Official construction data revisions pose yet more problems for GDP measurement

One of the themes of this website has been that Gross Domestic Product numbers are far less reliable than is commonly perceived. Indeed we have discovered that there is quite a litany of flaws in using them as a measure of economic wellbeing. Today I wish to simply concentrate on the issue of how accurate they are and in particular how accurate one of the building blocks of it which is the construction series is. Before I do so I would like to point out that under the new ESA 10 definition it would appear that GDP in Spain has seen a trim this morning.

Three tonnes of cocaine seized in Spain, 12 arrested: police

Has anybody informed the police services that they are the enemies of modern GDP measurement?

US Construction Output

Yesterday saw quite a change in the official US construction data released by the Census Bureau. In its own inimitable style Zerohedge reported it thus.

That all changed today when the US Census released its latest, November, construction spending data, which not only missed expectations of a 0.6% increase, but tumbled -0.4%, the most since June of 2014, while all the recent changes were mysteriously revised lower.

Of course this poses a question for the US Federal Reserve and its plans to continue to increase interest-rates. This was enhanced by John Williams of the San Francisco Fed who told CNBC this “I think something in that three to five rate hike range” can be expected in 2016. By the way John most people call that four…

As ever there is more than one way of looking at data so let us move on from the “end of the world as we know it” style of Zerohedge and look at the numbers themselves. If we do so we see that the construction output series in the year to October has been revised downwards from an annual growth rate of just over 13% to one of just under 11%. Accordingly we see that construction output was still growing strongly but there has been a trim. This continued into the November numbers which posted a decline of 0.4% on a monthly basis but were still 10.5% higher than a year before. So our first lesson is that there has been a slowing of the fast rate of growth but that care is needed with individual monthly numbers as what has just happened shows that they are unreliable.

The numbers can be looked at another way which was not in the Zerohedge article and they are illustrated by this. The old October 2015 output was 1,017,381 million dollars and the new is 1,127,040 million which last time I checked was a larger number! So the new series has in fact raised output by 1.8% compared to the past. If we look back we see that the boost came from December 2013 onwards as before the series was lower going back a couple of years rather than higher.

If we look into what caused this it was the private residential improvement spending series which has been revised back to January 2005 due to a “processing error”. If we return to GDP implications we see that the numbers for more than 10 years ago have just changed and back then they changed by around 1%. The precise series concerned changed by 1.7% back then and by 7% in October of this year. Thus in Taylor Swift terms there has been this.

Oh, oh, trouble, trouble, trouble
Oh, oh, trouble, trouble, trouble

So we see that there is a Good a Bad and an Ugly to this. The good is that the US Census Bureau has discovered its mistake and put it right so strike one for welcome honesty, and this is added too by output being higher than before. The bad is that even the numbers for 2005 have just been declared to be wrong and the whole series up to now has been changed by 1.7% with differing changes on its journey. The ugly message comes as we consider such a change (1.7%) when the GDP number it goes into is analysed in bites of 0.1%. Also as housing is a physical object it should be relatively easy to count as opposed to the inanimate concept of services which poses worries about what might be going on there.

If we move to the overall issue of housing and its impact on the US economy then the National Association of House Builders puts it thus.

Historically, residential investment has averaged roughly 5% of GDP while housing services have averaged between 12% and 13%, for a combined 17% to 18% of GDP.

There is another issue here as if we have trouble counting houses and bricks what about the impact of trying to impute rents? Also for 2015 construction output was of the order of 4% of GDP.

Industrial Output

The series here has seen its own problems as I pointed out on the 22nd of July last year.

Firstly the picture in the United States changed yesterday when the level of industrial production was revised down by 2.5% (2012 and 2013 both saw 1% falls).

I brought more detail to the table when I looked at the problems of first world manufacturing on the 2nd of December.

The downwards revisions exclude some high-tech industries where the numbers are oddly troubled and reduced 2012 by 1.7%, 2013 by 1.6% and 2014 by 0.5%. They also implied the full data set for 2014 was not yet in, begging the question if it might also see a more substantial revision.

We are in the zone of up down flying around here as we note that construction has overall been revised up as industrial production was revised down. Again as these are actual things one would think that they are easier to measure than the service sector which of course is much the largest piece of what we consider to be a modern economy.

UK construction series

This is a much more troubled series than the US version. Back on the 12th of June I pointed out another problem in what has become a long list when first quarter output was revised up from -1.1% to -0.2%.

the incorporation of late data, new seasonal adjustment parameters and the introduction of an interim solution for deflators.

So new seasonal adjustment and a replacement for the deflators which means that something serious is wrong here. Take a look at these numbers from the Construction Products Association also from last June.

In 2014 Q4, on a yearly basis, construction output was revised up to 8.9% from 4.5% in the March release.  Similarly, on a quarterly basis, construction output increased by 0.2%, an upward revision from the March release which reported a 2.2% fall in output.

The problems continued in 2015.

In Q1, according to the April release, total construction output rose 4.4% year-on-year compared to a 0.3% fall reported in the March release.

How can one have any confidence in numbers changing by such amounts? This was the road to where I suggested not entirely in jest that I would count cranes myself. If we switch to official language then there have been 3 signs of a change. There is an official steering group looking into it, the national statistics designation has been suspended and a new person has been appointed to oversee the data. Poor Kate Davies has received something of a rugby hospital pass although of course it is also true that the only way is up.

In October the Brickonomics blog reported more signs of trouble.

They added about £1.5 billion to the turnover in cash terms over the months March to July. That’s about 2.4% more over that period than was thought when the count was made a month earlier.

The “headscratching cause?” well ….

it appears in March there was a reallocation of a major business from the services sector to construction. When I say major I mean a firm turning over significantly more than £1 billion a year, probably more like £3 billion.

This poses more questions than answers as we reshuffle the GDP pack but if you think about it there is a danger that this masks the underlying problem by making the series look better without really changing anything. I also note that the planned analysis of why the official series differs so much from the Purchasing Managers Indices was cancelled on the 11th of September last year for “operational reasons” which have nearly lasted for 4 months now.


Today’s article has added to my series on how we should take GDP numbers we much more than a pinch of salt. If we struggle to measure with any degree of accuracy something that you would think would be easy such as construction and indeed industrial output where does that leave services? I did raise the issue of this with the Professor Sir Charlie Bean review of UK economic statistics and if I get a reply I will let you know. After all it must be four- fifths of our economy now.

Meanwhile if you want to take the blue pill then Bloomberg offer this by pointing out that in an analysis of 142 countries GDP data the data quality index has the US in 1st place and the UK 3rd. I would not want to be in 142nd place would you?! Still over to Morpheus from The Matrix.

Let me tell you why you’re here. You’re here because you know something. What you know you can’t explain, but you feel it. You’ve felt it your entire life, that there’s something wrong with the world. You don’t know what it is, but it’s there, like a splinter in your mind, driving you mad.




15 thoughts on “Official construction data revisions pose yet more problems for GDP measurement

  1. Shaun, I think I have aired my cynicism about official GDP statistics being measured to decimal places often enough not to repeat my comments again! The Bloomberg article is interesting and makes one wonder about the supposed growth in places like Spain. My contact there is very sceptical about the recent growth figures and believes there is significant political interference in the numbers to favour a PP government. I would be interested to know where Spain was on the list!

    By the way, gentlemen who are getting on a bit would always favour the blue pill!

    • Hi Pavlaki

      Steady, steady, have you seen the advert where the man of a certain age drops the blue pill out of the window? Panic then ensues! Perhaps he is a central banker….

      Thank you for asking about Spain as it made me look at the source data from World Economics which is rather different to the Bloomberg article. The US is 2nd and not 1st and the UK is 18th and not 3rd whilst Spain is 26th. So apologies for repeating a Bloomberg table where they have changed the data without saying why and it is a shame that an organisation like it has done this.

      I also note that 142 has become 154 which made me wonder how I had made such an odd mistake until I spotted this.

      “while Myanmar’s statistics are the least trustworthy, coming at 142nd place. ”

      Oh and according to WorldEconomics Switzerland is number one and Haiti is number 154.

  2. Hi Shaun,

    I wouldn’t be holding my breath in expectation of a reply from Prof Bean. If I may shamelessly misquote Immanuel Kant
    Out of the crooked timber of GDP data, no straight thing was ever made.

    So we’re in 3rd place are we? Can I safely assume that 142nd place is occupied by the good citizens of La-La Land?

    Welcome to the Desert of the Real.

    • Hi Jim M

      As I have just replied to Pavlaki the Bloomberg report on misleading GDP data was misleading. Ouch! We are 18th and the US is 2nd and to answer your question it is Haiti occupying 154th place as we wonder where 12 nations were found from?

      Oh and as to Charlie Bean well I have put my views forward both on paper and at the Royal Statistical Society meeting. To be frank he gave the impression of a man who was going through the motions. He did have time to point out he was boosting his pension though…

    • Hi therrawbuzzin

      I fear that you are right and I am sure that Sir Humphrey Appleby said something similar. Also let me repeat my best wishes to Mrs rrawbuzzin and also to you as I learned what it is like to sit by someones hospital beside last year. One of the reliefs was to use the hospital wi fi when the patient is asleep or being treated…

      • Hopefully SWMBO is coming home on Thursday or Friday.
        I weep for what has happened to Manchester Royal Infirmary in the seven years since we last had occasion to use its facilities.
        The skill and ability is breathtakingly brilliant, but the nursing is totally dysfunctional.
        The unfortunate problem with paying nurses what they deserve is this:
        When they were badly underpaid, the only people who became nurses were those with a calling.
        Raise the wages, and people of totally unsuitable temperament and commitment levels apply for the money.
        Patients in my wife’s (renal) ward, are viewed as an irritation at best.
        I say this with a heavy heart.

  3. Hi Shaun

    I think most would agree with you about the inaccuracy of these statistics, as the above comments show.

    These statistics have two broad uses: research and as a basis for economic policy decisions. Laying aside the research aspect their main use it seems to me is to inform policy decisions so if we have wrong information then we have the potential to make wrong decisions. But is this actually true?

    It seems to me, looking at the recent behavior of the Fed, BOE and ECB that the facts (right or wrong) have little to do with policy. The BOE “looking through” the inflation data in 2011 and looking through lower inflation now are two recent examples of this (both resulting in them doing nothing which is what they do best). The Fed saying that it is “data dependent” and then raising IRs when all the numbers are going south, and a strong case could be made for a cut in IRs, is another example. It would seem that policy makers only follow the data when it gives them the answer that they want.

    On the other hand they may be just as sceptical as you about the data you mention and, despite what they say, regard it as little more than background “noise” which is in any case subject to statistical confidence limits.

    However, and adverting to yesterday’s discussion, I think they find it much more difficult to ignore trends in the exchange rate, a number which is not statistically manipulated in any way but which almost certainly presages changes in inflation in the future, however that inflation is measured. In some ways it is a much easier, and arguably, better guide to the future inflation wise than might be provided by an arcane calculation of the “output gap”.

  4. Hi Shaun,

    From my experience in the industry, construction statistics have been volatile for as long as they have been collected. Index changes of several percent, up and down, month on month (or quarter on quarter) are not at all unusual and are nothing new. Some of this is probably down to the nature of the business, some to lack of accuracy in recording figures and some due to genuine trends. In my view, the main risk lies in believing that short term fluctuations in these figures are an accurate indicator of much at all. I believe that construction is about 6% of UK GDP these days, so this view does have implications for the cult of GDP worship which seems to be so popular.

    On the whole, your crane counting method may have merit as a measure of construction output. I can add sightings of quite a few tower cranes around the Midlands to your Battersea sample.

    • Hi RobinB

      Thanks for the perspective that to some extent it was ever thus. On that road the issue is that GDP statistics are under more pressure now after a setback than they were when the message was of the NICE decade. There is a point in that but I think that it is also true that in the modern technological era we should be “getting better all the time” to quote Lennon/McCartney whereas some of the revisions above highlight that we may even be getting worse.

  5. Nice to see I’m not the only one who reads ZH.

    Mish Shedlock has been banging on about the BLS employment stats in the USA for years,and having read his work,and having been an avid reader here for a long time,I can’t quite sqaure the USA at 1 and the UK at 3 with what I’m able to make of things.

    The worrying thing for me-and for the next generation-is that the majority of mainstream economic thinkers are all too happy to float nominal national debts off figures that are at best inaccurate and at worst,downright misrepresentations of the fact.

    As for how to accruately measure things,I think you may be more right than you realise about counting cranes or some other such measure.The last thing genius learns is simplicity.

    • Hi Dutch

      There are problems everywhere with the numbers. There is consumer inflation missing house prices, wages numbers omitting the self-employed and a litany of issues with GDP to name but three. It is on that basis that I am against overly activist policies as they are as likely to be wrong as right!

      As to ZH I dip in and out as they have some good sources and can be entertaining but can also lose the plot. I used (for fun) to imagine it was just one person with a massive supply of coffee and Red Bull……

    • ,…I can’t quite sqaure the USA at 1 and the UK at 3 with what I’m able to make of things.
      What kind of chicanery do you think is the preserve of our politicians?
      It is worth remembering that there are two forms of measurement;
      absolute and incremental.
      The table of the corruption of statistical data is, most certainly, the latter type.

  6. My big problem with GDP isn’t the inaccurate measurement, it’s the implication that “things are getting better”. If I borrow 10 times my salary to buy a house off a boomer+ who bought at 3 times his salary then GDP is up. Is this *better*? No, it’s a victory for the rentier filth.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s