75-year Projections based on calibration to SSA Trustees data

[baogorek]

Summary

This PR implements a 75-year projection capability for federal income tax revenue (2025-2100) by integrating PolicyEngine's economic microsimulation with Social Security Administration demographic forecasts. This enables quantifying the fiscal impact of population aging while preserving the full complexity of the tax code.

Key Features

Two-Stage Projection Methodology

Stage 1: Economic Uprating

• Sophisticated modeling of 17 distinct income categories uprated by economic fundamentals
• Complete tax code implementation with all brackets, credits, deductions, and interactions
• Calibrated to match IRS Statistics of Income aggregates

Stage 2: Demographic Reweighting

• Two calibration methods available:
  • IPF (Iterative Proportional Fitting): Traditional raking approach with KL-divergence
  • GREG (Generalized Regression): Modern calibration enabling continuous variables
• Age-specific targets from SSA Trustees Report (single-year groups 0-85+)
• Optional calibration to Social Security benefit projections (GREG only)

What's Included

New Scripts

• run_household_projection.py: Main projection engine with IPF/GREG support
• create_reweighting_matrix.py: Builds demographic transition matrices
• age_projection.py: Processes SSA age-specific population forecasts
• extract_ssa_costs.py: Extracts Social Security benefit targets
• transition_matrix_demo.py: Validation and methodology demonstration

Data Files

• SSPopJul_TR2024.csv: SSA Trustees Report demographic projections (2025-2100)
• social_security_aux.csv: OASDI cost projections from SSA Table VI.G9

Documentation

• Comprehensive README explaining methodology and theoretical foundation
• Technical notes on avoiding multicollinearity in GREG calibration
• Validation metrics showing IPF/GREG agreement within 0.2%

Impact

This enables unprecedented analysis of:

• Fiscal sustainability: Decompose revenue changes into economic vs demographic components
• Policy design: Evaluate reforms in context of future demographics
• Distributional analysis: Track tax burden shifts between age cohorts
• Scenario planning: Model alternative demographic and economic scenarios

Usage

# Traditional IPF approach (default)
python run_household_projection.py 2050

# GREG calibration with demographics only
python run_household_projection.py 2050 --greg

# GREG with demographics + Social Security benefits
python run_household_projection.py 2050 --greg --use-ss

Validation

• Population totals exactly match SSA projections
• Age distributions preserved through reweighting
• Social Security benefits match SSA Trustees Report (when using --use-ss)
• IPF and GREG produce equivalent results (within 0.2%) for identical constraints
• Validated against R's survey package

Future Extensions Needed

• Update population calibration from CBO (ending 2055) to SSA (through 2100)
• Incorporate CBO Long-Term Budget Outlook inflation projections
• Extend income category projections beyond current CBO horizon using consistent methodology

---

This provides a comprehensive foundation for evidence-based fiscal policy analysis in an era of demographic transformation.