By Tony Collins
Thanks to reports by the National Audit Office, the questioning of HMRC civil servants by the Public Accounts Committee, answers to FOI requests, and job adverts for senior HMRC posts, it’s possible to gain a rare insight into some of the sensitive commercial matters that are usually hidden when the end of a huge IT contract draws closer.
Partly because of the footnotes, the latest National Audit Office memorandum on Aspire (June 2016) has insights that make it one of the most incisive reports it has produced on the department’s IT in more than 30 years.
Aspire is the government’s biggest IT-related contract. Inland Revenue, as it was then, signed a 10-year outsourcing deal with HP (then EDS) in 1994, and transferred about 2,000 civil servants to the company. The deal was expected to cost £2bn over 10 years.
After Customs and Excise, with its Fujitsu VME-based IT estate, was merged with Inland Revenue’s in 2005, the cost of the total outsourcing deal with HP rose to about £3bn.
In 2004 most of the IT staff and HMRC’s assets transferred to Capgemini under a contract known as Aspire – Acquiring Strategic Partners for Inland Revenue. Aspire’s main subcontractors were Accenture and Fujitsu.
In subsequent years the cost of the 10-year Aspire contract shot up from about £3bn to about £8bn, yielding combined profits to Capgemini and Fujitsu of £1.2bn – more than double the £500m originally modelled. The profit margin was 15.8% compared to 12.3% originally modelled.
The National Audit Office said in a report on Aspire in 2014 that HMRC had not handled costs well. The NAO now estimates the cost of the extended (13-year) Aspire contract from 2004 to 2017 to be about £10bn.
Between April 2006 and March 2014, Aspire accounted for about 84% of HMRC’s total spending on technology.
Servers that typically cost £30,000 a year to run under Aspire – and there are about 4,000 servers at HMRC today – cost between £6,000 when run internally or as low as £4,000 a year in the commodity market.
How could the Aspire spend continue – and without a modernisation of the IT estate?
A good service
HMRC has been generally pleased with the quality of service from Aspire’s suppliers. Major systems have run with reducing amounts of downtime, and Capgemini has helped to build many new systems.
Where things have gone wrong, HMRC appears to have been as much to blame as the suppliers, partly because development work was hit routinely by a plethora of changes to the agreed specifications.
Arguably the two biggest problems with Aspire have been cost and lack of control. In the 10 years between 2004 and 2014 HMRC paid an average of £813m a year to Aspire’s suppliers. And it paid above market rates, according to the National Audit Office.
By the time the Cabinet Office’s Efficiency and Reform Group announced in 2014 that it was seeking to outlaw “bloated and wasteful” contracts, especially ones over £100m, HMRC had already taken steps to end Aspire.
It decided to break up its IT systems into chunks it could manage, control and, to some extent, commoditise.
HMRC’s senior managers expected an end to Aspire by 2017. But unexpected events at the Department for Work and Pensions put paid to HMRC’s plan …
Eight lessons from Aspire
1. Your IT may not be transformed by outsourcing. That may be the intention at the outset. But it didn’t happen when Somerset County Council outsourced IT to IBM in 2007 and it hasn’t happened in the 12 years of the Aspire contract.
“The Aspire contract has provided stable but expensive IT systems. The contract has contributed to HMRC’s technology becoming out of date,” said the National Audit Office in its June 2016 memorandum.
And Mark Dearnley, HMRC’s Chief Digital Information Officer and main board member, told the Public Accounts Committee last week,
“Some of the technology we use is definitely past its best-before date.”
2. You won’t realise how little you understand your outsourced IT until you look at ending a long-term deal.
Confidently and openly answering a series of trenchant questions from MP Richard Bacon at last week’s Public Accounts Committee hearing, Dearnley said,
“It’s inevitable in any large black box outsourcing deal that there are details when you get right into it that you don’t know what’s going on. So yes, that’s what we’re learning.”
3. Suppliers may seem almost philanthropic in the run-up to a large outsourcing deal because they accept losses in the early part of a contract and make up for them in later years.
“What we are finding is that it [the break-up of Aspire] is forcing us to have much cleaner commercial conversations, not getting into some of the traditional arrangements.
” If I go away from Aspire and talk about the typical outsourcing industry of the last ten years most contracts lost money in their first few years for the supplier, and the supplier relied on making money in the later years of the contract.
“What that tended to mean was that as time moved on and you wanted to change the contract the supplier was not particularly incented to want to change it because they wanted to make their money at the end.
“What we’re focusing on is making sure the deals are clean, simple, really easy to understand, and don’t mortgage the future and that we can change as the environment evolves and the world changes.”
4. If you want deeper-than-expected costs in the later years of the contract, expect suppliers to make up the money in contract extensions.
Aspire was due originally to end in 2004. Then it went to 2017 after suppliers negotiated a three-year extension in 2007. Now completion of the exit is not planned until 2020, though some services have already been insourced and more will be over the next four years.
The National Audit Office’s June 2016 memorandum reveals how the contract extension from 2017 to 2020 came about.
HMRC had a non-binding agreement with Capgemini to exit from all Aspire services by June 2017. But HMRC had little choice but to soften this approach when Capgemini’s negotiating position was unexpectedly strengthened by IT deals being struck by other departments, particularly the Department for Work and Pensions.
Cabinet Office “red lines” said that government would not extend existing contracts without a compelling case. But the DWP found that instead of being able to exit a large hosting contract with HP in February 2015 it would have to consider a variation to the contract to enable a controlled disaggregation of services from February 2015 to February 2018.
When the DWP announced it was planning to extend its IT contract with its prime supplier HP Enterprise, HMRC was already in the process of agreeing with Capgemini the contract changes necessary to formalise their agreement to exit the Aspire deal in 2017.
“Capgemini considered that this extension, combined with other public bodies planning to extend their IT contracts, meant that the government had changed its position on extensions…
“Capgemini therefore pushed for contract extensions for some Aspire services as a condition of agreeing to other services being transferred to HMRC before the end of the Aspire contract,” said the NAO’s June 2016 memo.
5. It’s naïve to expect a large IT contract to transfer risks to the supplier (s).
At last week’s Public Accounts Committee hearing, Richard Bacon wanted to know if HMRC was taking on more risk by replacing the Aspire contract with a mixture of insourced IT and smaller commoditised contracts of no more than three years. Asked by Bacon whether HMRC is taking on more risk Dearnley replied,
“Yes and no – the risk was always ours. We had some of it backed of it backed off in contract. You can debate just how valuable contract backing off is relative to £500bn (the annual amount of tax collected). We will never back all of that off. We are much closer and much more on top of the service, the delivery, the projects and the ownership (in the gradual replacement of Aspire).”
6. Few organisations seeking to end monolithic outsourcing deals will have the transition overseen by someone as clear-sighted as Mark Dearnley.
His plain speaking appeared to impress even the chairman of the Public Accounts Committee Meg Hillier who asked him at the end of last week’s hearing,
“And what are your plans? One of the problems we often see in this Committee is people in very senior positions such as yours moving on very quickly. You have had a stellar career in the private sector…
“We hope that those negotiations move apace, because I suspect – and it is perhaps unfair to ask Mr Dearnley to comment – that to lose someone senior at this point would not be good news, given the challenges outlined in the [NAO] Report,” asked Hillier.
Dearnley then gave a slightly embarrassed look to Jon Thomson, HMRC’s chief executive and first permanent secretary. Dearnley replied,
“Jon and I are looking at each other because you are right. Technically my contract finishes at the end of September because I was here for three years. As Jon has just arrived, it is a conversation we have just begun.”
“I would hope that you are going to have that conversation.”
Richard Bacon added,
“Get your skates on, Mr Thompson; we want to keep him.”
“We all share the same aspiration. We are in negotiations.”
7. Be prepared to set aside millions of pounds – in addition to the normal costs of the outsourcing – on exiting.
HMRC is setting aside a gigantic sum – £700m. Around a quarter of this, said the National Audit Office, is accounted for by optimism bias. The estimates also include costs that HMRC will only incur if certain risks materialise.
In particular, HMRC has allowed around £100m for the costs of transferring data from servers currently managed by Aspire suppliers to providers that will make use of cloud computing technology. This cost will only be incurred if a second HMRC programme – which focuses on how HMRC exploits cloud technology – is unsuccessful.
Other costs of the so-called Columbus programme to replace Aspire include the cost of buying back assets, plus staff, consultancy and legal costs.
8. Projected savings from quitting a large contract could dwarf the exit costs.
HMRC has estimated the possible minimum and possible maximum savings from replacing Aspire. Even the minimum estimated savings would more than justify the organisational time involved and the challenge of building up new corporate cultures and skills in-house while keeping new and existing services running smoothly.
By replacing Aspire and improving the way IT services are organised and delivered, HMRC expects to save – each year – about £200m net, after taking into account the possible exit costs of £700m.
The National Audit Office said most of the savings are calculated on the basis of removing supplier profit margins and overheads on services being brought in-house, and reducing margins on other services from contract changes.
Even if the savings don’t materialise as expected and costs equal savings the benefits of exiting are clear. The alternative is allowing costs to continue to soar while you allow the future of your IT to be determined by what your major suppliers can or will do within reasonable cost limits.
HMRC is leading the way for other government departments, councils, the police and other public bodies.
Dearnley’s approach of breaking IT into smaller manageable chunks that can be managed, controlled, optimised and to some extent commoditised is impressive. On the cloud alone he is setting up an internal team of 50.
In the past, IT empires were built and retained by senior officials arguing that their systems were unique – too bespoke and complex to be broken up and treated as a commodity to be put into the cloud.
Dearnley’s evidence to the Public Accounts Committee exposes pompous justifications for the status quo as Sir Humphrey-speak.
Both Richard Feynman and Einstein said something to the effect that the more you understand a subject, the simpler you can explain it.
What Dearnley doesn’t yet understand about the HMRC systems that are still run by Capgemini he will doubtless find out about – provided his contract is renewed before September this year.
No doubt HMRC will continue to have its Parliamentary and other critics who will say that the risks of breaking up HMRC’s proven IT systems are a step too far. But the risks to the public purse of keeping the IT largely as it is are, arguably, much greater.
The Department for Work and Pensions has proved that it’s possible to innovate with the so-called digital solution for Universal Credit, without risking payments to vulnerable people.
If the agile approach to Universal Credit fails, existing benefit systems will continue, or a much more expensive waterfall development by the DWP’s major suppliers will probably be used instead.
It is possible to innovate cheaply without endangering existing tax collection and benefit systems.
Imagine the billions that could be saved if every central government department had a Dearnley on the board. HMRC has had decades of largely negative National Audit Office reports on its IT. Is that about to change?
This morning (22 June 2016) on LinkedIn, management troubleshooter and board adviser Colin Beveridge wrote,
“Good analysis of Aspire and outsourcing challenges. I have seen too many business cases in my career, be they a case for outsourcing, provider transition or insourcing.
“The common factor in all the proposals has been the absence of strategy end of life costs. In other words, the eventual transition costs that will be incurred when the sourcing strategy itself goes end of life. Such costs are never reflected in the original business case, even though their inevitability will have an important impact on the overall integrity of the sourcing strategy business case.
“My rule of thumb is to look for the end of strategy provision in the business case, prior to transition approval. If there is no provision for the eventual sourcing strategy change, then expect to pay dearly in the end.”