So, if we were denied that tool, at the present time, we would not be able to easily raise the level of short-term rates.

So we would likely, to regain effective control of short- term interest rates, need to shrink our portfolio from this current large level, back to the kinds of levels we had before the crisis. And we have set over several years, a plan for how we would normalize policy that relies not on selling long-term assets but on adjusting short-term interest rates.

I believe that if we were to follow the plan of selling off long- term assets, it could prove very disruptive to the expansion. It is a strategy that I think could harm the economic recovery and it certainly is not what we have set out to the public. We said we would shrink our balance sheet in a gradual and predictable way so as to not be disrupted.

I also believe that continuing reinvestment until the federal funds rate reaches a higher level makes sense. We want to ensure that we have the ability to respond to adverse shocks by easing monetary policy by lowering the policy rate. Having more “dry powder” in the form of higher short-term interest rates seems more desirable than less dry powder and a smaller balance sheet.

Now the words “well underway” in the FOMC statement are vague—what does that mean in terms of the level of the federal funds rate? Reiterating the disclaimer that I am speaking for myself, my view is that we should not set a numerical tripwire for ending reinvestment. If the economy were growing very quickly and the risks of an early return to the zero lower bound for the federal funds rate were deemed to be low, then I could see ending reinvestment at a relatively low federal funds rate. In contrast, if the economy lacked forward momentum and the risks of a return to the zero lower bound were judged to be considerably higher, I would want to continue reinvestment until the federal funds rate was higher. Consistent with the general principles I mentioned before, the evolution of the overall monetary policy stance—both interest rate decisions and balance sheet developments—should be data dependent.

"The interesting question is what happens when the Fed starts to shrink our balance sheet," Dudley said. "We have said in our exit principles that we intend to end the process of investment only after we begin the process of raising short-term interest rates."

And, he added, the Fed has said "we expect the interest rate to be the primary tool of monetary policy, and the balance sheet will be in the background."

The Fed "will be doing the balance sheet shrinkage at some future point in time," Dudley added. "But we will be probably doing it in a way that we have some focus on market function and make sure that our activities don't have bigger consequences for the financial markets. So we will definitely have to take that into consideration in terms of how we shrink our balance sheet over time."

Dudley, who is vice chair of the policy-setting Federal Open Market Committee, said the committee does not intend "at this time to actually engage in asset sales."

Our colleague Sam Fleming recently interviewed New York Fed president Bill Dudley and asked him about the possibility of the Fed’s maintaining a larger balance sheet indefinitely in conjunction with the use of its reverse repo facility. His answer suggested that the central bank remained undecided:

Q: Do you think in the longer-term, the new normal for the Fed balance sheet in 2020 onwards would be a larger balance sheet than in the past?

A: It is really premature to say. We are going to learn a lot about running monetary policy with a large balance sheet, using the overnight RRP and interest on excess reserves as our primary tools to affect the federal funds rate target.

As we learn that that will probably affect our judgment about whether it would be better to go back to a corridor system, with a very small amount of reserves in the system, or whether it would be better to have a greater amount of reserves and rely on these interest rate instruments to guide the stance of monetary policy.

My personal opinion — not the committee’s opinion — is jury’s out. We are going to learn a lot. Let’s learn and then decide and not make any judgments today.

Q: I think it was Ben Bernanke who argued for sticking with the RRP and having a bigger balance sheet.

A: There are some benefits of having more excess reserves in the system. It is a supply of liquidity, of safe assets to the system, it probably provides a little bit more lubrication to make the system work more easily. But we also have to see how well this works in practice, not just in theory . . .

And the other question is how big is big enough. I would be shocked if we wanted to run a $4.5 trillion balance sheet . . . The question is how much excess reserves would you want in the system if you wanted to run with a system based on interest rates being your primary instrument of policy.

BARTIROMO: Let's talk about the balance sheet of the Federal Reserve right now. It has grown enormously. How are you going to unwind that balance sheet without impacting global markets?

PLOSSER: Well, I think that's a question we don't know yet. We're in uncharted territory, we've never had to do this. I think the Fed's inclination will be to allow the balance sheet to shrink very slowly, if we can, but policy choices may face us where we may have to raise interest rates more rapidly or sell off assets in the balance sheet.

We don't know what's going to happen, so we need to be prepared for that and I think it's still one of the looming risks of our sort of unconventional policies that we took over during this last eight years.

MCKEE: Well how do you shrink the balance sheet? Is the market big enough to absorb Fed sales?

PLOSSER: I think absolutely. Well that doesn't mean -- obviously we are not going to sell them all at once, but I think we can begin a process. I would have argued we should begin letting them stop reinvesting would be the first step, so quit buying. But I think we can let them run off and sell small quantities month by month without disrupting the markets. We bought them without disrupting the markets, so there's no particular reason why we shouldn't be able to sell them without disrupting the markets.

MCKEE: Well you have treasuries and mortgage bonds. Would you sell both?

PLOSSER: Yes. I would tend to want to sell both. I think we have a problem with all the mortgage-backed securities and agency debt that we have. As you know, I have sort of been a -- not been a big fan of that sort of fiscal policy, sort of tax strategy of us helping one market at the expense of the others. So I would be getting rid of both. We -- our long run should be to get back to something that looks like an all-treasuries portfolio.

ASMAN: How do you do it? Do you let the bonds expire? Do you actively sell them or what?

FISHER: Very slowly would be the answer. Now, again, I didn't want to go that far, but we did. It is a committee. It's a thoughtful group of people. That's what was decided.

But (INAUDIBLE) when we sell something we could sell.

ASMAN: OK.

FISHER: But I think that would be the last resort here because we have to be careful here. We've -- the markets have moved, according to us, you know the old saying, you don't fight the Fed. We have driven rates to historic lows.

So we all expect to our balance sheet to gradually decline over time after we regard it is appropriate to begin to tighten policy. We have not decided and will probably wait until we're in the process of normalizing policy to decide just what our long-run balance sheet will be. But clearly it will be substantially lower than it is now, and it will take a period of a number of years. This is -- this could happen simply by ending our reinvestment policy at some point. If we did that and nothing more, it would probably take somewhere in the neighborhood of five to eight years to get it back to pre-crisis levels.

For today, I will note that, in the view of most participants, the broad principles set out in June 2011 remain applicable. One difference is worth mentioning. While participants continue to think that, in the long run, the Federal Reserve’s portfolio should consist predominantly of Treasury securities, a strong majority now expects that the Committee will not sell agency MBS during the process of normalizing monetary policy, although in the longer run, limited sales could be used to reduce or eliminate residual MBS holdings. I emphasize that, given the outlook and the Committee’s policy guidance, these matters are unlikely to be relevant to actual policy for quite a while.

 BRADY: Thank you, Mr. Chairman.  If the economy were to accelerate, the Fed would have to start unwinding QE3. So what is the Fed's exit strategy, the steps you'll undertake? And when do you anticipate, again, executing this?

BERNANKE: Mr. Chairman, so first -- the first thing, of course, would be to wind down eventually the quantitative easing program, the asset purchases. As I've said, the program relates the flow of asset purchases to the economic outlook. As the economic outlook and particularly the outlook for the labor market improves in a real and sustainable way, the committee will gradually reduce the -- the flow of purchases.

I want to be very clear that a step to reduce the flow of purchases would not be an automatic mechanistic process of -- of ending the program. Rather, any change in the flow of purchases would depend on the incoming data and our assessment of how the labor market and inflation are evolving.

   So at some point, of course, we will end the asset purchase program. Subsequent to that we will follow the guidance that we've provided about interest rates. Our principal tool for raising interest rates will be the interest rate on excess (ph) reserves that we pay, which will induce higher money market rates and a higher federal funds rate. And we will complement that with other tools, including tools that we have for draining reserves. 

    We may or may not sell assets. At this point it does not appear that it is necessary for us to sell any assets -- or particularly not any mortgage-backed securities -- in order to exit in a way that doesn't endanger price stability. 

    So there are a number of steps. We are currently discussing further our exit strategy, and we hope to provide more information going forward. But we certainly are confident that we can exit over time in a way that will be consistent with our policy objectives. 

    BRADY: You anticipate allowing maturing securities to roll off the balance sheet before you begin selling securities themselves? 

    BERNANKE: As I said, we -- we could normalize policy by simply letting securities roll off, and I think there's some advantages to doing that. For one it wouldn't disrupt markets so much. It would avoid as much irregularity in our fiscal payments to the Treasury. But we will see, ultimately, in the very long run, I think there's a desire to get back to a predominantly Treasury security portfolio. 

    But, again, in the exit process, allowing assets to roll off would be sufficient to bring us to a more normal balance sheet within a reasonable period. 

In March 2011, I shared some principles I thought should guide the Fed's eventual exit from this period of extraordinary monetary policy and its attempts to normalize policy. I said at the time that an effective exit strategy had to begin by deciding on our destination — what monetary policy operating framework we will use after exit — and then articulating a systematic approach to getting there. I was gratified that in June of the same year, the Federal Open Market Committee (FOMC) adopted a similar set of exit principles.

…

In addition to stopping asset purchases, the Fed should take two other steps as precursors to exit. First, we should seek to normalize the spread between the discount, or primary credit, rate, the rate at which banks borrow from the central bank, and the target federal funds rate... More than three years later, the crisis has passed and the other temporary lending programs the Fed initiated during the height of the crisis have disappeared. Thus, it may be a reasonable time to restore the spread to a more normal level.

Second, another step that might be taken before exit begins is to rethink our reinvestment strategy. There are no longer any short-term Treasuries in the Fed's portfolio. Rather than reinvesting maturing assets and prepaid assets into longer-term assets, it might be prudent to begin reinvesting in shorter-term assets. That would provide more flexibility in managing our balance sheet as we move forward.

If we do not stop purchases soon, one part of the exit strategy that might need to be reconsidered is asset sales… Determining the optimal pace of normalization from a very large and long-duration balance sheet is a complicated task that will depend on the speed with which interest rates rise and the size of the balance sheet when normalization commences. One factor to consider is the fiscal implications... Over time as the economy improves, interest rates will rise. Since the Fed's portfolio holds predominantly longer-term securities, interest received will not rise appreciably, but the interest paid on reserves will have to go up. Roughly speaking, if reserves total $2 trillion, remittances to the U.S. Treasury will fall by $20 billion for each 100-basis-point rise in the IOR…

In addition, should the Fed choose to sell long-term assets in a rising interest rate environment, it could experience capital losses. This would further reduce remittances to the Treasury. The more aggressively the Fed sells off its assets, the higher the losses and the more likely remittances to the Treasury could turn negative for a number of years. Although negative remittances would not impair the Fed's ability to implement monetary policy, they may impose significant political risk for the institution.

If asset purchases continue at current levels, reserve balances could grow to $2.25 trillion or more. That may require the Fed to sell assets at a somewhat faster pace than contemplated in the principles adopted in 2011.

Federal Reserve Chairman Ben S. Bernanke said the Fed will raise the interest rate on excess reserves as its primary tool for tightening monetary policy rather than selling assets from its balance sheet.

“The principal tool that we contemplate is the interest rate paid on excess reserves,” Bernanke said today in response to audience questions at a conference in Stone Mountain, Georgia. During a tightening, money market rates will probably stay close to the interest rate on excess reserves, he said.

“We have said in our existing exit principles we may sell assets in a steady way,” he said. Still, “asset sales are late in the process and not meant to be the principal tool of policy normalization.”

I do think that the Fed needs to shrink its large balance sheet. But I see that as a longer-term mission that can take place over the next five or six years or so. I believe that this mission should be guided by two key principles. First, the Fed should commit itself to a viable path of shrinkage of its asset holdings. Second, that path should be sufficiently gradual that it will interact little with the effectiveness of monetary policy. Along these same lines, the FOMC should offer as much certainty as possible about the rate of shrinkage.

Once the Fed decides to pare its balance sheet, it would need to sell securities at a “gradual and predictable” pace to “mitigate the uncertainty” from selling longer-term assets. “I would expect such sales when they came to put some pressure on financial markets,” she said.

“I would not expect to see a significant financial-market reaction to the conclusion” of the $600 billion in purchases, Yellen said.