Opening up more data on public construction project spend could result in more realistic building budgets

A joint blogpost by the ODI and Spend Network

Spend Network and the Open Data Institute are working together to find potential savings in public sector spend without cutting services. Their analysis is based on open data, and they hope to show how opening up more data could also make it easier to find potential savings. This third in a series of blogs representing an open methodology approach to this research focuses on construction contracts, examining the overall impact of bringing additional spend across departments down to a median level.

null (Crossrail construction, Royal Oak, London. Additional spend can happen, but why do some projects go so far over budget? CC-BY-SA (3.0) Marcus Rowlands

In 2014, it was reported that for 11 recent major public sector building projects spend was, on average, 158% above the contracted value.

Construction projects can go over budget for a number of reasons. Aside from changes to specification once construction has begun or flaws in project planning or delivery, there are valid mitigations for additional spend being incurred. These could include discovery of archaeologically significant sites during tunnel digging, discovery of asbestos during school renovations, or delays due to criminal damage or poor weather conditions. It is impossible to determine how much total additional spend, which may include unplanned spend for the examples above, is ‘overspend’ with no clear mitigation.

While acknowledging that it would be unrealistic to argue for eliminating additional spend entirely, we believe that making use of available open data can reveal the basis for managing spend better – with a goal of bringing overspend in particular down to the median level across government.

In order to carry out this investigation, we are looking up Contract Award Notices for construction tenders on Tenders Electronic Daily (TED). We will then match these to data on spend with winning suppliers from contracting entities in Spend Network’s database.

Our ultimate aim is to show that opening up more data will make it easier to identify where spend is managed best – with a view to replicating this, as much as possible, on other projects in other departments.

How will we do this?

We will review Contract Award Notices to find the original contracted value of construction tenders. Then, we will work out the actual spend on the contract by using spend data from each contracting entity to find out how much was spent with the supplier. The additional spend on each tender is the difference between the original contracted value and the final spend. Next, we will calculate the median additional spend as a percentage of the original contracted value of the tender. Finally, we will figure out the savings that could be made if all spending on construction tenders was brought down to the median additional spend in 80% of cases.


Contract Award Notices are only published for 30% of tenders on TED, so this limits our sample. Spend data do not refer directly to the tender under which a supplier payment is made. We will be treating all spend with the supplier that wins a tender by the contracting entity as covered by the tender. We acknowledge that eliminating additional spend is impossible, which is why we will focus on median additional spend.

We also believe that using median spend as a baseline is applicable to other areas where government spend varies between departments and local authorities, and plan to use this in our final analysis on utilities management.

We like to work openly, so please feel free to let us know your feedback on our methodology. Please send your emails to [email protected]