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.

June 21, 2016

Wage Growth for Job Stayers and Switchers Added to the Atlanta Fed's Wage Growth Tracker

The Atlanta Fed's Wage Growth Tracker (WGT) moved higher again in May—the third increase in a row and consistent with a labor market that is continuing to tighten. At 3.5 percent, the WGT is at a level last seen in early 2009.

As was noted in an early macroblog post, when the labor market is tightening, people changing jobs experience higher median wage growth than those who remain in the same job. Median wage growth for job switchers has significantly outpaced that of job stayers in recent months. For job stayers, the May WGT was 3.0 percent, the same as in April, whereas for people switching jobs the median WGT increased from 4.1 percent to 4.3 percent in May (the highest reading since December 2007; see the chart).

Wage Growth Tracker

Because these patterns over time can help shed light on the relative strength of the labor market, we have added downloadable job stayer and job switcher WGT series to the Atlanta Fed's Wage Growth Tracker web page.

I should note that it is not possible to completely identify people who are in the same job as a year ago according to data from the Current Population Survey. Instead, we define a "job stayer" as someone whom we observe in the same occupation and industry as a year earlier, and with the same employer in each of the last three months. A "job switcher" includes everyone else (a different occupation or industry or employer). We'll be monitoring these data in coming months to see if discernable trends begin to emerge, and we'll discuss any findings here.

June 21, 2016 in Employment, Labor Markets, Wage Growth | 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

June 16, 2016

Experts Debate Policy Options for China's Transition

After nearly three decades of rapid economic growth, China today faces the challenge of economic rebalancing against the backdrop of slow and uncertain global growth. Although investment and exports have been a motor for growth, China is increasingly experiencing structural issues: widening inequality, overcapacity as a consequence of policy distortions, unsustainable environmental costs, volatile financial markets, and rising systemic risk.

On April 28–29, I attended the First Research Workshop on China's Economy, organized jointly by the International Monetary Fund (IMF) and the Atlanta Fed. The workshop, held at the IMF's headquarters in Washington DC, explored a series of questions that have emerged as China shifts toward a new growth model. Is this the end of the growth miracle? Will the Chinese renminbi one day be as important as the U.S. dollar? Should the rapidly increasing shadow banking activity in China be a source of concern? How worrisome is the rapid rise in China's housing prices?

Panelists shared their views on these and other issues facing the world's second-largest economy (or largest, if measured on a purchasing-power-parity basis). Plans are under way for a second workshop to be held in 2017.

The following is a nice summary of the research discussed at the workshop. It was originally published in the IMF Survey Magazine, and was written by Hui He, IMF Institute for Capacity Development, and Nan Li, IMF Research Department. Thanks to the IMF for allowing me to repost it here.

Is China's economic growth sustainable?
Understanding the source of China's tremendous growth was a recurring theme at the workshop. "China's economy combines enormous dynamism with huge distortions," observed Loren Brandt (University of Toronto). Brandt described his research based on China's firm-level data and emphasized that firm dynamics (entry and exit), especially firm entry, have been the main source of the productivity growth in the manufacturing sector.

Echoing Brandt's message, Kjetil Storesletten (University of Oslo) discussed regional growth disparities and showed that barriers preventing firms from entering an industry account for most of the disparities. Such barriers are more severe for privately owned firms in regions in which state-owned enterprises (SOE) dominate, he said.

In his keynote speech, Nicholas Lardy (Peterson Institute for International Economics) offered an upbeat view on China's transition to a new growth model, one in which the service sector plays a larger role than manufacturing. The bright side of the service sector, he noted, is its continued strong productivity growth. The development of financial deepening and the stronger social safety net are contributing to increased consumption, which helps to rebalance the economy.

However, he emphasized, SOE reforms remain critical as the service sector cannot provide a silver bullet for a successful transition.

Central bank's policy decisions
Several participants tried to discern how the People's Bank of China (PBC) conducts monetary policy. Tao Zha (of the Atlanta Fed's Center for Quantitative Economic Research and Emory University) found that the PBC reacts sharply when the gross domestic product's growth rate falls below its target, increasing the money supply by 11.5 percentage points for every 1 percentage point shortfall.

Mark Spiegel (Center for Pacific Basin Studies) discussed the trade-offs involved in Chinese monetary policy—for example, controlling the exchange rate versus maintaining inflation stability. He also argued that the heavy use of reserve requirements on banks as a monetary policy tool might have an unintentional consequence to reallocate capital from SOEs to more efficient privately owned firms and could therefore offset the resource misallocation caused by the easy credit to SOEs that banks granted in the high growth years.

Renminbi versus the dollar
Eswar Prasad (Cornell University and Brookings Institution) argued that China's capital account will become more open and the renminbi will be used more widely to denominate and settle cross-border transactions. But he also noted that legal and institutional constraints in China were likely to prevent the renminbi from serving as a safe-haven currency as the U.S. dollar does today.

Moreover, he said, the current sequencing of liberalization initiatives—that is, removal of capital account restrictions before appropriate financial market supervision and regulation and exchange rate reform—poses financial stability risks.

Shadow banking and the housing market
Recently, volatile Chinese financial markets and continued housing price appreciation have raised serious financial stability concerns.

Michael Song (Chinese University of Hong Kong) argued that rapidly rising shadow banking activity is an unintended consequence of financial regulation. Restrictions on deposit rates and loan-to-deposit ratios have led to the issuance by banks of "wealth management products" to attract savers with higher returns. Because these restrictions had a greater impact on small banks, the big state banks had more room to undercut the smaller banks by offering wealth management products with higher returns and then restricting liquidity to them in interbank markets, ultimately making the banking system more prone to liquidity distress and runs.

Hanming Fang (University of Pennsylvania) found that, except in big cities such as Beijing and Shanghai, housing prices in China's urban areas between 2003 and 2013 more or less tracked rising household incomes. In his view, the Chinese housing boom is thus unlikely to trigger an imminent financial crisis. He warned, however, that housing prices may fall rapidly if economic growth slows dramatically, and that such a development could, in turn, amplify the economic downturn.

Rising wage inequality
China's rapid growth over the past two decades has been accompanied by rising wage inequality, an issue highlighted by two conference participants. Dennis Yang (University of Virginia) explored the distributional effects of trade openness in China and found a significant impact on wage inequality of China's accession to the World Trade Organization in 2001.

Chong-En Bai (Tsinghua University) argued that the decline after 2008 of the skill premium—that is, the ratio of the skilled labor wage to the unskilled labor wage—can be explained by the Chinese government's targeted credit extension to unskilled labor-intensive infrastructure sector (as part of the fiscal stimulus following the global financial crisis). Such distortionary policies might have short-run growth benefits but could lead to long-run welfare losses, he said, especially when rural-to-urban migration has run its course.

June 16, 2016 in Asia, Economic Growth and Development, Labor Markets, Monetary Policy, Real Estate | 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

June 02, 2016

Moving On Up

People who move from one job to another tend to experience greater proportionate wage gains than those who stay in their job, except when the labor market is weak and there are relatively few employment options. This point was illustrated using the Atlanta Fed's Wage Growth Tracker in this macroblog post from last year.

Given that the Wage Growth Tracker ticked higher in April, it is interesting to see how much of that increase can be attributed to job switching. Here's what I found:

Wage Growth Tracker

A note about the chart: In the chart, a "job stayer" is defined as someone who is in the same occupation and industry as he or she was 12 months ago and has been with the same employer for at least the last three months. A "job switcher" is everyone else.

The overall Wage Growth Tracker for April was 3.4 percent (up from 3.2 percent in March). For job stayers, the Tracker was 3.0 percent (up from 2.9 percent), and for job switchers it was 3.9 percent (up from 3.7 percent). So the wage gains of job switchers do appear to have helped pull up our overall wage growth measure.

Moreover, unlike the wage growth of job stayers, job switchers are now tending to see wage growth of a similar magnitude to that experienced before the recession. This observation is broadly consistent with the improvement seen during the last year in the quits rate (the number of workers who quit their jobs as a percent of total employment) from the Job Openings and Labor Turnover Survey.

I think it will be interesting to continue to monitor the influence of job switching on wage growth as a further indicator of improving labor market dynamism. An update that includes the May data should be available in a few weeks.

June 2, 2016 in Employment, Labor Markets, Wage Growth | 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

June 01, 2016

Putting the Wage Growth Tracker to Work

The April pop in the Atlanta Fed's Wage Growth Tracker has attracted some attention in recent weeks, resulting in some interesting analysis. What is the tracker telling us about the tightness of the labor market and the risks to the inflation outlook?

We had earlier noted the strong correlation between the Wage Growth Tracker and the unemployment rate. Tim Duy took the correlation a step further and estimated a wage Phillips curve. Here's what he found:

The chart shows that lower unemployment generally coincides with higher wage growth (as measured by the Wage Growth Tracker), but wage growth varies a lot by unemployment rate. In the past, an unemployment rate around 5 percent has often been associated with higher wage growth than we currently have.

If the Wage Growth Tracker increased further, would that necessarily lead to an increase in inflation? Jared Bernstein suggests that there isn't much of an inflation signal coming from the Wage Growth Tracker. His primary evidence is the insignificant response of core personal consumption expenditure (PCE) inflation to an increase in the Wage Growth Tracker in a model that relates inflation to lags of inflation, wage growth, and the exchange rate.

However, I don't think the absence of a wage-push inflation connection using the Wage Growth Tracker is really that surprising. The Wage Growth Tracker better captures the wage dynamics associated with improving labor market conditions than rising labor cost pressures per se. For example, if firms are replacing departing workers with relatively low-wage hires, then the wages of incumbent workers could rise faster than do total wage costs (as this analysis by our colleagues at the San Francisco Fed shows). That said, as Bernstein also pointed out in the Washington Post, it's also pretty hard to find evidence of wage pass-through pushing up inflation in his model using more direct measures of labor costs.

I look forward to seeing more commentary about Atlanta Fed tools like the Wage Growth Tracker and how they can be part of the broader discussion of economic policy.

June 1, 2016 in Employment, Labor Markets, Wage Growth | 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

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 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

April 29, 2016

Is the Number of Stay-at-Home Dads Going Up or Down?

A recent Wall Street Journal post observed that most of the recession's "stay-at-home dads" are going back to work. Specifically, data from the U.S. Labor Department shows that the share of married men with children under 18 who are not employed (but their spouse is) rose during the recession and has since given back much of that increase, as the Journal's chart below indicates.

Rise and Decline of the Stay-at-Home Dad

Of course, being a stay-at-home dad in the sense defined in the previous chart (that is, not employed) can be either involuntary because of unemployment, or it can be the result of a voluntary decision to not be in the workforce. Most of the variation in the previous chart is cyclical, suggesting that it is related to the rise and fall in unemployment. But it also looks like the share of stay-at-home dads is higher now than it was a decade or so ago. So perhaps there is also an increasing trend in the propensity to voluntarily be a stay-at-home dad.

To explore this possibility, the next chart shows the annual average share of married men ages 25–54 who have children and who say the main reason they do not currently want a job is because of family or household responsibilities. (This reason doesn't necessarily imply that they are looking after children, but it is likely to be the leading reason.) The fraction is very small—about 1.3 percent in 2015, or 285,000 men—but the share has more than doubled during the last 15 years and would account for about half of the elevated level of the stay-at-home rate in 2015 relative to 2000.

Share of 25- to 54-year-old married men with children who don't currently want a job because of family or household responsibilities

So although large numbers of unemployed stay-at-home dads have been going back to work, it also appears that there's a small but growing group of men who are choosing to take on household and family responsibilities instead.

April 29, 2016 in Employment, Labor Markets, Unemployment | Permalink


what happens if we actually model this explicitly? Using controls for cyclical effects? My concern is that you are assuming that the most recent point on the chart is some kind of long run point. Not that I have substantial evidence to the contrary, but it appears to me the trend may well continue

Posted by: g | April 30, 2016 at 02:39 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

April 15, 2016

Labor Force Participation: Aging Is Only Half of the Story

The labor force participation rate (LFPR) is an important ingredient in projecting employment growth and the unemployment rate. However, predicting the LFPR has proven difficult. For example, in 2011 the Congressional Budget Office (CBO) projected that the LFPR in 2015 would be about 64.3 percent. In reality, the LFPR turned out to be 62.6 percent. Based on the CBO projection, the economy would have needed to create about 4 million more jobs to reach the 2015 unemployment rate of 5.3 percent.

Why is the LFPR so hard to predict? Leaving aside the challenge of projecting the size of the population, movements in LFPR primarily reflect shifts in the age distribution of the population as well as a number of behavioral factors. Although the aging trends are largely baked in, the behavioral factors vary over time. According to our estimates, about half of the 3.4 percentage-point decline in the LFPR between 2007 and 2015 is the result of the aging of the population, while behavioral factors account for the rest.

The complication is that the specific behaviors can change. The following chart shows a decomposition of the change in LFPR from 2007 to 2011 and from 2011 to 2015. Though the aging of the population contributed about the same amount to the decline in LFPR in both periods, the contributions from other factors have varied a lot. (We delve into the changes in the factors following the chart.)

Aging: The single largest factor contributing to the decline in the overall LFPR has been the rising share of older Americans in the population. In 2007, about one in five Americans were over 60 years old. In 2015, almost one in four were over 60. Moreover, this demographic force will continue to suppress the overall LFPR as the share of older Americans increases further in coming years. (For an in-depth discussion of the economic implications of an aging population—including changes in the labor market—please read the Atlanta Fed's 2015 annual report.)

Later retirement: One countervailing factor to an aging population has been the rising LFPR of older individuals. The retirement rate of those over 60 declined between 2007 and 2011 by a similar amount as it had before the recession. However, the trend toward later retirement has slowed considerably in recent years. The reason for this slowing is a puzzle and has important implications for the future course of overall LFPR.

Schooling among the young: The enrollment in educational programs by American youth has been generally rising over several decades, and this trend has put downward pressure on the overall LFPR. During the 2007–11 period, the share of 16- to 24-year-olds who do not want a job because they were in school or college accelerated relative to the prerecession trend. However, enrollment rates have since flattened out. The slowing may reflect enrollment rates catching up to the longer-term trend or may be a result of changes in the opportunity cost of education.

Not in the labor force but want a job: The share of the population saying they want a job but are not classified as unemployed by the U.S. Bureau of Labor Statistics definition is countercyclical—it tends to go up during bad times and down during good times. The relative size of this group increased between 2007 and 2011 and has since retraced about half of that increase as the economy has strengthened. We expect that this category will continue to shrink some more as the economy continues to expand.

Health: The share of individuals who do not want a job because they were too ill or disabled to work has increased over time. The relative size of this group increased between 2007 and 2011. Since 2011, the rate of increase has slowed, and it actually declined in 2015. It is not clear what drove the larger increase during the 2007–11 period, but there is some literature linking weak labor market conditions to poor health outcomes.

Prime-age reasons for not wanting a job (other than health): During the recession, the share of prime-age (ages 25 to 54) women not wanting a job because of household or family responsibilities decreased. One explanation is that some women entered the labor force to help make ends meet. At the same time, there was an offsetting effect from a rise in educational enrollment. Since the recession, nonparticipation because of household or family responsibilities has returned to near prerecession levels, and educational enrollment has leveled off.  

To summarize, we find that relative to the 2007–11 period there has been a:

  • flattening in the rate of retirement by older individuals,
  • flattening in the rate of educational program enrollment by younger individuals,
  • declining share of the population saying they want a job but not officially counted as unemployed,
  • smaller drag from nonparticipation because of health, and
  • larger drag for reasons other than health among prime-age individuals.

Where will LFPR be by the end of 2016? What about five years from now?
During the first three months of 2016, there has been an increase in the overall LFPR. This was largely the result of a decline in the share of prime-age people citing health reasons for nonparticipation, with some contribution from a decline in the share who want a job but are not "unemployed."

Though these boosts to participation may offset the effect of an aging population in the short term, most forecasts have the LFPR declining over the next several years. How much participation will actually decline depends on the answers to several difficult questions, such as: Will older individuals push retirement out even farther? Will school enrollment rates rise more rapidly again? Will the health status of the population improve? The difficulty of answering these questions helps explain why making accurate labor force projections is challenging.


April 15, 2016 in Labor Markets | Permalink


Excellent as always. But can the behavioral factors in the 25-54 main cohort and its specific lfpr be shown in a separate post. Do behavioural factors of the older cohorts effect the 25-54 cohort and its lfpr.

Posted by: am | April 15, 2016 at 04:03 PM

I compared YoY changes in the labor participation rate with YoY changes in real median household income (as a proxy for wages). The relationship appears to be a close one. Apparently, the idea of opportunity cost is a major part of the explanation for changes in the labor participation rate.

See http://www.philipji.com/item/2016-04-16/what-explains-changes-in-the-labour-participation-rate

Posted by: Philip George | April 16, 2016 at 08:11 AM

If I'm retired (which I am), I'm not in the labor force anymore. Why am I counted in the LFPR? Retirees shouldn't impact LFPR one bit.

Posted by: Mac McVicker | April 28, 2016 at 08:58 PM

Dated but useful showing the change in age distribution from 2000 to 2014.
By a bls economist.

Posted by: am | June 17, 2016 at 05:56 AM

Dated but useful showing change in age distribution from 2000 to 2014.
By BLS economist.

Posted by: am | June 17, 2016 at 05:58 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

April 04, 2016

Which Wage Growth Measure Best Indicates Slack in the Labor Market?

The unemployment rate is close to what most economists think is the level consistent with full employment over the longer run. According to the Federal Open Market Committee's latest Summary of Economic Projections, the unemployment rate is currently only 15 basis points above the natural rate. Yet, average hourly earnings (AHE) for production and nonsupervisory workers in the private sector increased a paltry 2.3 percent in March from a year earlier (as did the AHE of all private workers), and is barely above its average course of 2.1 percent since 2009.

In contrast, the Atlanta Fed's Wage Growth Tracker (WGT) suggests that wage growth has been increasing. The February WGT reading was 3.2 percent (the March data will be available later in April), considerably higher than its post-2009 average of 2.3 percent.

Wage Growth

Why is there such a large difference between these measures of wage growth? Besides differences in data sources, the primary reason is that they measure fundamentally different things. The WGT is an estimate of the wage growth of continuously employed workers—the same worker's wage is measured in the current month and a year earlier.

In contrast, the AHE measure is an estimate of the change in the typical wage of everyone employed this month relative to everyone employed a year earlier. Most of these workers are continuously employed, but some of those employed in the current month were not employed the prior year, and vice versa. These changes in the composition of employment can have a significant effect.

A recent study by Mary C. Daly, Bart Hobijn, and Benjamin Pyle at the San Francisco Fed shows that while growth in wages tends to be pushed higher by the wage gains of continuously employed workers, the net effect of entry and exit into employment tends to put a drag on the growth in wages. Moreover, the magnitude of the entry/exit drag can be relatively large, varies over time, and differs by the type of entry and exit.

For example, older workers who have retired and left the workforce tend to come from the higher end of the wage distribution, and their absence from the current period wage pool exerts downward pressure on the typical wage. The greater number of baby boomers starting to retire is having an even larger depressing effect on growth in wages than in the past. Because the WGT looks only at continuously employed workers, it is not influenced by these net entry/exit effects.

To the extent that firms adjust the pay for incumbent workers in response to labor market pressures to attract and retain workers, the WGT should reasonably capture changes in the tightness of the labor market.

Economists at the Conference Board modeled the relationship between different wage growth series and measures of labor market slack. One of the slack measures they use is the unemployment gap—the difference between an estimate of the natural rate of unemployment and the actual unemployment rate.

To illustrate their findings, the following chart shows the WGT and AHE measures along with the unemployment gap lagged six months (using the Congressional Budget Office estimate of the natural rate).

Wage Growth and the Unemployment Gap

The WGT appears to move more closely with the lagged unemployment gap than does the growth in AHE, and a comparison of the correlation coefficients confirms the stronger relationship with the WGT. The correlation between the lagged unemployment gap and the change in average hourly earnings is 0.75.

In contrast, the correlation with the wage growth tracker is higher at 0.93. Moreover, the unemployment gap-AHE relationship appears to be particularly weak since the Great Recession. The correlation since 2009 falls to just 0.08 for the AHE, whereas the WGT correlation is still 0.93.

Our colleagues at the San Francisco Fed concluded their analysis of the effect of flows into and out of the employment on wage growth by suggesting that:

"... wage growth measures that focus on the continuously full-time employed are likely to do a better job of gauging labor market strength, since they are constructed to more clearly capture the wage dynamics associated with improving labor market conditions. The Federal Reserve Bank of Atlanta's Wage Growth Tracker is an example."

That assessment is consistent with the Conference Board study, and suggests that labor markets may be tighter than is commonly believed based on sluggish growth in measures of average wages such as AHE.

April 4, 2016 in Economic conditions, Employment, Labor Markets, Wage Growth | Permalink


There are some similar issues in the UK. On the statistics the difference between the price change (here nominal wages KA25 ) and the quantity changes (here in employment KA26) is set out in UK's Av Weekly Earnings series http://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/datasets/averageweeklyearningsbysectorearn02

Comparisons of the stats suggest number of differences between the countries. First, UK nominal wage wrt to labour market variables is (much) less than in UK.

Secondly, in both countries stats on 'insiders' (eg WGT) tends to be more responsive to recessions - perhaps because the link between pay and jobs is more immediate so they accept lower pay - than in recoveries where they are not as interested in the need to attract outsiders.

Thirdly, there has been a very different effect of 'outsiders' employment growth on total current wage growth in the countries. A caricature until recently is that the UK there has been growth in employment of people with below average wages and vice versa in the US.

One reason for this might be the relative success of welfare to work policies. In UK there has been a successful 'activation' policy including for new groups such as lone parent and those aged 60-64. Whereas in the US there has been increased welfare dependency - including the extension of unemployment benefit duration in existence until the of 2013.

Posted by: Bill Wells | April 05, 2016 at 07:19 AM

Maybe also check the distribution of wage income, as trophy earners may be seeing greater growth, dragging up the average, but perhaps not the median.


Posted by: john | April 05, 2016 at 08:27 AM

Is there data that shows reduction in pay for a post when the previous occupant of the post retired rather than quit.

Posted by: am | April 09, 2016 at 03:06 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

February 17, 2016

Are Paychecks Picking Up the Pace?

From the minutes of the January 26–27 meeting of the Federal Open Market Committee, it's clear that many participants saw tightening labor market conditions during 2015:

In their comments on labor market conditions, participants cited strong employment gains, low levels of unemployment in their Districts, reports of shortages of workers in various industries, or firming in wage increases.

Based on the Atlanta Fed's Wage Growth Tracker (WGT), the median annual growth in hourly wage and salary earnings of continuously employed workers in 2015 was 3.1 percent—up from 2.5 percent in 2014 and 2.2 percent in 2013. That is, the typical wage growth of workers employed for at least 12 months appears to be trending higher.

However, wage growth by job type varies considerably. For example, the WGT for part-time workers has been unusually low since 2010. The following chart displays the WGT for workers currently employed in part-time and full-time jobs. For those in part-time jobs, the WGT was 1.9 percent in 2015, versus 3.3 percent for those in full-time jobs. The part-time/full-time wage growth gap has closed somewhat in the last couple of years but is still large relative to its size before the Great Recession. Note that full-time WGT is similar to the overall WGT because most workers captured in the WGT data work full-time (81 percent in 2015).

Median Wage Growth by Hours Worked

In addition to hours worked, median wage growth also tends to vary across occupation. The following chart plots the WGT for workers in low-skill jobs, versus those in mid- and high-skill jobs. (We define low-skill jobs as those in occupations related to food preparation and serving; building and grounds cleaning; and maintenance, protection, and personal care services.)

Notably, after lagging during most of the recovery, median wage growth in low-skill occupations increased 2.8 percent in 2015, versus 2.0 percent in 2014 and compared to 3.2 and 2.7 percent for other occupations in 2015 and 2014, respectively.

The improvement in wage growth for low-skill occupations seems mostly attributable to full-time workers; wage growth for people in low-skill jobs working part-time was about half that (1.6 percent versus 3.0 percent) of those working full-time (see the chart).

Median Wage Growth by Occupation

This pickup in low-skill wage growth fits with some anecdotal reports we've been hearing. Some of our contacts in the Southeast have reported increasing wage pressure for workers in lower-skill occupations within their businesses. One can also see evidence of growing tightness in the market for low-skill jobs in the help-wanted data. As the following chart shows, the ratio of unemployed to online job postings for low-skill jobs is always higher than for middle- and high-skill occupations. But the ratio for low-skill jobs is now well below its prerecession level, and the tightness has increased during the last two years.

Supply/Demand Rate by Occupation

The take-away? Wage growth for continuously employed workers appears to have picked up some steam in 2015, and the recent trend in wage growth is positive across a variety of job characteristics. Wage growth for people in lower-skill jobs has increased during the last couple of years, consistent with evidence of increasing tightness in the market for those types of jobs. The largest discrepancy in wage growth appears to be among part-time workers, whose median gain in hourly wages in 2015 still fell well short of those in full-time jobs.

February 17, 2016 in Economic conditions, Employment, Labor Markets, Wage Growth | Permalink


Whenever someone tries to point to stagnant median wages as evidence of rising inequality over the last few decades , they will quickly be reminded that you need to look at total compensation , not just wages - i.e., to capture the rising costs of health care benefits , etc. , etc.

It seems to me that we should demand the same for studies like the one above. I can easily imagine a situation in which the low-end workers who seem to be gaining in wage growth could simultaneously be losing on the benefits side , resulting in little or no net gain , if not a net loss.

Posted by: Marko | February 18, 2016 at 12:33 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

Google Search

Recent Posts

April 2017

Sun Mon Tue Wed Thu Fri Sat
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 29



Powered by TypePad