Abstract

Growing evidence suggests that internal migration experience shapes future internal migration behavior. However, it remains unclear what stage of the decision-making process past internal migration facilitates and whether the impact depends on the distance moved. To advance understanding of the role of past migration, we explicitly and dynamically link migration experiences to the formation and realization of future internal migration intentions by blending the aspiration–ability framework with the learned behavior hypothesis. We empirically test our proposition by fitting a series of logistic regression models to longitudinal microdata from the Household, Income and Labour Dynamics in Australia (HILDA) Survey, which has been conducted annually since 2001. We use a two-step approach by first modeling internal migration intentions and then modeling the realization of these intentions, distinguishing between residential moves, onward interregional migration, and return interregional migration. We find that migration experience is positively associated with both the formation and realization of migration intentions and that the effect of past migration increases with the distance moved and the number of past migrations. These findings suggest that migration experiences accumulate over the life course to predispose individuals toward subsequent migration. Finally, we show that the effect of past migration is not the result of a lack of social capital among repeat migrants—a finding that reinforces the importance of conceptualizing internal migration as a life course trajectory rather than a series of discrete events.

Introduction

Internal migration results from a complex process of maximizing opportunities, meeting personal aspirations, or surviving hardship. The decision to migrate internally is often triggered by a shift in individual and household needs and preferences (Mulder and Hooimeijer 1999) caused by key life course transitions, such as education completion, labor market entry, changes in employment status, and household and family formation (Cooke 2008; Kulu and Milewski 2007; Mulder 1993). These triggers filter through individual, household, and contextual factors; some of these factors (e.g., homeownership, family ties, economic recessions) constrain migration (Alvarez et al. 2021; Mulder 2018; Mulder et al. 2020a; Van Der Gaag and Van Wissen 2008), whereas others (e.g., educational attainment) facilitate it (Bernard and Bell 2018; Wagner 1990). Thus, a trigger or motive for migrating is a necessary but insufficient condition for the realization of migration (De Jong 2000). This multistep process is best conceptualized through the aspiration–ability framework (Carling 2002), which posits that migration is preceded by the formation of migration aspirations and intentions whose realization is modulated by resources and barriers. Originally formulated to explain international migration, this framework has been increasingly used in the internal migration literature, with growing attention devoted to the formation and realization of internal migration aspirations and intentions (Coulter et al. 2011; De Groot et al. 2011; Kley and Mulder 2010).

Despite progress over the last two decades in understanding internal migration decision-making, most empirical studies, including those drawing on longitudinal data, analyze migration as a year-to-year change in place of residence. This approach starkly contrasts with the life course perspective, one of the main paradigms in demographic research on internal migration (Kulu and Milewski 2007; McCollum et al. 2020). This perspective posits that “individual action is embedded in a process of decisions and behaviors that occurs over time” (Bernardi et al. 2019:3). Thus, experiences and resources accumulated over the life course, including migration experiences, shape subsequent decisions, including those related to internal migration (Myers 1999). The decision to migrate is therefore part of a long-term trajectory that unfolds over individuals' life courses rather than a series of discrete, independent events (Coulter et al. 2011).

To reconcile empirical research with the life course theory and clarify the internal migration decision-making process, a growing number of studies have applied sequence analysis to longitudinal microdata to trace internal migration pathways over sustained periods (Castagnone 2011; Coulter et al. 2011; Karhula et al. 2020; Stovel and Bolan 2004; Toma and Castagnone 2015; Vidal and Lutz 2018; Zufferey et al. 2021). This new line of inquiry has revealed heterogeneity in migration behavior that is overlooked in year-to-year analyses and has shown that migration trajectories are often contingent on how young adults began their migration careers. Individuals who left the parental home early appear to record more migrations throughout adulthood than individuals who left the parental home later (Bernard 2017). Building on this descriptive body of work, recent efforts have attempted to understand the impact of past internal migration on future internal migration. This work has identified a positive association between the number of past internal migrations in childhood and young adulthood and the probability of migrating internally later in adulthood (Bernard and Vidal 2020; Myers 1999). Early studies explained this association by arguing that internal migrants are more likely to return to their previous regions of residence because of location-specific capital, such as homeownership and social networks (DaVanzo 1981, 1983). However, recent studies have shown that internal migrants are also more likely to migrate to new destinations domestically and internationally (Bernard and Perales 2021). This finding indicates a learned behavior, whereby internal migration experiences shape future migration behavior in a recursive and self-reinforcing manner (Myers 1999).

However, the stage of the migration decision-making process at which migration experience plays a role remains unclear. Equally uncertain is whether particular migrations, such as the first migration in adulthood, are critical in shaping subsequent migration behavior or whether each successive migration cumulatively enhances the odds of migrating. It also remains to be established whether the impact of migration experiences is modulated by the distance moved. In this study, we advance understanding of internal migration decision-making by blending the migration aspiration–ability framework (Carling 2002; Carling and Schewel 2018; De Jong 2000) with theory viewing migration as a learned behavior (Bailey 1989; Bernard and Perales 2021; Morrison 1971). We contend that through internal migration experiences, individuals progressively acquire a set of skills, perceptions, and networks that facilitate the formation of internal migration aspirations and their realization by altering the perceived costs and benefits of migration.

Drawing on this framework, we empirically test our research hypotheses using Australia as an example. As a high-mobility country, Australia provides an ideal case study to explore the links between past and future internal migrations. As much as 40% of its population changes place of residence every five years, placing Australia on par with the United States and Scandinavian countries as one of the most mobile countries in the world (Bell et al. 2018; Bell et al. 2015). Drawing on data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey, a nationally representative longitudinal survey conducted annually since 2001, we model migration intentions and their realization in the following year as a function of the number of past migrations, distinguishing between short- and long-distance migration. We also distinguish between onward and return migration to establish whether past migrants simply return to their previous regions of residence or whether migration experiences increase their odds of migrating to a new destination.

Theoretical Framework

Internal Migration as a Learned Behavior

Past internal migrants are more likely to migrate than the general population (Bailey 1993; Bailey and Sly 1987; De Jong 1999; Morrison 1971). Despite a few exceptions (DaVanzo 1981, 1983; Morrison 1971; Myers 1999), research has explained the links between past and future internal migration mainly through the lens of return migration. Internal migrants are likely to return to their previous regions of residence because of location-specific assets, such as homeownership and social networks (DaVanzo 1981, 1983), which lower the cost of return. This is particularly the case for migrants who faced adverse circumstances at destination, such as job dissatisfaction (Lin et al. 1999; Niedomysl and Amcoff 2011) or marital separation (Spring et al. 2021). However, researchers have been increasingly recognizing that return migration should not be viewed solely as a corrective move and that it can be part of a planned strategy following the completion of tertiary education, for example (Thomassen 2021).

Emerging evidence suggests that the impact of internal migration experiences extends to onward internal migration: past internal migrants are more likely to migrate to a new region than individuals who never migrated and those who migrated fewer times (Bernard and Perales 2021; Myers 1999). In addition, recent evidence suggests interdependencies between different types of moves such that past residential mobility and internal migration exert a positive influence on each other (Bernard and Perales 2021) and on international migration (King and Skeldon 2010). Although further research is needed to confirm these processes, these findings suggest that it is the experience of moving itself rather than the distance moved that facilitates and enhances future migration.

Research on childhood mobility suggests that internal migration decisions in adulthood are also contingent on how young individuals begin their migration careers. The number of adult internal migrations by age 50 has been found to increase linearly with the number of childhood migrations in both the United States (Myers 1999) and a dozen European countries (Bernard and Vidal 2020), even when parental socioeconomic background is controlled for. Relocations in adolescence are particularly influential, and migration experience in preschool years appears to have a long-lasting effect on adult mobility in the Czech Republic, Denmark, France, Spain, and Switzerland. Childhood moves seem to be particularly important in facilitating the transition to the first adult internal migration, but their effect diminishes thereafter as individuals gain migration experience and develop related skills and know-how (Bernard and Perales 2021).

Mounting evidence suggests that internal migration experiences shape future internal migration decisions, but the mechanisms through which they operate remain unclear. Drawing on socialization theory, Myers (1999:871) proposed that internal migration “as a lifestyle is transmitted from parent to child” and concluded that individuals replicate migration behavior experienced in childhood. Similarly, in a study of second-generation Latvians' emigration intentions, Ivlevs and King (2012:128) argued that the “intergenerational transmission of family migration capital emerges as a migration driver distinct from other known determinants of migration and intrinsic to the migration process itself.” A proposed explanation for the association between past and future migration is that migration experiences, particularly in young adulthood, lower the perceived costs and constraints of moving (Huinink et al. 2014) by fostering migration-facilitating skills. For example, having been exposed to the intricacies of migration and learning to negotiate complexities involved in leaving home and entering new social contexts may increase migration ability and expertise.

Morrison (1971:179) was the first to suggest that migrants learn by doing, arguing that “decision thresholds are initially high for persons who never moved in their adult life . . . Once a move has been made, though, the experience may foster a learning process that blunts subsequent inertia.” This concept has been coined the learned behavior hypothesis (Bailey 1989; Cooke 2018) and has been expanded to emphasize the dynamic nature of this process. This approach is built around three principles suggesting that migration is a recursive, cumulative, and self-reinforcing process.

According to this framework, the positive association between past and onward migration indicates a process of learned behavior, whereby migration experiences cumulatively and continuously shape future migration behavior in a recursive and self-reinforcing manner. The strength of this effect, however, may be modulated by the broader societal context. For example, one's migration history may be more valuable to future migration in countries where internal migration is constrained by tighter housing and labor market regulations (Sánchez and Andrews 2011) than in high-mobility countries with lower barriers to migration (Long 1991). Although promising, this framework says little about the stage of the decision-making process at which migration experience operates.

Aspiration–Ability Framework

We argue that the migration aspiration–ability framework is ideal for more thoroughly exploring how migration experiences affect future migration behavior and illuminating the stages of migration decision-making that migration experiences shape. We view migration behavior as preceded by the formation of migration aspirations and intentions (Clark and Lisowski 2019; De Jong 2000), a concept grounded in the social psychological theory of planned behavior (Ajzen 1991). First formulated by De Jong (2000), formalized by Carling (2002), and refined by Carling and Collins (2018), this multistep model posits that the desire to migrate is a necessary but insufficient condition for migration to occur. Migration aspirations lead to migration intentions, which filter through individual, household, and contextual factors—some that facilitate and others that constrain the realization of migration aspirations. As a result, some individuals will not realize their migration aspirations, and others will migrate internally in lieu of an international relocation. Although developed for international migration, this simple yet useful approach has also been used in the internal migration literature.

Building on foundational work by Hughes and McCormick (1985), Sell and DeJong (1983), De Jong (1999), and Böheim and Taylor (2002), a growing body of work has examined the realization of internal migration aspirations. Whereas early studies drew on cross-sectional surveys, recent research has relied on longitudinal surveys and found that one half to two thirds of desired moves are realized. Some individuals—such as young adults, renters, and the native-born—are more likely to realize desired moves than other groups. (Clark and Lisowski 2018), although these associations are moderated by income (Coulter and Van Ham 2013; Coulter et al. 2011). Some life events, such as union formation/dissolution and childbirth, tend to precipitate the realization of planned moves; unanticipated events, such as job loss, have mixed effects and can sometimes precipitate an unplanned move (De Groot et al. 2011). Qualitative studies, which have dominated the international migration literature, have shown that rural dwellers are more likely to want to migrate but are less likely to realize their migration aspirations (Creighton 2013). However, the formation of international migration aspirations is also shaped by structural and contextual forces (Mondain and Diagne 2013). The societal context is also likely to shape aspirations and the ability to migrate internally. For example, in countries with higher income inequalities, individuals from lower socioeconomic backgrounds may have greater desire to migrate internally but less ability to realize these aspirations.

Linking Past and Future Internal Migration Over the Life Course

Although this body of work has been instrumental in improving the conceptualization of internal migration decision-making, the impact of migration experiences on the formation and realization of migration aspirations has received much less attention in the internal migration literature than in the international migration literature (Aslany et al. 2021). To our knowledge, the only exception is De Jong (2000), who hypothesized a link between past and future internal migration. His empirical analysis of the Thai context demonstrated the positive impact of having migrated on both the formation and realization of subsequent migration intentions. However, he did not differentiate between return and onward migration, making it difficult to ascertain the role of migration experiences and establish whether learned behavior is at play or whether repeat migrants simply return to previous places of residence. Furthermore, the use of a binary measure of past migration (i.e., having migrated or not) is likely to conceal the extent to which past migration affects future migration. Further, he did not consider the distance moved or the timing of past migrations, which may alter the relationship between past and future migration.

To advance understanding of the impact of past internal migration on future internal migration, we propose a framework that blends the aspiration–ability framework (Carling 2002; De Jong 2000) with the learned behavior hypothesis (Bailey 1989; Bernard and Perales 2021; Morrison 1971). The former offers a multistep perspective that highlights the importance of migration aspirations as the primary determinant of migration behavior and recognizes that the formation and realization of migration aspirations are shaped by separate drivers, whereas the latter provides a dynamic perspective by linking past and future migration in a feedback loop. Thus, we argue that through past internal migrations, individuals progressively acquire skills, perceptions, and networks that alter the perceived costs and benefits of migration, thereby facilitating the formation and realization of aspirations to migrate to both new and previous regions of residence.

Figure 1 depicts this model: migration experiences are explicitly and dynamically linked to both the formation and realization of migration aspirations, as indicated by the dotted arrows. Migration aspirations lead to migration intentions, which refer to a commitment to efforts toward migration and precede migration behavior. The proposed framework posits that internal migration intentions are triggered by a key life course transition (e.g., education completion, labor market entry, family formation) that creates a mismatch between actual and desired places of residence (Bernard et al. 2016; Coulter and Van Ham 2013; Mulder 1993; Mulder and Hooimeijer 1999). These events filter through contextual, household, and individual characteristics that facilitate or constrain migration aspirations and subsequently their realization. These characteristics include housing and labor market regulations and economic conditions at the macro level, household income and household structure at the household level, and educational attainment and employment status at the individual level.

Fig. 1

A multistep model of migration decision-making linking past and future migration

Fig. 1

A multistep model of migration decision-making linking past and future migration

Close modal

Studies of repeat internal migration have long postulated that individuals who have migrated will find it easier to migrate again (Goldstein 1954, 1964; Morrison 1971; Rogers 1969), but they have not specified the mechanisms at play. We argue that past migrants have higher aspirations and abilities to migrate because they face lower nonmonetary costs and higher perceived benefits of migration resulting from migration-facilitating skills, perceptions, and networks that are progressively acquired from one migration to the next.

First, past migrants have firsthand knowledge of the logistical and administrative hurdles of migrating to a new region and ways to navigate these challenges. These practical skills increase future migration ability. The more migrations experienced, the more versed individuals will be in dealing with the practical challenges of migrating internally. By learning how to negotiate the complexity of leaving their residence and entering new social contexts (Myers 1999), particularly when migrating over long distances, past migrants are also likely to develop social skills that they can draw on for future migration. These migration-facilitating skills are progressively developed and strengthened with each subsequent migration. Evidence from social psychology suggests that frequent internal migrants' exposure to diverse experiences makes them more socially skilled and adaptive to new environments and increases the value they place on autonomy and independence (Mann 1972; Oishi 2010). These skills will further reduce the perceived costs of migration, thus facilitating the conversion of aspirations into actual migration behavior.

Second, individuals who migrated in the past, particularly those who experienced an improvement in personal circumstances (e.g., better housing, improved access to amenities and services, enhanced employment opportunities, proximity to family members and friends), are likely to view migration as rewarding and enriching; individuals with no migration experience are more likely to view migration as a risky enterprise. Internal migration—particularly when related to unstable housing tenure, job loss, or marital separation (Huttunen et al. 2018)—may be a negative experience for some individuals. However, empirical evidence modeling average outcomes suggests a clear positive association between internal migration and a range of life outcomes: employment and occupational mobility (Fielding 1992), as well as subjective well-being and life satisfaction (Nowok et al. 2013). Having reaped the benefits of migration, past migrants are more likely to view migration positively than individuals without such experience and thus to have greater subsequent migration aspirations.

Finally, individuals with an internal migration history will differ in their trans-regional connections by having spatially dispersed social networks (Viry 2012) that are often oriented toward friends and siblings (Drevon et al. 2021; Oishi et al. 2013). These social ties are likely to facilitate the formation of migration aspirations as individuals typically prefer to relocate to regions where they have social ties (Büchel et al. 2019) and may seek to relocate closer to nonresident family members at particular life course stages, such as the birth of a child or marital separation (Mulder 2018; Mulder et al. 2020a; Mulder et al. 2020b; Winkels 2012). Having social networks at destination may also facilitate the conversion of migration aspirations into actual migration behavior by providing support and thus reducing the costs of migration. However, the strength of this effect is likely to depend on the size and composition of social networks and their geographic distribution in a country (Huttunen et al. 2018; Van Hear 1998).

Research Hypotheses

Drawing on this framework, we formulate five hypotheses. First, we anticipate that individuals who migrated in the past will have greater migration aspirations and will be more likely to realize these aspirations (Hypothesis 1). However, because the formation and realization of migration aspirations are shaped by separate drivers (Carling 2002), migration experience may shape aspiration and ability to a different extent. For example, having a social network in a prospective region of residence may lead to greater aspirations but not necessarily translate into migration behavior if those networks are too distant to provide support. From existing theoretical and empirical research, however, it is unclear which stage of the internal migration decision-making process is most likely to benefit from migration experience.

Migrants are more likely to return to a region they left because of location-specific assets (Davanzo 1981) but also are more likely to migrate to a new region because their monetary and other migration costs will be lower than those of individuals with no migration experience. This proposition leads us to expect that the impact of past migrations will extend to both return and onward internal migration but will be greater for return migration because of location-specific assets, such as homeownership and social networks (Hypothesis 2).

The learned behavior hypothesis highlights the cumulative effect of repeated migrations—not just recent migrations—on migration behavior. We therefore expect that each successive migration will increase migration aspirations and facilitate their realization (Hypothesis 3). Empirically, this hypothesis implies that migration experience should be measured not as binary (i.e., having migrated or not) but continuously (i.e., number of previous moves) to gauge the cumulative effect of past migrations. However, nonlinearity and threshold effects may exist: the feedback loop depicted in Figure 1 might ease or even stop after a certain number of migrations. For example, childhood moves appear particularly important in facilitating the transition to the first adult migration, but the effect diminishes thereafter (Bernard and Perales 2021). Similarly, the impact of adult migration experiences may diminish as individuals progress in their migration careers as migrants become accustomed to the challenges of relocating to a new region. We therefore posit that the impact of past migrations will remain positive but progressively decline with the number of past migrations (Hypothesis 4).

The impact of past migration on future migration decisions may also depend on the distance moved. Because longer moves sever social ties and networks more than shorter moves, we expect that interregional migration will more strongly influence subsequent migration than residential mobility (Hypothesis 5).

Data and Methods

The Household, Income and Labour Dynamics in Australia Survey

We use longitudinal microdata collected as part of the HILDA Survey, an annual longitudinal survey that has been following Australians aged 15 or older since 2001. Unlike some longitudinal surveys, the HILDA Survey does not collect retrospective migration histories. We therefore use migration history from Waves 1 to 17 to examine the impact of past migration on migration intentions and actual migration behavior in Waves 11 to 18 and update respondents' migration histories annually. We cannot consider the complete migration history of respondents, particularly migration experience in childhood, which is one of the drawbacks of using prospective longitudinal survey data.

In studying the impact of past moves, we aim to provide a more fine-grained account than previous studies, which measured migration experience with a binary variable (e.g., having moved or not). We measure past migration, our main explanatory variable, as a continuous variable indicating the number of moves. To establish whether the association between the number of adult migrations and the odds of migrating is linear, we construct a set of discrete variables (zero, one, two, three, and four or more migrations). We also distinguish between past residential mobility and interregional migration. The latter is defined as moves of 65 or more kilometers, the threshold established to distinguish employment-related moves from housing-related moves (Thomas et al. 2019). This distinction allows us to explore the extent to which the impact of past moves on future mobility decisions depends on the distance moved.

In the first step of the analysis, we examine migration intentions, which result from the formation of migration aspirations (see Figure 1). The HILDA Survey asks respondents to indicate how likely they are to change their address in the next 12 months using a five-point Likert-type scale (very unlikely, unlikely, unsure, likely, very likely). This assessment of the probability of migration is not strictly equivalent to intention, which refers to a commitment to efforts toward migration (Carling and Mjelva 2021). The latter better fits with our theoretical ambitions. However, the actual difference between the two concepts is probably smaller for internal migration than for international migration because barriers to internal migration are lower, given that most countries do not restrict migration within their national borders.1 Thus, we use likelihood to move as a proxy for intention. In addition, because likelihood is collected in relation to a change of address, we cannot distinguish between residential mobility and interregional migration intentions, which is another shortcoming of the HILDA Survey; we will return to this issue in the Discussion and Conclusion section.

In the second step, we examine actual migration behavior in the subsequent wave by modeling migration behavior conditional on the likelihood of moving reported in the previous year (Böheim and Taylor 2002; Coulter et al. 2011). To construct a dependent variable that robustly assesses the impact of past migration on subsequent migration, we distinguish between return and onward migration using respondents' previous Statistical Area Level 4 (SA4). The second level of spatial units after States and Territories under the Australian Statistical Geography Standard (ASGS), SA4s broadly correspond to labor markets. The HILDA Survey data contain 89 SA4s.2 A return migration is defined as migration to an SA4 where the respondent previously resided, whereas an onward migration is a move to an SA4 where the respondent did not previously reside. The distinction between onward and return migration allows us to differentiate between the two processes of learned behavior and location-specific capital outlined in the preceding section. We do not see these processes as competing mechanisms, and we expect past migration to have a positive impact on both return and onward migration.

Modeling Strategy

To establish the association between past migration, migration intentions, and actual migration behavior, we use two multinomial logistic regression models with year and state fixed effects. Consistent with our theoretical model, we first estimate migration intentions for the next 12 months as a five-category multinomial outcome (very likely to move, likely to move, unsure about moving, unlikely to move, and very unlikely to move) and use “very unlikely to move” as the reference category. This modeling strategy has the unique advantage of allowing the direct comparison of the estimated effects of the same variable (e.g., number of past moves) on four levels of migration intentions compared with “very unlikely to move.” Formally, this baseline model can be expressed as follows:

In PrMiw=cPrMiw=1=αc+Pi,w-1β1c+χi,w-1β2c+γw+ϑs+εiwc, 
(1)

where i, w, and s represent individuals, waves, and state of residence, respectively; M is the dependent variable, which captures migration intentions with c categories; and α is the model's intercept. In addition, P is a set of focal explanatory variables capturing the accumulated number of past moves, and X is a vector of control variables. The βs are (vectors of) model coefficients; ɛ is the usual stochastic regression error term; and γw and ϑs are wave and state fixed effects, which help mitigate the effects of business cycles and differences across states. To account for the nesting of observations within households and to safeguard the regression assumption of the independence of observations, we cluster standard errors on household membership (Cameron et al. 2012).

In the second set of regressions, we model actual migration behavior conditional on the likelihood to move reported the previous year. Again, we use a multinomial model as expressed in Eq. (1) based on four categories (no move, residential move, onward migration, and return migration) to estimate the impact of past moves on the realization of migration intentions. By recognizing that migration behavior forms a continuum in space and that individuals jointly consider the different forms of mobility available to them (Bell and Ward 2000), this approach provides a more nuanced and robust approach to mobility decision-making than is possible with a binary model (move/no move).

We include a series of control variables known to be correlated with migration: age, sex, educational attainment level, employment and educational statuses, marital and parental statuses, homeownership, income quantile, and a set of dummy variables capturing key life course transitions. The latter include marrying or forming a partnership, having a child, divorcing or separating, and changing jobs. We selected these life course events because of their long-established role as triggers of migration (Bernard et al. 2016; Kulu and Milewski 2007; Mulder 1993). We innovate by also controlling for personality traits as measured by the Five-Factor Personality Inventory: extroversion, agreeableness, conscientiousness, emotional stability, and openness to experience. These traits are unobservable in most survey data sets. A growing body of work from the psychological literature suggests that internal and international migrants exhibit different personality traits than the general population. Recent studies from Australia, the United States, and Europe found that individuals who score high on openness are more likely to migrate (Campbell 2019; Fouarge et al. 2019; Jokela 2009, 2014; Shuttleworth et al. 2020), although there is no agreement on the effect of other personality traits on migration behavior. The HILDA Survey captures the Big Five personality traits only every four years. To obtain data for waves with missing values, we therefore linearly interpolate observed values, a common practice (Cobb-Clark and Tan 2011; Schurer et al. 2015) given the stability of the Big Five personality traits (Cobb-Clark and Schurer 2012). All explanatory variables except sex are lagged by a year. Descriptive statistics for all variables can be found in Table A1 of the online appendix.

Finally, we conduct robustness checks to ensure that the association between migration experience and subsequent migration behavior is caused by repeat migrants' potential lack of social capital.

To aid interpretation, we express the estimated model coefficients as exponentiated coefficients, which gives relative risk ratios (RRRs): the ratio of the probability of being in a given outcome category (e.g., very likely to move) to the probability of being in the baseline category (i.e., very unlikely to move) associated with a one-unit increase in an explanatory variable. RRRs greater (smaller) than 1 indicate that a given explanatory variable is associated with an increased (decreased) likelihood of outcome realization relative to the baseline outcome. For ease of interpretation, for key regression results, we also provide visual representations through predicted probabilities with the covariates held at their observed value.

Results

Descriptive Statistics

Table 1 displays the migration histories collected from Waves 1 to 18. On average, respondents moved 2.4 times during the observation period. Less than one third of respondents never moved, 22% moved once, and close to 50% moved twice or more. In line with theoretical expectations, most moves occurred over short distances. Yet, 12% migrated at least twice over a distance of 65 or more kilometers. These summary statistics confirm the importance of repeat movement in Australia, with 15 being the maximum number of moves recorded over the observation period.

Table 1

Migration history

All Changes of AddressResidential Moves (<65 kilometers)Interregional Migration (65+ kilometers)
Average Number of Moves 2.40 1.80 0.50 
% Who Never Moved 28.88 33.48 74.70 
% Who Moved Once 21.59 23.44 13.17 
% Who Moved Twice 14.96 15.66 7.22 
% Who Moved Three Times 10.73 10.21 2.64 
% Who Moved Four or More Times 23.80 17.22 2.30 
All Changes of AddressResidential Moves (<65 kilometers)Interregional Migration (65+ kilometers)
Average Number of Moves 2.40 1.80 0.50 
% Who Never Moved 28.88 33.48 74.70 
% Who Moved Once 21.59 23.44 13.17 
% Who Moved Twice 14.96 15.66 7.22 
% Who Moved Three Times 10.73 10.21 2.64 
% Who Moved Four or More Times 23.80 17.22 2.30 

Source: Authors' calculations using data from Waves 1–18 of the HILDA Survey (n = 9,913 individuals).

Next, we turn our attention to mobility intentions and their realization. As shown in Table 2, 13% of respondents reported being very likely or likely to change address in the next 12 months, and less than 10% were unsure about their mobility intentions. In contrast, more than 57% of respondents indicated being very unlikely to move, and another 20% reported being unlikely. Although migration intentions may seem low, they broadly align with the level of mobility recorded in Australia, where approximately 15% of the population changes their address every year (Bell et al. 2015).

Table 2

Mobility intentions and mobility behavior, descriptive statistics

Mobility Behavior
Type of Move
Mobility Intentions: % of All Respondents (Wave w)Mobility: % Who Moved (Wave w + 1)Local Move Within an SA4Onward Migration to a New SA4Return Migration to an SA4 of Previous Residence
Very Likely to Move 7.35 65.18 36.83 20.35 8.00 
Likely to Move 5.32 40.39 23.96 12.23 4.20 
Unsure About a Move 9.60 23.62 14.87 6.47 2.28 
Unlikely to Move 20.21 9.49 6.53 2.11 0.85 
Very Unlikely to Move 57.52 3.61 2.52 0.83 0.26 
Mobility Behavior
Type of Move
Mobility Intentions: % of All Respondents (Wave w)Mobility: % Who Moved (Wave w + 1)Local Move Within an SA4Onward Migration to a New SA4Return Migration to an SA4 of Previous Residence
Very Likely to Move 7.35 65.18 36.83 20.35 8.00 
Likely to Move 5.32 40.39 23.96 12.23 4.20 
Unsure About a Move 9.60 23.62 14.87 6.47 2.28 
Unlikely to Move 20.21 9.49 6.53 2.11 0.85 
Very Unlikely to Move 57.52 3.61 2.52 0.83 0.26 

Note: SA4 = Statistical Area Level 4.

Source: Authors' calculations using data from Waves 11–18 of the HILDA Survey (n = 78,565 person-years).

As expected, there is a marked gradient in the realization of mobility intentions: the stronger intentions are, the greater the likelihood of having moved by the subsequent wave. Among individuals who reported being very likely to move, 65% moved the following year; the comparative figure for those who were unsure is 24%. However, not all individuals realized their mobility intentions. More than one third of individuals who reported being very likely to move did not migrate in the following 12 months, and close to 60% of those indicating that they would likely move stayed put during the following year. The fact that not all migration intentions are realized highlights the relevance of conceptualizing migration decision-making as a multistep process and demonstrates the importance of considering and identifying the barriers and facilitators that underpin the realization of migration intentions.

A few people who did not anticipate moving were forced to relocate the following year, but this proportion is small: less than 4% of individuals who reported being very unlikely to move ended up moving in the subsequent year. As expected, most realized moves took place locally, within the SA4 of current residence. Yet, a sizable proportion of realized moves represent onward migration to a new SA4, and the share of onward moves increases with the stated likelihood of moving.

Regression Models

Migration Intentions

Table 3 presents the results of our multinomial logistic regression models of mobility intentions. The estimated coefficients for the explanatory variables of interest—namely, the cumulative number of previous moves—and the control variables exhibit the expected signs. The RRRs for migration likelihood progressively decline with age. We also find evidence of a strong socioeconomic gradient: individuals in the labor force, those with tertiary education, and individuals in the top income quantile have RRRs greater than 1. Unsurprisingly, married and partnered individuals, those in dual-income households, and parents report a lower migration likelihood. Homeownership also displays a deterring effect. Conversely, having experienced a life course transition in the last 12 months, particularly divorcing and changing jobs, boosts migration likelihood. Finally, among the personality traits, openness increases all degrees of migration likelihood relative to individuals who are very unlikely to move; extraversion raises only the odds of being very likely to move. The remaining personality traits, particularly conscientiousness, decrease migration likelihood.

Table 3

Relative risk ratios from a multinomial logistic regression model of mobility intentions (baseline outcome is very unlikely to move)

Very Likely to MoveLikely to MoveUnsure About a MoveUnlikely to Move
Migration History (continuous)  
 Number of prior residential moves 1.328*** 1.221*** 1.174*** 1.061*** 
 Number of prior internal migrations 1.464*** 1.326*** 1.230*** 1.101*** 
Female 1.026 1.002 0.966 0.999 
Age (ref. = 15–24)     
 25–34 0.503*** 0.574*** 0.633*** 0.676*** 
 35–44 0.327*** 0.293*** 0.382*** 0.529*** 
 45–54 0.266*** 0.229*** 0.331*** 0.465*** 
 55–64 0.275*** 0.193*** 0.265*** 0.409*** 
 65+ years 0.192*** 0.113*** 0.213*** 0.386*** 
Marital and Parental Status     
 Married or partnered (ref. = single) 0.721*** 0.586*** 0.553*** 0.809*** 
 Have children (ref. = no child) 0.670*** 0.704*** 0.700*** 0.834*** 
Household Structure (ref. = not in dual-income household)  
 Dual-income household 1.037 0.769* 0.967 0.896* 
 Power couple 0.975 0.635** 0.868 0.960 
Tertiary Education (ref. = no tertiary education) 1.188*** 1.244*** 1.093 1.097* 
Income Quantile (ref. = 40% to 59%)     
 0% to 19% 0.823** 0.897 0.964 0.969 
 20% to 39% 0.889 0.955 0.979 0.992 
 60% to 79% 1.1110 1.129 1.004 1.043 
 80% to 100% 1.289*** 1.142 1.038 1.084 
Labor Force Status (ref. = not in the labor force)  
 Employed 1.126* 1.275*** 1.143* 1.133*** 
 Unemployed 1.0.62 1.413*** 1.486*** 1.184 
Life Course Transitions     
 Married or partnered 0.836 0.738 0.651*** 0.888 
 Divorced or separated 1.709*** 1.665*** 1.318*** 1.258*** 
 Birth or adoption 1.096 0.932 0.888 1.031 
 Changed job 1.300*** 1.381*** 1.268*** 1.060 
Homeownership (ref. = homeowner)     
 Renter 2.305*** 2.273*** 2.310*** 1.615*** 
 Rent-free 2.148*** 1.548*** 1.619*** 1.249*** 
Personality Traits     
 Extroversion 1.082*** 1.039 0.962* 0.984 
 Agreeableness 0.990 0.958 0.920*** 0.959*** 
 Conscientiousness 0.925*** 0.913*** 0.934*** 0.948*** 
 Emotional stability 0.956* 0.955 0.966 0.958*** 
 Openness 1.076*** 1.051* 1.040*** 1.038*** 
Number of Observations 68,711    
Log Pseudo-Likelihood −8.130e+07    
Very Likely to MoveLikely to MoveUnsure About a MoveUnlikely to Move
Migration History (continuous)  
 Number of prior residential moves 1.328*** 1.221*** 1.174*** 1.061*** 
 Number of prior internal migrations 1.464*** 1.326*** 1.230*** 1.101*** 
Female 1.026 1.002 0.966 0.999 
Age (ref. = 15–24)     
 25–34 0.503*** 0.574*** 0.633*** 0.676*** 
 35–44 0.327*** 0.293*** 0.382*** 0.529*** 
 45–54 0.266*** 0.229*** 0.331*** 0.465*** 
 55–64 0.275*** 0.193*** 0.265*** 0.409*** 
 65+ years 0.192*** 0.113*** 0.213*** 0.386*** 
Marital and Parental Status     
 Married or partnered (ref. = single) 0.721*** 0.586*** 0.553*** 0.809*** 
 Have children (ref. = no child) 0.670*** 0.704*** 0.700*** 0.834*** 
Household Structure (ref. = not in dual-income household)  
 Dual-income household 1.037 0.769* 0.967 0.896* 
 Power couple 0.975 0.635** 0.868 0.960 
Tertiary Education (ref. = no tertiary education) 1.188*** 1.244*** 1.093 1.097* 
Income Quantile (ref. = 40% to 59%)     
 0% to 19% 0.823** 0.897 0.964 0.969 
 20% to 39% 0.889 0.955 0.979 0.992 
 60% to 79% 1.1110 1.129 1.004 1.043 
 80% to 100% 1.289*** 1.142 1.038 1.084 
Labor Force Status (ref. = not in the labor force)  
 Employed 1.126* 1.275*** 1.143* 1.133*** 
 Unemployed 1.0.62 1.413*** 1.486*** 1.184 
Life Course Transitions     
 Married or partnered 0.836 0.738 0.651*** 0.888 
 Divorced or separated 1.709*** 1.665*** 1.318*** 1.258*** 
 Birth or adoption 1.096 0.932 0.888 1.031 
 Changed job 1.300*** 1.381*** 1.268*** 1.060 
Homeownership (ref. = homeowner)     
 Renter 2.305*** 2.273*** 2.310*** 1.615*** 
 Rent-free 2.148*** 1.548*** 1.619*** 1.249*** 
Personality Traits     
 Extroversion 1.082*** 1.039 0.962* 0.984 
 Agreeableness 0.990 0.958 0.920*** 0.959*** 
 Conscientiousness 0.925*** 0.913*** 0.934*** 0.948*** 
 Emotional stability 0.956* 0.955 0.966 0.958*** 
 Openness 1.076*** 1.051* 1.040*** 1.038*** 
Number of Observations 68,711    
Log Pseudo-Likelihood −8.130e+07    

Notes: Data are from Waves 1–18 of the HILDA Survey. Standard errors, clustered by household, are not shown here. The model includes state and year fixed effects.

*p < .05; **p < .01; ***p < .001

Two patterns emerge concerning migration experience. First, in line with Hypothesis 1, past moves positively influence migration intentions. As shown in Table 3, the number of past moves is associated with an increase in mobility intentions, and this effect is gradual. For example, the number of past residential moves bears more weight on the odds of being very likely to move (RRR =1.328, p < .001) than of being likely to move (RRR =1.221, p < .001) or of being unsure about moving (RRR =1.174, p < .001). Results from a Wald test confirm that these coefficients are statistically different from one another for both past residential moves and past interregional migrations (see Table A2, online appendix). Second, Wald tests show that interregional migration has a greater influence on migration intentions than residential mobility (see Table A3, online appendix), lending support to Hypothesis 5. The greater weight of long-distance migration most likely stems from the severance of social ties, forcing individuals to enter new social contexts. We contend that entering new contexts contributes to the development of perceptions, skills, and networks that facilitate the formation of migration intentions.

We now examine whether the estimated association between the number of earlier moves in adulthood and the perceived likelihood of moving is linear by estimating a second regression in which the number of past migrations is measured by discrete categories (zero, one, two, three, and four or more moves) rather than continuously. Figure 2, which displays predicted probabilities, shows that past migrations cumulatively increase the odds of being very likely to move in the future, with intentions to migrate progressively increasing with the number of past migrations. Results from a Wald test confirm that regression coefficients for residential moves are statistically different from one another for individuals who are very likely to move, likely to move, or unsure about moving (see Table A4, online appendix).

Fig. 2

Predicted probabilities from multinomial logistic regressions of the likelihood of moving. Data are from Waves 1–18 of the HILDA Survey. n = 68,711 person-years. The reference category is very unlikely to move. The number of past moves is entered as a series of categorical variables.

Fig. 2

Predicted probabilities from multinomial logistic regressions of the likelihood of moving. Data are from Waves 1–18 of the HILDA Survey. n = 68,711 person-years. The reference category is very unlikely to move. The number of past moves is entered as a series of categorical variables.

Close modal

The results for internal migration are not as clear-cut. Having migrated twice has the same effect on migration intentions as having migrated three times. However, having migrated twice has a significantly greater effect on migration intentions than one migration. Similarly, individuals who migrated four or more times are significantly more likely to report a greater likelihood of moving than those who migrated three times. Overall, these results suggest that repeated migration experiences accumulate over the life course to strengthen future migration intentions, supporting Hypothesis 3 but invalidating Hypothesis 4.

Actual Migration Behavior

We now turn to the conversion of migration intentions into migration behavior, conditional on migration intentions in the previous wave. Table A5 of the online appendix compares results with another model that does not control for migration intentions. It shows that the inclusion of migration intentions improves the overall model fit as measured by the Akaike information criterion while maintaining consistent regression coefficients.

The full set of results in Table 4 shows that many of the variables underpinning migration aspirations also facilitate their conversion into migration (e.g., tertiary education), whereas some attributes (e.g., being in a dual-income household, having children) constrain both steps of the decision-making process. However, some characteristics act in the opposite direction. For example, conscientiousness is associated with lower aspirations to migrate internally, but among those with aspirations to migrate, conscientiousness facilitates migration. This counterbalancing effect highlights the importance of distinguishing both steps of the decision-making process.

Table 4

Relative risk ratios from a multinomial regression model of mobility behavior (baseline outcome is no move)

Residential MobilityOnward MigrationReturn Migration
Migration History (continuous) 
 Number of prior residential moves 1.427*** 1.259*** 1.464*** 
 Number of prior internal migrations 1.140*** 1.608*** 1.661*** 
Female 1.037 1.145* 1.044 
Age (ref. = 15–24)    
 25–34 0.703*** 0.631*** 0.815 
 35–44 0.597*** 0.486*** 0.675** 
 45–54 0.679*** 0.586*** 0.582** 
 55–64 0.744*** 0.522*** 0.613** 
 65+ 0.932 0.758* 0.693 
Marital and Parental Status    
 Married or partnered (ref. = single) 0.991 0.944 1.221 
 Have children (ref. = no child) 0.964 0.698*** 0.564*** 
Household Structure (ref. = not in dual-income household) 
 Dual-income household 0.749** 0.705* 1.096 
 Power couple 0.892 0.786 0.822 
Tertiary Education (ref. = no tertiary education) 0.932 1.187* 1.189 
Income Quantile (ref. = 40% to 59%)    
 0% to 19% 0.950 0.864 1.014 
 20% to 39% 0.973 0.883 0.846 
 60% to 79% 1.054 0.973 0.796 
 80% to 100% 1.103 1.185 0.947 
Labor Force Status (ref. = not in the labor force) 
 Employed 1.317*** 1.131 1.232 
 Unemployed 0.987 0.791 1.260 
Life Course Transitions    
 Married or partnered 0.743* 0.863 0.518* 
 Divorced or separated 1.248* 1.106 1.432* 
 Birth or adoption 1.082 0.917 1.327 
 Changed job 0.953 1.050 0.940 
Homeownership (ref. = homeowner)    
 Renter 1.911*** 0.920 2.413*** 
 Rent-free 1.640*** 1.092 3.222*** 
Personality Traits    
 Extroversion 1.022 1.058* 1.051 
 Agreeableness 1.009 0.945 0.937 
 Conscientiousness 1.050* 1.059 1.155** 
 Emotional stability 0.976 1.029 0.958 
 Openness 0.932** 1.072* 1.124* 
Migration Intentions (ref. = very unlikely to move)    
 Very likely to move 22.890*** 42.580*** 35.230*** 
 Likely to move 7.440*** 13.50*** 8.519*** 
 Unsure about a move 3.929*** 5.881*** 4.566*** 
 Unlikely to move 1.849*** 2.204*** 2.056*** 
Number of Observations 67,809   
Log Pseudo-Likelihood −2.490e+07   
Residential MobilityOnward MigrationReturn Migration
Migration History (continuous) 
 Number of prior residential moves 1.427*** 1.259*** 1.464*** 
 Number of prior internal migrations 1.140*** 1.608*** 1.661*** 
Female 1.037 1.145* 1.044 
Age (ref. = 15–24)    
 25–34 0.703*** 0.631*** 0.815 
 35–44 0.597*** 0.486*** 0.675** 
 45–54 0.679*** 0.586*** 0.582** 
 55–64 0.744*** 0.522*** 0.613** 
 65+ 0.932 0.758* 0.693 
Marital and Parental Status    
 Married or partnered (ref. = single) 0.991 0.944 1.221 
 Have children (ref. = no child) 0.964 0.698*** 0.564*** 
Household Structure (ref. = not in dual-income household) 
 Dual-income household 0.749** 0.705* 1.096 
 Power couple 0.892 0.786 0.822 
Tertiary Education (ref. = no tertiary education) 0.932 1.187* 1.189 
Income Quantile (ref. = 40% to 59%)    
 0% to 19% 0.950 0.864 1.014 
 20% to 39% 0.973 0.883 0.846 
 60% to 79% 1.054 0.973 0.796 
 80% to 100% 1.103 1.185 0.947 
Labor Force Status (ref. = not in the labor force) 
 Employed 1.317*** 1.131 1.232 
 Unemployed 0.987 0.791 1.260 
Life Course Transitions    
 Married or partnered 0.743* 0.863 0.518* 
 Divorced or separated 1.248* 1.106 1.432* 
 Birth or adoption 1.082 0.917 1.327 
 Changed job 0.953 1.050 0.940 
Homeownership (ref. = homeowner)    
 Renter 1.911*** 0.920 2.413*** 
 Rent-free 1.640*** 1.092 3.222*** 
Personality Traits    
 Extroversion 1.022 1.058* 1.051 
 Agreeableness 1.009 0.945 0.937 
 Conscientiousness 1.050* 1.059 1.155** 
 Emotional stability 0.976 1.029 0.958 
 Openness 0.932** 1.072* 1.124* 
Migration Intentions (ref. = very unlikely to move)    
 Very likely to move 22.890*** 42.580*** 35.230*** 
 Likely to move 7.440*** 13.50*** 8.519*** 
 Unsure about a move 3.929*** 5.881*** 4.566*** 
 Unlikely to move 1.849*** 2.204*** 2.056*** 
Number of Observations 67,809   
Log Pseudo-Likelihood −2.490e+07   

Notes: The data are from Waves 1–18 of the HILDA Survey. Standard errors, clustered by household, are not shown here The model includes state and year fixed effects.

*p < .05; **p < .01; ***p < .001

Confirming Hypothesis 1, the results in Table 4 indicate that migration experience not only stimulates intentions but also facilitates their realization. The RRRs are positive and statistically significant for return migration, confirming that past migrants are more likely to migrate than the general population because they return to previous regions of residence, presumably because of location-specific capital. More importantly, past moves are also associated with increased odds of migrating to a new region, with an RRR greater than 1.608 (p < .001) for internal migration. This finding suggests a process of learned migration: individuals who previously migrated internally are more likely to migrate to regions where they have never lived before. Thus, in line with Hypothesis 2, the positive impact of past moves on future migration is the result of both processes of learned behavior and location-specific capital, and the effect of migration experiences is greater for return migration than for onward migration.

Consistent with migration intentions, past interregional migrations bear greater weight on actual migration behavior than residential moves. This result holds for both onward and return migration, confirming the importance of long-distance migration experiences on subsequent migration behavior and lending support to Hypothesis 5. On the other hand, past residential moves are associated with greater odds of moving locally than migrating interregionally. Thus, residential moves are more important than long-distance internal migration in facilitating future residential mobility, perhaps because repeated exposure to the logistical challenges of moving locally presumably lowers the perceived costs and constraints of local moves. These results suggest that different types of moves might contribute to the development of skills that facilitate one type of move over the other.

We next explore whether the impact of past migration accumulates as individuals progress in their migration careers. Figure 3 reports predicted probabilities of migrating, with past moves as categorical variables; results from associated Wald tests are reported in Table A6 of the online appendix. The results indicate that each successive residential move increases the odds of engaging in residential mobility in the future in a broadly linear fashion. For interregional migration, having migrated once exerts the same effect on onward migration as having migrated twice, but having migrated three times has a significantly greater effect on onward migration than having migrated twice. Similarly, individuals who migrated four or more times are significantly more likely to migrate than those who migrated three times, suggesting that past migrations have a multiplying but not linear effect on future migration.

Fig. 3

Predicted probabilities from multinomial logistic regression models of migration. The data are from Waves 1–18 of the HILDA Survey. n = 67,809 person-years. The reference category is no moves. The number of past moves is entered as a series of categorical variables.

Fig. 3

Predicted probabilities from multinomial logistic regression models of migration. The data are from Waves 1–18 of the HILDA Survey. n = 67,809 person-years. The reference category is no moves. The number of past moves is entered as a series of categorical variables.

Close modal

Robustness Checks

In a robustness test, we examine how the regression coefficients on the variables of interest behave under modified regression specifications. The association between migration experience and subsequent migration behavior may be confounded by the lack of social capital among repeat migrants. Internal migrants are known to have smaller social networks built around peers, whereas less mobile individuals tend to develop larger networks centered on family members (Drevon et al. 2021; Viry 2012). Past internal migrants may therefore be more likely to migrate again simply because they lack social ties to root them in place. To ascertain the role of migration experience, we use duration of residence as a proxy for the accumulation of social capital and place-based attachment and examine the extent to which the results hold for individuals with varying duration of residence. We replicate the analysis reported in Tables 3 and 4 separately for individuals who have been in their current place of residence for less than 5 years, 5 to 9 years, and more than 10 years. These cutoff points are based on evidence that the odds of migrating initially increase with the duration of stay, peak at 5 years, and decline with the duration of residence thereafter (Thomas et al. 2016). Because some individuals did not move in any of the HILDA Survey's 18 waves, we measure duration based on a retrospective question on the duration of residence to ensure that duration of residence is not underestimated (Andrews et al. 2011).

The results in Figure 4 show that the association between migration intentions and the cumulative number of past residential moves and past internal migrations is statistically significant for all durations of residence. This finding indicates that the impact of one's migration experience is not simply the effect of reduced social capital or lack of place-based attachment among repeat migrants, lending further support to the hypothesis that migration is a learned behavior. Further, the effect of past moves on intentions increases with duration of residence, particularly for past interregional migration. Among individuals who are very likely to move, RRRs increase from 1.26 for a duration under 5 years to 1.74 for a duration of 5 to 9 years and to 2.20 for a duration of 10 or more years. Thus, as individuals grow more rooted and attached to a place, their migration experience becomes increasingly important in preventing cumulative inertia. Figure 5 shows a similar gradient for migration behavior. Of particular interest is onward internal migration, for which RRRs more than double from duration under 5 years to duration of more than 10 years.

Fig. 4

Relative risk ratios with 95% confidence intervals from multinomial regressions of mobility intentions, with the baseline outcome being very unlikely to move. The data are from Waves 1–18 of the HILDA Survey. Sample sizes are 25,758 for a duration under 5 years, 11,869 for durations of 5–9 years, and 29,710 for a duration of 10 or more years. The models include state and year fixed effects. The control variables included are the same as those in Table 3.

Fig. 4

Relative risk ratios with 95% confidence intervals from multinomial regressions of mobility intentions, with the baseline outcome being very unlikely to move. The data are from Waves 1–18 of the HILDA Survey. Sample sizes are 25,758 for a duration under 5 years, 11,869 for durations of 5–9 years, and 29,710 for a duration of 10 or more years. The models include state and year fixed effects. The control variables included are the same as those in Table 3.

Close modal
Fig. 5

Relative risk ratios with 95% confidence intervals from multinomial regressions of migration behavior, with the baseline outcome being no move. The data are from Waves 1–18 of the HILDA Survey. Sample sizes are 25,758 for a duration under 5 years, 11,869 for durations of 5–9 years, and 29,710 for a duration of 10 or more years. Models include state and year fixed effects. The control variables included are the same as those in Table 4.

Fig. 5

Relative risk ratios with 95% confidence intervals from multinomial regressions of migration behavior, with the baseline outcome being no move. The data are from Waves 1–18 of the HILDA Survey. Sample sizes are 25,758 for a duration under 5 years, 11,869 for durations of 5–9 years, and 29,710 for a duration of 10 or more years. Models include state and year fixed effects. The control variables included are the same as those in Table 4.

Close modal

Discussion and Conclusion

By focusing on year-to-year changes in place of residence, most empirical studies have examined the decision to migrate at one point in time, providing only a snapshot of individuals' migration trajectories. This approach stands in stark contrast with the idea that individual lives are long-term biographies unfolding over many years (Halfacree and Boyle 1993) and does not fully capture the insights of the life course perspective. Given these deficiencies, growing efforts have been made in the last decade to reconcile migration research with a central dimension of the life course approach: time (Coulter and Van Ham 2013). Recognizing that “individual action is embedded in a process of decisions and behaviors that occurs over time” (Bernardi et al. 2019:3), scholars have increasingly conceptualized migration as a trajectory that unfolds over individuals' life courses rather than as discrete, independent events (Coulter et al. 2016). This approach has been tested empirically mainly through sequence analysis, which traces successive moves over prolonged periods (Castagnone 2011; Coulter et al. 2011; Karhula et al. 2020; Stovel and Bolan 2004; Toma and Castagnone 2015; Vidal and Lutz 2018; Zufferey et al. 2021). This burgeoning literature has revealed a positive association between the number of past childhood and adulthood migrations and the probability of subsequently migrating (Bernard and Vidal 2020; Myers 1999) both domestically and internationally (Bernard and Perales 2021). These findings suggest a feedback loop between past and future migration, but empirical evidence remains thin and often lacks theoretical underpinning.

To advance our theoretical understanding of the links between past and future migration, we combined the migration aspiration–ability model (Carling 2002; Carling and Schewel 2018; De Jong 2000) with the learned behavior framework (Bailey 1989; Bernard and Perales 2021; Morrison 1971) and argued that migration aspirations lead to migration intentions, which precede migration behavior. We contended that internal migration experiences support the acquisition of migration-facilitating skills, perceptions, and networks that enable the formation and realization of aspirations to migrate to both new and previous regions of residence by reducing the perceived costs and benefits of migration.

We empirically tested this proposition by applying multinomial regression models to longitudinal data from the HILDA Survey. The results confirm that past moves are associated with greater intentions of migrating and greater odds of migrating to a region of previous residence (return migration) or a new destination (onward migration), in line with Hypothesis 1. The positive association between the number of previous moves and onward migration suggests a process of learned behavior by which migration experiences facilitate future migration behavior, supporting Hypothesis 2. In Australia, where less than two thirds of mobility intentions are realized, the accumulation of internal migration experience contributes to explaining why some individuals realize their mobility intentions but others do not. A comparison of the regression coefficients for past moves with other well-established determinants suggests that the magnitude of these effects is not minor. It follows that empirical studies drawing on longitudinal data should at least control for the number of past moves when seeking to explain migration behavior.

The results also reveal that the influence of past moves increases with the distance moved, as Hypothesis 5 predicts. We interpret the greater weight of interregional migration (65+ kilometers) as resulting from the severance of social ties, which forces individuals to enter new social contexts, thus contributing to the development of skills and attitudes that facilitate the formation of migration intentions. Yet, the positive association between past residential moves and future interregional migration suggests that the experience of moving even over short distances matters. This association may be due to exposure to the logistical challenges of moving, which presumably lowers the perceived costs and constraints of moving. Equally important is the finding that past residential moves play a greater role in shaping future residential relocation, whereas past internal migrations are more closely associated with future long-distance migration than residential mobility. These results suggest that different types of moves might contribute to the development of different skills that distinctly affect the migration decision-making process.

The impact of migration experiences may also vary with the circumstances in which they occurred. Future work should explore whether some migrations more strongly influence future migration behavior because they align with key developmental periods in childhood, adolescence, and young adulthood or because they are linked to a significant life course event caused by adverse circumstances (e.g., parental separation, poverty-induced residential instability) or positive experiences (e.g., upward occupational mobility, residential upgrade). The role of childhood migration experiences cannot be tested with data from the HILDA Survey because they do not include migration history before the individual's entry into the panel. This drawback can be addressed with population register and retrospective survey data, which provide complete migration history since birth, although these data do not contain information on migration intention. An alternative approach would be to include a retrospective life-history module by asking individuals who enter a longitudinal survey about their migration from early childhood to the time of the interview, as is currently done in the Germany Family Panel. Such data would also help to shed more light on the progressive impact of past migrations. In line with Hypothesis 3, our results suggest that the impact of past residential moves and interregional migrations is cumulative and point to a possible snowball effect, in contradiction with Hypothesis 4. However, this process is not linear. Our understanding of this process is constrained by the HILDA Survey's partial migration histories, which reinforces the need to improve migration data collection practices.

Another potential avenue for future research is to explore whether the strength of these effects varies depending on the national context. Our empirical analysis could be easily replicated in the British context by drawing on the UK Household Longitudinal Study, on which the HILDA Survey is based. However, both surveys are limited in (1) collecting data on migration likelihood rather than intentions and (2) collecting information on the likelihood of changing addresses with no information on preferred destinations. Thus, our analysis could not differentiate the impact of past moves on intentions to move locally versus regionally. Future efforts should therefore be geared toward improving survey instruments and items by recognizing that internal migration aspirations are multidimensional and best gauged with multiple complementary questions (Carling and Mjelva 2021).

Our theoretical framework embeds the internal migration decision-making process within a wider migration history. Future work should examine the mechanisms underpinning the links between past and future internal migration. This line of inquiry will require bespoke surveys, such as the Family tiMes survey in Switzerland, which retrospectively collected data on migration histories and social networks (Drevon et al. 2021). However, qualitative research may better inform an understanding of how individuals' migration experiences shape their skills and attitudes toward migration over time.

Notes

1

A well-known exception is China, where the household (hukou) registration system constrains internal migration (Bell et al. 2020; Chan et al. 1999; Tan and Zhu 2021; Wang 2004).

2

There are 107 SA4 regions covering the whole of Australia without gaps or overlaps. These include 18 nonspatial SA4 special-purpose codes comprising Migratory–Offshore–Shipping and No Usual Address codes for each state and territory, which are not used in this article.

References

Ajzen, I. (
1991
).
Theory of planned behaviour
.
Organizational Behavior and Human Decision Processes
,
50
,
179
211
Alvarez, M., Bernard, A., & Lieske, S. N. (
2021
).
Understanding internal migration trends in OECD countries
.
Population, Space and Place
,
27
,
e2451
. https://doi.org/10.1002/psp.2451
Andrews, M., Clark, K., & Whittaker, W. (
2011
).
The determinants of regional migration in Great Britain: A duration approach
.
Journal of the Royal Statistical Society: Series A (Statistics in Society)
,
174
,
127
153
.
Aslany, M., Carling, J., Mjelva, M. B., & Sommerfelt, T. (
2021
).
Systematic review of determinants of migration aspirations
(QuantMig Report, Project Deliverable D2.2).
Southampton, UK
:
University of Southampton
.
Bailey, A. J. (
1989
).
Getting on your bike: What difference does a migration history make?
Tijdschrift Voor Economische en Sociale Geografie
,
80
,
312
317
.
Bailey, A. J. (
1993
).
Migration history, migration behavior and selectivity
.
Annals of Regional Science
,
27
,
315
326
.
Bailey, M., & Sly, D. F. (
1987
).
Metropolitan–non metropolitan migration expectancy in the Unites States, 1965–1980
.
Genus
,
43
(
3–4
),
37
60
.
Bell, M., Bernard, A., Charles-Edwards, E., & Zhu, Y. (
2020
).
Internal migration in the countries of Asia: A cross-national comparison
.
Dordrecht, the Netherlands
:
Springer
.
Bell, M., Charles-Edwards, E., Bernard, A., & Ueffing, P. (
2018
).
Global trends in internal migration
. In Champion, T., Cooke, T., & Shuttleworth, I. (Eds.),
Internal migration in the developed world: Are we becoming less mobile?
(pp.
76
97
).
London, UK
:
Routledge
.
Bell, M., Charles-Edwards, E., Ueffing, P., Stillwell, J., Kupiszewski, M., Kupiszewska, D. J. P., & Review, D. (
2015
).
Internal migration and development: Comparing migration intensities around the world
.
Population and Development Review
,
41
,
33
58
.
Bell, M., & Ward, G. (
2000
).
Comparing temporary mobility with permanent migration
.
Tourism Geographies
,
2
,
87
107
.
Bernard, A. (
2017
).
Cohort measures of internal migration: Understanding long-term trends
.
Demography
,
54
,
2201
2221
.
Bernard, A., & Bell, M. (
2018
).
Educational selectivity of internal migrants: A global assessment
.
Demographic Research
,
39
,
835
854
. https://doi.org/10.4054/DemRes.2018.39.29
Bernard, A., Bell, M., & Charles-Edwards, E. (
2016
).
Internal migration age patterns and the transition to adulthood: Australia and Great Britain compared
.
Journal of Population Research
,
33
,
123
146
.
Bernard, A., & Perales, F. (
2021
).
Is migration a learned behavior? Understanding the impact of past migration on future migration
.
Population and Development Review
,
47
,
449
474
.
Bernard, A., & Vidal, S. (
2020
).
Does moving in childhood and adolescence affect residential mobility in adulthood? An analysis of long-term individual residential trajectories in 11 European countries
.
Population, Space and Place
,
26
,
e2286
. https://doi.org/10.1002/psp.2286
Bernardi, L., Huinink, J., & Settersten, R. A. (
2019
).
The life course cube: A tool for studying lives
.
Advances in Life Course Research
,
41
,
100258
. https://doi.org/10.1016/j.alcr.2018.11.004
Böheim, R., & Taylor, M. P. (
2002
).
Tied down or room to move? Investigating the relationships between housing tenure, employment status and residential mobility in Britain
.
Scottish Journal of Political Economy
,
49
,
369
392
.
Büchel, K., Puga, D., Viladecans-Marsal, E., & von Ehrlich, M. (
2019
).
Calling from the outside: The role of networks in residential mobility
.
Journal of Urban Economics
,
119
,
103277
. https://doi.org/10.1016/j.jue.2020.103277
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (
2012
).
Robust inference with multiway clustering
.
Journal of Business & Economic Statistics
,
29
,
238
249
.
Campbell, P. (
2019
).
Dispositional traits and internal migration: Personality as a predictor of migration in Australia
.
Journal of Research in Personality
,
78
,
262
267
.
Carling, J. (
2002
).
Migration in the age of involuntary immobility: Theoretical reflections and Cape Verdean experiences
.
Journal of Ethnic and Migration Studies
,
28
,
5
42
.
Carling, J., & Collins, F. (
2018
).
Aspiration, desire and drivers of migration
.
Journal of Ethnic and Migration Studies
,
44
,
909
926
.
Carling, J., & Mjelva, M. B. (
2021
).
Survey instruments and survey data on migration aspirations
(QuantMig Report, Project Deliverable D2.1).
Southampton, UK
:
University of Southampton
.
Carling, J., & Schewel, K. (
2018
).
Revisiting aspiration and ability in international migration
.
Journal of Ethnic and Migration Studies
,
44
,
945
963
.
Castagnone, E. (
2011
).
Transit migration: A piece of the complex mobility puzzle. The case of Senegalese migration
.
Cahiers de l'Urmis
,
13
. https://doi.org/10.4000/urmis.927
Chan, K. W., Liu, T., & Yang, Y. (
1999
).
Hukou and non-hukou migrations in China: Comparisons and contrasts
.
International Journal of Population Geography
,
5
,
425
448
.
Clark, W. A. V., & Lisowski, W. (
2018
).
Examining the life course sequence of intending to move and moving
.
Population, Space and Place
,
24
,
e2100
. https://doi.org/10.1002/psp.2100
Clark, W. A. V., & Lisowski, W. (
2019
).
Extending the human capital model of migration: The role of risk, place, and social capital in the migration decision
.
Population, Space and Place
,
25
,
e2225
. https://doi.org/10.1002/psp.2225
Cobb-Clark, D. A., & Schurer, S. (
2012
).
The stability of Big-Five personality traits
.
Economics Letters
,
115
,
11
15
.
Cobb-Clark, D. A., & Tan, M. (
2011
).
Noncognitive skills, occupational attainment, and relative wages
.
Labour Economics
,
18
,
1
13
.
Cooke, T. (
2008
).
Migration in a family way
.
Population, Space and Place
,
14
,
255
265
.
Cooke, T. (
2018
).
United States: Cohort effects on the long-term decline in migration rates
. In Champion, T., Cooke, T., & Shuttleworth, I. (Eds.),
Internal migration in the developed world: Are we becoming less mobile?
(pp.
101
119
).
London, UK
:
Routledge
.
Coulter, R., & Van Ham, M. (
2013
).
Following people through time: An analysis of individual residential mobility biographies
.
Housing Studies
,
28
,
1037
1055
.
Coulter, R., Van Ham, M., & Feijten, P. (
2011
).
A longitudinal analysis of moving desires, expectations and actual moving behaviour
.
Environment and Planning A: Economy and Space
,
43
,
2742
2760
.
Coulter, R., Van Ham, M., & Findlay, A. (
2016
).
Re-thinking residential mobility: Linking lives through time and space
.
Progress in Human Geography
,
40
,
352
374
.
Creighton, M. J. (
2013
).
The role of aspirations in domestic and international migration
.
Social Science Journal
,
50
,
79
88
.
DaVanzo, J. (
1981
).
Repeat migration, information costs, and location-specific capital
.
Population and Environment
,
4
,
45
73
.
DaVanzo, J. (
1983
).
Repeat migration in the United States: Who moves back and who moves on?
Review of Economics and Statistics
,
65
,
552
559
.
De Groot, C., Mulder, C. H., Das, M., & Manting, D. (
2011
).
Life events and the gap between intention to move and actual mobility
.
Environment and Planning A: Economy and Space
,
43
,
48
66
.
De Jong, G. F. (
1999
).
Choice processes in migration behavior
. In Pandit, K. & Davies Whithers, S. (Eds.),
Migration restructuring in the United States: A geographic perspective
(pp.
273
293
).
New York, NY
:
Rowan and Littlefield Publishers
.
De Jong, G. F. (
2000
).
Expectations, gender, and norms in migration decision-making
.
Population Studies
,
54
,
307
319
.
Drevon, G., Viry, G., Kaufmann, V., Widmer, E. D., Gauthier, J. A., & Ganjour, O. (
2021
).
Analysing the effects of residential mobility behaviours on the composition of personal network in Switzerland
.
Population, Space and Place
,
27
,
e2472
. https://doi.org/10.1002/psp.2472
Fielding, A. J. (
1992
).
Migration and social mobility: South East England as an escalator region
.
Regional Studies
,
26
,
1
15
.
Fouarge, D., Özer, M. N., & Seegers, P. (
2019
).
Personality traits, migration intentions, and cultural distance
.
Papers in Regional Science
,
98
,
2425
2454
.
Goldstein, S. (
1954
).
Repeated migration as a factor in high mobility rates
.
American Sociological Review
,
19
,
536
541
.
Goldstein, S. (
1964
).
The extent of repeated migration: An analysis based on the Danish population register
.
Journal of the American Statistical Association
,
59
,
1121
1132
.
Halfacree, K. H., & Boyle, P. (
1993
).
The challenge facing migration research: The case for a biographical approach
.
Progress in Human Geography
,
17
,
333
348
.
Hughes, G. A., & McCormick, B. (
1985
).
Migration intentions in the UK: Which households want to migrate and which succeed?
Economic Journal
,
95
,
113
123
.
Huinink, J., Vidal, S., & Kley, S. (
2014
).
Individuals' openness to migrate and job mobility
.
Social Science Research
,
44
,
1
14
.
Huttunen, K., Møen, J., & Salvanes, K. G. (
2018
).
Job loss and regional mobility
.
Journal of Labor Economics
,
36
,
479
509
.
Ivlevs, A., & King, R. M. (
2012
).
Family migration capital and migration intentions
.
Journal of Family and Economic Issues
,
33
,
118
129
.
Jokela, M. (
2009
).
Personality predicts migration within and between U.S. states
.
Journal of Research in Personality
,
43
,
79
83
.
Jokela, M. (
2014
).
Personality and the realization of migration desires
. In Rentfrow, P. J. (Ed.),
Geographical psychology: Exploring the interaction of environment and behavior
(pp.
71
87
).
Washington, DC
:
American Psychological Association
.
Karhula, A., McMullin, P., Sutela, E., Ala-Mantila, S., & Ruonavaara, H. (
2020
).
Rural-urban migration pathways and residential segregation in the Helsinki region
.
Finnish Yearbook of Population Research
,
55
,
1
24
.
King, R., & Skeldon, R. (
2010
).
‘Mind the gap!’: Integrating approaches to internal and international migration
.
Journal of Ethnic and Migration Studies
,
36
,
1619
1646
.
Kley, S. A., & Mulder, C. H. (
2010
).
Considering, planning, and realizing migration in early adulthood: The influence of life-course events and perceived opportunities on leaving the city in Germany
.
Journal of Housing and the Built Environment
,
25
,
73
94
.
Kulu, H., & Milewski, N. (
2007
).
Family change and migration in the life-course: An introduction
.
Demographic Research
,
17
,
567
590
. https://doi.org/10.4054/DemRes.2007.17.19
Lin, J.-P., Liaw, K.-L., & Tsay, C. (
1999
).
Determinants of fast repeat migrations of the labor force: Evidence from the linked national survey data of Taiwan
.
Environment and Planning A: Economy and Space
,
31
,
925
945
.
Long, L. (
1991
).
Residential mobility differences among developed countries
.
International Regional Science Review
,
14
,
133
147
.
Mann, P. A. (
1972
).
Residential mobility as an adaptive experience
.
Journal of Consulting and Clinical Psychology
,
39
,
37
42
.
McCollum, D., Keenan, K., & Findlay, A. (
2020
).
The case for a lifecourse perspective on mobility and migration research
. In Falkingham, J., Evandrou, M., & Valchantoni, A. (Eds.),
Handbook on demographic change and the lifecourse
(pp.
200
212
).
Cheltenham, UK
:
Edward Elgar Publishing
.
Mondain, N., & Diagne, A. (
2013
).
Discerning the reality of ‘those left behind’ in contemporary migration processes in sub-Saharan Africa: Some theoretical reflections in the light of data from Senegal
.
Journal of Intercultural Studies
,
34
,
503
516
.
Morrison, P. A. (
1971
).
Chronic movers and the future redistribution of population: A longitudinal analysis
.
Demography
,
8
,
171
184
.
Mulder, C. H. (
1993
).
Migration dynamics: A life-course approach
(Doctoral dissertation). Department of Social and Behavioral Sciences, Amsterdam Institute for Social Science Research,
Amsterdam, the Netherlands
.
Mulder, C. H. (
2018
).
Putting family centre stage: Ties to nonresident family, internal migration and immobility
.
Demographic Research
,
39
,
1151
1180
. https://doi.org/10.4054/DemRes.2018.39.43
Mulder, C. H., & Hooimeijer, P. (
1999
).
Residential relocation in the life-course
. In Van Wissen, L. & Dykstra, P. A. (Eds.),
Population issues: An interdisciplinary focus
(pp.
159
183
).
New York, NY
:
Kluwer Academic/Plenum Publishers
.
Mulder, C. H., Lundholm, E., & Malmberg, G. (
2020a
).
Young adults' return migration from large cities in Sweden: The role of siblings and parents
.
Population, Space and Place
,
26
,
e2354
. https://doi.org/10.1002/psp.2354
Mulder, C. H., Lundholm, E., & Malmberg, G. (
2020b
).
Young adults' migration to cities in Sweden: Do siblings pave the way?
Demography
,
57
,
2221
2244
.
Myers, S. M. (
1999
).
Residential mobility as a way of life: Evidence of intergenerational similarities
.
Journal of Marriage and the Family
,
61
,
871
880
.
Niedomysl, T., & Amcoff, J. (
2011
).
Is there hidden potential for rural population growth in Sweden?
Rural Sociology
,
76
,
257
279
.
Nowok, B., Van Ham, M., Findlay, A. M., & Gayle, V. (
2013
).
Does migration make you happy? A longitudinal study of internal migration and subjective well-being
.
Environment and Planning A: Economy and Space
,
45
,
986
1002
.
Oishi, S. (
2010
).
The psychology of residential mobility: Implications for the self, social relationships, and well-being
.
Perspectives on Psychological Science
,
5
,
5
21
.
Oishi, S., Kesebir, S., Miao, F., Talhelm, T., Endo, Y., Uchida, Y., . . . Norasakkunkit, V. (
2013
).
Residential mobility increases motivation to expand social network: But why?
Journal of Experimental Social Psychology
,
49
,
217
223
.
Qiu, T., & Chu, Y. (
2020
).
Age patterns of China's repeat migration
.
Frontiers of Economics in China
,
15
,
433
471
.
Rogers, T. (
1969
).
Migration prediction on the basis of prior migratory behavior: A methodological note
.
International Migration
,
7
(
1–2
),
13
19
.
Sánchez, A. C., & Andrews, D. (
2011
).
To move or not to move: What drives residential mobility rates in the OECD?
(OECD Economics Department Working Papers, No. 846).
Paris, France
:
OECD Publishing
. https://dx.doi.org/10.1787/5kghtc7kzx21-en
Schurer, S., Kassenboehmer, S. C., & Leung, F. (
2015
).
Do universities shape their students' non-cognitive skills?
(Life Course Centre Working Paper Series, No. 2015-24).
Brisbane, Australia
:
University of Queensland
.
Sell, R. R., & DeJong, G. F. (
1983
).
Deciding whether to move: Mobility, wishful thinking and adjustment
.
Sociology and Social Research
,
67
(
2
),
146
165
.
Shuttleworth, I., Stevenson, C., Bjarnason, Þ., & Finell, E. (
2020
).
Geography, psychology and the “Big Five” personality traits: Who moves, and over what distances, in the United Kingdom?
Population, Space and Place
,
27
,
e2418
. https://doi.org/10.1002/psp.2418
Spring, A., Mulder, C. H., Thomas, M. J., & Cooke, T. J. (
2021
).
Migration after union dissolution in the United States: The role of non-resident family
.
Social Science Research
,
96
,
102539
. https://doi.org/10.1016/j.ssresearch.2021.102539
Stovel, K., & Bolan, M. (
2004
).
Residential trajectories: The use of sequence analysis in the study of residential mobility
.
Sociological Methods and Research
,
32
,
559
598
.
Tan, Y., & Zhu, Y. (
2021
).
China's changing internal migration: Toward a China variant of Zelinsky's transition thesis
.
Geoforum
,
126
,
101
104
.
Thomas, M., Gillespie, B., & Lomax, N. (
2019
).
Variations in migration motives over distance
.
Demographic Research
,
40
,
1097
1110
. https://doi.org/10.4054/DemRes.2019.40.38
Thomas, M. J., Stillwell, J. C., & Gould, M. I. (
2016
).
Modelling the duration of residence and plans for future residential relocation: A multilevel analysis
.
Transactions of the Institute of British Geographers
,
41
,
297
312
.
Thomassen, J. A. K. (
2021
).
The roles of family and friends in the immobility decisions of university graduates staying in a peripheral urban area in the Netherlands
.
Population, Space and Place
,
27
,
e2392
. https://doi.org/10.1002/psp.2392
Toma, S., & Castagnone, E. (
2015
).
What drives onward mobility within Europe? The case of Senegalese migrations between France, Italy and Spain
.
Population
,
70
,
65
95
.
Van Der Gaag, N., & Van Wissen, L. (
2008
).
Economic determinants of internal migration rates: A comparison across five European countries
.
Tijdschrift Voor Economische en Sociale Geografie
,
99
,
209
222
.
Van Hear, N. (
1998
).
New diasporas: The mass exodus, dispersal and regrouping of migrant communities
.
London, UK
:
UCL Press
.
Vidal, S., & Lutz, K. (
2018
).
Internal migration over young adult life courses: Continuities and changes across cohorts in West Germany
.
Advances in Life Course Research
,
36
,
45
56
.
Viry, G. (
2012
).
Residential mobility and the spatial dispersion of personal networks: Effects on social support
.
Social Networks
,
34
,
59
72
.
Wagner, M. (
1990
).
Education and migration
. In Mayer, K. U. & Tuma, N. B. (Eds.),
Event history analysis in the life course research
(pp.
129
145
).
Madison
:
University of Wisconsin Press
.
Wang, F.-L. (
2004
).
Reformed migration control and new targeted people: China's hukou system in the 2000s
.
China Quarterly
,
177
,
115
132
.
Winkels, A. (
2012
).
Migration, social networks and risk: The case of rural-to-rural migration in Vietnam
.
Journal of Vietnamese Studies
,
7
(
4
),
92
121
.
Zufferey, J., Steiner, I., & Ruedin, D. (
2021
).
The many forms of multiple migrations: Evidence from a sequence analysis in Switzerland, 1998 to 2008
.
International Migration Review
,
55
,
254
279
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).

Supplementary data