Importantly, these QE deposits add to the net deposits in the banking system. This is different from when a dealer acts as a private sector broker, buying a Treasury from one customer and selling it to another. In that case, the Treasury seller gets credited for a deposit that is, more or less, matched by a debit in the Treasury buyer's account. (The difference, if any, gets booked as a trading profit in the dealer's account.) There's no net deposit increase. It all balances.
It's also different from when a dealer sells a Treasury that it previously held (perhaps from the original issuance). In that case, the dealer sells to the Treasury, gets credited in reserves, but no new deposit is created. That's the scenario Coy was referring to. In that case, you only get a new deposit when the dealer makes a loan.
(Read more: Ray Dalio worries the Fed's QE may run out of gas)
But with a lot of QE transactions, there is new net deposit creation. There is no private sector deposit account debited from the sale of the Treasury because it is the Fed that is buying. What is recorded instead, is a debit of a private sector Treasury. The Fed records a liability—the new reserves—but doesn't debit these reserves from anywhere. They are created with a keystroke.
These deposits created through QE are real bank deposits. They are M-1. The funds in them can be withdrawn as cash currency, spent, invested, or held in the deposit accounts. They are real money, created through QE. And, if investors were so inclined, they can be used to purchase financial assets from high yield bonds to stocks. They could, theoretically, produce assets bubbles and even inflation, if the deposits were just spent on consumer goods.
This doesn't mean that this is happening now. I suspect at lot of QE created deposits simply remain as deposit assets of the Treasury sellers. Some of it, however, probably goes into assets that are close substitutes for Treasurys.
(Read more: The great fight over QE and deflation)
As I've written in the past, the exchange of a Treasury for a bank deposit doesn't suddenly make someone decide they'd rather spend instead of save. And although M-1 deposits grow, the amount of Treasurys in the private sector is relatively reduced. If you construe the money supply very broadly, so that Treasurys are included as money, there's no net creation of money, just a change in its composition. Unless or until the changes in the composition of this broad version of the money supply lead to changes in spending and investing tendencies, QE shouldn't be expected to have broad effects.
It's true that QE is an asset swap. But it's not just an "asset swap." It's actually a number if swaps of liabilities for assets. The liabilities: reserves from the Fed, deposits from the banks. The assets: Treasurys to the Fed, reserves to the banks, deposits to original Treasury owner.