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
November 28, 2016
Does Lower Pay Mean Smaller Raises?
I've been asked a few questions about the relative wage growth of low-wage versus high-wage individuals that are measured by the Atlanta Fed's Wage Growth Tracker. Do individuals who were relatively lower (or higher) paid also tend to experience lower (or higher) wage growth? If they do, then wage inequality would increase pretty rapidly as low-wage earners get left further and further behind.
The short answer is no. As chart 1 shows, median wage growth is highest for the workers whose pay was relatively low (in the bottom 25 percent of the wage distribution), and lowest for those who were the highest-paid (in the top 25 percent of the wage distribution). Median wage growth is reasonably similar for those whose pay was in the middle 50 percent of the wage distribution.
To understand what's going on, let's look at the construction of a Wage Growth Tracker sample. In simple terms, a person's wage is observed in one month, and then again 12 months later. But relatively low-wage workers are less likely to remain employed (and hence more likely not to have a wage when observed a second time) than other workers. Almost half of workers who are not employed 12 months later come from the lowest 25 percent of the wage distribution. For workers in a relatively low-wage job, a greater share who might otherwise have experienced a declining wage left their employment, resulting in a larger share of wage increases among those who remained employed.
In contrast, relatively high wage earners in the Wage Growth Tracker sample have a remarkably low median wage growth—zero in recent years. They also have a much greater chance of experiencing a wage decline than other workers (see chart 2).
However, getting a complete picture for high-wage individuals in the Current Population Survey is limited by the fact that observations are top-coded (or censored to preserve identifiable individuals' anonymity). For example, weekly earnings higher than $2,885 are currently simply recorded as $2,885. If a person in this circumstance gets a wage increase, it will still be reported as just $2,885, which would make it seem as if wages didn't increase, even if they did.
Top-coding itself has only a relatively small effect on the median wage growth for the whole sample because top-coded earnings aren't that common. But they are a reasonably large share of the upper part of the wage distribution, which makes the median wage growth pretty unrepresentative for people who were relatively high wage earners. In principle, one could try to surmount this problem by estimating the earnings for top-coded workers, but my experience has been that doing so is likely to add more noise than insight.
What about examining a worker's current wage instead of their prior wage? Is the median wage growth also higher for workers who are currently in the lowest part of the wage distribution? No. In fact, they are more likely than others not to have received a pay raise or even to have had the rate of pay reduced. Conversely, someone who is currently in the upper part of the wage distribution is more likely to have received a larger pay raise than other workers. Some workers move up the wage distribution—but not all.
The bottom line is that the point of reference matters a lot when looking at the tails of the wage distribution, and top-coding limits the ability to learn much about the wage growth of high wage earners. But for the middle part of the wage distribution, it doesn't matter so much. The median wage growth of the overall sample is pretty representative of the typical wage growth experience of workers in the heart the wage distribution.
November 22, 2016
Outside Looking In: Why Has Labor Force Participation Increased?
The labor force participation rate (LFPR) is an estimate of the share of the population actively engaged in the labor market. The LFPR has increased about 30 basis points over the past year (from the third quarter of 2015 to the third quarter of 2016)—a modest reversal in the precipitous decline in the LFPR that began in 2008. What accounts for this stabilization and—given the demographic and cyclical forces in play—how much longer can it last?
The following is perspective through the lens of the reasons people give for not participating in the labor force. Perhaps the component most responsive to changes in labor market conditions is what I will refer to as the "shadow labor force," which is made up of people who are not in the official labor force and are not actively seeking employment, but who say they want a job. (This group includes people discouraged over job prospects.) During tough times, the share of the population in the shadows rises, and during good times it falls. In the third quarter of 2016, about 2.3 percent of the population fell into this category—down from a high of 2.8 percent but still a bit above prerecession levels (see the chart).
But focusing solely on the decline in the shadow labor force to explain the recent reversal in the LFPR would be a mistake. In fact, high unemployment in the aftermath of the Great Recession was accompanied not only by a rise in the share of the shadow labor force, but also by an increase in the share of the population who said they didn't currently want a job—because of either a health issue or engagement in some other activity. Although some of this likely reflects trends already at work before the recession, some of it was also probably a cyclical response to weak job opportunities.
The chart below shows how these various factors cumulatively contribute to the decline in the LFPR between the third quarter of 2007 and the third quarter of 2016. It shows that, in addition to a larger share in the shadow labor force, the reasons for the decline between 2007 and 2016 also stemmed from a greater age-adjusted share who were too sick or disabled to work (purple) or in school instead of working (light blue). Interestingly, the share out of the labor force but wanting a job (dark blue) actually exerted the smallest downward force on LFPR of all of these three reasons. The green section represents the impact of the baby boomers: an increasing share of the population of retirement age. Partly offsetting this shift in the age distribution was a decrease in the propensity of these workers to actually retire (orange).
The next chart shows that almost all the nonparticipation factors that had put downward pressure on the LFPR since 2007 have reversed course and contributed positively to an increase in the LFPR during the past year. In particular, there was a decline in the share of the population who cited nonparticipation because of poor health or enrollment in school or were otherwise wanting but not looking for work. This decline in the schooling and illness nonparticipation rates is particularly noteworthy because it stands in contrast to the increasing trends that were in place prior to the recession (to read more, visit our LFP Dynamics page).
The only significant factor continuing to depress the LFPR during the past year is the impact of an increasing share of the population in age groups with relatively low labor force attachment. This factor brings me to the second question I posed earlier in this post: What will this picture look like going forward? Unfortunately, I think the answer is that it's very hard to say.
Other things equal, it seems reasonable to think that the nonparticipation rates attributable to age-adjusted schooling and poor health will eventually revert to the upward trends occurring before the recession, a reversion that will push down the LFPR. Probably the biggest wild card for the future is what will happen with decisions concerning retirement (and hence older individuals' LFPR). The trend toward retiring later in life has risen and fallen a couple of times during the past two decades. The positive role that later retirement has played in mitigating the overall decline in the LFPR in recent years—coupled with the steadily increasing share of the population approaching traditional retirement age—suggests that deferred retirement will be an especially important factor to keep an eye on. This point is nicely illustrated in this piece by our colleagues at the Kansas City Fed, who look at the role of later retirement in reducing the rate of outflow of people from the labor force.
Note: Data on the reasons for not participating are available in the data download section of the Labor Force Dynamics page, which we will update every quarter (data for the third quarter are coming soon).
November 22, 2016 | Permalink
November 15, 2016
Wages Climb Higher, Faster
The Atlanta Fed's Wage Growth Tracker is a three-month average of median growth in the hourly earnings of a sample of wage and salary workers taken from the Current Population Survey. Last month in a macroblog post, I noted that the Wage Growth Tracker reading for September, at 3.6 percent, was close to where it had been hovering since April. However, I also noted that the non-averaged median wage growth for September was at a cyclical high of 4.2 percent, and so it would be interesting to see what the October data revealed. Well, the October data are in, and they do confirm a sizeable uptick in wage growth over the last couple of months. The median wage growth for October was 4.0 percent, which brings the Wage Growth Tracker up to 3.9 percent—a percentage point higher than a year ago, and now the highest level since November 2008.
In addition, nominal wage rigidity, as measured by the fraction of workers reporting no change in their hourly rate of pay from 12 months earlier, declined to 13 percent—the lowest since April 2008.
The rise depicted by the Wage Growth Tracker is consistent with the recent trend in average hourly earnings from the payroll survey (up 2.8 percent from a year earlier in October—the fastest pace since June 2009). This increase is occurring even though the unemployment rate has changed little in recent months and is only 10 basis points lower than a year ago. Perhaps employers are finally catching up to the realities of a low unemployment rate. Larger wage gains may also be behind why we are seeing fewer workers leave the labor force. Labor force participation is some 30 or so basis points higher than it was a year ago, and this is primarily because the flow out of the labor force has slowed.
Note: The Wage Growth Tracker website now contains data for the smoothed and unsmoothed series going back to 1983. Previously, the historical data started in 1997. You will notice gaps in the time series in 1995–96 and 1985–86 because the Census Bureau masked the identifiers used to match individual earnings during those periods.
November 14, 2016
Is There a Gender Wage Growth Gap?
The existence of the "gender wage gap" is well documented. Although the gap in the average level of pay between men and women has narrowed over time, studies conducted in the past few years find that women still tend to make about 20 percent less than men. Researchers estimate that between one half and three quarters of the gap can be accounted for by observable differences between men and women in the workforce such as labor market experience, educational attainment, as well as job characteristics (see here , here, and here). This estimation leaves one quarter to one half of the gap that is the result of other factors. While some pin the remainder on discrimination or unfair hiring practices, others suggest the remaining gap may reflect subtle differences in work preferences, such as women choosing jobs with family-oriented benefit packages or flexible work arrangements.
A related question is whether there are differences between the average wage growth of men and women. Since 2010 the Atlanta Fed's Wage Growth Tracker has revealed a disparity between the pay raises of continuously employed men and women, as depicted in the following chart.
Between 1997 and 2010, wage growth of men and women was about equal. Since 2010 however, a gap has emerged. On average, men have been experiencing about 0.35 percentage points higher median wage growth than women. Can differences in characteristics such as experience and job choice explain this gap?
To answer this question, I aggregated individuals into groups based on their potential labor market experience (0–5 years, 5–9 years, 10–24 years, and 25–48 years) education (degree or no degree) family type (married, whether your spouse works, and whether you have kids); industry (goods versus services) occupation (low, middle, or high skill); sector (public versus private); and if the person switched jobs recently. I then computed the median wage growth for each unique group in each year. Using a statistical technique called a Oaxaca Decomposition, I separated out the difference between men and women's wage growth that can be pinned on differences in the way men and women are distributed among these groups (the "endowment" effect).
The following chart shows median wage growth after removing this endowment effect.
After removing the difference in wage growth that is the result of differences in gender-specific characteristics, wage growth of men and women is much more similar. In particular, these differences appear to almost entirely account for the gap that had emerged after 2009. What explains the gap in wage levels between men and women is still an open question, but this analysis suggests that much of the difference in wage growth through the years has to do with family/job choices and other individual characteristics.
November 07, 2016
The Price Isn't Right: On GDPNow's Third Quarter Miss
The U.S. Bureau of Economic Analysis's (BEA) first estimate of third quarter annualized real gross domestic product (GDP) growth released on October 28 was 2.9 percent. A number of nowcasts were quite close to this number, including the median forecast of 3.0 percent from the CNBC Rapid Update survey of roughly 10 economists. The Atlanta Fed's GDPNow model forecast of 2.1 percent? Not so close.
What accounted for GDPNow's miss? The table below shows the GDPNow forecasts and BEA estimates of the percentage point contributions to third quarter growth of six subcomponents that together make up real GDP. The largest forecast error, both in absolute terms and relative to the historical accuracy of the projections, was for the contribution of real net exports to growth. The published contribution of 0.83 percentage points was much higher than the model's estimate of 0.07 percentage points.
Why did GDPNow miss so badly on net exports? For both goods and services real net exports, the GDPNow forecast is a weighted average of two forecasts. The "bean counting" forecast uses the monthly source data on the nominal values and price deflators of exports and imports. The econometric model forecast uses published values of 13 subcomponents of real GDP for the last five quarters to predict real net exports for both goods and services. The statistically determined weights on the bean counting forecast increase as we get closer to the first GDP release and accumulate more monthly source data. (More details are provided here.)
For real net exports of services, 89 percent of the weight was given to the bean counting forecast. This weighting worked out well last quarter as the forecasts of the contribution of real services net exports to third quarter growth from both the bean counting and combined models were within 0.01 percentage points of the published value of −0.14 percentage points. But for real net exports of goods, the bean counting forecast received only 59 percent of the weight in the final GDPNow forecast. It projected that real net exports of goods would add 0.76 percentage points to growth—reasonably close to the BEA estimate. In contrast, the econometric model projected a subtraction of 0.58 percentage points from growth.
Since GDPNow had monthly price and nominal spending data through September on goods imports and exports, why didn't it place more weight on the source data? One of the important reasons is that it's difficult to match the quarterly inflation rate of the BEA's import price deflator for goods. The BEA constructs its price deflator with detailed price indices from the Bureau of Labor Statistics (BLS) producer price index and import/export price index programs as well as a few other sources. GDPNow uses the BLS's import price data at higher levels of aggregation than the BEA uses and also differs in the manner that it handles seasonality. The chart below plots the difference between the BEA's quarterly goods import price inflation rate and the GDPNow proxy. These inflation measures have differed by 5 percentage points or more on a number of occasions. Since goods imports are 12 percent of GDP, a miss of this magnitude on the price deflator would lead to a miss on the real net exports of goods contribution to growth of 0.5 percentage points or more, even if the other ingredients in the calculation were all correct.
Are there any lessons here for improving GDPNow? Ideally, GDPNow would be able to closely map the monthly source data to real goods net exports so that most of the weight would go to the bean counting forecast once all of the data are in—much as it does with nonresidential structures and residential investment. The BEA's estimates of real petroleum imports are based on similar data in the monthly international trade data publication. Because petroleum imports account for so much of the volatility of inflation for goods imports, it may be better to use the monthly real petroleum imports data directly and only worry about replicating the price index for nonpetroleum goods.
That said, a previous macroblog post illustrated that the method GDPNow currently uses has a reasonable forecasting track record for net exports when compared with several consensus estimates from professional forecasters. Net exports may remain difficult to nowcast even with refinements to GDPNow's methodology.
GDPNow has established a commendable track record. But sometimes when it misses the mark, an analysis of the error can provide insight into how GDPNow works and the limitations of the model.
- Labor Supply Constraints and Health Problems in Rural America
- Building a Better Model: Introducing Changes to GDPNow
- How Ill a Wind? Hurricanes' Impacts on Employment and Earnings
- When Health Insurance and Its Financial Cushion Disappear
- What Is the "Right" Policy Rate?
- Is Poor Health Hindering Economic Growth?
- Behind the Increase in Prime-Age Labor Force Participation
- An Update on Labor Force Participation
- Another Look at the Wage Growth Tracker's Cyclicality
- GDPNow's Second Quarter Forecast: Is It Too High?
- November 2017
- October 2017
- September 2017
- August 2017
- July 2017
- May 2017
- April 2017
- March 2017
- February 2017
- January 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