Assessing the health of US labor markets remains a challenge

We continue to see debate around the trajectory of US labor markets. Measures such as the employment-population ratio have completely diverged from the "headline" unemployment rate.

The explanation of course is reduced labor "activity", as individuals exit the labor force (see discussion). However, various studies indicate that the answer is not as clear-cut - due among other things to changing demographics. Historically older workers who are still of working age are far more likely to leave the workforce. As the US workforce gets older, there is some natural attrition that would have taken place even without the massive shock to the system from the financial crisis. Some researchers at the NY Fed did a good job in attempting to quantify that effect (see chart/link).

These same dynamics are responsible for a portion of the decline in labor participation rate. BMO has done some good work in the area, which we try to summarize here. The historical trend looks quite unsettling, as the chart below shows.

Source: BMO

But there is an enormous difference in the "natural" labor force participation rate of a 55-year-old vs. say a 35-year-old - that is unrelated to the Great Recession.

And as we get more 55-year-olds in the workforce, the participation rate "naturally" declines.

BMO: - An aging population will continue to press down on the part rate, as persons 55 and over participate at only half the rate of prime-age workers (25 to 54), while persons 65 and over participate at just one-quarter the rate. If the downward pull from aging overwhelms the expected upward push from discouraged workers and college students returning to the labor force, then the part rate will decline further.

BMO lists other reasons for declining participation that are unrelated to demographics. Instead these effects seem to be the result of government policy.

1. Since the onset of the Great Recession many more Americans have started collecting disability (from the Social Security Disability Insurance program) than in the past (see discussion).

BMO: - In addition to the downward pull from demographics, disability rolls could continue to climb. Macroeconomic Advisers estimates that the Social Security Disability Insurance program, by discouraging participation, could have reduced the part ["part" is short for participation] rate by about 0.1 ppts on average from 2006 to 2013 [notice that this represents some hefty numbers on an absolute basis]. 

2. Now that the extended unemployment benefits ended, more people are counted as having exited the labor force altogether because they are not officially "unemployed". See this chart (from WSJ) on what happened in North Carolina after it cut benefits.

BMO: - The expiration of the extended UI benefits program will also depress the part rate this year. Assuming one quarter of the 1.3 million people who lost emergency benefits on December 28 find work and another quarter leave the labor force, the part rate could fall 0.2% in 2014. This accords with the additional decline in North Carolina’s participation rate, relative to the national average, after the state decided to end extended benefits five months earlier. 

3. Many analysts are suggesting that Obamacare will detract from labor force participation as well by making health insurance cheaper for those who have minimal income and incentivizing some to leave the workforce.

BMO: - According to the CBO, the Affordable Care Act will likely also reduce the part rate, as new federally-subsidized private health insurance plans will reduce the incentive to seek employer-sponsored plans. Quantifying this impact, however, is virtually impossible. 

The hope is that the expected economic improvements over time will counteract some of these trends. The challenge for the FOMC of course is which employment measures to use for policy decisions and guidence. Clearly the traditional unemployment rate benchmarks are inadequate. The "activity" measures such as participation are driven by multiple factors, some of which are not directly related to the health of the economy. When is the employment situation in good enough shape to start raising rates for example and which measures do we rely on to make that call? This assessment challenge makes the Fed's "dual mandate" a particularly daunting task. 

SoberLook.com

From our sponsor:

Has the Great Recession created behavioral changes in the labor markets?

The US civilian unemployment rate, which clocked at 7.2% last month, has been declining at the fastest rate in nearly two decades. The focus however has been on other labor market indicators. One of those is the employment-population ratio, the proportion of the US working-age population that is employed (discussed here). With declines in the unemployment rate one would expect an increasing fraction of working-age population getting back to work. The ratio however has remained nearly constant since the recession.

The dispersion between the two measures is quite clear if one plots them on a year-over-year basis.

Source: FRED (h/t Robert Mellman/JPM)

Most would interpret this divergence as an indicator of Americans leaving the workforce in large numbers, particularly as their unemployment benefits run out. But that's not the full story. The question is whether the Great Recession had created "behavioral changes" in the labor markets. Here are a few points worth addressing with respect to the flows in and out of the workforce (based on recent work by Robert Mellman/JPM):

1. Are people more likely to get discouraged and leave the workforce after the Great Recession than in the past? The answer is quite surprising. The "dropout rate" is actually similar to historical trends.

JPMorgan: - Despite the background of high unemployment, drop-out rates of the unemployed are surprisingly similar to prior expansions. 

2. Then why have so many people left the workforce? The answer is simple. Taking roughly the same "dropout rate" as before but applying it to a much larger number of unemployed people than in the past will create a large total "drop-out" pool.

3. But are those who exited the workforce more reluctant to reenter the job market than has been the case historically? Once again it turns out that the "reentry willingness rate" is not materially different from history.

JPMorgan: - ... there has been no increased reluctance of those out of the labor market to enter the labor market in any given month (whether into employment or unemployment). An average of 7.6% of those out of the labor market entered every month in the prior expansion, and the figure is 7.5% for the current expansion.

4. Why has the employment-population ratio not budged since the recession? One of the persistent problems with the US labor markets is the current pool of unemployed people taking far longer to find work than in the past. This would suggest that the key for those without work is to enter the workforce as quickly as possible - even if it means part-time, temp, consulting, etc. The barriers to re-entry are much higher these days than any time in recent decades.

JPMorgan: - The labor flow data show that the chance of an unemployed worker finding a job in a given month fell dramatically during the last recession and has remained near its lows since.

SoberLook.com

From our sponsor:

The latest jobs number - a deeper dive

Guest post by Lee Adler (The Wall Street Examiner)


The BLS today reported a gain of 171,00 in nonfarm payrolls. The actual, not seasonally adjusted (NSA) number was a gain of 911,000. In the actual (NSA) data, October is always an up month. Last year the October NSA gain was 895,000. In 2010, it was 978,000. The 10 year average gain for October for 2002 to 2011 was 737,000. This month’s report was a good one, consistent with the trend.

There’s a problem with the seasonally adjusted number. The SA number for this month will subsequently be revised in each of the next 5 years as the BLS attempts to fit the SA number to the actual change. It will also have a major benchmark revision in February, when the annual benchmarking process is finalized.

The BLS headline number is really lousy data, but the market pays attention to it. In September, the August SA headline number was revised up by 46,000 and July was revised up by 40,000. Then this month both August and September were revised up, August by 50,000 and September by 34,000. The BLS will revise this month’s number, not only next month and the month after, but every year for the next 5 years as they hone the SA number to include a look back to pinpoint where this month’s number actually should have been. The truth is that the current SA number is a wild guess and a fraud. The BLS statisticians know it and have publicized that fact, but the mainstream media has ignored the warnings for years.

We need to look at the best data we can find to know the truth about what’s going on. That’s the NSA data. The withholding tax collections are actual and real time, but from time to time may be skewed by factors other than the number of employed persons, so we need to be alert for anomalies in that data.

The monthly employment numbers reported above come from the BLS the Current Employment Statistics Survey or CES, a survey of business establishments. The BLS also does a survey of households. To further complicate matters, the household survey or CPS — Current Population Survey– often tells a different story from the establishment survey. This month the two were consistent.

As with the CES, in the CPS October is a month in which the actual NSA number always increases. This year the number of persons reported as employed in October rose by 706,000 from September (Actual NSA). That compares with a gain of 485,000 in October 2011 and 34,000 in 2010. The average gain in October for the previous 10 years was 480,000. By this standard this was a very good month. The year over year gain was 2.2%.

Full time, as opposed to total employment, is a key measure. Part time jobs are nice, and for many that hold them, they are a lifeline, but the important metric here is full time jobs. Without those, we’re dead.

Full time employment in the CPS rose by 367,000. Part time jobs increased by 339,000. Last year full time jobs increased by 476,000 in October, but in 2010 they dropped by 43,000. The 10 year average gain in full time jobs for October was 81,000. This year’s performance wasn’t as good as last year, but it was significantly better than average.

The chart below shows the year to year trend line connecting the October data in full time and total employment along with the raw NSA data and the SA fiction. The seasonally finagled data shows the trend nearly catching up in both series due to the big upward revisions to the July through September data. Prior to this, from March through August the SA line had been diverging from the actual trend, particularly in full time employment.

Full Time Employment Short Term View – Click to enlarge

June or July is usually the peak month for both total and full time employment. This year the numbers broke last July’s level in April. The economy was a couple months ahead of schedule in affirming the uptrend in jobs. That uptrend is still firmly entrenched. The gains have accelerated in 2012 versus 2011. With QE3, the Fed is adding more fuel to a fire that was already beginning to burn hot.

Full Time Employment, Stocks, and The Fed – Click to enlarge

Stock market performance is at the mercy of the Fed (or over the past 12 months the ECB, not shown), and employment typically reflects them both. Over the past year, the SOMA has not reflected the impact of the Fed’s MBS purchases from the Primary Dealers, a subject which I cover in depth weekly in the Fed Reports. While SOMA has stayed flat pending the first settlements of the QE3 purchases on November 13, the graph of Fed purchases from the Primary Dealers (not shown) has been rising steadily since last September. By cashing out the dealers via these MBS buys, the Fed enables the dealers to buy more Treasuries. The next week, the Treasury spends that cash. That’s how Treasury debt is immediately transformed into economic activity and slow and steady job creation. With new QE, the Fed will be adding even more cash to power that trend.

The chart below shows that while the number of jobs is growing, the employment to population ratio has barely gained since the recovery began in 2009. The economy seems only to be keeping pace with population growth. Top line growth may satisfy the markets, but it is doing next to nothing to help the millions of people who remain unemployed. Their numbers are growing right along with the number of people who do have jobs. It is a sad state of affairs for the US, but markets don’t care about that.

Full Time Employment to Population Ratio – Click to enlarge

Many of the unemployed do not possess the skills that are in demand in the market. Mortgage application takers and processors, and construction laborers generally do not make good computer game programmers. Economic pundits must face the fact that the 10 million fake jobs spawned by the bubble are not coming back. The 7.9% unemployment rate is probably “normal.” The bubble unemployment rate of 5.5% was abnormal

SoberLook.com

The Beveridge curve shows a structural shift in US employment dynamics

The Beveridge curve, developed in the UK back in 1958, compares job vacancies as a fraction of total labor force with the unemployment rate. It allows one to study, among other things, labor inefficiency and labor mobility. If there are job openings in one part of the country or one industry, but the unemployed are unable to fill those openings due to geographical or skill mobility constraints, the Beveridge curve would show it.

Barclays Capital has recently looked at the Beveridge Curve for the US. The curve "regime" has shifted significantly in the post-recession environment.

Source: Barclays Capital

With the current job openings number, the unemployment would have been at 5.5% on the pre-recession curve, but it clearly isn't. What does this tell us?

This looks like a structural shift rather than a cyclical adjustment, with a completely new "equilibrium" level. That's why the Fed now believes that the new "natural" unemployment rate is closer to 6.7% rather than the 5% prior to the recession. This fact is critical as the Fed adjusts the monetary policy. If the unemployment rate begins to approach 6.7% (not the pre-recession 5%), the Fed Funds target rate will begin to rise.

So what's the explanation for this shift?

Geographic mobility constraints may explain some of it, as negative home equity makes it difficult for the unemployed to move. Economists however are skeptical that geographic "immobility" could explain the whole shift. If slow migration was the explanation, over time the Beveridge curves should start converging as migration picks up - people walk away from their homes, sell them, or find jobs locally. But so far the pre- and post-recession curves have been diverging.

Some economists point to a cross-sector skills mismatch. For example lost construction jobs may not be replaced with newly created manufacturing jobs because of different skill requirements. But as a fraction of the overall unemployment rate, construction unemployment is roughly back to where it was prior to the recession.

Barclays Capital: - ... sectors which experienced the largest relative increase in unemployment during the recession (particularly construction) have experienced the largest relative decline in the recovery, suggesting that workers laid off in construction have found jobs elsewhere. Indeed, the ratio of the construction sector unemployment rate to the total is now close to pre-recession levels.

There may be some evidence of a skills mismatch within industries. For example, within the manufacturing sector the Beveridge curves have diverged as well. But have skill requirements in US manufacturing really changed so much during the recession as to generate such a shift?

Source: Barclays Capital

There is a better explanation that points to a structural shift in employment dynamics. Some refer to it as hysteresis. In engineering the phenomena describes a system that responds not just to its current environment, but to the path it took. A thermostat switch will turn on at a different temperature than it turns off to avoid rapid switching. Built-in hysteresis will assure that the switch "event" depends not just on the current temperature but on whether the temperature got there by increasing or decreasing.

It seems that employment dynamics have a built-in hysteresis, driven by the effects of long-term unemployment. Data suggests that the longer someone has been out of work, the longer it takes them to find work.

Source: Barclays Capital

This means that after a deep prolonged recession, as job vacancies rise, they will be filled at a rate different from when jobs were originally lost.

Barclays Capital: - One reason is so-called hysteresis – a term to explain the long-lasting effect of a temporary shock, in this case, how a cyclical jolt to the economy transmits into persistently higher (ie, structural) unemployment. Long-term unemployment is the most powerful channel for this, in our view. In essence, this is because evidence suggests that the longer a person remains unemployed, the less likely they are to re-enter employment . 

Given that the long-term unemployment has been at record levels (chart below), the structural hysteresis effect is far more pronounced than in the past. The path to recovery on the Beveridge curve is very different from the path of decline. The higher "natural" unemployment rate is here to stay.

For more on the Beveridge curve shift, please see this 2010 post by the Cleveland Fed.

SoberLook.com

A hidden surprise in the US employment report

Today's employment report brought with it all sorts of negative economic surprises, creating a sharp drop in the Citi Economic Surprise Index.

Citi Economic Surprise Index

Bloomberg: - The Citigroup Economic Surprise Index for the U.S., which measures how much data is missing or beating the median estimates in Bloomberg surveys, fell to minus 53.6, the lowest since September. It turned negative this year in April after remaining above zero since October. The Federal Reserve announced a program dubbed “Operation Twist” to boost growth on Sept. 21, 2011, four months after the index turned negative.

One of the worst surprises buried in the report however was the fact that all the job growth came from increases in part-time jobs. The number of full-time jobs actually declined significantly.

WSJ: - ... the job growth is coming entirely from workers getting part-time jobs. The number of Americans working full-time fell by 266,000 in May, erasing all the gains of the past three months. The total employment figure only rose because 618,000 more people got part-time jobs. Many of those people would rather be working full-time: The number of people classified as “part time for economic reasons” — meaning they’re working part-time because they can’t find a full-time job — rose by 245,000 to 8.1 million.

SoberLook.com

The flatlining of the Employment to Population Ratio

Last Friday's poor employment number was a reminder that the US employment picture remains fragile. In spite of the abysmally low number of jobs created (120K vs. 205K expected), the unemployment rate continues to decline. The unemployment rate is a misleading indicator however. A more indicative measure for the overall employment picture is the US Employment Population Ratio, the proportion of the US working-age population that is employed (part time or full time). And that measure (closely watched by the Fed) actually declined slightly last month.

Source: JPMorgan

Over the longer term, the Employment Population Ratio has been fairly flat as the unemployment rate kept dropping. So where is the inconsistency? One measure that explains this disconnect is called the Labor Force Participation Rate. It measures the total US labor force, combining both the employed as well as the unemployed, as a proportion of the total working age population. This measure is showing that the total labor force  has been shrinking relative to the corresponding age group. And that explains a good deal of the reduction in the unemployment rate as the number of people officially employed has been measured against an increasingly smaller labor force (with many unemployed simply exiting).

US Labor Force Participation Rate Total SA (Bloomberg)

There are numerous factors contributing to the reduction in the labor force, including people staying out of the workforce (supported by their spouse for example), going to school and living on student loans, participating in "unreported employment", as well as workers going on disability (legitimately or not).

Going forward this ratio will be critical to watch because it will indicate whether people of working age are entering or leaving the official labor force. For now though the headline unemployment rate number is not meaningful and should not be used as a gauge of improving labor conditions.

SoberLook.com

Quarter of the workforce dropouts go on disability

As discussed in the recent post on low employment ratio in the US, declining labor participation continues to be an issue. The chart below shows that as the unemployment rate declined recently, the employment ratio has not budged. The Fed pays close attention to this discrepancy - as they should.

Source: Barclays Capital

But how does one survive after losing the unemployment benefits? Clearly people struggle. One way to pay the bills however is to file for and receive the federal disability benefits - assuming of course one has a disability. Interestingly enough, the Great Recession and the slow recovery somehow generated many more disability recipients.

JPMorgan: As of January over 8.5 million individuals were receiving federal disability payments (an additional 2 million spouses and children of disabled workers also received disability payments). Since the onset of the recession and the subsequent slow recovery, this figure has accelerated and grown faster than the overall size of the potential labor force— currently 5.3% of the population aged 25-64 is on federal disability, up from 4.5% when the recession began.

Source: JPMorgan

JPMorgan points out that increases in the number of disability benefits recipients account for about a quarter of the decline in employment participation. Furthermore during recessions the number of new disability claims actually increases, even though the number of jobs with higher injury incidence (such as construction) generally declines. Try explaining that one...

Half of the benefit recipients suffer from "mental disorders" and "musculoskeletal disorders" (such as back pain). "Mood disorders" alone account for over 10% of this group. And once someone starts receiving these benefits, it's almost impossible to take the off the program. In 2011 only 1% of the recipients lost their benefits because they were no longer deemed disabled. So how much is this program costing the US taxpayer? Apparently quite a bit.

JPMorgan: The cost to the federal budget of these programs has escalated along with the number of claimants, and now runs around $200 billion per year—more than the budgets of the Departments of Commerce, Energy, Homeland Security, Interior, Justice, and State combined.

Thus a quarter of people who drop out of the workforce and come off the unemployment benefits, simply move to receiving disability payments. And most stay there until they roll into the social security program when they retire - from their disability. The same source, a different program.

SoberLook.com

Employment participation rate shows a troubling trend

One reason to remain cautious on the recent strength in the US employment indicators is the rapid decline in the employment participation rate (chart). We are at the levels not seen since the early 80s.

Proportion of the US population that is in the labor force (Bloomberg)

This is telling us that Americans are dropping out of the workforce at a fairly rapid rate. Some may argue that this is simply a result of aging US population - as more people retire, a smaller portion of the population is working. The US is not Japan with respect to the speed of population aging, but the demographics certainly support this theory - the proportion of those who are 55 and older has been on the rise since 1999.

Source: Credit Suisse

 
However this does not tell the full story. Yes, the population is aging, but if one looks at the sources of employment participation decline, the older group is not the main contributor.

Employment participation rate by age group (Source: Credit Suisse)

This data is a bit dated, but the trend is unmistakable - the declines have been coming from the two younger groups. It is possible that the older group may even be contributing positively to employment participation as people end up having to work longer before retiring. This means that the sharp decline in the employment participation rate has to be taken into account when considering the recent improvements in the unemployment rate as well as the US jobless claims.

SoberLook.com