Add voluntary tax filer variable and update tax_unit_is_filer

[MaxGhenis]

Summary

• Add would_file_taxes_voluntarily input variable for voluntary tax filers
• Update tax_unit_is_filer to use propensity variables from microdata

Background

Previously, tax_unit_is_filer used a rules-based approach: file if required OR if income_tax < 0 (getting a refund). This deterministic approach didn't capture voluntary filing behavior.

The new approach uses propensity variables assigned during microdata construction:

1. Required to file - Income threshold logic from IRC § 6012 (unchanged)
2. Takes up EITC - Implies filing to claim refundable credits
3. Would file voluntarily - State requirements, documentation needs, habit

Changes

New variable: would_file_taxes_voluntarily

• Input variable with default_value = False
• Set to True in microdata for ~5% of tax units not taking up EITC
• Captures filing for reasons beyond refundable credits

Updated: tax_unit_is_filer

• Changed from value_type = float to value_type = bool
• Removed redundant unit = USD
• Now returns: required_to_file | takes_up_eitc | would_file_taxes_voluntarily

Related issues

• Closes Impute probability of filing taxes among tax units with zero net tax #4286 - Impute probability of filing taxes among tax units with zero net tax
• Related to How well does `(state tax + federal tax) != 0` reflect tax filing? #4283 - Discussion on how well tax != 0 reflects filing

Related PRs

• Add voluntary tax filer assignment policyengine-us-data#513 - Adds would_file_taxes_voluntarily to CPS microdata

Test plan

• Verify CI passes
• Confirm would_file_taxes_voluntarily variable is accessible
• Verify tax_unit_is_filer returns expected values

🤖 Generated with Claude Code