Tales of the unexpected

Tales of the unexpected seem to be written daily across the world at the moment. Many are spun with grandiose allusions to David and Goliath – Leicester licking the Premiership giants, Farage flaying the EU elites and now Trump’s triumph over the American establishment. Ignoring that spin, they have wider and more important lessons for us all.

Two in particular come to mind. One: life is not as predictable as many would have you believe. Two: the immediate outcome of an unexpected event is often less spectacular than imagined as they occur. There are many more.

If you want to understand what’s really going on in construction, it’s well worth bearing this in mind, because we view the world through distorted lenses ground and polished by cognitive bias that readily lead us to wrong conclusions.

Construction, as an industry, is extremely complex and ‘facts’ are hard to trust at the best of times. We very easily read into the data what we’d like to believe or try to fit the data to match a popular narrative.

With this in mind, let’s take a look at some of the latest construction survey data, before peeking at the most recent batch of forecasts.

One of the most obvious recent example of cognitive bias skewing the ‘facts’ can be seen in the Markit/CIPS UK Construction PMI – an indicator commonly used to measure actual construction activity, among other aspects of change within the sector.

Chart 1 shows its take on the impact of the EU referendum. Its measure of ‘activity’ drops sharply after the vote was counted and bounces back.

Let’s test this against a sensible judgment of reality. It seems rather farfetched that construction activity actually dropped rapidly in the month after the referendum as a result of a decision made to leave the EU at some unspecified point in the future. Even on the smallest projects, workload on site is planned and set up well in advance. It seems extremely unlikely that a contractor would slow down work on an active site for a few weeks after June 24th and then speed it up again when things didn’t seem as bad as originally thought.

That, however, appears to be what the data were suggesting. It would seem these figures are measuring changes in sentiment rather than changes in bricks laid or concrete poured. Interestingly, the Bank of England agents’ score for construction on the same chart shows a similar general downward path without the shock effect resulting from talk of Brexit. 

And we see evidence in the quarterly RICS construction survey (Chart 2) that sentiment was probably what caused the recorded slowdown in construction activity posted by the Markit survey. The work-done measure dips a bit, but the workload expectation dives a lot and rebounds. This suggests work continued pretty much as expected, but there was a sharp downward blip in sentiment as a result of the EU referendum.

The lesson here is to be careful how we read the data. There is a big difference between a fall in activity and a fall in sentiment. There’s nothing solid in the data that points to any impact on actual activity resulting from the EU referendum. There may be problems working their way through the pipeline, but that isn’t what these headline indicators are purportedly measuring.

All three of the surveys suggest activity is rising, albeit at a markedly slower rate than two years ago. This, however, appears to contradict the official Office for National Statistics construction output data. It suggests that activity has fallen over the second and third quarters of 2016, as we see in Chart 3.

All four indicators suggest that about two years ago growth slowed sharply. The difference is that the Markit/CIPS, Bank of England and RICS surveys suggest the industry continued to grow, though slowly, over the period. The ONS data suggest workload in construction has been pretty much flat across the period and falling of late, technically into recession.

There are plenty of explanations that might account for the difference. The Bank of England, Markit and RICS surveys may be subject to survivor bias. They may be too heavily weighted towards new work (black line), which has grown, and take less account of repair and maintenance (blue line) which has declined according to the ONS. Perhaps they are not discounting enough for inflation.

Then again, problems may lie in the output data, such as overestimating inflation and so reducing the ‘volume’ measure too much. They could be missing construction activity that had been measured before. For instance, the fall in repair and maintenance work looks rather suspicious, particularly the heavy falls in public sector housing repair and maintenance. It could be real, but then again it could be down to housing associations taking construction work in house and that work not being picked up by ONS.

If the repair and maintenance sector is shifting into fairly deep recession, this will in time be revealed in the labour statistics. Repair and maintenance is far more labour intensive than new build, so a loss of output here will disproportionately reduce labour demand. The danger is that it reduces the pressure to train new workers.

This amalgam of conflicting data is what faces those seeking to divine construction’s prospects. It can appear more of an art than a science. It is certainly very easy to form the patterns we want to see in our minds by taking selected messages from the mix.

The slew of unexpected events of late has made picking the likely facts from the possible fictions that much tougher. Inflationary pressures caused by the falling pound will cause sporadic and unpredictable effects, which will not necessarily be accounted for with accuracy by those surveying construction activity. This will send potentially false signals into the construction market.

We do, however, have forecasters within the industry who try to assess the vast array of data produced and come to some judgements on what that might mean for construction activity. The latest are now out from the Construction Products Association (CPA), Experian and Hewes & Associates.

They are summarised for all work and new work in Charts 4 and 5. As always there is a difference in terms of how they assess the future. Hewes tends to take account of more of the possible risks, so the forecast tends to be lower.

The Hewes forecast points to a deeper recession across three years with a fall in workloads as deep as 4.3% next year. The CPA is more optimistic, seeing no recession but a slowing to less than 1% growth next year and basically little or no growth in the following two. Experian sees a fairly shallow recession, dropping 1% next year, 0.3% in 2017 and a mild recovery by 2018.

Despite the CPA’s slightly more optimistic forecast, it sees private housing and commercial building dipping into recession, both sectors that provide much work for architects.

What is interesting is that both Experian and the CPA see new work activity performing slightly better than activity overall, while Hewes sees the reverse. The declines suggested by Hewes in new work are deep.

But the uncertainties surrounding these forecasts are great and the Trump factor may well have added a few more uncertainties.

Bringing it all back home, the RIBA workload expectation survey adds to the mix of data (Chart 6). Again, it shows a broad decline in the rate of growth in workload and it shows more recently the jitters in workload expectations caused by the referendum vote. It appears to share the view that in general workloads are currently growing slowly.

But, tellingly, it shows that the dial measuring the likelihood of employing more people is pointed very much at ‘wait and see’.

That is what we all need to do before the mists lift and we can see the future more clearly. In the meantime, with so many unexpected twists and turns in the road ahead, it is worth putting the fog lights on and driving with caution.