Challenges in using data across Government inquiry

Leadership of initiatives to improve data is fragmented and unclear. It is not clear who is responsible for planning and driving the changes needed to improve government’s use of data. DCMS is responsible for data policy and GDS for data standards. The Data Advisory Board is the senior oversight board across government. However, at July 2019, only 2 of 18 people attending the most recent meeting were permanent secretaries, despite these being the core members of the board. In 2017, government committed to appointing a Chief Data Officer by 2020 but has not done so yet. The Cabinet Office said this is because it wanted ‘conditions for success’ in place first. In our view government should appoint a Chief Data Officer before it develops the data strategy so he or she can lead on the strategic direction of data. Recommendation: As a matter of urgency Cabinet Office and DCMS should appoint a Chief Data Officer for government, to act as a single point of accountability for government’s use of data.

DCMS and Cabinet Office have made little progress in developing the data strategy since they announced it over a year ago. In June 2018, the Secretary of State for DCMS announced that the UK government would develop a national data strategy to ‘unlock the power of data across government and the wider economy’. DCMS issued a preliminary ‘open call for evidence’ on 10 June 2019—a year after it announced the national data strategy. It plans a public consultation for autumn

Rather than mandating an approach to good data use and sharing, DCMS and Cabinet Office are relying on winning the ‘hearts and minds’ of other departments. This is an approach that has failed in the past. We have seen many times that working together across departments is difficult, largely due to departmental funding, accountability and a culture of working within departments’ boundaries. Central departments such as the Cabinet Office have an important role identifying and seeing through changes that are beneficial for government as a whole. Similar central government initiatives such as Shared Service Centres and Verify have failed in the past due to inadequate buy-in from departments. DCMS has not thought through how to get departments on board to fulfil the data strategy. It will need continued and sustained pressure from Cabinet Office and DCMS if government is to use data more efficiently and effectively. If this is left to individual departments to do voluntarily, it is likely to be squeezed out by other priorities. Recommendation: The Cabinet Office and DCMS should check progress against their plans to improve government’s use of data and review the merits of mandating 6 Challenges in using data across government a consistent approach. They should make sure that the strategy builds in ways to monitor compliance and the Chief Data Officer should hold departments to account.

The Cabinet Office has not developed a useful set of standards to support effective use of data across government. It is difficult and expensive to bring data together without consistent ways of recording and presenting data. For example, there are over 20 ways of identifying individuals and businesses across government, with departments using many different reference numbers for the same organisation or person. The Cabinet Office sees data standards as the way forward but is behind in setting them and has not agreed a plan for them across government. GDS has approved only 16 data standards thus far. The Cabinet Office told us that these cover ‘obscure’ subjects, and they do not, for example, include fundamentals such as consistent ways of recording individuals’ names. Like so many cross-government initiatives, there is no mandate across government to comply with data standards. Recommendation: The Cabinet Office and the Government Digital Service should identify and prioritise the top 10 data standards that would benefit government. It should specify their use in new systems from 1 April 2020 and monitor implementation by departments.

Ageing IT systems across government make it difficult for it to use data effectively and efficiently. Ageing IT systems make extracting data expensive and sometimes impossible; whether for use within a department or for sharing with others. We have seen numerous projects where the importance of good quality data for monitoring progress of the programmes and policies they systems were intended to support was an afterthought. For example, in our report on Universal Credit (October 2018) we found that the Department for Work and pensions could not assess how it treated vulnerable claimants as its systems did not allow this. Replacing or modifying government’s systems will take time and be costly and needs careful planning. We were surprised that the government does not have a list of its ageing IT systems on which to base its strategy. Government needs to get its own house in order—to understand its ageing IT systems and the problems with them—before it can hope to take advantage of the opportunities from using and sharing data better. Recommendation: The Cabinet Office and DCMS should identify the main ageing IT systems that, if fixed, would allow government to use data better. They should ensure that whenever departments replace or modify these systems this is done with full consideration of how the systems will support better use of data in government.

Government officials’ concerns about protecting data can stand in the way of using it to coordinate services. Preserving public trust is a precondition for government to use data more effectively and it is right that officials are concerned about data protection. Departmental cultures can also discourage data sharing as civil servants are not used to working across departmental boundaries. DCMS’s work includes helping departments to use data ethically while also using data in new ways. Its Centre for Data Ethics and Innovation is part of this. The Digital Economy Act 2017 (the Act) has made gaining parliamentary approval to share data between departments or with external bodies easier but it is still quite new. DCMS’s Challenges in using data across government 7 efforts to increase awareness of the Act have had limited impact so far. There were 36 agreements under the new Act by July 2019, mostly aimed at reducing fraud and error and fuel poverty with slower progress in other areas. Front-line staff are afraid of making mistakes and do not always have the operational guidance they need to give them confidence and knowledge of when and how to share data appropriately. Recommendation: DCMS should review departments’ data-sharing guidance and standard operating processes; and report how well they support the wider use of data alongside data protection by 30 June 2020. 8 Challenges in using data across government 1 Government’s strategy and leadership on data

The new strategy is the 12th in a long line of strategies and reports over the last 20 years that were supposed to improve how government uses data. We asked DCMS whether it had a list of milestones for implementing the strategy that we could use to track its progress. It told us that it did not yet have an implementation plan as it had not yet written the strategy.13 DCMS and Cabinet Office’s approach to implementing the strategy across government

Lack of cost and performance data is a long-standing problem in government. In our 2016 inquiry on Accountability to Parliament for taxpayers’ money, the Cabinet Office and the Treasury recognised that government too often lacks good cost and performance data. We found there was a lack of urgency in dealing with this persistent and pervasive issue.14 8 Q 102, C&AG’s Report para 2.6 9 Qq 17–19, 25 10 Qq 18, 20, 22–23, 25, 33 11 Q 33 12 Qq 44, 77, 81 13 Qq 76–77, C&AG’s Report, para 1.9, Figure 6 14 HC Committee of Public Accounts, Accountability to Parliament for taxpayers’ money, 39th Report of Session 2015–16, HC 732, 4 May 2016 10 Challenges in using data across government

In our previous inquiries into cross-government projects we have found that working together across departments can be challenging. Unless there is clear governance and shared budgets for cross-departmental data projects, they are likely to be squeezed out by departments’ own priorities.15 We heard that departments can struggle to fund strategic data projects where the benefits accrue to departments other than those making the investment or to government more widely.16 The Cabinet Office told us that a cross- departmental funding bid for data projects was under-development. However, its approval depended on the next comprehensive spending review, which may cause further delay.17

In our 2019 report on Driving value in public spending, we found compelling evidence that departments work in separate silos, are not incentivised to plan for the longer-term, and are repeatedly over-optimistic. We recommended that the Treasury and the Cabinet Office should incentivise joint departmental plans and delivery; and should require government departments to show how their plans and funding bids for the next Spending Review delivered long-term, sustainable value for money. We also recommended that, in advance of the next Spending Review, the Treasury and the Cabinet Office should work with government departments to better challenge the realism of departments’ plans.18 Given the significant challenges that government faces, we were concerned that it was unlikely to be successful in improving how it uses data unless it can make sure that departments act together to implement the strategy. As we found in our examinations of the Rural Payments Agency and the GDS, without a good system in place progress can depend on individual personalities which can be a problem.19

DCMS told us that getting departments to put the strategy into practice was ‘not all about hard leverage … it is about persuasion, and hearts and minds’.20 DCMS told us that it was aware that departments can have a large number of significant priorities, and that its approach was to keep going until change started to happen.21 We have found in our previous inquiries that cross-government initiatives can fail if the lead departments have not managed to get buy-in from other departments. For example in our inquiry into Shared Services, Cabinet Office told us that government accountability structures made it inherently difficult to undertake cross-government programmes. Having decided not to make it compulsory for departments to join the Shared Service Centres programme, Cabinet Office failed to create an environment of collaboration among departments and then did not intervene in a timely and effective manner when things began to go wrong.22 In our report on Verify we found that the Cabinet Office recognised that departments did not have the right incentives to adopt Verify and we concluded that GDS’s inability to get buy-in from department’s ultimately led to Verify’s failure.23 15 Q 105 16 Qq 32, 92, 103; C&AG’s report, para 2.15 17 Q 26–28 18 HC Committee of Public Accounts, Improving government planning and spending, 78th Report of Session 2017–19, HC 1596, 8 February 2019 19 Q 24, 32; HC Committee of Public Accounts, The Common Agricultural Policy Delivery Programme, 26th Report of Session 2015–16, HC 642, 2 March 2016 20 Q 100 21 Q 48 22 HC Committee of Public Accounts, Shared service centres, 20th report of Session 2016–17, HC 297, 19 October 2016 23 HC Committee of Public Accounts, Accessing public services through the Government’s Verify digital system, 95th Report of Session 2017–19, HC 1748, 8 May 2019 Challenges in using data across government 11 2 Infrastructure and conditions needed for improved use of data across government Data standards

As government’s use of technology has increased, departments have each developed their own systems, often to deliver a specific policy. As a result, departments hold inconsistent data in incompatible systems both within departments and across government. For example, there are more than 20 different ‘identifiers’ (for example National Insurance numbers and tax references) used to identify people and organisations across just 10 departments.24

The Cabinet Office accepted that the government needed standardised ways of recording data so it can be shared between systems easily and people do not have to develop manual workarounds to match data.25 Without data standards, it is difficult and expensive to bring data together. The cost of this extra work is unknown but it is likely to be considerable.26 The Cabinet Office told us that it had made some “early steps” in agreeing and implementing data standards for new data, but it accepted that further work was needed. So far, GDS has published 16 data standards. These include standardised ways of recording country names, advertising government job vacancies and publishing information about government grants and contracts. The Cabinet Office accepted that some of the critical standards do not exist yet. For example, there is not yet a standard for how to record people’s names.27

Cabinet Office told us there was no part of government that could mandate compliance with data standards and that departments did not always use the standards that do exist. It told us “there is not a mandate, but there is an invocation to meet the standards”. Cabinet Office acknowledged that it needs to accelerate promotion of data standards once they have been approved. We asked the Cabinet Office whether, in the absence of the strategy, it had a plan for developing data standards. Cabinet Office told us that a plan was being worked on but had not yet been considered by the Data Advisory Board.28 Ageing IT systems

The NAO found that most of government’s data resides in its legacy systems, some of which can be inflexible and lead to poorly organised, out of date or inaccessible data that cannot be extracted or shared easily and is costly to combine with data from different systems.29 We asked DCMS how it planned to tackle government’s ageing IT systems which limit how government can use data. DCMS and Cabinet Office told us that this would take time as it was not realistic to replace all of government’s ageing IT in one go.30 24 C&AG’s Report, paras 3.4, 3.12–3.14, Figure 10 25 Qq 46, 49 26 Qq 84–86; C&AG’s Report, para 2.14 27 Qq 49–50 28 Qq 49–50, 53–54, 57 29 C&AG’s Report, paras 11, 3.8 30 Qq 38, 111, 117 12 Challenges in using data across government DCMS told us that when departments develop new systems, it is important that they consider from the very beginning how to design them so they do not create problems for the future.31 For example, avoiding systems that do not have easy access to comprehensive performance data; or systems that hold data in ways that make it difficult to reuse it easily within and between departments. We have seen numerous projects where the importance of good quality data for monitoring progress of the programmes and policies they systems were intended to support was an afterthought.32 For example, in our report on Universal Credit in 2018 we found that the Department of Work and Pensions could not monitor its treatment of vulnerable claimants, such as those with mental health problems, as it did not collect data on these groups within its systems.33

We asked about the new data systems government was developing to prepare for the UK’s exit from the European Union. We had already heard that some critical data standards were not in place yet, but Cabinet Office assured us that the new systems were being built to other GDS standards, namely its digital service standards and its technology code of practice.34 We pressed DCMS on how it would tackle the challenges of replacing government’s ageing IT and whether it had a list of government’s systems and the problems with them on which to base its strategy. Neither Cabinet Office nor DCMS has a government-wide list of the IT systems that government will need to upgrade or replace if it wishes to make real advances in how it uses data. DCMS told us that, as a result of information asset registers, it had a much better understanding of departments’ information assets and the data that they held, but it could not say whether this resulted in a comprehensive list of legacy systems. It accepted that, without this, it was not possible to identify opportunities to link together data systems to improve services.35 Data sharing and data protection

The Digital Economy Act 2017 (the Act) brought in an easier process for public bodies to gain legal authority to share data.36 The Cabinet Office told us that around 36 new data sharing agreements had been approved so far using the new process. These agreements were particularly aimed at reducing fraud and error and fuel poverty with slower progress in other areas. The Cabinet Office recognised that cross-departmental data-sharing presents government with a huge opportunity to improve, and that this depended on a cultural change across government as well as having the right technology. DCMS recognised the importance of data protection, concerns about safeguarding data and not wanting to take risks with people’s information. DCMS told us that it was working to ensure that government’s culture and overall approach recognised the benefits of innovation and how the risks attached to data sharing can be managed.37

We asked how the Department could challenge the fear among front-line staff about what data they can and cannot share. Cabinet Office told us that the more it was able to demonstrate the success, safety and security of data-sharing, the more comfortable people 31 Qq 37–38 32 HC Committee of Public Accounts, Third Annual Report of the Chair of the Committee of Public Accounts, Second Special Report of Session 2017–19, HC 1399, 5 July 2018, p.18 33 HC Committee of Public Accounts, Universal Credit, 64th report of Session 2017–19, HC 1183, 26 October 2018 34 Qq 49–50, 119–120 35 Qq 37–42 36 Q 45; C&AG’s Report, para 13 37 Qq 45, 62 Challenges in using data across government 13 will be using them in the future.38 We asked whether officials could pick up a phone to get advice about their specific situation or concerns about data-sharing. DCMS told us that there were a lot of sources of information and advice about how to use the Digital Economy Act. It also told us “it’s not just about telling, it’s about selling”.39 It told us that it had spent a substantial amount of time at roadshows explaining to staff how they can apply the Digital Economy Act with confidence.40

We pressed DCMS and the Cabinet Office about how they planned to encourage a data-sharing culture alongside compliance with the General Data Protection Regulation (GDPR). DCMS told us that it had worked with the Information Commissioner’s Office to make sure that the Act is consistent with GDPR but that it will be an ongoing task to communicate this.41 It told us that it had set up the Centre for Data Ethics and Innovation to support staff in government to use data safely and appropriately while also using it more widely and in new ways.42

The Cabinet Office told us that it needed to increase data awareness and capability in government. It told us it had started this through the GDS Academy, which had recently introduced two courses on data. Around 200 people have also been trained on the basics of data as part of a separate initiative known as the ‘data science accelerator’.43 Civil servants are unused to working outside departmental boundaries which is an extra hurdle to setting up and maintaining cross-departmental initiatives, including data-sharing. The NAO found that staff awareness and capability of how to share data properly can be also improved through detailed guidance, such as HMRC’s information disclosure guidance. The HMRC guidance is online so it is available to all staff and provides transparency for the public.44 38 Qq 58–59, 62 39 Qq 60–62, 64 40 Q 58 41 Qq 45, 58 42 Qq 60, 67 43 Q 53 44 C&AG’s Report, para 4.15, Figure 18 14 Challenges in using data across government