Introduction

Many cities wish to create more lively urban areas by planning for mixed use of new buildings (Gehl 2013). Mixed use involves different uses that occupy discrete parts of a building, block, or area (Rowley 1996, p. 87). Potential effects of urban planning regulations to allow and promote mixed use have been studied (Grant and Perrott 2011; Hirt 2012), but gaps in our knowledge remain as to what extent detailed requirements for non-residential ground floor premises have led to actual uses and to an increased diversity of uses. Furthermore, few researchers have studied infills of apartment buildings in low-density suburbs where apartment buildings already dominate the urban fabric. Encouraging non-residential ground floor uses in sparsely populated suburbs appears as a particular challenge, more often studied in cold climate zones (DeLisle and Grissom 2013; Xu 2017; Freemark 2018), which is relevant in northern Europe.

The empirical setting for this study is infill housing projects in suburbs in Gothenburg, Sweden’s second largest city with a population of 600,000. Here, local politicians have pursued a policy for mixed use in new residential projects, evident at least since 2000 in the comprehensive plans for the whole city area (Gothenburg 1999). Emphasis on mixed-use development was reinforced in the city development strategy document (Gothenburg 2014). The intentions were then that there should be a greater diversity, development of a mixed city with more life and movement, increased security and more meeting places. Such principles were mirrored in planning regulations for housing projects within a 2015 programme for producing 7000 new dwellings, almost only as infill projects with apartment buildings in suburbs dating from the 1960s and 1970s. It is the projects within this programme, BoStad2021, that are analysed in this investigation. The main purpose of this programme was to rapidly increase the housing supply in Gothenburg. The choice of infill as a policy was not primarily dictated by a general policy of densification (Charmes and Keil 2015); instead, densifying should be understood as a consequence of city policies for diversity combined with an avoidance of greenfield development.

Varieties of mixed-use regulations are vertical (Montgomery 1998) and horizontal (Ryckewaert et al. 2021) mixed-use zoning as well as hybrid vertical/horizontal zoning. Within the programme, zoning regulations were primarily for vertical mixed use, but there are examples of hybrid vertical/horizontal zoning in the plans. Parking, offices, schools and day-care facilities are among the uses that planners often situated separately.

Against this background, the research purpose is to study factors explaining gaps between plans and outcomes for non-residential uses of ground floors in a Swedish housing programme, with 30 development plans for apartment building infill in low-density suburbs already dominated by apartment buildings. The research question is how zoning regulations are mirrored in actual non-residential uses of ground floors. Our hypothesis is that the presence of ground floor activities is influenced by planning regulations intended to promote mixed use, urban design factors, mobility factors and socio-economic factors. As the focus is on infill projects, a further potential factor is the pre-existing supply of services and goods in the immediate surroundings of the project, an agglomeration effect that may influence the use of new non-residential ground floor premises.

This paper is organized so that after the review of literature, we present the method and then the Gothenburg case study, the housing programme in its local context, before analysing ground floor uses in plans and in reality. Three project examples of mixed use are analysed more in detail. The resulting pattern is compared to earlier studies, and conclusions are drawn.

Literature review

Setting the scene for an investigation whether vertical zoning for mixed use in suburban multi-storey apartment buildings is effective needs contributions from several types of earlier studies. The review includes arguments for mixed use, zoning and other mechanisms supporting mixed use, demand for non-residential uses of ground floors, and the role of local traffic systems including public transport.

Arguments for mixed use

More than 60 years of publications inspired by how Jane Jacobs (1961) defended mixed-use cities constitute a vast body of literature, often bringing up socio-economic effects. Researchers have identified a long list of arguments for mixed use. It is said to improve the ambience and attractiveness of areas (Coupland 1997), and it “creates an urban environment active at all hours, making optimum use of infrastructure” (Grant 2002, p. 72). The quality of an active frontage can significantly affect people’s perceptions of a public space “in terms of its safety, comfort, sociability and liveliness” (Heffernan et al. 2014, p. 92). Similarly, Kickert (2016, p. 55) finds that an open relation between buildings and public space at the ground floor level is crucial for making urban environments walkable by providing destinations, safety (“by expressing openness and hospitality”) and experiential quality. Furthermore, mixed-use development reduces the need for travel (Coupland 1997; Duany et al. 2000, p. 187; Grant 2002), builds the bonds of community (Duany et al. 2000, p. 187), generates social diversity (Grant 2007), lowers crime (Coupland 1997), contributes to sustainability (Coupland 1997), as well as stimulating economic growth and tax revenues (Duany et al. 2000, p. 187). Several studies indicate that higher household incomes in suburban neighbourhoods might be related to better local service supply, which affects the outlook for mixing of uses (Meltzer and Capperis 2017; Kang 2019). It is possible to speak of commercial gentrification (Meltzer 2023), which may occur to some extent with mixed use in urban infill projects that are associated with higher housing costs and household incomes, increasing the local supply of goods and services.

Supporting mixed use

Given the advantages of mixed use, planners and local politicians may wish to promote it relying on zoning regulations (Hirt 2007, 2012; Conrad 2010), while there are other policies that can be combined with zoning. Economic incentives (Rowley 1996; Rabianski et al. 2009) such as subsidies and taxes can be provided in order to support mixed use. Subsidies provided by developers can be needed for corner stores, as these “would not be economically self-sufficient before a neighbourhood matures” (Duany et al. 2000, p. 187). The dynamics of emerging service demand within a new development has seldom been studied, although Grant and Perrott (2011, p. 186) noted that “the retail uses in Cornell’s town centre struggled because they went in before the population arrived to support them”. Retail can be considered as a follower land use, with a time lag dependent on the influx of new residents.

Patterns of non-residential uses in mixed-use projects

For a more detailed understanding of what supports vital ground floors, it is necessary to identify patterns and trends of non-residential uses. Looking at the influence of urban density on social sustainability in the UK, Dempsey et al. (2012, p. 94) listed services as “supermarket, food shop, corner shop, post office, doctor’s surgery, chemist, bank and building society, restaurant/takeaway/café, recreational facilities, library and pub among others”. For comparative purposes, studies carried out in cold climates are relevant, since the average Gothenburg temperature in January is 3 °C. DeLisle and Grissom (2013) investigated Seattle, another cold city, relying on a detailed NAICS 2002 (North American Industry Classification System) classification for business activities and found a dominance of non-retail and non-restaurant tenants in mixed-use buildings. In the Toronto region, Xu (2017) studied urban fringes, making site visits to identify types of businesses in mixed-use units. For 59 buildings belonging to affordable housing in Chicago, Freemark (2018) found 19 publicly accessible commercial uses, led by Office and Hair/nail salon. The retail and services composition of four New Urbanist cores in North America (incl. Portland and Tacoma) was mapped by Novak (2019), again classifying according to NAICS. In eight European urban areas, Wandl and Hausleitner (2021) specified the number of different functions in each unit studied, applying the Statistical Classification of Economic Activities in the European Community (NACE). There is one recent study of retail vacancies in Turku, a Nordic region city with an average January temperature of – 3.3 °C, listing commercial spaces by 14 business categories and studying vacancy patterns (Marjanen et al. 2023), but this concerns an inner-city area; there have been no similar studies of suburban vacancies in the region. At least in cold regions, suburban non-residential uses of premises in suburbs are dominated by services to the human body: retail sales of food and beverages, hairdressing and other beauty treatments, restaurants, fitness and medical services are recurring and typical examples of what has been found. That these services remain or have expanded is partly a consequence of a larger shift among consumers towards the experience economy (Sundbo 2021).

It is often conceded that one reason for the difficulties to fill ground floors with non-residential activities is the decline of retail shops. There are many studies of the underlying mechanisms explaining the reduced demand for retail services (Kickert and Talen 2023), with dire consequences for both inner cities (Talen 2022) and suburbs, sometimes even seen as an existential crisis of shopping streets (Carmona 2022; 2023). Retail concentration has been going on for many decades, in Sweden supported by zoning (Nyström 1994; Kärrholm and Nylund 2011; Söderholm and Wihlborg 2015) until recently, while e-commerce is a later phenomenon. Defining a methodology for structurally measuring the interactivity of ground floor frontages over time, Kickert (2016) focused on two case study urban cores of Detroit and The Hague. Although focus was on decline in urban cores, the mechanisms identified should have a bearing on suburbs: modernization and consolidation of the economy, technological changes in transportation, “a culture of amnesia and the growing divide between social classes” (Kickert 2016, p. 68) explain frontage deactivation. A few years later, e-commerce, the experience economy, the Maker Economy, and regulatory, cultural and demographic shifts were invoked by Kickert (2021) for their impact on American urban storefronts. Talen and Park (2022) interviewed Chicago business leaders who indicated demographic change, urban context, and property owner behaviour as the main reasons for retail vacancy. They also point out the complex relationship between demographic changes and demand for retail. Relevant also in a Swedish context, it is the comments they received on parking and infrastructural issues, decline in retail and changing demographics. Street-level storefront vacancies throughout Xining, China, have been analysed by Li and Long (2024), distinguishing between 12 categories of store types. They have operated with several explanatory variables: distances to urban centre, to nearest bus stop and to nearest commercial zoning; road and population densities as well as neighbourhood housing prices, which can be seen as a proxy for income levels. There remains, however, a gap in the literature because suburban conditions and trends are seldom in focus.

Problems arising from vacant premises

Among main impacts of retail vacancy, respondents in the Chicago study by Talen and Park (2022) underlined adverse effects on neighbourhood cohesiveness: economic disempowerment, less foot traffic and “deteriorating blocks with no appeal” (p. 1420). Furthermore, it was thought that vacancies increase crime. The technical flexibility needed for vertical mixed use may imply higher construction costs for higher ceilings in ground floors, compared to exclusively residential buildings (Martani et al. 2018). Retail dismantling might spread through contagion in a neighbourhood (Kickert 2016), but reactivation of empty spaces can also have positive effects by spreading (Tamini 2018, p. 69). Obviously, property owners will suffer from lack of rental income from shops that are empty. If vacancies are believed to be an initial phenomenon for a new development, temporary uses of vacant premises can be considered (Patti and Polyak 2015; Madanipour 2018), otherwise, conversion to residential use should be an option, depending on zoning regulations in force.

Local traffic systems, parking and public transport

A variety of non-residential uses can be supported by both pedestrian and vehicular access. Looking at the challenge of making retail uses viable in mixed-use suburban developments in Canada, Grant and Perrott (2011) suggested that mixed use has a better chance of success in places with more traffic. A recognized challenge is to design traffic systems that minimize risks to children (Rothman et al. 2014; Gemmell et al. 2023) and other pedestrians while allowing for service access using public transport and private cars. Insights into how suburban areas can accommodate safely various types of traffic have been emerging over the years: the revival of streets as opposed to malls (Southworth 2005), and commercially active ground floors may form a special type of strip malls (Linovski 2012) with limited surface parking along the sidewalk. US initiatives for achieving complete streets (Anderson et al. 2015; Hui et al. 2018) belong here, with examples of how four-lane streets have been transformed to two-lane, how on-street parking and diagonal parking have been introduced, along with (sometimes protected) bicycle lanes, bus shelters so as to encourage multimodal travel: “the majority of roads with Complete Streets features had fewer collisions and fewer injuries after their retrofits than before” (Anderson et al. 2015, p. 7). As could be expected from the principles of movement economies (Hillier 1996), we find that for more than one example, higher retail sales were noted along redesigned corridors.

Traffic systems in Swedish housing areas developed around 1970 still reflect the SCAFT road safety principles imposed as norms by central government, including “separating different classes of traffic” and “differentiation within each road network” (SCAFT Group 1968; Hagson 2005, p. 3). Road space had to be organized into a hierarchical network, where external feeder roads provide mobility for cars, through-traffic is minimized, and priority on the local network is given to pedestrians and bicycles. Building entrances do not face roads, and there are often vast parking spaces between roads and buildings, serving as additional protection.

There are no simple relations between car ownership and sale of services and goods in suburban neighbourhoods, considering the contexts of how local traffic systems have been designed (with may either support or discourage car use), parking opportunities (which may encourage consumer reliance on cars) and public transport (which may reduce reliance on cars) (Tana et al. 2016; Jiang et al. 2017; Pengjun and Shengxiao 2018).

A conceptual framework

Relying on prior research, we assume that the presence of active non-residential uses of ground floors depends on a long range of factors besides planning regulations intended to promote mixed use. There are urban design factors, mobility factors and socio-economic factors. Examples of urban design factors are the scale of an infill project and the traffic system within an area, including separation of vehicular and pedestrian traffic as well as parking arrangements. Mobility factors include access to public transport and car ownership, whereas socio-economic factors refer to age groups, immigrants, household incomes and households being renters or owners. For an infill programme, another potential factor is the service supply in the immediate surroundings of projects, whether adjacent commercial spaces are filled with complementary or competitive services.

Method

Given the research question of how zoning regulations are mirrored in actual non-residential uses of ground floors, the sources that we have relied on are document analyses and field observations. Official municipal documents from Gothenburg (planning guidelines, comprehensive city plans and detailed development plans) have been analysed. Furthermore, a municipal staff member has given access to informal minutes from programme coordination meetings between politicians, officials and developers held in 2015, when the programme was in a formative stage.

During site visits to all projects between November 2022 and January 2023, “economic activities” in premises were recorded and classified according to NACE Rev. 2, based on the UN ISIC Rev. 4 classification. The percentage of ground floor frontage length taken up by non-residential functions in actual use or vacant was estimated from site observations and checked against plans and building permits when ambiguous. For non-residential functions, ground floor frontage has been parsed for windows and transparent entrance doors, and the width of these segments has been measured as well as the total length of ground floor frontage; the percentage has been estimated for premises actively used for non-residential purposes, and separately estimated for vacant premises. All ground floors with immediate pedestrian access and when relevant, facing bus stops or customer parking, have been included. Buildings within the new detailed development plan for the infill project are included, although at least part of a few buildings may have existed initially. As residential functions are also counted ground floor service spaces that are exclusively used by residents, such as laundries, waste handling, bicycle storage and repair, and district heating. For primary schools and child day-care facilities with ground floor windows raised above eye level, window width is not included. Moreover, the number of premises active with non-residential functions (outside the project area) visible from entrances to the new buildings within the project area were recorded in order to analyse whether new activities in infill buildings would compete or broaden the mix of activities within the immediate surroundings. The limited number of projects (30) and the high number of potential explanatory variables do not support a regression analysis with frontage length taken up by non-residential activities or vacant as dependent variables, which would have been desirable. Instead, a correlation analysis has been performed.

It is possible that our data exaggerate the future proportion of empty premises, as there may be an adjustment delay before they are used, partly also because of COVID-19 effects. However, the prevalent housing shortage in Gothenburg meant that completed buildings were filled quickly by residents, which should have led to a rapidly developing increased demand for services and goods in the area.

For each project, the detailed development plan which assigns functions to buildings has been analysed, and this plan is also the source for an assessment of to what extent the local traffic system and parking arrangements conform to the SCAFT principles of separating different classes of traffic and differentiating within road networks. This assessment has been made by the three authors independently, relying on a continuous scale, where 0.0 is no SCAFT separation while 1.0 corresponds to full adherence to SCAFT. Statistical data from 2017 for average annual income (in Swedish crowns, SEK) in the “primary areas” where the infill projects were to be carried out have been provided by the city statistical office. Gothenburg is divided into almost 100 primary areas with an average of about 5000 inhabitants.

The selection of three among the 30 BoStad2021 projects for an analysis that brings in a wider range of explanatory factors was because these projects had the highest number (per infill dwelling) of active non-residential uses of their ground floors. Here, parking arrangements constituted an added urban design factor, while car ownership and access to public transport were additional mobility factors; age groups, presence of immigrants as well as households renting or owning their dwellings were further socio-economic factors that were introduced.

Context of the programme

Taking the BoStad2021 programme as our case, or more precisely its 30 projects as a multiple-case study, there is a sprawl context, a housing initiative from developers that was transformed by city politicians, and successively a policy emphasis on encouraging non-residential use of ground floors. During the 1960s and 1970s, Gothenburg developed a dispersed pattern of suburbs with clustered apartment buildings, not unlike the suburban periphery of Seattle (Moudon and Hess 2000). When planning almost only for multi-storey apartment buildings, the low average density that resulted in Gothenburg is partly to be explained by the difficult geology of the region (Lundqvist 2016), partly by city policies for generous greenspaces and considerable spaces for parking.

In 2014, two Gothenburg housing developers and one architectural firm presented the city with a proposal for accelerating the residential planning process (Bröchner et al. 2021). The timeline for the process set in motion by this initiative is represented in Fig. 1. The political response was to launch a programme, later called BoStad2021, and the target was to have 7000 dwellings finished by 2021. These dwellings were intended to be an addition to the hitherto achieved about 2000 dwellings annually. The city was to select sites, almost only in existing suburbs to avoid greenfield development, and 31 infill sites were identified, one of which had to be left out for financial reasons. This corresponded to no less than 26 developers, of which four companies were owned by the city itself through its Framtiden Group. From the outset, the programme had a stated political emphasis on meeting two deadlines: one for the approval of the detailed development plans (end of 2017) and one (end of 2021) for the completion of all projects. Of the development plans, 24 met the original deadline. Almost all the dwellings produced are in apartment buildings. The first construction start was in 2017 with the Minister of Housing holding the spade, which underlined the national and local political importance of the programme. The first project (a tower block housing cooperative with 60 dwellings aimed at students) was finished in mid-2019. The plans for the infill projects included between 55 and 830 dwelling units (see Fig. 2). Only two projects included semi-detached dwellings, forming a total of 111 within the programme.

Fig. 1
figure 1

Source: Author’s compilation

The timeline of the BoStad 2021 programme

Full size image
Fig. 2
figure 2

Source: Author’s compilation

The 30 BoStad 2021 infill projects

Full size image

The detailed development plans within the programme often included regulations concerning non-residential uses, in accordance with the current Swedish Planning and Building Act (Ch. 4), which allows municipalities to determine the extent of the built environment above and below the ground surface, as well as the use of the buildings. The National Board of Housing, Building and Planning has prescribed which functions can be mandated or accommodated in plans, and relevant functions are Dwellings, Centre, Care, Retail trade, Offices, Visitor facilities, and School. Vertical zoning is allowed and possible. Before the 1931 City Planning Act which reflected modernist principles, there were basically no restrictions for mixed-use development in Sweden (Kalbro and Røstnes 2012), but successive legislative changes have empowered municipalities to perform detailed top-down planning.

The comprehensive plans for Gothenburg

Mixed-use ambitions have been expressed in successive comprehensive plans for Gothenburg since 1999 (Gothenburg 1999). In a 2008 policy document for urban planning qualities, the municipal Building Committee paid attention to the needs of small business, suggesting as one solution that ground floors of residential buildings could be planned for premises (Stadsbyggnadskontoret Göteborg 2008, p. 16). Furthermore, the importance of a mixture of functions was argued, and it was stated that ground floors can be designed for activities that populate the place or the street (p. 34); the starting point should therefore be to develop the neighbourhoods so that non-disturbing activities and housing can be integrated, based on the local conditions. With less precision, the 2009 comprehensive plan for Gothenburg stated that a dense structure offers conditions for a greater mix of functions through a better basis for local commerce and other activities. There was no analysis of demand in the comprehensive plan, which anyhow is not binding for subsequent detailed plans, unlike in the Norwegian planning system (Mäntysalo et al. 2015).

Findings

The findings reported here begin with an overview of municipal policy change, followed by non-residential uses present in the 30 projects and their surroundings, compared to the zoning regulations in the corresponding detailed development plans. Three projects with a comparatively high number of active non-residential premises have been selected and are analysed based on a wider range of factors that might influence mixed use within the programme.

Municipal policy change

We have observed city planners serving as mediators between politicians and developers in the case of BoStad2021, relying on the informal notes taken by a municipal staff member in 2015. It is clearly the politicians who have promoted the idea of lively ground floors even in areas with low demand for non-residential uses. The BoStad2021 programme coordination group held a meeting late in 2015 discussing active ground floors with nine developer representatives and seven city officials present. Officials reported that the municipal Building Committee wished to see detailed development plans with vertical zoning requiring ground floor activities. More than once during the year, the committee had questioned plans that did not require active ground floors; plans lacking provisions for active ground floors should now include a clear explanation as part of the official documentation from the City Planning Department. Three representatives of major developers reacted; one thought that it was necessary to check the commercial viability for an active ground floor in each particular project; another underlined that existing local retail analyses should be respected, and a third developer wished to engage directly with the Building Committee.

At a later coordination meeting the chairman of the Building Committee, an Environmental Party politician, stated initially that it was “incredibly important when transforming the city, with its major routes, into new urban environments that there must be active ground floors”. If there should be a developer loss of revenue, “we need to talk more about how to find a balance in that”. He could imagine that the city would be involved and fill the ground floor at first; during the first 10–15 years, “they can be difficult to fill, but then there is usually a shortage”, he added.

It was only in June 2017 that the Building Committee decided on city guidelines for planning for retail, as a partial update of the 2009 comprehensive plan, stating that commercial spaces in ground floors were "of great value". Nevertheless, misgivings were formulated in the guidelines: “if retail is too thinly spread, there is an imminent risk of ground floor vacancies, making projects more costly and causing more costly dwellings”. It was also noted that it is always an initial challenge for retail to be viable on ground floors of apartment buildings; there are essentially other locations with a greater commercial potential. Guardedly, the guidelines acknowledged that “active ground floors is a clear goal in the urban planning and a current issue to continue working on”.

When gaps between visionary planning and outcomes in Gothenburg occur, it has been explained in recent case studies by financial consequences of implementation having been brought in at a late stage (Brorström and Styhre 2021; Saldert 2021). Molnar (2022) chronicles a Gothenburg dialogue between the city and developers of a central city area, where the issue of municipal rent subsidies for ground floor facilities and a scheme for having facilities with higher rents to subsidize the rents of other facilities were left unresolved, and what remained were only specific ground floor regulations in the detailed development plan.

Zoning in plans

Typical examples of how the BoStad2021 detailed development plans formulated mixed-use regulations are:

  • “Centre activities shall correspond to at least x sq.m., only on the ground floor”

  • “Centre activities only on the ground floor, shall be along X-street to a depth of at least x metres”

  • “Centre activities are allowed, only on the ground floor”

  • “Centre activities allowed on a maximum of two storeys from the entrance level”

  • “Centre activities only at the entrance level”

The question is which types of zoning regulations that could be said to have been more efficient in supporting mixed use (see Table 1). Where buildings had been zoned for a mandated mix of residential and ground floor non-residential uses, there was a higher frequency of currently active non-residential establishments, but the clear majority of project plans just allowed mixed use and had a relatively lower number of active establishments. The disappointing outcome of the few project plans with buildings zoned for only non-residential uses should be interpreted as a low level of demand for non-residential uses, and plans including both buildings with mandated mix and buildings where mix was allowed were often found to have developers who were reluctant to prioritize the construction of buildings with a mandated mix.

Table 1 Comparison of types of plan regulations with current building uses
Full size table

Actual non-residential uses

The 30 plans represent a variety of infill configurations. Most projects comprised several buildings, and it was typical that plans required or made possible non-residential active ground floors for at least one of the buildings. Basic data for the 30 projects are displayed in Table 2. In general, site visits indicated that there were few actively used non-residential premises and many vacancies, but despite at least parts of a number of projects having been delayed, and possible COVID-19 effects, several actual uses of ground floor premises have been recorded (see Table 3). In a few cases, bicycle rooms, as it appeared not much used by dwellers, were found in ground floor premises. In two cases, food retailers had moved from adjacent separate one-storey buildings into ground floors within a project.

Table 2 Basic data for the 30 projects
Full size table
Table 3 Comparison of non-residential uses, within project areas and visible from projects
Full size table

“Surrounding services” in Table 2 means that active premises in the surroundings were visible from at least one building entrance within the project area. A comparison between types of ground floor activities found within the infill projects and those visible from the projects shows no great difference (see Table 3), except that there is a slightly higher frequency of restaurants, usually of the kebab/pizzeria variety, in project surroundings. Altogether, added non-residential functions in the infill development did not change the established pattern in the surroundings, and this confirms how limited the range of functions is likely to remain despite the inflow of new residents with possibly higher incomes. Only one unmanned convenience store could be identified, on the edge bordering on a project area.

Analysing non-residential ground floor frontages

We have used Pearson’s correlation analysis to study the relations between project size, frontage uses, types of zoning, surrounding services, income, and two design variables: SCAFT traffic system and distance to bus or tram stop. Table 4 shows that correlations between variables are generally low. Active non-residential uses of ground floor frontage, as a percentage, is significantly and negatively correlated with the walking distance to the nearest bus or tram stop. In other words, the shorter the walk, the more non-residential activity. Vacant frontage for non-residential functions, again as a percentage, is significantly correlated with average income in the area. This is surprising, just as that there is an equally strong negative correlation between income and the presence of a SCAFT type local traffic system. Vacant premises in higher income areas could be explained in several ways. One possibility is that urban planners and developers have overestimated an income effect, implying that few non-residential functions have been allocated to infill projects in low-income areas. Another explanation could be that the rent level also for premises in higher income areas is higher. Still another explanation could be that low-income areas were previously undersupplied with non-residential ground floors. There is a less surprising significant correlation between the scale of the infill, measured as number of dwellings, and the number of buildings zoned so that a non-residential ground floor is allowed. Summing up, what the correlations indicate is that public transport and area income do have an influence on how ground floors are actually used. Zoning regulations appear as less influential. Apart from a basic lack of demand and the COVID-19 pandemic, effects of e-commerce, and general slowness of households modifying their buying habits may have contributed.

Table 4 Pearson correlations
Full size table

Three projects with many active premises

Given the limitations of the set of variables studied in the correlation analysis, our three case studies draw upon a broader range of factors that may exert an influence on the presence of active ground floors without vacancies in the sense of unoccupied non-residential ground floor premises. Figures 3 and 4 show the three project examples, selected because they include a high number of non-residential ground floor activities, measured in relation to the size of the projects (number of dwellings). They also represent various patterns of zoning control.

Fig. 3
figure 3

Source: Municipal detailed plans, site observations

Three infill projects with active ground floors

Full size image
Fig. 4
figure 4

Source: Author photographs

The three infill projects

Full size image

The first project (Radiotorget, Fig. 3a) is a transformed and updated small community centre from the 1950s, now with 14 non-residential premises in use. Throughout the new detailed development plan, a mix of residential and non-residential uses (“centre activities”) was allowed. Ownership is now partly municipal, partly private. Average income in the area is slightly higher than for the city as a whole, while there are comparatively fewer inhabitants in the 0–30 age group; the proportions of rental and cooperative/owner-occupation is roughly the same as average for the city. The proportion of immigrants is lower here than the city average. As the area was planned originally before the SCAFT principles of separating traffic had been imposed, access to shops and other services for both pedestrians and drivers is easy. There is good public transport, and car ownership is 18% higher than the city average per 1000 inhabitants. The parking arrangements are diverse and scattered in the project area. They include different forms of both surface parking lots (on-street and off-street/in-lot) and structured parking facilities (deck car park and underground car parks integrated with residential buildings). There were no ground floor vacancies.

The second project (Fyrklöversgatan, Fig. 3b) modified an original SCAFT morphology by redeveloping a large oblong parking space, creating only three shops, all on the ground floor at the eastern end where there is student housing. The detailed development plan required at least 300 sq.m. of non-residential ground floor premises. There is a single private sector owner. Just as for the other two projects, average income in the area is slightly higher than for the city as a whole; demographically, there is a concentration on the 25–40 age group; owner-occupation is clearly less frequent than the average for the city. The proportion of immigrants is clearly higher here than the city average. Bus stops and short-term parking possibilities of the ground floor strip mall in this project appear to be important for raising local demand for services and goods. Again, the shops are easily accessed for both pedestrians and drivers; a bus stop faces the three shops. In contrast to the other two projects, car ownership in the area is 24% lower than the city average. Parking is concentrated in both ends of the project area and include surface off-street/in-lot parking and structured parking facilities. The structured parking facilities include multistorey and underground car parks, all integrated with residential buildings. There were no ground floor vacancies.

The third project (Gunnestorpsvägen, Fig. 3c) is a short ground floor strip mall 300 m from a 1970s shopping centre. There are 11 non-residential premises in use. The detailed development plan allowed a mix of residential and non-residential uses while prescribing that only ground floors could be used for “centre activities”. Here, there are several private sector owners. The average income in the area is close to that of the city as a whole, while there are comparatively fewer inhabitants in the 20–30 age group and more in the 75 + group; comparatively more dwellers are found in housing cooperatives or in owner-occupation than in rented apartments. The proportion of immigrants is as the city average. Car ownership is higher than average for the city. The new parking facilities and the strip mall now created make access to non-residential facilities convenient, and there is good public transport (a bus stop there). Here, the parking arrangements are diverse and scattered in the project area. They include different forms of both surface parking lots (on-street and off-street/in-lot) and structured parking facilities (deck car park and underground car park integrated with a street). There were two ground floor vacancies.

Adding to a broader understanding of influential factors, the three selected projects have in common that average income in their primary areas is at least as high as the Gothenburg average. This appears as a paradox, given that we have seen (Table 4) an association of vacant premises with higher incomes. The three projects have good access to public transport, and they do not reflect the SCAFT principles of traffic separation. Parking arrangements also appear to be influential for securing the viability of commercial uses of their ground floor premises. On the other hand, there was little agreement between these three projects when considering property ownership, age distribution of inhabitants, proportions of rental and cooperative/ownership dwellings, proportion of immigrants, and car ownership.

Discussion and conclusion

Our aim has been to study how zoning regulations are mirrored in actual non-residential uses of ground floors and which factors in general relate to non-residential frontage uses. The hypothesis has been that the presence of ground floor activities is influenced by planning regulations intended to promote mixed use, urban design factors, mobility factors and socio-economic factors. First, however, the analysis of city policy documents showed a successive orientation driven by politicians towards stricter and more comprehensive mandating of non-residential frontages, a transformation that took force at an early stage of the BoStad2021 programme, and especially in 2017, although mixed-use ambitions had been formulated during more than a decade in policy documents.

The comparison of plans and outcomes for non-residential uses of ground-floors in this housing programme with apartment building infill in low-density suburbs, already dominated by apartment buildings, has shown that several of the attempts to activate ground floors, and to introduce new uses to the existing urban areas through vertical zoning, have been less than successful. Judging by our correlation analysis of data from the 30 projects, we have found that a longer walking time to the local bus or tram stop is associated with a lower percentage of ground floor frontages for non-residential purposes. Higher household incomes in an area are associated with higher vacancy percentages of ground floor frontages. This income effect is unexpected. The link between incomes and vacancies, which anyhow are few, could be that demand in richer areas has been overestimated by planners and politicians. As in earlier comparable studies of peripheral suburbs in cold regions, retail shops except those selling food are a rarity, and otherwise the typical commercial use of ground floors, when allowed, is for services to the human body and hard or impossible to deliver digitally: hairdressing, beauty, nails, fitness facilities, and places serving food, as well as institutions for day care and schools. Grant and Perrott (2011) asked “Where is the café?”, and cafés are very seldom found in the Gothenburg project areas. Our three case studies reveal that further analysis using a wider range of variables than those entered into our correlation analysis is desirable. It is less obvious what larger samples would contribute to our understanding unless more variables catching, e.g., demography and parking conditions, are measured.

Attempts to introduce categories of retail trade other than for food are unlikely to succeed. Politicians and planners should accept that this is what can be achieved, unless in special circumstances. As acknowledged in the city’s own guidelines for retail in planning, the overriding political goal of active ground floors is hardly realistic if applied in low density areas in the periphery. Access to public transport and limited short-term parking possibilities of the ground floor strip malls do appear to be important for raising suburban demand for services and goods, but this may not be sufficient.

The comparison made here between ground floor uses in the Gothenburg infill buildings and what is already there in the neighbourhood does not support the belief that it is just a matter of years before a wider range of services will fill the empty premises. Nevertheless, it is possible that more flexible zoning regulations simply aiming at protecting dwellers from noise, smells and other nuisances could encourage more inventive types of mixed use, including pop-up shops and activities. Higher construction costs associated with extra ceiling height for ground floors may raise the issue of municipal subsidies, but the Gothenburg experiences reveal that visionary ideas for financial support may fail to materialize if planning and finance is not coordinated already at an early stage (Molnar 2022). Top-down regulation of building uses should be applied sparingly, as it is not self-evident that local politicians make better demand forecasts than experienced developers are able to do.