Although it dislikes the term, JPMorgan is a prime example of a universal bank. Others – Citigroup, Bank of America Merrill Lynch, Barclays, Deutsche Bank – also combine retail and investment banking but JPMorgan is the most prominent.

And universal banks have been, to put it politely, a disappointment. JPMorgan produced a return on equity of 9.4 per cent last year. That is barely adequate but it is the best of a bad bunch. None of the others made it past 5 per cent. And last year was not out of the ordinary. Those five universal banks together have managed an average return on equity of 5 per cent over the past five years. There is always an excuse – fines, new rules, restructuring charges, tough conditions in one market or another – but these are all part and parcel of universal banking.

Compare that with the specialists. Look at Wells Fargo, which is predominantly in the retail sector, with a 12 per cent five-year average ROE. Or even Goldman, a pure investment bank, with 11 per cent.

The universal banking model is broken, a fact some banks have realised. UBS and RBS have moved. Others – Deutsche and Barclays, for example – have been less radical so far and need to go further. The US universal banks are the most wedded to the model, promising better returns in the future. But shares in almost all of them trade at a discount to specialists. The message from investors is clear.

Paywalled piece from The Financial Times, via The Browser (which, if you're into more than just technology news, is a great, short, curated daily list of links to interesting reading online; another artisanal service I happily subscribe to).

How does that legendary Jim Barksdale quote go? “There are only two ways to make money in business: One is to bundle; the other is unbundle.”

Certainly feels like times are ripe for an internet only financial services play. Our arcane financial system has held up longer in its current form than I expected. I know there have been a few, though I have not tried any of them, but this article is more evidence that a horizontal (or specialized) attack might be fruitful. I'd probably start by trying to actually own the customer's money/wealth/assets, then branch out into figuring out ways to sell them financial products for those (e.g. insurance, credit cards). Customer acquisition is such a bear, though, that it might be easier for a trusted and well-known company like an Amazon to tackle.