 Okay, so it's quarter two so I think we'll make a start. Hello and welcome back to the fourth session of today's LFS and APS user conference. You are in session 4A, just in case I just wanted to check that you're in the right session for the presentations you wanted to join this afternoon. If you wanted to join session 4B then the link for that is in the chat. So I've got the pleasure of introducing two presentations in this session today. Our first one up is from David Owen from the University of Warwick. David is a principal research fellow at the Institute of Employment Research and a geographer and his research interests are in using large quantitative data sources such as the LFS and census to explore labour market differences between socioeconomic and ethnic groups and his talk today is about the armed forces veterans in the UK labour market. So thank you very much David and if you're ready I can hand over to you. Okay, thank you Martina. I'll just try sharing my screen now. Yes, that seems to be working. Yeah, we can see that. Thank you. Okay, right, thank you. Sorry the bit misleading on my screen. Okay, yes, thank you. As I said I'm going to talk about data on the armed forces veterans in the UK labour market and this is drawing upon data from the annual population survey accessed by the secure data, the INS Secure Research Service. The work is undertaken as part of a research project funded by the Forces in Mind Trust which is an armed forces charity looking into longer-term outcomes of employment for veterans, for ex-UK forces, the veterans in the UK. This part of the research refers to the quantitative part of the project. Most of the research is undertaken through interviews as an online survey kind of qualitative semi-structured interviews and focus groups because there isn't a lot of quantitative data on armed forces veterans. This is despite the fact that the veterans have been of high profile recently falling to long major conflicts in the Middle East and Afghanistan and Iraq and other conflicts. And the government has made commitments in the UK government and the Scottish government has made commitments of the veterans to their fair treatment and to help them into employment. But one of the problems with this is that in order to make such commitments you need data to monitor the lives of the careers of veterans and this is something which is lacking. The career transition, when people leave the services they're invited to contact the career transition partnership, to guide them into civilian employment. So the career transition partnership gathers data on the initial part of the initial move into civilian employment and this shows that veterans are a very high rate of economic activity and employment initially. They collect data up to six months after leaving the forces and show that veterans are 90 percent are economically active at that point and 84 percent are in employment at that point six months later. The annual population survey as far as I can tell is the only major UK survey which contains a question on veterans but this was only present in the survey between 2014 and the January-December 2014 and January-December 2018 and it was funded. This was funded by the Ministry of Defence. It's the most recent source of data on veterans. We've got to go through a percent or series of descriptive tables and charts which contrasts labour market participation by type of veteran and their occupational industrial employment patterns looking at the demographic variations between veterans and we're also going to cover look at the deficiencies of labour market data for veterans. So what sources of data exist from which veteran employment can be monitored? Well the most regular source of employment data in the UK and the GB and Great Britain are the business register and employment survey, UK business counts, annual survey, now known as none of these record veteran status. The labour force survey and annual population so the main source is a regular data on the demographic characteristics of people working in the workforce. Unfortunately the none of the data is no data from the ALFS and APS include data on none of the published data or unused license data from the ALFS and APS include data on veterans. Understanding society is the main source of longitudinal data. It too does not include a veteran question. Veterans can potentially be identified using data from the secure versions of the APS and the ALFS by using the most detailed versions of the previous industry and occupation identify identifying those people who had previously worked in the armed forces and in armed forces occupations and tracing them in you know through the waves or in using the questions on occupation the one year ago or in last occupation but this is very difficult because of a small number small sample sizes sizes. The other sources of possible sources of data or sensors of population it does the previous sensors have asked about current employment or currently serving in the armed forces these can that question can be used in the longitudinal study to trace the subsequent employment and labour market history of people who were serving in the forces and some data is available from commission ad hoc commission questions which show that there is some of the same patterns which will be revealed later on in this paper. The service the service leaders data base which collects data on all personnel of left the UK armed forces has been linked with the 2011 sensors of population and this has shown that the veterans have a very high economic activity and employment rates in 2011 confirming the CTP statistics but the problem is that there's no source of more recent data no source of regular data the annual population survey included the questions on veteran status because in between 2014 and 2018 because the Ministry of Defence funded the series of questions which were used to produce a report on veterans until up to 2018 which has been withdrawn that one was just one second after I'm sorry somebody at the door no very sorry no problem yes these are the questions which are in the annual population survey veterans status best served which are ever served in the branch of the armed forces which branch are currently in the armed forces or in the reserves and then a question which measures records the the year in which a person left the armed forces in this analysis I'm focusing now the Ministry of Defence report was focusing on the whole what the series of annual Ministry of Defence reports do produce employment data data on the characteristics of the veteran population but they come from the health branch of the MOD and their focus and their focus is on the whole veteran population the whole veteran population is elderly and they and there are over two million veterans most of whom are over the age over his time of age and clearly then the in terms of welfare the the needs are more towards health directed to all health rather than towards employment this the the difference of this paper as looking at the research project looking at employment history experience long-term employment experience of veterans therefore focusing on veterans age 18 to 64 assuming that very few veterans would be aged under 18 and so the that that population is about three quarters of a million over the and this this as I say this is over the period 2016 to 2018 the tables and figures present here use a three year APS for 2016 to 18 in order to get large enough sample sizes to analyze rather than looking at the annual APS the so there are about 13,000 people leave the forces every quarter at the moment over the period 2016 to 18 and there's about some of them three and a quarter of a million veterans in the population within that population as the chart shows that the veteran population is is skewed towards the older age groups which over half are being aged over 50 numbers declining in the success of the younger age groups most veterans tend to leave in the 30s and 40s about an eighth of veterans 13.3 percent are female reflecting the gender breakdown and forces the whole female veterans tend to be younger than male veterans geographical distribution of veterans veterans represent just under two percent of old people aged 18 to 64 living in Great Britain the largest number of veterans living in the south west south east and southwest of England but the veterans share of the population highest in the southwest and smallest by far in London London's it's it's um these numbers tend to reflect the pattern pattern in 2011 from the 2011 census which was data which was linked with the service leaders database so there's not there's no real change in geographical distribution over that period overall economic activity so veterans 82 percent of veterans were economically active in 2011 in 2016 to 2018 83.2 percent were were economically active so more again more stability over the period the sensitive economy active was high it was it's high the the over 90 percent look in the right hand axis of this chart for for most age groups in in the prime economic active age of the younger age group was lower age group lower percentage of economic active because of participation education but then when you reach over 50 economic and of the more nearer to retirement age economic activity falls off quite sharply unemployment rates are low 3.3 percent of the highest for those aged 18 to 18 to 24 the economic activity rate increases with age and uh near the very nearer a veteran is to you know fit for a time so they hire the economic the high high economic activity rates um through a differential by gender in terms in terms of employment rates uh men are more likely to be more likely to be employed than women the employment rates for men increased between age 24 and 25 29 year age groups and remained about 90 percent until the 40 45 year age group the afterwards falling sharply women's employment rate the small numbers mean that it's 100 percent for the younger stage group um then fall uh until and then rise again in the 40 in for female veterans in the fourth in their 40s the male female male male female differential was widest for veterans in the 30s and in the the 60s um the employment rate for women falling off more quickly than that for men from refined reflecting a higher the short low at a younger a younger rate average rate age of retirement um women were more likely to be employed than men if although in the 45 to 49 year old age group um educational high educational qualifications has a high as a very high strong impact on economic activity as you can see here the economic activity rate declines as the highest educational qualification declines um so from 90 90 percent for those with degree or equivalent qualifications down to around 60 percent for those with no qualifications um uh as a concomitant become the economic inactivity rate increases as the highest level of a uh of of educational qualification falls most most markedly between those between uh GCSE and uh other qualifications those in no qualifications again the the unemployment rate is also highest for those with the poorest educational qualifications the employment rates um again show a very stark contrast between uh by by educational qualification and also the the gradients in terms of employment rates is is steeper for women than for men in terms of as the level of highest educational qualification declines um so the the differential in employment rates between men and women at highest for those with the the poorest educational qualifications um in terms of regional variations employment rates are highest for veterans the highest in southeastern in eastern england and london for the lowest employment rates in wales and scotland the economic activity rate shows a less clear pattern but is the highest highest in london where the the veteran population is smallest um and also in the neighboring areas of the parts of the southeast there's not a lot of less variation between regions outside of other than this southeast corner of britain so the relative to the rest of britain uh contrast the economic inactivity was lower in the south also lower in the south and east of england than in the rest of great britain uh turning out to disability disability plainly uh of major reasons for leaving the forces can be so being just uh you know invalided out um and so and as we said earlier in the conference that uh the disability there are disabled people experience lower rates of employment in general than people on the able-bodied so um but the same is apparent for veterans disabled veterans the whole are much less likely than non-disabled to be working there's quite a large differential so that's over 34 percent 34 percent difference in the probability of being in in work that the percentage of disabled veterans employed it also decreases as their their age increases um but then until the age of 40 um so yeah it does with age um um female veterans are slightly less likely to be uh so slightly more likely to be employed employed than than men um and the impact of disability on the probability of being work is higher for the the army um than for the other forces so work the veterans do um the strong gender differential the the largest uh the uh sector and districters for men were manufacturing transport and kind of sections uh for men were manufacturing transport and storage construction and public administration defense well for veterans for email veterans this is very different nearly a third work in say health and social work uh followed by public administration defense education and wholesale retail wholesale and retailing and repair of vehicles so a strong um tendency to occupation the soccer major groups uh ranked by by percentage of all veterans so the largest of sub major groups are transports and drivers and that that's the that's really refers to men um the um and there's probably employed other ranks uh and the other the the next largest is corporate managers and directors who probably tend to employ officers so unfortunately we're not able to see the effective rank on previous rank on employment in the APS because rank isn't captured um so the the other large occupations for business and public sector associate professionals and elementary administration and service occupations uh for men skill men metal and electrical electronic trades and science engineering and technical professionals for for women administration health and caring personal service and major major sources of occupation um younger veterans are more likely to be in elementary administration and service occupations and skill metal electrical electronic and trades and older veterans much more likely to be as working as transport and drivers or corporate managers and directors occupations um I just wanted to conclude in my talk so the amount of data on veterans is limited um there's a prospect a much better um a much greater rate by elizio data because the 2021 sense 2021 census for england wales and 2022 census of scotland both include a question on veteran status uh we'll just move to the next slide to show that so this is the question that just who previously served in the armed forces yes uh though or no and whether you're currently or previously served in the reserves reserve forces um the prospect of a flexible table builder there's a potential to create detailed tables for veterans for small areas um and for a whole range of dimensions available in the census and there's also a prospect that uh with the data in the census of the longitudinal studies could produce very detailed time the time the information on transitions over time unfortunately the question is not about not asked in north nile and because the census test just revealed that actually asking that question would represent a serious threat to the viability of the census as a whole because of the level of opposition resistance to the question so we'd only be days with jerry britain um and the other another problem is that the census does not ask when the veteran left the armed forces the branches in which they serve or again or whether what rank they did though that can identify that from occupation question but you cannot break the other ranks into into you know a non-commissioned officers and ranks which you'll see is important so the lns does plan plan to link the census with administrative data at least in the same way as they did with 2011 to produce more up-to-date and regular data on the labour market it strikes me though that the best way of producing regular data on the veterans in the future is to include reinstate the veteran questions to the annual population survey or LFS so just concluding to repeat some of the points I've made that there's a lack of regular data on veterans the annual population so is it for 2014 to 2018 it's been the only source of detailed information on that for jerry britain on veterans which can be used to monitor the experience of veterans and in order to look at ongoing trends in veteran employment I've the re-instating question to the aps would be one of probably the best way of doing one way of doing this right okay yes so I think that concludes my my my talk fantastic thank you very much David it was a really interesting presentation and I appreciate and I actually really appreciate that and amongst that dilemma of not being much data out there to understand this very small or niche part of the population further that the pool data set the pooled aps file has been useful for that reason okay thank you very much David so now we move on to the second presentation of this session which is from Richmond again from the kinks college of london and economic stats research centre of excellence Richmond is the fourth year PhD student in economics at the kinks college in london and he holds an m fill in economics from the university of garner and a bachelor of arts degree in the economics from the kum necroma university of science and technology in garner and proud to join in the kinks college in london he worked as a principal research assistant at the institute of statistical social and economic research at the university of garner his research interests are in applied micro economics focusing on labour economics and developed economics and economics in education he's part of the economics economic statistics center of excellence a research network and his current project involves using administrative data to develop new labour force and migration statistics for the united kingdom so um Richmond if you're ready then um uh the floor is yours all right thank you okay i hope you can see my screen yes we can yeah thank you yes um so this uh my first work that i did when i started with my phd so um it's a work that we did um and we okay it's not been published yet it's it's it's it's still in the way uh as we've published on the on the escrow websites as a working as a working document so i'm looking forward to your comments so that i can make it very well and then i publish before i start um i'd like to issue a disclaimer that this work was produced using statistical data from onus the use of the onus that's called data uh that's not imply onus endorsements so um this work uses research data sets which may not exactly reproduce national statistics aggregate all right so um basically um we know that uk relies heavily on survey data to estimate migration uh statistics and um currently we know that there is no known immigration register in the uk so much of the assessments uh are done using the survey data and basically it's done using more they rely heavily on the annual population survey or the labour force survey and then the international passenger survey and we also know that uh because survey data sets has um the the uh there's issue with the sampling size we are not like there will definitely be a margin of uncertainty so using that to publish local area population estimates may actually be problematic and uh it will come with a margin of uncertainty so what we intend to do we in what we try to do is to explore whether an alternative source of information regarding local area immigration population uh could help improve the accuracy and the reliability of published a local area migration figures for uk so um what's uh we propose is to use the electoral rule register and uh what we try to do in this paper is to try and compare if we can actually get some information from the electoral register that can help us estimate migration figures from it so um that's what we do we try to calculate some estimates we drive some estimates from the um electoral register and then we compare it with it's a equivalent estimate from APS data and then the census and then we try to compare and then we try to make uh we try to conclude that's if possible maybe they can start considering the use of electoral register as a means of publishing and migration statistics um before I start I would like to talk a little bit about the definition because in the electoral register we don't have a variable that describe where name was born or whether the name was one abroad but then we have nationality and we know that's using UN definition for a migration uh they used uh whether you were born abroad where you were born uh to be able to estimate that statistic so that's um the only caveat other than that so for now for this paper we use nationality as our definition for immigration and we know from the APS and census um we also can get nationality from there so what we do is to compare the two and then um we make our suggestions and another issue would also be those we do our nationality which uh we note that it could be a problem but then perhaps if maybe in the electoral register when they are registering if they can include where you were born it could really go a long way to help um why are we thinking the electoral registration um register could be important because of its uh legality that's they if you don't register you could be fined and then the way and manner in which the local authority tries to update the register we think it's it's a good way of capturing everyone into the register um um for now we don't have the individual level data sets so what we use is the aggregate uh published data on the um on this website that is for the electoral rule so they publish the aggregate for each local authority so that's what we use and then for the census we use the 10 percent individual level sample and we use the annual population survey from 2004 to 2017 and i've i've done lots of work after uh we published this so for now i'm not going to talk about those ones i would limit myself to this so how do we uh then capture um because for now we don't have the individually uh individual level data sets but then how do we do what we intend to do um we know from uh the voting that um for instance those from the european union can vote in the local authority election but then they cannot vote in parliamentary election so then if we um difference these two then we know that the difference between those who are in the local authority uh those who can vote in the local authority election versus those who cannot vote will represent the european union so at least we able to capture the EU nationals from there and then we compare it with um a drive what a drive estimates from the APS which we do and then we compare to see if uh the APS is able to give something similar and then we see that when you look at the graph we see that from up to um i think 2014 they seem to be moving at the same level but then after that it drops and we we try to find the reason for why it's happened that way uh this could be that uh maybe because there was a change in the way registration was done in the electoral register that could be the reason so we tried to see uh who were those who could have dropped out could it be EU students or not so what we do is to try and add um um those uh EU students to return to see but then still there was still some gap so um our conclusion from that was um probably they may not only be the reason the students may not be uh only the reason it could also be a reduction in double entries linked to movers so we were not really clear as to what could be the reason but then we still went ahead to compare the three data sets so we started by listing the top 50 uh local authorities with the highest EU nationals so we compare across the three data sets and then what we find is that um for instance so we rank them and checking it individually uh we're able to see it looks like the electoral register moves closely with the census than the APS all right um yes and then uh the other thing that we do is to find the correlation between the two so we look at the correlation between the electoral register and the APS we also look at the correlation between the electoral and the census and the census and the APS and yeah all the uh correlations seem to be pretty high but then we we find that the electoral register seem to be much closer to the census obviously because it's also an administrative data it's it was expected and then the last thing that we tried to do was to look at how the distribution have have done in terms of over time as in the top 50 whether they account for how much and we find that um so for instance we find that um the top 50 uh the electoral register in 2004 accounted for about 61 percent whereas this accounted for about 63 percent and then with time um this one has been dropping steadily as compared to this and the other thing that we also tried to do was to also look at for instance we know that the common characteristics of immigration settlement pattern is the geographical concentration so we we were thinking that it could be that for instance if there is an influx of uh EU migrants at a certain point in time then we expect that um if we should calculate the geographic inequality we'll be able to capture at what point in time this inequality was what I mean is for instance um when people move early early on they would definitely try to leave closed by or they would try to go to where their people are so we expect that inequality would be high but with time then as they're able to assimilate then they can uh spread and then it becomes even so we start by looking at the 2011 data set because we have the census and then we compare but then when we do the comparison we find that the electoral register has a closer correspondence with the census than the APS except for for one so we also further have the conviction that it could be that if they use the electoral register in estimating migration statistics it could help the next term is we try to see um by looking at the inequality over time thus we didn't have enough data to to to to uh to justify what we really wanted to say for instance and we were thinking that's probably maybe in 2005 as a result of a eight uh moving to the UK uh it's good capture and then we find that um the year we were thinking that the year perhaps may have captured uh the immigration movements over time better than the APS the next thing that we tried to do was to um still with the differences in the local authority was to assess so another way we tried to assess the degree of correspondence between the APS and the EIL counts is to estimate a simple regression where the dependent variable is the APS and uh counts of the EU nationals and then um so the main explanatory variable of interest is the equivalent electoral register counts so if the population counts in the true uh data sets in area i in time t okay at the same we would expect that the estimated coefficients from a simple uh two variable regression uh would be zero on the intercepts and then unity on the slope but then if there is a negative intercept then it will suggest that um APS may underestimate the count of uh the count in the reference local authority okay in the first period so that is what we try to do and then that's what we find we find that um when we do um the regression without uh the local without year dummy and local authority fixed effect uh we find that um we find that the EIL slope is significantly different from one when we add the local authority dummy it relaxes the constraints a bit but it's still different from one but then and when we look at when we add the structural brick to find um whether maybe the changing register could have helped then we find that um adding the post year 2014 structural brick now we see that the year Lope is uh equal to one and the other thing that we tried to do was to look at um the local authority uh coefficient we find that there were a significant number of them which were negative and others were positive so we wanted to try and uh understand how or what could be the reason for distance so what we did was to extract these coefficients and then we ran the regression on them to try and see if we could find the explanation and what we did was to what we found was that it could be that's the population in those local authority could actually explain some of these stuff so in conclusion um we think that the electoral register appears to offer complementary and useful information on regional figures and trend and it appears to be very close to uh the census that is if you compare it to the 2011 figures uh than the others regional discussion uh probably maybe better measured through the electoral register than the APS yes even though nationality is not a good way of measuring immigration but I think we think it's quite close um um we think that if we get access to the electoral register as in getting the individual level data we could be able to do uh much and then uh advance our case than uh we said thank you thank you very much Richmond uh it's a really good presentation