The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.
- BLS Handbook of Methods
- Bureau of Economic Analysis
- Bureau of Labor Statistics
- Congressional Budget Office
- Economic Data - FRED® II, St. Louis Fed
- Office of Management and Budget
- Statistics: Releases and Historical Data, Board of Governors
- U.S. Census Bureau Economic Programs
- White House Economic Statistics Briefing Room
January 17, 2018
What Businesses Said about Tax Reform
Many folks are wondering what impact the Tax Cuts and Jobs Act—which was introduced in the House on November 2, 2017, and signed into law a few days before Christmas—will have on the U.S. economy. Well, in a recent speech, Atlanta Fed president Raphael Bostic had this to say: "I'm marking in a positive, but modest, boost to my near-term GDP [gross domestic product] growth profile for the coming year."
Why the measured approach? That might be our fault. As part of President Bostic's research team, we've been curious about the potential impact of this legislation for a while now, especially on how firms were responding to expected policy changes. Back in November 2016 (the week of the election, actually), we started asking firms in our Sixth District Business Inflation Expectations (BIE) survey how optimistic they were (on a 0–100 scale) about the prospects for the U.S. economy and their own firm's financial prospects. We've repeated this special question in three subsequent surveys. For a cleaner, apples-to-apples approach, the charts below show only the results for firms that responded in each survey (though the overall picture is very similar).
As the charts show, firms have become more optimistic about the prospects for the U.S. economy since November 2016, but not since February 2017, and we didn't detect much of a difference in December 2017, after the details of the tax plan became clearer. But optimism is a vague concept and may not necessarily translate into actions that firms could take that would boost overall GDP—namely, increasing capital investment and hiring.
In November, we had two surveys in the field—our BIE survey (undertaken at the beginning of the month) and a national survey conducted jointly by the Atlanta Fed, Nick Bloom of Stanford University, and Steven Davis of the University of Chicago. (That survey was in the field November 13–24.) In both of these surveys, we asked firms how the pending legislation would affect their capital expenditure plans for 2018. In the BIE survey, we also asked how tax reform would affect hiring plans.
The upshot? The typical firm isn't planning on a whole lot of additional capital spending or hiring.
In our national survey, roughly two-thirds of respondents indicated that the tax reform hasn't enticed them into changing their investment plans for 2018, as the following chart shows.
The chart below also makes apparent that small firms (fewer than 100 employees) are more likely to significantly ramp up capital investment in 2018 than midsize and larger firms.
For our regional BIE survey, the capital investment results were similar (you can see them here). And as for hiring, the typical firm doesn't appear to be changing its plans. Interestingly, here too, smaller firms were more likely to say they'd ramp up hiring. Among larger firms (more than 100 employees), nearly 70 percent indicated that they'd leave their hiring plans unchanged.
One interpretation of these survey results is that the potential for a sharp acceleration in GDP growth is limited. And that's also how President Bostic described things in his January 8 speech: "For now, I am treating a more substantial breakout of tax-reform-related growth as an upside risk to my outlook."
November 06, 2017
Building a Better Model: Introducing Changes to GDPNow
Among the frequently asked questions on GDPNow's web page is this one:
Is any judgment used to adjust the forecasts? Our answer:
No. Once the GDPNow model begins forecasting GDP growth for a particular quarter, the code will not be adjusted until after the "advance" estimate. If we improve the model over time, we will roll out changes right after the "advance" estimate so that forecasts for the subsequent quarter use a fixed methodology for their entire evolution.
This macroblog post enumerates a number of minor changes to GDPNow that were implemented on October 30, when it began forecasting fourth-quarter real gross domestic product (GDP) growth. Here is a summary of the changes, intended to improve the accuracy of the GDP subcomponent forecasts:
- Services personal consumption expenditures (PCE). Use industrial production of electric and gas utilities to nowcast real PCE on electricity and natural gas. Use international trade data on travel services to forecast revisions to related PCE travel data.
- Real business equipment investment. Use/forecast data from the advance U.S. Census Bureau reports on durable manufacturing and international trade in goods that, previously, hadn't been utilized until the full reports on manufacturing and/or international trade .
- Real nonresidential structures investment. Replace a discontinued seasonally adjusted producer price index for "Steel mill products: Steel pipe and tube" with a nonseasonally adjusted version. The index is used to construct a price deflator for private monthly nonresidential construction spending.
- Real residential investment. Use employment data for production and nonsupervisory employees of residential remodelers to help forecast real investment in residential improvements.
- Real change in private inventories. Use published monthly inventory levels in the U.S. Bureau of Economic Analysis's underlying detail tables 1BU and 1BUC after the third-release GDP estimate from the prior quarter to estimate inventory levels for a number of industries in the first month of the quarter forecasted by GDPNow.
- Federal, state, and local government spending. Forecast investment in intellectual property products for these subcomponents using autoregression models.
The first three columns of the following table decompose the official estimate of the third-quarter real GDP growth rate, and forecasts of the growth rate from the discontinued and modified versions of GDPNow, into percentage point contributions from the subcomponents of GDP.
As the table shows, the methodological changes did not have much of an impact on the final third-quarter subcomponent forecasts—apart from inventory investment, where the modifications lowered the contribution to growth from 0.80 percentage points to 0.60 percentage points—or on their accuracy. Nevertheless, the topline GDP forecast of the modified model (2.3 percent) was less accurate than the previous version (2.5 percent). In the discontinued version of GDPNow, an overestimate of the inventory investment contribution to growth partly canceled out underestimated contributions from each of net exports, government spending, and nonresidential fixed investment.
In the modified version, the inventory contribution was also underestimated and did not cancel out these other errors. The last two columns of the table show that all of the subcomponent errors of the modified model were at least as small as their historical average for the discontinued version. However, the topline GDP forecast was less accurate than average because of less cancellation of the subcomponent errors than usual. We hope that the cancellation of subcomponent errors in the modified model will be more similar to the historical average in the discontinued version in the future.
Although the methodological changes could have more of an impact than the table suggests, we do not expect them to have a substantial impact in general. For example, on October 30, the discontinued version of GDPNow projected 3.0 percent GDP growth in the fourth quarter, which was little different from the modified model forecast of 2.9 percent growth. We provide a more detailed explanation of the changes to GDPNow here . Going forward, this same document will document any further changes to the model and when we made them.
October 19, 2017
How Ill a Wind? Hurricanes' Impacts on Employment and Earnings
According to the Current Employment Statistics payroll survey, seasonally adjusted nonfarm payroll employment declined 33,000 in September. This decline was the first drop in employment since 2010 and followed a 169,000 gain in August. At the same time, seasonally adjusted average hourly earnings in the private sector increased 2.9 percent year over year in September. This increase in average wages was the largest since the end of the Great Recession in 2009. However, it seems likely that the decline in employment contributed to the rise in average hourly earnings. Why would a decline in employment contribute to an increase in average hourly earnings? We're glad you asked!
As noted by the U.S. Bureau of Labor Statistics, Hurricanes Harvey and Irma reduced employment in the payroll survey, whose reference period is the pay period that includes the 12th of the month. Hurricane Harvey first made landfall in east Texas on August 25 and again in Louisiana on August 30, and Hurricane Irma made landfall in south Florida on September 10. The storms forced large-scale evacuations and severely damaged many homes and businesses. For workers who are not paid when they miss work, being unable to work during the surveyed pay period means they are not counted in September payrolls.
To measure the size of Harvey and Irma's effect on payroll employment, we first looked at data from the Current Population Survey (CPS). We found that the bad weather forced about 1.5 million nonfarm workers who had a job during the September reference week to miss work. Of those, about 1.2 million were wage and salary earners, and about 760,000 of those were unpaid during their absence from work.
Our analysis indicates that September saw a shortfall in seasonally adjusted payroll employment between 200,000 and 300,000 jobs, suggesting that workers returning to work could result in a large rebound in payroll employment. (Not to get too far into the weeds, but our analysis involved regressing payroll employment growth on its lagged values as well as current and lagged seasonally adjusted changes in shares of workers who were not at work because of bad weather.)
What about average hourly earnings? Changes in average hourly earnings over time reflect both the effect of people getting pay raises and changes in who is working this month versus last month or last year. This latter effect can be large during recessions, when workers in lower-wage jobs are disproportionately more likely to be laid off. The absence of these workers from payrolls increases the average wage among the remaining employed workers, even if those remaining workers are not getting much of a pay increase (see this macroblog post for more discussion).
The September payroll survey depicted a particularly large decline in employment in the leisure and hospitality sector, which is significant because average hourly earnings in that sector are typically about 40 percent lower than overall average hourly earnings. In addition, from the CPS we see that the usual hourly earnings of workers not at work because of bad weather is much lower than for other workers. These data suggest that temporary absences from work because of bad weather likely put upward pressure on average hourly earnings, and some of that upward pressure could reverse itself as these workers return to their jobs. If the pace of average hourly earnings doesn't relax, however, then that would suggest more workers getting larger pay raises due to a tightening labor market.
September 07, 2017
What Is the "Right" Policy Rate?
What is the right monetary policy rate? The Cleveland Fed, via Michael Derby in the Wall Street Journal, provides one answer—or rather, one set of answers:
The various flavors of monetary policy rules now out there offer formulas that suggest an ideal setting for policy based on economic variables. The best known of these is the Taylor Rule, named for Stanford University's John Taylor, its author. Economists have produced numerous variations on the Taylor Rule that don't always offer a similar story...
There is no agreement in the research literature on a single "best" rule, and different rules can sometimes generate very different values for the federal funds rate, both for the present and for the future, the Cleveland Fed said. Looking across multiple economic forecasts helps to capture some of the uncertainty surrounding the economic outlook and, by extension, monetary policy prospects.
Agreed, and this is the philosophy behind both the Cleveland Fed's calculations based on Seven Simple Monetary Policy Rules and our own Taylor Rule Utility. These two tools complement one another nicely: Cleveland's version emphasizes forecasts for the federal funds rate over different rules and Atlanta's utility focuses on the current setting of the rate over a (different, but overlapping) set of rules for a variety of the key variables that appear in the Taylor Rule (namely, the resource gap, the inflation gap, and the "neutral" policy rate). We update the Taylor Rule Utility twice a month after Consumer Price Index and Personal Income and Outlays reports and use a variety of survey- and model-based nowcasts to fill in yet-to-be released source data for the latest quarter.
We're introducing an enhancement to our Taylor Rule utility page, a "heatmap" that allows the construction of a color-coded view of Taylor Rule prescriptions (relative to a selected benchmark) for five different measures of the resource gap and five different measures of the neutral policy rate. We find the heatmap is a useful way to quickly compare the actual fed funds rate with current prescriptions for the rate from a relatively large number of rules.
In constructing the heatmap, users have options on measuring the inflation gap and setting the value of the "smoothing parameter" in the policy rule, as well establishing the weight placed on the resource gap and the benchmark against which the policy rule is compared. (The inflation gap is the difference between actual inflation and the Federal Open Market Committee's 2 percent longer-term objective. The smoothing parameter is the degree to which the rule is inertial, meaning that it puts weight on maintaining the fed funds rate at its previous value.)
For example, assume we (a) measure inflation using the four-quarter change in the core personal consumption expenditures price index; (b) put a weight of 1 on the resource gap (that is, specify the rule so that a percentage point change in the resource gap implies a 1 percentage point change in the rule's prescribed rate); and (c) specify that the policy rule is not inertial (that is, it places no weight on last period's policy rate). Below is the heatmap corresponding to this policy rule specification, comparing the rules prescription to the current midpoint of the fed funds rate target range:
We should note that all of the terms in the heatmap are described in detail in the "Overview of Data" and "Detailed Description of Data" tabs on the Taylor Rule Utility page. In short, U-3 (the standard unemployment rate) and U-6 are measures of labor underutilization defined here. We introduced ZPOP, the utilization-to-population ratio, in this macroblog post. "Emp-Pop" is the employment-population ratio. The natural (real) interest rate is denoted by r*. The abbreviations for the last three row labels denote estimates of r* from Kathryn Holston, Thomas Laubach, and John C. Williams, Thomas Laubach and John C. Williams, and Thomas Lubik and Christian Matthes.
The color coding (described on the webpage) should be somewhat intuitive. Shades of red mean the midpoint of the current policy rate range is at least 25 basis points above the rule prescription, shades of green mean that the midpoint is more than 25 basis points below the prescription, and shades of white mean the midpoint is within 25 basis points of the rule.
The heatmap above has "variations on the Taylor Rule that don't always offer a similar story" because the colors range from a shade of red to shades of green. But certain themes do emerge. If, for example, you believe that the neutral real rate of interest is quite low (the Laubach-Williams and Lubik-Mathes estimates in the bottom two rows are −0.22 and −0.06) your belief about the magnitude of the resource gap would be critical to determining whether this particular rule suggests that the policy rate is already too high, has a bit more room to increase, or is just about right. On the other hand, if you are an adherent of the original Taylor Rule and its assumption that a long-run neutral rate of 2 percent (the top row of the chart) is the right way to think about policy, there isn't much ambiguity to the conclusion that the current rate is well below what the rule indicates.
"[D]ifferent rules can sometimes generate very different values for the federal funds rate, both for the present and for the future." Indeed.
July 07, 2016
Is the Labor Market Tossing a Fair Coin?
How important is tomorrow's June employment report? In isolation, the answer would surely be not much. The month-to-month swings in job gains can be quite large, and one month does not a trend make.
And yet, there seemed to be a pretty significant reaction to the May employment number, a reaction that did not escape the attention of MarketWatch's Caroline Baum:
So yes, the Fed does seem to be altering its macro view on potential growth (slower) and the neutral funds rate (close to zero) as a hangover from the Great Recession becomes an increasingly inadequate explanation for persistent 2% growth.
What comes across to the observer is a bad case of one-number-itis. The monthly jobs report does contain a lot of important information, including hiring, wages and a proxy for output (aggregate hours index). But the Fed talks out of both sides of its mouth, cautioning against putting too much weight on a single economic report, and then doing just that.
I get it. I don't speak for the Fed, of course—above my rank—but I am in fact one of those who regularly cautions against putting excessive weight on one number. And I am also one of those taken aback by the May employment report, so much so that my view of the economy changed materially as a result of that report.
Let me check that. My view of the risks to the economy, or more specifically the risks to my assessment of the strength of the economy, changed materially.
Here's an analogy that I find useful. Flip what you assume to be a fair coin. The probability of getting a heads, as we all know, is 50 percent. And if you weren't too traumatized by the statistics courses in your past, you will recall that the probability of two heads in a row is 25 percent, dropping to just about 13 percent of the coin coming up heads three times in a row.
Now, 13 percent is not zero, but it may be getting low enough for you to begin to wonder about your assumption that the coin is actually fair. If you have some stake in whether it is or isn't, you might want to take one more toss to get a little more evidence (since the odds of getting four heads in a row is, while not impossible, pretty improbable).
The point is that it wasn't just the May statistic that was striking in last month's report, but also the fact that the March and April numbers were revised downward to the tune of nearly 60,000 jobs. And if you step back a bit, you will see that the rolling three-month average of monthly job gains has been declining through the first half of the year (as the chart shows), even adjusting the May number for the Verizon strike:
Strike-adjusted, the May job gains were the lowest since December 2013. The three-month average (again strike-adjusted) was the lowest since the middle of 2012. In other words, although the year-over-year pace of jobs gains has been holding up, momentum in the labor market is decidedly softer—at least when measured by payroll employment gains.
I have been assuming that the U.S. economy will, for a while yet, continue to create jobs at a pace greater than necessary to maintain the unemployment rate at a more or less constant level. That pace is generally believed to be about 80,000 to 140,000 jobs per month, depending on your assumptions about the labor force participation rate. Another jobs report (including revisions to past months) that counters that assumption would, I think, cause a reasonable person to reassess his or her position.
Based on today's ADP report, the odds look good for some decent news tomorrow. On the other hand, if the June employment number does tick up, some observers will no doubt note that it is a pre-Brexit statistic. It may take a few more flips of the coin to determine if that consideration matters.
July 06, 2016
When It Rains, It Pours
Seasonally adjusted nonfarm payroll employment increased by only 38,000 jobs in May, according to the initial reading by the U.S. Bureau of Labor Statistics (BLS), and the total increase for the prior two months was revised down by a cumulative 59,000. Although the May increase was depressed by 35,100 striking workers at Verizon Communications, observers widely anticipated this distortion (the strike started April 13). Nonetheless, the median forecast of the May payroll gain from a Bloomberg survey of economists was 160,000, still well above the official estimate. The disappointing employment gain in May, I believe, is statistically related to the downward revisions to the seasonally adjusted gains made over the prior two months.
In contrast to the revision to the seasonally adjusted data, the nonseasonally adjusted level of payroll employment in April was only revised down by 3,000 in the May report. So most of the downward revision to the seasonally adjusted March and April employment gains was the result of revised seasonal factors (the difference between 59,000 and 3,000). In the chart below, the green diamond (toward the left) is the downward revision of 56,000 that resulted from the revised seasonal factors plotted against the Bloomberg survey forecast error for May (the difference between the actual estimate of 38,000 and the forecast of 160,000). The other diamonds represent corresponding points for reports from January 2006 through April 2016. The data points indicate a clear positive relationship and—based on the May Bloomberg forecast error—a simple linear regression would have almost exactly predicted the total downward revision to the March and April employment gains coming from revised seasonal factors.
To gain some insight into the positive relationship in the above chart, I used a model to seasonally adjust the last 10 years of nonfarm payroll employment data (excluding decennial census workers). Note that although I followed the BLS's procedure of accounting for whether there are four or five weeks between consecutive payroll surveys, I did not seasonally adjust the detailed industry employment data and sum them up, as the BLS does.
According to my seasonal adjustment model, the seasonally adjusted April employment level using data from the May employment report is 60,000 below the seasonally adjusted April employment level estimated with data from the April report. My seasonal adjustment model only using data through April from the May report predicts a nonseasonally adjusted increase of 789,000 jobs in May instead of the BLS's estimated increase of 651,000 jobs. The difference between these two estimates is similar to the Bloomberg survey forecast error noted above.
Further, when I replace the BLS's nonseasonally adjusted estimate for May with the model's forecast, the estimate of seasonally adjusted April employment is only 2,000 less than the model estimated with data from the April employment report. Hence, almost all of the model's downward revision to seasonally adjusted April employment appears to be the result of adding fewer jobs in May than the model expected.
The above analysis illustrates that, when it comes to looking at seasonally adjusted employment data, the number of jobs next month will affect the estimate of the number of jobs this month. This is not a very appealing notion, but when using seasonally adjusted data, it comes with the territory. Fortunately, analyzing the nonseasonally adjusted data allows us to gauge the impact of a surprise in the current estimate of seasonally adjusted employment growth on revisions to the prior two months. So when the June report is released on Friday, we will be paying close attention to both the seasonally adjusted headline numbers as well as the revisions to the nonadjusted data.
November 10, 2014
Wage Growth of Part-Time versus Full-Time Workers: Evidence from the CPS
Last week, our Atlanta Fed colleagues Lei Fang and Pedro Silos highlighted the wage growth trends of full-time and part-time workers in recent years. Using data from the U.S. Census Bureau's Survey of Income and Program Participation (SIPP), they showed relatively weak growth in hourly wages of part-time workers between 2011 and 2013. The Current Population Survey (CPS)—administered jointly by the Census Bureau and the U.S. Bureau of Labor Statistics—also contains wage information and has data through September 2014. We thought it would be interesting to see if the CPS data revealed a similar post-recession pattern, and if the more recent data show any sign of improvement. The short answer is that they do.
The following chart displays the median year-over-year growth in hourly earnings of wage and salary earners (shown as quarterly averages). The wage data are constructed using a similar methodology to that outlined in this paper by our San Francisco Fed colleagues Mary Daly and Bart Hobijn. The orange line is the median year-over-year growth in the hourly wages of all workers. The green line is the median wage growth of workers who worked full-time in both the current month and 12 months earlier (it is close to the orange line because most workers work full-time hours). The blue line is the median wage growth of workers who were part-time in both periods. Note that the median part-time wage growth is less precisely estimated (and thus demonstrates relatively more quarter-to-quarter variation) than its full-time counterpart because the CPS's sample size of wages for part-time workers is much smaller than for full-time workers.
Despite the noisy nature of the part-time wage data, it seems clear that the median wage growth of people usually working part-time fell dramatically behind that of full-time workers between 2011 and 2013. This finding is consistent with that of Fang and Silos. Interestingly, the other period when median part-time wage growth slipped behind was during the sluggish labor market recovery following the 2001 recession, albeit much less dramatically than the recent episode.
The SIPP data used by Fang and Silos ended in mid-2013. The more recent CPS data suggest that overall wage growth has picked up during the last year and that the wage growth gap has closed a bit, which are encouraging findings. But the wage growth of part-time workers, as a group, continues to lag well behind that of full-time workers. The relatively low wage growth of part-time workers heightens the importance of the fact that the number of people working part-time—especially involuntarily part-time—remains elevated.
TrackBack URL for this entry:
Listed below are links to blogs that reference Wage Growth of Part-Time versus Full-Time Workers: Evidence from the CPS:
November 06, 2014
Wage Growth of Part-Time versus Full-Time Workers: Evidence from the SIPP
Debates about the sluggish recovery in output, the low growth in labor productivity, and the actual level of slack in the U.S. economy are common within policy circles (see, for example, this speech by Fed Chair Janet Yellen and previous macroblog posts—here and here). One of the defining features of the recovery from the Great Recession has been the rise in the number of people employed part-time. As reported by the U.S. Bureau of Labor Statistics, roughly 10 percent more people are working part-time in September 2014 than before the recession. Part-time workers generally earn less per hour than full-time workers, so lower hours and lower per-hour earnings both contribute to their lower incomes. Despite those differences in wage levels, less is known about wage growth of part-time relative to full-time workers. Has wage growth been different? Has wage inequality increased across the two groups of workers?
To find out, we employ data from the Survey of Income and Program Participation (SIPP) to analyze the wage growth of part-time and full-time workers. The SIPP is a longitudinal survey designed to be representative of the U.S. labor force. It is constructed as a sequence of panels of households who are interviewed for three to five years. Designed and maintained by the U.S. Census Bureau, the first panel began in 1984, and the most recent panel started in 2008. Households are interviewed every four months during the time they remain in the sample, providing information on work experience (employment, hours, earnings, occupation, and industry, among other variables) for the months between interviews.
The 2008 SIPP panel data that we use cover the period from August 2008 to April 2013. We restrict the analysis to hourly workers, a group representing roughly half of all employed in the 2008 panel. The reason we focus on this group is that they provide the cleanest measure of the price of labor: a wage rate for each hour they work. The remainder of workers—those compensated with a monthly or annual salary—do not report such a measure, and it needs to be inferred from their responses about total earnings and total hours worked. Because hours reported in the SIPP include much missing data and are sometimes inaccurate, we discard salaried workers. We also exclude anyone whose wages or hours information was allocated or imputed and anyone at the top or bottom of the wage distribution.
We divide the sample into two groups: those whose usual hours are fewer than 35 hours a week (part-time workers) and those who usually work 35 hours or more per week (full-time workers). We then compare the distribution of wage growth for each group and compute the median wage growth rate. To eliminate short-term fluctuations and seasonal effects, we compute median hourly wage growth rates over a three year period, expressed as an annual rate. Since the data start from August 2008, our series for the wage growth rate starts from August 2011.
Chart 1 shows the median wage growth rate of individuals over time. During the recovery, the median growth rate of full-time workers has been higher than that of part-time workers. In particular, wage declines were more common among part-time workers.
To further analyze the wage growth pattern of full-time and part-time workers, we subdivide the sample by education. Chart 2 plots the median wage growth rates for those with at least a bachelor's degree and those with some college or less. The median wage growth rates for full-time workers are larger than for part-time workers within each education group and highest for college graduates working full-time. Also apparent is that the weak wage growth of part-time workers is significantly influenced by the sluggish wage growth among those with less than a bachelor's degree.
Overall, we find that part-time workers as a group appear to experiencing a lower average wage growth rate than full-time workers during the recovery from the Great Recession. Education matters for wage growth, but the pattern of lower wage growth for part-time workers persists for people with broadly similar educational attainment.
By Lei Fang, research economist and assistant policy adviser, and
Pedro Silos, research economist and associate policy adviser, both in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference Wage Growth of Part-Time versus Full-Time Workers: Evidence from the SIPP:
November 04, 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.
By Pat Higgins, senior economist in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference Data Dependence and Liftoff in the Federal Funds Rate:
September 29, 2014
On Bogs and Dots
Consider this scenario. You travel out of town to meet up with an old friend. Your hotel is walking distance to the appointed meeting place, across a large grassy field with which you are unfamiliar.
With good conditions, the walk is about 30 minutes but, to you, the quality of the terrain is not so certain. Though nobody seems to be able to tell you for sure, you believe that there is a 50-50 chance that the field is a bog, intermittently dotted with somewhat treacherous swampy traps. Though you believe you can reach your destination in about 30 minutes, the better part of wisdom is to go it slow. You accordingly allot double the time for traversing the field to your destination.
During your travels, of course, you will learn something about the nature of the field, and this discovery may alter your calculation about your arrival time. If you discover that you are indeed crossing a bog, you will correspondingly slow your gait and increase the estimated time to the other side. Or you may find that you are in fact on quite solid ground and consequently move up your estimated arrival time. Knowing all of this, you tell your friend to keep his cellphone on, as your final meeting time is going to be data dependent.
Which brings us to the infamous “dots,” ably described by several of our colleagues writing on the New York Fed’s Liberty Street Economics blog:
In January 2012, the FOMC began reporting participants’ FFR [federal funds rate] projections in the Summary of Economic Projections (SEP). Market participants colloquially refer to these projections as “the dots” (see the second chart on page 3 of the September 2014 SEP for an example). In particular, the dispersion of the dots represents disagreement among FOMC [Federal Open Market Committee] members about the future path of the policy rate.
The Liberty Street discussion focuses on why the policy rate paths differ among FOMC participants and across a central tendency of the SEPs and market participants. Quite correctly, in my view, the blog post’s authors draw attention to differences of opinion about the likely course of future economic conditions:
The most apparent reason is that each participant can have a different assessment of economic conditions that might call for different prescriptions for current and future monetary policy.
The Liberty Street post is a good piece, and I endorse every word of it. But there is another type of dispersion in the dots that seems to be the source of some confusion. This question, for example, is from Howard Schneider of Reuters, posed at the press conference held by Chair Yellen following the last FOMC meeting:
So if you would help us, I mean, square the circle a little bit—because having kept the guidance the same, having referred to significant underutilization of labor, having actually pushed GDP projections down a little bit, yet the rate path gets steeper and seems to be consolidating higher—so if it’s data dependent, what accounts for the faster projections on rate increases if the data aren’t moving in that direction?
The Chair’s response emphasized the modest nature of the changes, and how they might reflect modest improvements in certain aspects of the data. That response is certainly correct, but there is another point worth emphasizing: It is completely possible, and completely coherent, for the same individual to submit a “dot” with an earlier (or later) liftoff date of the policy rate, or a steeper (or flatter) path of the rate after liftoff, even though their submitted forecasts for GDP growth, inflation, and the unemployment rate have not changed at all.
This claim goes beyond the mere possibility that GDP, inflation, and unemployment (as officially defined) may not be sufficiently complete summaries of the economic conditions a policymaker might be concerned with.
The explanation lies in the metaphor of the bog. The estimated time of arrival to a destination—policy liftoff, for example—depends critically on the certainty with which the policymaker can assess the economic landscape. An adjustment to policy can, and should, proceed more quickly if the ground underfoot feels relatively solid. But if the terrain remains unfamiliar, and the possibility of falling into the swamp can’t be ruled out with any degree of confidence...well, a wise person moves just a bit more slowly.
Of course, as noted, once you begin to travel across the field and gain confidence that you are actually on terra firma, you can pick up the pace and adjust the estimated time of arrival accordingly.
To put all of this a bit more formally, an individual FOMC participant’s “reaction function”—the implicit rule that connects policy decisions to economic conditions—may not depend on just the numbers that that individual writes down for inflation, unemployment, or whatever. It might well—and in the case of our thinking here at the Atlanta Fed, it does—depend on the confidence with which those numbers are held.
For us, anyway, that confidence is growing. Don’t take that from me. Take it from Atlanta Fed President Lockhart, who said in a recent speech:
I'll close with this thought: there are always risks around a projection of any path forward. There is always considerable uncertainty. Given what I see today, I'm pretty confident in a medium-term outlook of continued moderate growth around 3 percent per annum accompanied by a substantial closing of the employment and inflation gaps. In general, I'm more confident today than a year ago.
Viewed in this light, the puzzle of moving dots without moving point estimates for economic conditions really shouldn’t be much of a puzzle at all.
By Dave Altig, executive vice president and research director of the Atlanta Fed
TrackBack URL for this entry:
Listed below are links to blogs that reference On Bogs and Dots:
- What Does the Current Slope of the Yield Curve Tell Us?
- Does Loyalty Pay Off?
- Immigration and Hispanics' Educational Attainment
- Are Tariff Worries Cutting into Business Investment?
- Improving Labor Market Fortunes for Workers with the Least Schooling
- Part-Time Workers Are Less Likely to Get a Pay Raise
- Learning about an ML-Driven Economy
- Hitting a Cyclical High: The Wage Growth Premium from Changing Jobs
- Thoughts on a Long-Run Monetary Policy Framework, Part 4: Flexible Price-Level Targeting in the Big Picture
- Thoughts on a Long-Run Monetary Policy Framework, Part 3: An Example of Flexible Price-Level Targeting
- August 2018
- July 2018
- June 2018
- May 2018
- April 2018
- March 2018
- February 2018
- January 2018
- November 2017
- October 2017
- Business Cycles
- Business Inflation Expectations
- Capital and Investment
- Capital Markets
- Data Releases
- Economic conditions
- Economic Growth and Development
- Exchange Rates and the Dollar
- Fed Funds Futures
- Federal Debt and Deficits
- Federal Reserve and Monetary Policy
- Financial System
- Fiscal Policy
- Health Care
- Inflation Expectations
- Interest Rates
- Labor Markets
- Latin America/South America
- Monetary Policy
- Money Markets
- Real Estate
- Saving, Capital, and Investment
- Small Business
- Social Security
- This, That, and the Other
- Trade Deficit
- Wage Growth