Our banks are private partnerships with unlimited liability for directors - this tends to increase the give-a-shit factor for reviewing assets and loans but also limits scale. Thus, we also have the incorporated UBS and Credit Suisse who, in proper fashion, proceed to blow up perennially.
I expect that a lot of the success of the capitalist model stems from the fact that most actors in this model are not too big to fail. (See "internet": To make a resistent system, minimize the impact of any single node going down.)
The idea of a "market" is based on something like a bazaar, where no individual actor matters. When any single actor is large enough to matter on their own, you are leaving the the market model behind. There are terms like "monopoly" and "oligopoly" for that.
When the issue is "we can't let these actors fail, they are so big that it would destroy the market", it sounds like the solution should be a simple size limitation.
Yes, the efficiency argument can be taken to justify anything. In truth, competition will always increase efficiency more than just scale alone.
The comparison with nodes is important and one which most people don't quite get. Banks should be limited in scale. If they need to undertaken large financings, then they can always form a syndicate that exists for a specific project, and is disbanded afterwards.
Everyone wants to sheet home the blame to the junk mortgages, or something else, but in reality any financial system is going to encounter shocks, whether they be man-made or perhaps natural disaster.
What is important is that the structure of the industry is robust enough so that one or two failures doesn't crater the industry.
Think of it like scaffolding - occasionally accidents happen on construction sites. But you don't want the entire scaffolding to collapse as a result. Any failure should be localised.
Warfare changed forever when the Generals couldn't sit in a bunker far from enemy fire and send cannon fodder in. It's time to have the finance generals out of the bunker. Ruin a bank, and you should be financially ruined as well.
There are some efficiencies that come with scale. For example, it's preferable to have very large utility companies with heavy regulation. I tend to agree that scale is unnecessary for investment banking. If an entity needs capital in bulk it can go public.
The trade off then becomes efficiency of scale vs. tolerance to failure.
I would argue that certain sectors, like infrastructure for public utilities, would be negatively impacted by treatment as a free market - the cost of building that infrastructure presents a barrier to competition, so the market isn't terribly 'free' anyway - and gain much more from the efficiency of having a single pipe going to everyone's home which all suppliers use, with everyone paying for the maintenance of their own infrastructure.
According to Carolyn Sissoko (syntheticassets.wordpress.com) this is how English banks used to operate as well. Back in the days when draughts were accepted or not depending on the creditworthiness of the bank they were written by, it would have been laughable for a banker to not be super-wealthy, because how else would anyone recover their money in case of significantly correlated demand for liquidity? (This is long before the days of government guarantees preventing bank runs.)
Not a lot of Swiss banks go bankrupt..