Is £40m write-off on a big software project normal?

By Tony Collins

On BBC R4’s “Week in Westminster” on Saturday morning (14/12/13)  guest presenter Isabel Hardman of The Spectator spoke to Conservative MP Richard Bacon and me about big government projects that go wrong.

Hardman mentioned that Bacon has co-written a book  on government failures Conundrum: Why every government gets things wrong and what we can do about it.

Referring to write-offs so far of about £40m on Universal Credit, Hardman asked me whether it was normal for such a write-off on a big project.

I said it wasn’t. The work and pensions secretary Iain Duncan Smith has said it was. When questioned by MPs of the work and pensions committee on 9 December, IDS implied that it was not unusual to write-off a third on large software-based projects. He suggested that research by Forrester supported this view.

Software coding for Universal Credit has cost about £120m so far (excluding hardware, infrastructure, consultancy or other IT-related costs). So IDS suggested that a write off of £40m was only about a third of the software coding costs.

But I haven’t seen any evidence that suggests write-offs of a third of the software costs on a big project are typical.   

I replied to Hardman that although there has been much trial and error on Universal Credit IT, £40m is a lot to write off.

[Trial and error included an attempt, from 2011 onwards, to adopt an agile approach but the National Audit Office said the DWP “experienced problems incorporating the agile approach into existing contracts, governance and assurance structures”. The NAO added that the Cabinet Office “did not consider that the Department (DWP) had at any point prior to the reset [Feb-May 2013]  appropriately adopted an agile approach to managing the Universal Credit programme”. The DWP has now introduced what it calls Agile 2.0, a hybrid approach incorporating elements of  agile with waterfall, though agile purists say it is impossible to combine the two.]

I told Hardman that the write-offs were largely because the DWP was unclear at the outset what the software was supposed to do.”With big IT projects it’s a bit like designing a bridge and you know where one side begins but you’re not sure where the other side ends. They have been learning as they go along and that’s probably why there have been large write-offs,” I said.

Hardman asked Richard Bacon whether it was normal to set out on these big projects without knowing where the bridge was going. Bacon agreed, citing the NPfIT which had led to large write-offs on failed work for England-wide electronic patient records. He said it was not at all abnormal for ministers to set off on big projects without knowing where they were going.  

The good news?

I told Hardman that IDS was at least well informed. He now has the NAO scrutinising the project as well as his own external consultants and the independently-minded Howard Shiplee as head of the project.

But I didn’t think UC would be complete until 2020 at the earliest given that the last big computerisation of benefit systems, Operational Strategy, took about 10 years to complete. Hardman said: “That would be a humiliation for IDS surely?”

I replied that IDS may not even be in politics in 2017. I also said that UC will probably not bring the financial benefits predicted, to judge from the last big computerisation of benefits.  But UC has wide support. Perhaps, I said, it has to work … eventually. 

BBC R4 Week in Westminster – 14/12/13