Reducing automation risk through career mobility: Where and for whom?

Abstract

Automation risk prevails less in large cities compared to small cities but little is known about the drivers of this emerging urban phenomenon. A major challenge is that automation risk is quantified by work-related tasks that allows for measurement through occupation, which is in turn implicitly related to local economic structure and to individual career paths. This paper examines the role of working in cities on changes in automation risk through individual career mobility. Using panel data on Swedish workers, we show that the metropolitan effect of reducing automation risk is mainly induced through inter-firm job mobility. Separate estimates for different groups show that this effect accrues mostly to native, high-skilled and male workers. El riesgo de automatización prevalece menos en las grandes ciudades que en las pequeñas, pero se sabe poco sobre los factores que impulsan este fenómeno urbano emergente. Una de las principales dificultades es que el riesgo de automatización se cuantifica mediante tareas relacionadas con el trabajo que permiten la medición a través de la profesión, que a su vez está relacionada implícitamente con la estructura económica local y con las trayectorias profesionales individuales. Este artículo explora el papel del trabajo en las ciudades en los cambios en el riesgo de automatización a través de la movilidad de trayectorias profesionales individuales. Se utilizaron datos de panel sobre trabajadores suecos, con los que se mostró que el efecto metropolitano de la reducción del riesgo de automatización está inducido principalmente por la movilidad laboral entre empresas. Las estimaciones separadas para los distintos grupos muestran que este efecto corresponde sobre todo a los trabajadores nativos, altamente cualificados y de sexo masculino. オートメーション(自動化)のリスクは小都市に比べて大都市では少ないが、この新たな都市現象の要因についてはほとんど解明されていない。オートメーションのリスクは職業による測定を可能にする仕事に関連するタスクによって定量化されるのであるが、逆にそれは、地域の経済構造や個人のキャリアパスに絶対的に関連するということが大きな課題である。本稿では、個人の雇用流動性によるオートメーションのリスクの変化における都市部で働くことの役割を検討する。スウェーデンの労働者に関するパネルデータを用いて、オートメーションのリスクを低減する大都市効果は、主に企業間の雇用流動性によって惹起されることを示す。異なる集団の各々の推定から、この効果の恩恵の大部分は、地元出身でスキルが高い男性の労働者が受けていることが示される。 Automation influences labour markets by replacing human workforce in certain tasks (Autor et al., 2003; Brynjolfsson & McAfee, 2011). Whether this threatens existing jobs or facilitates creation of new jobs depends on the type of investments into automation, worker skills and their potential renewal (Acemoglu & Restrepo, 2020a). Because investments into technologies as well as the skill level of human workforce typically differ across regions, the problem has a necessary spatial dimension. Accordingly, workers in small cities have been found to face higher automation risks than workers in large cities (Crowley et al., 2021; Frank et al., 2018). Further evidence suggests that investments into labour replacing technologies in metropolitan regions lead to employment growth in the long run because robots supplement automatable tasks that trigger upgrades in local skills (Leigh et al., 2020). Nevertheless, how individuals adopt their skills to avoid the threat of automation and how geography facilitates this process is one of the important questions in this quickly evolving, but still largely uncovered, field (Frank et al., 2019). In this paper, we expand on recent aggregate analyses (e.g., Crowley et al., 2021) by detailing the micro-mechanisms explaining the role of agglomeration in relation to regional automation vulnerability. Specifically, we look at how working in cities facilitates career upgrades and prevents automation risks. We argue that large labour markets create more favourable conditions for job mobility, through which individuals are able to reduce their exposure to automation. The role of cities in reducing automation risk therefore lies in facilitating upward career mobility. According to central tenets in urban economics, two mechanisms take place in cities in this regard. First, the demand for non-automatable tasks is high in cities due to their functional specialization (Duranton & Puga, 2005). Second, cities are arenas of learning that offer additional opportunities for individual workers to advance their career and perform tasks that are difficult to replace (De la Roca & Puga, 2017; Glaeser & Maré, 2001; Gordon, 2015). However, workers are not homogenous in their ability to learn in cities, which arguably depends on their individual skill base and other personal attributes that may determine career mobility. High-skilled workers, for example, can further upgrade skills by learning while low-skilled workers enjoy the demand for low-skilled non-automatable jobs (MacKinnon, 2017). The recent empirical analysis exploits employer-employee matched data covering a 10% random sample of the Swedish workforce over the 2005–2013 period. We apply the occupation-level risk measure introduced by Frey and Osborne (2017) to quantify how much an individual is exposed to automation. Descriptive statistics help us understand the role of cities in getting a low-risk job. Further, we apply multinomial logit models of occupation mobility that enables us to examine the role of large cities in helping workers to upgrade careers and prevent technology-driven displacement. Our results show that there are stark differences between small and large regions concerning the distribution of high- and low-risk automatable jobs. During this period, Sweden experienced a remarkable rise in the share of low-risk jobs that concentrated in metropolitan regions with a growing intensity. The largest cities of Sweden accounted for 58% of overall job creation between 2005 and 2013, which was 65% in the case of low-risk jobs. in large cities have higher of getting a low-risk which for individual We that workers automation risk on their and skill that high-skilled male workers a career upgrades that and learning in cities prevent automation However, for workers, workers than and are Our that is much for workers to jobs in empirical for that one of the major the spatial of workers is the of new jobs. this the paper in The a of the that is by an of in data and the for the analysis in The paper with recent in the of automation technologies their on the labour the of In a recent and argue that about of jobs across different face at a of by in the the and regional differences regional in et al., 2018). This further in employment and in in the by the labour of automation, is still a of how automation labour argue that the and of and other technologies the of workers tasks (Acemoglu & Restrepo, et al., 2003; Brynjolfsson & McAfee, & that automation the demand for labour in tasks (e.g., and other (e.g., personal through (Autor & Whether the of the of jobs is a aggregate data on have that automation reduce the employment share of workers tasks (Acemoglu & Restrepo, & & 2018). the of data on how may in the spatial that differences across local in of and ability to adopt technologies on the of that the labour of automation across and workers on their are to and by to on the of cities or in and are the the demand for and jobs is higher in cities as well as the demand for jobs (e.g., workers, workers, technology-driven to to urban related and have been et al., et al., as with the geography of small cities and are to by job the automation of while large cities or the employment of the et al., & 2017). A recent on metropolitan this by that small cities have higher automation risk than large cities (Frank et al., 2018). the automation risk for empirical analyses are at the occupation-level et al., Frey & the of Frank et and other et for Swedish on regional automation risks and Crowley et for with The of a an the that the local that the structure of a is by the local demand for spatial differences in automation risk a of attributes and with the local The demand for non-automatable tasks to higher in large cities because of functional specialization (Duranton & Puga, 2005). and technologies allows to their in tasks and and their in the largest of jobs not high but skills that are as of automation (Autor et al., 2003; Frey & 2017; & the overall share of jobs to higher in cities, workers to large cities are therefore to have a of low-risk jobs. We to this as the upgrade in the career of to a which prevails the worker a job in a large job or upgrades can as that a in the automation risk by workers to jobs in large cities with this of to large cities as career upgrades as because the structure of the local labour through which working in cities to the of automation risk is that workers in cities can in their career a of in cities depends on jobs that offer to of and 2015). exposed to at and personal and have on the and job of workers (De la Roca & Puga, 2017). individuals with urban to a of job or & We to the career as upgrades because over through the of and la Roca and (2017) evidence on upgrades by that working in cities is with and found that the urban growth is for workers. tasks at higher of the career (e.g., and with skills of different et al., mobility through upgrades a of automation that the mechanisms upgrades are much more than in the case of upgrades changes in automation risk the that low-risk jobs to in large upgrades in cities in the of learning opportunities (De la Roca & Puga, 2017; Glaeser & Maré, 2001; Gordon, through & and job the role of cities in reducing individual automation risk can through career However, to examine the to which upgrades help reduce automation one has to other worker for worker is important because the of getting a non-automatable job is to by by et and and determine and the career of workers. individual into upgrades is important because workers with different may urban career opportunities to different In this workers in cities are not to the by and upward career mobility is more workers due to their to to workers with high or are to the of the upgrades in human there are other individual and that determine potential for career mobility (e.g., of labour may to job and therefore differences between and workers and in career mobility. effect across workers with different attributes one can the less worker groups that to automation of their We with a in which changes in automation risk are to individual attributes through occupation mobility the to which workers are exposed to the threat of automation depends mainly on the of their of the changes in automation risk across jobs. job mobility that the of that influences job mobility has an effect on individual exposure to automation. in and job changes are to an problem in which workers for a of jobs that an career is a of potential but the of job mobility are by individual attributes (e.g., and a of other to the example, mobility is in the that job and are to the inter-firm mobility is by the of local job at in workers a of mobility this the of in the automation A to the of cities is data for workers not in or of the that the we can the effect of and on upgrades that the in the This that of and to the urban that workers a of job opportunities and therefore the career much et al., not of the in the of individual the estimates of to the and for are the we can to examine the role of cities in reducing individual automation has the that the distribution of the results are as this can to the of more than one urban a of on the distribution of automation risk and that low-risk jobs to in the Swedish metropolitan and a other large cities, in of the we urban and large urban We matched employer-employee data Sweden covering the between 2005 and In this the of individuals on the Swedish labour as well as their place of and place of at the level are with other individual attributes The 2005–2013 is due to two & have not labour in Sweden to the of which this for as we can to Second, the is due to a in occupation and which and the occupation is which is with the are to the of the The are into large urban and small The the largest cities of Sweden and urban other and cities as while to the mostly regions of Sweden between of which the and to and and the We on a small of because in models on for workers across in a of not due the small of worker mobility in the in in the of the in the metropolitan regions in and about of is the job growth has place while the of has by about in the small the are in the regions the employment is with the in as well as to (e.g., and to and has at a in but in the regions this is more with an in a This a in the as jobs not a are with new jobs a in the largest regions while that is not the case in the regions & the spatial of employment in Sweden as is in the regions but that is while more of the the and to a is found in the We measure automation risk the estimates of Frey and Osborne (2017) at the of to the of for of the In their Frey and Osborne (2017) a that and learning a a of to and the that the level of and and to perform to a learning that a of automation risk for the are in the of Frey and Osborne Automation matched to the Swedish data in two We the between and of the a in the case of automation a of of the a the on the of for the measurement of automation risk is data on job at the level of and not workers. the of an occupation may across regions, is that to type but in the occupation may in changes in the automation risk of the workers may new skills to less to automation. a more for the than the have to certain that the results may to the of the we the other but to the learning not of job mobility, the on therefore the of career in that in or occupation mobility. the and of the automation risk The and this of tasks as and The tasks and mostly This as workers, and low-skilled The spatial of labour that regions in of of are exposed to automation. a of the and for in the in regions the mainly to and as and while in large and small regions and are the low-risk in small and large urban regions employment in is mostly found different of and while in regions and workers in between the spatial of low-risk The the in of the the of low-risk 2005 with 2013, we that automation to in regions while in the largest the we in this paper, regions have for automation. This is related to the spatial of job creation and in as the creation of new jobs have been mainly to the large urban regions the & 2017). in employment has mainly in the largest regions, while the jobs in and regions to the and central of Sweden is with an automation risk over statistics of are in which between at across and in the in multinomial logit the of is on in is important to there are differences between the of and of the workers at the 2005 to The is than the between and for the job or are more The largest of can found in the in the in the the workers, are and have at one attributes differences can between and there are and of the overall sample have a while in the case of the of workers is higher of the are individuals working in or other large cities in and regions of the overall The regional differences between to in than while more in large cities compared to by and on mobility in Sweden that the are between small and large regions, and large regions and & 2017). of the to workers their automation risk by to low-risk jobs in The of changes in automation risk is in large urban regions and in small can found on occupation mobility that upward changes in automation in of the to risk in small regions this is According to statistics job mobility that changes in automation with the of the job mobility is more in type of In to the on automation risk more is of occupation mobility that changes in automation in the the for can changes are of and a of In workers to in which can their and (e.g., and this is of the or upward changes in automation We by individual but workers to In this multinomial logit are by a of individual as place of and a for working in that the of automation is the the automation risk of the occupation is into the to for further career workers in the of the automation risk distribution (e.g., and are able to further reduce their automation exposure by one that changes in automation risk between and is with the risk level in and of the of for the estimates show that individuals in are more to than in is with a in the risk of changes in automation risk and a in the risk of upward changes to the of individual attributes as and skills may important of job mobility. Because are with estimates are to and the estimates by the in the the not in the of by The results show that working in is with a in the of reducing automation risk by to less in the the the effect on the of upward changes and The in and for large urban but the results The of to jobs with automation risk is higher in to small and working in or other large urban regions not the of upward changes in automation in that workers in are more to reduce their automation through However, as in the a large of job changes are that take place between with more or less This the of to job mobility or career upgrades that in automation the is to this at we by the with a for risk at changes that this we can examine working in other large urban career The was to the of risk changes which the of in the as to a job automation risk is at to a job risk is at and that the results may to the of the we the other but to the estimates for in the two of that workers in are more to occupation that in changes in automation in is with a in the of automation at to large urban have a much on career upgrades which in the that allows for and This that working in the of reducing automation risk by at and the of automation not the of job mobility is higher in cities, but upward job mobility that individual to a there is evidence of a between the workers of the largest Swedish and in other of the in metropolitan are more to reduce their automation exposure through mobility. the models are of worker and the of to this can to the career or effect of metropolitan that data on employer-employee we can between two of job mobility through which automation risk can The is inter-firm mobility the workers across and the is mobility are or to job the in large labour markets more opportunities to reduce automation risk through inter-firm mobility, the between the of the labour and job mobility is less that workers in cities to job because large urban markets to in a of jobs and learn more quickly about their job with in cities, upward mobility more important as a of career that to automation. is not in the of the in urban their in to the other large urban labour markets to by workers at and are may new over and one of their existing the role of and inter-firm career upgrades in the of individual automation risk we the of the to a job with automation risk the to a job with automation risk in to a job with higher automation and We not between and mobility in case of automation risk because job changes that automation risk are in the a for lead to estimates & that the of potential the influences the of between we the of as a further job mobility in large cities is by the to and to new the of upgrades still higher in cities for the local of in of show that workers in are more to reduce their automation risk by and occupation at the in the largest the of inter-firm upgrades to the base by while effect on the of upgrades is results are the that at changes in automation risk According to this the of a large upgrade by while the of large upgrades by We can that risk through and mobility, the of career is mobility. we look at changes that to have a effect on mobility the the between the on upward and mobility in the is with while for the urban the is with the in the between the on and mobility is not The that upward mobility is more in that urban career opportunities for their and over new for higher Our that career upgrades are more in is in with the results of and However, this the of automation et found that to the of technologies by the tasks for their workers to avoid labour replacing technologies urban job mobility more in workers with different working in large to a sample estimates differences between worker In the of in the of cities is to groups for which automation exposure is is that are less than to the career opportunities by cities due to to differences in and labour & and that there are differences in between and which the to which and can career upgrades in are more to with automation risk (e.g., and different of are to in opportunities for career is career mobility is by the of human the of and to upward mobility in in through economic et al., & mobility at the lead to by for a of In the of and labour one that cities less of a to reducing automation risk is an important by which the career effect of cities across the of the In workers with have to and in an labour This potential for (MacKinnon, is because can apply to a of have to employment opportunities with the potential for career and therefore have a higher to low-risk jobs compared to their low-skilled to on urban that for workers the urban with job et al., This suggests that the career effect of cities is to mostly to workers. We examine effect across groups by the in for We the because changes that are for of the for individual as individual are models in to of the of in show differences in the of Separate estimates for and male workers the that can take more of their for the of reducing automation risk is by in for is by for large urban have a effect Our results for are in with the in the that upward mobility is of their the models is that estimates the of or other personal labour are to their human in the due to the of human the worker is in the of as local to employment in field & a automation risk may at a their workers not face can more enjoy the career of the of on the role of in career opportunities by large labour low-skilled workers than we that working in cities the of changes in automation risk and the of upward changes by the in the case of workers with are with a in the of changes in automation risk to the base of results that more workers can take of the career opportunities by cities to a less urban workers may have career in cities as low-skilled jobs in cities have automation is for less workers to the potential labour threat by and the of in small urban are as more exposed to the threat of by automation, because regions to more on skills and this is to in the (Frank et al., 2018). In this paper we and for large cities offer career opportunities which the of automation This was by of matched employer-employee data covering a 10% random sample of the Swedish labour the Our evidence that the of jobs is higher in regions and that the has over the the In that results of Crowley et that automation exposure with the of of the level of of we can show that the groups of workers at risk of automation across the regional the of workers in in are different of and workers, a share of workers are at risk the In that the spatial of labour that different groups of workers are at risk of in different of economic the that new tasks to in large cities with the higher potential for and learning the of new (Duranton & Puga, we can that spatial differences for groups of workers to to the of the labour markets local are over the of new jobs have been in the largest labour markets in Sweden & and the jobs in regions are as in the results this that to regions is the to because the of a job to in a is higher related jobs are et al., et al., we that the of new jobs in the are more to in large cities, the employment share of is The of this is that workers in large cities have of a non-automatable job as Our empirical results show that working in cities have a and effect on the of changes in automation risk for a of individual attributes occupation and labour However, of urban and on the of and workers for jobs that in their that the of reducing the risk of displacement. for example, the of to or labour face are less to the career effect of In this the workers are but and workers less the career cities The of further the of groups and therefore the of effect between worker how career in cities in different of the is a that further and empirical therefore is for further to technology-driven labour have to in to for different groups of workers. that cities (e.g., and or to but for the In the case of more workers, the on conditions for job mobility (e.g., by and on the of low-risk about of jobs in the face a risk of by in the as by and a to on the that this not across but This for and is to the with the between and and on the one and for a new on the other et al., 2018). The and the at the of in and the of the for in and was by the and has the

Topics & Concepts