Where Have Six Million Missing US Workers Disappeared To?

18th September 2013

Are the unemployment statistics giving a seriously misleading indication of the timetable for Fed policy normalization? Back in the 12th June weekly on this topic, I noted that: ‘The Fed has no idea how many of those workforce exiles have left for structural/demographic reasons rather than cyclical ones; if the latter is the key factor, then unemployment would probably rise in the initial stages of an accelerating recovery, as discouraged potential workers began actively looking again, and thus prolong easy policy. We won’t get a clear answer until y/y employment growth accelerates to 3-4% for several successive months, and then see if there is a strong participation rate response.’  The head of the SF Fed has recently pointed out that the unemployment rate has consistently been the best single measure of slack in the labour market for many decades, and that it remains very closely correlated with alternative measures derived from other sources such as private employment surveys/job openings etc. As it likely tests the 7% level by early 2014, the  Fed’s hand will be forced on exiting extreme monetary stimulus…

While markets have been cheered by Larry Summers having dropping out of contention as the next Fed Chairman, inspiring hopes for a slower QE exit, the bigger issue is the underlying state of the employment market and whether it will prompt the Fed to lower its 6.5% unemployment threshold for a rate hike, which on current trends looks likely by H1 2015. A key issue for global investors to watch is the falling participation rate but also rapidly slowing workforce growth. If the same percentage of adults were in the workforce today as in January 2009, the measured unemployment rate would be 10.8%, and the Fed would be launching QE4. Over the past three months, the US has averaged 148,000 new jobs, a slower pace than the previous six months and yet the unemployment rate dropped to 7.3% in August, the lowest since December 2008, because over the summer a further 312,000 people dropped out of the workforce.

Between 1960 and 2000, the proportion of Americans in the workforce surged from 59% to a peak of 67.3% of the population before dropping to the current 63.2%, driving overall economic growth and household incomes. That was largely due to women entering the labour force while improvements in healthcare allowed working lifespans to be extended.  For the Fed and investors, the question is how much of this downshift is cyclical (the still weak employment market discouraging potential workers) versus structural (i.e. demographic trends).  If the same trends in employment/population growth were in place as pre-recession, total employment would by now be about 6m higher.

The employment/population ratio has hardly changed at all since 2009, implying that the whole of the decline in unemployment has been due to a decline in the participation ratio. Since 2000, the labour force growth rate has been steadily declining as the baby-boom generation has begun retiring and under-25s enter the workforce at a later age. Bureau of Labour Statistics and Chicago Fed researchers have forecast the participation rate to trend even lower by 2020, regardless of how well the economy does in the interim. The extent to which this workforce shift is cyclical rather than structural has huge implications for trend US wage inflation, productivity and GDP growth.

According to the February 2013 CBO estimates, potential growth of the labour supply has been slowing from 2.5% annual growth from 1974-1981 to only 0.8% from 2002-12 and is projected to slow further to only 0.6% over the next five years. At current levels, working-age population growth is at multi-decade lows. Part of that decline in working-age population growth reflects lower immigration to the US as a massive post 9/11 increase in border security spending since 9/11 has curtailed Mexican illegal immigration, and tougher visa rules have slowed legal migrant flows.

The CBO has estimated that the non-inflationary unemployment rate (NAIRU) has risen for temporary reasons to 6% of the labour force, but in the long term it will drop back to 5.5%. At the rate of jobs market progress seen recently, this implies that slack will be wholly eliminated in about 24months, implying a fairly rapid exit from the Fed’s aggressively easy monetary stance. The standard way of estimating the number of people who have temporarily  withdrawn from the jobs market until it improves is to compare the actual participation rate to that predicted if the demographic composition of the workforce had remained unchanged after 2008. This method implies that about 1.3% of the labour force has dropped out since 2008, but should come back if the jobs market improves and that the amount of slack in the labour market might be at least a percentage point higher than implied by the unemployment rate. However, for several important demographic groups, there appears to have been a long term downtrend in participation rates for structural reasons unconnected to the economic cycle, while the potential supply of labour is now constrained, leaving a lot less lack in the system than many assume.


US Productivity Growth a Double Edged Sword for Labour Market…

In this US recovery, total payroll employment has only risen by 1.9%, versus over 8% on average in recoveries since 1961. Productivity has rebounded above its 2% LT trend, while real median income has been negative since the recession. That more than any Fed action explains why equities have performed so well in such a lackluster macro environment, as margins have swiftly rebounded on technology enabled cost-cutting, but that is also driving extreme income inequality and weak aggregate demand. The non-farm payrolls report on Friday showed unemployment declining to 7.8% in September after holding between 8.1% and 8.3% during the first eight months of the year. This is the lowest unemployment rate in more than four years. Employers added a seasonally adjusted 114k jobs, accompanied by upward data revisions indicating that 181k jobs were added in July and 142k in August, and Q3 job growth was far higher than in the spring, amid yet another US growth panic.

Despite gasoline prices flirting with $4/gallon and the pervasive uncertainty over 2013 tax levels, the jobs market has returned to its recent trend. Since the start of 2011, payrolls have grown by a pretty consistent average of 169k jobs a month. Payrolls have risen by more than 250k only three times since the start of 2011, and by less than 50k only once. The labour force rose by 418k in September and the participation rate rose, although it remains historically low (see chart below). In addition, the long term unemployment level has moved down to 40.1% from 41.9% in June. Naturally, the surprisingly low unemployment rate and large upward revisions to prior months drove Republican conspiracy theorists needlessly into overdrive.

Large revisions are a long-standing pattern, whether there’s an election looming or not. Almost all the revisions, however, came from an upward revision of 101,000 to local government education in August before seasonal adjustment – a repeated anomaly at this time of year. It’s likely something is wrong in the BLS collection process as there shouldn’t be a recurrent pattern of error like this. The real issue is less the quantity than quality of the jobs being created; the vast majority were part-time, continuing the general trend since 2009. After accounting for population growth the economy is on balance shedding full-time positions with benefits, offset by part-time positions at lower pay and no benefits. We need to see median real household income rising on a consistent basis to drive a stronger and less volatile recovery, rather than simply cheaper consumer financing as prescribed by the Fed.

The  gain  in  hourly  earnings  was  solid  in  September  at  1.8%  y/y  and  suggests  a corresponding increase in personal income, which would imply support for consumer spending, and further employment gains (the relationship as shown in the chart is that personal income/consumption leads employment by about 3 months). The minutes of the September FOMC meeting released last week point to a vigorous discussion about employment conditions. Those wondering when the Fed’s QE measures might end did not get the definitive decision rule they were hoping for, as the current US recovery is historically well past middle age, at 13 quarters, but has been a very weak one on almost every metric, making traditional policy short hand rules such as the capacity utilization rate dubious.

Stepping back from the data noise, the proportion of 16 to 64-year-olds who are employed fell sharply in the recession and has barely recovered since; while the unemployment rate has steadily fallen back towards its historic levels, labour force participation has fallen, keeping the employment-population ratio constant. One hypothesis is that the recovery has been so weak because of underlying adverse trends in the US labour market. In an academic paper entitled “Manufacturing Busts, Housing Booms, and Declining Employment: A Structural Explanation”, the authors show how the ongoing decline in the demand for men with no more than a high school education in increasingly IT intensive manufacturing has generated an increasingly negative employment trend for these workers for three decades.

This trend continued unabated during the years after the 2001 recession but was masked by the housing boom, which lifted employment for less-skilled workers for another five years. If we view the housing boom (when residential investment doubled from its long-term average at about 3% of GDP) as an aberration that is unlikely to resume, it is misleading to compare the current labour market with that just preceding the onset of the 2008 crash, just as output gap analysis is similarly misleading. In both cases, the starting level of activity was artificially boosted by an historic credit boom, flattering the apparent strength of the US economy at every level back in 2007. The decline in the employment to population and male workforce participation ratios has been a long-term structural trend, pre-dating the 2008/9 recession but accelerated by it.


The recent auto sales data, alongside the ISM surveys (particularly on services) suggest that we’re still stuck in that 2% or so growth zone. Whether the recent mediocre but steady performance of the jobs market and overall economy can be sustained into Q1 or even accelerate, depends increasingly on political events in Washington. As both the Presidential election and ‘fiscal cliff’ loom, there seems to be bipartisan agreement emerging in a couple of areas. Defense spending is slated to come in for $600 billion in cuts over nine years, which might compromise national security and would certainly eliminate thousands of jobs. It seems likely that at least some of that will be restored. In addition, there is bipartisan support for allowing the temporary payroll tax reduction to expire; this will help restore cash flow into Social Security and Medicare, but will cost the average household about $1,000 annually. The uncertainty has been a deterrent to economic activity, as reflected in weakening corporate investment trends, for example.

The  two  sides  have  been  in  meaningful  discussions,  which  are  encouraging;  Treasury Secretary Tim Geithner recently met with Congressional leadership and the Congressional Ways and Means Committee has stepped up its deliberations, but there remains fundamental ideological disagreement on the future course of tax policy and entitlement spending. Without a compromise in these two areas, a long-term budget agreement will remain difficult to achieve. Some observers suggest that a deal will be reached that postpones the crunch point on the debt ceiling, deferring hard discussions until later in 2013, or even that we will go over the cliff to avoid the Republicans breaking their bizarre but sacred principle of no tax rises, and then some taxes get cut again in January.