If all the job openings in the United States were to be filled today, an additional 5 million Americans would be employed. That total is higher than the population of twenty-eight states, as well as every American city other than New York. It is the most job openings at any one time in the United States since 2001—enough to provide work for nearly three-in-five of the 8.7 million Americans who are now categorized as unemployed.

Of course, the notion of instantaneously and simultaneously filling all of America’s vacancies is a nice thought exercise, but not particularly realistic. A healthy economy will always have openings as it grows and changes, as businesses open and close, and as workers leave jobs and begin new ones.

But the sheer magnitude of current job availabilities raises important questions. Why are employers apparently having more difficulty filling openings than in the past? Is it because applicants lack the skills required? Or are businesses feeling uncertain about the durability of the recovery? What industries have rebounded most strongly after the Great Recession, and which have lagged behind?

This report, the third in The Century Foundation’s Working Paper Series, will explore these issues complicating the demand side of the U.S. labor market. But our guiding question will be a simple one: Where are the jobs?

Our exploration will emphasize government statistics on labor market flows—the movement of people through various employment statuses—that complement and enrich the picture provided by more familiar stock indicators, such as unemployment, which tell us where workers reside at a fixed moment in time.

The five key findings:

1. Now is the best time in recent years to be a job seeker. After falling steeply during the Great Recession, the job openings rate has fully recovered; similarly, after spiking during the recession, layoffs have fallen back to their normal rate. The pace of hiring is also nearly back to pre-recession levels.
2. White-collar industries—led by finance, health care, education, and professional services—are the best places to be looking for work. These sectors enjoy low unemployment, lots of openings, and strong compensation.
3. Most blue-collar fields are suffering. The construction industry was particularly hard-hit by the recession, and the sector’s job openings remain scarce. The same is true of manufacturing, which has also failed to regain pre-recession employment levels. As a result, many low-skill workers have had to seek refuge in the retail and leisure industries, which tend to have the worst-paying, least-stable jobs.
4. Government remains among the most desirable employers. Pay is good and job security is unparalleled in public sector jobs—for highly skilled and less-skilled workers alike. It is also America’s largest industry, accounting for one-in-six jobs. But the pace of government hiring, which plummeted after the recession, remains lackluster. Finding government work is no easy task.
5. The rich are getting richer. Industries with the highest average earnings have seen the fastest earnings growth.

Most discussions of the labor market focus on unemployment. But as important as unemployment is, it gives us only a partial view of the labor market: namely, the excess supply of workers—those who cannot match themselves to jobs. (If you’d like to read more about the limitations of unemployment statistics, see the first installment in the Working Paper Series, “Uncovering the Labor Market Recovery.”) To fully understand the labor market’s health, we also need to consider its demand side.

Fortunately, there’s a useful (if underappreciated) report from the Bureau of Labor Statistics that allows us to do exactly that. It’s called the Job Openings and Labor Turnover Survey, or JOLTS, for short.

Released (usually) on the second Tuesday of every month, and with less fanfare than its more popular labor-supply counterparts, JOLTS gives us the less-discussed side of the story: How many positions are sitting empty, and how aggressively are employers trying to fill them? In other words, in response to the unemployment rate’s proclamation of excess supply, JOLTS answers with a measure of excess demand.

EXPLAINER: An Introduction to JOLTS

Lots of openings—high labor demand—are a positive sign for workers, because it means the bargaining position of qualified applicants is stronger. Not only are job seekers more likely to find work, but when they do, it is likely to be better paying. On the other hand, if openings are relatively scarce, jobs will be harder to come by. Firms can be more selective with who they pick, and they can be stingier in the compensation they offer.

There’s more to JOLTS than just job openings, though. As its name suggests, it also measures turnover. Whereas most labor market indicators focus on stock quantities, like numbers of jobs or counts of the unemployed, JOLTS prioritizes labor market flows: the dynamics of how people move from one employment status to another. It does this by decomposing monthly employment changes into hires and separations, and further categorizes the latter based on whether departures are voluntary (quits) or unwelcome (layoffs).

In other words, JOLTS gives us insight into the reasons why the jobs numbers are evolving as they are. If employment is increasing in aggregate, is it because more people are being hired, or because fewer are being fired? Though both explanations are consistent with the same observed outcome, they have potentially different implications for policy.

Similarly, if we notice an increase in separations, it could be because layoffs are up, which would be bad news. But if, instead, the rise were attributable to an unexpected spike in quits, we would interpret the evidence much differently: when employees are confident enough to voluntarily leave their jobs, they must think the labor market is pretty strong.

Perhaps even more importantly, focusing on these flows can alert us to developments that static snapshots will miss. Even when headline numbers aren’t changing, the underlying reality can be evolving. For example, consider an economy where the total number of people employed has remained the same for three straight months. What’s going on? One extreme possibility is that every person kept the exact same job: no hires and no separations. At the other extreme, everyone could have changed jobs, so that even while the headline total is the same, the composition of employment is entirely different.

Neither of those scenarios is realistic, but at which end of the spectrum the reality lies has major implications for the well-being of labor market participants. The pace of job turnover matters.

Movement can be a good thing. Jobs turning over quickly can indicate a vibrant economy. When one person leaves a job, it creates an opportunity for someone else. Change generates creative dynamism, and with fresh starts come new ideas and newfound energy. Productivity will likely increase, and wages may follow.

But taken too far, excessive turnover can be disruptive, with recruiting, job search, and training taking up too much of society’s productive resources. It’s a delicate balance. JOLTS provides a clearer picture of the job market’s state of flux.

Figure 1 shows the trends in the three most important JOLTS indicators from 2001 to 2014: openings, hires, and separations. Each is expressed as a rate, or percent, of total employment.

Download

Given that aggregate employment usually changes very little from month to month, it is not surprising that the hires and separations rates generally track each other very closely. In 2007, for example, both averaged 3.7 percent monthly, with hires just barely outpacing separations. This means that in any given month of 2007, 3.7 percent of workers were newly hired, and 3.7 percent left—about equal to the typical shares entering and leaving employment since 2001.

During and after the recession, however, those trends diverged, with the hiring rate falling to a low of 2.8 percent, while the separation rate continued to average around 3.5 percent. The obvious result was a huge loss of jobs.

Just as tellingly, the jobs openings rate (openings divided by employment), which had risen significantly in the years preceding the recession—to 3.2 percent in 2006 and 2007 (from its 2000–2005 average of 2.6 percent)—dropped steeply to 1.6 percent in July 2009. In other words, during the recession, the job openings rate was halved. Not only were firms getting rid of workers—they didn’t want new ones, either.

But since mid-2009, the job openings rate has rebounded quite steadily, rising to an average of 3.3 percent for August-October 2014. At this point, it’s actually a bit higher than it was in pre-recession 2007. By contrast, both the hiring and separation rates remain below their 2007 averages, by 3.8 percent and 4.8 percent, respectively.

Total separations, however, is a fairly imprecise labor market indicator, because it includes both voluntary exits (quits) and involuntary ones (layoffs). Individually, each type of departure provides a gauge of labor market health, but together they can muddle the picture. Figure 2 splits apart these two measures.

Download

The layoffs rate spiked during the recession, increasing 51 percent, from 1.3 percent to 2.0 percent, but has since fallen back to 1.2 percent—below its 2007 average. That firms today are not forcing their employees out the door at the pace they were a few years ago is an indication there is strengthened market demand for the products and services these employees provide.

The quits rate, on the other hand, dropped from 2.1 percent to 1.3 percent during the recession—a sign that workers were reluctant to leave their jobs. But recently, it has been back on the rise, up to 1.9 percent from August to October 2014. Workers confident enough to leave their jobs is an indicator of labor market strength.

As you can see, layoffs and quits tend to move in the opposite direction. When times are good, workers are more apt to change jobs because the risk of not being able to find employment are smaller. When the economy is in a tailspin, firms shed workers—and those fortunate enough to keep their jobs hold on to them tightly. Our present situation—fewer layoffs and more quits—is exactly what we hope for. The discretionary job termination behavior of both firms and employees is evidence that both sides of the labor market are sensing a strengthening economy, underscored by the recent drop in the unemployment rate.

We can visualize the post-recession evolution of job flows even more clearly by plotting their rates of change during the past few years. Figure 3 hones in on the post-recession recovery of the key labor market flows, using layoffs rather than total separations. It shows the percentage change from the 2007 average for the hires, layoffs, and openings rates, using a six-month rolling average since can data can be somewhat volatile.

The extent of recession-inflicted damage is obvious from the steep peak in the layoffs rate and almost equally deep valley in openings. Yet both are now fully recovered, with openings 3.7 percent above pre-recession norms and layoffs down 9.4 percent. The hiring rate was less dramatically affected, but it has also recovered more slowly.

Download

Another useful indicator or labor market health is hires relative to layoffs. This hires-to-layoffs (H-L) ratio is similar to the hires rate, except that rather than using employment as the denominator, it uses layoffs. In this way, it brings together in a single number the two flows that together comprise labor demand (while removing quits, which is an indicator of labor supply). This metric also conveys positive news. In the years leading up to the recession, as Figure 4 shows, this ratio was often above 3.0, and averaged 2.8 in 2007. But during the recession it fell to just 1.5—that is, only one-and-a-half workers were being hired for every worker being fired. Since 2009, progress has been more or less steady, and from August to October 2014, it is back up to 3.0, better than it was in 2007.

Download

So we’re seeing positive trends in some of the important rates and ratios that characterize labor demand. But what do they mean in tangible terms? A job openings rate of 3.4 percent, as it was it November 2014, means there are about 5 million vacancies waiting to be filled in the United States right now. That’s enough to provide work for more than half of America’s roughly 8.7 million officially unemployed (as of December 2014). But if so many jobs are available, why aren’t more people finding work?

Even in the best of times, some people are bound to be temporarily out of work as they switch jobs. Transitions are not always smooth, and it can take some time for prospective employees and employers to come together. One job does not always end with the next neatly in-hand; in the career relay race, baton drops are common. Interviews and paperwork can stretch out over weeks; sorting out schedules and juggling family obligations can lengthen the process. And some workers must undergo training or licensing before they complete transitions.

People experiencing these types of delays are known as the frictionally unemployed, because, like marbles rolling across a sandy surface, they are hitting small bumps that slow their journey.

We don’t usually worry about frictional unemployment. Healthy economies need flux and change to grow and innovate, and we accept that desires cannot always be instantaneously fulfilled.

What we do worry about is structural unemployment—the kind that results when the characteristics of workers are not well-suited to the demands of available jobs. A particularly troubling form of structural unemployment is skills mismatch. If, for example, all available job openings are for computer programmers and all applicants are construction workers, we have a problem. Even when plentiful, grossly mismatched openings will not help the unemployed find work. Retraining can take years, if it happens at all; in the meantime, the economy foregoes potential output.

Of course, mismatch can occur along other dimensions as well. Geography is one. A boom of job openings in the Midwest doesn’t do a whole lot for unemployed residents of the Northeast. Schedules can also be a factor. If firms are looking to hire for the night shift, but most applicants insist on day jobs, vacancies will persist. Sometimes, several structural barriers can compound—for example, accountant-rich New Jersey might be looking for teachers, while Tennessee’s adequately stocked school system might need more CPAs (this a hypothetical example).

As rapid technological progress and globalization have roiled the U.S. labor market in recent decades, mitigating structural unemployment has become a top policy concern. But in some cases, it’s hard to distinguish between deep structural mismatches and a third type of joblessness: the cyclical unemployment.

Cyclical unemployment fluctuates with the business cycle. It has to do with the strength of aggregate demand (economist-speak for people buying stuff) rather than the attributes of workers and jobs. When times are good and consumers are spending, businesses add jobs to keep up with demand. During downturns, by contrast, workers who would have been profitable to employ in a strong business climate become liabilities and are laid off.

Classifying the unemployed into these three neat buckets—frictional, structural, and cyclical—is not a simple matter. In many cases, multiple forces can be at play. During a recession, a bit of structural mismatch can be magnified, and frictions can be made rougher still. So rather than attempt a comprehensive categorization, we can more easily gain insights into the composition of unemployment by examining how the relationship between job openings and unemployment has trended over time.

Figure 5 traces job openings as a percentage of unemployment from 2001 to 2014. The higher the openings-to-unemployment (O-U) ratio, the more work availabilities there are for each unemployed worker.

As we might expect, the O-U ratio is very vulnerable during a recession. After starting the decade at an unusually high level near 90 percent, the O-U ratio declined quickly, to the mid-30s, during the early-2000s recession, before rebounding to near 70 percent in 2007. Then, during the Great Recession, the ratio took another huge tumble, dropping below 15 percent in July 2009. Since then, it has ticked back up, slowly at first, but more quickly in recent months. With the current ratio just above 50 percent, we are right back in average territory.

Download

The meaning of this recent rise in the O-U ratio is somewhat uncertain. On one hand, it clearly has a positive relationship with overall economic health; in years where economic growth has been strong historically, the O-U has been high. And, in general, more job openings for each job seeker suggest that the unemployed will have an easier time finding work.

But a rise in the O-U ratio can be indicative of less-sanguine trends as well. For example, if job openings increase rapidly while unemployment stays flat, it could suggest that structural mismatches are also increasing.

Figure 6, which plots the job openings rate and the unemployment rate separately (in this case, with job openings as a percentage of the labor force, to put the two in comparable terms), hones in on what is going on.

Download

A simple comparison of the trends over time for these two rates demonstrates their inverse tendencies: when the job opening rate falls, the unemployment rate rises, and vice versa. At their respective worsts, the openings rate fell 53 percent during the recession, while unemployment rose 117 percent, compared to 2007. Or, to put the two on more equal footing (there’s only so far the jobs opening rate can fall, since it’s already close to the zero lower bound), the openings rate lost about half its value, while unemployment more than doubled.

Since the recession, however, job openings have risen faster than unemployment has declined. While the openings rate has already improved beyond its pre-recession benchmark, the unemployment rate is still a quarter higher than it was in 2007 (and that’s not even factoring in the big drop in labor force participation we have seen in recent years). So while the growth in job openings is an encouraging sign, it should give us pause that unemployment hasn’t recovered as quickly. Perhaps this is evidence of a cyclical problem—in other words, that aggregate demand hasn’t recovered as much as we would like to think. Or, more forebodingly, structural imbalances may be on the rise.

Addressing this important question of demand versus mismatch is an important issue that has attracted the attention of many economists and policymakers in recent years, and we will be revisit it in the next installment in this Working Paper Series. But divining clear answers is difficult, and there remain important areas of disagreement.

For now, it is sufficient to recognize that nearly 5 million job openings means there is considerable room for more people to be employed. And for the remainder of this report, we will seek to answer somewhat less controversial questions. Where are the jobs? And, in particular, how do the distributions of jobs and unemployment vary by industry?

By drilling down to the industry level, we will be able to get some sense of the degree of mismatch in the current labor market, as well as a feel for how well different sectors have recovered from the cyclical downturn. Equally important, we will see which categories of workers have been doing well, and which have been struggling. To round out our picture, we will also look at trends in wages across industries: the quantity of jobs matters, but so does their quality.

Our answers will help inform policy, while also providing a compass for job seekers. When you know where to look for (good) jobs, you are more likely to find one.

Although there are numerous ways we could classify work, categorizing workers by industry is one of the most useful. As the BLS defines it, an industry “consists of a group of establishments primarily engaged in producing or handling the same product or group of products or in rendering the same services.”

In contrast to occupation, which describes the types of tasks a worker performs, or personal characteristics, such as age or race, the category of industry groups workers according to what their employers do. As a result, an industry label conveys not only information about workers, but also about the firms that employ them. That can help to illuminate the intersection between job seekers and job availabilities.

EXPLAINER: More about Industries

Table 1 provides a summary of the way industries are classified in JOLTS. These are the categories we will use in our discussion.

Download

What industries are best to be working in right now? Or, to be more precise, how do the places where people are looking for jobs (technically, the industry in which they are classified based on their current or most recent job) correspond with where the job openings are?

Figure 7 provides our first clue. It compares the job openings rate and the unemployment rate within and between industries, using “roughly seasonally adjusted” averages from August to October 2014 (see the “More about Industries” Explainer). To make the job openings and unemployment rates comparable, it expresses job openings as a percent of the labor force (rather than as a percent of employment, as it normally is). Thus, the ratio between the dark bar (unemployment rate) and the light bar (job openings) indicates the extent to which job seekers outnumber (or fall short of) job availabilities.

Download

The figure is arranged, from top to bottom, in order of decreasing unemployment-to-openings ratios. At the top is construction, whose ratio of 5.8 indicates there are 5.8 available workers for each construction job opening. Expressed another way, out of every hundred construction workers, 8.7 are unemployed (i.e., the unemployment rate). But for those same hundred workers, there are only 1.5 job openings, which means that even if every opening were filled, 7.2 out of every 100 construction workers would still be left out. Put simply, now is not a good time to be in construction.

At the other extreme, the financial services industry is doing quite well. It has no gap, with almost exactly the same amount of unemployed as openings. Out-of-work bankers shouldn’t be sending out resumes for long.

For the private sector overall, the unemployment-to-openings ratio is 1.6: an unemployment rate of 5.8 percent, offset by a job openings rate of 3.6 percent. The industries where work is toughest to find—where latent labor supply most exceeds unfulfilled labor demand—are exclusively low-skill or blue-collar industries: manufacturing, mining, transportation, wholesale and retail trade, leisure and hospitality, and other services.

In general, these industries tend to have unemployment rates near the private sector average, but with job openings rates that are especially low. The two exceptions are construction and leisure and hospitality, which have, by far, the highest unemployment rates. In the case of the latter, however, high unemployment is ameliorated by higher than average job openings.

In contrast, high-skill industries—finance, education and health services, professional services, and, to a lesser extent, information—have unemployment levels that are almost offset by vacancies. It is comparatively far less difficult for workers in these industries to find jobs.

Government—a typically white-collar industry that nevertheless can offer good jobs to the less skilled—also fares better than the private sector, with a ratio of 1.3, meaning that for every four unemployed government workers, there are about three available government jobs. What sets government apart from the private sector is its exceedingly low unemployment rate, just 2.9 percent.

Equally important is who government employs. While the public sector, like the private one, has placed a greater premium on education and intellectual skills in recent decades, government remains an important source of good jobs for less-educated workers as well. For example, occupations like police, sanitation, day care, and the postal service often do not require college degrees—but they do provide valuable public services. At a time when stable, well-paying jobs for the less-skilled are drying up, government can become increasingly important refuge in the market for blue-collar jobs.

EXPLAINER: Three Caveats About Job Openings

To make the picture a bit clearer, we can plot the industry-level unemployment-to-openings ratios themselves, rather than their components. We do this in Figure 8, with one modification: rather than plot the unemployment-to-openings ratio, we plot its inverse; that is, the same openings-to-unemployment ratio (O-U) we graphed for the economy as a whole in Figure 5. This eases the interpretation. The nearer the ratio is to one (or 100 percent), the closer an industry’s number of job openings to its number of unemployed. Of course, the ratio can go above one as well. The higher it is, the easier it should be for workers to find job.

This new perspective on the balance between unemployment and openings makes plain just how much better high-skill industries are doing than low-skill ones. Finance, education and health services, and professional and business services are the clear leaders, while manufacturing, and most obviously, construction, are the laggards.

Download

At this point, we have a pretty good sense of which industries have strong labor markets and which ones are weak. But how about the relative importance of each of these industries? A high (or low) job openings rate for mining, for instance, doesn’t mean very much, since it’s a tiny industry of just 925,000 employees to begin with. By contrast, a spike in openings in education and health services, where nearly 22 million Americans are employed, is a big deal.

Figure 9 provides a perspective on relative importance, displaying each industry’s share of job openings and unemployment. Just five industries—professional services; education and health services; leisure and hospitality; wholesale and retail trade; and government—account for 75 percent of all job openings. These industries are fairly diverse, with business, education, and health among the most skilled fields, while leisure and retail are among the least skilled.

However, those five industries cover only 64 percent of the unemployed, so there’s a pretty large gap—11 percentage points—between openings and unemployed. The entire difference is attributable to professional services and education/health—the two large, high-skill fields—which together account for 38 percent of openings but only 26 percent of unemployment.

Download

Conversely, construction and manufacturing represent nearly a fifth of the unemployed, but account for only 8.7 percent of job openings. And the two largest low-skill industries—leisure and trade—also bear a disproportionate share of unemployment.

What emerges is a clear pattern: the unemployed are concentrated in low-skill sectors, while the jobs are available largely in high-skill ones. The economy wants accountants and teachers, but the labor force is supplying factory workers and carpenters.

As we have emphasized, job openings are an imperfect indicator of labor demand, particularly if the pace from posting to hiring has slowed down (or otherwise changes over time). Fortunately, other JOLTS indicators can broaden our understanding of demand for workers.

One such useful measure is the hires-to-layoffs (H-L) ratio, which we discussed above in the context of the overall labor market. By bringing together employers’ voluntary expansions (hires) and voluntary contractions (layoffs), the H-L ratio gives us a feel for the speed of employment growth. The higher the ratio, the faster employment is growing. Figure 10 plots the H-L ratio for twelve-month period beginning in November 2013 and ending in October 2014 (the most recently available data at the time this report was written).

Download

Overall, the private sector’s H-L ratio is 2.9: about three people are hired for every one laid off. Above-average industries are clustered closely together, with four industries between government, at 3.5, and mining, at 3.2. As before, finance, education/health, trade, and leisure have strong demand for new employees.

Below-average performers are also clustered, with transportation, manufacturing, information, and other services all between 2.7 and 2.9. Once again, construction is the large adverse outlier, at 1.8—a relative hiring pace less than two-thirds of the average industry. Somewhat surprisingly, professional services has the second lowest ratio, at 2.4. This industry, which consists of lawyers, accountants, managers, and the like, has the highest hiring rate of any industry, at 5.3 percent; but it also has the second-highest layoffs rate, at 2.2 (trailing only construction, at 2.7).

In sum, the H-L ratio gives us a sense of which industries are doing the most hiring, relative to their propensities to fire. But clearly, the H-L ratio isn’t perfect either. The reason is that not all industries have identical labor supply conditions. In particular, employees in different industries tend to quit at different rates. Leaving voluntary exits (quits) out of the equation will have the effect of inflating the ratio in those industries where quit rates are high (because employers are forced to replace workers more frequently). But we don’t necessarily want to directly include quits in our demand measure either, because quits are about labor supply rather than labor demand, and it is an indicator of the latter that we are looking for.

One solution is to compute an adjusted H-L ratio, as shown in Figure 11. The adjusted H-L ratio is calculated by adjusting the hiring rate to reflect the extent to which an industry’s quits rate differs from the economy-wide (nonfarm) average. By subtracting out the portion of hiring attributable to above (or below) average quits, the adjusted H-L ratio puts industries on more even terms.

EXPLAINER: Adjusted Hires-to-Layoffs Ratio

Making this adjustment changes the picture somewhat. Government becomes the big outlier, with an adjusted H-L ratio of 6.7—two-and-a-half times that of the private sector.

The reason is that government has a quits rate that is less than a third of the private sector (0.6 percent versus 2.1 percent). It also has a (stereotypically and predictably) low layoffs rate (0.4 percent, compared to the private sector average of 1.4 percent). Thus, its low (but comparatively not as low) hiring rate of 1.4 percent (versus 3.9 percent) is more than enough to make government the clear leader in relative labor demand. When we consider that few government employees leave their jobs, it’s as if the government is hiring at a faster pace (relatively speaking) than it really is. Said differently, if government workers quit at the same rate as private sector workers, its H-L ratio would be 6.7.

Bear in mind, though, that this doesn’t mean government is hiring a lot of people—it’s not. What it does mean is that, relative to the number of people government lays off, it hires a lot. This bodes well if you have a government job, though not necessarily if you are trying to get one.

Download

The quits adjustment also makes a big deal for finance, boosting its H-L ratio to 4.3, making it the leader among private sector industries. Manufacturing’s low quits rate also gives it a big boost, from 2.7 to 3.8 hires per layoff, ranking it third with education/health (which also moved up 0.4). As with government, a high adjusted H-L ratio means job security for employed workers, even as it suggests the unemployed may struggle to find jobs in these industries.

Leisure and hospitality, with its relatively high quits rate, takes the biggest hit, falling from 3.3 to 2.3 hires per quit. Wholesale and retail loses some ground, falling from 3.4 to 3.0. The rest of the picture is largely unchanged.

By and large, the industries with low adjusted H-L ratios have above-average quits and layoffs rates. They also tend to have above average hiring rates, but this only partially offsets their propensity to shed workers. High separation rates suggest these are more volatile industries.

The volatility could be on the firm side: business performance may be more susceptible to cyclical or seasonal fluctuations. But it could also be on the employee side: workers in these industries may be more peripatetic, less reliable, or more motivated to change jobs frequently (perhaps because their jobs are low-paying, highly stressful, or otherwise undesirable). Whatever the reason, high turnover means less stability, both for firms and workers. It can also impose costs on both parties, in the form of recruitment, training, and job search.

On the other hand, industries with high adjusted H-L ratios have quite low quits and layoffs rates. Their hiring rates also tend to be below average. Government and education are good examples. These industries are stable to a stereotypical extent, offering strong job security. The downside is that job seekers in these industries can have difficulty finding work, because these industries tend to hire at a slow pace.

Together, these patterns produce an interesting story. When it comes to labor demand, the tail tends to lead the dog. Exit rates drive entry rates. In good times, firms hire at rates sufficient to replace workers they fire or who quit, plus a little extra, because individual industries, like the labor force as a whole, tend to be growing over time. Industries that are expanding have the most “excess” hiring (hiring above the replacement rate), and those that are contracting have the least.

Now that we have a sense of labor market movements, we can look at how these worker flows have translated into real job growth. The next collection of charts draws from the Current Employment Statistics (CES) program, and illustrates the industry-level jobs evolution since 2007. (Also known as the Establishment survey, the CES is what BLS uses to measure monthly trends in employment and earnings.) These measurements of the stock of jobs in different sectors yield similar insights to the flow indicators examined so far.

In looking at this data, it is useful to group industries by the average levels of education of their workforces. Among the population 25 years of age and older for whom the BLS has industry data (about 123 million people, which excludes agriculture and self-employment), 38 percent are college graduates and 62 percent are not. We will define industries that have a greater than average share of college graduates (above 38 percent) as white-collar, or high-skill industries. Similarly, those sectors with a greater than average proportion of workers with less than a college degree are defined as blue-collar, or low-skill. (This is obviously a vast simplification of the distributions of education and skills by industry, but simplifications, properly understood, can aid our comprehension of complicated phenomena.) To facilitate visual interpretation of our results, these industries are color-coded as white and blue, respectively. Figure 12 shows the education breakdown by industry.

Download

With this industry grouping in mind, let's examine job growth. As depicted in Figure 13, overall private employment is up somewhat since 2007, by 0.7 percent. In nominal terms, there are more private sector jobs today than there were in 2007—about 817,000 more, on average each month, between November 2013 and October 2014 . But in real terms—factoring in the growth of the labor force, which has been 1.7 percent—jobs have yet to regain pre-recession levels. Had the private sector expanded as quickly as the labor force, they would be an additional 1.1 million working Americans today. And even this understates matters, because, with the aging of the population and the slow recovery, labor force participation has dropped, from 66.0 percent in 2007 to 62.9 percent today.

Download

Again we see that the industries that have been hardest hit by the recession are, predictably, the goods-producing sectors involving manual labor: construction and manufacturing. The former has seen its payrolls shrink by 21.4 percent, while the latter has contracted by 12.7 percent. Collectively, that translates to 3.4 million jobs.

We saw earlier from the JOLTS data that the information sector has been perhaps the weakest of the high-skill industries, with fewer job openings and less hiring than its peers. Looking at jobs growth, the picture becomes even starker: the information sector has lost 11.8 percent of its jobs since 2007, the third-highest of any industry.

Among white-collar industries, finance has also contracted somewhat. Given that the recession was, in many respects, a creature of the financial industry, that should not be unexpected. Rounding out the (nominal) industry losers are the blue-collar retail and wholesale sectors; recall, again, that things look worse if we factor in population growth.

The best-performing industries in terms of employment growth have been mining (+24.8 percent) and leisure (+8.6 percent) among blue-collar fields, and education/health (+15.0 percent) and professional services (+6.5 percent) among white-collar sectors. These industries have rated well on multiple dimensions; job growth is the product of sufficient openings, high hires-to-layoffs ratios, and moderate turnover—though the particular mix of factors is somewhat idiosyncratic by industry. Part of the complexity is that job growth entails both labor supply and labor demand: not only must jobs be available, but people must be qualified for and willing to take them. (Bear in mind, too, that these proximate explanations of job growth are themselves the product of more fundamental economic forces, such as consumer demand, investment, and technological change.)

As with job openings rates, however, job growth alone doesn’t give us a sense of how important industries are in the grand scheme of our economy. For that, we need to look at each industry’s share of overall jobs. As Figure 14 shows, the two-fastest growing private sector white-collar industries—education/health and professional services—are also the two largest, accounting for about three-in-ten jobs overall. The strength of these sectors is one of the reasons high-skill workers have fared well in recent years.

Download

But notice the largest sector of all: government, which employs nearly one out of every six workers. Although classified as white-collar in the figure, government, like other white-collar sectors, has a fairly even split betiween college graduates and those with less than a college degree. This, combined with it's size, makes it a particularly important source of middle-skill jobs.

What is striking, if we return for a moment to Figure 13, is that government employment has actually shrunk 1.5 percent since 2007. This might not sound like a big deal—but it is in the context of government’s outsized importance. Translated into employment levels, a decline of this magnitude in government employment is worth 339,000 jobs. Had government employment kept pace with the growth of the labor force, 714,000 more Americans would be working for the government today. (The role of government employment in the labor market recovery will be the subject of a future chapter of the Working Paper Series. You can sign up here if you’d like to receive updates when new installments are released.)

Turning to blue-collar industries, we can see more clearly why times have been tough for low-skill workers. Four of the five largest blue-collar sectors (retail, wholesale, manufacturing, and construction), accounting for 28 percent of all jobs and 60 percent of blue-collar jobs, have lost workers since 2007. Of the major blue-collar industries, only leisure and hospitality has grown.

Industry size—job quantity—is not the only thing that matters. The quality of the job—in particular, how well its workers are compensated—is also extremely important. And by this measure, as Figure 15 shows, leisure and hospitality—the fastest growing major blue-skill industry—is the worst sector. The average leisure and hospitality worker makes just $18,900 a year (gross, before taxes). This is not even enough to keep a family of three above the poverty level ($19,790 in 2014). Similarly, retail, the largest blue-skill sector, is second-worst in terms of pay, with average annual earnings of $27,700.

The pattern extends to white-collar jobs as well. The largest private white-collar industry is education and health services, which is also the white-collar sector with the lowest average earnings—$42,300. That is less than the private sector average of $44,100, and even less than many blue-collar industries. (The CES doesn’t provide earnings data for government workers.)

Download

On the other hand, well-paying, less-skilled industries—wholesale, construction, and manufacturing, each with average earnings of more than $50,000—are also the industries that have been shrinking. Likewise, the best paying (on average) white-collar fields—information ($65,500) and finance ($59,900)—have also been contracting. The exceptions to the pattern—utilities ($78,700) and mining ($72,300)—are tiny industries with a small impact on the overall labor market.

EXPLAINER: Average Earnings

The pattern is similar if we take a slightly different perspective and consider how earnings have grown since 2007. Simply put, the pace has been tepid. For the private sector as a whole, as Figure 16 shows, average weekly earnings increased 3.1 percent from 2007 to the November 2013-October 2014 period after taking inflation into account. On an annual basis, that is a raise of about half a percent a year.

Download

Setting aside mining as an outlier (+11.9 percent), the fastest earnings growth has been in white-collar industries. Earnings in information has increased 9.4 percent, in finance by 8.1 percent, and professional services by 6.9 percent. It is worth re-emphasizing, however, that the largest wage growth has occurred in those white-collar fields that are high-paying, but comparatively small and/or contracting.

Similarly, the most important low-skill industries, leisure (-0.9 percent) and retail (-2.1 percent), have actually seen earnings shrink. Sectors such as construction (+5.2 percent) and wholesale trade (+6.1 percent) have fared better—but it is in these areas where finding jobs has become especially challenging.

Indeed, if we turn to Figure 17, we see the general pattern conforms to our impression: earnings growth has been faster in smaller industries. Workers who have enjoyed significant earnings growth have been the fortunate few. The points in the scatterplot cluster fairly neatly around a straight line; as it turns out, an industry’s employment share in 2007 can account for 39 percent of its earnings growth in the subsequent seven years. Of course, this is not to say size of an industry is causing earnings growth; just that the two are associated in such a way. In other words, the relationship is descriptive (telling us what has happened), rather than explanatory (telling us why it happened).

Download

These patterns also suggest a potentially more striking trend—namely, that higher-paying industries have seen faster relative earnings growth. The notion of widening inequality is one that has seeped into the popular consciousness in recent years; might wage growth by industry add a new bit of evidence to the debate?

We can examine the idea of industry-based inequality more formally by plotting a simple linear regression of earnings growth on average weekly earnings in 2007 (all in 2014 dollars), as in Figure 18.

Download

The picture confirms our second intuition, and the fit is remarkably close to a straight line. Although the relationship is, again, just a correlation, it is suggestive of feedback from earnings to earnings growth. The R² of 0.57 means that earnings level in 2007 “explains” (again, it’s just an association) 57 percent of real wage growth during the next seven years. (If we exclude utilities—the clear outlier with earnings of $78,700, and which accounts for only 0.7 percent of all jobs—the R² increases to 0.74.)

The rich are getting richer. For every $10,000 in additional average weekly earnings an industry had in 2007, its average earnings grew by an extra 2 percentage points during the six-year period. For example, the average employee working in retail, a low-paying industry, earned $28,300 for a full year of work in 2007; after adjusting for inflation, the same employee actually made less, just $27,700, last year. By contrast, the average information worker earned $59,900 in 2007, among the highest of any industry. During the next six years, the advantage of information workers over poorly-paid workers only grew, with average earnings increasing 9.4 percent, or $5,600 in real terms. The same is true of other white-collar sectors, like finance (a gain of $4,500) and professional services (a gain of $3,600). Like so many other labor market indicators, this link between earnings and earnings growth reflects rising inequality.

Creating meaningful, well-paying jobs for less-skilled workers has become one of the most significant and perplexing economic policy challenges confronting the United States. Education is clearly one important component of an effective response: making less-skilled workers more skilled is one essential path to promoting upward mobility. But even if all Americans were college educated, inequalities would remain; not everyone can be an above-average earner, and not everyone can have the corner office.

As important as improving absolute living standards is, some things that matter will always be measured in relative terms. The real challenge is to strike the right balance between providing the incentives the economy needs to grow and innovate, and ensuring that even the least among us have the chance to live and work with dignity, mobility, and contentment.

The good news is that now is the best time in quite some time to be a job seeker in America, provided you have the skills employers desire. What we make of this fortuitous moment—and, in particular, who gets to share in the opportunities it provides—will depend upon our policy choices.

We know where the jobs are. But we must decide where our shared values lie.

Written by

Mike Cassidy is a policy associate at The Century Foundation. He is a strong believer in the power of scientific analysis, and his research focuses on using economics to understand human behavior, especially at it relates to poverty, inequality, performance, and progress. A proud alum of Princeton’s Woodrow Wilson School, Mike holds a Master in Public Affairs with a concentration in economics and public policy. From 2007 to 2012, he worked at the New York City Office of Management and Budget, where he oversaw the city’s social service and criminal justice agencies.  In his spare time, Mike is a semi-professional distance runner and competed in the 2012 U.S. Olympic Marathon Trials.

Editors

Jason Renker and Greg Anrig

Project Manager, Graphic Designer and Developer

Abigail Grimshaw