Holding Period Effects in Dividend Strip Returns

Holding Period Effects in Dividend Strip Returns#

Last updated: Mar 16, 2026, 12:42:47 AM

Table of Contents#

Pipeline Charts 📈

Pipeline Dataframes 📊

Pipeline Specs#

Pipeline Name

Holding Period Effects in Dividend Strip Returns

Pipeline ID

P04

Lead Pipeline Developer

Jie Lin and Zimeng Yi

Contributors

Jie Lin, Zimeng Yi

Git Repo URL

Pipeline Web Page

Pipeline Web Page

Date of Last Code Update

2026-03-16 00:21:10

OS Compatibility

Linked Dataframes

P04:crsp_sp500_daily
P04:crsp_treasury_returns
P04:fred_treasury_rates
P04:fama_french_factors
P04:fama_french_monthly
P04:optionmetrics_spx_raw
P04:optionmetrics_spx_monthly
P04:optionmetrics_zero_curve
P04:clean_options
P04:clean_zero_curve
P04:clean_rates
P04:clean_crsp_sp500_monthly
P04:implied_rates
P04:implied_rates_1y
P04:zero_curve_1y
P04:strip_prices
P04:all_strip_prices
P04:monthly_returns

About this project#

This repository contains our final project for FINM 32900 at the University of Chicago.

We replicate key empirical results from:

Golez, Benjamin, and Jens Jackwerth.
Holding Period Effects in Dividend Strip Returns.
Review of Financial Studies 37, no. 10 (2024): 3188–3215.

Our assigned project is P04. Holding Period Effects in Dividend Strip Returns.

Using an end-to-end reproducible analytical pipeline, we replicate and extend the following objects from the paper:

  • Figure 1: 12-month interest rates

  • Figure 2: cumulative returns

  • Figure 3: annualized volatility across holding periods

  • Table 1: monthly return summary statistics

We also extend these results through the most recent available sample in our pipeline (through 2024).

This project is formatted using the cookiecutter chartbook template and automated using PyDoit.

Team responsibilities#

Both group members contributed to understanding the paper, designing the pipeline, validating results, and preparing the final report. Primary responsibilities were divided as follows.

Jie Lin#

  • Built and tested major parts of the automated data and calculation pipeline

  • Worked on strip return construction and diagnostics

  • Implemented Figure 2 replication and extensions

  • Implemented Figure 3 replication and extensions

  • Integrated figures and tables into the LaTeX report

  • Helped test end-to-end reproducibility with PyDoit

Zimeng Yi#

  • Worked on Figure 1 replication and extensions

  • Worked on Table 1 replication and extensions

  • Helped validate intermediate outputs and summary statistics

  • Contributed to the report writeup and empirical interpretation

  • Assisted with GitHub workflow, project organization, and debugging

Replication goals#

The paper studies dividend strip prices inferred from S&P 500 index options and emphasizes the importance of using option-implied interest rates rather than exogenous zero-curve rates. Our project reproduces the paper’s main empirical objects and then re-runs the same calculations on an updated sample.

Our replication pipeline performs the following steps:

  1. Pull raw data from WRDS, FRED, and the Fama-French Data Library

  2. Clean the raw data into tidy intermediate datasets

  3. Estimate option-implied interest rates

  4. Construct dividend strip prices using put-call parity

  5. Build monthly strip and market return panels

  6. Generate final figures, tables, and the LaTeX report

Quick start#

1. Install LaTeX#

You must have TeX Live (or another LaTeX distribution) installed and available in your path.

Installers:

2. Create and activate a virtual environment#

python -m venv .venv
source .venv/bin/activate      # On Windows: .venv\Scripts\activate

3. Install dependencies#

pip install -r requirements.txt

4. Create a .env file#

Copy .env.example into a new file called .env in the project root:

cp .env.example .env

Then edit .env with your own settings. A typical file looks like:

WRDS_USERNAME=your_wrds_username
START_DATE=1996-01-01
END_DATE=2025-12-31

5. Run the full pipeline#

doit all

This executes the project end-to-end: raw data pulls → cleaning → intermediate calculations → figure and table generation.

6. List available tasks#

doit list

Main PyDoit tasks#

Task

Description

pull_crsp_spindx

Pull daily S&P 500 index data from CRSP

pull_crsp_treasuries

Pull monthly Treasury returns from CRSP

pull_fred

Pull Treasury rates and Fama-French factors

pull_optionmetrics

Pull SPX options and zero curve from OptionMetrics

clean

Clean and standardize all raw data

calc_implied_rates

Estimate option-implied interest rates

calc_strip_prices

Construct dividend strip prices

calc_returns

Build monthly return panel

plot_figure1

Replicate Figure 1

plot_figure1_extension

Extend Figure 1 through 2024

figure2

Replicate Figure 2

figure2_extended

Extend Figure 2 through 2024

figure2_extended_winsorized

Robustness version of extended Figure 2

figure3

Replicate Figure 3

figure3_extended

Extend Figure 3 through 2024

table1

Replicate Table 1

table1_extended

Extend Table 1 through 2024

all

Run the full pipeline end-to-end

Data sources#

Source

Content

OptionMetrics via WRDS

European SPX options; zero-coupon curve

CRSP via WRDS

Daily S&P 500 index data; monthly Treasury bond returns

FRED

Treasury constant maturity rates

Kenneth French Data Library

Daily and monthly Fama-French factors; 1-month risk-free rate

Directory structure#

src/                  Python scripts for pulling, cleaning, calculating, and plotting
_data/                Raw, cleaned, and intermediate parquet files (not tracked in Git)
_data/calc/           Derived datasets (implied rates, strip prices, monthly returns)
output/               Generated figures and tables
reports/              LaTeX report source files
assets/               Static files not generated automatically

Why _data/ is not tracked in Git#

Most files in _data/ are either automatically reproducible by rerunning the pipeline, or they come from licensed academic databases (WRDS) and cannot be redistributed.

Naming conventions#

Prefix

Purpose

pull_*.py

Pull raw data from an external source

clean_*.py

Clean and standardize raw data

calc_*.py

Compute derived datasets

figure*.py / table*.py / plot_*.py

Generate final output figures and tables

Notebook tour#

The project includes notebooks that provide a tour of the cleaned data and selected parts of the analysis logic. These notebooks help the reader understand the structure of the data, how the paper’s methodology is implemented, and how key replication objects were validated before scripting.

The final official outputs are generated by the Python scripts and the PyDoit pipeline, not by manual notebook execution.

Unit tests#

This project uses unit tests to validate key intermediate logic and replicated values, including checks on Figure 1 summary statistics, Table 1 values, Figure 2 terminal values, and Figure 3 holding-period volatility values.

pytest                    # Run all tests
pytest --doctest-modules  # Include doctests

Setting environment variables#

If you want to export your .env variables into your shell session:

Linux / Mac:

set -a
source .env
set +a

Windows (PowerShell):

Get-Content .env | ForEach-Object {
    if ($_ -match '^([^=]+)=(.*)$') {
        [Environment]::SetEnvironmentVariable($matches[1], $matches[2], 'Process')
    }
}

Code formatting#

This project uses Ruff for linting and formatting:

ruff check . --fix
ruff format .

Notes on data limitations#

The original paper uses intraday CBOE data for part of the sample. Our replication uses WRDS-accessed OptionMetrics end-of-day data, which leads to some remaining quantitative differences, especially in strip return statistics:

  • Figure 1 is replicated closely

  • Market return results are generally close to the paper

  • Strip return results are directionally and qualitatively similar, but can differ more in magnitude

  • Post-2022 strip extensions become less stable and should be interpreted cautiously

These issues are discussed in detail in the report.

Repository hygiene#

This repository does not contain:

  • Copyrighted raw data

  • API keys or private credentials

  • Committed .env files

Use .env.example as the template for your local .env file.

Contents