State unemployment insurance programs are now being added as state-specific compute-only variables, starting with Pennsylvania UC and followed by New Jersey and New York UI.
The open modeling question is how and when those state-specific modeled benefits should feed the shared unemployment_compensation income variable, which then flows into taxable unemployment compensation, SPM income, and other downstream aggregates.
This needs a shared design because the program inputs are not simple annual income fields:
- PA UC uses high-quarter wages, base-year wages, credit weeks, weekly earnings, and weeks unemployed.
- NJ UI uses qualifying base weeks, base-period wages, weekly wages, and claimed weeks.
- NY UI uses high-quarter wages, second-highest-quarter wages, base-period wages, quarters with wages, weekly hours, weekly earnings, and weeks unemployed.
Naively deriving these from annual survey income could impose distributional assumptions that change eligibility and benefit amounts. For example, a uniform annual-to-quarterly allocation can fail or distort quarter-based monetary eligibility tests.
Scope for this follow-up:
- Decide whether state UI programs should auto-populate
unemployment_compensation, remain opt-in model variables, or feed a separate state-benefit aggregate.
- Define an allocation/data-input pattern for annual microsimulation data.
- If integration is approved, update
unemployment_compensation or the appropriate aggregate for all state UI implementations consistently.
- Add tests covering downstream taxable unemployment compensation and household/SPM income flows.
Related PRs:
State unemployment insurance programs are now being added as state-specific compute-only variables, starting with Pennsylvania UC and followed by New Jersey and New York UI.
The open modeling question is how and when those state-specific modeled benefits should feed the shared
unemployment_compensationincome variable, which then flows into taxable unemployment compensation, SPM income, and other downstream aggregates.This needs a shared design because the program inputs are not simple annual income fields:
Naively deriving these from annual survey income could impose distributional assumptions that change eligibility and benefit amounts. For example, a uniform annual-to-quarterly allocation can fail or distort quarter-based monetary eligibility tests.
Scope for this follow-up:
unemployment_compensation, remain opt-in model variables, or feed a separate state-benefit aggregate.unemployment_compensationor the appropriate aggregate for all state UI implementations consistently.Related PRs: