Public transport provision and social sustainability in Sweden

ABSTRACT

The provision of public transport is important for a sustainable society and is vital for many groups. Children, older people, and people from carless households are examples of such groups: they rely on public transport and risk social exclusion without it. The aim of this study is to identify areas in Sweden where an improved public transport system might enhance social sustainability. To this end, an index of public transport needs at a sub-municipality level was calculated along with an index of public transport provision using data from public transport authorities in Sweden. The results indicate that areas with a mismatch between the provision of and estimated demand for public transport are relatively heterogenous. High gaps exist both in rural and peripheral regions of Sweden as well as the suburbs of urban regions.

KEYWORDS:

• Public transport provision
• social sustainability
• public transport gaps
• Sweden

Introduction

Over the last two decades, sustainability has become an important dimension of transport planning. Sustainability as a concept is commonly described as ‘economic and social development that meets the needs of the current generation without undermining the ability of future generations to meet their own needs' (WCED 1987). This description emphasizes the three foundations of sustainable development: economic development, social development, and environmental development.

The social dimension of sustainability has seen a notable increase in attention within transport planning research over recent years (Lucas 2012; Bao et al. 2023). Publications that address the social dimension of sustainability cover a wide range of topics and include the following: enquiries into the impact of transport choices on people’s health (Khreis and Nieuwenhuijsen 2019; Mizdrak et al. 2023; Rojas-Rueda et al. 2012; Stankov et al. 2020); transportation barriers for marginalized groups (Boisjoly and El-Geneidy 2016; Farber et al. 2018; Guzman et al. 2017; Lo et al. 2011; Smart 2015); and the implication of ride-hailing on mobility and accessibility (Young et al. 2020; Young and Farber 2019; Schaller 2021). An additional salient theme relating to the social dimension of transport planning pertains to investigations that centre on the concepts of transport equity, social justice, and social exclusion (Lucas 2012; Lucas et al. 2007; Pereira et al. 2017; Roorda et al. 2010; Woodcraft 2015).

Public transport plays an important role in the achievement of social inclusion (Kamruzzaman et al. 2016; Saif et al. 2018; Stanley and Lucas 2008). One of the principal functions of public transport is to facilitate accessibility to employment, education, health care, social activities, and everyday amenities for people with restricted mobility choices (Manaugh et al. 2010). The compensatory role of public transport when it comes to transport-related social disadvantages is why equitable distribution of public transport provision is a point of focus of public transport planners and policymakers worldwide (Camporeale et al. 2019).

The mapping of equity in public transport provision was carried out in several countries in Europe (Carroll et al. 2021; Fransen et al. 2015; Mattioli 2014; Mattioli 2017; Rau and Vega 2012), South America (Jaramillo et al. 2012; Hernandez 2018), Australia (Currie et al. 2010; Delbosc and Currie 2012), and Asia (Wang et al. 2022). These studies not only contributed with empirical evidence but also added to the methodological discourse, particularly with regard to the measurement of public transport provision. For example, Delbosc and Currie (2012) estimated public transport provision in Melbourne by combining walkability to transit nodes and weekly service frequency to express service levels. Jaramillo et al. (2012) examined public transport provision in Santiago de Cali, Colombia, by investigating the number of stops, vehicle capacity, and service frequency to calculate transport density per inhabitant. Fransen et al. (2015) introduced spatial-temporal accessibility metrics in Flanders, Belgium, where they looked at trips at specific times and days to various types of facilities to study accessibility in the public transport network. Hernandez (2018) used a cumulative opportunity indicator to gauge potential job and education accessibility in relation to public transport provision in Montevideo, Uruguay.

However, it is worth noting that these studies often aggregated metrics of public transport provision to administrative areas. For instance, Delbosc and Currie (2012) combined data on the proximity of public transport stops to residential areas with data on departure frequency to determine service levels per census district. Similarly, Fransen et al. (2015) introduced a spatial-temporal accessibility metric by using real-time transport data and facility locations, computed at the level of Traffic Analysis Zones (TAZ). While aggregating public transport provision metrics to administrative areas might be appropriate in evenly populated regions, it can lead to an underestimation or overestimation of actual public transport provision in sparsely populated areas that have an uneven population distribution.

The research presented in this paper departs from previous research on public transport equity since it proposes a method of identifying public transport gaps - that is to say, areas with a spatial mismatch between the provision and estimated needs of public transport - using grid level data on population distribution. It relies on general transport feed data from Swedish public authorities and the spatial distribution of segments of the population that have the greatest need for public transport to calculate public transport levels over a 24-hour period in walking distance from the populated grid. The targeting of areas seen to have public transport gaps (i.e. investment and policy incentives in relation to public transport) may facilitate improvements in social sustainability and equity in financial resources. For groups more dependent on public transport, the absence of such services might lead to a deterioration in access to employment and educational opportunities and in some areas to a higher need to own a car: that is to say, forced car ownership.

Recognized for its commitment to environmental sustainability, Sweden offers an intriguing and challenging context in which to investigate public transport gaps as it is one of only a few European countries that frequently aligns with the social-democratic welfare model (Esping-Anderssen 1996). In a social-democratic welfare state like Sweden, emphasis is on social equality and extensive state intervention to provide a comprehensive social safety net that typically includes universal access to health care, education, and social services, as well as generous unemployment benefits and pension schemes. Despite Sweden, with its social-democratic welfare state model, having undergone significant changes since the 1990s, including increased income inequality and the introduction of market mechanisms in the welfare sector, it maintains a high degree of social equality. Insights gained from studying public transport gaps in Sweden can significantly contribute to international discussions on effective public transport policy.

This paper has a six-section structure. Section 1, the introduction above, is followed by Section 2, which provides a detailed summary of the research context of public transport provision, equity and public transport gaps. Section 3 presents the data sources used in the empirical estimation, while Section 4 outlines the methodology that was adopted. Results from the empirical examination are then presented in Section 5. Section 6 provides a closing discussion.

Research context

The ability to be mobile has always been important for humans since it enables connections with other people and places. The need for travel, the ability to travel, and choices relating to travel influence and are influenced by culture and by how society functions (Stanley and Stanley 2017). Public transport has an indisputable role to play in the level of mobility of any society. Individual mobility patterns are shaped by many factors, such as ethnicity, gender, age, income, and the physical ability to travel (Macedo et al. 2022; Smart 2015). Public transport is primarily intended to facilitate access to employment, education, health care, social activities, and everyday amenities for people with restricted mobility (Manaugh et al. 2010). The literature highlights the following groups as being most dependent on public transport: those with a limited budget, the unemployed, young people without a driving licence or car, and older people with physical disabilities that prevent them from driving. Nowadays, research is extensive on the diverse aspect of public transport accessibility (see, for example, Saif et al. 2018 for an overview). How well a public transport system is designed to create accessibility to job opportunities (Deboosere and El-Geneidy 2018; Tao et al. 2020; Bastiaanssen et al. 2020), to service facilities (Mattson 2011; van Gaans and Dent 2018), and to social activities (Currie and Stanley 2008, Lättman et al. 2016; Berg and Ihlström 2019) is one of the themes in the research. Within this body of literature, the social sustainability dimension and equity in the distribution of public transport are two important themes. Public transport provision is crucial in terms of the compensatory nature of public transport when it comes to social sustainability.

Research makes clear two approaches to the study of equity in public transport provision. The first approach concerns the horizontal distribution of public transport and assesses the concentration of public transport provision in relation to the entire population. A common strategy to quantify horizontal equity is to construct Lorentz curves and calculate Gini coefficients, and in a similar manner as with wage inequality, express the inequality of public transport distribution in relation to population using a single indicator. For example, for the metropolitan area of Melbourne, Australia, Delbosc and Currie (2011) reported a Gini coefficient of 0.68, indicating rather high inequity in terms of access to public transport. Using the Lorentz curve, Delbosc and Currie (2011) concluded that 70% of the population in the Melbourne metropolitan area is served by only 19% of the provision of public transport. Using a similar approach, Wang et al. (2022) reported the Gini coefficient for the total population of Shanghai, China, to be 0.141, indicating that inhabitants have relatively equal access to public transport services. Horizontal equity assessment has been used in several other studies to identify segments of transport networks that are vital for the efficiency of a public transport system and for estimating public transport provision (Mishra, Welch, and Jha 2012; Mishra et al. 2015; Pavkova et al. 2016; Sharma et al. 2020).

While a horizontal equity approach provides insight when it comes to mapping how well a public transport system covers an entire population in a given geographical area, it does not provide insight into how it serves more marginalized groups or, therefore, how it strengthens the social sustainability of that area. From the perspective of social sustainability, the second approach – vertical equity – is better. Vertical equity focuses on how well a public transport system serves population groups that need it the most. Low-income households, older people, and young people are commonly considered to be marginalized when it comes to accessibility (Camporeale et al. 2019). To compare the accessibility of specific groups in the population to public transport provision, the Gini coefficient and Lorentz curve are used in a similar manner to how they are used in horizontal equity studies. For example, in their study of social sustainability in Shanghai, China, Wang et al. (2022) examined Gini coefficient values for three disadvantaged groups and demonstrated that a relatively high vertical equity existed since most transport system services were available to a substantial proportion of these three disadvantaged groups.

Closely related to the vertical equity approach is the concept of public transport gaps. Also, in literature about public transport gaps, the focus of the assessment is on how well a public transport provision provides for those population groups with the greatest needs. A common strategy in the public transport gap analysis is to use GIS software to map geographical areas with a mismatch between the provision of public transport and the concentration of individuals in marginalized groups. Nowadays, there is empirical evidence on public transport gaps for both several European countries (Fransen et al. 2015; Mattioli 2014; Mattioli 2017; Rau and Vega 2012) and several non-European countries (Wang et al. 2022; Hernandez 2018; Jaramillo et al. 2012; Delbosc and Currie 2012; Currie et al. 2010).

When the public transport system fails to supply transport services to the population, other means of transportation must be used. A poor public transport service might ‘force' people into car ownership. Deprived groups that reside in areas from which the distance is long to services and everyday amenities often have no choice but to own a car irrespective of the potential financial stress this may mean for them. This has been acknowledged in the literature under the term ‘forced car ownership' (Lucas 2012; Curl et al. 2018). Forced car ownership contributes to further marginalization of low-income households since a large proportion of their disposable income must go towards car ownership and maintenance. As Currie and Delbosc (2011) demonstrate, the additional financial burden related to owning a car may result in a scenario where low-income groups travel shorter distances and undertake fewer trips to minimize the costs associated with maintaining a car. Recent studies show that the level of forced car ownership is high in many rural and peripheral areas in several countries. For example, Curl et al. (2018) found in a longitudinal study from Scotland that forced car ownership increased after the financial crisis and that households were reluctant to give up their car despite financial pressures. Carroll et al. (2021) used GIS to identify areas in rural Ireland that had high levels of forced car ownership. They identified several clusters where social deprivation was high and access to public transport low. In Sweden, Amcoff (2017) found that those who resided in administrative areas located far from a grocery store had a high level of access to a car. The researchers involved in the studies on forced car ownership believe that the identification of areas where people are ‘forced' to own a car can lead to improvements in transport policy that can address social sustainability and reduce forced car ownership in disadvantaged rural and peripheral areas.

Although the role of public transport in fostering social inclusion is indisputable, it is important to consider that poor public transport provision does not automatically lead to social exclusion. In her (2012) conceptualization of transport-related social exclusion, Lucas emphasizes how it is the interaction between transport disadvantage and social disadvantage that, in combination with other factors, can lead to transport-related social exclusion: ‘transport disadvantage and social disadvantage interact directly and indirectly to cause transport poverty. This in turn leads to inaccessibility to essential goods and services, as well as ‘lock-out’ from planning and decision-making processes, which can result in social exclusion outcomes' (Lucas 2012, p. 106). In other words, some individuals may experience transport disadvantage but not social exclusion, while others may experience social exclusion despite having good access to transport.

Taking into account the interaction between transport disadvantage and social disadvantage, this study aims to identify areas in Sweden where the overlap between these two factors might contribute to the phenomenon of transport poverty. To this end, we map the mismatch between public transport provision and estimated public transport needs, with the goal of pinpointing potential public transport gaps.

Data

To identify areas where the expansion of public transport might enhance social sustainability, two data sources were used in the analysis: these were Statistics Sweden and Trafiklab, which works with Sweden’s regional public traffic authorities.

First, register data from Statistics Sweden was utilized. Swedish register data contains information from various statistical registers about every citizen of Sweden. This information is aggregated into grids. To ensure privacy and data protection, the grid-size used in urban areas is 250 metres x 250 metres and in sparsely populated areas 1 000 metres x 1 000 metres. In this study, population size, aggregated to grids, was used in the calculation of public transport provision. By breaking down the data on population size into smaller geographical units, it is possible to identify local variations in public transport provision, which would not be possible using administrative area level data.

Second, demographic data aggregated to DeSO (Demographic Statistics by Small Areas) level was used to calculate the estimated need for public transport provision. DeSO was developed by Statistics Sweden in 2018 to enable the analysis of socio-economic changes at the sub-municipal level. The DeSO divides Sweden into 5984 areas, with population size varying between 700 and 2 700 inhabitants per area. Border delineation considers natural boundaries such as watercourses and railways, and fundamental components in the delineation are borders of urban areas and electoral districts (Statistics Sweden 2023). Demographic data used to estimate the need for public transport provision comprised the proportion of inhabitants with a low economic standard, the proportion of inhabitants with less than 12 years of schooling, the proportion of inhabitants not in paid employment, and the proportion of inhabitants living in rented housing (see Table 1).

Table 1. Data used in the analysis and their aggregation level.

Variable	Aggregation level	Source
Population size	Grids	Statistics Sweden
Proportion of inhabitants with a low economic standard	DeSo areas	Statistics Sweden
Proportion of inhabitants (aged 25-64) with ≤ 12 years of schooling	DeSo areas	Statistics Sweden
Proportion of inhabitants (aged 16-64) not in paid employment	DeSo areas	Statistics Sweden
Proportion of inhabitants living in rented housing	DeSo areas	Statistics Sweden
Number of departures within 24 h	X and Y coordinates	Trafiklab

Third, general transport feed (GTFS) data originating from Trafiklab was utilized in the analysis. In terms of Sweden, GTFS data contains information about national timetables, routes, and lines for more than 50 000 stops in the public transport network. Internationally, GTFS data has been criticized for being inaccurate, particularly when it comes to rural areas, since some checks have shown that data does not cover all available services (Benevenuto and Caulfield 2020). In Sweden, meanwhile, it is the regional public transport authorities that provide GTFS data. Based on random checks, the data is highly accurate.

Methodology

To identify areas where investment in public transport has the potential to improve social conditions, what is required is a way to calculate both the spatial provision of public transport in an area and the spatial distribution of those groups in the population that are most dependent on public transport. We propose a methodology such as that used by Currie et al. (2010) and Fransen et al. (2015) to combine information on the spatial distribution of public transport provision, using an index of public transport provision (IPTP), with information on public transport needs, using an index of public transport needs (IPTN) so as to identify public transport gaps.

Public transport provision mapping

Consistent with Currie et al. (2010), we opted to measure public transport provision in terms of transport service levels in walking distance for each DeSo area over a 24-hour period. Transport service levels were calculated using the following equation: (1) $\mathrm{TS}{L}_{\mathrm{Deso}}={\displaystyle \frac{{\sum }_{i=1}^n\mathrm{Departure}{s}_{\mathrm{grid}i}}{n}}$(1) where TSL_Deso is the service level of the DeSo area, n is the number of populated grids in the DeSo area under analysis, and Departures_{grid i} is the number of departures within walking distance for the populated grid i in the DeSo area in focus. The calculation of the transport service levels was performed in three steps using QGIS software. In step one, a buffer zone with a 400-metre diameter was delineated for each populated grid. This distance represents the maximum distance most people are prepared to walk to reach a public transport stop (Currie et al. 2010; Delbosc and Currie 2011). In step two, the public transport stop with the highest frequency of departures within a 24-hour period within the 400-metre buffer zone was identified, and the number of departures was assigned to the populated grid, representing the level of public transport service. In step three, the average number of departures per DeSo area was calculated, aggregating the number of departures within the 400-metre buffer zone of each populated grid and using the formula outlined above. Once we had obtained the service levels for each DeSo area, we calculated the IPTP by averaging the number of departures to unity: that is to say, the greater values of the IPTP indicate higher public transport provision, while the lower values of the IPTP indicate lower public transport provision.

In our calculation of IPTP, we used the information on the spatial distribution of population (in the 250-metre grids and 1 000-metre grids). Therefore, even though the index is aggregated to the DeSo area level, we consider the spatial distribution of the population within the DeSo area. Figure 1 provides a simplified example of calculating transport service levels in a DeSO area using population size information from grid data. In this illustration, the population in the DeSO area is concentrated in three grids. Two of these grids have public transport stops within walking distance, while the third has none within walking distance. In this simplified scenario, calculating transport service levels based on population size at the grid level yields significantly lower service transport levels compared with using data on population size at the DeSO level.

Figure 1. Simplified example of calculation of transport level in DeSo areas.

This differs then from Currie et al. (2010), who assumed population is distributed evenly in the DeSo area. We believe that the study of public transport stops within walking distance that have the highest frequency of departures depicts the real situation; however, so as to ensure the robustness of the IPTP calculation, we chose also to calculate two other variants of IPTP. The second variant of the IPTP considered departures from all public transport stops within the 400-metre buffer zone from each populated square, while the third variant considered all departures from the nearest transport stop only.

For the final analysis of areas where public transport has the potential to improve social sustainability, we retained the first version of the IPTP, using the total number of departures from the most frequently served stop within the 400-metre buffer zone from each population square.

Social disadvantage mapping

The identification of areas whose social sustainability could be improved through public transport requires the mapping of the spatial distribution of the population that needs public transport the most. Previous research (Currie et al. 2010; Jaramillo et al. 2012; Fransen et al. 2015) combined several indicators to express the level of social deprivation in administrative areas. Carroll et al. (2021) used the Townsend material deprivation index that developed from the 1981 Census in the UK (Townsend 1987) to identify hot spots of public transport disadvantage in the UK. Since the development of this deprivation index, new deprivation indexes have been developed, based on census data, in other countries and regions: for example, the USA, Canada, and France (Messer et al. 2006; Pampalon et al. 2009; Salmond and Crampton 2012; Guillaume et al. 2016).

Although there is no official deprivation index for Sweden, the deprivation index proposed by Strömberg et al. (2023), which has four indicators, was used for this study. The first indicator gives the proportion of individuals with a low economic standard, measured as individuals with less than 60% of the national median disposable income; the second indicator gives the proportion of individuals aged 25–64 who have less than 12 years of schooling; and the third indicator gives the proportion of inhabitants aged 16–64 who are not in paid employment. The last indicator, meanwhile, gives the proportion of inhabitants living in rented housing.

To calculate the IPTN for each DeSo area, we adopted a statistical approach in the form of a factor analysis. Factor analysis is a commonly used method to reduce a high number of variables into a small number of variables through the identification of latent components. In this study, a principal component analysis (a variant of factor analysis) was used to extract the latent component(s). In line with previous research (for example, Strömberg et al. (2023); Fransen et al. (2015)), factors with eigenvalues >1 were retained for further analysis along with factor loadings > 0.30. The principal component analysis resulted in one latent variable, which explains more than 66% of the total variance of the data (see Table 2).

Table 2. Extraction of principal components from the individual variables.

Component	Initial eigenvalues		Extraction sums of squares loadings
	Total	Variance explained (%)	Total	Variance explained (%)	Cumulative variance (%)
1	2.675	66.872	2.675	66.872	66.872
2	0.974	24.358
3	0.297	7.437
4	0.053	1.333

The factor loadings (see Table 3) indicated that the latent factor has a high loading on the proportion of individuals with a low economic standard (0.969), the proportion of inhabitants not in paid employment (0.936), and the proportion of inhabitants living in rented housing (0.856). The factor loading of the proportion of inhabitants with ≤12 years schooling was somewhat lower (0.355); note, this variable was retained on theoretical grounds.

Table 3. Component matrix showing factor loadings for each variable.

Variable	Component 1
Proportion of inhabitants with a low economic standard	0.969
Proportion of inhabitants aged 25–64 with ≤ 12 years of schooling	0.355
Proportion of inhabitants aged 16–64 years not in paid employment	0.936
Proportion of inhabitants living in rented housing	0.856

Subsequently, a four-indicator IPTN was calculated as a single latent variable, applying a linear combination of the factor loadings as weights for the individual variables. The IPTN reached an acceptable level of KMO¹ (0.57, p < 0.0019) and Cronbach alpha 0.67. A DeSo area with a high value of IPTN is considered to be more deprived in comparison to a DeSo area with a low value of IPTN.

Public transport gaps

To identify areas where the extension of public transport can potentially improve social sustainability, public transport gaps were delineated. To this end, in line with Fransen et al. (2015) and Currie et al. (2010), information on public transport provision was combined with estimated public transport needs to identify DeSo areas with a low and a high provision of public transport in relation to estimated needs. The following equation was used for the calculation: $\mathrm{IPTG}=\mathrm{IPTN}\text{–}\mathrm{IPTP}$Due to the standardization of both the IPTP and IPTN to unity, the resulting index of public transport gaps (IPTG) is in a range of – 1 and 1, which makes it possible to compare DeSo areas when it comes to the mismatch between public transport provision and estimated public transport needs. High values of IPTG indicate an estimated high need for public transport compared with provision, and low values of IPTG indicate a high provision of public transport compared with estimated needs.

Based on the calculated value of the IPTG, DeSo areas with low and high values of IPTG were identified. DeSo areas belonging to the highest quintile in terms of the IPTG values were identified as high gaps. These DeSo areas deserve attention when it comes to the distribution of public transport, since estimated public transport needs exceed the current provision of public transport. DeSo areas belonging to the lowest quintile of the IPTG values were labelled low gaps. These DeSo areas deserve further attention since the provision of public transport exceeds estimated public transport needs.

Results

Public transport provision and social disadvantage

Figure 2 shows the spatial distribution of the four indicators used for the construction of the IPTN – the proportion of inhabitants with a low economic standard (Figure 2(a)), with less than 12 years of schooling (Figure 2(b)), not in paid employment (Figure 2(c)), and living in a rented apartment/house (Figure 2(d)) for DeSo areas in Sweden. Dark areas on the maps indicate more socio-economic vulnerable DeSo areas compared to the light areas. A general pattern that is observable in all the maps is that the darker areas show sparsely populated areas in northern Sweden. It can also be noted that there are dark-coloured DeSo areas in southern Sweden as well as on the outskirts of the metropolitan areas of Stockholm, Gothenburg, and Malmö.

Figure 2. The spatial distribution of the indicators utilized in calculating the IPTN.

The four single indicators, outlined above, were combined using factor analysis to create the IPTN. Figure 3 shows the spatial distribution of the IPTN for Sweden.

Figure 3. The spatial distribution of IPTN in the DeSo areas in Sweden.

The general pattern in this map resembles the pattern shown in Figure 2. Social deprivation, modelled by the IPTN, is highest in the northern part of Sweden, barring the coastal areas bordering the Baltic Sea, while lower values of the IPTN are found in the metropolitan areas and central parts of the country. Considering that DeSO areas in metropolitan regions are significantly smaller geographically, three enlarged maps are presented for Stockholm, Gothenburg, and Malmö. These maps show fringe areas with an apparent concentration of very high or high and very low and low scores.

Using the calculated IPTN value, DeSo areas were classified into quintiles (see Table 4). In the first quintile, the DeSo areas with the lowest values of IPTN were placed, while the DeSo areas with the highest values of IPTN were placed in the fifth quintile. For each quintile, a median value of each of the four indicators (proportion of inhabitants with low economic standard, less than 12 years of schooling, not in paid employment, living in rented housing) was calculated. What was noticeable was that when the value of the IPTN increased, so did the median values of each of the indicators. In Q1 (that is to say, the quintile with the lowest estimated need for public transport), 4.6% of the population had a low economic standard, while in Q5 (that is to say, the quintile with the highest estimated need for public transport), the corresponding number was 25.4%. Similarly, the values of the other three indicators also vary, forming the IPTN.

Table 4. Characteristics of the IPTN.

DeSo areas classified into quintiles of the IPTN	Proportion of inhabitants with a low economic standard (median)	Proportion of inhabitants aged 25–64 with <= 12 years of schooling (median)	Proportion of inhabitants aged 16–64 not in paid employment (median)	Proportion of inhabitants living in a rented housing (median)
Q1	0.046	0.438	0.134	0.050
Q2	0.084	0.539	0.156	0.111
Q3	0.120	0.559	0.186	0.212
Q4	0.165	0.580	0.229	0.379
Q5	0.254	0.611	0.308	0.599

To estimate the provision of public transport for the DeSo areas, IPTP was used. Figure 4 shows the spatial distribution of the IPTP based on the average number of departures from the busiest transport stops within walking distance of each populated grid in each DeSo area. Darker areas on the map indicate DeSo areas with good public transport accessibility. DeSo areas with the worst accessibility are indicated by light colours. The greatest provision of public transport can be found in the central parts of the three metropolitan regions of Stockholm, Gothenburg and Malmö, as well as in urban areas located in central and southern Sweden. With the exception of the populated areas on the coasts, both northern Sweden and some central southern areas of the country are at the greatest disadvantage when it comes to public transport. It can also be observed that public transport provision decreases sharply outside the suburban regions of the three metropolitan areas.

Figure 4. The spatial distribution of IPTP in the DeSo areas in Sweden based on average number of departures from busiest transport stop within a 400-metre walking distance of each populated grid in each DeSo area.

To test the robustness of the calculation, we use Figure 5 to depict the results of IPTP based on departures from all transport stops (within a 400-metre walking distance) and Figure 6 to depict the results of the IPTP based on departures from the nearest transport stops (within a 400-metre walking distance). Essentially, the general pattern of the spatial distribution is very similar to what Figure 4 depicts: the highest provision of public transport is to be found in the metropolitan regions and in urban settlements in southern and central Sweden, while northern Sweden is characterized by substantially low public transport provision. Furthermore Figure 6 indicates that in southern and central Sweden, there are more public transport stops in walking distance. What this shows is that southern Sweden has a greater provision when departures from the nearest bus stop are considered.

Figure 5. The spatial distribution of IPTP in the DeSo areas in Sweden based on average number of departures from all transport stops within a 400-metre walking distance of each populated grid in each DeSo area.

Figure 6. The spatial distribution of IPTP in the DeSo areas in Sweden based on average number of departures from the nearest transport stop within a 400-metre walking distance of each populated grid in each DeSo area.

Table 5 shows the summary statistics based on the three ways to estimate the index of public transport provision. The index value is marginally lower for the method when only the stops with the highest frequency of departures within a 400-metre buffer zone is calculated.

Table 5. Summary statistics public transport provision.

	Index of public transport provision
	Departures from nearest public transport stop within a 400 metres buffer zone	Departures from public transport stop with highest frequency within a 400 metres buffer zone	Departures from all public transport stops within a 400 metres buffer zone
Min,	0	0	0
Q1	0.01	0	0
Median	0.04	0.02	0.03
Mean	0.06	0.04	0.06
Q3	0.08	0.05	0.08
Max	1	1	0.99

Public transport gaps

By combining the index of social deprivation and the public transport provision index, we can identify areas where there is a mismatch between the provision of and need for public transport. Figure 7 depicts the spatial distribution of areas with high gaps – that is to say, DeSo areas belonging to the first quintile in terms of IPTG, where the need for public transport greatly exceeds the provision of public transport, and low gaps – that is to say, DeSo areas belonging to the lowest quintile of IPTG, where public transport provision greatly exceeds the need for public transport.

Figure 7. The spatial distribution of highest quintile of the IPTG (high gaps) and lowest quintile of the IPTG (low gaps) in DeSo areas in Sweden.

Figure 7 shows that high gaps are found primarily in central places within rural and peripheral areas, both in northern Sweden and in central and southern Sweden. Interestingly, there is a disproportion between the provision of and need for public transport in the suburbs of the metropolitan regions of Stockholm, Gothenburg, and Malmö. DeSo areas designated as low gaps are located primarily in central Sweden, as well as in and around the metropolitan areas of Stockholm, Gothenburg, and Malmö. Large urban centres in central Sweden also show low gaps in relation to public transport provision.

Table 6 depicts descriptive statistics of the IPTP, IPTN, and IPTG for DeSo areas belonging to low gaps and high gaps. In the lowest quintile of the IPTG, low gaps, the IPTP is high compared to the value in the highest quintile of the IPTG, high gaps, indicating a higher provision of public transport in DeSo areas designated as low gaps. In a similar way, the IPTN is substantially lower in the DeSo areas belonging to the low gaps, indicating less estimated demand for public transport in these. Finally, it is noticeable that the IPTG is substantially lower in the DeSo areas designated as high gaps in comparison to DeSo areas that are considered low gaps.

Table 6. Descriptive statistics IPTP, IPTN and IPTG for low gaps and high gaps areas.

	IPTP	IPTN	IPTG
Low gaps
Mean	0.08	0.12	0.02
Min	0	0	−0.87
Max	1	0.83	0.08
St.Dev	0.12	0.08	0.08
High gaps
Mean	0.04	0.48	0.44
Min	0	0.30	0.30
Max	0.28	1	0.92
St.Dev	0.03	0.13	0.12

In addition, Table 7 displays descriptive statistics of variables used to calculate the IPTN. It can be observed that DeSo areas designated as low gaps have on average a lower proportion of inhabitants with a low economic standard compared to DeSo areas designated as high gaps. A similar pattern is also evident with respect to the other three variables that build the IPTN index – the proportion of inhabitants with less than 12 years of schooling, not in paid employment, and living in rented housing. The biggest difference between the low gaps areas and the high gaps areas is in the variables proportion of inhabitants with a low economic standard and the proportion of the population living in rented housing.

Table 7. Summary statistics for variables used to calculate the IPTN, for highest quintile of the IPTG (high gaps) and lowest quintile of the IPTG (low gaps).

	Proportion of inhabitants with a low economic standard (mean)	Proportion of inhabitants aged 25–64 years with <= 12 years of schooling (mean)	Proportion of inhabitants aged 16–64 years not in paid employment (mean)	Proportion of inhabitants living in a rent apartment (mean)
Low gaps	0.062	0.366	0.154	0.146
High gaps	0.306	0.642	0.338	0.632

Summary and conclusion

Public transport plays an indisputable role in facilitating access to essential services and opportunities, particularly for marginalized groups in the population. This study aims to identify public transport gaps – areas with a mismatch between the provision and estimated needs of public transport – in the context of Sweden.

To account for the spatial distribution of population with the greatest need for public transport provision, an IPTN was used. In the index, weighting was on the proportion of inhabitants with a low economic standard, less than 12 years of schooling, not in paid employment, and living in rented housing. To measure the provision of public transport, the IPTP was calculated using data from regional transport authorities and demographic grid data on the spatial distribution of the population. The utilization of grid-level data made it possible to check for uneven population distribution, which allowed for a more accurate estimation of public transport provision. Mapping the difference between the IPTN and IPTP reveals DeSo areas where the provision of public transport disproportionally exceeds the needs, labelled as low gaps. In these areas it may be the case that resources allocated to public transport are used in an inefficient manner. Meanwhile, in areas where there is a high need for public transport but where there is low provision, labelled high gaps, social sustainability can be enhanced by improving the public transport provision.

As illustrated in Figure 7, DeSO areas with the potential to improve social sustainability through public transport show notable heterogeneity, particularly in terms of their geographical locations. Interestingly, high gaps are evident in both central locations of rural and peripheral regions of Sweden, particularly in the northern part of the country, as well as in the suburban areas of urban centres. However, despite the fact that both categories of DeSo areas labelled as having high gaps experience a situation where provision of public transport fails to meet the estimated demand, the impact on social sustainability varies. For a population that has a high demand for public transport and that resides in suburban parts of the urban regions, a low provision of public transport may not contribute significantly to further social deprivation since most everyday amenities in Swedish suburbs are usually located within a short geographical distance from their place of residence. However, poor public transport provision in DeSo areas located in more peripheral parts of the country might lead to fewer possibilities for individuals who are more dependent on public transport. Children and adolescents especially might be dependent on parents when it comes to access to everyday amenities because of long distances (Berg and Ihlström 2019). Another potentially vulnerable group is much older people, as typically there are fewer other people in their proximity who can provide them with transportation. As shown in studies on public transport gaps in the UK (Carroll et al. 2021) and Australia (Delbosc and Currie 2012), residents in more peripheral parts of a country are to a greater extent forced to own a car because of long commuting distances and inadequate public transport. Although owning a car might initially seem to improve mobility, as shown in Currie et al. (2010), addressing the inadequacy of public transport provision through car ownership may inadvertently result in a scenario where individuals, constrained by limited resources, are compelled to restrict their social interaction due to budgetary constraints.

Although different methodologies were employed, the results of this study align closely with findings presented in other studies. Fransen et al. (2015) noted that areas with provisions exceeding anticipated needs are in suburban areas of Flanders, as well as in rural regions. Currie et al. (2010) reported significant disparities in the demand for public transport between outer and inner areas of Melbourne, Australia. Carroll et al. (2021) identified that areas at the highest risk of forced car ownership, indicating a substantial gap between public transport provision and estimated population needs, are situated in the most rural regions of Ireland. The empirical results of this study indicate that similar areas can be found in Sweden, despite its long history of extensive social welfare public policy. The reason for this might be that investments in public transport in the Swedish planning system are primarily a means to reduce traffic emissions since car traffic is replaced with public means of transportation: their principal aim is not to create social sustainability. The focus has therefore been on maximizing the horizontal equity (Mishra et al. 2012, 2015; Pavkova et al. 2016; Sharma et al. 2020) rather than on establishing vertical equity (Camporeale et al. 2019, Wang et al. 2022) in the public transport system. This study shows that public transport is unevenly distributed between different socio-economic groups in Sweden and that this is an important factor to take into consideration when planning public transport in order to avoid negative consequences in terms of social sustainability.

While this study offers insights into public transport gaps in Sweden, it is crucial to acknowledge inherent limitations. Firstly, methodology should incorporate time-variability of public transport needs. The estimation of public transport provision in our analysis is based on the number of trips within walking distance during a 24-hour period. As Kamruzzaman and Hine (2011) demonstrate, public transport availability is essential to specific segments of the population whose needs vary depending on the time of day and day of the week. On weekdays, it is essential that the unemployed can attend training and career counselling and that those in employment can make it to work, while at the weekend and in the evening, it is important that adolescents can rely on public transport for social activities (Berg and Ihlström 2019). The time-variability of public transport provision thus represents a potential avenue for future investigation. It might also be worthwhile extending the analysis to encompass accessibility through the public transport network to various destinations.

Secondly, in this present study, the estimation of public transport provision assumes a 400-metre walking distance. However, it is essential to recognize that the definition of walking distance varies considerably between demographic groups. For some, it may be substantially shorter, while for others, it may be longer. With this variability in mind, a potential avenue for future research could involve differentiating the analysis of public transport provision tailored for specific groups in the population.

Finally, the results of our research revealed heterogeneity of areas that are designated as high gaps – that is to say, areas with insufficient provision of public transport in relation to needs. It was found that high gaps areas exist in rural and peripheral areas of Sweden, as well as in suburban parts of urban centres. In light of this diversity, a more nuanced exploration could involve a detailed classification of areas based on the discrepancy between public transport provision and estimated demand. This classification could serve as a crucial foundation for tailoring effective solutions. The context, whether urban or peripheral, will dictate the appropriate remedy. For instance, in urban settings, addressing the provision and demand gap may involve improving services on existing routes. However, the challenge is more pronounced in areas with low population density, where extending the current public transport infrastructure becomes economically impractical due to a lack of existing facilities. Consequently, innovative and alternative solutions are imperative for these gaps to be bridged (Berg and Ihlström 2019).

In conclusion, the identification of public transport gaps highlights specific areas that need to be improved so as to enhance social sustainability. Addressing these identified gaps will not only improve transportation accessibility but also contribute to creating a more equitable and socially sustainable society, which aligns with Sweden’s enduring commitment to welfare and equality.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The Kaiser–Meyer–Olkin (KMO) test is a statistical measure to determine how suited data is for factor analysis.