For decades, lenders have treated employment as a proxy for stability. But more Americans are piecing together income from hourly work, contract jobs, commissions and gig platforms that can produce dramatically different paychecks from one week to the next.
The trend is reshaping one of banking’s oldest disciplines. Rather than asking only whether borrowers have jobs and acceptable credit scores, lenders increasingly have the ability to measure whether income itself is predictable.
As payroll deposits, transaction histories and cash-flow data become richer, underwriting can evolve from a snapshot taken at loan origination into a continuous assessment of financial behavior.
The Wage to Wallet™ Index report “Side Work Patterns in the Labor Economy” finds that roughly 60 million Americans work in what PYMNTS Intelligence, WorkWhile and Ingo Payments define as the Labor Economy, earning hourly wages of $50,000 or less annually. While job security has improved for many workers, financial confidence has not kept pace because cash flow has become far less predictable.
Nearly one in five Labor Economy workers performed regular side work during the previous six months, while many rely on several income sources that arrive at different times and in different amounts. Those workers typically maintain fewer income streams than higher earners, leaving them more exposed when one source slows or disappears.
Forty-one percent of Labor Economy workers report feeling financial stress often, compared with 27.2% of higher-income workers. More than two in five use side income simply to cover everyday living expenses rather than to build savings, and 22.7% experienced an unexpected household income interruption during the previous year.
Those figures point to an underwriting problem that conventional credit scores are ill-equipped to solve.
Credit scores summarize past repayment behavior. They reveal relatively little about whether a borrower whose income fluctuates weekly remains capable of managing those fluctuations responsibly. A worker who drives for a delivery platform, picks up seasonal shifts or supplements wages with freelance work may present irregular cash flow without presenting elevated credit risk.
For banks, distinguishing between those two situations is becoming commercially important.
From Static Underwriting to Agentic Credit
Financial institutions possess considerably more information than they did when traditional underwriting models were developed.
Recurring payroll deposits, direct deposit histories, payment timing, spending behavior and transaction data provide a far richer picture of financial stability than periodic income verification alone. There’s the raw material at hand to understand how consistently income arrives, how consumers manage uneven cash flow and whether temporary disruptions translate into lasting financial weakness.
That thinking underpins “The ABCs of AI Credit,” a PYMNTS Intelligence report with Thredd examining the emergence of agentic credit. Rather than relying primarily on fixed rules and historical scorecards, the report argues that credit decisions are moving toward real-time transaction intelligence embedded directly into payment flows.
AI agents are a boon in continuously evaluating behavioral context, transaction patterns and external data to support payment and credit decisions. Instead of producing only binary approval or declining outcomes based on static thresholds, these systems assess intent, spending patterns and changing financial circumstances as transactions occur.
That capability matters because volatile income should not automatically be interpreted as weak credit.
Consumers with multiple employers or project-based work may experience significant swings in weekly deposits while maintaining healthy financial habits over time. Looking beyond traditional scorecards allows lenders to identify borrowers whose earnings fluctuate but whose overall financial behavior remains consistent.
The potential benefits extend across lending operations.
Agentic credit can improve authorization rates by identifying legitimate transactions that conventional fraud rules might reject. It can reduce false declines, lower operating costs by limiting manual reviews and strengthen defenses against sophisticated fraud schemes by evaluating customer behavior continuously instead of periodically.
Institutions capable of separating income volatility from genuine credit deterioration may be able to expand lending responsibly, refine pricing, preserve customer liquidity and reduce unnecessary declines without accepting additional credit risk.
As more Americans earn their living through combinations of hourly work, digital platforms and contract income, underwriting may become less dependent on what borrowers earned last month than on how consistently they navigate the uncertainty of the next one.
