Til Work Do Us Part: The Social Fallacy of Long-distance Commuting

Abstract

This paper focuses on the social implications of long-distance commuting on commuters and their spouses in Sweden. In a nation-wide study, the extent to which long-distance commuting increases the odds that couples will separate is investigated through event history analysis. Discrete-time logistic regression models were employed with longitudinal data on Swedish couples in 2000 to explore the odds of separation following long-distance commuting during 1995 to 2005. As expected, the results show that separation rates are higher among long-distance commuting couples compared with non-commuting couples. More complex results show that for men the odds of separating are highest if the commuting is on a temporary basis, and that women’s odds decrease when they continue commuting for a longer time-period. The long-distance commuting effect on relationships also varies depending on residential context.

Keywords

long-distance commuting social costs household separations longitudinal study event history analysis

1. Introduction

Commuting is an important aspect of daily life for many people. The majority of the workforce travels to and from work on a daily basis. Presently in Europe there is an increasing trend of long-distance commuting, both daily and weekly (Barker and Connolly, 2006; Lyons and Chatterjee, 2008; Renkow and Hower, 2000).

This development of increased commuting over longer distances is likely to continue in the coming years and will have consequences on both societies and individuals. The environmental effects of commuting, such as air pollution, urban sprawl and traffic congestion, are well known (Brueckner, 2000; Tolley, 1996; Travisi et al., 2010), as is the importance of commuting for one’s economic performance in the labour market (Krieger and Fernandez, 2006; Russo et al., 2007). That commuting longer distances is often associated with higher incomes and career opportunities for the individual is also acknowledged (So et al., 2001; van Ommeren et al., 2000).

Long-distance commuting also imposes significant costs on people and their social environment (Cassidy, 1992; Green et al., 1999). Regardless of the individuals’ motives, commuters must often spend less time socialising with family and friends as they are away from home a longer time, either daily or several days a week if they commute weekly. It is not always unproblematic for couples to manage to balance work and everyday life in a sustainable lifestyle when one or both of them commute long distances on a daily basis. The picture of the (long-term) social implications of long-distance commuting on commuters and their households is relatively unclear, however. The main reason for this is that many studies have mainly focused on either the personal well-being of the commuter him/herself (for example, health, stress), or economic aspects regarding commuting, often in terms of income and career achievements or when commuting is an alternative to migration. While there are several studies that focus on the increase in commuter marriages in which couples live apart during the week (Green et al., 1999; van der Klis and Karsten, 2009; van der Klis and Mulder, 2008), there are few, if any, longitudinal studies that assess the long-term effects of commuting on family relationships in terms of divorce/separation. Therefore, this article focuses on the behaviour of couples when one or both spouses are long-distance commuting (30 km or more). The aim is to analyse how long-distance commuting affects the risk of separation among Swedish couples.

The paper utilises a unique dataset with longitudinal data. This enables the empirical analysis of the effect of long-distance commuting on couples, which can hardly be done using questionnaire or interview surveys. Discrete-time logistic regression models were employed to register data on Swedish couples (two adults, with or without children) in 2000 and their commuting behaviour over a 10-year period (1995–2005) to analyse whether long-distance commuting appears to affect the duration of marriages/cohabitations ending in divorce or separation.

The rest of this article is organised as follows. The next section gives a brief overview of literature regarding social implications of long-distance commuting on people and households. Following this section, the data as well as the regression methods used to estimate the risk of separation are described. In the fourth section, some descriptive results and estimation results are presented and interpreted. These results are finally discussed in the concluding section of the article.

2. Theoretical Framework: How Does Commuting Affect Commuters and Their Households?

From a classical urban economic perspective, extra monetary and mental costs of commuting longer distances must be compensated for on the labour and housing markets in order for the commuters’ well-being or utility to be equalised (Alonso, 1964; Muth, 1969). For example, a long commuting distance can be compensated for by lower rents or housing prices (for example, Renkow and Hoover, 2000), desired housing or neighbourhood characteristics (for example, Plaut, 2006) or an intrinsically or financially rewarding job (for example, So et al., 2001; van Ommeren et al., 2000; Zax, 1991). Accordingly, commuting is then determined to be in an equilibrium state in the housing and labour markets in which the commuters are fully compensated with either higher wages or lower rents. On average, commuters’ subjective well-being or utility would then be the same regardless of their commuting time.

Commuting is more than simply covering a physical distance between home and work. Besides taking time, it also involves travelling costs and can affect individuals’ personal well-being in several ways and intervene in personal relationships.

2.1 Commuting and Subjective Well-being

There are studies that show that commuters on average are less satisfied with their lives than non-commuters are. In a study by Stutzer and Frey (2008), commuters’ subjective well-being was studied as a proxy for utility to test whether they are fully compensated for the stress they incur. They found that longer commutes systematically decrease life satisfaction. For example, a commuter who travels one hour each way would have to make 40 per cent more in salary to be as ‘satisfied’ with life as a non-commuter is. They therefore conclude that most commuters are not fully compensated for the stress they pay—a so-called commuting paradox. A Swedish study by Fults (2010) also suggests that longer commuting times have a negative effect on people’s subjective well-being. Several other studies also indicate that commuters generally experience a stressful lifestyle impacting their own psychological and physical well-being (Evans et al., 2002; Kluger, 1998) and that longer commuting time increases commuters’ stress levels (Costa et al., 1988; Gottholmseder et al., 2009; Kluger, 1998). Besides the travel time, one’s control over the commute has an impact on perceived stress. Commuting is more stressful when there is a lack of control and the journey involves unpredictability, such as traffic stocks, the driving behaviour of other road users (for car users) and unreliability of public transport services (Evans et al, 2002; Koslowsky et al., 1996; Kluger, 1998). A questionnaire study by Gatersleben and Uzzell (2007) shows that car drivers have the most stressful commutes, but public transit users had the most negative attitudes towards their travel mode, mainly due to boredom. Kluger (1998) found that car commuters with lengthy journeys are likely to be in a negative mood in the evening, implying that commuting stress can spill over into family life and overall well-being in several ways. Having experience of commuting, however, makes the commute more predictable and is thus found to lower the stress induced by commuting (Kluger, 1998). Analyses by Gottholmseder et al. (2009) of the effect of commuting in Austria on individuals’ stress perception found that commuters living in a partnership could handle stress better than single commuters could, but there were no gender differences. On the other hand, Roberts et al. (2009) found that women in the UK were more sensitive to commuting stress than men were, even when considering working hours and occupations and that this stress is a result of the gendered division of their everyday household tasks, including childcare.

While commuting can cause stress and health problems, in several studies Mokhtarian and colleagues (Mokhtarian and Salomon, 2001a, 2001b; Mokhtarian et al., 2001; Ory and Mokhtarian, 2005) have shown that the time spent between home and work can also be utilised by the commuter for something positive. Beyond the obvious utility of reaching work, there are utilities for activities that can be conducted while commuting—for example, doing work, or social networking through mobile information and communication technologies, or simply sleeping or enjoying the scenery. For the conventional commuting worker, the commute can also offer a natural transition between being at work and at home. Time spent in the car or on the bus, train or bike, or otherwise making one’s way from work to home can serve as a decompression period for commuters. There are qualitative studies that suggest that commuting might be understood as a gift rather than a curse (Jain and Lyons, 2008) and that some people even want to have longer commuting times in order to be able to use the travel time productively (Redmond and Mokhtarian, 2001). Those who derive positive utility from commuting are also found to experience the commute as less stressful (Gottholmseder et al., 2009) and experience less disutility of commuting (Lyons and Urry, 2005; Ory and Mokhtarian, 2005). Nevertheless, when commuting times become too long, the willingness to commute decreases (Sandow and Westin, 2010). There are also studies that show that commuters would like to decrease their commuting time, regardless of mode used (Sandow and Westin, 2010; Páez and Whalen, 2010; Redmond and Mokhtarian, 2001).

2.2 Commuting a Mobility Strategy for the Household

Commuting can be a mobility strategy to maintain social relationships and avoid migration. It can give people the opportunity to maintain a social network built up during years of living at the same place, which would be lost if they moved (Fisher and Malmberg, 2001). In post-modern society, shifting labour market structures involving increased specialisation, make it more difficult for both partners in a dual-career household to accommodate both spouses’ careers close to home. Commuting for one or both spouses in a household can therefore make it possible for each of them to have a career without moving. For some households, these complexities of the changing geographical labour mobility result in a ‘commuter marriage/partnership’ in which both partners pursue their careers while one lives near work part of the time because the commuting distance is too great to travel on a daily basis (Green et al., 1999; van der Klis and Karsten, 2009; van der Klis and Mulder, 2008).

While commuter marriages can be a solution for dual-career couples as they theoretically present equal opportunities for couples’ careers, for workers within the mining, oil and gas industries this lifestyle has rather been the result of structural changes within these employment sectors. Within these industries, mainly in Australia and Canada, long-distance commuting (also referred to as fly-in commuting) has become the dominant approach to new mining, oil and gas developments (Costa, 2004; Houghton, 1993; Storey, 2001). Rather than developing new settlements with services to accommodate workers and their families, as a means to reduce costs many workers are transported from metropolitan areas to remote sites to work for a certain number of days, after which they return home for a number of days.

In households with children, this mobility strategy with commuter marriages often means that the home-based parent (often the woman) limits his/her work time in order to juggle work and family commitments when the other spouse is away from home. While at least one partner can pursue a career with this geographical strategy, van der Klis and Karsten (2009) found that among commuter families in the Netherlands the commuting parent experiences sacrifices due to missing vital parts of everyday family life.

Commuting is generally a more difficult lifestyle when there are children living at home. When a parent is away from home for longer periods of time during the day, or for several days if the commute is weekly, he/she can feel guilty for missing vital daily parts of the children’s development (Rotter et al., 1998). There is also a mental distance to consider when long-distance commuting. Even if access to fast modes of transport make it possible to choose to work further from one’s home without having to increase travel time, there is a mental distance. If something happens at school or a child gets sick, it is important for parents of small children to know that at least one of them can manage to get home quickly (Friberg et al., 2004).

2.3 Commuting and Citizen Participation

Lengthy commutes leaving fewer hours for spare-time activities also have a direct negative impact on people’s involvement in community affairs and informal social interaction (Pocock, 2003, cited in Flood and Barbato, 2005, p. 7; Vilhelmson, 2002). Studies on the relationship between commuting and citizen participation in the US suggest that commuting is likely to reduce the time available for political activism. Putnam (2000), for example, argues that there is a negative relationship between commuting and civic engagement. He estimates that for each additional 10 minutes commuting, a person’s involvement in community affairs (such as attending public meetings or church services and chairing committees) is cut by 10 per cent. Lidström (2006), on the other hand, has found that this is not the case in a Swedish context. On the contrary, Lidström found that different forms of citizen participation are actually more common among commuters than non-commuters. But the results also suggest that this positive relationship holds only as long as the commuting time is limited. If the trend of increasing long-distance commuting continues in Sweden, more people will face more obvious time limits for citizen participation.

Overall, these studies imply that lengthy commuting means less time for interaction and socialising with family, friends and neighbours as well as for engagement in community affairs. While commuter partnership is an exceptional example of how family structures can change, life for many other couples is also affected by long-distance commuting. One question is then what the social implications of long-distance commuting on relationships are; whether or not couples can handle the costs of long-distance commuting in the long run.

3. Data and Method

Empirically, the study is based on geo-referenced longitudinal individual register data for the entire Swedish population. The database ASTRID contains annually updated information from Statistics Sweden on many individual demographic and socioeconomic attributes including family situation and members, earnings, work, employment and unemployment, support income and co-ordinates for place of living and work with 100 metres’ resolution. Information on travel time and travel modes is unfortunately not available, however. The definition of what constitutes a long-distance commute is therefore based on travel distance as a proxy for travel time. What constitutes a long commuting distance is, however, not clear and unambiguous. In this study, a one-way distance (Euclidean distance) of 30 kilometres or more is defined as a long-distance commute. This corresponds to an average of at least 45 minutes by car in Sweden,¹ which has been found to be the threshold for what constitutes an acceptable travel time both nationally in Sweden and internationally, as longer travel times have costs that impact daily life (see van Ommeren, 1996; Sandow and Westin, 2010; Wachs et al., 1993). This Euclidean distance between the co-ordinates of home and work does not measure the actual distance, which is about 30 per cent longer (Reneland, 1998), but is likely to correspond to about 35–50 kilometres on the ground depending on context (physical infrastructure, transport mode, traffic conditions, etc.). As the data are on an annual basis, it is not possible to distinguish between daily and weekly long-distance commuting; therefore, both groups are included in the analysis.

In total, the data consisted of all employed individuals aged 20–60 years in 2000. Persons not living with a partner in the year 2000 were excluded. The focus of this study is nuclear household relationships; thus only couples living together² (married or registered as cohabiting³ and having the same residential co-ordinates) were accounted for in the analyses. This gives a total of 2,143,256 persons, of whom 186,156 (9 per cent) were long-distance commuters at the time of sample (year 2000). Information on a number of demographic and socioeconomic characteristics of all these individuals in the sample was extracted from the database back to 1995 and up to 2005.

In this nation-wide longitudinal study, the extent to which long-distance commuting increases the risk of separation is investigated through event history analysis using discrete-time logistic regression. As Allison (1982) and Singer and Willett (1993) have shown, this is an adequate method for studying the occurrence of an event, such as a separation. The method is based on a person–year dataset, in which each person has multiple records (one for each year), instead of a person-oriented dataset as in standard logistic regression. Also unlike standard logistic regression, which examines the overall (unconditional) probability of an event without taking into account the timing of that event, discrete-time logistic regression models the conditional probability that the event will occur (i.e. the probability of separating at time t ₁ given that the relationship has lasted until t ₁). More importantly, discrete-time models have the ability to examine the impact of time-varying covariates, such as type of place of residence or income from year to year, on the outcome of an event. For example, an individual’s household composition (i.e. being a mother or a father) or income may change over time and these changes may have an effect on the risk of experiencing the event. All covariates (i.e. independent variables) used here are time-varying. The demographic covariates (Table 1) include age and children living at home each year. The socioeconomic covariates include highest attained education level, employment sector and income each year, deflated according to the value in the year 2000 of the Swedish crown and commuting status each year (commuting or not). Employment is defined as having an annual income from work of at least 50,000 SEK in order to exclude those who are not part of the workforce—for example, students working part-time. Furthermore, as the geographical structure of the Swedish local labour market is very diverse, from three major metropolitan regions down to small and sparsely populated regions, it was also tested whether different residential contexts have different impacts on separation for commuters and non-commuters. The choice of regions was based on the Swedish Agency for Economic and Regional Growth’s (2010) classification of regions into metropolitan regions, urban regions and rural regions. This classification is based on functional labour market regions in which people can live and work without too much time-consuming commuting. These functional regions are then grouped together based on factors important to development and growth (population size, education level, share of businesses and accessibility to employment opportunities) into five groups of regions: metropolitan regions, urban regions, rural regional centres, rural regions—private employment and rural regions—public employment. Here the three smaller regions are grouped into rural regions. Thus, the analysis estimates the probability of separation for each year as a function of these covariates. The outcome (event) of the study was divorce for married persons and a change from cohabiting status to living alone for persons registered as cohabiting.

Table 1.

Descriptive characteristics of couples, Sweden, 2000 (N = 2,143,256)

Characteristics	Women		Men
	Long-distance commuter (n = 188,306)	Non-commuter (n = 868,545)	Long-distance commuter (n = 312,301)	Non-commuter (n = 774,104)
Gender (percentages)	18	82	29	71
Average age	43	44	44	44
Family situation (percentages)
Children aged 0–6 years^a	32	27	33	33
Children aged 7–17 years^a	40	44	45	45
No children	41	41	36	36
Education level (percentages)
Junior high school (low)	11	15	16	21
High school (medium)	63	67	64	63
University education (high)	26	18	21	17
Income level (percentages) ^b
Low	48	59	16	24
Medium	32	28	39	45
High	21	13	45	31
Employment sector (percentages)
Primary or secondary	15	14	40	48
Private service	29	21	38	31
Public service	51	60	19	17
Other	6	5	4	4
Residential region (percentages)
Metropolitan	46	45	45	45
Urban	37	39	39	39
Rural	17	16	16	16

Can have children in both age groups.

Annual income level: Low 50,000–200,000 SEK; Middle 200,000–300,000 SEK; High 300,000+ SEK. A high income corresponds to approximately 31,300 €.

The logistic regression model for the estimates of separation by person i in the year t is

log [P_{i t} / (1 - P_{i t})] = α_{t} + β^{'} x_{i t}

where, α_t (t = 1, 2, ….) is the constant term; x_it the explanatory variable; and β is the logistic regression coefficient.

All persons were followed from the year 2000 until separation, death, death of spouse or the year 2005, and were then removed from the dataset (censored). Only the recorded marriage/cohabitation in 2000 was considered. This could be a second- or higher-order marriage if the persons had separated or been widowed before 2000. Because information on family status only goes back to 1995, we cannot know the duration of those marriages/cohabitations so it was arbitrarily set to one year. Using an indicator (dummy) it was tested whether or not the marriage/partnership had already started in 1995, but the commuting effects were always the same (not shown). Models were also estimated separately for those marrying/cohabiting before and after 1995, but since they gave very similar results they are not shown.

Individuals or households engage in long-distance commuting for a range of different reasons. As the individual’s or household’s motives for long-distance commuting are not observable in register data this paper concentrates on actual commuting behaviour and those who are in fact long-distance commuting. Several specifications of the long-distance commuting variable were used. In one model, a distinction was made between not commuting and commuting each year to explore the overall effect. In another model, distinctions were made between whether the person had been involved in long-distance commuting for five years or more or less than five years, or had not commuted during 1995–2005. All models were estimated separately for women and men.

4. Results

4.1 Descriptives

In total, 9 per cent of the couples long-distance commuted in 2000. About half (49 per cent) of these people were still long-distance commuting five years later. Overall, it was relatively common to commute for a longer period: 34 per cent of the couples studied long-distance commuted at least five years during the period 1995 to 2005. It was more common among men than women to be a long-distance commuter (Table 1).

Table 1 shows socioeconomic and geographical characteristics for long-distance commuters and non-commuters divided by gender. Like most people in Sweden, the majority of the long-distance commuting couples live in metropolitan regions. The share of highly educated and high-income earners is greater among long-distance commuters compared with the non-commuters. There are, however, large gender differences between the commuters. The share of highly educated female commuters is greater than that of highly educated male commuters. Despite these educational differences between commuting women and men, only a minority of these women are high-income earners, while most long-distance commuting men have a high income. In line with another study of long-distance commuters in Sweden (Sandow and Westin, 2010), this reflects that men benefit economically from long-distance commuting more than women do, although long-distance commuting generally has a positive effect on a person’s working income.

While the average age indicates that many couples are middle-aged, it is more common to have children, especially of pre-school age, among commuters than the non-commuting couples. This may imply that children increase place attachment and that long-distance commuting is part of a strategic mobility choice or solution to avoid migration.

Separation rates

The total number of persons who had separated at the five-year follow-up in 2000 was 236,446 (11 per cent), of whom 110,253 were women and 126,193 men.⁴ On average, long-distance commuter couples have a 40 per cent higher risk of separating than do non-commuting couples (14 per cent of the long-distance commuting couples separated compared with 10 per cent of the non-commuting couples). As shown in Figure 1, long-distance commuters’ annual separation rates are higher than those of non-commuters, even though the share of separations decreases with time for the studied couples.

Figure 1.

Separation rates: percentage of separations each year among studied couples.

As shown by Sandow and Westin (2010), long-distance commuting in Sweden seems to be a long-term mobility choice rather than a short-term one. The majority of long-distance commuters in Sweden commute for five years or more, and many of them commute for more than 10 years. It was therefore tested whether separation rates differ between couples in which one or both partners have a long duration of long-distance commuting compared with other couples. It was found that among couples in which the commuter(s) have a long duration (≥5 years) of long-distance commuting, fewer couples separated (11 per cent) compared with those having a short commuting duration (15 per cent). Separation rates were about the same for non-commuting couples (10 per cent) and couples having a long commuting duration (11 per cent). That the risk of separation is lower among the persevering commuters may reflect a customisation process (Rüger and Ruppenthal, 2010) whereby the many years of long-distance commuting have made the commute part of one’s lifestyle and one has learned to live with the experiences of the social and economical costs and benefits of commuting in everyday life. The gained experiences of what it means to long-distance commute may then be seen as worthwhile. A selection process could then explain that those who lack the ability to manage the stress and other costs caused by long-distance commuting have separated or stopped long-distance commuting to avoid the risk of a separation. Those couples separating during the first years of long-distance commuting may have separated anyway, also reflecting a selection process.

Overall, these differences in separation rates between the two groups of commuters are the same for each year (Figure 2). No gender differences were found regarding who the commuter was in a couple. For those couples who did not separate, it did not matter whether it was the man or woman who was the long-distance commuter.

Figure 2.

Separations by experience of long-distance commuting before marriage/cohabitation (percentages).

Previous experiences of commuting

About 20 per cent of the long-distance commuters were already commuting when they married or moved in with their partner.⁵ For these people, commuting with its pros and cons was probably already part of their everyday life. When the effect on separation of previous experience of long-distance commuting before moving in together/getting married was controlled for, a positive relationship was found. Among commuting couples, separation was less common if at least one spouse had previous experience of long-distance commuting before marriage/cohabiting, and was long-distance commuting at the time of marriage/cohabitation (16 per cent compared with 20 per cent for those without experience). Separation was least common if it was the female spouse who had experience of long-distance commuting prior to the relationship.

4.2 Regression Results

Overall commuting effects

Married and cohabiting men who long-distance commuted had 4 per cent higher odds of separating than did men who did not long-distance commute, for whom all the other observed variables were the same (Table 2). No overall effect of commuting was seen for women. Estimates with a sub-division of commuting duration (Table 3) show higher separation rates among long-distance commuting men only for a duration of less than five years. Men long-distance commuting over a longer time-period do not separate to a significantly higher extent than non-commuting men do. In contrast to the male long-distance commuters, a long duration of long-distance commuting for women is associated with an 8 per cent reduction in the odds of separating (Table 3).

Table 2.

Results from discrete-time logistic regression on the estimation of the effect of long-distance commuting, education, income, children, age and duration of partnership on the odds of separating

	Women	Men
	OR^a (95% Cl^b)	OR^a (95% Cl^b)
Age*
20–24	0.14 (0.12–0.15)	0.13 (0.12–0.14)
25–29	0.28 (0.27–0.29)	0.23 (0.22–0.24)
30–34	0.51 (0.50–0.53)	0.40 (0.39–03.42)
35–39	0.88 (0.86–0.90)	0.78 (0.74–0.78)
40–44 (ref.)	1 (ref.)*	1 (ref.)*
45–49	0.90 (0.88–0.92)	0.89 (0.87–0.91)
50–54	0.62 (0.60–0.64)	0.55 (0.53–0.56)
55 +	0.38 (0.37–0.40)	0.31 (0.30–0.32)
Children 0–6 years living at home *
Yes	0.31 (0.31–0.32)	0.06 (0.05–0.06)
No	1 (ref.)*	1 (ref.)*
Children 7–17 years living at home *
Yes	1.11 (1.09–1.13)	0.37 (0.37–0.37)
No	1 (ref.)*	1 (ref.)*
Education level *
Low	1.26 (1.22–1.30)	1.44 (1.40–1.48)
Middle	1.36 (1.34–1.39)	1.48 (1.45–1.51)
High (ref.)	1 (ref.)*	1 (ref.)*
Income level
Low	0.86 (0.84–0.88)	1.01 (0.99–1.03)
Medium	1.19 (1.16–1.21)	1.03 (1.01–1.05)
High (ref.)	1 (ref)*	1 (ref.)*
Employment sector *
Primary and secondary	1 (ref.)*	1 (ref.)*
Private service	1.10 (1.07–1.13)	1.02 (1.00–1.04)
Public service	1.13 (1.11–1.16)	1.12 (1.10–1.15)
Other services	1.09 (1.05–1.13)	1.03 (0.99–1.07)
Duration partnership (per year)	0.47 (0.47–0.48)	0.49 (0.48–0.49)
Long-distance commuter *
Yes	1.00 (0.98–1.02)	1.04 (1.02–1.05)
No	1 (ref.)	1 (ref.)*
Log likelihood	606,391.282	623,992.339
Pseudo R ² Nagelkerke	0.308	0.367

Odds ratio, significant values (p ≤ 0.01) are marked with bold text. Values >1 show increased odds.

95 per cent confidence interval.

indicates that the whole variable is significant at p ≤0.01—i.e. it shows whether or not the model as a whole becomes better when the variable is included in it.

Table 3.

Results from discrete-time logistic regression on the estimation of the effects of the duration of long-distance commuting on separating

	Women	Men
Long-distance commuter	OR^a (95% Cl^b)	OR ^a (95% Cl^b)
No	1 (ref.)*	1 (ref.)*
1–4 years	1.02 (1.00–1.04)	1.04 (1.02–1.06)
5 + years	0.93 (0.90–0.96)	1.02 (1.00–1.05)
Log likelihood	606,366.877	623,990.564
Pseudo R ² Nagelkerke	0.308	0.367

Odds ratio, values >1 show increased odds. Significant values (p ≤ 0.01) are marked with bold text.

95 per cent confidence interval.

indicates that the whole variable is significant at p ≤0.01—i.e. it shows whether or not the model as a whole becomes better when the variable is included in it.

Note: Only the effects of the commuting variable are displayed in this table, but all the variables shown in Table 2 were included in the models and the estimates for these variables were the same in both models.

Models (results not shown here) were also run to test the effect of long-distance commuting on relationships when both partners were involved in long-distance commuting as well as the effect of previous experiences of long-distance commuting. If both spouses were long-distance commuting, separation rates were lower. Having previous experience of long-distance commuting when starting a relationship was also found to reduce separation rates for both women (16 per cent) and men (11 per cent).

While long-distance commuting affects the odds of separation, other factors also have a significant effect on the probability of separating. For example, when other factors are controlled for, the longer the duration of marriage/cohabitation the lower the odds are of separating for both women and men. Also, those living in a family with small children (0–6 years) at home, when other factors are controlled for, separate to a lower extent than do couples with no small children at home.

Children

Couples with children are generally found to be more stable than childless couples and the lowest risk of separation is seen in couples with very young children (Andersson, 1997; Hoem, 1997). As a greater share of the long-distance commuters than the non-commuters have pre-school children, it was tested whether the effects of long-distance commuting on separation differ between childless couples and those with small children (0–6 years) living at home. The model was estimated separately for couples with small children in the household and for couples having no children. The results (not shown here) confirm that long-distance commuting increase the risk of separating for all groups except for women with small children; they actually lower their risk of separating when long-distance commuting over a longer time-period.

Geographical effects

The long-distance commuting effect on separation varies in different geographical contexts (see Table 4). For example, men living in metropolitan regions have 4–5 per cent higher odds of separating if they are long-distance commuting compared with men who are not commuting (but who have the same income level, duration of marriage, etc.). The highest separation rates for both women and men were estimated for short durations of long-distance commuting in rural regions. After five years of commuting, the divorce rates were equal to those of the general population of men and women living in rural regions. For women living in urban regions, separation rates are 11 per cent lower among persistent long-distance commuters (≥5 years) than among non-commuters. The point estimates suggested low divorce rates for women having a long duration of long-distance commuting in the other two residential regions, but significance was not attained.

Table 4.

Effect of long-distance commuting (A) and the duration of long-distance commuting (B) on the odds of separating by geographical area of residence

	Women			Men
	OR^a (95%Cl^b)			OR^a (95%Cl^b)
Residential region	Metropolitan	Urban	Rural	Metropolitan	Urban	Rural
A. Long-distance commuter
Yes	0.98 (0.96–1.01)	0.99 (0.96–1.02)	1.05 (1.01–1.10)	1.04 (1.02–1.07)	1.00 (0.97–1.02)	1.05 (1.01–1.10)
No	1 (ref.)	1 (ref.)*	1 (ref.)*	1 (ref.)*	1 (ref.)	1 (ref.)*
Log likelihood	298,770.03	220,075.68	86,451.814	299,366.12	232,001.03	91,252.662
Pseudo R² Nagelkerke	0.297	0.316	0.326	0.352	0.381	0.385
B. Long-distance commuter
No	1(ref.)	1(ref.)*	1(ref.)*	1(ref.)*	1(ref.)	1(ref.)
1–4 years	1.00 (0.97–1.03)	1.02 (0.99–1.05)	1.08 (1.03–1.14)	1.04 (1.01–1.06)	1.02 (0.99–1.05)	1.07 (1.02–1.12)
5+years	0.95 (0.91–0.99)	0.89 (0.84–0.94)	0.95 (0.87–1.05)	1.05 (1.02–1.09)	0.96 (0.92–0.99)	1.02 (0.96n-1.09)
Log likelihood	298,766.59	220,057.93	86,445.375	299,365.34	231,992.69	91,251.2
Pseudo R² Nagelkerke	0.297	0.316	0.326	0.352	0.381	0.385

Odds ratio, values >1 show increased odds. Significant values (p ≤ 0.01) are marked with bold text.

95 per cent confidence interval.

indicates that the whole variable is significant at p ≤0.01—i.e. it shows whether or not the model as a whole becomes better when the variable is included in it.

Note: Only the effects of the commuting variable are displayed in this table, but all the variables shown in Table 2 were included in the models and the relationship between these variables and separation had the same direction in both models.

It is not possible to explain within this study why a few years of long-distance commuting significantly increase the risk of separating for those living in rural regions. Overall, because the reasons these people start (or stop) long-distance commuting—as well as transport mode—are unknown in this study, it is not possible to say why these geographical differences exist in how long-distance commuting affects household relationships. It could be the case that the reasons people start long-distance commuting on average differs depending on where they live and that this has an effect on how couples can handle the consequences of long-distance commuting. For example, when someone lives in a smaller region his/her employment opportunities in the local labour market are more limited compared with those in big city regions. The reason to start commuting can therefore be more of a necessity in order to acquire a job and avoid migration. The social costs of long-distance commuting can therefore be experienced as great and the utility as low, and could thus affect the relationship more negatively than would be the case if, for example, commuting was chosen to receive a higher wage.

5. Concluding Discussion

Commuting is salient in many people’s everyday life. Although it has its advantages, numerous disadvantages can make the long-distance commuting lifestyle difficult. As the numbers of long-distance commuters are growing, it stands to reason that more couples will face the pros and cons of commuting in their daily routines. While some couples have to handle the consequences of long-distance commuting temporarily, for only a number of years, others will face and even adapt to a more long-term commuting lifestyle.

There are several reasons why long-distance commuting might be expected to affect separation rates, in either direction, and it appears that the separation effect is ambiguous. First, it seems as if the first years of long-distance commuting may be the most challenging for a relationship. Secondly, for those couples for whom long-distance commuting has been part of their lives for more than a few years, separation rates are lower. This offers support for the customisation process, whereby couples with time manage to adjust their lifestyle and develop strategies to handle the many presumable costs (both social and economic) of a long commute. It may be the case that for many the mobility choice of long-distance commuting is strategic and more socially sustainable than other alternatives, such as migrating and losing social networks and/or is a result of migrating and keeping one’s old job, etc., and they therefore have a higher tolerance for the commuting stress. On the other hand, a selection effect can explain that those separating relatively soon after choosing this mobility strategy lack these abilities to customise to the daily costs or would have separated anyway.

When other factors in the event history analysis are controlled for, the results also reflect gender differences. Male long-distance commuters can expect to separate to a higher extent than non-commuting men when commuting less than five years. However, in general much fewer women long-distance commute at all, reflecting that the gender expectations and structural constraints about breadwinning and parenthood still prevail. These norms are likely to make it more problematic for women than for men to long-distance commute, both practically and emotionally, which can cause women to consider long-distance commuting as a threat to their relationship. Nevertheless, those women long-distance commuting for many years run a lower risk of seeing their relationship broken. It may be the case that these women manage better than men do to adjust to the commuting lifestyle. A possible explanation for this may be that these women are not in a relationship with traditional views on household roles—i.e. that the woman should shoulder the main part of the domestic work and childcare and be the second wage-earner. They may, then, in contrast to the traditional gender differences in commuting, place more weight on labour market aspirations and less on domestic responsibilities, have support in household-related duties at home and therefore tolerate longer commuting times. Further, this suggests a more modern lifestyle in which the traditional gender differences in long-distance commuting are abandoned.

The results also showed that the long-distance commuter effect on relationships not only varied depending on whether or not the long-distance commuter was a woman but also on what geographical context the couple lived in. An explanation for these geographical differences can be that there are different causes for starting long-distance commuting depending on where you live, which has different outcomes on relationships.

To summarise, one might expect the social costs of long-distance commuting to reduce the quality of a relationship in many ways and thus increase the risk of separation. The statistical results from these analyses not only confirm such assumptions about social costs but can also reveal other and more unexpected results regarding the effects of long-distance commuting on relationships. Depending on whether it is the man or the woman who long-distance commutes, couples seem to handle the social costs of long-distance commuting in different ways. When it is the woman who is the persistent long-distance commuter, couples seem to manage to create a sustainable work–life balance although one partner is a long-distance commuter.

The causality between long-distance commuting and separation was not possible to establish within this study, however. How the long-distance commuter and his/her spouse can tolerate or manage the costs of long-distance commuting is of course influenced by external circumstances and societal changes such as illness, unemployment and changes in household composition. Furthermore, in order to understand why some couples can adapt to a commuting lifestyle and others separate after a number of years, there is a need for future research applying other methods such as interviews and questionnaire surveys. How the social cost of long-distance commuting is experienced by both the commuter him/herself and the spouse is of importance for planners and policy-makers striving to promote socially sustainable transport patterns.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Allison

P. D.

(1982) Discrete-time methods for the analysis of event histories, Sociological Methodology, 13, pp. 61–98.

Alonso

(1965) Location and Land Use: Toward a General Theory of Land Rent. Cambridge, MA: Harvard University Press

Andersson

(1997) The impact of children on divorce risks of Swedish women, European Journal of Population, 13, pp. 109–145.

Barker

Connolly

(2006) Long distance commuting in Scotland. Scottish Household Survey/Transport Research Planning Group Topic Report, Scottish Executive Social Research.

Brueckner

J. K.

(2000) Urban sprawl: diagnosis and remedies, International Regional Science Review, 23(2), pp. 160–171.

Cassidy

(1992) Commuting-related stress: consequences and implications, Employee Counseling Today, 4(2), pp. 15–21.

Costa

Pickup

Di-Martino

(1988) Commuting—a further stress factor for working people: evidence from the European Community, International Archives of Occupational and Environmental Health, 60(5), pp. 371–376.

Costa

(2004) A review of long distance commuting: implications for northern communities. Paper presented at the Canadian Institute of Mining, Metallurgy and Petroleum (CIM) AGM and Conference, Edmonton, May.

Evans

G. W.

Wener

R. E.

Phillips

(2002) The morning rush hour: predictability and commuter stress, Environment and Behaviour, 34, pp. 521–530.

10.

Fisher

P. A.

Malmberg

(2001) Settled people don’t move: on life course and (im)mobility in Sweden, International Journal of Population Geography, 7, pp. 357–371.

11.

Flood

Barbato

(2005) Off to work: commuting in Australia. Discussion Paper Number No. 78, The Australia Institute, University of Canberra.

12.

Friberg

Brusman

Nilsson

(2004) Persontransporternas ‘vita fläckar’: om arbetspendling med kollektivtrafik ur ett jämställdhetsperspektiv. Centrum för kommunstrategiska studier, Linköpings University.

13.

Fults

K. K.

(2010) A time perspective on gendered travel differences in Sweden. Licentiate thesis in infrastructure, Department of Transport and Economics, Royal Institute of Technology, Stockholm.

14.

Gatersleben

Uzzell

(2007) Affective appraisals of the daily commute: comparing perceptions of drivers, cyclists, walkers, and users of public transport, Environment and Behavior, 39(3), pp. 416–431.

15.

Gottholmseder

Nowotny

Pruckner

G. J.

Theurl

(2009) Stress perception and commuting, Health Economics, 18, pp. 559–576.

16.

Green

A. E.

Hogarth

Shackleton

R. E.

(1999) Longer distance commuting as a substitute for migration in Britain: a review of trends issues and implications, International Journal of Population Geography, 5, pp. 49–67.

17.

Hoem

J. M.

(1997) The impact of the first child on family stability. Research Reports in Demography No. 119, Stockholm University (www.suda.su.se/SRRD/srrd119.doc).

18.

Houghton

D. S.

(1993) Long-distance commuting: a new approach to mining in Australia, The Geographical Journal, 159(3), pp. 281–290.

19.

Jain

Lyons

(2008) The gift of travel time, Journal of Transport Geography, 16(2), pp. 81–89.

20.

Klis

M. van der

Karsten

(2009) The commuter family as a geographical adaptive strategy for the work–family balance, Community, Work & Family, 12(3), pp. 339–354.

21.

Klis

M. van der

Mulder

C. H.

(2008) Beyond the trailing spouse: the commuter partnership as an alternative to family migration, Journal of Housing and the Built Environment, 23(1), pp. 1–19.

22.

Kluger

A. N.

(1998) Commute variability and strain, Journal of Organizational Behaviour, 19(2), pp. 147–165.

23.

Koslowsky

Aizer

Krausz

(1996) Stressor and personal variables in the commuting experience, International Journal of Manpower, 17(3), pp. 4–14.

24.

Krieger

Fernandez

(2006) Too much or too little long-distance mobility in Europe? EU policies to promote and restrict mobility. Technical report, European Foundation for the Improvement of Living and Working Conditions (www.eurofound.eu.int/docs/areas/populationandsociety/mobility4paper2006.pdf).

25.

Lidström

(2006) Commuting and citizen participation in Swedish city-regions, Political Studies, 54, pp. 865–888.

26.

Lyons

Chatterjee

(2008) A human perspective on the daily commute: costs, benefits and trade-offs, Transport Reviews, 28(2), pp. 181–198.

27.

Lyons

Urry

(2005) Travel time use in the information age, Transportation Research Part A, 39(2/3) pp. 257–276.

28.

Mokhtarian

Salomon

(2001a) How derived is the demand of travel? Some conceptual and measurement considerations, Transportation Research Part A, 35, pp. 695–719.

29.

Mokhtarian

Salomon

(2001b) Understanding the demand for travel: it’s not purely ‘derived’, Innovation, 14(4), pp. 355–380.

30.

Mokhtarian

P. L.

Salomon

Redmond

L. S.

(2001) Understanding the demand for travel: it’s not purely ‘derived’, Innovation, 14(4), pp. 355–380.

31.

Muth

R. F.

(1969) Cities and Housing: The Spatial Pattern of Urban Residential Land Use. Chicago, IL: University of Chicago Press.

32.

Ommeren

J. van

(1996) Commuting and relocation of jobs and residences: a search perspective. PhD thesis, Virje Universiteit Amsterdam.

33.

Ommeren

J. N. van

Berg

G. J. van den

Gorter

(2000) Estimating the marginal willingness to pay for commuting, Journal of Regional Science, 40(3), pp. 541–563.

34.

Ory

Mokhtarian

(2005) When is getting there half the fun? Modeling the liking for travel, Transportation Research Part A, 39(2/3), pp. 97–123.

35.

Páez

Whalen

(2010) Enjoyment of commute: a comparison of different transportation modes, Transportation Research Part A, 44, pp. 537–549.

36.

Plaut

(2006) The intra-household choices regarding commuting and housing, Transportation Research Part A, 40, pp. 561–571.

37.

Pocock

(2003) The Work/Life Collision. Sydney: The Federation Press.

38.

Putnam

R. D.

(2000) Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster.

39.

Redmond

Mokhtarian

(2001) The positive utility of the commute: modeling ideal commuting time and relative desired commute amount, Transportation, 28, pp. 179–205.

40.

Reneland

(1998) Befolkningens avstånd till service: GIS-projektet Tillgänglighet i svenska städer 1980 och 1995. Rapport 1998:5, STACTH Stads- och trafikplanering Arkitektur Chalmers Tekniska Högskola, Göteborg.

41.

Renkow

Hower

(2000) Commuting, migration, and rural–urban population dynamics, Journal of Regional Science, 40 (2), pp. 261–287.

42.

Roberts

Hudgson

Dolan

(2009) It’s driving her mad: gender differences in the effects of commuting on psychological well-being. Working Paper No. 2009009, Department of Economics, The University of Sheffield, May.

43.

Rotter

J. C.

Barnett

D. E.

Fawcett

M. L.

(1998) On the road again: dual-career commuter relationships, The Family Journal, 6(1), pp. 46–48.

44.

Rüger

Ruppenthal

(2010) Advantages and disadvantages of job-related spatial mobility, in: Schneider

Collet

(Eds) Mobile Living Across Europe II: Causes and Consequences of Job-Related Spatial Mobility in Cross-national Perspective, pp. 69–93. Opladen: Barbara Budrich Publishers.

45.

Russo

Reggiani

Nijkamp

(2007) Spatial activity and labour market patterns: a connectivity analysis of commuting flows in Germany, The Annals of Regional Science, 41(4), pp. 789–811.

46.

Sandow

Westin

(2010) The persevering commuter: duration of long-distance commuting, Transport Research Part A, 44, pp. 433–445.

47.

Singer

J. D.

Willett

J. B.

(1993) It’s about time: using discrete-time survival analysis to study duration and the timing of events, Journal of Educational Statistics, 18(2), pp. 155–195.

48.

K. S.

Orazem

P. F.

Otto

D. M

(2001)The effects of housing prices, wages and commuting time on joint residential and job location choices, American Journal of Agricultural Economics, 83(4), pp. 1036–1048.

49.

Statistics Sweden (2003) Single and cohabiting adults in the population register, tax register and in reality. Background statistics to population and welfare statistics 2003:11. Official Statistics of Sweden, Statistics Sweden.

50.

Statistics Sweden (2009) Tables on the population in Sweden 2008. Official Statistics of Sweden, Statistics Sweden.

51.

Statistics Sweden (2011) Kan yrket förklara skilsmässan?, SCB tidskrift Välfärd, 2, pp. 11–14.

52.

Storey

(2001) Fly-in/fly-out and fly-over: mining and regional development in Western Australia, Australian Geographer, 32(2), pp. 133–148.

53.

Stutzer

Frey

B. S.

(2008) Commuting and life satisfaction in Germany, Scandinavian Journal of Economics, 110(2), pp. 339–366.

54.

Swedish Agency for Economic and Regional Growth (2010) Classification of Sweden’s regions (www.tillvaxtverket.se/huvudmeny/faktaochstatistik/omregionalutveckling/regionfamiljer.se).

55.

Swedish Institute for Transport and Communications Analysis (2007) RES 2005–2006: the national travel survey. SIKA statistics, No. 2007:19, Swedish Institute for Transport and Communications Analysis.

56.

Tolley

(1996) Green campuses: cutting the environmental costs of commuting, Journal of Transport Geography, 4(3), pp. 213–217.

57.

Travisi

C. M.

Camagni

Nijkamp

(2010) Impacts of urban sprawl and commuting: a modelling study of Italy, Journal of Transport Geography, 18(3), pp. 382–392.

58.

Vilhelmson

(2002) Rörlighet och förankring: geografiska aspekter på människors välfärd. Kulturgeografiska Institutionen Handelshögskolan, Göteborg University.

59.

Wachs

Taylor

B. D.

Levine

Ong

(1993) The changing commute: a case study of the jobs–housing relationship over time, Urban Studies, 30(10), pp. 1711–1730.

60.

Zax

J. S.

(1991) The substitution between moves and quits, The Economic Journal, 101(409), pp. 1510–1521.