Lessons from Elkhart, Indiana

In the 2008 presidential campaign then-candidate Barack Obama visited Elkhart, Indiana twice.  He visited again as president in February 2009 to make the case for his $800 billion stimulus package.  President Obama came to Elkhart three times because the city was in the midst of an economic freefall and seemed to epitomize the plight of the U.S. economy in 2008-2009.  Although Elkhart’s unemployment rate was 4.6% in 2007, by March of 2009 it would reach 20.2%.  Elkhart has bounced back from the recession much better than other cities.  Its unemployment rate is now 8.3%, slightly above the national average but dramatically lower than it was just three and a half years ago.

Indiana is the most manufacturing-intensive state in the U.S. with one of six private sector workers employed in manufacturing.  In addition, over 7 out of 10 manufacturing jobs in Indiana are in the durable goods sector.  Arguably, no city in the U.S. is more dependent on the production of durable goods than Elkhart, which has been called the recreational vehicle capital of the world.  While only 6.7% of private sector jobs in the U.S. are in durable goods manufacturing in Elkhart 43.6% of private sector workers are employed in the manufacture of durable goods, with the majority of those jobs in the production of recreational vehicles and motor vehicle parts.  Elkhart has been one of the top job creating cities in the U.S. since the recession ended.  In the past three years jobs in durable goods manufacturing have increased by 42.7% in Elkhart compared to 5.6% growth in the U.S. overall.

As a durable goods manufacturing center Elkhart’s downturn and recovery looks much more like the typical pattern for a deep recession followed by a brisk recovery.  Other cities hurt most by the recession (such as Riverside, CA and Las Vegas NV) suffered collapses in their residential real estate markets and residential investment.  These cities have not yet experienced the rebound seen in Elkhart.  The following chart indicates how the unemployment rate in Elkhart spiked much above the rates in Las Vegas and Riverside but has fallen more quickly since then.


There has been much discussion about why this recovery has been so slow and so weak.  Reinhart and Rogoff relied on cross-country and historical comparisons to argue that a slow and weak recovery is to be expected after a systemic banking and financial crisis.  It is difficult, however, to make consistent cross-country comparisons of unemployment rate fluctuations because of differences in the way each country measures unemployment.  Bordo and Haubrich limit their analysis to American historical data.  They find that “general recessions associated with financial crises are generally followed by rapid recoveries.”  They also conclude that one reason for the slowness of the recovery “is the moribund nature of residential investment.”

Another way to understand the 2008-2009 downturn and the subsequent recovery is by comparing unemployment rate fluctuations across U.S. cities.  Such a comparison shows that the unemployment rate has remained stubbornly high in cities where the housing bubble burst.  Durable goods manufacturing centers like Elkhart saw a big drop in the demand for the products they produce and a large increase in unemployment during the recession.  But manufacturing-intensive cities have recovered somewhat more rapidly – more typical of previous downturns.  That gives the edge, for the time being, to the explanations provided by Bordo and Haubrich.

Comparing Half-time in America to Morning in America

The Bureau of Labor Statistics just released another solid jobs report, for February.  The labor market is recovering from the recession, but payroll employment is growing much more slowly than it was in early 1984.  Moreover, the unemployment rate remains stubbornly high relative to 1984 and previous recoveries.

Payroll employment grew by:

  • 926,000 jobs in the first two months of 1984 or about 1.0 percent
  • 511,000 jobs in the first two months of 2012 or about 0.4 percent

Employment grew 2.5 times faster in early 1984 than in early 2011.  Payroll employment needed to increase by 1.33 million jobs rather than 511,000 jobs in the first two months of 2012 to be comparable to 1984.

The current unemployment rate of 8.3% is slightly higher than the unemployment rate of 7.8% in February, 1984.  The natural unemployment rate in 1984 was higher because the workforce was less mature.  Younger and less experienced workers tend to have higher unemployment rates because they are often switching jobs.  It is therefore more accurate to compare unemployment rates by age group.

The table below shows that, age-group by age-group, the unemployment rate in 2012 is 0.7% to 4.4% higher than in 1984.

Age Group

Unemployment Rate February 2012

Unemployment Rate February 1984






























*Not Seasonally Adjusted for these Age Groups

 The unemployment rates of young workers and seniors are much higher today than in 1984.  The unemployment rate of prime working age adults, age 35-54, is about 1.5% higher than in 1984.

Despite a few solid jobs reports, the labor market recovery remains weak by historical standards.  Employment growth is not as robust, unemployment rates are higher, and a higher fraction of the unemployed have been out of work for at least six months compared to earlier recoveries

The Frozen Labor Market

The labor market recovery of the past two years ago has not gained enough strength to rule out the possibility of a lost decade.  The recovery is especially weak for less educated workers and for cities devastated by the housing market collapse.  The economy is in better shape than three years ago but we would need to add 10 million full-time jobs to return to pre-recession employment to population ratios.  It is doubtful that 2012 will be much better than 2011 because the labor market appears much less dynamic than it was just a decade ago.

The recovery is weaker than the headlines suggest.  The unemployment rate has fallen by 1.4% over the past two years but this is largely due to workers leaving the labor force; the labor force participation rate has declined by 1.1% since January 2010.  The participation rate has been trending down for years so a small part of this decline was expected.  Nonetheless we should not be encouraged by declining labor force participation during a recovery.

The following figure presents a disconcerting trend: hires and separations have fallen steadily relative to the size of the labor force over the past decade.  The labor market is much less dynamic than it was prior to the Great Recession.  Although companies are laying off fewer workers, they are also hiring fewer workers, and employed workers are reluctant to quit.  Consequently new labor force entrants and job losers face fewer job opportunities and struggle to find a job.

Gross labor market turnover, measured by the sum of hires and separations per labor force participant, has fallen 25% since 2002.  This indicates a labor market that is more sluggish and less dynamic than it was just a decade ago.  Companies may be reluctant to hire because they are uncertain about: (i) the strength of the economic recovery, (ii) health care costs, or (iii) whether temporary tax code provisions will be extended or curtailed.  Hiring rates may also be weak because jobless workers lack the skills and experience that companies demand, or because start-ups can’t obtain financing for their new ventures.  A robust recovery must include much more hiring so that enough jobs are created to employ new graduates and the long-term unemployed.  This won’t happen if the labor market remains as sluggish as it has been since 2008.


We are producing more with fewer workers; U.S. employment is below the pre-recession peak of January 2008 but output (real GDP) is 1.2% higher.  Employment in the private sector and state and local government has declined over the past four years while Federal employment has increased.  Since January 2008:

  • Private sector employment declined by 4.5%
  • Federal government employment increased by 3.4%
  • State government employment declined by 1.6%
  • Local government employment declined by 2.8%

Is the decline in state and local government employment a drag on the economy as Paul Krugman has argued?  Or have state and local governments, like the private sector, become more efficient and deliver more and better quality services with fewer employees?  This is impossible to tell from National Income Accounts data which only measure the cost of government spending and not the value of the government services provided.

It is important to put the recent declines in state and local employment into perspective.  In the past fifteen years local government employment has grown almost twice as fast (17.5%) and state government employment has grown slightly faster (10.3%) than private sector employment (9.1%).  State and local government has grown relative to the private sector for decades.

The recovery/stimulus legislation was designed to bolster state and government spending.  However, as John Taylor testified to Congress, state and local governments used stimulus funds largely to reduce borrowing rather than increase expenditures.

Paul Krugman argues for more state and local government spending on goods, services and investment.  (He admits that “safety-net spending … has soared in this slump.”)  Keynesians believe that if spending were $340 billion higher (about 2% of GDP), GDP would be 3% higher due to the “multiplier” effect and the unemployment rate would be 1.5% lower.   A “multiplier” argument is a smokescreen for the real debate about the appropriate size and scope of government.  Some state and local government spending/employment cuts make sense if they eliminate waste and duplication but that is the opposite of what “multiplier” calculations would conclude.

In addition many leading economists such as John Cochrane and John Taylor are skeptical of large multipliers for stimulus spending.  More importantly, the “multiplier” argument says nothing about which programs should be expanded and whether any programs should be cut.  The emptiness of a “multiplier” justifcation for government spending is clear when one recognizes that Keynesians believe that government reductions in waste and fraud would lower GDP and raise the unemployment rate.

One of the most important election issues is the debate over the proper size and scope of government.  The debate would be more informative if it focused on the direct costs and benefits of government programs, assumed that programs must be paid for even if financed through bonds, and did not rely on possible “multiplier” effects to justify spending.

New Jobless Claims Cast Doubt on Recovery Winter

Everyone seems to be impressed by the low levels of new unemployment insurance claims that have been filed in 2012.  I look at the numbers and notice how high new claims are relative to layoffs.  The data suggest that nearly everyone who has lost their job through a layoff or reduction in force files for unemployment benefits.  Before the recession, that was not the case.  Job losers are much more pessimistic about their chances of finding a new job than they should be if the economy was enjoying a recovery winter.

A little over two million new claims were filed in the first four weeks of the year.  That’s about 12% less than were filed over a comparable period in 2011, and 27% less than in 2010.  I am not surprised by this decline because there must be layoffs before there can be new jobless claims.  Unemployed workers must be job losers to be eligible for unemployment insurance.  The BLS labor turnover survey (JOLTS) indicates that layoffs have fallen even more dramatically than new jobless claims since the recession.

The following chart shows new unemployment insurance claims and layoffs each month between 2001 and 2011.  The figures are not seasonally adjusted – layoffs and UI claims spike up in January and are lowest in the fall every year.

Between 2001 and the fall of 2008 there were 19% fewer jobless claims than there were layoffs.  A disturbing pattern has emerged beginning in the fall of 2008.  Since November of 2008 about 11% more people filed for UI benefits than there have been job losses through layoffs.  Some of the excess new claimants were likely deemed ineligible for unemployment insurance.  Even so, the fraction of laid off workers who rely on unemployment insurance is at an all-time high.  This suggests job losers are pessimistic about their prospects of finding a new job and is a leading indicator that, despite what some are saying, it will be a while before hiring picks up.

Part-Time Recovery

The unemployment rate and the monthly change in total payroll employment are clearly the most widely watched labor market indicators.  Neither of these statistics measures an important consequence of the 2008 recession: adult men are working part-time at record rates.  As the labor market turns the corner in 2012 the most important leading indicator may be the fraction of adult men employed in part-time jobs.

The following figure shows that the fraction of employed men who work part-time nearly doubled between 1986 and 2011.   By 2011 about one third of employed men age 20-24 worked part-time, and 7.4% of employed men age 25-54 held part-time jobs.  About 80% of the increase in part-time work for the 25-54 age group occurred during the 2008 recession.  About 2.5 million full-time jobs were lost during the recession for men in the 20-24 and 25-54 age groups.  It is less well known that the recession caused about 1.5 million adult men in these age groups to switch from full-time to part-time work.  The patterns are somewhat different for women, but I will leave that for another post.

The following chart shows that the overall employment to population ratio dropped by 11.4% for men age 20-24 and 6.7% for men age 25-54 between January 2008 and January 2010.  These large employment declines understate the depth of the downturn because they treat all jobs the same, whether they are part-time or full-time.  The full-time employment to population ratio dropped by 13.8% for men age 20-24 and 8.9% for men age 25-54.  These precipitous drops in full-time employment rates were accompanied by increases in part-time employment rates of just over 2%.

The labor market has been slowly recovering since January of 2010.  The following chart shows that the overall employment to population ratio for men age 20-24 grew by 4% over the past two years.  The length of the workweek did not change much for these younger men because about two thirds work full-time, and full-time employment growth was about double part-time employment growth.

The employment to population ratio of men age 25-54 has grown by 1.5% since January of 2010.  The full-time employment rate increased by 2% while there was a slight decrease of 0.5% in the part-time employment rate.  The length of the average workweek increased slightly over the past two years these men as some part-time jobs were replaced by full-time employment.

A necessary component of a solid recovery is the transition to full-time work for the (at least) 1.5 million adult men who work part-time jobs because of the weak economy.  This change will not appear as an increase in payroll employment or a reduction in the unemployment rate.  A key leading indicator that that the economic recovery is finally gaining strength will be more substantial decreases in the fraction of adult men who work part-time.

Turning the Corner

When I saw last Friday’s encouraging jobs report I knew it would be controversial.  An increase in payroll employment of 243,000 is good news, but any January report contains a large seasonal adjustment because it is typically the weakest month of the year for employment.  The Bureau of Labor Statistics generates a new seasonal adjustment factor every month, to allow for changing economic conditions, but it means that no two January adjustments are the same.  Skeptics, such as Zero Hedge, correctly observed that the past two January seasonal adjustments have been especially large which might account for 100k of the job gain reported last week.

The Bureau of Labor Statistics is faced with an incredibly challenging task.  It must generate a near real-time count of the country’s total payroll and report it as if January was no different than June.  I know things are out of hand when Rush Limbaugh and Rachel Maddow are commenting on the appropriateness of the government’s seasonal adjustment process.  It is unfortunate that last week’s solid report was obscured by an opaque statistical methodology.   I prefer a more transparent method for presenting high-frequency changes in payroll employment.  My approach shows slow and steady improvement in the aggregate labor market throughout 2011.

First, I consider average payroll employment over a quarter year to mitigate the noise in any single month’s report.  Second, I compare year-over-year percentage changes in quarterly employment rather than use a confusing and complicated method for removing seasonal effects.  The following figure presents these changes in payroll employment for all of 2011 and January 2012.

Employment grew by 1% annually in the first half of 2011 and by 1.2% and 1.3% in the third and fourth quarters.  The January jobs report is encouraging because it reflects a 1.5% annual employment growth rate.  Of course the next two monthly reports will have to be equally strong to maintain a growth rate of 1.5% for the entire first quarter of 2012.

Quarterly year-over-year changes eliminate some of the noise in monthly reports and their evolving seasonal adjustments.  Conventional wisdom, fueled by noisy monthly reports, is that the labor market recovery sputtered in the second half of 2011.  In fact, the labor market has been improving slowly and steadily.  There are problems to be sure.  Annual employment growth of 1.5% is better than we have seen recently but it’s painfully slow given the deep recession.  More importantly, as I will show in future posts, the labor market remains especially weak for less skilled workers and in areas hit hard by the real estate crisis.

Could it be the Weather?

Friday’s jobs report showed that the economy lost almost 2.7 million jobs between December 2011 and January 2012.  That’s pretty good for a January.  In the world of labor statistics and seasonal adjustments it translates into a gain of 243,000 jobs.  How does that work?  The Job loss of 2.02% in January of 2012 was better than what we have seen recently.  If we exclude the change from December 2008 to January 2009, at the depths of the recession, the average December to January change since 2006 was a decline in employment of 2.11%.

We learned on Friday that employment is 9/100 of one percent higher than it would be have been if January 2012 was like the typical January in recent years.  With total employment in the U.S. of approximately 130 million, this means there were 117,000 more people working in January than we would have otherwise expected.

What does this mean for our economic outlook?  It could be that employment was slightly higher due to an unusually mild winter.  It could be that fewer seasonal workers were hired heading into the holidays and laid off in January.  The January employment declines in retail and leisure and hospitality were unusually modest; 3.4% compared to the typical 3.6% decline.  These two industries account for half of the 117k additional jobs.

Friday’s jobs report may be signaling that the labor market recovery is accelerating as we head into 2012.  Or it may indicate that consumer spending was a little higher during an unusually mild January.  If the latter is correct we should be prepared for smaller than usual employment gains as we head from winter into spring.

Women and the Economic Recovery

News media coverage of the 2008 recession and the subsequent recovery has focused on changes in total employment and movements in the overall unemployment.  Although the change in the total number of jobs is an important economic indicator, a focus on overall employment and unemployment can mask important changes in the composition of the workforce.  I look beyond changes in total employment to illustrate why the 2008 recession is unlike any in U.S. history.  Job losses in recessions previously were concentrated in traditionally male-dominated jobs.  The 2008 recession was different.  Women lost jobs in record numbers and women have seen small job gains during the recovery.  This stands in stark contrast to previous recessions and recoveries.

The following graph illustrates changes in the full-time equivalent employment (a part-time job is counted as ½ of a full-time job) of men.  Four years after total employment started to decline, men’s employment recovered and grew to 4 to 5% above the pre-recession peak in the 1970’s and 1980’s, but only about 1% above the pre-recession peak after the 1990 and 2000 recoveries.  The recovery over the past two years has been especially weak.

The U.S. labor force changed dramatically during the 1970s and 1980s.  The labor force participation rate of women increased from 43.3% to 57.5% between 1970 and 1990 but has leveled off since then.  The widespread reallocation of women’s human capital from household activities (not included in GDP) to paid market work had a profound impact on measured employment and GDP.  Employment and GDP increase, even though the same work is being done, when house cleaning and lawn care tasks are completed by paid service workers rather than by family members.

Cyclical employment fluctuations look very different for women.  The robust labor market recoveries of the 1970’s and 1980’s were fueled by the entry of women into the labor force.  Women’s employment grew by 15% between 1974 and 1978, and by 10% between 1981 and 1985, despite the deep recessions.

Job losses during recessions were largely confined to men until the 2008 downturn.  Women’s employment never fell by more than 1% in any recession prior to 2008.  In contrast, women’s employment fell by 5% between January 2008 and December 2009, and is still 4.6% below the pre-recession peak.

Future economic recoveries are unlikely to exhibit the robust employment gains that occurred during the 1970’s and 1980’s as millions of women moved from unpaid household work to jobs in the market sector.  This type of transformational change is unlikely to occur again.

The 2008 recession is unique because of the magnitude of employment losses sustained by women (although still smaller than for men) and the sluggish growth in women’s employment during the recovery.   In the past year both employment and labor force participation declined for women but increased for men.  Despite the labor market gains achieved by women over the past forty years, women face a somewhat new challenge – coping with substantial job losses in an economic downturn.

We’re Too Old For This Recession

The outcome of the 2012 election may well depend on the unemployment rate in November.  President Reagan was re-elected in 1984 when the unemployment rate was 7.2%, but Presidents George H.W. Bush and Jimmy Carter were defeated when unemployment rates were 7.4% and 7.5% respectively.

It is misleading to compare today’s unemployment rate to the rate 30 years ago because today’s labor force is older and more experienced.  The unemployment rate was high during the early 1980s in part because many baby boomers were in their 20’s when unemployment rates tend to be high.  The declining unemployment rate in the late 1990’s and early 2000’s was aided by baby boomers entering their prime working ages.

Voter sentiment in November may depend on whether unemployment rates by age group are unusually high.  To make this comparison, I calculate an overall unemployment rate that holds constant each age group’s share of the labor force.  Using this method, I find that age adjustments increase last month’s 8.5% unemployment rate to 9.1%, while the “age-adjusted” unemployment rates when Carter, Reagan, and George H.W. Bush faced re-election were 6.8%, 6.8%, and 7.5% respectively.

Unemployment rates in the early 1980s were “inflated” by young baby boomers entering the labor force.  Although last month’s 8.5% unemployment rate is high by historical standards, it would be even higher if it weren’t for all the baby boomers now in their 50’s and 60’s (when unemployment rates tend to be low).  A worker today faces an unemployment rate that is 2.3% higher than someone the same age faced in November 1980.

This means that the recession of 2008 is even more severe than we first thought.  The unemployment rate has averaged 8.4% over the past four years which appears lower than the 8.6% average rate during the 1981 recession, but really isn’t.  In fact, the average “age-adjusted” unemployment rate over the past four years is 8.9%, which is much higher than the average 8.1% adjusted rate during the 1981 recession.

The following graph compares unemployment rates for five recent recessions and recoveries, after adjusting for workers’ ages.  It is clear that the 2008 recession has been the deepest and longest downturn in decades.  The “age-adjusted” unemployment rate exceeded 10% for 19 straight months in 2009-2011 and remains stubbornly high.

There is no doubt that the 2008 recession is more prolonged and severe than any economic downturn since the Great Depression.  The current 8.5% unemployment rate is high by historical standards, even if I don’t adjust for the graying of the labor force.  I find that the “natural” rate of unemployment for today’s relatively experienced labor force is about 1.2% lower than it was in 1980.  Therefore unless the unemployment rate in November 2012 is 6.3% or less, voters will face a higher unemployment rate this fall than voters the same age faced when Ronald Reagan defeated Jimmy Carter.

%d bloggers like this: