People or place? Towards a system of holistic locational values for creative workers

ABSTRACT

Previous studies of the locational patterns of European creative workers typically neglect their lower mobility, focusing on a simplistic “attraction and retention” logic. They largely fail to consider the heterogeneity of places and people, disregarding the basic human functions that connect people to their environment. Our objective is to propose a system of locational values that adds depth to the dynamics between creative people and places. The model developed encompasses 37 indicators, creating a large dataset (N = 4,823) via a postal questionnaire of creative workers in Slovenia. A set of statistical methods was conducted to analyse the differences in locational values, with the urban-rural dichotomy employed as the key test. The results show that the type of place, rather than demographic factors, determines distinct urban and rural sets of locational values. This illustrates the heterogeneity of interactions between creative workers and places that should be considered by future research and policymaking.

KEYWORDS:

• creative economy
• cultural ecologies
• living environment
• urban-rural
• comparative statistical model

Introduction

Questions regarding why people live and work where they do have driven key debates in economic geography for a number of years – debates that are presently being revisited in the wake of the global pandemic. A particular focus has been on creative or knowledge economy workers due to their high levels of human capital and their central role in utilising other forms of capital to generate prosperity, in combination with perceived greater mobility than for other members of the labour force. This has broad implications for economic development policy, inequality, well-being and social justice. In this paper we seek to reconnect this field of enquiry – which has developed in a somewhat insular and piecemeal manner over the previous two decades or so – with a more fundamental consideration of the basic functions of human existence. To this end the concept of Social Geography as articulated by the German school of social geography is introduced, central to which is the idea that people’s interaction with their place of residence is defined in accordance to how successfully they can implement and satisfy their basic human functions in their everyday lives, commonly interpreted as work, residence, education, services, leisure, and transport. In this way we argue that the long term and bi-directional interaction between people and places can be better understood. In doing so our aim is to propose a system of locational values that adds more depth to the dynamics between creative people and places. This is important because the pervasive “attract and retain” logic of previous studies represents an essentially reductionist perspective and one that is unsuited to contexts of lower worker mobility, particularly those found in Europe. Thus, we aim to better capture the multifaceted interaction between creative people and places.

The following section traces the evolution in the literature of factors considered in creative workers’ location decisions, from hard neoclassical factors (i.e. people do not matter, territorial assets do) through Florida’s (2002) creative class thesis (people matter, but from a limited economy-centric logic) to post-Florida research (e.g. Musterd & Gritsai, 2013) on additional contextual factors and some personal trajectory dimensions. It concludes by highlighting the present “state of the art” as still limited by a fundamentally attract and retain logic, and thus an incomplete account of the full range of interactions between people and places. In response to these, the idea of locational values which aim to overcome the previous reductionist perspectives by reflecting deeper shared meanings in this set of interactions is then developed. The social geography concept (Elkins, 1986; Ruppert, 1984) is applied in order to provide a more holistic approach that focuses on “everyday” aspects of human behaviour (residence, work, education, services, leisure and transport) by which creative people shape their environments and vice versa. This in turn is linked to more recent thinking on cultural and creative ecosystems (de Bernard et al., 2022; Gross & Wilson, 2020), which emphasises typically overlooked interactions not motivated by, or directly instrumental to, economic value. This approach enables the development of a comprehensive set of 37 indicators of locational values; in order to operationalise these, we employ a unique dataset of 4,823 survey responses from creative workers in Slovenia. We argue that an improved understanding of these interactions is important as it can ultimately contribute to more appropriate and locally nuanced policymaking.

Employing a set of binomial logistic regression models as the key test of this holistic set of locational values, the empirical analysis investigates whether there are significantly different locational value systems across urban vs rural localities (i.e. we employ an urban-rural typology here as the key manifestation of place-specific differences). We show that there are, with demographic factors playing a lesser role, i.e. we find place matters more. In doing so, two contributions are highlighted relating to the theory and empirics of creative people and places; firstly, our indicators are more structured and originate from a comparative reference to original theory explaining the interaction between specific types of social groups and specific places (social geography), and secondly, we consider the heterogeneity of places – previous studies rarely encompassed different types of places (e.g. metropolitan areas or rural areas) simultaneously, and none to our knowledge did so via a bespoke primary dataset. In doing so we address the weakness of the fundamentally reductionist approaches of previous research and underpin the understanding of people and place interaction on more holistic foundations.

Rationale and theoretical framework

Evolution of research in the location of creative workers

The following section traces the evolution of the extant literature around the location of creative workers. In doing so it highlights how at the beginning studies just focused on the issue of attracting creative talent and how they have included only later the retention element, albeit essentially only as an extension of the prior attraction logic, and largely using the same variables to proxy. We argue that for the majority of people, retention is something more fundamental regarding their connection to place; in other words, this people-place interaction is more than just the absence of mobility, and that this is particularly relevant given the greater rootedness on average in the European context.

Creative workers are a particular focus due to an increasingly knowledge-based economy within which their presence is assumed as a key driver of innovation and productivity (Musterd, 2004). More generally, it is accepted that innovative capacity plays a key role in long-term economic prosperity (Moretti, 2012); it is also widely acknowledged that the outcomes of knowledge-based activity are becoming spatially more polarised (Glaeser, 2011), which in turn has recently been associated with the rise of populist politics and a crisis in one-size-fits-all models of development (Rodríguez-Pose, 2020).

In examining the location of knowledge workers, neoclassical explanations emphasised the “hard” location factors such as job opportunities and wages, and city size more generally as a proxy for “thick” labour markets, the presence of universities (as generators of human capital), good transport links and so on (see Musterd & Gritsai, 2013). Conversely, in his creative class thesis, Florida (2002) proposed a reversal of this traditional “people follow jobs” logic, with mobile creatives hypothesised to locate in high amenity metro areas, to which economic outcomes (innovation, growth, start-ups) then accrued. These amenities (or “soft” factors) included openness to varied lifestyles (ethnic groups, lifestyles, sexual orientation), and the presence of a wide range of cultural and leisure facilities. This model for urban success was catchily dubbed the “3Ts” – technology, talent, and tolerance, which quickly gained popularity with urban policy makers (Clifton, 2008; Martin-Brelot et al., 2010), while attracting extensive critique from the academic community regarding the fuzziness of concepts (Pratt, 2008) and assumptions on causality when employing a cross-sectional analysis using secondary data to infer revealed locational preferences (Storper & Scott, 2008).

Florida’s original research was undertaken in North America – a liberal market economy (Hall & Soskice, 2001) with high levels of flexibility and mobility. Whether or not one accepts the critiques noted above, it is reasonable to expect the creative class thesis to operate differently within the coordinated market economies (to a greater or lesser extent albeit with the significant exception of the UK) of Europe. An extensive programme of research investigated these issues in Denmark, England and Wales, Finland, Germany, the Netherlands, Norway and Sweden (see Andersen & Lorenzen, 2005; Boschma & Fritsch, 2009; Clifton, 2008; Clifton et al., 2013), with some additional indicators included to reflect this context – public provision and social cohesion indices (the latter albeit problematically proxied by unemployment). This research also differed from Florida’s by analysing data from all localities rather than selected metro areas only such as New York, San Francisco, Boston, San Jose, Silicon Valley, etc. in the North American context; the fundamental approach was nevertheless the same, i.e. revealed locational preferences from secondary cross-sectional data. Although results largely confirmed qualified support for the role of soft location factors, “hard” factors (human capital, regional job opportunities) were also found to be important. Findings also highlighted the effect of coordinated market economies in flattening the distribution of the creative class (Clifton et al., 2013), the confounding role of particular local socio-political contexts (Clifton & Macaulay, 2015), and revealed less applicability of the thesis beyond larger metro regions (Andersen & Lorenzen, 2005). It also noted previously neglected demographic or stage-of-life factors (ibid).

Moreover, as noted by Storper and Scott (2008) if workers are not particularly mobile, as is broadly the case in the European context (Martin-Brelot et al., 2010), it is hard to accurately infer revealed preferences of locational choice, because there is little spatial sorting. A revealed preference approach also has the problem that amenities providing different levels of utility (or indeed dis-utility) will be bundled together leading to false conclusions from cross-sectional secondary data; as noted by Storper and Manville (2006, p. 1252) “one person’s amenity is often the next person’s inconvenience”. There are a few studies that employ net migration data (Niedomysl & Hansen, 2010; Vossen et al., 2019) but these still rely on hypothesised preferences from secondary data. Relatively rare studies employing qualitative methods have confirmed the minor role played by soft factors in migration decisions, while stressing more esoteric biographical narratives (e.g. Alfken, 2015).

Europe is of course itself not homogenous with regard to institutional and cultural contexts (Favell, 2008)¹; this observation in combination with the need to address the limitations highlighted above was the motivation for further comparative research collectively known as ACRE.² The key contribution of this work was complementing prior analysis of hard and soft factors with the role played by personal trajectories in understanding the locational choices and indeed rootedness of creatives. These trajectories are essentially the attachment of a person to a geographical location resulting from their previous life course, and are suggestive of a deeper evolutionary perspective (Frenken & Boschma, 2007) rather than a “shopping list” of amenities and economic factors. The findings of Musterd and Gritsai (2013) suggest the need for more place-specific policymaking, and thus highlight the value of considering people-place interactions. However, a significant limitation of these studies is that they draw their conclusions from a small sample of metropolitan areas only, with no direct comparison of different types of places at the same scale. Subsequent studies have sought to augment the ACRE findings by applying larger samples (Escalona-Orcao et al., 2018), investigating small towns (Escalona-Orcao et al., 2017) or micro scales (Comunian et al., 2010; Montanari et al., 2018), but these are also by their nature partial accounts only. Specifically, rural settings have been largely neglected in the literature, barring a few notable exceptions (McGranahan & Wojan, 2007; White, 2010), with McGranahan et al. (2011) showing that rural creative class growth is particularly strong in locations with attractive outdoor amenities, a result confirmed by Argent et al. (2013).

As outlined above, there is extant research of urban and rural localities – but only using secondary revealed preference approaches, and omitting some key (people-based) factors. No existing study has undertaken a comparative analysis across both urban and rural localities using a comprehensive large-scale primary dataset in order to better understand the people-place interactions of creative workers. This is an important gap in the literature, which this paper aims to address. This gap matters because ultimately sub-optimal policy choices are likely to arise from incorrect or at least incomplete accounts of the interactions between people and places, with potentially significant and long-term consequences for both prosperity and well-being.

As noted by Woods (2009), prevailing city-region approaches to development carry the risk of addressing rural localities solely in terms of their relation to the urban; there is indeed existing evidence that such urban-centric policy models are less relevant (and thus less effective) for creatives in other places (Huggins & Clifton, 2011), which in turn raises the fundamental question as to whether these workers are themselves somehow “different” in such places (indeed Huggins and Thompson (2021) have recently shown that behavioural and cultural factors can shape long term development trajectories at the local level). These are issues that require further exploration. To do so suggests the value of broader primary datasets which can investigate the holistic, long term and bi-directional relationships between creative workers and their place of residence. In doing so we aim to address the weakness of the fundamentally reductionist approaches taken in previous research and to underpin the examination of creative worker location with due consideration of the “everyday” aspects of human behaviour (residence, work, education, services, leisure and transport) by which creative people shape their environments and vice versa.

Building locational values systems from a basic human functions approach

To reflect these requirements, we propose a set of locational values; the term “value” implies that something is preferred over something else and shared by a group of people. In ethical theory, values influence how people make decisions, not just their immediate wants and desires, but they reflect on deeper concerns about what is important (Dietz et al., 2005).

We do so by utilising indicators previously used in the “3Ts” and ACRE research (which were developed incrementally, addressing specific weaknesses as they became apparent) but augment these by mapping against a base level of “basic functions of human existence” as articulated by the German school of social geography (Elkins, 1986; Ruppert, 1984). The key idea here is to move beyond the lens of “place attraction/retention”, with people modelled solely as rational agents pursuing instrumental ends (Cardoso et al., 2022). In contrast, social geography offers a holistic examination of people/place interaction through all basic human functions for specific social groups.

The German school of social geography, which emerged after the Second World War, placed particular emphasis on the so-called “basic functions of human existence”. These are commonly listed as work, residence, education, services, leisure, and transport; other functions, such as reproduction and social relations can be added (Elkins, 1986). Put simply, people choose where to live in accordance to how successfully they can implement and satisfy their basic human functions in their everyday lives. Social geography defines people with the same or similar behaviour or the same characteristics as social groups, without the need for any direct connection between these persons. According to Weichhart (2008), social geography is thus a typology of behaviours, with these social groups as carriers of spatial processes (Ruppert, 1984). Such a conceptualisation is consistent with data collection at the level of the individual actor. Some authors have recently used the term “social geographies” and “geography of social geography” to reflect the evolution of thinking in the field (Hopkins, 2011; Smith et al., 2010). However, the different perceptions are essentially variations on the same two basic questions, which refer to the spatial constitution of society on the one hand, and to the spatial expression of social processes on the other (Werlen, 2008). Social geography therefore foregrounds the question: “How do people choose where to live (Knox & Pinch, 2010)?” To answer this question, it focuses on everyday life and the ways in which social groups interact with each other and with the spaces in which they live.

Thus, our structured approach goes beyond the traditional attraction and retention logic in the research of creative workers’ geographies. Most often, such research relied empirically only on the listing of amenities, without considering the heterogeneity of places and creative workers and ignoring human needs that bind them to their environment. Our way of thinking is closer to the framework of cultural ecology, in which culture and creativity are seen as interwoven in situated communities and places (Gross & Wilson, 2020). This approach implies that spatial values can be seen as a form of “cultural opportunity” (Gross & Wilson, 2020). We argue that places can offer cultural opportunities that are distinct; these are the mobilised cultural resources that include the economic profile of the area, professionals, demographic features, housing stock, individuals, landscape & geographic features, public services, the geographic shape & size of the ecology and transport systems. The interest is in the diversity of these resources and the two-way dynamics between people and places – how creative people shape their environments and vice versa, and how certain environments can enable or constrain creative people. However, ecological thinking is still in the early stages of its theoretical elaboration and remains fuzzy and empirically untested (de Bernard et al., 2022). Social geography suggests a framework to operationalise the cultural ecological approach. From an ontological perspective, it focuses on human functions and provides a set of holistic indicators that describe different facets of human activities. From an epistemological perspective, it starts from the idea that certain social groups (such as creative workers) shape their environment by “performing” common human functions, rather than being determined by it.

Data and methodology

To answer our research question, we designed a holistic model of 37 indicators encompassing locational values of creative workers. We began our model with a limited set of indicators applied by the “3Ts” and ACRE research but expanded it to a more holistic set of “basic human functions”, which are along the lines of the German school of social geography commonly listed as residence, work, services, education, leisure, and transport (Elkins, 1986; Ruppert, 1984). As indicated in Table 1, previous research to a large degree neglected some of the basic human functions such as living costs and services and paid only partial attention to some of the important locational values such as recreational activities or safety issues. Next to the quantitative nature of locational values, qualitative aspects were also underrepresented (e.g. quality of services, quality of environment). Moreover, some modern indicators such as internet accessibility were also previously omitted.

Table 1. Descriptive statistics of locational values of creative workers.

Basic human functions		Indicators of locational values	Mean	SD
Residence	Living costs	Living costs	.04	.199
		Housing costs	.08	.264
	Housing	Housing availability^2	.05	.216
		Housing affordability^2	.07	.255
		Housing equipment	.07	.257
		Housing quality^2	.09	.283
	Personal networks	Born here^2	.10	.294
		Family lives here^2	.24	.425
		Friends live here^2	.08	.270
		Studied here^2	.00	.061
		Local community	.02	.139
		Housing inheritance	.11	.310
	Physical environment	Private open space	.20	.401
		Quality of environment	.24	.426
		Diversity of built environment^2	.01	.088
		Weather/climate^2	.04	.193
		Proximity to nature^2	.40	.489
	Safety	Personal safety	.16	.364
		Safe for children^2	.23	.422
	Tolerance and openness	Openness to different people^1,2	.02	.125
Work		Job^1,2	.21	.405
		Higher wage^2	.02	.128
		Working conditions^2	.03	.179
Education		Secondary education centre	.01	.098
		Higher education centre	.01	.093
		Educated population	.00	.038
		Good schools/faculties^2	.02	.123
Services		Publicly owned services^1	.09	.291
		Privately owned services	.05	.217
		Higher quality of services	.01	.117
Leisure		Diversity of leisure^2	.02	.140
		Recreational activities	.07	.258
		Cultural activities^1,2	.03	.164
Transport		Regional centres accessibility^2	.03	.161
		Road accessibility^2	.06	.243
		Public transport accessibility^2	.03	.166
		Internet accessibility	.03	.162

Note: N = 4823, Min = 0, Max = 1.

Source: Own elaboration; structure derived from social geography concept (Elkins, 1986; Ruppert, 1984). Indicators marked with (¹) originate from the “3Ts’ research (Andersen & Lorenzen, 2005; Boschma & Fritsch, 2009; Clifton, 2008; Clifton et al., 2013) and indicators marked with (²) originate from the ACRE research (Martin-Brelot et al., 2010; Musterd & Gritsai, 2013)

The model was simultaneously tested in urban and rural areas of Slovenia – a country in the heart of Europe exhibiting a range of socio-economic and political orientations and which is thus considered as a good proxy for the European context. Although a post-socialist country, it represents a successful transformation to a market economy (Stanilov, 2007). The creative class in Slovenia represents 38% of the workforce (Kozina, 2018), comparable with countries in Northern and Western Europe (see Andersen & Lorenzen, 2005; Boschma & Fritsch, 2009; Clifton, 2008) where the labour force share of creatives typically accounts for over one third (Florida, 2002). Additionally, as a small country Slovenia enabled wide systematic sampling of creative workers representing all urban and rural places in the country (Kozina, 2022). A unique primary dataset was generated via a postal questionnaire which operationalised the indicators shown in Table 1 (the practicalities of its design and administration are described in detail in Tiran et al., 2020).³ Based on the Slovenian Standard Occupational Classification (SKP-V2), which is in turn derived from the International Standard Occupational Classification (ISCO-88), the following codes at the three-digit level were applied to define the target population as members of the creative class (knowledge using, applying and knowledge creating occupations): 1, 211–214, 221–223, 231–235, 241–245, 247, 31, 32, 341–347, 521.⁴ Respondents were persons in employment, and self-employed persons who have compulsory social insurance, regardless of whether they work full-time or part-time.

The systematic random sample was prepared by the Statistical Office of the Republic of Slovenia. The size of the initial sample was limited to 20,000 respondents, representing 6.5% of the creative workers in Slovenia. The number of returned and valid questionnaires was 4,823, which represents 24.1% of the original sample or 1.6% of the total population. The comparison of demographic variables (Age, Gender, Level of Education) proved a good match between the final sample and the total population (see Appendix 1), which shows a high degree of generalisation of our dataset. The questionnaire primarily asked respondents: “What are the reasons for your current residence in this place?”. In order to persuade people to reflect more deeply and prioritise among the most important ones, the respondents had to select up to three reasons that were the most important for them.

The questionnaire asked about the length of stay in the current place. The vast majority of respondents had lived in their current place for more than a decade and also do not intend to move elsewhere in the foreseeable future (see Appendix 2), which is in line with previous findings across Europe (Martin-Brelot et al., 2010). Current and expected length of stay varies little by type of place (urban, intermediate, rural), giving us a good reason to see creatives not as “mobile assets”, and to move beyond the instrumental attraction and retention logic of their locational preferences.

In order to examine differences in the locational values of creative workers in urban and rural areas of Slovenia, we employed the Eurostat’s Degree of urbanisation (DEGURBA), which uses population density as a measure of urbanisation (Figure 1). DEGURBA is a territorial typology used for classifying Local Administrative Units at level 2 (LAU-2 or municipalities).⁵ It is used by a wide range of policymakers, researchers, national administrations and international organisations (OECD, UN, EU) as an explanatory variable in analyses of labour markets, education, living conditions, welfare and tourism (Dijkstra & Poelman, 2014). The methodology classifies municipalities based on geographical contiguity and a minimum population threshold applied to 1 km² population grid cells, thus avoiding distortions caused by using municipalities varying in size and/or shape. DEGURBA distinguishes between densely-populated areas (cities, large urban areas), intermediate density areas (towns and suburbs, small urban areas) and thinly-populated areas (rural areas).⁶ Although the definition of an urban-rural continuum is based only on the population grid and neglects other criteria (e.g. concentration of jobs, services, built-up areas or night lights), its advantage lies in international comparability across Europe. At present there is no other harmonised method for delineating between urban and rural areas on a global level (European Commission, 2020).

Figure 1. Urban (densely populated) and non-urban (intermediate and thinly populated) areas in the Degree of urbanisation of Europe and Slovenia (DEGURBA). Sources: EFGS, JRC, Eurostat, LandScan, REGIO-GIS.

Among 212 Slovene municipalities, 171 are classified as thinly-populated (81%), 39 as intermediate (18%), while only 2 municipalities (1%) are classified as densely populated (Figure 1). These are Ljubljana (286,000) as the capital city and Maribor (112,000), the only two municipalities in Slovenia with more than 100,000 inhabitants. Our sample of 4,823 creative workers corresponds well with the national creative class distribution: 26.7% of respondents were from densely-populated, 33.8% from intermediate and 39.6% from sparsely-populated areas. The national creative class distribution is 25.3% in densely-populated, 33.6% in intermediate and 41.1% in sparsely-populated areas.

To analyse the differences in locational values of creative workers between urban and rural areas in Slovenia, we performed a bivariate association analysis and built a set of binomial logistic regression models that also helped us to control for the length of stay and demographic variables. In performing the bivariate association analysis, we applied a χ² test and post-hoc test to search for standardised residuals (z-scores) < –1.96 and > +1.96, p < .05. Z-scores that have a value above or below this interval mean that the cell is overrepresented or underrepresented in the actual sample, compared to the expected frequency. In a nutshell, the χ² test shows whether the association between two categorical variables is statistically significant or not, while the post-hoc test helps to break down a significant χ² test among the different degrees of urbanisation (densely-populated, intermediate, sparsely-populated) (see Sharpe, 2015). By using binomial logistic regression, we predicted the probability that an observation falls into the urban or rural category of a dichotomous dependent variable. Different iterations of modelling revealed that the highest share of explained variance exists when comparing urban (densely populated) and non-urban (intermediate and thinly populated) areas. For this reason, it makes sense to compare urban and non-urban areas, although we recognise that there are some significant results for intermediate areas as well. More detailed comparisons and analyses can be the subject of future research.

Results and discussion

Analysis of the differences between locational values of creative workers in urban and rural areas (see Appendix 3) reveals some key findings. First, we performed bivariate association analysis to individually examine indicators of locational values (Table 2). Results show that the type of place is very important when considering locational values of creative workers; 31 out of 37 indicators (84%) are significantly different when comparing the urban and the rural type. This finding shows the applied urban-rural framework has significant merit and serves as a solid base to distinguish between the urban and the rural type of creative systems. The most divergent values in the two types include those linked with basic human functions of work, education and services which are valued much more in urban places than in rural places. Conversely, certain residential preferences (living costs, personal networks and closeness to nature) are valued more in rural places. Here, we can observe two distinct value systems of creative workers connected to two types of places.

Table 2. The association analysis results (χ² test and post-hoc test) between indicators of locational values and the degree of urbanisation (DEGURBA).

Basic human functions		Indicators of locational values	χ^2	Standardised residuals (z-scores)
				Densely populated areas	Intermediate density areas	Thinly populated areas
Residence	Living costs	Living costs	16.898*	−3.35	0.68	2,12
		Housing costs	19.396*	−3.43	0.38	2,46
	Housing	Housing availability	6.548	0.05	1.81	−1,72
		Housing affordability	10.996*	−2.11	2.36	−0,45
		Housing equipment	31.349*	4.05	0.24	−3,55
		Housing quality	8.213*	2.22	−0.27	−1,58
	Personal networks	Born here	27.468*	−3.34	−0.92	3,59
		Family lives here	36.828*	−4.10	0.01	3,36
		Friends live here	1.811	0.84	0.27	−0,94
		Studied here	18.553*	3.51	−2.40	−0,67
		Local community	9.426*	−2.29	−0.13	2,00
		Housing inheritance	1.739	−0.70	−0.40	0,95
	Physical environment	Private open space	97.034*	−7.25	1.13	4,90
		Quality of environment	54.815*	−4.72	−0.56	4,39
		Diversity of built environment	26.446*	4.36	−1.35	−2,33
		Weather/climate	51.832*	−5.63	0.41	4,24
		Proximity to nature	91.119*	−6.34	2.14	3,23
	Safety	Personal safety	4.054	−1.37	−0.12	1,23
		Safe for children	41.648*	−4.74	1.10	2,88
	Tolerance and openness	Openness to different people	26.132*	4.34	−1.60	−2,08
Work		Job	144.914*	9,10	−2.74	−4.94
		Higher wage	53.044*	6,18	−2.60	−2.68
		Working conditions	57.854*	6,39	−2.22	−3.19
Education		Secondary education centre	31.525*	4,62	−0.82	−3.03
		Higher education centre	51.387*	5,92	−1.11	−3.83
		Educated population	12.668*	3,03	−1.54	−1.06
		Good schools/faculties	52.533*	6,12	−1.89	−3.28
Services		Publicly owned services	104.069*	7,73	−0.54	−5.85
		Privately owned services	24.833*	3,79	−0.08	−3.04
		Higher quality of services	93.224*	8,20	−3.03	−3.94
Leisure		Diversity of leisure	82.202*	7,68	−3.02	−3.52
		Recreational activities	12.820*	0,52	2.28	−2.54
		Cultural activities	107.462*	8,73	−3.08	−4.32
Transport		Regional centres accessibility	15.992*	−2,83	2.75	−0.21
		Road accessibility	2.906	−0,83	1.32	−0.54
		Public transport accessibility	78.530*	7,12	−0.94	−4.98
		Internet accessibility	3.385	1,52	−0.34	−0.93

Note: df = 2, * p < .05, N = 4658.

standardised residuals (z-score) > +1.96.

standardised residuals (z-score) < −1.96.

In the urban type, locational values pertaining to agglomeration characteristics seem more important: creative workers value access to specialised and public services, cultural and learning facilities (universities, schools) and especially the proximity of jobs, better working conditions and a diverse built environment. These values are in line with a predominant view that conditions for creating or stimulating creative economies in the context of a global world depend on urban history and the economic tradition of the territory (Pareja-Eastaway & Pradel i Miquel, 2015). Furthermore, it has been established that for creative people in urban agglomerations the primary driver is so-called “urban buzz” with high socioeconomic diversity as a result of density, connectivity, and proximity advantages among the heterogeneous groups and institutions (Arribas-Bel et al., 2016). Our findings are in line with empirical evidence from other countries where city centres have been found to be the location for many workers, especially from the creative industries (Coll-Martínez et al., 2019).

The rural type on the contrary has more emphasis on other locational values that create an entirely different value system; proximity to nature, close connections with family and the community, and the perceptions of a clean and open space. Place attachment is also strong in the sense that in rural places the indicator of “being born here” and “family lives here” is above average. More practical values include lower living/housing costs and climate conditions. Values connected to diversity, frequent interactions or proximity to economic or support institutions are less important in rural areas.

In addition to the divergences between the two types of places, it is useful to highlight any converging locational values. Certain indicators do not vary and are equally important regardless of the type of place. Infrastructure is one such important aspect, meaning factors such as the availability of stable internet connection and good road accessibility, are universal at least across the urban-rural dichotomy. Some personal aspirations are also equally important in different spatial contexts: closeness of friends, personal safety and housing quality are basic human needs. These values present the very basis of understanding personal and value-based locational preferences of creative workers regardless of the place type.

Bivariate association analysis uncovers two distinct types of place-specific values (i.e. urban and rural type) and is consistent with arguments that socio-spatial variables linked to creative processes vary from one context to another (Meusburger, 2009; Uršič & Imai, 2020). By doing this, our findings upgrade the results of Martin-Brelot et al. (2010) and Musterd and Gritsai (2013) which studied the locational preferences of creatives in a sample of large urban regions only, and without distinguishing different types of places within them. Our comparative statistical model clearly exposed the relevance of the urban/rural context in understanding locational preferences of creative workers. It also expands the relationship between creative workers and their environment beyond only economic or job-related functions and introduces the vital “everyday” aspects of human behaviour (leisure, transportation, education, etc.), to answer the seminal question of “Why here and not elsewhere” (Howkins, 2009, p. 43). In addition to previously studied “hard” and “soft” factors and personal networks, we also reveal the importance of the physical environment related to proximity of nature, quality of environment and availability of private open spaces. Although these locational values are significantly more important in rural areas, they also rank highly in urban areas, which prompts new research questions for spatial reorganisation induced by the Covid-19 crisis – for example, will these values now become even stronger drivers in the future, and what second order effects (around affordability for example) might this have for rural areas? Moreover, our analysis develops the work of Gülümser et al. (2010) who have shown that the application of urban-centric analyses to rural areas may lead to misrepresentation or misunderstanding of rural creative capacity.

Second, we performed a binomial logistic regression analysis to better understand the interdependency between the type of place (dependent variable urban/rural) and locational values. The model calculates odds ratios (OR) as a measure of association where OR describes how much more likely a locational value is to occur in either rural or urban places (Model 1). We can additionally control for demographic indicators (Model 2) to test whether locational values are related to individual factors (the length of stay, gender, age, education and household type) or if they are truly place-specific. The logistic regression model was statistically significant, χ²(37) = 871.38, p < .001. The model explained 25% (Nagelkerke R²) of the variance in urban/rural dichotomy and correctly classified 77.8% of cases. In a second model, we controlled for the length of stay and demographic variables of gender, age, education and household type. This logistic regression model was also statistically significant, χ²(48) = 1017.91, p = < .001, explaining 30% (Nagelkerke R²) of variance and improving case prediction to 79.6% (Table 3).

Table 3. Binomial logistic regression results (dependent variable is the level of urbanisation); 0 = rural (intermediate and thinly populated areas, 1 = urban (densely populated areas)).

Indicators of locational values	Model 1			Model 2
	B (SE)	OR	95% CI	B (SE)	OR	95% CI
Constant	−0.76 (0.20)***			−1.06 (0.82)
Living costs	−0.76 (0.23)**	0.47	0.30–0.74	−0.60 (0.24)*	0.55	0.34–0.88
Housing costs	−0.61 (0.17)***	0.54	0.39–0.75	−0.58 (0.19)**	0.56	0.39–0.81
Housing availability	−0.08 (0.18)	0.93	0.65–1.31	−0.04 (0.19)	0.96	0.66–1.39
Housing affordability	−0.37 (0.16)*	0.69	0.50–0.95	−0.26 (0.17)	0.77	0.55–1.09
Housing equipment	0.44 (0.14)**	1.55	1.17–2.05	0.45 (0.15)**	1.56	1.16–2.11
Housing quality	0.40 (0.13)**	1.49	1.15–1.94	0.34 (0.14)*	1.40	1.06–1.86
Born here	−0.36 (0.15)*	0.70	0.52–0.94	−0.28 (0.16)	0.76	0.55–1.04
Family lives here	−0.46 (0.11)***	0.63	0.51–0.79	−0.38 (0.12)**	0.68	0.54–0.87
Friends live here	0.13 (0.14)	1.14	0.86–1.51	0.26 (0.15)	1.29	0.96–1.75
Studied here	1.24 (0.58)*	3.44	1.10–10.76	1.52 (0.61)*	4.58	1.39–15.09
Local community	−0.73 (0.32)*	0.48	0.26–0.91	−0.68 (0.33)*	0.51	0.27–0.97
Housing inheritance	−0.02 (0.13)	0.98	0.75–1.27	0.03 (0.14)	1.03	0.78–1.37
Private open space	−0.69 (0.12)***	0.50	0.40–0.64	−0.65 (0.13)***	0.52	0.41–0.67
Quality of environment	−0.28 (0.11)*	0.76	0.61–0.93	−0.25 (0.12)*	0.78	0.62–0.98
Diversity of built environment	1.28 (0.38)**	3.60	1.71–7.60	1.30 (0.40)**	3.68	1.70–7.99
Weather/climate	−2.07 (0.36)***	0.13	0.06–0.25	−2.10 (0.37)***	0.12	0.06–0.25
Proximity to nature	−0.57 (0.10)***	0.56	0.46–0.69	−0.51 (0.11)***	0.60	0.49–0.75
Personal safety	0.03 (0.12)	1.03	0.82–1.30	0.05 (0.13)	1.05	0.82–1.35
Safe for children	−0.47 (0.11)***	0.62	0.50–0.77	−0.27 (0.12)*	0.76	0.60–0.97
Openness to different people	0.54 (0.28)	1.72	0.98–3.00	0.58 (0.30)	1.79	0.99–3.22
Job	0.50 (0.11)***	1.66	1.35–2.04	0.56 (0.11)***	1.75	1.40–2.19
Higher wage	0.92 (0.27)**	2.50	1.47–4.25	1.21 (0.30)***	3.36	1.87–6.03
Working conditions	0.61 (0.20)**	1.83	1.25–2.69	0.53 (0.21)*	1.69	1.12–2.56
Secondary education centre	0.85 (0.35)*	2.34	1.18–4.67	1.09 (0.36)**	2.99	1.47–6.09
Higher education centre	1.25 (0.40)**	3.49	1.60–7.60	1.03 (0.41)*	2.80	1.24–6.29
Educated population	1.83 (1.16)	6.26	0.65–60.36	2.06 (1.17)	7.84	0.79–77.88
Good schools/faculties	1.26 (0.28)***	3.51	2.02–6.09	1.42 (0.30)***	4.12	2.31–7.36
Publicly owned services	0.57 (0.13)***	1.77	1.38–2.27	0.68 (0.14)***	1.97	1.51–2.58
Privately owned services	0.25 (0.16)	1.28	0.93–1.77	0.28 (0.17)	1.33	0.95–1.86
Higher quality of services	1.59 (0.33)***	4.90	2.55–9.43	1.39 (0.34)***	4.03	2.05–7.92
Diversity of leisure	1.18 (0.26)***	3.26	1.98–5.38	1.13 (0.27)***	3.11	1.84–5.24
Recreational activities	0.16 (0.15)	1.17	0.87–1.57	0.11 (0.16)	1.11	0.81–1.52
Cultural activities	1.18 (0.21)***	3.25	2.14–4.94	1.05 (0.22)***	2.85	1.84–4.42
Regional centres accessibility	−0.89 (0.29)*	0.41	0.23–0.73	−0.87 (0.30)**	0.42	0.23–0.76
Road accessibility	−0.28 (0.16)	0.75	0.55–1.04	−0.25 (0.17)	0.78	0.56–1.09
Public transport accessibility	1.18 (0.20)***	3.25	2.19–4.83	1.27 (0.21)***	3.56	2.34–5.42
Internet accessibility	0.21 (0.22)	1.23	0.81–1.89	0.21 (0.23)	1.23	0.79–1.93
Gender (reference = male)
Female				0.00 (0.08)	1.00	0.85–1.17
Age in years				0.02 (0.00)***	1.02	1.01–1.03
Household or family (reference = one-member household)
Husband and wife/cohabiting couple without children				0.29 (0.16)	1.34	0.99–1.82
Husband and wife/cohabiting couple with children or lone father/mother and children				−0.27 (0.13)*	0.76	0.59–0.98
Two family units				−0.81 (0.19)***	0.45	0.31–0.65
Other				0.02 (0.21)	1.02	0.67–1.54
Education (reference = primary degree)
Secondary degree				−0.96 (0.77)	0.38	0.08–1.72
College degree				−1.18 (0.77)	0.31	0.07–1.38
University degree				−0.49 (0.77)	0.61	0.14–2.75
Master's degree or PhD				0.05 (0.77)	1.05	0.23–4.76
Length of stay in the current settlement (reference = less than 10 years)
More than 10 years				0.05 (0.10)	1.05	0.85–1.29
N	4658			4471
-2 Log-likelihood	4530.84			4150.89
Model χ^2(df)	871.38 (37)***			1017.91 (48)***
Cox & Snell R^2	0.17			0.20
Nagelkerke R^2	0.25			0.30
Classification accuracy	77.8%			79.6%

* = p < .05; ** = p < .01; *** = p < .001; B = regression coefficient; SE = standard error; OR = odds ratio; 95% CI = 95% confidence interval.

The results of Model 1 are complementary to our findings from the association analysis. However, they also offer a better understanding of the relationship between the type of place and locational values. Odds ratios tell us that in urban areas creative workers are much more likely to emphasise traditional agglomeration-based values such as higher quality of services (OR = 4.90), diversity of built environment (OR = 3.60), studied here (OR = 3.44), good schools/faculties (3.51), cultural activities (3.25), and public transport accessibility (3.25). On the other hand, locational values like living costs (OR = 0.47), local community (OR = 0.48), weather/climate (OR = 0.13), regional centre accessibility (OR = 0.41) are more likely to be expressed by rural creative workers.

There are no significant spatial implications when it comes to gender, education and length of stay in the current place in Model 2. This implies that the geography matters more than individual demography-related factors and reinforces the idea that there are two distinct values systems of creative workers, dependent on the type of place and independent of their internal demographic structure. Age is statistically very significant, however the effect itself is almost negligible (OR = 1.02), suggesting the need for further research into specific age subgroups to unpack this overall result for any practical significance (Kirk, 1996). The combined results give us strong evidence of an overemphasis in the European context on the importance of attracting and retaining creative workers, both in academic debate and in policy-making. When controlling for demographic variables, “housing affordability” becomes statistically insignificant, signalling that creatives with children and living in two-family units have their housing arrangements resolved, rendering the location value of housing affordability less relevant. The locational value of “being born here” also becomes statistically insignificant after introducing controlling variables. To better understand the dynamics underpinning these relationships we need to explore further the values of creative workers at an individual level, likely by conducting more qualitative-based research.

Locational values (at least in part) indicate different creative systems in which there are various aspirations, motivations and attachments of creatives to certain places. The focus of understanding spatial patterns of creativity should shift beyond the instrumental attraction / retention logic of creative capital towards emphasising place-specific values. Our results confirm the soundness of cultural ecology thinking in order to augment traditional locational choice models. By bringing in the notion of cultural capabilities of specific types of places (Gross & Wilson, 2020), we can make the locational models of creative workers more holistic and better adapted to the European context of lower mobility. This cultural ecology thinking assumes that creative workers make their locational decisions based on shared cultural, social values and not merely on the aggregation of their individual preferences (see Dietz et al., 2005).

The above-mentioned narrative has additional practical implications. Urban and rural places offer specific cultural capabilities, which should be considered by policy makers. For instance, it makes little sense to emphasise cultural diversity or higher quality of services in a thinly populated countryside, since there is a completely different set of values in place. The values of rural creative workers are more about promoting environmental amenities and the “slow” way of life, which should be acknowledged by context-sensitive policies to achieve sustainable rural development (Janc et al., 2020).

Conclusions

The overall aim of this research is to operationalise and test a holistic locational values system for creative workers that is better tailored to the European context and considers the multiple interactions between creative workers and places. To do this we built on previous locational work by applying the seminal concept of social geography as articulated by the German school of social geography, which emphasises the everyday aspects of human behaviour related to basic human functions. We employed an urban-rural dichotomy as the primary test of place-based divergence. Statistical analysis showed that the type of place does indeed matter, as it determines two distinct types of locational values systems. Urban creative workers emphasise values related to their work function, access to education and to services. In contrast, rural creative workers stress residential functions, particularly the cost of living, the benefits of the physical environment (weather, nature, open spaces) and personal ties such as proximity to family members and involvement in the local community. Our findings illustrate the nuances and heterogeneity of creative workers according to the type of place, which should be considered by future researchers and policy makers. Spatial context should not only be considered as a background factor, but as important – if not crucial – in developing the creative and cultural economy within regions and communities.

This narrative is corroborated by the binomial logistic regression model, which shows that demographic variables do not really influence or change the results of our research. Furthermore, it also serves to negate the prevailing attraction/retention logic, since the analysis shows that the time spent in a given urban or rural locality does not significantly impact the value system of creative workers. To put it bluntly, we find that geography plays a larger role than individual demographic factors in explaining locational values of creative workers. Herein lies the theoretical and practical implication of our study: future research on the distribution of creative workers should always consider the specificity of places. In addition to the urban vs. rural dichotomy presented here, other contrasting types of places should be explored, such as centre vs. periphery, metropolis vs. small or medium-sized city, or even in different socio-economic contexts altogether (post-socialist vs. traditional capitalist, global North vs. global South). There is a need for a more in-depth examination of place types that go beyond urban and rural, also in relation to the practical implications of this research, which include the need for policies that are place-sensitive. Thus, while we remain largely agnostic regarding specific sets of policy recommendations – and indeed whether one value system is in any way “better” than another – the key issue here is that place-based policy interventions should better consider place-specific contexts as a general principle.

Seeing creative workers as a social group with specific “needs” that bind them to their environment, we can also recognise the potential of cultural ecology thinking. Our research has confirmed that place does matter and that it shapes creative ecosystems not just through an occupational perspective, but is more complex and includes other dimensions, such as leisure, physical space, personal networks, etc. (de Bernard et al., 2022). A social geography framework merged with cultural ecology thinking overcomes the shortcomings of previous studies by going beyond partial attempts to understand creatives’ locational patterns. Instead, this framework offers to holistically measure interdependencies between creative people and places through everyday aspects of human behaviour (residence, work, education, services, leisure and transport). This puts into question the fundamental causal logic that needs further exploration: do creatives live in certain places because the environment is more suitable to their needs and aspirations or do certain groups of creative workers shape their environment according to their shared values? In the latter case, it is less important to study amenities of places and more important to research the (intrinsic) values of creative workers clustering in those places.

There are limitations to our research. While we argue that Slovenia represents a good proxy of the European context in general, we should also acknowledge that it typically offers relatively good access to the natural environment even from urban areas (Brezina et al., 2021). So, applying our methodology in different urban hierarchies, places with significant post-industrial economies and so on could offer further insights. There is also a growing recognition of the heterogeneity of rural and urban areas, so our criterion of population density as a proxy for urban/rural comparison could be expanded to include other, more nuanced typologies of place-based values, for example, marginalised, peri-urban rural areas, etc. One potential methodological limitation is the absence of the control group (i.e. other “non-creative” labour groups). Thus, we cannot claim the identified urban and rural sets of values can specifically be attributed only to creative workers in contrast to those in other types of occupation. Future research could shed more light on some contradictory discussions, where authors clearly differentiate between “creative” and “non-creative” urban preferences (Bereitschaft, 2017; Van Heerden & Bontje, 2014), as well as between different creative subgroups (e.g. creative core, creative professionals, bohemians).

Such questions are, of course, particularly relevant at the present time as the fallout of the Covid-19 crisis has precipitated an ongoing reassessment of the relative balance of positive and negative externalities of urban life (Florida et al., 2020). For example, increased (semi) rural living and access to restorative open space may remain attractive. If telecommuting becomes the norm for creative workers in the future, we might expect a shift in more work-related values, such as the availability of space for working from home or the relevance of daily commuting infrastructure. It may be fruitful to repeat the data-gathering exercise in the future; while deep-seated value systems are unlikely to change significantly in a relatively short space of time, it is likely that urban-rural spatial sorting may shift after covid, in combination with increased digitalisation and sustainability pressures (Asheim et al., 2023). On the one hand it may be easier for rural creatives to remain nearer their home communities, while conversely also more practical for urban creatives to relocate there. Parallel migration studies will be required to untangle these effects. As a practical intervention, localised coworking hubs offer a potential means to build cohesion particularly in rural areas if these places begin to attract incoming residents from urban areas post covid.

Further research should aim to capture these possible changes. Moreover, it is likely that more qualitative (including ethnographic and psychological) research will be required to unravel the motivations and beliefs that underlie locational values at the individual level, and how they vary across space. These are important considerations if we are to design more nuanced, place-based development policies that successfully build on existing local capabilities and avoid the econo-centric orthodoxy.

Disclosure statement

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

Additional information

Funding

This work was supported by the Slovenian Research and Innovation Agency: [grant number P6-0101].

Notes on contributors

Jani Kozina

Jani Kozina, PhD, is a Research Associate at the Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU) and Assistant Professor of Cultural and Economic Geography at the Postgraduate School ZRC SAZU. He is interested in the social and cultural manifestations of industrial and creative cities.

Nick Clifton

Nick Clifton, PhD, is Professor of Economic Geography and Regional Development at Cardiff Metropolitan University. His main research interests are in regional development, small business and entrepreneurship, networks, business strategy, innovation and creativity.

David Bole

David Bole, PhD, is Principal Research Associate and Head of the Department of Human Geography at the Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU). He is also Assistant Professor of Urban Geography at the Postgraduate School ZRC SAZU. His main research interests lie at the intersection of economic and urban geography.

Notes

1 Favell actually highlights a lack of transience as a strength in producing a “patchwork” of cultural diversity and distinctiveness.

2 ACRE: Accommodating Creative Knowledge – Competitiveness of European Metropolitan Regions within the Enlarged Union, see Musterd and Gritsai (2013) for a detailed description.

3 For the avoidance of doubt the survey was undertaken prior to the Covid pandemic, declared in early 2020. Given the central aim of the paper – ie to better understand the long term and bi-directional interaction between people and places – this is likely preferable to data collection during a period of extreme crisis or its immediate aftermath. We return to this point in the concluding section.

4 The same classification was used by Andersen and Lorenzen (2005), Clifton (2008), Boschma and Fritsch (2009) and Clifton et al. (2013) in the context of Denmark, England and Wales, Finland, Germany, the Netherlands, Norway and Sweden. Our approach is therefore comparable to these studies, which is the reason why we used SKP-V2 (based on ISCO-88) instead of SKP-08 (based on ISCO-08). The only difference is that we also took into consideration Customs, tax and related government associate professionals (344) which have similar characteristics compared to the other occupational groups included within Other associate professionals (34). This difference is negligible as their share of the bulk sample was only 1.7%.

5 https://ec.europa.eu/eurostat/web/nuts/local-administrative-units

6 In densely-populated areas at least 50% of the population lives in high-density clusters (contiguous grid cells of 1 km² with a density of at least 1,500 inhabitants per km2 and a minimum population of 50,000), while thinly-populated areas have more than 50% of the population living in rural grid cells (grid cells outside urban clusters of contiguous grid cells of 1 km² with a density of at least 300 inhabitants per km² and a minimum population of 5,000). Intermediate density areas have less than 50% of the population in rural grid cells and less than 50% in high-density clusters.