To unpack a new Mercator Advisory Group white paper released this week, Credit Scoring, Fintech, and Consumer Loans: Why AI Scoring Models Do Not Replace the FICO Score, PaymentsJournal sat with Mercator’s Director of Credit Advisory Services Brian Riley to hear more about the origins of FICO’s scoring system and his perspective on its contrasting properties compared with a host of novel alternative scoring models.
Regulated Data Is the Foundation
Riley was careful to note the baseline for understanding: they are a measure of risk derived from data required by the Fair Credit Reporting Act, and other regulatory requirements. “The FICO Score only uses data prescribed by the Fair Credit Reporting Act,” Riley explained. That data, which issuers supply to credit bureaus, does not contain any personal identification information (PII) or subjective content.
As Riley put it regarding the data’s cleanliness, as true at the launch of the scores in 1989 as today, “What they did was very clever; they focused the FICO Score on data that’s required by credit issuers to be submitted to credit bureaus for reporting. By doing that, they set an important precedent — putting only appropriate information into the score.”
Put another way, sex, age, ethnicity, religion, and other PII do not contribute to the score’s calibration.
The Five Components[i]
- Payment history: 35% is determined by a consumer’s track record of paying their accounts over time.
- Amount owed: 30% is determined by how much debt a consumer carries in total.
- Length of credit history: 15% is determined by the duration of credit history.
- Credit mix: 10% is determined by the various types of credit a consumer might have (e.g., credit cards, mortgage, installment loans).
- New credit: 10% is determined by the recency with which a consumer has applied for new credit.
Riley explained the evolution of FICO Scoring, first as a measure of risk, then transforming into a means of simplifying underwriting and account management. FICO keeps its score relevant by carefully adding in some variables, as it did with the recent FICO Score 9 and FICO Score 10 Suite. Some lenders are now testing FICO 9 and FICO Score 10, which reduces the reliance on medical account collections. It also picks up rent data when reported. The score used most widely is FICO Score 8.
The FICO Score Is an Adaptive, Independent, Future-Proof Metric
“What the FICO Score did was take those five elements [above] and develop predictive score cards to create a standard risk metric to be used across consumer credit decisions. That metric goes from 300 to 850, and consistently ranks risk across consumer lending products and over market cycles.
Suffice it to say, by consolidating credit bureau data into a single, trackable statistic, the FICO Score delivered enormous efficiency to card portfolio management. So useful in fact, that FICO Scores emigrated from a back-office utility to a front-end consumer-facing tool. As the metric became better understood by card issuers, its utility grew within the card-issuing banks. The efficiency of the scoring technique and dependability of its underlying data developed appealing consumer-facing applications, “the application [of FICO Scores] transcends the risk management side, until now it’s at the front end,” used by investors and countless financial intermediaries at the consumer-facing inauguration to examine credit readiness.
FICO Score: Don’t Get Fancy and Stick to the Facts
Proponents of furnished data might point to a growing list of innovations and adaptations that have been incorporated to create an entry point to credit access for those without scorable credit histories. UltraFICO Score expands on the score built on traditional credit data by empowering consumers to link checking, savings, or money market accounts to their scores. The firm estimates that 15 million consumers in the U.S. could benefit. FICO Score XD leverages alternative data such as mobile and landline phones, utilities, and subscription service payments are incorporated into FICO Score XD. The through-line of these evolutionary advancements is “reliable data” from trusted partners combined with a scoring technique that stays true to the original score’s design, 30 years successful.
In his closing remarks, Riley offered three poignant reminders:
- The role of a credit score is to provide a metric that consistently and reliably predicts a consumer’s credit risk across economic cycles.
- The score is fair and consistent. A consumer with a 720 FICO Score that has only credit card accounts has similar risk to one with multiple credit types. The same thinking applies to a consumer with a FICO Score of 660, whose risk profile will be more concerning than that of a consumer with a 729 FICO Score.
- Regarding the economic clouds on the horizon, “[FICO] has been around since 1989 and invested years of testing before that. It withstood many economic cycles, and is founded on reported data. As you go through a changing economy … all of a sudden, we don’t know how [alternative models will] react when inflation is at eight percent. We don’t know what happens when interest rates jump up 50 basis points.”
It is easy to underappreciate the standardization that the FICO Score when times are good. But markets, business and home buying are all built on confidence. When the world gets uncertain, it is reassuring that trusted, tested standards like the FICO Score can help rebuild confidence.