Nature Abhors an Output Gap

In The Washington Post, Neil Irwin highlights a shortcoming that I know all too well:

Throughout the halting economic recovery that began in 2009, the formal economic projections released by the Congressional Budget Office, White House Council of Economic Advisers, and Federal Reserve have displayed quite a consistent pattern: This year may be one of sluggish growth, they acknowledge. But stronger growth, of perhaps 3.5 percent, is just around the corner, and will arrive next year.

Consider, for example, the Fed's projections in November of 2009. Sure, growth would be slow in 2010, they held. But 2011 growth, they expected, would be 3.4 to 4.5 percent, and 2012 would 3.5 to 4.8 percent growth. The actual levels of growth were 2 percent in 2011 and 1.5 percent in 2012.

What's amazing is that the Fed's newest projections, released in December of 2012, look like they could have been copy and pasted from 2009, just with the years changed: They forecast sluggish growth in 2013, 2.3 to 3 percent, followed by a pickup to 3 to 3.5 percent in 2014 and 3 to 3.7 percent in 2015.

I, for one, am guilty as charged, and feel pretty fortunate that the offense is not a hanging one. In fact I don't think Irwin's indictment is overly harsh, and he is on the right track when he offers up this explanation for the last several years' persistently overly rosy projections:

Economic forecasters tend to look at past experience and extrapolate; in the past, when there has been a recession, the very forces that caused the recession become unwound, sowing the seeds for expansion...

Here is a basic fact about macroeconomic forecasting. The truly powerful driver of forecasts is mean reversion, which is the tendency of models to predict that gross domestic product (GDP) will move toward an average trend over time. This fact holds true whether we are talking about formal statistical analysis or the intuitive judgmental adjustments that all forecasters apply to their formal statistical models.

Forecasters are not completely robotic, of course. Irwin is correct when he says "forecasters tend to look at past experience and extrapolate, but forecasters do leaven past experience with incoming details that alter judgments about what is the mean—the "normal state," if you will—to which the economy will converge. But whatever is that normal state, our models insist that we will converge to it.

Nothing illustrates this property of forecasting reality better than this chart, which supplements the latest economic projections from the Congressional Budget Office:

The potential GDP line in that chart is the level of production that represents the structural path of the economy. Forecasters, no matter where they think that potential GDP line might be, all believe actual GDP will eventually move back to it. "Output gaps"—the shaded area representing the cumulative miss of actual GDP relative to its potential—simply won't last forever. And if that means GDP growth has to accelerate in the future (as it does when GDP today is below its potential)—well, that's just the way it is.

Unfortunately, potential GDP is not so simple to divine. We have to guess (or, more generously, estimate) what it is. That guessing game has been harder than usual over the past several years. Here is the record of the CBO's potential GDP since 2009:

I think this picture is a fairly representative record of how views about the potential level of U.S. GDP has evolved over the past several years. What has not been resolved is the debate over what conclusions should be drawn from persistent overestimates of potential and serial misses to the high side on GDP projections.

Irwin seems to be of two minds. On the one hand he offers very structural-sounding reasons for poor forecasting experience:

... the financial-crisis-induced recession of 2008–2009 was so deep that it had deep-seated effects that go beyond those explained by those traditional relationships. It messed up the workings of the financial system, and banks are still trying to figure out what the new one looks like.

On the other hand, he makes appeals to very traditional explanations tied to deficient spending and insufficient policy stimulus (though even here structural change may be one reason that stimulus has been insufficient):

Breakdowns in the financial system mean that low-interest rate policies from the Fed don't have their usual punch. An overhang of household debt means that consumers hold their wallets more than usual. Federal fiscal stimulus to offset those effects is now long-over...

This much, in any event, is clear: Given any starting point where the level of GDP is below its potential level—that is, given an output gap—forecasts will include a bounce back in GDP growth above its long-run average, at least for a while. That's just the way it works.

If, contrary to conventional wisdom, you believe that the true output gaps are much smaller than suggested in the CBO picture above, you might want to take the under on a bet to whether GDP forecasts will prove too optimistic once again.

By Dave Altig, executive vice president and research director of the Atlanta Fed

 

Still a Skeptic: Addressing a Few Questions about Nominal GDP Targeting

In a comment to last week's post on inflation versus price-level targeting, David Beckworth asks the following (referring back to an even earlier post on nominal gross domestic product [NGDP] targeting):

You refer back to your previous post on NGDP level targeting, but fail to take note of the comments that respond to your concerns about it. Specifically, see the ones by Andy Harless and Gregor Bush. Would love to see your response to those ones. Do you have a response for them? I am listening if you have one.

Here is an excerpt from the Harless comment...

Most people who advocate NGDP targeting today advocate level path targeting, not growth rate targeting. I don't believe that your "historical justification" applies in this case. Indeed, I think it makes the case for level targeting (of either the price level or NGDP, but there are reasons to prefer the latter) relative to the current system which centers on a growth rate target for the price level (in other words, an inflation target).

...and here is the Bush comment:

Just to add to Andy's point, advocates of NGDP level targeting argue that it's precisely because of uncertainty around estimates [of] potential output [that] NGDP targeting should be adopted. They argue that [as] long as the central bank keeps nominal spending on, say, a 5% trend line, there will be neither demand side recessions (mass unemployment) nor high inflation. In other words, AD will be stable and this will produce a stable macroeconomic environment. Whether inflation is 2% and real output [grows] at 3% or inflation is 3% and real output grows at 2% is of no concern.

In the post on NGDP targeting I was in fact thinking about level targeting, and Gregor Bush's last sentence gets to—in fact is—the heart of our disagreement. I am just not willing to concede that anchoring long-term inflation by saying something like "2 percent, 3 percent, whatever" is the path to sustaining central bank credibility. Over the longer term, inflation is the only thing that monetary policy can reliably deliver, as the Federal Open Market Committee (FOMC) has clearly articulated in its statement of longer-run goals and policy strategy:

The inflation rate over the longer run is primarily determined by monetary policy, and hence the Committee has the ability to specify a longer-run goal for inflation. The Committee judges that inflation at the rate of 2 percent, as measured by the annual change in the price index for personal consumption expenditures, is most consistent over the longer run with the Federal Reserve's statutory mandate...

The maximum level of employment is largely determined by nonmonetary factors that affect the structure and dynamics of the labor market. These factors may change over time and may not be directly measurable. Consequently, it would not be appropriate to specify a fixed goal for employment; rather, the Committee's policy decisions must be informed by assessments of the maximum level of employment, recognizing that such assessments are necessarily uncertain and subject to revision.

This excerpt does not imply, of course, that the Fed need slavishly pursue a numerical inflation target in the shorter run and, as I have pointed out before, in his last press conference Chairman Bernanke explicitly indicated that the FOMC does not intend to do so:

The Committee... intends to look through purely transitory fluctuations in inflation, such as those induced by short-term variations in the prices of internationally traded commodities, and to focus instead on the underlying inflation trend.

My price-level targeting post, co-authored with Mike Bryan, was exactly making the point that, over the past couple of decades, the FOMC has essentially delivered on a 2 percent longer-term price-level growth objective, while accepting plenty of shorter-term variability.

In the end, it is an open question whether credibility in delivering price stability, hard won in the '80s and early '90s, could be sustained if the FOMC says it does not care so much about the exact level of the average rate of inflation, even in the long run. To be truthful, I can't give you an answer to that question. But neither can the proponents of NGDP targeting. I just don't feel that this is an opportune time for an experiment.

Update: Scott Sumner responds.

By Dave Altig, executive vice president and research director of the Atlanta Fed

 

Nominal GDP Targeting: Still a Skeptic

In a few days the clock will run out on another year of disappointing economic growth in the United States and, generally speaking, in the world. It is inevitable and appropriate, then, that the year-end ritual of looking forward by looking backward will include an assessment of whether more or better policy can contribute to a pick-up in growth that failed to materialize in 2012.

To this discussion, Harvard professor Jeff Frankel brings some fresh thinking to the not-quite-fresh notion that the Fed should adopt a nominal gross domestic product (GDP) targeting approach as a replacement for existing central bank practice—described by Frankel and others as policy driven by an inflation-targeting framework. What I particularly like about Frankel's proposal is the fact that he offers up a practical roadmap for using the Fed's current communications tools to transition to an explicit nominal GDP targeting framework. If I were inclined to think such a move would be a good idea, I would view Frankel's proposal with some enthusiasm. Alas, I am not yet so inclined.

As to the case for skepticism on theoretical grounds, I commend to you this excellent post by Mark Thoma at Economist's View. But Professor Frankel suggests a case for nominal GDP targeting on practical grounds by appealing to this counterfactual:

A nominal GDP target for the US Federal Reserve might have avoided the mistake of excessively easy monetary policy during 2004-06, a period when nominal GDP growth exceeded 6 per cent.

Maybe. Average annual real GDP growth over those three years was just over 3 percent, compared to the Congressional Budget Office (CBO) estimates of potential GDP growth of just under 2.5 percent. That's not a big difference, but more importantly the average gap between the level of real GDP and the CBO estimate of potential was just 0.3 percent of average output—essentially zero. The importance of this so-called "output gap" becomes evident if you read the Michael Woodford interview referenced in the aforementioned piece at Economist's View. In that interview, Woodford says, "The idea was to talk about a price level, as opposed to the inflation rate, but a corrected price level target where you add to it some multiple of the real output gap." So for him, something like this measure would be a key element of his proposed monetary policy rule.

What if, rather than some measure of nominal GDP, the 2004–06 Fed had instead been solely focused on the inflation rate? You can't answer that question without operationalizing what it means to be "focused on the inflation rate," but for the sake of argument let's simply consider actual annualized PCE inflation over a two-year horizon. (In his press statement explaining the Federal Open Market Committee's (FOMC) latest decision, Fed Chairman Ben Bernanke suggested using a one- to two-year horizon for inflation forecasting horizon to smooth through purely transitory influences on inflation that the central bank would inclined to "look through.") Here's the record, with the period from 2004 through 2006 highlighted:

If you really do think that there was a policy mistake over the 2004–06 period—and in particular if you believe the FOMC during that should have adopted a more restrictive policy stance—I'm hard-pressed to see what advantage is offered by focusing on nominal GDP rather than inflation alone. In fact, you could argue that that the GDP part of the nominal GDP target would have added just about nothing to the discussion.

I add the observation in the chart above to my earlier comments on an earlier Frankel call for nominal GDP targets. To summarize my concerns, the Achilles' heel of nominal GDP targeting is that it provides a poor nominal anchor in an environment in which there is great uncertainty about the path of potential real GDP. As I noted in my earlier post, there is historical justification for that concern.

Basically, anyone puzzling through how demographics are affecting labor force participation rates, how technology is changing the dynamics of job creation, or how policy might be altering labor supply should feel some humility about where potential GDP is headed. For me, a lack of confidence in the path of real GDP takes a lot of luster out of the idea of a nominal GDP target.

By Dave Altig, executive vice president and research director at the Atlanta Fed

 

The (Unfortunately?) Consistent Record of the Recovery

In his last two posts (here and here), economist Tim Duy has done some yeoman work displaying and discussing the economic context of monetary policy decisions past and prospective. Though Wednesday's self-titled post "Data Dump" focuses on the incoming data as a set-up to the next meeting of the Federal Open Market Committee (FOMC), what strikes me is the consistency of the broad macroeconomic outcomes over the course of the recovery. Gross domestic product (GDP) growth has pretty clearly clocked in at about 2 percent...

(enlarge)

...and, looking through the quarterly ups and downs, payroll employment growth has clearly trended near 150,000 jobs per month after a slower start in 2010:

(enlarge)

The inflation picture shows more variation...

(enlarge)

...but in my view, that sort of variation is why it makes sense to think in terms of medium-term performance. "Medium-term" is more a measure of art than science, and I would concede the point that the recovery as a whole would be on the shorter end of that time frame. Suffice it to say that the pace of price-level growth over the past two and a half years wouldn't contradict the presumption that inflation is pretty close to the FOMC's stated longer-run objective.

Duy looks at this performance and sees pretty clear evidence of failure:

The economy continues to settle into a path that is not consistent with either part of the Fed's dual mandate. Moreover, there are very real downside risks to even a tepid outlook...

This is frustrating. What in the world is the point of making a big claim to affirm the nature of the dual mandate and then subsequently ignore any forecasts that indicate you have no faith the elements of the dual mandate will be met anytime soon?

That complaint is not really about the inflation part of the mandate, but the employment/growth part of it. But if you are willing to accept that employment growth remains on a pace of 150,000 jobs per month—and I see no clear evidence to the contrary—it is not at all obvious that the pace of the recovery is inconsistent with the FOMC's view of achieving its dual mandate. Here, for example, are the central tendency ranges of the unemployment rate projections from the FOMC's June Summary of Economic Projections (SEP) and the employment growth that would be required to meet those objectives (with some important assumptions, such as the labor force participation rate remaining at the current level).

(enlarge)

Here is the important statement of conditionality, as described in the SEP document:

The charts show actual values and projections for three economic variables [GDP growth, the unemployment rate, and PCE inflation] based on FOMC participants' individual assessments of appropriate monetary policy.

Under appropriate policy—which pretty clearly means mandate-consistent outcomes—the majority of FOMC participants don't seem to think that the unemployment rate will improve that quickly. And, to my point, it is not clear that the trend in payroll employment is inconsistent with that pace of improvement.

Of course, individual contributors to the SEP may have different assumptions about things like the labor force participation rate. More importantly, the SEP is silent on what, in each contributor's view, constitutes "appropriate policy."

And I am certainly begging the important issues. Would the economy have achieved even the somewhat unspectacular pace of 2 percent GDP growth, 150,000 jobs per month, and average inflation near the long-run objective absent large-scale asset purchases ("QE2"), forward guidance (statements indicating that policy rates are expected to be exceptionally low through at least late 2014), and maturity extension programs ("Operation Twist")? Does "appropriate policy" imply that more must be done to achieve even the modest progress in the unemployment rate implied in my calculations above? And could we have (looking backward) or can we (looking forward) do even better with an even more aggressive approach, as many Fed critics argue?

Good questions, those.

By Dave Altig, executive vice president and research director at the Atlanta Fed

 

How big is the output gap?

It's fair to say, I think, that the question posed in the title of this blog post is at the heart of any monetary policy debate.

Here's how the discussion went at the June meeting of the Federal Open Market Committee (FOMC):

"Meeting participants again discussed the extent of slack in labor markets. Some participants judged that the unemployment rate was being substantially boosted by structural factors such as mismatches between the skills of unemployed workers and those required for available jobs....One implication of the view that there is relatively little slack is that providing more monetary stimulus would be likely to raise inflation above the Committee's objective. Some other participants acknowledged that structural factors were contributing to unemployment, but said that, in their view, slack remained high and weak aggregate demand was the major reason that the unemployment rate was still elevated. These participants cited a range of evidence to support their judgment....These arguments imply that slack in labor markets remains considerable and therefore that a reduction in the unemployment rate toward its longer-run normal level would not have much effect on inflation."

If you want more specifics about these contrasting views, you might find recent speeches by some FOMC meeting participants helpful. Jeff Lacker, president of the Federal Reserve Bank of Richmond, is pretty clearly in the "relatively little slack" group:

"It's worth noting...that the effects of unemployment insurance benefits together with the effects of labor market inefficiencies could plausibly account for a quite substantial portion of our elevated unemployment rate. The quantitative estimates of labor market mismatch come from independent methods and datasets and, in principle, measure conceptually distinct inefficiencies. We shouldn't necessarily assume these effects are additive, but combining all three together yields a range of 2.9 to 5.9 percentage points, which is sizable relative to the increase in the total unemployment rate of 5-½ percentage points during the recession."

Vice Chairman Janet Yellen is also pretty clearly on the other side of the debate:

"A critical question for monetary policy is the extent to which these numbers reflect a shortfall from full employment versus a rise in structural unemployment. While the magnitude of structural unemployment is uncertain, I read the evidence as suggesting that the bulk of the rise during the recession was cyclical, not structural in nature.

"Consider...the difference between the actual unemployment rate and the Congressional Budget Office (CBO) estimate of the rate consistent with inflation remaining stable over time...[the] index of the difficulty households perceive in finding jobs...[and the] index of firms' ability to fill jobs....All three measures show similar cyclical movements over the past 20 years, and all now stand at very high levels."

The positions outlined above lay bare why estimates of the output gap command such weight in the discussion of monetary policy—both ends of the FOMC's dual mandate of maximum employment and price stability may run through it. If the output gap is large, that is, if the level of gross domestic product (GDP) is running significantly under potential GDP, the economy is obviously not in a position of maximum employment. And if that is the case, the inflation trend is likely to be headed lower and so the price stability mandate may also be in jeopardy.

Where do I come out in this debate? That isn't important since I don't get a vote. But my boss, Dennis Lockhart does, and he laid out his position in a recent speech to the Mississippi Economic Council.

"I think the output gap—the amount of slack in the economy—is neither as sizeable as the high-end estimates, nor is it zero. If there were no slack at all, 8.2 percent unemployment would represent full employment. If this were so, the economy would have undergone profound structural change over the last five years. As I weigh the findings of research by Federal Reserve economists and others, I do not think a compelling case has yet been made that structural adjustment has played a dominant role in slowing growth and progress against unemployment.

"If, on the other hand, slack in the economy were close to the high estimates, we should have seen more and more persistent downward pressure on prices and wages than has, in fact, been the case. Deciding on the extent of the output gap is not straightforward. I believe the truth is in the gray middle."

To emphasize, this "gray middle" isn't a compromise but a weighing of the available evidence. If the GDP gap really is close to zero, the profound structural change that the economy ought to have experienced hasn't found great support in the data. But if this is just a bigger version of gaps of recessions past, then where is the great disinflationary pressure such slack would ordinarily imply?

You may have another view and, again quoting Lockhart's recent speech, "reasonable people can consider the issues...and come to different conclusions. "And if you're having trouble getting a grasp on GDP, potential GDP, and why the measurement of our national potential isn't an easy task, perhaps we can help. We've recently produced an educational video on the issue. It's not a deep treatment of the issue—in fact, just the opposite. It's a jumping-in point for those who are interested in the policy debate but haven't a clue what a GDP gap is. As always, let us know what you think.

By Mike Bryan, vice president and senior economist in the research department at the Atlanta Fed