insurance-fairness

FCA Consumer Duty and Equality Act 2010 compliance auditing for UK insurance pricing models — proxy discrimination detection, exposure-weighted bias metrics, and FCA-ready audit reports.

Blog post: Your Pricing Model Might Be Discriminating

---

Your postcode rating factor probably correlates with ethnicity. The FCA's Consumer Duty (PS22/9) requires you to demonstrate this is not producing indirect discrimination under Section 19 of the Equality Act 2010. The FCA's multi-firm review of Consumer Duty implementation (2024) found most insurers' Fair Value Assessments could not do this. insurance-fairness produces the documented, exposure-weighted audit trail your pricing committee can sign off.

This is a compliance audit tool, not a methodology library. It produces evidenced, FCA-mapped analysis that will stand up to an FCA file review. Fairlearn and AIF360 help you satisfy a chosen fairness criterion — they do not answer: "Can I demonstrate to the FCA that this model does not constitute indirect discrimination under Section 19 of the Equality Act?"

---

Why this library?

Task	Manual approach	insurance-fairness
Proxy detection	Spearman correlation — misses non-linear relationships; postcode-ethnicity returns |r| ≈ 0.10, finds nothing	CatBoost proxy R² + mutual information; returns R² ≈ 0.62, unambiguously RED — 100% detection rate across 50 seeds
Exposure-weighted A/E by group	Custom code per model, often not exposure-weighted	calibration_by_group() — exposure-weighted, RAG status, 10-decile grid, Equality Act framing
Log-space demographic parity	Multiply predictions, take ratios — silently wrong for multiplicative models	demographic_parity_ratio() — log-ratio for GLM/GBM multiplicative worlds
Financial impact of discrimination	Not quantified	ProxyVulnerabilityScore — per-policyholder £ cost of proxy exploitation
Action vs outcome fairness (Consumer Duty Outcome 4)	Not possible with standard metrics	DoubleFairnessAudit — Pareto front, Tchebycheff scalarisation, FCA evidence pack
Audit report for pricing committee	Manual Word document, no regulatory mapping	report.to_markdown() — FCA-mapped Markdown with PRIN 2A, FG22/5 cross-references and sign-off table

---

Quickstart

pip install insurance-fairness

import numpy as np
import polars as pl
from catboost import CatBoostRegressor
from insurance_fairness import FairnessAudit

rng = np.random.default_rng(42)
n = 20_000

# Synthetic UK motor book with known proxy structure
postcode_diversity = rng.uniform(0, 1, n)   # ONS LSOA diversity index proxy
gender             = rng.choice(["M", "F"], n)
vehicle_age        = rng.integers(1, 15, n).astype(float)
ncd_years          = rng.integers(0, 9, n).astype(float)

# Postcode diversity drives claims non-linearly — and is correlated with ethnicity
true_rate    = 0.08 + 0.06 * postcode_diversity**2 + 0.01 * vehicle_age - 0.005 * ncd_years
claim_amount = rng.gamma(shape=1.5, scale=true_rate * 300, size=n)
exposure     = rng.uniform(0.5, 1.0, n)

df = pl.DataFrame({
    "claim_amount":        claim_amount,
    "exposure":            exposure,
    "gender":              gender,
    "postcode_diversity":  postcode_diversity,
    "vehicle_age":         vehicle_age,
    "ncd_years":           ncd_years,
})

# Fit a simple pricing model
X = df.select(["postcode_diversity", "vehicle_age", "ncd_years"]).to_numpy()
y = df["claim_amount"].to_numpy()
model = CatBoostRegressor(iterations=100, verbose=0).fit(X, y)
df = df.with_columns(pl.Series("predicted_rate", model.predict(X)))

audit = FairnessAudit(
    model=model,
    data=df,
    protected_cols=["gender"],
    prediction_col="predicted_rate",
    outcome_col="claim_amount",
    exposure_col="exposure",
    factor_cols=["postcode_diversity", "vehicle_age", "ncd_years"],
    model_name="Motor Model Q1 2026",
    run_proxy_detection=True,
)
report = audit.run()
report.summary()
report.to_markdown("audit_q1_2026.md")

See examples/quickstart.py for a fully self-contained example.

---

Installation

pip install insurance-fairness
# or
uv add insurance-fairness

For the Pareto optimisation subpackage (requires pymoo):

pip install "insurance-fairness[pareto]"

Dependencies: polars, catboost, scikit-learn, scipy, numpy, jinja2, pyarrow

---

Features

• Proxy detection — identifies which rating factors (postcode, vehicle group, occupation) are acting as protected-characteristic proxies using mutual information, CatBoost proxy R², partial correlation, and SHAP-linked price impact
• Exposure-weighted bias metrics — calibration by group (A/E ratio), demographic parity ratio (log-space), disparate impact ratio, equalised odds, Gini by group, Theil index; all weighted by earned car-years, not policy count
• Counterfactual fairness — flip protected characteristics and measure premium impact; supports direct flip and LRTW marginalisation
• Proxy vulnerability scoring (ProxyVulnerabilityScore) — per-policyholder £ difference between unaware premium and discrimination-free benchmark; TVaR overcharge, parity cost, fairness range
• Indirect discrimination audit (IndirectDiscriminationAudit) — five benchmark premiums from Côté et al. (2025): aware, unaware, unawareness, proxy-free, parity-cost; no causal graph required
• Multicalibration (MulticalibrationAudit, IterativeMulticalibrationCorrector) — audit and iteratively correct pricing models for multicalibration fairness (Denuit, Michaelides & Trufin, 2026)
• Marginal fairness correction (MarginalFairnessPremium) — closed-form Stage 2 adjustment of Expected Shortfall, Wang transform, or any distortion risk measure; no iterative solver (Huang & Pesenti, 2025)
• Double fairness (DoubleFairnessAudit) — joint action (Delta_1) and outcome (Delta_2) Pareto optimisation addressing Consumer Duty Outcome 4 (Bian et al., 2026)
• Discrimination-insensitive reweighting (DiscriminationInsensitiveReweighter) — training-data weights that achieve X ⊥ A without removing the protected attribute; KL divergence minimisation (Miao & Pesenti, 2026)
• Privatised audit (PrivatizedFairnessAudit) — fairness auditing when protected attributes are estimated from proxies or privatised via local differential privacy (Zhang, Liu & Shi, 2025)
• Optimal transport subpackage — discrimination-free pricing via Lindholm marginalisation, causal path decomposition, Wasserstein barycenter correction
• FCA-mapped audit reports — Markdown with PRIN 2A, FG22/5, Equality Act s.19 cross-references and a sign-off table; suitable for pricing committee packs and FCA file reviews

---

Expected performance

On a 20,000-policy synthetic UK motor portfolio with a known postcode-ethnicity proxy structure:

Metric	Manual Spearman (|r| > 0.25)	Library (proxy R² + MI)
postcode_area flagged as proxy	No (|r| ≈ 0.10)	Yes (proxy R² ≈ 0.62, RED)
Detection rate across 50 seeds	0%	100%
Non-linear proxy detection	No	Yes (CatBoost)
Financial impact quantified	No	Yes

Task	Time (n=50,000 policies)	Notes
Calibration by group (10 deciles)	< 2s	Primary Equality Act metric
Demographic parity ratio	< 1s	Log-space, multiplicative model
Proxy R-squared (per factor, CatBoost)	15–60s	Per factor; subsample for large books
Mutual information scores	< 5s	Catches non-linear relationships
Full FairnessAudit.run() with proxy detection	2–10 min	Produces FCA-ready Markdown report

---

Key modules

bias_metrics

from insurance_fairness import (
    calibration_by_group,
    demographic_parity_ratio,
    disparate_impact_ratio,
    equalised_odds,
    gini_by_group,
    theil_index,
)

cal = calibration_by_group(
    df,
    protected_col="ethnicity_group",
    prediction_col="model_freq",
    outcome_col="n_claims",
    exposure_col="exposure",
    n_deciles=10,
)
print(f"Max A/E disparity: {cal.max_disparity:.4f} [{cal.rag}]")

dp = demographic_parity_ratio(df, "gender", "predicted_premium", "exposure")
print(f"Log-ratio: {dp.log_ratio:+.4f} (ratio: {dp.ratio:.4f})")

proxy_detection

from insurance_fairness import detect_proxies

result = detect_proxies(
    df,
    protected_col="ethnicity_proxy",
    factor_cols=["postcode_district", "vehicle_age", "ncd_years", "driver_age_band"],
    run_proxy_r2=True,
    run_mutual_info=True,
    run_partial_corr=True,
)
print(result.flagged_factors)  # ['postcode_district']
print(result.to_polars())      # sorted by proxy R-squared

double_fairness

Action fairness (Delta_1) and outcome fairness (Delta_2) can conflict. Minimising premium disparity does not minimise loss ratio disparity. DoubleFairnessAudit recovers the full Pareto front:

from insurance_fairness import DoubleFairnessAudit

audit = DoubleFairnessAudit(n_alphas=20)
audit.fit(X_train, y_premium, y_loss_ratio, S_gender)
result = audit.audit()
print(result.summary())
print(audit.report())   # FCA Consumer Duty Outcome 4 evidence section

---

Limitations

• Proxy detection thresholds are not FCA-prescribed. Proxy R-squared thresholds (amber: >0.05, red: >0.10) are operationally derived. A factor below the red threshold may still constitute indirect discrimination — the statistical test is a trigger for investigation, not a compliance safe harbour.
• Proxy detection requires a protected characteristic column. Most insurers do not hold individual ethnicity data. Joining ONS 2021 Census LSOA ethnicity proportions to postcode is an area-level proxy, not individual-level.
• Calibration by group is not counterfactual fairness. Equal A/E within each protected group means the model correctly prices the proxy-contaminated risk profile — not that it prices independently of the protected characteristic.
• The audit report is evidence, not a compliance determination. It does not constitute legal advice or a regulatory safe harbour.

---

References

• FCA. (2023). Consumer Duty: Final rules and guidance (PS22/9). fca.org.uk
• Equality Act 2010, s.19 (indirect discrimination). legislation.gov.uk
• Lindholm, M., Richman, R., Tsanakas, A. & Wüthrich, M.V. (2022). "Discrimination-Free Insurance Pricing." ASTIN Bulletin, 52(1), 55–89. doi:10.1017/asb.2021.23
• Côté, O., Côté, M.-P. & Charpentier, A. (2025). "A Scalable Toolbox for Exposing Indirect Discrimination in Insurance Rates." CAS Working Paper.
• Denuit, M., Michaelides, M. & Trufin, J. (2026). "Multicalibration in Insurance Pricing." arXiv:2603.16317
• Miao, K.E. & Pesenti, S.M. (2026). "Discrimination-Insensitive Pricing." arXiv:2603.16720
• Zhang, Y., Liu, Y. & Shi, P. (2025). "Discrimination-Free Insurance Pricing with Privatized Sensitive Attributes." arXiv:2504.11775

---

Part of the Burning Cost toolkit

Takes a fitted pricing model and a dataset with protected characteristics. Feeds audit reports and proxy detection results into insurance-governance for pricing committee sign-off packs and FCA Consumer Duty documentation. → See the full stack

Related libraries

Library	What it does
insurance-causal	DML causal inference — isolate genuine causal price effects from proxy correlation before auditing for discrimination
insurance-causal-policy	SDID and Doubly Robust Synthetic Controls — establish whether a rate change caused an outcome shift
insurance-governance	Pricing committee sign-off packs and FCA Consumer Duty documentation
insurance-monitoring	Post-bind drift detection — flag when protected-group A/E ratios are shifting in the live book

All libraries | Discussions | Issues

Need help implementing this in production? Talk to us.