Are old regions less attractive? Interregional labour migration in a context of population ageing

Resumen

Regional demographic change is often conceptualized as a circular process, where out-migration continuously worsens conditions of population ageing and shrinkage. Thus, if migration acts as a consequence as well as cause of ageing, migration patterns should be influenced by the age structure of origin and destination regions. This paper analyses individual-level migration decisions of full-time employees across 326 German regions between 1997 and 2013 using binary choice models. The results show that individuals are more likely to migrate out of and less likely to migrate towards ageing regions. Moreover, the identified patterns are consistent with age-selective migration reinforcing ageing processes and polarization of demographic structure. El cambio demográfico regional se suele conceptualizar como un proceso circular, en el que la emigración empeora continuamente las condiciones de envejecimiento y reducción de la población. Por lo tanto, si la migración actúa tanto como consecuencia como causa del envejecimiento, los patrones de migración deberían estar influenciados por la estructura de edad de las regiones de origen y destino. Este artículo analiza las decisiones de migración a nivel individual de empleados a tiempo completo en 326 regiones alemanas entre 1997 y 2013 mediante la utilización de modelos de elección binaria. Los resultados muestran que las personas son más propensas a emigrar fuera de las regiones envejecidas y menos a emigrar hacia ellas. Además, los patrones identificados están en consonancia con la migración selectiva por edades que refuerza los procesos de envejecimiento y la polarización de la estructura demográfica. 地域の人口統計額的な変化は循環的なプロセスとして概念化されることが多いが、地域の外への移住は高齢化と人口縮小の問題を悪化させ続ける。すなわち、移住が高齢化の結果でも原因でもあるとすると、移住パターンはその出発地と目的地の住民の年齢構成の影響を受けるはずである。本稿では、1997~2013年のドイツの326地域を対象に、二値選択モデルを用いて、個人レベルでのフルタイムで働く労働者の移住に関する意思決定を分析した。結果から、高齢化が進む地域から移動する確率が高く、高齢化が進む地域に移動する確率が低いことが示された。さらに、確認された移住のパターンは、高齢化と人口統計学的構成の二極化に拍車をかける年齢選択的な移住と一致するものである。 Migration affects the size and composition of population and is, therefore, a fundamental component of demographic shifts. With fertility rates falling below replacement level in most industrial nations, migration has become the primary driver of population growth on a national and regional level. Moreover, since migration is often selective, for example, with respect to age, gender, or skill-level, it also shapes the composition of population and thus contributes to population ageing. However, although migration clearly affects demographic structure, regional demography could in turn influence migration behaviour. Despite the current and predicted future extent of population ageing in Europe and globally (Coleman, 2006), the question of how these changes in the age structure may affect regional migration patterns has so far received relatively little attention. Theoretically, population ageing may directly affect aggregate migration rates through shifts in the population composition: older regions have fewer individuals in the most mobile life stages. However, demographic change is also associated with changes in regional characteristics, which may influence a region's attractiveness for migrants on an individual level. Strongly ageing regions may simultaneously experience adverse economic and labour market conditions, which reinforces out-migration. Moreover, as public services and amenities are adapted to an older population, ageing regions may become less attractive for younger individuals specifically. The relationship between migration and population ageing may therefore be bi-directional as well as mutually reinforcing. This is reflected in conceptualizations of regional demographic change as a circular process (e.g., Bontje & Musterd, 2012; Hoekveld, 2012): regions with declining and ageing population may become continuously less attractive to migrants. The interplay between migration and population ageing could thus lead to a polarization of demographic structures. Demographic change has far-reaching socio-economic effects, for example, on economic growth and productivity, provision of public services, and social cohesion. In this sense, the potentially reinforcing roles of population ageing and migration are directly relevant for issues of regional development. However, despite progressing demographic change on a global scale, there is limited theoretical and empirical literature on how a region's demographic structure affects individual migration behaviour. Micro-level studies emphasize the relevance of an individual's age in migration decisions (e.g., Goetzke & Rave, 2013; Millington, 2000) but neglect the effect the regional age structure may have on individuals. Aggregate studies of migration often include the share of young population (e.g., Biagi et al., 2011; Hatton & Williamson, 2005; Rodríguez-Pose & Ketterer, 2012) but usually limit its discussion to that of a control variable. While the impact of cohort size on mobility has been analysed for example for the baby boomers (Plane, 1992; Plane & Rogerson, 1991), these studies focus purely on demographic effects and do not include regional characteristics. This paper therefore focuses on investigating the link between regional age structure and migration behaviour explicitly and on an individual level. Using German administrative labour force data, we test whether regional age structure affects employees' propensity to migrate as well as their destination choice. While descriptive in nature, the analysis allows testing whether the patterns of labour migration are consistent with migration and ageing as reinforcing or opposing processes, which is relevant in the discussion on policy responses to population ageing. With a median age of 45.9 in 2015, Germany ranks as the second-oldest country in the world (after Japan) (UN, 2019) making it a relevant case to better understand the impact of further ageing around the world. We analyse the effect of regional age structure on interregional migration decisions for employees in Germany between 1997 and 2013. Focusing on interregional migration is particularly interesting because it allows capturing the effects on both sending and receiving regions. In contrast to international migration, which represents an addition of external population, interregional migration flows illustrate the growth of some regions at the cost of depopulation in others. Moreover, because interregional migrants are more similar to the resident population, institutional factors that may constrict international migration (language, transferability of degrees) do not apply. This paper proceeds as follows. First, the theoretical and empirical background is summarized. Second, the data and methodology are described. Third, the results are presented for migration propensity and destination choice before, fourth, discussing them and offering concluding remarks. Demographic change in industrial countries is characterized by falling fertility rates and rising life expectancy. Jointly, these effects slow population growth and lead to ageing. In Germany, where demographic change is relatively far progressed, the number of deaths has exceeded births since 1973 (Destatis, 2015), indicating that, for almost 50 years now, population growth has been exclusively due to international migration. On a regional scale, the negative natural balance means that shrinking German regions do not necessarily experience out-migration—some regions shrink despite positive net migration. Simultaneously, the population is ageing: Between 1990 and 2013, the share of population above 65 years to and the median age to years (Destatis, Germany, of the population changes which is in international migration the in Germany for most of the to the net migration Germany to the is at although migration has been migration is not relevant due to its but also because of its the are relatively young and often & In with the which that individuals are by (e.g., et al., migrants are to be on et al., and & the driver of population growth in countries a and with its migration is relevant of regional development. the net effect of is a in the migration clearly a of with on both sending and receiving regions al., sending migration of young and individuals is a of but also that could most to or negative of demographic change are more likely to In more population (e.g., the or with may not be to migrate and thus in shrinking and ageing which may as et al., This with that interregional migration not regional economic (e.g., & and may lead to further migrants are et al., In this sense, if population ageing reinforces migration of young and demographic change could regional in economic and lead to polarization of age across regions. This is both because young individuals may be more likely to and because their further reinforces the process of ageing. is a in of migration, on an individual level. In of and migration is an by its and if the in the destination less the cost of the of in the origin and of migration may on the between origin and destination (e.g., the younger individual have a of and thus the of for a migration may with age, for example, because older individuals have a of social or or of on as well as that, be more and more for a younger an older Migration propensity thus in the and with age, that migrants on younger of the regional age In this sense, the of ageing as a consequence of migration is if migrants are regions out-migration are to in the of migration as migration, this of migration the young and the is However, if we migration decisions to be by origin and destination characteristics, the of ageing to a consequence of migration that these regional are of demographic structure. of migration that out-migration of young population rising and falling in the origin thus a in migration if regions by the demographic structure of their population, shifts in the population age structure by out-migration of the young or due to changes in fertility and may be by changes in regional characteristics, which future migration behaviour. In the of of migration, which the relevance of amenities that may be by (e.g., et al., ageing could (e.g., through changes in regional become a relevant or a consequence of migration. between regional age structure and migration, of influence be population cohort effects, and effects because the age of on the age composition of the population & with a relatively age structure should less out-migration in because there are fewer individuals in the most mobile age for this is in the literature on the of migration flows (e.g., Hatton & Williamson, although may on & On a regional Rodríguez-Pose and illustrate that young regions relatively more out-migration. While composition effects may in it should be that on the aggregate individual level. may older regions but this not in an individual's migration In the literature on mobility across that cohort size could affect individual migration behaviour. on this of literature that labour and market to an in the of which are less The could thus the in migration rates for the 1992; Plane & Rogerson, 1991), although this may not in (e.g., & Moreover, if also labour and market for could mobility for age & In this sense, a relatively regional age structure be to migration propensity migration may be influenced by regional age structure due to the and of migration with individual age, migration may also on an individual's life The literature on mobility labour market factors as both migration propensity and destination (e.g., & et al., older labour market in and amenities become more relevant in migration patterns & & 1992; & Millington, et al., Goetzke and show that age of and amenities or with particularly to and younger more to Thus, there may be across regions on the age for example, on labour market factors Moreover, age may also have for of for example, in of public services the provision of public services age structure, this could affect a region's attractiveness for age example, provision may be a for and thus in decisions for these age but not for older In and emphasize the relevance of for migration the we in how migration decisions with regional age structure for younger and older individuals. the relevance of ageing for migration is thus and in the studies test the of age structure as an in migration. is an analysis by et show that regions with of population may net migration but this a aggregate analysis with control This paper contributes to the literature by investigating whether an ageing population may influence regional attractiveness for migration. In contrast to et and most literature on migration this analyses individual-level should be that we do not a relationship between age structure and migration decisions but focus on the effects of regional ageing. the regional age structure could the of amenities that of individuals migration. using individual level data of of age structure that in aggregate analyses of migration Moreover, individual data allows not whether migration behaviour with the regional age composition but also whether this effect by may out regions where are as for migrants in et al., that migration reinforces population ageing, which may empirical results of demographic polarization (e.g., & et al., and for future polarization of age structures. This is an of the between ageing and migration that has so far not received in the literature but is explicitly in this The individual-level migration data is the of by the of the German at the of This represents a of administrative labour force in and is of employees by social the and public are not in this data, it represents the most labour market for to the data structure and because is the of the is limited to full-time employees to are as or for by the of full-time that include a number of full-time are the of these the the the the current the the individual is a We also age, gender, level and for their current to The almost for individuals. Between 1997 and 2013, interregional migration are to of the The of analysis is an of 326 as in the regions an of German regions where regions with less are The individual data is with regional the Regional Germany and the of the for and Regional The of regional age structure, is the share of population above 65 years The results are to using labour market regions of regions and of ageing age of the labour force and the for the and are in the We analyse individual-level labour migration in a for migration propensity and destination choice. of this is that it the whether to migrate may be with the destination choice. the literature on individual migration and as it is and & we are in whether migration patterns are by regional age and not by the fundamental migration we this and the of regional age structure for propensity to migrate and for destination choice. also contributes to the regional age structure, there is which is by the regional in by for effects, effects are although the has a structure, the results are using the with by We control for individuals by as a However, individual characteristics, a effects individuals are at as both a and which is we the as as an to the effects between the individual and regional are to test whether the effect of the regional age structure across age and a this paper the question whether regional age structure is to destination this the is to migrants has a choice of but the origin and a the that the destination we the to for First, in the may the analysis because may be more that their is more that of a Second, the choice has as the destination Thus, the choice for the migration analysis of of We therefore et in in to the and the of as individual is We by with the of and the of regional a of the are by which that individuals are to on current regional characteristics. The not individual directly because are across However, by individual with regional we test whether the effect of the regional age structure with an individual's age or should be that on the of of the of should not on in the choice test this the could be for of choice as et testing is for with to size on choice may also not be in empirical & Moreover, and that, between the as a of of with a share of population are in the and of Germany and these are also the regions a of ageing between 1997 and 2013 the in the share of population is in German regions to in the However, it should be that population ageing in almost German regions. regions show a in the share of population above 65 and the extent of the individual migration decisions in the it the net migration for the by population patterns are regions in the and of Germany show for and its However, interregional migration are not limited to the of While for example, around and experience positive net migration, more regions in the also experience out-migration. Moreover, the of migration are relatively the is with a median of The for interregional migration flows the German Migration The for the of individual are to in aggregate data and median Thus, the migration of employees in the data in to aggregate migration The for the share of population is positive across This that, individuals in regions with an older age structure are more likely to in younger regions. The effect of a in population is a in the of labour migration. The individual-level are in with and the empirical individual's age is negative for migration propensity with a positive and full-time employees are less likely to migrate Using the as individuals with and are more likely to at the current is which is consistent with to The for migration is, as positive and in which that migration is more likely if an individual has Regional population size and are negative for migration indicating that individuals are less likely to regions and with the negative for as adverse labour market conditions should as a However, the of individuals are and for migration be with in both the origin and destination in the origin may that the individual is thus the to of the is associated with and the share of employees with of out-migration. In with the literature on amenities (e.g., Rodríguez-Pose & Ketterer, we that individuals are less likely to regions with In individuals in regions with a share of population are more likely to are also more likely to and German regions and German although the not migration. The results show that individuals in older regions are more likely to whether these individual decisions are consistent with or migration, and include with the these not affect the or of the and the share The between the share of population and individual age, is individuals in older regions are more likely to the effect of the regional age structure with an individual's The effects of the share of population at individual are in the of individuals younger an in the share of population is to migration In for 50 and the change be to the of migration. This is consistent with age-selective migration and where young individuals are more likely to ageing regions older individuals are more likely to The results thus the that individual migration patterns are reinforcing regional ageing processes in German regions. the results of share and The are negative and the as The effects in the of illustrate that an in the share of population is to migration propensity for with but it for with the Thus, reinforcing the out-migration of ageing to out-migration of with the This the that population ageing could also and that it may of the effect of age structure on the of we turn to where migrants the results of destination choice for migrants in Germany using of The of share of population above 65 is for years with similar across The of as a destination with the share of regions with an older age structure are less likely to be younger we understand the of as a of regional attractiveness for this that, older regions are less attractive younger the relevance of the regional age structure, the results in also show of regional of migration, is The for population and the number of are which migration towards regions with in that we analyse the labour market not as a the are or positive across the In the share and the share of in the are to be as is the share of population the to the of population a migration this that with more international are less as destination by labour migrants and not necessarily that international and interregional migrants regions. The results show that, for employees are not more likely to to regions. the in the of Germany is Thus, regions in Germany are more likely to be as destination regions German regions. should be that this may be due to mobility mobility and not patterns of migration the to the on an aggregate level. it that, for regional and individual to the German regions are not less attractive in the the and testing for in the While the the with individual age is positive at of in of the negative and positive that the negative effect of age structure on destination choice with an individual's an in the share of population above 65 years in is associated with a in the of a if the is relatively further this the of show the and for the share of population for on individual results individuals younger 50 destination choice is associated with the regional While the is also negative for older individuals for some it is most of the results are in with of individuals by age, where younger migrants are towards regions. We therefore do not of the that migration could balance age across but for patterns of polarization of age structures. The results for level are less and clearly show changes the with individual are indicating that migrants with or are relatively more likely to older destination regions migrants with the effect of regional age structure on the of destination choice is in more for more individuals. In this sense, we some of migration on level. the effect of ageing not between or the migrants more in the years of the migrants relatively more to age structure in The that more individuals are less likely to regions with may towards an between population ageing and regional to This paper the relationship between population ageing and migration patterns on an individual level by employees' propensity to migrate and destination choice regions with demographic structures. for individual and regional characteristics, the share of population of the origin is to the of out-migration. the destination choice that with a share of population are less likely to as We further whether the results are consistent with patterns of and migration. between individual age and share of population above 65 years for both migration propensity and destination choice. This that younger individuals may be more to regional age structure both in to and in a destination The for effects in migration patterns is more with to be more likely to older regions which the of out-migration. the could be to the of in In the that individuals are not more likely to ageing regions is as it may that the in ageing regions is less However, there is some that in years individuals may be more to regional age structure in their destination choice. Thus, population ageing may not be associated with a of the results in individuals to ageing regions. the of population ageing in labour migration, the results First, population ageing is a relevant of migration for a of individual and regional characteristics, the share of population in we do not effects of age structure on an individual's to the results that population ageing is not a consequence of migration but may in turn be relevant in migration for example, through across regions with Second, the of the that, if migration to the processes of demographic The presented results show that migration flows relatively to relatively young regions. Moreover, the identified effects that the migration flows may be on age, with younger individuals more likely to older regions. this is older regions do not an of population but an of the younger age which the age of the we do not the net effect of the migration flows the illustrate that age-selective migration in Germany may lead to a polarization of age consistent with regional demographic change as a circular to the presented analysis should be the of the data and the focus on interregional migration. the administrative labour force data is of the allows to the of thus the to While a share of the population, it should be that individuals are not in the labour force or the may a of factors to whether and where to The of the analysis is that international migration is not a of the share of population as a control variable. the focus to labour migration to of migrants and and thus a of factors that be relevant for international migration. also and on migrants and their at a scale, which is to for international migrants. However, migration patterns German interregional migration as if it a In international of a in regional demographic structure, the primary of population With respect to the labour migration patterns the relevance of discussion for migration more on whether international regional decisions interregional we by or country of the is of regional labour and therefore also employees with a Moreover, analyses and for international in German and a negative impact of the share of population consistent with the results presented Thus, the empirical that both international and interregional migrants may younger older regions. On a national scale, not be to or population ageing (Coleman, but this further that population growth through may not these demographic at a regional In this sense, regions to the demographic effect of may more of and migrants. The results of this analysis on policy making to the of demographic change on a regional level. In of the patterns it that labour migration and population ageing are reinforcing processes to polarization of demographic structure. changes in of economic and social to some theoretical population ageing not to thus the question of whether policy should and influence migration patterns With progressing demographic change on a global scale, countries demographic of ageing and shrinking similar to the conditions in the of policy to regional attractiveness in ageing and shrinking particularly for age and most to and the impact of demographic polarization on regional more a relevant for future to for their and and the of the as well as and for destination choice of results ageing by for destination choice of The is not for the or of by the should be to the for the