National Statistical

The Office for National Statistics is increasingly looking to harness alternative data sources from both Government and the private sector, to improve the quality of statistics and reduce the burden for firms needing to fill out questionnaires. Here John Allcoat explains how we regained national statistics status for our statistics by using a wider set of private sector data sources to help us improve our estimates of the construction industry.

While making up only a relatively small part of GDP, the fortunes of the construction industry are regarded as something of a bellwether for the economy as a whole. But with a multitude of relatively small business and sub-contractors undertaking much of the work, it’s a challenging sector to measure.

Back in 2009 the ONS took control of construction statistics from the then Department for Business, Innovation and Skills. We began a long period of development to address shortcomings in the statistics. These improvements including introducing a monthly, real terms measure of growth in the sector, alongside the introduction of substantially improved Construction Output Price Indices which meant specific price indices were developed for each sub-sector in the industry. This method is considered among the best in Europe.  Additional improvements were made to enhance the methodology to address previous concerns over upward bias in the early survey returns, along with the incorporation of Value Added Tax turnover data to help improve the coverage of small and medium sized businesses.

The other important area of development was within our construction new orders figures. They show the future levels of demand and as such are used by analysts as a forward-looking indicator to help predict future activity in the sector. These data were previously collected through a survey. However, construction firms often struggled to provide the necessary level of detail to complete the questions correctly. It proved especially difficult for them accurately to report what regions their construction work was taking place in, as this information was often not easily accessible.

To improve the quality of the data, we decided that a better way to obtain information on new orders would be to identify through a competitive tendering process a partner in the construction industry which could both provide us with detailed data while also helping us understand it.

Since 2013 we have used data supplied to us by construction market experts Barbour ABI, which has detailed information on new orders in the construction industry, enabling us to address some of the previous challenges we faced. Previously the survey was sent to around 9,000 companies annually, whereas the new data source uses local authority planning applications on a daily basis to gather this information.

Each quarter we receive the new orders data, and after adjusting for inflation and for the normal season variation in the volume of new orders, we include these estimates in our regular Construction Output in Great Britain bulletin. It’s not just about crunching the numbers, however, it’s important understand what’s driving the changes.

For example, the data published on 8 December 2017 showed a huge increase in the value of new orders compared with the previous quarter. So, what was going on? The answer is this case was actually quite simple, but only when you know the sector inside out:  the period in question saw the awarding of the first high-value contracts for High Speed 2, which is almost wholly responsible for the unprecedented scale of the quarterly increase.

Using data from the private sector and other government departments will play an increasingly important role in the development of official statistics as we move away from traditional surveys and make more use of alternative data sources.

We are always keen to hear from our users who would like to work with us to make better use of our data or to discuss alternative data sources – if you are interested please get in touch.

John Allcoat – Head of Construction Statistics