Back to Growth Econometrics

Methodology

Data sources, variable definitions, and processing pipeline

Data Source

All data in this module comes from the Penn World Table (PWT) version 11.0, published by the Groningen Growth and Development Centre at the University of Groningen.

Citation

Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2015), "The Next Generation of the Penn World Table". American Economic Review, 105(10), 3150-3182.

Variable Definitions

VariablePWT CodeDefinitionUnit
GDP per capitargdpe_pcReal GDP at chained PPPs (expenditure side) per capita2017 US$ PPP
PopulationpopPopulationMillions
Human CapitalhcHuman capital index based on years of schooling and returns to educationIndex
TFP LevelctfpTFP level at current PPPs (USA=1)Index (USA=1)
Labor SharelabshShare of labor compensation in GDP at current national pricesFraction (0-1)
EmploymentempNumber of persons engagedMillions

Purchasing Power Parity (PPP)

All monetary values are expressed in 2017 US dollars at purchasing power parity (PPP). PPP adjustments account for differences in price levels across countries, making GDP comparisons more meaningful than exchange-rate based comparisons.

For example, a dollar buys more goods and services in India than in Switzerland. PPP adjustments correct for this, providing a better measure of actual living standards.

Growth Accounting

The productivity decomposition uses standard growth accounting based on the Solow growth model:

gY = α · gK + (1-α) · gL + gA

Where:

  • gY = GDP growth rate
  • gK = Capital stock growth rate
  • gL = Labor growth rate (adjusted for human capital)
  • gA = TFP growth rate (Solow residual)
  • α = Capital share of income (~0.33)

Data Coverage

185
Countries
1950–2023
Time Period
11.0
PWT Version

Data Processing

Raw PWT data is processed through a pipeline that:

  1. Downloads the official PWT Excel file from the University of Groningen
  2. Extracts relevant variables (rgdpe, pop, hc, ctfp, labsh, emp)
  3. Computes derived variables (rgdpe_pc = rgdpe / pop)
  4. Handles missing values (linear interpolation for small gaps)
  5. Assigns countries to regions based on World Bank classifications
  6. Exports to JSON format optimized for web consumption

Enhanced Analysis (v2)

Version 2 introduces additional analysis outputs for deeper exploration of regional growth patterns, distributional dynamics, and economic mobility.

New Data Files

FileDescriptionUse Case
regional-distributions.jsonBoxplot stats and KDE data for each region per yearBoxplot/Violin charts
decadal-growth.jsonGrowth rates by country and decade (1960s–2010s)Small multiples grid
mobility-transitions.jsonCountry quintile transitions 1960→2020Sankey diagram
regional-trajectories.jsonRegional average GDP with confidence bandsRegional comparison chart

Key Definitions

Quintile Classification

Countries are ranked by GDP per capita and divided into five equal groups (quintiles). Q1 = poorest 20%, Q5 = richest 20%. This classification is computed for both 1960 and 2020 to measure mobility.

Kernel Density Estimation (KDE)

A smoothed approximation of the GDP distribution within each region. Uses Gaussian kernels with Silverman's rule-of-thumb bandwidth selection. Provides richer distributional information than boxplots alone.

Confidence Bands

Regional trajectories show ±1 standard deviation bands around the mean GDP per capita. This captures within-region variation and divergence over time.

Regional Classification

Countries are assigned to 8 major regions for comparative analysis:

Advanced Economies

OECD high-income: USA, Germany, Japan, UK, France, etc.

East Asia

China, South Korea, Taiwan, Hong Kong, Mongolia

South Asia

India, Pakistan, Bangladesh, Sri Lanka, Nepal

Southeast Asia

Thailand, Vietnam, Indonesia, Malaysia, Philippines

Latin America

Brazil, Mexico, Argentina, Chile, Colombia

Sub-Saharan Africa

Nigeria, South Africa, Kenya, Ethiopia, Ghana

Middle East & N. Africa

Saudi Arabia, Egypt, Iran, Turkey, Morocco

Eastern Europe

Poland, Russia, Ukraine, Czech Republic, Romania

Resources

Limitations

  • • Historical estimates before 1970 have larger uncertainty margins
  • • PPP comparisons are most reliable within income groups
  • • TFP is a "residual" and captures measurement error alongside true productivity
  • • Some countries have significant data gaps requiring interpolation
  • • Recent years (2020+) may be revised as COVID impacts are reassessed

Data: Penn World Tables 11.0 • Last updated: December 2024