14586.pdf1.7 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.