February 26, 2014
The Pattern of Job Creation and Destruction by Firm Age and Size
A recent Wall Street Journal blog post caught our attention. In particular, the following claim:
It’s not size that matters—at least when it comes to job creation. The age of the company is a bigger factor.
The following chart shows the average job-creation rate of expanding firms and the average job-destruction rates of shrinking firms from 1987 to 2011, broken out by various age and size categories:
In the chart, the colors represent age categories, and the sizes of the dot represent size categories. So, for example, the biggest blue dot in the far northeast quadrant shows the average rate of job creation and destruction for firms that are very young and very large. The tiny blue dot in the far east region of the chart represents the average rate of job creation and destruction for firms that are very young and very small. If an age-size dot is above the 45-degree line, then average net job creation of that firm size-age combination is positive—that is, more jobs are created than destroyed at those firms. (Note that the chart excludes firms less than one year old because, by definition in the data, they can have only job creation.)
The chart shows two things. First, the rate of job creation and destruction tends to decline with firm age. Younger firms of all sizes tend to have higher job-creation (and job-destruction) rates than their older counterparts. That is, the blue dots tend to lie above the green dots, and the green dots tend to be above the orange dots.
The second feature is that the rate of job creation at larger firms of all ages tends to exceed the rate of job destruction, whereas small firms tend to destroy more jobs than they create, on net. That is, the larger dots tend to lie above the 45-degree line, but the smaller dots are below the 45-degree line.
As pointed out in the WSJ blog post and by others (see, for example, work by the Kauffman Foundation here and here), once you control for firm size, firm age is the more important factor when measuring the rate of job creation. However, young firms are more dynamic in general, with rapid net growth balanced against a very high failure rate. (See this paper by John Haltiwanger for more on this up-or-out dynamic.) Apart from new firms, it seems that the combination of youth (between one and ten years old) and size (more than 250 employees) has tended to yield the highest rate of net job creation.
By John Robertson, a vice president and senior economist in the Atlanta Fed’s research department, and
Ellyn Terry, a senior economic analyst in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference The Pattern of Job Creation and Destruction by Firm Age and Size:
February 06, 2014
A Prime-Aged Look at the Employment-to-Population Ratio
Trying to interpret changes in labor utilization measures such as the employment-to-population ratio is complicated by the fact that they do not refer to the same set of people over time. The age composition of the population is changing, and behavior can vary across and within age cohorts.
This issue is illustrated in a recent New York Fed study of the employment-to-population ratio by Samuel Kapon and Joseph Tracy. This ratio nosedived during the recent recession by about 4 percentage points and has barely budged since.
This measure of labor utilization is the clear laggard on any labor market recovery dashboard. But the authors show that it is not so clear that the employment-to-population ratio is really so far from where it should be, once you control for the fact the employment rates tend to be lower for younger and older people and that the age composition within the population has shifted over time. This idea is similar to the one used to estimate the trend labor force participation rate in this Chicago Fed study by Daniel Aaronson, Jonathan Davis, and Luojia Hu. The issue of controlling for dominant demographic trends is one of the reasons we at the Atlanta Fed decided not to feature either the overall employment-to-population ratio or the overall labor force participation rate in our Labor Market Spider Chart.
A simple, and admittedly crude, alternative to computing the demographically adjusted employment-to-population ratio trend is to look at a segment of the population that is on a relatively flat part of the employment (or participation) rate curve. A common standard for this is the so-called prime-aged population (people aged 25 to 54). These individuals are less likely to be making retirement decisions than older individuals and are less likely to be making schooling decisions than younger people. Of course, this approach doesn't control for within-cohort factors like educational differences.
So what do we find? The prime-aged employment-to-population ratio declined almost 5 percentage points between the end of 2007 and 2009 (versus 4 percentage points overall) and since then has recovered about 25 percent of that decline. Using the end of 2007 as reference, the Kapon and Tracy trend estimate has declined about 1.7 percentage points, which implies the overall employment-to-population ratio, by not continuing to decline, has improved by about 40 percent.
Then what does the analysis say about labor utilization in the wake of the recession? Once demographic factors are controlled for, both aforementioned measures indicate that labor-resource utilization has improved relative to trend. In fact, as Kapon and Tracy note, the relative improvement would be even greater if you believed that employment was above trend before the recession.
By John Robertson, a vice president and senior economist in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference A Prime-Aged Look at the Employment-to-Population Ratio:
January 17, 2014
What Accounts for the Decrease in the Labor Force Participation Rate?
Despite the addition of only 74,000 jobs to the economy in December, the unemployment rate dropped significantly—from 7 percent to 6.7 percent. The decline came mostly from a decrease in the labor force.
Since the recession began, the labor force participation rate (LFPR) has dropped from 66 percent to 63 percent. Many people have left the labor force because they are discouraged from applying (U.S. Bureau of Labor Statistics data indicate that a little under 1 million people fall into this category). But the primary drivers appear to be an increase in the number of people who are either retired, disabled/ill, or in school.
Certainly, the aging of the population accounts for much of the increase in the retired and disabled/ill categories. Still, there has been a lot of movement over the past few years in the reasons people cite for not participating in the labor force within age groups. Knowing the reasons why people have left (or delayed entering) the labor force can help us understand how much of the decline will likely halt once the economy picks back up and how much is permanent. (For more on this topic, see here, here, and here.)
The chart below shows the distribution of reasons in the fourth quarter of 2013. (Of the people not in the labor force, 1.6 percent indicate they want a job and give a reason for not being in the labor force. They are categorized here as "want a job" only.) Young people are not in the labor force mostly because they are in school. Individuals 25 to 50 years old who are not in the labor force are mostly taking care of their family or house. After age 50, disability or illness becomes the primary reason people do not want to work—until around age 60, when retirement begins to dominate.
How has this distribution changed over the past seven years? For simplicity, I've grouped people by age to show changes over time in the reasons people give for not being in the labor force. However, you can also see an interactive version of the same data without age buckets—and download the data—here.
Of the 12.6 million increase in individuals not in the labor force, about 2.3 million come from people ages 16 to 24, and of that subset, about 1.9 million can be attributed to an increase in school attendance (see the chart below). In particular, young people aged 19 to 24 are more likely to be in school now than before the recession. Among college-age people, those absent from the labor force because they are in school rose from 57 percent to 60 percent. Among people of high school age, the share not in the labor force because they are in school rose from 87 percent to 88 percent.
The number of middle-aged workers not in the labor force rose by 1.8 million (or 11 percent), with four main factors driving the increase.* "Wants a Job" increased 546,000 (34 percent). The "In School" category increased 438,000 (a 38 percent rise). "Disability/Illness" rose 393,000 (an 8 percent rise), and 302,000 more people said they were retired (a 43 percent rise; see the chart below).
Among individuals aged 51 to 60, those not in the labor force increased by 1.6 million (or 16 percent). This increase came almost entirely from the number of people who are disabled or ill, which rose by 1.3 million (a 33 percent increase). Interestingly, the number of retired individuals actually fell by 305,000 between the fourth quarter of 2007 and the fourth quarter of 2010. Since then, the number of retired people within this age group has risen 183,000 but remains 122,000 lower than fourth-quarter 2007 levels. So it seems more people in this age group were delaying retirement instead of leaving early (see the chart below).
About 6.8 million of the 12.6 million increase in those not in the labor force came from the 61-and-over category. An additional 5.3 million (a 17 percent increase) are retired, and 1 million more (a 34 percent increase) are not in the labor force because they are disabled or ill. The other categories were little changed (see the chart below).
In total, the number of people not in the labor force rose by 12.6 million (16 percent) from the fourth quarter of 2007 to the fourth quarter of 2013. About 5.5 million more people (a 16 percent increase) are retired, 2.9 million (a 23 percent increase) are disabled or ill, and 2.5 million (a 19 percent increase) are in school. An additional 161,000 are taking care of their family or house, and an additional 99,000 are not in the labor force for other reasons. The fraction who say they want a job has risen the most (32 percent) but has contributed only 11 percent to the total change. The chart below shows the overall contributions by reason to the changes in labor force participation for all age groups since the onset of the recession.
What further changes can we anticipate? It's hard to say, as many moving parts are at play. Most people currently in school will be approaching the labor market upon graduation. But increased college and graduate school enrollment could augur a permanent shift in the portion of the population who are in school instead of the labor force. We can also expect continued downward pressure on the LFPR from retiring baby boomers as well as boomers who exit the labor force because of disability or illness.
Last, the portion of people who want a job has increased the most since the recession began, and is currently 1.4 million above its prerecession level. People in this category tend to have greater labor force attachment, making them more likely to shift into the labor force. In fact, the number of people in this category has already started to decrease—and is down 709,000 from the fourth quarter 2012.
My Atlanta Fed colleagues Julie Hotchkiss and Fernando Rios-Avila in their 2013 paper "Identifying Factors behind the Decline in the U.S. Labor Force Participation Rate," looked at a range of LFPR projections for 2015–17 based on different labor market assumptions. Depending on the future strength of the U.S. labor market, the projections are highly varying—ranging between a decline of 2.4 percentage points and an increase of 2 percentage points from the 2010–12 average of 64.1 percent. So far, more factors are pulling down the LFPR than pushing it up; the latest reading for December 2013 is already 1.3 percentage points below the 2010–12 average. At that pace, the Hotchkiss et al. lower-bound estimate will be reached before the end of 2014, unless the dynamics change as the economy further improves.
By Ellyn Terry, an economic policy analysis specialist in the research department of the Atlanta Fed
* I've chosen to break the "middle-age" grouping at age 50 instead of 54 because the probability of retiring has changed in different ways over the past few years for the 25- to 50-year-old group and the 51- to 60-year-old group. See the chart mentioned earlier for more detail.
TrackBack URL for this entry:
Listed below are links to blogs that reference What Accounts for the Decrease in the Labor Force Participation Rate?:
December 27, 2013
Is the Labor Force Participation Rate about to Fall Again?
A few posts back my Atlanta Fed colleagues Tim Dunne and Ellie Terry offered up our latest contribution to the ongoing head-scratching over the rather spectacular decline in U.S. labor force participation (LFP) since the onset of the Great Recession in December 2007. “Rather spectacular” in this case means a fall in the participation rate from 66 percent (of the working age population either working or actively seeking work) to the 63 percent level reported for November. In people terms, that 3 percentage point decline represents a reduction of about 1.4 million participants in the U.S. labor market.
Like many other analysts, Dunne and Terry find that the drop in labor force participation appears to come from a combination of demographic factors—mainly the aging of the population—and other causes not specifically identified but generally interpreted to be associated with the weak economy in one way or another.
Two developing stories suggest the LFP may not be leaving the spotlight just yet. The first is this one, from USA Today:
Some 1.3 million Americans are set to lose their unemployment benefits Saturday...
Federal emergency benefits will end when funds run out for a program created during the recession to supplement the benefits that states provide. The cutoff will initially affect 1.3 million people, but 1.9 million more will lose benefits by mid-2014 when their 26 weeks of state paychecks run out, according to the National Employment Law Project.
What will those 1.3 million Americans do when their benefits run dry? According to a recent study by Princeton University’s Henry Farber and the San Francisco Fed’s Robert Valletta—also presented at a conference hosted here at the Atlanta Fed in October—on balance, the affected individuals are likely to leave the labor force:
We examined the impact of the unprecedented extensions of UI [unemployment insurance] benefits in the United States over the past few years on unemployment dynamics and duration and compared their effects with the extension of UI benefits in the milder recession of the early 2000s. We found small but statistically significant reductions in unemployment exits and small increases in unemployment durations arising from both sets of UI extensions. The magnitude of these overall effects is similar across the two episodes...
We find that the effect on exit from unemployment occurs primarily through a reduction in labor force exits rather than through exit to employment (job finding). This is important because it implies that extended benefits do not delay the time to re-employment substantially and so do not have first-order efficiency effects. The major effect of extended benefits is redistributive, providing income to job losers who would have exited the labor force otherwise (consistent with Card et al. 2007). [link mine]
In other words, if a significant decline in unemployment benefits comes to pass, we may well see another bump downward in the labor force participation rate. Although a decline in LFP associated with the expiration of extended UI benefits would fall in Dunne and Terry’s nondemographic category, the Farber and Valletta results suggest that we should interpret any such decline as structural. And structural in this case means not directly amenable to correction by policies aimed at stimulating spending.
The other important piece of recent news, however, is this one, which you probably heard about:
According to the Bureau of Economic Analysis, real gross domestic product—output produced in the United States—actually grew at a rate of 4.1% in the third quarter, up from BEA’s previous estimate of a 3.6% growth rate. The final results are also a gain over the second quarter’s 2.5% GDP growth.
Furthermore, as noted at Calculated Risk, the good news doesn’t stop there:
A little Christmas cheer...
Macroeconomic Advisers...[raised] its estimate for fourth-quarter growth. It now forecasts gross domestic product to expand at an annualized rate of 2.6% in the final three months of the year, up three-tenths of a percentage point from an earlier estimate.
And Goldman Sachs has increased their Q4 GDP tracking to 2.4% annualized growth.
That all adds up to pretty decent growth in the second half of the year. If it persists, and the long-awaited acceleration in the economic expansion finally arrives, better labor market conditions should follow. And if the six-year fall in LFP has in large measure been driven by weak economic conditions, we should at least see a pause in participation declines as economic activity picks up. Actually, we should probably see an outright increase.
The next several quarters, then, may well provide some clarity as to the persistent question of whether or not the large recent exodus of Americans from the labor force has been the result of a lackluster economy. In this period, we may get some clarity as to whether efforts to stem that exodus were justified by a correct diagnosis of the underlying cause.
By Dave Altig, executive vice president and research director at the Atlanta Fed
TrackBack URL for this entry:
Listed below are links to blogs that reference Is the Labor Force Participation Rate about to Fall Again?:
December 19, 2013
Labor Force Participation Rates Revisited
In an earlier macroblog post, our colleague Julie Hotchkiss examined the decline in labor force participation from the onset of the Great Recession into early 2012, concluding that cyclical factors likely accounted for most of the drop. In this post, we examine how labor force participation has changed since the start of 2012 (and admittedly, we’re much less ambitious in our analysis than Julie). Motivating our analysis, in part, is the observation that much of the recent decline in the labor force participation rate (LFPR) is related to rising retirements (see the November 19 Research Rap by Shigeru Fujita). This is not surprising, as the percentage of individuals aged 65 and older in the population has been increasing sharply over the last half decade. That said, our approach indicates that the LFPR of prime-age workers (ages 25–54) continues to fall, and this is an important source of the overall decline in LFPR in the recent data. Such declines in LFPR in these age categories should be less related to retirement decisions, keeping on the table the possibility that a weak overall labor market remains a key drag on labor force participation.
A straightforward decomposition illustrates that the decline in LFPR among prime-age workers is a major contributor to the overall decline in LFPR. To see this, we separate the change in LFPR into three components: one that measures the change due to shifts in the LFPR within age groups—the within effect; one that measures changes due to population shifts across age groups—the between effect; and one that allows for correlation across the two effects—a covariance term. It works out the covariance term is always very close to zero, so we will omit discussion of that term here. The analysis breaks the data down into five age groups: 16–24, 25–34, 35–44, 45–54, and 55+.
The chart presents the decomposition from Q1 2012 to Q3 2013. Over this period, the overall LFPR declined by half a percentage point, from 63.8 percent to 63.3 percent. The blue areas represent the change due to within-age-group effects, and the green areas represent the change due to between-age-group effects. The sum of the bars is equal to the overall change in labor force participation.
Three key results emerge. First, increases in labor force participation for the youngest age group boosted overall labor force participation by 0.075 percentage points. Second, the growing population share of the 55+ age group reduced LFPRs over the period by 0.21 percentage points, accounting for roughly 40 percent of the overall decline. Third, labor force participation for prime-age workers continued to fall. The combined within effect for the prime-age individuals (25–34, 35–44, and 45–54) reduced the participation rate by 0.28 percentage points—or a little over half of the overall decline in labor force participation. Additional declines in labor force participation were associated with the reduction in population shares of prime age workers.
From an accounting standpoint, the analysis shows that the fall in the LFPR for prime-age workers is a main contributing factor to the recent decline in labor force participation. Indeed, the LFPR of prime-age workers fell from 81.6 to 81.0 from Q1 2012 to Q3 2013, with similar declines for both men and women. Given that prime-age workers make up more than half of the population, it is not surprising that the drop in the LFPR for these age groups accounts for a substantial fraction of the overall decline.
To put this in perspective, we present the same decomposition from Q1 2010 to Q4 2011, where the decline in the LFPR is 0.8 percentage point. While the magnitude of the overall change is different, the decomposition results are quite similar. The decline in participation rates for prime-age workers accounts for a little over 60 percent of the overall decline, with a substantial drag from the rise in the share of older workers (accounting for a third of the drop). In short, the changes in participation due to within and between effects over the first two years look quite similar to that of the second two years of the labor market recovery.
A corollary to this analysis is that these sources of decline in labor force participation have allowed the unemployment rate to decline more sharply than expected, given the moderate employment growth observed. We will not take a stand on whether these are “wrong” or “right” reasons for unemployment rate declines. Rather, we note that the patterns observed early in the recovery are still in place (more or less) in the recent data.
By Timothy Dunne, a research economist and policy adviser,
and Ellie Terry, an economic policy analysis specialist, both in the research department of the Atlanta Fed
TrackBack URL for this entry:
Listed below are links to blogs that reference Labor Force Participation Rates Revisited:
November 14, 2013
Atlanta Fed's Jobs Calculator Drills Down to the States
In March 2012, the Federal Reserve Bank of Atlanta launched its Jobs Calculator, an application that illustrates the relationship between the unemployment rate, growth in payroll employment, the labor force participation rate, and a few other variables to boot. Most notably, it tells us how many jobs need to be created to achieve a specific unemployment rate within a given period of time. This tool has turned out to be a useful one for anchoring discussions about national employment growth and unemployment among policy makers and the media.
However, the national employment situation masks significant differences in state labor markets. For example, at the trough of the business cycle (June 2009), the national unemployment rate was 9.5 percent, but it ranged from 4.2 percent in North Dakota to 15.2 percent in Michigan. State policy makers, in managing the dynamics of their own employment situation, need to know the data on a state level.
We are pleased to announce that the Atlanta Fed recently unveiled the state-level Jobs Calculator. The same tool that has been used for national discussions is now available for state-level analyses (see the figure below).
Not only does this state tab allow a quick overview of the historical employment growth in each state (see, for example, Alabama's historical employment growth in the figure below), but it also has the same functionality as the national Jobs Calculator. (Because of the recent partial government shutdown, the data are updated only through August; state-level employment data for October will be available November 22.)
Like the national Jobs Calculator, the state-level version allows the user to input a target unemployment rate, choose the number of months desired to hit the target rate, and find out how many new jobs are required per month to get there. But the calculator is flexible enough to allow other interesting experiments as well.
Consider the case of Florida. During the recession, Florida experienced a significant decline in its population growth. It has gone from a high of about 0.2 percent growth per month (roughly 2.4 percent per year) to its current 0.115 percent growth per month (about 1.38 percent per year; see the figure below). Suppose policy makers in Florida want to know how a return to prerecession population growth might affect the number of jobs needed to maintain its current unemployment rate over the next 12 months. (Note that as of August, the unemployment rate in Florida was 7 percent.)
The calculator's default settings always answer the question, “How many jobs per month does it take to maintain today's unemployment rate over the next 12 months?” To answer our hypothetical policy makers' question, all they would have to do is enter a prerecession monthly population growth rate of 0.2 percent into Florida's state Jobs Calculator, leaving everything else the same. Given the current data in hand, we would discover that Florida would need to generate about 6,000 more jobs per month at the higher population growth than at the current—and lower—population growth to stabilize the unemployment rate at 7 percent.
The data behind the state-level Jobs Calculator come from the U.S. Census Bureau's Establishment Survey, the same data used for the national Jobs Calculator, combined with the Local Area Unemployment Statistics (LAUS) programs run by each state. The LAUS contain the regional and state employment statistics that are consistent with data from the Census Bureau's Current Population Survey. State-level population estimates are provided by the U.S. Census Bureau (and are described in more detail here). You'll note that the LAUS data, especially for very small states, look more erratic than national or larger states' numbers—the unfortunate consequence of small sample sizes.
LAUS data are generally issued about the third Friday of each month following the reference month, which means that the state-level Jobs Calculator statistics will be updated about two weeks after the national Jobs Calculator. The schedule of release dates is available from the U.S. Bureau of Labor Statistics.
By Julie Hotchkiss, a research economist and policy adviser in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference Atlanta Fed's Jobs Calculator Drills Down to the States:
October 09, 2013
Delving into Labor Markets
Though never far from the headlines, the Federal Reserve's dual mandate comes front and center again with the announcement today of President Obama's nomination of Fed Vice Chair Janet Yellen as the next chair of the Board of Governors. Inevitably, analysis will turn to discussions of who is a hawk and who is a dove, who cares relatively more about inflation, and who cares relatively more about growth and employment.
That's unfortunate, because such characterizations really do miss the point. The debate among different policymakers is not about whether person A is more concerned about jobs and unemployment than person B, but about legitimate and longstanding conversations about what accounts for the performance of labor markets and what role monetary policy might have in the event that performance is judged to be subpar.
As it happens, the Atlanta Fed's most recent contribution to this discussion came last week in the form of the annual employment conference sponsored by the Bank's Center for Human Capital Studies. Organized, as in past years, by Richard Rogerson (Princeton University), Robert Shimer (University of Chicago), and Melinda Pitts (Federal Reserve Bank of Atlanta), the conference explored the causes of the continued weak labor market recovery in the United States. The existing literature has suggested a number of possibilities: wage rigidities, mismatch between workers' skills and the skills required by new jobs, extended unemployment insurance benefits and other government policy changes, and firms' reorganizing and asking workers to do more. The papers sought to analyze and document the importance of these factors for the slow recovery.
One notable policy change in the recent recession was the unprecedented expansion of unemployment insurance (UI) benefits to as long as 99 weeks for a very large fraction of UI-eligible workers. Did this increase play an important role in high levels of unemployment? Two papers from the conference addressed this question from different perspectives. "Do Extended Unemployment Benefits Lengthen Unemployment Spells? Evidence from Recent Cycles in the U.S. Labor Market," by Henry S. Farber and Robert G. Valetta, assessed the extent to which extended UI benefits result in higher unemployment because workers choose to remain unemployed longer. They find a statistically significant effect of longer UI durations on the duration of unemployment spells, but they conclude that the overall contribution to the unemployment rate was less than half a percentage point. Because the aggregate unemployment rate rose by more than 5 percent, this effect accounts for less than 10 percent of the overall increase.
"Unemployment Benefits and Unemployment in the Great Recession: The Role of Macro Effects," by Marcus Hagedorn, Fatih Karahan, Iourii Manovskii, and Kurt Mitman, offered a different perspective. The authors look at the evolution of unemployment rates in counties that are adjacent but lie in different states. They use the fact that the timing of extended benefits occurs at different times across states to identify the effect of extended UI durations on country-level unemployment. They find that the effects are sufficiently large that the increase in UI duration can account for virtually all of the increase in unemployment.
While seemingly at odds, the results of these two studies are consistent. The first paper shows that the decrease in the job-finding rate for workers with relatively longer benefits did not increase that much compared with the rate for workers with shorter-duration benefits, holding the overall unemployment rate constant. The second paper argues that the job-finding rate decreases for everyone when benefits are extended. The authors find that when some workers have access to longer-duration UI benefits, being unemployed is not as painful for them, which puts upward pressure on wages. To the extent that firms cannot target their job openings toward workers without access to UI, firms may be less likely to create jobs, making it harder for all workers to get job offers. The impact on uninsured workers may be as large as the impact on insured workers, and so the microeconomic estimates in Farber and Valetta will not necessarily uncover UI's total impact on the unemployment rate.
The possible role of wage rigidities has figured prominently in many accounts of the large increase in unemployment during the recent recession. Two papers considered the importance of this explanation. "Wage Adjustment in the Great Recession," by Michael Elsby, Donggyun Shin and Gary Solon, used microdata from the U.S. Census Bureau's Current Population Survey to examine the extent to which wages are sticky. The paper finds that there has been less response in average real wages during the recent recession than in previous recessions, perhaps suggesting that real wage rigidity contributed to the large increase in unemployment. However, they also show that wages at the individual level are really quite flexible. Specifically, relatively few individuals have zero nominal wage growth from one year to the next, and many people experience decreases in nominal wage rates.
A key issue in the theoretical literature is the extent to which wage stickiness affects new hires versus existing workers. In "How Sticky Wages in Existing Jobs Can Affect Hiring," authors Mark Bils, Yongsung Chang and Sun-Bin Kim show that even if wages for new hires are completely flexible, they may nonetheless have large effects on unemployment fluctuations when one allows for an "effort decision" for existing workers. This decision means that in response to negative shocks, firms require existing workers to expend more effort given that their wage is fixed, decreasing the need to hire new workers. The authors show that this effect is quantitatively significant and can come close to resolving the unemployment volatility puzzle, which relates to the large fluctuations in unemployment relative to productivity.
An empirical regularity that has appeared in the last few years is an outward shift in the Beveridge curve, which relates the unemployment rate to the level of vacancies. One interpretation of this upward shift is that the matching of unemployed workers and vacancies has worsened. Yet there is a lot of variety in the job-search effort by workers with different characteristics, such as the length of unemployment, whether they are on temporary layoff, and so on. In "Measuring Matching Efficiency with Heterogeneous Jobseekers," Robert Hall and Sam Schulhofer-Wohl devise a method for incorporating this heterogeneity into the analysis and show that there has indeed been a decrease in the matching rate for workers during the last few years. It will be important for future research to determine how much this decrease reflects a decline in search intensity or whether the lower job-finding rates represent a decrease for a given level of search intensity.
Related to the two issues of nominal rigidities and mismatch, in the paper "Labor Mobility within Currency Unions," Emmanuel Farhi and Ivan Werning study the role of labor mobility in diminishing the effects associated with nominal rigidities. For example, some researchers have suggested that a key difference between the apparent success of the United States relative to the euro zone is U.S. labor is more mobile. Farhi and Werning argue that one should not assume the mobility necessarily reduces the effects of nominal rigidities. In particular, they conclude that mobility eases the effects of nominal rigidities only if goods markets are well integrated.
Two papers focused on the nature of worker mobility across firms in the recent recession. In "Worker Flows over the Business Cycle: The Role of Firm Quality," Lisa Kahn and Erika McEntarfer examine recent changes in flows of workers between firms that offer jobs of differing quality. They find that that lower-quality firms decreased both hiring and separations by large and equal amounts, whereas high-quality firms have much smaller declines in both hiring and separations. The net result is that the fraction of workers in lower-quality jobs tends to increase during recessions.
In closely related work, "Did the Job Ladder Fail after the Great Recession?" by Giuseppi Moscarini and Fabien Postel-Vinay, uses data from the U.S. Bureau of Labor Statistics' Job Openings and Labor Turnover Survey (JOLTS) to study the hiring and separation patterns across firms of different sizes. They determine that the pattern of firm growth across size classes was different during this recession than in previous recessions. In particular, they find that following the Lehman Brothers collapse, smaller firms actually fared worse than larger firms, perhaps because financing constraints had more severe consequences for smaller firms.
As the provisions in the Affordable Care Act (ACA) take effect in the coming months, there may be large effects not only on the market for health care but also on the labor market. In particular, the ACA will implicitly introduce taxes and subsidies that will differ across firms and workers of different types. In "Effects of the Affordable Care Act on the Amount and Composition of Labor Market Activity," Trevor Gallen and Casey Mulligan develop a framework to think about how these provisions will influence labor market outcomes across different sectors and worker types, and they use a calibrated version of the model to quantify the effects. The authors predict that the ACA will substantially reduce the return to market work for low-skilled individuals and that a large number of individuals who currently receive health insurance through their employers will end up purchasing insurance through the exchanges established as part of the ACA.
The conference also featured a presentation by Ed Lazear, "The New Normal? Productivity and Employment during the Recession and Recovery." The talk highlighted three themes from Lazear's recent research. First, productivity did not decline in the recent recession—as it typically had done in previous recessions—perhaps reflecting that workers expend more effort during periods of high unemployment since they fear unemployment more in a weak labor market. Second, the unemployment rate is a less useful indicator of the overall state of the labor market during the current recovery (in recent years the decline in the unemployment rate has not been accompanied by an increase in the employment-to-population ratio, since labor force participation has declined). The third theme is that the deterioration in labor market outcomes during the recent recession should be interpreted as cyclical rather than structural and, hence, a labor market recovery is likely once GDP growth is stronger.
We certainly wouldn't claim that the conference put to rest any of the relevant questions that will confront the Federal Open Market Committee and its new chair going forward. But we do believe that continuing to support the dissemination of the type of research presented at this conference gives us a fighting chance.
By Richard Rogerson of Princeton University and Robert Shimer of the University of Chicago, both advisers to the Atlanta Fed's Center for Human Capital Studies, and Melinda Pitts, director of the Atlanta Fed's Center for Human Capital Studies
TrackBack URL for this entry:
Listed below are links to blogs that reference Delving into Labor Markets:
September 23, 2013
The Dynamics of Economic Dynamism
Earlier today, Atlanta Fed President Dennis Lockhart gave a speech at the Creative Leadership Summit of the Louise Blouin Foundation. He posed the questions: Is the economic dynamism of the United States declining? Is America losing its economic mojo? He observed:
“... we see a picture in which fewer firms are expanding, and each expanding firm is adding fewer new jobs on average than in the past. Fewer firms are shrinking, and each is downsizing by less on average. Fewer people are being laid off or are quitting their job, and firms are hiring fewer people. In other words, the employment dynamics of the U.S. economy are slower.
The decline in job creation and destruction was also the theme of this recent macroblog post by Mark Curtis, which featured some pretty nifty dynamic charts of trends in job creation and destruction by industry and geography.
Identifying the policy implications of these slower dynamics requires careful diagnosis of the causal factors underlying the trends. The cutting edge of economic research looking at this issue was featured at the 2013 Comparative Analysis of Enterprise Data Conference hosted last week by the Atlanta Census Research Data Center (ACRDC), which is housed at the Atlanta Fed and directed by one of our senior research economists, Julie Hotchkiss. Through the ACRDC, qualified researchers in Atlanta and around the Southeast can perform statistical analyses on non-public Census microdata.
The agenda and papers presented at the conference are located here. Some of the papers, I think, were particularly relevant to what President Lockhart discussed. A few examples:
“Reallocation in the Great Recession: Cleansing or Not?” by Lucia Foster and Cheryl Grim of the Center for Economic Studies at the U.S. Census Bureau and John Haltiwanger at the University of Maryland looked at the so-called “cleansing hypothesis,” in which recessions are not only periods of outsized job creation and destruction, but they are also periods in which the reallocation is especially productivity enhancing. They find that while previous recessions fit this pattern reasonably well, they do not see this kind of activity in the most recent recession. In fact, they find that in the manufacturing sector, the intensity of reallocation fell rather than rose (because of the especially sharp decline in job creation), and the reallocation that did occur was less productivity enhancing than in prior recessions.
“How Firms Respond to Business Cycles: The Role of Firm Age and Firm Size,” by Javier Miranda, Teresa Fort, John Haltiwanger and Ron Jarmin, looked at the varying impact of recessions on firms by size and age. They show that young businesses (which are typically small) exhibit very different cyclical dynamics than small/older businesses and are more sensitive to the cycle than larger/older businesses. The paper also explores explanations for the finding that young/small businesses were hit especially hard during the last recession. They identify the collapse in housing prices as a primary culprit, with the decline in job creation at young firms especially pronounced in states with a large drop in housing prices.
As a side note, although not presented at the conference, “The Secular Decline in Business Dynamism in the U.S.,” a new paper by Ryan Decker, John Haltiwanger, Ron Jarmin and Javier Miranda, analyzes the overall secular decline in job reallocation across industries. They find that changes in industry composition (the decline in manufacturing and rise of service industries) are not driving the decline. Instead, the primary driver seems to be the decline in the pace of entrepreneurship and the accompanying decline in the share of young firms in the economy.
Finally, Steve Davis, from the University of Chicago, talked about his joint research with John Haltiwanger, Kyle Handley, Ron Jarmin, Josh Lerner and Javier Miranda on private equity in employment dynamics, Private equity critics claim that leveraged buyouts bring huge job losses. Davis shows that private-equity buyouts are followed by a decline in net employment at these firms relative to controls (similar firms that were not targets of a buyout). However, that net change pales compared with the amount of gross job creation and destruction that typically occurs within the target firm after the buyout. In particular, he finds that in addition to reducing employment at its existing establishments, including by selling some establishments to other firms, jobs are created at new establishments within the firm via acquisition and the opening of new establishments. Moreover, they show that this reallocation is generally productivity enhancing for the firm. Although the data used in the study go only through the mid-2000s, it seems reasonable to infer from the findings that the decline in private equity deals during and since the last recession has contributed to the overall lower level of employment dynamics in this recovery.
The Comparative Analysis of Enterprise Date Conference was an excellent representation of the type of high-quality research being conducted on questions that go to the heart of the cyclical-versus-structural debate about the future course of the U.S. economy. While this is an exciting and important time for researchers in this field, it is troubling to learn that the programs that collect the data used in these types of studies are being trimmed because of federal budget cuts.
By John Robertson, vice president and senior economist in the Atlanta Fed’s research department
TrackBack URL for this entry:
Listed below are links to blogs that reference The Dynamics of Economic Dynamism:
September 17, 2013
The ABCs of LFPR
As the Federal Open Market Committee (FOMC) meets amid much speculation about the next steps for monetary policy, it does so in in the context of an August 2013 Employment Situation report that was generally viewed as a mixed bag. The employment numbers undershot the consensus among most market observers, while the unemployment rate edged down again. But even the drop in the unemployment rate—a cumulative 0.8 percentage point over the past 12 months—failed to impress everyone. Martin Feldstein, for instance:
The official unemployment rate has declined sharply (to 7.3% last month from 10% in October 2009) only because so many people have stopped looking for work or are working part-time.
Part of what Professor Feldstein is referring to, of course, is the labor force participation rate (LFPR), which measures the share of the adult population that is in the labor force. LFPR includes those who are employed and those who are unemployed but looking for a job, but not those who are unemployed and are not looking for a job (which includes retirees and discouraged workers).
We generally refrain from direct commentary about issues related to monetary policy in the time surrounding FOMC meetings. I won't break with that tradition but am more than happy to highlight a resource that can help you draw your own conclusions about all things having to do with the labor market, including the LFPR.
Our Center for Human Capital Studies' Federal Reserve Human Capital Compendium is a collection of Federal Reserve System research published on topics related to employment, unemployment, and workforce development. Our latest update offers several entries that address the LFPR and its implications for the labor market. Two recent additions:
Will a Surge in Labor Force Participation Impede Unemployment Rate Improvement? Researchers at the Richmond Fed concluded that, in the short run, the LFPR and the unemployment rate are negatively correlated. This conclusion is derived from the fact that unemployed participants in the labor force are more likely to leave the labor force than those who are employed. Also, movement from unemployed non-participant to employed participant (basically skipping the unemployed-participant phase) is more likely in an improving labor market. They concluded that movements in the LFPR lag six months behind movements in the unemployment rate.
Cyclical versus Secular: Decomposing the Recent Decline in U.S. Labor Force Participation. Researchers at the Federal Reserve Bank of Boston found that since 2008, the decline in the LFPR largely reflects demographic effects of an aging population. Furthermore, the cyclical response of the LFPR during the latest recession and recovery period has been smaller than expected, so the unemployment rate would have been three-quarters of a percent lower if the LFPR had followed historical norms. They conclude that going forward, the unemployment rate should give an accurate read on labor market conditions and that further cyclical declines in the LFPR are unlikely if the labor market continues to improve.
But much more information on the LFPR and other topics including wages and earnings, outsourcing, and productivity is available. If you're looking for something to do while you await the FOMC's decision, one option is building a little human capital of your own with our Human Capital Compendium.
By Whitney Mancuso, a senior economic analyst in the Atlanta Fed's research department
TrackBack URL for this entry:
Listed below are links to blogs that reference The ABCs of LFPR:
September 13, 2013
Job Reallocation over Time: Decomposing the Decline
One of the primary ways an economy expands is by quickly reallocating resources to the places where they are most productive. If new and productive firms are able to quickly grow and unproductive firms can quickly shrink, then the economy as a whole will experience faster growth and the many benefits (such as lower unemployment and higher wages) that are associated with that growth. Certain individuals may experience unemployment spells from this reallocation, but economists, starting with Joseph Schumpeter, have found that reallocation is associated with economic growth and wage growth, particularly for young workers.
Recently, a number of prominent economists such as John Haltiwanger have expressed concern that falling reallocation rates in the United States are a major contributor to the slow economic recovery. One simple way to quantify the speed of reallocation is to examine the job creation rate—defined as the number of new jobs in expanding firms divided by the total number of jobs in the economy—and the destruction rate, defined likewise but using the number of jobs lost by contracting firms. Chart 1 plots both the creation and the destruction rates of the U.S. economy starting in 1977. These measures track each other closely with creation rates exceeding destruction rates during periods of economic growth and vice versa during recessions. The most recent recession saw a particularly sharp decline in job creation (you can highlight the creation rate by clicking on the line), but it is clear this decline is part of a larger trend that far predates the current period. A decline in these rates could indicate less innovation or less labor market flexibility, both of which are likely to retard economic growth. Feel free to explore the measures for yourself using the figure’s interactivity.
To better understand these important trends we create a common variable called reallocation, which is defined as total jobs created plus total jobs destroyed, divided by total jobs in the economy. This formula creates one measure that describes how quickly jobs are moving from shrinking firms to expanding firms. Using data from the U.S. Census Bureau’s Business Dynamic Statistics, we examine differences in this variable across sectors and across states. Furthermore, using some basic data visualization tools, we can see how reallocation has evolved over time across these dimensions.
Chart 2 plots reallocation rates by industry from 1977 to 2011. The plot highlights the reallocation rate for all industries, but you can also select or deselect any industry to more clearly view how it has changed over time. Scrolling over the lines allows you to view the exact rates by industry in any time period. A few interesting patterns emerge. First, sectors have different levels of job reallocation in the cross section. Manufacturing stands out as having particularly low reallocation rates, probably the result of the large fixed-cost capital requirements required in production. Second, not all industries experienced sharp declines during this period. If you highlight the finance, insurance, and real estate sector, it is evident that reallocation rates actually increased for this sector until the most recent recession. Retail and construction, on the other hand, have experienced steady and significant declines during the past 35 years.
Chart 3 maps reallocation rates across states for the year 1977. This figure provides us with a cross sectional view of geographical differences in reallocation rates. States with the highest reallocation rates are dark brown, and states with the lowest rates are light brown. You can click through the years to visually capture how these rates have changed overtime for each state. Compare the color of the map in 1977 with the color in 2011. Scroll the mouse over any state to view that state’s reallocation rate in the particular year.
As with industries, states display clear cross sectional differences in their reallocation rates. The highest rates are found in western states, Florida, and Texas, and the lowest are in the Midwest. Scrolling through the years shows that the decline in reallocation rates is common to the entire country.
Overall, these figures display a stark trend. The economy is reallocating jobs at much slower rates than 20 or even 10 years ago, and this decline is, with only a few exceptions, common across states and industries. Economists are just now starting to explore the causes of this trend, and a single, compelling explanation has yet to emerge. But some explanation is clearly in order and clearly important for economic policymakers, monetary and otherwise.
By Mark Curtis, a visiting scholar in the Atlanta Fed's research department
Please note that the charts and maps in this post were updated and improved on November 27, 2013.
TrackBack URL for this entry:
Listed below are links to blogs that reference Job Reallocation over Time: Decomposing the Decline: