 welcome back from the break and hopefully from a lunch time if you did manage to grab some food. If you want to stay here for Parallel Session 3A, it'll be in this section, this session here, chaired by myself. Alternatively, if you're looking for Session 3B, so that's Long-Term Plans and Structural Dynamics, that session is in a separate call. So Gillian has put the link in the chat bar. So if you're looking for Session 3B, follow the other link. Otherwise, we're heading into Session 3A here, looking at mobility and transitions. In which case, welcome back everybody. So heading into this next section, we have three speakers with three very interesting papers coming up. So I'll start off by introducing our first speaker in this section. So it'll be Xavier Sandini, who is an assistant professor at the Institut Nationale de la Recherche Scientifique in Montreal in Canada. His research explores the drivers of trends in job stability in OECD countries, the consequences of career instability, and on intergenerational income mobility, obviously with a focus in Canada. His research appears in the British Journal of Industrial Relations, Work, Employment and Society, the International Labour Review, and the Canadian Review of Sociology as well. So here today, he'll be telling us about how he's been using the five-quarter longitudinal LFS datasets to investigate demographic characteristics of people moving between occupations and potential obstacles in doing so. You can see the slides are there, so Xavier, over to you. Well, hello, everybody, and thank you so much, James, for this introduction. So I'm really thrilled to be here to present a paper co-authored with Edward Haddon, who's a postdoctoral fellow at ENRS. Okay, so this paper is interested in contributing to the debate on occupational transitions as they occur, or the motivation is that we're expecting to see high rates of occupational transitions in times of recession or economic or restructuring that occur. So, for example, during the COVID pandemic, we've seen a lot of discussions about workers leaving certain sectors that were shut down for to find jobs in new occupations and sectors that were not shut down or that were growing in size. And likewise, in times of technological change, or when there's a recession leading to a shrinkage in the size of a given sector, such as the manufacturing sector, many workers are faced with the need to find a job in a new occupation that they're not necessarily skilled for, that they're not necessarily prepared for. So this debate on occupational mobility has, at least in some policy circles, has focused on the issue of the mismatch between the skills workers have and the skills required for occupations that are looking for workers. So in a Canadian context, we currently have a pretty intense public policy discussion on how to deal with this issue and with the issue of worker or skill shortages on the labor market. And so large publicly funded programs have emerged focusing on retraining, reskilling. And this has led, not just in Canada, to the development of some tools to identify the least costly occupational transition pathways with tools such as one developed by the World Economic Forum, trying to identify occupations that are pairs of occupations that are the most skill similar and that have only small differences in wages. So these occupations would be, these pairs of occupations would be occupations between which it is easy to transition because it would require only limited rescaling and would lead to, would allow workers to keep similar wages when they undergo this transition. So the contribution of the paper I'm presenting today, what I'm seeking to do is to try to broaden a little bit that perspective to include the obstacles other than skills that workers may face when trying to change occupations. And so I'm going to draw on various literatures, but for, so for example, the literature in sociology, there's a rich literature identifying what what sociologists call class related obstacles, which is mostly occupation specific professional networks, social and cultural capital that facilitate transition between certain expectations that are part of larger class groupings, for example. There's also a possible role that's been documented in the literature of, of course, for discrimination or liberal market segregation based on gender and race, etc., that may make it hard for workers to transition between two skills similar occupations, but that may have different gender compositions, for example. And finally, there are social closure dynamics based on the overcredentialization of certain occupations, alliantencing and certification requirements necessary for entering certain occupations that restrict the number of workers that can enter an occupation. And so in short, the objective of this paper is to develop measures capturing a broad range of obstacles that that workers may face when they seek to change occupations beyond, beyond skills. So to provide basically a multidimensional look at obstacles to occupational mobility. And so in this paper, we provide a descriptive analysis of occupational transition flows and a broad set of drive of possible drivers of occupational transitions. And so just in short, we ask what are what factors structure occupational mobility and what may be the role of factors that are more rarely studied. So that's all I'm going to say for in terms of context. And I'm going to jump right into the data part of the presentation. So when we study occupations with statistical data, as most of you know, we usually rely on standard occupational classifications, such as the UK SOC classification. So here for this presentation, we're using SOC 2010. And the basic principle here behind the development of such classifications is that a job is defined as a set of tasks or duties carried out by one person. So a job is a job's become occupations and or different jobs are part of the same occupation if they bundle a set of tasks or duties or knowledge similarly. And so in the UK SOC documentation or classification, jobs are classified based on skill levels and types of skills that are used. With skill level being related to the level of education training experience that occupational incumbents usually have and the skill specialization or type of knowledge more related to the kind of knowledge of their views at work and the kind of tasks that are performed at work. So again, as most of you know, there are nine major groups in the UK SOC classification. And these are broken down into submajor groups at the two-digit level, minor groups at the three-digit level and then unit groups, which is the most detailed level of the classification at the four-digit level. And there are 369 units in the SOC 10. So each of these occupational categories are associated with textual description. And so the issue that we're trying to address today is that when using this classification, this year called classification, it's not always obvious to determine how dissimilar two occupations are. For example, there's no, and I'm simplifying a little bit, but there's no straightforward way without using, without relying on the text to quantify, for example, whether to move between a chemical scientist and biochemist. So from the 21-11 unit group and 21-12 unit group is a move between two occupations that are more similar than a move between 21-11 and, for example, 21-14 social and human scientists. So with our knowledge, reading the text, we know that chemical scientists are more similar to biochemists than social scientists. But it's not as straightforward to quantify that. And so what that means is, if we want to think about, in this case, in this framework, occupations are discrete categories. And when we want to think about how different those occupations are, or the distance between them, we have to think about it in a binary way, in or out, so the same or different. And so, for example, if a worker was to transition between the fourth major group and administrative and secretarial occupation to another occupation in the same major group, we would say that they're transitioning to a similar occupation. But if they're transitioning from the fifth major group to the fourth major group, or seventh major group to the fourth major group again, it's hard to say which worker transition or which type of transition has the most obstacles involved. In contrast, in this paper, we trotted up the multidimensional view of occupations, looking at a broad range of different characteristics that we quantify on continuous scales, where we would be able to quantify whether a move from the seventh to the fourth major group spans or involves a greater number of obstacles than a transition from the fifth to the fourth major group, for example. So in order to be able to do that, we need a few things. So we need a first data source that is longitudinal that captures, that measures occupations at two time points for a representative sample of adults. And so the UK labor force survey provides that information. I'll say a little bit more about that in the next slide. But we also need a data set of occupational characteristics to calculate distance indices. So measures of distance between various occupations for multiple dimensions of occupational skills and composition. And to do so, we we use the ONET that we map onto the SOC 2010 classification and data from the UK LFS again on occupational composition. So again, the UK labor force survey, as you probably all know, is a representative sample of the adult population. It's a large, quite large data set with a lot of information on worker characteristics and their jobs. One feature that we leverage in this paper is that it's sampling design uses rotation groups, which means that every respondent stays in the survey or is expected to stay in the survey for five quarters, which makes the LFS longitudinal. And so we can observe any occupational change that occurs within this window of five quarters. Now the ONET, some of you may know this data set. So this data set basically includes a large number of indicators on the work activities, knowledge, skills, activities performed in each detailed occupations from the US SOC classification. So it's collected from occupational experts and occupational incumbents. And so the part of the data set that we use is the 41 indicators under the work activities or tasks dataset. So how this data set is created is it's an aggregation of information on hundreds of detailed tasks that are performed by workers in different occupations that are classified in, as I said, 41 different indicators. One example is, for example, is inspecting equipment structures or materials, etc. So it's work activities or tasks, but there are also physical tasks as well. And so this is what it looks like. Basically, every indicator has two scales that we multiply to create a single scale for a given indicator, importance multiplied by level. And as an example, it reveals some similarities between occupations that wouldn't be captured by a hierarchical framework, such as the major, some major, minor and unit groups. So sociologists here estimate quantifiable characteristics of products, events, or information in a similar way or with a similar using a similar level as urologists, border makers, etc. So it reveals some similarities, but of course those occupations are different under other dimensions. And so what we're doing with the ONET is that we're conducting a principal component analysis on 41 ONET work activities indicators, where we basically extract eight principal components and develop a single distance measure or calculate the single distance measure based on these principal components using Euclidean distance. And so what this gives us is a skilled dissimilarity score between all 369 by 369 pairs of occupations in order to quantify the skill dissimilarity between all possible pair of occupations in the UK SOC. We do the same with the differences in sociodemographic composition. And so we calculate the proportion of workers with various characteristics in the from the UK LFS again, pulling a few cross sections together. So share of workers with graduate degrees, share of women, share of racialized workers in each occupations. And then we calculate the absolute value, the difference in those proportions between all possible pairs of occupations again. And so we estimate the log linear model without going too much into details. The dependent variable is the log of the frequency count of the number of workers who are in a given cell in our dataset, which is a dataset aggregated at the level of pairs of occupations. And so we know how many workers each line is a pair of occupations with a number of workers who underwent this transition over five quarters. And so we have a parameter for skill dissimilarity and we have three parameters for difference in sociodemographic composition. So here are very quickly our distributions for the distance measures. They're all standardized in the model that I'm going to present after. And they're not very correlated, so it's likely that they may have independent effects. So only in a descriptive way, this is basically all the transitions that we observe. And so the large points on the small diagonal are all workers who stayed in the same occupation at time one and time two in the same unit group. And then dots that are in the same similarly colored squares are workers who changed occupations, changed unit groups, but stayed in the same major group. So we see that a lot of workers transition between two occupations that are in the same major group. But what we see then in gray is all kinds of transitions where workers left their major groups. So it seems like it's fairly frequent that there are certain locations in this occupational transition matrix where transitions are quite likely and frequent. And so basically we want to know if these transitions are driven by the similarity between those occupations, even if they're not in the same major group, for example. So the estimates of our log linear model are presented in a stable. And I want to focus on model eight, which includes all of our distance measures plus our big class or occupational major group and microclass. So occupational minor groups, dummies. So these coefficients, when they're below one, it means that an increase in one standard deviation in a distance measure is associated with decrease in the rate of transition between a pair of occupations. And so we see that as the distance and skill similarity in education share and sex share and ethnicity share increases in all cases, it decreases the rate of transition between those occupations. So occupations that are more dissimilar along those four dimensions have fewer workers transitioning between them. And so this broadly supports our hypothesis that there's going to be beyond beyond skill dissimilarity. There are other factors that or other obstacles faced by workers who seek to transition between occupations. And so I will jump to my concluding slide now. So just restating the point that the factors that structure occupational mobility flows, occupational flows are really multiple. And in this paper, there are some important limitations to note, which is that we still take a limited number of dimensions into account. And so we haven't taken in this version of the paper into account the geographic concentration of occupations, differences in compensation or certification and licensing requirements. There may be other dimensions. And if you have ideas that be interested to hear them, we haven't taken also into account the direction of the distance or the move towards more skill intensive or less skill intensive occupations or between more feminized or less feminized occupations. And finally, we haven't really explored the difference between personal and occupational characteristics. So this would need to be modeled in a slightly different way. And so just to open on a discussion point, I think this paper points at the need to think about a broad range of possibly exclusionary factors that may represent significant obstacles to certain occupational transitions when we think about the development of interventions and policies focused on facilitating occupational mobility. So thank you. And I look forward to your comments and questions. Moving on to the next section. So bear in mind that there's a lot of chat in the chat bar, I know, but there is parallel session 3b if you wanted the other presentation going on. However, if you're still staying here, we're moving on to Fabian Petzi, who was a research fellow at the Science Policy Research Unit at the University of Sussex, and is also part of the Horizon 2020 Pillars Project. He's interested in labour economics, macroeconomics and behavioural economics with his research focused on labour markets, technology skills and intergenerational mechanisms. Today, we're going to hear about an investigations topic of social mobility and the ability for people to proverbially climb the ladder and develop their careers through middling jobs up towards high paying occupations. So Fabian, over to you. Today, talk is about social mobility and the basic link that it could have with job polarisation. So this is a joint work with Cecila Garcia-Penalosa and Tangy Vanipursel from X-Marsay School of Economics. And the question we ask is basically whether individuals can still climb the social ladder as the middling jobs are becoming scarce. And we use a British call stage to answer this question. So there has been a decline in social mobility over the past decade. This has been observed for the UK, but also for the US. And therefore, this has strengthened the link between individuals' background and also their social economic outcomes, such as like the occupation that you could get. The second point is that there has been a phenomenon that is the job polarisation that has been observed. And this phenomenon has also been in many developers and developed countries. So basically the job polarisation process is the idea that if you split occupations into three groups, the low paying occupations, the middling and the high paying occupations, the employment creation is done at the extreme of distribution. And so those jobs in the middle, they tend to disappear. The intuition behind that is that those jobs are routine jobs. And therefore they are automatised and there are fewer jobs available here. And so if you used to be in a world where you can start at the bottom and you can climb the social ladder to end up in high paying occupation, but those jobs are disappearing in the middle, then you might get stuck at the bottom. And so that is basically the kind of question that we answer in this paper. So to answer to this question, we use those two metro British core studies. So one is born in 58, this is the NCDS, one in board in 70, this is the British core study, the BCS. And so we exploit the fact that one cohort, the younger one, entered on the labour market, which was much more polarized compared to the older cohort. So we proceed in two steps. So today we mostly focus on the second one, which is the one that we improved recently. And I will just go through the first one. So our precast strategy goes this way. So we didn't think of first the changes in social mobility that are due to what we call the intra-generational component. So it's the job-to-job transition. So basically what Xavier was presenting in the previous presentation, so all people moved from one occupation to the other. And we will focus on the first part of the occupation. So when they are young and their mature occupation, when they are 42, for instance. And so we will focus on this mechanism. And we will have another component that will be the intra-generational component. So the role of family background, for instance, the role of parental income in explaining the transition and the social mobility of their children. So we'll play those two components. Then we will also go at the regional level, and we will estimate the effect of polarization on the role of parental income. So basically, if you are born in a region in which there has been a polarization, we will find that the role of parental income has increased, and this has basically increased the immobility in that region. So in terms of the main results, there are three of those. So first of all, we show that this intra-generational component, so the mobility that you get from one drop to the other, is very important in explaining the inter-generational mobility. We also find that those from better backgrounds, so those from the top of the parental income distribution, they have been more likely in the younger court to climb the social ladder compared to the older court. So the role of parental background has been reinforced from one court to the other. And lastly, based on regional studies, we show that the effect of parental income on occupational outcomes is stronger in the area with greater total polarization. So that is for the main results. In terms of related literatures, so we contribute to the literature on the determinants of inter-generational mobility. And here, mostly we show that this intra-generational component, so the job that you get as a first occupation and all the transition going on after that are very important in explaining the inter-generational mobility when you look at the effect of parental income on children's outcome. So that is our contribution to this literature. We also contribute on showing that there is an increased role on showing that the role of parental background has increased on children's outcomes. And so here we get rid of the mechanism of education and we still find that parental income has an effect on the transition afterwards. Lastly, and more importantly, the consequences of employment polarization. There is a literature about that with several consequences. We basically show that actually employment polarization has also an effect on polarization and the closest papers to ours are basically Enig and Google, in which they argue that the polarization has generated the polarization in education which reinforces the lack of mobility. The thing is that their explanation works pretty well for the US, for instance, where education costs are very high, but this cannot be applied to European countries. And with hall paper, we get a story based on basically the transitions from job to job that could apply for countries such as the UK or any other European countries. So let's dig into the data. So we have those two major British course studies. So one is the BCS. So they are born in 1970. The other one is born in 58. And we will focus on two periods. So their entry job, their first period occupation. So for the younger cohort, this will be 26. For the older, this will be 23. And we will look at the outcome when they are 42. And we'll look at the transition between this first period and the second period, but also at the first period occupation and the second period occupation. We will measure the average parental income as a proxy for basically the parental background. And we will get this parental income from underage interviews. We will look at this viable in logarithm. And then we will standardize that the course level. The intuition for that is that because there has been a rise in inequality, the variance of the distribution of parental income has increased for one cohort to the other. So we want to get rid of these effects. And so we standardize it. In terms of occupations, we will regroup occupations. So here we have UK data. So we have the social, the SOC classification. We use a crossword to get the ESCO HCI occupations. And we classify according to the literature on job polarization into three categories, plus an autophore category for people not being, not working basically. So here you have the occupations and we end up with those four types of jobs. So in terms of the empirical approach for the first part, basically, we estimate multinomial logistic regression. So for the first period occupation, for the mature occupation. And then here we look at the transition. So once you get this first period occupation, J, all the new trends to occupation K, or it might be that this is the same type of occupation and you stay in the same position. All the long for each regression, basically, we will have the effect of parental background and a set of control variables. And we will interact all variables with a dummy that that equals what for the younger cohort. And basically, by looking at the coefficient, interact with the dummy, we get the change from one core to the other in terms of the role of parental income. So I'm going to go a bit faster than that. But basically what you observe here is that if you are at the top of the parental income distribution, basically the likelihood that you start in a hybrid occupation has increased. Whereas if you are at the bottom, for male and female, this hasn't changed. It's the same picture with the second period occupation. So when individuals are 42. And so you see that basically people along the parental income distribution do not have the same chance to end up in different occupations. And this has worsened over time, or at least for those two courts. So I'm going to skip on the transition and I will go at the regional level directly. So here what we do is that we use the label for survey to build a polarization measure that will explain in just a few minutes. We have those 10 regions, which are Nets2 regions. And so what we are interested in is the change in the role of parental income between one court and the other. So this will be measured by the coefficients beta here. So this is for the younger court. This is for the older court. And we take the difference in the role of parental background to explain the occupation K at the age of 42. And we will get this estimation at the regional level. So we'll get basically a change in the role of parental income for each region in the UK. And what we are interested in is the link between this regional polarization, but this measure that I will explain in a minute, and the role of parental income. And so we will run thereafter this regression on which we try to look at the effects of this change in polarization on the change in the role of parental income background. We have a set of control here. I won't spend much time on it, but they make sense in some hold. The problem with this relationship is that you have to concern. You have the energy issues. So basically the regional structure of employment may have also been affected by the degree of social mobility. And you have a limited viable bias that is possible. So also factors may have affected both polarization and social mobility. So the strategy regarding this is to use a shift share measure that is based on national level changes. So basically we look at the change at the national level between 2004 and 2002, which is basically the middle of the career of the younger cohort and the middle of the career of the older cohorts. And we use the share. So it's fairly standard in the shift share literature. And so the share was before the period. So at this point, the polarization process was not much going on. And so we can basically get rid of the energy issue. And then we instrument those share here with basically the change in the occupations, but average across a set of European countries. And so this is consistent also with the literature. And because we know that this polarization process has taken place in several European countries. And so by doing so, basically, we get a shift share strategy and getting directly to the result. What we observe is that basically in the regions in which the change in employment polarization has been large, we also observe that the change in the parent-time coefficient for the second period occupation has also been large. There is a positive correlation between employment polarization at the regional level and the change in the role of parental income in those regions. So I want to spare a bit of time for questions. So in terms of insight that we can derive from this paper, basically we show that the intra-generational mobility is an initial aspect of the observed correlation between parents and child outcomes. And we also show that basically those from better hope backgrounds, so at the top of the parental income distribution, they have become more likely to climb the job ladder and those who are at the bottom, they are more likely to get stuck at the bottom of the income distribution because they are more likely to get jobs which are low paid. And lastly, at the regional level, basically we show that areas in which the polarization was important, was greater, are also the regions in which the role of parental income has played, has become more important. So thank you for your attention. If you have any questions, I would be more happy to answer. And you can always reach me by my email and you can find the draft of the paper on my website. Thank you for your attention. Very much Fabio. There's nothing in the Q&A box at the moment. So if anyone has any questions, please put them in. I'll ask you a question myself if that's all right. So when we're thinking about the parental income, how exactly is that defined? Is that the main earner or the total of a couple or household income? How about is that broken down into just the single variable? Yes, good question. Thank you. So what we do here is similar to what is done in the literature regarding parental income when they look at income or social mobility. So basically the papers of Blanton, Macmillan, and Coalfalls. So here we look at the parental income in underage interviews and we use the household income as a measure. So for instance, in the case of the BCS, this income is available at year 10 and 16. So we think basically the average. And here for the old accord, unfortunately, it's only available at age of 16. Of course, we run several web snatch shape to show that this is not what is driving the effect. That's whole household income. So it might be factoring in, say, a grandparent who happened to live there or an older child's part of the family as well. Sorry. So the parental income there you're saying is the household income. So if there are other workers in the household beyond just the parent or two parents, that might be factored into that total as well. That's a good point indeed. So here, yes, here, we do not have this. We do not have this. I agree that this would matter, especially because we could think that there would be also a channel for which having the grandparents around could help in building the future of the children. Yes. It's a good point that we should look at. Thank you. Sorry, I haven't indicated we'll work, but hopefully a useful thought. We have another question here as well from Christina. So what patterns you've found in the second cohort at age 42 have been influenced by the economic crisis that was happening at the time? Yes. So, okay, it's a very good point also. So there are two things. We have checked, I don't know if it's here in the appendix because I made a very short presentation, but this is important. So there have been several crises over this timeline in the 90s also. So what we do is that actually we have the full activity history. The reason why we pick up those specific years is because there are interviews at those years, and therefore the occupations that are reported during those interviews are more precise. But if we want, we can basically pick any year in between, but this won't be as accurate. So what we have done is that we have checked basically with respect to the business cycles and the crisis. If this was not driven by all the crisis that happened over all this timeline, and we find that basically the results are pretty robust to this 2008 crisis. Although this might also be part of the story, but there is something that is more about long term. But thank you for your question, Cristina. I don't think there are any other questions at the moment, in which case. Thank you very much Fabien. We're almost back on time. Fantastic. So I appreciate the presentation. Thank you very much indeed. In which case we'll head on to the third of our three presentations in this particular parallel session. So moving over to Marie. Marie Horton is a senior research analyst at Engineering UK, which is a non-profit organization working with the engineering community to increase the knowledge and interest in engineering careers among young people. Recent reports by Engineering UK have published looking at options in opinions on engineering, as well as educational pathways into engineering and the workforce itself, which have been used to inform educational providers, employers and policymakers. Specifically today, she'll be presenting her review of the representation of female workers in the engineering workforce and the effects of recent efforts to attract more female workers into the profession. So Marie, hopefully you're there ready to go. Yep. Can you see my slides? Okay. Yep, slides and we can hear you. Go for it. Thank you. Okay. So as James said, I'm going to talk through some work we published in the last year on the trends in the numbers and percentages of women in the engineering workforce. So to start a bit about Engineering UK, so we're not not for profit organization and we work in partnership with the engineering community. We do this with our programs such as energy quest, robotics challenge and the Big Bang Fair and they're designed to excite young people about the variety and opportunity presented by a career in modern engineering. We try to give them the chance to meet people already working as engineers. So I work in the research team and as well as analyzing survey data gathered from young people, part of my role is to explore the data on the engineering workforce and to do this, we use the labour force survey and we explore all different insights that we can gain. So last year I did a deep dive into the trends in female engineers because we know that although engineering remains a male dominated field, since 2010 we've seen both a proportional and absolute increase in the number of women working in engineering and we wanted to explore if there are particular roles within engineering where we've seen these increases, where they've been concentrated or perhaps some parts of engineering that haven't seen as much increase or where the numbers of women have decreased. So the analysis we've carried out looks at various aspects of what we call the engineering footprint. I'm just going to talk you through this diagram here to make it clearer what I'm referring to throughout the rest of the presentation and the way we define this is using the SOC codes that some other people have already talked about. I won't go into too much detail about what SOC codes are. Firstly the centre circle here, we have the core engineering occupations. So these are roles that are primarily engineering based and require consistent application of engineering knowledge and skills to execute the role effectively. So for example this might be civil engineers, mechanical engineers and technician, but also part of the engineering footprint we have what we call related engineering careers and that's represented here in the outer circle. So these are roles that require a mixed application of engineering knowledge and skills alongside other skill sets that are often of greater importance to executing the role. So for example that might be quantity surveyors, architect, IT operations technicians, web designers and developers and then for both core and related engineering occupations we can categorize them by industry which we would do using the SOC codes. So they could either be in the engineering industry which might be for example manufacturing or construction company but there are also engineers that work outside of the engineering industry. So that might be for example in retail and this is important in the context of this work because we know that there are more women working in areas outside of core engineering and also outside of the engineering industry than there are in the core occupations within the engineering industry. So it's important for us to try and understand why. Now before we start to look at the results I just want to briefly cover the way the data set was prepared. So for each of the data points calculated we took four quarters of the LFS quarterly data set from the UK data archive and we collated them to create an annual data set and then we took only waves one and five of the data so that we weren't double counting any respondents who would have appeared in more than one quarter of the data but in different ways. And then also just to reiterate what Martina was saying earlier about the mode change over the time period so when we're looking at the trends we know that the mode change from mainly face to face to telephone data collection and that might have had an effect on the sample demographics so we just need to take that into account when we're looking at the trends. Okay so moving on to the data this chart is showing the trends in women from 2010 to 2021 so the yellow line at the top of the chart shows all occupations and the grey line shows core and related engineering combined. So obviously starting with the good news there we can see there's an increase in the engineering occupations from 10.5% in 2010 to 16.5% in 2021 however we're still a long way off the 47.7% average for all occupations and so that's remained quite stable over the entire period and engineering has not seen much of an increase. Now if we bring in the breakdown of core and related separately so you'll notice that the blue line for core engineering remains consistently lower than the orange line for related engineering across the entire period and the difference is about 4 to 5%. So what that means is that the roles not traditionally associated with engineering may be more successful in attracting female engineers into the workforce than what may be considered the traditional engineering careers. So now here we have the chart representing the same data but in terms of actual number of employees rather than the percentages. The colour of the lines here represent the same groups as in the previous slide but note there's a double axis here due to the difference in the magnitude between engineering and overall obviously and so the the axis on the left shows the values for the engineering occupations and on the right for all occupations. So the important thing is that the increase in proportion of women represents an increase from 562,000 women in 2010 to 936,000 in 2021. If we're looking at core and related combined and that's the grey line there and this has coincided with an overall expansion of the workforce from 5.3 million workers in 2010 to 5.6 million in 2021. So what's also worth noting is that the rising women continued in absolute terms even when the total number of engineering roles fell in the first year of the COVID-19 pandemic and the numbers of women working overall also fell but we still saw the increase in engineering which is a positive sign. Now when we look at core and related engineering in the first half of the decade we see there's little difference between the numbers of females employees and actually in the latter half of the decade there are more female employees in core engineering despite there being a lower proportion of women in cores and related. So what that's saying is that the size of the workforce in core engineering has seen larger increases over the period and many of the additional roles have been filled by women. Now if we look at the data just for 2021 so this chart shows the percentage of women working in different parts of the workforce. At the error bars at the end are the 95 confidence intervals and by that we mean we're 95 confident the true value lies somewhere between those limits. As a reminder I'll bring in the chart at the top here from the beginning to explain the footprint. So first the core engineering roles so they're those in the central circle of the footprint so they're that they've got the lowest percentage of women at 15.2 percent of the workforce compared to 19 percent in the related engineering occupation and you'll remember from the diagram that's represented by the outer circle and if we combine those two that's where we get the 16.5 percent that we saw on the previous chart and obviously we're comparing to the 47.7 percent of all occupations at the top there. So that's really you can see from this chart the significant difference there. Then in this next section rather than looking at occupations we're looking at industries so that's everything on the left hand part of the footprint diagram shown in the purple there. So we see here that the sector has higher proportions of women when compared than when we can learn look at occupations the 23.9 percent of the engineering sector were women. However that's still low compared to all industries and then finally what we look at here is the number of women in engineering occupations within engineering industries rather than engineering occupations outside engineering industries. So we can see that there's many more women working outside of engineering industries although they're in engineering occupations. We turn now to look at the occupations within the engineering footprint and the different proportions of women working in them. So this table shows the female percentage of the workforce in 2010, 2015, 2019 and 2021. I appreciate there's a lot of numbers here so I'm just going to talk you through them a little bit. So across that time period culture media and sports roles consistently had the highest percentage of women working in engineering roles at around two-thirds while skill-based roles had the lowest at less than five percent. There's also been significant increases in the percentage of female engineers working within other occupational groups during this time in particular within the professional and associate professional groups highlighted here where the figures have increased by between eight and fifteen percentage points. In general this has coincided with an overall expansion in the number of people working within these SOC major groups suggesting that new roles are being created in these areas and they're attracting more women into engineering roles. So for example between 2010 and 2021 the number of engineering roles within the business, media and public service professionals occupational group increased by 30,000 and the percentage of women in these roles doubled but the picture is complex there's not really a clear trend across all groups while we saw an increase in the number of engineering roles at the business and public service associate professional group. The proportion who are women decreased from 25.2% in 2010 to 16.9% in 2021. Conversely despite the increasing overall numbers of engineering roles that are classed as process, plant and machine operatives there's been an increase in the percentage of women in this group from 17.7% to 20.7%. So now if we look at occupations in more detail starting with those that saw the increases. So encouragingly 61 of the 97 roles in the footprints saw an increase in the percentage of female workers between 2010 and 2021 and in 19 of these roles increases exceeded 10 percentage points and they were actually quite wide ranging in what they were covering. So for example there were rubber process operatives, TV, video and audio engineers and electronics and electrical engineers also saw more than 10% increase. In all but seven of the 61 roles the increase has been in both proportional and absolute so by that I mean we've observed an increase in both the percentage and the numbers of women working in them. For 27 it's coincided with an expansion of roles overall in that occupation. So for example IT and telecommunications professionals and the number of the total number of roles increased by more than 90,000 overall and of these new roles more than half of them were taken up by women. There are also cases however where the rising women has been amidst an overall contraction of the workforce. So for example the number of female electronics engineers increased by 2500 but the number of men decreased by 15,000 so that resulted in an overall decline of around 12,500. Going forward it's important to understand what's driving the opposing trends so and whether there are any differences in the way women and men are being treated in terms of recruitment pay contract type or retention for example across the different areas of the footprint. Now on the other end of the spectrum 23 roles have seen a decrease in the proportion of women since 2010 and in nine of these the change was five percentage points or more and again it was quite wide ranging so there was inspectors of standards and regulations, there was assemblers, environmental professionals, all of those saw a five percentage point or more decrease. For all but five of the 23 roles it also represented a decrease in absolute terms so there were actually fewer women working in the occupations in 2021 than there were in 2010. The five exceptions to this are listed here so these are the ones where the number of women between 2010 and 2021 grew just simply not at the same rate as men. And finally for the 13 roles from the footprint listed here the proportion of women has remained at zero percent so virtually no women are in these professions in 2010 and that's remained the case some 11 years later. It's worth noting that for the majority of these roles the overall size of the workforce has shrunk over the last decade but still to have no women working them in them is a reminder of the large gender gap in engineering. So to summarize since 2010 we've seen a large rise in the number of women across the majority of engineering roles and across both engineering and non-engineering industries and this is welcome news we encourage the engineering community to continue to celebrate and promote examples of women working in engineering roles and the sector especially to the girls who could be tomorrow's engineers so that we can continue to increase diversity. To do these engineering employers need to understand their own workforce focus on understanding practices that extend recruitment and retention to underrepresented groups they need to identify and promote practices that help to increase the appeal, recruitment, retention and progression of women in engineering and we will continue to research what works in terms of encouraging more girls into engineering and continue to monitor what's going on in the workforce so that we can see what progress is being made. That's it from me. Great thank you very much indeed Marie, very informative.