New Jobless Claims Aren’t As Good As They Look

The Department of Labor announced yesterday that new jobless claims have fallen to 305,000 for the week, a six-year low.  Unfortunately, the new jobless claims figure isn’t as reliable an indicator of labor market trends as it used to be.  It says more about the continued decline in layoffs than it does about a surge in hiring.  The biggest problem in today’s labor market is a weak hiring rate for employers.  So while a lower level of new jobless claims is better than a higher level, I don’t expect this to signal a boom in job creation.

To see why layoffs aren’t the problem, as many workers were laid off in the five years since the Lehman collapse as during the real estate boom of 2003-2007.  In the five years between 2003 and 2007, in the midst of a real estate boom and economic recovery, when the unemployment rate averaged 5.2%, there were 1.86 million layoffs per month, on average.  In the five years since the Lehman collapse in August 2008, including the biggest recession since the Great Depression, when the unemployment rate averaged 8.6%, there were also 1.86 million layoffs per month, on average.  While layoff rates were elevated during the second half of 2008 and all of 2009, over the past three years layoffs have been 8.8% lower than pre-recession levels and continue to decline.

Over the past three years during our tepid economic recovery, only 4.23 million people have been hired per month, on average, compared to 5.1 million people per month in 2003-2007. 870,000 fewer people have been hired each month over the past three years compared to 2003-2007.  This great slowdown in hiring, which represents a decline of 17% compared to the previous economic recovery, is the biggest challenge facing the labor market and the economy.

Unemployment insurance data are less relevant in 2013 because of the narrow focus on relatively recent job losers.  New labor market entrants and re-entrants to the labor force, comprise about half of the unemployed but are ineligible for unemployment insurance.  Longer term unemployed workers and jobless workers who have given up their job search and dropped out of the labor force are also ineligible for UI.  For unemployed new entrants and re-entrants to the labor force, and for the millions of jobless workers who have dropped out of the labor force, a surge in hiring is needed to bring them back to work.  So while there were 3.8% fewer layoffs in the past 12 months compared to one year earlier, the fact that hiring only increased by 1.1% over the same period means that employment rates will remain low until businesses increase the pace of their hiring.

Unemployment insurance data are also less relevant in today’s economy because the fraction of unemployed workers eligible for state unemployment insurance programs is at an all-time low.  The following chart shows the percentage of unemployed workers who are job losers and have been unemployed for 26 weeks or less.  These two conditions are good proxies for the key determinants of eligibility for most state UI programs.* Since 1980, in non-recession years, about 40% of unemployed workers would satisfy these requirements.  During recession years, about 45% of the unemployed are job losers in their first 26 weeks of unemployment.  Over the past 3 and one half years, however, only about one-third of unemployed workers have been relatively recent (26 weeks or less) job losers.

losers

The fraction of jobless workers who are eligible for UI is even smaller.  If one includes workers who have dropped out of the labor force in the past year but are available for work as “jobless”, only 26% of “jobless” workers satisfy the conditions required by most state UI programs.

While new jobless claims data provide some information about the rate at which workers are losing jobs, and whether job losers appear to be finding work before filing for UI benefits, yesterday’s new claims figures must be interpreted carefully.  The labor market of 2013 is very different from the pre-recession labor market.  Layoffs continue to decline but have not coincided with commensurate increases in hiring in a weak recovery.  Most jobless workers are long-term unemployed, new-entrants or re-entrants to the workforce unable to find work, or those who have given up job search altogether and are no longer labor force participants.  Employers have hired 870,000 fewer workers per month over the past three years than they were prior to the recession.  Until hiring levels approach the 5 million hires per month that were common in 2003-2007, job growth will continue to disappoint.  Slightly lower new jobless claims per week is just one small step in the right direction.

*Unemployed workers are eligible for Federal UI benefits if they are unemployed longer than 26 weeks and reside in states where unemployment rates are sufficiently high.  Most state UI programs allow up to 26 weeks of benefits but they need not be the first 26 weeks of a worker’s unemployment spell and benefits are only available for job losers.

Detroit’s Job Creation Problems

Detroit urgently needs job creation.  No American city faces more difficult economic problems than Detroit.  Even seemingly good economic news is misleading.  Detroit’s unemployment rate fell from 27.8% in the summer of 2009 to 16.3% in the most recent jobs report.  However, the apparent decline in unemployment is illusory.  Unemployment dropped because jobless residents either left the city or gave up looking for work; employment in Detroit actually declined by 1.8% between the first five months of 2009 and the first five months of 2013.

Some of Detroit’s economic problems are due to the mass exodus from the city.  Detroit’s population declined by 810,000 (about 54%) since 1970, or the equivalent of the population of Columbus, Ohio, the nation’s 15th largest city.  Enough residents have left so that the sprawling city of Los Angeles has a population density 63% higher than Detroit’s.  So many skilled workers have left Detroit that manufacturing businesses find it difficult to hire qualified workers even as the demand for their products has recovered in the past few years and the unemployment rate is 9 percentage points above the national average.

Detroit is No Longer the Motor City

In 1978 when the “Big Three” automakers had an 80% share of the domestic car market there were over a quarter of a million auto manufacturing jobs in Detroit.  Today, there are fewer than 38,000 auto manufacturing jobs in Detroit and suburban Wayne County, a decline of 85% since 1978.  Thirty five years ago Detroit was the home to one in four auto industry jobs in the U.S.  Today only 4.7% of U.S. auto industry jobs are located in Detroit and suburban Wayne County.  The decline in manufacturing jobs in Detroit is not limited to the auto industry.  Since 1978 there has been an 88% decline in employment in manufacturing industries other than motor vehicles.

A Longer Workweek in Manufacturing

At a time when many economists are concerned that most job creation consists of part-time jobs, the workweek in Detroit is getting longer.  The average length of the workweek in Detroit manufacturing establishments was 48.4 hours per week in 2012, about 6.7 hours longer (16.1%) than the U.S. average for manufacturing establishments.  The following chart shows that the average length of the workweek in Detroit’s manufacturing sector increased by 15.2% between 2003 and 2012, compared to a modest 3.2% increase for the U.S. overall.  (For ease of exposition the length of the workweek has been normalized to 100 in 2003.)

Detroit1

A substantially longer workweek means that during the economic recovery (from 2009 to 2012) about 44% of the increase in labor inputs in Detroit’s manufacturing sector were achieved by a longer workweek instead of job creation.  The following chart illustrates what might have happened had there not been increases in hours worked per employee.  Manufacturing employment in Detroit in 2012 would have been 15.2% higher had the length of the workweek remained constant (at 42 hours per week) from 2003 to 2012.  (Again for ease of exposition total labor inpouts in 2003 have been normalized to 100.)

Detroit2

Do Manufacturing Establishments in Detroit Face a Shortage of Skilled Labor?

There are a number of reasons why the length of the workweek in Detroit increased more substantially than in the remainder of the U.S.  If establishments in Detroit faced more economic uncertainty they may have chosen a longer workweek rather than incurring the costs of hiring additional workers.  If establishments in Detroit faced higher fixed costs of fringe benefits per worker, such as health insurance coverage, they may have also chosen to increase hours per employee instead of hiring more workers.  In addition collectively bargained agreements between manufacturing firms and unions may have specified that workers retained during previous reductions in force would see increased hours per week before new employees could be hired.

The sharp increase in the length of the workweek may also be evidence that manufacturing establishments in Detroit face a shortage of skilled workers.  If businesses find it difficult to attract qualified workers because of the exodus of skilled workers over the past decade, the best way to accommodate labor demand may well be to increase hours per employee instead of creating jobs.

Conclusion

The forecast for job creation in Detroit remains bleak despite the Federal government bailout of domestic auto manufacturers.  Detroit is no longer The Motor City as fewer than 5% of auto industry jobs are located there.  About 44% of the increase in labor inputs in manufacturing in Detroit has occurred due to a longer workweek instead of job creation.  Employers in Detroit are likely relying on a longer workweek rather than more job creation because of the exodus of skilled workers from the metro area as well as uncertainty about Detroit’s economic future.

Welch Consulting Employment Index Rebounds in January

The Welch Consulting Employment Index is 94.9 for January 2013, up from 94.5 in December.  The employment index increased despite the increase in the U.S. unemployment rate because more people were participating in the labor force in January, as a fraction of the total population, compared to December.  An index value of 94.9 means that full-time equivalent employment (from the BLS household survey) is 5.1% below its level in the base year of 1997, after adjusting for both population growth and changes in the age distribution of the labor force.  The index has recovered from sharp declines in the summer of 2012 and is the same as it was in March 2012.  This means that full-time employment has kept pace with population growth over the past ten months.  Over the past five years the Welch Consulting Employment Index has fallen 6.2% (it was 101.2 in January 2008).

The Welch Consulting Employment Index, disaggregated by gender, is 92.8 for men and 97.7 for women.  Both the indices for men and women are up sharply slightly from December.  Over the past ten months the men’s index is up 0.3% while the women’s employment index is down 0.2%.  Over the past five years the men’s employment index is lower by 6.5% and the women’s index is lower by 5.7%.

Welch_Index_Jan

 

Welch_Index_gender_Jan

Technical Note: Full-time equivalent employment equals full-time employment plus one half of part-time employment from the BLS household survey.  The Welch index adjusts for the changing age distribution of the population by fixing the age distribution of adults to the 1997 base year.  The Welch Index adjusts for population growth by fixing total population to its 1997 level.  Seasonal effects are removed in a regression framework using monthly indicator variables.

 

 

January Jobs and Seasonality

Earlier this week I expected the January jobs report to indicate that the economy “added” between 250,000 and 300,000 jobs.  My forecast was premised on BLS continued use of a much larger seasonal adjustment factor.  The BLS appears to be using a more conventional seasonal adjustment factor in 2013 than in 2011 and 2012.  This should mean fewer ups and downs in the jobs reports in 2013 due to seasonal noise.

The BLS seasonal adjustment procedure is a moving average process: seasonal factors change over time.  January is the weakest month of the year for jobs.  Therefore the seasonal factor for January inflates the payroll employment total in order to account for the anticipated drop in employment after the holidays.  The following chart shows how much the BLS seasonal factors have inflated employment in January from 2001 to 2013.  The lowest seasonal adjustment was +1.522% in 2009.  The ten-year average adjustment from 2001 to 2010 was 1.573%.

Jan1

The BLS attributed none of the severity of the downturn in January 2009 to a large seasonal effect at the time.  As the previous chart indicates the BLS actually viewed January 2009 as the mildest January in a decade; that is why employment was adjusted up by only 1.522%, much less than in other years.  The steep employment decline in January 2009 has an echo in two to three years.  The BLS X-12 ARIMA model for seasonal adjustment incorporates the loss of 3.7 million jobs between December 2008 and January 2009 as evidence that the seasonal “January effect” in employment had increased.  Consequently the seasonal adjustments for January 2011 and 2012 were +1.66% and 1.65% in order to offset the apparently larger “january effect”.  The 2011 and 2012 adjustments were the largest for January since 1965.

The following chart shows that the January seasonal adjustments added about 100,000 more jobs in 2011 and 2012 compared to the application of a longer run (10 year) average of seasonal effects.  The chart also shows that the January effect for 2013 (+1.598%) is only slightly above the ten-year average.  The large employment decline in January 2009 had a minimal impact on the adjustment factor last month, adding only about 33,000 more jobs.

Jan2

157,000 additional jobs in the past month, with 33,000 of the gain due to a generous seasonal adjustment, indicates that the job market is weak.  The good news on Friday came from the adjustments to the payroll reports for late 2012.  The January 2013 report is subject to more revisions.  At this point in the recovery, with the unemployment rate hovering near 8%, payroll employment growth of less than 200,000 per month is disappointing.

The January Effect and the Jobs Report

Expect tomorrow’s jobs report for January to indicate that the economy “added” between 250,000 and 300,000 jobs.  Also expect jobs growth in the spring of 2013 to be disappointing.  These expectations have nothing to do with budget negotiations in Washington, the impact of austerity measures in the Eurozone, or corporate earnings reports.  The pattern of strong employment growth in January followed by disappointing payroll reports in the spring is the result of the BLS seasonal adjustment procedure.

One year ago the BLS reported that nonfarm payroll increased by 275,000 between December 2011 and January 2012 — the largest increase in employment since the recession (excluding May 2010 when hundreds of thousands of seasonal Census workers were hired).  In fact, nonfarm payroll decreased by 2.67 million employees, but that decline was much smaller than the BLS statistical model projected.

The BLS seasonal adjustment procedure is a moving average process, which means that seasonal factors change over time.  The BLS seasonal  factors in January 2011 and January 2012 were unusually large because they reflected a historic decline of 3.7 million jobs between December 2008 and January 2009.  The BLS X-12 ARIMA model for seasonal adjustment viewed this massive decline in employment as a signal that the “January effect” in payroll employment had become more pronounced.  The procedure requires that future January’s would have larger projected employment declines.  To be clear, a larger “January effect” is measured relative to other months of the year.  If the new seasonal factors add 100,000 more jobs in January compared to the previous factors, a comparable number of jobs will be subtracted from other months of the year.  Hence bigger “January effects” mean smaller payroll gains in other months, most notably the spring.

The data from last January suggest that it was a particularly deep recession, and not changing seasonal factors, that accounted for the steep employment decline four years ago.  The legacy of the 2008-2009 decline is likely to still have an impact on BLS seasonal factors, but not by as much as in 2011 and 2012.  Nonetheless, it would not be surprising to see surprisingly strong job growth of 275,000 in tomorrow’s report, and weaker than expected growth in April and May.

Welch Consulting Employment Index Continued its Decline in December

The Welch Consulting Employment Index was 94.5 in December; down from 94.6 in November.  An index value of 94.5 means that full-time equivalent employment (from the BLS household survey) is 5.5% below its level in the base year of 1997, after adjusting for both population growth and changes in the age distribution of the labor force.  The index had recovered this fall from the sharp declines experienced in the summer of 2012, but has fallen for two straight months.  Moreover, the index is now below its level in March 2012.  Full-time employment has not quite kept pace with population growth over the past nine months.  The Welch Consulting Employment Index has grown by less than one percent per year over the past two years.

The Welch Consulting Employment Index, disaggregated by gender, is 92.4 for men and 97.1 for women.  Both the indices for men and women are down slightly from November.  Over the past nine months the men’s index is down 0.1% while the women’s employment index is down 0.8%.  Over the past two years the men’s employment index has grown by 1.4% per year and the women’s index has grown by 0.4% annually.

Welch_Index_Dec

Welch_Index_gender_Dec

Technical Note: Full-time equivalent employment equals full-time employment plus one half of part-time employment from the BLS household survey.  The Welch index adjusts for the changing age distribution of the population by fixing the age distribution of adults to the 1997 base year.  The Welch Index adjusts for population growth by fixing total population to its 1997 level.  Seasonal effects are removed in a regression framework using monthly indicator variables.

Williston, North Dakota: Boom Town

Employment is growing at a faster rate in Williston, North Dakota than anywhere else in the U.S.  Williston is located in northwest North Dakota, about 70 miles south of the U.S.-Canadian border.  Northwest North Dakota is the center of the Bakken oil boom, which has dwarfed any other energy booms in North Dakota and Montana.  The employment boom in Williston over the past two years is one of the most dramatic since the U.S. Department of Labor began collecting local employment and unemployment data.

Williston is a small city (a micropolitan area according to the Census Bureau) where employment grew modestly from 1990 to 2005 and by 10% per year from 2005 to 2010, despite the global recession.  Growth in Williston really began to take off about two years ago.  The unemployment rate in Williston is 0.7%, or 1/11 of the U.S. unemployment rate.  Over the past two years employment has grown by more than 40% per year which is 26 times faster than in the rest of the U.S. and over 29 times faster than growth in Williston prior to 2005.  To put this into perspective employment in Williston is now increasing by the same number of jobs every four days that used to be created each year from 1990 to 2005.

The following chart shows that employment has doubled in Williston in the past two years.  Between 1990 and 2005 employment grew in Williston at an annual rate of 1.37%.  Over the past two years employment in Williston has grown by 2.86% per month.

Willis1

There are growing pains associated with the employment boom.  The latest data from the Williston police department indicate that both property crimes and violent crimes have increased substantially in Williston, but roughly in proportion to the increase in employment.  The following charts compare the growth in crimes reported by the Williston Police Department to the growth in employment.

Willis2

Willis3

Williston is the quintessential energy boom town.  Employment is growing at an astronomical rate.  Wages are also rising — starting pay for high school graduates now tops $50,000 in the energy boom areas in North Dakota and Montana.  The average weekly wage in North Dakota has risen by 27.1% over the past three years, or 5.2 times faster than in he U.S. overall.  As employment has grown so quickly in such a short time the cost of rental housing has soared and the number of violent and property crimes have increased substantially.  These are just some of the growing pains associated with an economic boom.

The Welch Consulting Employment Index Continued to Rebound in October

The Welch Consulting Employment Index is 94.8 for October 2012, up 1.4% from October 2011 (seasonally adjusted).  An index value of 94.8 means that full-time equivalent employment (from the BLS household survey) is 5.2% below its level in the base year of 1997, after adjusting for both population growth and changes in the age distribution of the labor force.  The index is at its highest level since March 2012.  The index is 2.0% higher than its level two years ago and is 2.9% above its lowest level ever reached in July 2011.  The index had fallen sharply in the spring and early summer but has recovered all of those losses over the past two months.

The Welch Consulting Employment Index, disaggregated by gender, shows that the labor market recovery has been stronger for men than for women since 2009.  The index for men is 92.6 for October 2012, up 1.9% over the past twelve months, and up 3.0% since October 2009.  The index for women is 97.5 for October 2012, which is down slightly over the past month, and up much less (0.9%) over the past twelve months and past three years (0.7%) than it has been for men.  Despite the difference in trends over the past three years favoring men, the index for men remains well below the index for women.  This is because men lost many more full-time equivalent jobs (a decline of 9.5%) than women (a decline of 5.0%) from the beginning of 2008 through the fall of 2009.

Technical Note: Full-time equivalent employment equals full-time employment plus one half of part-time employment from the BLS household survey.  The Welch index adjusts for the changing age distribution of the population by fixing the age distribution of adults to the 1997 base year.  The Welch Index adjusts for population growth by fixing total population to its 1997 level.  Seasonal effects are removed in a regression framework using monthly indicator variables.

Why Even Mark Zandi Can’t Improve ADP’s Forecast of the BLS Jobs Report

This is the first month that the ADP national employment report is being produced in conjunction with Mark Zandi and Moody’s Analytics.  Previously, Macroeconomic Advisors helped ADP generate forecasts of the Bureau of Labor Statistics (BLS) nonfarm payroll employment report.  This month’s report was released with the announcement of a partnership with Mark Zandi and Moody’s Analytics and a newer and better methodology for predicting the key BLS jobs number.  The ADP report typically is released one or two days prior to the BLS report which is released the first Friday of each month.  ADP claims that, since 2001, the correlation between its forecast of the change in payroll employment and the BLS change in payroll employment is .96.  If this is true, as I explain below, the ADP report is either very close to being the best possible forecast of the BLS employment change or ADP has been very lucky in its forecasting over the past decade.  Mark Zandi and Moody’s Analytics will have a very difficult time improving on the previous track record for ADP.

Since 2001, the average change in BLS monthly payroll employment has been an increase of 7,000 employees per month.  There is considerable dispersion in monthly changes in employment, however.  The standard deviation of changes in monthly employment has been 241,000 employees per month.  The BLS reports that the standard deviation in monthly employment changes due to “sampling variation” is 61,000 employees per month.  In other words, even if the BLS drew an equally large and representative sample of employers and generated a second independent count of the monthly change in payroll employment, it would not be surprising if the second calculation differed from the first by 61,000 employees.

By converting these standard deviations into variances, and taking a ratio of sampling variation to total variation, it follows that 6.4% of the month-to-month variation in BLS payroll employment changes is due to “noise” or “sampling variation”, and 93.6% of the variation is due to “true” changes in employment.  Even if the BLS drew an equally large and representative sample of employers and generated a second count of the change in payroll employment in a given month, the correlation between the two different BLS employment changes would be .968.  (Because a correlation coefficient is simply the square root of the percentage of variation explained, and .968 is the square root of .936).

This means that even if Mark Zandi had access to the same data as the BLS, it is unreasonable to expect a forecast to be correlated more closely than .968 with the employment change reported by the BLS.  A forecast of the BLS change in payroll employment can’t be more reliable than the report itself.  The BLS acknowledges that their reports are imperfect.  Even if the BLS could make repeated independent calculations in the same month the correlation between independent BLS calculations would be only .968.  Consequently, if in the future ADP claims that their new model generates forecasts that are correlated .97 or more with the BLS jobs report we should all be dubious.

Losing Ground to Great Britain

It is hard to admit it, but Great Britain is doing a much better job at creating jobs than the U.S.  That has not always been the case, but it’s been true for the past decade.  What’s even worse is that we are falling further and further behind our British friends.  The United States economy is still the largest in the world, but we are no longer the preeminent job creator in the developed world.  While it is understandable that our job growth is slower than in China and parts of the developing world, there is no reason why employment in the U.S. should be falling relative to employment in Great Britain.

The latest employment figures were just released by the UK’s Office for National Statistics.  About 71.3% of adults age 16 to 64 were employed in Great Britain, in the three-month period from June to August.  That represents a one percentage point increase from the ratio in June to August 2011 (70.3%).  Moreover, this means that the employment to population ratio, the preferred measure of labor force activity by most labor economists, is almost 4 percentage points higher in Great Britain than in the United States for adults age 16-64.

The following chart shows the employment to population ratio for adults in the United States than Great Britain from 1984 to 2011.  The employment rate in the U.S. was higher than in Britain in 17 of 18 years from 1984 to 2001.  In a typical year 1.7% more adults were employed in the U.S. relative to Great Britain.  Since 2001 the Brits have been working more.  For 10 straight years the fraction of adults working in the U.S. has been lower than in Great Britain, and the gap is getting wider.

Changes over the past year have only widened the gap in employment rates between the two countries.  The employment rate increased by a full percentage point in Britain but by only half as much in the U.S.  Both rates are lower than they were prior to the recession but the U.S. employment rate is 4.1 percentage points lower while the British employment rate is only 1.6 percentage points lower than in 2005.  The employment rate of 71.3% in Great Britain and 67.4% in the United States means that 4 percent fewer adults are working in the U.S. than in Great Britain.  This translates into 8 million fewer jobs in the U.S. than if our policies led to employment rates more comparable to the U.K.

Economists and pundits in Great Britain have complained that fiscal austerity has reduced growth in their output and employment.  Whether or not these criticisms are valid, Americans would be happy to have the same record of job creation over the past four years as the British.  The decline in our employment rate in the past four years is troubling.  The employment rates illustrated above exclude individuals age 65 and above, so this is NOT due to the aging of the baby boom.[1]  It should not be too much to expect the employment rate in the U.S. to equal the rate in Britain, and that would mean eight million more jobs.  Eight million more jobs would solve a lot of problems in this country concerning the deficit, poverty and the long-term unemployed.  The government does not create jobs but tax and regulatory policy can get in the way.  Let’s hope that the presidential candidate who wins in November can work with the Congress to reverse the course we are on so that the U.S. can once again reclaim its position as the biggest job creator in the developed world.


[1] If one compares overall employment rates across countries it should be noted that there are actually a higher fraction of adults age 65 and above in Great Britain than in the U.S.

%d bloggers like this: