#

Last updated: Feb 10, 2026, 5:56:27ā€ÆPM

Table of Contents#

Pipeline Charts šŸ“ˆ

Pipeline Dataframes šŸ“Š

Pipeline Specs#

Pipeline Name

Pipeline ID

crsp_analysis_pipeline

Lead Pipeline Developer

Contributors

Git Repo URL

Pipeline Web Page

Pipeline Web Page

Date of Last Code Update

2026-02-10 17:56:12

OS Compatibility

Linked Dataframes

crsp_analysis_pipeline:crsp_returns

This repository contains our final project for replicating tables from the paper ā€œIntermediary Asset Pricing: New Evidence from Many Asset Classes.ā€ Our goal is to reproduce key tables (Table 2 and Table 3) from the paper using data from CRSP, Compustat, and Datastreamā€”and to update these results with the latest available data.

Project Overview#

The paper argues that capital shocks to financial intermediaries can explain cross-sectional differences in expected returns across various asset classes (stocks, bonds, options, commodities, currencies, and credit default swaps). Based on this idea, our project focuses on:

  • Building Risk Factors: We construct risk factors using financial intermediariesā€™ capital ratios.

  • Replicating Table 2: This table shows the relative size of major market makers by calculating monthly ratios of total assets, book debt, book equity, and market equity relative to different market groups, and then averaging these ratios over time.

  • Replicating Table 3:

    • Panel A: Uses data from 1970Q1 to 2012Q4 to compute the Market Capital Ratio, Book Capital Ratio, and AEM Leverage Ratio, and explores the correlations among these ratios and key economic variables (such as E/P, unemployment, GDP, financial conditions, and market volatility).

    • Panel B: Constructs risk factors from the ratios in Panel A and analyzes their correlations with each other and with the growth rates of various economic indicators.

Data & Methodology#

  • Data Sources:
    We modified the primary dealer list based on real data sources. The ticks.csv file was updated according to the corresponding gvkey codes.

  • Modifications to Calculations:
    We adjusted the calculation logic for important ratios and macroeconomic variables (e.g., key ratios and updated macro variable computations) based on the paperā€™s description, resulting in significant optimization of the replicated results for Table 3.

  • Output Generation:
    The reproduced table results are automatically generated as LaTeX (.tex) files and saved in the output directory. Further data analysis, including descriptive statistics, correlation analysis, and visualization of ratio trends, was also performed.

  • Automation:
    We implemented complete automation of the project workflow, storing all results in the _output directory.

Project Structure#

  • LaTeX Documents:
    Auto-generated LaTeX files include the replicated tables and additional figures from the paper.

  • Jupyter Notebook:
    A comprehensive notebook demonstrates data processing, ratio calculations, table generation, and analysis results.

  • Python Code Files:
    Scripts for data download, processing, and metric calculations are included. Each file contains detailed docstrings describing its functionality.

  • Automation (dodo.py):
    The dodo.py file automates the entire project workflow including data fetching, processing, table generation, and even running the final notebook.

  • Test Files:
    Unit tests for both Table 2 and Table 3 have been developed to ensure the replication results closely match the original paper.

  • Additional Files:
    The repository also includes supplementary files such as a README, an environment example file (.env.example), and other auxiliary scripts to ensure a clear and rigorous project process.(If you want to clone our repository, please ensure you have an .env file (or rename .env.example to .env) and specify the paths for DATA_DIR and OUTPUT_DIR so that the scripts can locate input data and store all generated output files correctly.)

Work Division#

  • Liu Junyuan:

    • Developed the code for auto-generating the LaTeX summary document.

    • Worked on replicating additional tables and figures from the paper.

    • Enhanced table analysis scripts and improved the README file.

  • Hanlu Ge:

    • Modified the primary dealer list and updated ticks.csv based on real data sources.

    • Developed the Jupyter Notebook showcasing data processing, ratio calculations, table generation, and analysis.

    • Implemented the automation script (dodo.py) and wrote tests to ensure accurate replication of the tables.

Regular and transparent communication was maintained throughout the project to ensure that both team members remained aligned and met project goals.

Setup & Usage#

  1. Clone the Repository:

    git clone https://github.com/Junyuanxx/Final-Project-32900.git
cd Final-Project-32900

  2. Create and Activate the Virtual Environment:

    python -m venv env
source env/bin/activate  # On Windows, use `env\Scripts\activate`

  3. Install Dependencies:

    pip install -r requirements.txt

  4. Run the Project:

    set PYTHONWARNINGS=ignore::FutureWarning
doit

Contact#

If you have any questions or suggestions, please feel free to reach out via GitHub issues or contact the project members directly.

Contents