Skip to content

About me

I am Reeyarn Li, currently a post-doc researcher at Paderborn University, Germany. Between July and December 2025, I was also a visiting scholar at the University of North Carolina at Chapel Hill.

Published papers (and dataset)

Journal articles

Working papers

  • Kim, J., Li, R., & Petacchi, R. (2025). "Measuring the Informativeness of Audit Reports: A Machine Learning Approach"
  • Landsman, W., Li, R., & Sievers, S. (2025). "Investor Inattention to Earnings Metadata." (Job Market Paper)

Working-in-progress

  • Li, R. (2025). "The Economic Role of Non-GAAP Earnings." Available at SSRN with Abstract ID: 5926002

  • Kaiser, S., Li, R., & Sievers, S. (2025). "Centralized Electronic Reporting and Disclosure Informativeness: Evidence from a Top-Down Regulatory Change in Germany" EAA 2025 Annual Congress.

  • Franke, B., Kaiser, S., Li, R., Schlackl, F. & Sievers, S. (2025). "Changes in Firms’ Business Models and Accrual Estimation" EAA 2025 Annual Congress.

  • As is seen on my github, I'm working on something with XBRL and deep learning.

Open-source

Claude Code Skills

repo-to-collab (github.com/reeyarn/skill_collection/tree/main/repo-to-collab)

A Claude Code skill that scaffolds a research project for genuine human–AI co-authorship. Run it once in any repo; it generates a CLAUDE.md hierarchy, a living STATUS.md working-memory file, and a suite of session-management slash commands (/session-start, /wrap-session, /git-commit, /lit-search, /reg-table). The core idea: make the collaboration state explicit in files so neither party loses context between sessions.

Python Packages

LeanRL (github.com/reeyarn/LeanRL)

A lightweight, memory-efficient, and (hopefully) fast Python library for extracting specific information from XBRL filings and taxonomies — without loading the entire DTS (Discovery Tree). Forget strict XBRL validation and complex object models. Extract only what you need into simple Python structures (e.g., dict, list, pandas.DataFrame)

Open-ESEF (github.com/reeyarn/openesef)

  • Open-ESEF is a Python-based open-source project for handling XBRL filings that follow ESEF, the European Single Electronic Format.

  • ESEF is a standard for electronic financial reporting developed by the European Securities and Markets Authority (ESMA).

  • This project contains open-source code from two repos:

    • XBRL-Model project.
    • SEC EDGAR Financial Reports project.

  • Project is under development.

LiberalCourt (github.com/reeyarn/LiberalCourt)

Python script to build the LiberalCourt (pnld) measure from FJC judges data, extending it to more recent years. Replication and extension of Huang, A., K.W. Hui, & R.Z. Li (2019). "Federal Judge Ideology: A New Measure of Ex Ante Litigation Risk" Journal of Accounting Research 57(2): 431-489. https://onlinelibrary.wiley.com/doi/full/10.1111/1475-679X.12260