Earnings Management Then and Now: Revisiting Jones (1991)
Website Data, Replication, and Research Design
About This Lecture
This lecture explores how modern data sources can breathe new life into classic accounting research. We’ll examine Jennifer Jones’s influential 1991 study on earnings management during import relief investigations, then see how contemporary website data could extend and validate her findings.
The lecture is structured in two parts:
Part 1: Three Papers introduces the methodological toolkit and theoretical foundation:
- Overview: Modern web scraping techniques for longitudinal data collection based on Haans and Mertens (2024)
- Website-based disclosure measures and their validation based on Boulland et al. (2025)
- The classic Jones (1991) earnings management study
Part 2: Research in Action demonstrates the research process:
- My interpretation of Jones (1991) with modern methods
- Preliminary exploration of website disclosure patterns during investigations
- Structured critique and discussion of research design choices
Beyond understanding specific methods and findings, this lecture aims to stimulate your own research creativity. By engaging with the materials—seeing how papers can be combined, observing the decision-making process, and practicing structured critique—you’ll develop pattern recognition for spotting research opportunities. The GitHub repository provides working code and data that you can adapt for your own questions, and the lecture format encourages you to think: “What would I do differently?” or “Where else could this approach work?”
Lecture Materials
📊 View Slides
Click above to view the complete presentation with examples from recent research papers.
📁 GitHub Repository
Check out the repository for all code and data used in this lecture.
🤖 Paper Summaries
Summaries of key papers generated using a local LLM:
Is this better than any of the free LLMs out there? Probably not!
BUT: It’s YOURS to adapt, extend, and improve as you see fit
As always: Check for accuracy and completeness! Compare to your own reading of the papers. We can tweak the prompts together if you want to generate your own summaries.
For another way to utilize LLMs in research, see Peter and Weinrich (2025)
Learning Goals
This lecture is designed to spark your own research ideas:
- During Part 1: Ask yourself: “What other classic studies could benefit from website data?”
- During Part 2: Think: “What would I do differently? What am I curious about?”
- After class: The GitHub materials are yours to adapt, extend, or apply to entirely different questions
The best outcome isn’t just understanding this project—it’s developing your ability to spot opportunities where new data sources can unlock fresh insights into established phenomena. Use the lecture materials as a launching pad for your own creativity.
Contact
For questions, ideas, or potential extensions, feel free to reach out.
License
This repository is made available under the MIT License. You are free to use, adapt, and extend the code and materials, provided that you include appropriate attribution.
How to cite
If you use this repository or lecture materials in your own work or teaching, you can cite it as:
@misc{peter2025phdlecture,
author = {Caspar David Peter},
title = {Earnings Management Then and Now: Revisiting Jones (1991) -- Website Data, Replication, and Research Design},
year = {2025},
howpublished = {GitHub repository},
url = {https://github.com/CasparDP/phd-lecture-dec}
}AI/LLM disclosure
AI tools (local and hosted LLMs) helped with brainstorming, structuring the lecture, and polishing some wording. The research ideas, empirical designs, and code are mine—any errors are proudly human.