Juiced Data

We have been watching, with no small degree of skepticism, a stream of improving Macro-economic data. Color us unconvinced. Many of the key releases have been fraught with misleading headlines obscuring much weaker data beneath, and last month was no different. From Inflation to Federal Deficit to Unemployment Rates to Industrial Output to recent GDP (and its revisions), nearly every data point comes with an asterisk.

When we look back at this period of economic home runs, we will call it the season of steroids. Like Major Leaguers, the Data is on the Juice.

Take the Leading Economic Indicators (and revisions) from the Conference Board. The changes to the LEI now register a flattening yield curve as a positive for future economic activity. Only in the alternative universe where the Conference Board lives is this considered a positive. The CB now requires the yield curve to actually invert before it bodes negatively for future economic growth.

The Board was apparently not pleased that 8 of the 10 past LEIs were negative. Hey, if you don’t like what the indicators are suggesting, than why not just change the model? And that’s exactly happened. Taking a page from the BLS handbook (Birth Death adjustment, anyone?), the Conference Board reduced the utility of LEIs for investors. Their work now falls into the category of economic cheerleading.

Kindly return your PomPoms to the gymnasium at the end of the semester.

Don’t care much for that private group? Then consider what BLS BEA hath wrought. Their GDP revisions for 2005 Q1 border on the absurd. In order to crank GDP from its disappointing initial reading of 3.1% to the more vigorous final 3.8%, the BLS BEA had to make some sketchy adjustments. Primary amongst their changes was (I am not making this up!) an actual decrease in Home Prices for Q1. Thus, by somehow emphasizing unit sales (as opposed to price appreciation), courtesy of the Price Deflator, GDP became higher in the final read.

Torture the data long enough, and it will confess to whatever you want it to.

Despite this gamesmanship – or perhaps because of it – much of the investing public knows only half the story when they read the economic headlines. The challenge to us is to not only attempt to discern reality, but to anticipate when the great masses do so also. Its what has caused us to title research in the past with phrases such as “Fundamentals Stink: Buy Stocks.”

When the charade finally ends – probably after the last of the Bears capitulates – the finale will be ugly.

>

UPDATE:  August 2, 2005 9:43pm

A few quick points: When GDP is calculated, one of the components is Structures (Residential). That’s the line where the new home construction supposedly dropped in price. The data comes from Census (part of BEA). Here is the relevant line from the Technical Note, under “Sources of Revisions”:

.  Investment in residential structures was revised up, mainly on the basis of a downward revision to the Census price index for single-family houses. (Emphasis mine).

If prices went down, unit production went up. So not only do we have more output (units built), but since prices theoretically declined, the price deflator does its thing. Voila! GDP revised from 3.1% to 3.8%!  (Hey kids, its magic)

Note also that part of GDP measures new — but not existing — home sales. The transaction of shifting title from one party to another isn’t considered production (nothing gets made), while building a new residential structure is.

Lastly, a Brain Freeze™ caused me to type BLS instead of BEA. Those responsible for this error have been sacked.


>

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. Ed commented on Aug 2

    Thinking about the textbook theory of why financial fraud occurs gives us an easy rationale for this statistical monkeying:

    “the purpose of financial reporting is to obtain cheap capital … If [this can be best achieved with] statements that measure financial condition inaccurately, [the logic of shareholder value] obliges management to publish that sort, rather than the type held up as a model in accounting textbooks.”

    Ie, if the choice is between taking a hit for telling the truth, and getting a reward for making it up, the latter is the path to go. Obviously, this approach heavily discounts future value, and fails the naieve “truth wins in the long term” test, because the kind of companies for whom fraud is a good option are choosing between bankruptcy now and bankruptcy later, when fraud is discovered.

    In the government’s case, it is to prove to the world that the US has a strong economy, so they will continue to lend to us at low interest rates.

    To this, I say:

    “The government has a setup that Bernie Ebbers will probably spend the rest of his life dreaming enviously about from prison: I get to report the numbers I think are meaningful, and I get to determine how the numbers I report are calculated.”

    http://ddo.typepad.com/ddo/2005/07/why_worldcom_an.html

  2. BondFund1 commented on Aug 2

    Barry, keep up the good work deciphering this stuff. We need you now more than ever…very few impartial pundits willing to tell it like it is.

  3. Jordan commented on Aug 3

    YES!

    I am so sick and tired of these financial professionals who wont or cant read into these grossly manipulated statistics. This piece proves you are way above the everyday wall street man.

    Im afraid its to late though. Real interest rates will continue to be negative as inflation is running at 5% even despite the cheap foreign made goods we import. To me, it looks like we are on the early road to hyperinflation.

  4. Marshall’s Ghost commented on Aug 3

    Barry,

    The GDP revision does not include a decrease in Home Prices for Q1. Rather the GDP revision is a downward revision in the rate of growth of housing prices. A slower rate of growth does not imply a negative rate of growth. You can easily verify this by looking at the GDP data:

    http://www.bea.doc.gov/bea/dn/home/gdp.htm

    Click on tables for new release. You can verify that the price index for Investment in residential structures did in fact increase by looking at table 4. You can also verify by comparing the growth of Nominal Investment in residential structures (Table 3A(2)) to the growth of Real Investment in residential structures (Table 3B(2)).

    I notice that you make the mistake of confusing growth rates and levels quite often. Whether comparing GDP deflators, the effect of housing on the CPI, or household liabilities you constantly compare levels and growth rates as if they are the same thing. They are not, and when you treat them as if they are you end up with nonsensical results which you then attribute to a conspiracy at the BLS or BEA. Do you really think the BLS and BEA economists are engaged in a criminal conspiracy to defraud the public? I know a number of BLS and BEA economists. They don’t seem like the criminal type to me.

    One last point. Do you mean to imply that there should be no adjustment to establishment payroll due to the birth and death of firms or do you just not like the current methodology? What would you change with respect to the current adjustments for the birth and death of firms?

  5. Ed commented on Aug 3

    You may have incorrectly typed “BLS”, but the BLS does this too.

    The same thing happened with the 1990s “productivity miracle”. Here’s how:

    1) investment in computers was added to labor output as a measure of productivity

    2) computers “output” was calculated as dollars spent on computers divided by price of the good.

    3) the price of the good was adjusted to reflect improved quality, even though the price didn’t change (ie, Dell computers have cost the same for about 5 years, but PCs get more powerful evey 18 months)

    4) hence, the “output” calculation has a rising numerator, and a shrinking denominator. A potent combo!

    The result is “record productivity, unlike everything we’ve ever seen” and employees can surf ESPN at high speeds…

  6. Barry commented on Aug 3

    Marshall’s Ghost

    Here’s the BEA quote: “The structures revision was due to downward revision to the one-family house price index which resulted in more real sales and less inflation.”

    This was first brought to my attention by MNSI:

    ANALYSIS: US Q1 GDP Rev +3.8%: Prices Dn, Expts & Res Inv Up>
    Wednesday, June 29, 2005 8:31:18 AM (GMT-04:00)
    Provided by: Market News

    –Commerce Dept Lowers Prices for 1-Fam Houses, Financial Services

    WASHINGTON (MktNews) – The U.S. Q1 real GDP revision was to +3.8% in yet another upward move that left the growth rate the same as the +3.8% of Q4:2004.

    Market News International estimates that about 2/3 of the revision from the prior +3.5% real growth rate was due to a downward shift in prices and only about 1/3 due to data that found additional growth. . . The structures revision was due to downward revision to the one-family house price index which resulted in more real sales and less inflation — an almost unbelievable result given that the housing market is on fire with home prices posting double-digit appreciation on the year in some areas.”

    Note that this is not a slowing in the rate of growth — but an ACTUAL DECREASE IN PRICES. Census reports the raw data to BEA, who then massage it.

    Draw your own conclusions — you know what mine are: Garbage in, garbage out.

  7. Marshall’s ghost commented on Aug 3

    Barry,

    Inflation is a rate of change, not a level. When MNSI says “The structures revision was due to downward revision to the one-family house price index which resulted in more real sales and less inflation” they mean what they say – less inflation not negative inflation. Same with the BEA quote, revision resulted in less inflation not negative inflation.

    The initial estimate of GDP was based on positive price increases in housing. The revision also resulted in a positive price increases in housing –not as large as the initial estimate but nonetheless still positive.

    Perhaps a pathological example would help. Suppose last quarter’s price index equals 100. Furthermore suppose the initial estimate of this quarter’s price index is 110 (10% inflation). If a revision in the initial estimate results in a revised price index of 105 we can say that prices were revised downward resulting in less inflation. Nonetheless, 105 is greater then 100 so even though the revision resulted in less inflation the inflation rate is still positive – just not as positive as initially estimated.

    Its all right there in the GDP data.

  8. TimW commented on Aug 3

    I highly doubt the gamesmanship that you allude to in this post. I work on an economic indicator… I can speak confidently about my data and say that what we put out is accurate. None of us here.. from the top administrators (who aren’t political appointees!!) to the lowest analysts have “pom poms” or are even remotely close to being cheerleaders. I can’t even imagine how someone would pull off such a stunt!

    I, like Marshall’s Ghost, would like to know how you would account differently for the birth and death of firms in our surveys. Also, MG is right about the home price measure. It was a downward revision from the initial report, but the estimate was still positive.

    Posts like this just fuel the fire of conspiracy theorists like Ed and Jordan here. There is no conspiracy brewing in the halls of BEA and Census. (as I roll on the floor laughing)

  9. Barry Ritholtz commented on Aug 3

    I don’t think you need to wear a tin hat to buy my argument that the headline is often rosier than the underlying data. Unemployment at 5%, for example, sounds great — until you look beneath at the labor participation rate. Industrial output’s upside surprise makes the economy sound robust — until you notice that the strong number was mostly due to utilities responding to a heat wave (more AC).

    As to the housing prices — whether you want to say they were up, down or sideways in that quarter — they were well KNOWN when the preliminary GDP report was issued.

    A few months later, these numbers were adjusted downwards. Thus, the only way they could get the res. inv. number to meaningfully impact GDP higher in the final is by altering the mix, i.e., by emphasizing more unit sales as opposed to price appreciation.

    One needn’t fear black helicopters to note that the data beneath often belies the headlines — and thats my key takeaway.

    Again, I dont need to convince anyome of this. If you think the data supports a robust and ongoing economy, buy some indices and put them away for a year or two . . . I vote with my (and my clients) dollars, and I am watching carefully for signs that this unwinds.

  10. Marshall’s Ghost commented on Aug 3

    Barry,

    It isn’t your claim that the headline is often rosier than the underlying data that I find offensive. Instead its your implication that the data was manipulated in a fraudulent manner and anyone who thinks otherwise is either too stupid to know better or in on the conspiracy. The fact that you base your accusations on your mistaken read of a BEA data makes you character assassinations all the more offensive. Do you really believe there is a conspiracy at the BLS or BEA?

    You often follow this pattern. The government releases data claiming 2+2 = 4. You respond that this is hopelessly optimistic and only a fool or a co-conspirator would believe such a lie. Then someone who is familiar with the data explains that 2+2 does indeed equal 4 to which you always respond “If you believe that buy stocks” and then change the subject to some new conspiracy. I see this pattern with your posts and comments on LFP, household dept, housing and the CPI and now GDP. Again I ask, do you think the government is intentionally defrauding the public?

  11. E Pluribus Unum commented on Aug 3

    Leading economic indicators: Is the data on steroids?

    From Cunning Realist:Barry Ritholtz is a financial professional who has a reputation as a no-BS straight shooter, which is rare on Wall Street these days. He has a good record of calling major market turns, and his writing is accessible…

  12. E Pluribus Unum commented on Aug 3

    Leading economic indicators: Is the data on steroids?

    From Cunning Realist:Barry Ritholtz is a financial professional who has a reputation as a no-BS straight shooter, which is rare on Wall Street these days. He has a good record of calling major market turns, and his writing is accessible…

  13. E Pluribus Unum commented on Aug 3

    Leading economic indicators: Is the data on steroids?

    From Cunning Realist:Barry Ritholtz is a financial professional who has a reputation as a no-BS straight shooter, which is rare on Wall Street these days. He has a good record of calling major market turns, and his writing is accessible…

  14. Ed commented on Aug 3

    I don’t think this at all resembles “2+2=4”.

    I spent a while reading through the BLS calculation of productivity – it spans almost 10 pages, and that isn’t all the detail.

    This makes some sense – our economy is a massive, nebulous entity, and measuring it requires a ton of assumptions.

    Since no one can know with certainty exactly what is going on, those responsible are required, to some extent, to hypothesize what they think is happening, and then adjust assumptions such that the results match this perception.

    Do I think they are maliciously fraudulent? No.

    I think it reasonable to believe that the people creating statistics are subject to the same interpretative biases about the world they live in that any American is.

    These biases can include: a) the late 1990s saw stupendous growth in the American economy. b) America is and continues to be the strongest economy in the world with low inflation and steady growth.

    Do I think we are a weak country? No way. But I do think that fooling ourselves with statistics allows us to perpetuate a lot of self-detrimental economic activity that will make us weaker.

    But don’t take my word for it. I get these ideas from much smarter and credible people than myself:

    Link to Bill Gross

  15. Econbrowser commented on Aug 4

    When should we worry about the yield curve?

    The slope of the yield curve is likely to become an increasingly bearish indicator as this year progresses, and recent changes in the calculation of the index of leading economic indicators should not be interpreted as in any way denying that fact.

  16. E Pluribus Unum commented on Aug 5

    U.S. payrolls recorded strong gains in July

    Finally! Hiring data were better than economists’ forecasts:Employers expanded their payrolls by 207,000 in July, the most in five months, while unemployment rate held steady at 5 percent, the government reported Friday… Analysts had predicted a gain…

  17. E Pluribus Unum commented on Aug 5

    U.S. payrolls recorded strong gains in July

    Finally! Hiring data were better than economists’ forecasts:Employers expanded their payrolls by 207,000 in July, the most in five months, while unemployment rate held steady at 5 percent, the government reported Friday… Analysts had predicted a gain…

  18. New Economist commented on Sep 8

    (Mis) leading indicators?

    This is a belated post, inspired by a 17 August blog by Mark Thoma on The Use of Leading Economic Indicators in Economic Forecasting. Mark cites a paper by Stock and Watson and a Bloomberg column by Caroline Baum, which asked why economic forecasters …

Posted Under