We use cookies on our website to give you the best online experience. Please know that if you continue to browse on our site, you agree to this use. You can always block or disable cookies using your browser settings. To find out more, please review our privacy policy.

About


Policy Hub: Macroblog provides concise commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues for a broad audience.

Authors for Policy Hub: Macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

Comment Standards:
Comments are moderated and will not appear until the moderator has approved them.

Please submit appropriate comments. Inappropriate comments include content that is abusive, harassing, or threatening; obscene, vulgar, or profane; an attack of a personal nature; or overtly political.

In addition, no off-topic remarks or spam is permitted.

September 8, 2016

Introducing the Atlanta Fed's Taylor Rule Utility

Simplicity isn't always a virtue, but when it comes to complex decision-making processes—for example, a central bank setting a policy rate—having simple benchmarks is often helpful. As students and observers of monetary policy well know, the common currency in the central banking world is the so-called "Taylor rule."

The Taylor rule is an equation introduced by John Taylor in a seminal 1993 paper that prescribes a value for the federal funds rate—the interest rate targeted by the Federal Open Market Committee (FOMC)—based on readings of inflation and the output gap. The output gap measures the percentage point difference between real gross domestic product (GDP) and an estimate of its trend or potential.

Since 1993, academics and policymakers have introduced and used many alternative versions of the rule. The alternative forms of the rule can supply policy prescriptions that differ significantly from Taylor's original rule, as the following chart illustrates.

Effective federal funds rate and prescriptions from alternative versions of the Taylor rule
(enlarge)

The green line shows the policy prescription from a rule identical to the one in Taylor's paper, apart from some minor changes in the inflation and output gap measures. The red line uses an alternative and commonly used rule that gives the output gap twice the weight used for the Taylor (1993) rule, derived from a 1999 paper by John Taylor. The red line also replaces the 2 percent value used in Taylor's 1993 paper with an estimate of the natural real interest rate, called r*, from a paper by Thomas Laubach, the Federal Reserve Board's director of monetary affairs, and John Williams, president of the San Francisco Fed. Federal Reserve Chair Janet Yellen also considered this alternative estimate of r* in a 2015 speech.

Both rules use real-time data. The Taylor (1993) rule prescribed liftoff for the federal funds rate materially above the FOMC's 0 to 0.25 percent target range from December 2008 to December 2015 as early as 2012. The alternative rule did not prescribe a positive fed funds rate since the end of the 2007–09 recession until this quarter. The third-quarter prescriptions incorporate nowcasts constructed as described here. Neither the nowcasts nor the Taylor rule prescriptions themselves necessarily reflect the outlook or views of the Federal Reserve Bank of Atlanta or its president.

Additional variables that get plugged into this simple policy rule can influence the rate prescription. To help you sort through the most common variations, we at the Atlanta Fed have created a Taylor Rule Utility. Our Taylor Rule Utility gives you a number of choices for the inflation measure, inflation target, the natural real interest rate, and the resource gap. Besides the Congressional Budget Office–based output gap, alternative resource gap choices include those based on a U-6 labor underutilization gap and the ZPOP ratio. The latter ratio, which Atlanta Fed President Dennis Lockhart mentioned in a November 2015 speech while addressing the Taylor rule, gauges underemployment by measuring the share of the civilian population working their desired number of hours.

Many of the indicator choices use real-time data. The utility also allows you to establish your own weight for the resource gap and whether you want the prescription to put any weight on the previous quarter's federal funds rate. The default choices of the Taylor Rule Utility coincide with the Taylor (1993) rule shown in the above chart. Other organizations have their own versions of the Taylor Rule Utility (one of the nicer ones is available on the Cleveland Fed's Simple Monetary Policy Rules web page). You can find more information about the Cleveland Fed's web page on the Frequently Asked Questions page.

Although the Taylor rule and its alternative versions are only simple benchmarks, they can be useful tools for evaluating the importance of particular indicators. For example, we see that the difference in the prescriptions of the two rules plotted above has narrowed in recent years as slack has diminished. Even if the output gap were completely closed, however, the current prescriptions of the rules would differ by nearly 2 percentage points because of the use of different measures of r*. We hope you find the Taylor Rule Utility a useful tool to provide insight into issues like these. We plan on adding further enhancements to the utility in the near future and welcome any comments or suggestions for improvements.

September 21, 2015

What Do U.S. Businesses Know that New Zealand Businesses Don't? A Lot (Apparently).

A recent paper presented at the Brookings Institute, picked up by the Financial Times and the Washington Post, suggests that when it comes to communicating their inflation objective, central banks have a lot of work to do. This conclusion is based primarily on two pieces of evidence.

The first piece is that when businesses in New Zealand are asked about their expectations for changes in "overall prices"—which presumably corresponds with their inflation expectation—the responses, on average, appear to be much too high relative to observed inflation trends. And the responses vary widely from business to business. According to this survey, the average firm in New Zealand expects 4 to 5 percent inflation on a year-ahead basis, and 3.5 percent inflation over the next five to 10 years. Those expectations are for the average firm. Apparently, about one in four firms in New Zealand think inflation in the year ahead will be more than 5 percent, and about one in six firms believe inflation will top 5 percent during the next five to 10 years. Certainly, these aren't the responses one would expect from businesses operating in an economy (like New Zealand) where the central bank has been targeting 2 percent inflation for the past 13 years, over which time inflation has averaged only 2.2 percent (and a mere 0.9 percent during the past four years).

But count us skeptical of this evidence. In this paper from last year, we challenge the assumption that asking firms (or households, for that matter) about expected changes in "overall prices" corresponds to an inflation prediction.

The second piece of evidence regarding the ineffectiveness of inflation targeting is more direct—the authors of this paper actually asked New Zealand businesses a few questions about the central bank and its policies, including this one:

What annual percentage rate of change in overall prices do you think the Reserve Bank of New Zealand is trying to achieve? (Answer: ______%)

The distribution of answers by New Zealand firms is shown in the chart below. According to the survey, the median New Zealand firm appears to think the central bank's inflation target is 5 percent. Indeed, more than a third of firms in New Zealand reported that they think the central bank is targeting an inflation rate greater than 5 percent. Only about 12 percent of the firms were able to correctly identify their central bank's actual inflation target of 2 percent (actually, the New Zealand inflation target is a range of between 1 and 3 percent, centered on 2 percent).


If this weren't embarrassing enough for central bankers, the study also reports that New Zealand households (like U.S. households) don't seem to know who the head of the central bank is. In fact, the authors show that there are more online searches for "puppies" than for information about macroeconomic variables.

OK, to be honest, we don't find that last result very surprising. Puppies are adorable. Central bankers? Not so much. But we were very surprised to see just how high and wide-ranging businesses in New Zealand perceived their central bank's inflation target to be. We're surprised because that bit of information doesn't fit with our understanding of U.S. firms.

In December 2011, the month before the Fed officially announced an explicit numerical target for inflation, we wanted to know whether firms had already formed an opinion about the Fed's inflation objective. So we asked a panel of Southeast businesses the following question:

150921-table

What we learned was that 16 percent of the 151 firms who responded to our survey had no opinion regarding what rate of inflation the Federal Reserve was aiming for. But of the firms that had an opinion, 58 percent identified a 2 percent inflation target.

But perhaps this isn't a fair comparison to the recent survey of New Zealand businesses. In our 2011 survey, firms had only six options to choose from (including "no opinion"). It could be that our choice of options biased the responses away from high inflation values. So last week, we convened another panel of firms and asked the question in the same open-ended format given to New Zealanders:

What annual rate of inflation do you think the Federal Reserve is aiming for over the long run? (Answer: ______%)

The only material distinction between their question and ours is that we substituted the word "inflation" for the phrase "changes in overall prices." (For this special survey, we polled a national sample of firms that had never before answered one of our survey questions.) The chart below shows what we found relative to the results recently reported for New Zealand firms.


Our survey results look very similar to our results of four years ago. About one in five of the 102 firms that answered our survey was unsure about the Fed's inflation target. But almost 53 percent of the firms that responded answered 2 percent. (On average, U.S. firms judged the central bank's inflation target to be 2.2 percent, just a shade higher than our actual target.)

Furthermore, the distribution of responses to our survey was very tightly centered on 2 percent. The highest estimate of the Fed's inflation target (from only one firm) was 5 percent. So again, our results don't at all resemble what has been reported for the firms down under.

Why is there a glaring difference between what the survey of New Zealand firms found and what we're finding? Well, as noted earlier, we've got our suspicions, but we'll keep studying the issue. And in the meantime, have you seen this?

Photo of Mike Bryan
By Mike Bryan, vice president and senior economist,
Photo of Brent Meyer
Brent Meyer, assistant policy adviser and economist, and
Photo of Nicholas Parker
Nicholas Parker, economic policy specialist, all in the Atlanta Fed's research department

Editor's note: Learn more about inflation and the consumer price index in an ECONversations webcast featuring Atlanta Fed economist Brent Meyer.

September 4, 2015

5-Year Deflation Probability Moves Off Zero

Since 2010, our Bank has regularly posted 5-year deflation probabilities derived from prices of Treasury Inflation-Protected Securities (TIPS) on our Deflation Probabilities web page. Each deflation probability, which measures the likelihood of a decline in the Consumer Price Index over a fixed five-year window, is estimated by comparing the price of a recently issued 5-year TIPS with a 10-year TIPS issued about five years earlier. Because the 5-year TIPS has more "deflation protection" than the 10-year TIPS, the implied deflation probability rises when the 5-year TIPS becomes more valuable relative to the 10-year TIPS. (See this macroblog post for a more detailed explanation, or this appendix with the mathematical details.)

From early September 2013 to the first week of August 2015, the five-year deflation probability estimated with the most recently issued 5-year TIPS was identically 0 as the chart shows.


Of course, we should not interpret this long period of zero probability of deflation too literally. It could easily be the case that the "true" deflation probability was slightly above zero but that confounding factors—such as differences in the coupon rates, maturity dates, or liquidity of the TIPS issues—prevented the model from detecting it.

Since August 11, however, the deflation probability has had its own "liftoff" of sorts, fluctuating between 0.0 and 1.3 percent over the 16-day period ending August 26 before rising steadily to 4.1 percent on September 2. Of course, this rise off zero could be temporary, as it proved to be in the summer of 2013.

How seriously should we take this recent liftoff? We can look at options prices on Consumer Price Index inflation (inflation caps and floors) to get a full probability distribution for future inflation; see this published article by economists Yuriy Kitsul and Jonathan Wright or a nontechnical summary in this New York Times article. An alternative is simply to ask professional forecasters for their subjective probabilities of inflation falling within various ranges like "1.0 to 1.4 percent," "1.5 to 1.9 percent," and so forth. The Philly Fed's Survey of Professional Forecasters does just this, with the chart below showing probabilities of low inflation for the Consumer Price Index excluding food and energy (core CPI) from each of the August surveys since 2007.


Although the price index, and the horizon for the inflation outcome, differs from the TIPS-based deflation probability, we see that the shape of the curves is broadly similar to the one shown in the first chart. In the most recent survey, the probability that next year's core CPI inflation rate will be low was small and not particularly elevated relative to recent history. However, the deadline date for this survey was August 11, before liftoff in either the TIPS-based deflation probability or the recent volatility in global financial markets. So stay tuned.

Photo of Pat Higgins
By Pat Higgins, senior economist in the Atlanta Fed's research department

November 4, 2014

Data Dependence and Liftoff in the Federal Funds Rate

When asked "at which upcoming meeting do you think the FOMC [Federal Open Market Committee] will FIRST HIKE its target for the federal funds rate," 46 percent of the October Blue Chip Financial Forecasts panelists predicted that "liftoff" would occur at the June 2015 meeting, and 83 percent chose liftoff at one of the four scheduled meetings in the second and third quarters of next year.

Of course, this result does not imply that there is an 83 percent chance of liftoff occurring in the middle two quarters of next year. Respondents to the New York Fed's most recent Primary Dealer Survey put this liftoff probability for the middle two quarters of 2015 at only 51 percent. This more relatively certain forecast horizon for mid-2015 is consistent with the "data-dependence principle" that Chair Yellen mentioned at her September 17 press conference. The idea of data dependence is captured in this excerpt from the statement following the October 28–29 FOMC meeting:

[I]f incoming information indicates faster progress toward the Committee's employment and inflation objectives than the Committee now expects, then increases in the target range for the federal funds rate are likely to occur sooner than currently anticipated. Conversely, if progress proves slower than expected, then increases in the target range are likely to occur later than currently anticipated.

If the timing of liftoff is indeed data dependent, a natural extension is to gauge the likely "liftoff reaction function." In the current zero-lower bound (ZLB) environment, researchers at the University of North Carolina and the St. Louis Fed have analyzed monetary policy using shadow fed funds rates, shown in figure 1 below, estimated by Wu and Xia (2014) and Leo Krippner.

Unlike the standard fed funds rate, a shadow rate can be negative at the ZLB. The researchers found that the shadow rates, particularly Krippner's, act as fairly good proxies for monetary policy in the post-2008 ZLB period. Krippner also produces an expected time to liftoff, estimated from his model, shown in figure 1 above. His model's liftoff of December 2015 is six months after the most likely liftoff month identified by the aforementioned Blue Chip survey.

I included Krippner's shadow rate (spliced with the standard fed funds rate prior to December 2008) in a monthly Bayesian vector autoregression alongside the six other variables shown in figure 2 below.

The model assumes that the Fed cannot see contemporaneous values of the variables when setting the spliced policy—that is, the fed funds/shadow rate. This assumption is plausible given the approximately one-month lag in economic release dates. The baseline path assumes (and mechanically generates) liftoff in June 2015 with outcomes for the other variables, shown by the black lines, that roughly coincide with professional forecasts.

The alternative scenarios span the range of eight possible outcomes for low inflation/baseline inflation/high inflation and low growth/baseline growth/high growth in the figures above. For example, in figure 2 above, the high growth/low inflation scenario coincides with the green lines in the top three charts and the red lines in the bottom three charts. Forecasts for the spliced policy rate are conditional on the various growth/inflation scenarios, and "liftoff" in each scenario occurs when the spliced policy rate rises above the midpoint of the current target range for the funds rate (12.5 basis points).

The outcomes are shown in figure 3 below. At one extreme—high growth/high inflation—liftoff occurs in March 2015. At the other—low growth/low inflation—liftoff occurs beyond December 2015.

One should not interpret these projections too literally; the model uses a much narrower set of variables than the FOMC considers. Nonetheless, these scenarios illustrate that the model's forecasted liftoffs in the spliced policy rate are indeed consistent with the data-dependence principle.

Photo of Pat HigginsBy Pat Higgins, senior economist in the Atlanta Fed's research department