Explore 75-Year Projection Datasets for Social Security Reform Analysis

[DTrim99]

Explore 75-Year Projection Datasets for Social Security Reform Analysis

Background

The current Social Security taxation reform analysis in this repository is limited to 10-year projections (2026-2035). New PolicyEngine datasets will soon enable 75-year projections, which are critical for:

• Long-term solvency analysis
• Intergenerational impact assessment
• Alignment with Social Security Trustees' 75-year projection window

Objective

Create a data exploration notebook (similar to analysis-notebooks PR #88) to:

1. Validate 75-year dataset coverage across the projection period
2. Explore Social Security-specific variables relevant to this analysis
3. Document data quality and limitations for long-term projections
4. Identify any gaps that would affect reform impact calculations

Variables to Explore

Core Social Security Variables

• taxable_social_security - Portion of SS benefits subject to income tax (critical for Options 1-4, 8)
• social_security - Total Social Security benefits received
• social_security_retirement - Retirement benefits specifically
• social_security_disability - SSDI benefits
• social_security_dependents - Dependent/survivor benefits
• age - Age distribution over 75 years (demographic shifts)
• is_senior - Senior population trends

Related Tax Variables

• adjusted_gross_income - AGI trends over 75 years
• taxable_income - Taxable income base
• income_tax - Federal income tax (for revenue calculations)
• standard_deduction - Changes over time
• itemized_deductions - Itemization patterns

Employment/Payroll Variables (for Option 5-6 Roth-Style Swap)

• employment_income - Wage base for employer-side payroll tax
• fica_ss_tax - Current Social Security payroll tax
• employer_social_security_tax - Employer portion (relevant to Options 5-6)

Analysis Structure

Following the RI dataset exploration approach:

1. Dataset Overview

• Load dataset: Microsimulation(dataset="<75-year-dataset-path>")
• Check temporal coverage (2026-2100)
• Validate household/person counts across years
• Document any missing years or data gaps

2. Social Security Benefit Distribution

# For each decade: 2030, 2040, 2050, 2060, 2070, 2080, 2090, 2100
for year in [2030, 2040, 2050, 2060, 2070, 2080, 2090, 2100]:
    taxable_ss = sim.calculate("taxable_social_security", period=year, map_to="household")
    total_ss = sim.calculate("social_security", period=year, map_to="household")

    # Distribution analysis
    # - Median/percentiles
    # - Share of households with SS income
    # - Taxable vs non-taxable portion

3. Age Distribution Trends

• Senior population growth (65+)
• Working-age population (18-64)
• Demographic dependency ratios
• Validate against SSA Trustees projections if possible

4. Income Component Trends

• How AGI composition changes over 75 years
• Growth rates of different income sources
• Real vs nominal values (inflation adjustments)

5. Reform-Specific Variables

• For Option 4 ($500 tax credit): Eligible population size
• For Options 5-6 (Roth Swap): Employment income base trends
• For Option 7 (Senior deduction): Senior population with standard deduction

6. Data Quality Checks

• Missing values by variable and year
• Outliers and extreme values
• Consistency with known demographic trends
• Comparison with SSA Trustees intermediate assumptions

7. Summary Tables

Export CSV summaries similar to RI exploration:

• 75_year_ss_variables_summary_weighted.csv
• 75_year_ss_variables_summary_unweighted.csv
• Include breakdowns by decade

Deliverable

A Jupyter notebook (analysis/75_year_dataset_exploration.ipynb) that:

• Mirrors the structure of ri_dataset_exploration.ipynb
• Focuses on Social Security and age-related variables
• Produces summary tables for documentation
• Identifies any data limitations for 75-year projections
• Does NOT run during CI (place in analysis/ folder, not jupyterbook/)

Related Work

• Current 10-year analysis: analysis/policy-impacts-dynamic.ipynb
• RI dataset exploration: PolicyEngine/analysis-notebooks#88
• This will enable extending current analysis to full 75-year window

Notes

• Timing: Wait for dataset availability announcement
• Location: Create in analysis/ folder (not jupyterbook/) to avoid CI timeouts
• Execution time: May be long (expect similar runtime to 10-year dynamic calculations)
• Commit outputs: Execute notebook and commit with outputs for documentation