Family Finance Surveys User Conference 2026 - Session 3

Hello and everyone. Welcome back from lunch. Um before we get started, there's just been a few changes to the program. So I'm just going to run through those quickly. Um so the first session we're going to start off with the ONS updates on household income and expenditure and we're also going to have a presentation on the analysis of household digital costs. That would be in session four has moved to session three. and um Eve Little who will be from the impact team will be taking part in session 4. Um but other oh and sorry there'll be a slight adjustment to the tea and coffee break. So rather than 2:35 it will be 2:40. Um otherwise I'm going to pass over to Becca Briggs now who's going to chair session three for us. Thank you. Um hi everyone. Um I'm Becca Briggs. I'm the deputy director for household resilience at the ONS and really pleased to be here today. Very much enjoyed the morning and I'm sure I'm very much going to enjoy the afternoon as well. So um well without much further ado um we like we said we've just switched the session slightly. So first up is going to be um an update from some of my lovely ONS colleagues uh focusing on household income and expenditure. So just to give a bit of background to our speakers, we've got Lee Culvin who will be joining us online. and he heads up our household income and expenditure analysis team. Has been doing that for the last six years. So looks after all of our outputs in that space. Um and prior to that he was a teacher for a long time as well. So brings a bit of a different background to our teams. Um and then uh joining him will be Eliza Swan who leads the living cost and food survey developments team at ONS. Um and has a long experience working in surveys. Prior to ONS, she worked at research organization Picker. Um, also the CQC's National Patient Survey Program and the NHS staff survey. So, I think Eliza, you might say a few words before we hand to Lee. >> Hi everyone. Um, right. So, I'm just quickly going to introduce the session, then I'll pass straight over to Lee for his se uh se section. Um so uh what we're going to be covering today um firstly Lee's going to do an update on um an overview of the household finance statistics at ONS and talk you through some of the highlights from um his team's very recent publication of the ONS um family spending um for uh financial year ending 2025 which was just published recently. Um, and then I'm going to take over and talk about um the Office for Statistics Regulation, the OSR review of those statistics, which was published earlier this year, and give you a bit of an update on what we're doing about the recommendations. The reason I'm taking that bit is because the living costs and food survey is um is the the data source for the for family spending and um most of the recommendations relate to uh the LCF. And then after that, I'll run you through an update on uh developments in the living cost and food survey. But without further ado, I'll pass straight over to Lee. >> Thank you, Eliza. Um, I apologize for not being able to attend in person today. Um, my name is Lee Culvin and I will be and I am the team lead of household income expansion analysis team in ONS. Um, and today I'm going to take you through what we've done over the last 12 months. So recently in ONS um we have published our family spending publication for financial year end 2025. Very recently the family spending has been reviewed by OSR and for the financial year ending 2025 publication we updated our equivalization method to use the modified OECD two person reference household rather than the single person approach which it was using previously and that was part of our ongoing quad work to better align our accredited household income uh and expense statistics the ons um journeys 12 months as well we engage the users cap the capture the user to needs through the completion of the offenses and taxes and benefits data table review in the next 12 months. Now we um um are going to be looking ahead and looking at um intend to publish a high level household uh financial recovery plan as part of the next quarterly update um to the ONS economic statistics plan and communicate further detail with users and that is going to be on the 15th of July. Um we are continue to improve our production pipelines for income and spend statistics moving away from legacy systems and to continue our work to improve the cost cutting insights for household income and expenditure stats. We are also looking to do two publications in the autumn this year looking at the average household income for the financial year end in 2025 um which will be published first and then not long after we're planning to publish the effects of taxes and benefits for the finance year in 2025 as well. Um the stakeholder engagement strategy will also be published uh in the next 12 months and engagement mechanisms for income and expenditure with stakeholders will be under review in that period as well. Um so we'll go straight into now um the highlights from the ONS's family spending publication for financial year 2025. We're going to look at what households are spending the most on. How's expenditure changing across the distribution? And compared with financial year in 2023, how was the spending of the top and bottom fifth of households changed? Um and also why stated both absolute and relative year changes in spending in this presentation are adjusted for inflation um and are presented in real terms. Now in the financial year 2025 publication, the chief sample was 5,000 households which was a big improvement on the previous year. So very welcomed. So let's have a look at what the average weekly household expendure was for financial year in 2025. After adjusting for inflation, it increased by 5%. Now it increased um up to 260 uh £676 in the financial year in 2025 which was a nominal increase of 9% um of £5330 compared to the previous year and after taking account for inflation um there was a real terms increase of £3510 which is 5% showing that a large proportion of this nominal increase was from price increases alone. Um when we even though we've had this 5% real terms increase um in the in financial year in 2025 the average weekly house expenditure still remained 10% or £7560 per week lower than in financial year in 2020. So it still hasn't fully recovered yet. If we then look at the household expenditure across the main KO cop categories for the fure in 2025 um we after for inflation um household spending increased across most expensure categories this year which is um which was very different from the previous year where we saw there was only certain areas that grew in the financial year in 2024. The largest increases financial year in 2025 were seen in transport, recreation and culture, restaurant, hotels, housing, net, fuel and power. Transport increased by 9% which was810 and this was mainly driven by the purchase of um vehicles which increased by 25 23% compared to the year before about 710. um and particularly that was on secondhand vehicles which increased by 25% or £550. There was a slight increase as well um in transport services uh of 5%. Recreation culture increased by 9% which was 6.70 in the finance year end 2025 and this was mainly driven by increases in recre recreational cultural services which increased by 9%. package holidays in particularly package holidays abroad which increased by 14% or £410 a week. Restaurants um and hotels increased by 6% um or 280 um which again was mainly driven by restaurant and cafe meals uh about 9% increase or 140 a week. Housing debt and fuel increased by 5% um um which was about560 um and this was mainly driven by an increase in net rent which was 6% or3 pound30 a week um and electricity gas and other fuels which increased by 7% compared to previous year of about 220. So where were households choosing to spend a large proportion of their expenditure on it? It it staying pretty consistent to the previous year which is housing net fuel and power and transport continued to be account for the highest portion of household spending. Um around 18% of all total expenditure was on housing net and fuel and power about £1840. Transport was around 14% of total expenditure 9640. Recreation and culture then is about 12% of total household expenditure about £8210 a week. And food and all alcoholic drinks then was 11% or £73.70 a week. So let's start looking at how this is different across distribution. So the difference in real-time weekly expenditure between the richest and the poorest fifer households has widened in the financial year in 2025. The richest fifth spent about £1,18360 a week which was 2.7 times that of the poorest fifth which had an expenditure of about £4730 in financial year in 2025. Now this widened this difference in expenditure widened from the previous year and it's reached its highest level since [snorts] 2021 increasing from 2.5 times to 2.7 times. The richest the richest fifth spending rose by about 98 sorry not 100 9810 or 10% which was twice the rate at which the poorest fifth increased theirs by only 5% which was £1810. a stark difference in terms of expenditure. We're going to have a look at then what the difference where the poorest richest households spent most of their to their expenditure as proportion of total expenditure. And we can see that the poorest households spent proportionally more on housing, fuel and power than the richer households. Now the poorest fifth spent about 25% of their total expenditure or £104.70 on housing net and fuel and power. And this was most of this was really on net rent. About 53.6% of the housing net and fuel and power expenditure was on net rent about £8330. They also then were spending more on food about 15% 60 £6180 and that has mainly come from food purchasing itself. About 91% of this figure comes from food purchasing which is £5650. And then transport makes up about 10 point 3% of their total expenditure and that is mainly made up from operational personal costs personal transport which was 39% of that figure at £1650. Now the richest households are slightly different um but they have a slightly different profile. They're more homeowners than they are renters. So their expense is focused in different ways. So their highest the large portion of their total expenditure goes on other expenditure which was 1780 or 16.5% of the total expenditure and that mainly was made up from mortgage interest payments about 40% of that figure was for mortgage interests which made was around £7180 a week. Then transport made up the second largest um spend for them which was £17460 or 16.1%. And the most of this figure about 45% of this figure came from purchasing of vehicles about £80 a week. Recreation and culture then was the third largest um direction of expenditure and this mainly was made up from um package holidays. About 45% of the £145 came from package holidays about at £64.70. So quite a stark difference between the rich poorest households where they're prioritizing the spending. So when we were going through the publication this year, we um saw that spending plans had changed significantly between financial year endings uh 2023. Now the analysis for the financial year 25 showed this and what it shown is that the spending shifted unevenly between the poorest and richest households where they were changing their spending. And if we look at the um that now we can see that spending increased far more for the richest uh fifth than the poorest households. Um if we can move on to the next slide please um Eliza. Um and they were just spending more. The richer households are just spending spending more. They've increased expenditure quite substantially where the poorest fifth had increased spending but mostly to keep pace with increases in in price rises. So the poorest fist largest real terms increase was in the housing fuel and power uh at about £5.90 and that was 6% real terms increase and this was mainly contributed by an increase in um electricity, gas and other fuels which increased by 6% or £420. They again increased the expenditure in food about 5% in real terms and again all this was mainly coming from just food purchases um maintaining their food consumption. Now with recreation and culture again that increased by 20% compar um compared to France year in 2023 and this was driven by pack and package holidays 50 54 there was a 53% increase in package holidays or£380 a week. So yeah, it sounds quite significant but it was only 3 pound80 a week that it increased that by um with the richest fifth the largest real terms increases came in the trans transport which is 40 um 46 a week increase or 36% and again it goes back to this purchasing of vehicles. Um it was a £33.34 uh p a week increase or 71% increase compared to 2023. And for recreation and culture again they had the 22% increase compared to financial in 2023 or £26 20 and again that's package holidays um 19% or £10 a week being spent on increase on that. their missed good and services um increased by 24%. And when we looked into this to see what was really driving this increase in expenditure, it was mainly that their increase in in insurance expenditure had gone up by 23% or £750 a week. So you can see there's real stark contrast there between the two charts showing increases in real terms expansion between the richest and poorest households. It's very different. Um it shows how they're choosing to spend the money and constraints different households have on their spending. If we move on to the next slide Eliza um the upcoming work we have now in the next 12 months is we're looking at the average household income publication as I said. So this will be looking at the estimates of the average household income in the UK with the analysis for how the measures have changed over time and that financial year in 2025 and that's going to come out in the autumn this year. We have the effects from taxes and benefit on households and that looks at the redistitutional effects of individuals and households and direct and indirect tax taxation and the benefits received in cash or kind from the UK government financial year in 2025 and that's also going to come out in the autumn of this year. We are also continuing our methods development which is again continuing our work to improve the combination of our income and expendure data and also to improve our co coherency with other um income and expendure um um producers and thank you for your time and also if you had any queries about the data please reach out to the family spending in inbox there. If any questions please ask. [applause] Thanks Lee. Um I'll move straight onto the LCF section and then we can take questions of everything at the end if that's all right. Um oh sorry yes so first of all talking about the office for statistics regulation review. Um so the OSR published their compliance review of family spending statistics in February uh and recognized the challenges we face in our data collection in the LCF um which is the family data spending data source as I've mentioned [clears throat] and acknowledged the range of interventions and developments we have in place. They concluded that while there there have been growing issues around achieved sample size and timeliness, considering the positive use of feedback and the current and planned improvements to the survey, these statistics remain of sufficient value and quality to meet users needs. Um, and they confirmed um that family spending continues to comply with the code of practice and can continue to be published as accredited official statistics. Um, so that was uh good news for us and for Lee's team. um and they made three very helpful recommendations which we are um working on at the minute and I'm just going to talk you through briefly now. So first of all um recommendation one was to help users mitigate data quality issues. ONS should provide more accessible explanations of data changes. So um we're we're taking a staged approach to addressing this. Um and our um first steps involve engaging directly with users regarding the previous variable change issues because we really need to understand exactly what information is needed and where where the issues are. So we've already begun engagement on this at our um uh expenditure steering group in March. Um and we'll be following up again with that group to seek their input and obviously if anyone in this room has has views on it, we're we're always um happy to receive information on that. Um we're also in the process of vazing with OSR to identify key users who have raised this concern so we can follow up with them directly. [snorts] Uh after that um based on these learnings we will review and improve signposting to existing materials and variable guides in place for um the UKDS and the SRS metadata and consider whether we need to produce additional information to improve clarity. Uh and then after that we will be mapping out upcoming anticipated variable changes um as a result of some of the work I'll be talking through in a minute and and plan how we'll we'll communicate these to ensure you are getting the information you need to understand our data changes. Um and finally then establish an ongoing approach so this doesn't become an issue again in future um to make sure we're providing the variable change information uh using for example the future quality and methodology information um survey guide documents. The second recommendation was that ONS should set out the steps it is taking to ensure that the speed um the speed of data processing uh sorry taking to ensure that we speed up data processing to uh deliver this family spending output closer to the target of one year after the close of survey data collection period. So as noted here we've made significant significant improvements in processing data handling and collaboration um which has already had a had a had a benefit. We um published family spending three months earlier uh this time around than last time, but we know there is still a way to go to meet that um one-year target. Uh and there are a range of um changes we've made, including closer collaboration with the uh between the survey and analysis teams, as well as some of the um technical changes I'll be talking through as part of my work, my um discussion of the LCF developments. So, I won't dwell on those right now. Finally then the third recommendation was that ONS should update the quality and methodology information and include explanations about survey bias representativeness comparability and the implications for the use of the data. Um so we are working on this at the minute. Uh we'll be rewriting the QMI um to include more focused analysis on those areas that that were identified as um as needing improvement. Um and uh we're looking to publish the updated QMI sometime in the autumn of this year. So hopefully that will address those needs. Moving on then to talk through the developments in the living costs and food survey. Um as I mentioned I lead um the sorry as Becca mentioned I lead the inputs development team on the LCF. I've got several colleagues working on uh other um development teams looking at the processing for example and then some of the kind of technical infrastructure that sits behind this. Some of these are that will come to my projects. Some of them are are other people's. Um but just to give a bit of wider context, I think um Jen Farell, my uh deputy director, spoke last year at this conference about the uh ONS's wider program to improve our our surveys. Um and this really fits into part of that and is some of the the sort of detail of what's actually going on to improve the quality, resilience, and sustainability of our economic statistics and the surveys that underpin them. Um so I I won't read through them all but we've listed across the bottom the drivers for LCF improvement obviously things like respondent and interviewer burden um coding and editing and processing demands the sessation of the survey on living conditions um and then some stuff around wider data quality concerns and and updated classifications. So here's an overview of the projects we're undertaking as part of the LCF development work. Um the the green ones are the projects that we have now fully implemented. Um and all the blue ones are ones that we're working on at the minute. I'm going to talk through each of these with a bit more detail now. So I won't I won't dwell on this slide. So, first of all, um I'm sure you'll be all pleased to hear if you don't know already that we're implementing a boost or we have implemented a boost, sorry, to the LCF sample size, increasing the GB sample from roughly 20,000 to 32,500 households annually, which is brilliant news. Um, and the sample design remains consistent with the prior approach. We did um look into that with methodology colleagues and concluded that was the right way forward. Um and so because that came in in April, we're already starting to see um increases in in achieved sample which is brilliant news. The next um intervention that we have implemented uh already uh is the record of spending tool. Um so this is very much something that uh as data users you won't see this is behind the scenes but this is a tool that um our interviewers used to input respondent spending information um into our systems. Uh prior to this interviewers collected receipts from respondents um manually entered them into an Excel spreadsheet and then we had a whole separate team of people who had to enter that into our our actual data processing systems. Um this cuts out that dual entry. Um which obviously improves data quality and management. Um it also paves the way for future data collection improvements and efficiencies um including the upcoming receipt assistant tool I'll talk about in a minute as well as um things like the blaze 5 uplift and a potential respondent facing tool which we're not actually working on now but this is it's something that's been talked about for a long time and wouldn't be possible without this this step. Um, and this is a long time in the the making. I wouldn't underestimate the complexity of the systems, the legacy systems we're dealing with on the LCF given how long the survey has been around. So, the other data collection systems improvements we're working on um currently uh I I mentioned briefly the receipt assistant tool. So this is a tool that is being developed by my colleagues in the ONS data science campus which uses AI and machine learning to support LCF receipt processing. So it it reads the images of the receipts, extracts the item level details and helps to classify um expenditure according to our extremely detailed LCF coding frame. Um this should improve the quality, the timeliness and the granularity of the household spending data because it would dramatically reduce the manual processing involved. We have this team of I think it's 30 people who are who spend their time typing in the information and coding it. This will mean it gets straight into the the quality assurance um stage. They'll be checking the codes. They'll be making sure the data is is entered correctly. But um that'll be brilliant when that comes in and we're looking to implement that from April um 2027. And then the on the other side of the screen we've got the Blaze 5 uplift. Um so Blaze 5 for those of you who don't know is a software owned by or produced by Statistics Netherlands. Um and it's what our surveys are most of the surveys I think in ONS not all of them are hosted on. Um and the we're currently operating in Blaze 4.8 eight and it's a major major project to uplift the whole of the survey to the to the new system which will improve um functionality but also user experience. It will make things much nicer for our interviewers. And then data quality by by ensuring that the systems are kind of on the right platforms and we can um we can we'll be freer to make changes to to uh improve efficiency and um yeah and so that we are looking to implement from April 2028 and and there's a huge team of people working on this at the moment. Then we've got um developments we are undertaking on our processing systems. So the first of these is complete um uh towards the end of last year. It's beginning of this year. Um colleagues finished upgrading all the data delivery and processing pipelines um replacing legacy systems. So these are now all in Python um and they're much more robust, transparent and efficient. We know what we're dealing with. We can change it. we can um we can properly quality assure it. So um this will improve the quality and the timeliness and also reduce the risks of our uh processing systems. So again that was something that had to be done really before a lot of other developments could take place. The second two items on here we're aiming to implement um at a similar time to the Blaze 5 uplift. Um so we've got uh the income alignment and processing up upgrades. So this is look analyzing the feasibility of implementing statistical imputation and editing for income variables across all LCF outputs that use income data. Um and then on the expenditure side we're looking to develop pipelines to edit and impute spending data replacing manual case by case processes. Um and we'll be doing comparative analysis of the data derived from the different methods to better understand the impact. So um there are differences currently in how our uh some of our data that uh used to come from the SLC. We've got pipelines in place for for the LCF mostly our editing and our editing particularly is all done um manually still. So that's what that project is looking into but they haven't made any decisions on that. They're very much a kind of analysis stage um to see to see how it affects things. Then we move on to the section that I'm I'm uh most familiar with. So there are two projects uh related to updating the the content of the data collection. Um the first of this questionnaire optimization um has been uh broadly paused um for the time being because KOP has is the priority because we we're working to a very strict deadline for getting that in place to support the econ wider economic statistics transition. Um, but the questionnaire optimization work, which I'll be picking up uh again soon, is about reviewing the the LCF questionnaire as a whole and identifying priority areas for improvement and rationalization again about about getting building up the stability of the the survey. Um, similar, I think, to some of the stuff that Tanzy was talking about earlier on the WAS. Um but the real focus of the team over the past year has been koi cop which I'm going to talk through in more detail because this is one of the areas that as data users you will actually start to see changes in in the data you're receiving. So just to give a bit of a background uh I won't assume knowledge. I'm sure plenty of you know this already but KOP stands for classification of individual consumption according to purpose and it's a um UN statistics um classification framework. I like to think of it as a classification of anything that a household can spend money on. Um and uh they introduced a new version of the framework in 2018 which all countries internationally are gradually moving to um to move over to um some are ahead of us, some are some similar stage to us. But the key difference between the two um frameworks, there are some things that have moved around but also they've added an extra level of granularity. Koiop uh 99 which was the previous one had three different levels into which things were classified. Coop 18 adds another another level. So we need to be able to reflect that in in in our data. Um and it also updates the framework to to reflect modern spending patterns which if you think about everything that's changed since 1999 in terms of online spending and everything that entails is quite substantial. So the ONS the ONS transition will take several years but LCF being one of the key data sources for say national accounts uh is the first step in the process. That's why we've been working at a very tight deadline to to get that in. So our approach to the changes has been focusing on the priority of enabling collection of koi cop 18 while retaining the ability to deliver koi cop 99 outputs with minimal discontinuity where feasible. Um, so essentially we've designed all our changes. So this is changes to the questionnaire and the coding frame as much as possible to mean that we can classify the the data into both the old outputs and the new. And I'm using koi cop 99 here as a bit of shorthand. I'm talking about all of the outputs we produce. We're not only thinking because obviously plenty particularly family spending plenty of the breakdowns are not restricted to the to the koi cop classification. So as far as possible, we're looking to still be able to deliver those, but obviously you know minimal discontinuity where feasible. There are absolutely some areas where there where there will be discontinuities and where it's not feasible to maintain consistency. Um so we did an extensive review of the existing LCF coding frame um and the questionnaire to uh to enable this um and thankfully in many cases the LCF data is already sufficiently granular. We just need to map it to the new framework. So if you think about the data that the LCF collects on food, if you've looked at that, it goes to an incredible level of detail because of because of deafers needs really. Um and so in those cases, we simply say, okay, in quick 99 it goes here. In Quick of 18, it goes here. That's all fine. Um but we have needed to make some changes to capture extra detail and those changes have just been introduced in this month's questionnaire in July. Um so the the coding frame is is going live in um five days time I think once we start getting uh uh receipt data in um and our focus has been on areas where the koi cop has changed between 99 and 18. So we've identified through this process various other changes that we kind of want to make but we've been putting them all on a backlog to focus on as part of the questionnaire optimization project. I won't go into detail on these, but here's a a sort of summary of some of the areas where we've had questionnaire changes. And to explain the color coding, in the top row, we've got things that are more minor changes. Um, season tickets, for example, we had to split out the the the transport types in that into more categories. um and um add some follow-up questions on remittance fees um and uh add a kind of new classification in for the for the non-academic courses where previously we just asked someone what type of course it was now we kind of asked them to to tell us according to different categories. We then have our um bigger changes in the middle row there. Um, so, uh, one of the bigger changes in in the QuickOP classification is that under QuickOp 99, delivery fees were counted with the product that was being delivered. The cost all just went in one together. Um, but in QuickOp 18, it splits it out. So, delivery is its own separate category. So, we've had to make some changes to our to our to various questions. Same with the carpets actually on that one to make sure we're getting that that data split out. um holiday accommodation. They've added a a new split between like sort of communal types of accommodation like camping and youth host and things like that versus hotels and stuff. Um so there are some quite complicated changes in there because we already were were delivering to quite detailed level. We wanted to retain all that granularity. And then finally regular payments on there. So, we've got questions in uh a question in the survey about um I think it's probably called standing orders for those familiar with the variables, but uh about things that have been paid for by standing order or direct debit. We've just modernized that because it was given quite old-fashioned examples. And it also didn't make clear that we essentially wanted any kind of automated regular payment, whether that was a standing order or whether that was kind of some um payment through an online wallet or something. And then finally, in the bottom row, we've got new topics. So we haven't previously asked about these. District heating for those not familiar is a uh is is where heating is produced centrally and then hot water is piped to properties. So this is particularly in some new developments. Um and it it's it's projected to become a much bigger thing. So we thought we future proof we'll get it in now. So it'll be very interesting to see what data we get there. Um but that data for for the small number of properties who were on district heating we just wouldn't have been capturing their spending before. Um and then school transport that's about sort of payments made to councils for school buses etc. So there are a few new questions on that coding frame changes. So um we've made we've increased the coding frame from 500 to 700 codes. Um and uh yeah where where categories need splitting or rearranging and very occasionally where no LCF code currently exists. So that that's that's pretty rare. Um there's an example there at the bottom. you'll see that we don't uh we don't match the Koi cop language exactly because our coding frame needs to make sense to our coding team um so that they're able to do to deal with it quickly. So we the the KOP framework is written for an international audience and we kind of make it more relevant to the UK. The record of spending tool is the other thing we've had to make a change in. So there there are actually two different koi cop changes that require the record of spending changes to fully collect and they rely on this because the distinction it is dependent on the identity of where someone bought it which is quite unusual in terms of these these uh uh this data. Um so uh the the example here so cantens in koi cop team we need to split the data depending on whether it's a canteen in a workplace versus an educational establishment. So we've made a change here in our record of spending tool um to to move universities and colleges into school and we kind of um agreed with key stakeholders that the discontinuity was was acceptable. That wasn't key a key split for them. Um, the one that we haven't yet been able to implement, and this is pretty much the only place that we know of so far that we're not going to be able to deliver the full detail of Koiop is the restaurants and cafes where Koi cop splits by something called full or limited service, which is basically whether a restaurant has waiter service. Um, and that would have required structural changes to this tool, which given the time scales we were looking at were just not feasible. So we're now revisiting that again with all of our technical colleagues to see whether we can make those changes in future. But also we're going to need to give serious thought to respondent burden and whether it's likely to be put put people off actually reporting this spending um which would be quite problematic. So this is going to need going to need a lot a lot more thought in the interim. Um colleagues in national accounts who do need this split are just looking to model the the distinction. Everyone else well can model it if they want to but otherwise just get the data at the the higher level. So we'll still collect all the data. Um so the effect on outputs um in our initial data processing we've set it all up so that we're collecting the new inputs but we'll we'll continue reporting to Koi COP 99 and all the other distinctions that we've talked about already. Um as I've said the design minimizes discontinuities but there will be some changes in the data. You can't make a change to a question and not expect at least a tiny change in the data. But obviously some of our changes are more fundamental where we're taking delivery out of something and you know we can't just put it back in for various reasons. There are also some areas where we've modernized bits of the LCF that were quite outdated. So things like spending on um streaming um you know your Netflix and whatever um also regular charges which I've talked about already. So that's areas where there will be changes and there'll definitely be changes to improve the quality of the data because we're reflecting what people are actually spending their money on. Um and we will be producing documentation. My team are working on it as we speak on the changes um to support everyone in in understanding interpreting the data. So just two more slides for me. Firstly, the timeline for the koi cop work when you'll actually start to see it. So as I've said changes to the LCF questionnaire encoding frame have now gone live. This starts to feed into our kind of uh data processing from November 2026. Um and uh it's initially kind of colleagues in our national accounts team and prices who will be using this data um straight away. Um then um there are there are range of outputs kind of in the longer term. So from a from an external perspective, you won't see the koi cop 18based data until 2029 onwards um when the ONS economic statistics transition will begin and um exactly where family spending feeds into that is still yet to be confirmed. And finally then just to bring it back out again to the um LCF development project timeline. So you'll see on the left hand side we've got the projects uh that we've already delivered and then there's some target dates for the rest of it just to kind of give you give you the wider sense. So the receipt assistant tool and possibly some of the questionnaire optimization will happen in 2027. Um hopefully uh target date for Blaze 5 uplift the pipeline um and processing upgrades is 2028 as well as again more of the priority questionnaire optimization. And then in 2029 is when you'll start to see this actually hit the data that's that's coming out um to you uh including having a full year of COP 18 data available in the the FY20 29 data um and the the ONS overall statistics transition to COP 18 and that is everything. Thank you. [applause] So, um, you've just heard quite a bit more about the LCF. So, we're now going to return to hearing about some analysis using the LCF. So, um, a huge welcome to Simeon Yates, who's professor of digital culture in the department of communications and media and joint director of the digital media and society institute and co-lead of the institutional AI for life research frontier at the University of Liverpool. Um, and you're going to tell us a bit more about some of your uh, recent research. >> Over to you. Thank you very much for for having me here and also thank you very much to the living costs and food team for producing this data. Um so just I was saying to some other people earlier this is my third conference in three weeks where I'm speaking to not my usual audience. So be nice to me. Um so I work predominantly on the social impacts of digital technology with a significant focus on digital inequalities. Who's online? Who's offline? Who's got skills? Who hasn't? Who's benefiting? and who's kind of losing out. Um, I also sit on the government's digital inclusion action committee which has been formed this last year. We now have a digital inclusion policy in Britain. We haven't had one for over a decade. Um, and the digital inclusion action committee, the DAK as it's called, reports to DC, DCMS, DWP, Department for Education, Department of Health and Local Government. And so we're in the process of trying to drive we are an independent committee made up of non uh political appointees um who are trying to drive forward a evidence-based policy agenda. So why am I here talking about digital inclusion? Um I I added this slide after seeing this morning just to make a point about why I think this is very important. Um there is not one aspect of modern life that doesn't now require some kind of digital component. And in particular, there are many aspects of modern life that are now digital first. A colleague of mine has a much more complicated version of this slide with about 30 different types of behaviors which are kind of the functions we want out of life from healthcare to just enjoying ourselves and meeting our friends. And down the page are little boxes indicating things that are now digital. And if you are digitally excluded in some level, you start crossing those boxes off as an easy thing to do. And we might think that this just it's an issue to bust some myths. People think digital exclusion is an older people issue. It is not. There is a lifestage issue for older people in terms of usability, access, and also shifting into poverty. There are an awful lot of young people who are low-level digital users. Um, and second, it's not a rural problem anymore. Outside of a small part of Scotland where we're still rolling out broadband, it is predominantly an urban poverty issue. So bear those things in mind. So as part of the work we do, I constantly get asked about the challenge of being digitally included and often people immediately focus on equipment. I'm going to quickly and briefly jump away from equipment in a second, but when we're just talking about costs, some kind of headline figures, Offcom find in their affordability tracker that in any one month, British households, 24% of British households do something to afford their digital access. Now if you dig under those figures like I do what we actually find and this then is going to be part of the story is that 19% you know or of of that 24% the n the first 19% are people in the lowest income quintile who indicate they have done the following not paid a bill put it on the credit card not eaten food not bought other services and in particular households with children because keeping digital access is incredibly important for uh school. 90 odd% of school homework is now sat online, right? Um it also affects lots of people in very marginalized circumstances of modern slavery. We're doing a project at the moment on the intersection between food poverty and digital because people are moving to digital services to allow you to access your food bank, right? Or food banks are giving out free SIM cards as long with the food. So you know lots of intersections there. What this paper is about is trying to find a measure of what's the point at which start people start to have affordability issues and therefore what is the size of the population that the government has to think about in terms of addressing digital inclusion. This has now become quite important. You may not notice but two Mondays ago so not last Monday before the government published the green paper on the future of television. It is no longer a question of if, and I was involved in all of the pre-work on this. There's no longer a question of if the transmitters will be turned off. The question is whether it's 2034 or 2044. At that point, every household in Britain will need to be digitally connected in order to get BBC1. So suddenly now this is a British culture challenge, you know, civic engagement challenge, not as well as the the costs, etc. We have built a measure of digital inclusion that goes beyond connectivity and I saw a slide earlier mentioning the minimum income standard. We work with the minimum income standard team at use their methodology which is participating and engaging and going through 50 and 60 focus groups in multiple rounds to get a definition of what it means to be digitally included. The verbal definition there is the British public after two rounds of doing this. So we did it just for household with kids well all household types and then focused on household with kids and refreshed it for all household types in depth again all funded by the Nfield Foundation and that's the verbal definition of what it means to be digitally included. A minimum digital living standard includes having accessible internet adequate equipment and the skills and knowledge people need. It is about being able to communicate, connect and engage with opportunities safely and with confidence. I should be able to do that with my eyes closed these days but I always have to check. Um, note there's nothing in there about affordability. There's nothing in there about specifics. That's the broad definition. We then went back to the public and said, "What do you need to meet that?" They gave us three groups of things. Kit, bottom layer of the house. Um, functional skills, I can use the kit, I can turn it on, I can make use of it, etc. And then critical skills, you'll be able to use it safely and confidently. And that's an old graphic because the the chimney pot is missing. The chimney pot now says well-being because that came out of the second round. Um I won't drag this out, but therefore we've been able to measure both qualitatively and quantitatively with surveys whether or not households meet this standard. Um we've done households with children. We're waiting on the funding for doing all household types. households with children. 45% of UK households with kids don't meet the standards set by UK households with kids. And people think that's a big number. I remind you that 40% of single parent households are in poverty and 22% of households with kids are in poverty. And that nearly 50% of the UK population at work don't have essential digital skills for work. And if we're measuring access, basic skills, and critical skills, it's no wonder we're on 45%. And this slide and the next one could be a copy of all the slides we've had earlier this morning about the factors that drive this, which are poverty, which are education, which and so on. Pretty much all the same variables that were presented in in talks earlier this morning. And for a start look at it, the probability of meeting the minimum digital living standard effectively drops by index of multiple deprivation or socioeconomic grade here under NRS rather than NSE. So buried in all of this then is a question of affordability. So I was asked to go by government to go off and think about how would we measure affordability and I came across a living costs and food survey and after a lot of staring at um flowcharts um we decided to think about well what would be a good cutoff. Historically 10% of income marked you as being in heating poverty if you spend more than 10%. It's now a more complicated measure which is about um the quality of your household and whether you've been pushed into poverty or not. We took those we took the offcom figure and we we looked at the spend that the kind of lowest quintile were spending on on on digital and we came up with 3% is a good measure. Oops, I've gone two slides. that 3% of income seems to be a reasonable starting point to think about pushing into costs such that you're having having to do something with the other costs in your income as we saw in the previous slides in order to stay digitally connected. Um and so if we look at the old fuel poverty that was 10% of income when we look at the costs of digital for a standard average household 3% it's about a third of what they spend on heating. So 3% seems a reasonable number. Food poverty as we've just run a workshop on recently is really hard to look at. Um but as was just shown on one of the the previous slides um for those in the lowest income quartile a very large percent of of of spend is spent on food. So if you're eating into your already large spend on food by going over 3% that seemed a reasonable cut. And then when we looked at the offcom, like I said, the 19% that are doing what I think are hard changes. The other bit that makes up 24%, that's me. That's me setting my expensive house going, "Oh, which deal can I get best out of talk talk or whatever." If you look at the distribution, it's people at the lower end and then people at the upper end who are doing affordability things. I'm hunting the market for a very expensive deal to do very nice things and have gigabit internet. Everybody else is worrying about their kids access. So using that 3% figure I went into the LCF survey categorized as much as I could about that was dead spend on digital split across mobile and other services etc and looked at the spend on whatever. So we find that 55% of the lowest income quintile are spending more than 3% on their access. Um and as you see it just drops off dramatically as you go up. people in the top quintile are spending less than 1.1%. If we look at the um kind of adjusted for household size, we see that those spending more than 3% are in the lowest income groups and we look at those on benefits, those pending more than 3% are in the um highest um groups. And if we then think about well what does this mean in terms of policy? So we've got about 18% of UK households are spending more than 3% but some of them are wealthy but that's still 5 million households are spending a lot of money on whatever. If we look at people who only have a mobile right and are still paying more than 3% right they're sort of really low income and they're on very low incomes and their income their spend pushes them into official poverty. That's um 2% of UK households, but there still remains 10% of households that are simply mobile only and have an affordability challenge. We've mapped it because you guys put oak in the um data set which my friend Alex Singleton at Liverpool that I work with geographer loves. So we we use that in a lot of our survey work to help do mapping. Um and we find it's urban poverty is the major predictor of not meeting this. And for the policy call to to government, we think that there's half a million households who simply even if we do give them the lowest possible cost internet, so social tariffs which only people on benefits can have, they still simply can't afford digital access no matter what we do with the market. There's another one and and a bit million households who could possibly save money. Some of them um and some of them could in fact still get all of the services but they're currently offline or mobile only with a social tariff and still say below 3% but they'll still have to spend some more. Um that's sorry that's the 1 million in the middle and there's 1.7 million people out there just spending too much and could get a bit of deal. So we think there's three levels of policy intervention. one is information better market information and so the market can do lift that kind of 1.7 million there's something around the group that's in the have to spend more category and especially with the TV transition that's really important but there is a group that all of the current policy mechanisms do not work and we will have to think about some other more radical policy interventions to support that half a million households online especially with the TV transition position coming. And so some final points, the 3% cut off seems to work as a good measure. Um the really low low income low internet group, we called it that and then we went and found that the new housing heating thing is also called Lily. Um so there we go. Just for confusion sake um identifies the really at risk groups and most of those are young people. Young people who are in employment age but not at home. are often in that group. So, not older people. A lot of older people and HUK that I work with don't like me saying this too much with the triple lock are well inside the affordability bracket. One of the challenges is that that group of people that was pointed out earlier in the day who could claim pension credit but don't. If they were claiming pension credit, they would be better off and therefore able probably to keep the costs under 3%. But they still might have to pay more. So we still might have to do something. But one of the policy interventions could be making sure they do. Um and the important point which I skimmed over at the start, this is not just about access. This just solves the making sure people can afford the devices thing. It doesn't solve the skills problem. Skills problem is massive. Another measure we use shows that about 40% of the UK population is safely online. There's another 30 odd percent who are missing skills or affordability or other challenges but stay online and then the rest of everybody else is a very low liimited digital user or intermittently online and all sorts of other circumstances. So this is only one bit of the policy kind of problem. That's me finished. I tried to keep to time. [applause]