When I’m working on strategy with financial organisations, I often find the Bodie and Crane classification (Harvard Business Review, 1996) a useful basis. This gives me a model that sees banks as a bundle of six basic functions: three core transfer functions and three interaction functions. The core functions are transfers in time (credit provisions, savings and loans with maturity transformation), transfers in scale (investments and funding large scale enterprise) and transfers in space (payments). The three interaction functions are information, risk and incentive.
(That final function is often overlooked, by the way, but it is strategically important to the economy. As my good friend Ron Shevlin pointed out in Forbes recently, almost all Americans with cryptocurrencies said that they would or might use their bank to buy and sell them. I strongly suspect that there are a substantial fraction of Americans who do not currently hold cryptocurrencies who would if the services was offered by their bank.)
This presents the incentive function of banking in context: that is, the existence of regulated financial institutions in a market means that transactions will take place. Now, this is not to say that speculating on DogeCoin is a good thing to do or not (I have genuinely no idea whether to buy it or not, and any references to specific cryptocurrencies in this article are for entertainment purposes only) but it is to say that transactions are enabled by banks and, as Ron points out, banks need information to set the dial on those incentive functions. But does it need to be bundled together with information and risk management functions? Probably not. Each of those individual functions could be unpacked. There is no obvious reason why this particular bundle is either optimal or invariant.