Casebook

Apple Card credit limits

The control case. The model was examined and cleared; the failure was that the person on the receiving end had no reasons and no route, and the people answering the phone had neither either. Most variables held. The two that did not are the two the Contestability standard exists for.

System
Automated credit-limit decisions on a consumer card issued by Goldman Sachs, with no route by which an applicant could learn why their limit differed from a spouse's or ask for the decision to be reconsidered.
Where
United States · New York Department of Financial Services
When
November 2019 to 2021
Scale
No unlawful discrimination found. The regulator's finding was that applicants, and the bank's own staff, could not explain individual outcomes, and that no reconsideration path existed.
Time to halt
About seventeen months to a policy change; the model was not withdrawn.
Halted by
Nobody. The issuer changed its policies after a regulator's investigation found the process, not the model, deficient.

The record

What happened

From the primary sources listed at the end. Where a finding is quoted, it is the inquiry's, not ours.

In November 2019, several applicants reported publicly that they had been offered credit limits many times higher than their spouses', despite shared finances and, in some cases, the spouse's better credit history. Customer service representatives could not explain the outcomes and, by the applicants' accounts, said the algorithm had decided. The New York Department of Financial Services opened an investigation.

The Department's report, published in March 2021, found that the underwriting model did not use sex or marital status and that the outcomes could be explained by differences in the applicants' individual credit files, including that a spouse who was an authorised user on the other's accounts had a thinner history. It found no violation of fair lending law. It also found that neither applicants nor the bank's staff had been able to obtain that explanation at the time, that there was no process to request reconsideration of a limit, and that the bank's reliance on individual credit data disadvantaged spouses whose finances were shared but whose credit histories were not.

The issuer subsequently introduced the ability for spouses to share an account and build credit jointly, and a reconsideration process. The model was not changed.

Scoring

The six state variables

One finding per variable, with the clauses whose obligation it falls under and the law it instantiates.

  1. held

    Capability

    What could the assembled system actually do?

    The model set limits. It did not do anything it had not been authorised to do, and its inputs were declared to the regulator.

    STD-07 §2.2 Law I

  2. held

    Authority

    Which actions was it permitted to take, for whom, until when?

    The issuer held the authority to set limits and did so within fair lending law. The regulator examined the authority and confirmed it.

    STD-07 §2.1 Law II

  3. drifted

    Evidence

    What propositions justified that authority?

    The evidence for each decision existed, in the credit file, and was produced to the regulator. It was not produced to the applicant, and the bank's own front line could not retrieve it. Evidence that the operator holds but cannot surface at the moment of the decision is evidence in name.

    STD-02 §1.2 STD-02 §2.2 Law III

  4. held

    Dependency

    How hard had withdrawal or substitution become?

    A credit limit is revisable and the issuer revised its policy. Nothing about the deployment made withdrawal or change costly.

    STD-06 §5.1 Law V

  5. failed

    Standing

    Who could challenge a decision, with what procedural force?

    There was no reconsideration path. An applicant could complain, and the complaint could be logged, but there was no route by which the complaint produced an answer against a stated standard or a possible change to the decision. The regulator's finding was about this, not the model.

    STD-02 §6.1 STD-02 §8.1 STD-02 §8.3 Law VII

  6. held

    Correction

    Which interventions stayed feasible, and on what clock?

    The issuer changed its policies within months of the report. The correction was institutionally feasible and was exercised without a court. That it took a regulator to prompt it is the drift; that it happened is the difference between this case and the other four.

    STD-02 §5.3 Law IV

Learning outcome

Where the failure trajectory ended

Handled or learned: did the sequence from failure to challenge end in case resolution, or did the exception change the machinery that produced it?

learned — exception absorption versus exception learning

The institution revised its machinery. The regulator found the process deficient, and the issuer responded by changing the process: it introduced a reconsideration path for credit-limit decisions, so that the error-bearing party could contest an outcome against a stated standard, and it changed the product so that spouses could share an account and build credit jointly. The model was not changed, but the workflow and the standing that surround it were. This is the only case in the casebook where the correction was exercised by the operator rather than imposed from outside, and the distinction is what made the episode end in learning rather than absorption.

Read the theory: exception learning · Corrective Learning eval suite

The missing record

The one object that was not there

Every case reduces to a record the standards require and the operator did not hold. This is that record.

A standing register entry for credit-limit decisions: that an applicant may ask for reasons, that the reasons name the inputs that moved the limit, and that a reconsideration is answered within a deadline. STD-02 §8.1 and §8.5 describe a process the issuer built after the fact.

Sources

The primary record

Courts, statutory inquiries, regulators, and parliaments, in that order of weight. No press coverage.

  1. Regulator New York State Department of Financial Services, Report on Apple Card Investigation · 2021-03-23