17 December 2012
This month’s Office for National Statistics construction output figures have provided a conundrum for commentators. Are they good or are they bad?
Chris Giles, the Financial Times Economics Editor tweeted that they were the “first semi-decent construction figures for some time”. Meanwhile BBC business tweeted “Construction output falls again”.
Those trying to take a view from these interpretations might be a bit confused. I suspect there might also be a little muttered frustration about lies, damn lies and statistics.
As to whether the figures are good or bad, I suspect that all depends on how you look at them. That’s not a cop out by the way.
Personally I’d suggest a different question we might ask. Do these figures really enlighten us about the direction of the construction industry?
Well not really, at least not on their own.
Certainly a rise in October on a non-seasonally adjusted basis is better than a fall. And a rise of 8.3% is not to be sneezed at. The figures did show that.
But if we look at the amount of construction work done (as measured by the ONS) over the 12 months to October there was about £500 million less than in the 12 months to September. So on that measure the industry is still shrinking fast, although not as fast as it was.
That’s the great thing about statistics. You can pick your measures and suggest very different things.
That said, one thing to bear in mind when picking your choice of measure for analysing construction data is that they are exceptionally volatile. It’s wise, if you can, to pick a measure that dampens some of the noise.
I’d be very wary about drawing too much from one month’s figures and I suspect that would be the advice you’d get from most seasoned watchers of construction data. There’s too much randomness and noise and the published figures are prone to fairly big revisions as new survey data arrives.
But also, for me and I suspect for others who keep a close eye on construction data, it’s hard to view the output figures in isolation of other data on construction. They are not, after all, created out of thin air, they relate to a real world which can be measured and is measured in various ways.
So, for instance, the new work element of these figures ultimately relates to the amount of work contractors win. The relationship may be fairly loose and there may be issues with the measuring of new orders, but there is a link.
Therefore it’s worth being at least conscious of the new orders data when considering the output data.
A few days ago the latest new orders figures were published. I concluded that if we pinned all our trust on those figures the prospects for new work in construction, as opposed to repair and maintenance, looked pretty hideous. There was little or no sign of growth in the sectors we pretty much rely on for growth, new private housing and commercial.
This was inevitably an invitation to the data gods to chuck out an oddity. And so it was. It was new work that enjoyed the biggest upward bounce in October, particularly the commercial sector and private housing, along with infrastructure.
The fact that there’s a pretty high level of infrastructure work going on is no secret. What’s of greater interest to me is how well the private housing and commercial sectors are doing.
Well on pretty much any measure you care to pick other than October’s output compared with September’s output on a monthly non-seasonally adjusted basis, the sectors are down.
On the grounds that you’d be insane to bet your house on one month’s construction data, I am drawn to conclude that the latest figures are interesting, but they do little to shift the reality that the overall picture is pretty ugly.