I suppose what OP was arguing is, when bank A transfers to bank B, does bank B check with the Feds that the money from bank A is "real"?
And is a transfer from A to B really a transfer from A's account at the Feds to B's?
In that case, why would the US have to act "indirectly", threatening bank A not to work with bank B? They could simply deny any transfers to and from bank B's account at the Feds, which would make any transactions impossible?
> when bank A transfers to bank B, does bank B check with the Feds that the money from bank A is "real"?
Yes, unless the banks have mutual accounts with each other that they can use to settle instead. So, if bank B is accepting to be owed money by bank A for any transfer from A to B, they don't need to settle – but there's practical limits to that, imposed by both risk and regulatory concerns. Eventually, they'll need to settle up if funds flows A -> B and B -> A don't largely cancel each other out.
> And is a transfer from A to B really a transfer from A's account at the Feds to B's?
For wire transfers and ACH, it is.
> They could simply deny any transfers to and from bank B's account at the Feds, which would make any transactions impossible?
Yes – that's in fact exactly what a freeze of that bank's Fed reserves is!
> unless the banks have mutual accounts with each other that they can use to settle instead.
So this is just based on trust that the other bank will keep a truthful score of the transfer, and there won't be a dispute at settlement time?
If the banks are often willing to trade with each other "on trust alone" like that, I suppose that shows why the Feds can't always directly block dollar transfers, but have to rely on threats of account freezes.
> So this is just based on trust that the other bank will keep a truthful score of the transfer, and there won't be a dispute at settlement time?
Exactly: Either there is trust (and external settlement is not required), or there isn't, in which case banks will usually only credit incoming transfers to their customers once the underlying funds have settled.
Practically, trust is a spectrum, and banks might only settle once the outstanding balance in either direction has become too large (that's then called "netting").
> If the banks are often willing to trade with each other "on trust alone" like that
Bank balances are ultimately only be useful to their accountholders if they can effect some payment with them. There is therefore also a spectrum of fungibility of USD balances.
If you think about it, what makes USD useful is the fact that they can be used to pay for imports from the largest economy in the world. Only very few US companies would be willing to trade with you when being paid in USD balances at some foreign bank that they can't, as one significant example, pay their taxes with.
In that case, why would the US have to act "indirectly", threatening bank A not to work with bank B? They could simply deny any transfers to and from bank B's account at the Feds, which would make any transactions impossible?