Gregory Miller


April 6, 2010

Deconstructing the Danger of One-off Events
By Gregory L. Miller
The March report on the Employment Situation was more fun for economists than it was for markets.   My model, designed to estimate current quarter real GDP growth based only on March data says 1Q10 real GDP grew in a range between 12% and 17%, centered on 9.5%.

OK, don’t fire off an instant email over that one. That one-month estimate is clearly flawed.  Last month, the model, based on frigid February data alone, estimated the first quarter economy contracted -2.4%.  The model using all three months of 1Q10 produces a more intuitively appealing estimate (and by “intuitively appealing,” I mean, consistent with my baseline forecast).  The labor plus capital inputs for the three months that comprise 1Q10 predict real GDP = 5.1%.

If that estimate is wrong, it is too low.  A wintry weather blast installed in the February data a notorious bias that affected, among other things economic, hiring US Census enumerators.  January data included the same, if less notorious, discontinuities. 

The Bureau of Labor Statistics (BLS) reported payrolls increasing 162K, about mid-way between the roughly 200K+ consensus and my 120K estimate. It was significantly beneath the extreme 400K estimates floating around the rumor mill.  It was the first material positive in three years.  Its composition suggests March is the beginning of the payroll survey recovery. 

Next consider the household survey, the unfortunately attired step-sibling of the employment series.  The household survey gets no respect.  Headlines rarely quote the household survey, although sound bites often embellish the importance of the unemployment rates, an offspring of the household survey.  The household survey sample is materially smaller than that for the employment survey.  But, the household survey does a better job of identifying turning points, partly because it picks up self employed, unpaid family workers, agricultural workers, private household workers, and small business job momentum.

The BLS distributes 40% of the payroll survey forms to establishments with 20 or fewer employees.  It is also true that the survey response rate is directly proportionate to firm size.

Here are the implications.  The household survey tends to lead the payroll survey on the recovery side of the cycle.  When favored jobs – those with large corporations – become scarce, workers shift to survival mode.  The observation that the rolls of the unemployed remain engorged does not imply those unfortunates do not work.  Many do, just off-the-grid.

They work for smaller firms, they work for themselves, they work from the house, they work for no explicit compensation to save household resources (i.e., staying home to avoid child care expenses).  But they work.  On the recovery side, off-the-grid employment matriculates.  What was a one-woman basement business takes on a helper and becomes a small business.  Small business expands to the next level.  Large business hires small business as “outsource” inputs until large business fear of hiring subsides.

The scale of the graph below may be hard to decipher, but it is a picture of household survey momentum leading payrolls once again.

 Conclusions from the March labor market report:

  • The labor market has clearly shifted to expansion mode. 
  • Is it robust? No, it is not. 
  • Does the March gain cut across important hiring sectors? Manufacturing and construction both posted material gains.  Of the majors, only nondurable goods manufacturing, information, and finance continued contraction.
  • Is it sustainable?  Appears so.
  • Is it inflated by public sector, taxed finance employment? Well, sort of … the 51K gain in Federal jobs is partly offset by a -10K from state and local.  However, Federal stimulus package employment promises mostly take the form of the elusive “jobs not lost”.  States and localities use Federal largess to sustain operations while tax receipts remain weak.
  • Do wages present an inflation risk? Not even close.  Total private wages are running at an annual rate of 2.1%. Only low consumer inflation keeps that paltry increment a net (1%) addition to household resources.







September 9, 2009

Good Enough
By Gregory Miller, Chief Economist, SunTrust Banks


The August employment report was, by all superficial standards, abysmal.  But, it is fitting the market should read this jobs report as good news; e.g., a report suggesting that labor market may well be recovering ahead of any reasonably expected schedule.  No, it’s not great, but for now, it’s good enough.


Payroll jobs fell for the twentieth consecutive month.  Private payrolls fell as government, for its third straight not-so-stimulating month, contracted [1].  The unemployment rate bounced 3-tics (economists view as large any monthly change more than 0.1).


Still, the market reacted with modest optimism.  The DJIA rose for the second consecutive day and the S&P 500 extended its rally to four days.  US treasuries sold off boosting the 2-year UST 8 BP and opening the 2/10 UST spread 23 BP to 265 BP.


The headline results are negative, but unremarkable (see Table 1 below).  August payrolls contracted 216K, the smallest loss since August 2008. The unemployment rate, from the household survey, rose to 9.7% -- a big increase but still only the 10th largest increment of the 21-month recession.  That rate increase separated another 460K workers from their jobs and brought to 7.4 million the increase in unemployment since the recession began. Average hourly earnings rose 6-cents.  That translates to 0.3%, just a rounding error from the recession average gain and modestly higher (an unusual occurrence) than average wage rate gains during the period since the 2001 recession.


Consensus payroll estimates [2] ranged as low as -365K and as high as -185K, centered on -275K.  My forecast for August payrolls was -196K jobs.


This is how to think about it.  Combining July and August data, it predicts 3Q09 real GDP = +3.1%, in a range of 2.8% to 3.3%.  That is a modest downward revision from my prior estimate of 3.4% -- likely due to minor downward revisions to July data.


For as comprehensively negative as the headline numbers are, this report is consistent with my expectation that recession is over.  Chart 1 illustrates a GDP estimating model based on quarterly productivity and successive month’s employment reports.  Standard production function theory drives the routine, e.g. total production (real GDP) is a combination of labor (measured by various series in the employment report) and capital (proxied by productivity).


Chart 1


 The model works because reduced negatives in the primary series (i.e. payrolls) become positive momentum.  In addition, in order to boost production in the early stages of rebound when recovery is uncertain, businesses substitute longer hours and overtime for permanent new hires.  In Chart 2, the uptick in the workweek is small and a bit obliterated by the “recession bar” but it is still up 0.25% in the past two months after declining 2.4% since the recession began.


 Chart 2


The scaling of overtime hours in Chart 3 makes those late-recession “turn-arounds” more evident. At or near the end of last four recessions, employers used extra hours to substitute for full-time hires.  Overtime has rebounded 11.5% this quarter after falling 30% during the first year of recession.

 Chart 3


Productivity is a volatile series that appears to defy the cycle.  But productivity resurgence is virtually always present during early recovery (see Chart 4).
Chart 4


 Chart 5 illustrates the labor market lag.  Compared with the time-weighted average of the past seven recessions, the current labor market is remarkably typical and there should be at least one more negative month before we even consider the prospect of positive job growth.

 Chart 5


Chart 6 illustrates the past year of payroll job deterioration.  Note, consistent with prior recessions, jobs were positive when the recession began.  It also shows a labor market past its nadir and moving steadily toward recovery.

Chart 6


Chart 7 displays the time it takes payroll jobs to return to labor market peaks.  It measures each recession’s temporal path from expansion peak – through recession – back to the previous peak.  This labor market collapse is deep but it length is about the middle of the pack compared with predecessor cycles.

Chart 7

Conclusion:  The commentary surrounding each month’s employment report has been fraught with emotion and exaggeration since this recession began.  Certainly there is significant excess capacity in the labor market. That employment dearth hurts individuals.  But, neither hyperbole nor government spending will solve this economic puzzle.


It is also likely that the labor force was “over-employed” before and even during the early months of this recession. Unemployment did not break above 5.5% until five months into the recession.  If the natural rate of unemployment (i.e., NAIRU: non-inflation acceleration rate of unemployment) was around 5.5%, it is fair to suggest that the labor market operated beyond its efficient frontier for 44 months.  It is likely that the “easy money” that pushed housing and mortgage markets to self-destruction also supported unsustainable levels of job creation. 


It is possible that, as this recession corrected financial market excesses, employers over-reacted to cost-cutting imperatives and reduced labor too much.  It that is true, an economy re-entering expansion phase could easily acceleration the process of re-equilibrating employment


I suspect too many commentators and analysts look way too hard for unique characteristics in this cycle.  Most of the time economic events are reasonably consistent and repair in the labor market is underway. In the meantime, production will recover whether new jobs are flush or few.  Maybe the cycle is like agent Oscar Wallace said to Eliot Ness, on discovering AL Capone’s tax evasion, “It ain’t sexy, but it’s got teeth.”




[1] In fairness, Federal jobs, x-US Postal Service, gained 3.9K.  State and local governments, straining under insufficient revenue streams and legally restrained from deficit spending, dropped 13K.  Included among the theoretical functions of government is hiring during recessions as an automatic stabilizer for private labor market weakness.  The truth is the Federal government contracted employment during each recession since 1960.


[2] There are many consensus estimates. This one came from the Bloomberg survey early during the Employment Report week.   A cautionary note about reliability of pre-release estimates: The well-promoted but rarely prescient ADP estimate attracts a good deal of employment-week attention as well.  Its private sector estimate was -298K; private payrolls reported -198K.  Since the recession began, the ADP consistently missed labor market losses by triple-digit (Ks).  Indeed the ADP predicted continued gains during five of the first eight months of recession.  Over the most recent six months, the ADP shifted to over-estimating labor market losses.  With no intent to sound pedantic, that means either the routine suffers a systemic bias or its survey participants reveal unwarranted optimism in early slowdown stages and miss the beginning of rebounds as well.  This suggests that the measure tends to exacerbate market uncertainty at critical turning points.

Author Bio

Gregory Miller is First Vice President and Chief Economist with SunTrust Bank, Inc.
He completed his undergraduate and graduate Economics degrees at Florida State University and has been a practicing economist, forecaster, and teacher for over 20 years.

Prior to joining SunTrust, Mr. Miller was on the faculty of the College of Business Administration at the University of South Florida in Tampa. Mr. Miller now serves on the USF Business College Board of Advisors. Before Joining USF, Mr. Miller served two Florida Governors in the Florida State Economist’s Office in Tallahassee.

As SunTrust Bank’s Chief Economist, Mr. Miller’s responsibilities include forecasting the national economy, particularly as it affects interest rates. He advises corporate and bank boards of directors. He sits on committees charged with interest rate setting, corporate investment, and benefits policy.

He is a policy advisor for Wealth Management and Corporate Investment Banking groups. Mr. Miller also evaluates markets for, and represents SunTrust in, regulatory matters concerning potential mergers and acquisitions.

In addition to his regular SunTrust duties, Mr. Miller is Past-President and Board Member of the Atlanta Economics Club and President and Founding Director of the Atlanta Economics Foundation.

Mr. Miller is also Past-Chairman of the Economic Advisory Committee of the American Bankers Association — a national group of economists who meet with the Federal Reserve Board of Governors to discuss the economy and monetary policy.