pay-equity-analysis

# Pay Equity Analysis ## Critical Legal Caveat Pay equity work has serious legal implications. **Conduct pay equity analyses under attorney-client privilege** when remediation will follow. Engage qualified employment counsel before running the analysis, not after. The agent flags; counsel decides. Disparate findings without privilege can be discoverable in litigation. Privileged analyses can be remediated discreetly. ## Why Run It The evidence on unexamined pay distributions is consistent: gaps emerge and grow without intentional examination. Hiring biases compound (salary history, negotiation premium, manager pressure). Performance rating biases affect comp adjustments. Time at level biases delay promotion-driven raises. Regular pay equity analysis is the audit; it doesn't fix the underlying mechanisms — but it surfaces them. ## Method (Regression-Based) 1. **Define comparable groups**: similar level, function, geography, tenure 2. **Variables to include**: - Comp (dependent variable; usually total cash, sometimes broken out) - Tenure - Level - Years of relevant experience (if recorded reliably) - Performance rating(s) over recent cycles - Geography - Demographic indicators (gender, race/ethnicity; legally protected per jurisdiction) 3. **Run regression**: comp ~ tenure + level + experience + performance + geo + demographic 4. **Examine demographic coefficients**: - Statistical significance - Magnitude (a 2% gap matters less than a 12% gap) - Direction (which group is paid less) 5. **Investigate before assuming**: legitimate explanations include role composition, recent hire timing, performance rating differences. *But*: those explanations may themselves contain bias (rating bias, promotion bias). ## Axes to Examine - Gender (incl. non-binary) - Race / ethnicity - Age (especially 40+ for ADEA) - Disability status - Parental / caregiver status (where data permits) - Intersectional cuts (with cell-size minimums) ## Cell-Size Minimums For statistical reliability and confidentiality: n ≥ 5–8 per cell. Intersectional cuts often run into cell-size limits — surface in aggregate or qualitatively when statistics aren't reliable. ## Remediation When unexplained gaps exist: 1. **Targeted adjustments**: bring under-paid employees up; don't cut over-paid down (illegal in some jurisdictions; demoralizing in all) 2. **Quartile-by-quartile**: adjust within band positions, not just absolute pay 3. **Communicate carefully**: don't tell each individual the reason if doing so would expose the analysis methodology; instead, frame as comp adjustment within normal cycle 4. **Re-audit**: after the next refresh cycle (6–12 months), confirm closure 5. **Address root causes**: if bias in performance ratings is the driver, fix the rating system ## Adjacent Audits Pay equity is one piece. Also examine: - **Promotion equity**: rates by demographic, time-in-level - **Performance rating equity**: distribution by manager × demographic - **Hiring equity**: offer-to-band ratio at offer - **Equity grant equity**: equity allocations by demographic - **Bonus / variable comp equity**: where applicable A pay equity audit that ignores promotion equity catches only half the picture. ## Common Failures - Running analysis without privilege; findings discoverable - DIY analysis using HRIS data extract without statistical rigor - Examining only gender, missing race / age / disability - Adjustment without addressing root causes (gaps return) - "We have no gap" claim based on a single t-test rather than regression - Vendor providing the analysis without context interpretation - Public commitments without privileged remediation plan ## What Reasonable Looks Like After a properly-run analysis: - Documented methodology - Documented findings (under privilege) - Remediation plan with budget - Implementation - Re-audit on cadence - Continuous monitoring at offer time and comp cycle ## Vendor vs. In-House For larger companies (1000+): in-house people analytics + counsel partnership For mid-size (100–1000): vendor (Syndio, Trusaic, similar) + counsel review For small (<100): often too small for statistically meaningful regression; use comparator review with counsel ## Cross-References - `compensation-strategist` agent - `dei-strategist` agent - `legal-and-jurisdictional-boundaries` rule - `compensation-philosophy` skill ## Key References - Bohnet, I. (2016). *What Works: Gender Equality by Design*. - Industry practitioner work from Syndio, Trusaic, etc. - Federal: Equal Pay Act of 1963; Title VII; Lilly Ledbetter Fair Pay Act 2009 - State patches: California Equal Pay Act, NY, MA, CO, WA, etc. — substantially expanded in past decade