America’s Lost Decade Already Happened

Once again, the United States faces the prospect of a jobless recovery.  Next month, the President is holding a conference on job creation.  The President seems genuinely distressed over the jobs situation.  He is seeking real answers.  He is asking the wrong questions. 

The President and his advisers view the job losses as a temporary cyclical problem, not as a structural problem with the U.S. economy.  Quick fixes for temporary downturns in the labor market are expensive but easy.  Last year (here,  scroll about half way down), I suggested a temporary social security tax holiday to boost employment and output.  Last week, Alan Blinder took up the call in an editorial in the Wall Street Journal.  At a cost of about $100,000 per job (I am not as optimistic as Blinder), the administration can create several million temporary jobs.  For far less money, the government could simply hire the same people at an average price of less than $40,000. 

Ultimately all of these programs are doomed to failure.  Quick fixes cannot solve structural problems, and a structural problem exists.  Before a solution can be devised, we must understand the source of the problem.  As of right now, the source is unknown and finding the source probably requires more data than is readily available on the structure of the job losses.  But we can make a start with the data in hand. 

In this post, I break out some of the key data, documenting the sectors of decline and the sectors of growth.  In my next post, I will try to explain the loss.  I will not succeed but perhaps we can get closer to an answer. 

The Problem

Between January 2000 and October 2009, employment in the United States rose by a paltry 67,000.  Including the already announced benchmark revision, the United States lost 757,000 jobs.  This loss was the first over a ten-year period since the Great Depression.  The post-war, record-low, ten-year job growth was 5.7 million achieved in December 1962. 

The job gap (the difference between the growth of the working age population and the growth in jobs) now exceeds 16 million.  The job gap in 1962 was 3.5 million jobs.  As a percent of employment, the current gap is 12 percent, the 1962 gap 6 percent, and the maximum gap during the Great Depression 19 percent.

And, the job gap is not merely an artifact of the current downturn.  At the peak of the labor market in December 2007, the job gap had already grown to more than 7 million jobs.  The economy was producing enough jobs to eventually overcome the difference, easing concerns into complacency, but the gap was still large.  The current downturn has simply exacerbated and made obvious an already existing problem. 

The current gap will almost certainly still exist when my fifth-grade daughter enters the labor force.  If we create 181,000 jobs per month (the average monthly job growth in the 1990s), the gap would take more than 12 years to close, given current projections of the growth of the overall workforce.  Even if job growth could be sustained at the maximum annual pace observed between 1940 and 2009 (392 thousand jobs per month), the jobs gap would persist for more than five years. 

Aren’t the losses “just” manufacturing jobs?

Over the past decade, job losses have been concentrated in manufacturing.  The goods-producing sector shed more than 6.3 million jobs as the Services sector added a roughly similar number.  In the middle of the last decade, many economists favored the idea that the loss of manufacturing jobs was a result of rapid productivity gains in the services sector.  They favored the idea that the loss of manufacturing jobs was part of the long evolution of modern economies:  from agriculture to manufacturing, from manufacturing to services.  The job losses were nothing more than a continuation of a long-term trend. 

The basic fact is verifiable.  As a share of output or employment, the services sector has outpaced the manufacturing sector since the early 1950s.  A shift to capital intensive technology in the manufacturing sector and rapid productivity gains in the services sector pushed and pulled workers.  The services sector was pulling labor out of the manufacturing sector, offering higher wages and easier work.  Between 1950 and 2000, the United States became an economic superpower and no one may question the relative change in our standard of living.  We survived the manufacturing losses and became stronger because of them. 

But, in the last decade, something changed:  manufacturing pushed workers out faster than the services sector could pull them in.  See, even though the share of manufacturing workers was falling, in absolute terms it was stable.  The level of manufacturing employment was approximately equal in 1950 and 2000. 

Job losses in manufacturing have been widespread.  Every major sub-category of manufacturing employment has declined.  Transportation, metals, computers, machinery, textiles, apparel, plastics, and printing have each lost more than 300,000 workers.  (I was amazed by the textiles figures.  When I went to NC State back in 1995, the textile industry was already rumored dead.  How can a dead industry still be shedding jobs 15 years later?)

Manufacturing is sick:  No surprise. 

Won’t Services save the day?

Maybe.  The services sector added almost 6.4 million jobs over the last decade.  The possibility that we remain in a period of rapid transformation exists.  It could be true, but the data stacks against the hypothesis. 

As manufacturing losses are broad, services gains are narrow.  Three sectors account for more than 100% of the service-sector gains.  The following chart shows the change in employment by major sector between January 2000 and October 2009.  Health led the charge.  The health services industry added more than 3.5 million jobs.  Part of this gain was funded by an expansion of Medicaid, but the gains look good.  Health is a growth sector. 

Unfortunately, the rest of the story is not as rosy.  Government employment accounted for one-third of the gains, 1.8 million people.  Worse, already in 2000, the government was the largest service-sector employer.  Government jobs do not promote growth.  Government jobs must be funded from other sectors.  Government jobs seldom innovate.  

Are the losses simply a symptom of an emerging new economy?

In the 1990s, the innovative services sector pulled workers in, but it did not pull just any worker.  Young workers are more flexible and adapt to new industries quicker than old workers.  They more flexible because they have less to lose in leaving the old industries; they have not yet built a stock of industry-specific capital.  The dynamic led to an increase in participation for workers younger than 45 and a decrease in those who were older.

Over the past decade the trend has reversed and it has reversed to a shocking extent.  The following graph shows the participation rate of workers by age group.  In all groups younger than 55, the participation rate has fallen.  In all groups over 55, the participation rate has risen. 

In 2003, the social security retirement age increased by a year, and that change may explain a bit of the shift.  But the change does not explain the increase in participation of the 70+ crowd.  The increase in 75+ almost doubles the participation rate of the very old.  The increase is not driven by better health, ten years is too short to effect that change. 

No, the data indicates two things:  older workers feel poorer and are working longer, and young people can’t get jobs.  (I am including everyone below 55 in the young category.  I will let you know when I need to change the definition again.)  I am surprised by the breakdown.  I had thought the picture would tilt the other direction, at least for prime-age workers (25-45). 

Push and Pull

The manufacturing sector pushes workers out whenever its productivity growth is sufficiently high.  The symptom of this push is rising output with or without gains in employment.  Since the 1950s, manufacturing output has increased steadily despite stagnant employment.  Beginning in 2000, this stopped being true.  Even the surge between 2004 and 2006 was mild.  The maximum growth rate of manufacturing output between 2001 and 2009 was lower than its average growth rate between 1992 and 2000. 

Likewise, the services sector pulled workers with high rates of productivity growth.  And, these high rates persisted over the last decade, but the average growth rate was a full percentage point lower in the last decade relative to the 1990s.  The services sector is not growing fast enough on average to offset manufacturing losses. 

Takeaways

The more I look at the data, the more I am convinced that the current downturn and the downturn in 2000 are related episodes.  At least from the labor-market perspective, they seem closely connected.  Treating the current job losses in isolation is a mistake.  Whatever forces devastated the labor market in the early 2000s remain in effect today. 

The President wants to solve the “jobs problem”, of this I am certain.  We are lucky to have a President who genuinely wants to solve the problem.  But his advisers do not seem aware of the structural issues.  The first step is admitting the problem.  Economists, especially those named Christina Romer or Larry Summers, need to stop thinking of the job losses as an unavoidable bad outcome associated with all recessions.  This time really is different.

Resolving the Foreclosure Crisis: What Can We Do?

Almost two years have elapsed since the beginning of the recession, but the foreclosure crisis continues.  At the end of the second quarter, residential mortgages held by commercial banks reached a new record default rate of nearly 9 percent, and according to recent news reports, this number continued to rise through the third quarter.  The high default rates do not owe to subprime borrowers alone.  The majority of delinquencies are no longer subprime mortgages.  That honor now belongs to prime and near-prime mortgages. 

The crisis has continued despite innumerable programs, at both the state and federal level, to alleviate the crisis.  The FHA has introduced a string of programs to help homeowners refinance into more affordable loans.  Together, various supervisory authorities have put pressure on banks and lending companies to modify the terms of residential mortgages—more than 1.5 million mortgages have been modified in the last year.  Several states have tried temporary foreclosure moratoriums.  In two indirect efforts to reduce foreclosures, the Federal Reserve purchased close to $800 billion in mortgage-backed securities and Congress passed an $8,000 new home-buyer tax credit. 

None of these programs has worked because none attack the source of the problem.  Households owe more on their mortgage than their home is worth.  They have economic incentives to default. 

Background

The mortgage crisis started because too many households borrowed too much and bought houses they could not afford.  Real money flowed into the housing market and residential investment increased. Because we couldn’t build enough, especially in urban areas, the price of houses rose. In this sense, we had borrowing bubble not a housing bubble.  While subprime mortgages are the poster children of the borrowing bubble, prime mortgages also flowed freely.

The borrowing bubble has burst. The money that flowed into the housing market is gone. There are now too many houses, and house prices have to fall.

This adjustment process is well under way. Housing starts have fallen off a cliff, and house prices are falling. Prices are anywhere between 5 and 40 percent below their peak depending on where you live and which measure you believe. They are going to be lower.

Falling prices are the root cause of mortgage foreclosures.  The more prices fall, the more households are under water (the value of the mortgage exceeds the value of the home).  Under water households are more likely than any other class of borrower to default.  They may continue making payments for a time, but the incentive to make payments is diminished.  If their house price falls more or if they lose even a little bit of income, they are likely to default.  From the household’s perspective, this default can be optimal. 

The relationship between being under water and defaulting is simple and direct.  Think of a household that has lost its income.  Households that have sufficient equity in their homes will always, under this circumstance, choose to sell the home (even at a loss) rather than undergoing foreclosure.  By selling, these households profit financially and socially—they go forward with their credit unblemished.  Without sufficient equity, they do not have this option.  Without income they cannot make payments.  But, they also cannot sell because they owe more than the home is worth.  Default is the only option.

The Plan

Any successful foreclosure-reduction plan must address under-water households first.  Reducing a household’s monthly mortgage payment reduces the probability of default but only slightly (see this post).  For most under-water households, a lower interest rate or an extended term does not change the default incentives. 

My plan is simple:  a voluntary program to purchase all mortgages with a loan-to-value ratio greater than 90 percent and issue a new mortgage with a 90 percent LTV to the household.  To reduce moral hazard and capricious program take up, issue an equity claim with the new mortgage.  Optimally, this claim amounts to 30 percent of the difference between the ultimate selling price and the origination value of the mortgage. 

The devil is in the details, but this plan would solve the foreclosure crisis. 

A Few Details

Why 90 percent?  This ratio gives homeowners an immediate financial stake in their property. A household with 10 percent equity does not have a financial incentive to default.  Even after paying closing costs and paying the equity claim (see below), selling is more profitable than defaulting.  A few households will still default, households make mistakes, but a few defaults are not a problem.

Why issue an equity claim?  The equity claim reduces the redistributive aspects of the program and reduces moral hazard.  As with any government intervention in markets, this plan is a transfer between households.  Responsible households are subsidizing the houses of those who over borrowed.  The program also encourages future households to borrow more in the hope they will receive a bailout if things go bad. Issuing an equity claim reduces both distortions.

Why 30 percent?  The equity claim recovers 30 percent of the difference between the origination value of the new mortgage and the eventual selling price of the home. With this percentage, the household stands to gain about 1 percent of the value of his home after closing costs at the time of origination.  That is, even at origination, the household has an incentive not to re-default.  If the claim were larger, households would remain under water, and the program would not meet with success.  Moreover, with a 70 percent stake in their property, homeowners have incentives to maintain their house.  

The equity claim, which is worth 3 percent of the home’s value at origination, also reduces take up of the program and provides an incentive for households in the program to increase the declared value of their property, thereby taking on a larger mortgage.  No one who does not need this program will voluntarily forfeit 3 percent of their home’s value.  

How do we determine home values?  The most difficult part of this plan is determining the house value.  But this is a macro program—macroeconomic policy is designed to care for the health of the economy not individuals—the program only has to be correct on average.  The current value of the home can be estimated by using the price of the home when it was last sold combined with the average change in house prices for the metropolitan area as measured by any good house price index.  The program should permit the household to increase but not decrease the current declared value.

Because no price index is perfect, especially when volumes are low, the government should reevaluate home values periodically, say quarterly.  If homeowners cannot sell their houses at the declared value, the declared value is too high and must be lowered.  If take up rates are high and homes are selling easily, the value may be too low.

Who would own the mortgage?  The mortgages would be held, initially, by the government.  But these notes do not have to be held.  Since the mortgages are now above water and since the government is committed to repeating the plan if values should fall, the mortgages could be sold to private investors.  If the plan is followed in its entirety, the mortgages would not even need a government guarantee. 

How much would the plan cost?  The cost of the program depends on take up rates, but it need not be expensive.  Outstanding mortgage debt increased almost $4 trillion between 2005Q1 and 2008Q2.  The average LTV for these mortgages was far less than 80 percent.  Most of these households remain above water; indeed, the best guess is that 3 out of 4 of these households remain above water.  With a 100 percent take-up rate amongst under-water households, the cost of the program would be roughly $100 billion.  Most likely, the final bill would fall between $50 and $100 billion.

Takeaways

This is the simplest plan for resolving the foreclosure crisis.  It requires little private information, and the government does not need to make an affordability determination.  The household’s income does not matter.  If the household can afford their payments, they will make them.  If not, they will sell, making a small profit. 

The plan has the added advantage of minimizing forward-looking housing market distortions.  Households can freely sell their house, and housing market adjustment is not hindered by negative equity households.  Labor market adjustment can also proceed.  An under-water manufacturing worker in Michigan is no longer forced to look for work in his local labor market alone.  He is free to sell his house and conduct a national search. 

My plan is scalable and can easily incorporate the good elements of other plans.  For example, with ongoing rising job losses, in some areas, too many houses could flood the market, hindering adjustment.  In this case, a temporary payment holiday for high LTV households or even an outright foreclosure moratorium could be combined with the mortgage purchase plan, effectively metering the flow of houses onto the market.  

No matter the bells and whistles, a successful program must ensure that households remain above water.

Households must have a stake in their property or they will default—at some point.

Separations and Hires: Has the Recovery Stalled?

This post revises and updates my views from this post last March. 

The JOTLS data (find the data here) produced by the BLS gives the best insight into the current state of the job market. As Robert Shimer, a professor at the University of Chicago, showed some time ago, unemployment can go up either because workers become more likely to lose their jobs (the separation rate) or because unemployed workers have a more difficult time finding new jobs (the hires or matching rate). The BLS only began collecting data in late 2000, much too late for us to compare the current downturn to previous episodes. Bob Shimer, however, has computed separation and matching rates going back to 1947 (his data is here). The data is not strictly comparable but I think we can use the lessons from Shimer’s data and apply them to the current episode.

I have spent a lot of time working with his data lately. The cyclical behavior of matching and separation rates is remarkable and should provide the key to the next level of understanding in business cycle research.  The more I work with this data the more I feel like I am beginning to understand consumer behavior during recessions. 

Matching rates, the probability of finding a job conditional on unemployment, begin to fall well before recessions begin and continue to fall well after the recession ends. Separation rates tend to rise at the beginning of recessions and tend to fall well before the end of the recession. Not surprisingly, the worst recessions in the post-war era (1958, 1982) are characterized by large changes in both rates.

In every post-war recession, the separation rate returned to more-or-less its long term average 4-to-6 months before the trough. The fall in separation rates also coincides with a rise in consumption. Apparently, consumption begins to rise once employed households no longer fear unemployment – a rational outcome. Consumption rises before unemployment falls.  Unemployed workers continue to have trouble finding work long after the recession ends.  But, their consumption is small and stable.  Employed worker consumption rises. 

As a result of this research, I am beginning to have more faith in the signals emitted by the JOLTS data. First, take a look at the picture below. The picture shows the number of hires each month in the JOLTS data from late 2000 to January 2009. Amazingly, the number of hires began to fall as early as January 2006, the same month the housing market turned sour.  This data is consistent with the duration of unemployment calculated from the household survey.  The average duration of unemployment is now at a record high, implying a record low probability of finding a job conditional on unemployment. 

 This is the clearest piece of data I have yet come across to indicate that the collapse of the housing market was not a random event. The decline in hire rates reduces the permanent income of households. People realize that conditional on losing their job, new work will be harder to find. Households also seem to know that this trend tends to have long cyclical properties – a decline in the series today is likely to signal a long period of increasingly lower matching rates.

Of course, I want to know if the recession is over, or if the recession has yet to end, when it is likely to end.  Take a careful look at the very end of the hires graph.  Hires spiked upward July but have since fallen back.  Granted the fallback is only two months worth of data, but it is consistent with a labor market that tried to improve and then suffered a setback.  This is consistent with employment data (discussed here) and it is consistent with the picture from the separation rate. 

As I showed in March, the separation rate the total number of separations has been steadily falling since early 2007. This data alone would indicate that flows into unemployment should be falling, quite the opposite of our experience over this period.  Again, note the July bobble in separations.

To understand the labor market, separations must control for the voluntary versus involuntary separations. If I quit my job today, knowing I had a new job in the bag, I would show up first as a separation then as a hire.  We care only about involuntary separation. To get a better picture, subtract the number of monthly quits from total separations.  The resulting picture, shown below, gives a completely different view of the state of the labor market.

The level of separations in January 2009 was 35 percent higher than its 2001-07 average level.  Keeping in mind that half of that time period was during bad labor markets, this statistic is quite stunning.  The labor market has improved since January.  However, the recovery seems to have stalled and over the past 4 or 5 months the number of involuntary separations has achieved a plateau 17 percent above pre-recession average. 

This plateau also indicates a recovery stalled.  While we do not have a sufficiently long time series to know the behavior of this series in previous recessions, Shimer’s separation rates fall sharply before the end of recessions and remain low thereafter.  The high level of involuntary separations is not consistent with recovery.  This data is giving the same signal as initial claims data.  Initial claims are down sharply from their peak but remain extremely high compared to their historic average. 

Casey Mulligan, a Chicago economist, notes in his blog (and more recently here) that consumer spending is rising as is disposable income even as the job market continues to deteriorate.  In particular, he has been keen on noting the ongoing increases in personal income.  He does realize that personal income includes transfers (at record highs) from the government.  I don’t think Casey Mulligan would really believe transfers accompanied by an increase in debt are an actual increase in income. 

Nonetheless, even as current income continues to rise, the high separation and low matching rates have sharply reduced permanent income for households – they are faced with an ongoing high probability of job loss and amazingly low odds of getting a new job if they become unemployed.  And, labor income is far and away the largest portion of permanent income for the vast majority of Americans.

The Unemployment Rate: Moderation through Participation

Here is a long-form answer to NorthGG’s recent comment.
In October, the unemployment rate breached double digits for the first time since 1983. This number, 10.2 percent, seems bad—one out of every ten Americans is out of work but the number is deceptively benign. In this recession, more than at any other time since the early 1970s, declines in labor-market participation are moderating the unemployment rate.

The unemployment rate including all workers who have left the labor force in the last year is currently about two percentage points higher than the official rate. That is, the drop in participation is currently contributing about 2 percentage points to the unemployment rate. Under this measure, the unemployment rate is currently at a record high.

The contribution from flows out of the labor force is about the same as in the early 1970s. There is an important difference in the current situation and the difference is critical for the long-term outlook for the U.S. economy. In the early 1970s, the baby boomers were just entering the workforce. The drop in participation came as boomers left the labor market to remain in school. They went to school both for economic reasons and to avoid the draft. But, this education pool of workers has been a boon for the U.S. economy and is likely, in part, responsible for the emergence of the U.S. as an economic superpower.

Now, though, the decline in participation is not, for the most part, being driven by the young. It is being driven by the old. The workers leaving the workforce are older. The largest contribution to the decline in participation is among workers between the ages of 45 and 55.

These workers are in the prime earning years. They do not leave the labor force lightly. I suspect that the majority of these workers had jobs that no longer exist. Many of these workers will have to find jobs in new industries. A lot of their industry-specific human capital has been destroyed. Almost certainly, at least for a time, the new job will pay less than the old job. Even when they eventually find work, they will be a drag on growth.

What do we do with the large mass of dislocated workers over the age of 45? They are too young, and too poor, to retire. I don’t know the answer but I have a feeling that without this answer the U.S. economy is not going to remain an economic superpower for long.

To formulate policy, we need data. We do not know why these workers are not working. We need to find out who these workers are. What jobs did they hold? Why are they no longer working? What skills do they have? What skills do they need? Are there similar workers in similar circumstance that have managed to stay working? Why did one group perform well and another poorly?

Once we have the answers to these questions then, and only then, we can begin to formulate a policy response. With funding the BLS and the Census Department could answer these questions in a few months. There is no point in throwing money desperately at job creation programs until we understand the source of the jobs problem.

The Employment Situation: Bad News for a Recovery

The economy lost 190,000 jobs in October, a significant improvement from early this year, but losses remain near the peak of the last two recessions. More importantly, several pieces of the jobs report point to a possible deterioration in the labor market and are unambiguously bad for the economic outlook.

Although the labor market has improved substantially since early this year, over the past three months, job losses stabilized around 200,000. We have never had an economic recovery with job losses at this level. I find it beyond belief that the economy is in the midst of a recovery with these losses.

Most forecasters a jobless recovery has already begun. But a jobless recovery is characterized by a weak but stable labor market, a market where losses have ended but gains have yet to occur. An economy can, apparently, muddle along without job growth; it cannot grow with large job losses.

So, even at this level of losses, an economic recovery is not in the cards. But, more worrisome, other indicators of the labor market are much weaker than the establishment survey and some of these indicators point to accelerating losses.

Initial Claims Remain Weak

I wrote almost a year ago on the strong long-term link between initial claims per month and job losses per month. At the time, claims were accelerating sharply and were pointing to unheard of job losses. Initial claims have, of course, improved. But they have not fallen quickly or robustly. Initial claims seem persistently stuck above 500,000 per week.

These initial claims are consistent with job losses between 300,000 and 450,000 per month.

The Household Survey is a Disaster

Once again, job losses as measured by the household survey are outpacing job losses from the establishment survey by a substantial number. From February to June, the household survey and the establishment survey were, on a twelve-month change basis, showing essentially the same job losses. However, since mid-summer, the household survey has pulled ahead by 878,000 about 292,000 extra losses a month.

As I wrote (here), the month-to-month changes in household survey employment are not a reliable indicator of labor market conditions. The survey is subject to substantial sampling variation and month-to-month changes can be absurdly misleading. I have found, however, that whenever the household survey jumps ahead of the establishment survey, in either direction, it tends to predict both the direction of the labor market and the sign of future revisions to the establishment survey.

The household survey, consistent with the claims data, is pointing to a much weaker labor market than is the establishment survey.

A Recession Dummy in the BEDS Model?

I don’t know why the establishment survey is so much stronger than other labor-market indicators. But, I do have a suspicion. A long time ago, I wrote about the Birth and Deaths model used by the BLS to adjust the establishment data. Essentially, the BLS uses this model to control for the number of businesses being created and destroyed every month. I suspect, but do not know, that the BLS is uses a recession dummy in the model. The recession dummy, if it exists, is important and likely substantially improves the performance of the model.

I suspect the BLS turned off the recession dummy in the third quarter. Without the recession dummy in place, the model will, for any given read of the source data, produce fewer job losses. Remember, I do not know that they use one. But, if I were using a model to estimate losses, I would include one. So, I suspect that they have one.

Takeaways

We cannot have a recovery, jobless or otherwise, if the economy continues to shed jobs at a rate of 200,000+ per month. The best indicators do not at the moment point to any further near-term improvement in the labor market. Until we see some substantial improvement in the labor market (at least 0 losses), the economy cannot recover.