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 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.
October 24, 2016
Is Wage Growth Accelerating?
The Atlanta Fed's Wage Growth Tracker came in at 3.6 percent in September, up from 3.3 percent in August and 3.4 percent in July, but the same as the 3.6 percent reading for June. By this measure, there are no obvious signs of an acceleration in wage growth for continuously employed workers during the last few months.
However, the headline wage growth tracker is a three month moving average of each month's median wage growth. Interestingly, for September, the median wage growth (using data that are not averaged, sometimes called "unsmoothed") was 4.2 percent, up from 3.6 percent in August, and the highest since late 2007. This pop in median wage growth can be seen in the following chart, which compares the median wage growth (smoothed using a three-month average) with the unsmoothed monthly median.
Even though this looks like a pretty large increase, the standard error on the difference between the unsmoothed August and September medians is also quite large at 0.5 percentage points. So the 0.6 percentage point difference in medians is not statistically significant—it could just as easily be sampling noise. But it is definitely something to keep an eye on going forward. As noted in a previous macroblog post, the correlation between the unemployment gap and the Wage Growth Tracker suggests that we should be seeing the wage growth tracker level off if the economy is stabilizing at full employment.
October 18, 2016
Unemployment Risk and Unions
A recent paper by the Economic Policy Institute (EPI) argues that increased unionization would have broad economic benefits and, in particular, could help improve the wage stagnation facing many lower-skilled workers. Yet union membership has been declining, down by about 3 million between 1983 and 2015, and membership is down 4.5 million in the private sector. (Union membership in the United States is discussed in this U.S. Bureau of Labor Statistics report and in this database, maintained by Barry Hirsch at Georgia State University.)
The overall membership decline in private-sector unions reflects a combination of lower employment in some traditionally unionized industries such as the steel and auto industries and lower unionization rates within industries. For example, the rate of unionization for goods-producing industries (largely manufacturing and construction) is down from 28 percent to 10 percent, and the rate in service-producing industries has declined from 11 percent to 6 percent. In contrast, union membership in the public sector has increased, mostly as a result of broad unionization among public safety, utility, and education occupations coupled with the fact that employment in these occupations has tended to grow over time.
For goods-producing industries in particular, unionized employment is down by about 4.2 million since 1983, and nonunionized employment is up by around 2.5 million. Many factors may have contributed to this shift away from union membership. A possibility I explore here is the role of wage rigidity. In particular, if union wage contracts prevent employers from adjusting wages in the face of an unexpected decline in output demand, then employers may adjust along the employment margin instead. The monopoly power of unions leads to higher wages for continuously employed union workers but also makes layoffs more frequent.
It is the case that unionized workers tend to earn more than their nonunion counterparts. For 1983 to 2015, I estimate that prime-age union workers in goods-producing industries earn an average of about 25 percent more (on a median hourly basis) than comparable nonunion workers (about 50 percent more in construction and about 10 percent more in manufacturing). In addition, the median wage growth of union workers is less cyclically sensitive. The following chart uses the Atlanta Fed's Wage Growth Tracker data, and it shows the annual median wage growth of continuously employed prime-age workers in goods-producing industries, by union status.
Not only is wage growth among union workers less variable over time as the chart shows, research has noted that union wages are less dispersed—even controlling for differences in worker characteristics. Joining a union leads to wages that tend to be higher, wages that vary less across workers, and wage growth that responds less to changes in economic conditions.
But what about unemployment risk? Do union workers get laid off at a greater rate than nonunion workers? Using matched data from the Current Population Survey, the following chart shows an estimate of the probability that a prime-age worker in a goods producing industry is unemployed 12 months later, by union status.
The probability of unemployment rises during economic downturns for both union and nonunion workers, but is higher for union workers. The union worker displacement rate reached 13 percent in 2009 versus 8 percent for nonunion workers.
However, recall provisions are often built into collective bargaining agreements, so perhaps looking at the total unemployment flow overstates the permanent job loss risk. To investigate, the following chart shows the likelihood of being on temporary layoff (expected to be recalled within six months) versus indefinite (permanent) layoff.
The likelihood of being recalled by your previous employer is much higher for union than nonunion workers, whereas the incidence of permanent layoff is about the same for both types of worker.
Admittedly, I'm not controlling for all the things about workers and employers that could influence employment and wage outcomes. But taken at face value, it appears that the likelihood of permanent job loss is no greater for union workers in goods-producing industries than for nonunion workers. At the same time, union workers are more likely to experience a spell of temporary unemployment. I view this as some evidence in support of my wage rigidity story, which holds that unionized firms use layoffs more intensively because wages are less flexible (I find that this same result holds if I look at the manufacturing and construction industries separately). However, this mechanism itself isn't able to account for much of the secular decline in union participation. The decline seems to be more about where the jobs are created than where they are lost.
September 30, 2016
A Quick Pay Check: Wage Growth of Full-Time and Part-Time Workers
In the last macroblog post we introduced the new version of the nominal Wage Growth Tracker, which allows a look back as far as 1983. We have also produced various cuts of these data comparable to the ones on the Wage Growth Tracker web page to look at the wage dynamics of various types of workers. One of the data cuts compares the median wage growth of people working full-time and part-time jobs. As we have highlighted previously, the median wage growth of part-time workers slowed by significantly more than full-time workers in the wake of the Great Recession. The extended time series allows us to look back farther to see if this phenomenon was truly unique.
The following chart shows the extended full-time/part-time median wage growth time series at an annual frequency.
The chart shows that the median wage increase for part-time workers is generally lower than for full-time workers, with the average gap about 1 percentage point. The reason for the presence of a gap is a bit puzzling. Could it be that part-time workers have lower average productivity growth than full-time workers? It is true that a part-time worker in our data set is more likely to lack a college degree than a full-time worker, and the median wage level for part-time workers is lower than for full-time workers. But interestingly, a reasonably systematic wage growth gap still exists after controlling for differences in the education and age of workers. So even highly educated prime-age, part-time workers tend to have lower median wage growth than their full-time counterparts. If it's a productivity story, its subtext is not easily captured by observed differences in education and experience.
Changes in economic conditions might also be playing a role. The wage growth gap exceeded 2 percentage points in the early 1980s and again between 2011 and 2013, both periods of considerable excess slack in the labor market, as we recently discussed here. In fact, in each of 2011, 2012, and 2013, half of the part-time workers in our dataset experienced no increase in their rate of pay at all.
To explore this possibility further, it's useful to separate part-time workers into those who work part-time because of economic conditions (for example, because of slack work conditions at their employer or their inability to find full-time work) from those who work part-time for noneconomic reasons (for example, because they have family responsibilities or because they are also in school). The following chart shows the median wage growth for full-time, voluntary part-time, and involuntary part-time workers.
Admittedly, there are not that many observations on involuntary part-time workers in our data set. But it does appear that their median wage growth has tended to slow by more after economic downturns than those working part-time for a noneconomic reason—at least prior to the Great Recession. After the last recession, however, the wage growth gap was just about as large for both types of part-time workers. In that sense, the impact of the last recession on the median wage growth of regular part-time workers was quite unusual.
Since 2013, median wage growth for part-time workers has been rising, which is good news for those workers and consistent with the labor market becoming tighter. With the unemployment rate reasonably low, employers might have to worry a bit more about retaining and attracting part-time staff than they did a few years ago.
September 27, 2016
Back to the '80s, Courtesy of the Wage Growth Tracker
Things have been a wee bit quiet in macroblog land the last few weeks, chiefly because our time has been devoted to two exciting new projects. The first is a refresh of our labor force dynamics website, which will feature a nifty tool for looking at the main reasons behind changes in labor force participation for different age groups. More on that later.
The other project has been adding more history to our Wage Growth Tracker. The tracker's current time series starts in 1997. The chart below shows an extended version of the tracker that starts in 1983.
Recall that the Wage Growth Tracker depicts the median of the distribution of 12-month changes of matched nominal hourly earnings. In the extended time series, you'll notice two gaps, which resulted from the U.S. Census Bureau scrambling the identifiers in its Current Population Survey. For those two periods, you'll have to use your imagination and make some inferences.
As we have emphasized previously, the Wage Growth Tracker is not a direct measure of the typical change in overall wage costs because it only looks at (more or less) continuously employed workers. But it should reflect the amount of excess slack in the labor market. This point is illustrated in the following chart, which compares the Wage Growth Tracker with the unemployment gap computed from the Congressional Budget Office's (CBO) estimate of the long-run natural rate of unemployment.
As the chart shows, our measure of nominal wage growth has historically tracked the cyclical movement in the unemployment rate gap estimate fairly well, at least since the mid-1980s. We think this feature is potentially important, because the true unemployment rate gap is very hard to know in real time and hence is subject to potentially large revision. For example, in real time, the unemployment rate was estimated to have fallen below the natural rate in the fourth quarter of 1994, but it is now thought to have not breached the natural rate until the first quarter of 1997—more than two years later. The Wage Growth Tracker is not subject to revision (although it is subject to a small amount of sampling uncertainty) and hence could be useful in evaluating the reliability of the unemployment rate gap estimate in real time.
This also is important from a monetary policy perspective if we are worried about the risk of the economy overheating. For example, President Rosengren of the Boston Fed described why he dissented at the most recent Federal Open Market Committee meeting in favor of a quarter-point increase in the target range for the federal funds rate. His dissent, he said, arose partly from his concern that the economy may overheat and drive unemployment below a level he believes is sustainable.
Currently, the CBO estimate of the unemployment rate gap looks like it is plateauing at close to zero. The fact that the Wage Growth Tracker for the third quarter slowed a bit is consistent with that. But it's only one quarter of data, and so we'll closely monitor the Wage Growth Tracker in the coming months to see what it suggests about the actual unemployment rate gap. We'll discuss what observations we make here.
August 15, 2016
Payroll Employment Growth: Strong Enough?
The U.S. Bureau of Labor Statistics' estimate of nonfarm payroll employment is the most closely watched indicator of overall employment growth in the U.S. economy. By this measure, employment increased by 255,000 in July, well above the three-month average of 190,000. Yet despite this outsized gain, the unemployment rate barely budged. What gives?
Well, for a start, there is no formal connection between the payroll employment data and the unemployment rate data. The employment data used to construct the unemployment rate come from the Current Population Survey (CPS) and the payroll employment data come from a different survey. However, it is possible to relate changes in the unemployment rate to the gap between the CPS and payroll measures of employment, as well as changes in the labor force participation (LFP) rate, and the growth of payroll employment relative to the population.
The following chart shows the contribution of each of these three factors to the monthly change in the unemployment rate during the last year.
A note about the chart: The CPS employment and population measures have been smoothed to account for annual population control adjustments. The smoothed employment data are available here. The method used to compute the contributions is available here.
The black line is the monthly change in the unemployment rate (unrounded). Each green segment of a bar is the change in the unemployment rate coming from the gap between population growth and payroll employment growth. Because payroll employment has generally been growing faster than the population, it has helped make the unemployment rate lower than it otherwise would have been.
But as the chart makes clear, the other two factors can also exert a significant influence on the direction of the unemployment rate. The labor force participation rate contribution (the red segments of the bars) and the contribution from the gap between the CPS and payroll employment measures (blue segments) can vary a lot from month to month, and these factors can swamp the payroll employment growth contribution.
So any assumption that strong payroll employment gains in any particular month will automatically lead to a decline in the unemployment rate could, in fact, be wrong. But over longer periods, the mapping is a bit clearer because it is effectively smoothing the month-to-month variation in the three factors. For example, the following chart shows the contribution of the three factors to 12-month changes in the unemployment rate from July 2012 to July 2013, from July 2013 to July 2014, and so on.
Gains in payroll employment relative to the population have helped pull the unemployment rate lower. Moreover, prior to the most recent 12 months, declines in the LFP rate put further downward pressure on the unemployment rate. Offsetting this pressure to varying degrees has been the fact that the CPS measure of employment has tended to increase more slowly than the payroll measure, making the decline in the unemployment rate smaller than it would have been otherwise. During the last 12 months, the LFP rate turned positive on balance, meaning that the magnitude of the unemployment rate decline has been considerably less than implied by the relative strength of payroll employment growth.
Going forward, another strong payroll employment reading for August is certainly no guarantee of a corresponding decline in the unemployment rate. But as shown by my colleagues David Altig and Patrick Higgins in an earlier macroblog post, under a reasonable range of assumptions for the trend path of population growth, the LFP rate, and the gap between the CPS and payroll survey measures of employment, payroll growth averaging above 150,000 a month should be enough to cause the unemployment rate to continue declining.
July 29, 2016
Men at Work: Are We Seeing a Turnaround in Male Labor Force Participation?
A lot has been written about the long-run decline in the labor force participation (LFP) rate among prime-age men (usually defined as men between 25 and 54 years of age). For example, see here, here, here, and here for some perspectives.
On a not seasonally adjusted basis, the Bureau of Labor Statistics estimates that the LFP rate among prime-age males is down from 90.9 percent in the second quarter of 2007 to 88.6 percent in the second quarter of 2016—a decline of 2.3 percentage points, or around 1.4 million potential workers.
Many explanations reflecting preexisting structural trends have been posited for this decline. But how much of the decline also reflects cyclical effects and, in particular, cyclical effects that take a while to play out? We don't really know for sure. But one potentially useful approach is to look at the Census Bureau's Current Population Survey and the reasons people give for not wanting a job. These reasons include enrollment in an educational program (especially prevalent among young individuals), family or household responsibilities (especially among prime-age women), retirement (especially among older individuals), and poor health or disability (widespread). In addition, there are people of all ages who say they want a job but are not counted as unemployed. For example, they aren't currently available to work or haven't looked for work recently because they are discouraged about their job prospects.
To get some idea of the relative importance of these factors, the following chart shows how much each nonparticipation reason accounted for the total change in the LFP rate among prime-age males between 2012 and 2014 and between 2014 and 2016. The black bars show each period's total change in the LFP rate. The green bars are changes that helped push participation higher than it otherwise would have been, and the orange bars are changes that helped hold participation lower than it otherwise would have been.
A note on the chart: To construct the contributions derived from changes in nonparticipation rates, I held constant the age-specific population shares in the base period (2012 and 2014, respectively) in order to separate the effect of changes in nonparticipation from shifts in the age distribution.
Notice that the decline in the prime-age male LFP rate between 2012 and 2014 has essentially fully reversed itself over the last two years (from a decline of 0.53 percentage points to an increase of 0.55 percentage points, respectively). The positive "want a job" contribution in both periods clearly reflects a cyclical recovery in labor market conditions. But the most striking change between 2012–14 and 2014–16 is the complete reversal of the large drag attributable to poor health and disability. Other things equal, if nonparticipation resulting from poor health and disability had stayed at its 2012 level, prime-age male participation in 2014 would have only declined 0.10 percentage points. If nonparticipation due to poor health and disability had stayed at its 2014 level, prime-age male participation in 2016 would have increased only 0.14 percentage points.
The incidence of self-reported nonparticipation among prime-age men because of poor health or disability has been declining recently. According to the Current Population Survey data, this reason represented 5.4 percent of the prime-age male population in the second quarter of 2016. Although this is still 0.7 percentage points higher than in 2007, it is 0.3 percentage points lower than in 2014. Some of this turnaround could be the result of changes in the composition of the prime-age population. But not much. Around 90 percent of the LFP rate change because of poor health and disability is due to age-specific nonparticipation rather than shifts in the age distribution, suggesting that some of the turnaround in the incidence of people saying they are "too sick" to work is a cyclical response to strengthening labor market conditions. We've yet to see how much longer this turnaround could continue, but it's an encouraging development.
For those interested in exploring the contributions to the changes in the LFP rate by gender and age over different time periods, we're currently developing an interactive tool for the Atlanta Fed's website—stay tuned!
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.
June 22, 2016
Was May's Drop in Labor Force Participation All Bad News?
The unemployment rate declined 0.3 percentage points from April to May, and this was accompanied by a similar drop in the labor force participation rate. It is tempting to interpret this as a “bad” outcome reflecting a weakening labor market. In particular, discouraged about their job-finding prospects, more unemployed workers left the labor force. However, a closer look at the ins and outs of the labor force suggests a possibly less troubling interpretation of the outflow from unemployment.
To get a handle on what is going on, it is useful to look at the number of people that transition among employment, unemployment, and out of the labor force. It is not that unusual for an individual to search for a job in one month and then enroll in school or assume family responsibilities the next. In fact, each month millions of individuals go from searching for work to landing a job or leaving the labor force, and vice versa.
The U.S. Bureau of Labor Statistics (BLS) publishes estimates of these gross flows. Analyzing these data shows that there was indeed an unusually large number of unemployed persons leaving the labor force in May. Curiously, the outflow was concentrated among people who had only been unemployed only a few weeks. It wasn't among the long-term unemployed. Therefore, it seems unlikely that discouragement over job-finding prospects was the main factor. Although it is plausible that people who say they are now doing something else outside the labor market feel disheartened, the number of unemployed who said they gave up looking because they were discouraged was largely unchanged in May.
So why was there an increase in the number of short-term unemployed who left the labor force in May? One clue is provided by the fact that the short-term unemployed tend to be relatively younger than other unemployed. Moreover, the single most common reason that unemployed young people leave the labor force is to go to school. Hence, there is a very distinct seasonal pattern in the outflow. It tends to be relatively low around May when school is ending and high around August when school is starting. Seasonal adjustment techniques correct for these patterns by lowering the unadjusted data in the fall and raising it in late spring.
The following chart shows the seasonally adjusted and unadjusted flow from unemployment to departure from the labor force. Although the trend has been declining during the last few years, a relatively large increase in the seasonally adjusted outflow took place in May of this year.
When I looked at the unadjusted microdata from the Current Population Survey (CPS), I found that the number of people who were unemployed in April 2016 but in May said that they were not in the labor force because they were in school did not exhibit the usual large seasonal decline. Therefore, when the seasonal adjustment is applied, the result is an increase in the estimated flow from unemployment to out of the labor force.
Taking the seasonally adjusted data at face value, it's not obvious that this is bad news. We know that people who leave unemployment to undertake further education tend to rejoin the labor force later. Moreover, they tend to rejoin with better job-finding prospects than when they left. Alternatively, it could be just a statistical quirk of the May survey. After all, the CPS has a relatively small sample, so the estimated flows have a large amount of sampling error. Either way, I don't think it is wise to conclude that the decline in the labor force participation in May reflected a marked deterioration in job-finding prospects. In fact, the job-finding rate among unemployed workers improved in May from 22 to 24 percent, contributing to the decline in the unemployment rate.
- The Fed’s Inflation Goal: What Does the Public Know?
- Going to School on Labor Force Participation
- Bad Debt Is Bad for Your Health
- Working for Yourself, Some of the Time
- Gauging Firm Optimism in a Time of Transition
- Can Tight Labor Markets Inhibit Investment Growth?
- More Ways to Watch Wages
- Unemployment versus Underemployment: Assessing Labor Market Slack
- Does a High-Pressure Labor Market Bring Long-Term Benefits?
- Net Exports Continue to Bedevil GDPNow
- April 2017
- March 2017
- February 2017
- January 2017
- December 2016
- November 2016
- October 2016
- September 2016
- August 2016
- July 2016
- 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