I have talked about all of these issues in great detail in the posts above. If you have not read them, please do so before reading the rest of this post. I will do a little summarization here of the ideas I wrote in my previous posts.
The two views of why some think that QE is inflationary usually boil to down to:
1) The Fed is Printing Money ($600 Billion in the case of QE-II) and this is obviously inflationary, or 2) An increase in reserves will stimulate lending in our Fractional Reserve banking system (via the Money Multiplier Model).
These might on their faces seem like a perfectly legitimate statements and be the bases for argument. But they are incorrect. In order to understand why, you need to understand what the primary function of a Central Bank is: Maintain control of its policy interest rate (for the Federal Reserve, this is the Fed Funds Rate). Open Market Operations (OMO) are designed to either add or drain reserves from the system so that interbank lending competition for the lending and borrowing of reserves (which affects rates) matches the target interest rate. The Fed controls the supply of reserves in the system, and adjusts them so that they equal the net demand at the target interest rate.
Most people however make no distinction between bank reserves (which are used to clear transactions either interbank, to the Treasury, or to the Fed) and the cash that is out in the private sector that you and I use everyday. They have completely different roles. Bank reserves are not lent out to the private sector. They cannot used for 'speculation'. A bank reserve is nothing more than either cash sitting in a bank vault (if the manager opted to keep the reserves as printed dollars), or it is stored electronically as a checking account at the Fed. But it's role is different than private sector cash.
Before I go on, let me ask you: What is a Treasury Bond?
You might say "A United States Treasury security is government debt issued by the United States Department of the Treasury through the Bureau of the Public Debt.". This is a true statement, but it isn't basic enough.
A US Treasury Bond is a savings account at the Federal Reserve. That's all it is.
So when a bank chooses to hold a Treasury vs. a bank reserve, they are holding an interest bearing claim vs. a non-interest bearing one (or low-interest bearing in the case of a bank reserve, 25 bp). When a bank decides to sell a Treasury to buy a reserve, or vice versa, they are simply exchanging a holding with the Treasury's account at the Fed for one directly with the Fed.
Next question: How do reserves as a whole change in amount in the banking system?
Let's simplify the scenario in that there are no Treasury Bonds. As I stated before (see The Matter of Deficits, Sovereign Default, and Modern Monetary Theory) there is no longer any reason for the US Treasury to issue bonds. Under the Gold Standard the Treasury bonds allowed the government to run a deficit by by borrowing more money than it had backed by Gold. When the Gold window was closed in 1971, that constraint was severed. In today's completely fiat currency system, US Government debt finances *nothing*. Treasury issuance's only role now is strictly to facilitate monetary operations [NOTE! I am *not* saying this is good, I am not saying this is bad. There is no judgment in this paragraph. I am simply recognizing the reality of our current monetary system.].
In reality where there are Treasury bonds, whether the bank holds reserves and Treasuries (which are the most liquid asset on the planet next to cash) or it holds just reserves, a swap between Treasuries and reserves is just an asset swap. Moving back and forth from one to the other is not inflationary and does not change net financial assets in the banking system. Therefore, absent any vertical money creation, at any point the amount of Treasuries held by banks + Bank Reserves is a fixed number. This means the banking system by itself cannot increase or decrease the amount of net reserves in the system. It can swap reserves for Treasuries, or vice versa. It can lend reserves amongst each other. But it cannot affect the net amount of bank reserves in existence.
All transactions within the banking system net to zero (for every asset on somebody's balance sheet, there is a corresponding liability on someone else's balance sheet. When a loan is created, a corresponding deposit is also created). This is why all transactions within the banking system are called 'horizontal'. This includes actions taken by the central bank. The Fed cannot arbitrarily print money and give to a bank [maybe it will be given that power in the future, but it does not have that power today. However you may rightly point out that the Fed bought essentially worthless MBS's at phoney market rates from banks. This is tantamount printing the money, buying worthless paper in exchange for new reserves, so that the bank now has reserves that it could buy Treasuries with and the Fed is stuck with worthless paper. So while the technicality of the asset swap was maintained, the Fed knowingly overpaid. I would argue that instances like these are inflationary if the Fed never tries to unload that paper.]. Whenever the Federal Reserve 'prints money', it is always buying something that exists already. So just like interbank transactions net zero, so do most central bank transactions [again, there are some exceptions to this rule, but OMO is the policy tool of choice and nearly all activity engaged by the CB is OMO. And the point being that OMO is an asset swap].
The way that net reserves in the banking system change is through Congressional spending.
The Treasury is the only party that can create money 'vertically' within our banking system. Treasury spending is the only transaction in which there is no corresponding liability. I am not talking about Treasury bond issuance here. Like I said above, Treasury bonds 'fund' nothing. When the Congress wishes to spend, they authorize the projects and direct the Treasury to spend. When the Treasury spends, it simply spends. It doesn't wait for 'bond revenues', it doesn't wait for 'tax revenues'. Both of this concepts had applicability under a Gold Standard. They have absolutely no applicability in a fiat currency system.
When the Treasury spends, money is created out of nothing. There is no asset swap. The Treasury either credits private bank accounts directly or issues checks that transact through the banking system. The result is a net increase in reserves. When the Treasury taxes, the result is a net decrease in reserves. In no way do taxes 'fund' anything in a fiat currency system.
... [ However government spending *MUST* be in some proportion to private sector productive capacity otherwise inflation will ensue. There is no 'free lunch' here. But the point of this section is to discuss the specifics of monetary operations, which is highly relevant to this post ]
The point of this background was to address the two ideas that somehow QE is inflationary.
Central bank Open Market Operations do not increase net financial assets within the system. They are managing the amount of bank reserves to ensure enough reserves exist for smooth interbank transaction settlement and the supply of reserves meets the demand for reserves at the central bank's target policy rate. So when the Fed does buy a Treasury bond from a bank, the Fed issues bank reserves to that bank (to be carried on the Fed's balance sheet). But from the bank's perspective (and by extension, the banking system as a whole), the bank had to sell a Treasury bond in order to receive reserves. The net financial assets within the banking system are unchanged.
You need to understand the concept of vertical transactions versus horizontal transactions to determine at what point the money supply either expands or contracts. All transactions within the banking sector (not the Fed and Treasury) are horizontal in nature. Every asset created has a corresponding liability. Nearly all Central bank transactions are also horizontal in nature, and OMO which is the central banks preferred tool for effecting monetary policy, produces horizontal transactions (it changes the composition of overall assets, but not the amount). The only way money is either created or destroyed is vertically, which happens by Treasury spending or taxation (spending as detailed above, bond issuance is *not* spending and it funds nothing. Bond issuance effects monetary operations, it does not finance anything in a fiat currency system).
That tackles why Central Bank OMO operations (of which QE is nothing but a subset) is not inflationary.
Regarding the Money Multiplier model, I explained in detail in my last post why building bank reserves was not inflationary. I will not regurgitate it here, but here is the quick synopsis (read my last post for a very detailed description):
Banks are not reserve constrained. When a bank wants to lend, it simply lends and tries to find reserves after the fact. The amount of bank reserves it has does not figure into lending decisions (although capital requirements do). In fact there are quite a number of banking systems that have no reserve requirements at all. So a Federal Reserve induced increase in the amount of bank reserves by a reduction in the amount of Treasuries (which is the asset swap mentioned above) will not make a bank more prone to lend, since it was never reserve constrained to begin with. In short, the increasing reserves targets only the supply side of the lending equation (and is never a constraint operationally), and says nothing about the demand side (the private sector's willingness to take on more debt).
The conclusion of my last post is that QE is not inflationary, and I stand by the conclusion. Further, I outlined my thoughts as to why the rally since the QE-II announcement was speculative (not driven by inflationary pressures). It is either completely speculative, or there is a liquidity composition change due to Treasury purchasing by the Fed from those that would want to speculate. However the reserves that the Fed issues in exchange for those Treasuries cannot be used for speculation, so the Treasury sales by speculators are likely not 'sales'. PDs might be lending out cash (again, not the reserves) in exchange for the Treasuries for speculation purposes (through the repo market), and holding reserves essentially as collateral. This would make for a carry trade. A carry trade is also an asset swap. And what should be more alarming (if this is what was actually happening), is that carry trades always have to be unwound eventually. However, whether this pet theory of mine is correct or not, all causes for this rally are speculative. There are no inflationary drivers behind this massive rally.
There is one distinction that ikkyu2 made on my Caps blog, which deserves to be brought up. Like I say above, one of the purposes of reserves is to ensure the smooth settlement of all interbank transactions. What would happen in the case of a massive wave of defaults, which would cripple the bank system (sort of the opposite of 'smooth') and tend to be highly deflationary. Could the preemptive building of bank reserves be an offset to future potentially (perhaps even likely?) deflationary activity? Please see comment #9 and comment #10 for this discussion.
But, getting to the real point of this post:
I just did my synopsis above of how the banking system actually works from the standpoint of interest rate management (which is essentially reserve management) and money creation. It is adapted from Follow Up: QE is not Inflationary, Thoughts on Risk Asset Instability [see: MTaA / Caps]. That is the 'long' version (this is actually a very complicated subject and these posts are by no means definitive. There are much more in-depth an comprehensive resources out there if you want to research this topic further).
But for a very succinct version of QE operations and why they are not inflationary, please read Mitch83's comment on TPC's post: Bernanke helps fuel an increasingly expensive market:
The point is: As a bank there is nothing you can’t do with treasuries that you can with reserves (or as you say “pure green liquid cash” = Dollars) in financial terms.
Regarding the operational realities involved in QE2 treasuries are “money” the same way reserves (“cash”) are. Of course you as an individual can’t go to the grocery store and pay with treasuries. You have to have cash. But banks don’t go to the mall with “green liquid cash” and buy stuff, they do their “payments” and other operations in a different way:
First, as Scott Fullwiler explained, if banks want to buy, they just buy, no matter if they have the same amount of reserves (a.k.a “cash”) or treasuries on the asset side of their balance sheets. If necessary they clear an overdraft with the Fed and later get reserves on the interbank markets from other banks, use their treasuries in the repo market (treasuries get leveraged), borrow at the discount window at the Fed (using the treasuries as collateral) or simply sell the treasuries on the bond market to raise the needed reserves (“cash”).
Second, if banks lend, they just create deposits (“out of thin air”) and corresponding debt claims by expanding their balance sheets (and that of their customers). Banks don’t lend out reserves or deposits from other customers. They even don’t have to have reserves on their balance sheets to create loans, meaning they are only capital constrained in lending (with ratios of 10% you see lending is also a form of leveraging), not reserve constrained. If a country has reserve requirements (RR), then the banks get the reserves they need afterwards (if necessary) to meet these requirements. But in fact RR are completely unnecessary (many countries even don’t have RR). Reserves play a predominantly unimportant role in bank lending.
So you can see banks / HFs etc. could do/buy the same things without restriction before and after QE2, as QE2 just replaces treasuries with reserves (the Fed buys in the secondary market, not in the primary market -> no debt monetization, QE2 = asset swap). The “spending power” of the private sector is exactly the same before and after QE2, meaning no net financial assets are added. The only difference is that an interest bearing asset (treasuries) is replaced with an almost non-interest bearing asset (reserves), reducing private sector income, which is in fact slightly deflationary.
The hope of the Fed is the releveraging of the private sector by impacting investors/consumers/lenders psychologically. As Scott Fullwiler said monetary policy is all about increasing the leverage of the private sector (horizontal level), whereas fiscal policy deleverages the private sector by actually adding net financial assets (vertical level). The Fed thinks that if confidence of the private sector is restored by QE2-QEn the releveraging game will continue, no matter how the balance sheets of the private sector look like. I think with QE1 the Fed as lender of last resort/”market maker” succeeded in restoring confidence in and of the credit markets and partly the “shadow banking system” (although as I see it it didn’t recover greatly or started releveraging in a big way, I think this is part of QE2-QEn).