Rural Migration

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Cities 130 (2022) 103850

Contents lists available at ScienceDirect

Cities
journal homepage: www.elsevier.com/locate/cities

Mover stayer winner loser


A study of income effects from rural migration
Lina Bjerke *, Charlotta Mellander
Jönköping International Business School, Box 1026 551 11, Jönköping, Sweden

A R T I C L E I N F O A B S T R A C T

JEL classification: Rural-urban migration theory suggests that one gains economically by moving from a rural to an urban area. The
R23 popular juxtaposing notion is: “If you stay, you lose.” But given the rapid increase in housing costs in bigger
J1 cities, are losses from staying in rural areas still necessarily the rule? If so, how big is the economic loss for those
O18
who stay in rural regions? Using Swedish micro data, we focus on the income effects of rural-urban migration
Keywords: among young individuals. We find that staying in a rural region is negatively related to an individual’s income
Rural-urban youth migration
levels, but the loss is mostly insignificant. After controlling for housing costs, it is even beneficial for many people
Income
Mover
to stay in their rural areas, except for the most highly educated individuals who can benefit by moving to an
Stayer urban area.

1. Introduction research also highlights the greater diversity of consumer services,


culture, and other experiences that create additional influential factors
“What if cities are no longer the land of opportunity for low-skilled prompting young adults to move from rural to urban areas (Florida,
workers? Dense cities have long promised higher wages, but now 2002; Glaeser et al., 2001). Current studies equally identify the role of
[this] is primarily true for workers with more education.” So stated a economic and social gains in rural-urban migration, though they are
recent New York Times article (Badger & Bui, 2019), which was based on based mostly on data in developing countries (e.g. Glaeser & Kohlhase,
a presentation by M.I.T. economist David Autor. His point is well taken. 2003; Lucas and Robert, 2004) or using a historical perspective of the
While urban areas have long experienced a large inflow of young, skilled data among Western nations (Gollin et al., 2016). Other empirical
individuals, rural areas have often been ‘left behind’ (Rodríguez-Pose, studies on the financial gains of rural-urban migration focus on other
2018). As far back as Marshall (1890), both theory and practice have contexts as well, but what seems to be consistently significant in the
shown how agglomeration effects from larger labor markets provide more developed parts of the world are the labor market benefits to be
higher wages. This wage premium has been a good reason for many to gained (Hansen & Niedomysl, 2009; Storper & Scott, 2009).
move to larger cities for a long time. The result for many rural areas has However, these gains are often moderated by housing costs. Several
been a depopulation of the educated, leaving an increased share of recent studies have examined the strong effects that migration into
elderly and a less skilled labor force. This work contributes to this field attractive urban areas has had on housing costs and the competitive
of research by asking the question whether the wage premium from dynamics of the housing market. Edlund et al. (2015) show how demand
larger cities still is high enough to cover the extra cost for housing. for centrality and reduced tolerance for commuting have been drivers of
Further, do the wage benefits differ depending on the level of education? increased urban housing values since the 1980s. Glaeser et al. (2001)
Our study positions the paper in a growing literature that has begun to argue that urban rents have increased faster than wage levels. In the
question the conventional wisdom of whether moving up the most attractive urban areas, rising housing values have vastly outpaced
geographical hierarchy is a generic solution to better income. national averages (Gyourko et al., 2013). Even if wage levels are
Human capital theories suggest that highly skilled workers are more significantly higher, far from everyone earns enough to cover the extra
likely to move than low skilled. The former benefit from larger relative housing costs of living in a big city (Florida, 2017).
wage differences across regions and have potentially more to gain from In light of these trends, the migration patterns of young and highly
migrating (e.g. Mincer, 1977; Nakosteen & Zimmer, 1980). More recent skilled individuals have stirred up an intense debate in Sweden. Some

* Corresponding author.
E-mail addresses: lina.bjerke@ju.se (L. Bjerke), charlotta.mellander@ju.se (C. Mellander).

https://doi.org/10.1016/j.cities.2022.103850
Received 15 December 2020; Received in revised form 10 June 2022; Accepted 24 June 2022
Available online 13 July 2022
0264-2751/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
L. Bjerke and C. Mellander Cities 130 (2022) 103850

claim that individuals are motivated to leave rural areas due to a considered to understand migration patterns. For example, the rural lack
decreased supply of services affecting their quality of life (Swedish of amenities may be compensated by higher wage levels for some
Government, 2015). Rural interest organizations have suggested that workers who remain at home, influencing their decision not to migrate.
young people flee from rural areas based less on better economic op­ This correlation is partially found in Swedish relocation conditions
portunities in cities than on a cultural notion that staying in rural areas (Backman & Bjerke, 2014).
makes them feel like “losers” by not moving. In this view, the perceived The problem is that migration patterns based on just these factors are
gains reaped from the rural-urban move are partly constructed from a inconsistent, as there are also alternative reasons behind the choice to
questionable, but widespread “urban norm structure” (Rönnblom, 2014; migrate or not. This was observed decades ago, starting with Lowry’s
Stenbacka, 2011). On the other hand, a more recent study suggests that (1966) gravity migration study, which showed that economic conditions
not all rural regions can be bundled together as losing propositions; in in the origin region do not affect migration decisions. These results were
fact, some are well-equipped for positive economic development and confirmed by Rabinski (1970) and also more recently by Schleicher
innovation that could entice youth to stay home (Rodríguez-Pose and (2017). In fact, numerous studies on US conditions examine the pull of
Wilkie, 2019). natural amenities to explain migration patterns into rural areas. Vias
The purpose of this paper is to identify the real extent of income gains (2012) notes that some micropolitan areas offer a small-town lifestyle
if one decides to move from a rural area or not. The conventional truth but still with commuting access to city amenities. This is also in line with
has been that bigger cities pay better due to agglomeration effects McGranahan and Wojan (2007) who show how natural amenities
(Glaeser, 1999; Rauch, 1993) and that the resulting higher wages are a explain migration patterns to rural parts of the US. Partridge et al.
migration pull factor. We aim to contribute to this field of research by (2010) find support for individuals who move to rural areas for better
answering two questions: Do bigger cities pay better to the extent that it quality-of-life reasons, though they also agree that migration equally
covers the extra housing costs, and do they do so for everyone? could be caused by monetary factors. Studies on counter-urbanization in
To do the study, we use Swedish microdata to track all individuals Sweden also recognize this quality-of-life effect, where career and
with a rural residential location at the age of 20. We follow them to the earnings are of minor importance for some (Lundgren, 2003).
age of 30 and identify their residential location and their incomes while Introducing yet another factor in the choice to migrate, a recent
controlling for regional and individual characteristics. With this data, stream of Swedish studies (Rönnblom, 2014; Svensson, 2006) perceive
we examine if it really pays off financially for young people to move from migration as a function of social pressure. These suggest that young
a rural to an urban region, or if cities are no longer the land of oppor­ people’s migration from rural areas is driven by an “urban norm” rather
tunity for everybody. than by economic incentives. By portraying cities as places of success,
rural areas are indirectly thought of as less advantageous. This may
2. Literature review affect young individuals, particularly those in rural areas who might
relocate so as not to be perceived as being “losers.” Currently, this
The migration literature has a long history and stretches over many research is primarily based on case studies of specific locations. For
disciplines, indicating that the causes and effects of people changing example, in one study of a rural region in Sweden, Svensson (2006)
residential locations can be looked at from many angles. One primary shows that only about 30 % of the young stayers thought local policy­
cause of relocation stems from advantages related to an upwards move makers considered them to be an asset. Among the young movers from
in the geographical hierarchy (Jacobs, 1969; Marshall, 1890). this area, nearly 70 % said they would be considered a local asset if they
Theory suggests that people relocate when another location better moved back home after their urban experiences. In other words, stayers
satisfies their financial needs and that there is a direct connection be­ perceived themselves as less valuable than movers.
tween incomes and migration decisions (Sjaastad, 1962). A vast amount Mover-Stayer models (Blumen, 1955) propose that certain groups
of literature also shows how wages increase with population size and are more likely to move, such as the young and the highly educated, who
density, due to agglomeration effects (Glaeser, 1999; Glaeser et al., have far more to gain from relocating. They have larger incentives in
1995; Glaeser & Mare, 2001; Rauch, 1993) which in turn can function as terms of higher wages and optimized value in a tight labor market
a migration pull factor. These effects increase the economic divide be­ (Schwartz, 1976). Migration can also be seen as a path trajectory, such
tween urban and rural areas, which has been occurring for decades. as an individual who decides to move to an urban area to attain higher
education, and thereafter is reluctant to move back again (Faggian et al.,
2.1. The urban-rural divide and the factors of migration 2007a; Faggian & McCann, 2009; Schwartz, 1976).
Factors such as marital status (Graves & Linneman, 1979), age
Urban areas have grown substantially in population by attracting (Pandit, 1997), and gender (Faggian et al., 2007b) can also play a role.
young, highly educated people, mostly due to strong labor markets, Using Swedish data, Bjerke and Mellander (2016) show that mobility
better educational opportunities, and a more diverse set of lifestyles. peaks around the age of 25 and that the likelihood to move drops sharply
Rural areas, on the other hand, have generally experienced a loss of after 35. These older stayers may have developed more location-specific
young people over time, resulting in several knock-on economic and advantages related to personal networks, social capital, labor market,
demographic consequences for them. As their population size decreases, and infrastructure – assets that would lose much of their value in
so have their public service offerings and other consumer amenities. This relocation.
multi-faceted downward slide makes remaining in rural regions even
less attractive, especially for younger individuals. 2.2. Costs of housing
The rural-urban migration is almost entirely among younger adults.
A study from Bjerke and Mellander (2016) illustrates how almost 90 % The vast inflow of individuals in bigger cities did not have the same
of moves across municipal borders are made by individuals between 18 response in housing supply, which in turn led to a rapid increase in
and 35. Most of these shifts are from smaller, less dense to bigger, more housing costs in many of those places (Glaeser, 2008; Gyourko et al.,
dense regions. Another study confirmed that only 15 % of Swedish 2013). In a more general context, previous studies show that housing
municipalities had more 25-year-olds in the year 2012 than they had 18- costs could have a negative impact and dampen the in-migration to high-
year-olds seven years earlier (Mellander, 2013). This demonstrates that cost areas (Cameron and Muellbauer, 1998; Plane et al., 2005; Buzar
the locational preferences of the younger generation very much deter­ et al., 2007; Berger & Blomquist, 1992). On the other hand, others
mine population growth or decline. suggest that in-migration can be positively related to increasing housing
Besides wage differentials, differences in amenities also weigh in the prices (e.g Bitter & Plane, 2012; Graves, 1983; Roy & Flood, 1992).
migration decision. Roback (1982) suggests that all factors must be Stawarz et al. (2021) build on the former literature and their results

2
L. Bjerke and C. Mellander Cities 130 (2022) 103850

suggest that with increasing housing costs (rents), a decline in residen­ of these moves, but also labor market conditions and other lifestyle-
tial inflows follows. They further show that this effect is strongest on the related factors such as family compositions (e.g. Bjerke & Mellander,
intensity of migration for the age groups of 18 to 24 and 30 to 49, times 2016).
when individuals either migrate for university studies or begin to settle There are 290 municipalities in Sweden, characterized as rural or
down. urban.2 The data do not contain information about an individual’s birth
Foulkes and Schafft (2010) further examine the role of financial region. As a proxy for “home region,” we use the residential region at the
means on migration and show how low income individuals are as in­ age of 18, i.e., one year before most youth finish high school. If it is a
clined to migrate long distances as their richer counterparts – contra­ rural municipality, the individual has a “rural origin.” This timing may
dicting the human capital theory. However, attracted by more be an even better proxy for the home region than the actual birth region,
affordable housing, their target destination is often another poor area, since the parents may have relocated over time. Only those 20-year-olds
which in turn increases poverty rates there. In developed nations, with a rural home region are selected in the analysis and they are 28,604
housing cost is especially an important factor. Differences in housing in number, with slightly more men (M) than women (W).
values across regions can hinder migration, mostly in rural-to-urban We then study these individuals ten years later at age 30, mapping
directions. Regions with higher housing demand have stronger real es­ where they live and their income as well as other personal characteris­
tate markets, which means that movements from rural areas to urban tics. Fig. 2 breaks down the locational choices of these individuals,
areas are hampered foremost by access to capital (Bover et al., 1989). reflecting migration or not.
The rapid increase in housing values in urban areas (e.g Gyourko et al., At 30 years old, we see that 11,583 have moved away from their
2013; Florida, 2017) may imply that a relatively higher wage level still home region, while 12,845 still (or again) live in their “home region.”
does not cover the extra costs of housing. Out of the 12,845 individuals who still live in the same region in 2015,
Taken together, however, much of the current research suggests that the data show that more than 7 out of 10 never lived in another region
there is indeed a positive financial reward from upwards migration in during the studied period, while the remaining 3 out of 10 moved but
the urban hierarchy. Bigger cities offer higher wages due to agglomer­ returned home. The term “Same LLM” (local labor market) refers to
ation effects. This research has a long history and there are relatively few 4176 individuals who moved out of their home region but at age 30 live
alternative hypotheses. Nevertheless, with the rapid increase in housing geographically close enough for it to be in the same LLM.
costs in bigger cities, we argue that the extra wage premium from big The second set of boxes in Fig. 2 shows gender differences for each
city labor markets might no longer be enough to cover the extra rent. locational choice. The largest difference appears in the category of those
This gap between wages and rent can potentially become a divider be­ still living in their home region when they are 30 years old. Men are in
tween those winning and those losing when moving to bigger cities. It majority, which is in line with previous literature showing that more
raises a key question in the current rural-urban migration debate: Is the women than men move to urban areas (Edlund, 2005).
wage premium in bigger cities big enough to cover their extra housing The bottom set of boxes in Fig. 2 illustrates educational attainment
cost, and is it true for everyone? The answer to the question positions (minimum 3 years of university education for a BA or above). Only about
this paper at the tipping point between the long-prevailing literature on 15 % of those in their home region at age 30 have a higher education.
agglomeration forces pulling people towards dense areas, and the new The corresponding number for those with higher education not living in
stream of literature suggesting that cities are no longer the land of op­ their home region is about 44 %, almost three times as many as those
portunity. The paper thus adds an important rural-urban perspective to who stayed in their home region. About 26 % of those living in the same
the literature on housing and internal migration. LLM, but not in the same municipality, have a higher education. Overall,
women have higher education. The largest difference in educational
3. Methodology attainment appears among those staying in their rural home region in
2015, where about 85 % do not have higher education. These descriptive
In the remaining part of the paper, we examine to what extent the results are possibly explained by results from several research reports (e.
locational choices of young rural individuals influence their income. The g., Lovén, 2016) that suggest that people who acquire a higher education
hypothesis tested concerns whether staying in a rural region is an have a generally higher tendency to move away from rural areas.
economically winning or losing choice for young adults, once other in­ However, comparing men and women, women are in general more
dividuals and regional factors have been controlled for. We consider educated (on a municipal level) but comparing all who study, men are
whether individuals decide to 1) move away from their home rural re­ more apt to move away than are women.
gion, 2) move and come back to it, or 3) not move at all.1
3.2. Rural movers vs. rural stayers
3.1. Data and background
As stated, we identify individuals who at the age of 20 (the year
We employ register micro data from Statistics Sweden. The data 2005) lived in a rural area and thereafter we examine their income level
contain information about all individuals in Sweden, such as residence 10 years later. We aim to capture them at a stage in life when they tend
location, gender, age, educational length, working profession, civil to have decided on a more permanent residential location. We estimate
status (single or not), and more. the regression as a linear cross-sectional model:
The analysis first tracks all individuals who were 20 years old in the
year 2005. This age is the starting point of what could be described as Y = Incomei = β1 Same residential municipalityi + β2 Same residential LLM i
the most significant migration wave. Fig. 1 illustrates the overall + β3 House pricei + [Individual characteristicsi ]ˊ β4
migration frequency across municipal borders and the age when these
+ [Regional characteristicsr ]ˊ β5
moves take place. Three peaks are noted, two smaller and one larger.
The first peak represents children moving with their parents, and the (1)
third peak is those moving around retirement age. The largest peak, The dependent variable, Y (ln) is a compound measure of gross
however, occurs between 20 and 35 years old. It starts to rise just after wages from employment and/or net income from self-employment in
finishing high school, and increases to the age of 25, then drops sharply the year 2015. Self-employment can be a necessity and may especially
around ages 30 to 35. Educational opportunities motivate a major part

2
Categorization according to The Swedish Agency for Growth Policy Analysis
1
Fig. A1 shows the distribution of rural and urban regions in Sweden. (2015).

3
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Fig. 1. Interregional migration for all ages of individuals in Sweden 2015 (Bjerke & Mellander, 2016).

Fig. 2. Locational and educational choices of rural origin individuals in 2003.

be true for some in rural areas, being located farther away from larger demand. Hort (1998) shows this for the Swedish housing market. This
labor markets.3 We do not separate employment and self-employment leads us to assume that housing costs affect the migration decision,
incomes since we are interested in total monetary gain. though indirectly through the relationship between incomes and hous­
Since urban regions generally have higher costs of living, we also ing costs. We try to capture this indirect relationship in a two-step-
take into account housing costs. Price dynamics in the housing market model. We build our dependent variable by estimating incomes as a
are highly related to population density. Prices tend to vary with in­ function of regional average house costs (house pricei) and create the
comes in the long run but are also short-term influenced by supply and vector of fitted values, Y
̂ . So, in the main empirical model, we use the
dependent variable Y
̂ as the (orthogonal) projection of Y onto the space
of average housing costs.
3
Self-employed in areas with such challenges tend not to be significantly
further job-creating (Stockdale, 2006).

4
L. Bjerke and C. Mellander Cities 130 (2022) 103850

̂ = β1 Same residential municipalityi + β2 Same residential LLM i


Y markets (Johansson et al., 2003).4
(2)
+ [Individual characteristicsi ]ˊ β3 + [Regional characteristicsr ]ˊ β4
3.3. Findings
Fig. 3 illustrates average regional income levels (to the left) as well as
the dependent variable Y. ̂ Starting with Y,
̂ the white and the patterned We now turn to the results of the analysis of income effects of in­
municipalities are those with negative values whereas those in shades of dividuals staying in their rural region versus moving away from it.
grey have positive values. The latter appears within and around larger Standardized and unstandardized coefficients are presented to get
cities. Income levels mirror this pattern, also related to population around the different units of our predictors. Five models were run: the
density and/or strong labor markets. first three use income (in logged form) as the dependent variable. Model
1 (Eq. (1)) is the base model and is followed by Models 2 and 3 that
3.2.1. Independent variables control for the effects of access to denser, urban areas – one based on
Details on definitions and source of data are provided in Appendix C. access to thicker labor markets (Model 2) and one based on housing costs
Below follow the general descriptions and motivations. (Model 3). Models 2 and 3 intend to capture the individual’s ability to
Local residence at the age of 30: This captures whether an individual commute to work from a rural location into a denser area. From previous
resides in his or her rural home municipality at the age of 30 years old, or research, we also know that such accessibility is a proxy for other labor
if he or she resides somewhere else. market characteristics, e.g., urban-related industries (see, for example,
Regional residence at the age of 30: This indicates whether an indi­ Andersson & Karlsson, 2007). The high interrelation between popula­
vidual lives in his or her home labor market region (LLM), but not in tion density and housing costs (a Pearson bivariate correlation coeffi­
their home municipality. In effect, these people live close to home or in cient of 0.86) leads to multicollinearity issues when combined in the
reasonable reach of it, but not exactly in the same municipality that they same model and is therefore used separately in Models 2 and 3,
started from. Some of the attraction of remaining in the home region are respectively. The last two Models (4 and 5) still employ income as the
familiar factors, such as old networks and family ties, sometimes called dependent variable, but this time after controlling for regional housing
family motivation of migration (Mitchell, 2004). costs (Eq. 2). Table 2 illustrates the results:
Migration frequency: As an attempt to see whether migration patterns
affect income, this captures the number of times the individual changed 3.4. Focus variables - locational choice
his or her residential municipality during the ten-year period. Kennan
and Walker (2011) show that migration decisions are highly driven by We start with Model 1 and the relation between income and loca­
income prospects, which means that repeated previous migration could tional choice. Our two focus variables are binary and take the value 1 if
have influenced income up or down. We add the effect of this potentially after ten years the individual is (i) residing in their home rural munic­
not being a linear relationship, with a negative effect on too many moves ipality, or (ii) in another municipality located within the same LLM. We
between different locations. find a significant negative effect from staying in the local rural home
Beyond this, the data reveal gender (man = 1), civil status (in a municipality, i.e., a rural wage loss, compared to those people who left
relationship = 1), background (foreign-born = 1), educational length their rural region to reside somewhere else. However, the impact of
(higher education = 1), and labor market experience. Civil status is staying in another municipality still within their home LLM is not
included since being single can decrease transaction costs of moving and significantly related to the individual’s current income level.5
may thereby indirectly affect income. Higher education reflects whether In Model 2, we control for accessibility to a bigger labor market.
the individual has a university education equal to or longer than three When this is accounted for, neither of our locational choice focus vari­
years. We assume this affects migration patterns in Sweden (Bjerke & ables becomes significant. The intuitive explanation for this is that if the
Mellander, 2016). Labor market experience captures the number of rural home municipality is within a large labor market within a
employment years between 20 and 30 years old. reasonable commuting distance, then the rurality does not affect the
Regional characteristics: The main regional character is average individual’s income level. In other words, the “rurality” is now of minor
housing costs (proxied by the housing values expressed in SEK) in the importance in finding labor market opportunities, though in terms of an
residential municipality. A vast part of the housing literature argues individual’s locational choice, it can matter for other reasons. Rural
there is a cointegrated relationship between housing costs and incomes. regions on the outskirts of larger metropolitan regions have an in-to-out-
There is also some relationship between disposable incomes and housing commuter ratio6 way below 1, which means that they are less dependent
costs trends, but the cause and effect are not easy to disentangle (Gallin, on their own labor market than a region in less dense areas with lower
2006; Geng, 2018). accessibility. This seems to be specifically true for individuals with
We also control for other factors related to one’s regional location at university degrees, since the interaction variable “long edu-stay” now
the age of 30. We assume that incomes could be affected by the industry becomes significant only at the 10 % level (compared to 1 % in Model 1).
structure at the location, primarily since the demand for educated labor In Model 3, we remove the accessibility measure and examine the
varies, but also because wage levels vary across industries due to other more direct reflection of the housing market in the specific municipality.
external factors, such as consumer demand and global competition. In this version, the income effect from staying is insignificant. The re­
Given this background, we include shares of employees within the sults are consistent, but the interaction between staying and having a
manufacturing, agriculture, mining, and hospitality sectors, respec­ higher education is now completely insignificant.
tively. Rural industry structures are primarily dominated by these sec­ We now move on to Models 4 and 5, where we change the dependent
tors (see Fig. A2). Income varies vastly even across rural areas, but the variable from total income to the income-housing residual (Eq. (2)),
agriculture, mining, and hospitality sectors are significant ones in the which is the individual’s income given the municipal housing costs. This
labor market in the northern parts of Sweden. The south is also char­ dependent variable is, in other words, the residual from a regression
acterized by rural areas, but there manufacturing is the dominating where the individual income is explained by housing values.
sector. We also control for the creative, knowledge-based sector, which
has primarily urban locations. A measure of accessibility to population is
also added to the empirical model to capture access to thicker labor 4
See the Appendix B for accessibility calculations.
5
All models have been estimated controlling for whether the move is to
another rural municipality. This variable is robust and follows the results of
staying. So, moving to another rural place has the same effect as staying.
6
SOU 2017:01.

5
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Fig. 3. Average Incomes and Estimated Incomes when Housing Values have been controlled for; To the left: Unstandardized residuals, used as the dependent variable
in Model 4 and Model 5. To the right: Average income 2015.

Starting with Model 4 (without accessibility), we see that staying in settled, may be reluctant to move. This implies that being rooted in one
the rural home region is now positively related to income, given house place is not the equivalent of being stuck (Florida, 2008). It may explain
costs. It seems that lower income levels resulting from staying in the the negative figure of mobility rate in the model, even though the in­
rural home region (shown in the earlier models) can be compensated for come effect is negative.
through lower housing costs. However, as in Models 1, 2 and 3, we only
get these significant results from the locational choice when individuals
decide to be in their same local rural home municipality. The choice of 3.5. Individual characteristics
staying within the same LLM but in another municipality than their
origin is not significantly related to income given the housing costs. In The individual characteristics variables we studied are consistent
Model 5, with accessibility, the effect from staying decreases in across all five model permutations. Starting with gender, men have
magnitude compared to the result in Model 4. This suggests that higher income, and looking at the standardized coefficients, gender has
accessibility to a larger market could compensate for the negative effect a relatively high impact on income. In Fig. 2, we noted that there are
related to income. significant differences between the migration patterns for men and
Our variable on migration frequency captures residential mobility women, but the difference is foremost related to educational levels. In
and the extent to which it relates to an individual’s income. In our result, general, women are more educated than men and gender differences in
higher mobility is negatively related to income. This is not in line with income seem to be even larger when focusing on those staying in the
previous literature on job mobility where mobility is shown to be posi­ home region. Out of those highly educated who decide to stay, about
tively related to upward career movements. However, our focus is on 70 % are women. The estimated positive effect on the income of being a
residential migration rather than a job change, though these are some­ man may therefore reflect regional structures rather than being an ed­
times but not always intertwined with each other. Our result may be ucation premium. These structures may be related to labor market dif­
explained by the fact that mobility due to job change in Sweden is ferences and the fact that agriculture, mining, and manufacturing in
infrequent. Going back to Table 1, we can see that the residential many rural regions suffer from low diversity and are dominated by men.
mobility rate is relatively low, with an average just above one move for (Bjerke et al., 2014).
this specific ten-year cycle. Some individuals may move for their studies As an extension of these gender differences, Table 3 illustrates the
but thereafter settle down quickly. There is a decreasing mobility trend employment distribution of women and men in rural areas. The table
in Sweden, as in many countries, which is often regarded as an shows the six largest industries for men and women, respectively. For
increasing problem for labor market dynamics (Fischer & Malmberg, women, these six industries accumulate to a bit more than 64 % of all
2001). People can be tied to a place for reasons beyond a job, and once women employed. For men, the top industries only employ a bit more
than 36 %. Women are concentrated in just a few industries, mostly in

6
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Table 1
Descriptive statistics (for variables descriptives, see Appendix C).
Variables Mean Std. dev Min Max

Dependent variables
Income 2478 1547 0.1a 16,630
6
Income - Housing cost residual − 1.23 e− 1538 − 11,348 13,959
Locational choice variables
Same residential municipality at the age of 30 (1 if same municipality) as at the age of 18 0.45 0.50 0 1
Same residential LLM at the age of 30 (1 if LLM, but not municipality) as at the age of 18 0.15 0.35 0 1
Migration frequency 1.33 1.40 0 10
Individual characteristics
Gender 0.52 0.50 0 1
Married 0.49 0.50 0 1
Experience 9.60 2.55 0 10
College Grad 0.28 0.45 0 1
Foreign-Born 0.06 0.23 0 1
Regional characteristics
Housing costs 2372 1742 307 10,182
Manufacturing 0.13 0.33 0 1
Agriculture 0.02 0.13 0 1
Mining 0.00 0.07 0 1
Hospitality 0.03 0.16 0 1
Creative 0.07 0.26 0 1
Accessibility 210,414 274,099 3084 994,733
N = 28,604
a
Those with negative or zero income has been assigned a value of 0.1. These are 8.7 % of the observations. Controls for the robustness of this data modification show
no major changes in results, either by removing them or adding a dummy.

the public sector. The third and sixth columns illustrate the average education is a monetary loss. However, adding it does not drastically
income in these industries. The columns to the right of these show the change the remaining variables. Having an immigrant background is
distance from the mean values of incomes in the group. About 13.7 % of also not significant in this model set-up.
women in rural areas are employed in the industry of residential care and Individual characteristics are similarly related to the dependent
activities. They have an average yearly income of 192,200 SEK, which is variables in all permutations of the model, with the statistically rela­
about 8 % lower than the average income of all women in rural areas. tively weak exception of having an immigrant background. It is negative
Turning to the right side of the table, it shows how men have higher and significant on our income-housing cost residual. One needs to
income but also how the “most popular” industries pay off. About 12.7 % emphasize that accessibility itself is now negative in this model. A
of men are employed in the industry of specialized construction activities. location with high accessibility implies higher income but this also
They have an average income of 312,500 SEK which is nearly 6 % higher drives up housing prices in the area.
than the average income of men in rural areas. All in all, the estimation
results on gender presented in Table 2 are all expected, given the
occupational and income structures presented in Table 3. 3.6. Regional characteristics variables
Turning our focus to Table 2 again, being married is negatively
related to income. Our individuals are 30 years old, which is just above The results for regional industrial structure are consistent across
the age when women on average give birth to their first child, but there models, with a few exceptions. One of these is that the relatively strong
is a relatively large regional variation. The oldest first-time parents are result for agricultural sector employment is only significant in Models
to be found in the largest cities (average age of 32.1 for mothers and 34.3 1–3, while being insignificant in relation to the housing cost residual.
for fathers) while the youngest first-time parents live in rural areas The mining employment variable is consistently strong and positive and
(average age of 25.2 for mothers and 28.2 for fathers).7 Marriage is by no indicates that in municipalities where this sector is strong, wages are
means a precondition to having children but studies show that the higher; in fact, high enough to cover the possible extra cost of housing.
transition to being a parent is highly correlated to also being in some As opposed to the mining industry, the income impact in a region
union (cohabitation or marriage) and that this is relatively high in with a relatively large share of employees within the hospitality sector is
Sweden (Baizán et al., 2004). only positive and significant if housing values are not accounted for. The
Labor market experience and university education are both positive coefficients are negative and significant when housing relative incomes
and highly significant. Looking at the standardized coefficients, labor are used as a dependent variable. This suggests that rural areas that
market experience outperforms higher education in terms of the create job opportunities provide higher wages, but these higher wages
magnitude of effects. This may have several explanations; one may be do not cover the extra cost of housing that tourism brings.
the low educational premium in Sweden for a university degree, which Being in a municipality with a larger share of employees in the
is, in fact, the lowest in OECD.8 Accumulated life income may therefore manufacturing sector is positive for income, even when housing values
be higher for those starting work immediately after high school. This are considered. This suggests that a strong manufacturing industry pays
may be expected specifically in a thinner labor market where matching relatively well, but the same municipalities may be less attractive places
between labor supply and labor demand is less efficient than in denser to live in given their relatively low levels of housing values. While the
areas (Duranton and Puga, 2004). This is also supported by the inter­ creative sector is more dominant in urban regions, the results suggest
action variable from Table 2 above that combines the effect of higher that having more employees in this sector is positively related to in­
education with also staying in their rural home region. This relation is comes also in rural areas.
negative, which suggests that staying when one also has a higher
4. Summary and concluding discussion

7
Source: Statistics Sweden. The starting point of this paper was the question of whether cities are
8
Education at a glance 2017 - OECD indicators. still the land of opportunity and if the wage premium from bigger cities

7
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Table 2
Regression results, unstandardized β-coefficients, t-statistics in italics and standardized β-coefficients in bold.
Dependent variable: Dependent variable:
Total income (ln) 2015 Housing cost residual

Model 1 Model 2 Model 3 Model 4 Model 5

Focus variables -Locational Choice


Same municipality − 0.14*** − 0.04 − 0.01 126.26*** 71.27***
(− 3.54) (− 0.86) (− 0.30) (5.71) (3.00)
− 0.023 − 0.006 − 0.002 0.041 0.023
Same LLM − 0.02 1.68e− 3 0.02 23.96 13.23
(− 0.52) (− 0.04) (0.46) (1.02) (0.56)
− 0.002 − 0.002 0.002 0.006 0.003
Migration freq. − 0.10*** − 0.11*** − 0.10*** − 42.29*** − 39.00***
(− 3.65) (− 3.87) (− 3.87) (− 2.74) (− 2.53)
¡0.023 ¡0.050 ¡0.049 ¡0.038 ¡0.035
(Migration freq.)2 0.008 0.01 0.01 − 2.75 − 2.94
(1.42) (1.49) (1.38) (− 0.86) (− 0.92)
0.017 0.018 0.016 ¡0.011 ¡0.012
Individual characteristics
Gender 0.63*** 0.62*** 0.62*** 889.32*** 891.00***
(22.28) (22.18) (22.10) 0.105 (55.59) (55.72)
0.106 0.106 0.289 0.289
Married − 0.19*** − 0.17*** − 0.16*** − 286.23*** − 294.62***
(− 6.86) (− 6.27) (− 5.81) (− 18.37) (− 18.86)
¡0.032 -0.029 ¡0.027 ¡0.093 -0.096
LM experience 0.71*** 0.71*** 0.71*** 238.35*** 239.14***
(131.01) (130.74) (130.56) 0.611 (77.45) (77.69)
0.613 0.612 0.395 0.396
Long education 0.52*** 0.48*** 0.46*** 393.26*** 416.39***
(13.41) (12.15) (11.69) 0.070 (17.77) (18.57)
0.079 0.073 0.115 0.122
Immigration background 0.06 0.04 0.04 − 57.12* − 44.99
(1.12) (0.73) (0.76) (1.75) (1.38)
0.005 0.003 0.003 ¡0.009 ¡0.007
Interaction: Longedu*stay − 0.17*** − 0.12* − 0.11 − 99.65*** − 123.61***
(− 2.46) (− 1.81) (− 1.53) (− 2.55) (− 3.15)
¡0.014 ¡0.010 ¡0.009 ¡0.016 ¡0.020
Regional characteristics
Housing cost 8.24e-5***
(8.63)
0.049
Manufacture 0.49*** 0.50*** 0.52*** 385.47*** 379.68***
(11.68) (11.93) (12.30) 0.058 (16.05) (15.81)
0.055 0.057 0.083 0.082
Agriculture 1.10*** 1.13*** 1.13*** 48.32 31.75
(10.53) (10.82) (10.88) 0.050 (0.81) (0.54)
0.048 0.049 0.004 0.003
Mining 0.66*** 0.75*** 0.69*** 1195.59*** 1145.70***
(3.32) (3.78) (3.48) (10.63) (10.17)
0.015 0.017 0.016 0.053 0.005
Hospitality 0.60*** 0.60*** 0.59*** − 240.55*** − 241.11***
(7.16) (7.18) (7.00) (− 5.03) (− 5.04)
0.033 0.033 0.032 ¡0.025 ¡0.025
Creative 0.49*** 0.47*** 0.46*** 617.80*** 626.13***
(9.12) (8.83) (8.69) (20.32) (20.59)
0.043 0.042 0.041 0.105 0.106
Accessibility 0.10*** − 52.15***
(6.58) a (− 6.28)
0.037 0.039
Constant − 0.37*** − 1.51*** − 0.60*** − 2798.10*** 2180.82***
(− 5.70) (− 8.17) (− 8.50) (− 75.58) (− 20.76)
Adj R2 0.415 0.415 0.416 0.304 0.305
N = 28,604

Accessibility and housing costs are highly correlated (0.86) and are therefore used separately.

is big enough to cover the extra housing cost? Much research has shown in the intersection point between the extensive literature on agglomer­
that bigger cities pay higher wages, which in turn would act as a ation forces—stronger labor markets and wage premiums on one
migration pull factor. But at the same time, bigger cities have become hand—and a newer stream of literature suggesting that cities are no
expensive to live in, and it may be the case that the extra wage premium longer is the land of opportunity, at least not for everybody. Our paper
from the agglomeration effects may not be enough to cover the extra cost contributes by examining the income effect based on the choice of
of housing – at least not for everybody. This work thereby positions itself staying or moving away from the home rural region, both with and

8
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Table 3
Distribution of employment, the average income of men and women in rural areas and the difference between the average income for the gender group.
Women in rural areas 2015 Men in rural areas 2015

Industry Distribution of average Difference from Industry Distribution of Average Difference from
employment (%) Income (1000 average group employment (%) Income (1000 average group
SEK) income (%) SEK) income (%)

Residential care 13.67 192.2 − 8.0 Specialized 12.74 312.5 5.8


activities construction
activities
Education 13.45 184.9 − 12.2 Manufacture of 5.67 323.9 9.1
fabricated metal
products
Social work activities 12.93 179.0 − 15.9 Land transport and 4.93 327.6 10.2
without transport via
accommodation pipelines
Retail trade 11.04 190.3 − 9.0 Retail trade 4.45 268.3 − 9.7
Human health activities 7.73 233.3 11.1 Construction of 4.40 329.6 10.7
buildings
Public administration 5.29 246.8 16.0 Wholesale trade 4.30 316.1 6.9
and defence
Accumulated sum 64.12 Accumulated sum 36.48

without housing values being considered. We focus only on individuals of a regions, which are now in a situation searching for highly educated
rural origin, following them from the age of 20 to the age of 30. human capital to keep up with the increased global competition.
We assumed that individuals migrate if they expect the income value An additional point to be made is that incomes can also be related to
of moving to be positive and that higher human capital increases the investment in housing. For those who move to bigger cities and can
incentives for such moves. Considering this, our results may add to afford to buy their housing, it can be an opportunity to cash in on the
understanding why people choose to stay in a rural region rather than appreciated value of their investment to buy cheaper housing in a more
move away from it. First, if one is an individual with a rural background, rural region later in life. While this aspect is not covered in our paper, it
one has choices to make, just like the urban counterparts. In terms of is an alternative source of income that would be beneficial for those who
one’s ability to increase one’s income, opportunities may be fewer if one migrate to cities (Fielding, 1992).
stays than if one moves away. There are also fewer educational oppor­ Third, our results suggest that regional characteristics are of
tunities and if one graduates, there may also be fewer job opportunities. importance for incomes, both in terms of sectorial structure and overall
If one does not wish to move, the choices become rather restricted. labor market. Following agglomeration theory and the evidence on the
However, our findings suggest that staying in a rural region is rarely urban wage premium, this is not at all surprising (Yankow, 2006). Given
negatively related to an individual’s income level and is most often that we have narrowed our focus to individuals with a rural home re­
insignificant. If we also consider housing values, it is instead consistently gion, our results are unique and highly policy relevant. Subsidies
financially beneficial to stay in a rural region for less educated directed to rural regions are today distributed with little attention on
individuals. “degrees of rurality” and how these may require diverse levels of sup­
Second, we find that if one stays in their rural region and attains a port, if any at all. Rural areas can be close to urban areas, but they can
higher education, a monetary loss from that choice of location can be also be located very far away from labor market opportunities. One
expected. On the one hand, rural regions in Sweden may not have a labor might expect that rural areas located nearby urban areas are in any need
market to offer jobs that match different sorts of human capital. Labor of support or only to a lesser extent. Those individuals can access urban
demand and supply are both larger in cities, which increases labor amenities without losing rural amenities.
productivity through better matching (Glaeser & Mare, 2001). In such However, one cannot forget that individuals may stay in their home
cases, a job at a rural location would generally be related to lower in­ rural areas for many reasons. Rural areas can potentially compensate for
come even if one has a higher education simply because the matching in their lower incomes with lower housing costs. The role and magnitude of
the labor market is less effective. Because of this, the wage differentials this differential seem to vary with the economic nature of the region,
between high and low educated individuals become smaller in rural specifically its industry structure. Rural regions in Sweden are highly
areas. In comparison, the human capital wage premium in urban regions heterogenous in this respect and this also causes differences in the
is bigger and large enough to cover the extra housing cost (Florida, housing market. Due to industry structures, some regions have high
2017). incomes, which is not completely reflected in their housing costs. Other
On the other hand, not all of income is related to higher education. regions have relatively low incomes but have housing costs driven up by
As shown in a recent report (Eklund & Pettersson, 2017), the overall external demand.
picture is that the educational system is incapable of saturating the labor Is this situation unique to Sweden then? We believe that this is not
market. Also, the education premium is overall very low in Sweden. This the case. A large percentage of the world’s countries have experienced
is partly supported by our results that show how labor market experi­ far-reaching urbanization in recent decades. Many of the cities that have
ence seems to have a higher pay-off than education. Since we know that had an increase in net migration have also had sharply increased
knowledge tends to agglomerate in space, the decision to stay in a rural housing values. The migration to cities has not only been driven by
region may be better combined with the decision not to achieve higher highly educated individuals, but individuals with different levels of
education, but rather to begin employment immediately after high education and different opportunities to get a job that pays well. In other
school. Unfortunately, the result of this may be fatal for many rural words, cities bring together individuals with different backgrounds and

9
L. Bjerke and C. Mellander Cities 130 (2022) 103850

thereby different probabilities to find acceptable housing and jobs. We descriptions are accurate and agreed by all authors.
also know from several studies (Florida, 2017; Florida & Mellander, Lina Bjerke*: conceptualization, methodology, formal analysis,
2015) that, for example, segregation has increased in large, attractive writing-review & editing, visualization, data curation.
cities over time, which is most likely a function of this very process. The Charlotta Mellander: Writing-original draft, writing-review &
issue has also been raised in other contexts than the Swedish, for editing, conceptualization.
example, by Autor (Badger & Bui, 2019) as well as by Florida (2017).
In sum, this research suggests that there are winners and losers, but Declaration of competing interest
they differ depending on the focus. Getting a university degree almost
consistently adds to income, with some exceptions, and with such an All authors have participated in (a) conception and design, or anal­
education level, leaving a rural area will likely generate a higher ysis and interpretation of the data; (b) drafting the article or revising it
paycheck. In our selected group of individuals, about 30 % had a uni­ critically for important intellectual content; and (c) approval of the final
versity degree by the age of 30; for them moving to an urban area would version.
be a better financial choice – even when the higher housing costs are This manuscript has not been submitted to, nor is under review at,
considered. However, among the remaining 70 %, it may be wiser to stay another journal or other publishing venue.
in a rural location since the possible higher wage levels of urban loca­ The authors have no affiliation with any organization with a direct or
tions most likely will not cover their extra costs of housing. Cities may indirect financial interest in the subject matter discussed in the
still be a land of opportunity for high-skilled workers but less so for those manuscript.
without a university degree. The following authors have affiliations with organizations with
direct or indirect financial interest in the subject matter discussed in the
CRediT authorship contribution statement manuscript.

The corresponding author* is responsible for ensuring that the

Appendix A

Fig. A1. Rural and urban regions.

10
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Fig. A2. Municipal employment shares within the agricultural sector, manufacturing sector, hospitality sector, and mining sector respectively 2015.

11
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Appendix B. Accessibility measure

We define accessibility to population as follows: Each municipality (i) offers access to a Y, where;
Y ∈ {No of Population}
But there is also an access to Y in the closely located municipalities (j). This is the case for all municipalities in the set N {1, …,n}. The total Y in
municipality i is defined as follows:
( )
AYi = Yi f (cii ) + Yj f cij + … + Yn f (cin )

where f(c) is a distance decay function to determine how the accessibility value is influenced by the related cost of reaching these individuals. We
approximate this specific relationship by an exponential function:
( ) { }
f cij = exp − λtij

where tij is the time distance between municipality i and j. λ is a time-sensitivity parameter that determines how the accessibility changes in t. λ has
been estimated by Johansson et al. (2003) and reflects the implicit value of daily time use.
When we combine the two equations, we get that the Y accessibility in municipality i is defined as follows:

290
{ }
AYi = Yj exp − λtij
j=1

Appendix C. Variables

Variable Definition Year

Income Gross wage from employment and/or net income from self-employment 2015
Same municipality 1: If the individual lives in the same municipality at age 30 as they did at age 18 2005/2015
0: Otherwise
Same LLM 1: If the individual lives in the same local labor market region (LLM) but not the same municipality at age 30 as they did at age 18. 2005/2015
0: Otherwise
Migration freq. Number of times the individual changed residential municipality between 2005 and 2015 2005–2015
(Migration freq.)2 Above squared 2005–2015
Gender 1: Man All
0: Woman
Married 1: Married and/or living together with another person 2015
0: Otherwise
LM experience The number of years the individual has been working between age 20 and age 30 2015
Long education 1: At least a year at university or equivalent 2015
0: Otherwise
Immigration background 1: Foreign born All
0: Otherwise
Housing cost Median house price in the municipality 2015
Manufacturing Share of employees in the municipality within the manufacturing sector 2015
Agriculture Share of employees in the municipality within the agricultural sector 2015
Mining Share of employees in the municipality within the mining sector 2015
Hospitality Share of employees in the municipality within the hospitality sector 2015
Creative Share of employees in the municipality within the creative sector 2015
Accessibility Accessibility (see Appendix B) to population 2015
Source: Statistics Sweden (SCB) for all variables.

References Bjerke, L., & Mellander, C. (2016). Moving home again? Never! The locational choices of
graduates in Sweden. The Annals of Regional Science, 1–23.
Blumen, I. (1955). The industrial mobility of labor as a probability process. Cornell
Andersson, M., & Karlsson, C. (2007). Knowledge in regional economic growth- the role
University.
of knowledge accessibility. Industry & Innovation, 14(2), 129–149.
Bover, O., Muellbauer, J., & Murphy, A. (1989). Housing, wages and UK labour markets.
Backman, M., & Bjerke, L. (2014). Returns to higher education: A regional perspective. In
Oxford Bulletin of Economics and Statistics, 51(2), 97–136.
Knowledge, Innovation and Space. Edward Elgar Publishing.
Buzar, S., Ogden, P., Hall, R., Haase, A., Kabisch, S., & Steinfiihrer, A. (2007). Splintering
Badger, E., & Bui, Q. (2019). What if cities are no longer the land of opportunity for low-
urban populations: Emergent landscapes of reurbanisation in four european cities.
skilled workers?. Jan 11th 2019, Available at: New York Times https://www.nytimes.
Urban Studies, 44(4), 651–677.
com/2019/01/11/upshot/big-cities-low-skilled-workers-wages.html.
Cameron, G., & Muellbauer, J. (1998). The housing market and regional commuting and
Baizán, P., Aassve, A., & Billari, F. C. (2004). The interrelations between cohabitation,
migration choices. Scottish Journal of Political Economy, 45(4), 420–446.
marriage and first birth in Germany and Sweden. Population and Environment, 25(6),
Duranton, G., & Puga, D. (2004). Micro-foundations of urban agglomeration economies.
531–561.
In J. V. Henderson, & J. F. Thisse (Eds.) (1 ed.,Handbook of regional and urban
Berger, M. C., & Blomquist, G. C. (1992). Mobility and destination in migration decisions:
economics (1 ed., Vol. 4, pp. 2063–2117). Elsevier.
The roles of earnings, quality of life, and housing prices. Journal of Housing
Edlund, L. (2005). Sex and the City. Scandinavian Journal of Economics, 107(1), 25–44.
Economics, 2(1), 37–59.
Edlund, L., Machado, C., & Sviatschi, M. M. (2015). Bright minds, big rent: Gentrification
Bitter, C., & Plane, D. (2012). Housing markets, the life course, and migration up and
and the rising returns to skill (No. w21729). National Bureau of Economic Research.
down the urban hierarchy. In The SAGE handbook of housing studies (pp. 295–312).
Eklund, J., & Pettersson, L. (2017). Högskola i otakt. Dialogos Förlag.
Bjerke, L., Backman, M., Wixe, S., Johansson, S., & Wallin, T. (2014). Platsbunden
Faggian, A., & McCann, P. (2009). Universities, agglomerations and graduate human
innovationsförmåga och arbetskraftens sammansättning. Report from Jönköping
capital mobility. Tijdschrift Voor Economische en Sociale Geografie, 100(2), 210–223.
University.

12
L. Bjerke and C. Mellander Cities 130 (2022) 103850

Faggian, A., McCann, P., & Sheppard, S. (2007a). Human capital, higher education and Lundgren, U. (2003). International Journal of Population Geography, 9, 399–418.
graduate migration: An analysis of scottish and welsh students. Urban Studies, 44 Marshall, A. (1890). Principles of economics. London, UK: Macmillan.
(13), 2511–2528. McGranahan, D., & Wojan, T. (2007). Recasting the creative class to examine growth
Faggian, A., McCann, P., & Sheppard, S. (2007b). Some evidence that women are more processes in rural and urban counties. Regional Studies, 41(2), 197–216.
mobile than men: Gender differences in U.K. Graduate migration behaviour. Journal Mellander, C. (2013). De ungas flykt till staden (Eng: The youth escape to the city).
of Regional Science, 47(3), 517–539. Retrieved from http://vertikals.se/charlotta/2013/12/02/de-ungas-flykt-till-stade
Fielding, A. J. (1992). Migration and social mobility: South East England as an escalator n/.
region. Regional Studies, 26(1), 1–15. Mincer, J. (1977). Family migration decisions. NBER working paper no, W0199.
Fischer, P. A., & Malmberg, G. (2001). Settled people Don’t move: On life course and (Im- Mitchell, C. J. (2004). Making sense of counterurbanization. Journal of Rural Studies, 20
)Mobility in Sweden. International Journal of Population Geography, 7(5), 357–371. (1), 15–34.
Florida, R. (2017). The new urban crisis: How our cities are increasing inequality, deepening Nakosteen, R. A., & Zimmer, M. (1980). Migration and income: The question of self-
segregation, and failing the middle class—And what we can do about it. New York: Basic selection. Southern Economic Journal, 46(3), 840–851.
Books. Pandit, K. (1997). Cohort and period effects in U.S. Migration: How demographic and
Florida, R. (2008). Who’s your city?: How the creative economy is making where to live the economic cycles influence the migration schedule. Annals of the Association of
most important decision of your life. A member of the perseus books group New York. American Geographers, 87(3), 439–450.
Florida, R. (2002). The rise of the creative class: And how it’s transforming work, leisure, and Partridge, M. D., Ali, K., & Olfert, M. (2010). Rural-to-urban commuting: Three degrees
everyday life. New York: Basic Books. of integration. Growth and Change, 41(2), 303–335.
Florida, R., & Mellander, C. (2015). Segregated city: The geography of economic segregation Plane, D. A., Henrie, C. J., & Perry, M. J. (2005). Migration up and down the urban
in America’s metros. Available at: Martin Prosperity Institute https://www.diva-porta hierarchy and across the life course. Proceedings of the National Academy of Sciences,
l.org/smash/get/diva2:868382/FULLTEXT01.pdf. 102(43), 15313–15318.
Foulkes, M., & Schafft, K. A. (2010). The impact of migration on poverty concentrations Rabinski, J. (1970). unpublished thesis. In Real earnings and the present value of future
in the United States, 1995–2000. Rural Sociology, 75(1), 90–110. earnings in a model of human migration (p. 111). Urbana: Department of Economics,
Gallin, J. (2006). The long-run relationship between house prices and income: Evidence University of Illinois.
from local housing markets. Real Estate Economics, 34(3), 417–438. Rauch, J. E. (1993). Productivity gains from geographic concentration of human capital:
Geng, N. (2018). Fundamental drivers of house prices in advanced economies. Evidence from the cities. Journal of Urban Economics, 34(3), 380–400.
Glaeser, E. L. (2008). Cities, agglomeration, and spatial equilibrium. OUP Oxford. Roback, J. (1982). Wage, rents, and the quality-of-life. Journal of Political Economy, 90
Glaeser, E. L. (1999). Learning in cities. Journal of Urban Economics, 46(2), 254–277. (6), 1257–1278.
Glaeser, E., & Kohlhase, J. (2003). Cities, regions and the decline of transport costs. Rodríguez-Pose, A. (2018). The revenge of the places that don’t matter (and what to do
Papers in Regional Science, 83(1), 197–228. about it). Cambridge Journal of Regions, Economy and Society, 11(1), 189–209.
Glaeser, E. L., Kolko, J., & Saiz, A. (2001). Consumer city. Journal of Economic Geography, Rodríguez-Pose, A., & Wilkie, C. (2019). Innovating in less developed regions: What
1(1), 27–50. drives patenting in the lagging regions of Europe and North America. Growth and
Glaeser, E. L., & Mare, D. C. (2001). Cities and skills. Journal of Labor Economics, 19(2), Change, 50(1), 4–37.
316–342. Rönnblom, M. (2014). Ett urbant tolkningsföreträde? En studie av hur landsbygd skapas i
Glaeser, E. L., et al. (1995). Economic growth in a cross-section of cities. Journal of nationell policy. Umeå centrum för genusstudier. Retrieved from.
Monetary Economics, 36(1), 117–143. Roy, J., & Flood, J. (1992). Interregional migration modeling via entropy and
Gollin, D., Jedwab, R., & Vollrath, D. (2016). Urbanization with and without information theory. Geographical Analysis, 24(1), 16–34.
industrialization. Journal of Economic Growth, 21(1), 35–70. Schleicher, D. (2017). Surreply: How and why we should become un-stuck!.
Graves, P. E. (1983). Migration with a composite amenity: The role of rents. Journal of Schwartz, A. (1976). Migration, age and education. The Journal of Political Economy, 84
Regional Science, 23(4), 541–546. (4), 701–719.
Graves, P. E., & Linneman, P. D. (1979). Household migration: Theoretical and empirical Sjaastad, L. A. (1962). The costs and returns of human migration. The Journal of Political
results. Journal of Urban Economics, 6(3), 383–404. Economy, 70(5), 80–93.
Gyourko, J., Mayer, C., & Sinai, T. (2013). Superstar cities. American Economic Journal: Stawarz, N., Sander, N., & Sulak, H. (2021). Internal migration and housing costs—A
Economic Policy, 5(4), 167–199. panel analysis for Germany. Population, Space and Place, 27(4), Article e2412.
Hansen, H. K., & Niedomysl, T. (2009). Migration of the creative class: Evidence from Stenbacka, S. (2011). Othering the rural: About the construction of rural masculinities
Sweden. Journal of Economic Geography, 9(2), 191–206. and the unspoken urban hegemonic ideal in swedish media. Journal of Rural Studies,
Hort, K. (1998). The determinants of urban housing price fluctuations in Sweden 27(3), 235–244.
1968–1994. Journal of Housing Economics, 7(2), 93–120. Stockdale, A. (2006). Migration: Pre-requisite for rural economic regeneration? Journal
Jacobs, J. (1969). The economy of cities. New York, NY: Random House. of Rural Studies, 22(3), 354–366.
Johansson, B., Klaesson, J., & Olsson, M. (2003). Commuters’ non-linear response to time Storper, M., & Scott, A. J. (2009). Rethinking human capital, creativity and urban
distances. Journal of Geographical Systems, 5(3), 315–329. growth. Journal of Economic Geography, 9(2), 147–167.
Kennan, J., & Walker, J. R. (2011). The effect of expected income on individual Svensson, L. (2006). Vinna och försvinna? Drivkrafter bakom ungdomars utflyttning från
migration decisions. Econometrica, 79(1), 211–251. mindre orter.
Lowry, I. S. (1966). Migration and metropolitan growth: Two analytical models. Chandler Swedish Government. (2015). Service i glesbygd. SOU, 2015, 35.
Pub. Co. Vias, A. C. (2012). Micropolitan areas and urbanization processes in the US. Cities, 29,
Lucas, J., & Robert, E. (2004). Life earnings and rural-urban migration. Journal of Political S24–S28.
Economy, 112(S1), S29–S59. Yankow, J. J. (2006). Why do cities pay more? An empirical examination of some
Lovén, I. (2016). Vem stannar kvar?-närhet till högskola och val av bostadsort. AgriFood competing theories of the urban wage premium. Journal of Urban Economics, 60(2),
working paper (2017: 4). 139–161.

13

You might also like