AI for Economic Research
Large language models, synthetic agents, and AI tools transforming how economists conduct research • 25 papers
LLMs for Economic Research
Using large language models to augment ideation, writing, coding, and data analysis in economics
Generative AI for Economic Research: Use Cases and Implications for Economists
The foundational paper describing dozens of LLM use cases across six domains: ideation/feedback, writing, background research, data analysis, coding, and mathematical derivations. Published in Journal of Economic Literature.
AI Agents for Economic Research
NBER Working Paper 34202. Demystifies autonomous AI agents for multi-step research tasks, including 'vibe coding' with Claude Code and building research assistants with LangGraph.
Large Language Models: An Applied Econometric Framework
Framework for valid LLM use in empirical research. Addresses when and how LLMs can be appropriately used for econometric analysis.
Large Language Models: A Primer for Economists
BIS Quarterly Review. 16-page primer with GitHub code for central banking applications of LLMs.
How to Learn and Teach Economics with Large Language Models
Practical guide for GPT use in economics education from two prominent economists.
Synthetic Agents & Homo Silicus
Using LLMs as computational models of humans for economic experiments and simulations
Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus?
Foundational paper introducing homo silicus—using LLMs as implicit computational models of humans. Replicates classic experiments from Charness & Rabin (2002), Kahneman et al. (1986), finding qualitatively similar results. Published in ACM EC'24.
Automated Social Science: Language Models as Scientist and Subjects
NBER Working Paper 32381. Framework using LLMs to both generate research hypotheses and serve as experimental subjects.
Out of One, Many: Using Language Models to Simulate Human Samples
Introduces 'silicon sampling'—conditioning LLMs with demographic backstories to simulate subpopulation opinions. Develops 'algorithmic fidelity' concept measuring simulation accuracy. Published in Political Analysis.
Synthetic Replacements for Human Survey Data? The Perils of Large Language Models
Critical validation study: 48% of regression coefficients differ significantly from human data; 32% sign flips. Important cautionary findings.
Large Language Models Empowered Agent-based Modeling and Simulation: A Survey and Perspectives
Comprehensive survey covering agent design, environment construction, and personalization for LLM-based simulations. Published in Nature Humanities and Social Sciences Communications.
ML Methods for Economists
Foundational papers on integrating machine learning with economic research methodology
Machine Learning: An Applied Econometric Approach
THE foundational paper explaining ML's role in economics: ML solves prediction problems while economics focuses on parameter estimation. Over 1,500 citations. Published in Journal of Economic Perspectives.
Machine Learning Methods That Economists Should Know About
Comprehensive review including supervised/unsupervised learning, matrix completion, and ML-econometrics methods. Published in Annual Review of Economics.
Deep Learning for Economists
Comprehensive tutorial on classifiers, regression models, generative AI, and embedding models with demo notebooks. Published in Journal of Economic Literature. Companion site: econdl.github.io
Machine Learning as a Tool for Hypothesis Generation
Systematic ML procedure for generating novel hypotheses. Application: defendant mugshots explain ~50% of predictable variation in judge decisions. Published in Quarterly Journal of Economics.
Causal Machine Learning
ML methods for heterogeneous treatment effects and causal inference
Recursive Partitioning for Heterogeneous Causal Effects
Introduces 'honest' causal trees using sample splitting. Published in PNAS.
Estimation and Inference of Heterogeneous Treatment Effects using Random Forests
Develops causal forests with asymptotic theory and valid confidence intervals. Published in JASA.
Generalized Random Forests
Framework extending random forests to solve local moment equations. Published in Annals of Statistics.
Double/Debiased Machine Learning for Treatment and Structural Parameters
Establishes Double ML using Neyman orthogonal scores and cross-fitting. Published in The Econometrics Journal.
Text as Data
NLP and text analysis methods for economic research
Text as Data
THE seminal survey providing introduction to text-as-data methods in economics. Published in Journal of Economic Literature.
Measuring Group Differences in High-Dimensional Choices: Method and Application to Congressional Speech
ML methodology for measuring partisanship in congressional speech (1873-2016). Published in Econometrica.
Text Algorithms in Economics
Overview covering word embeddings, topic models, and transformers for economics research. Published in Annual Review of Economics.
Satellite Imagery & Computer Vision
Using satellite and street-level imagery to measure economic activity
Measuring Economic Growth from Outer Space
Foundational paper using nighttime lights to proxy GDP. For poor-data countries, optimal growth estimate weights lights equally with official statistics. Published in American Economic Review.
Combining Satellite Imagery and Machine Learning to Predict Poverty
Landmark study: CNNs trained on satellite imagery explain 75% of variation in local economic outcomes across 5 African countries. Published in Science.
Using Publicly Available Satellite Imagery and Deep Learning to Understand Economic Well-Being in Africa
Models predict 70% of variation in village wealth across held-out African countries. Published in Nature Communications.
Using Deep Learning and Google Street View to Estimate the Demographic Makeup of Neighborhoods across the United States
Car detection infers income, race, voting patterns from 50M Google Street View images. Published in PNAS.