Public Audit Committee

Meeting date: Thursday, January 20, 2022

Official Report 623KB pdf

---

Section 22 Report: “The 2020/21 audit of National Records of Scotland”

The second item on our agenda is an evidence session with representatives of the National Records of Scotland. The session follows a section 22 report that the committee considered at its meeting on 9 December 2021 with the Auditor General for Scotland and a team from Audit Scotland.

I am pleased to say that we are joined by Paul Lowe, who is the registrar general for Scotland and keeper of the records of Scotland. Also from the National Records of Scotland are Linda Sinclair, who is the director of corporate services and accountable officer; Anne Slater, who is the director of operations and customer services; and Peter Whitehouse, who is the director of statistical services.

I mentioned that Linda Sinclair is the accountable officer, but I invite Paul Lowe to give us some opening remarks before the committee asks questions.

Thanks very much, convener, and good morning, committee members. Thank you for the invitation to attend today’s session. I will dispense with introductions, as the convener has covered them already.

Today marks two months to census day, which is on 20 March. I am pleased to confirm that we are in a strong position to deliver Scotland’s 2022 census. In a few weeks’ time, our public awareness campaign will launch and, from 28 February, every household in the country will receive a letter that advises how to participate.

We have only one opportunity to deliver the census and to get it right. The census uniquely asks the same questions of Scotland’s population with reference to the same points in time. Its insights provide rich data at the national and local levels, and provide benefits for many years after the census collection. It informs the allocation of budgets, the delivery of infrastructure, research activities and the provision of services.

For the census to be effective, an extremely high response rate is required. It is vital that we enable everyone to participate in it.

Following the lockdown in March 2020, NRS conducted a comprehensive impact assessment of the risks of the pandemic to the delivery of the census. Through that, we concluded that the only option that secured a high level of response rate and a high quality standard was moving the census to 2022. That recommendation was not made lightly. We considered how to balance the risks of delivering a poor-quality census in 2021 and the long-term impacts that that would have against the impacts of delaying for a year but having increased confidence in achieving a successful outcome.

The census is subject to an extensive programme of external assurance, and that programme points to high confidence in our readiness. Indeed, the most recent Scottish Government technical assurance review provides a delivery confidence assessment of amber/green, which is a significant achievement for a programme of the census’s scale and complexity. It is noteworthy that our United Kingdom counterparts went live at amber status. That reflects the strong position that Scotland’s census is in.

Thank you for your attention. I am happy to take questions from the committee.

Thank you very much, Paul. If you want members of your team to field any of the questions, please let us know who the appropriate person is. If members of your team want to come in, they should type R in the chat function to make sure that I call them.

I will give a bit of context to this morning’s session. A section 22 report by Audit Scotland is quite a serious matter; it happens when alarm bells have been or still are ringing. This morning, we want to probe a little into the journey that you have been on and to consider how things look over the next few weeks because, as you said, the census has still to be delivered.

I will start by reflecting on the second paragraph of the Audit Scotland report, which contains this note from the Auditor General:

“I have prepared this report to draw the Scottish Parliament’s attention to the challenges facing NRS in the delivery of the census programme. This includes the significant impact the decision to delay the census until March 2022 has had on NRS’s costs. There are also ongoing risks to delivery of the programme, including resourcing and financial pressures, which NRS will need to continue to manage so that the census can be delivered successfully in line with the revised budget and timetable.”

Committee members will return to those themes over the next hour.

I want to reflect on the evidence session that we had with the Auditor General on 9 December 2021. You might have seen that one of the issues that were brought to our attention was that an options appraisal report led you, through evidence, I presume, to the decision—or the recommendation to ministers for their decision—to postpone the census for a year. That was a very big decision to make. We asked the Audit Scotland representatives whether they had had sight of the options appraisal report. They might have had sight of it, but they were not in a position to offer it to us. Would it be possible for the committee to see the full options appraisal report?

I am happy to provide that, convener. I will follow that up in writing after this session. Do you want me to say a bit more about the options appraisal at this point?

Yes. I think that it would be useful if you could take us through the broad themes of that and why you arrived at the decision to postpone.

As good practice in any programme, it is important to assess risks and issues in the environment that might threaten delivery. The appraisal is an appropriate example of where that has been done.

At the advent of lockdown in March 2020, we started to assess the issues that the pandemic and lockdown might present to the delivery and success of the census programme. A number of factors drove that assessment. Obviously, the first was that our staff, virtually overnight, went from working in teams in offices to working remotely from home. Our contractors and suppliers were put into the same position, and our health services and others were focused on dealing with a national crisis. Therefore, we started to see impacts on the ability to collaborate with other organisations, which is very important in the census, and on productivity in the delivery of the census, and great uncertainty about the length, duration and significance of Covid-related disruption to the programme.

At that stage, we were less than 12 months from delivery. If we can avoid it, it is not wise to cancel a programme at the 11th hour, because the costs are significant. People had been advised and informed of the date of the census and how to participate. All of that has to be unwound. We need to make decisions on those matters with adequate time, to proceed in a safe and controlled manner.

I mentioned collaboration. When we deliver a census, we have to work closely with a range of partners, and local government and the national health service are very significant partners. As we all remember, at that point, the health and care sector was, understandably, facing unprecedented pressures. We had to ask ourselves whether it was appropriate for us to have detailed engagement with those partners about delivery of the census when they were being abstracted to deal with a significant national crisis.

We took into account a number of other factors. We were also mindful of what conditions would exist in Scotland were the census to be taken in March 2021 and the risks that we could be in lockdown or other equivalent conditions during that time.

A significant element of the census collection, even with the digital delivery that we plan, is a field force—teams of staff who will remind or assist people who have not completed the census at the census date. We were mindful of the public health concerns and the great uncertainty that existed at the time. There was no vaccination and there was limited information about how transmissible the virus was and what was safe to do at the time. The question was whether it would be appropriate—and, indeed, whether it would be within the law—to run a field force operation in March 2021. That was a risk-based assessment.

It transpired that, in March, Scotland was in lockdown. England came out of lockdown in early March, but Scotland was in lockdown throughout March, and it did not until into April move to the requirement to stay local, which was still a significant level of restriction in the country. Our risk assessment for that indicated some of the factors on collaboration and our ability to work with organisations. We have to enumerate hospitals, care homes, prisons, hotels and other places. The question was whether that activity would be appropriate at the peak of the pandemic. Our assessment was that there were lots of risks around that. However, we followed that exercise up.

It is a significant step to consider the possibility that action may need to be taken with the census, so we undertook a comprehensive risk assessment. We considered a number of different options for delivering the census that would allow us to retain the existing March 2021 delivery date: delivering it in the way that we had anticipated that we would; retaining the delivery date, but reducing the scope of delivery so that we would reduce the complexity of, and risk around, the census; and options that would involve delay. We considered those options through different lenses. We looked at them through the lenses of financial impact, risk, feasibility and viability, and then with a number of different criteria. We can expand on that when we send you the risk assessment document.

When we looked at our programme plans, the timescales and the activities that we had to undertake, our conclusion was that the disruption from the pandemic would exhaust our contingency time in the programme—we simply would not have the time to deliver the activities that we had planned. As you will appreciate, although the census is a programme that is run over a number of years, the activity in the year leading up to the census is the most intense. In light of that conclusion, we considered different options. We had conversations with our colleagues in the Office for National Statistics to see whether they could provide mitigations, steps or support. Those conversations were helpful but, for reasons that I would be happy to explain later if that would be helpful, we concluded that that was not viable.

We then examined a range of different options for descoping elements of the census to retain the March 2021 delivery date. Intrinsically, those had to involve the removal of field force elements because that involves a large-scale recruitment of about 3,500 to 4,000 people to work across Scotland. There is a mass recruitment and training exercise and then a mass deployment. It is a significant piece of work and would require significant mitigations to be put in place for Covid. Having recognised that we did not have contingency time left in the plan, the way to get us back on track would be to remove that element. However, I bring us back to quality. The expectation with a census is a high response rate. A low response rate leads to an ineffective census and one that is not of value.

We considered running our digital platform only, running paper-based form approaches only, and running digital and paper forms simultaneously, which is what we did in our rehearsal. Our modelling indicated that, with all of those, we would get a response level much lower than we require for a census.

For clarity, we would expect, and need, a response rate that exceeded 90 per cent. In 2011, we secured a 94 per cent response rate. Our modelling indicated that the different options that I have just set out would provide response rates of between 60 and 80 per cent, which would be too low to meet the quality threshold for the census.

09:15  

It is not just—[Inaudible.]

Mr Lowe has frozen. I do not know whether we are able to—

—predominantly—

Sorry, Mr Lowe—you froze, so we did not get the last 60 seconds of your answer.

Sorry, convener. I was just saying that the high quality threshold is really important. Nationally, we look for a response rate in excess of 90 per cent, but we also have to be mindful of response rates at the local community level, because that local data is part of the census’s unique nature.

The final aspect that we considered was the public appetite to interact with field force staff and participate in the census during a lockdown period. People were obviously very concerned—some were still shielding at that time, and the vaccination programme was very much in its infancy. At the time that we did the assessment, no vaccines for Covid had yet been identified.

I hope that that gives the committee a sense of the range of factors that we took into account.

Yes, that is helpful. Nevertheless, I go back to the fact of the matter, which is that in England, Wales and Northern Ireland, the census went ahead in March 2021 and, as I read it, there was a 97 per cent participation rate. You made the point that Scotland had a different status at various points in the month of March 2021. That is clearly a factor but, notwithstanding that, we have been told that one reason why the ONS and the Northern Ireland Statistics and Research Agency were able to go ahead was that they had access to other sources of administrative data, which I presume would be there for them to rely on to add texture to the returns from census day. Can you explain why NRS did not have access to those same sources of administrative data?

To be clear, with regard to the delivery of the census in other parts of the UK, which is relevant, the ONS is responsible for the delivery of the census in England and Wales, but it also provides most of the capabilities that Northern Ireland uses to deliver the census. It delivers the online systems and various sources of support for Northern Ireland, along with a field force and a public information campaign. Therefore, most of the census delivery across the rest of the UK is administrated by the ONS—it is a massive programme.

There are a number of factors in relation to administrative data. First, I think that a bit of confusion about administrative data perhaps crept in during the evidence session to which you referred. Administrative data was not developed as a contingency for the ONS’s census; it is a programme of research work that the ONS has been working on for many years as part of its broader functions.

In NRS, I am running not a mini office for national statistics but a very different organisation. With regard to statistics, I am responsible for demographic and population statistics in Scotland, whereas the ONS has a much broader remit. It undertakes a range of social surveys, economic and workforce statistics reporting and various other things that are not part of our organisation. My organisation, in turn, has a large number of functions that the ONS does not have. For example, I am responsible for the registration of births, deaths and marriages in the country, along with the national archives, the public records system and various other things. The functions in our organisation are quite different.

The ONS has for some time been considering how it might use data that exists in other Government organisations that would assist it in its broader statistical functions. It has a legal gateway to do so through the Statistics and Registration Service Act 2007, elements of which might potentially allow other bits of the UK, or the devolved Administrations, to enter into discussions with UK Government departments about access to data. However, I have to stress that administrative data is experimental statistics; it is not confirmed national statistics. The ONS has had a programme of discussion over many years with large Government departments about how they might get access to that data. Those discussions are not yet fully successfully concluded, but are in progress.

From the ONS’s assessment of the impact of Covid on its programme, it reached a view that an available option would be to take work from its separate programme on administrative data and potentially fill in gaps in responses if it had a poor response rate due to Covid. The ONS had sufficient confidence that access to that data would allow it to do so.

That said, there were risks. Administrative data is information that sits in a range of Government systems, often for operational purposes. The ONS would not know, until it had run the census, whether it had the right data available to it or how current that data was. There were also risks of gaps. However, the ONS felt that that provided sufficient assurance, and I am absolutely certain that that was the right call.

For data sharing with organisations, I need to have a legal gateway, but I also need to have a purpose. My purpose is to produce demographic statistics. Crudely, I therefore need to know three things. I need to know how many people are born, die and marry in Scotland. I have that data, because I control the registration system. I also need to know how many people migrate to Scotland. An existing national statistical model is used across the ONS, Wales, Northern Ireland and Scotland to measure migration, so I do not have an immediate justification to approach the Department for Work and Pensions, Her Majesty’s Revenue and Customs and others to ask for data that I do not require.

I am sorry to interrupt, but I want to check something. We were told in the evidence session with Audit Scotland that you were now planning to access some of those data sharing sources. Are you saying that you have not done that or that you have no legal basis for doing it, and that it is not happening?

For the 2022 census, we have been in discussion with Scottish organisations to use elements of administrative data—as well as data sets that NRS owns—in order to quality assure. It is not being used to fill in data gaps; it is a far less broad coverage of administrative data. It is not access to DWP or HMRC case information on individual citizens. It is for a very specific purpose as part of our quality assurance arrangements, but it is not how we would deal with no response or a biased response rate.

Thank you for clearing that up.

I must ask you to try to keep your answers a little briefer. I suspect that you will be able to do that in the forthcoming sections. Again, please feel free to bring in members of your team if you think that they can add useful evidence to the gathering of information that we are embarked on.

I invite Craig Hoy to ask a number of questions.

Good morning, Mr Lowe, and welcome to your colleagues.

Will you give us a bit of a flavour of the work that you did to assess the impact of the delay of the census on the delivery of public services in Scotland, and on the allocation of funding to councils and other public bodies?

Obviously, we undertook a range of investigations on that as part of our decision making. I will say at the start that Professor Sir Ian Diamond, the national statistician and head of the ONS, has stated that there are no insurmountable challenges to Scotland running the census a year behind other parts of the UK. I hope that I can provide that reassurance to start.

The census is a really important exercise. It gathers data, but it does so for a fixed point in time. We then need to use data on population that goes beyond that census year. Each year, we and other parts of the UK generate statistics called mid-year population estimates. We use the last census and make adjustments.

As I explained to the convener, in Scotland, that is about taking the 2011 data, the data for births and deaths in Scotland, which I have anyway as the registrar general, and migration data—we have a UK protocol for how that is developed. Therefore, each year, we can develop statistics based on that census baseline that cover the needs, whether that is allocation of funds to Scotland or regional allocation of funds by Government to the NHS and local authorities. It is about having a robust model.

There is a parallel case. In 2011, the results of the Scottish census were delivered by what was then the General Register Office for Scotland. The publication of the results from the census was a few months out of sync with publication in England, Wales and Northern Ireland, which meant that Scotland went for an additional year using baseline data from the 2001 census. When the results from the 2011 census were compared with that model data, there was less than 1 per cent difference at population level between the two, so we know that that data set and that approach to modelling between years are robust. We also undertook a range of discussions with the ONS and other statistical providers to work through that, and we continue to engage productively with all the UK statistics bodies.

That answer was quite focused on process. We are going to be out of sync. You identified that there are challenges but said that they are not insurmountable. Can you let us know what some of the challenges are?

I will hand over to my director of statistical services, Pete Whitehouse, to answer the detail of that, but the main point is about the use of funding allocations. As I explained, it is about using a methodology that we use each year between censuses and extending it for one year extra, so I do not see the impact as significant.

Population data is one of the fundamental deliveries of the census, and part of that is to drive, support and inform allocations in the country. However, it is also about working with colleagues across the UK to deliver UK population data. As Paul Lowe has set out, we work closely with the ONS, the Welsh Government and Northern Ireland to bring together our data.

We always look at population data from the point of view of bringing the best data that we have. After the 2011 census, there was a gap between when the data was made available for the rest of the UK and when Scotland brought its census estimate online. That was managed through the Government Statistical Service and by considering when retrospective adjustments could be made. In essence, the work was done with colleagues across Government to ensure that the allocations were done as they could be and, when better data came on stream, if there was a need to revise, that was done.

Paul Lowe mentioned Sir Ian Diamond, who has been very strong in his view on how we work together as statistical organisations to deliver the quality of data that is required by our users. We do that by looking at our mid-year estimates, rolling forward from the last census, bringing on the new census data when it comes online, and working with each other to understand our births, deaths and migration information. We do that collaboratively. That is overseen by the Office for Statistics Regulation, which has a keen interest in how UK statistics and population data are used and how that informs our formulas, particularly around that allocation.

There is a great deal of work going on. That will continue, and we will deliver the best available data at the time. As you probably know, in Scotland, health allocations and local authority allocations make great use of population data, but they also build on allocations from previous years and how we roll those forward. We feed into that and, at all times, we deliver the best data that we have at that point.

To go back to one of Paul Lowe’s earlier points, we know that the quality of the data is so important that we have to get it right. The challenge is the risk of going with data that does not meet the needs as against the obvious downside of the census taking place a year later than it otherwise might have done. Our conclusion is that delivering the census one year later, when we have in place processes and methodological solutions, and engagement with our colleagues, is not as bad an option as producing a census that is not fit for purpose, is biased and does not have enough coverage.

I hope that that is helpful.

09:30  

When Stephen Boyle was before the committee, we talked with him about the potential impact of a further delay in the census. The decisions of the past 48 hours, and the Covid data that has emerged, probably mean that that is less likely to happen. Nevertheless, could you give me a flavour of the options that NRS has considered for the census if lockdown restrictions return? In addition, have any resources or money been committed to contingency planning in that regard?

Over the past year, we have used the time, and some of the additional money, to ensure that we have flexible ways of delivering the census, based on different Covid scenarios. There is a range of things that we can do. We do not intend to delay, even without the very positive data on Covid that has, as you say, been received in recent days, although, sadly, the Covid situation can get worse in quite a short period.

We have a headquarters function that directs the operation of the census. We have developed capability to run that function remotely if we need to do so, in order to reduce the risk of key people being taken out of commission through being ill with Covid. We have capability to run either a physical or a remote set-up for the headquarters function. We have a contact centre, and the vast majority of its capability will now be delivered by operators working in different locations, rather than in large teams on a single site, which massively reduces the impact on them.

We have made changes to the delivery of the census. In 2011, we had community hubs that people could attend to get help, support and advice. We have taken a decision not to run those hubs this time, but we have massively increased the size of our contact centre so that we have staff who can speak to people. They can help people by giving them advice and talking them through the process, and, if necessary, helping them to complete their census return. We have invested in that, and in a range of other technologies such as a web chat function. There is also a lot of advice and help in the online system.

The other changes that we have made relate to the field force. In the 2011 census, we had 7,000 field force staff, and the figure was higher in prior census exercises. As part of our digital design, we are moving to a much lower number of field force staff. If people complete the census online, they do not need to interact with anyone else—they can do it safely in their own home. If they want a paper form, they can go online and request one, or phone us, and they will be provided with one by post. Nobody from NRS will turn up on their doorstep to hand them a census form.

Technology also allows us, rather than sending out field force staff en masse across Scotland, to target those staff in places where people have not responded. We will know that from our systems. That will mean that there will be many fewer interactions on the doorstep.

Finally, we have invested a lot in training our field force staff and in their kit and equipment, so interactions will be at the doorstep, from a distance, and they will be short term. We are engaging closely with the chief medical officer’s office on protocols and arrangements to ensure that it is content with our plans and proposals.

I hope that that helps.

It does, thank you.

I have one final question before we move on to the financial impact of the delay. First, I want to put it out there that the cost of the delay was £21.6 million. Given your previous answer, would it be right to say that, even though there is a perception that that money may be lost, you have actually used it to invest in greater flexibility, agility and efficiency in the system? Perhaps crisis begets opportunity in that respect.

Absolutely, Mr Hoy. None of that money is wasted. It has been about making sure that things such as the technology that we have developed continue to be refined and improved. We have used the time to do additional engagement with users and stakeholders, to improve the systems and services. We have also used it to learn lessons from the ONS and NISRA censuses. As part of the close collaboration that Peter Whitehouse has referenced, we have gathered lots of lessons learned on the experience and have made sure that we built that into our programme to make it stronger.

In addition, during lockdown, there has been a significant uptick in cybersecurity incidents. People will have seen in the press the range of global impacts of that. We have therefore taken the time to do additional testing and to strengthen our cybersecurity arrangements.

I agree that £21.6 million is a significant proportion of the programme cost. However, we have to balance the risk of failure of a census that costs £117 million with the additional costs of achieving the benefits of the census. International and UK studies indicate that, for every £1 invested in a census, the equivalent of between £5 and £6 is returned in wider economic value.

As part of the ONS’s decision making on not delaying its census, it calculated that the cost of delaying by a year would be £360 million. The cost of delivering its census programme was in the region of £906 million. That is a relative cost of well in excess of 35 per cent. That probably helps to situate the £21.6 million, which is an 18 per cent increase. That shows that we have taken things seriously and have tried to keep down the running costs of the additional year. However, we cannot be too stringent because, if we were to cut too deep, we would just not get good-quality census data.

I hope that that helps.

Thank you very much.

We will return to the issue of the digital dimension, which Mr Lowe raised in his answers to Craig Hoy’s questions. I will come to Willie Coffey to ask questions on that. However, before that, I will bring in Colin Beattie to continue on the theme of the financial implications of the delay to the census.

I will direct my questions to Paul Lowe; if it is appropriate, he can bounce them on to others.

On the face of it, the financial impact of £21.6 million seems like a lot of money. You have mentioned that some of that was used positively to invest in other preparations. I am struggling a bit with that, because we see that the increased supplier costs are £14.4 million, because of the extension of supplier contracts and an increase in the cost of goods. What goods are so significant that they contribute to that £14.4 million, and how has that been turned into a benefit, when it is the suppliers that are getting the increased costs?

I will hand over in a second to the accountable officer, Linda Sinclair, who can answer in greater detail.

The reschedule allowed us to do a number of things. As did census programmes in other parts of the UK, the census programme in Scotland used a hybrid of organisational resources and capability and a chain of contracts and services for its delivery. Inevitably, if the duration of the census is extended, there is an increase for suppliers, because they have to remain on that contract and to be involved in the work for a year longer.

However, I go back to the point about benefit. We used that time well. We increased additional testing and used it to make sure that we were in a robust place to deliver the census. We used the opportunity to make some improvements to the census; those are an element of the supplier costs that are referenced.

The supplier costs totalled around £14.4 million of the £21.6 million. We went through two rounds of costing with the suppliers to ensure that we were getting value for money from the different contracts. We have about 20 separate contracts on the census programme, which provide a range of different services to the programme. Within the time that we had available, we could also look at the design—Paul Lowe touched on this—and the things that we were asking our suppliers to provide for us. We worked through the learnings from the ONS and NISRA, looked at specific contracts and then worked out what that would mean for the activities that they were delivering for us.

As an example, some of the activities are volume and demand led, so within the time between March and April when the census is running live, we can look at—

I am sorry; I am having a problem with my network. I am not sure that you can hear me. Can you still hear me?

I can hear you perfectly.

For example, we were looking at how much we needed to draw down on those contracts at different points and different times with our different suppliers, such as APS, and working through the detail of the activities that they were undertaking, so that we could be sure that we were getting maximum value from each contract. I would be happy to write to the committee with more detail on that, because it is a very complex picture to capture in a short time. It was all about looking at the value that we can achieve from additional costs and the design of what is in the contracts. That changes over time, as we learned from the likes of the ONS and NISRA.

The impression that I got from Paul Lowe was that the £21.6 million is mitigated by the benefits that you have been able to achieve in additional experience and being able to look more closely at the programme and enhance it and so on, but that still does not actually mitigate the £21.6 million. It is still costing £14.4 million more than it would have had the census gone through.

It says specifically in the report that it is due to the cost of extending suppliers’ contracts. I do not know whether you negotiated with the suppliers to change their contracts, but that would obviously create an additional cost element. It is also mentioned that part of the additional cost is the increase in the cost of goods. What are those goods that are so significant?

The contracts had enough flexibility that we could extend the time period, so the major cost drivers were the continued resourcing.

We have an arrangement with different suppliers whereby we draw on their resources to work with us to deliver the census, and we then allocated that work across the additional year, which meant an additional resource requirement for the suppliers. That was a major part of the cost, together with continued licensing. We had to extend some of our digital equipment licences and other digital licences that we have in place for delivering the census and other aspects such as hosting and support costs.

There were underpinning costs of what the suppliers were delivering to allow us to then deliver the census in a collaborative environment with the suppliers. A range of things within the design of that allowed us then to look at how we get better value from how we work and how we deliver the census with our suppliers.

Apparently the better value came at a cost of £14.4 million.

That is the total cost.

I am still struggling a bit with the positivity around this. You had a bit of extra time, which gave you a chance to spend more time enhancing and improving delivery of the census. I presume that that had a cost in itself, and that it is rolled up into the £14.4 million.

You still have not answered the question about what the goods are that are so significantly increased in price that they warrant a mention.

09:45  

On the pricing aspects, we have a range of different contracts. Some are framework contracts and are therefore based on Scottish Government pricing. There are flexibilities within those that increase the pricing of the contracts. For example, changes to Royal Mail prices impact our overall pricing because of how much it costs us to post out paper forms. That is one impact that we had to price in as part and parcel of the additional costs. There are other impacts, such as changes in licensing costs where there are flexibilities under individual contracts that we need to put in place.

Part of the £14.4 million other costs was additional costs for manual coding. As part of the rehearsal for the programme, we considered how we run the manual coding element. Around £1 million-worth of the £14.4 million was for manual coding and additional activities associated with that from learnings from the rehearsal. There are costs for how long it will take to code individual paper forms that come back in. Those are based on learning from the rehearsal that it would take slightly longer than anticipated, so that is also a cost driver.

There are a range of different cost drivers. It is hard to encapsulate that in one simple response.

You are telling me that the original specifications for delivery of the census were underbudgeted, because you are saying that you found out that some things would take longer and added it into the cost subsequently. It appears to me that your original costing was not accurate.

It is the impact of adding an additional year and asking our suppliers to work with us for that year. The costs of their resourcing, the costs of the supplier and the hosting costs for that additional year make up the vast majority of that difference. It is about the length of time.

You were talking about the coding of pieces of paper.

That is one other element.

You mentioned it specifically and said that the exercise that you did showed that it would take longer than you anticipated. That must indicate that you would have had a budget overrun had the census gone ahead as originally planned.

We would have. That is one element of the £14.4 million. There were two elements of additional goods to which I drew attention: one is the Royal Mail pricing under a framework contract and the other relates to the additional costs of manual coding, which is an additional cost that we would have incurred in any case. We have set that out previously.

The convener asked for a copy of the options appraisal. Will you confirm that that will contain details of the additional costs that would have been incurred had the census gone ahead in 2021?

Yes. In the options appraisal, we considered a range of different aspects had the census gone ahead in 2021. Paul Lowe set out the different options, such as using no field force. We also considered costings around mitigations such as not using a field force or going paper only and how that might impact the manual processing of, and the increased postage costs associated with, paper forms, for instance.

Will you confirm also that the options appraisal contains details of the estimated additional expenses that were expected due to Covid mitigations if you had gone ahead with the census in March 2021?

There was an element of Covid-related mitigation cost but, at that point, we did not price whether we needed to use personal protective equipment and suchlike. It was difficult at that point to get a sense of the exact pricing. However, we can look at what information we had at the time and set it out for you.

Okay. You said that supplier costs increased, which was a substantial portion of the cost increase of the programme. What actions did you take to mitigate those costs?

We went through two rounds of discussions with our suppliers about the overall costs; we worked through the impact if we ran the programme forward for an additional year and then we interrogated the costs. We had an indicative set of costings from suppliers, which we interrogated, and we went back to suppliers on the detail, to ensure that sufficient contingency was in place and, equally, that no cost overruns were included. We also did a significant review of the programme resourcing, which is the other main driver of cost, to ensure that that was sufficient but was not overplaying the additional resource requirement for the programme. We scrutinised the detail on that to ensure that we were achieving value for money.

Earlier in 2021-22, you identified additional financial pressures of £1.5 million in the census. You have reduced that to £0.5 million through implementation of mitigating actions. What are those mitigating actions?

We have undertaken a range of activities this year around the census budget. There is fluidity to the delivery of the census, which means that the budgets will change over time. Partly, we have gone through a deep dive to ensure that we prioritise spend on all the things that we absolutely need and that there are no nice-to-haves, so we have been able to recycle savings. We have looked again at supplier contracts and recycled savings on things such as travel and subsistence costs, which have not been required in the Covid environment.

Particularly for March and April next year, a lot of the census delivery is based on volumetrics. It is volume based and demand led, because of the number of people who will submit a response online versus the number who will submit a response on paper, and the different associated processing costs. Therefore, we have assumptions built into the census that are based on the numbers and proportions that might work in different ways and how many forms we need to post out versus how many we need to process manually. We have taken learning from ONS and NISRA numbers, and we have reassessed all our volumetrics and assumptions in the live period that crosses the end of the financial year. We need to look carefully, on a day-by-day basis, at when some of those costs will land, from an accrual accounting perspective.

We have gone through the detail of those volumetrics and worked down some of the costs to make sure that we are in a more realistic position. We have looked carefully at the alignment of costings at the year end to ensure that we do not load large sums of money for postal drops, for example, in the last days of the financial year and that it is appropriately timed. That has allowed us to bring down the costings.

We have also continued to keep a clear and careful view of our overall resourcing costs, so there is a very strict control regime in relation to resourcing to ensure that we keep those costs managed and affordable over the programme.

Can you confirm that you have now achieved financial balance?

We are still working through some uncertainties. I mentioned the Royal Mail pricing, for example, and we were waiting for that confirmation to come through from Royal Mail. That has been coming through in the past couple of days and is being costed out within the programme, so that we can look at where we sit within the balance around that and whether we have sufficient funding. We are very close to that financial balance, and we are confident that we will bring the budget into balance at the end of the financial year.

The figure that I saw was that there was still a gap of £0.5 million. Does that gap remain?

It is around that—it is between balance and £0.5 million. A fair number of aspects are still moving on that. We are also going through some detailed resource costings towards the end of the financial year, so another updated set of numbers will come through, but we are very close to balance, and that is our confident position.

Obviously, one of the main costs is people. What you are doing to manage the workforce costs, particularly your numbers of temporary staff? What are you doing to make sure that you keep a tight hold on that?

The census takes up around 30 to 40 per cent of NRS’s current workforce, so we have to grow quite significantly, particularly in the years that surround the census, in order to deliver the census. It is not sustainable for us to carry that level of permanent staff, so we go through a range of recruitment resourcing approaches and look to bring on board individuals on the most affordable basis. Our workforce plan allows us to forward plan the costs and consider the financial implications of the resourcing plan. We go through a work-based control process to allocate the resources and keep a tight view of our overall costings. We also go through regular resource reviews and will continue to do so. We completed one at the back end of last year and have another planned for May or June this year so that we go through that regular cycle of refreshing the information around our resources.

We work on the basis of trying to use the most affordable route first, but that is not always possible because of the nature of the very specialised roles that we bring in for the census, and we are at the mercy of the wider prevailing contracting and resourcing market. Some of the more specialised digital skills are hard to reach and can attract a premium. There are tests all the way through the process before we agree to go to the external market to bring in those harder-to-reach and higher-cost premium resources.

I remind you that if you want to augment the oral evidence that you have given us in writing, please feel free to do that. That may be useful and you alluded to that at the start of that round of questions. It would be helpful for the committee if you could give us the updated figure on the balance when you have it.

I turn to a series of questions on the management of the census programme. I invite Willie Coffey and then Sharon Dowey to come in.

Good morning. I will start by asking Paul Lowe a few questions about the mechanism for getting the census—[Inaudible.]—flip between digital and paper. You talked about that earlier, but could you give us some more details about how that will work?

Hitherto, everybody got a paper copy delivered to their house, so you could be pretty sure that everyone had it. From what you said, this way, everybody will get a letter and people who arenae connected to the digital world can phone you and get one posted to them. What happens if they do not do that? How will we know who has not bothered to engage with the process? Will you have people in place who can check and determine that, and then deliver the census to people? Can you talk us through how you will get the census into everybody’s hands?

As you said, the letter that will go out across Scotland will provide a range of information about how to participate. It will include a unique code for each household that will be the key to accessing the online platform and completing the census return. However, paper forms will be retained for those who do not wish to, or who are unable to, provide a digital response.

Behind the scenes, we have systems that will track how many requests for paper forms we have had and whether we have received the completed forms. When people complete paper forms and post them back to us, those forms are scanned and the information is read off those scanned forms, so we have a record of forms that we have sent out, forms that have been returned and the data that we have captured from them. We will be able to use that to generate reports, so if we know that people have not completed their census online or on paper, there will be a chain of reminder letters. Ultimately, if there is no response after a period of time, the field force will attend the address and make contact with the household. I hope that that helps.

10:00  

Yes. So, if people ignore the invitation to complete the census, they will still get a visit and a paper copy. That covers that.

What did you say earlier about the expected likely percentage? You talked about an all-paper or all-digital model, or a mix of paper and digital. Did you say that there is a drop-off in the response rate if you mix those models of inviting contributions from people?

I am sorry if I was not clear about that earlier. In the census rehearsal that we ran in 2019, we provided people in three local authority areas with the opportunity to participate voluntarily. They were given access to the digital system as well as the ability to request paper forms.

In our assessment of how we might have tried to deliver the census in 2021, one option was to replicate the rehearsal and provide both paper and digital methods. We anticipated that that would have provided the highest return rate without using a field force. The paper-only option was likely to have had the next highest return rate, and the response rate for digital only was likely to have been lower, for the reasons that we talked about earlier. I hope that that helps.

If a huge number of responses to the census are completed digitally and electronically, who has access to that data? Who owns it, and how secure is it? Can you talk us through that, please?

That is an important question. NRS owns the data and it is kept confidential for 100 years. Later this year, we will publish the results of the 1921 census, which will be of great interest to historians, genealogists and others. Everyone who works on the census programme signs a confidentiality agreement, and disclosing census data would also potentially be a breach of the criminal law.

With regard to protecting the data, we have built rigorous Government information standards, and cybersecurity and data security standards, into the organisation. Our online systems have been subject to independent information assurance reviews by an organisation called Bridewell Consulting, which is accredited for assessment by the National Cyber Security Centre.

We published our first assessment prior to the census rehearsal in 2019. The next one for the census itself will be published at the end of January, and that will be made public so that anyone can see the robustness of our security arrangements. We have done a lot of work with the National Cyber Security Centre, Police Scotland, the Scottish Government and others to put in place robust cybersecurity measures to protect census data and preserve the online systems.

The data that is ultimately submitted by a person in Scotland, either digitally or on paper, is not used for any purpose other than the census, is it? It is not given to third parties, and they would not have any access, or any rights, to it. It is fairly secure and is for that single purpose. Is that correct?

If you are interested in hearing more detail on that, Pete Whitehouse can probably help, or we could write to the committee to follow up on it.

A range of statistical approaches are applied to census data, and data is anonymised so that it is not possible to identify individuals. We are not interested in individuals—we are interested in the accumulation of data. A lot of work is done to anonymise the data, and various other steps are taken, such as disclosure control, which means that it would not be possible, if somebody were to hack us, for individualised data to be extracted. That is the kind of processing work that goes on after the census data is submitted.

We do not sell census data; it is used to inform the statistics for the purposes of the census.

With regard to online submission, can people do that in little bits and bobs rather than completing the form from start to finish in one go? Can they pick up a bit of it and review it later on, and then submit the whole thing at the end when it is complete? Can they take their time about it?

That is absolutely correct, Mr Coffey. When you access the platform, you are able to set up a password. You can complete a bit of the census and then come back to it later on. As long as you remember your password and have your code, you can go through and complete it in stages.

How do you envisage the census developing in the future? I know that we might be 10 years away from the next one, but do you see us continuing with the digital solution, mixed with paper? Will we continue to use that model in future?

It is difficult to have a telescope to the future because so much happens in 10 years. We need only think about the technology that was around in 2001 and how that compares with what is available to us now.

Pete Whitehouse and his team are leading on some work to consider the future options for delivering the census. It would be hard to anticipate delivering a future census that did not have a significant digital component, but there is still a recognisable issue with digital exclusion, even though digital participation has increased during the pandemic.

Our colleagues in ONS are looking at whether we can use other existing surveys and data that exists elsewhere to completely replace the census. They are reporting on that in 2023. It will be interesting to see the results of that work.

However, in NRS, we are considering at a high level whether we will deliver the next census in a way that is broadly similarly to the way that we are doing it this year or whether we will use a hybrid of a survey combined with other data sources. A lot of work and a lot of investigation needs to go on in the next few years in relation to that.

That is very helpful and reassuring.

The census programme has been subject to a number of reviews. Challenges were first identified in 2018 and, in March 2019, the programme status was marked as red. Following the implementation of a recovery plan, the subsequent reviews noted significant improvements. A review in February 2020 gave the programme an amber rating and concluded that the census would have been able to proceed in 2021 had it not been for the pandemic. The most recent review, at the end of November 2021, gave the programme an amber/green rating.

I have two questions, which, in the interests of time, I will ask together. How is NRS managing risks, particularly around finance and resourcing, in the run-up to the census going live? What further review activity is anticipated over the remainder of the programme?

We have a well-developed risk management approach in the programme. We have a dedicated risk group and manage risk at individual project level, programme level and strategic level. That is reported to our programme board and to our NRS audit and risk committee. There are deep-dive sessions on risk, risk management, current risks and issues in the programme, so we have a well-developed risk management regime.

In his oral evidence and his section 22 report, the Auditor General recognised that we were doing the right things but said that we needed to keep doing them. That is where we are.

On resources, we have the people we need to deliver the census collection. To be frank, with a matter of weeks to go, there are not opportunities to get other people into the programme. We have an understood and managed position for the resources that we need to deliver the census collection.

There are other phases of the programme. We get the data and then we need to do lots of complex statistical work on it and prepare it for the release of the census results, so we will bring in resources to focus on that. Equally, people who are involved in the collection will, as we move through the year, no longer be required and will drop off the programme. Some new people will come in but, at the same time, quite a lot of people will leave, because they will no longer be required for that phase of work.

Does that help?

Yes, that is fine. You are happy with the process that you have in place.

We have robust workforce planning arrangements. Linda Sinclair and I chair a resources group in NRS, so we keep a close eye on census resourcing and cost.

Thank you. I hand back to the convener.

Thank you, Sharon.

I conclude the session by thanking the witnesses—Paul Lowe, Linda Sinclair, Peter Whitehouse and Anne Slater—for joining us. I do not think that we heard from Anne Slater, but she has been on hand to give evidence.

Reflecting on what Paul Lowe said at the beginning, we understand that NRS is not the Office for National Statistics, but it is the body that is charged with delivering the census in Scotland. We are concerned about the efficiency and effectiveness of that process. It is exactly two months to the day until census day on 20 March, and the committee may well invite the witnesses back after the census later in the year—it will be for the committee to decide that—to reflect on how it went and any broader lessons that can be learned, as well as to consider the future.

Paul Lowe said that 10 years is a long time. I am not sure whether the next census will be in 10 years’ time or nine years’ time. We have not touched on that question, but I am sure that we can return to it.

I reiterate that we welcome your commitment to give us a copy of the options appraisal. If any other aspects of the evidence that you have given this morning would be supported well by any written submissions, the committee would welcome that. Thank you for your time and your willingness to answer the questions that we have put.

I suspend the meeting while we change witnesses.

10:11 Meeting suspended.  

10:14 On resuming—