Macro Roundup

> “…While it is difficult to prescribe what the optimal pace should be, > there is accumulating evidence from multiple datasets and a variety of > methodologies that the pace of business dynamism in the U.S. has fallen over > recent decades and that this downward trend accelerated after 2000….To get > a sense of the pace of business dynamism in the U.S. note that, over the > 1980 to 2011 period, annual rates of gross job creation and gross job > destruction for private sector businesses averaged 16.7 and 15.2 > respectively. That is 16.9 percent of U.S. private jobs are created at new > and growing businesses, and 15.2 percent of jobs are lost at shrinking and > closing business. Net employment growth is the difference between creation > and destruction. Figure 1 shows that these rates exhibit both cyclical and > secular patterns. The magnitude of the decline is significant. Between > 1985 and 2005, the rate of gross job creation fell by 2.9 percentage points, > and the gross job destruction rate fell by 1.5 percentage points. A 1.0 > percentage point increase in the rate of gross job creation for the U.S. > private non-agricultural sector would represent roughly 1.1 million > additional jobs created in recent years….. a critical factor in the > decreasing pace of business dynamics is lower business start-up rates and a > decreasing role in the dynamic young businesses in the economy… In this > paper, we explore these trends further and discuss potential explanations. > For this purpose we use the Bureau of the Census Longitudinal Business > Database, a new database that covers all non-farm private sector > establishments and firms with paid employees in the U.S. between 1976 and > 2011. We analyze the extend to which compositional changes in U.S. > businesses across detailed industries, states, size classes, age classes and > firm structure account for the decline in dynamism. We find that the shift > in economic activity from smaller and younger firms toward larger more > mature firms over this period help explain the decreasing pace of business > dynamism. Changes in the industrial composition toward more dynamic sectors > have a muting effect but are not sufficient to reverse the firm age and size > effects. Overall, we find that such composition effects explain no more > than a quarter of the decline in dynamism even when we consider rich > interactions across these firm characteristics. The failure of these > effects to account for the aggregate decline partially reflects offsetting > composition effects but suggest that the real driving force is to be found > in factors working within detailed industry, firm size and age and > geographic groupings….Business dynamism as measured by the pace of job > creation, job destruction and/or job reallocation, has exhibited a > pronounced secular decline in the U.S. The most important factor we have > identified that helps to account for this decline is the decline in the pace > of entrepreneurship and the accompanying decline in the share of young > firms. Young firms have the highest pace of both job creation and job > destruction in the U.S. The large decline in the share of economic activity > by young firms accounts for a substantial share of the overall trend decline > in the pace of job reallocation. Changes in the age composition of U.S. > business activity accounts for about 26 percent of the overall decline in > the pace of job reallocation. In some sectors such as services, it > accounts for over 40 percent of the decline in the pace of job reallocation. > We do not have an explanation for the decline in the pace of > entrepreneurship, but it is apparent that this is one of the primary > contributing factors in accounting for the decline in the pace of business > dynamics….Shifting age composition accounts for 32 percent of the observed > decline in job creation, 20 percent of the decline in job destruction, and > 26 percent of the decline in job reallocation…. Ultimately, most of the > declines in these sectors is in the “unexplained” within component. There > are apparently changes in the conditions facing businesses in these sectors > that cannot be accounted for by the joint contribution of firm age, firm > size, detailed industry, geography and multiple location status….” > > Decker, Ryan and John Haltiwanger, Ron S Jarmin and Javier Miranda, “The > Secular Decline in Business Dynamism in the U.S.” NBER, 2014. Available at: > http://econweb.umd.edu/~haltiwan/DHJM_JEP_5_17_2013.pdf

"...The striking finding here is thatlong-term non-employment in the immediate aftermath of the Great Recession was not much worse than it was in the early 1980s for most cohorts, with the notable exception of young men. Moreover, it seems as if there has been a modest decline in long-term non-employment rates since the economy hit bottom in 2009....There were two major findings.First, about 10 per cent of men who are laid off en masse are never employed again. Intriguingly,the overall health of the economy at the time of getting laid off does not seem to play much of a role, although age does:...Second, the long-term employment effect of being laid off is remarkably constant across the four recessions for which we have data:..In a worse case scenario,these numbers thus imply that 10% of the population experienced a 10% permanent drop in employment, and hence the employment-population ratio permanently declined by 1 percentage point...These calculations would imply that the remainder of the gap of the current employment-population ratio to this implied steady state can be considered as ‘slack,’ i.e., workers that are not employed due to cyclical conditions that will return to employment as the labor market continues improving..."This time is not that different, long-term unemployment editionMatthew C Klein | Aug 26 14:16 |1 comment

But this is not all good news:

Based on the average survivor curve for the more recent period, about half the gap should have disappeared five years after the initial decline, and roughly 60% of the gap ten years after. Thus, five to ten years after the Great Recession the employment population ratio would be predicted to be 1 to 2 percentage points lower than it was before the recession.

If past patterns hold — and they mostly seem to be — we should expect some additional members of the ultra-long-term non-employed who are sitting on the sidelines of the job market to eventually re-enter the labor force as the economy continues to grow. Cold comfort for everyone else, although it should be reassuring to policymakers concerned that something unusual might prevent the labour market from reabsorbing the jobless.

That said, we are somewhat worried that the the lavender curve, which represents the odds that someone who stopped working in the Great Recession remains non-employed, seems to have stopped declining at a level above 50 per cent. Could the pattern have broken after the first few years of recovery?

(A footnote does warn that “care has to be taken in interpreting the final point of each episode in Figure 7, Panel B” because of the way months are averaged during recession years, so we should probably reserve judgment until we get a few additional years of data.)

It’s also worth noting the consistent upward shift in the odds of remaining outside the labour force since the 1980s of about 10-15 per cent. Song and von Wachter do not attempt to explain this phenomenon, although we wonder if it could be partly attributable to changes todisability insurance, possibly combined with therising population of ex-convicts.

The second part of the paper focuses on the long-term employment effects of being laid off.

To make their analysis easier, Song and von Wachter first narrowed their sample to men who worked for at least three consecutive years at the same employer, which had at least 50 employees.

These filters, including the focus on men only, arenot uncommonin the literature on layoffs and are designed to exclude young parents taking time off to spend with children, early retirees, and those fired for cause, all of whom would corrupt the sample.

The data made it easy to see when a worker separated from his employer — John Doe used to earn taxable income from the Cleveland Widget Company and then stopped — but not why. Lots of people quit, for example.

Song and von Wachter isolated the impact of layoffs by only counting separations that occurred at the same time a company cut its workforce by at least 30 per cent over a two-year period. (The exact conditions needed to define a layoff are a little more complex but interested readers can find the details on pages 15-16 of the paper.)

There were two major findings.

First, about 10 per cent of men who are laid off en masse are never employed again. Intriguingly, the overall health of the economy at the time of getting laid off does not seem to play much of a role, although age does:

Second, the long-term employment effect of being laid off is remarkably constant across the four recessions for which we have data:

As you can see in the chart above, men who were laid off during the teeth of the Great Recession were about as likely to be employed four years later as men laid off in 1990 or 2001. The distinguishing feature this time around was that many more people were victims of layoffs.

An optimist would see that chart as a demonstration that the labour market performed about as well as it always has given the magnitude of the downturn. The pessimistic interpretation is that many of the people who lost their jobs during the crisis and remain “non-employed” will never work again for almost no fault of their own.

Song and Von Wachter conclude by estimating the permanent employment costs of the most recent recession:

In a worse case scenario, these numbers thus imply that 10% of the population experienced a 10% permanent drop in employment, and hence the employment-population ratio permanently declined by 1 percentage point.

These calculations would imply that the remainder of the gap of the current employment-population ratio to this implied steady state can be considered as ‘slack,’ i.e., workers that are not employed due to cyclical conditions that will return to employment as the labor market continues improving. This entry was posted byMatthew C Kleinon Tuesday August 26th, 2014 14:16. Tagged withJackson Hole 2014, US Unemployment. Klein, Matthew, "This time is not that different, long-term unemployment edition," FT Alphaville, August 26, 2014. Available at:http://ftalphaville.ft.com/2014/08/26/1942831/this-time-is-different-long-term-unemployment-edition/

Put another way, as much as three percent of thecivilian population— about 7.5 million Americans — may have to get jobs before the US economy can be said to have fully recovered.

There is no evidence that the rate of exit from long term nonemployment has slowed during the Great Recession compared to the patterns in all episodes since the 1990s.

As the paper puts it:

As you can see in the chart below, the chances of getting back into the workforce after being non-employed for a full calendar year improve pretty dramatically (the lines all go down). Intriguingly, the odds of re-entering the workforce appear relatively constant over time and regardless of whether the economy is growing or shrinking. That said, many people who stop working for a single year never again earn taxable income:

Song and von Wachter’s solution to this problem was to analyse the probability that someone who joined the ranks of the long-term non-employed in a given year remained jobless (as long as they are under the age of 65).

The two-year measure supports the claim that the Great Recession was about as severe as the contraction of the early 1980s as well as the finding that things have modestly improved since the trough. Again, though, this measure is imperfect insofar as it fails to capture people who would be willing to work but have not been employed for at least three full consecutive calendar years.

One way around this problem is to lengthen the minimum duration of non-employment considered. Song and von Wachter calculated non-employment rates for people who earned no income for at least two consecutive years. Here is a chart comparing this measure against the one-year measure, both for men and women:

We should be somewhat cautious about the extent of the decline, however, since anyone who has been out of work for at least two full consecutive calendar years doesn’t get counted after the first full year — even if she were still interested in finding a job.

The striking finding here is that long-term non-employment in the immediate aftermath of the Great Recession was not much worse than it was in the early 1980s for most cohorts, with the notable exception of young men. Moreover, it seems as if there has been a modest decline in long-term non-employment rates since the economy hit bottom in 2009.

Their results by age and gender are below:

They begin by looking at the number of workers who made no money for at least a full calendar year after having earned taxable income the year before. Dividing this by the sum of the number of people working and the number of people who just became members of the long-term non-employed turns it into a proportion of the effective labour force.

In the first part of their paper, Song and von Wachter address what they call “long-term non-employment”, which, unlike unemployment, is sometimes voluntary.

We were surprised that the red line washigherthan the blue line given that the BLS survey is supposed to capture a broader swathe of workers than the Social Security system. We asked von Wachter about this and he said one possibility is that “the SSA data will capture any formal employment, no matter how short or low paid” while “short or low paid employment relationships may be under reported” in the BLS surveys, although he suspected there were probably other factors at work as well.

For example, a comparison of the BLS measure of employment against one constructed using the Social Security data indicates that the slightly narrower sample does not meaningfully decrease accuracy. Both were divided by the same BLS measure of the civilian population:

While there are some drawbacks with this approach — it excludes workers who aren’t forced to contribute to the Social Security system and it doesn’t distinguish between people who stop working involuntarily and those who cease toiling for pay of their own volition, among others — there are good reasons to think that these tax data can tell us useful things about labour market behavior.

Song and von Wachter get around both of these problems by using confidential income data from the Social Security Administration. By focusing on how much people actually earned each year rather than the answers tosomewhat arbitrary questionsabout how hard they were looking for work, they teach us facts about the labor market that we could not have learned by relying solely on the public data.

First, it isn’t possible to closely track individuals over time using publicly-available data. ThePanel Study of Income Dynamicsis attractive for tracking long-term trends but the sample size is small and low-frequency, for example. Second, employment data generally come from surveys and the ways people respond to those surveys may not be consistent over time. (Think of changes to unemployment benefits.)

Before we get into Song and von Wachter’s findings, it is worth taking a few moments to understand their methodology, which gets around the two big limitations of most studies of the job market.

But an intriguingpaperby Jae Song and Till von Wachter presented at this year’s Jackson Hole symposium suggests that the impact of the most recent recession on long-term unemployment was not actually that unusual given the number of jobs lost at the outset. The paper also presents some encouraging evidence that many of those who appear to have given up hope of finding a job could rejoin the labour force if the economy keeps expanding, as well as sobering demonstrations of the permanent costs of being laid off.

It looks as if people who lost their jobs in 2007-2009 are much more likely to remain unemployed — 5 years after the recession officially ended — than their predecessors. On top of that, of course, is the massive and sustained decline inthe share of working-age people with a job:

If we were asked to make the Great Recession look radically different from all other postwar US recessions, we would point tothis chart: