Replace SOEP 'not applicable' codes with 0 for income/wealth variables

[MImmesberger]

Summary

• Add _replace_not_applicable_with(series, value) utility in data_manipulator.py that maps SOEP code -2 ("trifft nicht zu" / "does not apply") to a concrete value (typically 0) for income variables, while preserving genuinely missing codes (-1, -3 to -9) as NA
• Apply to all income, pension, benefit, transfer, and cost variables across 6 clean modules:
  • pequiv.py: HH income, all pension types, benefits (kindergeld, wohngeld, ALG, etc.), transfers (unterhalt, BAföG), labor earnings, bonuses, costs
  • pgen.py: gross/net labor income
  • pl.py: net income average, maternity benefit, child support, disability degree
  • hgen.py: heating costs
  • hl.py: kindergeld, kinderzuschlag, wohngeld, ALG2, grundsicherung im alter
  • kidlong.py: childcare/school costs
  • pkal.py: unemployment months

Motivation

Previously, all SOEP negative codes (-1 through -8) were uniformly converted to NA. This lost the distinction between "no answer" (genuinely missing) and "does not apply" (code -2, logically means 0 for income).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!