Module 06: Panel Data Analysis​‌​‌​‌​​‍​‌​​​‌​‌‍​​‌‌‌​‌​‍​​‌‌​‌​​‍​‌‌​​‌​‌‍​​‌‌‌​​​‍​‌‌​​‌​​‍​​‌‌‌​​‌‍​‌‌​​​‌​‍​​‌‌​​​‌‍​‌‌​​​​‌‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌​‌‌​‍​‌‌​​‌​‌‍​​‌‌​‌‌‌‍​‌‌​​​‌‌‍​​‌‌​​‌‌‍​‌‌​​‌​​‍​​‌‌‌​‌​‍​​‌‌​​‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​‌‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​​​‌‍​​‌‌​​‌‌‍​​‌‌‌​‌​‍​‌‌​​​‌​‍​​‌‌​​‌​‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌‌​​‌‍​​‌‌‌​​​‍​​‌‌​‌​‌‍​‌‌​​‌‌​¶

Status: Available

General Description¶

We will learn panel data econometrics: the combination of cross-sectional data​‌​‌​‌​​‍​‌​​​‌​‌‍​​‌‌‌​‌​‍​​‌‌​‌​​‍​‌‌​​‌​‌‍​​‌‌‌​​​‍​‌‌​​‌​​‍​​‌‌‌​​‌‍​‌‌​​​‌​‍​​‌‌​​​‌‍​‌‌​​​​‌‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌​‌‌​‍​‌‌​​‌​‌‍​​‌‌​‌‌‌‍​‌‌​​​‌‌‍​​‌‌​​‌‌‍​‌‌​​‌​​‍​​‌‌‌​‌​‍​​‌‌​​‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​‌‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​​​‌‍​​‌‌​​‌‌‍​​‌‌‌​‌​‍​‌‌​​​‌​‍​​‌‌​​‌​‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌‌​​‌‍​​‌‌‌​​​‍​​‌‌​‌​‌‍​‌‌​​‌‌​ (countries, individuals) with time series (years). We will work with Fixed Effects, Random Effects, and Two-Way Fixed Effects models applied to real social science problems.

Level: Advanced Technologies: Python, linearmodels, pandas, Altair Prerequisites: Basic statistics, linear regression

Learning Objectives¶

• Understand the structure of panel data (unit x time)
• Distinguish between Pooled OLS, Fixed Effects, and Random Effects
• Apply the Hausman Test to choose between FE and RE
• Implement Two-Way Fixed Effects (unit + time effects)
• Interpret Odds Ratios in logistic models
• Calculate marginal effects in nonlinear models

Module Content¶

The complete module is located at:

ejercicios/06_analisis_datos_de_panel/
├── 01_analisis_guns.py              # Panel: gun laws and crime
├── 02_analisis_fatality.py          # TWFE: beer tax vs mortality
├── 03_dashboard_educativo.py        # Interactive dashboard with 4 tabs
├── conceptos_visuales_panel.py      # Conceptual visualizations
├── GUIA_PANEL_DATA.md               # Complete theoretical guide
├── grafico_panel_guns.png           # Analysis result
└── requirements.txt                 # linearmodels, altair, etc.

Practical Exercises¶

01 - Guns Analysis: Gun Laws and Crime¶

Question: Do gun-carry laws reduce violent crime?

• Dataset: Guns (Stock & Watson) - 50 U.S. states, 1977-1999
• Dependent variable: log(violent) - violent crime rate (logarithm)
• Key variable: law - whether the state has a "shall-carry" law (mandatory carry)
• Controls: income, population, density
• Models: Pooled OLS vs Fixed Effects vs Random Effects
• Methodology: Comparison of the 3 models + Hausman Test

python ejercicios/06_analisis_datos_de_panel/01_analisis_guns.py

02 - Fatality Analysis: Beer Tax and Mortality¶

Question: Does raising the beer tax reduce traffic accident deaths?

• Dataset: Fatalities (AER) - 48 states, 1982-1988
• Dependent variable: fatality_rate - deaths per 10,000 inhabitants
• Key variable: beertax - beer tax
• Controls: minimum drinking age (drinkage), unemployment, income
• Models: Entity FE vs Two-Way Fixed Effects (state + year)
• Innovation: TWFE controls for temporal trends (safer cars each year)

python ejercicios/06_analisis_datos_de_panel/02_analisis_fatality.py

03 - Interactive Educational Dashboard (Panel + Altair)¶

Local dashboard with 4 interactive tabs to visually explore the concepts:

1. Pooled OLS: Heterogeneity slider that shows Simpson's Paradox in action
2. FE vs RE: Explanation and decision table for the Hausman Test
3. Odds Ratios: Sliders to explore probability vs odds vs odds ratio in real time
4. Marginal Effects: Lin-Lin, Log-Lin, Log-Log comparison with dynamic charts

panel serve ejercicios/06_analisis_datos_de_panel/03_dashboard_educativo.py
# Open http://localhost:5006 in your browser

Additional Files¶

• conceptos_visuales_panel.py - Generates static charts explaining panel concepts
• dashboard_educativo_panel.py - Alternative version of the educational dashboard
• GUIA_PANEL_DATA.md - Complete theoretical guide: Pooled OLS, FE, RE, Hausman, TWFE

Theory: Key Concepts¶

What is panel data?¶

Data that combines cross-sectional (N units) with time series (T periods):

| country | year | democracy | gdp_pc |
|---------|------|-----------|--------|
| ESP     | 2000 | 0.85      | 24000  |
| ESP     | 2001 | 0.86      | 24500  |
| FRA     | 2000 | 0.88      | 28000  |
| FRA     | 2001 | 0.89      | 28500  |

Data Engineering: Wide vs Long Format¶

Pooled OLS vs Fixed Effects vs Random Effects¶

The Naive Model: Pooled OLS¶

Unobserved Heterogeneity¶

Hausman Test¶

Decides between FE and RE:

• H0: RE is consistent and efficient (prefer RE)
• H1: Only FE is consistent (prefer FE)
• If p-value < 0.05: use Fixed Effects

Dashboards¶

Educational Dashboard (local)¶

The main dashboard for this module runs locally with Panel (HoloViz):

panel serve ejercicios/06_analisis_datos_de_panel/03_dashboard_educativo.py

Includes 4 interactive tabs: Pooled OLS, FE vs RE, Odds Ratios, Marginal Effects.

QoG Dashboard - Advanced Analysis (GitHub Pages)¶

As a complement, you can explore a dashboard with panel analysis applied to the QoG dataset (4 research lines with Spark + PostgreSQL + ML):

View QoG Dashboard - Applied Panel Data

Resources¶

Documentation¶

• linearmodels Documentation
• Altair Documentation

Theoretical References¶

• Wooldridge, J. M. (2010). Econometric Analysis of Cross Section and Panel Data (2^nd ed.). MIT Press.
• Stock, J. H., & Watson, M. W. (2019). Introduction to Econometrics (4^th ed.). Pearson.

​‌​‌​‌​​‍​‌​​​‌​‌‍​​‌‌‌​‌​‍​​‌‌​‌​​‍​‌‌​​‌​‌‍​​‌‌‌​​​‍​‌‌​​‌​​‍​​‌‌‌​​‌‍​‌‌​​​‌​‍​​‌‌​​​‌‍​‌‌​​​​‌‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌​‌‌​‍​‌‌​​‌​‌‍​​‌‌​‌‌‌‍​‌‌​​​‌‌‍​​‌‌​​‌‌‍​‌‌​​‌​​‍​​‌‌‌​‌​‍​​‌‌​​‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​‌‌​‍​​‌‌​​​​‍​​‌‌​​‌​‍​​‌‌​​​‌‍​​‌‌​​‌‌‍​​‌‌‌​‌​‍​‌‌​​​‌​‍​​‌‌​​‌​‍​​‌‌​‌​‌‍​‌‌​​‌‌​‍​​‌‌‌​​‌‍​​‌‌‌​​​‍​​‌‌​‌​‌‍​‌‌​​‌‌​---

Course: Big Data with Python - From Zero to Production Instructor: Juan Marcelo Gutierrez Miranda | @TodoEconometria Hash ID: 4e8d9b1a5f6e7c3d2b1a0f9e8d7c6b5a4f3e2d1c0b9a8f7e6d5c4b3a2f1e0d9c Methodology: Progressive exercises with real data and professional tools

Academic references:

• Wooldridge, J. M. (2010). Econometric Analysis of Cross Section and Panel Data (2^nd ed.). MIT Press.
• Stock, J. H., & Watson, M. W. (2019). Introduction to Econometrics (4^th ed.). Pearson.
• Baltagi, B. H. (2021). Econometric Analysis of Panel Data (6^th ed.). Springer.