Dr. Krugman has another good smackdown of the inflationistas on his blog Monday. But his explanation is lacking in the details that definitively prove the inflationistas wrong. So let's round out the details.
First, the Fed sets rates by manipulating rates HIGHER. This isn't the case sometimes. It is the case ALL THE TIME. If the Fed didn't set the rate on reserves by paying interest (as they do today) the banking system would be flooded with reserves and the rate would naturally be driven down to zero as banks would attempt to get rid of these reserves in the overnight market. Because the banking system as a whole cannot control the amount of reserves, the process inevitably ends at zero. That is, the overnight rate is zero without Fed intervention. So, the Fed ALWAYS manipulates rates HIGHER in trying to achieve an interest rate target. And it does so by manipulating the amount of reserves or by setting a floor on the rate as they do today via IOR (interest on reserves). So, get over the whole "Fed manipulation" meme. The Fed always manipulates rates higher.
Since the overnight rate has obviously hit a floor at the current target of 0-.25% the Fed has been forced to implement policy in unusual ways. But don't mistake fiscal policy for monetary policy (as many have done in recent years). What the Fed is doing at the long end of the curve via QE is what they always do at the short end. In other words, the Fed is manipulating the amount of reserves in the banking system to try to influence interest rates. This isn't nearly as unusual as many people will make you think. For instance, in 2006 the Fed flooded the banking system with $50B in reserves and no one cried about "monetization" or anything like that. In the case of QE, since they can't manipulate short rates lower they are targeting other rates on the yield curve. And they're doing this like they always do - by altering reserves in the banking system.
But the most important piece of this puzzle is understanding the difference between primary and secondary markets in implementing monetary policy (not fiscal policy). The Fed is buying bonds on the secondary market. This is a lot like you buying shares of stock in your brokerage account. This is a simple exchange. Who really buys the bonds that fund the Treasury's spending? Not the Fed. It's the banks via their Primary Dealers. The Dealers bid in the primary market at auction as they're REQUIRED to do. In other words, the funding has already been allocated and purchased before the Fed ever does anything. Whether they were buying or not the funds would get spent and the bonds would get bought. Don't confuse the Fed's monetary policy for fiscal policy. They're two distinctly different things.
In addition, as Paul Krugman rightly notes, there was no decline in Treasury demand auctions following QE2. So the idea that the Fed is backstopping the Treasury markets has been proven wrong. Many of us who understood this called it in real-time and said that famous investors like Bill Gross would be wrong about how this would play out. That was definitively correct and proved, without a doubt, that the Fed's buying on the secondary market was not backstopping the Treasury market.
Anyways, long rates are ultimately a function of current economic conditions. The Fed sets short rates based on expectations of future economic conditions and long rates are an extension of short rates. In fact, if the Fed wanted to pin the 10 year t-bond at 0% it would just do it, but that's a different matter). And bond traders front-run the Fed in trying to outguess the future economic conditions. So, it's best to think of this whole relationship like a person walking a dog through traffic. The Fed walks the bond market around and the bond market tries to steer the Fed by guessing where traffic is headed. But the Fed can always control the rate and the leash if they want. The dog ultimately knows this and so doesn't steer too far from its master (though it doesn't want to be behind its master!). So it's all a delicate guessing game because there's no telling when the traffic might become faster or slower than we expect.