Rural Scotland - trajectories of young people and young adults: report

Chapter 2: The demographic context: age profiles of mobilities

This chapter aims to provide some demographic context for the policy and empirical evidence discussed elsewhere in this report. It charts some of the diversity of rural migration patterns by age across different parts of the country and begins to show how variations in age/gender-specific patterns of migration affect the changing sizes and age profiles of populations in particular places over time.

2.1. Sources of data

Unlike in much of Northern Europe, where there is a legal requirement to report all changes of address and where the resulting statistics are regularly made available for analysis of migration flows, in Scotland, NRS, and its predecessor the General Register Office for Scotland (GRO(S)), have for many decades used a variety of survey and administrative sources (including school and GP registrations) to make their own annual estimates of in- and out-moves, down to quite small geographies. It is these, along with the much more robust vital registration statistics on births and deaths, that they then use each year to produce their published estimates by age and sex, of the populations of the top-level administrative areas (local authority/Health Board (LA/HB)), and also of populations down to ‘datazones’ (most of which have between 500 and 1,200 people).^[6]

Unfortunately for our present purposes, however, NRS has only ever published a limited range of estimates of their underlying migration flows. These are normally confined to LA/HB geographies. As a result, in the areas being considered here, we only have separate NRS migration flow data for the three island authorities. In addition, the published data have normally been limited to in-, out- and net movements for the total populations of each LA/HB, plus net moves for five-year age-groups broken down by sex, and net figures by single year-of-age, but these are for ‘all persons’ only, with no breakdown by sex or any other variable.

So, although, as we shall show below, there must have been major difference in migration patterns between different sub-populations within each LA, we cannot use the NRS published migration data to get direct insights into these variations. Of particular relevance to this report, we cannot get even the limited published statistics for just the more rural parts of Highland, Argyll and Bute, and Dumfries and Galloway – and it is clear from what we show below that the LA-level statistics for these authorities, dominated numerically as they are by their urban and suburban areas, in no way reflect the migration patterns of their most rural areas.

It is for these reasons that for our more localised analysis we have had to fall back on inference from changes in the estimated age profiles of populations over time, and on comparisons between the age profiles of places at the 2011 and 2022 Censuses.

In addition, our ability to discuss the most recent years is affected by the impact of Covid-19 on urban to rural movements, some of which may turn out to be more than just temporary blips if they are continued into new patterns of learning and working in the future.

Nevertheless, despite these limitations, we believe useful broad conclusions can be reached.

2.2. Island local authorities

Until now, separate in-, out- and net-migration estimates have only been made public once, in a special chapter on population diversities in the Registrar General’s Annual Review for 2006 (Anderson and Wright, 2007, 77-8; Anderson and Roughley, 2018, 181-4). These figures show the average of four years of data for in- out- and net moves between 2002 and 2006 for each of the 32 LAs (Anderson and Wright, 2007).^[7] For this report, however, NRS has made unpublished data available to the Expert Advisory Group which show a similar set of figures across the last pre-Covid-19 years of 2017-19 and also for 2020-22.

In the Mid-2021 Population Estimates, Scotland (p.12), NRS identified clearly unusual patterns of migration for the 2020-2021 year. They point out that the estimates were, as usual, in part derived from changes in GP registrations; and that this may have accentuated a rise in ‘migrations’ through people moving from cities to new or second homes in rural areas to minimise the impacts of Covid-19 and registering with a local GP to ensure access to treatment and vaccinations. NRS also notes that some students remained at their parents’ rural addresses rather than moving to an urban place of study.

While comparison of the 2020-22 and 2017-19 figures in the unpublished migration data supplied to the Expert Advisory Group on Migration and Population (EAG) by NRS shows a general overall similarity in patterns of movement across the two periods, it does suggest some noticeable differences. In 2020-22 both gross and net late-teenage outflow figures were lower compared with 2017-19, most noticeably in the Western Isles. In-movements at older ages were also markedly higher in 2020-22 in Shetland, though less so in the Western Isles and with no significant differences apparent in the Orkney figures.

Faced with a choice of which data to use, to avoid our more detailed analysis of estimates-based migration being influenced by the unusual pattern of late teenage migration in 2020-22, we have chosen in this section to focus mainly on 2017-19 data, in preference to the most recent figures.

Figures 1, 2 and 3 show migration-by-age patterns for the three island LAs for these years and reveal some interesting differences between them. Crucially, they also show how much having separate in- and out-migration estimates can add to our understandings, compared with just the usual net figures.

In all three island LAs, there was both in- and out-movement at almost all ages in 2017-19. That a small net inflow of children was occurring everywhere (especially in Orkney) is presumably a sign of family migration; however, by age 16 a small net outflow is apparent in all three LAs. In 2017-19, the first significant gross and net outflows only started at age 18, when gross flows of 10 per cent of the age group left Shetland, 11 per cent left Orkney and 16 per cent left the Western Isles. Age 19 then saw the peak gross and net outflows in all three island groups, the gross out-flows involving a quarter of the age group in the Western Isles, 23 per cent in Shetland and 19 per cent in Orkney (and the nets 18, 19 and 14 per cent respectively). Strong outflow continued at age 20 (still 19 per cent in the Western Isles), and even though there was more inflow, the Western Isles still lost another 11 per cent of the age group at this age.

Thereafter in the Western Isles, gross out-flows remained above or around a tenth of the remaining age-group right through to age 25. But from 24 onwards these were at least roughly matched by in-flows, so, averaging across all years, a very small positive balance (around one per cent per year) occurred right through to age 65. The Shetland pattern was rather similar, with a slightly lower outflow pattern also being accompanied by lower inflows, but there the sum of the post-age 30 net figures is balanced at best, and from 45 marginally negative rather than positive. By contrast, Orkney show a clearly different pattern, with numbers of out-movers down to between three and six per cent of those aged from 27 to 45 and equalled or exceeded at every age by in-movers. Orkney thus experienced net inflows of between two and four per cent in almost every year from ages 30 through to 38 (and with never another negative year through to age 65).

Comparison of the 2017-19 data with the figures published in the 2006 Report for 2002-06 appear to show that some important changes in age-patterns of migration took place in the first two decades of the twenty-first century (Anderson and Wright, 2007; compare Figures 4 to 6 with Figures 1 to 3). In the 2017-19 data, in all three island LAs (and especially in Orkney) there was only a modest increase in gross or net outflows at age 17 compared with younger ages, with significant rises then at 18 and clear peaks only at 19, with these peaks nowhere exceeding 25 per cent of the age group. In 2002-06, by contrast, there was a clear rise in out-movements among those aged 17, with around one in ten in all three areas of young people leaving even at this age. There was then a big jump at 18, with departures of between 26 and 28 per cent of the age group in all three areas, and further very high levels of outflow at age 19, with the Western Isles actually peaking at this age, having out-movement by 29 percent of the age group; this compared with out-flows of 26 per cent in Shetland and 23 per cent in Orkney. In the Western Isles, 17 per cent left at age 20, and departures of more than ten per cent continued there through to age 24 and at more than five per cent to age 31. Out-flows of more than five per cent also continued to the late-twenties or early 30s in the other two island groups.

Importantly, however, in 2002-06 these large out-movements were accompanied in all three island groups by substantial in-moves at the same set of ages: more than five per cent in all three LAs at age 18, rising to 13 per cent peaks at ages 21 and 22 in Orkney, and at ages 22 and 23 in Shetland; in the Western Isles they were above ten per cent at every age between 20 and 26, where the peak inflow was 16 per cent at age 22. The result was that in all three LAs the net flows were clearly positive right through the mid- to later-twenties of people’s lives, and they remained modestly so in most years right through to age 60 and beyond in Orkney and the Western Isles, though no better than a rough balance in Shetland.

2.2.1. Impact of migration on population change, 2002-06 and 2017-2019

Figures 1 to 6 show how the patterns of migration have influenced populations in the three island authorities over the two most recent decades. They particularly point up the large numbers of late teenagers and people in their early twenties who left at all periods (proportionately larger in each case than from any non-island LA except Perth and Kinross in these years). But a key question for anyone interested in overall population change is to what extent out-movers at these ages have been replaced by in-movers at other ages, and what impact this might have had on the age profiles of the populations.

Table 1, now using numbers not percentages for clarity, shows some quite major contrasts between the three LAs both in the age profiles of migration and over time. The marked reduction between 2002-06 and 2017-19 in net outflows at ages 16-17 comes through very clearly in all three LAs, with the main concentration of net loss in both sets of years being at ages 18 to 20. In the 2002-06 data, net flows at older ages then became, often quite strongly, positive in all three LAs at almost every age up to the mid-forties, and for most ages thereafter. But the 2017-19 data suggest that a more differentiated was occurring in those years compared with 2002-06. In Orkney, rather like in the years 2002-06, the 2017-19 net flows were positive at almost every age after 21 and this resulted in a significant cumulative positive flow in people’s twenties, thirties and thereafter. In the Western Isles, by contrast, flows only went clearly positive in most years after the age of 28, and in Shetland, taking one year with another, the net flow pattern was clearly outwards for the whole of the rest of people’s lives.

The results are that in 2017-19 only Orkney had enough adult inflows before age 45 to counterbalance its out-migration at ages 16 to 20. Only Shetland, however, had overall net losses by migration in both 2002-06 and 2017-19, and this was in large part because it continued to lose out-migrants at older ages, while both Orkney and the Western Isles attracted significant inflows of middle-aged and older people up to retirement age and beyond – but, in the latter case in particular at some detriment to their overall age profile.

However, in all three island groups one other, often under-emphasised, factor was significant as a positive contributor to migrant flows: the arrival of in-migrant children (presumably as part of parental inflows from their later twenties onwards). In-flows of children of 15 and under provided a third of the overall net inflow into Orkney in 2017-19, and two thirds of the flow into the Western Isles – and, in doing so, also made an, admittedly modest, contribution to the balance of the age profiles of the islands. But what appear to have been markedly lower numbers of in-migrant children as a share of the population of Shetland, has been a further factor feeding its population decline.

2.2.2. Further exploration, using age-cohort depletion

No more sub-national data on separate in- out- and net-flow migration patterns by single year of age are currently available from NRS. Net flows, however, can be inferred by comparing population age distributions over two or more periods of time, using data from each year’s Mid-year Population Estimates, Scotland. This is done by using ‘age-cohort depletion’, which is based on counts of the number of people at each year of age who either ‘disappeared from’ or ‘appeared in’ that age group over a chosen period of time.^[8] So, if we are estimating net flow numbers between any two years, we subtract the number in the earlier year-of-age from the number in that same birth cohort (now one year older) in the same place a year later. An example of this kind of one-year ‘depletion’ would be the difference between the number of people counted at age 20 in 2022, and the number who would have been ‘expected to be’ present in that year if there had been no migration or deaths of people who were aged 19 in 2021. It should be noted that the figures cited below do therefore include deaths, but up to the early 50’s these are minimal, and do not significantly affect our results. One advantage of these data is that they allow us to undertake our analysis for males and females separately, and this is done below.

As noted above, Covid-related behaviour will have affected the data for 2020 and 2021, and quite probably on a limited scale for 2022. We therefore show in Figures 7, 8 and 9 analysis of net migrations, for the island authorities, by sex, for the years 2016 to 2019.

The graphs show the mean net percentage estimated gain or loss in numbers present across the years 2016 to 2019, for each year of age from the first to the 49th, for females and males separately.

At most ages, there were no major differences in net migration patterns between women and men, but with two marked exceptions.^[9] First, in the late teens, there was everywhere a clear excess of female net outflows over male; this excess was modest in Orkney but very marked in the Western Isles and especially in Shetland, where it seems likely to be the result of the large number of well-paid (but not requiring qualifications) jobs for young men in resource-based industries. Second, in the Western Isles, the net outflow for men was almost as large at age 20 as at age 19 (but only half as big across the two age groups among women), and in Shetland the net outflows suggest that, while significant numbers of women but fewer men left the islands at age 20, net flows of women were then positive at each age through to 27, while flows for men remained negative in these years. Just what caused these relatively small differences in migration patterns by age and gender must await further research.

2.3. Islands and mainland rural areas; use of Census data

As noted above, NRS has not published any recent data which would enable us to identify year-by-year patterns even of net migration movements below LA level. This gap is important, because there is huge variation in settlement patterns within LAs. For example, taking just the three island LAs that were examined above, more than two fifths of the population of the Orkney LA live in the town of Kirkwall, more than a third of people living in the Shetland LA are resident in Kirkwall, and more than a quarter of the Western Isles LA’s population live in Stornoway.^[10] All three towns have a wide range of good standard commercial, administrative, medical, educational and leisure infrastructure, providing services not only to direct residents but also to more immediately surrounding populations, many of whom commute into the towns regularly for work, shopping and/or education. As a result, the social and economic situation of both these urban residents and of those living nearby is dramatically different from that of people living in the more distant and sparsely populated rural places, whether on the mainland of Scotland, the largest islands, or the smaller inhabited islands.

It is therefore important that separate data be made available for analysis of migration in places further from these significant towns – and the same applies to places nearer and further from towns like Oban, Fort William, Wick and Thurso, which play a similar role in their parts of otherwise rural west Argyll and north Highland. Though not ideal, NRS has published Census data down to island/island group and civil parish (CP) levels. These at least allow us to compare single year-of-age counts at the 2011 census with the counts of the number of people eleven years older in the same spatial units in 2022. And this makes it possible to get some insights into the net cumulative impacts of migration flows over the previous eleven years, as experienced by each successive year-of-age; it also provides a useful perspective on the effect of migration on the communities concerned.

Figures 10 to 13 show the very marked contrasts in the patterns of these movements, across different sub-sets within the same broad spatial areas, between the 2011 and 2022 censuses. The underlying data are, for each place and single year-of-age, the differences in numbers counted in 2022 and numbers counted in 2011 who were eleven years younger at that time. The figures thus show cumulative patterns of net gains and losses over the previous eleven years. To reduce uncertainty in the estimates due to random variation, the plots in the graphs are three-year moving averages across ages.

Figure 10 shows that in all three Orkney areas (especially in the five northern islands most distant from Kirkwall, less in the rural areas of West Mainland, and only minimally in Kirkwall), the numbers of children under the age of 16 were higher than expected in 2022 compared with numbers in 2011 eleven years earlier. This is clear evidence of some net family in-movement over the eleven-year period. At older age groups, by contrast, there had been very significant net out-movement from the outlying northern islands, particularly marked among those in their mid-late twenties in 2022, more than 55 per cent of whom had disappeared since 2011. Losses among those living on rural West Mainland were somewhat lower, and there had been some quite significant in-movement there among those in their later thirties/early forties in 2022. In Kirkwall, however, inflows at best did little more than equal outflows for most age groups through to age 45.

A rather similar pattern of contrasts between town, surrounding parishes, and more distant parishes can be seen in Figure 11 for Oban and the west of Argyll.^[11] Oban itself had had tiny net inflows among children and for those who were reaching their early 30s in 2022, but otherwise modest patterns of net loss. By contrast, there had been substantial in-movements into Oban’s surrounding parishes, of children and even quite young middle-aged adults. These are all areas from which large numbers of people commute to Oban for work and/or depend on its significant shopping, educational and other infrastructure. Further south, in rural Kintyre, the picture was very different. In all but one of these parishes, populations fell between the two censuses, with huge outflows of people in their early and mid-twenties, and very modest in-flows thereafter.

Figure 12 shows the differing migration profiles of three of the largest non-LA island groups. Mull, and much more, Skye, had marked growth in population between 2011 and 2022. Nevertheless, both had quite marked net outflows among those in their teens and early twenties; in Mull this even involved some children of secondary school age, probably because of moves (some on a week-time only basis) to mainland secondary education; this is a pattern of education also seen in other Argyll islands. However, Mull experienced net inflow among people from their late twenties (peaking at over 60 per cent around the age of 30), and Skye was similar, with inflow rates having approached 40 per cent among those around the age of forty in 2022. The contrasts with Bute are marked, with net losses there apparent at almost every year-of-age up to the late thirties.

Finally, Figure 13, shows something of the marked contrasts between different mainland areas. All the ‘North-west coast’ parishes (those located southwards from Durness to Lochbroom) had population decline between 2011 and 2022, and all had experienced very severe net outflows (peaking at over 50 per cent at all ages among those from 13 to 30 in 2022), with only modest reversal of flows among those who were by then in their thirties and early forties. By contrast, the coastal parishes further south (from Gairloch down to Morvern) all experienced population growth in the same period, and although their teenage/young adult net migration outflows had been on a similar scale to those further north (but for fewer years and slightly less extreme), they clearly had had a very different pattern of net family and middle-aged inflows, at well above 40 per cent among those in their mid-30s and over in 2022. The contrast with the predominantly agriculture-based rural parts of the Machars area of Wigtownshire (the parishes south of Newton Stewart) could not be more marked. Only one of these parishes had population increase between 2011 and 2022, and Figure 13 shows this area as having had significant and continuing net outflows at all ages through to the early forties.

2.4. Overall impact

The graphs and data in the previous section show a widespread, but not universal, pattern of net inflows among children (and therefore in most cases of their parents), followed almost everywhere by net outflows, as teenagers (in a few cases clearly a significant majority) left home parishes or islands for education and or training elsewhere. In this they were actually joining people from similar age groups right across Scotland, because, for example, in 2016-19 only seven local authority areas, all with significant higher education provision, had net inflows in the 17-20 age groups – and, of course, many teenagers whose homes were in one of them actually went elsewhere for their studies/training. This is an important point, too often overlooked in writings focused just on rural areas.

By contrast with these earlier years of life, there has been huge variation in the direction and scale of net migration flows at older age groups. This raises two further key questions for the present and future demography of these areas; first, to what extent have switches at older ages from net outflow to net inflow (where they have occurred) maintained total population sizes today, and second, what impact have they had on population age structures for the future. There will be big differences in the consequences depending on whether the incomers are of childbearing age or bring pre-school or school-age children of with them, or are concentrated very much among older groups.^[12]

Table 2 attempts to throw some light on these two issues by summing net gains/losses over groups of ages, presented now as a proportion of the 2011 population (as opposed to of age groups as in Figures 10 to 13). It includes the places for which data are plotted in the graphs and some additions, notably in Shetland and the Western Isles.

Overall, the data suggest that total net in-migration by age groups involving children and those aged 31 to 65 only offset net outflows in the age group 17 to 30 in a small majority of the areas covered. Even more crucially, net total positive inflows before age 46 only occurred in Kirkwall, the surrounds of Oban, the rural parishes of Orkney’s West Mainland, Mull (by a large margin), Islay, and the west coast parishes between Gairloch and Morvern; Skye was just about in balance at these ages. In the other areas, therefore, even if there were total net migration inflows overall, they are unlikely to have improved the age profile of the population.

On the other hand, all of the urban areas included here, except Kirkwall, experienced overall population declines between 2011 and 2022; in particular, Stornoway and Lerwick had migration losses in all four age categories distinguished in the table. It is worthy of note however, that this pattern is not dissimilar from trends in many towns elsewhere in Scotland away from the cities of the Central Belt. Away from urban areas, very large numbers of in-migrants among those who were 46-65 in 2022 were particularly important in offsetting or nearly offsetting net outflows at earlier ages in Arran, Bute, the north-west mainland parishes, North Uist, and the distant Shetland and Orkney islands. Interestingly, Barra, the most distant of the Outer Hebrides from Stornoway, offset its very high late-teenage and early-twenties losses (presumably caused by the need to seek almost any advanced training/education off the island) by high levels of in-migration among people only a few years older, presumably stimulated by its strong economic growth from tourism, fish processing, and inshore fishing.

2.5. Conclusion

The above analysis highlights three important issues for research and policy development on mobilities of young to middle-age adults in rural Scotland.

The first is that, if we are to understand the quite complex interactions between in- and out-movements among people of these age-groups, it will be very important to find a way that NRS can allow researchers to obtain the best possible separate data on inflows and outflows, by age and preferably by sex, and for smaller geographies than just LAs. Such data are anyway needed because there is no such person as a net-migrant (Rogers, 1990).

The second point is that, as elsewhere in Scotland, high levels of outflow among rural teenagers and those in their early twenties are, and for good reasons must be expected to be, a normal part of very many individuals’ life courses. This suggests that attempts to discourage these outflows are not only likely to fail but could well lead to less qualified and more narrowly experienced labour forces in these areas, which would be an inhibiting factor on economic, cultural and social services development. More positively, it also suggests that policy interventions could aim more effectively to maximise the combined benefits of these processes of youth out-migration and middle-age return. Of particular interest would be policy development specifically focused on facilitating and encouraging in-moves among slightly older people (and not just returners), especially those in their late twenties to early forties who are most likely to bring or have children which will help balance population age structures.

The third point is to stress the key importance in both analysis and policy development of understanding the diversity of demographic patterns in different rural areas, and of their drivers. Across Scotland’s more than 90 inhabited maritime islands there are huge contrasts in size, recent patterns of growth/decline, age profiles, ease of access to major educational, medical, commercial and other infrastructures, opportunities for, or restrictions on, day-commuting to school or work, and underlying local economies and cultures. The same is true of different parts of the rural mainland. All these variations will significantly affect patterns of mobilities at all ages. Any notion that universally applicable policies can work to cover all these diversities is bound to fail. Instead (but bearing in mind that within this diversity there are similar economies and demographies, so few places are unique), the best way forward seems to be to develop a set of possible initiatives that can be tailored to carefully analysed, and particular, kinds of local needs. It is encouraging to see recent Scottish Government policy initiatives such as the establishment of Community Settlement Officer posts within a number of local authorities and support for community-based and/or community-led organisations, such as community councils moving in the direction of supporting place-based solutions (see Appendix 1).