No. 499: January Labor Data and Revisions, M3 and December Construction Spending
COMMENTARY NUMBER 499
January Labor Data and Revisions, M3 and December Construction Spending
February 2, 2013
Positive Benchmark Revision to Payrolls Overstates Current Circumstance
Annual Upside-Bias in Birth-Death Model Increased to About 620,000
January Unemployment: 7.9% (U.3), 14.4% (U.6), 23.0% (ShadowStats)
Month-to-Month Headline Unemployment Rates Are Not Comparable
January M3 Year-to-Year Growth at 4.5%
PLEASE NOTE: The next regular Commentary is scheduled for Friday, February 8th, covering the monthly December 2012 and annual 2012 balance-of-trade reporting.
The Special Commentary analyzing the GAAP-based 2012 Financial Statement of the United States Government will be published on Monday, February 4th.
Best wishes to all — John Williams
Opening Comments and Executive Summary. The broad economic outlook has not changed; after U.S. economic activity collapsed into 2009, there was no actual economic recovery, and there is no recovery pending in the near future (see No. 485: Special Commentary). Following a downside surprise in the headline “advance” estimate of fourth-quarter GDP growth, the first major report on January 2013 economic activity was somewhat softer than had been expected. Headline unemployment rose to 7.9% in January, from 7.8% in December, instead of declining, and nonfarm payroll employment rose by a below-consensus 157,000 jobs, versus a revised 196,000 jobs gain in December. The jobs numbers, were in the context of a roughly 0.3% upside annual benchmark revision to nonfarm payrolls, but much of the benchmark news already had been released.
The revamped seasonally-adjusted payroll data increasingly were flawed, with variations in seasonal adjustments—unpublished by the Bureau of Labor Statistics (BLS)—pushing prior jobs growth into the most-recent reporting periods and bloating the most-recent month-to-month payroll gains (the benchmark revision is detailed in the next section).
On the unemployment front, the annual revision to population estimates made a number of series reported with January 2013 household survey inconsistent with earlier data, but the headline unemployment rate went unscathed, there. Unfortunately, though, the usual concurrent-seasonal-factor adjustments to the unemployment rate continued, without the publication of consistently revised historical data.
An important point to remember, here, as discussed in the Concurrent Seasonal Factor section in Reporting Detail, is that with the clouded transparency of BLS reporting, published month-to-month comparisons of seasonally-adjusted unemployment rates and of month-to-month changes in nonfarm payroll levels generally are not meaningful.
In other reporting, year-to-year M3 growth in January 2013 rose to a 43-month high of 4.5%, reflecting an ongoing upturn in annual M3 growth (see the Hyperinflation Watch, section). At the same time, reflecting increasing liquidity stresses in the consumer area, January consumer sentiment and confidence both trended lower, showing low-level readings generally not seen outside of the depths of the most-severe modern recessions (see the last subsection of the Opening Comments).
December 2012 construction spending rose by 0.9% in the month, but the change, as usual, was statistically insignificant. Unseasonably mild weather and reconstruction resulting from Hurricane Sandy’s destruction likely provided some relative boost to the seasonally-adjusted construction numbers (see the second-to-last subsection of the Opening Comments and the Reporting Detail section).
January Employment and Unemployment. In the context of a 0.3% upside annual benchmark revision to nonfarm payrolls, January 2013 headline nonfarm payroll growth was 157,000 jobs, down from a benchmark-revised 196,000 gain in November. Given serious reporting flaws, discussed in the Concurrent Seasonal Factor Distortions section of Reporting Detail, the current numbers continue to be so far out of balance as to be effectively meaningless indicators of underlying economic activity.
The seasonally-adjusted January 2012 headline U.3 unemployment rate increased to 7.92% from 7.85% in December. The broader headline January U.6 rate held at 14.4% for the second month, where the ShadowStats alternate unemployment rate held at 23.0%, also for a second month.
Benchmark Revisions to Payroll Employment. In September 2012, the BLS announced a tentative upside benchmark revision to not-seasonally-adjusted March 2012 nonfarm payrolls of 386,000 jobs. In the benchmark revision published on February 1, 2013, the upside revision actually was 424,000 jobs. With other adjustments, the benchmark revisions restated historical payroll data back to 1990.
Each year, the BLS benchmarks its estimate of unadjusted nonfarm payrolls to the March level of employment suggested by employer state unemployment tax filings, in the same period, a less-than-perfect benchmarking device. Drawing a conclusion as to whether its March estimate was understated or overstated, the BLS restates its reporting history so as to adjust the March levels to the benchmark.
As is usual, the unadjusted March 2012 payrolls were reset higher by the desired amount, with the benchmark increase backed into the numbers—at a declining pace—for the prior year since the March 2011 benchmark. In addition, the revision was carried forward to current reporting, adding 723,000 unadjusted, 647,000 seasonally-adjusted jobs to the previously-estimated December 2012 payroll levels. The current treatment of the data here likely is overstating payrolls and growth, as also suggested by the upside revision to the annual bias, added by the Birth-Death Model, from 535,000 as of December, to over 620,000 post-benchmark (see Birth-Death/Bias-Factor Adjustment section in Reporting Detail).
Reflected in the first-graph following are the patterns of the revised reported levels of nonfarm payrolls, in place as of the referenced benchmark revision and subsequent reporting, until they are replaced by the next benchmark revision. The just-published 2012 revision is reflected by the heavy blue line.
The second graph plots revised seasonally-adjusted payroll levels since 2000, showing detail of the current employment level holding well below its pre-2007 recession peak. The third is a longer-term graph of revised employment levels, showing historical detail back to 1940 and, in perspective, that current levels still are minimally above levels in 2000. Of note, the plotted payroll levels are as reported by the BLS, and include inconsistent historical data resulting from concurrent seasonal adjustments.
Reflected in the next graph following are comparative benchmark revisions and subsequent reporting, for 2010, 2011 and 2012. The plotted patterns are of not-seasonally-adjusted (NSA), year-to-year change in place as of the referenced benchmark revision and subsequent reporting, until each series is replaced by the next benchmark revision. Unlike the distortions seen in the seasonally-adjusted series of nonfarm payroll levels, the year-to-year change is based on not-seasonally-adjusted data, and the data remain consistent over time. Growth patterns, however, are affected by the extended benchmark adjustments and by the Birth-Death Model’s increasing upside impact.
As shown in the longer-term graph (historical detail back to 1940), with the bottom-bouncing of recent years, current annual growth has recovered from the post-World War II record 5.06% (unrevised) decline in August 2009, which remains the most severe annual contraction since the production shutdown at the end of World War II (a trough of a 7.59% annual contraction in September 1945). Disallowing the post-war shutdown as a normal business cycle, the August 2009 annual decline was the worst since the Great Depression. Still, even with pick-up in annual growth in the series since mid-2010, the current level of employment is far from supporting a full economic recovery.
December 2012 Construction Spending Gain Was Statistically Insignificant. The trend of month-to-month stagnation in headline construction spending continued in December. The upside trend in the levels of revised monthly activity, as measured in current dollars (nominal, or not adjusted for inflation), and as seen in the graphs in the Reporting Detail section, largely is accounted for by rising construction costs. Despite the lack of a significant monthly gain in December, the latest numbers received some relative boost from unseasonably mild weather in the month, along with likely increasing reconstruction activity resulting from Hurricane Sandy’s damages.
The total value of construction put in place in the United States during December 2012 was $885.0 billion, on a seasonally-adjusted—but not inflation-adjusted—annual-rate basis. That estimate was up for the month by a statistically-insignificant 0.9% +/-1.9% from November. In turn, November showed a revised 0.1% gain (previously a 0.3% contraction) versus October. December construction spending was up year-to-year by a statistically-significant 7.8%, but a fair portion of that annual gain reflected inflation.
December’s 0.9% monthly construction-spending gain reflected a 1.4% drop in public construction spending, offset by a 2.0% gain in private construction.
As discussed in Hyperinflation 2012 and No. 485: Special Commentary, consumer liquidity remains structurally impaired, and that forestalls the possibility of sustainable economic growth. That circumstance also puts the lie to the government’s claims of economic recovery since June 2009, and dashes any expectations or hype of any developing recovery in the broad economy, including in housing or aggregate construction spending sectors. Placing likely further downside pressure on near-term consumer spending, reporting of both consumer sentiment and confidence has weakened anew.
January 2013 Consumer Sentiment and Confidence Suggest Ongoing Trouble. On February 1st, the University of Michigan released its consumer-sentiment index for January 2013. The not-seasonally-adjusted series was little changed, following a monthly plunge in December. In contrast, as discussed in Commentary No. 498, the Conference Board’s seasonally-adjusted consumer-confidence index fell sharply month-to-month in January 2013.
The numbers are plotted in the preceding two graphs, where both consumer sentiment and confidence remain at levels not seen outside of the depths of the most-severe historical recessions. Intensifying consumer liquidity problems remain primary constraints on sentiment and confidence, and on the ability—indeed the willingness—of the consumer to fuel sustainable growth in broad economic activity.
[More-complete reporting on the January labor data, and more detail and graphs on
December construction spending are found in the Reporting Detail section.]
January 2013 Broad Money Growth Rose to 4.5%. Based on roughly three weeks of reported data and on annual benchmark revisions published by the Federal Reserve, the preliminary estimate of annual growth in the ShadowStats Ongoing-M3 Estimate for January 2013 is 4.5%, as published in the Alternate Data tab at www.shadowstats.com. January annual growth was up from an unrevised 4.4% in December.
Where growth has been on the upswing in recent months, the January 2013 reading was the strongest level of annual growth seen since July 2009, when growth was headed lower. The last time that annual growth broke through 4.5%, on the upside, was in September of 2004.
Any prior-period revisions in the following numbers are due to Federal Reserve revisions to underlying data. The seasonally-adjusted, preliminary estimate of month-to-month gain for the January 2013 money supply M3 is 0.5%, down from a revised 1.0% (previously 1.1%) in December. The estimated month-to-month M3 changes, however, remain less reliable than the estimates of annual growth.
For January 2013, early estimates of year-to-year and month-to-month changes follow for the narrower M1 and M2 measures (M2 includes M1, M3 includes M2). Full definitions are found in the Money Supply Special Report. M2 for January is estimated to show year-to-year growth of about 7.5%, versus a revised 7.9% (previously 8.1%) gain in December, with month-to-month growth estimated at roughly 0.3%, down from a revised 1.1% (previously 1.2%) in December. The early estimate of M1 for January is year-to-year growth of roughly 11.6%, down from a revised 12.9% (previously 14.2%) in December, with month-to-month change a likely gain of 0.7%, versus a revised 1.5% (previously 3.3%) in December.
Hyperinflation Outlook: Summary. The following text largely is unchanged from the prior Commentary. It will be fully rewritten following the pending Special Commentary on the U.S. government’s financial statements. These comments are intended particularly for new subscribers, as well as for those who otherwise are not familiar with the hyperinflation report or the recent special commentary, linked below. Those documents are suggested as background reading on the financial turmoil and currency upheaval facing the United States in the next year or two.
The November 27, 2012 Special Commentary (No. 485) updated Hyperinflation 2012 and the broad outlooks for the economy and inflation, as well as for systemic stability and the U.S. dollar. These remain the two primary articles outlining current conditions and the background to the hyperinflation forecast. The economic and systemic solvency crises of the last seven years continue. There never was an actual recovery following the economic collapse into 2009, just a protracted period of business stagnation that began to turn down anew in second- and third-quarter 2012.
Subsequent to Special Commentary (No. 485), neither new economic data nor fiscal developments have altered the broad outlook. The expansion of QE3 by the Fed, on December 12, 2012, now has begun to impact the monetary system, spiking the monetary base and beginning to boost annual growth in the broad money supply. That circumstance should begin to contribute to the inflation outlook, even in the absence of normal bank lending.
Despite the near-term political hype that Congress will come up with a plan to balance the budget in a ten-year time frame, little but gimmicked numbers and further smoke-and-mirrors are likely to come out of current negotiations. Ongoing economic woes assure that the usual budget forecasts—based on overly-optimistic economic projections—will fall far short of fiscal balance and propriety. Furthermore, chances remain nil for the government addressing the GAAP-based deficit that hit nearly $7 trillion in 2012, instead of the popularly followed, official cash-based accounting deficit in 2012 of $1.1 trillion.
The ongoing unwillingness and political inability of the current government to address seriously the longer-range U.S. sovereign-solvency issues, only continue the regular unfolding of events that eventually will trigger a domestic hyperinflation, as discussed in Commentary No. 491.
The Fed’s current liquidity actions can be viewed as a signal of deepening problems in the banking system. As discussed by Mr. Bernanke, the Fed can do little to stimulate the economy, but it can create inflation. Nonetheless, the Fed’s move here was to prop-up the banking system and to provide back-up liquidity to the U.S. Treasury in the months ahead. The renewed direct monetization of Treasury debt will tend to savage the U.S. dollar’s exchange rate, to boost oil and gasoline prices, and to boost money supply growth and domestic U.S. inflation.
The primary issue, however, remains the failure of the U.S. government to make any serious effort at bringing the nation’s extreme and dangerous fiscal conditions into balance. Efforts at delaying meaningful fiscal action, and at briefly postponing conflict over the Treasury’s debt ceiling, have bought the politicians in Washington a little time in the global financial markets, but the patience in the global markets is near exhaustion. Market tranquility likely will not last much longer, despite the tactics of delay by the politicians. This should become increasingly evident as the disgruntled global markets begin to move against the U.S. dollar.
EMPLOYMENT AND UNEMPLOYMENT (January 2013, Payroll Benchmark Revision)
Latest Payroll and Unemployment Data Remain Seriously Flawed. The broad economic outlook has not changed, despite revised but still seriously-flawed data out of the Bureau of Labor Statistics (BLS).
Along with the February 1, 2013 reporting of January labor conditions came the upside annual benchmark revisions to payroll employment, which have had the effect of significantly bloating recent payroll reporting, revised and otherwise. The just-published benchmark actually is the fourth-version of that calculation published (three earlier versions were calculated internally by the BLS, starting in October 2012), as a result the just-published November 2012, December 2012 and January 2013 numbers are not consistent with or comparable to earlier data. The just-published formal benchmark revisions go back to 1990.
The latest BLS release also included the annual population revisions to the household survey, affecting only January 2013 and later. Those changes, however, mean that the January 2013 and later household-survey data are not comparable with the unrevised December 2012 data, either seasonally-adjusted, or not-seasonally-adjusted.
At the same time, the concurrent-seasonal-adjustment process has moved beyond December, the one-time-per-year period when seasonally-adjusted household-survey numbers (unemployment-rate data) are published showing consistent month-to-month reporting. Accordingly, seasonally-adjusted comparisons of monthly unemployment rates are not consistent and are not meaningfully comparable.
The details of current reporting distortions and the benchmark revisions to nonfarm payrolls are covered in the Opening Comments and the New Concurrent Seasonal Factor Distortions section. The issues with the household survey are covered in the Household Survey Detail, later in this general section.
Given the distortions from unstable concurrent seasonal factor adjustments used by the BLS in adjusting both the payroll and household surveys, as discussed in the Opening Comments and in Special Commentary (No. 485), the monthly headline changes reported for both the headline payroll growth and the headline unemployment rate continue to lack statistical significance.
To the extent that there is any meaning to the monthly reporting, it is that the economy is not in recovery; even payroll levels simply do not show the full economic recovery propagandized in the GDP reporting. Further, unemployment—as viewed by common experience (the ShadowStats Alternate Measure)—remains at an all-time high for the series, a level that rivals any other downturn of the post-Great Depression era.
PAYROLL SURVEY DETAIL. In the context of its annual benchmark revisions to nonfarm payrolls, the BLS reported on February 1st a seasonally-adjusted, month-to-month headline payroll employment gain of 157,000 for January 2013. Where the standard 95% confidence interval on monthly headline payroll employment reporting is +/- 129,000, circumstances suggest that a much wider confidence interval could be justified. The current numbers continue to be so far out of balance as to be absolutely meaningless here, due partially to concurrent-seasonal-factor distortions (discussed in the Opening Comments and in the Concurrent Seasonal Factor Distortions section).
The seasonally-adjusted December 2012 month-to-month jobs increase was revised to 196,000, where the pre-benchmark gain had been 155,000. The adjusted November 2012 month-to-month increase was reported 247,000, where the last pre-benchmark gain had been 161,000. The BLS publishes only two prior months of consistent data with concurrent-seasonally-adjusted payrolls. Accordingly, if the November change were counted on a basis consistent with the latest concurrent-seasonal-factor calculations, the revised November gain would have been 205,000.
These numbers, and data from a couple of months before appear to have been inflated by early use of the 2012 benchmark results, both in pre- and post-election reporting. Even with the benchmark revisions, the concurrent seasonal adjustments already are hiding shifting growth patterns that are never published by the BLS. As discussed in the New Concurrent Seasonal Factor Distortions section, the inconsistent use of concurrent-seasonal-adjustment factors, last year, had the effect of pushing relative payroll gains into the pre-election reporting period. The new pattern shows some shifting of jobs growth from the third-quarter to the November-to-January period.
Trend Model. As described generally in Payroll Trends, the trend indication from the current BLS seasonal-adjustment model is for a 158,000 monthly payroll gain in February 2013, based on January’s reporting. While the trend indication often misses actual reporting (the indication for January was for a 150,000 monthly gain, which was reasonably close to the official 157,000 headline gain), it nonetheless usually becomes the basis for the consensus outlook.
Annual Change in Payrolls. In terms of year-to-year change, the not-seasonally-adjusted growth in January 2013 payrolls was 1.57%, somewhat slower than the 1.69% annual pace now indicated for December. Pre-benchmark, the year-to-year change for December 2012 was estimated at 1.40%.
In addition to plots of the old versus revised payroll series, all the revised regular graphs of nonfarm payroll levels and year-to-year growth are shown and discussed in the benchmark-revision detail in the Opening Comments Section.
New Concurrent Seasonal Factor Distortions. There are serious and deliberate reporting flaws with the government’s seasonally-adjusted, monthly reporting of employment and unemployment. Each month, the BLS uses a concurrent-seasonal-adjustment process to adjust both the payroll-employment and unemployment-rate data for the latest seasonal patterns. The headline payroll gain and unemployment rate are so-calculated, but the adjustment process also revises the history of each series, recasting prior reporting on a basis that is consistent with the new headline numbers. The BLS does not publish the revised history, even though it calculates the data each month. As a result, headline reporting generally is neither consistent with nor comparable to earlier reporting, and month-to-month comparisons of these popular numbers usually are of no substance, other than for market hyping or political propaganda.
For January 2013 the headline unemployment rate was 7.9%, and the headline monthly payroll change was a gain of 157,000 jobs. Yet, the reported January headline unemployment rate was neither consistent with nor comparable to the headline December 2012 unemployment rate of 7.8%. While the 157,000 jobs gain for January was consistent with the 196,000 jobs gain now estimated for December 2012, those increases were not consistent with the new 247,000 jobs gain reported for November. The consistent and comparable monthly gain for adjusted November payrolls was 205,000, as noted in the Payroll Survey Detail section.
Unemployment Numbers Simply Are Not Comparable Month-to-Month. Except for the once-per-year December release—as seen last month—the BLS publishes no revised seasonally-adjusted data on a monthly basis for the household survey, even though those revisions are made and are available for publication every month, as part of the concurrent-seasonal-factor process.
As discussed frequently (see Commentary No. 473, Commentary No. 461, and Commentary No. 451, for example), the revisions to earlier data from the concurrent-seasonal-factor process can be significant. As a result, month-to-month changes in seasonally-adjusted unemployment rates are meaningless—not determinable under current BLS reporting policies—and use of monthly comparisons simply should be avoided. At this time, the BLS does make usable comparative data available to the public.
Payroll Growth Is Consistent Only One-Month Back, With Heavy Distortions Usual. With the payroll series, the level of payrolls is released for the headline month, and for the two prior months, on a consistent basis. That means that only the current headline month-to-month change and the change for the prior month are consistent and comparable. Unlike with the household survey, however, the BLS makes available the models and data so that others can calculate the payroll revisions, and we have done so below. The first graph below reflects that, but it was the final one for the distortions following the 2011 benchmark revision. All these data were reset with current benchmark revisions.
Conceivably, the shifting and unstable seasonal adjustments could move 90,000 jobs or more from earlier periods and insert them into the current period as new jobs, without there being any published evidence of that happening. The graph suggests that something along those lines happened as recently as the headline November 2012 reporting.
Current Benchmark Revision. As shown in the next graph, however, the distortions in the post-March 2012 benchmark environment already are showing, even though the data are based on the initial public reporting of the benchmark revision. The reason for this is that the benchmark revision actually was run internally by the BLS, based on October 2012 numbers. With subsequent internal runs in November, December and January, where the January version was published on February 1st, three months of revision already have skewed the data, as shown in the next graph. Without distortions, the plotted line would be flat and at zero.
Reflecting the benchmark revisions and unusual patterns of reporting in the revised data, as unpublished, seasonally-adjusted monthly payroll numbers currently are shifting jobs growth out of the July-to-October period into the current November-to-January period, bloating reported payroll growth in the most-recent months. If the concurrent seasonal factors were stable, the line in the new graph would be flat and at zero, as used to be the case when seasonal factors were fixed in advance of the monthly reporting. As shown in the graph, though, the month-to-month changes in the concurrent seasonal factors are highly volatile, consistent with unstable seasonal factors in a still highly unstable economy.
The issues with the BLS’s concurrent-seasonal-factor adjustments and related inconsistencies in the monthly reporting of the historical time series are discussed and detailed further in the ShadowStats.com posting on May 2, 2012 of Unpublished Payroll Data.
Note: As discussed in prior writings (see Hyperinflation 2012, for example), seasonal-factor estimation for most economic series has been distorted severely by the extreme depth and duration of the economic contraction. These distortions are exacerbated for payroll employment data based on the BLS’s monthly seasonal-factor re-estimations and lack of full reporting.
Where the BLS recalculates the monthly seasonal factors each month for payroll employment, going back a number of years, it only publishes the revised data for the last two months of reporting. The benchmark revision that accompanied the release of January 2013 payrolls, in theory, included a full update of the revised concurrent seasonally-adjusted data (actually it is off by three months). In the preceding graph, though, the latest revised (but not published by the BLS) adjusted payroll data show volatile, monthly seasonal-adjustment distortions of more than 40,000 jobs per month, with previously-reported payroll employment being shifted from the third-quarter 2012 into the three most recent months of reporting. If seasonal-adjustment factors were stable in month-to-month reporting, which they should be under normal circumstances, then the graph of differences would be flat and at zero.
Note: A further issue remains that the month-to-month seasonally-adjusted payroll data have become increasingly meaningless, with reporting errors likely now well beyond the official 95% confidence interval of +/- 129,000 jobs in the reported monthly payroll change. Yet, the media and the markets tout the data as meaningful, usually without question or qualification.
Birth-Death/Bias Factor Adjustment. Despite the ongoing and regular overstatement of monthly payroll employment—as evidenced usually by regular and massive, annual downward benchmark revisions (2011 and 2012, excepted)—the BLS generally adds in upside monthly biases to the payroll employment numbers. The process was created simply by adding in a monthly “bias factor,” so as to prevent the otherwise potential political embarrassment of the BLS understating monthly jobs growth. The “bias factor” process resulted from an actual such embarrassment, with the underestimation of jobs growth coming out of the 1983 recession. That process eventually was recast as the now infamous Birth-Death Model (BDM), which purportedly models the effects of new business creation versus existing business bankruptcies.
January 2013 Bias. The not-seasonally-adjusted January 2013 bias was a negative 314,000, versus a negative 4,000 in December 2012, and a negative 367,000 in January 2012. The aggregate upside bias for current year appears to have been upped to about 620,000, or a monthly average of roughly 52,000. That likely will increase in the months ahead. As of the twelve months ended December 2012, the bias had been 535,000, or a monthly average of roughly 45,000 jobs created out of thin air, on top of some indeterminable amount of other jobs that are lost in the economy from business closings. Those losses simply are assumed away by the BLS as part of the BDM, as discussed below.
Problems with the Model. The aggregated upside annual reporting bias in the BDM reflects an ongoing assumption of a net positive jobs creation by new companies versus those going out business. Such becomes a self-fulfilling system, as the upside biases boost reporting for financial-market and political needs, with relatively good headline data, while often also setting up downside benchmark revisions for the next year, which traditionally are ignored by the media and the politicians. Where the BLS cannot measure meaningfully the impact of jobs loss and jobs creation from employers starting up or going out of business, on a timely basis (within at least five years, if ever), such information is estimated by the BLS along with the addition of a bias-factor generated by the BDM.
Positive assumptions—commonly built into government statistical reporting and modeling—tend to result in overstated official estimates of general economic growth. Along with happy guesstimates, there usually are underlying assumptions of perpetual economic growth in most models. Accordingly, the functioning and relevance of those models become impaired during periods of economic downturn, and the current downturn has been the most severe—in depth as well as duration—since the Great Depression.
Indeed, historically, the BDM biases have tended to overstate payroll employment levels—to understate employment declines—during recessions. There is a faulty underlying premise here that jobs created by start-up companies in this downturn have more than offset jobs lost by companies going out of business. So, if a company fails to report its payrolls because it has gone out of business (or has been devastated by a hurricane), the BLS assumes the firm still has its previously-reported employees and adjusts those numbers for the trend in the company’s industry.
Further, the presumed net additional “surplus” jobs created by start-up firms, get added on to the payroll estimates each month as a special add-factor. These add-factors are set now to add an average of about 52,000 jobs per month in the current year, but the actual overstatement of monthly jobs likely exceeds that number by a significant amount. With the economy slowing anew, with growth below consensus expectations, the next hope for relief in current over-reporting would be the 2013 benchmark revision, due to be published in February of 2014.
HOUSEHOLD SURVEY DETAILS. As discussed in the Opening Comments and Executive Summary, the January 2013 household-survey data are inconsistent with December 2012 reporting. First, revised annual population controls were introduced, with the effect of destroying December-to-January data comparability. As noted by the BLS in its February 1st press release, “Data users are cautioned that these annual population adjustments affect the comparability of household data series over time.”
Specifically, per the BLS, the population re-estimation simply “… increased the estimated size of the civilian noninstitutional population in December by 138,000, the civilian labor force by 136,000, employment by 127,000, unemployment by 9,000, and persons not in the labor force by 2,000. The total unemployment rate, employment-population ratio, and labor force participation rate were unaffected.” No parallel revisions were made to historical data.
Second, the BLS resumed its unconscionable practice of revising previous estimates that are consistent with current reporting, but then publishing only the current number, not the consistent prior-period revisions. As discussed in upcoming sections, seasonally-adjusted month-to-month comparisons of components in the household survey usually have no meaning.
Headline Household Employment. Based on the household survey, which counts the number of people with jobs, as opposed to the payroll survey that counts the number of jobs (including multiple job holders more than once), January 2013 employment declined by 110,000 for the month, after adjusting for month-to-month distortions created by the update of population controls. Ignoring the resetting of the base data for January 2013, which added 127,000 people to the official employment count, the non-comparable month-to-month numbers showed employment increased by 17,000 for the month. Even so, as discussed in the Unemployment Rates section, the seasonally-adjusted household numbers in January were not otherwise comparable with December’s reporting.
Unemployment Rates. The reported January 2013 seasonally-adjusted headline (U.3) unemployment rate of 7.92% simply was not comparable to the reported 7.85% (rounds to 7.8%) unemployment rate of December 2012. Although the updated population estimates reset the base data of the January 2013 household survey, the January headline number still would have been 7.92% if consistent population estimates were used. Instead, the problem in unemployment-rate comparability is tied to the use of concurrent-seasonal-factor adjustments.
When the seasonally-adjusted January 2013 unemployment data were calculated, consistent, new seasonal factors also were recalculated for December 2012 and prior months. Based on the new seasonal factors, there was revised December unemployment rate that would be consistent with January’s headline reporting. Although the BLS recalculates the December unemployment rate, it does not publish it; it leaves intact the now-inconsistent number that had been reported for December. This process is repeated every month, except in December when a revised and consistently seasonally-adjusted series is published. The misreporting process begins anew with the reporting of the unemployment data for each January (see the discussions in Commentary No. 451 and Commentary No. 487 for further detail).
Given these reporting issues, the official +/- 0.23 percentage-point 95% confidence interval for the monthly headline U.3 number generally is meaningless, in the context of comparative month-to-month reporting. On an unadjusted basis, however, the unemployment rates are not revised and are consistent. January’s unadjusted U.3 unemployment rate was 8.5%, versus December’s 7.6%.
The broadest unemployment rate published by the BLS, U.6 includes accounting for those marginally attached to the labor force (including short-term discouraged workers) and those who are employed part-time for economic reasons (they cannot find a full-time job).
Reflecting a seasonally-adjusted gain in individuals working part-time for economic reasons, as well as something of an offsetting decline in short-term discouraged workers, the January 2013 U.6-unemployment rate held at a seasonally-adjusted 14.4%, versus December, but, again, the monthly seasonally-adjusted numbers were not comparable. The unadjusted January U.6 rate rose to 15.4%, versus 14.4% in December.
Discouraged Workers. The count of short-term discouraged workers (never seasonally-adjusted) was 804,000 in January 2013, but that was not comparable with the 1,068,000 of December, the 979,000 in November, or the 813,000 in October, thanks to the change in population assumptions.
The current official discouraged-worker number reflected the flow of the unemployed, or the balance of the headline unemployed—increasingly giving up looking for work—leaving the headline U.3 unemployment category and being rolled into the U.6 measure as short-term “discouraged workers,” versus those moving from short-term discouraged-worker status into the netherworld of long-term discouraged-worker status. It is the long-term discouraged worker category that defines the ShadowStats-Alternate Unemployment Measure.
In 1994, “discouraged workers”—those who had given up looking for a job because there were no jobs to be had—were redefined so as to be counted only if they had been “discouraged” for less than a year. This time qualification defined away a large number of long-term discouraged workers. The remaining short-term discouraged workers (those discouraged less than a year) were included in U.6.
Adding back into the total unemployed and labor force the ShadowStats estimate of the growing ranks of excluded, long-term discouraged workers, broad unemployment—more in line with common experience, as estimated by the ShadowStats-Alternate Unemployment Measure—held at a series record-high of 23.0% in January 2012, for a second month. That same level in December was up from 22.9% in November and October, and up from 22.8% in September, reflecting the increasing toll of unemployed leaving the headline labor force. The ShadowStats alternate estimate generally is built on top of the official U.6 reporting, and tends to follow its relative monthly movements. Accordingly, the alternate measure often will suffer some of the same seasonal-adjustment woes that afflict the base series, including underlying annual revisions.
As seen in the following graph, however, there continues to be a noticeable divergence in the ShadowStats series versus U.6. The reason for this is that U.6, again, only includes discouraged workers who have been discouraged for less than a year. As the discouraged-worker status ages, those that go beyond one year fall off the government counting, even as new workers enter “discouraged” status. With the continual rollover, the flow of headline workers continues into the short-term discouraged workers category (U.6), and from U.6 into long-term discouraged worker status (a ShadowStats measure). The movement of the discouraged unemployed out of the headline labor force has been accelerating. See the Alternate Data tab for more detail.
CAUTION: With the publication of January 2013 data, month-to-month comparisons of the various unemployment rates are meaningless, again, due to deliberate inconsistencies in BLS reporting. The unemployment data are comparable only once per year, in December, when the seasonally-adjusted data are revised and reported on a consistent monthly basis.
As discussed in previous writings, an unemployment rate around 23% might raise questions in terms of a comparison with the purported peak unemployment in the Great Depression (1933) of 25%. Hard estimates of the ShadowStats series are difficult to generate on a regular monthly basis before 1994, given the reporting inconsistencies created by the BLS when it revamped unemployment reporting at that time. Nonetheless, as best estimated, the current ShadowStats level likely is about as bad as the peak actual unemployment seen in the 1973 to 1975 and in double-dip recession of the early-1980s. The Great Depression unemployment rate of 25% was estimated well after the fact, with 27% of those employed working on farms. Today, less that 2% of the employed work on farms. Accordingly, a better measure for comparison with the ShadowStats number would be the Great Depression peak in the nonfarm unemployment rate in 1933 of roughly 34% to 35%.
CONSTRUCTION SPENDING (December 2012)
December Construction Spending Continued to Stagnate. The trend of low-level stagnation in construction spending was seen again in December, along with a continuing pattern of upside revisions to prior-period estimates. The latest numbers likely received some relative boost from unseasonably warm weather in December, along with increasing reconstruction activity necessitated by Hurricane Sandy’s destruction. Some temporary upside effects from that rebuilding are likely to continue for the next quarter or so. Where the month-to-month gain in spending was statistically-insignificant, the circumstance was more negative after adjustment for inflation.
Official Reporting. The Census Bureau reported February 1st that the total value of construction put in place in the United States during December 2012 was $885.0 billion, on a seasonally-adjusted—but not inflation-adjusted—annual-rate basis. That estimate was up for the month by a statistically-insignificant 0.9% +/- 1.9% (all confidence intervals are at a 95% level), from an upwardly revised $876.9 (previously $866.0) billion in November. Before prior-period revisions, the monthly December gain was 2.2%. In revision, November spending rose by 0.1% (previously a 0.3% contraction).
Aggregate December 2012 construction spending was up year-to-year by a statistically-significant 7.8% +/- 2.1%, with November annual growth revising to 9.1% (previously 7.7%).
A fair portion of December’s 7.8% annual gain, however, was covered by increases in actual construction costs. The Bureau of Economic Analysis (BEA) continued to underestimate year-to-year inflation in nonresidential and residential investment at 1.9% in its “advance” estimate of fourth-quarter 2012 GDP. Closer to real-world experience, McGraw Hill’s Engineering News-Record currently estimates year-to-year change in January 2013 construction costs at 2.9%, with labor costs up by 3.1%.
The statistically-insignificant 0.9% gain in monthly December 2012 construction spending included a 1.4% drop in public construction spending, which had revised to a 0.1% (previously 0.4%) decline in November. December private construction rose by 2.0% in the month, versus a revised 0.2% monthly gain (previously a 0.2% decline) in November. The accompanying graphs show the 0.9% monthly increase in December total construction, with private residential construction up by 2.2%, private nonresidential construction up by 1.8% and public construction down by 1.4% for the month.
The next two graphs reflect total construction spending through December 2012 (again, not adjusted for inflation), along with the January 3013 reporting of, and benchmark revision to, construction employment, per the Bureau of Labor Statistics. Construction employment was one of the payroll employment series revised to the upside. As seen in the second graph, previosly flat construction employment now has taken on a more-positive slope.
As with total construction spending number, the recent gain in activity most likely was influenced heavily by Hurricane Sandy-related reconstruction and by unseasonably good weather. The seasonally-adjusted employment level for January 2013 was 5.731 million, up by 0.5% for the month, versus 5.703 million in December 2012, which also was up by 0.5% month-to month. Pre-benchmark, the December jobs level was 5.564 million, but that still was up by 0.5% for the month. Year-to-year employment growth was 2.1% in January 2013, versus a 1.0% gain December 2012, which had been up by 0.8% pre-benchmark.
The preceding two graphs cover private residential construction, along with housing starts as reported for December 2012 (see Commentary No. 495 for detail). The difference in the graphs is the smoother pace of actual spending (not-adjusted-for-inflation), instead of the more-irregular monthly variation in the count of physical monthly starts.
The last two graphs in this series show the patterns of activity in private nonresidential construction spending and in public construction spending. The public construction spending is 98% nonresidential.
Weaker Economic and Stronger Inflation Data Are Likely. Beyond the dissipating effects of the repair, replacement and reconstruction activity generated by Hurricane Sandy, and in anticipation of the likely negative impact of expanded QE3 and the ongoing fiscal crisis/debt-ceiling negotiations on the currency markets, reporting in the months and year ahead generally should reflect higher-than-expected inflation and indicate weaker-than-expected economic results. Increasingly, previously unreported economic weakness should continue to show up in prior-period revisions.
Significant reporting-quality problems continue with most major economic series. Headline reporting issues remain tied largely to systemic distortions of seasonal adjustments, distortions that have been induced by the still-ongoing economic turmoil of the last five years. The recent economic collapse has been without precedent in the post-World War II era of modern economic reporting. These distortions have thrown into question the statistical-significance of the headline month-to-month reporting for many popular economic series. In any event, where reported numbers are too far removed from common experience, they tend to be viewed by the public with extreme skepticism.
Still, recognition of an intensifying double-dip recession continues to gain, while recognition of a mounting inflation threat has been rekindled by the Fed’s monetary policies. The political system would like to see the issues disappear, and it still appears to be trying to work numerical slight-of-hand with series such as the GDP and related projections of the federal budget deficit. The media do their best to avoid publicizing unhappy economic news or, otherwise, they put a happy spin on the numbers. Pushing the politicians and media, the financial markets and related spinmeisters do their best to avoid recognition of the problems for as long as possible, problems that have horrendous implications for the markets and for systemic stability, as discussed in Hyperinflation 2012 and No. 485: Special Commentary.
U.S. Trade Balance (December 2012 and Annual). The December 2012 trade deficit detail will be released on Friday, February 8th. Where the U.S. trade deficit continues in fundamental deterioration, the December numbers should be the last to have been impacted meaningfully by disruptions from Hurricane Sandy. Issues likely involved the flow of goods as well as the flow of paperwork on imports and exports through the Port of New York and surrounding facilities. The goods and reporting flows should be back to normal or otherwise caught up in the December reporting.
Any significant narrowing or widening of the December trade deficit—beyond market expectations of a slight narrowing in December’s monthly trade shortfall—respectively would tend to boost or impair the outlook for the first revision to the slightly negative initial estimate of fourth-quarter 2012 GDP. That GDP revision is due for release on February 28th.
The nominal (not adjusted for inflation) annual merchandise trade deficit for 2012 is on track to widen by a small amount versus 2011, with imports increasing at a faster pace than exports. Including the largely guessed at “surplus” in services, the nominal, annual deficit in goods and services is on track for a minimal narrowing.