Brett Caines is the Co-founder and CEO of Lumos Technologies, a leading small business credit risk analytics firm. Driven by a mission to expand access to financing, Lumos equips lenders with advanced predictive models to accurately assess risk during origination and monitor portfolio migration. Prior to launching Lumos in 2021, Brett served as the Chief Financial Officer of Live Oak Bank. Joining at its inception in 2008, he played a significant role in the bank’s growth and helped guide the company through its IPO in 2015.
When a Bull Walks into a Small Business Credit Portfolio
Relying on outdated, static underwriting methods leaves financial institutions exposed to rising default rates in small business credit.
Brought to you by Lumos Technologies

The phrase “a bull in a china shop” brings to mind a chaotic and unstoppable disaster. This is not a clumsy customer simply bumping into a display. This is hooves slipping on polished floors and a 1,000-pound rear end knocking over fine teacups. It’s total chaos.
In commercial lending portfolios, every small business loan is valuable, meticulously curated and sometimes fragile. Lenders spend years building the shop, carefully evaluating applicants and placing each performing loan neatly on the shelf. Then, a bull walks into the shop.
When the dust settles, portfolio managers stare hopelessly at shattered porcelain. They ask themselves how much time, energy and money it will take to clean it up. The answer is: a lot. But, they are asking the wrong question.
The correct question is: How did a bull get into the shop in the first place?
On quiet days, the shop feels relatively safe. Folks walk the aisles admiring the inventory. Default risks sit quietly in a corner with a smudge or crack here and there, which is nothing a little polishing or epoxy cannot fix.
But between the pandemic hangover of 2022 and the early months of 2026, the economic winds picked up, the market grew volatile and rogue bulls started wandering down Main Street. Default rates did not knock; they walked right in. In fact, the risk of small business loans going into default has increased significantly in recent years. An analysis of the active Small Business Administration 7(a) portfolio by Lumos shows defaults more than doubled from 2022 to 2025. Similarly, conventional small business loan portfolios have seen default rates increase by approximately 80% over the same period.
The bulls did not pick the locks. They walked through the open doors of outdated approaches to predicting risk. When institutions rely solely on slow, inconsistent manual underwriting or static, point-in-time scorecards, they leave the door wide open. A lender approves a loan because the applicant looks like a pristine teacup today, but they fail to measure how fragile that teacup will be the moment an economic bull charges through the room.
A large portion of risk sitting in portfolios right now comes from unseasoned loans originated a few years ago. Now, that porcelain portfolio is maturing in a very different economic environment. If banks do not have proactive, dynamic credit monitoring that spots these risks before the borrower misses a payment, they have left the door open for the bulls.
The good news is that financial institutions do not have to run the shop like this. They do not have to accept shattered porcelain as the unavoidable cost of doing business. They just need to fortify the front door and mind the shop better, employing predictive tools that can wrangle the stampede.
Imagine if a credit team could stand at the entrance, turn away just the riskiest handful of small business loan applicants, and in return, prevent smashed inventory in the future. That is the power of predicting risk over the full life of a loan. It defends the shop from the bulls that inevitably venture into the market from time to time.
By catching the riskiest applications at origination and keeping a watchful eye on the loans already sitting on the books, banks can capture all of the origination volume they want, without inviting disaster.
Manual underwriting must be complemented with data-driven predictive models. Advanced modeling delivers the needed consistency to evaluate every application objectively, alongside the efficiency to move much faster. When lenders can accurately quantify risk right at the front door, they can say yes to more loans with confidence.
But the job is not done once the loan is on the shelf. The right predictive models allow teams to proactively assess existing portfolios, identifying and quantifying hidden pockets of risk before the damage is done. Ultimately, this builds a high-quality, resilient portfolio that leadership deeply understands.
It is time for the industry to stop sweeping up broken teacups. Lock out the risk, and let the bulls doze somewhere else.
