What Tricolor Revealed About Risk That Wasn’t Gone

The Federal Reserve’s evidence on buy-here-pay-here (BHPH) lending presents a conditional advantage, not a clean safety story. The question is how much confidence to place in lower bank-assessed risk when the underlying consumer loans show higher delinquency and repossession, and what changed after Tricolor’s bankruptcy. Reading the matched sample carefully lets us distinguish three things that are easy to conflate: what the data directly compares, what structural protections might explain, and what a stress event can reveal about risks that the structure did not eliminate.

What the matched sample can actually compare

The Federal Reserve cannot identify every BHPH dealer directly in the consumer credit registry. For its corporate-loan analysis, it instead name-matches loan obligors in the FR Y-14Q data to individually known BHPH dealerships and their local subsidiaries. The resulting set covers approximately a dozen BHPH auto lenders and 82 obligors, includes more than a dozen banks over time, and represents more than $2 billion in commitments as of the end of 2025. It includes large dealers such as DriveTime, Byrider, and America’s Car-Mart, as well as the now-defunct Tricolor.

That construction is useful because the corporate data contains bank-assessed risk characteristics that do not appear in the consumer loan records. It also creates a selection boundary. A name match can find a facility when the obligor’s corporate identity is recognizable; it can miss a line of credit held through an entity with an opaque name or a complicated structure. The Federal Reserve specifically notes that special-purpose entities can make identification difficult. The authors therefore warn that the matched list is not an exhaustive inventory of BHPH dealers or bank funding.

The comparison group has its own definition. It consists of other auto-dealer obligors that received an auto floor-plan loan at least once in the data and had an industry classification associated with auto dealers or sales financing, while excluding the identified BHPH dealers. That is a more relevant comparator than an undifferentiated sample of every corporate borrower because it places BHPH facilities alongside other businesses with auto-dealer inventory financing. But it is still a control sample constructed from the reporting data, not a perfect twin for each BHPH dealer.

The Federal Reserve’s interpretation also relies on an assumption: the facilities it identifies are broadly representative of overall bank lending to BHPH dealers. That assumption may be reasonable enough to make the pattern informative, but it should remain visible. The evidence establishes what happens in the matched facilities and the defined control group. It does not establish the risk profile of every BHPH credit line, every nonbank funding source, or every dealer that the matching process could not identify.

The distinction is similar to looking through a window rather than removing the wall. The window can show a real part of the system with enough detail to compare it. It cannot show whether the unseen part has the same composition. A lower risk assessment in the observed facilities is meaningful evidence; it is not evidence that no different or riskier facilities exist outside the sample.

Lower assessed risk is not lower consumer risk

Within the identified bank facilities, the Federal Reserve reports that BHPH dealers had a slightly lower predicted probability of default than other auto dealers from 2020 onward.