The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

« April 2016 | Main | June 2016 »

May 23, 2016

Can Two Wrongs Make a Right?

In a recent macroblog post, I showed that forecasts from the Atlanta Fed's real gross domestic product (GDP) nowcasting model—GDPNow—have been about as accurate a forecast of the U.S. Bureau of Economic Analysis's (BEA) first estimate of real GDP growth as the consensus from the Wall Street Journal Economic Forecasting Survey. Because GDPNow essentially uses a "bean-counting" approach that tallies the forecasts of the various main subcomponents of GDP, the total GDP forecast error can be broken up into the forecast errors coming from each piece of GDP. For most of the subcomponents of GDP, the contribution to total GDP growth is approximately its real growth rate multiplied by its expenditure share of nominal GDP (the exact formulas are in the working paper for GDPNow). The following chart shows the subcomponent contributions to the GDPNow forecast errors since the third quarter of 2011. (I want to note that the forecast errors are based on the final GDPNow forecasts formed before the BEA's first estimates of GDP are released.)

The forecast errors for the subcomponents can sometimes be quite large. For example, for the fourth quarter of 2013, GDPNow underestimated the combined contributions of net exports and inventory investment by nearly 2 percentage points. However, these misses were nearly offset by overestimates of the other contributions to growth (consumption, business and residential fixed investment, and government spending).

The pattern of large but largely offsetting GDP subcomponent errors has been attributed to the work of a fictional "Saint Offset," as former Fed Governor Laurence Meyer noted in a 1998 speech. Unfortunately, "Saint Offset" doesn't always come to the forecaster's aid. For example, in the fourth quarter of 2011, GDPNow predicted 5.2 percent growth—well above the BEA's first estimate of 2.8 percent—and the subcomponent errors were predominantly on the high side.

A closer look at the chart also reveals that GDPNow has had a tendency to overestimate the contribution of business fixed investment to growth and underestimate the growth contribution of inventory investment. Although these subcomponent biases have nearly offset one another on average, we really don't want to have to rely on "Saint Offset." We would like the subcomponent forecasts to be reasonably accurate because the subcomponents of GDP are of interest in their own right.

Have the subcomponent biases been a unique feature of GDPNow forecasts? It appears not. Both the Survey of Professional of Forecasters (SPF), conducted about 11 weeks prior to the first GDP release, and Blue Chip Economic Indicators, conducted as close as three weeks prior to the first release, provide consensus forecasts for some GDP subcomponents. The following table provides an average forecast error (as a measure of bias) and average absolute forecast error (as a measure of accuracy) of the subcomponent growth contributions for the two surveys and comparably timed GDPNow forecasts.

We see that the biases in GDPNow's subcomponents have been fairly similar to those in the two surveys. For example, all three sources have underestimated the average inventory investment contribution to growth by fairly similar magnitudes.

The relative accuracy of GDPNow's subcomponent and overall GDP forecasts has also been similar to the accuracy of the two surveys. "Saint Offset" has helped all three forecasters; the standard errors of the real GDP forecasts are 20 percent to 40 percent lower than they would be if the forecast errors of the subcomponents did not cancel each other out.

Finally, notice that some GDP subcomponents appear to be much more difficult to forecast than others. For instance, the bias and accuracy metrics for consumer spending are smaller than they are for inventory investment. This differential is not really that surprising, because more monthly source data are available prior to the first GDP release for consumer spending than for inventory investment.

Can we take any comfort in knowing that private forecasters have mirrored the biases in GDPNow's subcomponent forecasts? An optimistic interpretation is that the string of one-sided misses are the result of bad luck—an atypical sequence of shocks that neither GDPNow nor private forecasters could account for. A more troubling interpretation is that there have been structural changes in the economy that neither GDPNow nor the consensus of private forecasters have identified. Irrespective of the reason, though, optimal forecasts should be unbiased. If biases in some of the subcomponents continue, then forecasters will need to look for a robust way to eliminate them.

May 23, 2016 in Forecasts, GDP | Permalink


Two wrongs can make a right. But it is not always the same two wrongs.

Posted by: Robert F. Dieli | May 24, 2016 at 08:53 AM

since you're projecting the real change in GDP from one quarter to the next, the BLS price indexes that the BEA uses to compute the deflators for the various components of GDP are an integral part of that change, yet there's no indicaton that your model takes those prices indices into account...every month you'll adjust your estimate of the change in real PCE when the retail sale report comes out, but the retail sales report tells us very little about PCE until the CPI report is released several days later...

i also have a problem with your use of the ISM surveys, unweighted diffusion indexes derived from samplings of subjective executive opinion....any correspondence those survey results have with the output of goods and services for the same month is pure chance..

your forecasts could be improved if you hewed closer to the methods that the BEA uses to compute GDP that are described in the NIPA Handbook...

Posted by: rjs | May 24, 2016 at 10:30 PM

Post a comment

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

If you have a TypeKey or TypePad account, please Sign in

May 19, 2016

Are People in Middle-Wage Jobs Getting Bigger Raises?

As observed in this Bloomberg article and elsewhere, the Atlanta Fed's Wage Growth Tracker (WGT) reached its highest postrecession level in April. This related piece from Yahoo Finance suggests that the uptick in the WGT represents good news for middle-wage workers. That might be so.

Technically, though, the WGT is the median change in the wages of all continuously employed workers, not the change in wages among middle-income earners. However, we can create versions of the WGT by occupation group that roughly correspond to low-, middle-, and high-wage jobs, which allows us to assess whether middle-wage workers really are experiencing better wage growth. Chart 1 shows median wage growth experienced by each group over time. (Note that the chart shows a 12-month moving average instead of a three-month average, as depicted in the overall WGT on our website.)

Wage growth for all three categories has risen during the past few years. However, the timing of the trough and the speed of recovery vary somewhat. For example, wage growth among low-wage earners stayed low for longer and then recovered relatively more quickly. Wage growth of those in high-wage jobs fell by less but also has recovered by relatively less. In fact, while the median wage growth of low-wage jobs is back to its 2003–07 average, wage growth for those in high-wage jobs sits at about 75 percent of its prerecession average.

Are middle-wage earners experiencing good wage growth? In a relative sense, yes. The 12-month WGT for high-wage earners was 3.1 percent in April compared with 3.2 percent and 3.0 percent for middle- and low-wage workers, respectively. So the typical wage growth of those in middle-wage jobs is trending slightly higher than for high-wage earners, a deviation from the historical picture.

Interestingly, this pattern of wage growth doesn't quite jibe with the relative tightness of the labor market for different types of jobs. As was shown here, the overall WGT appears to broadly reflect the tightness of the labor market (possibly with some lag).

In theory, as the pool of unemployed shrinks, employers will face pressure to increase wages to attract and retain talent. Chart 2 shows the 12-month average unemployment rates for people who were previously working in one of the three wage groups.

Like the relationship between overall WGT and the unemployment rate, wage growth and the unemployment rate within these wage groups are negatively correlated (in other words, when the unemployment rate is high, wage growth is sluggish). The correlation ranges from minus 0.81 for low-wage occupations to minus 0.88 for middle-wage occupations.

However, notice that although the current gap between unemployment rates across the wage spectrum is similar to prerecession averages, the current relative gap in median wage growth is different than in the past. In particular, the wage growth for those in higher-wage jobs has been sluggish compared to middle- and lower-wage occupations.

Nonetheless, it's clear that the labor market is getting tighter. Wage growth overall has moved higher over the past year, driven primarily by those working in low- and middle-wage jobs. Is firming wage growth starting to show up in price inflation? Perhaps.

The consumer price index inflation numbers moved higher again in April, and Atlanta Fed President Dennis Lockhart said on Tuesday that—from a monetary policy perspective—recent inflation readings and signs of better growth in economic activity during the second quarter (as indicated by the Atlanta Fed's GDPNow tracker) are encouraging signs.

May 19, 2016 in Economic conditions, Employment, Labor Markets, Wage Growth | Permalink


Could the jump in wage growth for low-wage earners in the top graph be attributed, in part, to the statutory increases in minimum wage in several states & metro areas in that time period? For several states the statute went into effect 1/1/16, but employers could have moved prior to that.

Posted by: JDT | May 20, 2016 at 09:11 AM

Very interesting post. Any thoughts on whether increase in Low Wage driven by changes in states' minimum wages? Or any breakdown by industry?
Thank you

Posted by: Igor | May 20, 2016 at 10:14 AM

Thanks for the nice post! What you say seems to confim the job-ladder model proposed by Melvin Reder in 1964 (Wage Structure and Structural Unemployment, The Review of Economic Studies, Vol. 31, No. 4. pp. 309-322). His view implies that skill margins move counter-cyclically. Unfortunately, Reder's contribution is largely forgotten.

Posted by: Ekkehart Schlicht | May 21, 2016 at 03:30 AM

Post a comment

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

If you have a TypeKey or TypePad account, please Sign in

May 16, 2016

GDPNow and Then

Real-time forecasts from the Atlanta Fed’s real gross domestic product (GDP) nowcasting model—GDPNow—have been regularly updated since August 2011 (the model was introduced online in July 2014). So we now have a nearly five-year history to allow us to evaluate the accuracy of the model’s forecasts. The chart below shows forecasts from GDPNow (red dots) alongside actual first estimates of real GDP growth (gray bars) from the U.S. Bureau of Economic Analysis (BEA). For comparison, the blue dots in the chart are the consensus (average) forecasts from the Wall Street Journal Economic Forecasting Survey (WSJ Survey).


The initial estimate of real GDP growth for a particular quarter is usually published at the end of the subsequent month. The WSJ Survey consensus forecasts plotted above were released about two weeks before these estimates. To maintain comparable timing with the WSJ Survey, the GDPNow forecasts shown in the chart are those constructed on or before the 12th day of the same month.

Occasionally, there has been relatively large disagreement between GDPNow and the WSJ consensus. For example, GDPNow predicted that GDP growth would be below 0.5 percent for five out of 19 quarters between 2011 and 2016, and the lowest WSJ Survey consensus forecast for any of those quarters was 1.3 percent. Nonetheless, the average accuracy of the GDPNow and WSJ Survey consensus forecasts has been similar: the average absolute forecast error (average error without regard to sign) for GDPNow was 0.56 versus 0.60 for the WSJ Survey consensus.

Studies have shown that the average or median of a set of professional forecasts tends to be more accurate than an individual forecaster (see, for example, here and here). Therefore, it’s surprising that GDPNow has been about as accurate on average as the WSJ Survey consensus. To see just how surprising this result is, I used the fact that the WSJ Survey provides both the names and forecasts of its respondents. From these, I constructed a panel dataset with each respondent’s absolute forecast errors and their absolute disagreement (difference) from the consensus forecast. Using a standard econometric technique (a two-way fixed-effects regression), we can then calculate each panelist’s average absolute GDP forecast error and their average absolute disagreement with the WSJ Survey consensus. These points are shown in the scatterplot below.


There is a clear inverse relationship between average forecast accuracy and average disagreement with the WSJ Survey consensus. However, GDPNow’s accuracy and disagreement statistics do not fit the general pattern. GDPNow (the orange diamond in the chart) was more accurate on average than all but six out of 49 WSJ panelists, though at the same time it differed from the consensus by more on average than all but four of the panelists.

What should one infer from all of this? Differences in forecasting method could be part of the explanation. GDPNow differs from many other approaches to nowcasting in that it is essentially a “bean counting” exercise. It doesn’t use historical correlations of GDP with other economic series in the way that commonly used dynamic factor models do, and it also doesn’t incorporate judgmental adjustments (see here for more discussion of these differences). During a period when the economy has been giving very mixed signals, perhaps it doesn’t come as a surprise that GDPNow’s forecasts occasionally deviate quite a bit from the WSJ Survey consensus. Time will tell if GDPNow continues to perform at least as well as the consensus.

May 16, 2016 in Forecasts, GDP | Permalink


It would be nice to know who are the three forecasters with average absolute error below the 0.5 line.

Posted by: Rafael | May 23, 2016 at 09:28 AM

Post a comment

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

If you have a TypeKey or TypePad account, please Sign in

May 04, 2016

What's behind the Recent Uptick in Labor Force Participation?

The labor force participation rate had been generally declining since around 2007. However, that trend has partially reversed in recent months. As noted in the minutes of the March meeting of the Federal Open Market Committee, this rise was interpreted as further strengthening of the labor market. But will the increase persist?

As shown in a previous macroblog post, the dominant contributor to the decline in participation during the last several years has been the aging of the population. To see what's behind the increase in participation during the last few months, the following chart breaks the participation rate change between the first quarters of 2015 and 2016 into a part that is the result of shifts in the age distribution (holding behavior within age groups fixed), and the parts that are the result of changes in behavior (holding the age distribution fixed).

During the last year, the negative effect on participation attributable to an aging population (0.22 percentage points) has been offset by a 0.23 percentage point decline in the share of people who want a job but are not counted as unemployed (including people who are marginally attached). This decline is an encouraging sign, and consistent with a tightening labor market.

How much more can the want-a-job category improve? We don't really know. But that category's share of the population is currently about 0.3 percentage points above the prerecession trough of 2.0 percent. So at the current pace we would be at prerecession levels in about a year.

Despite the recent uptick, projections over the next decade or so have the labor force participation rate moving lower, chiefly because of an aging population. But how much farther participation actually declines will also depend on the evolution of various behavioral factors. The employment report for April will be released this Friday by the U.S. Bureau of Labor Statistics, and it will be interesting to see whether the number of people on the margin of the labor force continues to shrink.

May 4, 2016 in Employment, Labor Markets, Unemployment | Permalink


Post a comment

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

If you have a TypeKey or TypePad account, please Sign in

Google Search

Recent Posts

February 2017

Sun Mon Tue Wed Thu Fri Sat
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28        



Powered by TypePad