A recent "SMBC US Rates Strategy" by Joseph Abate asks, "Will AI wake up cash?" It explains, "AI agents make it easier to search, compare and move cash between alternatives. It simplifies these tedious tasks and removes frictions that keep balances from moving to better, more efficient allocations. Until now, comparison shopping across rates was easy but re-aligning payment flows, like direct deposits or automatic bill pay was more difficult. As a result, cash tends to 'sleep' -- lingering in sub-optimal allocations because it is too costly to move into better alternatives. But AI could mean that it awakens." (Note: Register soon for our upcoming Money Fund University, which will take place Dec. 17-18 in Greenwich, Conn.. Subscribers and friends are welcome to join us for our Holiday Cocktail Party Thursday night, Dec. 17, at the Hyatt Regency Greenwich!)
Abate says, "Roughly 20% of US domestic deposits (or about $4trn) sits in non-interest-bearing accounts despite the Fed's aggressive 2022-23 tightening campaign. The departure of this cash for higher yields has been the focus of several articles. Departure risk may not be limited to banks; money funds hold $8trn despite expenses of around 25bp and 7d-returns below 1m bill rates."
He speculates, "If widely adopted, AI agents could increase cash efficiency -- effectively 'waking it up'. Personal assistants could be programmed move balances to banks with the highest deposit rates or to regularly sweep idle cash in a checking account to a higher-yielding savings deposit and then return it when payments are made. But while the focus has been on yield enhancement, we think there are other aspects of somnolent cash management that could be improved."
SMBC's piece continues, "AI agents could optimize payments and make it easier to secure deposit insurance above the FDIC's cap. Assistants could automatically restock and pay for groceries while selecting the credit card with the most rewards all without direct human interaction. Payments might still run on bank rails and through deposits, but banks would become 'dumb warehouses' of pooled cash."
It states, "Banks already use reciprocal deposit networks to expand insurance coverage by swapping balances with other institutions in the network. Participating banks provide higher insurance coverage all while the customer maintains a single balance relationship. Since deposit insurance homogenizes deposit risk across institutions, depositors might have a stronger incentive to use AI agents to yield shop. Reciprocal balances exceed $400bn but are a small portion of the $8trn in insured balances."
The article tells us, "By reducing frictions, AI agents will increase competition within cash management. Hortaçsu and Yoo illustrate the competitive effects in money funds. Assuming all investors are awake, they found that spreads between money funds would compress and market concentration among money fund providers would fall 20%. We expect AI agented cash will become more mobile and higher-yielding."
It says, "Last year, McKinsey estimated that if banks adjusted just 5-10% of their checking account rates to match money market levels, 'total deposit profits' could shrink by 20%. Indeed, if banks were pushed to pay money fund-like rates on all the deposits, their deposit costs would double and their net interest margin would fall 44%. However, we think the actual effects on deposit rates and flows may be considerably smaller."
The update comments, "[An acedemic paper by] Hortaçsu and Yoo uncover[s] significant inertia in mutual funds' holdings of cash despite the sophistication of the investor base and the presence of higher yielding alternatives. Indeed, only 3-5% of mutual funds change their money fund holdings from month-to-month. And, over the 2010-24 period, over half (53%) never changed their primary money fund. This level of inertia is remarkable as there is a 10-20bp spread between the highest and lowest yielding money funds when the fed funds rate exceeds 1%." (See Crane Data's Oct. 29 News, "NBER Paper on Institutional Money Funds with AI 'When Cash Wakes Up'.")
It explains, "Some of this stickiness is institutional. Mutual funds prefer to keep their cash 'in house' or within the same management company complex. That said, our sense is that it may also be a rational choice. There is only a small incremental gain from shifting their balances.... For such a low payoff, it may not be worth the effort –- even with an AI agent. At the retail level where balances are considerably smaller, the extra yield pick up might be worth less than lunch.... Instead, our sense is that using AI agents to yield shop makes more sense for institutional investors with significantly larger balances."
Abate writes, "But, even here, the incremental yield may have only a nominal value. Instead, institutional investors with balances above the insurance limit may not be willing to cede control over their cash management to a software program. Higher deposit rates might point to a bank experiencing funding strains. A cash manager may wish to override an agent that only seeks yield. We think AI agents could be designed to detect this risk.... Without access to high frequency data on deposit flows, it might difficult to program yield shopping agents to avoid riskier banks. Our sense is that it might be easier to program yield-maximizing money fund agents as funds publish daily outflow data."
Finally, he adds, "Overall, our expectation is that while AI agents may gain in popularity, they are unlikely to significantly drain deposits from banks. Banks provide key services in additional to payments and deposits -- particularly lending. As there are few substitutes for bank loans, our sense is that higher deposit rates would be accompanied by higher lending rates. Instead, it is possible that agentic AI and yield hopping is more of a threat to money funds. But this depends on how much incremental yield there is from shifting balances from fund to fund. And, importantly, it assumes that money funds are willing to tolerate 'hot money' flows that create risks for their more stable shareholders."
