Chief Strategy Officer / United States
Choices that cost income now to avoid losses later
- Total assets
- US$58bn
- Varrowfield Financial Corporation, June 2026
- Uninsured share of deposits
- 46% → 33%
- December 2022 to June 2026, after the deposit franchise review
- Seller deposit attrition before closing
- 3%
- Kessling Bancshares, July 2024 to June 2025; 12% at Dunmarrow in 2019
Decisions & consequences
01Investing half of the deposit surge, and short
At the end of 2021 Varrowfield held US$9.6 billion of cash, mostly from deposits that arrived after early 2020. Total deposits stood at US$43 billion, against US$35 billion just after the Dunmarrow acquisition, and our net interest margin had fallen to 2.71%. The treasurer proposed investing US$6 billion in agency mortgage-backed securities with a duration of about five and a half years, which at the yields then available would have added roughly US$70 million of annual income.
Three months into my new role, I had my team compare each commercial and private-banking client's balance with its average before 2020, the method I first tried at Sorrelton. About US$5.1 billion of the increase sat in accounts holding more than twice their usual balance, with no new treasury service attached. I argued that this money would leave as rates rose, and that investing it for five years meant funding long assets with balances we had no reason to keep.
The treasurer disagreed in writing. He pointed out, fairly, that our deposit models had understated stability for a decade, and that holding cash at near-zero rates was a certain cost set against an uncertain risk. The asset-liability committee settled on a compromise: US$2.5 billion invested in a ladder of securities maturing within three years, and the remainder held in cash and short Treasury bills. We gave up about US$34 million of 2022 net interest income compared with the full proposal, and I spent much of that year explaining why our margin trailed the peer group.
By March 2023 after-tax unrealised losses on our securities equalled 19% of tangible common equity, against a peer median near 31%. When US$2.9 billion of deposits left in eight business days, we paid it from cash and maturing securities without selling at a loss. Our interest-rate model now carries a separate, shorter assumed life for surge balances.
What I would do differently is the funds-transfer pricing. In 2021 our business lines were still credited for the surge deposits as though they were long-lived, so relationship managers had every reason to defend them as core. I should have pressed to change that credit at the same meeting as the investment decision. It took until mid-2023, which made the later deposit review harder.
Sorting deposits by the work they do, and leaving escrow balances
The deposits that left Varrowfield in March 2023 were not the ones our models had marked as volatile. Of the US$2.9 billion withdrawn in eight business days, about 60% came from commercial and private-banking clients who had banked with us for more than seven years. Most held balances well above the insured limit and used us for little beyond the account. The largest single group was title, escrow and property-management firms: thirty-six clients holding US$2.6 billion between them, priced through earnings credits that moved with market rates.
In May 2023 our chief executive asked me to lead a review of the deposit franchise. We classified every commercial balance by the services attached to it and found that only 59% sat in relationships using at least one treasury-management service. I recommended three changes: credit business lines through funds-transfer pricing according to that classification; invest US$48 million over three years in treasury-management products and implementation teams; and exit the escrow and title concentration within twelve months.
Our head of commercial banking, Teodoro Lisk, opposed the exit. He argued that these clients had stayed through the worst week and that losing billions of deposits while funding was expensive would damage our margin. He was right about the cost. Replacing the balances with home loan bank advances and brokered deposits added about US$31 million of interest expense across 2023 and 2024, and the business gave up roughly US$11 million a year of fee and float income.
We kept nine of the thirty-six firms, those whose payroll and operating accounts ran through us, and moved US$2.2 billion out by June 2024 with at least ninety days' notice to each client. By June 2026 uninsured deposits had fallen from 46% of the total to 33%, the share of commercial balances in relationships with a treasury-management service had risen to 78%, and treasury-management fees had grown from US$94 million in 2022 to US$131 million in 2025.
What I would do differently is the order. We announced the exit before the new funds-transfer pricing reached relationship managers' scorecards, so for two quarters bankers were told to shed deposits while still being paid to gather them. Two of our stronger commercial teams spent that period defending clients instead of winning new operating accounts.
Kessling Bancshares: an acquisition built for an eleven-month wait
In July 2024 Varrowfield agreed to acquire Kessling Bancshares of Albuquerque, a US$8.9 billion bank with thirty-four branches in New Mexico and southern Colorado and a strong book of business-banking clients, for US$1.18 billion in stock. Review periods for bank mergers had become longer and harder to predict. At Dunmarrow an expected five months had become eleven, and the seller had lost 12% of its deposits and twenty-three of sixty-one commercial bankers while it waited, partly because our agreement had restricted its hiring and lending.
I proposed designing the transaction for a twelve-month review rather than hoping for six. That meant three decisions. We would offer, before filing, to divest seven branches in two New Mexico counties where our combined deposit share was likely to attract competitive objections. We would loosen the interim operating covenants so that Kessling could keep hiring bankers, renewing loans within its own credit policy and pricing deposits without our consent. And we would publish a five-year community lending and investment plan with the application.
Our chief financial officer opposed the early divestiture, which surrendered US$610 million of deposits and a negotiating position. Our chief credit officer opposed the looser covenants, because Kessling could make loans we would not have made and we would inherit them. The chief executive and the board's risk committee accepted both after my team modelled the cost of a slow closing against the value of the branches.
Approvals arrived in May 2025 and the deal closed on 30 June, eleven months after announcement. Kessling's deposits fell 3% over that period, and four of its fifty-two commercial bankers left. The seven branches went to a community bank based in Santa Fe. We later added about US$14 million to loss provisions for loans Kessling made under the looser covenants, which I count as part of the price of the approach. Systems conversion followed in February 2026.
What I would do differently is the community plan. We wrote it with our own staff and published it with the filing, and a coalition of Albuquerque community organisations submitted a comment letter arguing, with some justice, that nobody had asked them what their neighbourhoods needed. Meeting them added about six weeks to the review and produced a better plan. Those meetings belonged before the announcement.
Closing four partner-banking programmes I had approved
In 2019, as head of corporate development, I proposed that Varrowfield hold deposits for the customers of fintech companies, earning a share of card interchange and a steady deposit stream. By June 2022 we had six programmes, 2.1 million end-user accounts, US$1.3 billion of deposits and about US$19 million of annual revenue. Four of the programmes relied on a middleware provider that kept the detailed ledger of each end user's balance, while we held one pooled account per programme.
In the second quarter of 2022 internal audit reported that for those four programmes we had been unable to reconcile end-user balances to our pooled accounts on eleven business days in the quarter, and that complaints routed through the partners reached us an average of nineteen days late. The partners and the provider offered remediation plans. Our head of digital banking argued for giving them six months, noting that the programmes were profitable and that rival banks were courting the same partners.
I recommended winding down all four within nine months and keeping the two programmes in which we held the end-user ledger ourselves. My reasoning was that we were responsible for customers' money we could not see, and that remediation would leave that structure unchanged. I also told the risk committee that the contracts had been mine, and it accepted the recommendation. We gave up about US$12 million of annual revenue and US$820 million of deposits, which left over the following nine months as funding costs rose.
The wind-down was completed in March 2023 with every end-user balance reconciled before transfer and no customer losing money. The two remaining programmes now operate under daily reconciliation, partner-funded compliance budgets and rehearsed exit plans, and had grown to US$740 million of deposits by June 2026. During 2024 several sponsor banks elsewhere came under enforcement actions, and the collapse of one middleware provider left end users unable to reach their money for months.
What I would do differently is the original design. In 2019 I priced the programmes on interchange and deposit value and treated compliance as a cost the partners would carry. I should have required our own ledger from the first day and built the full compliance cost into the price. That would have made two of the six uneconomic, which was probably the right answer.