Redlining by postcode: the credit practice hiding in plain sight.
Redlining has an unhappy history. It began as literal red lines on a map, marking the neighbourhoods where banks would not lend. We would like to think it belongs firmly to the past. In credit scoring, a quieter version of it has proved harder to leave behind.
A modern decision rarely says, in so many words, that a suburb is unwelcome. It does not need to. When a scoring model draws on area-level patterns, a postcode can shape the outcome, standing in for information the model does not actually hold about the person.
How it happens
The mechanism is straightforward. A traditional score is built from a limited set of signals: credit taken, applications made, repayments met. When those signals are thin, models lean on proxies, and geography is a convenient one. So where a person lives can influence the offer they receive, even when their own finances tell a more favourable story.
Two applicants with the same income, the same savings and the same repayment record can be treated differently because of the suburb attached to their file. That is not a judgement of either of them. It is an average, applied to an individual.
Why it lingers
It lingers partly because it is easy to miss. No one signs off a “postcode penalty”. It is built into the mathematics, and the mathematics is assumed to be neutral. It lingers, too, because there is usually something louder to attend to, so a slow, structural unfairness is left where it sits.
On the record
In 2022 ABC News reported that one of the major credit bureaus had begun using patterns in how people in an area behave with credit, in individual credit scores. In its own words, the data forms “a small component” of its credit scores, used “in limited circumstances”. The consumer group Choice warned that it risks proxy discrimination, since a postcode tends to correlate with income, age and background. A small component, then, and we should be precise about that. But the principle is the point: where you live, and not only how you behave, can shape the offer you receive.
The cost
For borrowers, capable applicants in the wrong postcode receive weaker offers, or are turned away. For lenders, decisions rest on a proxy rather than the person, which is a fairness risk and a commercial one at once. You decline customers you would gladly have approved, had you been able to see them clearly.
A better way
The answer is not to debate which poor proxy variable is fairest. It is to remove the need for poor proxy variables. When you can see an applicant's real income, cashflow and commitments from their banking, geography stops standing in for character, and you assess the person themselves.
That is what TaleFin is built to do: replace the proxy with the real thing, so an address is simply an address, and not a verdict.