JWSGS - DECEMBER 2004 EDITION
Issue Number Two
December 8, 2004
_____
STAGFLATION SIGNALS INTENSIFY
RECESSION WARNING POSSIBLE IN NEXT TWO MONTHS
THE DUAL DEFICITS AND THE DOLLAR DO MATTER
With October CPI inflation raging at an annualized 8.1%, and with November payrolls unchanged within the recognized scope of statistical significance, the popular, poor-quality numbers moved in tandem with the better-quality reports for a month, showing stagflation. Even so, in recent weeks,the more-reliable leading indicators have generated intensified warning signals of slowing activity and rising prices. A period of stagflation has been signaled and is in place, but something more troubling may be in the works: an inflationary recession.
Signals from the good-quality (and some of the other) leading indicators can be combined to generate early warnings of major upturns and downturns in business activity. If current trends persist, the first solid warning signals of a new recession are possible in the next month or two. Annual growth patterns in the November new orders component of the purchasing managers survey, October building permits and October real average weekly earnings already are borderline.
Such a warning, if signaled, would indicate a protracted period of contracting commercial activity beginning in first- or second-quarter 2005. While the downturn would be reflected in vital indicators such as federal tax receipts, corporate profits, payroll employment and industrial production, GDP growth would slow but not turn negative, unless the contraction were particularly severe. The early warning system will be detailed at such time as a signal is generated.
Despite the slowing economy, inflation fundamentals continue to deteriorate. The signaled inflation is not driven by strong economic demand, but rather by commodity price distortions (oil), a record low in the U.S. dollar and something close to criminal fiscal mismanagement of the federal government, which eventually will force inflationary monetary policy.
In that former malarial swamp on the Potomac, there is a popular belief that the budget deficit, the trade deficit and a weak dollar "don't matter." Such a misperception is only a temporary artifact of global forbearance of decades of fiscal abuses and economic shortsightedness by a number of U.S. administrations. That forbearance reflects the U.S. dollar's reserve-currency status and the United States' dominant economic and military power.
The trade and budget deficits, however, both have evolved to levels that are out of control, and the recent intensification of the dollar's weakness reflects this. With the U.S. credit and equity markets dependent on foreign capital for liquidity, a precarious vulnerability has developed, tied to the above circumstance. Accordingly, any particularly negative surprises on the budget or trade front have the potential for triggering unusually volatile reactions in the currency and credit markets. Risks would be to the downside for the dollar and to the upside for interest rates.
Despite a couple of major reports to the contrary, financial market misperceptions of solid economic growth and contained inflation continue. As a result, the broad reporting risk to upcoming economic releases continues to weigh in favor of indicators of economic growth coming in below, and indicators of inflationary pressures coming in above, market expectations. Where our analysis offers particular insight into likely reporting biases in the next releases of key series, such is noted in the text accompanying the specific series analysis.
This month's "Reporting Focus" is on the Trade Balance in Goods and Services, a series key to assessing a primary systemic imbalance in the U.S. economy.
THE BIG THREE MARKET MOVERS
(Each of these series is explored in the background article "A Primer On Government Economic Reports," available on the home page JWBGN.)
Employment/Unemployment -- The popularly followed November 2004 unemployment rate U-3 eased to 5.41% from 5.46% in October, seasonally adjusted, well within the published +/- 0.2% margin of error of the household survey. The broader U-6 measure, also seasonally adjusted, reversed its previous move, dropping a significant 0.3% to 9.4% for November. Including the long-term "discouraged workers" defined away during the Clinton administration, total unemployment is roughly 12.4%.
The November payroll survey showed a weaker-than-expected, seasonally-adjusted gain of 112,000 (58,000 net of revisions). Given a published 90% confidence interval of +/- 108,000 jobs, November's gain was indistinguishable from "unchanged." October's previously-reported gain of 337,000 jobs was revised to 303,000. The prior month's seasonal-factor machinations appeared to stabilize in the most recent report.
Next Release (January 7): Despite what likely will be a general lowering of consensus forecasts for December payrolls, odds still favor a weaker-than-expected gain in payroll employment and some rebound in the unemployment rate. The December payroll bias factor, once again, should be below average.
The bias factor, however, turns sharply negative in January, suggesting a very sharp downside surprise (probably a contraction) in January payrolls (February 4), with an offsetting, upside surprise rebound for February payrolls (March 4).
Upcoming revisions to the household survey (January 7) will restate seasonal factors back five years, but the underlying, unadjusted data will not change. The upcoming benchmark revision to the payroll survey (February 4) will revise the level of March 2004 payrolls upward by about 236,000, with guesstimated proportionate changes going both backward and forward in time. Seasonal factors and bias estimates also will get revamped in this annual re-jiggering of the numbers.
Gross Domestic Product (GDP) -- The "preliminary" estimate of third-quarter 2004 annualized real GDP growth revised upward to 3.9% from the "advance" 3.7%, a change that was little more than statistical noise in this otherwise statistically worthless series. Although the official third-quarter growth of 3.9% is up from first-quarter growth of 3.3%, reported annual growth is softening, where year-to-year gains in the first three quarters of 2004 have eased from 5.6% to 4.7% to 3.9%.
Published along with the "preliminary" estimate were first cuts of real growth in Gross National Product (GNP) and Gross Domestic Income (GDI) for the third-quarter. The "advance" estimate is so much of a guesstimate, the Bureau of Economic Analysis does not even attempt to publish these less-followed series at the time.
While both the GNP and GDI showed similar but weaker patterns of softening annual growth, compared to the GDP, quarterly changes were greater (GNP = 4.2%) or the same (GDI = 3.9%). The stronger quarterly GNP growth is unusual but happens on occasion, given special occurrences and/or peculiar reporting of international interest and dividend flows.
As discussed in the background articles, current GDP reporting is overstated by about 3%, which would place actual annualized growth at roughly 0.9% plus or minus about 3%, given published 90% confidence intervals.
Next Release "Final" Estimate (December 22): At present we have noted no particular bias that would affect the next revision beyond the realm of statistical noise. The "advance" report on fourth-quarter 2004 GDP is due for release on January 28, however, and is a good bet to show weaker growth than will be commonly expected. As an aside, the term "final" revision is a misnomer, since these data are subject to many years of annual and benchmark revisions, likely starting in July 2005.
Consumer Price Index (CPI) (No new reporting since the November BGN Addendum) -- After gaining just 0.2% in September, the seasonally-adjusted CPI-U rose a stronger-than-expected 0.6% (0.5% unadjusted), reflecting some reporting catch-up. October's year-to-year inflation jumped to 3.2% from 2.5% in September, which still remains far short of reality. The 3.2% is the change in the not-seasonally-adjusted index. Suggesting some seasonal-factor distortion, however, the year-to-year change in the seasonally-adjusted CPI was 3.3%, up from 2.4%.
Using the CPI's original (pre-Clinton Era) methodological approach of a fixed basket of goods (vs. substitution of hamburgers for steak as estimated by geometric weighting) would leave year-to-year inflation at about 5.9% instead of 3.2%.
The "experimental" Chained Consumer Price Index (C-CPI-U), the fully substitution based CPI that presumably is the eventual replacement for current CPI reporting, showed a 2.7% year-to-year gain, up from 2.1% in September. Interestingly, the C-CPI-U did not have its regular highlighting in the monthly press release.
Next Release (December 17): While inflation pressures will continue to offer upside surprises in many reports, given October's reporting catch-up and the usual year-end patterns, look for the next report to come in at or below consensus forecasts, which should be relatively low. Seasonally-adjusted November CPI-U would have to contract by more than 0.3% for the unadjusted 3.2% year-to-year inflation rate to soften. Changes above or below a 0.3% contraction will directly add to or subtract from October's annual growth.
The CPI, in theory, is a monthly average, surveyed throughout the month. Invariably, oil price volatility tends to have unrealistic impact on eventual CPI reporting. Where cost pressures flow through to the retail sector, common experience is that businesses are much faster in raising prices than they are in lowering them. In the happy world of CPI surveying and methodology, retail firms appear to raise prices reluctantly but lower them eagerly. Accordingly, any drop in monthly retail-level energy costs will show up quickly in monthly CPI reporting.
OTHER TROUBLED KEY SERIES
To varying degrees, the following series have significant reporting problems. Each series will be addressed in a monthly "Reporting Focus," with the monthly Trade Balance report covered this month. In addition to the big three, other series that have been detailed are The Federal Deficit (background articles) and Consumer Confidence (November 2004 BGN). New series added this month to this section are: Real Average Weekly Earnings and the Purchasing Managers Survey (Non-Manufacturing).
Federal Deficit (While there has been no new reporting on the monthly deficit since the November BGN, the federal debt ceiling was raised, and month-end November federal debt data are available.) -- President Bush signed the $8.2 trillion federal debt ceiling into law on November 19. Not so coincidentally, the $800 million increase over the prior $7.4 trillion ceiling is roughly what can be expected as the GAAP-based accounting deficit (ex-Social Security and Medicare) for fiscal year 2004.
The official deficit for the fiscal year ended September 30, 2004 was $412.3 billion, up from $377.1 billion the year before. For the twelve months ended October, the rolling deficit was $400.0 billion. Gross federal debt as of the end of September was $7.379 trillion, up $596 billion from a year earlier; debt at the end of October was $7.430 trillion, up $557 billion, but constrained by the $7.4 trillion debt ceiling in effect at the time.
With the increased debt ceiling in place, month-end November gross federal debt rose to $7.525 trillion, up $600 billion from November 2003.
GAAP-based reporting still is likely to show an annual shortfall of $800 billion for fiscal 2004, net of Social Security and Medicare accounting, $4.3 trillion including Social Security and Medicare, as discussed in the background articles.
Producer Price Index (PPI) No new reporting since the November BGN) -- The seasonally adjusted October Finished Goods PPI jump of 1.7% (2.2% unadjusted) was more than expected by the markets and reflected catch-up reporting on energy prices. Year-to-year PPI inflation rose to 4.4% in October, up from 3.3% in September.
Next Release (December 10): The monthly variations in this series have a large component of random volatility. Nonetheless, PPI inflation reporting over the next three-to-six months generally should be stronger than consensus forecasts.
Real Average Weekly Earnings -- Often, economic series ignored by the popular financial media turn out to be of surprisingly good quality, in that they generally are not heavily massaged for political considerations. The only basic flaw in this series is its deflation by the CPI-W (which suffers the same underreporting biases as the CPI-U).
In constant, inflation-adjusted dollars, average weekly earnings peaked in 1972. Then, President Nixon abandoned gold and floated the U.S. dollar. As the persistent trade deficit began its ever-intensifying deterioration (see this month's "Reporting Focus"), higher paying jobs moved offshore and real average weekly earnings entered a long-term downward spiral. Eventually, many families that had been supported by one breadwinner needed multiple earners in order to attempt to keep up with the actual cost of living.
Underlying reality has been continued real earnings decline, although such has been partially masked since 1996, when the change in shifting CPI methodology became enough to offset the declines. Nonetheless, reported real average weekly earnings have not grown since the last recession. October's seasonally-adjusted reading of $277.78 (1982 Dollars) was 0.5% below the level in December 2001. October 2004 also was down 0.4% from September and down 0.4% from October 2003. Allowing for the CPI biases, annual change in this series is borderline for generating a recession warning signal.
Retail Sales (No new reporting since the November BGN) -- The monthly gain in October retail sales was 0.2% +/- 0.8%, up a strong 7.6% from October 2003. Inflation-adjusted growth in retail sales below 1.8% (using the standard CPI for deflation) signals recession. Annual growth is closing in on that level.
Next Release (December 13): If the report reflects a soft start to the holiday shopping season--consistent with anecdotal evidence--CPI-adjusted annual growth could slip below 1.8% by January, signaling recession in the first half of 2005.
Industrial Production (No new reporting since the November BGN Addendum) -- Helped by utility usage, seasonally-adjusted October industrial production rose 0.7% (0.8% net of revisions) with year-to-year growth bumping up from a revised 4.5% in September to 5.2%. As a coincident indicator to the economy, it suggests ongoing growth in the manufacturing sector, reasonably consistent with the still positive readings out of the overall purchasing managers survey.
Next Releases(December 14/22): A generally milder-than-normal November should add a downside risk to consensus estimates of November production (December 14), since reporting depends heavily on electricity usage. An annual benchmark revision, incorporating new data for 2002 and 2003, will be published on December 22. Historical growth patterns should be weaker than previously reported. Details on the revision and highlights of series reporting problems will be covered in next month's "Reporting Focus."
New Orders for Durable Goods -- October seasonally adjusted orders were down 0.4% (up 0.3% net of revisions) from September, which was up 0.9% (previously 0.2%) from August. Year-to-year growth in October slowed to 2.8% (less than the CPI) against an upwardly-revised 10.4% for September. The widely followed Nondefense Capital Goods Orders fell by 3.3% for the month. Monthly volatility is high for this series, and another two weak months would be needed to send off a recession warning signal.
This series used to be one of the better leading indicators of broad economic activity, when smoothed using a three-month moving average. Then the semi-conductor industry stopped reporting new orders, and the series' quality fell apart.
The related Factory Orders report, published usually a week or so after the durable goods report, adds little of value and is worth ignoring other than for any revisions to the durable goods orders.
Purchasing Managers Survey (Non-Manufacturing) -- Published by the Institute for Supply Management (ISM), there is nothing unusually wrong with this survey of the service industry, except it does not have much meaning. Unlike its older counterpart, the manufacturing survey, if service companies such as law firms, hospitals or fast-food restaurants have "increased orders," that does not necessarily mean the economy is picking up.
The Prices Paid component diffusion index, however, is a general indicator of inflationary pressures, and the November measure was up 22.2% year-to-year. On a three-month-moving-average basis, the Prices index was up 20.6% from November 2003, versus a 21.4% annual gain in October.
Trade Balance (No new reporting since the November BGN) -- The September trade deficit in goods and services narrowed to $51.6 billion from August's revised $53.5 billion (previously $54.0 billion), leaving the third-quarter deficit 3.6% wider than the second quarter's and 27.3% worse than third-quarter 2003. Continued sharp deterioration is likely. (See this month's "Reporting Focus.")
Consumer Confidence -- Both the November Consumer Confidence and Consumer Sentiment surveys turned negative year-to-year (-2.2% and -1.0% respectively) for the first time in over a year. As lagging, not leading, indicators, these numbers confirm the economy slowed sharply in the late third quarter or early fourth quarter of 2004.
BETTER-QUALITY NUMBERS
The following numbers are generally good-quality leading indicators of economic activity and inflation that offer an alternative to the politically hyped numbers when the economy really is not so perfect. In some instances, using a three-month moving average improves the quality of the economic signal and is so noted in the text.
Economic Indicators
Purchasing Managers Survey (Manufacturing) - New Orders -- Although the November reading rose to 61.5 from 58.3, the numbers were seasonally adjusted with government estimates. Seasonal factors often can be misleading in the reporting of month-to-month change, an issue overcome by using year-to-year change on a three-month-moving-average basis. The November index was down 13.9% from the reading a year before, compared with October's 9.8% drop. Using a three-month moving average, November was down 9.6% versus October's 4.5%.
Published by the Institute for Supply Management (ISM), the New Orders component of the Purchasing Managers Survey is a particularly valuable indicator of economic activity. The index is a diffusion index, where a reading above 50 indicates rising new orders. The current year-to-year downturn in the three-month moving average is a borderline recession warning signal.
Help Wanted Advertising Index (HWA) -- Help wanted advertising bottom-bounced in October, rising to 37 from September's 43-year low of 36. The year-to-year change in the three-month moving average had been negative for two months in August and September, with a bounce to zero annual growth in October.
Published by the Conference Board, the HWA is a reliable leading indicator of employment activity. The series continues to be borderline in signaling a pending recession.
Building Permits (No new reporting since the November BGN Addendum) -- Seasonally adjusted October building permits fell 0.7% from September (down 1.0% net of revisions), with year-to-year change down by 1.5%. On a three-month moving average basis, annual growth dropped from September's 4.0% to 0.5%, the lowest level since November 2001.
As a leading indicator of economic activity, the series is showing a meaningful slowdown and has become borderline for generating a recession warning.
Inflation Indicators
Purchasing Managers Survey (Manufacturing) - Prices Paid -- The November Prices Paid diffusion index softened to 74.0 from October's reading of 78.5. The index has been signaling a strong rise in inflation since late in 2003. On a three-month moving average basis, the year-to-year change in the Prices Paid Index rose by 28.0%, down from October's 40.9% growth.
Published by the Institute for Supply Management (ISM), the Prices Paid component of the Purchasing Managers Survey is a reliable leading indicator of inflation activity. The prices paid index is a diffusion index, where a reading above 50 indicates rising inflation. While reduced in magnitude, the November inflation signal remains strong.
Oil Prices -- West Texas Intermediate Spot (St. Louis Fed) tumbled back by 8.8% in November, after soaring 15.7% for October. The year-over-year spot price is up by 55.9%. While oil prices have backed off recent highs, broad inflation costs still will rise more than the consensus expectations.
Oil price changes permeate costs throughout the economy, ranging from transportation and energy costs, to material costs in the plastics, pharmaceutical, fertilizer, chemical industries, etc. Where the effects of recent oil price volatility will be picked up quickly in the CPI, on the downside (see the CPI), broader inflation -- even as currently understated --will be stronger than commonly predicted for the next six-to-nine months, as a result of the net run-up in oil prices, to date.
U.S. Dollar -- November's dollar average (the Federal Reserve's Major Currencies U.S. Dollar Index) was down 2.1% from October and down 10.1% from November 2003. December's opening currency trading already is 1.7% below the November average, matching the all-time low seen for the U.S. currency in 1995, and down 16.6% from the disorderly sell-off levels that helped trigger the financial-market break in October 1987. New record lows for the dollar are likely in the months ahead, despite any near-term official or unofficial supportive intervention by any central bank(s). Generally, the weaker the dollar, the greater will be the ultimate inflation pressure and the eventual liquidity squeeze in the U.S. capital markets.
DECEMBER'S "REPORTING FOCUS" -- TRADE BALANCE IN GOODS AND SERVICES
The monthly reporting of the U.S. International Trade in Goods and Services has two components of widely varying quality. The goods component is of eventual good quality, backed by a full paper trail, in theory, of all import and export transactions. Since the import detail has a better paper trail (collection of duties, etc.), U.S. export numbers are honed through a series of monthly crosschecks with trade data and better paper trails published by major U.S. trading partners. The services component is largely a guesstimate provided by the same people who give us the GDP report. Services trade reporting is of significantly worse quality than the goods reporting.
Although now published jointly by the Census Bureau and the Bureau of Economic Analysis, in the pre-Clinton Era, the monthly trade data were published only for merchandise and only by the Census Bureau. At the time, the services trade balance was guesstimated only quarterly, for purposes of GNP and current account balances. Chances that monthly guessing would be more meaningful than quarterly guessing were zero, but since the merchandise trade deficit was constantly getting worse, and the services sector was in surplus, publishing the two series together would make what would still be a net monthly deficit appear smaller. That was done. The services surplus still is meaningless on a monthly basis and has been shrinking since the late-1990s, often contributing to the widening of the monthly trade shortfall.
In early reporting on the merchandise side, there was a substantial problem in the recorded timing of exports and imports. Although an import actually took place in April, for example, it got counted only in the month in which the paperwork cleared U.S. Customs, in April, perhaps, or June or even December. The reported monthly trade flows were nothing of the kind; they were instead a record of U.S. Customs paperwork flows, which varied sharply month-to-month and quarter-to-quarter. Distortions were so bad that the dependent quarterly GNP reports were thrown off sharply.
The problem also was bad enough to allow an easy manipulation of the paperwork flows and reported trade deficit in late-1987 and early-1988, as part of the effort by the Federal Reserve and U.S. Treasury to stabilize the U.S. dollar in the air pocket that followed the disorderly markets of October 1987.
At that time, the Census Bureau recognized and addressed the problem, reducing paperwork flow variations from double-digit billions of dollars to less than $2 billion per month, at present.
The paperwork "carryover" is estimated each month to be the same as was seen the month before, and it is revised and corrected for known carryover in the next month's reporting. In the annual benchmark revision, all known residual carryover effects are restated into the correct months.
The services sector also is revised monthly, but its quality does not improve. Recently, for example, methodology was changed so as to introduce "smoothing" techniques in insurance payment and premium flows. This was done primarily to remove the volatility seen in the services sector and GDP in the aftermath of the Sept. 11 terrorist attacks.
In general, though, since the better-quality merchandise numbers still outweigh the poor-quality service sector numbers, the reported monthly deficits tend to be meaningful. At a record level now, they are going to get worse.
The decline in the dollar's value that follows a deepening trade deficit is not quite the self-correcting mechanism economic theorists would like to think. More than price drives demand for non-commodity goods manufactured outside the United States. Quality and features are important. With the use of creative hedging tools, importers can do much to mitigate the impact of a weak dollar.
Does the trade deficit matter? In broad economic terms, it is a direct subtraction from GDP. In magnitude, the elimination of today's record trade deficit would add over 5% to current GDP and create a net 7.3 million new jobs. That would be enough to eliminate the discouraged workers, who cannot find jobs in areas of the country where manufacturing has shifted offshore. That is enough to bring anactual unemployment rate of about 12.4% down to around 7.8%, and the official 5.4% U-3 unemployment down to around 3.5%. Unfortunately, the outlook for the U.S. trade deficit is not for a turnaround.
SCHEDULED "REPORTING FOCUS" FOR JANUARY -- INDUSTRIAL PRODUCTION
The oldest, regularly published government economic series, the Federal Reserve's Index of Industrial Production was viewed at one time as the equivalent of today's GDP. Where electricity consumption is used as a measure for the product output of some industries, unusual weather patterns can inflate or deflate reported production artificially. Weather patterns over the last several years have become increasingly unstable.
The January BGN is scheduled for release on Wednesday, January 12, 2005. Subscribers will be advised immediately by e-mail of its posting on the website, as well as being advised of any interim alerts that might predate the January edition.
All of us at "Behind the Government's Numbers" extend our best wishes to you and your families for a joyous holiday season and for a happy, healthy and prosperous New Year!