A financial research study across 60 regional credit unions examined the implementation of machine-learning credit risk algorithms and subsequent small-business loan default rates. The study observed that credit unions implementing machine-learning algorithms experienced a 40 percent lower default rate on small-business loans over a three-year period than credit unions relying strictly on traditional manual underwriting. Skeptical financial analysts contend that the algorithm itself did not cause the lower default rate. Instead, they argue that credit unions adopting the algorithm had systematically raised their minimum credit score requirements for all loan applicants immediately prior to software installation, thereby selecting an inherently lower-risk applicant pool.
Based on the information provided, select the statement that most strengthens the financial analysts' alternative explanation, and select the statement that most weakens the financial analysts' alternative explanation (thereby supporting a direct causal relationship between algorithm adoption and reduced defaults).
- Strengthens Analysts' Alternative ExplanationCredit unions that raised minimum credit score requirements without installing the machine-learning algorithm achieved small-business loan default reductions equal to those of credit unions that installed the algorithm.
- Weakens Analysts' Alternative ExplanationCredit unions adopting the machine-learning algorithm observed substantial default reductions even among approved borrowers whose credit scores were below the newly raised minimum threshold via special waiver programs.