Navigation – Plan du site

AccueilRubriquesSystèmes, Modélisation, Géostatis...2003Who are the long-distance commute...

2003
243

Who are the long-distance commuters? Patterns and driving forces in Sweden

Qui sont les navetteurs de longues distances ? Formes et forces motrices en Suède
Marianne Öhman et Urban Lindgren

Résumés

Le sujet de l’étude est d’explorer les facteurs déterminants des déplacements pendulaires sur de longues distances en comparaison de ceux effectués sur de plus courtes distances. Les navetteurs de longue distance sont les individus dont les coordonnées de leur lieu de résidence et lieu de travail sont séparées d’une distance minimum de 200 kilomètres. En réponse à l’augmentation du nombre de ménages double-actifs, les déplacements pendulaires sur une grande distance sont apparus comme une stratégie de mobilité pour gérer les distances croissantes entre le foyer et le lieu de travail. L’analyse empirique est basée sur un échantillon extrait d’une base de données longitudinales et individuelles qui regroupe les informations démographiques, socio-économiques et géographiques de l'ensemble population suédoise. L’échantillon comprend tous les individus, qui en 1994 appartiennent à la classe d’âge 19-64 ans, ont un emploi rémunéré, et dont les coordonnées géographiques de leur lieu de résidence et de travail sont valides. L’estimation d’un modèle logit binomial a permis de prendre en compte les effets des variables liées à l’individu, au foyer, et au contexte géographique. Les résultats de l’analyse montrent que le navetteur de longue distance a plus de chance d’être un homme, jeune, ayant effectué au moins trois ans d’études supérieures, qui a eu une expérience antérieure de déplacement domicile-travail sur de longues distances, a de très bas ou de très hauts revenus, a un conjoint diplômé d’études supérieures, n’a pas d’enfant, et a moins de chance de vivre dans une maison pavillonnaire. Il ou elle est plus susceptible de vivre dans une petite ou moyenne ville, et de travailler dans certains secteurs professionnels. Plus particulièrement, l’expérience précédente de déplacement domicile-travail sur de longues distances apparaît avoir un effet positif important sur la probabilité d’être un navetteur de longue distance. En conclusion sur les stratégies de mobilité et sur les styles de vie des navetteurs de longue distance, il est possible de suggérer que la diversité des forces motrices doive être trouvée dans les préférences de l’individu et celles de son foyer, et dans les facteurs structurels.

Haut de page

Notes de l’auteur

TITRE SUEDOIS : Vem är långpendlaren? Mönster och drivkrafter i Sverige

Syftet med studien är att belysa de faktorer som ligger bakom långpendling i jämförelse med pendling över kortare distanser. Långpendlare definieras i studien som individer med ett avstånd mellan de geografiska bostads- och arbetsplatskoordinaterna på 20 mil eller mer fågelvägen. I och med ökningen av tvåförsörjarhushåll har långpendling växt fram som en mobilitetsstrategi för att hantera de allt längre avstånden mellan hemmet och arbetsplatsen. Den empiriska analysen baseras på ett urval från en longitudinell mikrodatabas innehållande demografisk, socioekonomisk och geografisk information för hela Sveriges befolkning. Urvalet innehåller alla förvärvsarbetande individer i åldern 19 till 64 år som har korrekta geografiska koordinater för bostad och arbetsplats 1994. En binomial logitmodell har använts för att fånga upp effekterna av attribut relaterade till individen, dennes hushåll samt delar av den geografiska kontexten. Resultatet av analysen visar en högre sannolikhet för att långpendlaren ska vara man, vara yngre, ha minst 3-årig universitetstutbildning, ha tidigare erfarenhet av långpendling, ha en mycket låg eller mycket hög årsinkomst, ha en partner med universitetsexamen, ej ha barn, och bo i lägenhet. Långpendlaren bor också sannolikt i en liten eller medelstor stad och arbetar inom vissa näringsgrenar. Tidigare erfarenhet av långpendling har en starkt positiv inverkan på sannolikheten att långpendla. En sammanfattande diskussion om mobilitetsstrategier och långpendling som livsstil antyder att drivkrafterna bör sökas i kombinationen av individuella preferenser, hushållspreferenser och i strukturella faktorer.

Sökord: långpendling, mobilitet, binomial logit

Texte intégral

The authors would like to thank Mirjana Boskovic for assistance in the data process-ing.

Introduction

1In most western countries, mobility is closely connected to welfare. Almost unreflect-ingly, we seem to believe that increased mobility generates improved welfare, and that mobility is the norm. As a growing number of people are able to travel longer distances, services and jobs can benefit from economies of scale by more concen-trated location patterns. In turn, this trend reinforces the need for even more mobility in order to maintain levels of service etc. Accessibility to a wide range of jobs and services as a result of extensive mobility is regarded as an efficient way of utilizing resources but in surveys, people often state other factors to contribute considerably more to their welfare (Vilhelmson 2002). This hints at the importance of activities and relations close to home. The creation and cultivation of social relations to other indi-viduals and different associations at home or in the local community have been shown to be of importance (Low & Altman 1992). Mobility, as well as place attachment, are essential aspects of the welfare of individuals.

2Changes in society have profoundly altered the conditions of mobility. In some respects, this development is paradoxical since increasing mobility exists in parallel with more immobility. Almost by definition long-distance commuters contribute a great deal to mobility, but this very mobility is often triggered by the unwillingness of carrying out another type of mobility, namely migration. Improved infrastructure and structural changes on the labour market creating new working conditions are two fac-tors triggering increased mobility (Öberg 1997). Today, travelling is a natural part of many jobs. However, factors like increased female participation rates and employers’ growing dependence on the unique expertise of their employees may reduce mobility. Companies and organisations heavily dependent on human capital tend to be more willing to accept arrangements where employees do not necessarily have to live in close proximity to their work place. From the household’s point of view, the choice of residence is influenced by dual careers where attention must be paid to two parties. All these aspects hint at the complex system in which the phenomenon of commuting is embedded. We do not intend to model the entire mobility system of the household but will focus on long-distance commuting, which seems to be located in the inter-section of mobility and immobility. Shedding more light on this part of the system is important for the understanding of the system dynamics and its impacts on society.

  • 1 http://www.scb.se/sm/AM32SM0002_kommentarer.asp (January 14, 2003)

3The distances between home and work, as well as the number of commuters, have increased considerably in the post-war period (e.g. Mitchelson & Fischer 1987, Hogarth & Daniel 1988, Fransson 1991, Rouwendal & Rietveld 1994, Green et al. 1999, van Ommeren 2000, Horner 2002, King 2002). In Sweden, the number of commuters crossing municipality borders has, with a few exceptions, been steadily increasing every year since the mid-80s.1 Improvements in transportation technology and infrastructure have increased the reach of individuals searching for jobs in ex-tended geographical spaces (Yapa et al. 1971). From the individual’s point of view, increased mobility generates benefits as well as detriments; commuting is time-consuming and restrains the everyday life of the family, but it may also produce or maintain a higher household income and an enhanced standard of living. These and other factors influence the behaviour of individuals/households and their decision-making concerning mobility strategies, which creates a selection of individuals into different commuting distances. From this point of departure, the aim of the paper is to explore the determinants of long-distance commuting. Who are those people and in what ways do they differ from short-distance commuters? This question is analysed employing a binomial logit model estimated on a longitudinal micro database, which contains information about demography, education, household situation, employment, income, and geographic location for the entire Swedish population.

4In the next section, a brief discussion of commuting as a mobility strategy and conceptual model is provided, followed by a section presenting a definition of long-distance commuters and some descriptive statistics on commuting in Sweden. There-after follows the empirical analysis with a presentation of the dataset, the sampling procedure, the statistical model, and the chosen variables. The paper ends with a pres-entation of the results of the analyses and some concluding remarks.

Commuting as a mobility strategy

5The increase in commuting is likely to be related to new flexibilities in working prac-tices and work place locations (e.g. working hours and working at home part of the time). One reason why employees might be positive to this flexibility is related to the difficulties in combining family and work. Van Ham (2001) states that households are becoming less tolerant to migration because of the increase in the number of dual-earner and dual-career households. Sweden is one of those European countries where the male breadwinning system is weak and the share of female participation in the labour force is high (Sommestad 1997). The dual-earner household is very common, a development supported by government policies through a long-term expansion of the public sector. Concurrently, with the entrance of women on the labour market, day care facilities for children were provided.

6Generally, the difficulties in considering two careers vary with occupations. Oc-cupations requiring higher education are mostly concentrated to larger cities, and met-ropolitan areas are often considered to be the optimal location of dual-career house-holds (Snaith 1990, Fielding 1992, Green 1995, Green 1997). Other occupations, e.g. female-dominated occupations within the public sector, have a more even distribution across space, which influences the mobility strategies of households in other ways (Halfacree 1995).

7A growing number of jobs are becoming more mobile. Many employees fre-quently go to conferences and meetings for different periods of time, or travel be-tween two or more work places every week. Individuals with more creative kinds of professions (as opposed to traditional, stationary jobs in production) are travelling between company head quarters and offices around the country, dealing with informa-tion, communication, managerial and analytical work. This complicates the definition of long-distance commuting; should the working practices of travelling be considered as actual commuting between several work places at different locations, or as business trips within the work. Though the incentives behind the travels might differ, the time-geographical outcome may be the same.

8The housing situation plays a vital role for the choice of mobility strategy. As compared to tenants, it has been shown that house owners have strong local bonds reflecting their mobility strategies (Stjernström 1998, Fischer & Malmberg 2001). Successively, time, money and social engagements are invested in their houses. The longer the family has been living in the same place, the stronger is their place-specific attachment. Moving elsewhere would imply a loss of this multi-dimensional invest-ment. In addition, there are other problems related to the housing issue. In ‘hot spot’ areas like larger cities, houses and flats are expensive and difficult to find, whereas in areas characterized by out-migration and weak labour markets, flats are empty and sometimes even torn down. A family moving from a rural area to a larger city may well have to face a situation of having to sell their house at an underrated price and not being able to find a new house of equivalent standard at a reasonable price within commuting distance to the new work in the large city.

9This mismatch between the housing market and the labour market partly ex-plains why weekly commuting may sometimes substitute migration (e.g. Hogarth & Daniel 1988, Rouwendal & Rietveld 1994, Green et al. 1999). When it is not possible to find work near home and when facing the choice between migration and unem-ployment, some people choose a third option – long-distance commuting – often in the form of weekly commuting with dual residence. The commuting distance should not merely be assumed to be a substitute for moving, however. Migration might not be an option or a consideration to the household at all. The decision might be between commuting short and long distances only. Long-distance commuting in different guises is not a new phenomenon (Gramling 1989, Skeldon 1990, Houghton 1993), but the level of transportation and communication technology might make it a more at-tractive alternative for households in order to combine the limited time for work, ca-reer, children, partners, domestic chores, and social life.

10Some people would argue that long-distance commuting is a certain way of life. Long-distance commuters are commonly regarded as highly mobile because they do not move. Their local attachments are supposed to lead to more mobility and longer travelling times. A long-distance commuter can reside in the countryside, enjoying the benefits of fresh air, tranquillity and lower housing costs and, at the same time, access better career opportunities and diversity of entertainment in a larger city. The price appears to be long, tiresome travels, periodical separation from other household members, shallower local attachments, and a feeling of rootlessness (Zelinsky 1971). We do, however, know little about the motives behind the commuting decision. One way of looking at the individual motives and structural driving forces behind this spe-cific mobility pattern is to divide mobility into what individuals can, must, and want to (Holm et al. 1989, Vilhelmson 2002). What people can do depends on transporta-tion and communication technology, their physical ability, and individual income and information resources. Modern infrastructure has enlarged the local labour markets; thereby making longer commuting distances a possibility. What people must do in order to earn their living and utilize different services depends on the dispersed local-ization of work places, housing, and shopping. What the individual perceives to be a must is also largely affected by norms and values. If income from work is a must, and the individual cannot find a home and work that are rather closely located, very long commutes can be acceptable if the time and money spent travelling are reasonable or the travels are undertaken less frequently. What people want to do might be to have a job, a better career opportunity, or simply visit other places for recreation. However, the issue of what people want to do is somewhat tricky. Individual preferences also mean that want differs between people. As in the case of must, individual want is in-fluenced by social norms. Living in nuclear families and returning home every night are often taken for granted as the individual’s own wish. In the junction point of want, must, and can, some individuals have found long-distance commuting and dual loca-tion to be their lifestyle and mobility strategy.

A conceptual model

11Together with the characteristics of the individual, the factors just mentioned regard-ing mobility strategy incentives constitute the main macro, meso, and micro level ex-planations and driving forces behind the choice of long-distance commuting. The most influent factors, as shown in Figure 1, can be summarized into seven categories representing personal characteristics, preferences and norms, household composition, social ties, labour market conditions, housing market conditions, and transportation.

Figure 1: A conceptual model of the driving forces behind long-distance commuting.

12Previous research shows individual characteristics to affect the commuting distance. Younger adults are generally more migratory, while the preference to stay because of local attachments seems to increase with age. The unwillingness to move and start over again at a new place probably makes long-distance commuting an increasingly attractive solution with age. This is not to say that young people leaving home have no local attachment, quite the opposite, but their goals and preferences at that stage in their life-course differ from those of households with children. Fransson (1991) showed the average age to be higher for those choosing to commute, as compared to movers, when faced with a job opportunity in another municipality. Others have found age to be negatively correlated to commuting distance (Rouwendahl & Rietveld 1994, Kullenberg & Persson 1997). All investigations concerning travels to work state that men travel longer than women (Madden 1981, Green et al. 1986, SAMPLAN 1995, Statistics Sweden 1996, Kullenberg & Persson 1997, Green et al. 1999, van Ham 2001, SIKA 2001). This gender difference is explained, for example, by men having less domestic responsibilities, and better-paid full-time jobs to compensate for the travel expenses (Madden 1981, Fagnani 1993, Hanson & Pratt 1995, MacDonald 1999). The length of the commuting distance is also related to the education level of the individual. University educated individuals are generally willing to travel longer to work (Gerstel & Gross 1984, Hogarth & Daniel 1988, Kullenberg & Persson 1997, Green et al. 1999, van Ham 2001). One of the reasons for this is that higher education often leads to better-paid positions which, in return, compensate for the travel costs. Foreign-born individuals commonly have weaker location-specific attachments in Sweden compared to natives, although this differs with the country of origin and the duration of stay in Sweden (Fischer et al. 1998). Foreign-born individuals are there-fore expected to have less obstacles to migration, and less incitements for long-distance commuting.

13Already Ravenstein claimed the immobility of families as one of his ‘laws of migration’ (Grigg 1977), showing the importance of household composition. The presence of a partner and children increases the total location specific attachments of a household. With the dual-earner household as a norm in Sweden, any move will create a search for two work opportunities in the same area. Children also cause an obstacle to migration, as a change of school and friends is rarely desired (Green 1997). Fransson’s study (1991) shows long-distance commuting to be more common among households with children in school age. Although single person households are not without local ties, the presence of partner and children should undoubtedly lead to a more complicated decision on where to live and work.

14Besides the ties to partner and children, an individual has social ties to friends, relatives, colleagues etc. at the location. Those social contacts can be just as valued, or even more, by the individual as having a job. Few people move to places where they do not know anyone. Information on the social network of an individual gives an un-derstanding of the degree of motivation to stay or move.

15Preferences and norms include both individual and societal preferences and norms. These are based on individual life values, some of which are shared by society while others are unique for each individual (Sobel 1981, Johansson & Miegel 1992, Hakim 2000). For example, in this case, they deal with the acceptance of how much time the individual can spend on work and commuting away from home, the prestige in jobs involving a great deal of travelling, and what kind of environment is suitable for raising children. The preferences and norms explain why some individuals strive for stability while others search flexibility and changes in life.

16The impact of the labour market on commuting behaviour concerns the kind of occupation and regional distribution. The capital and other major cities have a much more varied range of vacant positions to offer and lower unemployment rates as com-pared to the rest of Sweden. The major cities also hold a greater number of the more qualified occupations, which leads to dual career households having a preference for larger cities (Snaith 1990, Green 1997, Nilsson 2001). Other occupations, like educa-tion and healthcare, are more evenly distributed over the country. The flexibility in working practices also varies with the occupation. Some occupations and positions provide the opportunity to work from home one or more days a week. The issue of costs and income is inevitable, as travelling and maybe also dual living costs must be compensated for by a high-salary job or the costs covered by the employer (Hogarth & Daniel 1988, Fransson 1991). As the trend goes towards more short-term and pro-ject employments within several occupational sectors, the monetary and social costs of moving an entire household will probably be regarded as not being worthwhile. Labour market policies, manifested through removal subsidies, tax deduction for trav-els to work and dual living, etc. also influence people’s mobility behaviour.

17The housing market goes hand in hand with the labour market; where the la-bour market is ‘hot’, so is the housing market, and vice versa, consequently followed by the difficulties in buying or selling houses or flats, respectively, at acceptable prices. The fact that an individual is living in a small, detached house suggests that there is probably some money and time invested in the residence. House-ownership is a local attachment in itself and thus, an obstacle to migration, but a house generally requires more caretaking than a flat, which would obstruct longer absences from home. Changes in housing policies since the 1990s affect the market both in the indi-vidual commuter’s home region, as well as in the region of work. The state is no longer willing to buy migrant households out of their unsaleable houses in regions of weaker labour and housing markets. Cuts in construction subsides, as well as an in-creasing conversion of rental housing into tenant-owned co-operative housing, reduce the availability of rental flats, especially in over-crowded larger cities (Abramsson 2003).

18The probability of starting long-distance commuting should increase with the access to fast transportation, e.g. the closeness to a high-speed train station or car ownership. The time and money invested in commuting is probably traded-off against the salary check or the career progress achieved. The accessibility and choice of transportation mode are affected by infrastructure investments and governmental poli-cies regarding travel subsidies.

19The focus of this study is on personal characteristics and household composition in combination with some labour market and housing market attributes. The investiga-tion does not include social ties, preferences & norms, or transportation attributes.

Definition of the long-distance commuter

  • 2 http://www.scb.se/regioner/reg_tjanst/kom_fakt/kommunfaktadef.htm (2002-05-03)
  • 3 http://www.scb.se/statinfo/2000/be0205.asp (2002-08-28)

20What is then commuting, and what is the difference between short-distance and long-distance commuting? There does not seem to be any common agreement in mobility studies on how commuting should be defined. Statistics Sweden, for example, uses several alternative definitions. In labour market statistics, commuting is defined as travelling between a place of residence and a work place located in different municipalities.2 Regarding census data, Statistics Sweden defines commuters as those with their work place and their place of residence located in separate places.3 The former definition is also used by e.g. Fransson (1991) on the decision whether to migrate or commute.

21There is no definite division between short and long distances either. According to previous research, the commuting distance defined as long ranges from 45 km and more (van Ham et al. 2001), to 100 km and more (Abrahamsson 1993, Statistics Swe-den 1996). The diverging definitions are probably due to the difference in population density and country area. Another way is to define short distances as when commut-ing is possible on a daily basis, and long distances as exceeding the reach of daily commuting (e.g. weekly commuting).

22Commuting can be measured both in time and kilometres and, in fact, individu-als seem to be more sensitive to the time spent on travelling than the geographical distance covered (Vilhelmson 1994, Wang 2001). A study on professional Israeli women (Pazy et al. 1996) indicated that the respondents were more willing to increase their commuting distance than their commuting time. Any improvement in faster transportation may result in longer trips, using the same general amount of travelling time. According to previous studies, the time people are willing to travel varies, as does the distance. Some studies consider about 45 minutes up to an hour to be the maximum limit of time a person is willing to spend on commuting one way (Abra-hamsson 1993, Green et al. 1999), while others limit the range to between 25 to 35 minutes (Kenworthy & Laube 1999, van Ham et al. 2001 Schwanen & Dijst 2002). Schwanen & Dijst (2002) base their assumption on the fact that, on average, individu-als spend about ten percent of their daily working hours on commuting, which would make almost 30 minutes single way for an 8-hour workday.

23As the long-distance commuter spends plenty of time travelling each week, the mode of transportation becomes an important issue. The access to fast transportation modes and communication links is important in explaining the distances travelled (Wang 2001, Schwanen & Dijst 2002). There seems to be general agreement in trans-portation research that cost is one of the most important factors in explaining the choice of transportation mode (e.g. Ullman 1973, Hoyle & Smith 2000, Turton & Black 2000). Travelling costs both time and money. Unless the employer pays the trips, the work must be well paid to compensate for the expensive travels. The faster the transportation – the more expensive travelling generally is. Other important fac-tors concern the overall length of the trip, logistics, and convenience. Few travels in-clude just one undivided trip with only one mode of transportation. The accessibility to train stations, bus stops, and airports also includes the time and mode of getting there. Air travels always require another transportation mode to get to and from the airport. The puzzling of two or more transportation modes causes a logistic problem with the timing of schedules. Even though the car has been the fastest growing mode of transportation in the world (Giuliano 2000, Turton & Black 2000) and the most important one in explaining extended mobility, long distance commuters are travel-ling distances which are sometimes too far for car transportation. Airplanes, trains and express coaches are faster and more comfortable for longer trips, and combined with a laptop and a mobile phone they can sometimes be used for office work or even social contacts (Hogarth & Daniel 1988).

24This study is based on register data and commuting distance is calculated as the Euclidean distance between the coordinates of residence and the coordinates of work-place. Thus, it does not measure the actual distance, nor the amount of time spent travelling. The modes of transportation could unfortunately not be included in this study. It is not possible to assess the importance of transportation modes, because there are no records on infrastructure or transportation in the database. The limit be-tween short and long-distance commuting is set to 200 kilometres. The distance is deliberately set very high to differentiate between commuting on a daily and a weekly basis. Naturally, the distance limit for daily commuting is linked to different modes of transportation. This is why the generally used limit of 100 kilometres in Sweden (Abrahamsson 1993) was found to be too short. Many daily commuters in the more densely populated areas cross several municipality borders on their journey to work, but in some northern inland municipalities it is, at least theoretically, possible to travel 200 kilometres between home and work without crossing the municipality border. 200 kilometres was considered sufficiently long to disregard most commuting on a daily basis. However, the concept of long-distance commuting instead of weekly commut-ing will be used, as the register data do not tell when or how often the individuals travel.

Commuting in Sweden

25About 30 percent of all travels in Sweden in 1994 were related to work/school or business trips (Statistics Sweden 1996). On average, Swedish employees travel about five kilometres a day to and from their work place (Öberg & Springfeldt 1991), and every third employee must cross at least one municipality border to get to work (Öberg & Springfeldt 1991, Statistics Sweden 2001). According to labour market sta-tistics, commuting has increased during the 1980s and 1990s. In 1985, about 21.5 percent were commuting to another municipality and in 1999, the share had increased to 29.5 percent out of the almost four million people in the labour force (Statistics Sweden 2001). There was, however, a temporary decline following the economic re-cession in the early and mid-90s.

  • 4 Information given by Annika Persson, Utvärderingsenheten, The National Tax Board, April 26, 2002. T (...)

26There has been a number of studies dealing at least partly with long-distance commuting (for example, Vilhelmson 1990, Fransson 1991, Abrahamsson 1993, Sta-tistics Sweden 1996, Kullenberg & Persson 1997, Söderlind 1998, Eliasson et al. 2003, Lindgren & Westerlund 2003), but hardly any research has focused on long-distance commuting exceeding the reach of daily basis commuting in Sweden. Ac-cording to Fransson (1991) and Jansson (1999), there are approximately 200,000 long-distance weekly commuters who constitute about five per cent of the labour force. Costs due to dual living and semi-permanent residence are deductible in Swe-den, if the distance between the place of residence and the place of work exceeds 50 kilometres. The costs are only deductible for three years for cohabiting partners and one year for others (RSV 2001). Almost 80,000 people got a deduction for dual living or temporary work in 2000.4

Empirical analysis

Data source

  • 5 Total Population of Sweden Individual and Geographical Database. The TOPSWING database is held at t (...)

27The model estimations are based on register data compiled by Statistics Sweden into a longitudinal micro database called TOPSWING5, designed for research in the social sciences. The data is obtained from a number of different administrative records, for example, the censuses of 1985 and 1990, Register of the Total Population, and Regis-ter of Income and Capital. This anonymous database contains over one hundred demographic, social, economic, and geographic attributes of all registered Swedish citizens for eleven years during the period 1985 to 1995. The attributes are, for exam-ple, related to age, gender, educational level, occupational status, civil status, income and transfers, migration, immigration, emigration, housing, ethnicity, and household conditions. Members of the same household are related through a coded family iden-tity number in the database. Thus, attributes of spouses and children could be added to the analysis. Research on mobility is made possible by the geocoded records on resi-dential and work locations, available down to a 100 square metre resolution grid covering the entire territory. The possibility of linking geographic, economic and demo-graphic data at the household level gives the opportunity to obtain a more complete picture of the life situation of every individual and long-distance commuter. This is important as long-distance commuting is not merely assumed to be an individual decision, but a household strategy to handle work and family life. The municipality codes in the database on where individuals reside and work make it possible to attach addi-tional data from Statistics Sweden on average house prices, and regional unemploy-ment rates. Through the geographical coordinates, it has been possible to add a hierarchical classification of regions based on population size to define the place where the individual lives (Lindgren 2003).

Sampling procedure

28The analyses of long-distance commuters are based on a sample for 1994. The reason for choosing 1994 is that the TOPSWING database only holds the coordinates of the work place in that year. The sample has been derived from a stepwise procedure. In the first step, every individual between 19 and 64 is selected (5.2 million). In Sweden, citizens become legally competent at the age of 18 and most young people will leave upper secondary school at the age of 18 or 19. In the TOPSWING database, individu-als get their own household identity number the year they turn 19 (registration is thus made the year after they become legally competent), regardless of whether they live on their own, cohabit but are unmarried, or still live with their parents. Nineteen is also the age when most young people will look for their first job. Everyone from 19 years of age and upwards is therefore to be considered as his or her own household and economic unit. The upper limit is selected due to the administrative retirement age, 65 years. The age group chosen here is therefore 19 to 64.

29In the second step, the sample is restricted to those gainfully employed (remain-ing: 3.7 million). The definition of employment used in our study is the same as the one applied in ÅRSYS (Annual Employment Statistics). To be counted as employed in a specific year, the individual must, on average, have been working at least one hour per week during November. Furthermore, it must be possible to relate the employment to a principal workplace during that month.

30In the third step, individuals lacking residence or work coordinates were re-moved (remaining: 3.1 million). A number of individuals could not be included in the analysis, mainly because of non-valid coordinates. Some of the missing cases consist of registered individuals working in another country. For example, the known group of long-distance commuters working at oilrigs in the North Atlantic will not be in-cluded.

A brief description of long-distance commuting

31In 1994, forty eight per cent of the commuters have less than 5 km between their places of work and residence, and the average commuting distance is 2.0 kilometres. Most individuals obviously live close to where they work. The distribution of dis-tances is shown in Figure 2. In 1994, the share of long-distance commuters amounts to 1.5 per cent. In absolute numbers, there are about 47,900 long-distance commuters in 1994.

Figure 2: Commuting distance.

Description of variables and statistical model

  • 6 No strong correlations between the independent variables were found.

32For the modelling of who is a long-distance commuter, a binomial logit model was used to find any possible explanatory differences between the group of long-distance commuters and the group of commuters over shorter distances. Based on the concep-tual model in Figure 1, somewhat more than twenty attributes accessible in the data-base, concerning individual characteristics, household composition, labour market and housing market characteristics, were found to influence the commuting behaviour of the individual.6 The descriptions of those variables and the mean of each variable are presented in Table 1. The variables are grouped into long-distance commuters (LDC) and short-distance commuters (SDC).

Dependent variable

33The dependent variable Y is equal to 1 if there is a distance of 200 km or more be-tween the individual’s residence and work place coordinates. The logit formula for the probability of being a long-distance commuter is described below

.

Independent variables

34Most of the previous studies mentioned on long-distance commuting strongly indicate the typical long-distance commuter to be male, middle-aged, more highly educated, and working in certain well-paid occupations, e.g. managerial and professional. Those variables were also tested in this model, with the exception of type of profession that was not available in the database. Instead, six different sectors of employment were tested. Long-distance commuting is supposed to substitute for unwillingness to mi-grate, for reasons related to the partner’s work, children’s school, the housing situa-tion, and local attachments including social ties. Regarding the previous research re-sults, independent variables included in the model were supposed to give information on the partner’s occupational status and education, the age of children, type of dwell-ing, and regional housing and labour market situation. Added to the list of variables were experience of previous LDC, experience of unemployment, country of birth, and size of city.

Variable

Definition

LDC

SDC

MALE

Dummy variable=1 if the individual is male

0.682

0.512

NONSWEDE

Dummy variable=1 if the individual was born outside Sweden.

0.067

0.087

AGE19to26

Dummy variable=1 if the individual is between 19 and 26 years of age.

0.206

0.128

AGE27to45

Dummy variable=1 if the individual is between 27 and 45 years of age.

0.506

0.483

AGE46to64

Dummy variable=1 if the individual is between 46 and 64 years of age.

0.288

0.388

COMPSCHOOL

Dummy variable=1 if the individual has maximum comprehensive school education.

0.148

0.222

UPPERSEC

Dummy variable=1 if the individual has maximum upper secondary school or <3 years at university college.

0.650

0.641

UNIVERSITY

Dummy variable=1 if the individual has a university education of minimum 3 years.

0.197

0.134

COMMUTEXP

Dummy variable=1 if the individual was defined as LDC (>=200 km) one or two years previously.

0.476

0.013

HOUSE

Dummy variable=1 if the individual is living in a detached house.

0.564

0.614

CONSTRUC

Dummy variable=1 if the individual is employed in the construction sector,

sni 45100-45230.

0.038

0.027

DATACONS

Dummy variable=1 if the individual is employed in the data consultancy sector,sni 72200-72201.

0.012

0.006

MILITARY

Dummy variable=1 if the individual is employed in the military command sector, sni 75221-75225.

0.033

0.008

HEALTHCARE

Dummy variable=1 if the individual is employed in the healthcare sector,

sni 85100-85140.

0.047

0.103

MEDICALSALES

Dummy variable=1 if the individual is employed in the medical promo-tion/selling sector, sni 51460.

0.012

0.001

TEACHING

Dummy variable=1 if the individual is employed in the comprehensive or upper secondary school sector, sni 80100-80220.

0.013

0.055

INCOME1QUAR

Dummy variable=1 if the individual has an annual work income within the first quartile.

0.279

0.249

INCOME2QUAR

Dummy variable=1 if the individual has an annual work income within the second quartile.

0.148

0.252

INCOME3QUAR

Dummy variable=1 if the individual has an annual work income within the third quartile.

0.163

0.251

INCOME4QUAR

Dummy variable=1 if the individual has an annual work income within the fourth quartile.

0.411

0.247

UNEMPEXP

Dummy variable=1 if the individual has received any unemployment benefits the current year or the previous year.

0.306

0.144

CHILD

Dummy variable=1 if the individual has child/ren.

0.308

0.397

PRESCHOOL

Dummy variable=1 if the individual has a child of pre-school age, i.e.

0-5 years.

0.155

0.165

SCHOOL

Dummy variable=1 if the individual has a child of school age, i.e. 6-18 years.

0.153

0.232

SPOUSE

Dummy variable=1 if the individual has a spouse.

0.457

0.581

SPOUSEEMP

Dummy variable=1 if the spouse is employed.

0.368

0.488

SPOUSECOMP

Dummy variable=1 if the spouse has maximum comprehensive school educa-tion.

0.077

0.145

SPOUSEUPSEC

Dummy variable=1 if the spouse has maximum upper secondary school or <3 years at university college.

0.285

0.348

SPOUSEUNIV

Dummy variable=1 if the spouse has a minimum of 3 years at university.

0.092

0.085

REGIONUNEMP

Average unemployment rate in the labour market region of the individual’s residence.

11.175

10.967

HOUSEPRICE i

The logarithm of the average price of small houses sold in the residence municipality of the individual.

2.834

2.827

HOUSEPRICE j

The logarithm of the average price of small houses sold in the work place municipality of the individual.

2.904

2.843

BIGCITY

Dummy variable=1 if the individual is living in a city of minimum 200,000 inhabitants.

0.236

0.220

HINTERLAND

Dummy variable=1 if the individual is living within 40 km radius from a BIG-CITY.

0.166

0.171

OTHERCITIES

Dummy variable=1 if the individual is living in any city or city hinterland <200,000 inhabitants.

0.598

0.609

N

47,959

3,072,830

Table 1: Variable definitions and sample means.

Results

35In this section, the results of the binominal logit model are presented. First, the vari-ables with positive beta-coefficients are commented upon, followed by those with negative values. Besides the beta-coefficients, the change in the probability of being a long-distance commuter was calculated for each variable (see Table 2). The probabil-ity increases or decreases by the percentage point given for each independent variable. The model used to calculate the probability is presented in the Appendix. Last, the non-significant variables are briefly discussed.

36The variables with a positive effect on long-distance commuting in 1994 are: MALE, AGE19to26, UNIVERSITY, COMMUTEXP, MILITARY, MEDICALSALES, INCOME1QUAR, INCOME4QUAR, UNEMPEXP, SPOUSEUNIV, REGIONUNEMP and HOUSEPRICE j. The results of the variables related to the individual confirm earlier research results on males and the highly educated standing an increased prob-ability of being long-distance commuters (Hogarth & Daniel 1988, Wachs & Taylor 1993, Green et al. 1999, van Ham 2001, Wang 2001). The reasons why women gen-erally travel shorter distances are often part-time work, segregated labour markets, and their responsibilities connected to the domestic sphere, including child care (Madden 1981, Fagnani 1993, Hanson & Pratt 1995). The results in this study on edu-cational level were also in line with the majority of the LDCs, identified in other stud-ies, having an occupational profile requiring higher education (Gerstel & Gross 1984, Hogarth & Daniel 1988, Green et al. 1999).

37An important variable in explaining LDC is the previous experience of long-distance commuting, which increases the probability of being a long-distance com-muter by as much as 42 percentage points. Despite the intention of LDC being a tem-porary solution to a matching problem of housing and labour market, Green et al. (1999) found the average length of LDC among the respondents to be three and a half years. The result demonstrates that once you start long-distance commuting, there is a very high probability of continuing.

Variable

Beta

Sign.

Exp(b) prob*

Individual

MALE

.377

.000

.672

NONSWEDE

-.127

.000

-.177

AGE19to26

.139

.000

.220

AGE46to64

-.139

.000

-.192

COMPSCHOOL

-.131

.000

-.182

UNIVERSITY

.181

.000

.292

COMMUTEXP

3.928

.000

42.120

HOUSE

-.325

.000

-.412

Work and labour market

CONSTRUC

.042

.141

.063

DATACONS

.050

.313

.076

MILITARY

1.734

.000

6.440

HEALTHCARE

-.557

.000

-.635

MEDICALSALES

1.081

.000

2.796

TEACHING

-1.262

.000

-1.071

INCOME1QUAR

.184

.000

.298

INCOME2QUAR

-.140

.000

-.193

INCOME4QUAR

.704

.000

1.487

UNEMPEXP

.707

.000

1.496

Household

CHILD

-.145

.000

-.200

CHSCHOOL

-.158

.000

-.216

SPOUSE

.002

.929

.003

SPOUSEEMP

-.175

.000

-.238

SPOUSECOMP

-.145

.000

-.200

SPOUSEUNIV

.163

.000

.261

Region

REGIONUNEMP

.077

.000

.118

HOUSEPRICE i

-1.947

.000

-1.283

HOUSEPRICE j

3.428

.000

30.439

BIGCITY

-.084

.000

-.119

HINTERLAND

-.054

.002

-.078

Constant

-10.104

.000

LOG-L

N

-339 208

47 959

Table 2: Estimates of the logit model on who is a long distance commuter.

* See Appendix.

38Age is assumed to correspond to stages in the life course (Warnes 1992, Boyle et al. 1998), e.g. having children, and is therefore represented by three age category dummies. The first group intends to catch the age when individuals are at the top of their mobility, generally leaving the nest and busy with university studies or entering the labour market (Bengtsson & Johansson 1992, Johansson & Persson 2000). The average age of first time Swedish mothers was 27.2 in 1994 (Statistics Sweden 1994), and the second age group thus represents the age when people generally start families and are busy with both child-rearing and career. The last group represents the age when children are older or have left home. The youngest age group has a positive re-lation to LDC, while that of the oldest is negative. Age obviously has a negative linear effect on long-distance commuting. Individuals generally have their peak of migration between the age of 20 and 30 (Nilsson 2001). This high mobility of young people may very well include commuting. Many young people continue to live in their parental home during their 20s because of free residence and the domestic service provided by parents, or the inability to find their own dwelling at an affordable price, and then choose longer commuting distances because of the difficulty in finding a first job nearby.

39It can, however, also be suspected that there is a hidden problem with registra-tion of residence. The dominance of such young individuals being LDCs is probably related to the negligence in updating the national registration of their home address. Every year, many adolescents move from more sparsely populated areas to Stockholm and other larger cities to find work. The housing situation in Stockholm makes it dif-ficult to get a first-hand contract, however. Instead, most in-migrants to Stockholm and other larger cities end up with black market second- or third-hand contract flats, or moving around between friends’ places. This insecurity of mailing address proba-bly means that many individuals keep their residence registration at their parents’ ad-dress, while actually living and working in another city.

40Hogarth & Daniel (1988) as well as Green et al. (1999) found the occupation of long-distance commuters to have shifted from being dominated by construction work-ers, to largely involving managerial and sales persons. For the occupational sectors included, individuals working in the defence sector as well as in the medical sales sector had a higher probability of commuting. MEDICALSALES is the branch of sell-ing medical drugs and equipment. This has been a growing business within traditional commercial travelling. Mobility and organisational transfers are part of the occupa-tional advancement system within the military defence sector, MILITARY. With the end of the era of an accompanying wife, it can be assumed that many high positioned militaries are in fact long-distance commuters. Working within the branches of pro-moting and selling medical equipment, and within military defence, increases the probability of being an LDC by about 2.8 and 6.4 percentage points, respectively.

41Any unemployment experience − both short- and long-term unemployment are included − turns out to increase the probability of LDC, as do the high regional unem-ployment rates. An individual usually starts looking for work close to her residence, but after a period of unemployment, she will stretch her search area.

42Long-distance commuting and dual living increase the expenses for transporta-tion and housing and thus, the salary should be rather high to compensate for the costs involved. As expected, the higher income in the fourth quartile has a positive effect on long-distance commuting, as compared to the reference group of income in the third quartile. However, the first quartile also has a positive effect. The reason for this non-linear effect could be that many LDCs have been unemployed part of the current year before starting long distance commuting and thus, their annual income will seem low. Another possible explanation might be that many of the LDCs are, or have been pre-viously in the current year, owners of small firms, as the income variable does not include income from business but only wage earnings. The effect of age decreasing LDC could also help explain the positive relation between LDC and the lowest in-come quartile. Young individuals with low salaries or who have finished school and then entered the labour market late that year might be one explanation.

43The only household-related variable with a positive effect was if the spouse had an education at the university level. As mentioned in the introduction, dual career households where the spouses are highly educated have a somewhat limited labour market at the national level, which seems to promote longer commutes instead of mi-gration. Thus, education has a positive linear effect on long-distance commuting.

44The average price of small houses in the municipality of the work place has a considerable impact in the model. An increase in the average price increases the prob-ability of long-distance commuting by 30 percentage points. The inability to find the same living standards at an affordable price in the new work place area was one of the major reasons for weekly commuting given by the respondents in two British studies (Hogarth & Daniel 1988, Green et al. 1999). This is confirmed by the result in this study, and most certainly reflects the fact that most long-distance commuters bound for larger cities where housing is more expensive.

45The variables with a negative effect on long-distance commuting are: NONSWEDE, AGE46to64, COMPSCHOOL, HOUSE, HEALTHCARE, TEACHING, INCOME2QUAR, CHILD, SCHOOL, SPOUSEEMP, SPOUSECOM, HOUSEPRICE i, BIGCITY and HINTERLAND. Concerning the variables on individual characteris-tics, being born in another country turns out to have a negative effect on long-distance commuting. One explanation is that immigrants generally have a preference for living in the labour markets of the largest cities (Öberg & Springfeldt 1991, Andersson et al. 1998). Moreover, during an economic recession as in 1994, less people have an employment to commute to, and the foreign born are particularly exposed to difficulties in the labour market.

  • 7 The variable HOUSE is based on the database variable HUST (type of housing). HUST is not avail-able (...)

46As stated earlier, age has a decreasing effect on LDC, which is confirmed by the oldest age group being negative. Age should also claim its due, making long travels more exhausting and maybe not worth the cost. Those with few years in school are less likely to be long-distance commuters, as education at the comprehensive school level is negatively correlated to LDC. House ownership also decreases the probability of long-distance commuting, which is in line with the hypothesis that a house requires considerable work, and constitutes a burden if commuting means frequent absences.7

47Among the variables related to the labour market, working in the sectors of teaching and healthcare has a negative effect on long-distance commuting. Employ-ment in teaching, for example, decreases the probability of being a long-distance commuter by 1.1 percentage points. Occupations in healthcare and teaching are both within the public sector and equally distributed all over the country, thereby reducing the need to commute to other labour markets (Kullenberg & Persson 1997). Income in the second quartile also has a negative correlation.

48Several of the household-related variables have a negative effect. Having chil-dren and children of school age are both negative, which is probably related to the involvement required at home when you have children. Compared to the reference group with children of pre-school age, having children of school age is negatively related to long-distance commuting. Households with pre-school children quite often have one partner at home, for example on parental leave, which makes it easier to deal with household activities and domestic chores. When the children start school, more time must be devoted to their local leisure activities, thereby making it more difficult to commute long hours or frequently being absent. The probability of long-distance commuting decreases if the spouse is gainfully employed. Even if the dual earner household is the norm in Sweden, the fact that the spouse is employed might take some pressure off the individual to find work. The individual can afford to wait for a suitable work close to the residence, and does not need to stretch the searching area to find a work with high pay. The negative effect of the spouse only having comprehen-sive school confirms that the probability of long-distance commuting increases with the level of education.

  • 8 The definition of cities and hinterlands is taken from Lindgren 2003.

49Among the variables related to geographical attributes, living in a large city, liv-ing in a large city hinterland, and the average small-house prices in the home munici-pality were all negative. The conclusion follows that the long-distance commuter is more likely to live in a small or medium sized city or its hinterlands, and work in a larger city or its hinterland.8 This goes hand in hand with the findings of higher aver-age house prices in the municipality of residence decreasing the probability of long-distance commuting by 1.3 percentage points.

50Three variables were not significant and thus, poor indicators of long-distance commuting over 200 kilometres: having a spouse, and working in the construction and data consultancy sectors. Apparently, having a spouse or not has no significant influence on commuting behaviour. It is possible that different couples have different views on the issue of marriage and long-distance commuting. Some do not want to commute long distances because they value their time together with their partner. Others choose long-distance commuting just because they want to preserve their mar-riage, instead of their dual located careers causing a divorce. This variable might have shown a slightly different result if it had been possible to include non-married, cohab-iting partners in the analysis. Two of the six occupational sectors tested were not sig-nificant. Construction work has traditionally been the best-known occupation in-volved in long-distance commuting. On the other hand, Hogarth & Daniel (1988) stated the construction sector as being a declining recruitment base of long-distance commuters. Data consultancies could be expected to travel a great deal since they are often hired in short-term project form but in the model, this occupational group showed no significant differences.

51The model was also tested with a commuting definition of 150 km or more be-tween residence and work place coordinates. The result was similar, the only differ-ence being that the independent variables of working in the sectors of data consul-tancy and construction were found to be significant. This hints at the stability of the estimates and the possibilities of transferring the results to other countries and spatial contexts.

Summary and conclusions

52The aim of this study has been to examine the determinants of long-distance commut-ing. To summarize the results of the logit model, the long-distance commuter can be characterized as a university educated, Swedish born, young male with a very high, or very low, income and some experience of unemployment and previous long-distance commuting, living in a flat in, or near, a small- or medium-sized city, with a univer-sity educated spouse. The probability increases if the individual is working in a municipality with high-priced small houses, and is living in a municipality of low-priced small houses, and a local labour market region with high unemployment rates. The long-distance commuter is less likely to have children, but if he or she does, the child is probably of pre-school age. The result can be exemplified by the probability of being a long-distance commuter, as calculated for fictitious individuals belonging to different groups in society.

53The young, single male: this male in his early 20s is living on his own in a flat in the centre of a large city. He has a university degree and works in medical sales and promotion. As he has just started working, his annual salary is in the first quartile. Given all this information, the probability of the young, single male being a long-distance commuter is about 4 percent.

54The married female with children: this 35-year old female is living with her husband and two children in a house in a middle-sized city. She is working as a schoolteacher, and has an income in the second quartile. Her husband is also em-ployed. Both adults have a university diploma. One of the children is of school age. The probability of this family-oriented female being a long-distance commuter is close to zero (0.102).

  • 9 The geographical attributes of regional unemployment rates and the logarithm of the average munici- (...)

55The married, middle-aged male: a male in his late 50s is married, with two chil-dren in their late teens. He has an upper secondary school education and works in the defence sector with an income in the fourth quartile. His spouse has an upper secon-dary school diploma but no employment. He lives in a house in a small city and has previous experience of long-distance commuting because of his work. The probability of the middle-aged male is 72 percent9.

56The single mother: this 25-year old female is living in a flat in the suburb of a larger city. She is single with one child of pre-school age. She has comprehensive school only and works in healthcare, with an income in the third quartile. The prob-ability of the single mother is less than 0.5 percent.

57As will appear from the four stereotype examples, previous experience of long-distance commuting has a tremendous effect on the probability of being a long-distance commuter. Even though other attributes, like sector of employment, also have an impact on the probability, the experience of previous LDC is what makes the overwhelming difference between the ‘married, middle-aged male’ and the other ex-amples. This is shown by the fact that if the ‘married, middle-aged male’ had been set to having no previous experience of LDC, all other attributes the same, the probability would drop to five percent. Concerning the other examples, the married parent with children is less likely to commute as compared to the young single male and the sin-gle mother. The often-made assumption that long-distance commuting is a substitute for migration found no clear support in this study. Attributes like being older, having children, and living in a house are often considered as factors attaching the individual to the local place. All three of those variables did, however, have a negative effect on long-distance commuting.

58Returning to the conceptual model, the results certainly stress the combined push- and pull forces of the labour and housing markets, and there is certainly a re-gional perspective. The LDCs find their opportunities in the largest cities with a greater demand, especially for professional and managerial occupations. They are more likely to reside in smaller cities constrained by some unemployment and less diversified labour markets, pushing them out to seek for work elsewhere. But the sim-ple choice of moving to larger cities from the rest of the country is constrained by the housing markets. The difficulty in selling small houses and owner-occupied flats in economically weaker regions pulls the household back from leaving. The problems in finding a comparable standard of housing in larger cities push the household away from entering that labour market. Considering the household composition and per-sonal characteristics, there are also regional push and pull factors involved in the deci-sion where to live. Young, single individuals are probably more attracted to the op-portunities of life in the large city and temporary housing solutions, than households with children. The latter seem to seek more children-friendly environments and small-house residence outside the larger cities (Kupiszewski & Rees 1999). This study does not include transportation, social ties, and norms and preferences, but part of the driv-ing forces behind long-distance commuting must be sought there.

59Disregarding transportation, this leads us back to our discussion on can, must, and want. The driving forces behind our life-style and mobility strategy do not only include individual preferences, but also the wishes of other individuals and the struc-tural factors in our society. Obviously, it is possible for quite a large number of indi-viduals to commute very long distances. They only constitute a fraction of the total labour force, but in numbers they are almost 50,000, and in reality most likely many more. Just the very possibility that long-distance commuting (and dual living) can be carried out as a mobility strategy, thanks to transportation techniques and economic prerequisites, should not be underestimated as an underlying driving force. Not all individuals can commute long distance, however. Other factors such as occupational sector are also of importance. Many occupations and employers allow for flexible working hours and a flexible working place, thereby facilitating commuting. Other work places are less flexible in working hours and have shift work, for example health care and schools, which should therefore decrease the probability of long-distance commuting.

60How mobility is influenced by what an individual must do is somewhat less clear. The basic needs of sleep and food require a certain amount of hours every day. Besides that, adult individuals must make a living, i.e. must have an income. There is, however, some uncertainty between what an individual must do, and what she feels or interprets she must do. Earning a living is such a thing. Living in a welfare state, the individual would certainly not starve even if rejecting a job opportunity, and waiting a few more months to find work closer to home.

61Most likely, an individual accepts a job opportunity in another city because she or he wants to. The work might be well paid, a great career opportunity, or there might be any other reason attractive to the individual. Long-distance commuters could be individuals valuing the life-style of travelling and dual living more than the local attachment. The place of the job opportunity could, however, be unattractive to the individual or other members of the household, who therefore do not want to move.

62As this study is carried out from register data, the intentions, preferences, and driving forces behind individual decisions can only be speculations. The combination of can and must probably constitutes the foundation of what an individual wants. Liv-ing in a social context, structural factors such as the national economic situation, la-bour market conditions, and societal norms regulate what is possible and desirable to do. Periods of economic recession, as the beginning of the 1990s, cause a lack of va-cant positions in all kinds of work, even in larger cities. In a time where tenured ap-pointments become scarce, and temporary time-limited employments common, more insecurity is added to the individual’s decision on where to locate home and work. Finally, changes in societal norms should not be forgotten. The very knowledge of other individuals being long-distance commuters may have a great influence on what an individual interprets that she or he can and want to do. Many occupations and posi-tions requiring considerable travelling are also commonly regarded as prestigious. Even though no numerous increases in long-distance commuting are expected, the norms might be changing, regarding the must to return to the ‘nest’ every day, and where to locate work and home. Further research is necessary to find out what the wishes and preferences of long-distance commuters are. Register data is too limited for this purpose, and interviews or questionnaires are needed to examine the life style and values of long-distance commuters.

Haut de page

Bibliographie

Abrahamsson, K.V. (ed.) (1993): National Atlas of Sweden: Work and Leisure. Alm-qvist & Wiksell International, Stockholm.

Abramsson, M. (2003): Housing Careers in a Changing Welfare State. GERUM 2003:1, Umeå University.

Andersson, E., Borgegård, L-E. & Hjort, S. (1998): Boendesegregation i de nordiska huvudstadsregionerna. GERUM 1998:2, Umeå University.

Bengtsson, T. & Johansson, M. (1992): Befolkningens geografiska rörlighet – min-skar den? In: 90-talets befolkningsfrågor. ERU, Ds 1992:16. Stockholm.

Boyle, P., Halfacree, K. & Robinson, V. (1998): Exploring Contemporary Migration. Addison Wesley Longman, Harlow.

Eliasson, K., Lindgren, U. & Westerlund, O. (2003): Geographical Labour Mobility: Migration or Commuting? Regional Studies (forthcoming).

Fagnani, J. (1993): Life Course and Space. Dual careers and residential mobility among upper-middle-class families in the Île-de-France region. In: Katz, C. & Monk, J. (eds.): Full circles. Geographies of women over the life course. Routledge, London.

Fielding, A.J. (1992): Migration and social mobility: South-East England as an escala-tor region. Regional Studies, 26, 1-15.

Fischer, P.A., Holm, E., Malmberg, G. & Straubhaar, T. (1998): Why Do People Stay? The Insider Advantages Approach. Empirical Evidence from Swedish La-bour Markets. Reprint of CEPR-discussion paper, No. 1952. University of Bundeswehr, Hamburg.

Fischer, P.A. & Malmberg, G. (2001): Settled People Don’t Move: On Life Course and (Im-) Mobility in Sweden. International Journal of Population Geography, 7, 357-371.

Fransson, U. (1991): Flytta eller pendla. Aspekter på hushållens rörlighet. Statens institut för byggnadsforskning, SB:39, Gävle.

Gerstel, N. & Gross, H. (1984): Commuter Marriage. Guilford Press, New York.

Giuliano, G. (2000): Urban Travel Patterns. In: Hoyle, B. & Knowles, R. (eds.): Mod-ern Transport Geography. 2nd edition. Wiley, Chichester.

Gramling, R. (1989): Concentrated Work Scheduling: Enabling and Constraining As-pects. Sociological Perspectives, 32, 47-64.

Green, A.E., Coombes, M.G. & Owen, D.W. (1986): Gender-specific Local Labour Market Areas in England and Wales. Geoforum, 17, 339-351.

Green, A.E. (1995): The Geography of Dual Career Households: A Research Agenda and Selected Evidence from Secondary Data Sources for Britain. International Journal of Population Geography, 1, 29-50.

Green, A.E. (1997): A Question of Compromise? Case Study Evidence on the Loca-tion and Mobility Strategies of Dual Career Households. Regional Studies, 31, 641-657.

Green, A.E., Hogarth, T. & Shackleton, R. (1999): Long Distance Living. Dual Loca-tion Households. The Policy Press. Bristol.

Grigg, D.B. (1977): E.G. Ravenstein and the “laws of migration”. Journal of Histori-cal Geography, 3, 41-54.

Hakim, C. (2000): Work-Lifestyle Choices in the 21th Century. Oxford University Press, Oxford.

Halfacree, K.H. (1995): Household migration and the structuration of patriarchy: evi-dence from the USA. Progress in Human Geography, 19, 159-182.

Hanson, S. & Pratt, G. (1995): Gender, Work, and Space. Routeledge, London.

Hogarth, T. & Daniel, W.W. (1988): Britain’s New Industrial Gypsies. Policy Studies Institute, London.

Holm, E., Mäkilä, K. & Öberg, S. (1989): Tidsgeografisk handlingsteori. Att bilda betingade biografier. GERUM rapport 8, Umeå University.

Horner, M.W. (2002): Extensions to the concept of excess commuting. Environment and Planning A, 34, 543-566.

Houghton, D.S (1993): Long-distance Commuting: a new Approach to Mining in Australia. The Geographical Journal, 159, 281-290.

Hoyle, B. & Smith, J. (2000): Transport and Development: Conceptual Frameworks. In: Hoyle, B. & Knowles, R. (eds.): Modern Transport Geography. 2nd edition. Wiley, Chichester.

Jansson, B. (1999): Migrate or Commute. Strategic household choices in Sweden dur-ing the 1990s. Paper presented at Association of American Geographers, 95th Annual AAG-99 Meeting, Honolulu, March 23-27 1999.

Johansson, T. & Miegel, F. (1992): Do the Right Thing. Lifestyle and Identity in Con-temporary Youth Culture. Lund University.

Johansson, M. & Persson, L-O. (2000): Lokala arbetsmarknader i konkurrens – arbetskraftens rörlighet under 1990-talet. SOU 2000:36, rapport 1. Stockholm.

Kenworthy, J.R. & Laube, F.B. (1999): Patterns of automobile dependence in cities: an international overview of key physical and economic dimensions with some implications for urban policy. Transportation Research A, 33, 691-723.

King, R. (2002): Towards a New Map of European Migration. International Journal of Population Geography, 8, 89-106.

Kullenberg, J. & Persson, L-O. (1997): Tänjbara räckvidder. Lokala arbetsmarknader i förändring. SIR, Östersund.

Kupiszewski, M. & Rees, P. (1999): Lessons for the projection of internal migration from studies in ten European countries. Statistical Journal of the United Nations ECE 16, IOS Press, 281-295.

Lindgren, U. (2003): Who is the Counter-Urban Mover? – Evidence from the Swedish Urban System. International Journal of Population Geography (forthcoming).

Lindgren, U. & Westerlund, O. (2003): Labour market programmes and geographical mobility: migration and commuting among programme participants and openly unemployed. Institute for Labour Market Policy Evaluation, IFAU, Working Paper 2003:6.

Low, S. & Altman, I. (1992): Place Attachment - A conceptual inquiry. In: Altman, I. & Low, M. (eds.): Place Attachment. Plenum Press, New York, 1-12.

MacDonald, H.I. (1999): Women’s Employment and Commuting: Explaining the Links. Journal of Planning Literature, 13, 267-283.

Madden, J.F. (1981): Why Women Work Closer to Home. Urban Studies, 18, 181-194.

Mitchelson, R.L. & Fischer, J.S. (1987): Long-Distance commuting and Population Change in Georgia, 1960-80. Growth & Change, 18, 44-65.

Nilsson, K. (2001): Migration Among Young Men and Women in Sweden. GERUM 2001:2, Umeå University.

Öberg, S. (1997): Theories on Inter-Regional Migration: An Overview. In: Blotevo-gel, H. & Fielding, A.J. (eds.): People, Jobs, and Mobility in the New Europe. Wiley, Chichester.

Öberg, S. & Springfeldt, P. (eds.) (1991): National Atlas of Sweden: The Population. Almqvist & Wiksell International, Stockholm.

Pazy, A., Salomon, I. & Pintzov, T. (1996): The Impact of Women’s Careers on their Commuting Behaviour: A Case Study of Israeli Computer Professionals. Trans-portation Research A, 30, 269-286.

Rouwendal, J. & Rietveld, P. (1994): Changes in Commuting Distances of Dutch Households. Urban Studies, 31, 1545-1557.

RSV (2001): Traktamenten och andra kostnadsersättningar. RSV 354 utgåva 11.

SAMPLAN (1995): Hur vi reser och varför. Om olika samhällsgruppers resmönster. SAMPLAN 1995:11, Stockholm.

Schwanen, T. & Dijst, M. (2002): Travel-time ratios for visits to the workplace: the relationship between commuting time and work duration. Transportation Re-search A, 36, 573-592.

SIKA (2001): RES 2000. Den nationella reseundersökningen/ National Travel Survey 2000. Stockholm.

Skeldon, R. (1990): Population Mobility in Developing Countries. Belhaven Press, London.

Snaith, J. (1990): Migration and Dual Career Households. In: Johnson, J.H. & Salt, J. (eds.) Labour Migration, 155-71. Fulton, London.

Sobel, M.E. (1981): Lifestyle and Social Structure. Concepts, Definitions, Analyses. Academic Press, New York.

Söderlind, J. (1998): Stadens Renässans. SNS Förlag, Stockholm.

Sommestad, L. (1997): Welfare State Attitudes to the Male Breadwinning System: The United States and Sweden in Comparative Perspective. In: Janssens, A. (ed.): The Rise and Decline of the Male Breadwinner Family? International Re-view of Social History, Vol. 42, supplement 5. Cambridge University Press.

Statistics Sweden (1994): Befolkningsstatistik 1994, del 4. Stockholm.

Statistics Sweden (1996): Riks-RVU: Svenskarnas resor 1994, resultatrapport. Stock-holm.

Statistics Sweden (2001): Arbetsmarknad – AM 32 SM 0102.

Stjernström, O. (1998): Flytta nära, långt borta. GERUM 1998:1. Umeå University.

Turton, B. & Black, W.R. (2000): Inter-Urban Transport. In: Hoyle, B. & Knowles, R. (eds.): Modern Transport Geography. 2nd edition. Wiley, Chichester.

Ullman, E.L. (1973): Geography as Spatial Interaction: Studies in Regional Devel-opment Cities, and Transportation. University of Washington Press.

Van Ham, M. (2001): Workplace Mobility and Occupational Achievement. Interna-tional Journal of Population Geography, 7, 295-306.

Van Ham, M., Mulder, C.H. & Hooimeijer, P. (2001): Spatial flexibility in job mobil-ity: macrolevel opportunities and microlevel restrictions. Environment and Planning A, 33, 921-940.

Van Ommeren, J. (2000): Commuting and Relocation of Job and Residence. Ashgate, Hants.

Vilhelmson, B. (1990): Vår dagliga rörlighet. TFB-report 1990:16. Allmänna För-laget, Stockholm.

Vilhelmson, B. (1994): Att undersöka geografisk rörlighet som en aspekt på rela-tionen livsstil – miljö. Occasional Papers 1994:2. Göteborg University.

Vilhelmson, B. (2002): Rörlighet och förankring. Geografiska aspekter på män-niskors välfärd. CHOROS 2002:1. Göteborg University.

Wachs, M. & Taylor, B.D. (1993): The Changing Commute: A Case-Study of the Jobs-Housing Relationship Over Time. Urban Studies, 30, 1711-1729.

Wang, F. (2001): Explaining intraurban variations of commuting by job proximity and worker’s characteristics. Environment and Planning B, 28, 169-182.

Warnes, T. (1992): Migration and the life course. In: Champion, A. & Fielding, A. (eds.): Migration Processes and Patterns. Vol. 1. Research Progress and Pros-pects. Belhaven, London, 175-187.

Yapa, L., Polese, M. & Wolpert, J. (1971): Interdependencies of Commuting, Migra-tion, and Job Site Relocation. Economic Geography, 47, 59-72.

Zelinsky, W. (1971): The Hypothesis of the Mobility Transition. The Geographical Review, 61, 219-249.

Haut de page

Annexe

Ceux qui souhaiteraient des références en français à l'étude des distance de migration et de navettes pourront en trouver dans les publications suivantes:

BACCAÏNI Brigitte, (1996), "Les trajets domicile-travail en Ile-de-France. Contrastes entre catégories socio-professionnelles", Economie et Statistique, n° 294-295, 109-126.

BACCAÏNI Brigitte, (1997), "Commuting and residential strategies in the Ile-de-France. Individual behaviour and spatial constraints", Environment and Planning A, vol 29, 1801-1829.

BACCAÏNI Brigitte, (1997), "Les navettes des périurbains d'Ile-de-France", Popula-tion, n°2, 327-364.

BACCAÏNI Brigitte, PUMAIN Denise, (1998), "Les migrations dans le système des villes françaises de 1982 à 1990", Population, n°5, 947-978.

BACCAÏNI Brigitte, (2002), « Navettes domicile-travail et comportements résiden-tiels dans l'espace francilien ». in L'accès à la ville. Les mobilités spatiales en ques-tion, Paris, L'harmattan, 121-140.

Haut de page

Notes

1 http://www.scb.se/sm/AM32SM0002_kommentarer.asp (January 14, 2003)

2 http://www.scb.se/regioner/reg_tjanst/kom_fakt/kommunfaktadef.htm (2002-05-03)

3 http://www.scb.se/statinfo/2000/be0205.asp (2002-08-28)

4 Information given by Annika Persson, Utvärderingsenheten, The National Tax Board, April 26, 2002. The National Tax Board, RSV, does not separate between deduction for “dual living” and “temporary work” in their statistics. The estimation of long-distance commuters is not necessarily that overrated though, as long-distance commuting may well be of a temporary kind.

5 Total Population of Sweden Individual and Geographical Database. The TOPSWING database is held at the Spatial Modelling Centre in Kiruna, Sweden

6 No strong correlations between the independent variables were found.

7 The variable HOUSE is based on the database variable HUST (type of housing). HUST is not avail-able for all years, and must be taken from the year 1993 for this purpose.

8 The definition of cities and hinterlands is taken from Lindgren 2003.

9 The geographical attributes of regional unemployment rates and the logarithm of the average munici-pality house prices were set neutral by the average value of the variable, i.e. the same value for all four individuals. Thus REGIONUNEMP was set to 11, HOUSEPRICE i to 2.83, and HOUSEPRICE j to 2.84.

Haut de page

Pour citer cet article

Référence électronique

Marianne Öhman et Urban Lindgren, « Who are the long-distance commuters? Patterns and driving forces in Sweden », Cybergeo : European Journal of Geography [En ligne], Systèmes, Modélisation, Géostatistiques, document 243, mis en ligne le 01 août 2003, consulté le 28 novembre 2020. URL : http://journals.openedition.org/cybergeo/4118 ; DOI : https://doi.org/10.4000/cybergeo.4118

Haut de page

Auteurs

Marianne Öhman

Spatial Modelling Centre, P.O. Box 839, SE-981 28 Kiruna, Sweden

Urban Lindgren

Department of Social and Economic Geography, Umeå University, SE-901 87 Umeå, Sweden

Articles du même auteur

Haut de page

Droits d’auteur

Licence Creative Commons
La revue Cybergeo est mise à disposition selon les termes de la Licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 3.0 non transposé.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search