With apologies to Bonddad:>)
Looking through Bonddad's diary last night on the payroll employment differences between Clinton and Bush, I've decided its time to clarify something about the governments economic reporting.
The best way to do this is to point all of you to this website, John Williams', Shadow Government Statistics. For those that are interested in government economic reporting and just how wrong it really is then I suggest you read through the Government Reporting economic primers on this website for a complete background.
You can see from the first chart on the site that the CPI, using pre Clinton era statistical methodology that something is amiss.
But this diary is specifically about the July payroll data and how a little statistical manipulation can make a big difference in the data.
The whole article is at this link, I'm sorry its a subscription only link, but I highly recommend that those of you that are interested, that you consider paying for a subscription, or signing up for a trial subscription.
As reported the July payroll employment rose by 207,000. This was a surprise to those of us who follow the data closely because of the weak reported gains in the previous months, in addition to the fact that the monthly bias factors turned negative. Frankly there were expectations that the data would turn negative, I.E. a payroll contraction.
So what happened?
Well, lets see what John has to say about this.
A number of key economic indicators ran counter to our forecasts last month, particularly July employment. Nonetheless, the broad economic outlook has not changed, and the 2005-2007 inflationary recession continues to unfold.
Instead of the predicted outright contraction, July payrolls jumped by 207,000, the strongest monthly gain since April. What happened with the employment, CPI and trade numbers is discussed in each respective section. In the case of payrolls, consistent application of seasonal adjustments would have shown a jobs gain of 44,000, still not a contraction.
Emphasis mine.
Consistent application of seasonal adjustments, hmmmmmmm, that sounds like someone is manipulating the data doesn't it?
As suggested in the July 14th Alert "June CPI Hanky-Panky", more than poor quality reporting may be at work in current government statistical releases. Occasionally, government calculations of seasonal factors can suffer an across-the-board major distortion that will throw off monthly results for a month or two. Such appeared to be the case in April. This remains a possibility for June, given the number of series where seasonal-factor distortions helped skew reported results. Seasonal-factor rigging, however, also has been used historically as a tool for near-term political manipulation of data.
Again emphasis mine.
Given flagging presidential ratings, a more likely explanation than accidentally-warped seasonals for too-good data is that the Bush administration has moved into a direct, monthly political-manipulation mode, much as was seen during the better part of the Clinton administration. Another month of reporting should answer that possibility beyond a reasonable doubt.
Some background on Presidential manipulation of the data is warranted.
For example, during the Clinton administration, reporting of payroll growth had been targeted at three million new jobs per year, 250,000 jobs per month. For an extended period of time, the reported monthly changes -- usually after allowing for monthly revisions -- came out at exactly 250,000 per month, or exactly 500,000 for two months. What the obvious process indicated was that the seasonally adjusted numbers were being set, and the supposedly raw, unadjusted data were backed into. The practice continued until we called them on it. At least they stopped using exact increments of 250,000. The issue received minor press. Some other creative data tactics used in the 1990s have been described in the background articles available on the home page, but the statistical hanky panky was not limited to the Clinton administration.
When the first President Bush was facing his difficult reelection bid, a senior Commerce Department official met with the head of major corporation and asked that computer sales be overstated in reporting to the Bureau of Economic Analysis. The request was honored and reported GDP growth was spiked artificially. That story came from a participant in the original meeting and has been confirmed separately by an official at the BEA. As an aside, the involved corporation has been a major sponsor of the Conference Board.
Not only does a non-recession outlook help the current Bush administration, in theory, it also is a big plus to Wall Street, where financial pressures and resulting potential conflicts of interest often impact the way the economic environment is presented to the public. That brings us to the bizarre timing of the Conference Board in revamping the index of leading economic indicators (LEI) last month.
The conference board revamped its LEI last month, how strange that this would be done in July!
Once published by the government, the LEI was widely followed as a signal of pending economic activity. When the government moved to privatize the series in 1995 -- during the Clinton administration -- the Conference Board won the bidding.
Economic series occasionally get redefined. When the effort is to improve accuracy, that is one matter. Overhauling a series when it is signaling imminent recession and changing the signal to one of healthy economic growth is another. By redefining its series, the Conference Board accomplished the latter. Such is reasonably consistent with the government's approach to changes in statistical reporting methodologies. The final product usually ends up enhancing reported economic growth or reducing the rate of reported inflation.
The traditional recession signal from the LEI was generated by three consecutive month-to-month contractions. While such a signal sometimes proved to be false, usually the economy at least slowed. Prior to the current overhaul, the LEI had contracted for the last three months, through May, and had fallen in nine of the last 12 months. As of the June restatement, the LEI now has risen in each of the last three months.
The Conference Board's primary mission is to "help businesses strengthen their performance." Apparently, reporting bad economic news has not always been part of that mission. The late Al Sindlinger often told the story of his early days, as an outsider under contract, in setting up and running the Conference Board's Consumer Confidence Index. It seems that whenever he reported a gain in confidence, he would hear the news shortly thereafter on the radio. Whenever he reported a decline in confidence, he always would get a call asking him to recheck his numbers, with the clear message that a positive result was desired.
So I guess your getting the picture that the Government is just plain lying to you about all of this!
But lets look specifically at the payroll data for July!
The July gain was despite a negative bias factor that subtracted 76,000 jobs from the underlying unadjusted data (80,000 jobs were subtracted last July). In August, the fudge factor swings back to the plus side, where it added 123,000 jobs to last August's payroll reporting.
Once again, patterns of year-to-year growth in the payroll series have been highly irregular and highlight unconscionable games being played with the month-to-month seasonal factors. On an unadjusted basis (which is the statistically better way to view the annual changes), annual growth was 1.66% in June, 1.63% in July. Seasonally adjusted, the slightly declining unadjusted annual growth swings to an increase, with June annual growth of 1.59% surging to July's 1.69%. Properly adjusted for seasonal factors, the annual growth rates should be within a point or two, each month, to the second decimal point. Reported on an adjusted basis that is consistent with the unadjusted annual growth, July payrolls would have risen by 44,000, some 163,000 jobs less than the official estimate.
Lets dig into that for clarification.
2004 to 2005 June to June and July to July not seasonally adjusted payroll reporting shows the following,
June
- June 2004, 132,537,000
- June 2005, 134,732,000
Difference in unadjusted data for June is 2,195,000, (2,195,000 / 132,537,000) = 1.66% growth.
Seasonally adjusted data,
- June 2004, 131,479,000
- June 2005, 133,579,000
Difference in adjusted data is 2,100,000 (2,100,000 / 131,479,000) = 1.59% growth.
July
- July 2004, 131,384,000
- July 2005, 133,531,000
Difference in unadjusted data for July is 2,147,000, (2,147,000 / 131,384,000) = 1.63% growth.
Seasonally adjusted data,
- July 2004, 131,562,000
- July 2005, 133,786,000
Difference in adjusted data is 2,224,000 (2,224,000 / 131,562,000) = 1.69% growth.
So in June the data appears to be off by .07% and in July by .06%, because as John stated Properly adjusted for seasonal factors, the annual growth rates should be within a point or two, each month, to the second decimal point
What is the affect?
Applying the 1.66% percent to June's data would show 2,182,551 for June, and applying 1.63% to July's data would show 2,144,460 for July, thus adding 2,182,551 to the June 2004 number would show 133,661,000 and adding 2,144,460 to July's 2004 number would show 133,706,000. Thus (133,706,000 (July's 2005 #) - 133,661,000 (June's 2005 number) leaves us with 45k (rounded) additional jobs added in July.
Additionally, we have to remember that the bias factors are very problematic. In 2004 a full switch was made to a model method of determining the monthly bias adjustments. In the previous method an additional 145k was added to the survey data to account to business births. However there is a realization that this is problematic because in downturns adding in 145K jobs when you really should be subtracting them leads to significant differences in the reality of the situation. Thus the BLS decided to implement the birth/death model instead, which makes some wild assumptions as well as adding in a number from the model, however as the BLS itself states this is problematic in and of itself because there is no way for modeled data to capture a downturn. This is because the data period used to build the model is based on historical data which could be a period of growth, this applying a bias factor to the model from a period of growth when in reality you are experiencing a downturn leads to significant errors. I'm not trying to disparage the use of bias factors, clearly there needs to be some adjustment, but the method the BLS introduced in 2004 (which was really a phased in period that completed in 2004) is very problematic and does nothing other than introduce unneeded volatility into the series.
There are limits to manipulation, however, ranging from what even the most gullible analysts cannot swallow, to reporting of corporate profits and tax receipts. Also, standard economic series, either from the private sector, such as the Conference Board's help-wanted advertising index, or less followed series, such as real average weekly earnings, often remain free of undue outside influence. Therein lie the potential data shocks for the markets.
As an example, the unchanged June CPI pushed reporting to the brink of rousing the gullible. Accordingly, anything but a hefty July CPI gain is going to have serious credibility problems.
A number of inflation measures, such as those seen in the purchasing managers survey, are suffering shock on a year-to-year basis against last year's extremely strong inflation indicators. Those extraordinary comparisons will pass after August. Distortions from extremes of weather and unusual seasonal factors tend to reverse in a matter of a month or two. As an example, look for a fair tumble in August industrial production, as will be reported in mid-September.
Fortunately, businesses employ people that are bound to what they tell management about economic conditions, and as much as the government tries to manipulate the data, there are people that seek to provide businesses with realistic forecasts of business conditions and realize that relying on government data is plain idiocy.
Despite recent gains in key series, most economic data already have started to soften, and the trend will accelerate sharply, with monthly contractions beginning for payroll employment and industrial production in the near future. Political manipulation, if used, can keep the payroll data afloat for a while longer. Significant deterioration also will be seen in federal tax receipts (a widening budget deficit) and corporate profits. This outlook is predicated on economic activity that already has taken place and does not consider any risks from exogenous factors such as renewed terrorist activity in the United States.
I guess we'll just get to see how far really the governemt can go in manipulating the data eh?
The roots of the current difficulties are structural in nature. A consumer starved of income growth and overburdened with debt cannot sustain the real (inflation-adjusted) growth in consumption needed to keep GDP growth in positive territory. The income weakness is a direct result of the loss of a significant manufacturing base to offshore locations and the ensuing explosive, perpetual growth of the U.S. trade deficit.
Well, now there is some data that we all know to be true, pretty basic.
Exacerbating economic and financial woes will be unusually high inflation during this contraction. Inflation, fueled by high oil prices and weakness in the U.S. dollar, will not be brought under control simply by weakness in economic demand. Instead, persistently high prices only will serve to intensify the 2005-2007 recession, making it unusually long and protracted. Ongoing inflation woes and dollar problems will maintain upside pressure on long-term interest rates, inhibiting the traditional flattening of the yield curve expected with a recession.
An inflationary recession remains a nightmare for the financial markets. Particularly hard hit will be the U.S. dollar, with downside implications for both equity and bond prices.
So things are not nearly as rosy as the Bush admin would like you to believe. Personally, I think we all know that politicians of any creed will manipulate data to make it look like they are doing a good job. The issue is that over the years the manipulation of the data becomes bad for the people and the economy because businesses will make bad decisions based on totally incorrect data and whitewashing the data so much prevents us from working on the real problems we have as a society.
We all know from looking at the reality of the economic conditions that there are serious problems that need to be addressed and if we continue to avoid them then things will turn very grim for us. The reality is things are already grim for us, we have significant disparities in income and wealth in this country that is completely masked by those within the median attempting to make up the difference by borrowing the money from the people who actually have it.
Without a growing income base relative to inflation for those people to accommodate the debt repayments there will be a contraction in consumption equal to the debt load at some time and this reaction will not be very much fun at all.
And I'll not even attempt to go into the already fantastic diaries Bonddad and Sterling have on the serious economic issues that they have been keeping us abreast of for sometime now.
But I will say this, economic conditions are very much different from the conditions of the financial markets although they are certainly related. In a real world the financial markets will adjust themselves to reflect economic conditions in most circumstances. In today's pundit driven political manipulation of the data the hope seems to be that we can adjust economic conditions to financial markets perception of them.
Of course that is the black and white world of the conservative mind, unfortunately as well all know too well it bears little resemblance to reality on the ground as we know it.
Thanks for reading through this incredibly long diary, and I encourage all of you to consider subscribing to John Williams' work as it is incredibly important that we reward folks like him for all the hard work they do for us real people!
Subscribe to Shadow Government Statistics
Please note I am not affiliated with Gillespie Research in any form or fashion aside from being a subscriber.