ONS defends its construction statistics

Claims that official statistics have misrepresented the performance of the construction industry are unfounded, the Office for National Statistics said today.

There was disbelief in many corners of building and construction when ONS figures for the first quarter of 2011 showed a decline in output at a time when industry surveys showed that it was rising. The most recent estimate from ONS, at constant prices and seasonally adjusted, is that construction output fell  by 3.4 per cent between Q4 2010 and Q1 2011. That had an impact on the disappointing GDP figures for the first quarter.

This contrasted with evidence from the market survey by the Royal Institute of Chartered Surveyors, which suggested rising output. Its 2011 Q1 survey showed that respondents reporting a rise in output exceeded those reporting a fall. The actual figures were 24 per cent reporting a rise, 18 per cent a fall, which the RICS records as the balance, ie +6. In both Q4 2010 and Q1 2011,  58 per cent of respondents reported that business was flat. Two other business surveys also suggested improved business.

The suggestion was made that companies who participate in the ONS survey were responding differently as a result of changes introduced in January 2010. Before that, according to this explanation, the industry was asked to provide the value of new orders won, while the new form asked for the volume of output. Wrong, said Graham Sharp of the ONS today. “We have always used output and not new orders.”

So if that’s not the explanation, what is? The ONS analysis suggests the discrepancy arose because the surveys were not measuring the same thing. The ONS asks a very wide sample of the industry to fill in its questionnaires, and has moved from quarterly to monthly outputs. It surveys 8,000 businesses a month – representing 37 per cent of all construction employment and over 50 per cent of construction turnover.

Some businesses - those with more than 100 employees, or fewer than 100 employees but a turnover of more than £60 million - are surveyed every month. The rest of the industry is sampled – more than 3,300 of the smallest businesses with less than five employees, and around 3,600 of those with between 5 and 99 employees.

Something would have to be seriously wrong for a survey this size to get the wrong answer. But analysis by company size, carried out to try to discover why the discrepancies arose, turned up the fact that the biggest companies did less badly than the smaller ones between Q4 2010 and Q1 2011.

The smallest firms fell by 9.7 per cent, but the largest saw only a 4.3 per cent fall.  And when the ONS examined data from over 3,000 businesses that had returned data in all six months of the two quarters, it found that the largest businesses were reporting increased business – 52 per cent said they were doing more business in Q1 2011, against 48 per cent saying they were doing less. That gives a positive balance of 4 per cent, quite close to the RICS figures.

But the ONS further found that while the big companies felt output was increasing, when measured by value it was not. Improving sentiment was not matched by improving value.

This may help to square the circle between the ONS and the RICS. The RICS survey is much smaller – 166 respondents, though as they are chartered surveyors they represent a wider constituency – and, it is surmised, is weighted more heavily towards the bigger companies. The RICS survey gave a positive balance of +6, the ONS figures +4 if only the bigger firms are counted – no great difference.  

So is the ONS in the clear? The industry may take some persuading, but Mr Sharp said today: “We still feel confident that the survey we have is the best we could possibly have as an input into GDP.” He added that there was no evidence that the changes introduced in 2010 had affected outputs.