OneMain Holdings, Inc. OMF

Key data

The business

OneMain is a personal installment lender to nonprime and subprime households, sold through about 1,400 branches across 44 states and online. Roughly half the book is secured by the borrower's car, which lifts recovery when a loan sours. It charges rates far above a bank's, funds itself entirely in the wholesale bond and securitization markets rather than with deposits, and hands most of the spread back to shareholders through a dividend plus buybacks. The moat is distribution and decades of underwriting data on a customer the banks won't touch, married to a funding scale that keeps its cost of money below smaller rivals'.

The numbers

The quarters show a business that built steadily through 2025 and then gave a little back in the spring.

(a) Q4 2025 net income derived as full-year less prior three quarters; EPS not disclosed in filed data reviewed. (b) Q2 2026 net income not disclosed in filed data reviewed; EPS from earnings release.

Q1 is always the strong quarter, when tax refunds pull charge-offs down; Q2 and Q3 run heavier. So the drop from $1.93 to $1.31 is partly the calendar. What matters is the year-over-year line: Q2 2026 landed below Q2 2025 on both profit and per-share, the first down-quarter comparison in the recovery, and a sign credit cost ticked up rather than down. It still edged past the Street's estimate, extending a run of small beats, but the beat was on a lower bar.

(c) 1H 2026 figures not disclosed in filed data reviewed this run; Q1 and Q2 results shown in quarterly table above.

The 2021 peak is a mirage: pandemic cash kept losses artificially low and will not repeat. The real arc is the slide into the 2024 trough, when normalizing charge-offs cut earnings in half, and the 2025 rebound. Cash tells the calmer story underneath: operating cash flow climbed every single year, from $2.25B in 2021 to $3.13B in 2025, even as reported profit halved and recovered. The earnings swing is provisioning, not the cash engine. Normalized earnings power sits near $7, which is what the consensus $7.09 for 2026 and $8.59 for 2027 assume. At about 9x forward and 4.3x tangible net worth, the market is not pricing a re-rating; it is pricing a high-return lender it distrusts, and paying itself to wait. My read: roughly fair, with the whole argument resting on where charge-offs go next.

Management

CEO Doug Shulman has run this as a capital-return machine, and the record shows it: buybacks jumped to $107M in Q1 2026 alone against just $141M for all of 2025, and every repurchase over the past two years was struck below today's price. Insiders, though, have only sold: zero buys and twelve sales worth $7.86M over the last year, the largest a $2.08M Shulman sale last November and two more in January totaling $2.91M. Plan status is not disclosed in the filings reviewed, so read the selling as neither alarm nor comfort.

How it fails or surprises you

Credit cost re-troughs. The one number that decides the stock is the net charge-off rate. If the labor market softens and losses push back toward 2024 peak levels, provisioning eats the spread and EPS slides toward $4–5, tightening dividend coverage. Q2 2026's year-over-year profit dip is the first tremor. Watch the quarterly charge-off and the 30-to-89-day delinquency trend.

Funding freezes. OneMain has no deposits; it lives on the bond and securitization markets. A shock that widens spreads or shuts new issuance, as briefly happened in 2020 and 2022, spikes its cost of money and caps new loans. Fresh asset-backed note deals in August 2026 prove the window is open today, but windows close fast. Watch new-issue spreads.

Cuts land while credit holds (right tail). If the Fed cuts and charge-offs settle or fall from recent levels instead of climbing, the spread widens from both ends: funding gets cheaper as loan yields hold. EPS then runs past the $8.59 penciled in for 2027, the accelerating buyback shrinks the share count into it, and 4.3x tangible book re-rates on visible high returns. None of that is in a 9x multiple.

Closing thoughts

No single quarter's numbers settle this. The question is whether OneMain's borrowers keep paying through the next twelve to twenty-four months, which the economy decides, not the company. The left tail is a recession that spikes charge-offs and pressures the dividend, though OneMain held that dividend through prior stress. The right tail is a soft landing plus rate cuts, where earnings run and a distrusted 9x multiple lifts. On balance the fatter outcome is benign: entering at 9x forward earnings with a covered dividend and a mid-to-upper-range book multiple, the downside is a dividend-paid wait, not a permanent impairment, absent a funding freeze.

The bet is still that OneMain's working-class borrowers keep paying, that losses ease rather than spike, and that the bond markets keep funding the book. What breaks it is the net charge-off rate and the 30-to-89-day delinquency trend read together each quarter. If those two turn up together for two straight quarters, the earnings-power story is wrong and the dividend math gets tight; if they hold or fall, you are being overpaid to own a 23%-return lender.