How QE works

By Felix Salmon
November 4, 2010

Gawker’s John Cook asks me a question about how the Fed’s quantitative easing is supposed to work:

So the Fed is going to by $600 billion in U.S. Treasuries. It will presumably buy these Treasuries from private investors and institutions who had already purchased them–in other words, it won’t be handing $600 billion to the U.S. Treasury in exchange for bonds.

The purchases will be in increments of $1 million. Now, the kind of people who own $1 million and more in U.S. Treasuries tend to be people with a lot of money. And that money was kind of sitting there, and for some reason or another they decided to put it into treasuries, right?

So now along comes the Fed and says to those private investors and institutions, “Hey, I’d be happy to convert those treasuries into cash for you!” And they negotiate over price or there’s an auction or whatever, and the investors get their cash and the Fed gets its treasuries.

And so then these private institutions and investors are sitting there with a pile of cash. So why wouldn’t they just buy treasuries with it, which is what they had previously decided would be the wisest thing to do with that money?

The idea is to get those people to spend that cash in stimulative ways, right? But shouldn’t we assume that people who are sitting on large quantities of treasuries are sitting on them for a reason, and would likely continue to sit on them, even if they suddenly came into some cash?

John has a few of the details wrong, but at heart he’s absolutely right. The way that QE works is that the Fed will publish a schedule of how many Treasury bonds it intends to buy and when. It will then go out and buy those bonds from “the Federal Reserve’s primary dealers through a series of competitive auctions operated through the Desk’s FedTrade system.”

In English, what that means is that the New York Fed has a direct line to the biggest banks in the world (Goldman Sachs, Morgan Stanley, Deutsche Bank, etc — 18 in all). And it gets all those banks to compete with each other, either directly or on behalf of their clients, for who will sell the Fed the Treasury bonds it wants at the lowest price. The winners of the auction get the Fed’s newly-printed cash*, and give up Treasury bonds that they own in return.

The people selling Treasury bonds to the Fed, then, are big banks, who are told in advance exactly how many Treasury bonds the Fed wants to buy. As a result, they’re likely to buy Treasuries ahead of the auction, with the intent of selling them to the Fed at a profit. This is pretty much what John said would be going on, only they buy the bonds before the auction, rather than afterwards. Once the banks have made that profit, it’ll get paid out in bonuses to the people on the bank’s Treasury desk, with the rest going to their shareholders. We’re not exactly helping the unemployed here.

More generally, the Fed isn’t going to be buying any more bonds than the Treasury is issuing — so it’s not going to be lifting a lot of holders of Treasury bonds out of their long-term investments. But insofar as the Fed is forced to offer such high prices that investors simply can’t say no, those investors are probably just going to take the proceeds and invest them in agency debt instead from Fannie Mae and Freddie Mac. That debt is just as safe as Treasuries, and it even yields more than Treasuries, to boot.

What’s emphatically not going to happen is that the people who used to own Treasury bonds will take the Fed’s billions and suddenly turn around and spend them buying croissants at their local family-owned bakery. We’re talking about monetary policy here, not fiscal policy: the aim here is to bid up the price of Treasury bonds, which means that the yield on Treasuries will fall, and that those lower interest rates will somehow feed through into greater economic activity. The aim is not to take $600 billion and spend it on stuff in the real economy. That would be a second stimulus, and the chances of a second stimulus right now are hovering around zero.

Which is why Brad DeLong puts the value of buying $600 billion in Treasury bonds at about $7 billion in total, rather than anything near the headline $600 billion figure. The Fed is playing around with interest rates here — that’s its job. It’s not trying to directly stimulate demand.

*I should also take this opportunity to answer a question from CJR’s Dean Starkman, who asks where the money is coming from. The answer is that in a fiat-money system such as ours, the central bank can simply print as much money as it likes. If it wanted, it could literally go down to the local printing press, print out a bunch of $100 bills, put them in armored trucks, and send them over to JP Morgan or whoever sold them those Treasury bonds.** But that would be silly. So instead it simply increases the amount registered as on deposit at JP Morgan’s bank account at the New York Fed.

If JP Morgan had $100 billion in that bank account before, and then sells the Fed another $50 billion of Treasury bonds, then the Fed will just credit that $50 billion to JP Morgan, and the new balance in JPM’s account is $150 billion. Central banks can do that, which is why they’re so powerful. The amount of money in the system has just increased by $50 billion, and the Fed hopes that somehow that increase will feed through into higher inflation. Whether it will or not, however, depends on the degree to which JP Morgan can take that $50 billion and lend it out into the real economy. So far, banks have been bad at boosting their lending. And there’s not a lot of evidence that they’re getting any better.

**Update: Alea tells me I’m wrong on this: it’s the Mint which prints paper money, not the Fed, and all paper money is backed by Treasury-bond collateral.

13 comments

Comments are closed.