QE: Greg Ip answers my questions

By Felix Salmon
November 8, 2010
The Little Book of Economics: How the Economy Works in the Real World -- whether he could answer a few questions for me.

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Are you still confused about quantitative easing, what it is and how it works? I certainly was, and so I asked a genuine expert on the matter — Greg Ip, the author of The Little Book of Economics: How the Economy Works in the Real World — whether he could answer a few questions for me. Greg’s been writing some great blog entries on this subject at the Economist, like the one I linked to this morning, but sometimes you need to take a few steps back to clear up some very basic questions first. Thank you, Greg, for these fantastic answers! If you like them, go buy his book!

FS: Does the Fed print money? If so, how?

GI: Yes, and I’m surprised to see Pragmatic Capitalist dispute this.

It’s true that the Fed is not literally printing the $20 bills that end up in your wallet. As a commenter on your own blog has noted, that’s the job of the Bureau of Printing and Engraving. But money includes both currency in circulation and the reserves that commercial banks keep on deposit at the Fed. By that definition, the Fed is indeed printing it.

Here’s how QE works. The Fed buys a $100 bond from Bank of America. The bond gets added to the Fed’s assets. Bank of America has an account at the Fed. The Fed, with a keystroke, puts a $100 into B of A’s account. Where did the money come from? Thin air. Bank of America can visit its friendly neighborhood Fed branch and withdraw that $100 in the form of bills and coins. So for practical purposes the distinction between currency and reserves is meaningless; the monetary base includes both.

Incidentally, it makes no difference whether the bond belonged to Bank of America, or a customer of Bank of America; the mechanics are identical. When the Fed buys the bond from someone who isn’t a bank (e.g. a primary dealer), the transaction is settled through that person’s bank.

The Fed can also unprint money. Suppose its sells the $100 bond back to Bank of America. It then deducts $100 from B of A’s account at the Fed. The money disappears from existence.

You could argue at a more abstract level that this is not printing money. Normally we treat the Fed as independent of the government. Its liabilities, namely currency and reserves, are therefore not liabilities of the government, like treasury bills and bonds. When the Fed buys a Treasury bond, the liabilities of the Fed (reserves) grow but the liabilities of the government stay the same.

If instead you treat the Fed as an integral part of the federal government and merge their balance sheets, then QE simply takes one liability (a Treasury bond) out of circulation, and replaces it with another (reserves). This is actually a core criticism of QE: that in a liquidity trap, investors don’t really care if they own cash or Treasury bills and replacing one with the other accomplishes nothing.

In the real world I think the two should be, and are, treated separately. The Fed as far as we can tell is acting independently. It doesn’t have to buy Treasury debt; it could theoretically conduct QE by buying private sector or foreign debt.

Is there any difference between QE and the Fed’s normal open market operations, beyond the duration of the assets being bought?

Please allow me a moment of naked self promotion; my book, The Little Book of Economics: How the Economy Works in the Real World provides a layman’s explanation of how both conventional monetary policy and quantitative easing are implemented, on pages 71-72 and 156-161. (It has lots of other cool stuff, too.)

In theory, the two are less different than you’d think, as Ben Bernanke said over the weekend. In both cases, the Fed is buying and selling securities in an attempt to influence interest rates. Under conventional monetary policy, the Fed uses open market operations – buying and selling Treasury debt for holding periods of 14 days or less – to target the Federal funds rate. Since the Federal funds market is so small, the Fed doesn’t have to buy or sell much to get the rate to where it wants. And normally, if it’s targeting, say, a 3% funds rate, it will have to periodically sell, to keep the funds rate from dropping below target, as well as buy to keep the funds rate from rising above target.

Under QE, the Fed shifts its focus to long-term rates from short-term rates, and buys as much debt as it needs to get long term rates down. If it ever gets long-term rates to where it wants, it will stop buying. If it thinks long-term rates have gone too low, it could sell. Since the bond market is so huge the Fed has to buy way more debt than when it’s trying to sway the Fed funds rate.  That’s why its balance sheet has grown so much.

This discussion, however, masks an important practical difference. Under conventional monetary policy the Federal funds rate rises and falls with the supply and demand for reserves. The Fed targets the fed funds rate by buying debt, which adds to reserves, or selling debt, which shrinks reserves. Note that the quantity of debt involved, i.e. assets on the Fed balance sheet, is irrelevant: the Fed is trying to fine-tune the supply of reserves, i.e. its liabilities, to get the Federal funds rate on target.  

It’s just the reverse under QE: the Fed is explicitly targeting the amount of debt, i.e. the asset side of its balance sheet, and the amount of reserves it creates in the process is irrelevant. A lot of people have a monetarist-like conviction that the $1 trillion-plus of excess reserves on the Fed’s balance sheet represent a lake of inflationary gasoline just waiting for a match. My advice is to ignore it. Except when banks are liquidity constrained, reserves aren’t an important determinant of how much they lend; the demand for credit and banks’ own lending standards are far more important. QE will affect the demand for credit by driving down interest rates and the dollar and by driving up asset prices, not through the quantity of reserves.

Interestingly, when the Bank of Japan tried QE a decade ago it used a monetarist framework: it had a target for the quantity of reserves it sought to create which it then hoped banks would lend out, expanding the money supply. It didn’t work. So the Fed deliberately shunned this framework, even trying to supplant the damaged brand of “quantitative easing” with the term “credit easing.” Despite the different terminology, the practical results of the policies is the same: more reserves, bigger central bank balance sheets, lower bond yields.

Is the Fed trying to decrease medium-term interest rates and increase medium-term inflation expectations at the same time? Is that possible?

Monetary policy stimulates demand by lowering the real interest rate – that is, the nominal rate minus expected inflation. It could do this by lowering the nominal interest rate or raising expected inflation. The Fed, by implementing QE and telling us it thinks inflation is too low, is trying to do both. So far, it’s working: since August nominal rates have fallen and expected inflation has risen, so real interest rates are quite low, even negative at shorter terms. It’s conceivable, though, that it could fail: if inflation expectations really take off, then selling by panicked bond investors will overwhelm the Fed’s purchases and drive nominal and perhaps real rates up sharply.

What is a currency war? How would we know if the U.S. were engaging in one?

“Currency war” is one of those phrases that, like “shock and awe” seems destined to leave a bigger mark than the person who coined it (for the record, Guido Mantega, Brazil’s finance minister). It seems to be journalistic short hand for competitive devaluation, a dynamic where every country tries to lower the value of its currency in an effort to get an advantage for its exports over everyone else. Clearly, that’s impossible: currencies are a zero-sum market.

Currency war, or competitive devaluation, had a clearer meaning in the gold standard world where countries had implicitly agreed to keep their exchange rates fixed against each other. Countries that broke ranks by devaluing against gold got an immediate competitive advantage.

In a world of  floating exchange rates, it’s a little less clear. Win Thin of Brown Brothers Harriman says that Switzerland, Japan, Brazil, Mexico, Peru, Korea, Taiwan, and Israel have all intervened, but only to slow the pace of appreciation, not drive their currencies down. More prominently, several countries have tried capital controls to slow the pace of foreign buying of their assets and alleviate upward pressure on their currencies. I don’t think this, as yet, can be equated with currency intervention: it seems to me more a cousin to “macroprudential regulation,” by which central banks deliberately try to tamp down speculative flows in the markets in hopes of heading off a more painful speculative bust later on.

I take a relatively sanguine view of these steps. In their book “The End of Influence,” Stephen Cohen and Brad DeLong make an astute observation about China’s use of an undervalued exchange rate to spur export-led growth. It is, they note, a far less distorting form of industrial policy than subsidies, tariffs and quotas with their attendant corruption and rent-seeking. I feel the same way about the latest round of currency intervention and capital controls: they’re a retreat from the ideal world of perfect capital mobility, but a second best solution if the alternative is naked protectionism.

The U.S., I suppose, could be construed as having declared currency war if it started selling dollars on the open market against other currencies. QE alone doesn’t qualify.

What’s the difference between the Fed’s QE and the BoJ’s unsterilized fx intervention?

First, some Econ 101. When a central bank buys foreign currency, it usually pays for it with domestic currency that it borrows from someone else. Thus, the supply of domestic currency doesn’t change (in the jargon, the impact of intervention on the money supply is “sterilized.”) If, however, a central bank pays for the foreign currency with newly-printed domestic currency, the domestic supply of the currency increases; this is unsterilized intervention.

Buying either bonds or foreign currency with newly printed money are both QE. A monetarist would argue their impact is the same: by increasing the domestic money supply, they ultimately raise the price level and lower the purchasing power of the currency, which should drive down its foreign exchange value. The Swiss National Bank carried out unsterilized purchases of euros with Swiss francs because its domestic bond market was too small for a large QE programme.

In practice, however, the two policies have very different effects, especially by altering expectations. Domestic QE influences what investors think the government wants bond yields to be, while FX intervention does the same for the currency’s value. Not surprisingly, the latter is much more divisive globally.

If every country were to try unsterilized intervention, it would be equivalent to every country devaluing against gold at once: relative exchange rates wouldn’t change but everyone’s price level would rise, pushing down real interest rates. Barry Eichengreen, however, has argued that a more effective and less politically divisive way to achieve the same result would be for the Fed, BOJ and ECB to do domestic QE. It’s noteworthy that after its half-hearted, unsterilized intervention in September, the Bank of Japan has changed tack, accelerating its own QE programme to offset the effect on the yen of the Fed’s QE. “This kind of follow the leader response by central banks is part of the solution and not part of the problem,” Eichengreen says.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

Props to Greg for noting the distinction between monetary policy that targets liabilities versus rates. This usually gets lost in economic theory as it is taught and used. This paper from the BIS has a nice breakdown of theory versus practice: http://www.bis.org/publ/work269.pdf

Posted by kentropy | Report as abusive

Greg has this wrong. The Fed does not create new financial assets in the private sector. Only treasury does this through deficit spending. Via Warren Mosler:

“When the government “spends,” the Treasury disburses the funds by crediting bank accounts. Settlement involves transferring reserves from the Treasury’s account at the Fed to the recipient’s bank. The resulting increase in the recipient’s deposit account has no corresponding liability in the banking system. This creation is called “vertical,” or exogenous to the banking system. Since there is no corresponding liability in the banking system, this results in an increase of nongovernment net financial assets.

When banks create money by extending credit (loans create deposits), this occurs completely within the banking system and results in a liability for the bank (the deposit) and a corresponding asset (the loan). The customer has an asset (the deposit) and a corresponding liability (the loan). This nets to zero.

Thus vertical money created by the government affects net financial assets and horizontal money created by banks does not, although its use in the economy as productive capital can increase real assets.”

The Fed only changes the composition of already existing assets. Only the US treasury can “print money” through deficit spending. It might sound like semantics, but it most certainly is not.

Posted by trader34342 | Report as abusive

It’s also important to note that banks do not lend out reserves. This is another notion that Greg gets totally wrong. Banks are never reserve constrained. That’s why adding 1.2 trillion in reserves in QE1 did not result in an explosion in lending. The Fed can add as many reserves as they want. Banks won’t run out and lend them.

Posted by trader34342 | Report as abusive

Sorry, my first comment should read that “the Fed does not print new money that goes into the private sector.”

Posted by trader34342 | Report as abusive

Just read this article again. It’s totally 100% wrong. For instance, Greg Ip writes:

“Here’s how QE works. The Fed buys a $100 bond from Bank of America. The bond gets added to the Fed’s assets. Bank of America has an account at the Fed. The Fed, with a keystroke, puts a $100 into B of A’s account. Where did the money come from? Thin air. Bank of America can visit its friendly neighborhood Fed branch and withdraw that $100 in the form of bills and coins. So for practical purposes the distinction between currency and reserves is meaningless; the monetary base includes both.”

No. Where did the original money come from that was invested in the bond? It was already spent into existence by the US treasury. Then Bank of America went out and bought the bond. Then the Fed bought the bond from B of A and credited their reserve account with reserves. It was just an exchange. The Fed didn’t create any new money. The money in circulation wasn’t increased. Ip doesn’t understand how QE works.

Posted by trader34342 | Report as abusive


Doesn’t that ignore that Treasury’s deficit spending is enabled by the Fed’s issuance of Treasuries? Without some issuance, the government is unable to meet its cashflow obligations, and Treasury shuts down. And the bad debt would be written off, thereby “destroying” money, no? Thus, if you follow the chicken or the egg notions here, it is indeed the Fed that’s doing the creating. Treasury is somewhat further down the supply chain, one of a handful (BofA is another player in this example) who are “executing” the money into the economy.

Granted, I guess you’re right in that without a demand for loans, BofA can’t force the money into circulation, and I suppose a contributor to the liquidity trap. Am I missing something from your diddling with semantics?

Posted by MujiSamovar7 | Report as abusive

Well, you’re making the textbook assumption that the government borrows and then spends. But that’s not how it actually works. The government spends the money into existence and then that same money “funds” bond sales. Just think of it logically. If the government doesn’t issue any currency or spend any money then where do the banks get the money to buy the bonds?

The fact is, treasury has to spend the money first and that’s where Ip and most mainstream economists get it wrong.

Posted by trader34342 | Report as abusive

It doesn’t matter what you call it, it’s not going to work. There is enough money out there to finance the investment needed to drive a recovery, but the distribution of that money is such that those with it are not interested in using it. If the QE or any monetary policy the government or the Fed implements actually got money in the hands of people who are willing to use it, there would be a chance of a recovery.

All this talk is called re-arranging the deck chairs on the Titanic.

Posted by OnTheTimes | Report as abusive

Very nice explanation by Ip; kudos for him taking the time to write it and you for posting it, Felix.

A few quibbles:

“if it’s targeting, say, a 3% funds rate, it will have to periodically sell, to keep the funds rate from dropping below target”
Now that the Fed pays interest on reserves, this is not exactly true. The Fed will still wish to reduce reserves from time to time but the rate it pays is an effective floor on the fed funds rate.

“If it ever gets long-term rates to where it wants, it will stop buying.”
That’s OK if the Fed frames its target as a rate, but if it announces a quantity (as it has), it must follow through or it will impair its future ability to influence expectations. Remember, the market has already acted on the announcement.

“This is actually a core criticism of QE: that in a liquidity trap, investors don’t really care if they own cash or Treasury bills”
I think that Ip has not given sufficient consideration to this problem. Although you cannot buy real goods directly with T-bills, in the financial markets they are very close money substitutes because they can be posted as collateral. In the repo market they act much like money. Some people with monetarist inclinations, like Gorton, think that there has been a shortage of this collateral-money over the past decade and interpret AAA ABS paper as privately manufactured collateral, needed to satisfy this excess demand. I am skeptical of this interpretation with respect to past events, but obviously it is a logical possibility that at some point the supply of treasuries could be too small to support the repo market. The effect this would have is hard to predict. On one hand, the repo system is the banking system for large depositors who do not have access to insured banking. It fulfills the same valuable economic function that any other banking system does. Yet perhaps impaired access to the repo system would cause corporations to shed as much cash as possible, either in investments or dividends. That might generate the demand the Fed is hoping for. But maybe uninsured deposits in “too big to fail” banks would be seen as close enough substitutes.

Posted by Greycap | Report as abusive