 Hi and welcome back to the Family Finance Survey's user conference 2023. We're in our fifth and final session now where we're going to be hearing from some researchers on poverty and benefits. So starting us off is going to be Mary Alice Doyle. So if you'd like to start sharing your slides while I do the intro that would be great. So Mary Alice is going to talk to us about the policy and the reality of effective marginal tax rates and if there's a difference. Mary Alice leads the research team at policy and practice and is a PhD student at the London School of Economics. Her research focuses the interaction of benefit policy with labour market and health outcomes and as previously if you do have any questions if you could pop those in the chat and start off with the word question so I know what I'm looking for. Thank you very much over to you Mary Alice. Hi everyone. Thanks for taking around to the last session. So I'm presenting today on effective marginal tax rates or I'll explain what that is in a minute but really thinking about work incentives and looking at the difference between hypothetical cases where you might look at incentives for a given type of household versus what we actually see when we look at the FRS data. So let me get into it. So this is a summary of what I'm going to do today. Just give you some background and then get into my estimates of effective marginal tax rates based on the policy and then based on the reality of actual benefit take-up rates. Okay so effective marginal tax rates are really trying to answer the question of if I earn an extra pound how much of that pound do I keep? And so if we think about you know someone who might work an extra shift or an extra hour in a given week then they'll have part of their income taken through tax as you'd expect and then part of it taken through the benefit taper if they're receiving benefits and they'll keep a portion of that. So in this example that I'll walk you through in a minute we're talking about someone earning an extra pound and keeping 30p of that pound. So EMTRs are based on a range of different inputs instead of looking just at the tax system we bring in the tax and the benefit system. So looking at income tax, national insurance, the benefit taper and child benefit withdrawal and what I'm going to talk to you about today is also including other types of benefits that we don't usually see in estimates of EMTRs. So pass putter benefits like free school meals or healthy start and social tariffs like discounted water and broadband for low-income households. For example this pie graph on the side shows you for a single person who's earning £20,000 a year in claiming Universal Credit then their EMTR is going to be 70% and just to sort of clarify that's different from their average tax rate. So the average amount that they're paying in tax as a share of that £20,000 which if you include the total amount of Universal Credit that's tapered away at that level of income then it's 40% so average tax rate of 40% but an EMTR of 70%. So why do we care about EMTRs? Well we can think about primarily about work incentives. It's a separate empirical question as to how much EMTRs actually affect work incentives. I'm not going to get into that with this presentation but it is kind of a key input to work incentives to think about how much you are going to keep of the extra money that you earn and we can also just think about you know generally about fairness and I'm going to get into that but just sort of those are reasons why you might care about EMTRs. So what do we already know about effective marginal tax rates? Well we know that they're really high in the UK for some households so there's been several reports including one from Policy and Practice one from IFS and others that we've seen that look at EMTRs of 90% or more for some types of households so that means that you earn an extra pound and you keep 10p or less of that pound and that's because of the interaction of taxes the UC taper and the child benefit withdrawal that happens particularly at the the earnings window of 50 to 60k a year which is when the child benefit gets being withdrawn. So this has been described by researchers as a disincentive desert so a range of income which it really makes no sense financially for you to work a marginal amount more. We also know that people who receive certain benefits are then passported into other benefits so you know free school meals and healthy start being two of the the key ones that also different councils will have council tax support schemes and support for you know parents to pay for uniforms and school transport that is all based on benefits that you're already receiving so if you're getting universal credit or other similar benefits and those likely are going to increase effective marginal tax rates because it means that you get more of the benefits taken away when you when you sort of leave eligibility for the passported benefit and then we can we also know that a lot of benefits are not actually claimant people who are eligible for them and that's going to decrease effective marginal tax rates because it means that you haven't got that additional disincentive of having your benefits taken away because you're not receiving the benefits. So that's what we know but what we don't know is how many households are actually in this disincentive desert of having really high effective marginal tax rates. How much do passported benefits affect effective marginal tax rates and how does benefit take up effective effective marginal tax rates? So what I'm showing you today is my first pass at trying to answer these questions. So what I'm doing essentially is combining the FRS survey data with our benefits calculator. So we need representative data on distribution of household income for that and using the FRS. I use the 2020-21 data and looking at it at the benefit unit level. And then we also need information on actual benefit receipt which is what we have in the survey versus benefit eligibility for households who might be eligible benefits but not taking them up. And for that I'm using policy and practice as benefit calculator. We've recently developed a API which means you can interact with the calculator without having to you know go to the website and then click through and enter the data manually. You can just send a whole a whole table of data of FRS respondents for example and then send to the calculator and then that will calculate for all of those households their benefits and eligibility and all the details broken down into universal credit legacy benefits local benefits and other details of that even the elements of universal credit. And we also need eligibility for a lot of different types of benefits and not just you know the the key income replacement benefits. So as I said we've got all this detail in our benefits calculator. Okay so what does that tell us first off on benefit take-up which is a kind of a bi-product of this analysis of trying to understand how benefit take-up affects effective marginal tax rates. So this is my graph that's showing on the x-axis we've got household earnings in 10k bands that's annual earnings so if it's a two adult household that's going to be both adults earnings within these bands and then we've got the number of benefit units who are in each of these categories and we've got the number who are eligible for means test of benefits that's the orange bars and the green bars are the number who are actually claiming those benefits. So you can see that most people who are eligible do claim benefits but particularly at the lower end of the income distribution there are a lot of households who appear to be eligible but are not claiming. So based on this estimate it suggests that 13% of working age households are eligible for benefits but not claiming them. And then I think a question that seems interesting after that is okay so these households who are not claiming benefits how much could they be getting do they have a good reason for not claiming them in that they might not get very much if they did bother with making a claim. So to try and get at that I looked at the average value of a potential benefit claim for households who are working age and would be eligible for universal credit or equivalent benefits and I find that households that are already claiming benefits they're eligible for about one and a half thousand pounds a month less if the household hasn't got any children but then for households who are not claiming but eligible the amount that they could claim is still substantial on average but it's quite a lot less than those who are claiming. So part of the story doesn't appear to be that there's some households who are eligible benefits but not claiming because it's not worth it for them although the fact that these numbers are still quite large suggests that that's certainly not the full story but that's sort of an aside to try to to use this information to calculate EMTIs. So the other key element of EMTIs that guest talked about a lot is the child benefit withdrawal and so this graph has a lot of detail on it that I'll try and talk you through to to understand what's happening there in terms of like how big of a problem is this aggregate. So here we've got the number of households who are taking up child benefit by income band and then we've got the the share the green lines who are not taking it up and you can see that you know there's some households at all income levels but particularly at the high income levels household not taking it up which makes sense because they're not eligible for it most of the time and then we also see that once you hit households that have 50,000 pounds or more so the the eligibility is based on one earner earning 50,000 pounds or more but we do have households who are increasingly taking up child benefit but having that withdrawal so that's showing us that there are you know certainly a large number of households but not not in terms of a proportion of households who are receiving child benefit who are affected by the withdrawal. Anyway so let's get on to talking about the EMTR estimates. So this graph is showing you the breakdown of effective marginal tax rates by income band of the household so it's going to be not based on the individual who's earning the income but based on both there and if they have a partner that person's income as well that leads them into being in any of these bands. So you can see that so I've done this separately for the primary earner and secondary earner in the FRS data and you can see that as you might expect for the lowest levels of income the only thing that affects EMTR is the UC taper that's because people who are earning as well incomes aren't going to be paying any tax and then increasing as you get to higher incomes it's more income tax and the UC taper is falling away and we see a similar picture with slightly lower levels of the EMTRs for a secondary earner in the data. So that's showing you what contributes to EMTRs and how it varies at different income levels and I think the other interesting point to take away here is that these EMTRs like none of them are 80 90% or more so on average we're not seeing very high EMTRs but I'll get to that in a minute in terms of thinking about who is it who's paying these high EMTRs and then this graph is these graphs are showing you actual EMTRs versus those potential ones that I showed you before so sorry I should have said these ones I just showed you that's based on assuming that households have full take up of benefits that they're eligible for whereas here I'm showing you the comparison between what it would look like based on full take up versus actual take up and so here you can see the contribution that benefit take up makes to EMTRs and it's showing us that particularly for low income households the EMTRs are actually lower than they should be based on the policies that apply to them because they're not taking up benefits that they're eligible for and if I haven't got those benefits being tapered away. Okay and then the question of which households are actually facing high EMTRs that given we don't see on average households having very high EMTRs so then I split the data into groups of ranges of EMTRs and let's just focus on households that are having EMTRs of 80% or more there's about 950,000 of those households according to the 2020 FRS and those households generally have they have children they have earnings that are less than the median income and all the receiving needs as a benefit which of course makes sense most are receiving child benefit but it seems like none of them are receiving passported benefits so it's sort of these households who you know there's a large chunk of them but they are a small portion of the total number of households affected by EMTRs. Okay that's what we see on EMTRs but then I think the other interesting thing is so I tried to include free school meals and other passported benefits into the EMTRs but then we just get you know negative 1000% EMTR at the threshold at which you lose eligibility for those types of benefits because it usually is an income threshold and so instead here I'm showing you the for free school meals and healthy start these graphs are showing you on the left axis the number of households at each income level that are sorry the average value of the benefits for households who are at each income level and then the number of households is the the green lines which is the right hand axis and so you can see that it's really a very small number of households who are just sort of at the threshold of not being eligible for either of these benefits which is around 8k free school meals and 4k for healthy start and but for those households it is you know a huge fall in in your the value of your benefits and income if you do pass that threshold so just to think about a worth example here with free school meals so free school meals are worth on average around 460 pounds a year per child that's just an estimate for two children that would be 920 pounds a year so say that a parent is moving from not working into working two days a week at minimum wage and that would mean that they lose eligibility for free school meals they then need to earn an extra 2200 pounds a year to regain that value that they'd lost on free school meals and that's because of the combination of EMTRs and this passporting of benefits so because the UC TAPA and lost eligibility of council tax support mean that even though the household doesn't pay tax or national insurance they keep only 42% of their earnings which means that they need to work a whole lot more to earn back the value of free school meals so it would be on average an extra half a day or week at minimum wage that you need to work to earn back that value okay a lot from from what I found so in practice I found that average EMTRs aren't as high as we might expect but they are high for some households and I estimate that there's about 950,000 households that have EMTRs of over 80% for at least one of the earners in their household their largely households with children earning median income or less and receiving means tested benefits and also as it's not a surprise but just sort of confirms that millions of households are not taking out benefits of that that they could although they have lower average claims than those who are taking them up and the households with lower incomes those cliff edges in support interact with the EMTRs mean that they lose thousands of pounds worth benefits from working more and that comes through free school meals and other sort of passported benefits of course that's not in practice an issue at the moment because transitional protection that means that households that eligible free school meals don't move their legibility but it will be an issue when that traditional protection ends in 2025 and then there's a whole lot of limitations the type of analysis that I'll just sort of touch on so one is that we're assuming a static income for households and of course lots of low income households don't have static incomes so their EMTRs likely to vary across the year and their total income is perhaps not not well summarized as saying you know your last month's income at the time of the survey is going to be one-twelfth of your annual income so that's just a limitation of looking at any kind of survey data like this and then we also I haven't taken into account here households that don't claim benefits for good reasons like not being eligible for clean benefits even if they meet criteria of having low enough income and assets and of course our households that haven't reduced public funds will fall into that category so that's a next step to make sure that I do take it into account with my estimates of eligibility. In terms of other next steps I want to estimate participation tax rates as well as EMTR so not just earning an extra pound but if you you know take on an extra shift or start working part-time then what does that mean for your marginal tax rates and I want to include local benefits so particularly council tax reduction and social tariffs which are different depending on which utility you're with and see what that means for EMTRs and then do simulations of household earning an extra £100,000, £500,000 a year to see what that means for work incentives and again we can also do simulations on other metrics like looking at removal of child benefit withdrawal, lower UC taper rates etc we can do all of that in our API to then look at what that means for EMTRs for different types of households but this is a work in progress so any feedback is very welcome as I'm trying to figure out which of these next steps to focus on and what's most useful for for anyone who you know thinks about these positions as well so thanks very much looking forward to your questions. Wolf is a researcher at Loughborough University Centre for Research in Social Policy. He's worked on living standards and livelihoods in both the global north and global south exploring datasets and issues of resource access in a wide range of contexts. Thank you very much I'll pass over to you. Hi so yeah thank you very much everyone for sticking around until the end and yeah thanks to all the presenters earlier for some really interesting presentations and so yeah this presentation stems from my PhD research completed about a year ago very much looking at the limitations of what existing finance survey data in the UK can tell us and very much valuing the insights we can get from it but aware of where it isn't necessarily fully representing the circumstances of people particularly at sort of the bottom end of the income distribution so yeah I'll be talking through first of all a little bit more of that background context and I'll summarize the methods looking particularly at income sources and costs which are missing from surveys at the moment and then touching briefly on some issues to do with subjective food insecurity units of analysis and some of the implications of using longer reference periods for collecting the data as well. So to begin with sort of building on what Peter told us earlier there have been sort of large increases in the depth of poverty and one of the sort of indicators that's particularly come to sort of symbolize a lot of that is the large increase in trust-to-trust food bank parcels over the last sort of 15 years or so starting out at just over yeah around 26,000 parcels a year in 2009 up to three million as of the latest figures and these are sort of only indicative of a much wider increase in food aid as well and all of this has very much demonstrated the insufficiency of many people's formal incomes and we know that inadequate and unreliable financial resources are very much the key drivers of food aid usage and indeed much wider food insecurity and survey datasets like family resources survey very understandably focus on sort of formal regular income sources like wages and formal benefits and housing costs but beyond that a lot of the more piecemeal income streams or less regular costs that low income families might face are often overlooked and it's great that as of this year the family resources survey does include more details of food bank usage but this still is relatively top level in that it's great to know that people are going to food banks or not but it doesn't really quantify their usage of the services in great detail and then there are also issues around units of analysis in that in FRS and by extension HBAI the grouping of people is very much based on sharing and address and cooking facilities and without any more explicit information on resource sharing which is sort of a very reductive interpretation of the idea of sort of a household as being a group that shares a common pot and so it's kind of sharing the pot but not necessarily what's in the pot and then finally among other issues household residents can only be either fully present or fully absent without sort of more of the sort of nuances of people maybe being part time in different households or sort of temporarily resident and the social metrics commission measure which was developed sort of roughly the same time as I was doing the fieldwork for this study and does sort of very much go part of the way to addressing these issues but there are some emissions from that measure still that are potentially quite important to people on low incomes and in many cases that's just because perhaps the data isn't there to accommodate those aspects and so the way that I approached this study very much grew out of the use of household economy methods for researching incomes and broader livelihoods of people in developing countries in the global south where in many cases a lot of the core components of people's incomes overlap very much with incomes in the global north but it's particularly striking in some cases how you need to expand that sort of fairly that more standard formal framework and particularly where if you're thinking about rural areas of sub-Saharan Africa for example and the need to also look at food income sources and so for example if you're trying to predict a famine what happens to people's own crops and how that impacts on their food consumption is absolutely critical but then also there are a wider range of incomes and costs that also and sort of different dynamics involving them that also need to be taken into consideration so for example sort of season the seasonality of different income sources so like what people might do when they're not working in the fields on crops or if people are migrating for work for some of the year but then returning home for other periods and then this longer term period also allows for a bit more perspective and nuance than if you're just looking at a very short snapshot because you can look at how these things might change over time or else how they sort of accumulate to affect people's net income comes over that longer term period but I did also integrate these approaches with lots of aspects of the family resources survey partly to acknowledge how income from employment or benefits in the UK are often in many other regards more complex than you might have in some of these societies and then also just being very aware of trying to avoid reinventing the wheel where it wasn't necessary and being able to sort of simulate how the FRS might model the incomes that I was recording as well and then I also used minimum income standard data as part of the research partly as a sort of resource adequacy benchmark and Peter's already given a bit of an introduction to this but also as an implicit price for the food so you collect lots of very detailed information about what people are receiving but then you can give it a price based on the sort of implied price from the minimum income standard food baskets and then there are also various qualitative aspects that I also explored to put these incomes and how they related to people's overall livelihood strategies in a bit more context and the interviews were all conducted with people recruited in London in the borough of Southwark and recruited at one trustful trust food bank which is very much the sort of dominant model of food banking and they're certainly the highest profile and but then also two separate community meal services to also look a bit beyond just the food bank and the limitation main limitations of this research are partly that when looking at sort of wider issues of food insecurity we know that the majority of people who are food insecure don't actually use food banks for a variety of reasons around inaccessibility or stigma but yes so this this this data just relates to people who are going to food banks and then also of course it was just one location and in London and we know that people in London tend to have higher housing costs but this was sort of mitigated for my interviews by virtue of the fact that most of them couldn't afford to be renting privately or paying mortgages and so the vast majority of people were either in social housing or various different forms of homelessness and so turning now to the findings of the research this chart shows how HBAI might represent the net incomes after housing costs of these interviewees and so we can see that based on the sort of dotted line thresholds that over 80 percent of them would be below relative income poverty measures but there is a group whose incomes appear to be slightly above it which might look quite puzzling particularly at first glance but then to sort of unpick some of these a bit more to turn that orange line into a column chart you can then add in all the various different dimensions that are overlooked by HBAI and FRS and so we can see that there's a lot of extra income sources that sort of come in for the families at the bottom end of the distribution on the left and so income from social transfers in particular in blue and then in-kind incomes such as food aid usage and then also loans for this household here which help to sort of just about keep these families heads above water and then while also impacting on the family's resources at the upper end of the distribution as well but then also at the upper end there tend to be more additional costs which aren't necessarily being reflected in FRS so a lot of debts in the brown costs from disability benefits or costs the disability benefits would cover costs of childcare, costs of legal costs and then also costs of education for some people who are often sort of experiencing relatively temporary hardships so that they can do a vocational course of some description as part of a career change but very much experiencing a lot of deprivation as a result of that and then sort of looking at the cumulative impacts of all of these different resources and costs it has a very significant impact on what people's overall income looks like so overall the immediate median change is 40% of the original net income and in some cases that's sort of unimaginably bigger because of how close those incomes are to zero and so this sort of gives us yeah it raises a lot of questions for how exactly we should interpret some of these people's circumstances if the people might be considerably poorer than they would otherwise appear to be in particular and the one sort of exception is the richest household in the distribution where his income would still be very high but that is primarily because of receiving recent support from a relative and without which it would have fallen down much lower and the impact of this on the distribution is to create a slightly narrower range of net incomes but the overall impacts are slightly muted by the canceling out of sort of individual level increases and decreases but that individual level sort of changeability sort of shows how much is going on under the surface still and so turning to the implications of some of these emissions and many of the informal incomes in particular charity food aid are actually quite unsuitable for inclusion in poverty measurement in the UK and so I'm very much not suggesting that this should be included in HBAI poverty measures in particular because things like having to recourse having to resort to turning to a food bank is very much a symptom of acute poverty and often represents exclusion from mainstream consumption practices. It can't sort of really meet people's needs in food security terms because of how current definitions of food security emphasize the need to obtain food that's consistent with people's requirements and preferences and social acceptability but where these details can be very valuable still is for other purposes in particular helping to explain what might otherwise be considered as implausibly low formal incomes and also to monitor these sort of coping strategies that people are having to resort to and so for example one of the third poorest family in the study their money income or they didn't have any money income at all which might be considered fairly suspect perhaps if you were just looking at the net income in itself but we can see by looking at the food income that actually they were sort of just about surviving solely through meeting all of their food needs in this case through food banks, community meals and also food that was provided in their hostel but this is quite an extreme case because the median total food income was only about a fifth of the food budget so the median family in this study would still have had to meet four fifths of their food energy needs with purchases or probably more likely cutting back on food and we know from lots of other studies but also some of the qualitative aspects of this study that people really don't want to be having to turn to food aids to meet their needs so for example in order to reach that kind of level of food income people had to be constantly moving from place to place trying to find food often queuing up for a long time at different services where perhaps there might not be any food left by the time they got there or else there wouldn't be food that they wanted or they might be having to sort of scavenge from sort of leftover sandwiches or whatever at bus stops so this really isn't something that we should be taking into consideration when thinking about whether people are reaching poverty thresholds or not but on the other hand the missing costs are perhaps a bit more useful in poverty terms in the sense that these additional costs can really help to explain things phenomena like food bank usage and other experiences of deprivation and it would give us a much better sense of what people's levels of food access or ability to meet other needs actually are as well as making clearer relationships between net income and deprivation and some of these costs are similar to the social metrics commission recommendations in particular the debt payments and the indications of costs from disability benefits but this study also found that there were quite significant impacts from other areas of costs such as extra health needs not not covered by disability benefits and for example we know well there are there are a lot of interviewees for example who who did have quite significant health needs but these but they weren't even receiving disability benefits and others who thought that the disability benefits weren't meeting their needs and then also other costs that might be involved with education home moving or homelessness or legal fees which might encompass things like immigration fees or divorces or things like that and so if we if we look only at the costs the extra costs in the green and what as well as a few other perhaps less contentious income sources like perhaps loan income to balance out the debt payments we end up with only a few people very narrowly above the 60% of median income poverty threshold and even deeper poverty than otherwise was indicated so now I'll try I know that there's not all great amount of time so I'll try to whiz through these quickly and the last slides but the findings also sort of revealed some of the limitations of how of the established subjective measures of subjective food insecurity in that it was clear that by not explicitly covering the acceptability of food that people were receiving it seemed that people could then be sort of judged as food insecure which in ways that seemed a bit perverse and the fact that in FRS and HBAI there's not an extra level between the household which is just the people at the address who share cooking facilities and a more sort of nuclear family benefit unit and this groups together people who might be lodging or sofa surfing or in other shared accommodation as mentioned in the presentation first thing today about sort of financial cognitive models and so they're for people who really wouldn't be sharing any resources necessarily apart from perhaps a cooker in ways that could increase the apparent poverty and so for this study I didn't collect any details of the incomes of people who people were living with who they weren't related to in this sort of familial way but I could sort of look at what the impacts might have been if you for example separated out people who are in the same in different benefit units within the same family unit which tended to be the case where there was an adult child living with their parents and so for example in the in the example here to the right of the chart the sort of the biggest orange bar that's £336 a week was the combined income from both of these people but if you had separated out it would be very clear that this was two sort of different standards of living which in this case would be misleading because they were sharing resources but if they weren't sharing resources and both of them seem to have a higher income that would be slightly misleading. Hi Walt I'm afraid we're running out of time now. Oh okay no problem um but yes so yeah I can I can finish there and be happy to take any any questions.