Communicating Monetary Policy at the Zerobound

Eric Rosengren, president and chief executive officer at the Federal Reserve Bank of Boston, discusses monetary policy.

This meeting is part of the C. Peter McColough Series on International Economics, presented by the Corporate Program and the Maurice R. Greenberg Center for Geoeconomic Studies.

LIU: OK. Can I—can I have your attention? Thank you so much. I hope you enjoyed your lunch. I wanted to just interrupt you for a moment, because we want to begin the program.

Thank you very much, by the way, to the Council on Foreign Relations for hosting us this afternoon and for the C. Peter McColough Series on International Economics. I'm Betty Liu. I'm the host of the Bloomberg Television morning program "In the Loop," 8:00 to 10:00 a.m., by the way. Make sure you tune in.

(LAUGHTER)

I'm very pleased to introduce our features speaker for this program. He is the president and CEO of the Federal Reserve Board of Boston, Eric Rosengren. He's going to be presenting some insights, not only into monetary policy and the economy, but who's going to win, the Red Sox or the Tigers?

(LAUGHTER)

Eric Rosengren has been—he took office as the president and CEO back in July of 2007, so he's been there throughout the financial crisis. He's been with the Federal Reserve since 1985 when he joined as an economist in the research department. He's the author of over 100 articles and papers on the economy and the financial system. He's a voting member, as you know, of the FOMC.

Mr. Rosengren, thank you so much for joining us here at the Council on Foreign Relations. And I know you're going to start off with some comments about monetary policy.

(APPLAUSE)

ROSENGREN: Well, thank you very much, Betty, for that nice introduction. And thank you to the council for organizing this. This was meant to be conversational, so I'm not going to use the podium, going to keep it a little bit more informal. If somebody actually wants more formal remarks, they can see it on the Boston Fed website, since there's a full formal statement that goes with this, as well as a full PowerPoint. And I'm actually going to talk from the PowerPoint, which is the PowerPoint that's on our website, but you only have two slides, which has two figures on it that I'll talk about.

The topic I want to talk about is Federal Reserve communication. It's certainly something that's gotten a lot of attention since the September meeting, but actually has been really important over the entire time since 2008, when we hit the zero lower bound. So I want to provide a little context to the challenges that we're facing with communication and give you a few thoughts on how I think we should be communicating both with the type of people that are in this room, but, really, with the general public at-large.

One of the challenges is, even during normal times, it can be difficult to communicate about Federal Reserve policy. So talking about a fed funds rate and how it transmits through the entire economy can sometimes be challenging even during those good times. But now that short-term interest rates are at zero, we've been using a number of policy tools that are both a little bit new and also the public at-large is not as used to the kinds of communications that goes around those tools. I think that's presented a particular challenge, and one of the reasons it's such a challenge is that it's highly dependent on how people interpret some of our words.

So there are two main things that we're doing right now. One is, we're making purchases of long-term mortgage-backed securities and long-term Treasury securities. And the other is that we're providing some forward guidance. Both of those tools are designed to basically flatten the yield curve.

Now, if we want to affect people's behavior, then we have to have some understanding of if we use one or the other of these tools or if they're used in conjunction, what is the likely reaction of the investor public? What's the reaction of households and firms? We don't have much historical data, so unlike the fed funds rate that we have lots of that on, so we have a pretty good idea of how people respond to a monetary policy tightening or easing, we don't have that same kind of information with the tools that we're using now. We have a little bit of international evidence. We're not alone in using these tools. The Bank of Japan has been doing it for some time; Bank of England, ECB are doing variance of some of these tools.

But nonetheless, I think there's a great deal of uncertainty about exactly how the communication channel actually works. So this lack of historical precedent makes it difficult both to anticipate how people are going to behave and understand, really, what the impact of any one of our actions are going to be.

So if you'd look at the chart that's in your handout, the first figure that I have there is just the daily 10-year rate on the Treasury bond. Now, I'm not focusing on the Treasury bond because I think that's the most important rate; it's really more of a benchmark of long-term rates more generally. So the rates we really care about are the rates that are faced by households and businesses, so that corporate bond rates, that's mortgage rates, that's the rate that you'd pay on auto loan. But since many of those rates are priced off the 10-year Treasury or off intermediate Treasury securities, it's important to understand how those are moving.

And what you can see from this figure, a couple things to take away, is, one, since the beginning of May, I think you're all aware that there's been quite a movement in the 10-year Treasury rate. And I would highlight it's during a time where most private forecasters have actually been lowering their forecast. So the rates aren't going up because the economy is doing so much better than people anticipated; it really seems to be more generated by some of the actions that are occurring at the Federal Reserve or particular data releases that people put a lot of emphasis on.

And you actually see that in the bottom half of that figure, which lists the biggest movement in the 10-year Treasury rate over this approximately six-month period. And just a couple of the events that are associated around the times of those movements are highlighted there. And there's clearly a pattern. It tends to be around FOMC pronouncements, and it tends to be around employment reports, partly reflecting the fact that the Federal Reserve has been giving a fair amount of attention to labor markets.

So the sizable reaction we've seen to various policy announcements, I think I would take a couple lessons from that. First, some have had the view that unconventional policies don't really have any impact. That's hard to square with the kind of large movements we've seen in asset prices and interest rates. The way monetary policy works is by altering interest rates. And so I would be worried if we were changing our policies, particularly when it was less than fully anticipated, and there was no movement at all on interest rates. That's not the case. We're actually seeing that asset prices seem to be, if anything, surprisingly sensitive to any one of our announcements about these tools.

The second thing that I would emphasize is that those interest rates seem to be mattering. So if you look at what part of the economy is doing well over the last year, it's the interest-sensitive sectors. Housing has been the real leader in the recent last year of economic data. It's averaged roughly 15 percent growth. That is offsetting some of the fiscal austerity that we're seeing, the cuts in government spending and tax increase that we saw earlier in the year, and auto sales have been quite strong.

So if you think about those sectors that are going to be most sensitive to the interest rate movements, those are the sectors that seem to be doing very well in the economy right now. So what I would take from that previous chart is, one, these tools actually are reasonably powerful. The problem is they're somewhat imprecise, and we don't necessarily predict exactly how they're going to move when we decide to have a change in one of our policy tools. So that makes it much more difficult, not only to think about how we should change those instruments, but also how to communicate about it.

So how should we best communicate, given the challenges that we have with these tools? First, I think the primary objective is to make sure that we make the public at-large understand that, when we move either currently or with future movements and with forward guidance, that we're trying to do that in the context of getting maximum employment and price stability, our dual mandate.

So there actually is a goal in mind when we're moving these tools, and so that it isn't as if we're as focused sometimes on the short term. We're really thinking about the trajectory. So monetary policy is inherently forward-looking; it's not backward-looking. And we have to be thinking about how the economy's actually progressing.

Now, monetary policy is one element in that, so when we do a forecast of GDP and employment and inflation, obviously we think about what the right policy is to get to full employment, what the right policy is to get to a 2 percent inflation target, but we're not the only player in town. That's been made really clear over the last two weeks. Clearly, we have no impact on fiscal policy, but fiscal policy does have an impact on the economy, and so we have to take that into the context that we're thinking about.

Now, it's not just our fiscal policy that matters. It matters what the fiscal policy is in Europe. It matters what the fiscal policy is in Japan. Again, we don't control those factors, and we certainly don't control foreign governments and the kind of decisions they're making. And we also don't control whether a tsunami hits Japan.

So there are lots of things that can buffet the economy, and that means we need to be making adjustments, as incoming data tells us, that we're not on the path we thought we were going to be on. So when something hits that makes us alter the path, then we need, based on that data, to start making an adjustment.

So let me talk briefly about what I would think an optimal policy is and then talk a little bit about a clearly communicated policy and talk about the challenges between those two. Optimal policy, we should be looking at all the components of what's happening in the economy. We shouldn't just focus on one or two variables. We have an awful lot of economic data. Anybody who watches Bloomberg sees all the high-frequency data and people are changing their forecast on what's going to be happening with GDP, what's going to be happening with employment, what's going to be happening with inflation.

So there's a huge amount of data that is the context for us coming up with our forecasts for how the economy's going to evolve. And so we're going to be responding to that data as it comes in, so we want to be data-driven, but we're not just focused on that one element. We don't just focus on the unemployment rate; we want to think about labor markets more generally. We don't just look at one component of GDP; we want to look at all the components of GDP.

So the problem is, if you want to do a communication around that, if I tell you that I look at all available data, and when I see a shock, I try to make an adjustment, that probably isn't providing the kind of clarity that you all want. It's probably not providing the kind of clarity that someone trading in a financial market wants. So the communication strategy around that is, to some extent, conflicting with what the optimal policy would be, which would be taking all this into account.

The other challenge is that everybody interprets this data slightly different. So we basically know the direction, and most people will interpret the direction of the data in a similar way, but the magnitudes matter and the magnitudes are likely to be very different, both for market participants, but even for various members of the FOMC who are trying to decide what the right policy is.

At the opposite end would be a very clear and transparent communication approach, which would be to use calendar days. They're very observable. Everybody understands what a calendar date is. And as a result, there's no communication problem with a calendar date. The problem with a calendar date is, it may be inappropriate, given what happens in the economy. So if I announce that we're going to stop purchases on a particular day, and we have a huge shock to the economy, well, now I'm going to be stopping the policy, but it hasn't taken into account that there's been a huge shock to the economy. I actually ought to be doing something very different. And if I say, well, it's a calendar date, but it depends on what I see with the economy, well, then I'm right back to the original optimal policy decision-making.

So while something like a calendar date is very clear, it's not very flexible. And for monetary policy, it actually is pretty important to have the flexibility to make adjustments if unexpected things actually happen. So we don't want to be locked into an inappropriate policy, and the problem with calendar dates is that you at least run the risk that if the economy evolves in a different way than you anticipate, that you've locked yourself into an inappropriate policy.

What's an intermediate approach? Well, one is, you could tie yourself to an economic variable that's very important. Again, it's pretty transparent and easily communicated, particularly if it's something that's observable and understandable to the public, so something like an unemployment rate or an inflation index is pretty well understood by the public, easily observable. We have—under normal times, we have data that will tell us what's happening to that variable.

And so it has some of the attributes that we want, that it should in general be tied to the overall economy, but there is a potential downside, and that is, you're hoping that whatever variable you pick is a pretty good proxy for the entire economy. So let's take the unemployment rate. If the reason the unemployment rate is going down is because GDP is growing faster than potential, that firms are doing a lot of hiring, so you see a lot of payroll employment growth, then that's exactly what we want to see for the unemployment rate. It actually is a pretty good proxy for what's going on in the economy, pretty good proxy for what's going on in labor markets more generally.

Alternatively, if the only reason the unemployment rate's going down is because people are no longer looking for jobs, that they're discouraged workers and pulling out of the labor force, well, that's actually the opposite story. That's actually consistent with weak GDP, not particularly robust hiring, and as a result, that unemployment variable during times when other things are affecting it other than GDP and hiring practices may be sending a somewhat mixed signal.

So then, again, you're back to the problem, that while it's relatively clear and transparent, it's pretty easy to communicate. At times, you may have the inappropriate policy, if you've locked yourself to a particular variable. So that is a challenge, and I think the added challenge is frequently, when we've used things like the unemployment rate—and a good example is our 6.5 percent threshold for when we would raise short-term rates potentially—there are caveats around that. So once that—we go through that threshold, we're going to take a look at all of the things that are going on in the economy, and we're going to be making projections about what's happening and whether we expect the unemployment rate to continue to go down.

Well, those caveats tend not to get as much focus, and those caveats actually really do matter. So one of the challenges, when we're even using this intermediate approach, is that the caveats get lost and people kind of focus on the variable, focus on a date, focus on a time. And I think that's particularly true for financial market participants. When you're engaged in a futures contract, an options contract, calendar dates matter. Well, from a perspective of somebody at a central bank, I want to get the right economic outcome. The translation to the calendar date is probably less critical to me than it is to somebody who's trading a financial instrument.

So if I make an announcement of a 6.5 percent threshold for the unemployment rate, immediately you see people making a translation, based on their own forecasts or forecasts from private-sector forecasters, that translates it into a calendar date. Well, that's a bit of a problem, because then I'm back to the calendar date that all of a sudden the market is focused on and is somewhat invariant to the kinds of shocks buffeting the economy.

So let me just briefly talk about what happened in September, and then I'll turn it over to Betty. The second figure that's on that sheet of paper that should be in front of you is from the primary dealer survey. We do a primary dealer survey that's done prior to the FOMC meeting. We have it available to us at the FOMC meeting, and the New York Fed publicly releases it the day after the minutes are released, so this was released yesterday afternoon at 2:00.

And in this, it shows that primary dealers were asked, when do you—what do you think the probability is that we'll reduce the amount of purchases of Treasuries and agency securities at the September meeting, the October meeting, the December meeting, and after the December meeting?

And you can look at these probabilities that it's more than 50 percent for the September meeting, so there was an anticipation going into the September FOMC meeting that we would be making an adjustment, and there was an expectation that it was slightly more likely we'd do it for Treasuries than mortgage-backed securities. But I'd highlight that there was an awful lot to the right of that, as well, that when you look at agency MBS, there was a 52 percent probability of primary dealers that we would be doing something at the September meeting, but that means there was a 48 percent that we'd be doing it after the September meeting.

So after the meeting, a lot of the press kind of highlighted what a big surprise that was. Well, the difference between 52 percent and 48 percent shouldn't be a huge surprise. It matters in elections. It matters a little bit. But in terms of a huge surprise, that's well within the bounds of how much we can predict what people can expect us to do at any particular meeting.

And I think that's particularly true when you think about what the FOMC is, which is a group of 19 people, when we have all the governors and presidents, and they come to a meeting at—in Washington, D.C., for a reason. We want to listen each other. We want to hear the competing views of what's happening in the economy and what we should do. And then we should—we need to come to a consensus every meeting about what that policy would be.

But the primary dealers had a fair amount of uncertainty about what we do. It's not surprising that people going in with an open mind would say that, you know, depending on how you weight certain data, you might come up with a different conclusion. That means that we can't perfectly signal what we're going to do, because until we get in the room, we may not know exactly what we're going to do. That's why it's important to actually meet in Washington every six weeks.

So my own view of the September FOMC meeting, which clearly surprised some people, and some market participants were not positioned in the way they would have hoped to going into that—into our decision for the September meeting.

For me, I think there were three things that were particularly relevant for why I thought we should not cut back on our purchase program at that meeting. The first is that we did get weaker data over the course of the summer than we had been anticipating. So if you look at our summary of economic projections from the June FOMC meeting and compare that to the September forecast, there was a change, particularly if you look at GDP.

So GDP had been forecast for 2013 to be 2.3 percent to 2.6 percent. That was the consensus among the FOMC members. That was reduced to 2.0 percent to 2.3 percent. That's a pretty big shift when it's an average over four quarters, so it indicates that most FOMC participants actually read the data as indicating GDP wasn't as strong as anticipated at the June meeting.

The second thing that I would highlight is interest rates, market interest rates, including long-term rates, went up more than I would have anticipated between June and September. So my first chart highlights how much interest rates went up. Well, that really matter if I think the main driver of the economy right now and the reason that we're getting 2.2 percent growth in GDP over the recovery period, and not something less, is because the interest-sensitive sector has been doing quite well. Well, if interest rates go up unexpectedly, I have to be concerned that my expectations for housing, my expectations for auto sales may actually not be as good as what I was hoping for at that June meeting, so I have to take into account that interest rates moved up a little bit more than I thought were justified by the actions we were taking.

And, finally, the risk of a fiscal disruption. So at the September meeting, we certainly didn't know what was going to happen in the course of the beginning of October, but I think everybody knew it was a risk that there would be a somewhat dysfunctional debate about what we should be doing with fiscal policy. We all hoped that they would come to a conclusion fairly quickly. My base forecasts for both June and September was that there would be an agreement. That seemed to be the sensible solution for the U.S. economy. So my baseline forecast assumed that wouldn't happen, but it turned out to be wrong. They ended up having more difficulty coming to a consensus than I anticipated.

That's one of the inherent problems with any forecast, is there are things that are unexpected, and economists are no better than anybody else at forecasting political economy. There are probably people in this room much better than economists, actually, at forecasting political economy types of questions.

So given those three things, I thought it was particularly appropriate to do what we did. I would emphasize, as the purchases are not on a preset course, that I think we have to be highly focused on how the data comes in, but I would also highlight that the Federal Reserve, like everybody else, is learning how to communicate in a zero bound economy—or zero bound interest rates.

The—so we're looking to what—how markets respond, both here and abroad, when there are changes in policy. I think there's a lot to learn. We're learning by doing. That's a little bit unfortunate. It would be nice to have a little bit more historical data on this type of situation.

But nonetheless, I think if we keep our policy focused on getting full employment and getting inflation back to 2 percent, we'll end up with the right policy by doing that.

So that concludes my opening remarks. Slightly longer than I anticipated, but thank you very much, everybody.

LIU: Thank you so much. That was fantastic. And certainly, you are communicating, so you're out here. The fact that you're out here is communicating to the public.

I'm just kind of curious in the audience here, how many of you actually thought the Fed was going to taper in September? Can I just see hands? OK. So I would say that is still a majority of the audience thought that the Fed was going to taper in September, so does that surprise you?

ROSENGREN: I mean, in my chart for Treasury securities, it was 58 percent.

LIU: But these are—yeah.

ROSENGREN: So it seems roughly in line with what was going on in the surveys that we were taking. And what it indicates is somebody else could look at that same set of data, particularly if you were looking backwards, and said, well, they're focused on labor market conditions. The unemployment rate is 7.3 percent. But it was 8.1 percent a year earlier. That's a fair amount of progress.

So one individual may interpret that as strong labor markets or much stronger labor markets. Another person looking at that same data could say, but a lot of that is because of what happened with the labor force participation rate. We're not getting the kind of job growth, and GDP for the first half of the year came in at roughly 1.8 percent. It looks like the third quarter and the fourth quarter now are going to look a lot more like the first and second quarter than we anticipated, so that's not the kind of economy that gets us to where we want to go.

So people process the same kind of information differently. It doesn't surprise me that some people think either we would taper or that we should taper. And part of that's dependent on communications we're making. I think from—as several participants at the FOMC have highlighted, it was a close call for them.

LIU: Yeah.

ROSENGREN: Many of them chose at that meeting that it was not the appropriate time to do it. But I think the hands kind of reflected the fact that it was a close call. And for close calls, you're probably not going to be able to perfectly predict how a close call is going to come out, unlike the Red Sox.

LIU: Hearing what you're saying, though, the fact that there's two separate goals here, right? As you say, the Fed is concerned with, how are they going to influence the economy? How do they support the economy? And the markets—the investors are concerned with their profits. How are they going to take monetary policy and make money off of it?

So the fact that there are two goals going on here, does that mean that communication, interpretation is always going to be imperfect?

ROSENGREN: Well, I think communication always will be imperfect. And there are competing interests. So when a trader's trying to decide what kind of decision's going to come out of a September FOMC meeting, they're not necessarily deciding what should happen for the economy, what should happen to get 2 percent inflation, what should happen to get back to full employment. They're trying to forecast what will happen or make a prediction about what will happen. And the difference between 52 and 48, they have to position themselves one way or the other.

And so that is a very different situation than somebody who's trying to make the policy decision, which is less focused on getting exactly right a precise timing for when we should do something, and think really about over the course of the next two years, what is the path of policy that'll get us back to full employment? What'll get us back to a 2 percent inflation rate?

So while a trader has to be really focused on getting something exactly right at a particular point in time, we're trying to get something approximately right over a longer period of time. So there is some natural differences in terms of how we're thinking about the policy.

LIU: Well, and what you were presenting, which was just some very clear outlines of what could work and what couldn't work, it was more what doesn't work right now, right? A calendar date doesn't work. Even having specific indicators, whether it's the jobless rate or it may be labor force participation, who knows, other indicators, those are always imperfect, as well, because you need to see the entire picture.

So what is the solution, then, to a more clear and concise monetary policy communication? What is—is there a real answer to this? Or is there not?

ROSENGREN: I don't think we have the answer yet. And I'm a little skeptical we'll get the answer. Communication is challenging, even for somebody who's in the business like yourself, that thinking about a simple way to convey—so how I would convey something to this room might be very different than if I was talking to a group of college students. It might be very different than if I was talking to a retirement community.

So that if I only get one opportunity to communicate, different audiences are going to interpret those remarks in different ways, so I don't think there's one communication style or one set of ways of communicating that necessarily is going to be conveyed to everybody in the same way. And that's particularly true when you're talking about a very complicated topic.

And so I don't think there's going to be an easy answer, because those easy answers will frequently get you the wrong policy. So we should start with what is the right policy and then communicate that as best we can. Some people think that we should focus on getting the communication very precise, but that has an implication for whether we get the policy exactly right. And of the two, I'd much rather get the policy right and get the communication as right as we can. And I think that's where we are right now.

LIU: Janet Yellen, who is the nominee for Fed chairman, how is she going to be as a communicator? Will she be very similar to Ben Bernanke? How is she going to communicate to the market?

ROSENGREN: I think she—well, one, being able to communicate to the market's really important now. So 20 years ago, you didn't have to worry about whether somebody could do a press conference at the Federal Reserve. Now that's increasingly important. Around two-day meetings, where there's a press conference, you need the chairperson to be able to go out and clearly explain actions that may have only been decided a couple hours earlier.

I think Janet will be very good at speaking very precisely, very clearly, and very calmly. Those are attributes that I think Ben Bernanke had, and I think Janet has those same attributes. If you read her speeches, she's very precise in her language. She's clearly thought it through. She's an exceedingly good economist. So she understands the context, the economic theory, but I think she also spends a lot of time thinking about how to communicate it effectively.

So I think she will be very effective in the press conferences. I think she'll be very effective in forums like this, the more informal setting. And she's very good in the classroom, as well.

LIU: Is she for a more transparent Fed?

ROSENGREN: She is. She was responsible for a lot of the—she was in charge of the communications committee that has had some of the specific proposals on how we try to become more transparent in our communication.

LIU: All right. I'll let you drink a sip of water while I ask you this next question.

ROSENGREN: OK.

LIU: You talk a little bit about what happened—what's been happening in Washington. First off, you're not getting any data right now. The Labor Day, the Commerce Department all closed. You're not collecting data. How difficult is that going to make it for you in the next meeting?

ROSENGREN: It's certainly more challenging. So there is—ideally, we'd have really good data that would be available at the time we're making the decision. It looks like more than likely we're not going to have really good data at the time we're making the decision. It's not that we don't have any data, so we don't have the unemployment report, we don't have the labor market situation report, but we do have the ADP report, we do have initial claims, so we do have some information about the labor market.

Similarly, for a lot of other data series, many of the data series are provided, at least some elements of it, by the private sector.

LIU: But you wouldn't want to be making a decision on half data, though, or even partial data?

ROSENGREN: So you never have complete data when you're making a decision. It would be better if we had more complete data. I think it'll make it a little bit more difficult to have a decision that would alter policy, because the confidence we'll have about the incoming data will be less.

LIU: I know we've got until 2 p.m., so I know we've got about 25 minutes or so. I'm sure the audience has plenty of questions for you, so I want to turn it over to the audience for questions, and somebody will come around with a microphone.

Yes, sir?

QUESTION: Nisab West (ph), Pace University. Based on the statements that FOMC members and the chairman has made, it seems that you were surprised by the market's response to the tapering comments in June or in May. So my question is, A, what kind of models do you have to gauge that response? And, B, what do you personally believe? Does the market respond to stocks or to flows? Thank you.

ROSENGREN: So, many of the economic models have emphasized stocks, and including many of the models that we use at the Federal Reserve. So obviously, a small difference in tapering shouldn't have a huge difference in the stock. So if the model you're using is a stock-based model, you would certainly be surprised by the kind of movement that we've seen.

Even if you had a flow-based model, most of the private market forecasters were expecting relatively modest tapering in September. They weren't expecting a lot of tapering, and that was true for the next couple of meetings. Excuse me. So even with a flow-based model, you probably wouldn't have expected 100 basis point roughly movement in the rates, like we've seen since May.

I think the challenge is that there's a behavioral economics aspect to thinking about how the markets are going to interpret our actions. So we may be thinking that we're focusing on a fairly modest change that's dependent on the data. That can either be reversed or we might move very slowly afterwards. But market participants may assume that we're going to move much more dramatically after our first move.

We can try to communicate about that. We can try to make clear that we think accommodative policy, given that the unemployment rate is still at 7.3 percent, given that the PCE inflation index is at 1.2 percent, that that is a situation where we're missing on both elements of the mandate, and as a result, we need accommodative policy.

But markets may interpret it differently. And I would say that the degree of market movement, based on what happened over the last five months, was more than I would have anticipated and certainly more than what someone would anticipate if they didn't think quantitative easing was having any impact, because otherwise you wouldn't have expected the kind of dramatic reaction that we've actually gotten.

So I was surprised by the magnitude of the effect. I don't think we have yet a good model of exactly how the investing public is going to interpret relatively small changes. This is QE3, so after QE1 and QE2, we didn't have as much trouble anticipating how the market was going to think about our actions, in part because it was basically tied to calendar dates, that you could translate when the program was going to end, so there was a lot more certainty around that.

But it's also how people interpret the forward movements we're going to do. So even if we say we're going to move right now for a tapering, and we're not planning on doing anything else, do we see more data that support it, market participants may think that there's enough already that they're going to put into the—their own forecast much more tightening than we're anticipating?

So I think it's challenging for us to understand exactly how markets are going to react to our announcements. We have some idea, but I myself was surprised by how dramatic the movement was over the last five to six months.

LIU: Eric, by the way, what would get you—what would get you to vote yes on tapering? What would—what would you need to say to say, yes, I'm going to go vote for it?

ROSENGREN: So at the June meeting, I anticipated I would be voting for tapering in September, but I expected GDP to be growing faster than 2 percent, I expected payroll enrollment growth to be growing much faster than 150,000 jobs a month, and I expected there would be a fiscal agreement. And all three of those factors turned out not to be true. And as a result, it was not appropriate when we got to September, in my view, to actually make that change.

So somewhat stronger data on employment, somewhat stronger data on GDP, somewhat better fiscal policy than what we actually turned out would have been the right components for doing something in September, but we have to deal with the data we get, not with the data we wanted to get.

LIU: Next question, yes?

QUESTION: (inaudible) Partners. You mentioned an employment figure and how—about 50 percent, just focused on that number, 50 percent, focused on the drivers and the labor force participation as being an issue. When you think about an employment, GDP growth rising in salaries, what are those factors like the labor force participation that we should think about and focus on that have come up in past discussions that you anticipate will become a focus, as you really try to delve down on the drivers that have posted the headline numbers when you make your decision?

And separately, you mentioned interest rate-sensitive sectors, like autos and real estate. Can you help us understand a little bit how you assess what the impact will be of rates being at various level on those sectors and how they become a factor in GDP growth and—and employment increase and other of those factors?

ROSENGREN: So we're trying to get a sense of labor market conditions more generally. So we look at the entire employment report. We don't just look at any one element. We look at what happens with wages and salaries, as well.

So from the employment report, we certainly look at the unemployment rate, but we should also look at what payroll employment growth is. It gives you a sense of what's happening with hiring. We should look at labor force participation rate to get a sense of whether the reason the unemployment rate's improving or not improving is because people are coming in or out of the labor force. We should look at the employment to population ratio.

So I wouldn't say that there's one variable. We really want to look at all the variables. The reason, for example, the threshold's tied to an unemployment rate is, again, it's hard to communicate that we want to look at all these variables, all of which would have somewhat different thresholds, and sometimes that may give very mixed signals. So we want to put the whole labor market situation in context, including wages and salaries and the other things that would be tied to labor market conditions.

So I think that's one of the challenges in communicating. The ideal policy would be to say we look at all these variables and we try to get the right answer to get back to full employment in a reasonable period of time. But that's not a precise communication that you can trade on, so, hence, some of the challenge.

In terms—how do we think about the interest-sensitive sectors and how that is reflected in GDP? We have a lot of statistical models at the Fed. We're a data-driven organization. We have an awful lot of economists. They spend a lot of time trying to come up with statistical models that look at how movements in interest rates affect household decisions to buy cars, to buy houses, so we're very reliant on those models. They do a reasonable job, but at the end of every equation is an error term, so we don't get it perfectly right.

And there are periods when we have to make adjustments. So 2008, 2009, the housing sector—if you just did the typical housing model, it didn't perform particularly well. Well, that's not surprising, because during this period, housing prices were going down. During most periods, housing prices are going up. During most periods, you don't have credit conditions tightening quite dramatically. That was happening during this period. And frequently our models don't have a great proxy variable—for example, credit conditions and underwriting standards—and so those types of things aren't going to be picked up. That's why there's an error at the end of those equations.

And, in fact, our models, if you just use the historical models and didn't make any adjustments at all, they would have been off-track for 2009, 2010. They're much more on track now than they were at that time. So there's an art to this. There's not just science. And we spend a lot of time with the models. We spend a lot of time understanding the statistics behind the models. But we always have to think about, are there things not in that model that are really important?

It kind of gets to the previous question. We can have a model of how we think quantitative easing impacts interest rates and how those interest rates impact the economy, but there's a great deal of uncertainty even at the best of times around that, so we have to take it with a grain of salt, but it is better than nothing at all. It's much better than nothing at all.

LIU: On this side of the room, yes. I think you had—ma'am, you had—yes, you were the next one.

QUESTION: I wanted to ask you about how you feel your decisions on maintaining a zero interest rate policy affect the ability to reach fiscal deals. There are many people who believe that there's a lot of moral hazard that's been created and that we probably would have moved toward much more fiscal stability in the sense of forcing people to—to the table, if we hadn't allowed them to have this suppressed interest rates.

ROSENGREN: So my own personal view would be that the kind of discussion we've had over the last week probably has not been driven by monetary policy, that the kinds of arguments that they're having tied to the Affordable Care Act, tied to deficit reduction, is not primarily being driven by the fact that low interest rates make it a little bit easier to hit fiscal targets.

So I understand the point that you're trying to make, in terms of the moral hazard and in terms of low interest rates, make it a little bit easier, because the cash flow constraints on the fiscal government are somewhat different. To me, that doesn't seem to be the driving force. And when I watch the various news organizations, monetary policy doesn't seem to be the main reason that they're coming up with the kind of decisions that they're coming up with. I would hope they'd be coming up with more rational decisions than what they're coming up with if they were focused on what we were focused on, because one of the—we care about getting back to full employment, and both the legislature and the executive branch should, too.

LIU: But it's not the main reason at all, but I think the point is, is it's allowing the conditions for which the two sides can continue to fight over things like the Affordable Care Act and fight over entitlement spending, because nobody's pressured at this point to do something.

ROSENGREN: So I think there has been a fair amount of pressure to do something not generated by the monetary authority. But in the end, we're given a dual mandate. We're supposed to focus on inflation, and we're supposed to focus on what's happening in labor markets. And to us, fiscal policy is kind of the shock that we have to react to.

So ideally, we would have a fiscal policy well aligned with monetary policy, and both of them would be designed to get us back to full employment in a reasonable period of time. But if fiscal policy is taking actions that don't do it, we can't completely offset that, but we can certainly mitigate it. Failure to mitigate it would be inconsistent with our dual mandate, but I think it would also be inconsistent with the welfare of the public at-large. It would be arguing that big shocks that result in higher unemployment rate—there's no monetary policy reaction at all.

So monetary policy has to take fiscal policy in the context, in the same way that we have to take a shock from Europe or a shock from China or something that happens in Japan. And I think that is appropriate, and it's very consistent with the mandate that Congress has given us.

QUESTION: Benn Steil, Council on Foreign Relations. Before the financial crisis, inflation targeting was perhaps as close as one could get to orthodoxy among economists, in terms of approach to economic policymaking. Since the financial crisis, there's really been a flowering of debate, particularly among economists about what the optimal target is or what the appropriate targets are.

I've been quite struck with how much coalescing there has been among economists on the right of the spectrum and the left of the spectrum around one particular target—you probably know where I'm going—on nominal GDP targeting. I was wondering what your thoughts were about the appropriateness or effectiveness of NGDP targeting.

ROSENGREN: So nominal GDP targeting has been discussed quite a bit in the economics literature, as you pointed out, and there are a lot of academic economists that think it would be a very useful way to conduct monetary policy.

I think one of the challenges is explaining it to the public at-large. So GDP doesn't resonate with people the same way as unemployment and inflation. Unemployment and inflation are things that are pretty tangible to people, and that's people on Main Street. It's not just the investing public.

GDP, talking about all the goods and services in the economy, and particularly when I'm talking about a path and it's not a real path, it's a nominal path, is not as easy a concept, I think, to talk about to the general public. So I think one of the challenges is a communications challenge with nominal GDP targeting.

There are other characteristics of nominal GDP targeting in terms of thinking about how you hit that path and how you react to various shocks. That can potentially also be a bit of a challenge. I think it's certainly something that we should continue to think about. To some degree, you get some elements of that, when you have both inflation and unemployment to be thinking about, so nominal GDP tries to, in effect, boil that down to just one variable, rather than two.

I'm pretty comfortable with the two variables that we're focused on right now. I think we get most, but not all the benefits of nominal GDP targeting.

LIU: I think there were a question—yes, sir, in that middle table. Yes.

QUESTION: Yes, Darren Regas (ph), Credit Suisse. In your prepared comments, you made the phrase—or stated the phrase about zero bound economy. And then you quickly corrected that. Was that a Freudian slip on your forecast? And also...

ROSENGREN: It was a mistake.

(LAUGHTER)

QUESTION: Or is that an indication of your view on how this fiscal impasse and the potential shutdown is going to impact GDP growth over the coming, you know—the coming quarters? And, you know, if not, can you speak to that, please?

ROSENGREN: So I would hope we would do much better than zero percent growth going forward. And we were expecting more than 2 percent growth prior to the last three or four months. So that's what I would like to get back to is an economy that's growing faster than potential, faster than 2 percent growth. We're not there yet. It's not clear that we're going to get there in this quarter. We'll see.

We'll see how—I think it's a little uncertain to know exactly what's going to happen with the fiscal situation at this point. Hopefully there's an agreement, because it's probably premature to declare victory. We'll have to see if they actually come to a conclusion that's reasonable.

I actually think that we should be seeing GDP growing faster than 2 percent. I think it can grow faster than 2 percent. And the basic logic is pretty clear. We've had a lot of wealth improvement, in terms of housing prices have gone up, stock prices have gone up. As a result, you add more people getting back to work, with payroll employment growth growing not as fast as I would like, but growing.

You should have conditions for consumption to be picking up. You should also expect that at some point the fiscal headwinds are going to abate. That combination of fiscal headwinds abating and consumption picking up should be enough to get growth faster than 3 percent. Growth at 3 percent would start bringing down the unemployment rate. We'd get an unemployment rate going down for the right reason. Those conditions just haven't occurred, because we've had a series of negative shocks. Some of them have been self-induced. I think the last week has been an example of a self-induced shock that we might have had a better outcome by the fourth quarter if hadn't had this self-induced shock.

LIU: Question here.

QUESTION: I'm Padma Desai. I'm the Harriman Professor at Columbia University. I have a question regarding your table, second table. It would seem that toward the end of the year, the Fed would reduce its purchases of assets in the market, the probability suggests that, and interest rates are likely to go up, but that would attract funds into the U.S. from emerging market economies from around the world, right, from Japan to China to Brazil to India.

You talked about competing interests. Would you worry about the impact of this tapering toward the end of the year, how it would impact the economies and the exchange rates of these emerging market economies?

ROSENGREN: So we have to think about how our actions are going to affect the world economy, as well as our own economy. So does the Bank of Japan. So does the ECB. So does the Bank of England. And we're a big economy, so we can have a big impact on the rest of the world.

But do we focus primarily on what's happening to the rest of the world? No, we think about it more in terms of, if we're growing more slowly, they're likely to grow more slowly, and that's going to have a feedback loop that we have to take into account. So we do take into account, but our primary focus is trying to get our domestic economy to be appropriately balanced, to get back to full employment.

In the long run, I think that's going to be a better outcome for emerging markets, as well. Many of those emerging market economies—one of the primary determinants of how they're doing is their exchange rate, but another determinant is how much they're exporting, and we're one of the major export markets for many of these economies. So a strong, healthy U.S. economy in the end will be—have positive effects on many of these emerging market economies.

So I think we need to get it right for our own domestic reasons, and over time I think that's going to get you the right answer for some of the emerging markets. There may be a fair amount of volatility between now and where we want to be. We have to take that into account. We certainly have to be concerned about whether there are financial cracks, that there may be more of a shock than we're anticipating that could reverberate back to the United States.

So we certainly have to take it into account. But we can't use that as our primary determinant. We are setting policy for the United States in the same way that when Bank of Japan makes their decisions, I think they're doing it because they want to get the right outcome for Japan, the same thing with the Europeans when they're setting their own monetary policy.

LIU: Yes?

QUESTION: (inaudible) most recently at Blackstone. I have a question that you probably have thought about yourself, and that is because so much economic activity is done in the last—last couple months of year, you guys are probably really worried about this fiscal—this fiscal crisis. And if there is no fiscal agreement, the market is going to look to the Fed, once again, to sort of fill in the gap. And what are you guys thinking about from a contingency planning perspective to fill in the gap of economic activity, if there can be no fiscal agreement?

ROSENGREN: So it depends on what no fiscal agreement actually means. So whether, is it the continuing resolution? Or is it the debt ceiling?

So the debt ceiling has the possibility of having a much broader impact. In some sense, it's a shutdown on steroids, so that it would immediately force the U.S. economy to have a balanced budget and it would be in conjunction with the likelihood that interest rates on U.S. borrowing would go up, not only now, but for the foreseeable future. And that wouldn't just be interest rates on U.S. debt, it would be when you go to get your car loan, when you get your home loan. Those are all tied to the Treasury yield curve. So it would raise rates for all Americans, not only as taxpayers, but also just in terms of other types of items that they're purchasing.

So that obviously has much broader ramifications, so if we were to get into a debt ceiling situation, the idea that we would have draconian fiscal austerity immediately, in conjunction with the financial concerns, that interest rates might go up not just at that point in time, but for the foreseeable future, that is a place that I would hope we would not get to. So I'm hoping you're holding that off and just focusing really on the continuing resolution and just the shutdown itself, which does have an impact, and the impact is important, and the longer it goes on, the more important it becomes, so we have to take that into account.

Monetary policy can mitigate some of these shocks, but it can't offset it. So if we really have bad fiscal decisions that get made, and it really has a big impact on GDP, we can try to reduce some of that impact by lowering interest rates, but we can't lower interest rates enough and get individuals to be confident enough to be making the purchases of autos and houses to offset this really large shock.

So we have to be humble with how much monetary policy can do. We would try to lean against the wind in terms of getting the policy to be closer to what we want to get to be on the right path. But we can't offset big shocks. We can only mitigate them.

So we take it into account. We certainly think about it. We spend a lot of time thinking about financial stability and kind of—what kind of cracks could occur. And particularly for the debt ceiling, I think one of the challenges is, in the same way that around the failure of Lehman there were certain things that we just didn't perfectly anticipate. One of that was the runs on the money market fund industry.

So similarly, if we have a debt ceiling problem, it's going to be hard to predict how market participants are affected, which parts of financial markets might become much more illiquid, and how that may affect the balance sheets of a wide variety of firms and investors.

So I think that uncertainty is an unfortunate thing. And even if we only—so take the debt ceiling off. If you only thought about government shutdowns, if government shutdowns continually happen, it does start having an impact on how people think about the stability of our own government, how people think about what implications that has for their own spending decisions.

So even when we don't go over fiscal cliffs, even when we don't have some of these things, if people think it's going to become more common to have these kind of situations, that alone starts to have an impact on the economy, only some of which we're likely to be able to completely offset.

LIU: Yeah, I mean, there were reports already about banks stocking up on their ATM machines just in case people were going to go to their—their ATMs to withdrawal more cash, because they're just panicking, you know? They're hearing all this in the news.

I think we have time for one more question. So, yes?

QUESTION: Juan Ocampo, Trajectory Asset Management. I want to get back to your comment, how you thought in the June meeting that by September you would vote for tapering, and what happened was that the—the actual economic growth was not as strong as you had expected. And if you look at the central tendency forecast, actual growth, I think, has been pretty consistently missing what the forecasts have been throughout the whole QE period, if you want to talk about that, that way.

In medical terms, one might say this is failure to thrive. It's not a dying patient. You're not at zero. You know, it's not truly crisis. But on the other hand, you don't want to release the patient from the hospital yet, because they're not thriving. And it seems to be constantly just missing this target.

The question I'd have for is to respond to some of the criticism that people have about QE, and it goes along the following lines, that liquidity stimulus, whether it's traditional or nontraditional, can be extremely effective as long as done for a short burst, you know, in a true crisis and—but like steroids, which you mentioned before, if you keep administering them for too long, the side effects become bigger and bigger and bigger, and it gets to be an issue of whether that itself is contributing more than it's curing.

Now, two...

LIU: So the question? I'm sorry. We're running out of time. So what is your question?

QUESTION: No, the question here is whether, in fact, some of the side effects are creating a maximum speed limit, if you will, to the growth of the economy that is taking you below the 2 percent threshold that you've got. There are two that people talk about. One of them is, with the very low interest rates, you don't have the kind of flushing out of the economy of bad loans, businesses that should have been taken under, the creative restructuring, if you will, and the second one is just the fact that you've got uncertainty in terms of when QE is going to be lifted and what the true, if you will, you know, economics are, so businesses will tend to hold back until they get what they think is a cleaner read. Those are two of many that people talk about.

LIU: Right.

QUESTION: And you guys meet this issue. And how do you react to that?

ROSENGREN: So monetary policy works through moving interest rates. And that's both short rates, and it's long rates. As a result of just the actions that have occurred over the last six months, we've seen interest rates can move very substantially to relatively small changes in expectations about what monetary policy will be.

So I think we do have the ability to move interest rates and asset prices, even though we've been doing quantitative easing for quite a while. And I do think that the interest-sensitive sectors have actually responded quite substantially to that. In part, it's not just the fact that the interest-sensitive sectors are doing better. It's something like housing prices going up and not down, which is something that was much less certain a year ago than it is now, so we've had a full year where housing prices were improving. That's one reason why residential investment is doing well.

So low rates at a time when housing prices are going down doesn't induce much behavior that we'd want. The reason's clear, because it's a leveraged transaction that even at lower interest rates you don't want to buy an asset that you can buy in six months for a much lower price.

So in an environment where housing prices are going down, interest policies aren't going to be as effective at changing housing demand. I think part of what we got in this last year was, people now have the expectation that housing prices are more likely to go up than to go down. In combination with low interest rates, that has stimulated people to start purchasing houses and to think about their two costs, if they decide to wait and continue to rent rather to purchase a home.

There's a cost in that interest rates may be higher, and there's a cost in that the price may be higher. So I think that has been a very beneficial aspect of what the quantitative easing program has done.

Now, that's not to say that there are no collateral impact. And one of the benefits of having higher long-term rates now than we had back in May is some of those concerns about very low interest rates for a long period of time have abated. The rates are not nearly as low as they were six months ago. And so some of the concerns about these abnormally—some of the concerns that you've raised about keeping rates low for a long time, I think, have partly been abated by they're not nearly as low as they were.

We need to look for what's happening in other asset markets. We have to worry about asset bubbles. My own view is that right now I'm not seeing the fragility in many of those markets that would argue that we should have a policy that would result in us getting to full employment in a much longer period of time. It's already been too long. The unemployment rate's still too high. The inflation rate's still too low. That's the kind of context that I think accommodative policy is important.

And the worst outcome would be an outcome that Japan faced. So in 1997, they decided to tighten policy, because they were worried inflation was picking up too fast. The result was that they had an extended period where GDP was too slow, inflation was too low, and labor markets were not strong enough.

So the goal is to try to make sure that we avoid that outcome. That means trying to get in a reasonable period of time back to an economy that's closer to that inflation target of 2 percent and closer to full employment.

So I think we have to weigh the costs and benefits while we're thinking about these decisions. But ideally, we're going to try to get back to full employment more quickly. I think it'll help both the economy overall. It'll help with the fiscal situation. There are a lot of things that could be helped if we get back to full employment more quickly than what we've been doing.

LIU: All right. Thank you very much, Eric Rosengren, for joining us on this program. Thank you.

ROSENGREN: Thank you, Betty.

(APPLAUSE)