The Math That Broke the System
Ten lenders expelled from the USDA's rural lending program last week held $620 million in delinquent loans, 47% of the agency's total defaults, despite representing just 1.3% of participating lenders. Secretary Brooke Rollins called it proof of "absolutely no tolerance for the irresponsible and noncompliant actions of these lenders." But the numbers tell a different story: this isn't accountability working. It's a structural failure finally becoming visible.
The removed lenders, including Bank of Montgomery, Byline Bank, Celtic Bank, and ReadyCap Commercial, operated within the OneRD Guaranteed Lending Program, where the government absorbs default risk while private lenders collect origination and servicing fees. When nearly half your program's bad debt concentrates in ten institutions, the question isn't whether those lenders behaved badly. The question is what incentive structure made that behavior profitable enough to pursue at scale.
How the Guarantee Machine Works
Government loan guarantees separate risk from reward in a way that creates predictable problems. A lender originates a loan to a rural business, farmer, or homeowner. The lender collects fees for processing and servicing that loan. If the borrower defaults, the USDA guarantee means taxpayers cover the loss, not the lender who approved the loan.
The profit logic becomes straightforward: maximize loan volume, minimize underwriting scrutiny, extract maximum fees before defaults materialize. By the time a loan goes bad, the lender has already been paid. The taxpayer eats the loss. The expelled lenders didn't break this system, they optimized for it.
Over 750 lenders remain in the OneRD program, operating under identical incentives. Removing ten bad actors doesn't change the architecture that rewarded their behavior. The guarantee structure that allowed $620 million in concentrated defaults remains untouched.
The Pattern Across Programs
This isn't the first time government guarantees have produced this outcome. FHA mortgage insurance in the 2000s created similar dynamics: lenders originated subprime loans, collected fees, then watched from a safe distance as defaults cascaded through the housing market. Student loan guarantees produced a generation of for-profit colleges that maximized enrollment and federal loan volume while graduation rates collapsed. The rural lending program is the same mechanism with different borrowers.
The USDA's announcement frames expulsion as decisive action, "strengthening commitment to responsible lending practices and program integrity," according to the agency. But integrity requires more than removing the worst performers after losses accumulate. It requires designing systems where the entity making lending decisions shares meaningful exposure to the consequences of those decisions.
When $620 million in guarantees get called, taxpayers fund the difference between what borrowers can repay and what lenders originally disbursed. That's not an abstraction, it's budget authority that could have funded other rural development priorities, now redirected to cover losses from loans that should never have been approved.
What Expulsion Doesn't Fix
The ten expelled lenders are barred from future participation in USDA guaranteed lending programs. Their existing loan portfolios remain in place. Borrowers who took loans from Celtic Bank or Genisys Credit Union still owe that money. The delinquent loans still sit on USDA's books. The guarantee obligations still exist.
What changes is that these particular institutions can no longer originate new guaranteed loans. The 750 remaining lenders can. They operate in the same incentive environment that produced the problem. They face the same profit opportunity from high-volume, low-scrutiny lending. They enjoy the same guarantee protection if loans go bad.
Secretary Rollins positioned the expulsions as evidence that irresponsible lending "has no place in government programs." But tolerance wasn't the variable that mattered. Incentives were. As long as lenders profit from origination while taxpayers absorb default risk, the structural logic that produced $620 million in concentrated bad debt remains operational.
The USDA maintains a public database called Lender Lens where participants in the OneRD program are listed. The ten expelled institutions no longer appear there. The 750 others do. The guarantee structure that made their removal necessary continues unchanged, waiting to reveal which lenders will optimize for volume over quality, and how long it will take before the next round of expulsions becomes necessary to address defaults that are profitable right up until they're not.