Case Study: ANZ New Zealand achieves improved, explainable mortgage credit-risk performance with FICO's machine-learning solution

Leading New Zealand bank pioneers explainable machine learning models to assess credit risk

ANZ New Zealand, the country’s largest full-service bank with a heavy mortgage and retail-lending portfolio, wanted to determine whether robust, explainable machine‑learning (ML) techniques could improve credit risk decisions. The bank’s challenge was to prove ML could deliver measurable uplift while meeting strict requirements for explainability, accuracy, ease of implementation and stakeholder confidence.

ANZ partnered with FICO to build ML scorecards and used FICO’s Explainable AI to interpret results, running proof‑of‑concept, full development and parallel production tests alongside its traditional regression models. The ML approach produced measurable lift over traditional models, generated intuitive reason codes, revealed opportunities to simplify policy rules, and is being progressed toward production and wider use across other portfolios.

ANZ

Vimit Kapoor

Head of Retail Models