Interpreting the construction output figures has become a herculean task for analysts of late.

It’s tough to interpret any construction data at the best of times. Construction work is lumpy, heterogeneous, blurred in definition and geographically dispersed.

But making sense of the Office for National Statistics year-end construction output figures has left me stumped.

It’s no secret that the output series has proved a headache for the ONS. The series has been stripped of its National Statistics status and the interim review of the UK statistics by Professor Sir Charles Bean (pdf) released late last year wagged a stern finger at the construction output data.

Intense work is underway to improve the series. But issues remain, not least the effect of a spate of quite significant revisions and how some of the revisions might be incorporated into the back data. We are left wondering how close to its current levels will the time series be when everything has settled.

There are three particular issues, I think worth mentioning, that are hard to factor in. Firstly there has been a problem with the way price changes are factored in to provide a “like-for-like” measure of activity. This has led to uncertainty over whether adjusted data properly reflects the volume.

Secondly, there is a rather odd problem with the seasonal adjustment process that has created a large gap between the seasonal adjusted (table 2a) volume data and the non-seasonally adjusted (table 2b). This leaves doubts over where the problems lie and which table better reflects the actual path of construction.

The third problem is what I regard as a structural break (see previous blog for detail). This was caused, as I understand it, by the inclusion of a new large company into the sample. The effect was, from March 2015 onward, to add about £2 billion or more (my estimate) annually to output. This has had a profound impact particularly on the infrastructure series.

Trying to read the underlying trends through these quirks in the data makes for much less confident interpretation.

Sod it, let’s have a go anyway.

So when we say that construction output increased 3.4% in 2015, as the data on the page suggests, there is no way we can be confident. In fact it seems that this may well be a long way from what actually happened.

Taking out the effects of adding in a new company would leave output looking pretty flat across the year, with new work rising and repair and maintenance falling.

But other data from other sources suggest that activity has been rising. And there are plenty of folk in construction or prepared to talk about construction that are rather bullish. There are many who think construction is on a roll. You’ll, also, hear lots of people completely pooh-poohing the output data.

Still, we can probably be pretty confident that new work is performing better than repair and maintenance, because that tends to happen when the industry perks up from recession. That’s good news. It should also mean the industry is increasing its productivity.

If we are looking for other official data that might give us a better clue of construction activity we might turn to the numbers employed. Well, both the jobs measure and employment measure put the number employed in the third quarter of 2015 at about the same as a year earlier – about 2.2 million. No great growth in employment and the average hours worked seems to be about the same as a year ago.

So, if our gut feel is right and there is more new work in the mix, this most likely means there has been a bit of a rise in productivity. This in turn suggests, on the basis of stable employment, that there has been a rise in activity within construction over the year. It doesn’t however suggest a huge rise. (I am here missing out huge complications in how productivity might be determined, I know, but let’s keep it simple.)

Anyway, the more you look at the figures from different angles, the more sense you get of a flattening out of the recovery with perhaps a bit of a dip at the end to the year. Nothing to get freaked out about, but likewise nothing to make you feel the industry is on that much of a roll.

Put it this way, if I were at the helm of a Government that proclaimed itself to be “the builders”, I’d be keeping a close eye on this data. It is very hard to interpret and it may be telling us things that we are missing.

Two things are certain in my mind. One: construction needs a period of healing after a ravaging recession. Two: if the future the construction sector is in doubt, then this risks damaging future improvements to UK productivity.

I’d put construction on the watch list, George. I certainly wouldn’t take it for granted.