Smoot–Hawley Tariff Act & the Great Depression: A Quantitative Analysis

A quant portfolio project that tests whether the Smoot–Hawley Tariff Act (June 1930) amplified the Great Depression through trade contraction, structural breaks, and market reactions—using econometric and time-series methods.

---

Table of Contents

• Overview
• Research Question & Hypotheses
• Historical & Theoretical Foundation
• Project Structure
• Prerequisites
• Installation
• Data
• How to Run
• Run on Google Colab
• Methodology Summary
• File Descriptions
• Outputs & Deliverables
• Interpreting Results
• Data Sources for Real Research
• Optional Extensions
• How This Positions You
• References & Further Reading
• License & Contact

---

Overview

This repository contains a complete, reproducible analysis of the link between the Smoot–Hawley Tariff Act and the Great Depression. It combines:

• Data preparation (macro + market, 1925–1935)
• Exploratory data analysis (trade trends, pre vs post 1930, volatility)
• Econometric testing (structural break, regression, event study)
• Causality & robustness (Granger causality, variance shift)
• Advanced (Section 8): CUSUM stability test, Newey–West HAC standard errors, VAR + Granger + impulse response, placebo event study, difference-in-differences (treatment vs synthetic control)

The project can be run with synthetic data (for demo) or real historical data (see Data and Data Sources).

---

Research Question & Hypotheses

Primary research question

Did the Smoot–Hawley Tariff Act amplify the Great Depression?

Supporting questions:

• Did trade contraction significantly impact GDP and industrial output?
• Did markets anticipate or react to tariff implementation?
• Was there a structural break in trade after 1930?

Testable hypotheses

ID	Hypothesis
H1	Trade volume declined significantly after June 1930.
H2	Industrial production decline correlates with trade contraction.
H3	Stock market volatility increased post-enactment.

Economic transmission channel

Tariff ↑ → Imports ↓ → Retaliation ↑ → Exports ↓ → Output ↓ → Unemployment ↑

The analysis tests whether this channel shows up in the data as a regime change around June 1930.

---

Historical & Theoretical Foundation

• Post–WWI: Agricultural overproduction; protectionist political climate in the US and abroad.
• Policy: President Herbert Hoover signed the Smoot–Hawley Tariff Act in June 1930, raising tariffs on thousands of goods.
• International retaliation: Trading partners (e.g. Canada, UK) raised their own tariffs; global trade collapsed.
• Collapse in global trade: The project asks whether 1930 represents a structural break and whether trade is associated with output and market volatility.

The discussion section in the notebook addresses: Did tariffs cause the trade collapse, amplify an existing one, or were they largely reactive?

---

Project Structure

Quant/
├── README.md                          # This file (A–Z documentation)
├── requirements.txt                  # Python dependencies
├── run_analysis.py                   # One-shot script to run full analysis
│
├── data/                             # Data directory
│   ├── macro_1925_1935.csv           # US macro (or synthetic)
│   └── dow_1925_1935.csv             # Dow Jones (or synthetic)
│
├── notebooks/
│   ├── smoot_hawley_analysis.ipynb   # Main analysis notebook
│   └── eda_output.png                # Generated EDA plots (after run)
│
├── src/
│   ├── data_prep.py                  # Data loading, Post_1930, returns, YoY, control series
│   └── advanced_analysis.py          # CUSUM, HAC, VAR, placebo, DiD
│
├── scripts/
│   └── generate_demo_data.py        # Write synthetic data to data/
│
└── docs/
    └── data_sources.md               # Data sources for real research

---

Prerequisites

• Python: 3.9 or higher (tested on 3.11, 3.13)
• OS: Windows, macOS, or Linux
• Optional: Jupyter (for interactive notebook); otherwise use run_analysis.py

---

Installation

1. Clone or download the project

cd path/to/Quant

2. Create and activate a virtual environment

Windows (PowerShell):

python -m venv .venv
.venv\Scripts\activate

macOS / Linux:

python3 -m venv .venv
source .venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. (Optional) Generate demo data

If data/macro_1925_1935.csv and data/dow_1925_1935.csv do not exist, the notebook will generate synthetic data automatically. To pre-generate via script:

python scripts/generate_demo_data.py

---

Data

Variables (1925–1935)

Category	Variables
Macro	US imports, exports, GDP, industrial production, unemployment
Market	Dow Jones index, log returns, 12-month rolling volatility
Constructed	Post_1930 (1 after June 1930, else 0), YoY % changes for macro series

Expected CSV format

• data/macro_1925_1935.csv
  Columns: date, imports, exports, gdp, ind_prod, unemployment
date in a standard format (e.g. YYYY-MM-DD).

• data/dow_1925_1935.csv
  Columns: date, dow_jones (or close).

Synthetic vs real data

• Synthetic: Used for demonstration; generated by generate_synthetic_data() in src/data_prep.py (structural break and volatility increase post-1930).
• Real: For actual research, replace the CSVs with historical series; see Data Sources for Real Research and docs/data_sources.md.

---

How to Run

Option A: Interactive notebook (recommended for exploration)

jupyter notebook notebooks/smoot_hawley_analysis.ipynb

Or with Jupyter Lab:

jupyter lab notebooks/smoot_hawley_analysis.ipynb

Run all cells from top to bottom. If the data files are missing, the first code block will create synthetic data in data/.

Option B: One-shot script (reproducible run + saved plots)

From the project root (Quant/):

python run_analysis.py

This will:

• Load or generate data
• Print EDA (means, variances, pre vs post 1930)
• Run Chow test, regressions, and Granger causality
• Save EDA plots to notebooks/eda_output.png

Run on Google Colab

1. Upload the project to Google Drive (zip the Quant folder so it includes src/data_prep.py and src/advanced_analysis.py, plus data/, notebooks/; upload, then unzip so you have My Drive/Quant/ with src/, data/, notebooks/ inside).
2. Open colab.research.google.com → File → Open notebook → Google Drive → open Quant/notebooks/smoot_hawley_analysis.ipynb.
3. Run the first code cell. It will mount Drive (approve when prompted), set the project root to /content/drive/MyDrive/Quant, and install dependencies. If your folder is elsewhere (e.g. My Drive/Projects/Quant), edit the ROOT = Path(...) line in that cell.
4. Run all (or run each cell in order). The notebook runs Sections 1–7 and Section 8 (Advanced). If data/ has no CSVs, synthetic data will be generated.

If you use GitHub instead of Drive: open the notebook from GitHub in Colab, then replace the first code cell with the “clone repo” version in COLAB.md so Colab has the full repo (including src/advanced_analysis.py).

Full step-by-step and troubleshooting: COLAB.md.

---

Methodology Summary

Step	Content
1. Data prep	Load macro + market CSVs; add Post_1930, log returns, YoY % changes, rolling volatility.
2. EDA	Trade and industrial production trends; Dow level and volatility; pre vs post 1930 means/variances and % contraction.
3. Structural break	Chow test at June 1930; variance shift (residual variance post/pre); rolling regression (e.g. GDP on imports).
4. Regression	GDP and industrial production on imports, exports, Post_1930.
5. Event study	Dow returns and volatility around June 1930.
6. Causality	Granger causality (e.g. imports → GDP).
7. Discussion	Interpret cause vs amplification vs reactive policy; link to modern trade wars.
8. Advanced	CUSUM (parameter stability); Newey–West HAC SEs; VAR + Granger + IRF; placebo event (non-event date); DiD (US vs synthetic control).

---

File Descriptions

File	Purpose
README.md	Full project documentation (this file).
requirements.txt	Python package list (pandas, numpy, scipy, matplotlib, seaborn, statsmodels, jupyter).
run_analysis.py	Runs full pipeline and saves EDA figure; no Jupyter required.
notebooks/smoot_hawley_analysis.ipynb	Main analysis: hypotheses, data, EDA, econometrics, discussion.
src/data_prep.py	prepare_analysis_dataset(), generate_synthetic_data(), generate_control_series() (for DiD), loaders, Post_1930, returns, YoY.
src/advanced_analysis.py	CUSUM test, OLS with Newey–West HAC, VAR + Granger + IRF, placebo event study, DiD estimation.
scripts/generate_demo_data.py	Writes synthetic macro and Dow CSVs to data/.
docs/data_sources.md	Suggested sources (NBER, FRED, etc.) and prep checklist for real data.

---

Outputs & Deliverables

• Notebook: Clean, sectioned analysis with hypothesis, EDA, structural break tests, regressions, event study, Granger, and discussion.
• Figures: EDA plots (trade, industrial production, Dow, volatility) with June 1930 vertical line; saved as notebooks/eda_output.png when using run_analysis.py.
• Console/notebook output: Descriptive statistics, Chow test (F, p-value), OLS regression tables, Granger p-values.
• Discussion template: Section 7 in the notebook for your narrative (cause vs amplification, reactive vs destructive, modern parallels).

---

Interpreting Results

• Chow test: Rejecting H0 (no break) at 5% suggests a significant regime change at June 1930. With synthetic data, the result is illustrative only.
• Regressions: Significant coefficients on imports/exports support a link between trade and output; significant Post_1930 captures a level shift after the tariff.
• Event study: Negative abnormal returns or higher volatility around June 1930 support market reaction to the policy.
• Granger: Low p-values for “imports → GDP” suggest trade helps predict output (consistent with the transmission channel).

Always interpret in light of real data and identification (e.g. confounding factors, endogeneity).

---

Data Sources for Real Research

For replicable research, replace synthetic data with historical series. Summary:

Variable	Suggested source
GDP	NBER Macrohistory, FRED (e.g. GDPC1 if long series)
Industrial production	FRED INDPRO, NBER
Imports / Exports	NBER, Census (Historical Statistics of the United States)
Unemployment	NBER, BLS historical
Dow Jones	Yahoo Finance (^DJI), NBER, Global Financial Data

See docs/data_sources.md for details and a preparation checklist (frequency alignment, column names, Post_1930, log returns, YoY).

---

Optional Extensions

• FRED/yfinance: Use fredapi or yfinance to pull series programmatically (see commented lines in requirements.txt).
• More countries: Compare with a country with lower tariff escalation for robustness.
• Lagged trade effects: Add lags of exports/imports in regressions.
• Academic format: Turn the notebook narrative into an abstract, literature review, methodology, results, and policy implications.

---

How This Positions You

• Quant roles: Macro regime thinking, structural break analysis, time-series modeling, policy shock propagation.
• MSc / applied stats: Applied econometrics, hypothesis-driven empirical work.
• Interviews: Clear story on how policy shocks propagate, how trade wars affect markets, and how structural breaks change risk regimes.

---

References & Further Reading

• Smoot–Hawley Tariff Act (1930); Great Depression history and trade collapse.
• NBER Macrohistory Database; FRED (Federal Reserve Economic Data).
• Structural break tests (Chow, CUSUM); event study methods; Granger causality.

---

License & Contact

• License: cite data sources when using real data.
• Contact: Email- arjun.lakhanpal.in@gmail.com/lakhanpxl.arjun@gmail.com .

---

End of README — project documentation A–Z.