Thursday, January 20, 2011

What is Quantitative Easing?

Over on Coordination Problem, Steve Horwitz explains why he opposed both QE 1 and QE 2. Basically, he is responding to attacks by Rothbardians claiming that he has shown excessive quasi-monetarist tendencies. (Really, the problem is that Horwitz has deviated from the plumb-line that any increase in the nominal quantity of money, other than an increase in the quantity of gold, is EVIL.)

What is quantitative easing?

Perhaps it is just wishful thinking, but to me "quantitative easing" refers to efforts to increase the quantity of money. Quantitative tightening would be efforts to decrease the quantity of money.

This is as opposed to interest rate targeting, where easing means lowering the target for the federal fund rate and tightening means raising that target.

When the target for the fed funds rate had been reduced to about 2 percent, and many economists, journalists, and politicians were saying monetary policy was out of ammunition (seeing zero so close,) and so we needed fiscal policy, it became clear that this tool of manipulating short term rates, that appeared to work well during the late eighties, ninties and the first part of the 2000s, was breaking down. It was time to focus on the quantity of money rather than interest rates.

I must confess that I have never liked interest rate targeting, but its apparent success during the Great Moderation led me to hold my tongue. I never favored the 2 percent inflation target either, but disinflation to zero just wasn't a priority for me. ( I did write Mark Sanford about it when he was in Congress.)

Unlike Horwitz, I don't support Bagehot's approach to lender of last resort. Rather than have the central bank lend to sound banks at a penalty interest rate if there is an increase in the demand for base money, I favor having the Fed make open market purchases of government bonds to accomodate the increase in the demand for base money. The base money would go to whatever banks sell the bonds, or whatever bank is utilized by the nonbank firms or households that sell the bonds. The market can handle moving reserves to sound banks that need reserves. Banks that cannot obtain reserves will have to default. They should be closed and reorganized. Rapidly.

I suppose I don't trust the Fed to follow rules about only lending to sound banks. Years ago, I was persuaded by Friedman's argument that the discount window should close. I don't think "penalty interest rates" should be part of the process. Interest rates should depend on market forces. If Fed purchases of governement bonds results in lower rates for government bonds, that should not be a worry. Whether the interest rates at which various banks can borrow on the market rise or fall should also not be a concern.

To some degree, my approach is that the Fed needs to offset changes in the money multiplier (in the currency deposit ratio or reserve deposit ratio) by changes in base money, to prevent changes in the quantity of money. However, I don't really favor stablizing the quantity of money, and believe that the quantity of money should change to offset changes in the demand to hold money.

Sometimes rather than break things down between base money, money multiplier, the quantity of money in the hands of firms and households, and the demand to hold that money, I just focus on the quantity of base money and the demand to hold it. It has some value. If the demand for base money rises, the central bank should increase the quantity to match. The market can determine the quantity of deposits people want to hold. On the other hand, the relationship between the quantity of deposits and the demand to hold them plays a key role in the process by which inbalances in the demand for base money and the quantity of base money actually impacts spending on output, prices, and production.

The actual policy of the Fed in 2008 was mostly focused on supporting loan securitization. It did involve an increase in base money, which was what was needed. But it also was directed at supporting markets where banks made loans, sold them, investment banks bundled them, and then sold asset backed securities to mutual funds and other investors. The theory was that the demand for these securities was low because they had become illiquid--they couldn't be sold. If the Fed jump started this market, then everything would go back to where it was before 2008. If people were confident they could sell the securities when they wanted, they would be willing to buy and hold them. If they would hold the securities, then the investment banks could sell them. If the investment banks could sell them, they would buy loans from banks. If the banks could sell the loans, they would make more loans.

From a monetary disequilibrium approach, this would work to solve the problem because people were holding money rather than these asset backed securities, perhaps indirectly through mutual funds. If they went back to holding the asset backed securities, then the demand for money would fall to pre-crisis levels.

The benefit to the Fed's approach had to do with lending markets. If banks hold loans on their balance sheets, regulations require that they fund it partly with capital. And so, by making loans and selling them, the loans can be funded by investors without banks (or anyone) providing capital.

The alternative would be for banks to make the loans, hold them on their balance sheets, fund them with deposits, and sell new stock to meet capital requirements. This process would take time, and in the meantime, lending would be disrupted. The price response to this would be high loan rates and low deposit rates. The increase bank earnings both provides an incentive for investors to purchase new issues of stock and allows banks to build capital with retained earnings.

Of course, banks held large portfolios of asset backed securities, particularly mortgage-backed securities. They suffered losses, and had less capital already, which further caused problems with lending. Because there is no capital requirement when banks hold reserves or government bonds, banks that have suffered losses and so have less capital can meet their capital requirements by cutting back commercial and consumer lending As old loans are repaid, they can hold reserves or purchase government bonds. This reduces their capital requirement.

So, there was method to the Fed's madness. But I was always skeptical that it would work. And it didn't. Asset securitization has not recovered, and the problem remains that the demand for money is extremely high by historic standards. The crash in the stock market also has resulted in a higher money demand. The extremely low interest rates on government bonds has raised money demand. So, there is more going on than just the collapse of securitization. Which is, of course, one reason why I think the Fed's focus on fixing that market was an error.

In the past, my view had been that the Fed should buy T-bills when there is an increase in the demand for base money. If the Fed bought all of the T-bills that exist, and there was still too little base money, they should start moving up the yield curve-- 2 year bonds, 3 year bonds, and so on. If they end up with the entire national debt, then start looking to agency debt, AAA municipal bonds, AAA private securities and so on. But I didn't expect to ever get there. It is difficult for me to imagine that a monetary base of $7 trillion or so would be too small.

Because of the crisis, I have changed my mind. When the yield on an asset gets so low that holding currency is better (which is probably slightly negative,) then the Fed should stop buying and move up the maturity curve to something that still has a positive yield. (I still have trouble imagining that $7 trillion could be too little, I just don't think the Fed should bother buying government bonds with really low yields.)

Also, it is embarrassing to admit this, but I didn't realize to what degree the Fed's asset portfolio was already heavy with longer term government bonds. Open market operations in T-bills was apparently something from textbooks.

So, I think the quantity of base money has been too low for the last several years. And I think the Fed should continue to purchase government bonds with positive yields (which all of them have, but the 4 week ones and even 6 month ones have pretty low yields.)

And so, I support quantitative easing. And, when the demand for base money falls, as I expect it will, and hope it will soon, I will support quantitative tightening.

As I, and all the quasi-monetarists have explained, quantitative easing would be much better in the context of an explicit target for the growth path of aggregate money expenditures. I favor final sales of domestic product. Further, the Fed should stop paying interest on reserve balances, and in fact, charge a bit for them to reflect the interest rate risk it is bearing on its asset portfolio. But the Fed should also stand ready to expand base money however much is needed to get money expenditures to target. And then to contract it again when the demand to hold base money falls.


  1. A very good post, indeed. I have a question: ¿has been an target of the Fed to recapitalize banks with the retributive reserves?
    I don´t understand well the rationality of this question, except that the banks can have a liquidity base to repair their capital & liquidity ratio.
    If so, the monetary base (or HPM) should be augmented mucho more, to compensate money that is hoarded in reserves.
    Perhaps the Fed had two big problem: the liquidity in general and the sistemic risk, for which perhaps massive liquidity was not sufficient.

  2. "When the yield on an asset gets so low that holding currency is better (which is probably slightly negative,) then the Fed should stop buying and move up the maturity curve to something that still has a positive yield."

    Yields on Treasuries get low when markets perceive high risk of monetary policy errors. If the Fed buys Treasuries when their yield is low because of the abovementioned reason, the Fed will suffer losses if the monetary policy is successful. The actions of the Fed may reduce the yield on Treasuries further, thus increasing the market estimate of risk of monetary policy failures. As the increases in monetary base are subject to reversal, the net result of Treasury purchases at low yields might be contractionary (depending on the Fed's reaction function).

    "It is difficult for me to imagine that a monetary base of $7 trillion or so would be too small."
    As the increase of the monetary base to $7 trillion is tentative and subject to reversal, if Fed's reaction function is wrong, $7 trillion might be too small. This is why level targeting is important.

  3. "I just don't think the Fed should bother buying government bonds with really low yields." Scott Sumner claims that *money* is importantly different even from T-bills with zero yield, and that using the former to buy the latter would have a significant effect. Do you disagree, or are you saying only that buying a long-term T-bond with a significantly positive yield would be *even more* effective? (And perhaps you would agree that buying farmland or penny stocks would be even more effective yet?)

  4. I disagree with Sumner. I don't think open market operations with zero interest T-bills are effective. (Well, the limit is the storage cost of currency, and so slightly less than zero.) I don't really care about scenarios where there is a permanent increase in the quantity of money, so the price level will eventually rise, and so the expected real interest rate will be negative. I am primarily interested in the scenario where money expenditures are on target, there is no inflation, and the demand for money has increased. Once base money is no worse than T-bills as a store of wealth, creating base money by purchasing T-bills doesn't correct the excess demand for base money. Those who give sell the T-bills are just going to hold the base money.

    In the status quo situation, where monetary policy has already failed, purchasing zero interest T-bills might work, but it works by causing expections of a recovery of real output and maybe prices a bit. A constraint that the Fed will only hit the target, which is good, and that it will only buy T-bills with zero yields will not help with those expectations. It is essential that the Fed make it clear that it is not limited to purchasing T-bills. Purchasing longer term bonds now will have that effect. And, of course, the Fed has been buying longer term government bonds for decades. But remember, monetary policy has already failed. Money expenditures are below target. A system that requires that money expenditures fall below target, and then recover just doesn't make sense.

  5. I do not pretend to grasp all the nuances of all of this. But what I do grasp is that whatever the Fed does if it is wrong the consequences are likely to be very severe. Monetarists have made a compelling case - at-least to me, that the root of past recessions and depressions has always been monetary. Even conceding they are not entirely correct, it still seems there are significant monetary contributions. Even Bernanke says that the Fed will have to undo what it is doing. Quantitative tighting during a nascent recovery seems far trickier than easing is now. Further the Fed is acting at the end of an elastic string. The more rapidly they can react the less they will need to do. But they are positioned in the market such that their responses must be after - sometimes significantly after the signals they are trying to capture. If we are lucky they can slowly dampen oscillations until things are stable. But it is entirely possible that sudden changes will cause panic amplifying rather than dampening swings. Many many people are currently worried that Quantitative easing is a recipe for substantial inflation. They are looking at the economy through a microscope anticipating inflationary signals. It is very likely that at the first clears signs inevitably the market will react first and aggressively.

  6. Mr. Woolsey,

    I would like to ask a question to better understand your concept of accomodating movements in the demand for money by changes in the supply of money.

    How does the central bank actually know what the demand for money is? Isn't there a knowledge problem here? I mean, in the market the demand for goods and services are signalled through the price system. How would the amount of _money_ demanded be signalled to the central bank then?

    I apologize if the question was "dumb", I am no macroeconomist.

  7. You didn't take the bait: you didn't answer my (parenthetical) question about farmland and penny stocks. You say that zero-interest (strictly speaking, *slightly-negative-interest*) T-bills are no better as safe stores of value than is base money, so the Fed's buying the one with the other would be ineffective as monetary policy. You would have the Fed ease by buying longer term T-bonds with positive yields. These would, indeed, be more different from money than are zero-interest T-bills. But illiquid or speculative assets are *even more* different from money; would not Fed easing be *even more effective* if it bought such assets, rather than treasury securities of any kind?

  8. Philo:

    Perhaps having the Fed purchase penny stocks or farmland would be "more effective" in some sense than purchasing long term bonds. It would also be more risky to the Fed, and so to taxpayers.

    I am all for purchasing T-bills until their yields go below the cost of storing currency. I don't think minimizing the change in the quantity of money makes sense.

  9. Anonymous:

    There is no way to know what the amount of money people want to hold might be other than to observe the consequences. While I favor having the market solve this problem by index futures convertibility, I don't expect it to work perfectly. I just think it will work better than having everyone adjust their money prices and wages. Fundamentally, it has to do with the ability to harness specialized entrepreneurship to the problem. Frankly, the everyone speculating on the supply and demand for money along with microeconomic coordination is such a bad solution that having the Fed nationalize the relevant problem is better.

  10. Mr. Woolsey, thank you for your answer.
    If I may ask you another question:

    I see how a decentralized system could better solve the knowledge problem of ascertaining the demand for money than a central bank.

    However, given the institution of a central bank and its imperfect current methods for gathering information on the matter, do you think that a central bank does more harm than good (or viceversa) trying to supply the amount of money that it thinks is demanded? I.e., can the knowledge-problem in the existing institutional setup lead the central bank to act pro- rather than countercyclically?

    I agree that if this were the case, the solution - given the central bank institution - would be to improve the available statistics on the matter, as you have suggested in a current post. However, I am curious on your opinion on the ability of the current institutional setup to do more good than harm if it followed the rule of supplying money demanded, rather than to adhere to a strict quantity of money-rule.

  11. It is very tough to say much about Fed, I believe if we are really wishful to get success in a business like Forex then we must be certain of many things, as that’s the only way we could get fair bit of success. I am trading with OctaFX broker, it is a really solid company which has got brilliant 50% bonus on deposit offers, it is not just show piece, but life turning stuff given we can use this amount easily without any struggle a lot.