Steve Keen has an interesting piece up over at Forbes that describes the folly of QE and how it doesn't actually do what people think it does.
The central points are correct. The amount of reserves in aggregate existing in the banking system is determined largely by the Central Bank, they don't get lent out and overall they can't really go anywhere other than back and forth.
However there are few things in there that perhaps could be misinterpreted.
The first is whether QE is a loan. Possibly. The question is who to. A loan is often repaid and generally receives interest income. QE ends up paying money to the banks, so it can't be a loan to them.
By removing the bond purchases from the model you actually get rid of the mechanism by which Interest on Reserves (IoR) is actually funded. The coupon on the bond pays the interest on Reserves and the coupon on the bonds comes from ... the Government. (A standing feature of Steve's models is the lack of a Government). There may be a few corporate bonds in there as well paying interest, but mostly QE was government bonds.
The monetary circulation that pays IoR is that the government pays the interest to the Central Bank (CB) on the bonds, the CB pays IoR to the banks and then the government taxes the banks to close the loop. Or if it isn't taxing the banks enough then it taxes the rest of us or sells more bonds (which as you can probably work out ends up being a redistribution to the banks and bank shareholders from the rest of us).
If it is to be a loan, then it should be a loan to Treasury so that Treasury can pay the interest to the CB that represents the net coupon on the bonds to cover Interest on Reserves. Of the central banks I've looked at I can see none of them that is increasing its assets to cover Interest on Reserves. You can see from this picture at the BoE that reserve balances are going nowhere.
The main difference from a loan of QE purchasing bonds is that it creates deposits in the accounts of those selling bonds, which tend to be insurance companies and the like. The created deposits are not owned by the lending banks.
The next issue jolted the most. And that's the idea that banks 'can buy assets' with reserves. No they can't. The surprising fact is that bank reserves have next to no meaning in the banking process. They are quite literally superfluous to requirements. You don't need them. You don't need a central bank. It would all work quite happily without it (mostly). The central bank is there, like all central clearing houses, to eliminate a chunk of risk and make the process work more smoothly with less overall expense. In particular when the commercial banks stop trusting each other.
Bank transfers have particular requirements. In order to move a deposit to another bank that other bank has to take your place in the source bank. The clearing process can be seen as just a way of trading those obligations between banks so that money moves freely and easily.
If reserves don't buy assets for banks then what does? The answer is simple. The bank takes the asset and creates a matching deposit. Just as it does when it creates a loan asset. It's good old balance sheet expansion again. (And to be fair Steve's piece does mention money creation in the next paragraph).
If you take a bank purchasing shares from a entity that is also a deposit customer then what happens is simply this:
That's why banks are desperate to be 'universal' banks (combined commercial and investment banks). It saves the investment part all that messing around with external overdrafts and oversight. When universal banks talk about matching deposits to risk assets, this is what they mean.
So if banks can work such magic what stops banks owning everything?
Regulation risk weights assets to the bank's risk capital and requires risk capital to be allocated to various asset classes. The change in Basel regulations reduced the capacity of banks to load up on high yielding risky shares and worse. Of course we could just regulate what assets a bank can hold and create instead of doing crazily silly calculations.
What QE did was shrink the free float of income generating assets in the private market. And unsurprisingly the race for yield amongst the participants drove up asset prices to their new clearing price. There was no extra buying as such by banks, in fact probably a bit less because banks had their restraints tightened a little. Most of all there was just less stuff to buy. A lot less.
QE didn't really get into the money supply via asset purchases - those people were selling anyway and will sell just enough to get the liquidity they require. What it did was displace those who wanted to save - forcing them to pay more for assets and get less income. It's a transfer from those that want to invest to those that are currently invested. From income earners to capital owners. Beyond that it is a standard wealth effect, which may make some people borrow more or spend more because they feel wealthier. Perhaps.
The final point is the ways that reserves can reduce. The central bank can take them back, and the public can withdraw them as cash. But there is a third option. The government can tax them away. That's what running a surplus does - destroys deposits and extracts reserves.
And of course doing that would also drive up private debt to even higher levels. The main aim of QE was to kickstart private borrowing even though it was already too high for comfort. So we should be glad it didn't achieve its aim, but we should be really concerned that those in charges thought that aim was sensible.