Introduction

Coworking spaces are mushrooming worldwide as a new type of corporate real-estate market that is making traditional office spaces obsolete (Echeverri et al., 2021). The rapid proliferation of the digital economy (Ayres & Williams, 2004), creative economy (Florida, 2004), shared economy (Wu & Zhi, 2016), and gig economy (Stewart & Stanford, 2017) into global economic space gave birth to the innovative system of spaces organising digital labour (Dorschel, 2022) that is popularly known as coworking spaces. Undoubtedly, coworking is a promising and prosperous segment of corporate real-estate at present estimated to offer a workplace to 5 million digital workers at 42 thousand locations worldwide by 2024, with a 158 percent increase since 2020 (Coworking Resources, 2020).

A decade ago, the first modern co-working space was introduced at Mohan Estate in Delhi, NCR, in the year 2013 (Raju, 2020).We chose Delhi to understand the locational pattern of co-working spaces, because Delhi is the second largest populated city in the world where the commercial real-estate market is highly expensive (Business Line, 2021). Apart from that, the city traffic is highly congested in comparison to many Asian and European cities (TomTom International, 2021). India (as well as the Indian capital Delhi) is one of the countries where information cost is cheapest (Hindustan Times, 2022) that is an integral locational determinant for coworking spaces (Mariotti, 2015). However, there is a lack of studies that explore the geography of coworking spaces in Delhi, hence this study is an attempt to fill the knowledge gap.

Therefore, using the global or Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) model (Zhou, 2019), and secondary data for 117 coworking locations distributed over 280 municipal wards of NCT-Delhi; this paper addresses two main questions: (1) Where are the main locations of co-working spaces in Delhi?

(2) To what extent, are these co-working spaces associated with other urban socio-economic, services and lifestyle factors?

Literature review

Sharing economy

Sharing Economy (SE) facilitates shared access to facilities and commodities in a peer-to-peer (P2P) mode through digital platforms (Puschmann & Alt, 2016; Richardson, 2015). In a Shared economic model, individuals share access to a variety of assets and services (Lovic, 2020). Such kinds of access are usually convenient in nature (Eckhardt et al. 2019) which has expanded across a diverse range of products and services, including shared-lodging i.e. Airbnb (Zervas et al., 2017) and co-living (Mellner et al., 2021), shared-mobility i.e. Uber (Kim et al., 2018), food-deliveries (Tandon et al., 2021), shared-clothing (Eckhardt et al., 2019), food deliveries, non-bank payment providers, and co-working (Weijs-Perrée et al., 2019).

Many studies have indicated that the sharing economy has emerged because of the growing need to exploit the value of underutilised assets and provide them to individuals in need through a digital platform, resulting in accessibility to assets without owning them (Stephany, 2015).

Scholars have also argued that the sharing economy is highly promising because of its intention to use underutilised assets in a manner that can enhance competence, society, and sustainability (Heinrichs, 2013). Another study asserts that the sharing economy stimulates sustainable communities (Mi & Coffman, 2019). The sharing economy illustrates a radical change in the behaviour of consumers (Harmaala, 2015; Nerinckx, 2016) that lessens resource uses in general and encourages sustainable consumption of assets (Heinrichs, 2013; Heo, 2016) leading to creating opportunities of savings, income and employment generation (Fang et al., 2016), sustainable economic growth (Bonciu & Bâlgărm, 2016), and strengthening the social bonding and cohesion within urban communities (Kopnina, 2017; Laamanen et al., 2015; Munkøe, 2017). On the contrary, some studies argue that the sharing economy is merely a swift and innovative form of the market economy (Curtis & Lehner, 2019).

Links of new economic approaches with the coworking

The eminence of coworking has been growing exponentially over the years (Bouncken & Reuschl, 2018; Parrino, 2015) in the large metropolitan cities of the world, which could be explained by the advent of the creative economy (Waitt & Gibson, 2009), shared economy and digital economy; and the novel way of organising workers under it (Moriset & Malecki, 2009). Cities in both developed and develo** economies are rapidly espousing the coworking culture of the internet age (Fuzi, 2015; Wang & Loo, 2017). Coworking spaces could be termed as essentially shared workspaces in which the providers offer office amenities, and infrastructure on rent to the individuals who wish to avail benefits of freelancing together with like-minded folks (Arora et al., 2013; Capdevila, 2013). The desire for enhanced communication, seamless collaboration, and innovative design are the core reasons for producing contemporary coworking spaces (Khazanchi et al., 2018). Lessons from artists’ incubator projects reflect that coworking has advantage of economies of scope rather than economies of scale (Whitaker, 2021), it supports the thought of working alone together, and this has become possible only because of the rapidly growing number of distant workers, self-employed, and freelancers who are basically knowledge workers and dependent on high-speed communication and computation (Clifton et al., 2019). As of today, knowledge professionals have the freedom to choose their place and time of work due to the pervasive computing facilitated in modern coworking spaces (Armondi & Di Vita, 2017).

Digital economies have differently worked on different economic activities as far as their spatiality is concerned (Mariotti et al., 2017). The rise of co-working or work-from-anywhere culture has yielded higher geographic flexibility and productivity (Choudhury et al., 2021). It has been proving beneficial to start-up founders, minorities, females, free and open market supporters, and overseas entrepreneurs (Howell, 2022). Though evidence on the geographies of coworking spaces is mixed. On one hand, cloud computing and faster means of communication have dissociated knowledge workers from the static job-locations (Moriset & Malecki, 2009), even some scholars claim it as the end of geography (O’Brien, 1992) or the end of distance (Pons-Novell & Viladecans-Marsal, 2006). However, on the other hand, geographical clustering and branding of co-working locations are evident in most of the urban regions (Nathan et al., 2019) in which mobilities play a crucial role (Shearmur, 2021).

Coworking spaces

Coworking spaces are well known as the hubs of internet entrepreneurs (Wang & Loo, 2017), and researchers have been linking them to hackers (Pauline, 2005) while some others deem it a disruptive phenomenon in the urban real-estate market (Sutriadi & Fachryza, 2021). Authors claimed that such spaces encourage entrepreneurship even in sparse regions (Fuzi, 2015).

A literature review (Akhavan, 2021) of coworking spaces defined it as the ‘third place of work’ and reported that literature on coworking spaces belongs to multiple disciplines, including planning, geography, sociology, economics, and business management, among others; and in the past decade it covered three broader categories of analysis i) geography of coworking spaces, ii) communities and lifestyle of co-workers and iii) impact on coworking spaces on varying size of cities. The concept of third place (Oldenburg & Brissett, 1982) was initially introduced as public spaces that are different from the traditional dualism of places, namely home as the first place and workplace as the second place.

New coworking spaces are branded as the funky (informal) offices that are the epitome of the new economy, and designed to remove the monotony of the traditional office spaces (van Meel & Vos, 2001). Most of the authors agree that coworking spaces are vibrant, affordable, motivating and offer a workplace for professionals from different backgrounds to interact with each other while aspiring to knowledge sharing and co-creation (Fuzi, 2015; Parrino, 2015). Some countries are also conducting trials to explore the possibilities for public servants to work at co-working spaces (Houghton et al., 2018). Next, studies have suggested that locational determinants of manufacturing firms (Mariotti, 2015; Özdemir, 2002) could also be applied to service sector enterprises of which coworking spaces are an innovative example. Based on the above, locational determinants of coworking spaces might be assembled into 4 categories (Mariotti et al., 2021) including, (i) traditional locational factors, (ii) social, environmental and institutional context, (iii) policy framework and (iv) information cost. We used traditional locational factors, social, environmental, and institutional contexts to model the geography of coworking spaces in Delhi. We also note that the Startup India Initiative (Government of India, 2016) might be thought of as a supportive policy measure for coworking spaces in India that offers funding support and incentives, and industry-academia partnership and incubation. However, this policy is in its initial implementation stage and commenting on it will be hasty. The fourth location determinant is ineffective herein due to cheaper data tariffs.

Coworking spaces at global cities

An earlier study from Brisbane and Melbourne (Felton et al., 2010) recommended that unique supportive strategies and policies are required to sustain creative industries rooted in communication technologies at outer sub-urban locations, such policies and strategies should be different from those dealing with the creative industries operating in inner cities. Although these authors did not use the ‘coworking’ term.

In the case of Milan, authors found that coworking spaces are an urban product that is often developed in the vicinity of creative clusters, having a profound impact on the work, leisure, and culture of the city (Mariotti et al., 2017).

A study in Manhattan (Zhou, 2019) concluded that coworking locations are significantly correlated with transport links, random discoveries and events, and lifestyle-related amenities. The study further suggests that coworking spaces are highly clustered and agglomerated into mixed-use areas permitting uninterrupted access to urban amenities and resources.

An empirical study of coworking spaces in Bei**g (Huang et al., 2020) reported that these spaces are typically agglomerated near the concentration of creative and knowledge enterprises in high-density mixed-use areas, bank upon underutilised commercial spaces. Another study in Barcelona (Coll-Martínez & Méndez-Ortega, 2020) detected 4 key features of coworking spaces as follows:

  • The density of coworking spaces is higher in the central areas of the city where the probability of meeting with clients and vendors is much greater.

  • Coworking spaces are in proximity to urban amenities.

  • Coworking spaces are allied with distinct place-image.

  • Coworking spaces are coagglomerated with the creative industries inclusive of those epitomic and artificial knowledge based.

Coworking in India

Co-working is not entirely new to India, and it has been reintroduced and branded with new features, new design, and new amenities (Das et al., 2019). In India, co-working spaces are a sprouting concept that is in high demand because of their cost-effectiveness, and flexibility (Banerjee, 2021), it is also anticipated that employers would diminish permanent workspaces and would utilise co-working spaces to cut-down their expenses over traditional office spaces (Bhattacharyya & Nair, 2019). Vibrant and attractive cities including Bengaluru, Pune and Delhi have a conducive environment to develop creative classes that have drawn the attention of co-working space providers like European cities, i.e., London, Milan, Paris, and Berlin (Berbegal-Mirabent, 2021). At present, co-working spaces hold 3% of the market in India’s overall office inventory that is projected to attain a worth of approximately $2.2 billion and surpass an aggregated floor area of 50 million square ft by 2022–23 (Business Line, 2020). However, the co-living spaces in India are mostly available inside small-scale refurbished buildings that are an outcome of adaptive-reuse projects (Jain, 2020). Market analysts are optimistic that the coworking model will continue to rule even in the coming years (Chadha, 2021).

Also, hybrid models of working in the form of coworking spaces are in demand in the post COVID-19 period as businesses are now compelled to hunt for flexible spaces that might open new avenues for this innovative workplace (Ranjan, 2022). A recent study (Pacchi et al., 2022) suggests that local policymakers can play a vital role in assisting coworking spaces to encourage local employment and protracted place sustainability after COVID-19 ends up.

Materials and methods

Data

Only a few studies till date have explored the geography of co-working spaces in global cities, including Manhattan-New York (Zhou, 2019), Barcelona (Coll-Martínez & Méndez-Ortega, 2020), and Milan (Mariotti et al., 2017) among others. A pertinent study has modelled co-working spaces through Geographically Weighted Regression (GWR) using parameters, neighbourhood social atmosphere (population density, age, race, household income, education level), neighbourhood environment (medium house rent, proximity to park) transportation connection (subway stations, bus stops) Discovery and events (place of interest, theatres, museums), Life convenience (grocery, restaurants, coffee shops, fitness centres, drinking places), and Promoters of innovations (University and research institutions and creative enterprises),and co-working spaces as dependent variable (Zhou, 2019). Another study explored the locational pattern of co-working places through the identification of geographical clusters (Mariotti et al., 2017) while a recent study used the kernel density function to analyse agglomeration and co agglomeration of co-working spaces and creative industries (Coll-Martínez & Méndez-Ortega, 2020). As mentioned earlier, i) traditional locational factors, ii) social, environmental, and institutional context, iii) policy framework, and iv) information cost might determine the location of coworking spaces in an urban area. We used the first two determinants to determine the geographical distribution of coworking spaces in the study area. Whereas traditional locational factors consist of transportation connections and promoters of innovation that assure supply of a qualified workforce; social, environmental, and institutional contexts are reflected in neighbourhood social atmosphere, neighbourhood environment (medium house rent, proximity to park), transportation connection, discoveries and events, and life convenience.

Fine scale data at municipal ward level is not available for Delhi. So, based on data availability and a thorough review of the above mentioned studies, this study obtained data on dependent variables through Cofynd (https://cofynd.com/) which is a leading online portal in India to list co-working and co-living spaces with the tariffs, amenities, photographs, and locational info (FPJ Web Desk, 2021). Additionally, this study used a total of 16 parameters, including population density, median house rent, proximity to park, bus stops, theatres, museums, places of interest, grocery shops, restaurants, coffee shops, fitness centres, bars, universities & research institutions, and concentration of creative enterprises to understand the geographical variability of co-working spaces in Delhi (Table 1). Data collection for this study was carried out during January–February 2022, when the city had almost recovered from the COVID-19 pandemic. Many corporate real-estate providers claim their spaces as coworking spaces, but this study included only those who offer at least 10 essential amenities listed in Table 2. claims All independent variables were also scrutinised for multicollinearity that might affect the quality of estimators in regression models; we used multicollinearity variance inflation factor (VIF) to detect multicollinearity (Grekousis, 2020). No multicollinearity was present in the data as the VIF values for each of the independent variables were below 5 (Table 1).

Table 1 Data sources and VIF scores
Full size table
Table 2 Minimum available amenities at sampled co-working spaces
Full size table

Methods

GWR proved useful in detecting spatial variability of parameters to estimate an outcome variable. Researchers have used it frequently in detecting determinants of a spatial event in several cases (Aljoufie & Tiwari, 2021; Tiwari & Aljoufie, 2021; Zhang et al., 2019).

Study area

The National Capital Territory of Delhi (NCT) or Delhi is the capital of India with 16.8 million people (Registrar General & Census Comissioner of India, 2011). It is the second largest metropolitan city in the country, based on population size. Delhi is the largest commercial hub in northern India, and the seat of the union administration, national parliament, supreme court, and many other institutions of national importance.

Delhi is among the quickest growing states with an appealing real-estate market (IBEF, 2021) and an emerging hub of new economies in India (Chaudhry, 2021). The service sector is the sector where the largest number of workers are employed (ET Bureau, 2022), this is probably the reason why the size of coworking spaces in Delhi might be likely to increase. This sector contributes 84.59% of the total GDP in Delhi and more than 95% workers work in this sector (Planning Department Delhi, 2020).

Results

First, the study measured the individual effects of independent variables through OLS and GWR methods (Table 3) to determine which variables are retrainable for further models based on the significant p-values (< 0.01) and higher variance explained through R2 (> 0.20).

Table 3 p-value and R2 values for 16 variables
Full size table

The results of the initial regression models indicate that attractions, population density, and distance to park are not significant because of their higher p-values (p > 0.01), also eliminated estimators like bus stop, art galleries and museums from the models because of their lower R2 values.

Subsequently, final OLS and GWR models were developed using the remaining 9 variables.

Results of the regression model (Table 4) reveal that ‘density of bars’ is a significant parameter at the global level, and it is significant for 45.1% of the wards at the local level with a positive sign (Fig. 1). This sign is logical as bars are one of the integral factors in recent urban lifestyle for recreation (Karsten et al., 2015).

Table 4 OLS and GWR regression results with 9 variables
Full size table
Fig. 1
figure 1

Geographical distribution of bars by local t-scores and beta coefficients in the GWR model

Full size image

Additionally, ‘Median rents’ is also a significant estimator at the global level, and it is significant for 41.3% of the wards at the local level with a positive sign (Fig. 2). The sign is convincing as the wards consisting of co-working spaces are also associated with smart living locations (Baiardi, 2018) seeking higher rents.

Fig. 2
figure 2

Geographical distribution of median house rents by local t-scores and beta coefficients in the GWR model

Full size image

Moreover, ‘density of fitness centres’ is another significant variable at the global level, and it is significant for 59.6% of the wards at the local level with a positive sign (Fig. 3). The sign is obvious as a previous study found a positive association between fitness centres and coworking spaces (Gruenwald, 2020).

Fig. 3
figure 3

Geographical distribution of fitness centres by local t-scores and beta coefficients in the GWR model

Full size image

Furthermore, ‘density of metro stations’ is also a significant parameter at the global level, and it is significant for 28.6% of the wards at the local level with a positive sign (Fig. 4). This sign is obvious as earlier research found a positive correlation between fitness centres and coworking spaces (Zhou, 2019).

Fig. 4
figure 4

Geographical distribution of metro train stations by local t-scores and beta coefficients in the GWR model

Full size image

Whereas ‘density of restaurants’ is not a significant parameter at the global level while it is significant for 16.2% of the wards at the local level with a negative sign (Fig. 5). This sign seems logical as food delivery applications are becoming increasingly popular by which consumers can order food from restaurants that are located far from their workplace (Pandey et al., 2021).

Fig. 5
figure 5

Geographical distribution of restaurant densities by local t-scores and beta coefficients in the GWR model

Full size image

Also, ‘density of cinema’ is not a significant estimator at the global level while it is significant for 10.7% of the wards at the local level with a negative sign (Fig. 6). The sign is convincing as the majority of coworking locations offer big screens with mini cinema inside their premises (Morgan, 2020).

Fig. 6
figure 6

Geographical distribution of cinema densities by local t-scores and beta coefficients in the GWR model

Full size image

Similarly, ‘density of café’ is not a significant parameter at the global level while it is significant for 16.2% of the wards at the local level with a positive sign (Fig. 7). The sign is convincing as some coworking locations are positively correlated with the cafes in global cities (Zhou, 2019).

Fig. 7
figure 7

Geographical distribution of cafe densities by local t-scores and beta coefficients in the GWR model

Full size image

Likewise, ‘density of grocery shops’ was not a significant parameter at the global and local levels.

Finally, the presence of ‘creative enterprises’ was a significant factor in the location of coworking spaces. This is logical as many of the previous studies conclude that coworking spaces and creative enterprises are tangibly coagglomerated (Coll-Martínez & Méndez-Ortega, 2020).

Model diagnostic

A diagnosis for both the OLS and GWR regression models using the sum of the squared estimate of error (RSS or SEE), corrected Akaike information criterion (AICc), Adj. R2, and Log-likelihood indices. RSS measures the unexplained variance in a dataset used in a regression model (Hurvich & Tsai, 1993). In addition, AICc quantifies estimation errors, and marks the quality of the regression models for a dataset (Cavanaugh, 1997), a lower AICc score shows a better model fit, eventually (Oshan et al., 2019). Moreover, the log-likelihood function indicates the fitness of given data in the regression models, this criterion indicates the appropriateness of a variable intended for explaining an observed value (Ishiguro et al., 1997). Furthermore, Adj.R2 assesses the explanatory power of the linear regression models modified for the number of parameters used in the model (Miles, 2005).

The model diagnosis reveals that the RSS values for both models are lower in the GWR model (137.174) than the OLS model (164.370). Again, the AICc value is lower for the GWR model (659.801) than the OLS model (679.122). Whereas the adjusted R2 value was higher for the GWR model (0.485) in comparison to the OLS model (0.417); and the log-likelihood values were higher in the GWR model (− 302.640) than the OLS model (− 329.167). In brief, the GWR model outperformed the OLS model.

Discussion

This investigation into coworking space pointed out several associations between the coworking spaces and urban socio-economic, services and lifestyle factors. Results of the OLS model indicate that the density of bars, median house rent, fitness centres, and metro train stations have statistically significant correlations with the co-working spaces. GWR regression explained the spatial heterogeneity of those associations and provided some additional information, and found that restaurants, cinemas, and cafés also correlated with coworking spaces. The results of this study suggest that the geographical distribution of co-working spaces is not random in Delhi, which reinforces findings of some earlier studies in Milan (Mariotti et al., 2017) and New York (Zhou, 2019).

Additionally, this study also adds to the findings of Coll-Martínez and Méndez-Ortega (2020) who established that coworking spaces are agglomerated at central locations that have greater probabilities for connecting vendors and consumers of creative industries, and coagglomerated with the urban amenities and image of the places.

Moreover, coworking spaces in Delhi are coagglomerated with creative enterprises. It is in line with the findings of Coll-Martínez and Méndez-Ortega (2020) who reported similar findings in the case of Barcelona.

Furthermore, development of coworking spaces in India is still at a nascent stage, and demand for such spaces is expected to increase after the pandemic as the large-size caproate are also changing their work models and chasing for flexible working spaces, and embracing the culture of coworking spaces (Ranjan, 2022). Hence, the results of this study will be useful for government, urban planners, developers, and real-estate professionals who are interested in the planning and development of coworking spaces. In the post COVID-19 era, a localised policy on coworking spaces in Delhi may enhance place sustainability in the long term, and accelerate opportunities for local jobs.

Conclusions and policy implications

This study explored the location patterns of co-working spaces in Delhi, which is an emerging sub sector of commercial real estate. Using OLS and GWR models, this study supports the results of some previous studies (Mariotti et al., 2017; Zhou, 2019). This study observed a significant role of creative enterprises in the clustering of coworking spaces. It means coworking spaces are following the spatial pattern of creative enterprises (Table 4). Therefore, in the future, it might be beneficial to develop coworking spaces at locations close to creative industry clusters.

In the post COVID-19 period, coworking spaces might provide a hybrid model of flexible working or working with leisure and optimal safety in nearby places. Hence, in the circumstances after the pandemic, the government should consider the possibilities of a coherent work policy for Delhi so that aspiring workers and companies can opt for the coworking route. Also, after COVID-19, a localised policy intervention is essential to boost local employment and protracted place sustainability, like the recommendations made by authors after analysing coworking spaces in European cities (Pacchi et al., 2022). The government should also amalgamate incentivisation of coworking spaces in their policies so that coworking spaces can bloom and decongest the traffic scenario of Delhi. Further studies are recommended on the topic. Aforesaid policy measures could be added to the Startup India Initiative (Government of India, 2016).

The findings of this study might also be advantageous for urban planners, developers, and real-estate professionals who are working to develop the commercial real-estate sector. They should consider the location of creative industries while planning coworking spaces in the future.