Where Inertia Goes When Agents Do the Checking

In December 2023, the UK's Financial Conduct Authority published rate tables for easy-access savings accounts. On 4 December, Lloyds' Easy Saver paid 1.40% on £1,000. Metro Bank paid 5.22% for much the same product. Differences in funding needs and brand explain some of that. A gap that wide, though, only lasts if a great many savers never look.

Customers who never look are, in practice, a business line, even if no annual report lists them as one. In 2018 Britain's competition regulator, responding to a complaint from the charity Citizens Advice, cited estimates that this "loyalty penalty", the extra cost of staying put, came to about £4 billion a year across mobile, broadband, home insurance, mortgages and savings. The customers most exposed were older, poorer, or less comfortable online.

On 9 October, Singapore's central bank chief, Chia Der Jiun, suggested that customers' AI agents could make "money move around faster for smaller differences in benefits and incentives." Banks, he said, would need stronger offers and ways of communicating their value to those agents. He gave no numbers, so this is a forecast. The mechanism is plausible, though: if software does the checking, the people who never looked would stop subsidizing the people who do.

Britain has tried to make moving easier before. It built a guaranteed service for switching your everyday bank account, yet in 2016 the competition regulator found that only 3% of customers had switched banks in the previous year, and more than 40% feared something would go wrong. The service handled the move itself. It did much less about what comes before: noticing that a better deal exists, comparing, and bracing for the hassle people expected anyway. That earlier work is what an agent might take over.

Britain also offers a glimpse of what prices do when one route to the margin closes. Since 2022, home and motor insurers cannot quote renewing customers more than new ones. The FCA's review found that the average new-customer home premium rose from £177 to £222, while existing customers stayed near £272. The FCA doesn't treat these nominal, unweighted averages as causal, and it estimated the rules' own effect on premiums as small: under £7 per motor policy, and not statistically significant for home. Still, the home insurance gap roughly halved, and nearly all of the narrowing came from new customers paying more rather than loyal ones paying less. That doesn't prove the old penalty survived as profit, but it suggests that banning a gap doesn't automatically return money to the people who were paying it.

So where else might the money go? One candidate is the bank's front door. Citi says it limits agents to read-only open-banking access. Chase's terms require agents to identify themselves; Wells Fargo's new agreement, effective 24 November, lets it require agents to register. The banks cite security, which is plausible while the rules on who pays for agent-initiated fraud remain unsettled. Even so, friction at the door can do the work that customers' own hesitation used to do. Asymmetry would be the sign to watch for: extra checks when an agent moves money out, few when it brings money in.

Another candidate is the instruction itself. "Keep my cash in the best easy-access account" is the kind of thing people set once and rarely revisit. Chia's point about banks making their case to agents suggests where the old margin could end up: in becoming whatever the agent counts as "best," through paid prominence or through offers tuned to what the agent measures. The assistant platforms' published terms disclose no such deal for savings, though private arrangements wouldn't necessarily appear there. If one does arrive, people who stopped checking their bank may not check their agent either.

What would test the forecast? Where agent use is common, the gap between the best and worst easy-access rates should narrow. If it doesn't, there are two readings. One is that inertia persisted: few people delegated, because staying put was more about fear than effort. The other is that inertia relocated, which would show up as heavier checks on money leaving than on money arriving, or as agent-managed cash settling somewhere below the best rates.