awesome-ai-for-economists

Awesome AI for Economists [Awesome]

[PRs Welcome] [Last Commit]

A curated list of AI tools, agent skills, libraries, and resources for economics research, teaching, and policy analysis.

Scope: AI tools and resources with specific value for economists, plus research on the economics of AI. General-purpose AI assistants and coding agents are in the Appendix.

Highlights: Projects from the OpenEcon team, which maintains this list. Entries in bold below are the team's own.

• AI-research-setup - Step-by-step guide to setting up Claude Code and Codex for economics research on macOS and Windows. [GitHub stars]
• OpenEcon Data - MCP server and web app for plain-English queries over 330,000+ indicators from FRED, World Bank, IMF, Eurostat, and more. [GitHub stars]
• Stata-MCP - Runs Stata from VS Code, Cursor, Claude Code, Codex, and Copilot with real-time output, a data viewer, and graph display. [GitHub stars]
• Econ Writing Skill - Economics paper-writing skill synthesizing 50+ guides by Cochrane, McCloskey, Shapiro, Head, and others. [GitHub stars]
• Econ Paper Review Skill - Writes referee reports on economics papers with verified comments, editing notes, and a step-by-step revision plan; free for non-commercial use. [GitHub stars]
• Econ Slides Skill - Turns a paper into a Beamer talk with a timed speaker script, including job-market and discussant decks. [GitHub stars]
• overleaf-sync-now - Agent skill and CLI that pull recent Overleaf web edits into a Dropbox-synced copy before AI agents edit it. [GitHub stars]
• Econ Auto Research - Vision-stage plan (no code yet) for an economist-directed AI pipeline from research question to working paper, with adversarial self-refereeing. [GitHub stars]

Contents

• MCP Servers for Economic Data
• Agent Skills for Economists
• Coding Tools for Economists
• AI Agents for Research
• Causal Inference and Econometrics
• Simulation, Forecasting and Macro Modeling
• Literature Review and Research Discovery
• Economic Data and Analysis
• Academic Writing and LaTeX
• Document Processing and OCR
• NLP and Text as Data
• Policy, Labor and Alternative Data
• Finance-Specific AI
• Data Collection Tools
• Benchmarks and Evaluation
• Papers and Books
• Courses, Conferences and Community
• Appendix: General-Purpose AI

MCP Servers for Economic Data

Model Context Protocol (MCP) servers let AI assistants query economic databases directly. Connect Claude, Cursor, or other MCP-compatible tools to live data from FRED, World Bank, IMF, and more.

• Alpha Vantage MCP - Official server for stocks, forex, commodities, and economic indicators.
• Data Commons MCP Server - Official Data Commons server, hosted with a free key or run locally, serving Census, World Bank, BLS, and WHO statistics. [GitHub stars]
• ECB MCP Server - Query European Central Bank exchange rates, policy rates, yield curves, and euro-area monetary statistics. [GitHub stars]
• Equibles MCP - Self-hosted, open-source server for SEC filings, XBRL financials, 13F holdings, insider trades, short interest, CFTC positioning, and federal contracts. [GitHub stars]
• Eurostat MCP Server - Search, inspect, and query Eurostat datasets, including Comext detailed trade, PRODCOM, and NUTS regional data. [GitHub stars]
• FRED MCP Connector - St. Louis Fed's official connector letting AI assistants search, fetch, and chart FRED series; requires only a free account.
• FRED MCP Server - Community open-source FRED server run with your own free API key; a self-hostable alternative to the official connector. [GitHub stars]
• IMF Data MCP Server - WEO forecasts, balance of payments, and more from the IMF. [GitHub stars]
• Invokely - Hosted server for official U.S., EU, UK, Canadian, and Australian statistics, down to U.S. counties and Spanish municipalities.
• MoSPI eSankhyiki MCP - Official Indian statistics ministry server for PLFS, CPI, national accounts, NSS surveys, and more. [GitHub stars]
• Nasdaq Data Link MCP - Access Nasdaq/Quandl financial and economic datasets. [GitHub stars]
• OECD MCP Server - Access 5,000+ OECD statistical datasets via SDMX API with prompt templates. [GitHub stars]
• OpenEcon Data - MCP server and web app for plain-English queries over 330,000+ indicators from FRED, World Bank, IMF, Eurostat, and more. [GitHub stars]
• SEC EDGAR MCP - Connects AI assistants to U.S. SEC EDGAR filings, financial statements, and insider-trading data. [GitHub stars]
• StatCite - World Bank, IMF, and ECB statistics where every value carries a full citation, plus verification of claimed figures. [GitHub stars]
• U.S. Census Bureau Data API MCP - Official Census Bureau server connecting AI assistants to demographic, economic, and population statistics. [GitHub stars]
• UN Comtrade MCP Server - Access UN Comtrade international trade statistics with country and HS commodity code lookup. [GitHub stars]
• World Bank Data360 MCP - Official World Bank server with a public hosted endpoint (no key needed) for searching, fetching, and charting Data360 indicators. [GitHub stars]
• World Bank MCP Server - Community server for the legacy World Bank Open Data API (the official one is World Bank Data360 MCP above). [GitHub stars]

Browse the MCP Registry to discover more servers.

Agent Skills for Economists

Agent Skills are folders that an AI agent loads when a task needs them: a SKILL.md file of instructions, plus optional scripts and reference files. The open format works in Claude, Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and other agents.

• AI Research Feedback - Claes Bäckman's (SAFE Frankfurt) skills for multi-agent referee reports plus pre-analysis-plan, grant, and replication-code reviews. [GitHub stars]
• Awesome Journal Skills - Per-journal skill packs for AER, QJE, the AEJs, and hundreds of other venues, from topic choice to referee response. [GitHub stars]
• awesome-econ-ai-stuff - Antonio Mele's (LSE) SKILL.md collection for Stata, R, and Python econometrics, data fetching, LaTeX tables, literature review, and modeling. [GitHub stars]
• Claude Code Academic Workflow - Pedro Sant'Anna's (Emory) ready-to-fork template with 60+ skills, multi-agent review, quality gates, and replication protocols. [GitHub stars]
• dynare-copilot - Skill turning macro model ideas into validated Dynare .mod files, drawing on 149 Macroeconomic Model Data Base models. [GitHub stars]
• Econ Paper Review Skill - Writes referee reports on economics papers with verified comments, editing notes, and a step-by-step revision plan; free for non-commercial use. [GitHub stars]
• Econ Slides Skill - Turns a paper into a Beamer talk with a timed speaker script, including job-market and discussant decks. [GitHub stars]
• Econ Writing Skill - Economics paper-writing skill synthesizing 50+ guides by Cochrane, McCloskey, Shapiro, Head, and others. [GitHub stars]
• MixtapeTools - Scott Cunningham's (Baylor) skills for independent code audits, cross-language replication, bibliography checks, and Beamer decks. [GitHub stars]
• Stata Skill - Dylan Moore's (Hawaii) Stata reference skill covering econometrics, causal inference, Mata, and 20+ community packages such as reghdfe. [GitHub stars]
• World Bank Reproducibility Skill - Official World Bank skill that audits and assembles replication packages, starting with a data-privacy check. [GitHub stars]

More skills: Academic Research Skills (research-to-publication pipeline, CC BY-NC) | Anthropic Skills (Word, Excel, PowerPoint, PDF) | Posit Skills (Quarto, tidyverse, Shiny) | Scientific Agent Skills (statistics, FRED and World Bank lookups). Browse skills.sh for more. Skills can run code on your computer, so read a third-party skill's SKILL.md and scripts before installing it.

Coding Tools for Economists

AI assistants, agent connectors, and setup guides for Stata, R, MATLAB, Python, and notebooks. General-purpose coding agents are in the Appendix.

• AI-research-setup - Step-by-step guide to setting up Claude Code and Codex for economics research on macOS and Windows. [GitHub stars]
• Jupyter AI - Official JupyterLab extension bringing Claude, Codex, Gemini, and other agents into notebooks through a chat panel. [GitHub stars]
• Jupyter MCP Server - MCP server and Claude Code plugin letting AI agents edit and run live Jupyter notebooks. [GitHub stars]
• marimo - Reactive Python notebook stored as pure .py files with built-in AI assistant. [GitHub stars]
• MATLAB MCP Server - Official MathWorks server letting Claude Code, Codex, and Copilot run, lint, and test code in your local MATLAB. [GitHub stars]
• MCP-for-Stata - MCP server and CLI, with bundled Stata skills, letting Claude Code, Codex, and other agents run local Stata. [GitHub stars]
• mcptools - Posit's R package letting Claude Code, Copilot, and other MCP clients run code in your live R sessions. [GitHub stars]
• Posit Assistant - Posit's data-analysis agent for RStudio, Positron, and the terminal, working in live R and Python sessions with major model providers.
• Positron - Free Posit data science IDE for R and Python, with the Posit Assistant agent built in. [GitHub stars]
• Stata-MCP - Runs Stata from VS Code, Cursor, Claude Code, Codex, and Copilot with real-time output, a data viewer, and graph display. [GitHub stars]

AI Agents for Research

AI Agent Frameworks

Libraries for building your own LLM pipelines and agents, from data collection to analysis and reporting.

• Agno - Open-source Python framework with memory, knowledge, and 100+ toolkits including MCP support. [GitHub stars]
• chatlas - Posit's Python interface for LLMs with streaming, tool calling, and structured output. [GitHub stars]
• CrewAI - Open-source Python framework for orchestrating role-based, collaborative multi-agent systems. [GitHub stars]
• DSPy - Stanford NLP framework for building LLM pipelines from declarative Python modules, with optimizers that tune prompts and weights. [GitHub stars]
• ellmer - Tidyverse R package for calling LLM APIs with tool calling and structured data extraction. [GitHub stars]
• Google ADK - Google's modular multi-agent framework with MCP and A2A protocol support. [GitHub stars]
• LangGraph - LangChain's low-level framework for stateful, durable agent workflows built as graphs. [GitHub stars]
• Microsoft Agent Framework - Microsoft's successor to AutoGen and Semantic Kernel for multi-agent workflows in Python and .NET. [GitHub stars]
• OpenAI Agents SDK - Lightweight Python framework for multi-agent workflows with handoffs, guardrails, and tracing. [GitHub stars]
• Pydantic AI - Type-safe Python agent framework with validated structured outputs across all major model providers. [GitHub stars]
• smolagents - Hugging Face's minimalist agent library where agents write actions as Python code. [GitHub stars]

Autonomous Research Agents

Systems that turn a research question into a cited report, an analysis, or a draft paper.

• DeerFlow - ByteDance's research SuperAgent for long-horizon tasks with sandboxed code execution. [GitHub stars]
• Econ Auto Research - Vision-stage plan (no code yet) for an economist-directed AI pipeline from research question to working paper, with adversarial self-refereeing. [GitHub stars]
• GPT Researcher - Autonomous deep research agent producing cited reports from web sources. [GitHub stars]
• Kosmos - Edison Scientific's AI scientist that reads literature, analyzes datasets, and generates hypotheses in long autonomous runs.
• Project APE - University of Zurich experiment in which AI agents wrote 1,000 open policy-evaluation papers, ranked by LLM judges against AER/AEJ papers. [GitHub stars]
• The AI Scientist - Sakana AI system that generates ML research ideas, runs experiments, and writes and reviews LaTeX papers (Nature 2026; template-free v2). [GitHub stars]

Causal Inference and Econometrics

Core Libraries

• awesome-causal-inference - Curated list of causal inference resources, courses, and tools. [GitHub stars]
• causal-learn - Causal discovery algorithms (PC, FCI, GES) from the PyWhy ecosystem. [GitHub stars]
• CausalMatch - ByteDance's Python package for propensity score matching and coarsened exact matching, with machine-learning propensity models. [GitHub stars]
• CausalML - Uber's uplift modeling and causal inference with machine learning. [GitHub stars]
• CausalPy - Bayesian causal inference for quasi-experiments (DiD, synthetic control, RDD) powered by PyMC, with bundled agent skills. [GitHub stars]
• ddml - Double/debiased ML with stacked learners for Stata (Ahrens, Hansen, Schaffer, Wiemann), with a sister R package. [GitHub stars]
• DoubleML - Double/debiased ML (Chernozhukov et al., 2018) in Python and R; the Python package adds DiD, RDD, and panel fixed-effects models. [GitHub stars]
• DoWhy - End-to-end causal inference: model, identify, estimate, refute. [GitHub stars]
• EconML - Microsoft/PyWhy library for heterogeneous treatment effects via Double ML and causal forests. [GitHub stars]
• grf - Generalized Random Forests for heterogeneous treatment effects (Athey, Tibshirani, Wager). [GitHub stars]
• marginaleffects - Predictions, slopes, marginal means, and contrasts for 100+ model classes in R and Python, with an official agent skill. [GitHub stars]
• policytree - R package for learning shallow-tree treatment-assignment rules via doubly robust empirical welfare maximization (Athey and Wager). [GitHub stars]
• pyfixest - Fast high-dimensional fixed-effects regression in Python with IV, Poisson, and modern difference-in-differences estimators. [GitHub stars]
• xtdml - Stata module for double machine learning in static panel models with fixed effects. [GitHub stars]

Difference-in-Differences and Synthetic Control

• CImpact - Causal impact for time series with frequentist and Bayesian methods. [GitHub stars]
• contdid - DiD with continuous treatment, by Brantly Callaway. [GitHub stars]
• diff-diff - Python DiD library with Callaway-Sant'Anna, imputation, synthetic DiD, continuous treatment, and Honest DiD, benchmarked against R. [GitHub stars]
• differences - Python implementation of Callaway and Sant'Anna staggered difference-in-differences with multiple periods and triple differences. [GitHub stars]
• mlsynth - Python library bundling dozens of synthetic control estimators under one API, cross-checked against original implementations. [GitHub stars]
• moderndid - Modern DiD estimators (staggered, continuous, triple, Honest DiD) under one API, scaling to GPUs and Spark or Dask clusters. [GitHub stars]
• pysyncon - Python synthetic control suite with classic, augmented, penalized, and robust estimators plus placebo inference. [GitHub stars]
• scpi - Synthetic control with prediction intervals for single, multiple, and staggered-adoption units, in R, Python, and Stata. [GitHub stars]

Inference with AI-Generated Data

Valid estimates and standard errors when regression variables come from LLM or ML predictions.

• dsl - Egami, Hinck, Stewart, and Wei's R package for valid regression inference using LLM-annotated variables plus a small expert-coded sample. [GitHub stars]
• ppi_py - Prediction-powered inference (Science 2023): valid confidence intervals combining a small labeled sample with many ML or LLM predictions. [GitHub stars]
• ValidMLInference - Python package for bias-corrected inference with AI/ML-generated covariates (Battaglia, Christensen, Hansen, Sacher); the R port is MLBC on CRAN. [GitHub stars]

LLM and Foundation-Model Causal Tools

• Causal AI Scientist - LLM agent that answers causal questions by choosing and running DiD, IV, RDD, or adjustment estimators, with assumption checks. [GitHub stars]
• Causal-Copilot - LLM agent automating end-to-end causal discovery, inference, and reporting from natural-language queries. [GitHub stars]
• CausalFM - Foundation models for causal inference; pre-trained models replace per-problem estimation (ICLR 2026). [GitHub stars]
• CausalPFN - Prior-fitted transformer for amortized, in-context ATE and CATE estimation without per-dataset retraining. [GitHub stars]

Simulation, Forecasting and Macro Modeling

Agent-Based Models and LLM Simulation

• AgentSociety - Tsinghua's LLM-agent platform for social-science experiments (AgentSociety 2, 2026), plus the legacy city simulator with economic modules. [GitHub stars]
• BeforeIT.jl - Bank of Italy's agent-based macroeconomic model in Julia. [GitHub stars]
• Concordia - Google DeepMind library for generative agent-based modeling of social, economic, and market interactions. [GitHub stars]
• EconAgent - LLM agents with personality traits simulating macroeconomic activities (ACL 2024). [GitHub stars]
• LLM-ABM Survey - Tsinghua's paper list accompanying its LLM agent-based modeling survey (Humanities and Social Sciences Communications, 2024). [GitHub stars]
• LLM-Economist - Mechanism design with 3–1,000+ LLM agents for optimal taxation research. [GitHub stars]
• Magentic Marketplace - Microsoft Research environment simulating two-sided markets of LLM buyer and seller agents to study welfare and manipulation. [GitHub stars]
• MarS - Microsoft engine simulating financial markets via a generative Large Market Model of order flow. [GitHub stars]
• Mesa - Python agent-based modeling framework; its Mesa-LLM extension adds language-model decision-making to agents. [GitHub stars]

Forecasting and Nowcasting

• Chronos - Amazon's pretrained forecasting models; Chronos-2 adds zero-shot multivariate and covariate-informed forecasting. [GitHub stars]
• Durbyn.jl - Julia port of R's forecast package. [GitHub stars]
• Moirai (uni2ts) - Salesforce's library for Moirai time-series foundation models (latest Moirai 2.0), with pre-training, fine-tuning, and evaluation code. [GitHub stars]
• neuralforecast - Neural forecasting models (N-BEATS, NHITS, Transformers) for economic time series. [GitHub stars]
• Project Spectrum - BIS/ECB/Bundesbank project using GenAI embeddings to classify 34M products for inflation nowcasting (2026 report).
• statsforecast - Fast statistical forecasting for econometric time series. [GitHub stars]
• Sundial - Generative time-series foundation models giving probabilistic forecasts via flow matching without discrete tokenization. [GitHub stars]
• tabpfn-time-series - Zero-shot time-series forecasting that repurposes the TabPFN-v2 tabular foundation model, no training needed. [GitHub stars]
• TimeGPT - Nixtla's closed-source, API-based time-series foundation model with a free trial; TimeGPT-2.1 adds multivariate forecasting. [GitHub stars]
• TimesFM - Google's time-series foundation model; TimesFM 3.0 (2026) adds zero-shot multivariate forecasting with covariates (self-hosted weights are non-commercial). [GitHub stars]
• TiRex-2 - NX-AI's zero-shot forecaster extending TiRex to multivariate series with past and future-known covariates (NeurIPS 2026, Apache-2.0). [GitHub stars]
• Toto - Datadog's Apache-2.0 time-series foundation models; Toto 2.0 (2026) spans 4M to 2.5B parameters. [GitHub stars]

Game Theory and Mechanism Design

• Axelrod - Iterated Prisoner's Dilemma research with 200+ strategies and tournament simulations. [GitHub stars]
• Nashpy - Two-player game solver with support enumeration, vertex enumeration, and Lemke-Howson. [GitHub stars]
• OpenSpiel - Google DeepMind library of games with equilibrium-finding and multi-agent learning algorithms such as CFR and PSRO. [GitHub stars]
• pygambit - Game theory computation for extensive and strategic form games with a Python API. [GitHub stars]

DSGE and Structural Models

• Deep Learning for Solving and Estimating Dynamic Models in Economics and Finance - Simon Scheidegger's (Lausanne) open course with runnable notebooks applying deep equilibrium nets and PINNs to DSGE, OLG, and heterogeneous-agent models. [GitHub stars]
• econpizza - JAX-based solver for fully nonlinear heterogeneous-agent (HANK) macroeconomic models via automatic differentiation. [GitHub stars]
• gEconpy - DSGE toolkit in Python. [GitHub stars]
• MacroModelling.jl - Julia DSGE toolkit. [GitHub stars]
• OpenSourceEconomics - JAX-compatible DC-EGM, Kalman filters, and econ-project-templates.
• optimagic - Numerical optimization for economists (formerly estimagic). [GitHub stars]
• Structural Reinforcement Learning - JAX tutorials by Yang, Wang, Schaab, and Moll solving heterogeneous-agent models with aggregate risk via policy gradients. [GitHub stars]

Literature Review and Research Discovery

• alphaXiv - Line-by-line discussion on arXiv papers with AI-powered analysis.
• ASReview - Active learning for systematic reviews, reducing screening time by up to 95% (Nature Machine Intelligence). [GitHub stars]
• Asta - Ai2's agentic-science assistant for paper finding, cited literature reports, and data analysis; successor to the OpenScholar demo. [GitHub stars]
• Connected Papers - Visual graphs of related papers from a seed paper.
• Consensus - AI academic search with quote-linked citations, a multi-step research agent, and API and MCP access.
• Elicit - Research agent and PRISMA 2020-aligned systematic reviews over 138M papers, with data extraction and an API and MCP server.
• Google Scholar Labs - Free AI mode in Google Scholar that breaks down research questions and explains how each paper addresses them; sign-in required.
• Inciteful - Free citation network analysis using graph algorithms to find key papers and connections.
• Litmaps - Dynamic multi-seed citation mapping with monitored searches and daily alerts for new papers.
• Meta-Mar - Free meta-analysis platform where AI reads PDFs and extracts quantitative outcomes.
• MindMap AI - Turns research papers, PDFs, and notes into structured visual mind maps for literature synthesis.
• OpenAlex - Fully open catalog of scholarly works with a free REST API and semantic search.
• PaperQA2 - RAG system for question answering and summarization over scientific literature. [GitHub stars]
• Rayyan - AI-assisted screening and data extraction for systematic reviews, with team collaboration, PRISMA flow diagrams, and a free basic tier.
• ResearchRabbit - Citation network and semantic similarity discovery rebuilt with Litmaps.
• SciSpace - Deep review with iterative search and Zotero integration across 280M papers.
• Scite - Smart Citations classifying 1.6B+ citations as supporting or contrasting, plus an AI assistant, Reference Check, and MCP access.
• Semantic Scholar - AI-powered academic search with citation analysis, TLDR summaries, and free API across 220M+ papers.
• Suppr - AI literature search, document translation, and deep-research reports over PubMed and OpenAlex, with a Zotero plugin.
• Undermind - Autonomously reads hundreds of papers and produces structured reports with timelines and categories.
• Zotero MCP - Connects your Zotero library to Claude, ChatGPT, Codex, and Cursor to search, read, annotate, and cite papers. [GitHub stars]
• Zotero PapersGPT - Zotero AI plugin supporting ChatGPT, Claude, Gemini, and more for PDF chat and summaries. [GitHub stars]

Economic Data and Analysis

AI Data Analysis Platforms

• Julius AI - Upload datasets and ask questions in natural language; returns charts, regressions, and reports.
• Microsoft Data Formulator - Describe charts in English and get publication-quality output with Python code. [GitHub stars]

Economic Data Sources

• FRED API v2 - Bulk retrieval of all series in any release across 800,000+ time series.
• FXMacroData - API and MCP server for macroeconomic releases, calendars, FX, commodities, and bond yields, queryable in plain English.
• Global Macro Database - Macroeconomic dataset, free for academic use, covering 239 countries from 1086 to 2025 with forecasts to 2031, plus MCP access. [GitHub stars]
• UN System Data Commons - Official UN statistics portal (successor to UNdata) built on Data Commons, with natural-language search and MCP access.

Academic Writing and LaTeX

Agent skills for paper writing, refereeing, and slides are listed under Agent Skills for Economists above.

• bibverify - Verifies BibTeX entries against Crossref, OpenAlex, Semantic Scholar, and arXiv via CLI, MCP server, or agent skill. [GitHub stars]
• OpenAI Prism - Free LaTeX workspace with citation management, Zotero sync, sketch-to-equation, and real-time collaboration.
• Overleaf AI - LaTeX error fixing, table and equation generation, and writing feedback, included in all Overleaf plans with a basic free allowance.
• overleaf-sync-now - Agent skill and CLI that pull recent Overleaf web edits into a Dropbox-synced copy before AI agents edit it. [GitHub stars]
• Paperpal - Academic writing assistant on Overleaf, Google Docs, and Chrome.
• Refine.ink - AI technical verification flagging errors in reasoning, calculations, and references; used by AEA and Econometric Society journals.
• Thesify - AI reviewer evaluating argumentation, methodology, and rigor.
• Typst - Markup-based typesetting system and LaTeX alternative with simpler syntax, incremental compilation, and a collaborative web editor. [GitHub stars]
• Underleaf - Chrome extension for image-to-LaTeX and smart citation generation.

Document Processing and OCR

• Chandra - Datalab's open-weight OCR model converting images and PDFs to Markdown or JSON, handling handwriting, forms, and complex tables. [GitHub stars]
• CHURRO - Stanford's open-weight OCR model for handwritten and printed historical documents across 46 language clusters (EMNLP 2025). [GitHub stars]
• DeepSeek-OCR - Open-weight vision model converting document images to Markdown with formula and table recognition, succeeded by DeepSeek-OCR 2 in 2026. [GitHub stars]
• Docling - Open-source parser, created at IBM, converting PDFs, Office files, scans, and XBRL financial reports into Markdown or JSON. [GitHub stars]
• eScriptorium - Open-source platform for segmenting, transcribing, and OCR-ing historical and handwritten manuscripts.
• Marker - Converts PDFs, Office files, and images to Markdown or JSON with tables, equations, and forms; LLM mode merges cross-page tables. [GitHub stars]
• Mathpix - Screenshot-to-LaTeX OCR for equations, tables, and diagrams with PDF-to-Overleaf conversion.
• MinerU - Converts PDFs, scans, Office files, and e-books into LLM-ready Markdown or JSON with tiered local parsing (v4, 2026). [GitHub stars]
• Mistral OCR - Pay-per-page document OCR API returning text with bounding boxes, block types, and confidence scores in 170 languages.
• olmOCR - Ai2's fully open OCR toolkit and model (olmOCR 2) for tables, equations, and handwriting, plus the olmOCR-Bench benchmark. [GitHub stars]
• PaddleOCR-VL - Compact 0.9B vision-language OCR (v1.6, 2026) extracting layout, formulas, tables, and charts, including from historical documents. [GitHub stars]
• Pix2Text - Open-source tool recognizing layouts, tables, and math formulas in images, exporting LaTeX and Markdown. [GitHub stars]
• Surya - Open-source OCR toolkit for 90+ languages with layout, reading-order, and table recognition. [GitHub stars]
• Transkribus - Cooperative-run platform transcribing handwritten and printed historical documents with 300+ community AI models; 50 free credits monthly.

NLP and Text as Data

• CentralBankRoBERTa - Fine-tuned RoBERTa for sentiment classification in central bank communications. [GitHub stars]
• ChronoGPT - Open language models with yearly checkpoints (1999–2024), each trained only on text available through that year to avoid lookahead bias.
• EconBERTa - World Bank DeBERTa trained on 9.4B tokens from 1.5M economics papers with NER. [GitHub stars]
• FinBERT - Fine-tuned BERT for financial sentiment, widely cited in economics and finance research.
• GABRIEL - OpenAI toolkit turning text, images, or audio into quantitative measures for social scientists. [GitHub stars]
• LinkTransformer - Arora and Dell's (Harvard) Python package for transformer-based record linkage, deduplication, and classification of messy text in dataframes. [GitHub stars]
• OccCANINE - Model converting free-text occupational descriptions in 13 languages into HISCO codes with about 96% accuracy. [GitHub stars]
• WorldCentralBanks - Annotated corpus and per-bank models classifying monetary policy stance in central bank communications worldwide. [GitHub stars]

Policy, Labor and Alternative Data

Policy and Evidence Synthesis

• EUROMOD - EU tax-benefit microsimulation covering all 27 member states (open-source, EUPL-1.2).
• ImpactAI - World Bank DIME tool synthesizing RCT evidence with effect sizes across interventions.
• OpenFisca - Open-source microsimulation engine for computing taxes and benefits on population data. [GitHub stars]
• PolicyEngine - Open-source tax-benefit microsimulation for the U.S., UK, and Canada with a free web app, API, and Claude Code plugin. [GitHub stars]

Labor Market Data

• Lightcast - Global job-postings and skills data across 165 countries, plus the free Open Skills taxonomy of 34,000+ skills.
• LinkedIn Economic Graph - Research and interactive data on hiring, skills, and AI's effect on work, including the monthly LinkedIn Workforce Report.
• Revelio Labs - Workforce data with free monthly RPLS U.S. labor statistics, a monthly AI Labor Market Tracker, and academic access via WRDS.

AI Adoption and Exposure Trackers

• AI Economic Indicators - Stanford Digital Economy Lab's monthly dashboards on AI's effects on employment, adoption, and consumer surplus.
• Anthropic Economic Index - CC BY dataset and recurring reports on Claude use by O*NET task, occupation, and region, queryable through a Claude connector.
• Census BTOS - Census Bureau's biweekly business survey tracking current and expected AI use by sector, state, and firm size.
• GenAI Adoption Tracker - Bick, Blandin, and Deming's nationally representative survey of U.S. generative AI use at work and home.
• GPTs are GPTs - Task- and occupation-level LLM exposure scores for O*NET from Eloundou et al. (Science, 2024). [GitHub stars]
• Indeed Hiring Lab AI Tracker - Daily share of Indeed job postings mentioning AI in nine countries since 2019, under CC BY 4.0. [GitHub stars]
• Microsoft Global AI Diffusion Report - Quarterly telemetry-based estimates of generative AI use among working-age adults in 147 economies.
• OpenAI Signals - OpenAI's economic research hub with quarterly downloadable data on ChatGPT adoption and use by country, age, and task.
• Ramp AI Index - Monthly share of U.S. businesses on Ramp paying for AI tools, by industry, size, and provider; also on FRED.
• Yale Budget Lab AI Tracker - Regularly updated CPS-based analysis, with data, of whether AI is shifting U.S. occupations and unemployment.

Spatial, Satellite and Alternative Data

• AlphaEarth Satellite Embeddings - Google DeepMind's annual global 10-meter, 64-dimension satellite embeddings since 2017 on Earth Engine.
• Clay Foundation Model - Open foundation model for Earth observation generating embeddings from global multi-sensor satellite imagery. [GitHub stars]
• MOSAIKS - Precomputed satellite-imagery features and API for predicting local socioeconomic outcomes, including HDI for 61,530 municipalities (Nature Communications, 2026).
• Overture Maps - Open global geospatial data of places, buildings, addresses, and transportation, released monthly. [GitHub stars]

Finance-Specific AI

• ai-hedge-fund - Educational multi-agent system mimicking famous investors to research stocks and propose trading decisions. [GitHub stars]
• awesome-quant - Curated quantitative finance resources. [GitHub stars]
• Dexter - Autonomous financial research agent with task planning and real-time data. [GitHub stars]
• EdgarTools - Python library turning SEC EDGAR filings and XBRL financials into DataFrames and LLM-ready text, with a built-in MCP server. [GitHub stars]
• Fin-R1 - Financial reasoning LLM by Shanghai University of Finance and Economics. [GitHub stars]
• FinGPT - Open-source financial LLMs for sentiment, forecasting, and reports. [GitHub stars]
• FinRobot - AI agent platform for financial analysis combining LLMs, RL, and quantitative methods. [GitHub stars]
• OpenBB - Open-source data platform connecting financial and economic data to Python and AI agents via MCP, stewarded by the non-profit OpenBQ. [GitHub stars]
• TradingAgents - Multi-agent LLM framework simulating analyst, researcher, trader, and risk-manager roles for trading decisions. [GitHub stars]

Data Collection Tools

Web Scraping

• Crawl4AI - Open-source web crawler producing LLM-ready markdown for RAG and data pipelines. [GitHub stars]
• Crawlee - Open-source crawling and scraping library that extracts web data for AI, LLM, and RAG pipelines. [GitHub stars]
• Firecrawl - Open-source API turning websites into LLM-ready Markdown or structured JSON, with an AI agent for schema-based extraction. [GitHub stars]
• ScrapeGraphAI - LLM-powered scraping that adapts to website changes automatically. [GitHub stars]

Survey and Qualitative Research

• ATLAS.ti - Qualitative data analysis software with AI coding, document chat, and transcription, plus a free trial and student licenses.
• Conversations at Scale - Geiecke and Jaravel's (LSE) platform for AI-led qualitative interviews, free for non-commercial research. [GitHub stars]
• EDSL - Open-source Python library from Expected Parrot (co-founded by economist John Horton) for surveys and experiments with LLM agents. [GitHub stars]
• QualCoder - Open-source qualitative coding app for text, images, audio, and video, with an AI agent that searches data and proposes codes. [GitHub stars]
• Twin-2K-500 - Columbia dataset of 2,058 U.S. respondents answering about 500 survey and behavioral-economics questions, for testing LLM digital twins.

Warning: Nature reports AI chatbots infiltrating online surveys. Implement detection for MTurk/Prolific studies.

Synthetic Data

• MOSTLY AI - Open-source SDK generating synthetic tabular, multi-table, and time-series data locally, with optional differential privacy. [GitHub stars]
• REaLTabFormer - World Bank library using transformers to generate synthetic single-table and relational data. [GitHub stars]
• SDV - Open-source synthetic data generation with CTGAN, CopulaGAN, and GaussianCopula models. [GitHub stars]
• Synthcity - Generates and benchmarks synthetic tabular, time-series, and survival data with privacy and fairness metrics. [GitHub stars]

Benchmarks and Evaluation

Benchmarks and leaderboards for judging how well AI handles economic reasoning, forecasting, finance, replication, and real work tasks.

• ForecastBench - Forecasting Research Institute's live benchmark of LLM vs. superforecaster accuracy, including FRED-based questions (ICLR 2025). [GitHub stars]
• GDPval - OpenAI benchmark of real work tasks in 44 occupations from the top nine U.S. GDP sectors, with 220 tasks open-sourced.
• Open Financial LLM Leaderboard - Public leaderboard benchmarking LLMs across financial NLP, question answering, and forecasting tasks.
• Remote Labor Index - Scale AI and CAIS benchmark of 240 real freelance projects, measuring the share AI agents complete at professional quality.
• REPRO-Bench - Benchmark of 112 social-science papers, mainly economics and political science, testing whether AI agents can assess reproducibility (ACL Findings 2025). [GitHub stars]
• STEER - UBC benchmarks of LLM economic rationality and microeconomic reasoning (ICML 2024, NeurIPS 2025). [GitHub stars]

Papers and Books

Key Papers

Grouped by theme, newest first.

Using AI in Economic Research

• AI-Assisted Teams Outperform AI-Led Teams but Not Human-Only Teams in Assessing Research Reproducibility in Quantitative Social Science - Brodeur et al. (2026), Proceedings of the National Academy of Sciences.
• Teaching Economics to the Machines - Chen et al. (2026), NBER Working Paper 34713.
• Performing Valid Inference with AI/ML-Generated Covariates: A Guide for Empirical Practice - Christensen and Hansen (2026), AEA Papers and Proceedings.
• A Practitioner's Guide to Using Large Language Models and Generative AI in Economic History - Ferrara (2026), NBER Working Paper 35374.
• Large Language Models: An Applied Econometric Framework - Ludwig, Mullainathan, and Rambachan (2026), Annual Review of Economics.
• Machine Learning Who to Nudge: Causal vs Predictive Targeting in a Field Experiment on Student Financial Aid Renewal - Athey, Keleher, and Spiess (2025), Journal of Econometrics.
• Artificial Intelligence in Economics Research: What Have We Learned? What Do We Need to Learn? - Bahoo et al. (2025), Journal of Economic Surveys.
• Deep Learning for Economists - Dell (2025), Journal of Economic Literature.
• Difference-in-Difference Causal Forests with an Application to Payroll Tax Incidence in Norway - Gavrilova, Langørgen, and Zoutman (2025), Journal of Applied Econometrics.
• AI Agents for Economic Research - Korinek (2025), NBER Working Paper 34202.
• Economics in the Age of Algorithms - Mullainathan (2025), AEA Papers and Proceedings.
• Machine Learning as a Tool for Hypothesis Generation - Ludwig and Mullainathan (2024), Quarterly Journal of Economics.
• Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus? - Horton, Filippas, and Manning (2023), NBER Working Paper 31122.
• Generative AI for Economic Research: Use Cases and Implications for Economists - Korinek (2023), Journal of Economic Literature.

Economics of AI

• The Emerging Market for Intelligence: How Firms Buy and Sell AI - Demirer, Fradkin, and Tadelis (2026), Journal of Economic Perspectives.
• AI and Our Economic Future - Jones (2026), Journal of Economic Perspectives.
• Economic Scenarios for Transformative AI - Korinek et al. (2026), CEPR Discussion Paper 21939.
• The Simple Macroeconomics of AI - Acemoglu (2025), Economic Policy.
• Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence - Brynjolfsson, Chandar, and Chen (2025), Stanford Digital Economy Lab working paper.
• A Research Agenda for the Economics of Transformative AI - Brynjolfsson, Korinek, and Agrawal (2025), NBER Working Paper 34256.
• Generative AI at Work - Brynjolfsson, Li, and Raymond (2025), Quarterly Journal of Economics.
• How People Use ChatGPT - Chatterji et al. (2025), NBER Working Paper 34255.

Books

• Applied Causal Inference Powered by ML and AI - Free textbook by Chernozhukov, Hansen, Kallus, Spindler, and Syrgkanis on double/debiased ML, with R and Python notebooks.
• The Economics of Transformative AI - Agrawal, Brynjolfsson, and Korinek's NBER volume of 16 studies on AI agents, labor, growth, and risk (UChicago Press, 2026).
• The Means of Prediction - Maximilian Kasy on how AI works and who benefits (UChicago Press, 2025).
• The Scaling Era - Dwarkesh Patel and Gavin Leech's oral history of AI from interviews with Amodei, Hassabis, Zuckerberg, and others (Stripe Press, 2025).

Courses, Conferences and Community

Courses, Guides and Tutorials

• AI for Economic Theorists and Mathematicians - Markus' Academy series by Pietro Ortoleva and Fedor Sandomirskiy (Princeton) on AI for modeling and proofs.
• AI for Economists - Jesse Lastunen's curated, filterable directory of papers, courses, tools, and talks on AI in economics.
• AI Tools for Economists and Policy Analysts - Will Rinehart's (AEI) regularly updated guide mapping AI tools and prompts to economics and policy tasks.
• Claude Blattman - Chris Blattman's (UChicago) open-source guide to AI research workflows with Claude Code, no coding required, plus an installable skills plugin. [GitHub stars]
• Claude Code for Applied Economists - Paul Goldsmith-Pinkham's (Yale) eight-episode video series on data work, scraping, skills, sandboxing, and Overleaf.
• Data Analysis with AI - Gábor Békés's (CEU) eight-session open course using AI coding agents for data wrangling, econometrics, text as data, and DiD.
• EconDL - Melissa Dell's companion site to Deep Learning for Economists, with demo notebooks, software resources, and lectures.
• Economics of Transformative AI - Stanford Digital Economy Lab summer course (2025) with open slides, exercises, and recorded lectures.
• genaiforecon.org - Korinek's companion site to his JEL survey, with periodic updates on LLM tools; the latest (August 2025) covers AI agents.
• How to Learn and Teach Economics with LLMs - Cowen and Tabarrok's (GMU) practical guide to using GPT for learning, teaching, and researching economics.
• Machine Learning and Causal Inference - Video course from Susan Athey's Stanford GSB lab on ML for treatment effects, heterogeneous impacts, and targeting, with R tutorials.
• ML and Economics at Oxford - Maximilian Kasy's reading group.
• Modern AI for Economics Research - Markus' Academy talk by Benjamin Golub on Cursor, AI refereeing, and prompting for research.
• NBER Methods Lecture: AI-Generated Data - Melissa Dell and Ashesh Rambachan's 2026 Summer Institute lectures on estimation and inference with AI-generated data.
• Reflections on Vibe Researching - Joshua Gans (Toronto) on where LLMs aid theory work and where human judgment stays essential.
• Research in the Time of AI - Paul Goldsmith-Pinkham (Yale) on where AI compresses the research timeline and where taste matters.
• The Economics of AI - Anton Korinek's (UVA) Coursera course on AI's implications for economic growth, labor markets, inequality, and policy.
• Using AI in Research and Teaching - Paul Goldsmith-Pinkham's (Yale) 2024 practical guide to Copilot in VS Code, chat LLMs, and local models for research and teaching.

Conferences

• ESIF Economics and AI+ML Meeting - Econometric Society's biennial interdisciplinary meeting on economics and AI/ML (latest: Cornell, June 16–17, 2026).
• NBER Digital Economics and AI Meeting - Stanford, January 21–22, 2027; paper submissions and PhD tutorial applications due November 3, 2026.
• NBER Economics of AI Conference - Annual fall conference in Toronto organized by Agrawal, Gans, Goldfarb, and Tucker (latest: September 23–24, 2026).

Community

Newsletters: Autonomous Econ (Wong) | Frankly, the Counterfactual Was Worse (IU O'Neill) | genaiforecon (Korinek) | One Useful Thing (Mollick) | paulgp (Goldsmith-Pinkham) | Scott's Mixtape (Cunningham)

Podcasts: Dwarkesh Podcast | Exponential View | Justified Posteriors (Benzell and Fradkin)

People: Erik Brynjolfsson | Scott Cunningham | Melissa Dell | Anton Korinek | Antonio Mele | Sendhil Mullainathan

Institutions: EconTAI | NBER Digital Economics and AI | OECD.AI | Stanford DEL

Appendix: General-Purpose AI

Widely known tools listed for completeness with notes on what economists specifically value.

• ChatGPT - Deep Research that can focus on sites you trust, a Work agent for spreadsheets and decks, and an Excel add-in.
• Claude - Coding help for Stata, R, and Python, 1M-token context for long documents, Research mode, and Claude for Excel.
• Claude Code - Anthropic's agentic coding tool for the terminal, IDE, desktop app, and browser, extensible with MCP servers and agent skills.
• Cline - Open-source coding agent for VS Code and the terminal that works with many model providers, including local models. [GitHub stars]
• Cursor - AI code editor with agent mode, parallel Cloud Agents, and a free Hobby tier.
• DeepSeek - MIT-licensed open-weight models with 1M-token context, plus a low-cost API for large-scale text classification.
• Gemini - Deep Research reports from hundreds of sites (optionally your Gmail and Drive), with Canvas for drafts and visuals.
• Gemini Notebook - Formerly NotebookLM: cited Q&A over your papers, Audio and Video Overviews, and code execution for data analysis on paid tiers.
• GitHub Copilot - Free and Student plans, agent mode in VS Code, a cloud coding agent, and completions-only support in RStudio.
• Ollama - Free local runner for open-weight models (gpt-oss, Qwen, Gemma, DeepSeek), so confidential microdata never leaves your machine. [GitHub stars]
• OpenAI Codex - Coding agent in the ChatGPT desktop app, open-source CLI, and IDE, running parallel local and cloud tasks with reusable skills. [GitHub stars]
• OpenCode - Open-source coding agent for the terminal and desktop that works with 75+ model providers, including local models. [GitHub stars]
• Perplexity - Cited answers with an Academic source filter for scholarly papers, plus multi-step Research reports.

Contributing

Contributions welcome! Read the contribution guidelines first.